Supramolecular Chemistry: From Molecules to Nanomaterials 0470746408, 9780470746400

Table of contents :
Concepts
Definition and Emergence of Supramolecular Chemistry
Supramolecular Interactions
Complementarity and Preorganization
The Thermodynamics of Molecular Recognition
Cooperativity and the Chelate, Macrocyclic and Cryptate Effects
Multivalency
Solvation Effects in Supramolecular Recognition
Competition Experiments
Supramolecular Information/Programming from a Boolean Perspective
Self-Assembly and Self-Organization
Introduction to Surfactant Self-Assembly
Reversible Covalent Bond Toolbox
Chirality
Techniques
Binding Constants and Their Measurement
NMR Spectroscopy in Solution
Luminescent Spectroscopy in Supramolecular Chemistry
Uses of Differential Sensing and Arraysin Chemical Analysis
Transmission Electron Microscopy (TEM)
Computational Techniques (DFT, MM, TD-DFT,PCM)
Isothermal Titration Calorimetry in Supramolecular Chemistry
Diffusion Ordered NMR Spectroscopy (DOSY)
Solid-State NMR Studies on Supramolecular Chemistry
Mass Spectrometry and Gas-Phase Chemistry of Supermolecules: A Primer
Supramolecular Electrochemistry
Circular Dichroism Spectroscopy
Transport Experiments in Membranes
Vesicles in Supramolecular Chemistry
Rheology
Langmuir–Blodgett Films
Affinity Capillary Electrophoresis as a Tool to Characterize Intermolecular Interactions
Ion Chromatography and Membrane Separations Using Macrocyclic Ligands
Dynamic Light Scattering in Supramolecular Materials Chemistry
Brewster Angle Microscopy
X-Ray Diffraction: Addressing Structural Complexity in Supramolecular Chemistry
Small-Angle X-ray Scattering (SAXS) and Wide-Angle X-ray Scattering (WAXS) of Supramolecular Assemblies
Atomic Force Microscopy (AFM)
Scanning Electron Microscopy
Scanning Near-Field Optical Microscopy (SNOM)
Scanning Tunneling Microscopy (STM)
Molecular recognition
Crown and Lariat Ethers
Azacycloalkanes and Azacyclophanes
Hetero-Crown Ethers—Synthesis and Metal-Binding Properties of Macrocyclic Ligands Bearing Group 16 (S, Se, Te) Donor Atoms
Cryptands and Spherands
Schiff Base and Reduced Schiff Base Ligands
Calixarenes in Molecular Recognition
Cyclotriveratrylene and Cryptophanes
Carcerands and Hemicarcerands
Cyclodextrins: From Nature to Nanotechnology
Cucurbituril Receptors and Drug Delivery
Podands
Porphyrins and Expanded Porphyrins as Receptors
Supramolecular Phthalocyanine-Based Systems
Guanidinium-Based Receptors for Oxoanions
Anion Receptors Containing Heterocyclic Rings
Amide and Urea-Based Receptors
Synthetic Peptide-Based Receptors
Biological Small Molecules as Receptors
Receptors for Nucleotides
Receptors for Zwitterionic Species
Ion-Pair Receptors
Metal Complexes as Receptors
Hydrogen-Bonding Receptors for Molecular Guests
Boronic Acid-Based Receptors
Self-processes
Template Strategies in Self-Assembly
Self-Assembly of Coordination Compounds: Design Principles
Self-Assembly of Coordination Chains and Helices
Self-Assembly of Coordination Cages and Spheres
Self-Assembly of Organic Supramolecular Capsules
Self-Assembly of Supramolecular Wires
Synthesis of Supramolecular Nanotubes
Organic Foldamers and Helices
Self-Assembly of Macromolecular Threaded Systems
Self-Assembled Links: Catenanes
Templated Synthesis of Knots and Ravels
Self-Assembly of Nucleic Acids
Viruses as Self-Assembled Templates
Peptide Self-Assembly
Supramolecular devices
Photoprocesses of Relevance to Supramolecular Chemistry
Photoinduced Electron Transfer Processes in Biological and Artificial Supramolecules
Molecular Devices: Energy Transfer
Molecular Devices: Molecular Machinery
Molecular Logic Gates
Single-Molecule Electronics
Molecular Redox Sensors
Ion-Selective Electrodes With Ionophore-DopedSensing Membranes
Colorimetric Sensors
Luminescent Sensing
Photoswitching Materials
Supramolecular Chemistry in Medicine
Magnetic Resonance Imaging Contrast Agents
Supramolecular Chemistry for Organic Photovoltaics
Supramolecular materials chemistry
Crystal Engineering
Concepts and Nomenclature in ChemicalCrystallography
Crystal Structure Prediction
Noncovalent Interactions in Crystals
The Cambridge Structural Database System and Its Applications in Supramolecular Chemistry and Materials Design
Cocrystals: Synthesis, Structure, and Applications
Polymorphism: Fundamentals and Applications
Mechanical Preparation of Crystalline Materials. An Oxymoron?
Physico-Chemical Aspects of Inclusion Compounds
Clathrate Hydrates
Synthetic Clathrate Systems
Network and Graph Set Analysis
Coordination Polymers
Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties
Supramolecular Isomerism
Interpenetration
Templated [2 + 2] Photodimerizations in the Solid State
Gas Storage and Separation in Supramolecular Materials
Crystal Growth and Molecular Crystal Growth Modification
Soft matter
Soft Matter Science—a Historical Overview with a Supramolecular Perspective
Multicomponent Self-Assembled Polymers Based on π-Conjugated Systems
Functions Based on Dynamic Structural Changes of Coordination Polymers
Self-Healing and Mendable Supramolecular Polymers
Assembly of Block Copolymers
Molecularly Imprinted Polymers
Supramolecular Dendrimer Chemistry
Hyperbranched Polymers in Supramolecular Chemistry
Supramolecular Chemistry in Polymer Networks
Stimuli-Responsive and Motile Supramolecular Soft Materials
Self-Assembling Fibrillar Networks—Supramolecular Gels
Self-Assembly of Facial Amphiphiles in Water
Self-Assembly of Surfactants at Solid Surfaces
Physisorption for Self-Assembly of Supramolecular Systems: A Scanning Tunneling MicroscopyPerspective
Chemisorbed Self-Assembled Monolayers
Self-Organization and Self-Assembly in Liquid-Crystalline Materials
Liquid Crystals Formed from Specific Supramolecular Interactions
Covalent Capture of Self-Assembled Soft Materials
Designing Peptide-Based Supramolecular Biomaterials
Self-Assembly of Polymers into Soft Nanoparticles and Nanocapsules
Self-Assembled Polymer Supermolecules as Templates for Nanomaterials
Supramolecular reactivity
Supramolecular Organocatalysis
Replication Processes—From Autocatalysis to Systems Chemistry
Artificial Enzyme Mimics
Dynamic Covalent Chemistry
Reaction Networks
Reactivity in Nanoscale Vessels
Reactions in Solid-State Inclusion Compounds
Functional Polymers
Reactions in Dynamic Self-Assemblies
Supramolecular aspects of chemical biology
Rational Design of Peptide-Based Biosupramolecular Systems
Nucleic Acid Mimetics
Supramolecular Approaches to the Study of Glycobiology
Porphyrinoids: Highly Versatile, Redox-Active Scaffolds for Supramolecular Design and Biomimetic Applications
Supramolecular Chemistry of Membranes
Membrane Transport
Supramolecular Bioinorganic Chemistry
The Role of Supramolecular Chemistry in Responsive Vectors for Gene Delivery
Supramolecular Chemistry in In Vitro Biosensors
Supramolecular Chemistry in Biological Imaging In Vivo
Aptamer Moieties in Biochemical Applications
Supramolecular Approaches for Inhibition of Protein–Protein and Protein–DNA Interactions
Supramolecular Approaches to Medicinal Chemistry
Supramolecular Systems for Tissue Engineering
Chemical Biology Using Fluorinated Building Blocks
Nanotechnology
Atomic Force Microscopy Measurements of Supramolecular Interactions
Biologically Derived Supramolecular Materials
Advances in Supramolecular Chemistry of Carbon Nanotubes
One-Dimensional Nanostructures of Molecular Graphenes
Magnetically Responsive Self-Assembled Composite Materials
Nanoelectronics
Nanolithography
Supramolecular Nanoparticles for Molecular Diagnostics and Therapeutics
Nanotechnology: The “Top-Down” and“Bottom-Up” Approaches
Photochemically Driven Molecular Devices and Machines
Self-Assembled Nanoparticles
Supramolecular Hybrid Nanomaterials as Prospective Sensing Platforms
Two-Dimensional Supramolecular Chemistry
Rotaxanes—Self-Assembled Links
Glossary

Citation preview

Definition and Emergence of Supramolecular Chemistry∗ Jonathan W. Steed1 , Jerry L. Atwood2 , and Philip A. Gale3 1

Durham University, Durham, UK University of Missouri, Columbia, MO, USA 3 University of Southampton, Southampton, UK 2

1 Introduction 2 Emergence 3 Conclusion References

1

1 2 5 5

INTRODUCTION

Although the word “supramolecular” made an early appearance in Webster’s Dictionary in 1903, “Supramolecular chemistry” in its modern sense was introduced only in 1978 by Lehn, who defined it as the “. . .chemistry of molecular assemblies and of the intermolecular bond.”1 Classic explanations of supramolecular chemistry describe it as “chemistry beyond the molecule,” “the chemistry of the noncovalent bond,” and “nonmolecular chemistry,” or even “Lego chemistry.” The early work in the field concerned the formation of supermolecules comprising two components, a host and a guest, which interact with one another in a noncovalent manner (Figure 1). The host is a large molecule or aggregate such as an enzyme or synthetic cyclic compound possessing a sizeable, central hole, or cavity. The guest may ∗

Adapted in part from Supramolecular Chemistry, J. W. Steed and J. L. Atwood, Wiley: Chichester, 2nd Ed., 2009.

be a monatomic cation, a simple inorganic anion, an ion pair, or a more sophisticated molecule such as a hormone, pheromone, or neurotransmitter. More formally, the host is defined as the molecular entity possessing convergent binding sites (e.g., Lewis basic donor atoms, hydrogen-bond donors, etc.). The guest possesses divergent binding sites (e.g., a spherical, Lewis acidic metal cation, or hydrogenbond-accepting halide anion). In turn, a binding site is defined as a region of the host or guest capable of taking part in a noncovalent interaction. The host–guest relationship has been defined by Donald Cram2 as follows: Complexes are composed of two or more molecules or ions held together in unique structural relationships by electrostatic forces other than those of full covalent bonds . . . molecular complexes are usually held together by hydrogen bonding, by ion pairing, by π-acid to π-base interactions, by metal-to-ligand binding, by van der Waals attractive forces, by solvent reorganising, and by partially made and broken covalent bonds (transition states). . .High structural organisation is usually produced only through multiple binding sites. . . A highly structured molecular complex is composed of at least one host and one guest component . . . A host–guest relationship involves a complementary stereoelectronic arrangement of binding sites in host and guest. . . The host component is defined as an organic molecule or ion whose binding sites converge in the complex . . . The guest component as any molecule or ion whose binding sites diverge in the complex . . .

This description might well be generalized to remove the word “organic,” since more recent work has revealed a wealth of inorganic hosts, such as zeolites3 and polyoxometallates,4 or mixed metal–organic coordination compounds, such as metal–organic frameworks (MOFs)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc002

2

Concepts Molecular chemistry Host

Supramolecular chemistry

Molecular precursors

Specific characteristic, function or properties: Recognition Catalysis Transport

+ +

+

Covalent molecule: Chemical nature Shape Redox properties HOMO–LUMO gap Polarity Vibration and rotation Magnetism Chirality

Figure 1

Guest

Definition of traditional supramolecular “host–guest” chemistry according to Lehn.5

(see Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry), which perform similar functions and may be thought of under the same umbrella.

2

Supermolecule (complex): Degree of order Interactions between subunits Symmetry of packing Intermolecular interactions

EMERGENCE

The original supramolecular host–guest complexes involve a host molecule that possesses an intrinsic molecular cavity into which the guest fits; hence, they are, in principle, stable in all forms of matter (solid, liquid/solution, and the gas phase). The host–guest concept is much older than the work by Pedersen6 on hosts for alkali metal ions in the late 1960s that gave birth to modern supramolecular chemistry and can be dated back to the extensive body of clathrate or solid-state inclusion chemistry. This field begins with the twin descriptions of zeolites or “boiling stones” discovered by Axel Cronstedt in 1756 and clathrate hydrates or “anomalous ice” prepared by Joseph Priestley in 1778. The evolution of this area is elucidated later in this work by Bishop (see Synthetic Clathrate Systems, Supramolecular Materials Chemistry) and forms much of the early part of our subjective timeline of supramolecular chemistry (Table 1). Interspersed among these milestones is the parallel birth of self-assembly as in the formation of self-assembled monolayers first observed as the spreading of oil on water by Benjamin Franklin in 1774, and the birth of nanochemistry (the 1818 recognition of the particle size-dependent color of colloidal gold). We can also see the evolution of crystal engineering from the early topochemical postulate and molecular engineering of von Hippel in the 1960s to

the supramolecular synthon approach of Desiraju in 1995. The years 1989 and 1995 mark milestones in the design and synthesis of coordination polymer systems that have brought about the explosion of porous MOF chemistry over the past decade. Biological receptor–substrate supramolecular chemistry and, by generalization, the whole of modern host–guest chemistry has its roots in three core concepts: 1. The recognition by Paul Ehrlich in 1906 that molecules do not act if they do not bind, Corpora non agunt nisi fixata; in this way, Ehrlich introduced the concept of a biological receptor. 2. The recognition in 1894 by Emil Fischer that binding must be selective, as part of the study of receptor–substrate binding by enzymes. He described this by a lock -and -key image of steric fit in which the guest

Substrate

+

(a)

Lock and key

Enzyme

Complex

+ Induced fit (b)

Figure 2 (a) Rigid lock and key and (b) induced fit models of enzyme–substrate (and hence host–guest) binding.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc002

Definition and emergence of supramolecular chemistry

3

Table 1 An illustrative timeline charting the development of supramolecular chemistry from its roots in solid-state inclusion compounds, through the birth of macrocyclic host–guest chemistry in the 1960s to its modern incarnation in self-assembled materials and nanoscale chemistry. 1756 1774 1778 1810 1818

— — — — —

1823 1841 1849 1891 1891

— — — — —

1893 1894 1906 1937

— — — —

1939 — 1940 — 1945 — 1949 1953 1956 1958 1959 1961 1964 1965 1962 1967 1968 1968 1969 1969 1969 1971 1973 1978

— — — — — — — — — — — — — — — — — —

1976 1979 1981 1986 1987

— — — — —

1989 1989 1991 1994 1995 1995 1996

— — — — — — —

1996 1998 1999 2004

— — — —

Axel Cronstedt: description of “boiling stone” (zeolite) Benjamin Franklin: spreading of oil on water Joseph Priestly: “anomalous ice” Sir Humphrey Davy: discovery of chlorine hydrate Jeremias Benjamin Richters: particle size explanation for the color of “drinkable gold”; colloidal gold known since antiquity (e.g., Lycurgus cup, fourth century AD) Michael Faraday: formula of chlorine hydrate C. Schafh¨autl: study of graphite intercalates F. W¨ohler: β-quinol H2 S clathrate Villiers and Hebd: cyclodextrin inclusion compounds Agnes Pockles: the first surface balance, leading to the development of the Langmuir trough and the Langmuir–Blodgett technique Alfred Werner: coordination chemistry Emil Fischer: lock-and-key concept Paul Ehrlich: introduction of the concept of a receptor ¨ K. L. Wolf: the term Ubermolek¨ ule is coined to describe organized entities arising from the association of coordinatively saturated species (e.g., the acetic acid dimer) Linus Pauling: hydrogen bonds are included in the groundbreaking book The Nature of the Chemical Bond M. F. Bengen: urea channel inclusion compounds H. M. Powell: X-ray crystal structures of β-quinol inclusion compounds; the term “clathrate” is introduced to describe compounds where one component is enclosed within the framework of another Brown and Farthing: synthesis of [2.2]paracyclophane Watson and Crick: structure of DNA Dorothy Crowfoot Hodgkin: X-ray crystal structure of vitamin B12 Daniel Koshland: induced fit model Donald Cram: attempted synthesis of cyclophane charge-transfer complexes with (NC)2 C=C(CN)2 N. F. Curtis: first Schiff’s base macrocycle from acetone and ethylene diamine Busch and J¨ager: Schiff’s base macrocycles Olga Kennard and J. D. Bernal: The Cambridge Structural Database von Hippel: birth of crystal engineering Charles Pedersen: crown ethers Park and Simmons: Katapinand anion hosts F. Toda: “wheel and axel” inclusion compound hosts Jean-Marie Lehn: synthesis of the first cryptands Jerry Atwood: liquid clathrates from alkyl aluminum salts Ron Breslow: catalysis by cyclodextrins G. M. J. Schmidt: topochemistry Donald Cram: spherand hosts produced to test the importance of preorganization Jean-Marie Lehn: introduction of the term “supramolecular chemistry,” defined as the “chemistry of molecular assemblies and of the intermolecular bond” Deliberate clathrate design strategies; “hexahosts” D. D. MacNicol and later in 1982 “coordinatoclathrates” E. Weber Gokel and Okahara: development of the lariat ethers as a subclass of host V¨ogtle and Weber: podand hosts and development of nomenclature A. P. de Silva: fluorescent sensing of alkali metal ions by crown ether derivatives Award of the Nobel prize for Chemistry to Donald J. Cram, Jean-Marie Lehn, and Charles J. Pedersen for their work in supramolecular chemistry G. M. Whitesides: self-assembled thiol monolayers on gold R. Robson: 3D coordination polymers based on rod-like linkers G. M. Whitesides: a chemical strategy for the synthesis of nanostructures M. Brust: synthesis of thiol-stabilized gold nanoparticles O. M. Yaghi: first MOF; key coordination polymer papers by M. J. Zaworotko and J. S. Moore G. Desiraju: supramolecular synthon approach to crystal engineering Atwood, Davies, MacNicol, and V¨ogtle: publication of Comprehensive Supramolecular Chemistry containing contributions from many key groups and summarizing the development and state of the art J. K. M. Sanders: the first example of a dynamic combinatorial chemistry system Rowan and Nolte: helical supramolecular polymers from self-assembly J. F. Stoddart: molecular electronics based on interlocked molecules J. F. Stoddart: the first discrete Borromean-linked molecule, a landmark in topological synthesis

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc002

4

3.

Concepts has a geometric size or shape complementarity to the receptor or host (Figure 2a). This concept laid the basis for molecular recognition, the discrimination by a host between a number of different guests. The fact that selective binding must involve attraction or mutual affinity between the host and guest. This is, in effect, a generalization of Alfred Werner’s 1893 theory of coordination chemistry, in which metal ions are coordinated by a regular polyhedron of ligands binding by dative bonds.

Receptor–substrate chemistry underwent a huge paradigm shift in 1958 with Koshland’s “induced fit” model (Figure 2b), and these concepts have since permeated throughout biological and abiotic supramolecular chemistry. Supramolecular chemistry as we understand it today has evolved to encompass not just host and guest chemistry but also all aspects of self-assembly. It includes the design and function of molecular devices and molecular assemblies, noncovalent polymers, and soft materials such as

Larger molecule (Host) Crystallization

Smaller molecule (Guest)

(a)

Lattice inclusion host–guest complex or clathrate (Solid-state only)

Covalent synthesis

Small molecular ‘‘guest’’

Small molecules Host–guest complex

Large ‘‘host’’ molecule (b)

Covalent synthesis

Small molecules (c)

Spontaneous

Larger molecule Self-assembled aggregate

Figure 3 Key paradigms in supramolecular chemistry. (a) Solid-state clathrate paradigm, (b) molecular host–guest paradigm, and (c) self-assembly paradigm. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc002

Definition and emergence of supramolecular chemistry liquid crystals, informed nanoscale chemistry, and “bottomup” nanotechnology. In 2002, Lehn added a functional definition: “Supramolecular Chemistry aims at developing highly complex chemical systems from components interacting by noncovalent intermolecular forces.”7 Hence, the current emphasis is on increasing complexity and hence increasingly sophisticated functionality and on the information stored in molecular components that allows this complexity to be achieved. Modern supramolecular systems are beginning to display complex emergent properties based on the nonlinear interactions between the molecular component parts. It is clear that there are certain properties and features that emerge according to the length scale on which a system assembles, and indeed on which it is studied. Thus, the way in which ostensibly easily understood molecular-level supramolecular interactions scale up into the nanoworld is not always predictable and represents the frontiers and future of supramolecular science. As direct microscopic imaging and manipulation on the multinanometer scale become increasingly technologically feasible, it is increasingly possible to study the fascinating consequences of chemical emergence —the “arising of novel and coherent structures, patterns, and properties during the process of self-organization in complex systems.”8 Fundamentally, supramolecular chemistry concerns the mutual interaction of molecules or molecular entities with discrete properties. This interaction is usually of a noncovalent type (an “intermolecular bond” such as a hydrogen bond, dipolar interaction, or π-stacking). Key to many definitions of supramolecular chemistry is a sense of modularity. Supermolecules, in the broad sense, are aggregates in which a number of components (of one or more type) come together, either spontaneously or by design, to form a larger entity with properties derived from those of its components. These aggregates can be of the host–guest type in which one molecule encapsulates the other or they can involve mutually complementary, or self-complementary, components of similar size in which there is no host or guest. We can thus trace the evolution of supramolecular chemistry from the original solid-state “clathrate” paradigm (Figure 3a), through the molecular host–guest paradigm (Figure 3b) to the self-assembly paradigm (Figure 3c). As it is currently practiced, supramolecular chemistry, with its emphasis on the interactions between molecules, underpins a very wide variety of chemistry and materials science impinging on molecular host–guest chemistry, solid-state host–guest chemistry, crystal engineering and the understanding and control of the molecular solid state (including crystal structure calculation), supramolecular

5

devices, self-assembly and self-organization, soft materials, nanochemistry and nanotechnology, complex matter, and biological chemistry. Dario Braga has summed up the impact of supramolecular concepts in the following way9 : The supramolecular perception of chemistry generated a true “paradigm shift”: from the one focused on atoms and bonds between atoms to the one focused on molecules and bonds between molecules. In its burgeoning expansion the supramolecular idea abated, logically, all traditional barriers between chemical subdivisions (organic, inorganic, organometallic, biological) calling attention to the collective properties generated by the assembly of molecules and to the relationship between such collective properties and those of the individual component.

3

CONCLUSION

It is clear that the molecular-level approach to understanding binding phenomena that gave rise to supramolecular chemistry has found application in a vast array of phenomena and is to a great extent fueling the concepts and growth of a vast swathe of chemically related science. For example, future applications of supramolecular chemistry in biological systems may include new treatments for disease by the inhibition of protein–protein interactions or by the perturbation via synthetic channels or carriers of chemical and potential gradients within cancer cells triggering apoptosis. From molecules to supramolecular assemblies, to nanomaterials and complex molecular biosystems, the ensuing chapters in these volumes capture in detail the backdrop and current state of the art in all of these fields that are driven or informed by supramolecular concepts.

REFERENCES 1. J.-M. Lehn, Angew. Chem. Int. Ed. Engl., 1988, 27, 89. 2. D. J. Cram, Angew. Chem. Int. Ed. Engl., 1986, 25, 1039. 3. R. Szostak, Molecular Sieves, Van Nostrand Reinhold, New York, 1989. 4. A. Muller, E. Krickemeyer, J. Meyer, et al., Angew. Chem. Int. Ed. Engl., 1995, 34, 2122. 5. J.-M. Lehn, Supramolecular Chemistry, 1st edn, Wiley-VCH Verlag GmbH, Weinheim, 1995. 6. R. M. Izatt, Chem. Soc. Rev., 2007, 36, 143–147. 7. J.-M. Lehn, Proc. Nal. Acad. Sci. U.S.A., 2002, 99, 4763. 8. J. Goldstein, Emergence: Complex. Organ., 1999, 1, 49. 9. D. Braga, Chem. Commun., 2003, 2751.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc002

Supramolecular Interactions Dushyant B. Varshey, John R. G. Sander, Tomislav Friˇscˇ i´c, and Leonard R. MacGillivray University of Iowa, Iowa City, IA, USA

1 2 3 4 5 6 7 8

Introduction Supramolecular Chemistry Supramolecular Interactions Construction of Supramolecular Compounds Host–Guest Chemistry Molecular Recognition Self-Assembly Supramolecular Structures via Molecular Recognition and Self-Assembly 9 Conclusions References

1

9 9 10 16 16 16 16 17 21 21

INTRODUCTION

To achieve the impeccable ability of nature to construct molecules (e.g., proteins), chemists have traditionally employed approaches at the molecular level. Molecular chemistry, since the synthesis of urea by W¨ohler in 1828,1 has relied on building molecules via stepwise formation and breakage of covalent bonds. Molecular techniques of chemists have culminated into the total syntheses of sophisticated molecules (e.g., vitamin B12).2 However, nature routinely utilizes noncovalent interactions to organize molecules to form aggregates that perform specific functions. Chemists now recognize advantages of the synthesis paradigm of biology that can facilitate the

construction of complex molecules, otherwise unavailable via traditional approaches. An early transition toward this approach was realized when Emil Fischer, in 1894, proposed the “lock-and-key” model for enzyme–substrate interactions.3 The elegant mechanisms of enzymes provided basic principles for the new subject, namely, “Supramolecular Chemistry,” from which principles of molecular recognition and supramolecular function evolved.4, 5

2

SUPRAMOLECULAR CHEMISTRY

The term supramolecular chemistry was coined by JeanMarie Lehn in 1969. Lehn defined supramolecular chemistry as “the chemistry of molecular assemblies and intermolecular bonds,” which is more commonly referred to the “chemistry beyond the molecule.”6 The Nobel Prize was awarded to Lehn, Charles Pedersen, and Donald Cram in 1987 for pioneering contributions to supramolecular chemistry.7 As molecules are built by connecting atoms by covalent bonds, supramolecular compounds are built by linking molecules with intermolecular forces (Figure 1).8 Thus, in molecular chemistry, precursor molecules undergo covalent-bond making or breaking to produce a target molecule A. In contrast, in supramolecular chemistry molecule A can act as a host that interacts with a guest via noncovalent forces (e.g., hydrogen bonds) to form a supermolecule B.

2.1

Development of supramolecular chemistry

The concepts and roots of supramolecular chemistry can be traced to the discovery of chloride hydrate by Sir Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc003

2

Concepts Pedersen (1967),14 which were compared to natural macrocycles (e.g., ionophores, heme). Seminal contributions were then made by Pedersen and Lehn on crown ethers and cryptands (1960s),7 respectively, and Cram on spherands (1970s) (Figure 2).15 Moreover, the development of synthetic receptors introduced molecular recognition as an area that blossomed into supramolecular chemistry by collaborative concepts that stemmed from biology and physics. Supramolecular chemistry is now a major interdisciplinary field that embodies the expertise of synthetic organic chemists, inorganic, and solid-state chemists, theorists, physicists, and biologists that strive to develop new molecules and materials with unique properties and applications.

Molecular chemistry

Molecular precursors

A Supramolecular chemistry

A Host

Guest B

3

Figure 1 An illustration of molecular versus supramolecular chemistry.

SUPRAMOLECULAR INTERACTIONS

Supramolecular compounds are formed by additive and cooperative noncovalent interactions. The noncovalent interactions include a wide range of attractive and repulsive forces. The most common noncovalent interactions, along with approximate energies, are listed in Table 1.8 A detailed understanding of the origins and scopes, as well

Humphrey Davy in 1810.8, 9 Development was initiated through the understanding of selective binding of alkalimetal cations by natural,10 as well as synthetic, macrocyclic, and macropolycyclic ligands11–17 as described by Curtis (1961),11 Busch (1964),12 J¨ager (1964),13 and

H3C

O O

H3C

N O

O O

CH3 H3C

O

O H3C

O

O

O O

O

O

O

CH3 O

O

O

CH3

O H 3C

N

CH3

CH3

H3C

(a)

Figure 2 Table 1

(b)

(c)

CH3

Early developments in supramolecular chemistry: (a) crown ether (Pedersen) (b) cryptand (Lehn) (c) spherands (Cram). Common supramolecular interactions.8

Supramolecular interactions

Directionality

Ion–ion van der Waals Closed-shell metal–metal Ion–dipole Dipole–dipole Coordination bonds Hydrogen bonds Halogen bonds π –π interactions Cation–π and anion–π interactions

Nondirectional Nondirectional Nondirectional Slightly directional Slightly directional Directional Directional Directional Directional Directional

Bond energies (kJ mol−1 ) 100–350 105 M−1

13 N+ N H H H N

NO2 11 OEt

11 OEt

OEt

HN H

N H

O H N H

N

N

N

OEt

14

O

O HN H

N H

NH H

N

N

N

N O 15 Ka = 7 × 106 M−1

Ka = 2 × 107 M−1 16H+ B(C6H3(CF3)2)4− H

N H

N+ N H H H

N

N

Ph 12

Ka > 5 × 105 M−1

NO2

O

N

16

K2CO3

N

HN H

N

N

N

+

NH H N

14

14

Ka = 3 × 1010 M−1

Ka < 10 M−1

O O O O O O S S S N H

N H

N H

N

N

N

Ka = 1 × 105 M−1

17

17

Figure 4 Chemical structures of DDD–AAA triply H-bonded pairs 11.12, 12.13, 11.14, 11.15, 14.16H+ , and 17 studied by the Zimmerman, Anslyn, Leigh, and Wisner groups.

UV–vis spectroscopic methods.26 Once again, the Ka value exceeded the range that was accessible by UV–vis measurements and only a lower limit of Ka > 5 × 105 M−1 could be given. Very recently, the Wisner group reported a stable double helical complex through an AAA–DDD array and determined the Ka value for 17 (Ka = 1 × 105 M−1 ) in CDCl3 by 1 H NMR.27 Also recently, the Leigh group synthesized 14 and 15 which features a DDD array and which is also highly fluorescent. Accordingly, the Leigh group used fluorescence spectroscopy as the analytical technique—which greatly expands the range of Ka values that are accessible—to determine the Ka values for the 11.14 (Ka = 2 × 107 M−1 ), 11.15 (Ka = 7 × 106 M−1 ), and

14.16H+ (Ka = 3 × 1010 M−1 ) complexes. Clearly, secondary electrostatic interactions can play a dramatic role in determining the overall affinity of H-bonded complexes. Leigh and coworkers conclude that “in this series each incremental increase of two cooperative secondary interactions increases the stability of the neutral triple hydrogen bonded complex by roughly 3 kcal mol−1 .”28 Addition of solid K2 CO3 to 14.16H+ deprotonates 16H+ to give 16 which has a DAD H-bonding array. Proton NMR titration experiments between 14 and 16 in CD2 Cl2 at millimolar concentrations do not reveal any interactions between 14 and 16 (Ka > 10 M−1 ). The introduction of a single noncomplementary H-bonding interaction reduces binding affinity by 109 -fold! Such stimuli-induced changes in Ka and the corresponding changes in G provide a potent driving force for the current generation of molecular machines.7 Given the analysis described above, which shows that the value of Ka for hydrogen-bonded assemblies increases as the number of hydrogen bonds increases, it is perhaps not surprising that a number of investigators have constructed assemblies driven by a multitude of hydrogen bonds. One example comes from the work of Ghadiri, who prepared 18 (Figure 5).29 Compound 18 is a cyclic decapeptide composed of alternating hydrophobic D- and L-amino acids, and adopts a circular structure with Hbond donors and acceptors oriented perpendicular to the plane of the macrocycle. Accordingly, 18 undergoes Hbond-mediated assembly to form nanotubular assemblies. Addition of 18 to an aqueous solution of phosphatidylcholine liposomes results in their assembly in the membrane ˚ diameter. In a related experiment, into channels of ≈10 A the addition of 18 to a solution of glucose-entrapped unilamellar lipid vesicles results in efflux of glucose as monitored by the absorbance of NADPH (nicotinamide adenine dinucleotide phosphate-oxidase) at 340 nm produced in an enzyme-coupled assay.29 A particularly interesting example of association driven by the formation of four H-bonds that features issues of complementarity and preorganization was reported by Zimmerman.30 Corbin and Zimmerman synthesized ureidodeazapterin 19 which—unlike the H-bonding systems described above—has the potential for prototropic equilibria (Figure 6). Compound 19 presents an AADD-H-bonding array which is self-complementary and therefore undergoes dimerization to yield 19.19 in C6 D5 CD3 and CDCl3 . However, 19 can undergo prototropy to yield 20–23 which present AADD-, ADAD-, ADDA-, and ADDA-H-bonding arrays, respectively. Protomers 20 and 21 are, of course, also self-complementary and form homodimers 20.20 and 21.21. Because 19 and 20 both possess AADD-H-bonding arrays, they are also complementary to each other and are capable of forming the heterodimer 19.20. In total, three homodimers and one heterodimer are observed in C6 D5 CD3

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc004

Complementarity and preorganization

H2N O H N D L

HN O HN

O

O NH

O

L

D

NH

HN O

D

O H NH N

L

HN N O H

L D

HN

L

O

D NH O

N H O

18

NH

Phosphatidylcholine liposomes R

OH HO OH OH HO N N N N N N N N N OH HO OH HO

N HO

R

OH R

R

HO OH HO OH N N N N N N N N N HO OH HO OH HO

N

OH HO N N

R

OH HO OH N N N N N O OH H

R

HO OH HO OH N N N N N N N N N HO OH HO OH HO

R

N HO

N N OH HO

5

of protomers 22 and 23, dramatically alters the prototropic equilibrium with the exclusive formation of a mixture of the 22.24 and 23.24 heterodimers. In this way, 22 and 23 can be viewed as being complementary to 24 but are not preorganized for heterodimerization because of the energetic costs associated with breaking up dimers 19.19–21.21 and 19.20 as well as tautomerization to 22 and 23. Complementarity is a complex property because most molecules exist in a variety of conformations and some even possess prototropic forms each of which possesses a specific complementarity. Beyond this, some compounds possess two or more potentially overlapping binding locations31 which can further complicate the situation.

2.2.3 Electrostatic interactions Electrostatic effects can play a very large role in determining the overall strength and geometry of noncovalent complexes. Most generally, positively charged regions of molecules are attracted to negatively charged regions of their partners. In supramolecular systems, such interactions most commonly take the form of ion–ion, ion–dipole, ion–quadrupole, and quadrupole–quadrupole interactions. Of course, the strength of such interactions falls off as one proceeds from ion to dipole to quadrupole.32

OH R

R

N

OH HO N N

N HO

OH HO N N

N N OH HO

OH

N N OH HO

N

R

OH R

R

HO OH HO OH N N N N N N N N N HO OH HO OH HO

N

OH HO N N

N HO

OH HO N N

N N OH HO

R

OH

N N OH HO

N

OH R

HO OH HO OH N N N N N N N N N HO OH HO OH HO

R

N

R

Figure 5 Chemical structure of Ghadiri’s cyclic peptide 18 which undergoes self-association in phosphatidylcholine liposomes to form nanotubes that transport glucose across the membrane.

to the exclusion of monomer as a result of the high Ka value (Ka > 107 M−1 ) for this quadruply H-bonded DDAAbased system. Interestingly, protomers 22 and 23 feature ADDA-H-bonding arrays that are not self-complementary and not observed in C6 D5 CD3 solution. Addition of diamidonaphthyridine 24, which features a DAAD-H-bonding array that is complementary to the ADDA-H-bonding array

Ion–dipole interactions The pioneering work of Pedersen on the crown ethers constitutes the premiere example of ion–dipole interactions in supramolecular chemistry.2 Figure 7 shows idealized chemical structures of 12-crown-4, 15-crown-5, and 18crown-6, which are prototypical members of the crown ether series of macrocycles discovered by Pedersen.33 By a combination of methods, most notably solubility measurements and UV spectroscopy, Pedersen showed that the crown ethers form complexes with a variety of alkalimetal cations (e.g., Li+ , Na+ , K+ , Cs+ ) and also with ammonium ions mainly in CH3 OH as solvent. Figure 7 shows a representation of the geometry observed in the Xray crystal structure of 18-crown-6 subsequently determined by Dunitz and Trueblood,34 which shows that two of the CH2 groups turn inward and fill the cavity of the receptor. In the presence of KSCN (potassium thiocyanate), 18crown-6 undergoes a conformational change that results in the formation of the 18-crown-6·K+ complex depicted in Figure 7. In this manner, 18-crown-6 is complementary toward K+ ion but is not preorganized for binding. The driving force for the formation of the 18-crown-6·K+ complex is ion–dipole interactions between the K+ ion and the dipole associated with the ether O atoms. Through the combined efforts of a large number of researchers, the structural features (e.g., number of coordinating atoms, identity of coordinating atoms, types and geometry of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc004

6

Concepts O

N H

N

O

N H

N N 19

H

O N H

Bu

Bu

N H

N Bu

N

N N H 20

O

N

H

N

N N H 21

O

O

N Bu

N

H

N H

O N 22 H

O H

N

N N N 23 H

O

H

N

Bu O

N H

N N

Bu

O

N

H N

H N

H

O

N H

N H

N

N

O

Bu

Bu

O N H

N

24

N H

N

Bu

24

24

O

N H 19 20 O

O N N

Bu

N

H N

H N O

H

O

O

N H

N H

N

N

Bu

Bu

N

N H

N H

N H

O

N

N

N O

N

H N

H N N

H 19 19 O

Bu O H O

Bu

H

N

N

N H

Bu H N

N N

O

O H

N

O

N H O

N

N

H N

H N N

N

H Bu N 21 21

H N 20 20

O

O

O

N

H

N H

N

Bu N

O N

22 24

H N

Bu Bu

H

N H

N H

N

N

N

Bu O H N

Bu O

O 23 24

Figure 6 Chemical structures of tautomeric forms of monomeric ureidodeazapterin (19–23), homodimers (19.19, 20.20, and 21.21), and heterodimers (19.20, 22.24, 23.24). O

O O

O

O

O

(a) 12-Crown-4

O

O

18-Crown-6

O

CH2-groups fill cavity

O

O

O O

O O

15-Crown-5

O

O

O

O

O O

H

O H O

O

O

O H

H

(b)

O

O

KSCN

+ O K O O O − SCN 18-Crown-6 K+ O

Figure 7 (a) Chemical structures of 12-crown-4, 15-crown-5, and 18-crown-6, and (b) illustration of the conformational change that occurs upon binding of K+ ion.

bridges) toward the complexation of cations have been delineated. For the purpose of this discussion, one of the most interesting features of the complexation behavior of crown ethers is their selectivity toward alkali cations based on size. For example, among the series of alkali cations (M+ , G kcal mol−1 ; Na+ , −5.89; K+ , −8.27; Rb+ , −7.26; Cs+ , −6.06)35 18-crown-6 displays highest affinity toward K+ in CH3 OH because the size of this cation matches best to the size of the cavity. Interactions with quadrupoles Molecules that are centrosymmetric (e.g., benzene) do not possess a molecular dipole moment. The distribution of

C 6H 6

C 6F 6

Figure 8 Electrostatic potential surfaces for benzene and hexafluorobenzene spanning the range from −60 (red) to +50 (blue) kcal mol−1 .

electrons within a benzene molecule, however, is not symmetrical, with the region above the plane of the aromatic ring constituting a region of negative electrostatic potential and the H-atoms around the aromatic ring constituting a region of positive electrostatic potential (Figure 8). For this reason, benzene has a nonzero quadrupole moment and is capable of noncovalent interactions of, for example, the ion–quadrupole and quadrupole–quadrupole type. The research group of Dougherty36, 37 extensively studied the binding properties of anionic cyclophane 25 (Figure 9). They found that 25 displays high affinity toward cationic species in water. For example, N-methylquinolinium 26 and N-methylisoquinolinium 27 bind to 25 with G◦ values of −8.4 and −7.3 kcal mol−1 , respectively, at pH 9.0 in 10 mM borate-buffered water. In contrast, 25 binds far more weakly with quinoline 28 (G◦ = −5.3 kcal mol−1 )

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc004

Complementarity and preorganization Host

of similar magnitude but of opposite sign.38 Interestingly, C6 H6 (m.p. 5.5 ◦ C) and hexafluorobenzene (m.p. 4 ◦ C) form a 1 : 1 cocrystal (m.p. 24 ◦ C) which consists of stacks of alternating molecules. A very interesting example of a quadrupole–quadrupole interaction was utilized by Dougherty et al.39 to promote polymerization of diynes in the solid state. They synthesized 32–34 and obtained 1 : 1 cocrystals of 32 and 33 as well as crystals of 34 (Figure 10). As expected, the cocrystals of 32 and 33 adopt an alternating stacking geometry in the crystal which brings the diyne ˚ units in close proximity (center-to-center distance ≈3.7 A; ◦  ≈ 75 ). Photolysis of the crystals or powdered samples resulted in the alternating copolymerization of 32 and 33 as confirmed by fast-atom bombardment mass spectrometry.

Guests

− + CO2 Cs

+ − O2C Cs

N+ 26

O

O

27

O

N 28

O

O

− + CO2 Cs + − Cs O2C 25

N + + N(CH3)3

O 29

+ S(CH3)2

+ N(CH3)3 30

7

31

Figure 9 Chemical structure of cyclophane 25 and some of its guests (26–31).

2.2.4 Size and shape

which is neutral in borate buffer. Tight binding is not restricted to aromatic ammonium ions, but is also possible with aliphatic ammonium ions. For example, acetylcholine (29) and 1-trimethylammonium adamantane (30) are very good guests (G◦ = −6.2 and −6.7 kcal mol−1 ) for 25. Intriguingly, sulfonium cations are also excellent guests for 25, and 31 which is the S-analog of 30 binds nicely (G◦ = −5.7 kcal mol−1 ). The patterns of 1 H NMR chemical shifts for 25.30 and 25.31 are similar, which suggests that these complexes adopt a common geometry. The results discussed here, along with numerous other host–guest complexes discussed by Dougherty as well as theoretical calculations, establish that these complexes benefit from a cation–π interaction that is predominantly electrostatic (e.g., ion–quadrupole) in origin. Similar to benzene, hexafluorobenzene possesses no molecular dipole moment because it is centrosymmetric. As a result of the high electronegatively of F atoms, however, the fluorinated rim of C6 F6 constitutes a region of high negative electrostatic potential, and the regions above and below the plane of the aromatic rings are in fact electrostatically positive (Figure 8). Benzene and hexafluorobenzene have sizable molecular quadrupole moments

Complementarity between the size and shape of a guest and the size and shape of the cavity of its cognate host is of critical importance in determining the binding strength. When the molecular surfaces of the host and guest are able to embrace but not overlap, the noncovalent interaction between them will be maximized. Excellent examples of the importance of size and shape complementarity in host–guest complexation are available with the cucurbit[n]uril (CB[n]) family of macrocycles (see Cucurbituril Receptors and Drug Delivery, Molecular Recognition).40, 41 CB[n] compounds (n = 5, 6, 7, 8, 10) are readily prepared by the condensation of glycoluril (1 equivalent) with formaldehyde (2 equivalent) in hot (e.g., 23 vibrational entropy (Svib ) > entropy of symmetry

5

◦ ◦ (Ssym ) ≈ entropy of mixing (Smix ). As a rule of thumb, ◦ ◦ Strans and Srot for a small solute are approximately 30 eu (entropy units, kcal mol−1 K−1 ). Both will be lost if a small molecule undergoes a reaction. In contrast, a loss of ◦ —which can be mostly attributed to changes in internal Svib bond rotations—is usually an order of magnitude smaller. It should be noted that of these three, only translational entropy is concentration dependent. The additional entropy ◦ ◦ and Smix are generally quite small. Thus, the factors Ssym entropy of symmetry is defined by Ssym = −R ln σ , where σ is the symmetry number characteristic of the point group of the molecule. For molecules of low symmetry (e.g., the C1 point group) σ = 1, whereas for higher symmetry molecules, for example, dodecahedrane (Ih point group) ◦ usually lies between zero and σ = 60. In other words, Ssym 8.3 eu, with a distinct bias toward zero. The entropy  of mix◦ = −R i ni ln ni , ing of i components is defined by, Smix where n is the mole fraction. Thus, for an equimolar two◦ = −R(0.5 ln 0.5 + 0.5 ln 0.5) = component system Smix R ln 2 = 1.38 eu. So again, any change in these types of entropy as a result of reaction (or complexation) is usually small. Hence, for a host–guest complexation event, it is the changes in translational and rotational entropy that dominate, although in some cases a loss of the other forms of entropy for the complexed host and guest may also play a role. As a final note on entropy, it is also worth recalling that as a solution is made more dilute, the entropy of the system increases. In the case of the dilution of a solution of host, guest, and host–guest complex (19), this entropic change will be larger if the distribution of species shifts toward free host and guest (two species) rather than the host–guest complex. Hence the observed decomplexation of a host–guest complex as a solution is diluted. How do we determine the enthalpy and entropy contributions to the overall free energy change of a binding event? We should recall that there are two general approaches. The most accurate one is to determine the enthalpy change directly using a calorimetric approach. ITC measures the amount of heat liberated by a binding event as aliquots of the guest are added to the host (or vice versa). As the titration proceeds, the amount of free host decreases and so the amount of heat liberated with each addition of guest also decreases. The result of an ITC experiment is therefore the overall enthalpy change for complexation and, equally as important, a curve of how the amount of heat liberated or consumed decreases as a function of the host/guest ratio. This latter curve defines the equilibrium constant for the process, and hence using (18) and (22) the complete thermodynamic profile (G◦ , H ◦ , S ◦ ) at constant pressure is obtained. Errors in H ◦ can be as low as 1%, with attendant errors in free energy and entropy changes ideally 1010 M−1 ). As we have discussed, ITC does not spectroscopically examine the free and the bound states in a host–guest system but instead determines the overall enthalpy change upon addition of multiple aliquots of guest to a solution of the host. Hence, the aforementioned discussion of experimental timeframes and their relationship to the exchange rates do not apply to ITC. Instead, it is important to determine whether upon addition of each aliquot the mixture has equilibrated before an addition aliquot has been added. In other words, in ITC it is important that the time to equilibration for the system be faster than the “pause” between each injection.

2.4

Medium of study: organic solvent or water

The role that solvent plays in chemistry can never be overstated, and in supramolecular chemistry we are particularly interested in how solvent interacts with the free and bound solutes, and consequently how this influences the noncovalent interactions between them. The bulk of research in the field carried out thus far has been in organic solvents, with a bias toward nonpolar and aprotic solvents that maximize electrostatic interactions between the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition molecules. This has given practitioners a firm understanding of both the relationship between molecular structure and noncovalent interactions, and how the different properties of solvents influence noncovalent forces. That stated, in the natural world most supramolecular chemistry occurs in water. Consequently, aqueous supramolecular chemical research that has also focused on water as a solvent, particularly with cyclodextrins,19–23 cyclophanes,24 and, most recently, dynamic molecular capsules.25 This research has highlighted many differences between aqueous and nonaqueous supramolecular chemistry which are important to appreciate. At a very general level, working in water rather than organic solvent affects the thermodynamics of binding in two important ways. First, all other things being equal, binding in water is stronger than binding in organic solvents.14a Second, whereas heat capacity changes of binding processes in organic solvents are usually small or not observed at all, this is generally not the case in water. In other words, in aqueous solution it is frequently observed that the H ◦ and S ◦ for complexation changes as a function of temperature. We will expand on these points in due course, but for now let us briefly discuss the root cause of these two phenomena, the hydrophobic effect. Most simply stated, the hydrophobic effect26–30 is the reason why oil and water do not mix. Many details of the hydrophobic effect are still to be identified and enumerated, but the key noncovalent interaction behind it is hydrogen bonding. More specifically, the strength of the water–water interaction gives it a high cohesive energy and a high surface tension, which leads to a sizable energetic penalty when forming a cavity in water. Hence, the dissolution of a solute, particularly a hydrophobic one, involves significant changes to the local dynamical structure of the water. Key to the hydrophobic effect is therefore the dynamical structure of the hydration shell around solutes, how this changes according to the size, shape, and nature of the solute, and how these changes alter the enthalpy and entropy of the solute/solvation shell. Broadly speaking, these hydration shells vary from solute to solute, but at a fundamental level the hydration of solutes is still poorly understood. Some of the best insight has come about from in silico studies examining how the solvation shell changes as a function of the size27 and shape31, 32 of the hydrophobe. In addition, empirical studies have also been of immense importance. For example, the state of the art in studying solvation shells is quite advanced for protons and small hydrocarbons such as methane,33, 34 and is improving both for inorganic salts35 and organic molecules such as β-cyclodextrin.36 Whatever the rules that govern the hydration of solutes, a key point is that when considering a binding event in water we should be mindful that in water the host–guest equilibrium depicted in (19) is a considerable oversimplification.

9

The host, guest, and host–guest complex each has a particular solvation shell, and the change in solvation in forming the complex plays a large part in the overall thermodynamics of binding. A much more accurate equation would therefore account for the hydrating water molecules. Regardless of our poor understanding of these hydration shells, a comparison of many host–guest complexes reveals that the complexation of most organic molecules receives a thermodynamic boost from the hydrophobic effect. That said, because of the shielding properties of highly polar water, noncovalent interactions that primarily involve electrostatic forces cannot be relied upon to the same extent as they can be in organic solvents. Hence, although the strongest of noncovalent interactions—metal coordination—has proven to be effective drivers of complexation (and assembly) in water, hydrogen bonding has proven so far to be of limited utility. However, more often than not, the thermodynamic boost from the hydrophobic effect more than compensates for any loss of attractive electrostatic interactions between molecules and, as a result, binding constants in water tend to be at least 1–2 orders of magnitude larger. That this is true is perhaps not so interesting in its own right. After all, there are many situations where strong binding is not a requirement, and many systems where strong binding is a detriment. However, that binding is usually stronger means that by and large any particular host is able to bind a wider range of potential guests; and if selectivity is required, it is always easier to prevent binding than create it. In addition to stronger binding, the desolvation of surfaces on the host and guest as they form a complex also leads to a characteristic decrease in the heat capacity of the solution. The standard heat capacity of a substance at constant pressure (Cp◦ ), is the amount of energy a substance absorbs per unit change in temperature (29): ◦

Cp =

∂H ◦ ∂T

(29)

The characteristic decrease in heat capacity for a binding event in water demonstrates that the free host and the guest are able to absorb more energy per unit temperature than the corresponding host–guest complex. Much is still to be learned why this is so, but the current understanding is that the ordered (and this word is used in the loosest possible terms) solvation shells around the hydrophobes can act as heat sinks because, as the temperature is raised, many less ordered states become available in which energy can be stored. However, desolvation of the hydrophobic surfaces on the host and guest reduces the total number of salvationshell waters and attenuates this sink.29, 37 As a result, the heat capacity of the solution decreases. Indeed, this decrease in the heat capacity is one of the best hallmarks

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

10

Concepts

of the hydrophobic effect, and is a much more reliable hallmark than the often observed increase in entropy upon binding usually attributed to the release of “ordered” water molecules from the solvation shell. A change in heat capacity for a binding event indicates that its associated enthalpy as well as entropy change changes as a function of temperature, and this can lead to considerable inaccuracy in van’t Hoff plots if the thermodynamic parameters for complexation are being sought. The relationship between the standard enthalpy change (H ◦ ) and the standard heat capacity change (Cp◦ ) of a reaction or a complexation is given by (30), where H◦ is the reference enthalpy at 298 K: ◦

◦

H = H◦ + Cp

(30)

A similar equation (31) can be derived for the relationship between the standard entropy change (S ◦ ) and the standard heat capacity change: ◦

◦

S = S◦ + Cp ln T

(31)

where S◦ is the reference entropy at 298 K. Combining (30) and (31) with (23) leads to (32): ◦

◦

R ln Ka = −H◦ (1/T ) + Cp ln T + (So − Cp ) (32) This equation allows us to carry out van’t Hoff plots even when H ◦ and S ◦ change as a function of temperature. By fitting the equation using the three variables H◦ , S◦ , and Cp◦ and then using (30) and (31) we obtain H ◦ and S ◦ at the temperatures sought. It is useful to recall that Cp◦ is a second derivative when determined with NMR and other spectroscopic techniques; Ka values must be determined at different temperatures and van’t Hoff plot performed in order to get the enthalpy change, and it is how this enthalpy change changes as a function of temperature that gives Cp◦ . On the other hand, Cp◦ is a first derivative of an ITC experiment. The enthalpy change is determined directly, and a series of experiments at different temperatures yields Cp◦ as the gradient of the graph of H ◦ versus T . Hence, ITC is a much more accurate technique for determining Cp◦ . We have now introduced the necessary basics for determining the association constant (Ka ), standard free energy change (G◦ ), standard enthalpy change (H ◦ ), standard entropy change (S ◦ ), and standard heat capacity change (Cp◦ ) for the complexation of a host and a guest. In subsequent sections, we detail practical aspects of these measurements.

3

PRACTICALITIES

Beyond the basic thermodynamics that we have just discussed, there are many considerations regarding how we actually perform experiments to determine Ka , G◦ , H ◦ , S ◦ , and Cp◦ of complexation in 1 : 1 host–guest systems (19). In determining thermodynamic data for a complexation event, there are, as there are with any physical chemistry problem, two goals. The first goal is to define the system mathematically with a basic mathematical model; the second is to fit the obtained data to the mathematical model. The first goal is, of course, independent of the analytical technique we are going to use, whereas the second is very much dependent on it. Irrespective of the technique used, the overall aim is to quantify the formation of the host–guest complex for a given initial concentration of host and guest. This can be accomplished by directly measuring the amount of host–guest complex, or indirectly by measuring the remaining free host or guest and using mass balance equations to calculate the concentration of the complex. Many analytical techniques are available to the experimentalist for this task, but we focus here on the most widely used techniques, NMR, UV, and fluorescence spectroscopy, and ITC, and give a succinct account on how to conduct such experiments with these techniques. For more detailed descriptions of the individual techniques, as well as details of other techniques used, the reader is directed to some of the many excellent reviews available in the literature.5, 38–41 Each technique has its advantages and limitations, and our intent here is to provide enough information to allow the experimentalist to choose the most suitable technique for his/her particular research. This section begins with highlights of how the timeframe of a technique and the concentration of a sample have important practical ramifications. Subsequently, we discuss the base mathematical model for 1 : 1 complexations before looking at how this model is tailored to each analytical technique. Finally, we discuss a common approach sometimes used to confirm 1 : 1 binding, as well as very briefly highlight higher stoichiometry systems.

3.1

Timeframe of analysis

We have discussed the importance of timeframe of analysis with regard to both the timeframe by which equilibrium in a system is attained and the timeframe of the exchange process of the complexation under investigation. Practical aspects of the former simply involve double-checking that equilibrium has in fact been attained. Practical aspects of the latter are a little bit more complex. As we have discussed, different techniques operate at different timeframes, and it is important to appreciate how this relates to the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition exchange rates between the free and bound states (25). Is the instrumental technique detecting an average signal of the species involved, or are the individual signals from each species apparent? In most spectroscopic techniques including UV–vis and fluorescence, the exchange between free and bound species is much slower than the timeframe of the technique. In other words, the binding event under study is slow on the instrument timescale and the instrument detects each of the species individually. In contrast, this is not the case with NMR spectrometry where complexations that are slow as well as fast on the NMR timescale are both observed. We will return to this point when discussing the individual techniques.

3.2

Concentration range

Another important practical aspect is the concentration under which the complexation is being studied. Each technique has its own limitations defined by its sensitivity, and we discuss these as we examine the individual approaches. In addition, there is the common factor arising from the asymmetry inherent to the host–guest equilibrium (19) and how, for a fixed binding constant, the initial concentration of the host and guest dictates the extent of complexation. Figure 3 shows the state of affairs with a total (initial) concentration of host (Ht ) and guest (Gt ) of 1 mM and a range of binding constants (Ka ) from 10 to 108 M−1 . Ideally, for accurate measurements, equilibrium should result in somewhere between 20 and 80% complexation. If the

11

concentration chosen is too high, then the concentrations of host and guest will be too low and sizable errors will ensue. Likewise, if the initial concentrations are too low, then there will be an insufficient amount of the complex formed and again large errors will arise. In the case shown in Figure 3, the optimum range of Ka is observed to be between 5 × 102 and 2 × 104 M−1 . Of course, the distribution of species can be shifted easily by changing the ratio of Ht /Gt ; for example, if Ka = 10, and Ht = 1 mM and Gt = 100 mM, then 50% complexation is attained (compare with first column in Figure 3). The best way to determine the concentration at which an experiment needs to be run is to know the binding constant; but this is of course the first step of a circular argument. The only option therefore is to take an educated guess at the strength of association and then to perform the required experiment. A second experiment can subsequently be run if binding was weaker or stronger than anticipated. An alternative viewpoint is expressed in Figure 4, which shows percentage complexation against total host and guest concentration (Ht and Gt ) for a host–guest complexation (Ka = 1 × 103 ). This graph represents the effect of dilution upon complex formation. At high guest concentration (1 M), the system is close to full saturation with 97% complexation. In contrast, if the working concentration is too low (Ht = Gt < 1 × 10−4 M) essentially no complexation is observed. We can define a concentration range for this model system of between 0.5 to 50 mM, and whether a particular technique is suitable for the task at hand depends on its inherent sensitivity. Furthermore, there may be solubility

1E−03

Concentration (M)

1E−03

8E−04

6E−04

4E−04

2E−04

0E+00 1E+ 01

1E + 02

1E + 03

1E + 04

1E+05

1E +06

1E + 07

1E + 08

K a (M−1)

Figure 3 Graph of the concentration of host [H ] (blue), guest [G] (red), and host–guest complex [HG] (green) against equilibrium constant (Ka ) where the total (initial) concentration of host and guest (Ht = Gt ) = 1 mM. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

12

Concepts 100

% complexation

80

60

40

20

0 1E + 00

1E − 01

1E −02

1E − 03

1E−04 G t (M)

1E −05

1E − 06

1E − 07

Figure 4 Graph of the percentage complexation against total guest concentration (Gt = Ht ) where the equilibrium constant (Ka ) is set to 1 × 103 M−1 .

limits of the host, guest or complex that may also need to be considered.

3.3

Mathematical model for 1 : 1 complexation

With the exception of just one scenario, all of the techniques commonly utilized by supramolecular chemists to determine thermodynamic data yield too many unknowns if just a single host/guest ratio is studied. The exception, slow exchanging systems studied by NMR, allows the direct measurement of the individual components of the mixture (see below) from just one Ht : Gt ratio. For the other approaches, we must carry out a titration in which the Ht : Gt ratio is varied systematically and construct a mathematical model based on the mass balance equations for the equilibrium and (20). The corresponding mass balance equations are (33) and (34): Ht = [H ] + [HG]

(33)

Gt = [G] + [HG]

(34)

where Ht and Gt are again the total amount of host and guest in solution. For each of the analytical techniques we are going to discuss, we define equations that relate a measurable to three unknowns: the equilibrium constant; a constant specific to the complex and the technique used (δ max , ε, and H ◦ for NMR, UV–vis, and ITC respectively); and the concentration of free guest [G]. We therefore need an expression of [G] that relates it to a series of known quantities and the equilibrium constant for the process. To do this we will combine (20), (33), and (34) to give (35): Ka [G]2 + (1 − Ka Gt + Ka Ht )[G] − Gt = 0

(35)

The solution of this quadratic, that is, (36), is our base equation for 1 : 1 complex formation, which we will combine with equations tailored for each analytical technique relating a measurable to Ka , δ max /ε/H ◦ , and the concentration of free guest [G]. It is useful to note that the right-hand side of (36) is composed of only Ht and Gt (which can be calculated) and the unknown Ka . [G] =

−(1 − Ka Gt + Ka Ht ) ±

 (1 − Ka Gt + Ka Ht )2 + 4Ka Gt 2Ka

(36)

3.4

NMR spectroscopy

Most binding constant determinations in supramolecular chemistry have been performed using NMR, and, in particular, 1 H NMR. Protons are almost ubiquitous to organic chemistry, and the 1 H nucleus is of high abundance. This means a particularly fast analysis time relative to other popular nuclei such as 13 C. In general, NMR will permit the determination of binding constants of between 0.1 and 104 M−1 , although stronger binding constants can be determined if a longer acquisition time is possible or a competition experiment is performed (see below). In these cases, binding constants of up to 1 × 106 M−1 represent the absolute upper limit of the technique before analyses are beset by large errors. The normal limit of Ka ≈ 104 M−1 in NMR spectroscopy arises from the technique’s relative insensitivity. As a result, most proton NMR spectra will be recorded at sample concentrations of 1–5 mM. If the binding constant is high, then it will be necessary to run the sample at concentrations of 100 µM or less, and in such cases the signal-to-noise ratio is small and an extended acquisition

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition time is required. This may be possible in slow exchanging systems (see below), but in fast exchanging systems this approach becomes unwieldy. A critical point in NMR determinations of thermodynamic data is to have a proton whose chemical environment changes sufficiently upon binding to lead to a change in chemical shift, that is, δHfree = δHbound . As has been discussed above, binding determinations by NMR can be divided into two cases: those that are slow on the NMR timeframe and those that are fast. The observed complexation and decomplexation rates (25) are those of unimolecular processes (units s−1 ), and it is the slower of these two processes that we must compare with the rate of NMR data acquisition. Our previous discussion emphasized how the timeframe of the NMR experiment was dependent on the external field strength of the instrument, but it is also important to note that whether a complexation is observed to be fast or slow on the NMR timescale depends on the chemical shift difference between the free and the bound state. The key equation is (37), which defines the boundary between fast and slow timescales, that is, the observed rate constant for coalescence (kcoal ) of the signals for the proton in question in the free and the bound state: kcoal = 2.22δ

(37)

Thus, the larger the δ the higher the observed rate constant [kb in (25)] needs to be to switch from slow to fast on the NMR timescale. In certain cases, where there is a spread in δ values ranging from the very small to very large (e.g., 3 ppm or more), it is entirely possible for some pairs of signals to be fast on the NMR timeframe while others appear slow. Furthermore, in such cases, intermediately shifted signals may be close to coalescence, resulting in a broad or unobserved signal. In most cases, however, all the signals are usually either fast or slow on the NMR timeframe. If an equilibrium process is slow on the NMR timescale, the spectrum will show distinct peaks for the free and the bound state. Hence, by knowing the initial concentration of Ht and Gt , it is then straightforward to determine the concentration of HG, H , and G by integration. In such cases, the Ka values (and using (18) the G◦ value) are quickly determined, and if enthalpy and entropy change or, indeed, heat capacity changes are also sought, it is simply a matter of recording NMR spectra of the same sample at different temperatures. That said, for a number of reasons it is advisable to perform the determination of the Ka and G◦ at two or three different ratios of Ht and Gt . Changing this ratio will help confirm which signals belong to the formed complex (increase in intensity with increased titrant), as well as confirm slow kinetics (the peaks will

13

change in intensity but not shift). Doing so also avoids any major error since the Ka and G◦ values obtained should be identical. Determinations obtained this way and repeated 3 times with new stock solutions will give determinations with errors of less than 10%. During the course of the development of the field, hosts (as well as guests) have tended to become structurally more elaborate, in which case they often exchange slowly on the NMR timescale. Nevertheless, structurally more open hosts that exchange guests rapidly on the NMR timescale still account for the majority of host molecules. The determination of binding constants in these systems involves more effort, both in terms of data collection and fitting. Regarding the latter, we will need (36) to build our mathematical model for Ka determinations with NMR. We begin by first noting that, in fast exchanging systems, the observed frequency δHobs of the proton of interest becomes the weighed average of the free and bound states (38): δHobs = xfree δHfree + xbound δHbound

(38)

where xfree and xbound are the mole fractions of the free component and the complex, respectively. In our determination, we are going to change the ratio of host and guest by titrating in the guest (or host) and monitor how δHobs changes. In other words, we are going to plot a binding isotherm. Typically, a host solution is titrated with a stock guest solution in the NMR tube, and the shift of the proton most affected by binding and not obscured by other signals is then recorded for a minimum of 10 different ratios. It is critical to cover a large range of host/guest ratios to ensure that the system is close to saturation at the end of the titration. In other words, the resulting binding isotherm should reach a plateau. Now that we have our data, if we set δ obs = δHobs − δHfree and δ max = δHbound − δHfree , then from the mass balance equations (33) and (34) and the equilibrium equation (20), the NMR binding isotherm can be expressed as δ obs =

δ max Ka [G] 1 + Ka [G]

(39)

which can be rearranged to (40): δ obs =

δ max 1 Ka [G]

+1

(40)

where δ obs is the shift in parts per million of the observed proton, and δ max is the maximum shift of the observed proton at full complexation. Note that we have too many unknowns in this equation (Ka , [G], and δ max ) to solve for Ka with a single Ht : Gt ratio. To solve this problem,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

14

Concepts

we must combine (40) with our quadratic expression for [G] (35) to get a theoretical expression of δ obs versus Gt (41):

δ obs =


resulting binding isotherm is too sharp and approximates to a step rather than a curve. In such cases, too few points define the binding event and a large error ensues. δ max

 2 +1 Ka Gt − Ka Ht − 1 + (1 − Ka Gt + Ka Ht )2 + 4Ka Gt

A simple iteration process of Ka and δ max using the spreadsheet software such as Origin or the solver in Excel will fit the experimentally derived binding isotherm for δ obs versus Gt and will yield the remaining two unknowns Ka and δ max . As mentioned, the titration must cover the full range of complexation and ultimately bring about the saturation of the host. In addition, to avoid large errors, the obtained binding isotherm must be neither too shallow nor too sharp. Figure 5 shows three theoretical binding isotherms, where a guest is titrated to a host solution of 1 mM, inducing a δ max of 0.2 ppm. The binding isotherm obtained for the Ka = 1000 M−1 system is ideal. It covers the full complexation range, and will allow a good fit of the data. With the weaker binding guest (Ka = 50 M−1 ) the binding isotherm is more akin to a straight line and insufficient guest has been added to saturate the host. In such cases, there are two options: (i) titrate the host further with more guest to reach a plateau close to the δ max = 0.2 ppm, in which case unless a highly concentrated guest solution can be made the dilution of the host must be taken into account39 ; (ii) work at higher concentrations of the host, which may or may not be possible depending on its solubility. On the other hand, when the binding is too strong for the working concentration (Ka = 104 M−1 ), each addition of guest results in it being fully complexed. The



(41)

The easiest way to obtain an ideal curve is to lower the working concentration, which is often not possible with NMR because of its relatively low sensitivity. Alternatively, a less direct approach with larger attendant errors is to perform a competition experiment in which a complex of the host and a relatively weakly bound guest is titrated with the stronger binding guest (see below). Once a Ka value has been determined, this leads directly (18) to the corresponding G◦ of complexation, but to ascertain the H ◦ and S ◦ values of a fast exchanging system it is necessary to determine a van’t Hoff plot by repeating the titration procedure at different temperatures. Note that, because measured Ka values in fast exchanging systems have larger intrinsic errors than analogous determinations in slowly exchanging systems, errors in ascertaining the H ◦ and S ◦ values via a van’t Hoff plot can be quite substantial. In summary, the relative insensitivity of NMR spectroscopy means that the solution under study must be relatively concentrated, a factor that attenuates the range of binding constants that can be determined. In NMR examinations of supramolecular events, the exchange process can be faster than, slower than, or on the NMR timescale. Slow exchanging systems provide more structural information and allow a more straightforward and accurate determination of Ka , G◦ , H ◦ , and S ◦ than

0.20

∆d (ppm)

0.15

0.10

0.05

0.00 0

0.002

0.004

0.006

0.008 G t (M)

0.01

0.012

0.014

0.016

Figure 5 Three theoretical binding isotherms for 1 mM host titrated with guests of association constants, Ka = 50 ( ), 1000 ( ), and 104 M−1 ( ). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition fast exchanging systems. Finally, it should be noted the sensitivity weakness of NMR is countered by the large amount of structural information provided by the technique, particularly when the exchange process between free host and guest and the host–guest complex is slow on the NMR timescale.

Furthermore, for common 1 cm cells we can write (43), which expresses the total observed absorbance as a function of the respective molar absorptivities of the host, guest, and complex at a given wavelength (εH , εG , and ε HG ) and their respective concentrations: Aobs = εH [H ] + ε G [G] + ε HG [HG]

3.5

UV–vis and fluorescence spectrometry

(43)

It is helpful to now define the change in absorbance ε upon complex formation:

After NMR, UV–vis and fluorescence spectrometry are probably the most widely used techniques for thermodynamic data collection. Both require UV–vis absorbance (and in the case of fluorescence, emission) of the host and/or the guest. It is also important that upon complex formation the absorbance or emission change. Both analytical techniques can be approached in a very similar fashion: because of its wider popularity, we emphasize here UV–vis spectroscopy. In UV spectrometry, the excitation of electrons by absorption is accompanied by changes in vibrational and rotational quantum numbers. As a consequence, a broad signal of vibrational and rotational fine structure is produced rather than a single absorption line corresponding to a particular electronic transition. Furthermore, interactions between the solute and the solvent further complicate the data and smooth this broad absorption band. As a result, the amount of structural information provided by these techniques is less than that provided by NMR. One consequence of broad absorption bands is that there is inevitably extensive band overlap from the different species in solution: a fact that can sometime complicate data interpretation. Even if this is not the case, data gathering must rely on titration studies analogous to those described for the NMR analysis of fast exchanging systems; there is no direct analogy to systems exchanging slowly in the NMR timescale even though exchange is slow on the UV–vis timescale. The reason for this is that, with UV–vis determinations, it is signal intensity that we are measuring as a function of host–guest ratio and this amplitude is dependent on the unknown ε of the solution. Hence, there are too many unknowns to determine Ka values from a single mixture of the host and the guest. For a host–guest complexation event, there are potentially three species that can absorb and, because the absorbance of each is additive, we can write (42):

ε = εHG − ε H − ε G

(42)

where Aobs is the total observed absorbance, and AH , AG , and AHG are the respective absorbances of H , G, and HG.

(44)

Equation (44) actually pertains to a relatively rare system, because in most cases we select the absorption to monitor so that one of the species does not absorb. In rare cases where all species do absorb at the wavelength examined, it is necessary to determine εG and εH in a separate experiment to reduce the number of unknowns. Hirose’s excellent practical guide on binding constant determinations describes the methods to treat the collected data (Aobs vs Gt ) of this more complex regression analyses.39 For the majority of cases where only one species absorbs at the observed wavelength (ε H or εG = 0), we can write simplified equations (45 and 46). In the case of these equations, we assume it is the guest that does not absorb: Aobs = AH + AHG

(45)

ε = εHG − ε H

(46)

Combining the mass balance equations (33) and (34) with (45), we can express (47): Aobs = εH Ht + ε[HG]

(47)

which when combined with (20) gives (48): A = Ka ε[H ][G]

(48)

And using the mass balance equations again, we can then obtain the binding isotherm (49): A =

Aobs = AH + AG + AHG

15

εKa Ht [G] 1 + K[G]

(49)

We now have an expression similar to that obtained by NMR (39) and we can insert (49) into our expression of [G] for 1 : 1 complexations (36) to give (50):

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

16

Concepts

Aobs =


εHt  2 +1 Ka Gt − Ka Ht − 1 + (1 − Ka Gt + Ka Ht )2 + 4Ka Gt

Equation (50) is the mathematical expression that we must iterate with the experimentally determined data. We can see that we are dealing with an equation similar to that obtained for NMR in fast exchanging systems (41). The experimental data will in this case be A versus Gt (rather than δ obs vs Gt ), but the fitting of the binding isotherm is exactly the same except that now Ka and ε are iterated to fit the curve. Remember that we are dealing with changes in absorption intensity rather than a shift in absorption (as we were with NMR). As a result, we cannot change the host/guest ratio simply by adding one to the other because this will change the concentration of the host and hence the absorption intensity. Rather, the experimentalist must prepare separate solutions with increasing amounts of guest (or host) at a constant concentration of host. Although these determinations require more effort, because we are typically working at lower concentrations (about 0.01 mM), UV–vis spectrometry allows the determination of thermodynamic data that cannot be measured by NMR because the binding constant is too high. The typical range of Ka that UV–vis experiments cover is 1–1 × 106 M−1 , but higher binding constants can be determined if the extinction coefficient of the molecules is high. As an example, if ε = 3 × 105 , this would give an absorbance of 0.3 at 1 µM, allowing Ka values as high as 1 × 107 M−1 to be ascertained. On the other hand, the determination of small binding constants by UV–vis is facilitated by weakly absorbing species. Of course, the general point that studies with UV–vis require working at lower concentrations also means that each set of experiments require less host and guest: a fact that reduces pressure on synthesis to form the molecules in question. The study of binding by fluorescence spectrometry is performed with the same experimental and data analyses as UV–vis spectrometry. Thus experiments usually involve either a fluorescent host or a fluorescent guest, and the change in fluorescence is monitored at different host/guest ratios. The UV–vis equations described above can easily be adapted to fluorescence to provide the association constant for a binding event.5 In summary, UV–vis and fluorescence spectroscopies do not provide as much structural detail of a formed host–guest complex. They do, however, allow a wider range of binding constants to be determined. In addition, although each determination requires more practical work, the amount of material required for a determination is—all



(50)

other things being equal—less that that required for NMR experiments.

3.6

Isothermal titration calorimetry (ITC)

The development of sensitive isothermal titration calorimeters first impacted the biological sciences. However, over the last decade or so the technique has become increasingly utilized within supramolecular chemistry. One of the major reasons behind its proven popularity is the fact that Ka , G◦ , H ◦ , and S ◦ are obtained in a single automated experiment. In a typical ITC titration, small aliquots of a concentrated solution of the guest are added to a solution of the host in the ITC cell. Upon each addition, a measured amount of heat is given off, and this decreases during the titration and reaches zero upon saturation of the host. The total heat liberated in this titration yields the enthalpy change, while the shape of the curve for heat release as a function of host/guest ratio provides the equilibrium constant and hence the free energy of binding. As a result, the entropy change for complexation can be directly calculated. This more direct approach to garnering thermodynamic data has much smaller associated errors, particularly for H ◦ , and avoids the issues associated with van’t Hoff plots such as the possibility that H ◦ varies with temperature, or whether the obtained data has chemical “roots” or is simply artifactual. ITC also allows a more direct and accurate determination of heat capacity changes. For straightforward cases of 1 : 1 complex formation, there is a linear relationship between H ◦ and T , the gradient of which is the Cp◦ associated with binding. Thus, a series of five or more experiments run at different temperatures accurately yields any heat capacity change. An ITC titration yields an amount of heat released or absorbed (Q) for each aliquot of added guest solution. The sum of these heats (Q) can be defined by (51): ◦

Q = [HG]V0 H = (Gt − [G])V0 H

◦

(51)

where V0 is the volume of the reaction (ITC cell), and H ◦ is the molar heat of the ligand binding (in cal mol−1 ). This expression is similar to that obtained by NMR (39) and we can insert it into our expression of [G] for 1 : 1 complexations (36) to give (52):

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition

 ◦

Q = Gt V0 H − V0 H

◦

−(1 − Ka Gt + Ka Ht ) ±

Modern instruments generate a plot of Q (cal mol−1 ) versus molar ratio of host and guest using a spreadsheet software that takes into account the change in volume in the cell as the titrant is added. This plot is equivalent to the binding isotherm of NMR or UV. Thus, the theoretical expression of Q (52) can be fitted to a number of preset models of differing stoichiometry to generate Ka (and hence G◦ ) and H ◦ and calculate S ◦ .40 In addition to the goodness of the fit of the obtained curve, ITC also generates the stoichiometry number (N), which, in a 1 : 1 complex, should be close to 1 (0.95 < N < 1.05). Any major deviation from this, for example, N = 0.5, suggests another binding stoichiometry, in which case the fitting model should be reconsidered. If the stoichiometry of complexation can be confirmed by another method, then the ITC instrument allows the N value to be numerically fixed, which can reduce the error of the obtained thermodynamic data, especially in weakly binding systems. If the stoichiometry is in doubt, it is often a good idea to confirm it by NMR or another spectroscopic technique. Ideally an S-shaped titration curve should be obtained. If binding is too weak, then the obtained curve will tend toward linear, whereas if it is too strong, a “step” will instead be observed. The steepness in the titration curve can be described by the Wiseman parameter c, defined as c = Ka × [X], where [X] is the concentration of titrate.40 It is often quoted that, in order to obtain an isotherm corresponding to a smooth S-shaped curve and minimal errors in both H ◦ and Ka (±2–3%), the Wiseman parameter should lie in the range of 100–500. However, most recently it has been shown that it is possible to obtain accurate determinations of Ka with c values as low as 5.42 ITC is capable of determining a wide range of binding constants, typically between Ka = 5 and 109 M−1 . Of course, the spectroscopic character of the molecules involved is not important in such experiments. Instead, the ability to measure Ka is dictated by the extent that the complexation liberates or absorbs heat. For the lower limit, the closer a complexation is to being enthalpically neutral, the larger the error in both H ◦ and Ka , and events that are truly enthalpically neutral across a temperature range cannot be measured. Such events are relatively rare, and consequently the lower practical limit of ITC is more often than not defined by insufficient solubility of the host and guest, which limits the amount of heat generated and results in a flat curve. The upper limit of Ka determination in ITC



(1 − Ka Gt + Ka Ht )2 + 4Ka Gt 2Ka

17

 (52)

pertains to the shape of the titration curve. If the binding is so strong that a smooth S-curve in not obtained but rather a “step” is observed, and this cannot be remedied by dilution because insufficient heat is given out by complexation, then although the H ◦ of complexation can be determined, the obtained Ka will have an unacceptably large error. For c values >500, the steepness of the curve is such that errors in Ka ensue. In summary, ITC is capable of determining a wider range of binding constants than either NMR or UV–vis spectrometry, and generally does so with much smaller errors in Ka , G◦ , H ◦ , S◦ , and Cp◦ : the only prerequisite being that the binding event is not enthalpically neutral at the temperature range being studied. One drawback of the approach that has perhaps inhibited the growth of the technique in supramolecular chemistry is the relatively large volume of the ITC cell (1.4 ml). However, recently sample requirements have decreased considerably, with most recent instruments having a cell volume of only 200 µl. Automated ITC instruments can also be purchased, which makes an in-house ITC a rapid and accurate means to determine thermodynamic data.

3.7

Should NMR, UV–vis, and ITC measurements give the same results?

It is important to note that the study of a single binding event using multiple analytical techniques will not necessarily give the same thermodynamic data. Yes, they are measuring the same complexation process, but their “viewpoints” are not the same. In ITC, for example, even if heats of dilution or heats of protonation are taken into account by the subtraction of data from reference titrations, what is being measured is a global change in enthalpy. Thus, the data obtained will include, for example, changes in the solvation of the host and guest. In contrast, a technique such as NMR relies on measuring subtle changes in electron distribution in and around one atom. This more local viewpoint is sometimes apparent from thermodynamic data generated by monitoring the shift of different protons in a host (or guest). In such instances, the obtained binding constants may differ somewhat from proton to proton. The different viewpoints of the various techniques noted, the obtained thermodynamic data should be similar. If not, problems with the determinations must be examined for.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

18

3.8

Concepts

Does the complexation possess 1 : 1 stoichiometry?

3.9

The focus of this chapter has been on 1 : 1 binding events summarized by (19), and the tacit assumption has been that the stoichiometry of the binding event is known. For binding events that are slow on the NMR timescale, it is trivial to determine the ratio of host and guest in the complex. However, in fast exchanging systems, and when using UV–vis spectrometry and ITC, we rely on fitting programs to confirm the stoichiometry of complexation from the binding isotherm. But what do we do if a poor fit is obtained? Are we dealing with a 1 : 1 complexation and poor data, or are we dealing with a higher order process? One way for ascertaining the stoichiometry of binding is to perform a so-called Job plot (also called the continuous variation method). The theory behind, and practice of, this approach has been described elsewhere,38, 39 so we only briefly discuss the practicalities of the method. The approach is straightforward: a series of at least 10 experiments are carried out in which a metric proportional to [HG]—chemical shift, absorption, enzyme activity—is measured as a function of different ratios of host and guest at constant concentration. The plot of this metric against the mole fraction (of either host or guest) is hyperbolic in the case of 1 : 1 complexation, with a maximum at the 1 : 1 ratio of the host and guest (Figure 6). If a higher stoichiometry is under investigation, the hyperbolic curve will be asymmetric, with a maximum for the plot of the mole fraction of the host (xH ) at xH /[xH + xG ]. Thus for a 1 : 2 complex or a 2 : 1 complex, the maxima will be at 0.666 and 0.333, respectively.

Arbitrary units

3E − 05 2E − 05 2E − 05 1E − 05 5E − 06

Competition experiments

Each of the analytical technique discussed above has an upper limit for determining a binding constant (104 , 106 , and 109 M−1 for NMR, UV–vis, and ITC, respectively). What can be done if a host and guest are found to associate more strongly than this limit? The strategy to overcome this problem is to perform a competition experiment, whereby the guest that binds too strongly to the host (GA ) is titrated to a complex between the host and a more weakly binding guest (GB ) of known association constant (Ka(B) ). Such a titration will allow an apparent reduction of the binding constant of GA (Ka(A) ). We can write (53) and (54): H + GA

ka(A) −

− −− −− − −

HGA

(53)

H + GB

ka(B) −−

−− −− − −

HGB

(54)

where Ka(A) > Ka(B) and the latter is known. The mass balance equations then become: Ht = [H ] + [HGA ] + [HGB ], GAt = [GA ] + [HGA ], and GBt = [GB ] + [HGB ], where Ht is the total host and GAt and GBt are the totals of GA and GB , respectively. If we take the example of slow exchanging system by NMR, the determination of Ka(A) is accomplished using these mass balance equations, the relative integration of HGA and HGB , and the following equation (55): Ka(A) [HGA ][GB ] = Ka(B) [HGB ][GA ]

(55)

The mathematics necessary to fit a binding isotherm arising from a competition experiment utilizing a fast exchanging process in NMR or UV–vis spectroscopy or ITC is considerably more involved. These have been extensively described in the literature,43, 44 particularly in regard to the use of displacement assays for the determination of association constants of spectroscopically silent guests.45 As a rule of thumb, competition experiments will extend the upper binding constant limit of a technique by at most 3 orders of magnitude. Put another way, they define the upper limits of NMR, UV–vis, and ITC respectively to 107 , 109 , and 1012 M−1 , respectively. The down side of competition experiments is that the errors associated with the determination of the weaker binding guest are propagated into the competition experiment such that errors in the determination of the larger association constant are typically at least 10–15% (and often higher).

0E + 00 0.0

0.2

0.4

0.6 Mole ratio

0.8

1.0

Figure 6 Continuous variation method (Job Plot) for a hypothetical 1 : 1 complex formation.

3.10

A word about higher order systems

Higher order systems, such as ternary binding/assemblies involving a host and two guests, are also frequently

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition encountered in supramolecular chemistry. Assuming one is interested in the microscopic (individual) binding events rather than the macroscopic (overall) binding process, these systems are intrinsically more complicated than 1 : 1 binding systems: not only because of the higher stoichiometries but also because there are variations within each system of defined stoichiometry. For example, in a 1 : 2 host–guest system the two binding sites may be the same or different, may be independent of each other or coupled (the system possesses cooperativity), and may bind their guest at the same time or sequentially. This means that there is no standard mathematical model for each binding/assembly of defined stoichiometry, but rather models have to be customized for each variation. The result is a considerable broadening of the range of base mathematical models which is beyond the scope of this introduction. Nevertheless, it is instructive to consider some of the general principles. In the 1 : 1 binding systems, we devised a base mathematical model (36) that gave an exact solution to the association constant determination. Such a mathematical model is often called a closed-solution approach. There is in fact an alternative to this, and that is to carry out a open-form approach. We have not mentioned this up until this point because the mathematics of the 1 : 1 complexation is simple enough that an exact solution can be readily sought. In the case of higher stoichiometries, however, an exact (closed) solution is often difficult or impossible. As we discuss below, the only option therefore is to take an open approach to such systems. Let us briefly consider stoichiometry of higher order systems. Not unexpectedly, the overarching theme here is that, the greater the number of entities involved in the final chemical entity, the more complicated the mathematics needed to describe the system in question. Let us begin by expanding (19) to give a general equilibrium equation for higher order assembly: mH + nG  Hm Gn

(56)

where Ka = [Hm Gn ]/[H ]m [G]n . The corresponding mass balance equations are: Ht = [H ] + m[Hm Gn ] and Gt = [G] + n[Hm Gn ], where Ht and Gt are respectively the total host and guest present in solution. It is possible to again obtain an expression of free guest ([G]) such as was carried out for the 1 : 1 complexes (36), but as the stoichiometry increases, the order of the polynomial generally increases as well. For example, in most ternary host–guest systems, the base mathematical model, or closed-form expression of [G], takes the form of a cubic equation (general form: f (x) = ax3 + bx2 + cx + d), whereas a binding event involving four species generally takes the form of a quartic function (general form: f (x) = ax4 + bx3 + cx2 + dx + e). Hence,

19

even for a “simple” ternary complex, the exact (or closedform) solution of the required cubic equation is relatively involved. As just mentioned, in a host with two binding sites, the nature of the binding sites, whether there is cooperativity in binding, and whether there is an order to guest complexation, all modify the base mathematical model. The nature of the recognition sites is straightforward; they can either be identical or different. Regarding cooperativity in the system, there are three possibilities: (i) the two binding pockets are independent of one another and there is no cooperativity; (ii) the net free energy change of binding for the overall process is more negative than the sum of the individual free energy changes arising from binding each single guest: in other words the system displays positive cooperativity; (iii) the system is negatively cooperative system, that is, the net free energy change for the overall process is less negative than the sum of the free energy changes of the individual binding events. Finally, binding can occur randomly or sequentially. The simplest combination of all these variations is where the two binding sites are identical and independent of each other, in which case the exact solution base model simplifies to a quadratic equation. If the binding sites are different, and/or there is cooperativity involved, and/or there is an order to complexation, then the base mathematical models are generally cubic equations. As with 1 : 1 complexations, the next step after a base model has been chosen is to modify it in order for the model to fit the chosen analytical technique. Software that models the different systems arising from combinations of these factors is available in modern ITC instruments, and a number of researchers have provided their own in-house software for those techniques such as NMR that are not specifically designed for binding constant determinations.46, 47 As alluded to above, an alternative to closed solutions of higher polynomials is to use an open, or iterative, approach. This more general strategy to parsing out the thermodynamic data for each microscopic binding event within higher order systems requires that each equilibrium constant expression and corresponding mass balance equation be identified. From this, the concentration of each species at set Ka values can be formulated by iteration, and this process “layered” on top of the normal iteration process that fits the species distribution to the experimentally obtained data.48 The advantage of an open solution approach is that it can be readily expanded for higher stoichiometric systems that are too difficult, if not impossible, to analyze via closed-form solutions. It is worth noting that many researchers make available in-house software for these calculations.49–51 A significant disadvantage of this approach, however, is that this dual iterative approach requires an initial estimate of the association constants at each microscopic step; and the more complicated the system is, the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

20

Concepts

more the accuracy required for these estimations. The reason for this increased accuracy requirement is that different initial Ka estimates can lead to different final Ka values. It is therefore prudent to begin with a good estimate of the Ka values and make sure that the final Ka values represent a “global minimum” by changing the starting point to see if this affects the outcome. It is also prudent to use chemical intuition when considering the final data. For example, the garnering of four association constants from a binding isotherm possessing only one inflection point should be treated with considerable suspicion. Likewise, if the magnitude of the obtained data seems overtly high or low relative to published data, the practitioner should be cautious. Of the numerous practical issues that can arise in studying higher order systems, perhaps the most important is the strength of cooperativity. The degree or strength of cooperativity can be determined using the Hill equation.12 For practical purposes, however, we simply need to be mindful of cases where binding is strongly cooperative because this will affect our ability to observe selected intermediates. For example, if it is strongly positively cooperative (Ka2  Ka1 ), then the addition of host to guest may not allow the observation of the intermediate HG because the excess guest present will promote the full conversion of HG to HG2 . In cases where exchange is slow on the NMR timescale, this would mean that no peaks corresponding to HG would be apparent; whereas if exchange is fast on the NMR timescale (or we are using UV–vis or ITC), no inflection point corresponding to this complex would be apparent in the binding isotherm. If this is the case, then only the overall (macroscopic) thermodynamic data can be obtained. One possible way to obtain the microscopic data is to perform a reverse titration of guest into host, but only if the positive cooperativity is not too strong. Similarly, if strong negative cooperativity is observed, it may not be possible to observe the formation of HG2 even if the host is titrated into excess guest. These influences of cooperativity mean that it is often instructive to perform both a forward and a reverse titration when studying higher order systems.49

4

CONCLUSION

We have focused here on the theoretical and practical aspects of determining the thermodynamic profiles of 1 : 1 host–guest complexations. The determination of binding constants, the change in free energy, the change in enthalpy, and the change in entropy for complexation processes has had a profound influence on our understanding of how structure and solvent effects influence intermolecular reactions. Quantifying these thermodynamic parameters has

also allowed researchers to identify previously underappreciated noncovalent forces, and how they can play a role in the properties of molecules. As a result, en masse, these studies have been responsible for defining and uniting the field of supramolecular chemistry. As the field moves forward and begins to address new challenges of a second phase, it is likely that binding constant determinations will continue to play a significant role. That role may change, but the need to quantify thermodynamic parameters of association is of such importance in both synthetic and biological contexts that a good grounding will always be essential. With that in mind, we hope that readers have found this introduction to the topic useful.

REFERENCES 1. E. Fischer, Ber. Deutsch. Chem. Ges., 1894, 27, 2985–2993. 2. D. J. Cram, Angew. Chem. Int. Ed. Engl., 1988, 27, 1009–1020. 3. J.-M. Lehn, Angew. Chem. Int. Ed. Engl., 1988, 27, 89–112. 4. C. J. Pedersen, Angew. Chem. Int. Ed. Engl., 1988, 27, 1021–1027. 5. C. A. Schalley, Analytical Methods in Supramolecular Chemistry, Wiley-VCH, Weiheim, 2007. 6. E. V. Anslyn, J. Am. Chem. Soc., 2010, 132, 15833–15835. 7. R. F. Ludlow and S. Otto, Chem. Soc. Rev., 2008, 37, 101–108. 8. B. C. Gibb, Nat. Chem., 2009, 1, 17–18. 9. B. C. Gibb, Nat. Chem., 2009, 1, 252–253. 10. A. C. Balazs 1632–1634.

and

I. R. Epstein,

Science,

2009,

325,

11. G. M. Whitesides and R. F. Ismagilov, Science, 1999, 284, 89–92. 12. E. V. Anslyn and D. A. Dougherty, Modern Physical Organic Chemistry, University Science Books, Sausalito, 2006. 13. P. Atkins and J. de Paula, Physical Chemistry, W. H. Freeman and Company, New York, 2006. 14. (a) K. N. Houk, A. G. Leach, S. P. Kim, and X. Zhang, Angew. Chem. Int. Ed., 2003, 42, 4872–4897; (b) R. R. Krug, W. G. Hunter, and R. A. Grieger, Nature, 1976, 261, 566–567. 15. S. R. de Groot and P. Mazur, Nonequilibrium Thermodynamics, Dover, New York, 1984. 16. I. R. Epstein and K. Showalter, J. Phys. Chem., 1996, 100, 13132–13147. 17. I. R. Epstein, J. A. Pojman, and O. Steinbock, Chaos, 2006, 16, 037101. 18. J. Rao, J. Lahiri, L. Isaacs, et al., Science, 1998, 280, 708–711. 19. K. A. Connors, Chem. Rev., 1997, 97, 1325–1358.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

The thermodynamics of molecular recognition

21

20. M. V. Rekharsky and Y. Inoue, Chem. Rev., 1998, 98, 1875–1917.

36. D. Harries, D. C. Rau, and V. A. Parsegian, J. Am. Chem. Soc., 2005, 127, 2184–2190.

21. E. Engeldinger, D. Armspach, and D. Matt, Chem. Rev., 2003, 103, 4147–4586.

37. K. A. Dill, Biochemistry, 1990, 29, 7133–7155.

22. Y. Liu and Y. Chen, Acc. Chem. Res., 2006, 39, 681–691. 23. H. Dodziuk, Cyclodextrins and their Complexes, WileyVCH, Weinheim, 2006. 24. C. S. Wilcox, N. M. Glagovich, and T. H. Webb, Designing Synthetic Receptors for Shape-Selective Hydrophobic Binding, ACS Symposium Series, ACS (American Chemical Society), Washington, DC, 1994, vol. 568 (Structure and Reactivity in Aqueous Solution), pp. 282–290. 25. Z. R. Laughrey and B. C. Gibb, Chem. Soc. Rev. 2011, 40, 363–386. 26. S. C. B. S. Granick, Science, 2008, 322, 1477–1478. 27. D. Chandler, Nature, 2005, 437, 640–647. 28. Y. Levy and J. N. Onuchic, Annu. Rev. Biophys. Biomol. Struct., 2006, 35, 389–415. 29. N. T. Southall, K. A. Dill, and A. D. J. Haymet, J. Phys. Chem., 2002, 106, 521–533. 30. J. Israelachvili, Intermolecular & Surface Forces, Academic Press, San Diego, CA, 1991. 31. P. Setny and M. Geller, J. Chem. Phys., 2006, 125, 144717. 32. J. Ewell, B. C. Gibb, and S. W. Rick, J. Phys. Chem. B, 2008, 112, 10272–10279. 33. E. S. Stoyanov, I. V. Stoyanova, and R. A. Reed, J. Am. Chem. Soc., 2010, 132, 1484–1485. 34. S. F. Dec, K. E. Bowler, L. L. Stadterman, et al., J. Am. Chem. Soc., 2006, 128, 414–415.

38. K. A. Connors, Binding Constants: The Measurement of Molecular Complex Stability, 1st edn, John Wiley & Sons, Inc., New York, 1987. 39. K. Hirose, J. Incl. Phenom. Macrocycl. Chem., 2001, 39, 193–209. 40. M. M. Pierce and B. T. Nall, Methods, 1999, 19, 213–221. 41. A. Valazquez-Compoy, S. A. Leavitt, and E. Freire, Methods Mol. Biol., 2004, 261, 35–54. 42. W. B. Turnbull and A. H. Daranas, J. Am. Chem. Soc., 2003, 125, 14859–14866. 43. B. W. Sigurskjold, Anal. Biochem., 2000, 277, 260–266. 44. N. J. Buurma and I. Haq, Methods, 2007, 42, 162–172. 45. E. V. Anslyn, J. Org. Chem., 2007, 72, 687–699. 46. P. J. Munson and D. Rodbard, Anal. Biochem., 1980, 107, 220–239. 47. A. P. Bisson, C. A. Hunter, J. C. Morales, and K. Young, Chem.—Eur. J., 1998, 4, 845–851. 48. J. Huskens, A. Mulder, T. Auletta, et al., J. Am. Chem. Soc., 2004, 126, 6784–6797. 49. J. C. D. Houtman, P. H. Brown, B. Bowden, et al., Protein Sci., 2007, 16, 30–42. 50. P. Gans, A. Sabatini, and A. Vacca, Talanta, 1996, 43, 1739–1753. 51. O. Raguin, A. Gruaz-Guyon, and J. Barbet, Anal. Biochem., 2002, 310, 1–14.

35. K. J. Tielrooij, N. Garcia-Araez, M. Bonn, and H. J. Bakker, Science, 2010, 328, 1006–1009.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc005

Cooperativity and the Chelate, Macrocyclic and Cryptate Effects Richard W. Taylor1 , Rowshan Ara Begum2 , Victor W. Day2 , and Kristin Bowman-James2 1 2

University of Oklahoma, Norman, OK, USA University of Kansas, Lawrence, KS, USA

1 Introduction 2 Supramolecular Coordination Chemistry 3 Transition Metal Coordination Chemistry 4 Supramolecular Chemistry 5 Conclusions Acknowledgments References

1

1 2 3 8 24 25 25

INTRODUCTION

Cooperativity and the chelate, macrocyclic and cryptate effects are terms that were coined at different times during the twentieth century. Cooperativity involves a process where multiple (two or more) binding sites interact to bind a guest. It is considered to be positive when the stability of the resulting complex is greater than the sum of the individual interactions. However, there are also examples of negative cooperativity, where the process of interaction of the binding sites gives energetically unfavorable results, usually from undesirable steric or electronic effects. This chapter is devoted primarily to a discussion of positive cooperativity that involves the chelate, macrocyclic, and cryptate effects, all of which utilize the interaction of two or more binding sites in tandem to achieve more stable

host–guest complexes. A key historical example of positive cooperativity is found in the metalloprotein hemoglobin. This exceptionally efficient oxygen transport protein binds up to four O2 molecules sequentially at different sites. Each additional O2 binds with higher affinity than the previous one due to changes induced in the tertiary structure of the metalloprotein by the previous binding events.1 The chelate effect refers to enhanced stabilities achieved in complexes where binding of a ligand (potentially referred to as a host) to a guest (traditionally a metal ion) is stabilized by the presence of more than one binding site on the ligand. The macrocyclic and cryptate effects build on the properties found for the chelate effect as a result of increased dimensionality and structure that is provided to the binding process. The macrocyclic effect reflects the elevated stability of macrocyclic complexes by virtue of a closed ring system that binds a metal ion or other guest at multiple sites. In this case, the ligand is even less readily released or dissociated because of the constraints placed on movement of any one binding site by virtue of the closed ring. The cryptate effect involves the highest form of complex, or host–guest, stability as a result of the increased dimensionality provided by the bicyclic (or higher order cyclic) cage. In the beginning of the increased cognizance of the different types of chemical influences that multiple binding sites can impart, the focus was totally on transition metal coordination complexes. Indeed, for the majority of these effects (with the exception perhaps of the cryptate effect), supramolecular chemistry was not even on the radar screen. Now, however, decades after the term “supramolecular” was coined by the now Nobel Laureate Jean-Marie Lehn,2

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

2

Concepts Increasing organization

Acyclic (podand) No preorganization

Macrocyclic effect

Chelate effect

Cryptand effect

= Donor site = Acceptor site

Figure 1 A pictorial representation of effect of increasing host organization with increasingly restricted binding of a guest within chelate, macrocyclic, and cryptand ligands.

it is evident that the same phenomena can be attributed to supramolecular chemistry. Another influence on binding, both in transition metal chemistry and supramolecular chemistry, is the role of preorganization. Preorganization tends to enhance cooperativity, since it infers that a host is already conformationally set in place for the most efficient binding (Figure 1). In general, preorganization is mandated for the constrained cyclic hosts, both macrocyclic and macrobicyclic, that force a preordained structure, such as in porphyrins. However, it can also occur at the chelate level, in a conformational rigidity (e.g., donor groups appended to a phenyl ring in cis positions) and/or proximal H-bonding effects (e.g., in 2,6diamidopyridine groups, where the pyridine NH groups are drawn inward to H bond with the pyridine nitrogen atom). Preorganization is discussed in detail in another chapter (see Complementarity and Preorganization, Concepts), but is still an important contributor to the effects described here. Prior to exploring various aspects of cooperativity and its influence on the chelate, macrocyclic and cryptate effect, it is important to understand the role of coordination chemistry as it relates to both transition metal and supramolecular chemistry. What follows is an explanation of this relationship that includes both bonding similarities and differences.

2

SUPRAMOLECULAR COORDINATION CHEMISTRY

In the late 1800s, the visionary Alfred Werner predicted the actual structures of transition metal coordination complexes in the absence of X-ray crystallography or other definitive structural tools.3 He put forth the hypothesis that transition metal ions had not just one, but two valencies. The first would be the oxidation number of the metal ion, +1, +2, +3, and so on, that would require a sufficient complement of counterions to satisfy the neutrality principle. However,

beyond the necessity of achieving neutrality, Werner proposed that transition metal ions possessed a secondary valence. This would be a “coordination number” governed by neutral species (e.g., H2 O or NH3 molecules), anions (e.g., Cl− , CN− , CH3 CO2 − ), and even more complex ligand frameworks that could donate a lone pair (or pairs) of electrons to the metal ion, forming dative or coordinate covalent bonds. This proclamation revolutionized transition metal chemistry and provided the basis for the new and still expanding field of coordination chemistry. In the mid-twentieth century with the birth of supramolecular chemistry, scientists began exploring the influence of weaker (compared with covalent bonds) “supramolecular” interactions. A new and vibrant field of chemistry has now evolved from almost unnoticed beginnings. Supramolecular chemistry includes a plethora of possible host–guest systems held together by noncovalent forces, and in many instances H bonds. These systems represent the primary focus of this chapter, although, because of the groundwork laid by the chemistry of transition metal complexes, aspects of traditional coordination chemistry are also included. Although binding modes are considerably different between transition metals, cations, anions, and even in the rarer case, neutral molecules, there are nonetheless striking similarities if a broader view of binding is considered. Some specific examples are shown in Figure 2. In transition metal complexes, coordinate covalent bonds are formed

M n+

R3N:

M′n +

R2O:

(a)

(d) R2O:

M = Transition metal M′ = Nontransition metal A = Anion

:An −

R3NH

(b) HNH2

R+

(c)

Figure 2 Comparison of binding modes for transition metal ions, cations, and anions.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

Cooperativity and chelate, macrocyclic and cryptate effects between the ligand electron pair donors and the metal ion (Figure 2a). These donations lead to interesting stabilizations (known as crystal field stabilization) because of the (usually) unfilled d orbitals on the metal ions.4 For other cationic species, a similar donation of the lone electron pairs can occur, either to a nontransition metal ion (potentially electrostatic but still involving electron pair interactions) or to a multiatomic cation such as ammonium ion (electrostatic and H bonding) (Figure 2b and c).5 For anions, the electron pair donation is reversed and the flow proceeds from the anion to the “ligand” hydrogen atoms, that is, H bonding (Figure 2d). This chapter addresses cooperativity and the chelate, macrocyclic and cryptate effects by examining the corollaries and differences between transition metal and supramolecular coordination. These four effects have been responsible for many of the exciting developments in supramolecular chemistry, from simple sensor and sequestration agents to more complex molecular self-assemblies and functional molecular machines and devices. The following sections describe the evolution and the interrelations of the four effects, beginning with the transition metal basics that laid the groundwork starting in the 1940s (Section 3) as shown in Figure 2(a), and progressing to the supramolecular aspects (Section 4) as shown in Figure 2(b)–(d). The latter section begins first with nontransition metal (ionic) examples, which started to materialize in the 1970s, and progresses to nonmetal (H bond) hosts and guests that are still being formulated. Throughout this chapter, the terms host (receptor) and ligand are used interchangeably, where the term ligand refers to the species making up the secondary valence of transition metals in transition metal coordination chemistry.

3

3.1

TRANSITION METAL COORDINATION CHEMISTRY Chelate effect

In traditional coordination chemistry, chelates (from the Greek word for claw, χηλη, ´ chel`e) refer to complexes with a ligand that contains more than one donor atom. The number of donor atoms in a given ligand is referred to as the denticity (from the Latin dentis for teeth). The chelate effect was first coined in the 1940s6, 7 led by Schwarzenbach, in some of the formative years of coordination chemistry. The area began to flourish in the 1950s, when researchers, such as Martell and Calvin,8 Bjerrum,9 and Schwarzenbach,10 were able to examine complex solution equilibria.11 As noted above, a chelator is a ligand that has more than one donor atom that is capable of binding to a metal

3

ion simultaneously. For example, ethylenediamine (en) or bipyridine (bipy) can form a complex with Cu2+ where both amine nitrogens are coordinated to the metal ion forming a five-membered chelate ring. In a majority of cases, the numerical value of the complexation constant for a chelate ligand with n donor atoms is larger than the comparable overall stability constant for a complex consisting of the same number of unidentate ligands with the same donor atom. This phenomenon (enhanced stability constant) has been termed the chelate effect.6, 7 For example, the formation constant (K1 ) for the reaction of Cu2+ with en (K1 = 2.5 × 1010 ) may be compared with overall constant (β 2 ) for the reaction of two molecules of the monodentate ligands NH3 (β 2 = 6.8 × 107 ) or CH3 NH2 (β 2 = 3.2 × 107 ) as shown in (1–3). Cu2+ + en ⇔ Cu(en)2+ K1 = [Cu(en)2+ ]/[Cu2+ ][en] 2+

Cu

+ 2NH3 ⇔ Cu(NH3 )2

(1)

2+

β 2 = [Cu(NH3 )2 2+ ]/[Cu2+ ][NH3 ]2

(2)

Cu2+ + 2CH3 NH2 ⇔ Cu(CH3 NH2 )2 2+ β 2 = [Cu(CH3 NH2 )2 2+ ]/[Cu2+ ][CH3 NH2 ]2

(3)

Figure 3 shows some simple chelators and their comparable monodentate reference ligands. The ligands chosen are the simple mono- and bidentate amines often used in supramolecular and self-assembled structures, especially the pyridine derivatives. The energetics of the chelate effect become more clear upon examining both the formation constants and thermodynamic parameters of the two nickel(II) reactions12 (4 and 5): [Ni(H2 O)6 ]2+ + 6NH3 = [Ni(NH3 )6 ]2+ + 6H2 O

(4)

−1

β ∼ 10 , G = −51.8 kJ mol , 9

H = −100 kJ mol−1 , S = −163 J mol−1 K−1 [Ni(H2 O)6 ]2+ + 3en = [Ni(en)3 ]2+ + 6H2 O

(5)

β ∼ 1018 , G = −101.8 kJ mol−1 , H = −117 kJ mol−1 , S = −42 J mol−1 K−1 There are many factors to be considered in examining the enhanced stability provided by chelating ligands. These include both thermodynamic and kinetic effects. Thermodynamically, an early explanation was that the origin of the effect was entropic in nature. This can be seen by comparing reactions in (4 and 5). The number of species in solution remains the same before and after the reaction in (4), and the overall G is favorable at −51.8 kJ mol−1 . In (5), however, the four reactant species are replaced by seven product

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

4

Concepts Monodentate NH3 Ammine

CH3NH2 Methylamine

N Pyridine py

Bidentate CH3 O

H2N

O

Acetate

H3C

O NH2

−

N

N

O

Ethylenediamine 2,2'-Bipyridine en bipy

N Picolinate

Tridentate

CH3

N N − HO O Dimethylglyoximate DMG

Tetradentate

NH

HN

NH

HN

NH2

H2N

NH2

H2N

N N

N

2,2',2''-Terpyridine terpy

Figure 3

1,4,7,10-Triethylenetetramine 1,4,8,11-Triethylenetetramine trien 2,3,2-tet

Monodentate and chelating ligands and abbreviations.

ions/molecules, contributing to more disorder and an even more favorable (more positive) entropy situation. (Counterions remain the same in both so are not included in this count.) Thus, the less negative S for (5) compared to (4), in addition to a slightly more favorable H , results in almost doubling the G (G = H − T S). Another consideration is the cooperative influence of chelating ligands. In the 1970s, Busch termed this effect “multiple juxtapositional fixedness” (MJF).13, 14 When a group of monodentate ligands is attached to a metal ion, dissociation becomes rather simple, since the individual ligands are not tied to each other (Figure 4a). However, in bi- and multidentate ligands, such is not the case. For a bidentate ligand, upon dissociation of one donor, the freed end is still held in proximity due to the coordination of the second donor (Figure 4b). Hence, there is more opportunity for the ligand to recombine with the metal ion as opposed to dissociating. By increasing the denticity, this effect becomes even greater, since complete dissociation would require multiply tethered donors to be released from the metal ion (Figure 4c). Thus, binding sites in multidentate ligands are considered to be fixed to their “juxtapositioned” counterparts. The MJF effect, which involves the kinetic aspect of the stability of these complexes, is also operative upon the initial binding of ligands. In both cases, cooperativity is involved, because the binding of the donor atom under consideration depends on the binding of previous donor atoms in the same ligand. The process of losing or dissociating the entire macrocycle, which is a totally closed ring, thus becomes even more difficult, and is the origin of

NH3 H3N M NH H3N 3

NH3 H3N H3N M NH3

NH3 H3N H 3N M

NH3

(a)

H N H2N N M N H H

H N N M N H H

NH2

H N M N H N 2 H

(b)

H N N HN N M N H H

NH2

N N N M N H H

N N M N H H

H N N N M H

(c)

H N N HN N M N

(d)

H H N N M N N H H

NH2

N N N M N

N N M N

H N N N M HN

Figure 4 Schematic diagrams showing the dissociation pathways and the influence of the MJF effect in the case of (a) monodentate (no effect), and increasing effect in (b) bidentate, (c) tetradentate, and (d) macrocyclic ligands.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

Cooperativity and chelate, macrocyclic and cryptate effects the macrocyclic effect to be discussed in the next section (Figure 4d). It should be noted, however, that a fundamental problem with quantitatively calculating the chelate effect from Kchelate /β mono is that the equilibrium constants do not have the same molecularity when expressed in molar units (i.e., K1 = M−1 and β 2 = M−2 ).15 Indeed, the (numerical) chelate effect almost disappears when the concentrations are expressed as mole fractions, with similar findings observed in gas-phase measurements. This same problem arises in the supramolecular chelate effect, and will be described in greater detail in Section 4. However, in the solution phase, the reaction of complexes containing monodentate ligands with chelating ligands usually results in a favorable binding constant, as observed for the displacement of ammines in [Ni(NH3 )6 ]2+ with en (6).11, 12 [Ni(NH3 )6 ]2+ + 3en = [Ni(en)3 ]2+ + 6NH3 log K = 8.76

(6)

Table 1 lists formation constants with selected metal ions as an illustration of the chelate effect for transition metal ions. The structures of the ligands and their abbreviations are shown in Figure 3. In the table, the chelate effect is defined as log K = log K1 − log β n , which is equivalent to log Kexch for the exchange reaction (7)

M(L)n + Ln ⇔ M(Ln) + nL Kexch = [M(Ln)][L] /[M(L)n ][Ln] n

Reaction: Two-coordinate

1 2 3 4 5 6 7 8 9

Ni2+ + 2NH3 ⇔ Ni(NH3 )2 2+ Ni2+ + en ⇔ Ni(en)2+ Ni2+ + 2py ⇔ Ni(py)2 2+ Ni2+ + bipy ⇔ Ni(bipy)2+ Cu2+ + 2NH3 ⇔ Cu(NH3 )2 2+ Cu2+ + 2CH3 NH2 ⇔ Cu(CH3 NH2 )2 2+ Cu2+ + en ⇔ Cu(en)2+ Cu2+ + 2py ⇔ Cu(py)2 2+ Cu2+ + bipy ⇔ Cu(bipy)2+

LogK1 (polydentate ligand) = 1.152 log β n (unidentate ligand) + (n − 1) log 55.5

n = denticity

10 11

Ni + 3py ⇔ Ni(py)3 2+ Ni2+ + terpy ⇔ Ni(terpy)2+

12 13 14 15

Cu2+ + 4NH3 ⇔ Cu(NH3 )4 2+ Cu2+ + 2en ⇔ Cu(en)2 2+ Cu2+ + trien ⇔ Cu(trien)2+ Cu2+ + 2,3,2-tet ⇔ Cu(2,3,2-tet)2+

Log Kn (β n )a (Kcalc )b

log K c (#)d

5.08 7.35 3.10 7.04 7.83 7.51 10.40 4.45 8.00

— 2.27 (1) — 3.94 (3) — — 2.57 (5); 2.89 (6) — 3.55 (8)

(β 2 ) (K1 ) (7.58) (β 2 ) (K1 ) (β 2 ) (β 2 ) (K1 ) (10.76) (β 2 ) (K1 )

Reaction: Three-coordinate 2+

3.71 (β 3 ) 10.7 (K1 )

— 7.0 (10)

Reaction: Four-coordinate

aK

13.0 (β 4 ) 19.6 (K1 ) 20.05 (K1 ) (20.20) 23.2 (K1 )

— 6.6 (10) 7.0 (10) 10.2 (10)

and β n defined in (1 and 2). refers to the K calculated using (8). c log K = log K chelate − log Kunidentate . d (#) refers to the number of reaction that is being compared. bK

(7)

where Ln is a chelating ligand with n identical donor atoms and L is the monodentate reference ligand. Table 1 covers many of the important considerations of the chelate effect in transition metal complexes. It is divided to show examples of two-coordinate, three-coordinate, and four-coordinate binding constants. First, examination of the data shows that the chelate effect is not confined to a particular metal–ligand system, as seen for Ni2+ and Cu2+ , and is operable for both aliphatic and aromatic amines. For simple bidentate ligands such as en and bipy, the chelate effect ranges from 2.27 for Ni(en)2+ to 3.94 for Ni(bipy)2+ , and is larger for bipy than en with the same metal ion. As noted earlier, the chelate effect also increases with increasing number of rings. Transition metal chemists have applied a very simplified approach to the chelate effect for linear polyamines, expressed in (8), that corrects for inductive effects (pKa (CH3 NH2 )/pKa (NH3 )).11

Table 1 Complex formation constants for reactions of metal ions with multidentate ligands and their unidentate analogs.16 Number

5

1

calc

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

(8)

6

Concepts

The latter term, (n − 1) log 55.5, derives from the fact that when coordinated water molecules are replaced by chelating ligands, the increase in the number of molecules in solution causes an increase in the entropy in the amount S = nR ln 55.5 = 33.4n J mol−1 K−1 , where n refers to the number of chelate rings. As noted for the en and trien systems in Table 1 for both Cu2+ and Ni2+ ions, this approximation works quite well for the linear polyamines.11, 12 Also note as seen in Table 1, the chelate effect increases with an increase in the number of chelate rings for a given metal ion, but this can be offset by cumulative ring strain for some metal ion complexes. This is illustrated by comparison of the formation constants for Cu2+ complexes with NH3 , en, trien, and 2,3,2-tet given for #12–15 in Table 1. The log K values, compared to log β 4 (NH3 ), are 6.6, 7.0, and 10.2 for two en ligands, trien, and 2,3,2tet, respectively. However, the increase in log K is much greater for 2,3,2-tet, where a six-membered chelate ring is formed by coordination of the metal ion to the two interior nitrogen atoms. The observed increase in stability has been ascribed to the lessening of ring strain due to the expanded six-membered chelate ring in 2,3,2-tet.11, 12 It should be noted, however, that the actual strain associated with ring size will also depend on the size of the metal ion. For transition metal ions, in considering ring sizes from four to seven, a four-membered ring is the most strained (as in acetate); a five-membered ring (as in en) is optimal for larger transition metal ions; and six- and sevenmembered rings allow for increasing flexibility, but they are more favorable for binding with smaller ions. These observations derive mainly from differences in bond lengths and bond angles that give rise to greater or lesser strain in the cyclic systems.11 Additionally, it should be noted that the distances from the transition metal ions to the ligand donors are greater than distances between atoms in the ˚ compared with 1.4–1.5 A), ˚ ligand (usually around 2.0 A which influences the strain introduced by various ring sizes. Another level of cooperativity can be observed for the chelate effect in terms of bonding of additional chelating ligands, as seen in the examples below. In coordination chemistry, a simple form of cooperativity can be observed in the successive complex formation constants (KMLn) for certain metal–ligand complexation reactions (9 and 10). Mx+ + Ly− ⇔ MLx−y KML1 = [MLx−y ]/[Mx+ ][Ly− ]

(9)

x−2y

MLx−y + Ly− ⇔ ML2 x−2y

KML2 = [ML2

]/[MLx−y ][Ly− ]

(10)

In addition to sequential binding effects of multiple chelates, other mitigating factors can play a role in the

O

H

Im O

N

N Fe

N O (a)

Fe N

Im

O H Im =

NH N

(b)

Figure 5 (a) Structure of the Fe(DMG)2 (Im)2 complex and (b) electron density map for the Fe(DMG)2 pseudomacrocyclic ˚ ring around the iron atom with contours at increments of 0.5 e− /A ˚ starting at 0.2 e− /A.

binding of ligands to transition metals. For example, the magnitude of the equilibrium constants generally decreases as successive ligands coordinate to the metal center due to a combination of statistical, electrostatic, and steric factors; that is, KML1 > KML2 .17 For example, when M = Ca2+ and L = picolinate anion, a bidentate ligand, K1 = 380 for binding of the first ligand, and K2 = 24.3 for the second ligand.18 On the other hand, the Cu2+ and Zn2+ complexes with dimethylglyoxime (DMG, also a bidentate ligand), show cooperative behavior in metal complex formation. In this case, the ligand readily deprotonates and forms two H bonds, which results in the formation of a “pseudo” macrocyclic ligand. K2 ≥ K1 by factors of 32 and 16 for the Cu2+ and Zn2+ complexes, respectively.19 Two intramolecular hydrogen bonds between the oxygen atoms of the coordinated ligands are responsible for this effect and have been verified by X-ray crystallography.20, 21 The electron density diagram for the di-imidazole iron(II) complex with dimethylglyoximate,22 synthesized as an early model of the heme iron proteins, nicely illustrates the pseudomacrocyclic effect (Figure 5).

3.2

Macrocyclic effect

Macrocyclic chemistry had its beginnings in the 1960s. In 1962, Curtis published the first tetraaza macrocycle.23 However, the first planned synthesis of a macrocyclic ligand (and complex) came two years later, when Busch reported the use of a kinetic template effect and nickel(II) ion to achieve a mixed aza-, thia-donor macrocyclic ligand24 (Figure 6). Early macrocycles by Busch and others were generally based on nitrogen donor groups and were often used as models for the naturally occurring macrocycles such as the porphyrins. As the field of macrocyclic chemistry grew, so did the realization that macrocyclic complexes, particularly transition metal complexes, exhibited enhanced stabilities over noncyclic systems, even those with multiple chelate rings.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

Cooperativity and chelate, macrocyclic and cryptate effects

3.3 R

N

S Ni

N

S

+

Br

CH2Cl2 R

N

Cryptate effect

S Ni

Br

N

S

R = CH3, C2H5, C5H11

Figure 6 Final step of the reaction sequence resulting in the formation of the first planned macrocyclic ligand by a kinetic template effect.24

This meant that transition metal complexes that were kinetically labile and therefore not very easy to study on normal time scales could be made increasingly inert so as to enable room temperature study of the chemistry. This finding led Margerum and Cabbiness to coin the term “macrocyclic effect,” for the increased stability of macrocyclic complexes over their acyclic counterparts.25 The origins of the macrocyclic effect have been a subject of discussion for a number of years, and aspects of the following four factors undoubtedly play a role.26 Macrocyclic ligands are often preorganized in a fashion that readily allows for complexation. Solvation of the donor atoms is possibly less in the more limited macrocyclic cavity. The basicity of the macrocyclic ligands is influenced by the inductive effects of the bridges between the donor atoms, which increases the donor capabilities in macrocyclic ligands. Last but not least, the electron repulsion from the constrained donor lone electron pairs in macrocycles is eased upon metal ion coordination. The macrocyclic and cryptand ligands discussed in this chapter are depicted in Figure 7 along with their common names. In terms of thermodynamics, the role of entropy versus enthalpy has been hotly debated. Log K and thermodynamic values are provided in Table 2. As can be seen from the thermodynamic parameters provided in the table, it becomes evident that the macrocyclic effect is primarily enthalpic in origin. This is dependent, however, on comparing systems without steric strain, which naturally adds other mitigating factors to the thermodynamics.

Probably the most famous cryptands are those first reported by Lehn in 1969.27, 28 These will be described in the section on supramolecular chemistry (Section 4). However, transition metal cryptands were reported several years later, the clathrates and sepulchrates of Sargeson and coworkers29, 30 (Figure 7). The sepulcrates derive from a hexamine cagelike structure that encloses around a metal ion. They are actually a class of clathrochelates, the term clathro being derived from the Latin word meaning lattice. However, Sargeson named the ligand sepulchrates, more or less in keeping with the tone of the “crypt” in cryptand. This group of complexes has been studied at length because of the ease with which the captured metal ions undergo rapid redox changes in addition to a high complex stability.30 Many of these complexes are also excellent oxidizing agents in the higher oxidation states. A number of other cryptandlike macrocycles and their transition metal complexes have been synthesized, including the lacunar (dry cave) ligands of Busch used to bind small molecules such as oxygen31 (Figure 4). There are a number of variations on this theme, with polycyclic macrocycles of many shapes and varieties. However, since the focus of this review is on supramolecular chemistry, the reader is directed to a review of some of these interesting transition metal complexes.32 R′′ R′

N

NH HN

NH NH HN

N

N

NH HN

NH NHHN

N

N

R′

N

R

Cyclam Sepulchrate

R′′ Lacunar ligand

Figure 7

Macrocyclic and cryptand ligands.

Table 2 Complex formation constants and thermodynamic parameters for the macrocyclic effect for Cu(II), Ni(II), and Zn(II) with cyclic and acyclic tetraamines.a Complex Cu(2,3,2-tet)2+ Cu(cyclam)2+ Ni(2,3,2-tet)2+ Ni(cyclam)2+ Zn(2,3,2-tet)2+ Zn(cyclam)2+

7

log K

(log K)

G

(G)

H

(H )

−T S

23.2 26.5 15.5 19.4 12.6 15.5

— 3.3 — 3.9 (3) — 2.9 (5)

−132.1 −151.8 −90.4 −110.9 −72.3 −87.1

— −19.7 — −20.5 — −14.8

−115.9 −135.6 −77.9 −100.9 −49.8 −61.9

— −19.7 — −23.0 — −12.1

−16.2 −16.2 −12.5 −10.0 −22.5 −26.2

(−T S) — 0 — 2.5 — −3.7

at 25 ◦ C11 ; values of the G, H , and −T S in kJ mol−1 ; macrocyclic effect = (log K) = log Kcyclam − log K2,3,2-tet ; (X) = Xcyclam − X2,3,2-tet .

aH O 2

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

8

Concepts

4

SUPRAMOLECULAR CHEMISTRY

Various aspects of cooperativity and the chelate, macrocyclic and cryptate effect are described in this section for supramolecular complexes. As noted in Figure 2, there are two general types of interactions, electrostatic and H bonding, included in this section—however, in all cases these are noncovalent in nature. Three types of guests will be described: metal ions of the nontransition metal variety (primarily electrostatic interactions, Figure 2b), and nonmetallic cationic and anionic guests (potentially electrostatic and H-bonding interactions, Figure 2c and d). Both nonmetallic cationic and anionic guests utilize their H-bond donor as well as acceptor sites. Each of the sections below, on chelate, macrocyclic, and cryptate effects, will include several examples of two or more different types of guests.

4.1

Background

In supramolecular chemistry, stable host–guest complexes are built on multiple, simultaneous noncovalent interactions.33 This is because a single supramolecular interaction, being “beyond” the covalent or dative bond, is inherently weaker than the coordinative covalent or dative bond that is found in transition metal chemistry. Hence, stability is enhanced by the additive effect of multiple host–guest interactions. However, as Schneider points out, referring to the “additivity of noncovalent bonding” there is also a chelation influence when the interactions emanate from a single host (or ligand), as seen in transition metals. In essence, polytopic hosts can bind to either monotopic or polytopic guests, the latter situation of particular importance in host–guest chemistry, where an organic or inorganic guest may have more than one binding site. These principles are abundant in biology, where host–guest chemistry is the modus operandi for enzyme and protein interactions with their substrates, among other important processes including antibody–antibody interactions. Complementarity, as seen in the chelate effect for transition metal complexes, also plays an important role. The “fit” of the guest to the host should be as favorable as possible; otherwise, the full benefit of the chelate enhancement effect in binding by multiple interactions is lost. Thus, for example, in a simple system where the polytopic host and guest match each other (Figure 8), the total free energy of the interaction, Gtot , is obtained by the sum of the free energies of the individual interactions where H1 G1 refers to the binding site H1 of the host and G1 of the guest, and so on (11). However, the overall binding constant Ktot is not the product of the individual binding constants, because Ktot is in units of M−1 , while if each individual constant is multiplied, the Ktot would have dimensions

Host H1

H2

H3

H4

H5

G1

G2

G3

G4

G5

Guest

Figure 8 Schematic representation of multiple host–guestbinding site interactions.

M−n where n would be the number of pairwise contacts, in this case five. In actuality, the same problem occurs for Gtot as well, since if treated separately, each of the G values would involve a significant entropy term, where only one entropy term would be involved in the Gtot . In order to circumvent this problem, the same solution as applied to the transition metal quandary can be applied to supramolecular chemistry. Namely, when expressed as molar fractions instead of molarities, the equilibrium constants become dimensionless. This involves multiplying the molar concentrations of an 1 M aqueous solution of the solute by a factor of 1/55.5 to obtain molar fractions. When substituted in the multiple equilibrium equations and solving, the resulting dimensionless Ktot is obtained (12). G = GH1 G1 + GH2 G2 + GH3 G3 + GH4 G4 + GH5 G5

(11)

Ktot = (55.5)−1 KH1 G1 KH2 G2 KH3 G3 KH4 G4 KH5 G5 (12) Note first that (11) assumes complete additivity of the effects. Nonetheless, the snowballing effect of multiple interactions can be easily seen. It should be noted, however, that there are a number of factors that weigh in when determining the “additivity” of multiple binding events, especially for entropic considerations. These include, for example, ion pair versus neutral interactions; solvation and desolvation effects; and the phase under study, gas or solution among others. Nonetheless, it does appear that the experimental values of G additivity hold in general when rotational entropy is not compromised on complexation, solvation and desolvation effects are comparable for each binding site, and partnering host and guest sites are able to bind without inducing significant strain.33 Over and above the chelate effect in supramolecular chemistry, macrocyclic and cryptand hosts take advantage of a combination of effects that serve to leverage their binding capability (Figure 1). For example, because of preorganization effects of the closed cyclic systems, immediately more contact with the guest—especially spherical guests—occurs. Solvation effects are also important. In acyclic hosts, the binding sites are readily available

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

Cooperativity and chelate, macrocyclic and cryptate effects to the solvent, which then requires additional rearrangement upon complexation of a subsequent guest. As the “dimensionality” increases, the binding sites can be more shielded, allowing for less rearrangement upon guest binding (Figure 9). Even so, acyclic chelates benefit from the chelate effect once they start wrapping or binding a guest in their multiple sites. Macrocyclic and macrobicyclic or polycyclic hosts not only benefit from the chelate effect, but have the macrocyclic and, for the higher order hosts, the cryptate effect in their favor. On the other hand, with organization comes more repulsion, for example between the lone electron pairs in the crowns and cryptands in Figure 9. Repulsion is subsequently eased upon binding of a guest. The binding constants (log K) for a series of ionophores are provided in Table 3. They are listed in order of increasing preorganization as shown in Figure 10. The following sections provide examples of cooperativity as it relates to the chelate, macrocyclic, and cryptate effects for supramolecular complexes of both cations and anions.

4.2

Cations

Perhaps the most apparent similarities with transition metal complexes are the main group metal ions and their supramolecular complexes. Here, the progression from the chelate to macrocyclic to cryptate effect can be readily recognized and the thermodynamics are, for the most-part, established. The term ionophore is often used for this class of hosts, originating from its use in biology for lipid-soluble molecules that serve as vehicles for transporting ions across membranes.

4.2.1 Chelate effect Nature utilizes the chelate effect often in the naturally occurring ionophores. For example, the polyether carboxylic acid ionophore A23187 (calcimycin) behaves as a monoanionic tridentate ligand (Figure 11a). For Ca2+ the equilibrium constants KML1 and KML2 are 2.5 × 106 O

O

H3CO

O

O

Acycle = podand

OCH3

O

O

Relaxed conformation results in high solvation of lone pairs and minimal electron – electron repulsion.

O O O CH3 H3C O

O

O

O O

O

Lessened solvation of lone pairs O and increased repulsion of lone pairs pointing inward.

O

9

O O

O O

O

Macrocycle (crown ether) O N O O

O O N

Limited solvation of lone pairs in cavity but lone pair repulsion in the cavity is retained.

O

O N O O

O O N O

Macrobicycle (cryptand)

Figure 9 Preorganization effects in the binding of a spherical ion by the non-preorganized pentaglyme, and the effect of increasing organization progressing down the series to macrocyclic and cryptand hosts. (Redrawn from Ref. 34.  John Wiley & Sons, Ltd., 2009.) Table 3 Stability constants (log K) in methanol at 25 ◦ C for the binding of alkali metal ions with ionophores that have increasingly complex design and dimension.33 Ionophore Pentaglyme Tripod Valinomycin [18]Crown-6 [2.2.2]Cryptand Spheranda a

Li+

Na+

K+

Rb+

Cs+

— 12.4

(1) (3) (5) (5)

= log Kcyclic − log Kacyclic . Number of reaction that is being compared in parenthesis. c In methanol.55 d In CHCl3 saturated with H2 O, K values obtained from K = kf /kd .35 a (log K) b

Table 6 Comparison of the macrocyclic effect and differences in the thermodynamic parameters for complexes of selected alkali and divalent metal ions with pentaglyme and [18]crown-6.a Metal ion Na+ K+ Rb+ Cs+ Ba2+ Pb2+

(log K)b

(G)c

(H )c

2.9 4.0 3.4 2.6 4.7 4.8

−16.1 −23.0 −19.5 −15.3 −25.4 −27.2

−15.5 −2.5 −2.89 −11.0 −23.3 −18.6

(−T S)c 0.63 20.1 16.4 3.89 2.01 8.37

In methanol at 25 ◦ C, taken from Table A4 in Ref. 33, Ref. 55, or calculated from G. (log K) = log K[18]cr-6 − log Kpenta . c (X) = X −1 [18]cr-6 − Xpenta , (G), (H ), (−T S) in kJ mol . a b

effect enhances the binding by more than several orders of magnitude. An especially convincing case is that of the spherand, which is possibly the epitomy of the concept of preorganization, showing incredible binding despite its two dimensionality, that is, monocyclic as opposed to bicyclic or polycyclic. It is particularly interesting to compare the thermodynamic aspects of supramolecular chemistry with transition metal chemistry (Tables 2 and 6). If similar chemistry holds, the macrocyclic effect should be primarily enthalpic in origin. As can be seen from Table 6, the (H ) values show that the effect is primarily enthalpic for Na+ , Cs+ , Ba2+ , and Pb2+ , ranging from negative teens for the

mononegatively charged ions to the negative twenties for the dipositively charged ions. On the other hand, for K+ and Rb+ the trend is reversed, and it would appear that the favorable (G) is primarily entropic in origin. These values can be compared to the values for 2,3,2-tet and cyclam that illustrate this effect is primarily enthalpic in origin for transition metal ions (Table 2). Crown ethers and their derivatives also display a cooperative, multisite binding of molecular cations as well as metal ions. Here, H-bonding and electrostatic interactions play a role. The number of H bonds will also influence the affinities, as evidenced in the simple ammonium series, with NH4 + about equal to CH3 NH3 + (log K = 4.27 and 4.25,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

Cooperativity and chelate, macrocyclic and cryptate effects

O O

six potential donor groups (Figure 18). For more information about molecular cation guest complexes, the reader is urged to see reviews by Gokel et al.44 and Schneider.56 Another interesting example of cooperativity and the macrocyclic effect involves the taco complex of Gibson and coworkers.57 Inspired by the formation of pseudorotaxanes (threaded macrocycles) using paraquats reported by Stoddart and coworkers in the late 1980s,58 Gibson and coworkers capitalized on the more flexible bending capability of larger crown ethers to drive rotaxane formation in a cooperative-like binding sequence (Figure 19). They found that when they closed the crown to capture a guest inside, a 100-fold increase in the binding constant, K, was achieved. Two of the structures of the crown are shown in Figure 20, one with a H-bonding group linking the two hydroxyl sites, and one with a covalent chain closing off the cavity. As can be seen in the latter structure, a water molecule also manages to fit in the cavity, pushing the paraquat guest to the back of the host.

O

N O

O

O O

(a)

(b)

(c)

Figure 18 (a) Chemdraw diagram of the pyridine-containing crown, and (b) overhead and (c) side views of the crystal structures of benzylammonium ion bound to the crown.

respectively) > (CH3 )2 NH2 + (log K = 1.76), reflecting the cooperative additivity of binding sites.55 As can be seen from the crystal structure of a simple benzylamine complex with a pyridine-containing crown, the amine group dips into the crown, and forms H-bond interactions with all

Figure 19

15

Schematic representation of the complexation of paraquat by the flexible difunctionalized crown ether.

O

O

O

O

O

R R O

O

R = CH2OH R/R = O(CH2CH2O)4

O

O

O

+

H3C N

+

N CH3

(a)

(b)

(c)

Figure 20 (a) Chemdraw diagram of the reaction of the large crown ether with the paraquat guest, and crystal structures of (b) the paraquat guest in the crown complex (R = CH2 OH), closed off by H bonding between the hydroxyl groups and a CF3 CO2 − ion, and (c) a covalently closed cryptand/crown (R/R = O(CH2 CH2 OCH2 CH2 O)2 complex. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc007

16

Concepts

Table 7

Stability constants (log K)a for chloride salts of alkali and alkaline-earth metal ions with cryptands in H2 O. ˚ Metal ion (ionic radius, A)

Cryptand

[2.1.1] [2.2.1] [2.2.2] [3.2.2] [3.3.2] [3.3.3]

Cavity ˚ c size (A)

Li+ (0.86)

Na+ (1.12)

K+ (1.44)

Rb+ (1.58)

Cs+ (1.84)

Mg2+ (0.87)

Ca2+ (1.18)

Sr2+ (1.32)

0.8 1.15 1.4 1.8 2.1 2.4

4.30 2.50 100 kV to get good resolution. Ultrahigh-voltage TEM with the acceleration voltage >1000 kV is very useful for scanning a specimen with high thickness because of high transmissibility of electron beam, but it may damage a specimen by microstructural defects in it during observation. In general, supermolecules for TEM observation are put on a copper grid with a diameter of 3 mm, coated with a thin support film, to prepare the specimen. This specimen must be placed in a specimen holder, as shown in Figure 2(c), in order to insert into the TEM column. The specimen holder is a sophisticated device being capable of tilting the specimen in an appropriate angle inside the TEM column. As in the case of data recording unit, we can observe the magnified images on a fluorescent screen at the bottom of the TEM column. The images can be transferred onto photographic films and/or stored on a computer hard disk via CCD (charge-coupled device) camera. In principle, contrast of the TEM image arises because of the differences in electron density of the elements constituting the specimen. Thus, for the sample with less contrast originally, several methods have been developed.2, 4 In this chapter, we describe three types of specimen preparation

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Transmission electron microscopy (TEM)

100 nm (a)

Figure 3

(c)

Typical TEM images of lipid bilayer vesicles taken by (a) negative staining, (b) freeze-fracture, and (c) cryogenic method.

techniques that can be used for the analysis of various supermolecules: negative staining, freeze-fracture, and cryogenic (Cryo) method. The superimposing of structures of supermolecules in two-dimensional TEM images makes it difficult to understand the three-dimensional (3D) superstructures. TEM tomography is a powerful method to explore supermolecules in three dimensions. The 3D image can be created by using computer from a series of TEM images taken by tilting the sample stage. For the supermolecules with periodic ordered structure, electron diffraction is an indispensable analytical method of TEM. Electron diffraction pattern contains information on the crystal structure, lattice repeat distance, and specimen shape. The pattern is always related to the image of the area of the specimen. There are basically two types of electron diffraction methods, selected area electron diffraction (SAED) and nanobeam electron diffraction (NBED), in which size of the area is of submicrometer scale in the former and of nanometer scale in the latter. In addition, two spectroscopic measurements done by using the TEM apparatus, EELS and EDS, are effective for structural analysis of the specimen.

3

200 nm

100 nm (b)

3

SPECIMEN PREPARATION TECHNIQUES FOR TEM IMAGING

As mentioned above, TEM is a powerful method for the visualization and analysis of the supramolecular structures with a resolution of nanometer scale. However, in general, we are able to evaluate the sample by using TEM only under limited conditions due to theoretical or technical restrictions. TEM has two significant limitations on the observation of supramolecular structures. First, popular TEM apparatus can visualize only heavy elements, which scatter electron beam strongly, whereas most of the supramolecular structures is formed by organic molecules without heavy elements. Secondly, the specimen for TEM observation must be placed in a high-vacuum condition, in which case the sample cannot be inserted into the TEM

in solution state. Consequently, a variety of techniques have been developed for the preparation of TEM specimen up to the present time. To obtain a reliable result on TEM observation, an appropriate method for the specimen preparation must be chosen, because each technique has different advantages and limitations. In this section, we describe the characteristics and protocols of three preparation techniques: negative staining, freeze-fracture, and Cryo method. Figure 3 shows the typical TEM images of lipid bilayer vesicles obtained by using these methods.

3.1

Negative staining TEM

For the evaluation of the supramolecular structure, negative staining is a convenient method for specimen preparation. In principle, the contrast of the image is enhanced by difference in location of heavy metal ions as the stain, which has high electron density, around the specimen. This method affords the negative picture in which the object is highlighted as a white image by the exclusion of stain. In this method, it is important to choose an appropriate stain, which does not interact with the object, to prevent positive staining and unexpected structural disruption. The characteristics of well-known stains are summarized in Table 1. The spatial resolution of negative staining is in the range of several nanometers depending on the type of stain used. This method does not require any special equipments and techniques. In addition, the object, which is unstable when irradiated using the electron beam, can be observed using TEM. However, the structure in solution may not be maintained during the preparation, because in this method the specimen is dried finally.

3.1.1 Protocol: preparation of the specimen for negative staining TEM Tools and materials Tweezers Micropipet (10 µl) Filter paper

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

4

Techniques

Table 1

Comparison of stains for negative staining TEM.

Stain Phosphotungstic acid [P2 O5 · 24WO3 · nH2 O] (anionic stain) Uranyl acetate [UO2 (CH3 COO)2 · 2H2 O] (cationic stain) Ammonium molybdate [(NH4 )2 MoO4 ] (anionic stain) a

Concentration (wt%)

pH

0.5–2

6.5–7.5a

0.5–2

4.0–4.2b

2–3

6.0–8.0c

Characteristics Small particle, but less contrast than uranyl acetate Large particle, but the highest electron density (good contrast) Applicable to wide targets, but low electron density

The solution pH can be adjusted by aqueous sodium hydroxide. solution pH can be raised to 6.0 by addition of aqueous sodium carbonate. The solution pH can be adjusted by aqueous ammonia.

b The c

Microcentrifuge tube Copper grid with 200 mesh, covered with thin support film such as carbon substrate. Preparation procedure There are two methods for the specimen preparation by negative staining. The researchers are recommended to test both methods to diminish the artifact that arises from the specimen preparation. It is also better to test different stain methods to evaluate the interactions between the object and stain. Premixed method 1.

2.

3.

4. 5.

Prepare an aqueous solution of the stain, and its pH should be adjusted using an appropriate acid or base as indicated in Table 1. Mix equal amount of sample and stain solutions in a microcentrifuge tube. Twenty microliters of each solution is sufficient for the specimen preparation. Place a drop (about 10 µl) of the mixture onto a grid held by a pair of tweezers and leave for 30 s. The grid must be pretreated with glow discharger to increase the hydrophilicity. Remove excess solution on the grid using filter paper. Gently dry the grid in a desiccator containing silica gel for overnight.

Sequential two-droplet method 1. 2.

Make the grid hydrophilic by glow discharger. Place a drop (about 10 µl) of the sample solution onto a grid and leave for 30 s. 3. Remove excess solution on the grid using filter paper. 4. Place a drop (about 10 µl) of an aqueous solution of stain and leave for 30 s. 5. Remove excess solution on the grid using filter paper. 6. Gently dry the grid in a desiccator containing silica gel for overnight.

As an example, the microscopic image of small unilamellar lipid vesicles prepared by the negative staining method is shown in Figure 3(a). The sample solution was stained with 2% of uranyl acetate on a continuous carbon support. The specimen is observed as a negative image, in which the object and background were highlighted as white and black images, respectively. The image provides enough resolution and high contrast to evaluate the microscopic structure of lipid bilayer membrane. The thickness of the lipid bilayer is estimated to be about 5 nm, which corresponds to the theoretical thickness with regard to the molecular length of lipid.

3.2

Freeze-fracture replica TEM

In contrast to negative staining, freeze-fracture replica method provides three-dimensional information of the microscopic structure of the object. In this method, we observe the replica of the frozen object by TEM. Specimen for freeze-fracture replica TEM is prepared by combination of several special techniques as schematically shown in Figure 4. First, the sample solution is rapidly frozen using coolant such as slush nitrogen. Next, the resultant ice in the amorphous state is fractured by a cryo knife, followed by the replica formation by vacuum deposition of heavy metal such as platinum and carbon. The heavy metal is deposited from an angle to make a contrast along the height of the object, or shadowing. Finally, the replica is removed from the frozen sample by melting the solution and transferred onto a TEM grid. The rapid-freezing process enables the observation of dynamic structure, whereas the negative staining is usually able to visualize the morphology in equilibrated state. Since the replica prepared by this method is stable against the intense irradiation of electron beam, the microscopic image can be obtained in high magnification. On the other hand, this method has two disadvantages: (i) the specimen preparation requires special machine (Figure 5) and techniques and (ii) in contrast to

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Transmission electron microscopy (TEM)

5

Shadowing Carbon layer Sample Matrix Rapid freezing

Figure 4

Fracture

Replica formation

Schematic representation of the TEM specimen preparation by freeze-fracture replica method.

5. Knife

6. Sample stage (b)

7.

8.

(a)

(c)

Figure 5 (a) Frozen specimen preparation machine, (b) inside the specimen chamber, and (c) sample holder.

9. 10.

the detailed view of an object’s surface, it is difficult to observe the inner structure of a complex object.

3.2.1 Protocol: preparation of the specimen for freeze-fracture TEM Equipment Frozen specimen preparation machine (for example, JEOL JFD-9010) Slush nitrogen generator (for example, JEOL JFD-380P). Tools and materials Tweezers Micropipet (10 µl) Copper grid with 200 mesh Liquid nitrogen.

11. 12. 13. 14.

Figure 3(b) shows an example of the microscopic image of small unilamellar lipid vesicles prepared by the freezefracture replica method.

3.3 Preparation procedure 1. Turn on the device (Figure 5) and wait until the pressure reaches less than 2 × 10−3 Pa. 2. Fill the dewar with liquid nitrogen until the temperature of the knife reaches a value less than −170 ◦ C. 3. Place a set of sample holder in liquid nitrogen for precooling. 4. Place 2 µl of sample solution on a specimen carrier held by a pair of tweezers, and then immediately put it into the slush nitrogen for rapid freezing.

Secure the specimen carrier to the holder using a pair of tweezers in liquid nitrogen. Attach a transfer rod to the set of specimen holder, and immediately insert into the prevacuum room of specimen preparation machine. Transfer the specimen holder from prevacuum room onto the sample stage in the main chamber, remove the transfer rod, and maintain the temperature of the stage at −120 ◦ C. Shave a couple of micrometers of the frozen sample by knife installed in the machine. This is the fracture process. The temperature of the knife must be less than −110 ◦ C. Keep the knife away from the sample stage, and tilt the stage at 60◦ . Expose 1.5 mm of the Pt–carbon rod and increase the beam current to 100 µA for 10 s, for the vapor deposition of platinum; this process corresponds to shadowing. The vapor deposition must be carried out at a pressure of 10−5 Pa. Tilt the stage to 0◦ , and deposit carbon for 20 s by rotating the sample stage. Remove the set of sample holder by a transfer rod. Keep the specimen carrier in water to remove the replica, and skim the replica using copper grid. Remove the water on the copper grid using filter paper.

Cryogenic TEM

Cryogenic transmission microscopy (Cryo-TEM) is an excellent method, by using which the objects in solution can be visualized without fixing and staining. As mentioned above, negative staining is a convenient method for the preparation of TEM specimen; however, the object can be visualized only in dry state, which is sometimes much different from the structure in solution. In addition, using Cryo-TEM the structural information inside the object can be obtained in contrast to freeze-fracture replica method.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

6

Techniques contrast and high magnification even by using zero-loss imaging mode. Because of the same reason, there is a limitation on the combination with analytical methods, which requires the irradiation of objects using strong electron beam.

3.3.1 Protocol: preparation of the specimen for Cryo-TEM Equipment Rapid-freezing device (for example, Leica EM CPC) Cryotransfer system (for example, Gatan Model 626.DH) Hydrophilic-treatment device (for example, JEOL HDT400).

(a)

Tools and materials Tweezers Micropipet (10 µl) Copper grid with 200 mesh, additionally coated with microgrid Filter paper Liquid nitrogen Propane gas. Preparation procedure Preparation of cryochamber

(b)

Figure 6 Equipment for Cryo-TEM specimen preparation: (a) rapid-freezing device and (b) cryotransfer holder, and their workstation.

The specimen for Cryo-TEM is prepared by rapid-freezing device, as shown in Figure 6(a), to form amorphous ice at Cryo or liquid nitrogen temperature without the coexistence of any stains. Since the supramolecular assemblies in solution generally consist of organic molecules, it is difficult to gain the good contrast without stain. To this end, the combination of amorphous ice formation and energy filter TEM are employed to minimize the background noise arise from ice crystal. The formation of amorphous ice also has a contribution over the preservation of the hydrated objects. The specimen preparation by this method requires special technique and equipment to form fair amorphous ice. In addition, a special specimen holder, as shown in Figure 6(b), must be employed to maintain the specimen at Cryo temperature during observation. To prevent the electron-beam-induced damage of organic sample and the melting of amorphous ice, the intensity of the beam should be minimized. Thus, it is difficult to gain the significant

1. Fill the 20-l reservoir with liquid nitrogen in the rapidfreezing device. 2. Turn on and run the device at least 30 min prior to the sample freezing. 3. Fill a secondary cryogen container with a coolant such as propane, when the chamber becomes sufficiently cold, that is, the temperature is less than −160 ◦ C. 4. Place a grid carrier and filter paper in a cryochamber for precooling. Sample freezing 1. 2. 3. 4. 5.

6.

Make a grid hydrophilic by hydrophilic-treatment device. Lift up and hold the grid with a pair of tweezers. Apply an aliquot (3 µl) of a sample solution to the faceside of grid. Set and secure the tweezers onto the rapid-freezing device (Figure 7a). Remove excess solution, by holding a piece of filter paper parallel to the grid (Figure 7b and c). Thickness of the solution should be less than 200 nm. Plunge the tweezers into a coolant such as liquid propane as soon as possible after the removal of excess solution (Figure 7d).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Transmission electron microscopy (TEM)

4.

5. 6.

(a)

(b)

3. 4.

(d)

Figure 7 Preparation of Cryo-TEM specimen: (a) fixation of the grid using a pair of tweezers, (b) removal of excel solvent, (c) close-up of the specimen grid, and (d) freezing of the specimen in a coolant.

7. 8. 9. 10.

11. 12.

Keep the tweezers in a coolant for about 1 min so that bubbles are not generated. Remove the tweezers from the freezing device. Absorb the coolant on the surface of the grid by filter paper. Put the grid into a holder placed in a cryochamber. Do not remove the grid from the cryochamber to prevent dew condensation. Cover the grid holder by transfer tool and put them into liquid nitrogen. The frozen specimen can be kept in liquid nitrogen up to one year.

Loading of a grid into cryoholder 1. 2.

3.

Evacuate and heat the cryoholder at 100 ◦ C for at least 2 h to obtain enough thermal insulation. Open the shutter and remove the clip ring on the holder before insertion, and insert the cryoholder into the workstation. Fill the dewar and the workstation with liquid nitrogen and wait until the temperature reaches below −165 ◦ C.

The stuffs used for the grid transfer must also be cooled down by liquid nitrogen. Transfer the grid in the grid holder to a cryoholder using tweezers. At this time, static electricity on one’s body needs to be removed to prevent unexpected motion of the grid. Secure the grid by clip rind and close the shutter. Wait until the temperature of the holder becomes −190 ◦ C.

Cryotransfer to microscope 1. 2.

(c)

7

Remove the cryoholder from the workstation. Insert the cryoholder into the microscope immediately. The transfer time should be minimized to prevent dew condensation. Once the cryoholder is installed to the microscope, add liquid nitrogen to dewar on the cryoholder. The observation must be carried out at less than −146 ◦ C.

Figure 3(c) shows an example of the Cryo-TEM image of small unilamellar lipid vesicles. The object is highlighted as a black shadow in the image. Although the contrast of the object is lower than the specimen prepared by negative staining or freeze-fracture methods, enough resolution is obtained to distinguish the characteristics of the membrane such as thickness of the lipid bilayer. Particles of ice crystal, which may be formed in the process of specimen preparation, give black shadows as an artifact.

3.4

TEM tomography

TEM provides microscopic observation or analysis of nanostructures. In general, TEM image shows the twodimensional projection of a three-dimensional object, so it is difficult to obtain the information with respect to the direction of height with high resolution. The electron tomography technique gives an insight into this problem by the computer-based reconstitution of a three-dimensional structure from a series of projected images for a specimen. The tomography technique was developed over 30 years ago and has been recently used for various biomolecules in the field of biological science. It is also expected that this technique has a potential for the three-dimensional evaluation of supramolecular structures. For the reconstitution of three-dimensional image of the specimen, a series of images are taken by tilting the sample stage. Usually, 121 images are acquired at the angle from −60◦ to +60◦ with equal angular increments of 1◦ . In principle, tilting the sample stage can be carried out manually; however, fully automatic software is available for a series of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

8

Techniques

the tomography processes such as image collection, alignment, and reconstitution. In general, the alignment of the individual image is performed by cross-correlation or by tracking markers such as gold nanoparticle embedded in a specimen. For the reconstitution of the three-dimensional image, the tilting angle of the specimen stage must be greater than 60◦ . The missing wedge, which is the unsampled volume in the Fourier space due to the limited tilt, can be minimized by the acquisition at higher tilt angle or dualaxis tomography technique. In addition, the object placed at the center of the TEM grid can be observed by tomography, because there is a limitation on the tilting angle at the edge of the grid. Since a number of images must be taken for a specimen in the tomography, the damage of the electron-sensitive specimen is a serious problem. Thus, the image acquisition should be carried out under extremely low electron illumination conditions. Zero-loss imaging, which is described later, or STEM is an effective approach to obtain fine image, which diminishes the influences of chromatic aberration and background. Cryo-TEM and EELS imaging can be combined with tomography to observe the solution sample and to obtain the elemental information, respectively. Practically, the magnification is limited up to ×200K in maximum, which is lower than the magnification limit for two-dimensional TEM images. In addition, the spatial resolution in height direction is not too high as compared to that in the horizontal direction. As an example of the TEM tomography, three-dimensional images of an organic–inorganic hybrid vesicle, cerasome, coated with the magnetic metal layer of FeCoNi alloy53 is shown in Figure 8. Tomography technique provides the whole capsule image with a diameter of 350 nm. In addition, its sliced image exhibits both the inner and outer surface clearly, and the thickness of the metallic layer is estimated to be 30–40 nm. The images obtained using

TEM tomography may be not as clear as the corresponding two-dimensional TEM images with high resolution; however, it provides enough resolution to understand their nanostructure.

4

ELECTRON DIFFRACTION

When an electron beam passes through a specimen with ordered periodic structure, a diffraction pattern is formed on the back-focal plane of the objective lens. Electron diffraction is not only useful to generate images of diffraction contrast but also for structural analysis of supermolecules forming crystal or self-assembly. Electron diffraction pattern gives us information on the crystal structure, lattice repeat distance, and specimen shape. The pattern is always related to the image of the area of the specimen. Thus, there are basically three types of electron diffraction methods based on the size of the area of the specimen and the illumination modes of electron beam: SAED, NBED, and convergent-beam electron diffraction (CBED). SAED provides structural information on the area of the specimen on a submicrometer scale. Selector aperture is inserted into the plane that contains the first magnified image of the specimen, the image plane of the objective lens. This aperture is used to limit the region of specimen from which an electron diffraction pattern is recorded. Electrons are transmitted through the aperture with a diameter D, which corresponds to the diameter of D/M at the specimen plane, where M is the magnification of objective lens. The values of D and M are usually set around 20 µm and 100 times, respectively. Thus, diffraction information is obtained from the specimen region whose diameter is as small as 0.2 µm. The diffraction angle θ in TEM is very small, commonly less than 1◦ , because the reflecting lattice planes are nearly parallel to the primary beam. Using the approximation sinθ ≈ θ for a small angle, we can rewrite Bragg’s law as follows: λ = 2dθ

(1)

Here, λ is the electron wavelength and d is the spacing of lattice planes. Noting the relationship between geometry and the formation of a diffraction pattern in the TEM column, shown in Figure 9, we have the following equation as the basic relationship for electron diffraction using TEM: λL = Rd (a)

(b)

Figure 8 Three-dimensional reconstruction of TEM images of magnetic metal-coated hybrid vesicle without staining: (a) the whole image and (b) its sliced image.

(2)

where L is the camera length, the distance between the specimen and photographic plane of the camera, and R is the distance from the central spot of the transmitted beam to a diffraction spot in a photographic film. The value λL

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Transmission electron microscopy (TEM) Electron beam λ Specimen

2q

L

Central spot

ANALYTICAL TEM FOR EVALUATION OF SUPRAMOLECULAR STRUCTURES

Film

Diffraction spot

Figure 9

by illuminating the specimen using an electron beam with a fine probe of diameter, in place of insertion of the selector aperture in the optical pass. Parallel electron beams with a probe of diameter of a nanometer scale illuminate the specimen in NBED method, whereas convergent electron beam illuminates the specimen in CBED method.

5

R

9

Formation of an electron diffraction pattern in a TEM.

is called the camera constant of a TEM. Thus, we can calculate the spacing of the lattice planes by measuring R. While precision of SAED in TEM is relatively low than that of X-ray diffraction, the former method has an advantage in diffraction intensity much higher than the latter. As an example of the SAED applied to supramolecular system, it has been reported that the inverted hexagonal (HII ) phase formed with peptide lipids was successfully characterized by this method.54 The peptide lipids having an amino acid residue, interposed between an ionic head moiety and a hydrophobic double-chain segment, generally form bilayer membranes in aqueous media. However, nonionic peptide lipids having hydroxyl groups on the head moiety afford nonbilayer phase depending on the differences in the number of hydroxyl groups. Thus, it was suggested that a mixed peptide lipid system, in which there are strong hydrogen-bonding interactions among the head moieties, formed the HII phase in the negative staining TEM images. Three types of images, which are typical of HII phase, were observed: two kinds of striped patterns with different layer thickness and a network array of small internal aqueous compartments were observed in different areas of the specimen. The formation of HII phase was strongly supported by means of SAED in TEM; 5.2 and 3.1 nm of the repeating distances of layers and 6.2 nm of the repeating distance of aqueous compartments. In addition, the image and the corresponding electron diffraction pattern originating from one of the layers of HII phase were converted into those from the other type of layers by rotating the specimen 30◦ . Thus, the first example of HII phase formed by synthetic lipids has been confirmed by TEM measurements. On the other hand, NBED and CBED give structural information about the nanometer region of the specimen. In these methods, electron diffraction pattern was observed

The benefit of TEM in the investigation of supramolecular chemistry is not only imaging but also a microscopic analysis with a subnanometer resolution. EELS and EDS or EDX analyses provide quantitative elemental information about the specimen on a microscopic scale. These techniques utilize the inelastic scattering of an electron beam in contrast to the normal imaging and electron diffraction that rely on the elastic scattering of an electron. There are several pathways with respect to various energy levels in the inelastic electron scattering such as excitation of the inner-shell, ejection of secondary electrons, plasmon excitation, and phonon excitation. EELS and EDS mainly focus on ionization of the inner-shell electron by detecting the energy loss of inelastically scattered electron and characteristic X-ray, respectively. Although EELS and EDS target at obtaining the elemental information about microscopic structure, each method has significantly different characteristics derived from the detection mechanism as summarized in Table 2. Thus, the researchers need to choose an appropriate method for the evaluation depending on the target material. In this section, we briefly describe the principles and practical information of these two methods from the viewpoint of the application in supramolecular chemistry.

5.1

Electron energy-loss spectroscopy (EELS)

EELS detects the energy distribution of electrons, which passes through the specimen with the loss of energy caused by the inelastic interaction. EELS spectrum consists of three major types of signals with different energy levels depending on the pathways as shown in Figure 10. At the origin of energy loss, spectrum at 0 eV, or the zero-loss peak, is usually observed intensively. This peak corresponds to electrons without energy loss, which passed through the specimen without any interactions or with an elastic interaction. The zero-loss peak is not applicable in the microscopic elemental analysis; however, it is applied to improve the quality of bright-field images by combining with energy filter as described later. The region between zero loss and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

10

Techniques

Table 2

Comparison of EELS and EDS.

Method

EELS

EDS

Detection target Available information Target element Energy resolution Special resolution Elemental mapping Data analysis

Energy loss of inelastically scattered electron Elemental, chemical, and dielectric Low atomic number 1 eV >3 nm Available Complex processing to subtract high background intensity

Characteristic X-ray Elemental High atomic number 150 eV >0.2 nm Available but slow scanning Simple

The background derives mainly from the multiple scattering of electron and the extension of adsorption edges. The background to be subtracted from the EELS spectrum is empirically described as

Intensity (a.u.)

Zero-loss peak

Ib = AE −r Plasmon peaks Interband excitation Intensity × 10 0 1

Figure 10

Ionization edge with fine structure Intensity × 104

20 200 Electron energy loss (eV)

Schematic image of electron energy-loss spectrum.

100 eV, the so-called low-loss region, reflects the information about the excited valence electrons such as interor intraband transitions, or plasmons. Beyond the low-loss region, the background intensity is smoothly falling concomitant with the appearance of edges corresponding to the inner-shell ionization of atoms. This region is called high-loss region. Since the onsets of these ionization edges appear at well-defined positions for specific elements, lowloss region is an important feature in EELS spectrum for the determination of local elemental compositions. The intensity of the edge depends on the excitation probability of the inner-shell electron, which increases with decreasing atomic number. Thus, light atoms yield relatively high intensity of ionization edge than heavy atoms. In addition, the energy resolution of EELS (1 eV) is significantly higher than that of EDS (150 eV), yielding detailed information about electron state. Accordingly, the EELS is a suitable method for the elemental analysis of the supramolecular structure, which consists of organic molecules, rather than EDS. The analysis of local composition by EELS spectrum requires complex processing due to the low ratio of peak to background (P /B ratio) especially in the high-loss region.

(3)

Here, Ib is the background intensity, E is the energy loss, and A and r are fitting parameters. The local elemental composition can be determined by the integration of ionization peaks after the background subtraction and the ratio of scattering cross sections. As compared to EDS, however, determination precision is poorer because of multiple scattering, which strongly depends on the thickness of the specimen. In addition, the ionization edge in very high-loss region greater than 1000 eV is hard to be detected by EELS, because the edge intensity significantly decreases with increasing energy loss. For the EELS analysis, specimen must be prepared thin, less than 100 nm, to diminish the influence of the background that results because of multiple scattering by plasmon. Please note that the carbon signal that originate from the supporting film on TEM grid may appear as an artifact, because the electron beam passes through the grid. EELS provides the elemental mapping feature of a specimen using the energy filter or STEM as described later. The combination with Cryo-TEM technique also reveals the elemental information about the sample in solution state, which cannot be achieved using EDS. Therefore, EELS will be a potent approach to clarify the detailed structure of supermolecules. The selection of specific energy windows on the EELS spectrum enables obtaining extra information from TEM image such as elemental distribution. The energy filter installed in TEM is the key part to extract the two-dimensional information about electrons with specific energy. Magnetic prisms and optical columns in an energy filter separate the electrons passed through the specimen with specific energy. STEM is another method to obtain the energy-loss TEM image by the serial acquisition of EELS spectrum.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Transmission electron microscopy (TEM) Since the zero-loss peak is derived from elastically scattered electrons or transmitted electrons, the elementspecific information cannot be obtained from it. However, the zero-loss imaging significantly increases the quality of bright-field images by omitting the noise caused by the incorporation of inelastically scattered electrons. It can diminish chromatic aberration, which is a serious problem especially for a thicker specimen. Thus, zero-loss imaging is effectively used to gain enough contrast in the Cryo-TEM observation. The core-less imaging with a defined energy window yields an element-specific image theoretically. Since the background intensity is extremely high in EELS spectrum as mentioned above, the subtraction of background is an important process to gain contrast in the image of elemental mapping. Here, we describe the procedure of three-window method, which is one of the common methods for the background subtraction. In the high-loss region, two images are additionally taken below the window of target edge. These two images are used for the calculation of parameters A and r in (3) to obtain the background image. The elemental mapping is an effective method to obtain the elemental distribution on a microscopic scale. However, owing to the low intensity of the ionization edge in the high-loss region, core-less imaging requires relatively high intensity of electron beams. Thus, the sample must be stable against the irradiation of strong electron beams. Figure 11 shows an example of EELS mapping of an organic–inorganic hybrid vesicle, cerasome, coated with the Ni-nanometal layer.26 The specimen was prepared by

50 nm (a)

casting an aqueous dispersion of vesicle on a carbon-coated copper grid. The experiment was carried out using a JEOL JEM-3100FEF analytical TEM with an integrated -filter at an acceleration voltage of 300 kV. Core-loss images are taken at E = 284, 99, and 855 eV with the energy width of 16, 10, and 30 eV, corresponding to carbon, silicon, and nickel elements, respectively. The background was subtracted by the three-window method described above. The image clearly revealed the presence of these elements on the vesicular surface, reflecting that the lipid membrane is covered with silicate and metallic nickel layers.

5.2

Energy dispersive X-ray spectroscopy (EDS)

EDS is another analytical method for the evaluation of elemental composition on a microscopic scale by utilizing the characteristics of inelastically scattered electrons. In the de-excitation process, the excited atom loses its energy by two different pathways: emission of characteristic X-ray and ejection of Auger electrons. The characteristic X-ray, which is a detection target of EDS, has a unique energy depending on the atomic number and the quantum numbers of the energy levels involved in the electron transition. In general, EDS is suitable for the analysis of heavy atoms, because the fluorescence yield of X-ray increases with increasing atomic number. Light atoms display one or two digits smaller fluorescence yield than heavy atoms due to the increase of probability of Auger electron-involving process. Thus, EDS is an effective method, especially for the detection of heavy atoms that cannot be detected using EELS. EDS provides highly quantitative analysis of the local composition of elements, because the background intensity of the characteristic X-ray is very low in contrast to electron energy loss detected by EELS. The concentration ratio of two different elements (CA /CB ) can be quantitatively determined by the ratio of peak intensities (IA /IB ) and k-factor or Cliff–Lorimer factor (kA,B ) as follows:

(b)

CA /CB = kA,B (IA /IB )

(c)

11

(d)

Figure 11 TEM image of Ni-coated cerasome (a) and the corresponding EELS mapping for C (b), Si (c), and Ni (d). (Reproduced with permission from Ref. 26.  Ceramic Society of Japan, 2008.)

(4)

The k-factor, which is determined theoretically or experimentally, is the key parameter to obtain reliable result by the quantitative analysis of EDS data. In contrast to EELS, EDS is unable to reveal the information about the electron state due to their low energy resolution (150 eV). To apply the thin film approximation, thinner specimen is preferred for the quantitative EDS measurements. However, thin specimens require long-time irradiation of the electron beam to gain enough signals due to their weak X-ray intensity that may damage the sample. The ideal thickness of the specimen should be in the range of 10–100 nm to obtain

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

12

Techniques

an EDS spectrum with good signal-to-noise ratio. Generally, the specimen holder made of beryllium is preferred to diminish the unexpected background from the elements used in the holder. Be sure not to touch the beryllium holder by naked hand at the time of its use, because beryllium is a highly toxic metal! The characteristic X-ray corresponding to copper, iron, and cobalt is usually observed as an artifact, because these elements are frequently used in microscope and grids. Several ghost peaks such as escape peaks and sum peaks appear on the EDS spectrum due to the characteristics of the detector. Elemental mapping is available by two-dimensional scanning of electron beam on the specimen. The special resolution of the mapping using EDS depends on the diameter of the beam and accelerating voltage of the incident electron; consequently, the use of field emission gun yields a fine resolution less than 1 nm. The elemental mapping using EDS yields a highly quantitative image due to significantly lower intensity of background than EELS-based mapping. The scanning of the specimen takes a long time and damages the specimen by the electron beam. This practical problem makes the combination of EDS with Cryo-TEM or electron tomography difficult. On these grounds, it is suggested that both EDS and EELS are used in combination to obtain the local information one needs.

staining. TEM tomography is a method using which threedimensional structure of the sample can be visualized. Although there are not so many examples of supermolecules showing the tomographic image, this method will be popular in TEM imaging with promotion of the apparatus. In addition, more precise structural evaluation of the TEM samples is possible by using the imaging with electron diffraction showing periodic nanostructure and EDS or EELS exhibiting elemental mapping. It should be noted that the TEM imaging is a method to visualize selected tiny areas of the sample specimen. Thus, it is essential to clarify supramolecular structures by using TEM with other microscopy, such as SEM and scanning probe microscopy, and additional physical measurements that are capable of providing the structural information about the whole sample.

ACKNOWLEDGMENTS We are grateful to Ms. Sakiko Fujita of Nara Institute of Science and Technology for her comments and expert technical assistance in the TEM measurements.

REFERENCES 6

CONCLUSION

TEM is currently used for structural characterization of various supermolecules. In general, we can employ three types of the specimen preparation techniques: negative staining, freeze-fracture, and cryo methods. The negative staining is the most convenient method and widely used for imaging the cross-sectional view of samples with less difference in electron density toward the background of the sample specimen. However, it is necessary to pay attention so as to prevent the formation of artifacts, arising because of the interactions of the sample with heavy metal ions as the staining agent, and structural change, possibly occurring in the sample drying process. On the other hand, the freeze-fracture is the method that does not require staining, and is capable of imaging the surface structure of the sample replica. The latter method requires a higher familiarized technique than the former. Although negative staining and freeze-fracture techniques have been widely applied for TEM imaging in the past, the Cryo-TEM is the most employed method by many researchers in the field of supramolecular chemistry in the present. By filtering inelastic scattering of specific elements, the Cryo-TEM can provide direct image of the sample in frozen state showing good contrast without

1. R. F. Egerton, Physical Principles of Electron Microscopy: An Introduction to TEM, SEM, and AEM, Springer, New York, 2005. 2. J. Kuo, ed., Electron Microscopy: Methods and Protocols, 2nd edn, Humana Press, Totowa, 2007. 3. Y. Leng, Materials Characterization: Introduction to Microscopic and Spectroscopic Methods, John Wiley & Sons (Asia) Pte Ltd, Clementi Loop, 2008, pp. 79–119. 4. D. B. Williams and C. B. Carter, Transmission Electron Microscopy: A Textbook for Materials Science, 2nd edn, Springer, New York, 2009. 5. E. Dujardin and S. Mann, Adv. Mater., 2002, 14, 775. 6. F. Caruso, Adv. Mater., 2001, 13, 11. 7. C. Sanchez, H, Arribart, and M. M. G. Guille, Nat. Mater., 2005, 4, 277. 8. A. D. Bangham and R. W. Horne, J. Mol. Biol., 1964, 8, 660. 9. H. Ringsdorf, B. Schlarb, and J. Venzmer, Angew. Chem. Int. Ed. Engl., 1988, 27, 113. 10. T. Kunitake, Angew. Chem. Int. Ed. Engl., 1992, 31, 709. 11. J.-H. Fuhrhop and J. K¨oning, Membranes and Molecular Assemblies: The Synkinetic Approach, The Royal Society of Chemistry, Cambridge, 1994. 12. A. Mueller and D. F. O’Brien, Chem. Rev., 2002, 102, 727. 13. J.-K. Kim, E. Lee, Y.-B. Lim, and M. Lee, Angew. Chem. Int. Ed., 2008, 47, 4662.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Transmission electron microscopy (TEM)

13

14. A. Polidori, N. Michel, A. S. Fabiano, and B. Pucci, Chem. Phys. Lipids, 2005, 136, 23.

35. K. Yoosaf, A. Belbakra, N. Armaroli, et al., Chem. Commun., 2009, 2830.

15. J.-F. Gohy, B. G. G. Lohmeijer, B. D`ecamps, et al., Polym. Int., 2003, 52, 1611.

36. R. Matmour, I. D. Cat, S. J. George, et al., J. Am. Chem. Soc., 2008, 130, 14576.

16. T. Shimizu, R. Iwaura, M. Masuda, et al., J. Am. Chem. Soc., 2001, 123, 5947.

37. R. S. Johnson, T. Yamazaki, A. Kovalenko, and H. Fenniri, J. Am. Chem. Soc., 2007, 129, 5735.

17. A. Kaplun, F. M. Konikoff, A. Eitan, et al., Microsc. Res. Tech., 1997, 39, 85.

38. I. S. Choi, X. Li, E. E. Simanek, et al., Chem. Mater., 1999, 11, 684.

18. Y. Murakami and J. Kikuchi, Bioorganic Chemistry Frontiers, ed. H. Dugas, Springer-Verlag, Berlin, 1991, vol. 2, pp. 73–113.

39. D. Pantarotto, R. Singh, D. McCarthy, et al., Angew. Chem. Int. Ed., 2004, 43, 5242.

19. L. C. Palmer and S. I. Stupp, Acc. Chem. Res., 2008, 41, 1674. 20. Y. Sasaki, K. Matsui, Y. Aoyama, and J. Kikuchi, Nat. Protoc., 2006, 1, 1227. 21. O. Francescangeli, V. Stanic, L. Gobbi, et al., Phys. Rev. E, 2003, 67, 011904. 22. B. Pitard, N. Oudrhiri, J.-P. Vigneron, et al., Proc. Natl. Acad. Sci. U.S.A., 1999, 96, 2621.

40. Y. Liu, Z.-L. Yu, Y.-M. Zhang, et al., J. Am. Chem. Soc., 2008, 130, 10431. 41. C. Ehli, G. M. A. Rahman, N. Jux, et al., J. Am. Chem. Soc., 2006, 128, 11222. 42. V. Georgakilas, F. Pellarini, M. Prato, et al., Proc. Natl. Acad. Sci. U.S.A., 2002, 99, 5075. 43. F. Liu, J. Y. Choi, and T. S. Seo, Biosens. Bioelectron., 2010, 25, 2361.

23. R. Qiao and P. C. Ke, J. Am. Chem. Soc., 2006, 128, 13656.

44. K. J. C. van Bommel, A. Friggeri, and S. Shinkai, Angew. Chem. Int. Ed., 2003, 42, 980.

24. C. Richard, F. Balavoine, P. Schultz, et al., Science, 2003, 300, 775.

45. M. A. Alam, Y.-S. Kim, S. Ogawa, et al., Angew. Chem. Int. Ed., 2008, 47, 2070.

25. K. Katagiri, M. Hashizume, K. Ariga, et al., Chem. Eur. J., 2007, 13, 5272.

46. E. Pouget, E. Dujardin, A. Cavalier, et al., Nat. Mater., 2007, 6, 434.

26. F. Gu, M. Hashizume, S. Okada, et al., J. Ceram. Soc. Jpn, 2008, 116, 400.

47. L. Du, S. Liao, H. A. Khatib, et al., J. Am. Chem. Soc., 2009, 131, 15136.

27. J. J. E. Moreau, L. Vellutini, M. W. C. Man, et al., Chem. Eur. J., 2005, 11, 1527.

48. C.-P. Chak, S. Xuan, P. M. Mendes, et al., ACS Nano, 2009, 3, 2129.

28. Y. Guan, S.-H. Yu, M. Antonietti, et al., Chem. Eur. J., 2005, 11, 1305.

49. C. Han, L. Zhang, and H. Li, Chem. Commun., 2009, 3545.

29. R. Resendes, J. A. Massey, K. Temple, et al., Chem. Eur. J., 2001, 7, 2414.

50. Y. Liu, H. Wang, Y. Chen, et al., J. Am. Chem. Soc., 2005, 127, 657.

31. B.-B. Wang, X. Zhang, X.-R. Jia, et al., J. Am. Chem. Soc., 2004, 126, 15180.

51. S. J. Pennycook and Y. Yan, Z-Contrast imaging in the scanning transmission electron microscope, in Progress in Transmission Electron Microscopy 1: Concepts and Techniques, eds. X.-F. Zhang and Z. Zhang, Springer, Heidelberg, 2001, pp. 81–111.

32. S. Kawano, N. Fujita, and S. Shinkai, Chem. Eur. J., 2005, 11, 4735.

52. T. S. Balaban, A. D. Bhise, G. Bringmann, et al., J. Am. Chem. Soc., 2009, 131, 14480.

33. J. Gao, H. Wang, L. Wang, et al., J. Am. Chem. Soc., 2009, 131, 11286.

53. D. Minamida, S. Okada, M. Hashizume, et al., J. Sol-Gel Sci. Technol., 2008, 48, 95.

34. S.-L. Zhou, S. Matsumoto, H.-D. Tian, et al., Chem. Eur. J., 2005, 11, 1130.

54. Y. Murakami, J. Kikuchi, T. Takaki, and K. Uchimura, Bull. Chem. Soc. Jpn., 1987, 60, 1469.

30. T. M. Hermans, M. A. C. Broeren, N. Gomopoulos, et al., J. Am. Chem. Soc., 2007, 129, 15631.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc022

Computational Techniques (DFT, MM, TD-DFT, PCM) Rosemary Sheehan and Peter J. Cragg University of Brighton, Brighton, UK

1 Introduction 2 Computational Techniques 3 Applications in Supramolecular Chemistry 4 Conclusions References Further Reading Reviews Commonly Used Commercial and Academic Software Packages

1

1 3 11 16 17 18 18 18

INTRODUCTION

Computers are ubiquitous in contemporary science, from the word-processing functions used to create research articles and the databases which hold the information described in those articles, to interactive experiments and appealing graphics. Nowhere have the advances in computerized methods been more informative than in the simulation of existing or postulated molecules and predictions made about their properties or behavior. Computational approaches can be used to simulate molecular and atomic behavior based on fundamental descriptions of atomic and molecular orbitals (ab initio quantum mechanics), experimental data (a priori molecular mechanics, MM), or a combination of both (semiempirical methods). The choice of method depends on the task in

hand and the computational resources available. The most computationally intensive methods are based on calculations of molecular orbitals from solutions to the Schr¨odinger equation. This ab initio method uses fundamental mathematical principles to derive the structures and properties associated with molecular orbitals through a consideration of the effect of every electron in the molecule. The number of calculations and their complexity limits the size of molecules that may be studied by these means using desktop machines. The basis for computational chemistry lies in the mathematical descriptions of atoms and molecules provided by theoretical chemistry. While these descriptions are not perfect, they were good enough over a decade ago for Hehre to state the following: The theories underlying calculations have now evolved to a stage where a variety of important quantities, among them molecular equilibrium geometry and reaction energetics, may be obtained with sufficient accuracy to actually be of use.1 The advances in the efficiency and accuracy of current algorithms are complemented by ongoing developments in hardware, and have allowed molecular modeling to become a viable analytical tool for the chemical sciences. This is reflected by its widespread use in both academia and industry, with major pharmaceutical companies such as Unilever and GSK as well as some smaller companies incorporating computational modeling departments as a standard part of their infrastructure. Indeed, advances in software and hardware development in recent years have been such that the sizes of ab initio simulations now possible are similar to those computed by MM in the 1960s. The fact that Kohn and Pople were jointly awarded the Nobel Prize in 1998 is also indicative of the increasing

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

2

Techniques

recognition for the potential value of in silico molecular modeling2 ; Kohn for his work on density functional theory and Pople for development of computational methods in quantum chemistry. The way in which computational chemistry is used is worth exploring within industry and academia. Industrial applications, particularly by pharmaceutical companies, focus on drug design and discovery where existing compounds can be modeled to derive parameters such as polarizability, water/octanol partition coefficients (log P values), molecular mass, number of hydrogen bond donors and acceptors, and the disposition of pharmacophore centers. Analysis of this nature has two functions. First, it allows a database of existing compounds to be set up, which can then be searched by computed parameters. This in turn allows similarity analyses to be undertaken so that the data can be mined for compounds with properties similar to known pharmaceuticals. Second, it allows the Lipinski “rule of five” (RO5) to be used to determine if a given compound has the potential to be turned into a druggable form.3 The RO5 predicts that compounds with a molecular weight of less than 500 Da, a log P below 5, and five or fewer hydrogen bond donors (with no more than 10 hydrogen bond acceptors) are the most promising candidates.4 The extensive use of molecular modeling in the industry has proven its validity as a modern analytical technique. Its useful impact has clearly been proven. Industrial applications have included the development of bioactive materials (Merck, Novartis, Takeda Chemicals, DuPont, and Sumitomo Chemicals), homogeneous and heterogeneous catalysts (Ford, Haldor, Topsøe, and Ube Industries), food and hygiene products (Colgate Palmolive, Unilever, and Kellog), polymers, glass and structural materials (Asahi Chemicals, Owens Corning, Rhˆone Poulenc/Rhodia, and W. R. Grace), films and imaging (Fuji Photo Film, and Xerox), fuels and automotive chemicals (Chevron, TotalFina/TotalFinaElf, and Lubrizol), and electronic and photonic materials (Motorola, Toshiba, and Lucent). Computational chemistry has had its most significant commercial successes in drug discovery, homogeneous catalysis, and computational thermochemistry. Computational methods have been applied to drug discovery: the HIV protease inhibitor Agenerase was created by Vertex Pharmaceuticals as a result of molecular modeling and Aricept an ACE inhibitor used to treat Alzheimer’s patients was developed by Eisai Co., Ltd. Homogeneous catalysts are routinely screened by computational quantum chemistry with the result that in 1995 homogeneous catalysts contributed to 85% of the polymers market, despite the fact that they were predicted to be at 20% twenty years earlier, before the widespread use of molecular modeling in the industry. Companies now employing this

approach include BP (Amoco), Mitsubishi, BASF, DOW Chemical, and Phillips Petroleum. In ideal gas thermochemistry, some quantum chemical predictions have less uncertainty than the traditional calorimetric measurements. To illustrate this, Dow Chemical estimated that calculating
Hf for a molecule of interest would cost US$100 000 by calorimetry, but only US$2000 for a computational G3 calculation with comparable accuracy. The value of computational methods in this instance is clear.5 Molecular modeling provides the potential to investigate a huge range of chemical structures that would not be possible using traditional methods. The pharmaceutical industry typically uses modeling as a cheap and effective means of identifying lead compounds for further study in drug development, while in areas such as thermochemistry, prediction of spectroscopic frequencies, or protein crystallography the results of computational modeling can be more precise than those achieved experimentally. Molecular modeling is now an industry standard technique and is gaining popularity in the academic world as a tool for modern analytical chemistry. It can be extremely useful in the prediction and interpretation of infrared and ultraviolet spectra, particularly in cases where the spectra are hard to interpret due to contamination by energetically equivalent tautomers, for example, cytosine and other nucleotide bases. Computational techniques can also be useful in elucidating reaction mechanisms and for 3D structural visualization. To this end, nearly one-quarter of the articles in the American Chemical Society’s flagship publications, the Journal of the American Chemical Society and the Journal of Organic Chemistry, that are concerned with structural prediction or interpretation make use of molecular simulations in some way. Non-pharmaceutical applications tend to be aimed at accurate structure prediction together with properties such as electrostatic potential maps and surface hydrophobicity. One consequence of this is a desire for easily understood graphical representations of both the molecules and the calculated properties. An example of this can be seen in Figure 1, which shows a geometry-optimized computed structure of the potassium complex of the macrocycle 18crown-6. A representation of the complex’s electrostatic potential on a van der Waals surface is overlaid on top of this. A cutaway view is depicted so that the connectivity and identity of each atom can be seen clearly and correlated with the local electrostatics. In the example, blue represents an electropositive region, red is electronegative, and green is neutral. As the complex can be rotated on a computer screen, the structure can be considered to be pseudo three dimensional. The representation of physical properties adds a further dimension to this. Snapshots of molecules and associated properties, perhaps modeled at precise intervals

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM)

3

from a full ab initio model. The choice of which to use generally depends on the type of information required from the calculations, the size and constituents of the system to be analyzed, and the computing resources and software available.

2.1

Figure 1 18-Crown-6 binding a potassium cation, with an electrostatic potential mapped onto a van der Waals surface (blue, electropositive; green, neutral; red, electronegative).

during a conformational change, could therefore be represented as a movie sequence to illustrate how the properties change over time with variation in structure. The same principle can be applied to visualize a chemical reaction or the formation of a host–guest complex. All the computational methods discussed below have their respective strengths and weaknesses, but through an informed use of combinations of methods, with modification where appropriate, highly accurate results can be obtained with huge savings in terms of time and money. Their uses are not restricted to nanoscale or supramolecular chemistry; however, where possible examples are drawn from these fields.

2

COMPUTATIONAL TECHNIQUES

Many computational methods exist, and, as yet, no one technique is ideally suited to all problems. For the sake of simplicity, these computational methods can be subdivided into molecular mechanical and quantum mechanical approaches, although each may encompass a wide range of techniques. The model used in computational chemistry refers to the type of approach—mechanics, semiempirical, or ab initio —that is adopted. The simplest model is that based on MM, in which atoms are considered to be solid spheres with ideal interatomic bond distances and angles; the most complex models currently in use are those that attempt to solve the Schr¨odinger equations associated with molecular orbitals. In between are semiempirical models in which only valence electrons are considered in the quantum mechanical treatment, which in turn is greatly simplified

Molecular mechanics (MM)

Molecular mechanical methods are useful for structural elucidation, quick geometry optimization of neutral molecules, and visualization and comparison of different conformers. Currently, molecular mechanical methods are typically applied to large biological systems such as proteins,6 segments of DNA,7 transmembrane simulations in cells,8 and enzymes.9 Owing to their speed, molecular mechanical energy minimizations are also commonly used to obtain initial low-energy structures, prior to refinement with more rigorous, quantum mechanical calculations. MM treats atoms as hard spheres joined by elastic bonds that behave like harmonic oscillators. In addition, electrostatic charges (formal and/or partial) may be included in the calculation along with other, long range interactions such as van der Waals forces. The total energy of the system may be broken down into the following components: Etot = Ebond + Eangle + Edihedral + Enonbond These contributions to the overall energy of the molecule are optimized for all the interatomic interactions present. The parameters used to describe each interaction together with optimization algorithms make up the “force field,” which is used to compute the overall structure of a molecule. Each individual parameter is based on an ideal value for each element in a specified environment. The corresponding values used to describe the different bonds made by an element are known as atom types. For example, carbon has different ideal values for sp, sp2 , and sp3 geometries, and may also have further corrections if it is an sp2 hybridized atom in an amide or ester bond. Ideal values are usually determined by comparing bond lengths and angles from spectroscopic or crystallographic sources in the hope that a single set of force field parameters for the atom type will accurately reproduce all molecules where that type of bond exists. This is rarely the case, leading to the generation of numerous force fields including usermodified versions. In the early years of MM, it was not uncommon to encounter force fields designed for specific uses: general organic versions would be parameterized differently from those used for proteins, and those for coordination compounds were even more different. Most force fields are now universal in that they can accurately describe the optimum geometries of most compounds that

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

4

Techniques

+

−

Figure 2 Components of an MM force field: (from left) bond stretch, angle bend, dihedral twist, van der Waals and electrostatic interactions.

are likely to be encountered in the laboratory, as well as those that have yet to be synthesized. Force fields are usually composed of several sets of ideal values for elements as shown in Figure 2. These include bond stretching and bending parameters, as well those for less obvious effects such as torsion and “through space” interactions. The best description of a bond stretch is the Morse function; however, as this is computationally expensive, a simpler harmonic function is usually used in molecular modeling as shown below: Ebond = kl (l − l0 )2

(1)

where kl = stretching force constant; l = bond length; l0 = reference bond length. The contribution of the bond energy to the overall structure is derived at by summing the energy for all bonds. Many force fields use extra terms in the equation to improve the accuracy of the function. Bond angles are treated in a similar way to bonds but use a harmonic function: Eangle = kθ (θ − θ 0 )

2

(2)

where k = angular force constant; θ = angle; θ 0 = equilibrium angle. Further parameterization may be necessary when modeling square planar and octahedral geometries around transition metals. Historically, problems arose when ideal angles of 90◦ were involved as minimization algorithms using an iterative approach allowed two ligating atoms separated by less than 90◦ to collapse into each other. The consequences for inorganic and supramolecular coordination chemistry were profound; however, force fields are now able to reproduce coordination complexes accurately. Torsion (dihedral) interactions are usually described by a Fourier series: Edihedral = Vn (1 + scosnω)

(3)

where Vn = rotational barrier height; s = 1 (staggered minima) or −1 (eclipsed minima) n = periodicity of rotation; ω = torsion angle

The so-called “nonbonded interactions” between atoms separated by greater than two others, or “through space”, can be split into van der Waals and electrostatic components. There are many ways of describing van der Waals interactions; however, the most common methods employ either the 6–12 (or Lennard–Jones) potential (4) or the Buckingham potential (5): Evdw = ε[(rm /r)12 − 2(rm /r)6 ]

(4)

where ε = well depth; rm = minimum energy interaction distance Evdw = Aexp(−Br) − Cr −6

(5)

where A, B, and C are constants. The electrostatic term is usually calculated using partial charges by applying Coulomb’s law: Eel = qi qj /Drij

(6)

where qi , qj = partial charges on atoms i and j; D = dielectric constant; rij = distance between i and j. Once a model has been created and a force field defined, the overall energy needs to be minimized. In practice, this is done by optimizing all the energy terms to give the lowest energy structure. The minimization routines themselves involve moving atoms to approach their ideal bond lengths, angles, and so on and analyzing their effect on the energy. As all the functions that describe the atomic interactions are continuous and differentiable, the effect of moving every atom can be analyzed by the derivatives of these functions. Moving any atom will change the energy, and once each move is complete a reassessment of the energy is possible. A decrease in energy indicates that the movement was in a beneficial direction and may be continued until the energy starts to rise again. This process is repeated for every atom and all functions until a geometry corresponding to an energy minimum is obtained. The minimization programs are usually based on the Newton–Raphson method, which requires the calculation of the first and second derivative, or Hessian, matrices. Different approximations (e.g., steepest descent or conjugate gradient) of these matrices are used to speed up the computational process. The main problem

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM)

Energy

Property 2 Property 1

Figure 3

A potential energy surface map.

in minimizing structures is in identifying the differences between local and global energy minima, which can be found by inspecting a potential energy surface as illustrated in Figure 3. The energy calculated is an expression of the differences between ideal values for each atom type and the actual values that result from the necessity of invoking nonideal bond lengths, angles, and other terms to generate a suitable structure. The resulting energy for a molecule, often termed the steric energy, is therefore relative to those calculated for other molecules but has no meaning in thermodynamic terms. If heats of formation, binding constants, or predicted spectra are required, the calculation must employ semiempirical or ab initio methods. Given the excellent results that MM gives for optimized geometries, it is usual to start any simulation with this method and import the atomic coordinates into programs that can either refine the geometry further or simply determine the molecule’s thermodynamic properties with a single-point calculation based on the MM geometry.

experimental conditions. The method neither guarantees the finding of the global minimum geometry nor does it necessarily find all the known conformers, but a simulation run for long enough does generate a highly representative set of data for the conformers available to the target compound. Dynamics simulations are often carried out with Monte Carlo methods that randomly sample the resulting conformers to reduce the numbers used in analysis or, where only a small number of conformers exists, each simulation could start from a different geometry. The latter approach could be used to investigate the barriers to rotation between conformers of calixarenes and calix[4]resorcinarenes. Molecular dynamic methods are based on classical Newtonian mechanics. As yet, a complete many-particle, timedependent quantum method has not been developed, although hybrid methods such as Car–Parinello MD are emerging,10 which involve a quantum mechanical calculation of electronic structure at every time step, followed by classical dynamic motion of the atoms. Both MD and Monte Carlo techniques use Maxwell– Boltzmann averaging for thermodynamic properties, and are suitable for use with bulk systems, or ensembles, due to their basis in statistical mechanics. Both techniques are useful for conformational searching; however, Monte Carlo methods cannot track time-dependent properties. They make random changes to the molecules and only state the probability that a proposed configurational change should be made, not the route by which one configuration evolves to the next. The strength of stochastic Monte Carlo search methods, however, is that they have been found to sample conformational space more efficiently, producing a greater number of conformers in a shorter time period than MD.11, 12

2.3 2.2

5

Conformational analysis

Molecular dynamics (MD)

MD describes a process where the heating and cooling of molecules is simulated. The effects are seen in increased bond vibrations and molecular motion as the temperature increases and the reverse when it is quenched. If enough energy is transferred to the molecule, it may be enough to overcome rotational or inversion barriers and generate a different conformation to that of the starting structure. Simulations run for defined periods of simulated time so that different conformers can be obtained. These can be analyzed for relative abundance at particular temperatures or may be “cooled” to ambient temperatures and have structures optimized by MM. The latter approach will generate fewer conformers, as high-energy structures collapse into more favorable geometries at lower temperatures, but will represent those conformers available to the molecule under normal

Almost all molecules can exist in more than one conformation and it is possible to construct all the major conformers using their geometries as starting points to generate energetically minimized structures. The final energies can be compared and the conformer with the lowest steric energy should be the most stable. The problem in supramolecular simulations is the sheer number of conformers that may exist. To simplify the process of analysis, the number of potential conformers must be reduced while retaining those of most relevance to the experimentalist. Although theoretically every bond can rotate through 360◦ given enough energy, in practice it is assumed that double bonds are either cis or trans with respect to substituents generating torsion angles of 0◦ and 180◦ rather than the continuum of angles. Single bonds, through a similar logic, give rise to

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

6

Techniques

Rel. E (kJ mol−1)

41

27.33

13.67

0 M01

M14 Molecule

Figure 4

Conformational analysis of 14 salen structures (M01–M14), four conformers shown.

three possible conformers with torsion angles of +120◦ , 0◦ or −120◦ . Conformers could be generated at 60◦ , 30◦ , or even 1◦ increments but this would increase the number of results and computational time required. Conformational data can be generated by allowing every bond in the molecule to rotate sequentially until all conformers have been obtained but this rapidly generates vast numbers of structures as the size of the molecule increases. Alternative methods use Monte Carlo methods or MD to obtain a sample of structures. These methods cannot guarantee that the lowest energy conformer has been detected; nevertheless, they are often the only reasonable route to generating a statistically important sample of the conformers possible. An example of conformational analysis of salen, a Schiff base ligand commonly used in supramolecular and coordination chemistry, is shown in Figure 4. Analysis of the conformers generated requires some thought. Some conformers will have extremely high energies as remote substituents are forced into close proximity and they are unlikely to reflect a high proportion of molecules in the gas or solution phase. Thus, they may be ignored. Alternatively, a geometry optimization of each could be undertaken. This will remove high-energy conformers from the data set and determine if any collapse into the same optimum geometry or if several low-energy conformers exist with diverse geometries.

2.4

Periodic boundary simulations

In crystallography, each unit cell is a replica of another and may be used to reconstitute the entire sample through

Figure 5

A periodic boundary simulation.

extrapolation in three dimensions. A similar minimalist approach to MD simulations can be envisaged: the periodic boundary model. As all cells are identical, any molecule leaving the simulation through a cell wall will reenter from the opposite wall as molecules cannot be lost or gained as shown in Figure 5. Periodic boundary conditions keep the contents constant but also give consideration to the effects that neighboring cells may have on the cell of interest. The size of the cell is of key importance: the larger it is, the more hosts, guests, and solvent molecules can be included to give a more accurate representation of the system’s complexity but there is a direct cost in the time it takes

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM) for these calculations. Most commercial software is limited in the range of solvents that can be included in the cell, often restricted to water, but the method is as close as it is possible to come to simulating time-dependent phenomena such as the steps in host–guest complex formation.

2.5 Ab initio quantum mechanical methods Quantum mechanics is one of the oldest formalisms of theoretical chemistry, using established physical constants and differential equations to calculate molecular properties and geometries directly.13 Computational methods in quantum chemistry include both semiempirical and ab initio calculations. These methods acknowledge the effects of electrons to varying degrees, and are based on finding approximate solutions to the manyˆ = E, where  is the electron Schr¨odinger equation, H ˆ is the Hamiltonian opermany-electron wavefunction, H ator that performs an operation on the function to yield that same function multiplied by a constant, E. A range of mathematical approximations is used to overcome the fact that it has not been possible yet to solve the Schr¨odinger equation for more than one electron, and the reliability and accuracy of quantum mechanical methods varies according to the specific approximations used. Ab initio methods use simulation directly from first principles, and do not incorporate any empirical parameterization, with the advantageous result that they would eventually converge to the exact solution if all the approximations could be made sufficiently small. The most commonly used ab initio methods are based on the Hartree-Fock (HF) model and density functional theory.

2.5.1 Hartree Fock (HF) The HF model attempts to resolve the many-electron problem by writing the many-electron wavefunction in terms of a product of one-electron wavefunctions. The method has a number of weaknesses resulting from the approximations inherent in the model. Not least of these is that it does not account for the coupling motion of electrons (electron correlation) so the electrons tend to interact more than they should, leading to an unrealistically high electron–electron repulsion energy. Despite its weaknesses, the HF model is the foundation for most quantum chemistry, although it is usually used in modified forms, which correct for the effects of the approximations used. The HF approach to solving a Schr¨odinger equation associated with a molecule is to replace the manyelectron wavefunction with a single-determinant wavefunction calculated as the product of many single-electron

7

functions, or spin orbitals. The method assumes that the Gaussian functions for electrons in constituent atoms can be combined to give a description of the molecule as a whole, leading to the linear combination of atomic orbitals (LCAO) principle. The resulting coupled differential equations for all the electrons can then be solved. Clearly, this is a major undertaking for anything but the smallest molecules but other approximations can be introduced to simplify the calculations. Combining the HF approach with the LCAO approximation leads to the generation of Roothaan–Hall equations, which consider the energies of the electrons and, separately, the interactions between them. Computational methods are then applied to solve these equations so that molecular properties, including geometries, can be determined. To find a quantum mechanical solution that describes the electron distribution in a molecule, one or more mathematical functions have to be defined. Solutions to the Schr¨odinger equation for the hydrogen atom, the only case where an exact answer can be attempted, are based on polynomials with Cartesian and exponential components. These polynomial equations are best modeled by Gaussian-type functions. It is assumed that if the hydrogen atom with its single electron can be modeled in this manner then manyelectron systems can be modeled through a combination of Gaussian functions, hence the use of the LCAO approximation. The accuracy of the resulting molecular geometry and orbital related information is determined by the complexity of the basis set used. However, even after the accuracy is improved, the HF method is still expensive, time consuming, and scales poorly. For these reasons, it is best used for systems with a maximum of 100–200 atoms. Modest basis set HF models, such as 3-21G∗ and 6-31G∗ , described below, are acceptable for thermochemical/kinetic calculations, excellent for molecular equilibrium and transition state geometries, and are outstanding in the description of hydrogen-bonded systems. HF methods perform less successfully for reactions involving explicit bond breaking and formation, particularly for comparisons involving energy differences between reactants and transition states. They also perform very poorly in structural and energetic descriptions of transition metal inorganic and organometallic compounds. As with all attempts to solve multiple body problems, ab initio methods have drawbacks. As noted above, in the HF method, the effects of electron–electron repulsion are not treated explicitly. Subsequent developments have incorporated electronic correlation and spin orbit coupling terms. The so-called post-HF methods have arisen alongside improved computing power and more efficient algorithms to improve the accuracy of ab initio calculations.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

8

Techniques

2.5.2 Basis sets A basis set is a set of mathematical basis functions describing an atomic orbital. For example, a description of the bonding and antibonding orbitals available to a hydrogen molecule are given as follows: √ (in phase) = 1/ 2[1s (HA ) + 1s (HB )] √ (out of phase) = 1/ 2[1s (HA ) − 1s (HB )]

(7) (8)

where  is a function that satisfies the Schr¨odinger equation for hydrogen atoms HA and HB . Approaches to the solution of these equations that describe molecular orbitals using the LCAO approximation for multiple atoms can be of varying complexity. STO-3G The minimal basis set used in ab initio calculations is the Slater-type orbital description in which each function is a combination of three Gaussians with coefficients, and exponents are in turn determined through an iterative best fit to Slater-type functions. The description of this basis set is therefore STO-3G in Pople’s notation. The weakness of the method is that it describes hydrogen and helium with one function and the first row main group elements with five functions (one each for 1s, 2s, 2px , 2py , and 2pz ). This simplistic approach yields spherical basis sets for s-orbitals and, when the three orthogonal solutions are combined, for p-orbitals also. Thus, isotropic molecules are described well but those with large degrees of anisotropy are poorly modeled. Furthermore, the solutions are atom centered and give little useful information about internuclear electronic distribution, which affects both formal bonds and the weaker interactions that are of such immense importance in supramolecular chemistry. Split valence basis sets Core and valence electrons can be considered separately to take account of differences in their contribution to bonding. The basis sets are described in terms of the Gaussian functions contributing to the electronic interactions. A 321G basis set assumes a core in which each atomic orbital is described by a single function constructed from three Gaussians. The valence atomic orbitals are described by two different functions comprising two and one Gaussian functions, respectively. If more computational resource is available, then higher levels of basis set are available such as 6-311G. The accuracy of these descriptions of electrons can be improved if it is assumed that they can adopt more than one characteristic; hydrogen-centered s-electrons could take on some p-electron character when incorporated in a C–H sigma bond. To do this, a set of polarization

functions are added to the basis set description so that selectrons can invoke p-type behavior and p-electrons can behave like d-electrons. Polarization of hydrogen atoms is especially important in hydrogen-bonded systems and those with agostic interactions. Occasionally, it is advantageous to add one further level of complexity to the calculation: diffuse functions. Here, heavy atoms are given access to diffuse s- and ptype functions so that their more diffuse orbital regions are considered in calculations. This has relevance for supramolecular chemists interested in anion recognition as diffuse and polarizable orbitals associated with halides are likely to be involved in anion recognition and binding. In Pople notation, these basis sets are denoted by a “+” sign as in 6-311+G∗ .

2.5.3 Møller–Plesset, configuration interaction, and coupled cluster methods Several post-HF models have been developed to give more accurate results based on better treatment of electron correlation. Møller–Plesset (MP) models include an approximation of electron correlation by adding further terms to the HF approach. The first-order solution calculates ground and excited states separately without interacting as in HF methods but the addition of second- and higher-order terms introduces perturbation. The secondorder calculation, or MP2, therefore explicitly includes electron–electron interactions through the effects of electron promotion.14 Configuration interaction (CI) methods expand the HF description to allow mixing of different electronic configurations. One benefit is the ability to calculate ground and excited states. For sensor molecules that function upon guest binding to a host attached to a fluorophore, or a similar signaling group, the relative energies of ground and excited states for guest-bound and free ligands are essential to interpret the mechanism by which they function.15 Coupled cluster (CC) methods also include electron correlation through the creation of “cluster functions” that correlate motions of electrons in pairs or higher-order groupings. The method originated in a pairwise treatment of electrons in a many-electron system invoked by Cizek and Paldus.16

2.5.4 Density function theory (DFT) Density function theory (DFT), proposed by Kohn and Sham, is in principle a formally exact theory, in which energy is evaluated from the electron density, rather than as a solution of the many-electron wavefunction.17 This approach avoids making many of the mistakes involved

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM) with the HF model, although it still involves a varying degree of approximation, depending on the model and basis set used. Density functional methods are based on the theorem that a molecule’s energy minimum may be derived from a functional (or function of a function) of its electron density. The approach reduces the complexity of the equivalent HF calculation with multiple variables to one that requires only three. As a consequence the computational complexity scales with the number of electrons involved, N, rather than 3N and has opened up computational methods to larger molecules, including supramolecular assemblies. DFT is potentially more accurate than HF methods, and accounts for electron correlation by including into the Hamiltonian terms that derive from exact solutions of idealized many-electron systems. In contrast with HF models, DFT methods are able to satisfactorily describe the structure and energetics of transition metal compounds, as well as nonisodesmic thermochemistry. Finally, DFT generally scales better than HF, and therefore can be applied to much larger systems, including simulation of solids and bulk liquids. DFT has some limitations; for example, the weak interactions central to supramolecular and host–guest complexes are not described particularly by the B3LYP functionals often used in DFT calculations,18 but recent developments by Becke have led to much better descriptions of intermolecular complexes.19 The strength of this method is its speed of calculation compared to equivalent HF models. Time-dependent DFT (TD-DFT) Following on from the introduction of the Kohn–Sham DFT concept came Runge and Gross’ DFT for timedependent systems.20 Here the properties of molecules are determined as a snapshot in time and, consequently, can be modeled along a timeline to predict the evolution of a chemical species or reaction. The application of such a method to model supramolecular host–guest interactions and other complex phenomena, such as light-harvesting supramolecular complexes, is clear.

2.5.5 Composite methods Composite methods are used in programs such as GAUSSIAN that combine several types of calculations in pursuit of highly accurate solutions. Thus, the first step may be an MP2 geometry optimization using a 6-31G(d) basis set that accounts for d-orbitals. Polarization functions will then be included by a further calculation of the MP/6-31G(d) output by using it as the input geometry for a MP4/6-311+G(2df, p) basis set. These combinations of methods are sometimes known as thermochemical methods as they have widest uses in the prediction of thermodynamic parameters such as heats of formation.

9

2.5.6 Semiempirical methods Semiempirical methods are rooted in quantum mechanics, but incorporate experimental parameters. Several approaches have been applied to reduce the complexity of quantum mechanical calculations yet give results that are in good agreement with experimental data. The first method, devised by H¨uckel,21 considered π-systems, later extended to all valence electrons by Hoffman.22 Pople proposed methods based on partial or complete neglect of electron repulsion terms, thus they only consider valence electrons and assume localized atomic orbitals, such as the so-called neglect of diatomic differential overlap (NDDO) approximation.23 Under the NDDO approximation, the number of electron–electron interactions scale as N 2 rather than N 4 , where N is the number of mathematical functions used in the calculation. Other parameters derived from experiment, as seen in MM methods, are incorporated to give further timesaving approximations. Despite these shortcuts, it is still possible to generate well-founded models but, as with ab initio methods, it is best to assume that a good MM simulation will generate a suitably accurate structure for which a single-point energy calculation can be calculated. Other thermodynamic data can be determined from a combination of matrices calculated during the simulation, which in turn can be used to generate atomic partial charges (useful in determining charge complementarity between host and guest), molecular orbitals, potential hydrogen bonds, and so on. Semiempirical calculations, although more time consuming than molecular mechanical methods, are still faster and cheaper than simulation from first principles. Most computational chemistry program suites now make use of semiempirical methods based on Dewar’s improvements to the original NDDO approach. One of the first successful methods was Austin Model 1 (AM1), published in 1985, which works well for organic compounds.24 A similar method, Parameterized Model 3 (PM3),25 has gained wider acceptance due to the greater number of elements for which it has been parameterized. Variations such as PM3(tm) implemented in SPARTAN, which includes parameters for numerous transition and heavy metals, are particularly attractive in supramolecular complexes involving organic hosts and metal or anionic guests. A recent advancement in AM1, the Recife Model 1 (RM1) method, where many elemental parameters have been improved, has shown good reproducibility of experimental thermodynamic properties for almost 1500 organic systems and biomolecules.26 With the most advanced semiempirical methods, equilibrium geometries can be calculated, including geometries of transition metal inorganic and organometallic compounds, and they have moderate success with calculation of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

10

Techniques

Desolvation Solvation

Solvated guest dissociation

Solvated guest association

Gas-phase guest dissociation

Gas-phase guest association

Complex desolvation Complex solvation

Figure 6

Free-energy perturbation of a solvated supramolecular complex.

transition state geometries; however, they are less suitable for thermochemical and kinetic evaluations or for conformational assignments.

Most supramolecular systems exist in solvents, therefore any attempt to model supramolecular phenomena should, in theory, include the effects of solvent. As noted below, some methods such as free-energy perturbation can be applied to mimic the effects of solvents; however, one approach that is becoming quite widely adopted is the polarizable continuum model (PCM).27 Rather than model explicit solvent molecules, PCM assumes a global dielectric or conductance due to the solvent in which the molecule of interest inhabits a cavity. A supramolecular analogy would be that the molecule is the guest and the solvent continuum is the host. A van der Waals molecular surface is constructed, which includes an approximation of the solvent molecules’ interaction with the molecular surface. The free energy in solution is the sum of electrostatic and repulsive effects together with a cavitation energy.

single simulation at anything other than the level of MM. A reasonable argument also exists for the use of gas-phase calculations when the solvent involved is effectively noncoordinating. The free-energy perturbation method attempts to consider the effects of explicit solvent molecules, as illustrated in Figure 6. At its simplest, it requires that solvated host, solvated guest, and solvated complex geometries and thermodynamic properties are calculated. The same calculations are carried out on the nonsolvated system. It is important to note that the effects of the solvent on the host geometry when the latter is flexible as the solvated conformation that is most stable in the uncomplexed state is rarely the same as that found in the solvated host–guest complex. Other issues include the affinity of the solvent for the host. For example, a macrocyclic cavity containing several hydrogen bond acceptors, appropriately spaced and directed toward a site for a guest, must be able to allow trapped water to escape so the guest can bind. Similarly, the guest must be able to lose solvent so that it can interact effectively with the host. The energy of desolvation may be crucial in the binding process as it needs to be relatively low if the target guest is to displace the solvent.

2.5.8 Free-energy perturbation

2.5.9 Visualization

Many calculations occur in the gas phase, essentially because it is computationally expensive and time consuming to include host, guest, counterions, and solvents in a

For decades, it was possible to solve quantum mechanical calculations and determine the positions of atoms in dynamic flux using computational methods. However, for

2.5.7 The polarizable continuum model (PCM)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM) most experimentalists, the importance of the results is hard to understand. Modern graphical user interfaces now allow the abstract results to be visualized so that molecules can be displayed in various formats, from sticklike connectivities to electrostatic potentials mapped on to van der Waals surfaces to illustrate host–guest complementarity. It is possible to move molecules on screen so that particular features can be highlighted, which is of particular use when submitting illustrations to journals or explaining complex aspects of geometry. Graphical interface programs are now ubiquitous, and often freely available, so that results, whether of simple molecular connectivities or the changes to electrostatics during complex reactions, can be visualized with ease.

3

APPLICATIONS IN SUPRAMOLECULAR CHEMISTRY

In supramolecular chemistry, it is often necessary to determine the optimum conformation and resulting geometric parameters of a host molecule to determine its likely affinity for specific guests. Initial data may come from crystallographic studies or from structures generated by ChemDraw, ISIS, or similar programs. Once an initial structure has been determined, it is usual to use MM to generate a first best guess of the equilibrium geometry. Thereafter, other methods can be used to interrogate the features of interest to the experimentalist, such as the host’s specificity for a series of guests. The object of the simulation may be merely to determine structural parameters or may include thermochemical data and other simulated information that requires the use of higher-level methods. The following examples have been chosen with supramolecular chemistry in mind and thus focus largely on well-known, host–guest systems.

3.1

Crown ethers

One of the earliest applications of computational methods to supramolecular phenomena was a theoretical investigation into the widely acknowledged size selectivity of crown ethers for particular alkali metals (Figure 7). As early as 1975, STO/3G quantum mechanical simulations of [Li(12crown-4)]+ had been attempted,28 to be followed by 18crown-6 complexes of Na+ , K+ , and NH4 + in 1979.29 Yanabe’s calculations, using the semiempirical CNDO/2 method, gave a good correlation with experimental photoelectron spectra and, importantly noted that: The stability of the complex was reasonably explained by considering the hydrated species of the cation and the complex, indicating the important role of the solvation effect in the selectivity of the crown ether to the cation.29

11

NH HN

NH HN

NH HN

NH HN

NH HN

NH HN

NH HN

NH

HN

NH HN HO O O

O O

O

O

O

O O

O O

N

OH N

O O

Figure 7 Macrocycles investigated in pioneering MM studies: (top, from left) 12-aneN4 , 14-aneN4 , 16-aneN4 , hexaaza-18crown-6, (bottom, from left) 12-crown-4, 15-crown-5 and (N ,N  2-hydroxyethyl)-1,10-diaza-18-crown-6.

MM had been used by Sutherland to investigate the equilibrium geometries of conformations for 9-crown-3, 12-crown-4, and 18-crown-6.30 The simulated conformational preferences of 18-crown-6 were compared with those observed in the X-ray structure of its benzylammonium complex. Shortly thereafter, Kollman applied AMBER to the study of 18-crown-6 in its alkali metal complexes.31 Agreement was found between predicted crown geometries and those found in X-ray crystal structures. Hancock used an in-house MM approach to analyze macrocyclic coordination complexes as early as 198032, 33 before applying a similar methodology to crown ethers and related ligands.34 The principle was quite standard for its time; complexes were geometry optimized so that the complex with the lowest relative strain energy could be identified. Kollman’s study was later extended to the benzocrowns, dibenzo18-crown-6, and dibenzo-30-crown-10, where free-energy perturbation theory was used to reveal the importance of solvation in the calculations.35 Santos used MM (MM2 and AMBER) to compute eight low-energy conformers of the all-nitrogen-containing analog of 18-crown-6, hexaaza-18crown-6, which were then used as initial structures from which the triprotonated macrocycle could be generated.36 This species was investigated for its affinity toward silicic, borate, and carbonate anions. The interaction between the ammonium cation and 18-crown-6 in the gas phase was reinvestigated by Chakraborty using DFT methods.37 The most stable com˚ out of the macroplex was shown to have the cation 0.4 A cyclic plane with low barriers to guest rotation but higher barriers to tilting out of the plane perpendicular to the macrocycle. In a more ambitious study, Hay combined MM2 mechanics calculations for the K+ complexes of 11 crown ethers with HF/STO-3G calculations using GAUSSIAN 90. The results indicated that in selecting crown ethers based on apparent fit to target cations it was important to note the following:

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

12

Techniques

Consideration of M–O bond length preferences alone is a poor basis for crown ether design. We conclude that it is also necessary to consider the orientation of the C–O–C groups, relative to the metal ion, to understand the relationship between structure and reactivity. Hay later made use of the MM3 force field to accurately model 51 supramolecular complexes of macrocyclic and acyclic ligands with alkali and alkaline earth metal cations.38 Again, parameterization of MM3 to yield an augmented force field, MM3 , was informed by X-ray structures and HF calculations. Similar ab initio informed parameterization of an MM force field was reported by Thompson as a method to predict explicit solvation of the [K(18-crown-6)]+ complex.39 Restricted HF and the more advanced MP2 calculations were performed with GAUSSIAN 92 and GAMESS programs using the 6-31+G∗ basis set. The resulting bond distances and angles were used to create a hybrid QM/MM method that accurately described the [K(18-crown-6)]+ complex, or at least was in good agreement with other computational approaches. Glendening made a wider study of 18-crown-6 complexation but extended the metals investigated to include lithium through to cesium.40 Despite many papers following up these results and improving slightly on the data that come from theoretical models, there have been no major advances where simple crown ethers are concerned. However, the benchmarking of the 1990s has found wider applicability through greater implementation of higher-level ab initio codes and improved MM force fields. Consequently, researchers interested in more unusual crowns or guests have been able to create useful models. An alternative method can be found in Drew’s conformational analysis of 1,4,7-trithiacyclononane, where an “elastic bond” was invoked when generating conformers. A bond was broken, conformers generated, and the bond reformed before optimizing the geometry of the structure using the universal force field within the CERIUS suite of programs. Analysis of 729 structures showed that they fell into a small number of low-energy conformers (within 35 kJ mol−1 of the lowest energy structure found). Examples of these, as shown in Figure 8, have been found to match many of the predicted geometries with the exception of the lowest energy conformer, which has yet to be found.41 Wang used DFT analysis of a macrotricyclic tetramine hexaether and its protonated form, using B3LYP/6-31G methods implemented in GAUSSIAN 2003W, to predict the macrocycle’s binding affinity toward halide anions.42 The same group extended the method to model the bipyridylincorporating 15-crown-5, which operates as a molecular machine as a consequence of metals binding in the macrocyclic cavity or externally to the bipyridyl cleft.43 As an added complexity, the crown was modeled with and without externally bound tungsten tetracarbonyl, which required

0.0

5.4

14.9

16.0

21.4

Figure 8

8.1

12.0

17.6

19.4

27.1

28.5

Relative energies (kJ mol−1 ) of [9]ane-S3 conformers.

accurate modeling of a late transition metal. The DFT method was chosen, as implemented in GAUSSIAN 03, and the B3LYP hybrid functional used with the 6-31G(d) basis set. The gas-phase binding energies of the sodium and potassium complexes were calculated. Results showed that the allosteric effect of the tungsten moiety reduced the ligand’s affinity for both potassium and sodium. High-level computational methods are valuable in determining geometries, electron densities, and binding affinities; however, one of the most important aspects of the use of crown ether is in phase transfer catalysis. Here it is necessary to model binding affinities in one or more solvents, a task for which ab initio techniques are unsuited at the present time due to the large number of molecules necessary for an accurate simulation. Wipff has been in the forefront of MD studies of solvated crown ether complexes since 1985.44 The approach is to place the solvated ligand and target ion pair in a box, which is subject to periodic boundary conditions. MD simulations are then carried out for a set period of time, usually picoseconds to nanoseconds. Snapshots of the simulations are taken during the binding trajectory so that the process may be better understood.45 More recently, this method has been applied to 18-crown-6 extractions from ionic liquids, which helped illuminate the roles of water and the ionic liquid.46 One of the benefits of computational chemistry is the ability to model compounds and complexes that are hard to synthesize or are inherently difficult to study. Schreckenbach used a relativistic DFT to model 18-crown-6 complexes of the actinide species UO2 2+ , NpO2 + , and PuO2 2+ and found good agreement with experimentally determined bond lengths for Np(V). Experimental uncertainties were

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM)

13

Figure 9 Cyclodextrin inclusion complexes: (from left) β-cyclodextrin and benzoic acid, methyl-β-cyclodextrin and omeprazole, and 2-hydroxypropyl-β-cyclodextrin and ampicillin.

invoked for the less clear cut agreement with the other two examples. The simulations explained the binding preference for actinides in lower oxidation states. Gas-phase binding energies calculated for the penta aquo complexes and crown ether complexes of the actinides studied show that there is no intrinsic preference, or “better fit,” for actinyl(V) cations as compared to actinyl(VI) ones. Rather, the ability of NpO2 + (Np-V) to form in-cavity 18crown-6 complexes in water is traced to solvation effects in polar solvents. Thus, the effective screening of the charge provided by the macrocycle leads to destabilization of the An(VI) crown complexes relative to their An(V) counterparts.47 The simulation was thus able to give some insight into the electronic, as well as geometric, influence of the macrocycle.

3.2

Cyclodextrins

Cyclodextrins have had valuable industrial uses for a considerable time, particularly as agents to bind or release volatile molecules. Accurate predictions concerning the selectivity and stability of cyclodextrin–guest complexes are therefore of considerable interest both academically and practically.48 MD was used to simulate cyclodextrin hydrates49–51 as a test of the applicability of the GROMOS program package to systems beyond proteins and nucleic acids. Other early MD simulations focused on interactions with guests such as enantiomers of methyl-2-chloropropionate.52 Comparisons between calculated thermodynamic properties for complexes formed by α-cyclodextrin with para-substituted phenols and the results of MM simulations led to improvements in force fields that described the interactions.53 MM2 simulations were used to support NMR data for the β-cyclodextrin inclusion complex with benzoic acid.54 The well-known catalytic effect of cyclodextrins has been modeled. For example, the relative rate increase of hydrolysis of S over R phenyl ester stereoisomers in the presence of β-cyclodextrin

was found to correlate with a MD, free-energy perturbation simulation.55 The MD investigation of the interaction between a β-lactam antibiotic and ampicillin with β-cyclodextrin and its 2-hydroxypropyl derivative has been of greater application.56 Results suggested that the inclusion complex could be a successful drug delivery method. Similarly, MD studies of the complex formed by methyl-βcyclodextrin and omeprazole, a drug used to treat duodenal ulcers, indicated that the cyclodextrin derivative could act as a valuable excipient for omeprazole.57 These examples are illustrated in Figure 9. While simple MM and dynamics approaches can give useful insights into cyclodextrin complexation, thermodynamic and spectroscopic information requires higher-level calculations. Cyclodextrin complexes with dialkyl tartrates were modeled in a range of solvents using DFT methods and were compared with experimental data.58 The calculations supported the experimental data if solvated tartrate clusters were invoked in the simulation, thus giving a useful visual representation of the complexes. The complex of β-cyclodextrin and the pyrethroid insecticide cypermethrin was modeled by DFT (B3LYP/631G(d)) to show that the guest inserted at the wider opening of the macrocycle with its phenyl ring within the cavity. The simulation indicated that the driving force was the formation of an intermolecular hydrogen bond. Calculated data were in good agreement with spectroscopic determinations and the thermodynamic analysis was consistent with guest inclusion being a spontaneous enthalpy-driven process.59

3.3

Calixarenes

The explosion in interest in calixarenes in the 1980s following the publication of Gutsche’s improved synthesis of 4-t-butylcalix[4]arene60 led, unsurprisingly, to computer models. One of the most interesting features of the calix[4]arenes was the through-the-annulus rotation mechanism that generated the four well-known conformers: cone, partial cone, 1,2-alternate, and 1,3-alternate, depicted in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

14

Techniques

OH OH

OH OH OH OH HO

Figure 10

OH OH OH

HO HO

OH

OH OH OH

Conformers of 4-t-butylcalix[4]arene: (from left) cone, partial cone, 1,2-alternate, and 1,3-alternate.

Figure 10. The aromatic rings of the unsubstituted 4-tbutylcalix[4]arene are free to rotate once they have broken the strong hydrogen bonds between phenols in the cone conformer, but once O-alkylation occurs the rotation becomes harder. Under ambient conditions, substituents larger than ethyl groups effectively freeze the calixarene conformer into one of the available four. Early attempts to model the calix[4]arenes by Andreetti employed MM to determine the stability of pyridine inclusion complexes.61 Also, using mechanics, McMurry was able to show that the crystallographically determined structure of unsubstituted calix[4]arene did not in fact represent an energy minimum.62 A molecular mechanical analysis of all four isolated conformers of 1,3-diethyl-2,4-dimethyl ether 4-t-butylcalix[4]arene using QUANTA/CHARMm was essential to determine their relative stabilities.63 Wipff pioneered the use of MM/MD simulations to investigate the behavior of tetra(diethyl)amido-4-t-butylcalix[4]arene, the 4-t-butylcalix[4]arene anion, the 1,3-alternate 1,3dimethoxy-4-t-butylcalix[4]arene-crown, and calix[4]arenebis-crown-6 at the water–chloroform interface.64 An example of this interfacial simulation is shown in Figure 11. Later, in collaboration with Harrowfield, Wipff used a similar strategy to investigate alkali metal salts of 4-tbutylcalix[4]arene in acetone and acetonitrile.65

Figure 11 Simulating water:chloroform interfacial extraction by a crown ether.

More recently, it has become possible to extract useful information from ab initio studies of calixarenes as is evidenced in a review by Schatz.66 DFT calculations on tetrapropoxy-4-t-butylcalix[4]arene with H3 O+ helped Kˇr´ızˇ to interpret spectroscopic data related to the structure of the complex,67 and the conformational equilibria of tetraaminoand tetramercaptosulfonylcalix[4]arenes were determined by Magalhaes by the same method.68 Where the calixarene is larger and more flexible, other approaches are necessary. O-Substituted 4-t-butyloxacalix [3]arenes are easily prepared from the parent macrocycle, which combines phenolic and ethereal moieties in its 18-membered ring. The macrocycle’s flexibility allows full rotation of the phenols through its annulus, but, upon O-alkylation, with substituents larger than ethyl groups, becomes frozen out in either the cone or partial cone conformer, as illustrated in Figure 12. The former conformer is preferred for most molecular recognition applications, hence, to gain an insight into the factors affecting conformational preferences of the O-alkylated compounds, a study was undertaken using the Chem-X ALLATOM forcefield. The oxacalix[3]arene derivative structures were created and a single C–C bond removed. A full conformational search generated 120.9 million structures. Of these, only structures in which the distances and angles between the terminal carbon atoms were within normal C–C bonding distances and angles were retained. The C–C bond was replaced and each structure geometry optimized and assigned as a cone or partial cone based on positions of the substituents relative to the plane forms by the macrocyclic ring. The tri(picolyl) derivative showed that the expected distribution of conformers (25 : 75 cone: partial cone) was replicated with a 24 : 76 ratio but the tris(N,N-diethylacetamide) derivative had a 34 : 66 ratio. Experimentally, it was shown that the partial cone tri(picolyl) derivative could be obtained in 68% yield, in line with prediction. The cone tris(N,N diethylacetamide) derivative was isolated in 44% yield, which was clearly much greater than the theoretical 25% and closer to the computational prediction.69 Computational methods were also able to explain the differences between binding affinities and extractability of numerous metals by this calixarene derivative.70 Another derivative, with three

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM)

O OH

15

O O OH OH

OH O OH

Figure 12 derivative.

O OH

O

4-t-Butyloxacalix[3]arene: (left) cone and partial cone conformers and (right) a zinc complex model of an adamantyl

lower rim adamantyl methyl ketone substituents, was shown to be a weak extractant for most metals except Na+ . A combination of MM and ab initio (HF/6-31G) calculations generated structures that helped in the interpretation of the complexes’ NMR data.71

3.4

Miscellaneous host–guest systems

The hexaaza macrocycle known as a torand was modeled by Bell and Drew in order to better understand the conformers present in the X-ray structures of its alkali metal complexes.72 An investigation of the conformers available to tri(n-butyl)torand generated all eight by manually changing N–C–C–N torsion angles around the central 18-membered ring. The analysis ignored the complications of the flexible n-butyl substituents as the focus was on the macrocyclic cavity. The study, using both AMBER and SYBYL forcefields, showed that a conformer in which nitrogen atoms alternated with three pointing above the plane of the macrocycle and three below (+ − + − +−) was the lowest in energy. The same conformer was observed for the solidstate structures of both the potassium and rubidium complexes.73 Host–guest complexes can be successfully modeled using the simple techniques of conformational searching and geometry optimization (Figure 13). A 2-(5-methylpyridin-2-yl)-3-(6-(pyridin-2-yl)pyridin-2-yl)pyrazine ligand reported by Heirtzler formed complexes with copper(I) initially as a C2 stereoisomer of two metals and two ligands before relaxing to a less sterically congested meso form. Modeling both structures suggested that the latter was more stable by 33 kJ mol−1 and 1 H NMR indicated

the meso form to be the more stable in solution.74 Interestingly, the C2 stereoisomer was isolated and studied by X-ray crystallography! Another example of host–guest interactions being predicted by simple MM can be seen in Steed’s chloridebinding tripodal ligand. The podand was based on benzene with alternating ethyl and 3-pyridiniumferrocenylmethylamine substituents and was known to give an electrochemical response to halide anions. Binding constants were found to favor chloride over other anions by almost an order of magnitude. In the absence of crystallographic evidence, a conformational search was undertaken using Monte Carlo sampling to reduce the number of hits. The pseudo-cone geometry formed more stable complexes with hexafluorophosphate and chloride anions than the pseudo-partial cone and chloride was preferred by a significant margin. The calculated structure indicated unusually close interactions between the chloride guest and a pyridyl proton. It transpired that this was supported by a shift in the 1 H NMR of almost 1 ppm, which occurred while the first equivalent of the anion was added, thereafter no further changes were observed. Both experimental and theoretical data pointed to an initial 1 : 1 binding of chloride within the cavity of the podand.75 Most computational methods attempt to determine optimized geometries or calculate thermodynamic properties. Using an approach closer to the quantitative structure– activity relationship (QSAR) methods, widely used in industrial drug discovery divisions, Sheehan discovered a relationship between experimental binding affinities and guest LUMO energies in supramolecular complexes.76 Equilibrium geometries of the complexes were generated by a PM3 refinement of MM-derived structures, which were

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

16

Techniques

N N

N K+

N

N N

N N

N N Cu+ N + N Cu N N N N

Figure 14 4-t-Butyloxacalix[3]arene complex with a cavitybound quaternary ammonium cation: initial restrained geometry (top) and final geometry following MMFF/PM3 optimization (bottom). N+ Cl−

N+

N+

Figure 13 Examples of host–guest complexes: (from top) Bell’s torand, Heirtzler’s copper dimer and Steed’s podand.

followed by single-point energy calculations at the HF/631G∗ or BP/6-31G∗ level in a trade-off between computational time and accuracy. The protocol was used to examine host–guest complexes between 4-t-butyloxacalix[3]arene and quaternary ammonium cations, shown in Figure 14, as well as the more familiar complexes between a range of crown ethers and simple cations or small ammonium species. Speiss used ab initio methods (HF/6-31G∗ ) and the gauge-invariant atomic orbital (GIAO) approach to generate NMR chemical shifts to unravel the behavior of a naphthylene-bridged molecular tweezer and a 1,4dicyanobenzene guest. The benefit of augmenting experimental NMR data (in the solid-state and solution phase), and the X-ray structure of a 1 : 1 complex, with computational methods was clear to the authors who stated the following: All experiments were performed on only 10 mg of a powdered sample, without isotopic labeling. Therefore, it

is envisaged that solid-state NMR combined with quantum chemistry can become as valuable to the chemist as solutionstate NMR is today.77 Even some of the most intricate supramolecular systems can be probed by appropriate simulations. Jaime examined the translational barriers in [2]rotaxanes using MM3 and was able to show that the nonbonded interactions were the main cause.78 It has since been possible to interrogate these systems by ab initio methods. Goddard used DFT models (B3LYP/6-31G∗∗ ) as a step toward understanding the switching mechanism of a cyclobis(paraquatp-phenylene) shuttle as it moves between tetrathifulvalene and 1,5-dioxynaphthalene stations in a [2]rotaxane.79 Calculations of molecular orbital energies for different shuttle positions strongly suggested that an electron-tunneling mechanism was the key that led to the possibility of designing future [2]rotaxanes using computational methods.

4

CONCLUSIONS

The concept of the host–guest complex is a the heart of supramolecular chemistry. It is the formation of this chemical entity that allows us to extract specific chemical species and sense others. Complex formation can be detected through the changes in numerous parameters such as color, fluorescence, NMR spectra, mass spectra, and electrochemical response. The complex can be imaged by

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM) single-crystal diffraction methods so that the influence of steric effects, hydrogen bonding, and electrostatics can be assessed. One problem inherent in observing host–guest complexation is that most spectral and electrochemical methods are dynamic but reveal little of the atomic–scale mechanisms, whereas solid-state techniques give an accurate snapshot of the complex at the atomic level but tell us nothing about the processes involved in complex formation. Computational simulations are able to bridge the gap. By generating models that reflect both the changes in experimental parameters and the correct structures of the complex and its components, the process of host–guest binding can be studied in detail. The key to the success of computational methods is the accuracy of the model and the methods used by which it is generated.

17

19. A. D. Becke and E. R. Johnson, J. Chem. Phys., 2005, 15, 154101. 20. E. Runge and E. K. U. Gross, Phys. Rev. Lett., 1984, 52, 997. 21. E. H¨uckel, Z. Phys., 1931, 70, 204. 22. R. Hoffmann, J. Chem. Phys., 1963, 39, 1397. 23. J. A. Pople and D. Beveridge, Approximate Molecular Orbital Theory, McGraw-Hill, NY, 1970. 24. M. J. S. Dewar, E. G. Zoebisch, E. F. Healy, J. J. P. Stewart, J. Am. Chem. Soc., 1985, 107, 3902.

and

25. J. J. P. Stewart, J. Comput. Chem., 1989, 10, 209. 26. G. B. Rocha, R. O. Freire, A. M. Simas, and J. J. P. Stewart, J. Comput. Chem., 2006, 27, 1101. 27. B. Mennucci and J. Tomasi, J. Chem. Phys., 1997, 106, 5151. 28. A. Pullman, C. Gissener-Prette, and Y. V. Kruglyack, Chem. Phys. Lett., 1975, 35, 156. 29. T. Yanabe, K. Hori, K. Akagi, and K. Fukui, Tetrahedron, 1979, 35, 1065.

REFERENCES 1. W. J. Hehre, J. Yu, P. E. Klunziger, and L. Lou, A Brief Guide to Molecular Mechanics and Quantum Mechanical Calculations, Wavefunction Inc., CA, 1998.

30. M. J. Bovill, D. J. Chadwick, I. O. Sutherland and D. Watkin, J. Chem. Soc., Perkin Trans. 2, 1980, 1529. 31. G. Wipff, P. Weiner, and P. Kollman, J. Am. Chem. Soc., 1982, 104, 3249.

2. http://www.nobelprize.org/nobel prizes/chemistry/laureates/ 1998/press.html (accessed 14/06/11).

32. R. D. Hancock and G. J. McDougall, J. Am. Chem. Soc., 1980, 102, 6553.

3. C. A. Lipinski, F. Lombardo, B. W. Dominy, P. J. Feeney, Adv. Drug Delivery Rev., 1997, 23, 3.

33. R. D. Hancock, Prog. Inorg. Chem., 1989, 37, 187.

and

34. R. D. Hancock, J. Chem. Educ., 1992, 69, 615.

4. T. H. Keller, A. Pichota, and Z. Yin, Curr. Opin. Chem. Biol., 2006, 10, 357. 5. P. R. Westmoreland, P. A. Kollman, A. M. Chaka, et al., WTEC Panel Report on Applications of Molecular and Materials Modelling, Kluwer Academic Publishers, Dordrecht, 2002. 6. C. A. Schiffer, J. W. Caldwell, P. A. Kollman, R. M. Stroud, Mol. Simul., 1993, 10, 121.

and

7. N. Foloppe and A. D. MacKerell, J. Comput. Chem., 2000, 21, 86. 8. S. Berneche and B. Roux, Nature, 2001, 414, 73. 9. M. Garcia-Viloca, J. Gao, M. Karplus, and D. G. Truhlar, Science, 2004, 303, 186. 10. R. Car and M. Parinello, Phys. Rev. Lett., 1985, 55, 2471. 11. M. Saunders, K. N. Houk, Y. D. Wu, et al., J. Am. Chem. Soc., 1990, 112, 1419. 12. T. Noguti and N. Go, Biopolymers, 1985, 24, 527. 13. P. W. Atkins and R. S. Friedman, Molecular Quantum Mechanics, OUP, Oxford, 1997. 14. J. S. Binkley and J. A. Pople, Int. J. Quantum Chem., 1975, 9, 229.

35. P. D. J. Grootenhuis and P. A. Kollman, J. Am. Chem. Soc., 1989, 111, 2152. 36. M. A. Santos and M. G. B. Drew, J. Chem. Soc., Faraday Trans., 1991, 87, 1321. 37. Y. L. Ha and A. K. Chakraborty, J. Phys. Chem., 1992, 96, 6410. 38. B. P. Hay and J. R. Rustad, J. Am. Chem. Soc., 1994, 116, 6316. 39. M. A. Thompson, E. D. Glendening, and D. Feller, J. Phys. Chem., 1994, 98, 10465. 40. E. D. Glendening, D. Feller, and M. P. Thompson, J. Am. Chem. Soc., 1994, 116, 10657. 41. J. Beech, P. J. Cragg, and M. G. B. Drew, J. Chem. Soc., Dalton Trans., 1994, 719. 42. X. Zheng, X. Wang, S. Yi, et al., J. Comput. Chem., 2009, 31, 871. 43. Y. Miao, X. Wang, X. Jin, et al., J. Comput. Chem., 2010, 32, 406. 44. G. Ranghino, S. Romano, J. M. Lehn, and G. Wipff, J. Am. Chem. Soc., 1985, 107, 7873. 45. L. Troxler and G. Wipff, Anal. Sci., 1998, 14, 43.

15. P. J. Knowles and H. J. Werner, Chem. Phys. Lett., 1988, 145, 514.

46. P. Vassiere, A. Chaumont, and G. Wipff, Phys. Chem. Chem. Phys., 2005, 7, 124.

16. J. C´ızek and J. Paldus, Int. J. Quantum Chem., 1971, 5, 359. 17. W. Kohn and L. Sham, Phys. Rev. B, 1965, 140, 1133.

47. G. A. Shamov, G. Schreckenbach, R. L. Martin, P. J. Hay, Inorg. Chem., 2008, 47, 1465.

18. H.-J. Schneider, Angew. Chem. Int. Ed., 2009, 48, 3924.

48. K. B. Lipkowitz, Chem. Rev., 1998, 98, 1829.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

and

18

Techniques

49. J. E. H. Koehler, W. Saenger, and W. F. van Gunsteren, Eur. Biophys. J., 1987, 15, 197.

76. R. Sheehan and P. J. Cragg, Supramol. Chem., 2008, 20, 443.

50. J. E. H. Koehler, W. Saenger, and W. F. van Gunsteren, Eur. Biophys. J., 1987, 15, 211.

77. S. P. Brown, T. Schaller, U. P. Seelbach, et al., Angew. Chem. Int. Ed., 2001, 40, 717.

51. J. E. H. Koehler, W. Saenger, and W. F. van Gunsteren, Eur. Biophys. J., 1988, 16, 153.

78. X. Grabuleda and C. Jaime, J. Org. Chem., 1998, 63, 9635.

52. J. E. H. Koehler, M. Hohla, M. Richters, and W. A. Konig, Chem. Ber., 1994, 127, 119. 53. A. E. Mark, S. P. van Helden, P. E. Smith, et al., J. Am. Chem. Soc., 1994, 116, 6293. 54. D. Savatierra, C. Jaime, A. Virgili, and F. Sanchez Ferrando, J. Org. Chem., 1996, 61, 9578. 55. V. Luzhkov and J. Aqvist, J. Am. Chem. Soc., 1988, 120, 6131. 56. H. Aki, T. Nilya, Y. Iwase, et al., Thermochim. Acta, 2004, 416, 87. 57. A. Figueiras, J. M. C. Sarraguca, R. A. Carvalho, et al., Pharm. Res., 2007, 24, 377.

79. Y. H. Jang, S. G. Hwang, Y. H. Kim, et al., J. Am. Chem. Soc., 2004, 126, 12636.

FURTHER READING G. H. Grant and W. G. Richards, Computational Chemistry, OUP, Oxford, 1995. A. Hinchliffe, Chemical Modeling: From Atoms to Liquids, WileyVCH Verlag GmbH, Chichester, 1999. J. H. Jensen, Molecular Modeling Basics, CRC Press, Boca Raton, FL, 2010.

58. P. Zhang and P. L. Polavarapu, J. Phys. Chem. B, 2007, 111, 858. 59. W. Li, B. T. Lu, F. F. Chen, et al., J. Mol. Struct., 2011, 990, 244.

REVIEWS

60. C. D. Gutsche, B. Dhawan, K. H. No, and R. Muthukrishnan, J. Am. Chem. Soc., 1981, 103, 3782.

N. L. Allinger, Y. H. Yu, and J. H. Lii, J. Am. Chem. Soc., 1989, 111, 8551.

61. G. D. Andreetti, O. Ori, F. Ugozzoli, et al., J. Inclusion Phenom., 1988, 6, 523.

D. A. Perlman, D. A. Case, J. W. Caldwell, et al., Comput. Phys. Commun., 1995, 91, 1.

62. J. E. McMurry and J. C. Phelan, Tetrahedron Lett., 1991, 32, 5655.

C. J. Cramer and D. G. Truhlar, J. Comput. Aided Mol. Des., 1992, 6, 629.

63. L. C. Groenen, J. D. Van Loon, W. Verboom, et al., J. Am. Chem. Soc., 1991, 113, 2385.

W. Kohn, A. D. Becke, and R. G. Parr, J. Phys. Chem., 1996, 100, 12974.

64. G. Wipff, E. Engler, P. Guilbaud, et al., New J. Chem., 1996, 20, 403. 65. R. Abidi, M. V. Baker, J. M. Harrowfield, et al., Inorg. Chim. Acta, 1996, 246, 275. 66. J. Schatz, Collect. Czech. Chem. Commun., 2004, 69, 1169. 67. J. Koˇr´ızˇ , J. Dybal, E. Makrl´ık, et al., Supramol. Chem., 2008, 20, 487. 68. A. Suwattanamala, A. L. Magalhaes, and J. A. N. F. Gomes, J. Mol. Struct: Theochem, 2005, 729, 83. 69. P. J. Cragg, M. G. B. Drew, and J. W. Steed, Supramol. Chem., 1999, 11, 5. 70. P. M. Marcos, J. R. Ascenso, and P. J. Cragg, Supramol. Chem., 2007, 19, 199. 71. P. M. Marcos, J. R. Ascenso, M. A. P. Segurado, et al., Tetrahedron, 2009, 65, 496.

COMMONLY USED COMMERCIAL AND ACADEMIC SOFTWARE PACKAGES ADF : the Amsterdam density functional program, available since 1995, uses STO basis functions to calculate spectroscopic data and can model elements up to 118. CASTEP : uses DFT to calculate electronic properties of molecules, liquids, amorphous, and crystalline solids. ChemOffice: a suite of programs incorporating drawing (ChemDraw, the standard format for chemistry journals) and modeling (Chem3D, using MM and semiempirical methods) with interfaces to other programs.

72. P. J. Cragg, T. W. Bell, A. D.-I. Kwok, and M. G. B. Drew, Abstr. Pap. Am. Chem. Soc., 1991, 202, 297.

CRYSTAL: designed to apply HF and DFT methods to periodic materials such as crystals, surfaces, and linear polymers.

73. T. W. Bell, P. J. Cragg, M. G. B. Drew, et al., Angew. Chem. Int. Ed. Engl., 1992, 31, 345.

DMol : uses DFT to calculate properties of molecules, surfaces, and crystals including organometallics and catalysts.

74. P. J. Cragg, F. R. Heirtzler, M. J. Howard, et al., Chem. Commun., 2004 280.

GAMESS (UK/US): the general atomic and molecular electronic structure system program, which split into two different variants in 1981, for HF, MP2, MP3, CC, CI, and DFT methods.

75. L. O. Abouderbala, W. J. Belcher, M. G. Boutelle, et al., Proc. Natl. Acad. Sci. U.S.A., 2002, 99, 5001.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Computational techniques (DFT, MM, TD-DFT, PCM) GAUSSIAN : initially released by Pople as Gaussian70, and originally using HF models with Gaussian functionals, the package now includes MM forcefields, semiempirical methods, MP2 to MP5 calculations, DFT, and high-accuracy composite methods. GROMOS : a MD program predominantly for biomolecules in solution or the crystalline state. MOLCAS : intended for ground- and excited-state calculations, with a focus on excited-state potential surfaces, available methods include HF, DFT, MP, and PCM. MOPAC : originally released as Dewar’s Molecular Orbital PACkage, the program works at the semiempirical level of AM1, PM3 (and later improvements), and RM1.

19

Northwest National Laboratory and encompasses MM, MD, HF, DFT, TD-DFT, and post-HF methods. Q-Chem: a comprehensive package that incorporates HF, DFT, post-HF, and excited-state methods, such as TD-DFT. SPARTAN : a wide ranging package that includes MM, semiempirical (MNDO, AM1, PM3(tm), RM1), HF, DFT, CC, MP, excited-state methods (TD-DFT, CI), and composite methods, together with advance graphical outputs. TURBOMOLE : a quantum chemistry package incorporating MM, MD, HF, DFT, and MP2 together with CC, TD-DFT methods for excited states. VASP : the Vienna ab initio simulation package is designed for quantum mechanical MD using DFT and HF methods.

NWChem: designed to run in parallel as well as single node configurations, the package was developed at the Pacific

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc023

Isothermal Titration Calorimetry in Supramolecular Chemistry Franz P. Schmidtchen Technische Universit¨at M¨unchen, Munich, Germany

1 Introduction 2 Thermodynamic Foundations 3 Calorimetric Data Acquisition 4 Data Evaluation 5 Extending the Measurement Range 6 Expanding the Scope of ITC 7 Conclusion Acknowledgments References Further Reading

1

1 2 4 9 13 17 20 21 21 22

INTRODUCTION

Molecular interactions, by necessity, involve an exchange of energy and momentum between the noncovalently bonded interaction partners. Measuring such an exchange therefore can report on the immediate events taking place at the molecular scene provided a trustful conceptual link connects the microscopic world to the macroscopic platform of physical determinations. Isothermal titration calorimetry (ITC) offers this possibility based on one of the best established fields in physics: thermodynamics. Ultimately, ITC measures the transfer of heat to the outside on perturbing an enclosed system by incremental addition of one component to all the others in a solution phase. In solution, all molecular species can interact with one another. Rapid collisions among them lead to an equidistribution of

energy over all accessible degrees of freedom, furnishing eventually a macroscopically time-invariant state, the equilibrium, that represents a time-averaged situation. On the molecular level, the collisional exchange continues, leading to a redistribution of all independently movable particles on a timescale spanning about 20 orders of magnitude (judged on the basis of human perception). The huge spread of lifetimes of transient species reflects the quite different tendency of molecules to form associated species: that is, complexes comprising a variety of individual molecules including solvents that stand out above the rest by their sheer probability of occurrence and thus can be identified macroscopically. Despite the reversibility of all bilateral molecular relationships, the intrinsic stickiness of peculiar molecules for each other generates uneven amounts (concentrations in solution) of the aggregated complexes that represent minima in standard free energy
G◦ . As detailed below in the section on association energetics, free energies are composites of the change in the total number of populated energy levels at the temperature of measurement T
S ◦ and the overall change in direct mutual interactions encompassing all participants
H ◦ . It is the latter quantity that is determined in a time-dependent fashion in an ITC experiment. By summation over time, the heat released or taken up by the system at a fixed temperature on addition of aliquots of an interaction partner can be related to the nominal change in the composition, allowing the characterization of the equilibrium state. Thus, ITC measurements provide ready access to a full evaluation of the energetics of reversible binding of interaction partners in solution using heat as a universal experimental probe. Quite a number of in-depth discussions describe the conceptual basis and experimental details of ITC mainly from the perspective of biological applications.1–26

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

2

Techniques

As supramolecular chemistry generically addresses the high-end regime of molecular interactions, it particularly profits from the universality and independence from material peculiarities (the absence or presence of labels or indicator probes, transparency, homogeneity, etc.), rendering the measurement of heat energy (calorimetry) traded in solution processes an indispensable tool to learn about and characterize noncovalent interactions.27 With the advent of ultrasensitive and fully computeroperated calorimeters two decades ago, the ITC method evolved into the most rapid and accurate means to determine equilibrium energetics. The extraordinary expansion in scope, which is documented in very useful yearly reports written by senior experts in the field,28–34 comes primarily from biological studies catering to the need to characterize the binding relations in biopolymers. Thus, most fundamental elaborations of the original technique to cope with limited substrate availability, outrageous binding affinity, or minute signal-to-noise ratios were tested using proteins or nucleic acids, although the complexity of such systems imposes a great obstacle to any meaningful evaluation. The great benefit of near-universal applicability of calorimetry turns into a drawback when the disintegration of the measured signal into the contributing components is ultimately desired, for instance, to derive some mechanistic detail. In fact, the heat change determined is indeed a global response function of many processes occurring simultaneously. The direct mutual interaction of the binding partners as the process of prime interest is accompanied and partly covered by supplementary side reactions like proton transfers, counterion release, conformational rearrangements, and so on, which add an initially unknown share to the heat output. Also, less specific contributions such as the heats of dilution and mixing may intervene and must be taken into account. Naturally, the deconvolution of all intertwined processes is easier when there are fewer components participating in the specific process under study and less severe interference by nonspecific side reactions. On both counts, ITC studies of artificial host–guest systems, which are more readily controlled than the biological pendants, should be particularly suited to find out about the general factors governing supramolecular interactions. They can also be readily extended to nonaqueous solvents and thus unfold the energetic basis for technical applications such as in extractions, gel formation, or assembly processes. Inspecting the binding energetics of the same host–guest system in various solvents may also shed light on the role played by solvation. Despite being the most fundamental, yet unspecific, supramolecular interaction empirically known to affect binding, there is little quantitative knowledge available that reaches beyond general polarity concepts to explain the influence of solvation on host–guest

affinity and eventually is suitable in assisting rational host design. Another fertile domain of ITC investigations addresses the interactions of partners leading to structurally less defined complexes. Amphiphilic compounds aggregating into micelles or vesicles22, 35–38 and supramolecular polymers21, 39, 40 belong to this class. The assignment of quantitative number tags, for example, the molar heat or critical micellar concentration (cmc) characterizing reversible association, are easily performed using ITC and represent reproducible ensemble averages in such rapidly equilibrating systems. Compared to most spectroscopic techniques, ITC is physically a slow method that, however, also contains kinetic information. For instance, the rate of approach to the binding equilibrium can be taken to derive the rate constants of association and dissociation.41 Moreover, ITC measurements can be employed to follow reaction rates as in the determination of standard enzyme kinetics even under conditions that undermine the use of spectroscopic probes.42, 43 Clearly, ITC is a very versatile, sensitive, destructionfree, label-free, and very rapid (takes 1–3 h) instrumental method that requires typically less than micromolar amounts of material to learn about the thermodynamic parameters of reversible molecular associations. The accuracy and reliability of ITC for this purpose is unsurpassed, making it the gold standard in the characterization of intermolecular interactions in solution.

2

THERMODYNAMIC FOUNDATIONS

Supramolecular interactions by definition refer to the reversible formation and cleavage of bonds encompassing all participants starting from the species in prime focus, but also involving all changes in the molecular environment. The heat evolved or consumed in the shuttling of bonds, thus, is the integral result of manifold simultaneous events that add up to a truly global response. Because of the huge number of individual species present even in the most diluted sample solution, the energetic output is constant for defined compositions and framing conditions (p, T ) and potential fluctuations are many orders of magnitude smaller than observable by our most sensitive detectors. In essence, such a system behaves much like a black box where input and output can be controlled and determined; yet, the distribution of influx inside or the origin of the eflux eludes experimental inspection. Computer chemical simulations though can be of great help in this respect.44 However, correlating the observable energetic output with changes happening at the molecular level represents the ultimate goal and objective of supramolecular science.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

ITC in supramolecular chemistry

◦


G = −RT ln Kassoc ◦


S =

◦


H
G − T T

(1)

◦

(2)

∆G, ∆H, T∆S (kcal mol−1)

With enthalpy
H ◦ and Gibbs energy
G◦ at hand, the change in standard entropy
S ◦ is easily calculated from the Gibbs–Helmholtz equation (2). From a single calorimetric experiment at constant temperature, the main state functions
H,
G, and
S of the binding process are

accessible, if presumptions with respect to the singularity and stoichiometry of the process apply. Conducting such measurements at a range of temperatures furnishes the heat capacity
Cp (3). ◦


Cp =

d
H ◦ T d
S ◦ = dT dT

In principle,
Cp itself can be a function of temperature, but in the narrow span of temperatures of interest in supramolecular binding there is little risk to approximate
Cp by (4), where T1 and T2 denote two different absolute temperatures furnishing the respective standard enthalpies
H1◦ and
H2◦ . ◦


Cp =


H2◦ −
H1◦
S2◦ −
S1◦ = T2 − T 1 ln T2 − ln T1

(4)

The heat capacity
Cp occupies a pivotal position in understanding supramolecular interactions because it represents the temperature gradient of the energetic components composing the Gibbs enthalpy
G and thus allows calculating the affinity at various temperatures.45 Quite unlike most covalent bond formations in preparative chemistry, which are largely enthalpy-dominated and frequently possess vanishing heat capacities, all supramolecular processes come along with a substantial variation of the standard enthalpy with temperature. Associations, in general, feature negative
Cp values, which in the peculiar case of aqueous media can be correlated with the change in surface area buried from solvent on complexation. A typical temperature plot of the state functions is depicted in Figure 1.

30

1.4 × 109

20

1.2 × 109 1.0 × 109

10

8.0 × 108 0

6.0 × 108

−10

4.0 × 108

−20

2.0 × 108 10

(a)

(3)

Ka (M−1)

The role of calorimetry in this theater is to provide a reliable basis of experimental observables, leaving the interpretation in a molecular scenario open to the creativity and sceptical evaluation by the inquirer. Unlike other useful methods employed in supramolecular chemistry (e.g., mass spectroscopy, chemical force microscopy), calorimetry reports on ensembles averaged over time and individual energies of their members. This allows making use of thermodynamics for a full energetic characterization of the system under study. Heat as the primary observable in calorimetry is commonly measured at constant (atmospheric) pressure and thus represents an enthalpy change
H . If enthalpy is measured in response to a change in total composition of the system, the output depends on the extent of complex formation between the components, permitting access to molecular affinity. In the simplest case, just one 1 : 1 stoichiometric binding process dominates the molecular bond rearrangement, allowing the description of the entire process by the corresponding constant of mass action Kassoc . The affinity constant Kassoc relates to the Gibbs enthalpy of association
G◦ according to (1):

3

15

20 25 30 35 Temperature (°C)

40

10 (b)

15

20 25 30 35 Temperature (°C)

40

Figure 1 Temperature profiles of the association constant Kassoc (dash–dotted line), the enthalpy
H ◦ (dashed line), the entropy as T
S ◦ (dotted line), and the resultant Gibbs energy
G◦ (solid line) for a host–guest complexation characterized by (a) Kassoc (298K) = 109 M−1 ,
H ◦ (298K) = +20.9 kJ mol−1 ,
Cp◦ = −210 J K−1 mol−1 ; (b) Kassoc (298K) = 109 M−1 ,
H ◦ (298K) = −20.9 kJ mol−1 ,
Cp◦ = −210 J K−1 mol−1 . Both enthalpy and entropy possess substantial temperature dependencies, yet
G◦ does not change by more than 8 kJ over the range 10–40 ◦ C, rendering the estimation of binding enthalpies or heat capacity changes by noncalorimetric methods quite problematic. (Reproduced from Ref. 16.  Elsevier, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

4

Techniques

The two panels refer to supramolecular complexations having endothermic (a) or exothermic (b) enthalpies at room temperature. Either case is characterized by the same constant and negative heat capacity (
Cp◦ = −210 J mol−1 K−1 ), causing the enthalpy
H to change sign in the temperature span investigated. At
H ◦ = 0, the complexation is solely entropy-driven and the association constant is at a maximum at this point. From the diagram, it is obvious that
H ◦ and the entropic component T
S ◦ compensate each other, leading to a very flat temperature dependence of the Gibbs enthalpy
G◦ . Such behavior is a form of enthalpy–entropy compensation that appears as an intrinsic property of weak (i.e., supramolecular) interactions.46, 47 The similarity in slopes of the enthalpy and entropy components implies that the change in heat capacity
Cp◦ which defines the temperature gradients according to (3) and (5) must be much greater than the entropy
S. dT
S d
H =
Cp =
Cp +
S dT dT d
H dT
S ≈ only, if
Cp 
S dT dT
H (Tr ) − Tr
Cp Kassoc ln = Kassoc (Tr ) R

Cp 1 1 T − × + ln Tr T R Tr

(5)

systems with a goodness of fit (i.e., with binding strength) that is often derived and supported by energy-minimized structures delivered by molecular mechanics. Irrespective of the consideration of solvent influence, which adds another level of complexity, such comparisons are bound to lead to erroneous conclusions because the contribution of the entropy component is excluded a priori. In some instances this may be justifiable; however, as a rule, the omission of the entropic influence means an amputation of an essential property that distinguishes supramolecular from covalently connected systems. Because of weaker structural definition of the former, the energetic signature represents a welcome marker for additional characterization. Two host–guest systems having identical affinity (
G◦assoc ) may greatly differ in composition of the enthalpic and entropic contribution and thus possess widely different suitability for certain supramolecular functions. The deeper understanding of structure and function in supramolecular complexations mandates the qualitative and quantitative appreciation of all thermodynamic state functions. ITC currently is the most accurate, sensitive, fast, and convenient technique to obtain these quantities.

3

CALORIMETRIC DATA ACQUISITION

(6)

Consequently, there is no justification for disregarding this quantity as is frequently executed in the spectroscopic determination of the binding enthalpy using rudimentary   1
S van’t Hoff relationships e.g., ln Kassoc = −
H R T + R . Instead, a formula (6) must be used for calculation that honors a nonzero heat capacity
Cp◦ and allows the derivation of
H ◦ (Tr ) from the affinity constant Kassoc (Tr ) at the reference temperature Tr and a series of binding constants Kassoc obtained at various temperatures T by a nonlinear fitting process. The substantial discrepancies found between association enthalpies
H ◦ obtained by direct calorimetry or via the temperature dependence of the association constant using a van’t Hoff treatment most likely arise from lack of precision in the original measurements eventually combined with inadequate data evaluation.48, 49 Because calorimetry is the only method yielding enthalpies as direct experimental observables, the values derived on this basis appear more credible than from any other alternative method. Another result apparent from Figure 1 is the lack of correlation between the change in enthalpy
H and Gibbs energy
G. Again, this disparity emerges as a consequence of enthalpy–entropy compensation, yet it seems fair to state a severe neglect of appreciation of this fact in supramolecular chemistry. Common customs try to correlate measured affinities, for example, in abiotic host–guest

The measurement of heat conveniently takes temperature as an indicator. In modern calorimeters, the instrumental design employs two principal approaches that are easily distinguished with respect to the effect on the temperature output: In the adiabatic mode, heat evolution or consumption by the chemical process under investigation leads to a permanent increase or decrease of temperature. The extent of the change depends on the heat capacity of the instrument, which must be calibrated in separate experiments. Alternatively, the heat effect may be dissipated to a heat sink, so that the measurement following an initial perturbation falls back on a constant temperature baseline (isothermal mode). The heat flow may be observed directly using a pile of thermocouples (heat conduction device) or may be actively regulated to maintain a fixed level (power compensation). Both instrumental designs reach sensitivities in the nanowatt regime by means of a differential measurement relative to an internal reference. The schematic blueprint of a power compensation titration calorimeter is shown in Figure 2. Two coin-shaped identical cells (each holding about 1.4 ml) are permanently seated in an insulated compartment typically regulated 5–10◦ above the environmental temperature to allow a cooling heat flow. Both cells are completely filled; the reference cell contains the pure solvent and the measurement cell is filled with the solution of one partner of the binding reaction to be studied. The other reactant, usually prepared in 10–20-fold higher

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

ITC in supramolecular chemistry

Syringe holding the guest Mixing device

50 Cell feedback current 40

Sample cell containing host

[µcal sec−1]

Reference cell containing solvent

5

30 20 10 0

∆T

Sample cell reference cell power feedback

−10 0 (b)

20 40 60 80 100 120 140 160 180 200 Time (min)

(a)

Figure 2 (a) schematically depicts the instrumental setup of a power compensation calorimeter. Both cells are completely filled, the reference cell with pure solvent, the sample cell with a solution of one of the host–guest partners (e.g., the host). On addition of microliter aliquots of the guest solution delivered from the computer-driven syringe, a heat effect occurs that is counter-regulated by the cell feedback current to maintain
T at zero. (b) illustrates the data output consisting in a number of heat pulses that decrease in magnitude following the progressive saturation of the host binding site by the incremental addition of the guest species. (Reproduced from Ref. 27.  Wiley-VCH, 2007.)

concentration, is delivered from a syringe that is coaxially inserted through the long access tube. The tip of the syringe is deformed to a paddle to allow rapid mixing of the cell contents when the syringe device is rotated at a stirring rate of 200–400 min−1 . The reference cell is continuously heated to set a temperature difference of about 0.01◦ over the nominal temperature of the insulating jacket. A similar electrical power heater is attached to the sample cell and is automatically regulated by a feedback mechanism to minimize the temperature difference
T between the cells. On injecting aliquots of several microliters from the syringe, the association of the binding partners produces a heat effect that raises or lowers the temperature in the sample cell. The deflection of temperature triggers the feedback regulator to adjust the electrical power needed to maintain identical temperatures in both cells. The change in the respective feedback current is the primary signal observed and corresponds to a heat pulse (heat production over time). Integration with respect to time gives the energy that was traded on injecting the known amount of the reaction partner to the sample cell. If a series of injections are made, the compound in the cell is progressively converted to the molecular complex, leading to diminishing heat effects as the association approaches completion. A typical output picture showing the exothermic encapsulation of benzoate into the bistriazolo-strapped calixpyrrole 1 in acetonitrile at 303 K is depicted in Figure 3.50 The upper panel shows downward directed pulses, indicating the diminution of the feedback current necessary to keep a zero temperature difference to the reference cell as the heat from the exothermic association reaction makes up for the rest. The integration of the heat pulses when

plotted versus the nominal molar ratio of the injected compound over the one contained in the cell yields a titration curve that exhibits a characteristic shape. In the case shown in Figure 3, the sigmoidal appearance reflects the adequate choice of absolute concentration relations to allow the determination of the molar enthalpy
H ◦ from the extrapolated step height of the curve, the stoichiometry n of the binding process from the position of the inflection point along the molar ratio axis (tetraethylammonium benzoate (TEA)/1), and the affinity constant Kassoc from the slope in the inflection point. Modern calorimeters offer the comfort to have these quantities determined by software routines that use nonlinear curve-fitting to find the most probable parameters describing the supramolecular association with regard to the specifics of the instrument (cell volume, volume, displacement, etc.). Of course, a decisive prerequisite in any meaningful evaluation of calorimetric data is the judicious choice of experimental conditions, the appropriate correction of the data with respect to ubiquitous nonspecific contributions like the heat of dilution or mixing, and, above all, the adequate choice of a model representing the relevant processes in solution. Similar to many other data evaluations where several individual contributions combine to generate a singular output (as, e.g., in kinetics), adherence of the experimental data to a certain model does not ultimately prove the model but surely disproves all nonfitting alternatives. The various options are discussed further below. At this point, we shall focus the attention on the production of good quality data that in the end will be all decisive on the success of interpretative attempts.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

6

Techniques Time (min) N

N H3C CH3

N N

Et4N O

NH HN H 3C

NH HN

CH3

O

CFB (µcal sec−1)

N N

−10 0 10 20 30 40 50 60 70 80 0 −5 −10

∆H (kcal mol−1)

−15 0 H3C CH3

1

Molecular complex

−2 −4 −6 −8 −10 −12 0.0

0.5

1.0

1.5

2.0

Et4N+ benzoate /1

Figure 3 303 K.

ITC titration of 0.279 mM tetraethylammonium benzoate with 25 × 12 µl, 2.43 mM strapped calixpyrrole 1 in acetonitrile at

Paramount to the experimental setup is the purpose of the calorimetric titration: that is, whether the association constant and stoichiometry along with the standard enthalpy are to be determined or only the latter is the final goal. Such decision dictates the selection of the dimensionless c-value (7), which should lie within a span from 5 to 500 in order to render the titration curve sigmoidal. In this case, the step height between the asymptotic values at 0 and ∞ with respect to the molar ratio axis is read from a nonlinear least-square fit of the experimental data points and represents the standard enthalpy
H ◦ . The position of the inflection point in the sigmoidal titration curve defines the association stoichiometry, whereas the slope of the curve at this point translates as the association constant, which can be converted into the free energy
G◦ (1). c = n × [A]Kassoc

(7)

where [A] = concentration of the titrate compound in the cell [M], Kassoc = affinity constant, [M−1 ] and n = stoichiometric factor. In the case depicted in Figure 3, the c-value amounts to 15 and falls well within the range of 5–500, which displays clear sigmoidal curvature and is best suited for the calculation of
H, Kassoc , and n in a single experiment. Balancing the c-value frequently is a game, with bold restrictions emerging from instrumental sensitivity or unspecific interferences alike. If affinity is high (in artificial host–guest systems Kassoc > 106 M−1 ), the concentration required to place the c-value in the prospected

range may be too small and cause detection problems and insufficient signal-to-noise ratios especially when the molar enthalpies are not great either (|
H | ∼ 0–10 kJ mol−1 ). In Figure 4, the titration of a congener 2 of compound 1 (Figure 3) having a slightly smaller cavity with TEA chloride is shown corresponding to a c-value of over 1200. The paucity of data in the vicinity of the inflection point, in combination with the steep slope here, prevents a reliable derivation of Kassoc . On the contrary, if affinities are quite limited, high concentrations of the interacting compounds are needed, leading to saturation of the responsiveness of the instrument. In addition, this bears the risk of covering the effect of interest by an overwhelming unspecific background response. In many cases, the problems at either borderline can be relieved by adjusting the temperature, as modern calorimeters can be used between 0 and 80 ◦ C without extra equipment. The substantial change in heat capacity
Cp endemic in supramolecular interactions may easily shift the enthalpy and Gibbs enthalpy into the desirable range. In some instances, the precise estimation of the interaction enthalpy
H ◦ or the stoichiometry n rather than the affinity constant Kassoc is desired. Then, raising the c-value well over 1000 by an increase of the initial concentration is a beneficial option. The titration curve will then appear as a step (jump) function as in Figure 4 because the titrant added in aliquots from the syringe will be totally converted to the complex in each addition until the reaction partner in the cell is consumed completely. The subsequent injections will only show the spurious heats of dilution and mixing and thus will end in parallel to the molar ratio axis. The jump event marks the molar ratio of the components in the complex, and the step

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

ITC in supramolecular chemistry

7

Time (min) −10 0 10 20 30 40 50 60 70 80

H3C CH3

N N N

NH HN H3C

NH HN

Et4N Cl

+

CH3

H3C CH3 2

−5 −10 −15 2

−

0 ∆H (kcal mol−1)

N N N

CFB (µcal sec−1)

0

Molecular complex

−2 −4 −6 −8 −10 −12 −14

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Molar ratio

Figure 4 Titration of 2.9 mM bis-triazolo-calixpyrrole 2 (25 × 12 µl) into a 0.147 mM solution of tetraethylammonium chloride in acetonitrile at 303 K. The abrupt step appearance indicates a c-value of about 1200.

height gives the molar enthalpy referring to the compound delivered from the syringe. This procedure is more accurate than the previous case because no extrapolation (fitting) is required. The opposite situation is met when the concentration of the interaction partner contained in the calorimetric cell is well below the nominal dissociation constant of the binary complex. The shape of the titration plot then changes from sigmoidal into a monotonous curve, as is exemplified in Figure 5. In principle, the binding constant, molar heat, and stoichiometry can also be derived here in the same experimental run by curve-fitting procedures provided the binding saturation is extended to high levels. This may require excessive molar ratios of one binding partner over the other (up to 200),51 causing additional problems with unspecific interferences. Furthermore, small curvatures of the fitting functions may lead to parameter correlations that prevent finding the error minimum in the fit. If extra-calorimetric knowledge, for example, about the stoichiometry of the interaction, can be included in the analysis, fitting may converge much more readily using (n ×
H ◦ ) as a derived parameter.52 The fit function then needs to be adjusted accordingly. Since the c-value (7) determines the various regimes in the evaluation of ITC titrations, its correct setting is very important to minimize errors. An experimental study comparing the reports from different laboratories on ITC determinations (and ultracentrifugation

and surface-plasmon resonance analysis) of the same host–guest system found an impressive correlation of the error in Kassoc with the c-value, strongly suggesting the control of this parameter to the range of 20–100.53 A helpful visual tool to adjust the variable parameters is now provided as a spreadsheet.54 Of course, the empirical experimental approach is more reliable, since it is not confined to the ideal 1 : 1 binding model and may immediately reveal a more complex binding scheme. The meticulous error analysis of ITC titrations unfolded an unexpected discrepancy between the intrinsic statistical error in this method and the de facto experimental uncertainty of the thermodynamic parameters obtained. Employing optimal conditions, the statistical error level in the association constant or the molar enthalpy can be as small as 1% of the main uncertainty arising from the volume measurements.55–57 With respect to the volume delivery from the syringe, the aliquots titrated into the cell are not independent from one another, but accumulate the errors of the preceding additions resulting in a non-Gaussian error distribution. Thus, increasing the number of titration steps does not help in error reduction. Rather, on this basis the limitation to 10–15 injections is indicated. Similar arguments disclose the necessity to dramatically increase the guest to host molar ratio, if the host concentration applied is well below the anticipated dissociation constant. Even at a c-value as low as 0.1, the energetic parameters have been successfully derived when the titration was extended

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

8

Techniques O +

NH2

N

0

40

80

Time (min) 120 160

200

240

0 +

+

NH2

I

Na

N

O

−

−

I

N

H2N

+

+

3

4F

CFB (µcal sec−1)

O

N

−

O NH2 H2O

−10

−20

30 °C

−30

O

NH2 O

NH2

N

kcal mole−1 of injectant

+

+

−

N

I

N

O

H2N

+ +

N

O NH2

0

−1

−2

0.0

0.5

1.0 1.5 Molar ratio

2.0

2.5

Figure 5 ITC titration of the macrotricyclic cavity host 3 (8.3 mM) with sodium iodide (0.1 M) in water at 303 K. The c-value (7) in this case is around 1. (Reproduced from Ref. 27.  Wiley-VCH, 2007.)

to reach a final molar ratio RM in accord with the empirical formula given by (8)52, 58 : RM =

6.4 13 + 0.2 c c

(8)

The accuracy and precision as predicted by statistical theory are not nearly met in routine ITC experiments. However, using specially selected benchmark reactions and applying a Gaussian error approximation, the error of data-fitting can result in high precision of the derived association constants Kassoc (± 5%) and
H ◦ (± 1%).59, 60 In all practical cases, the repeatability between different runs may be considerably lower. Benchmarking studies on protein–ligand interactions in water53 or artificial host–guest binding in organic solvents61 involving different laboratories and calorimeter makes59 arrive at a much less optimistic reliability. Fair estimates assign the general error of experimental repeatability comparing the results of different laboratories to about 3–4 kJ each in free energy
G◦ and enthalpy
H ◦ , whereas the entropy
S ◦ as a derived quantity must be set

at 6–8 kJ.27, 62 Considering these limits, there is generally no point in discussing association constants differing by less than a factor 2–3! The origin for the less-than-optimal repeatability of ITC titrations can be traced to systematic errors in sample treatment as well as to difficulties in data evaluation. First in line of the factors that interfere with the reproducibility of calorimetric results is the purity of the compounds used. Owing to the ubiquity of heat effects, even small deviations from a nominal composition of a compound may result in dramatic differences in the calorimetric output. A frequent problem of this kind in abiotic host–guest binding is the presence of the solvent of crystallization. Ordinarily, this is not considered an impurity, as it is often present in a fixed stoichiometric ratio and can be accounted for in elemental analyses and in the spectroscopic evaluation. In calorimetry, however, solvent of crystallization adds a heat contribution of unknown size, which does not emerge from the interaction under study and thus tends to falsify the results. The problem is amplified by the polarity

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc024

ITC in supramolecular chemistry difference between the solvent of recrystallization and the one used in the supramolecular investigation. Thus, the worst case is met when compounds in a hydrated form are employed in nonpolar or aprotic solvents such as chloroform, dichloromethane, or acetone. Experience tells us that, even in polar solvents such as acetonitrile or dimethylsulfoxide (DMSO), the heat evolution on introduction of protic solvents such as water is strongly nonlinear with concentration. Moreover, many host–guest interactions in organic solvents involve hydrogen bonding and eventually respond in a nonmonotonous heat output on a gradual increase of the water content at low concentrations ( 0.5) dipole couplings in its REAPDOR (rotational-echo adiabatic-passage double-resonance) version.28 Beside the multitude of techniques based on magnetization transfer through the homonuclear dipolar coupling to obtain through-space 13 C chemical shift correlations, in the last decade sequences to determine through-bond carbon-13 connectivities have been developed. Often, the through-bond connectivies are necessary to establish unambiguous assignments of the NMR spectra, before throughspace interactions can be used to determine structure. The INADEQUATE experiment, well-known liquid-state technique to establish direct scalar connectivites in the 13 C skeleton, has been successfully applied in the solid state in its standard and refocused versions that yield antiphase and in-phase correlations, respectively.29 The calculation of chemical shifts using quantumchemical methods affords new insights in the extraction of structural and dynamics information allowing reliable assignments of the experimental data. Supramolecular systems present packing effects that can be tackled by a combined approach that combines experimental 1 H data

with quantum-chemical calculations Such a validation is obtained preferentially on systems that have known structures and a limited complexity. An example of comparison of experimental and computed 1 H NMR chemical shift spectra for alanine in the periodic lattice is reported in Figure 2.30 A sufficient agreement with the experimental data is found, especially for the hydrogen-bonded protons. In the static calculation, different NMR chemical shifts at 14.3, 6.4, and 5.6 ppm are expected for the three NH3 protons characterized by different geometry of HBs. In the real sample, the NH3 + group is rotating, resulting in a single NMR line for the amino protons at 8.6 ppm. About the same value (8.8 ppm) is obtained when averaging the calculated chemical shifts. In the case of quadrupolar interaction the signals appear severely distorted and often the center of gravity is changed. A combination of pulse schemes (multiple quantum magic angle spinning (MQMAS)) allows the separation of the chemical shift and the quadrupolar coupling.31 In the last decades many other 1D and 2D pulse sequences have been proposed that enable scientists to obtain structure and dynamic information about supramolecular systems that were previously very hard to study with conventional X-ray crystallography and neutron scattering.7 This multiparameter, multinuclear approach represents a formidable tool for structural investigation since each NMR signal can be univocally assigned to the respective nucleus in the precise environment as a report at molecular level

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Solid-state NMR studies on supramolecular chemistry

5

1A

P3A

1

O11A GC

2

P1A

3

5 4

6

O33A GE

15

3

4 2

O11B

25

1

P3B

20

5

P1B GD

30

30 (a)

1B

(b)

25

20

(ppm)

6

O33B

15

(ppm)

Figure 3 (a) X-ray structure of the sample [CH6 N3 ]3 [C9 H13 O9 P3 ][C9 H14 O9 P3 ]. (b) 1 H decoupled 31 P 2D through-space SQ–SQ (single quantum–single quantum) spectrum of mixture of single crystals and microcrystal of compound [CH6 N3 ]3 [C9 H13 O9 P3 ][C9 H14 O9 P3 ] obtained by a shearing transformation of the experimental SQ–DQ spectrum. (Reproduced from Ref. 32.  Royal Society of Chemistry, 2004.)

of the proximities and/or the nature of the strong and weak interaction involved. Of course almost all the nuclei of the periodic table are NMR active and then complementary data can be combined by a multinuclear study. For example, the very sensitive 31 P nucleus represents a valuable method for detecting supramolecular arrangements in phosphorus containing systems. The supramolecular structure obtained by cocrystallization of benzene1,3,5-tris(methylenephosphonic acid) [C9 H15 O9 P3 ] with guanidinium chloride [CH6 N3 ]Cl, yields the compound of formula that possesses six crystallographically inequivalent phosphorus atoms, and then six different 31 P peaks in the 31 P SSNMR spectrum.32 Their assignment has been performed by through-space single quantum–double quantum (SQ–DQ) correlation experiments (Figure 3) leading to two possible assignments of the six crystallographic P sites to the corresponding resonances in the 1D31 P MAS NMR spectrum.

3

STRUCTURE CHARACTERIZATION

Supramolecular chemistry utilizes weak and reversible noncovalent interactions, such as HB and aromatic π –π interactions, metal coordination, hydrophobic and van der Waals forces, and/or electrostatic effects to assemble molecules

into multimolecular complexes.33 Therefore, a complete understanding of the supramolecular forces (preferred geometries, competitive bonds, strength, and recognition pattern) is a prerequisite for a rational design of desired solid architectures.

3.1

The hydrogen bond

The most powerful organizing force in molecular assembly is certainly represented by the strong, highly selective, and directional HBs. Several techniques have been used for the HB detection and characterization, but due to their intrinsic limitation, methods of choice are mainly IR and NMR spectroscopy. Since in the solid-state signals are not averaged by solvent effects or by rapid exchange processes present in solution, the SSNMR approach allows an accurate evaluation of the HB local environment and strength. An increasing number of 1 H studies34 is focused on the direct chemical shift measurement of protons in HB obtained at very high spinning speed (up to 70 kHz). In particular, the proton is increasingly deshielded with increasing the HB strength, which leads to 1 H high frequency shifts far from aliphatic and aromatic signals (Scheme 1).35 The main feature of this technique is that the magnitude of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

6

Techniques Aromatics Aldehydes

Alkines

Alkenes

Alkanes

Hydrogen bond 20

10

15 Strong 20

18

16

5

ppm

Weak 14

12

10

8

Scheme 1 1 H chemical shift scale with strong and weak HBs highlighted. (Reproduced from Ref. 35.  Royal Society of Chemistry, 2008.)

shift is directly correlated with the length and the strength of the HB.36 Moreover, the 1 H chemical shift is very sensitive to the location of the hydrogen atom in these interactions. Direct relationships between δ 1 H and HB strength and between δ 1 H and X–H distance for different classes of hydrogenbonded compounds have been reported.37 Thus, NMR shift data can be used to detect and to estimate the presence and the strength of HBs (Scheme 1). For instance, the HB strength in adducts between dicarboxylic acids and diamine 1,4-diazabicyclo–[2.2.2]octane (DABCO) have been evaluated as a function of the acid chain lengh, that is, of the pKa . For these compounds it has been shown that intra- and intermolecular O–H· · ·O or N· · ·H–O HBs are strong interactions with proton chemical shifts of around 16 ± 1.5 ppm, while weaker N+ –H· · ·O− interactions are characterized by a δ 1 H of about 12.3 ppm.1, 37

The 1 H spectrum of the adduct between malonic acid and DABCO, reported in Figure 4(a), clearly shows in the high frequency region two well defined HB signals at 17.6 and 12.4 ppm that can be assigned to the O–H· · ·O and the N+ –H· · ·O− protons, respectively (Figure 4b).37 Owing to the intrinsic difficulty of X-ray techniques in detecting the hydrogen atom position, an effective method that combines 1 H fast MAS NMR and density functional theory (DFT) calculation for the determination of the O–H distance has been proposed.38 It consists in refining the hydrogen atom position within the HB, in which the computed 1 H δ are reported as a function of the O–H distance while all other atoms are retained at their positions. Using the experimental δ value of 16.5 ppm (Figure 5a), the plot for the succinic acid-DABCO adduct ˚ which is (Figure 5b) gives an O–H distance of 1.043 A, in good agreement with the value obtained by geometry ˚ 37 optimization (1.039 A). 1 The H chemical shift is not the only NMR parameter that can provide insights in HB interactions but also the 13 C CSA and the 15 N chemical shift are often used. Nuclear 13 C-NMR shielding tensors of carboxylic groups significantly change with the protonation state of the group, that is, carboxylic or carboxylate form.39, 40 For example, the carbon chemical shift tensors of the COOH signal obtained by sideband analyses of low speed spinning spectra in supramolecular adducts [N(CH2 CH2 )3 N]-H[OOC(CH2 )n COOH] (n = 1–7)41 afforded information on the character of the COOH group. δ 33 is usually not very sensitive to the protonation state of the carboxylic group, whereas δ iso increases in shielding upon protonation, but unfortunately the information is limited by the fact that δ 11

N+ H O− O H O

(ii)

(b) (i) 20 (a)

10

0

−10

dH /ppm

Figure 4 (a) 1 H MAS NMR spectrum (ωr = 35 kHz) of the adduct between malonic acid and DABCO obtained at 500 MHz (i) and at 300 MHz (ii). The different shape of the N+ –H· · ·O− signal is due to second order quadrupolar effect of the 14 N nucleus on the 1 H spectrum that is magnetic field dependent. (b) Single-crystal X-ray structure of the malonic acid-DABCO adduct. (Reproduced from Ref. 37.  American Chemical Society, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Solid-state NMR studies on supramolecular chemistry

7

Calculated 1H diso (ppm)

25

N H O

20 15 10 5 0 −5

30 (a)

25

20

15

10 ppm

5

0.8

−5

0

(b)

1.0

1.2

1.4

1.6

1.8

2.0

O – H distance (Å)

d33

Chemical shift tensors components (ppm)

Figure 5 (a) 1 H MAS NMR spectrum (500 MHz, ωr = 35 kHz) of the adduct between succinic acid and DABCO. (b) Plot of the calculated 1 H chemical shift versus the O–H distance in which the experimental 1 H chemical shift has been used in order to obtain a reliable O–H distance. (Reproduced from Ref. 37.  American Chemical Society, 2005.)

d22 O

R C

d11 O

X

H

260 240 220 200 180 160 140 120 100 80

d11

d22

d33

0.00 0.02 0.04 0.06 0.08 0.10 0.12 (a)

(b)

R(C–O)–R(C O) distance (Å)

Figure 6 (a) Orientation of chemical shift tensors (δ 11 , δ 22 , and δ 33 ) in carboxylic groups. (b) Chemical shift tensors versus difference between C–O and C=O bond lenghts for dicarboxylic acid-DABCO adducts. (Reproduced from Ref. 41.  Wiley-VCH, 2003.)

and δ 22 change their values in opposite direction. δ 22 , the chemical shift tensor that lies perpendicular to the plane of symmetry of the C=O group, is the most diagnostic parameter that reflects the HB strength (Figure 6). On the other hand, 15 N chemical shifts are very sensitive to the protonation state of the nitrogen involved in the HB interaction due to the wider range of chemical shift with respect to 13 C: intermolecular HBs produce a high frequency or low frequency shift in the 15 N values according to the type of nitrogen atom and to the type of synthons involved (Scheme 2).42 The diagnostic value of the 15 N chemical shifts has been demonstrated by exploring the effects of acid–base interactions on the 15 N spectra in dry solid poly-L-lisine with different acids.43

Azobenzene

Pyridine

−160

−102

500

400

300

200

Scheme 2 Influence of HBs on the nitrogen-containing group.

15 N

Aniline

NH3

−4

+25

100

ppm

chemical shift of some

Combined experimental and computational (DFT) results reveal low-field shifts of the amino nitrogen upon interaction with HX acids (HX=HF, H2 SO4 , CF3 COOH, (CH3 )2 POOH H3 PO4 , and HNO3 ). 15 N chemical shifts are maxima when the hydrogen is located in the HB center and then decrease again upon full protonation (as found also for aqueous solution at low pH). The combined use of 1 H and 15 N chemical shift data allows a distinction to be made between N+ –H· · ·O− interactions (with proton transfer) and N· · ·H–O interactions (without proton transfer) and between strong and weak HBs. Correlation of the isotropic 15 N chemical shift with the internuclear distances of the heavier atoms (N–O distance) involved in the HB interaction has been ascertained in the adducts of formula [N(CH2 CH2 )3 N]-H-[OOC(CH2 )n COOH] (n = 1 − 7) (Figure 7).41 By comparing 13 C chemical shift and CSA and 15 N chemical shift data it has been demonstrated that it is possible to evaluate the cocrystal or salt character of an adduct. An other case is the supramolecular complex 1,10dipyridylferrocene-anthranilic acid {[Fe(η5 -C5 H4 -C5 H4 N)2][(C6 H4 )NH2 COOH]}2 . The comparison of the free pyridine nitrogen chemical shift with that of the pyridine nitrogen

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

8

Techniques

3.2

2.72 2.70

1C3

1C5

N – O distance (Å)

2.68 2.66 2.64 2.62

1C9

2.60 2.58

1C7

1C8 1C6 1C4 1C8

2.56 2.54 −10

−8

1C7 1C4 1C9

−6

−4

1C6

−2

Experimental

15N

0

2

4

6

diso (ppm)

Figure 7 Plot of the experimental 15 N chemical shift of the nitrogen atoms involved in HB versus the N–O distance for dicarboxylic acid-DABCO adducts. (Reproduced from Ref. 41.  Wiley-VCH, 2003.)

in the adduct resulted in the confirmation of the presence of a strong O–H· · ·N interaction with no proton transfer from the acid to the nitrogen base and of a weak N–H· · ·N interaction.44 In the HB characterization 1D spectra give information concerning the HB presence, the hydrogen-bonded atom position, and the strengh of the interaction, while 2D 1 H DQ MAS experiments provide elucidation of HB geometries and networks. This is the case of the three crystal forms of the cocrystal 4,4 -bipy/pimelic acid (bipy = bipyridine), [NH4 C5 C5 H4 N]-[HOOC(CH2 )5 COOH], where their relationship have been investigated by comparing single-crystal X-ray diffraction and 1 H DQ MAS SSNMR experiments.5 Xray diffraction supplied packing and conformation of the molecules while 1 H DQ MAS spectra provided new parameters for differencing the polymorphs and for elucidating the HB network. In another case rotor-synchronized 1 H DQ MAS spectra were used to elucidate the supramolecular structures adopted by two different alkyl-substituted benzoxazine dimers [N,N-bis(3,5-dimethyl-2-hydroxybenzyl) “R” amine], where “R” = methyl or ethyl.4 Figure 8 shows the 1D and 2D 1 H (500.1 MHz) MAS (ωr = 35 kHz) spectra of the methyl (solid line) and ethyl (dashed line) dimers together with their HB network. Of most importance are the clear differences between the two spectra in the HB region. The analysis of the DQ signals in the 2D 1 H DQ MAS experiments allowed to establish that the methyl and the ethyl compounds are arranged in dimers and chains, respectively, as elucidate in Figure 8(d) and (e).

Distances and constrains determined by solid-state NMR

As already said, many applications of SSNMR for structure determination rely upon the dependence of the dipolar 3 where rIS is the distance between the coupling on 1/rIS two spins I and S. All these techniques provide distances which compare very well with those obtained by X-ray diffraction (XRD) if the samples contain isolated spin pairs (typically ±0.05 nm and often better). Spiess, Schnell, and coworkers45 were able to demonstrate the potential of the combined use of advanced SSNMR pulse sequences, (HDOR, heteronuclear dipolar-order rotor encoding and REREDOR, rotor-encoded REDOR), and quantum-chemical calculations for investigating the multiple HB network of N-butylaminocarboxyl6-tridecylisocitosine in the pyrimidone and pyrimidinol forms (Scheme 3). By performing solid-state 1 H– 15 N dipolar recoupling experiments, the authors were able to providing N–H distances of up to about 250 pm with an accuracy level of ±1 pm for short distances (around 100 pm) and ±5 pm for longer distances (180–250 pm). Also vibrational effects have been taken into account and the zero-point vibrations were found to enlarge the apparent distance determined for a typical N–H bond (103.5 pm) by about 3 pm. However, in most cases, where multiple spin systems are present and the number of spins and/or their geometrical arrangement is completely unknown is highly unlikely that reliable distances can be directly obtained from REDOR measurements. Furthermore the technique is complicated by rapid motion of the molecular structure. This is the case, for example, of the distance evaluation of the intermolecular distance between host and guest molecular components in supramolecular compounds such as p-tertbutylcalix(4)arene fluorobenzene where the 13 C NMR signal is modulated by heteronuclear dipolar interaction with the 19 F containing guest in redox experiments.46 A recent review from Brown22 elegantly collects some examples of the use of SEDOR (spin-echo double resonance), REDOR, and REAPDOR techniques for the determination of 13 C– 17 O and 15 N– 17 O dipolar couplings.47 However, these represent rare cases since for the oxygen atom there is only one NMR active isotope, oxygen17, which has a natural abundance of 0.038% and is a spin I = 5/2. Thus its study absolutely requires isotopic enrichment. Concerning the opportunities offered by constrain measurements through 2D SSNMR it is worth spending some words about their application in the biological fields. Indeed the number of articles devoted to the supramolecular structure elucidation by means of SSNMR of biocompounds is increasing.48 Like its liquid-state analogue, SSNMR

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Solid-state NMR studies on supramolecular chemistry

15

10

5

(a)

0

ppm A

A

BC

CC AD AB

B C

10

BD

10

15

AD

15

20

AC AB

20

25

(b)

15

10

H

CH3

H3C H3C

CH3

H3C O

O

(d)

N CH3

CH3

H3C O

(e)

H H N

CH3

CH2

O

H CH3

H

O H N

H

5 CH3

O H3C

O H

10

H3C

H

H 3C

15

CH3

N O

25

(c)

5

CH3

H3C

H3C

9

H 2C

CH3 CH3

Figure 8 (a) 1 H MAS spectra (500 MHz, ωr = 35 kHz) of the methyl (solid line) and ethyl (dashed line) benzoxazine compounds. (b, c) HB regions of 1 H DQ MAS spectra of the methyl (b) and ethyl (c) b compounds. (d, e) Schematic arrangement of the dimer (methyl) and chain-like (ethyl) structure. (Reproduced from Ref. 4.  American Chemical Society, 1998.)

spectroscopy does not automatically provide long-range structure-symmetry information as known from X-ray crystallography. As a consequence, large amounts of data on several differently labeled samples have typically been required to obtain sufficient long-range information. These requirements in some way have limited its diffusion. However, it has been demonstrate49 that it is possibile to identify the supramolecular conformation of fibrils directly from symmetry-induced resonance patterns in 2D heteronuclear 15 N– 13 C OC NCO and 15 N– 13 C OC NCA (superscript OC designates experiments50 designed using optimal control theory) correlation and homonuclear 13 C – 13 Cα DARR51 correlation SSNMR spectra for a single 13 C,15 N-labeled hIAPP20–29 (SNNFGAILSS) decapeptide from the human

islet amyloid polypeptide (hIAPP), which is believed to form the fibrillation core domain of fibrils in the pancreas of type 2 diabetes patients. Again, several types of oneand two-dimensional SSNMR techniques have been used to obtain constraints on the peptide conformation and supramolecular structure in amylin fibrils, a 37-residue peptide also called islet amyloid polypeptide or IAPP, and to derive molecular structural models that are consistent with the experimental data.52 SSNMR measurements on a series of isotopically labeled samples indicate a single molecular structure within the striated ribbons which contains four layers of parallel β-sheets, formed by two symmetric layers of amylin molecules.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

10

Techniques

R1

R1

H1 O

2

N3

3

H

O

O

H2

H3

N3

N

2

N

N3

9

8 CN

10

Ha

N2

O

H2

H

O

N2

28

N3 1

N

N H3

H

R2

1: R1 = n -C13H27, R2 = n -C4H9

2: R1 = n -C13H27, R2 = n -C4H9

1′: R = CH3, R = CH3

2′: R1 = CH3, R2 = CH3 (b)

Scheme 3 Schematic representation with HB network of the N -butylaminocarboxyl-6-tridecylisocitosine in the pyrimidone (a) and pyrimidinol (b) forms. (Reproduced from Ref. 45.  Wiley-VCH, 2005.)

3.3 π –π interactions Although much less strong than hydrogen bonding, aromatic π –π interactions and ring-current effects associated with the presence of aromatic moieties represent an other important source of structural information. In the last two decades π-stacking interactions between aromatic systems have been extensively studied. This interaction has been proposed to be a pivotal assembly force in many important supramolecular systems including various protein and enzyme substrates,53 nucleic acids,54 and molecular catalysts.55 A deeper understanding of the nature of the interaction is the key step for developing the synthesis of controlled nanoscale structures. Intra- or intermolecular ring currents due to nearby through-space aromatic moieties produce variation of the NMR chemical shift in “exposed” part of the molecule. Although all the nuclei are sensitive to this kind of effects, the main application are found for high-resolution 1 H solid-state spectra since a more pronounced effect can be visualized on the small chemical shift range of the proton nucleus. A clear example of ring-current effects in 1 H MAS SSNMR has been reported by Brown, Spiess, and coworkers56 in the study of the host–guest interaction between naphthalene-spaced tweezer as host molecule and 1,4-dicyanobenzene, as aromatic, electron-deficient guest (Figure 9). In CDCl3 solution formation and dissociation processes occur in the NMR timescale at room temperature and the aromatic protons are shown as a single resonance in the

26

22 7

NC

13

15 14

Ha 2 2 3

23 6

Hb

16 12

O

R1

(a)

Hb

R2

R1 2

11 17

H1

1

O H1

1

21

1

H N1

H2

H2

H

20

18 N1

N1

O

H N

R2 N

3

R

H N2

27 19

24 1

25 5

4

Figure 9 Schematic representation of the naphthalene-spaced tweezer-1,4-dicyanobenzene host–guest complex. (Reproduced from Ref. 56.  Wiley-VCH, 2001.) 1

H solution NMR spectrum, shifted by 4.35 ppm relative to that observed for the guest molecule alone. X-ray structure shows that the two guest aromatic protons Ha and Hb are differently involved in the ring currents due to the host molecule, with a distance of Hb to the center of inner benzene ring of only 260 pm. SSNMR investigation reveals that the guest remains complexed on the time scale of the NMR experiment. Accordingly a large difference in the proton chemical shift (Ha = 5.6 ppm, Hb = 2.0 ppm) have been detected as shown in the rotor syncronized 1 H DQ MAS NMR spectrum (Figure 10). Such splitting, that is not present in solution, is a strong indication of ring-current effects. All the NMR data are strongly supported by combining a theoretical approach by performing quantumchemical calculations to determine the structure and NMR chemical shifts of the host–guest complex. The approach presented by these authors clearly exploits the sensitivity of 1 H chemical shift to aromatic ring currents and can be in principle applied to crystalline and amorphous systems. A interesting example of the importance of ring-current effects in 1 H SSNMR is provided by the polycyclic aromatic molecule, hexa-n-dodecylhexa dodecylhexa-perihexabenzocoronene.57 By means of the 1 H MAS spectrum, in which three distinct aromatic resonances are identified (Figure 11a), it has been shown that hexa-ndodecylhexa dodecylhexa-peri-hexabenzocoronene forms a columnar mesophase with a very high one-dimensional charge carrier mobility. The observation of these distinct aromatic resonances is explained in terms of the differing degrees to which the aromatic protons experience the ring current of adjacent layers. Using the rotor-synchronized DQ MAS method (Figure 11b), definite proton–proton proximities are identified, which are shown to be in agreement with the known crystal structure of unsubstituted hexabenzocoronenes (Figure 12).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Solid-state NMR studies on supramolecular chemistry

11

1ar I 2a 1arII 1al 2b

×5

−4

4

2a 2b 1al 1al 1arII 1ar II 1arI 1ar II 1arI 1ar I

dDQ

0

8

16

14

12

10

8

(a)

6 4 ppm

2

0 −2 −4

−6

A B C

12 16 20

−5

8

(a)

6

4

2

0

−2 0

dSQ 2 a2 b

1ar(1) 1ar(2) 1ar(3)

5 10

0

CC 15

AB

2b

Double quantum

10

2

4 1arII 2a

d (1H)

20

6 1arI 8

10 132 (b)

128

124

120

10

8

6 4 2 0 Single quantum

− 2 −4 ppm

Figure 11 (a) 1 H MAS (500 MHz, ωr = 35 kHz) spectrum and (b) rotor-synchronized 1 H DQ MAS NMR spectrum of hexa-ndodecylhexa dodecylhexa-peri-hexabenzocoronene. (Reproduced from Ref. 57.  American Chemical Society, 1999.)

116

d (13C)

Figure 10 (a) Rotor-synchronized 1 H DQ MAS spectrum (700 MHz, ωr = 30 kHz) of the naphthalene-spaced tweezer-1,4dicyanobenzene host–guest complex. (b) 1 H– 13 C REPT-HSQC NMR correlation spectrum of the naphthalene-spaced tweezer1,4-dicyanobenzene host–guest complex. The notation 1ar and 1al refers to host aromatic and alkyl protons, respectively, while 2a and 2b to the two distinct guest aromatic protons. (Reproduced from Ref. 56.  Wiley-VCH, 2001.)

4

12 (b)

DYNAMICS IN SUPRAMOLECULAR SYSTEMS

The possibility of dynamic behavior for single groups or entire molecules in the solid state when weak noncovalent interatomic forces are present is quite general and this phenomenon has a pronounced influence on the macroscopic physical properties of the supramolecular systems.22

This is typical, for example, of host–guest adducts, where lower activation energy barriers are expected for the motion of molecules within the cavity or the channel of a matrix. These motions can cover a wide range of correlation times from very fast processes that occur in nanoseconds down to very slow motions of the order of seconds.58 Since the aim of the dynamic investigation is the quantification of the time scale of the motion and the geometric interpretation of the molecular process, the type and the correlation time of dynamics determine the choice of experiments and techniques. The SSNMR represents a powerful tool for the investigation of molecular dynamics due to the possibility offered by the use of different parameters to cover a wide timescale of a fluxional process from 102 to 10−10 s. When the motion occurs it determines variations in the NMR spectra that are dependent from the strength of the interaction tensors, the correlation time of the motion and the orientation of the interaction tensors

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

12

Techniques Chemical shift

Ig TC

HA

Reactions 2

HB HC

Quadrupole

Slow 2D-exchange 0

Slow spin alignment

T1D

−2

T1Q

−4

T1r

Int.

T2

MAS broadening Int.

−6 −8

T1 Anisotropy

T1 Fast

Fast

−10 Relaxation time analysis

Lineshape analysis

Scheme 4 Schematic representation of the correlation time range covered by the NMR parameters.

Figure 12 Representation of the proposed stacking of the aromatic cores in hexa-n-dodecylhexa dodecylhexa-perihexabenzocoronene. Three molecules are shown; the molecules above and below the central molecule are indicated by dashed and dotted lines, respectively. (Reproduced from Ref. 57.  American Chemical Society, 1999.)

with respect to the external magnetic field. Several SSNMR experiments in low resolution (wideline) and high resolution (lineshape analysis at VT, CSA analysis, 2D-EXSY, relaxation time studies, etc.) are able for detecting and quantifying molecular motions involving individual groups or entire molecules. Spin-lattice relaxation time (T1 ) measurements obtained by inversion recovery or saturation recovery59 have been used to investigate dynamic processes of the order of microseconds, whereas slower motions can be tackled by relaxation time in the rotating frame, T1ρ .60 In this case relaxation measurements are related to the spin lock field of the order of tens of kilohertz, allowing the extension of dynamic studies to lower frequencies. Alternatively by means of CP pulses, dynamic information can be achieved by heteronuclear T1ρ measurements.61 Further available methods are the measurement of relaxation parameters such as heteronuclear T1 , heteronuclear T1ρ , X–H crossrelaxation time and proton relaxation time in the dipolar state (T1D ).62 In Scheme 4 the relationship between the correlation time for a molecular motion and NMR parameters is reported. Deuteron NMR has been long known as a powerful tool for probing molecular dynamics.63

Isotopic substitution in a supramolecular system enables site-selective investigation of dynamics. For example, substitution of protons atoms with deuterium atoms allows the use of VT low resolution 2 D NMR investigation where the motionally induced reorientation of the 2 H quadrupolar tensor will affect intensities and linewidths of the 2 D spectra. Deuteron NMR line shapes and relaxation rates are usually dominated by the interaction of the nuclear electric quadrupole moment with the electric field gradient at the site of the nucleus. The deuteron has a relatively small electric quadrupole moment, which makes it easy to work with experimentally. Available deuteron NMR techniques cover a very broad range of time scales, from pico- to milliseconds.64 The basic formula describing the angular dependent quadrupolar coupling is given by ω = ωL ± 0.5 ωQ (3cos2 θ − 1)ηsin2 θcos2 φ where ωL is the Larmor frequency and ωQ represents the strength of the quadrupolar coupling. The asymmetry parameter η describes the deviation from axial symmetry while the angles θ and φ are the polar angles of the magnetic field Bo in the principal axes system of the quadrupolar coupling tensor. For rigid C–H bonds, its unique axis is along the C–H direction. In a powder, the maximum frequency splitting is then 2 ωQ . However, in the presence of rapid molecular motion, the quadrupolar interaction can be partially averaged, thus yielding an averaged quadrupolar coupling tensor. Computer fitting of the lineshape based on an assumed model for the dynamic process affords the types as well as the rate constant k extracted over a range of temperature, allowing the calculation of the activation energy. Though in difficult cases experimental lineshapes may be

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Solid-state NMR studies on supramolecular chemistry reproduced by several distinct models,65 2 H NMR still provides a fairly easy access to local geometries of mobile species. In the case of MAS conditions, the 2 H spectrum has a sufficient resolution to resolve distinct deuteron sites with different chemical shifts and the SSB pattern reflects the envelope of the static quadrupolar lineshapes. Guest molecules included into cavities created by host molecules such as cyclodextrin (CD), zeolites, alluminosilicates, and so on are often involved in motions with relatively low activation energies. The mobility in a solidstate inclusion environment represents a different dynamic regime in comparison with both solution and crystalline situations. In several cases the rotation of the entire molecule occurs only along certain axis. The shape of the guest, the size of the host and the strength of the host–guest intermolecular forces play a fundamental role in determining the kind of process involved and the regime of the motion. A typical example of dynamics in supramolecular systems has been reported by Garcia-Garibay and coworkers for MOF-5 that has a cubic framework composed of Zn4 O clusters (vertices) bridged by 1–4 phenyldicarboxylates (Figure 13).66 Dynamics of phenylene deuterated d4 -MOF-5 have been determined using VT 2 H NMR. Spectra at 300 K shows deuterons in the slow exchange regime, whereas the spectra recorded between 363 and 435 K are related to the intermediate regime of motion (Figure 14). By fitting the experimental data with a twofold flipping model (180◦ rotation), the Arrhenius analysis of the exchange rate affords an activation energy of 11.3 ± 2.0 Kcal mol−1 .

R2

12 MHz

411

5.0

387

2.6

363

0.9

300

0.001

−100

0 kHz

100

−100

0

100

kHz

Figure 14 Experimental (left) and calculated (right) quadrupolar echo solid state 2 H NMR of phenylene deuterated d4 -MOF-5 sample sealed at 3 mTorr. (Reproduced from Ref. 66.  American Chemical Society, 2008.)

Solid-state guest dynamics of dimeric capsules of tetratolyl urea calix[4]arene filled with different aromatic guests such as benzene-d6 , fluorobenzene-d5 and 1,4difluorobenzene were studied. Upon inclusion, all guest moieties revealed complexation-induced shifts varying from 2.8 to 5.1 ppm. All guest molecules were shown to undergo

R2 O

R1 R2 O

R1

O

R1

R1 R1 R2 R1

435 K

13

O

O

R2 R2

O

1a: R1 = H, R2 = H 1b: R1 = OCH3, R2 = H 1c: R1 = t - Bu, R2 = t - Bu

2

Figure 13 Left and center: Structures that emulate macroscopic toy gyroscopes with open (1a–c) and triply bridged topologies (2). Right: a fragment of the structure of MOF-5. (Reproduced from Ref. 66.  American Chemical Society, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

14

Techniques

distinct motions, ranging from simple C6 -rotations of benzene-d6 at low temperatures to rather ill-defined 180◦ phenyl flips of fluorobenzene-d5 .67 The large CSA associated with the carbonyl resonances in the 13 C cross polarization magic angle spinning (CPMAS) spectrum at room temperature of crystalline Cr(CO)3 (η6 -C6 H6 ) indicates that the carbonyl moieties are rigid due to crystalline packing constraints. Conversely 13 C carbonyl and deuterium data of Cr(CO)3 (η6 -C6 H6 ) included in β-CD selectively enriched in 13 C (carbonyl ligand) and 2 D (arene ligand) respectively,68 show that a fast motion of the guest molecule is present. The entire molecule does not rotate isotropically inside the cavity, but with a C6 rotation around the main molecular axis of symmetry. Interestingly the VT 13 C SSNMR spectra of (η5 -C5 H5 )2 Mo2 (CO)6 included in γ -CD can only be interpreted by fast motion of one half of the dimer inside the CD cavity, whereas the nonincluded (η5 -C5 H5 )Mo(CO)3 part is rigid due to strong interatomic interaction with neighboring molecules.69 Fast motion of the (η5 -C5 H5 )Mo(CO)3 part inside the hydrophobic cavity is explained by the weaker constrain forces.

5

CONCLUSIONS

In this chapter we demonstrated how SSNMR spectroscopy can contribute to understanding the structure and properties of supramolecular architectures. This topic is a very broad area with an extensive literature, and for this reason and the limited length of this chapter we have chosen to illustrate the potential role of SSNMR with select examples. The use of a multinuclear and multiparametric approach in the SSNMR investigation allows to obtain information at the local level opening new perspectives in the prediction and design of supramolecular systems. In particular, HBs, polymorphism, intermolecular packing arrangements, and dynamics of molecular segments can be investigated in great detail, in most cases without the need for special sample preparation. Correlations between the chemical structure and the SSNMR data have been extensively established. Especially the isotropic chemical shift, the CSA and the dipolar correlations with specific structure features like internuclear distances and dynamics could be established for an increasing number of complex supramolecular architectures. Moreover quantitative correlations between the SSNMR parameters and specific bond angles and bond lengths have been obtained. Concerning the HB we reported how 1 H, 13 C, and 15 N chemical shifts and chemical shift tensors can provide information on the presence and the strength of the interaction, data which are not easily amenable to obtain by diffraction techniques.

Several correlations among NMR chemical shifts, geometrical parameters and computed data were also reported. We showed the versatility of the NMR technique in covering a wide timescale of fluxional processes from 102 to 10−10 s by means of the analysis of its parameters such as the relaxation times T1 , T2 , and T1ρ . We highlighted also that when weak forces are involved in a supramolecular structure there is a decrease of the energy barriers associated to the motion of groups or of entire molecules in the crystal packing. Often NMR data represent the best evidence to demonstrate that the extent of guest dynamics included in the host cavity is dependent on the degree of symmetry associated with the included molecule as well as on the strength of the host–guest interactions. Activation energy calculations for molecular processes in high-resolution SSNMR studies can reveal many interesting and sometimes controversial factors related to molecular reorientation and static or dynamic disorder found by single-crystal X-ray diffraction studies.

ACKNOWLEDGMENTS We are indebted with Dario Braga and Fabrizia Grepioni for the helpful discussions.

REFERENCES 1. R. Gobetto, C. Nervi, M. R. Chierotti, et al. Chem.—Eur. J., 2005, 11, 7461. 2. R. K. Harris, Solid State Sci., 2004, 6, 1025. 3. D. Braga, L. Maini, C. Fagnano, et al. Chem.—Eur. J., 2007, 13, 1222. 4. I. Schnell, S. P. Brown, H. Yee Low, et al. J. Am. Chem. Soc., 1998, 120, 11784. 5. D. Braga, G. Palladino, M. Polito, et al. Chem.—Eur. J., 2008, 14, 10149. 6. J. Brus and J. Jakes, Solid State Nucl. Magn. Reson., 2005, 27, 180. 7. M. J. Duer, Solid-State NMR Spectroscopy, Blackwell Publishing Ltd., Oxford, 2004. 8. J. W. Steed and J. L. Atwood, Supramolecular Chemistry, John Wiley & Sons, Ltd, Chichester, 2009. 9. A. Pines, M. G. Gibby, and J. S. Waugh, J. Chem. Phys., 1973, 59, 569. 10. E. Vinogradov, P. K. Madhu, and S. Vega, Chem. Phys. Lett., 1999, 314, 443. 11. B. C. Gerstein, R. G. Pembleton, R. C. Wilson, L. M. Ryan, J. Chem. Phys., 1977, 66, 361.

and

12. E. Vinogradov, P. K. Madhu, and S. Vega, New Techniques in Solid-State NMR, Springer, Berlin, 2005, vol. 246, p. 33. 13. P. K. Madhu, Solid State Nucl. Magn. Reson., 2009, 35, 2.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Solid-state NMR studies on supramolecular chemistry 14. E. Vinogradov, P. K. Madhu, and S. Vega, Chem. Phys. Lett., 2002, 354, 193. 15. A. Lesage, D. Sakellariou, S. Hediger, et al. J. Magn. Reson., 2003, 163, 105. 16. S. Dusold and A. Sebald, Dipolar recoupling under magicangle spinning conditions, Annual Reports on NMR Spectroscopy, Academic Press, Inc., San Diego, CA, 2000, vol. 41, p. 185. 17. M. H. Levitt, Symmetry-based pulse sequences in magicangle spinning solid-state NMR, in Encyclopedia of Nuclear Magnetic Resonance, eds. D. M. Grant and R. K. Harris, Wiley, Chichester, 2002, vol. 9, p. 165. 18. W. Sommer, J. Gottwald, D. E. Demco, and H. W. Spiess, J. Magn. Reson., 1995, 113A, 131. 19. H. Geen, J. J. Titman, J. Gottwald, and H. W. Spiess, J. Magn. Reson., 1995, 114A, 264. 20. S. P. Brown and H. W. Spiess, Chem. Rev., 2001, 101, 4125. 21. I. Schnell, A. Lupulescu, S. Hafner, et al. J. Magn. Reson., 1998, 133, 61.

15

40. W. Veeman, Prog. NMR Spectrosc., 1984, 16, 193. 41. D. Braga, L. Maini, G. de Sanctis, et al. Chem.—Eur. J., 2003, 1, 5538. 42. G. C. Levy and R. L. Lichter, Nitrogen-15 Nuclear Magnetic Resonance Spectroscopy, John Wiley & Sons, Inc., New York, 1979. 43. A. Dos, V. Schimming, S. Tosoni, and H.-H. Limbach, J. Phys. Chem. B, 2008, 112, 15604. 44. D. Braga, S. L. Giaffreda, K. Rubini, et al. CrystEngComm, 2007, 9, 39. 45. M. Schultz-Dobrick, T. Metzroth, ChemPhysChem, 2005, 6, 315.

H. W. Spiess,

et al.

46. E. B. Brouwer, D. M. Gougeon, J. Hirschinger, et al. Phys. Chem. Chem. Phys., 1999, 1, 4043. 47. A. Brinkmann and A. P. M. Kentgens, J. Am. Chem. Soc., 2006, 128, 14758. 48. R. Tycko and Y. Ishii, J. Am. Chem. Soc., 2003, 125, 6606.

22. S. P. Brown, Macromol. Rapid Commun., 2009, 30, 688.

49. J. T. Nielsen, M. Bjerring, M. D. Jeppesen, et al. Angew. Chem., Int. Ed. Engl., 2009, 48, 2118.

23. M. G. Munowitz, R. G. Griffin, G. Bodenhausen, T. H. Wang, J. Am. Chem. Soc., 1981, 103, 2529.

and

50. C. Kehlet, M. Bjerring, A. C. Sivertsen, et al. J. Magn. Reson., 2007, 188, 216.

24. D. McElheny, E. de Vita, and L. Frydman, J. Magn. Reson., 2000, 143, 321.

51. K. Takegoshi, S. Nakamura, and T. Terao, Chem. Phys. Lett., 2001, 344, 631.

25. T. Gullion and J. Schaefer, J. Magn. Reson., 1989, 81, 196.

52. S. Luca, W.-M. Yau, R. Leapman, and R. Tycko, Biochemistry, 2007, 46, 13505.

26. D. D. Laws, H.-M. L. Bitter, and A. Jerschow, Angew. Chem., Int. Ed. Engl., 2002, 41, 3096.

53. S. K. Nurly and G. A. Pestko, Science 1985, 299, 23.

27. D. P. Raleigh, M. H. Levitt, and R. G. Griffin, Chem. Phys. Lett. 1988, 146, 71.

54. K. Guckian, B. A. Scheitzer, X.-A. Ren, et al. J. Am. Chem. Soc., 1996, 118, 8182.

28. T. Gullion and A. J. Vega, Prog. Nucl. Magn. Reson. Spectrosc., 2005, 47, 123.

55. E. J. Corey and M. C. Noe J. Am. Chem. Soc., 1996, 118, 319.

29. A. Lesage, M. Bardet, and L. Emsley, J. Am. Chem. Soc., 1999, 121, 10987.

56. S. P. Brown, T. Schaller, U. P. Seelbach, et al. Angew. Chem., Int. Ed. Engl., 2001, 40, 717.

30. J. Schmidt, A. Hoffmann, H. W. Spiess, and D. Sebastiani, J. Phys. Chem. B, 2006, 110, 23204.

57. S. P. Brown, I. Schnell, J. D. Brand, et al. J. Am. Chem. Soc., 1999, 121, 6712.

31. L. Frydman and J. S. Harwood, J. Am. Chem. Soc., 1995, 117, 5367.

58. M. R. Chierotti and R. Gobetto, Eur. J. Inorg. Chem., 2009, 18, 2581.

32. J. Sopkova-de Oliveira Santos, V. Montouillout, F. Fayon, et al. New J. Chem., 2004, 28, 1244.

59. M. E. Smith, Multinuclear Solid State NMR of Inorganic Materials, Pergamon Materials Series, Pergamon-Elsevier Science, Oxford, 2002, vol. 6.

33. B. Moulton and M. J. Zaworotko, Chem. Rev., 2001, 101, 1629. 34. J. Brus, J. Dybal, P. Sysel, and R. Hobzova, Macromolecules, 2002, 35, 1253. 35. M. R. Chierotti and R. Gobetto, Chem. Commun., 2008, 1621. 36. G. A. Jeffrey, An Introduction to Hydrogen Bonding, Oxford University Press, Oxford, 1997. 37. R. Gobetto, C. Nervi, E. Valfr`e, et al. Chem. Mater., 2005, 17, 1457. 38. R. K. Harris, P. Y. Ghi, R. B. Hammond, et al. J. Chem. Soc., Chem. Commun., 2003, 2834. 39. Z. Gu and A. McDermott, J. Am. Chem. Soc., 1995, 115, 4262.

60. D. Allion and C. P. Slichter, Phys. Rev. Lett., 1964, 12, 168. 61. K. Schmidt-Rohr and H. W. Spiess, Multidimensional Nuclear Magnetic Resonance and Polymers, Academic Press, London, 1994. 62. J. R. Lyerla, High-resolution NMR of glassy amorphous polymers, in Methods in Stereochemical Analysis, High Resolution NMR Spectroscopy of Synthetic Polymers in Bulk, ed. R. A. Komoroski, VCH Deerfield Beach, 1986, vol. 7, p. 63. 63. R. R. Vold, Deuterium NMR studies of dynamics in solids and liquid crystals, in Nuclear Magnetic Resonance Probes of Molecular Dynamics, ed. R. Tycko, Kluwer Academic Publishers, Dordrecht, 1994, pp. 27–112.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

16

Techniques

64. M. Cutaijar, S. E. Ashbrook, and S. Wimperis, Chem. Phys. Lett., 2006, 423, 276.

67. C. Moon, G. Brunklaus, D. Sebastiani, et al. Phys. Chem. Chem. Phys., 2009, 11, 9241.

65. Y. J. Lee, T. Murakhtina, D. Sebastiani, and H. W. Spiess, J. Am. Chem. Soc., 2007, 129, 12406.

68. S. Aime, H. C. Canuto, R. Gobetto, and F. Napolitano, Chem. Commun., 1999, 281.

66. S. L. Gould, D. Tranchemontagne, O. M. Yaghi, and M. A. Garcia-Garibay, J. Am. Chem. Soc., 2008, 130, 3246.

69. S. Aime, M. R. Chierotti, R. Gobetto, et al. Eur. J. Inorg. Chem., 2008, 2008, 152.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc026

Mass Spectrometry and Gas-Phase Chemistry of Supermolecules: A Primer Egor V. Dzyuba, Johannes Poppenberg, Sebastian Richter, Ralf W. Troff, and Christoph A. Schalley Freien Universit¨at Berlin, Berlin, Germany

1 Introduction 2 Methods and Mindsets 3 The Secondary Structure of Supermolecules: The Topology of Mechanically Bound Molecules 4 Chiral Recognition: Homochiral Serine Octamer 5 Reactivity of Supermolecules in Solution: Metallo-Supramolecular Complexes 6 Reactivity of Supermolecules in Solution: Self-Sorted Pseudorotaxanes 7 Reactivity of Supermolecules in the Gas Phase 8 The Best Fit Model of Alkali Metal Ion Binding to Crown Ethers: Valid in the Gas phase? 9 Conclusion Acknowledgments References

1

1 2 4 7 11 17 21 29 30 30 31

INTRODUCTION

During the last 40 years, supramolecular chemistry has matured into an independent field of chemical research, with focus on the noncovalent bond and the chemistry beyond the molecule.1–3 Noncovalent bonds are usually weak and often render complex formation reversible. However, it is not only the dynamic nature of complex formation

but also the high dependency of the weakly bound aggregates on the environment, for example, competitive solvents and the like, that make the analysis of supermolecules difficult. Therefore, their detailed characterization is often challenging and requires the combination of several complementary analytical methods. Noncovalent complexes of smaller ions have been investigated by mass spectrometric methods since the early 1960s.4 Nevertheless, it would certainly be fair to say that mass spectrometry (MS)5 is a late bloomer among the instrumental methods used for the routine characterization of supermolecules, because the soft ionization methods have not been routinely available before the 1990s. Although several quite sophisticated techniques existed for the preparation of gaseous noncovalent ions early on, the routine ionization of intact noncovalent complexes was quite difficult to achieve before the advent of electrospray ionization (ESI) and matrix-assisted laser desorption ionization (MALDI) MS. Meanwhile, MS offers a huge arsenal of gas-phase chemistry experiments that allow a much more detailed view on the structure of noncovalent complexes in solution and gas phase, their solution reactivity as well as their reactivity in the gas phase, and their gas-phase energetics.6 Thus, it is much more than just a simple characterization tool and can add valuable insight into many aspects of supramolecular chemistry, such as molecular recognition or host–guest chemistry, self-assembly, self-sorting, and the structures of mechanically bound molecules. The transition from solution to the highly diluted gas phase inside a mass spectrometer represents a quite drastic change in environment. Since the properties of noncovalent complexes depend much more, for example, on the nature of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Techniques

the solvent than those of covalent molecules, the changes caused by this transition may well add new insight into the role of the environment. The second aspect is that the ions in a mass spectrometer are isolated from each other. Consequently, no dynamic exchange reactions can occur and the gas phase offers a completely different view on their reactivity. This chapter aims at illustrating the scope and limitations of the method by discussing a number of more recent examples. To illustrate how the experiments are carried out and what aspects are important to take into account in the interpretation of the results, we added experimental details for a selection of experiments, which are described in this chapter.

2 2.1

Since ESI is the most commonly used ionization method, let us focus on this method here (Figure 1). It represents a very soft ionization method that directly transfers the complex ions from solution into the gas phase.8–10 The sample is dissolved in an appropriate solvent and this solution is transferred with a flow rate of a few microliters per minute through a metal capillary into a 2–5 kV electric field. The high voltage affords charge separation at the capillary tip and a so-called Taylor cone is formed. From the tip of the Taylor cone, a jet of small droplets is ejected; the droplets have a large excess of positive or negative charges depending on the high voltage polarity. Droplet formation is usually supported in an airbrush-like manner by a concentric stream of nebulizer gas (e.g., N2 ). Solvent evaporation from the droplets is aided by a stream of heatable desolvation gas. During solvent evaporation, the droplets shrink concentrating the charges in a smaller volume until the Rayleigh limit is reached, at which the droplet cannot support all charges anymore. Two models on how desolvated ions finally form exist: the charge residue model suggests that the droplets undergo Coulomb fission at the Rayleigh limit and form a number of smaller droplets. This process continues until only one ion is left in a nanodroplet, from which the residual solvent evaporates to yield the bare ion. The ion evaporation model, on the other hand, takes into account the fact that single ions can be evaporated from multiply charged droplets that still contain many ions. CSI is similar to ESI.11 In this case, the source housing and the desolvation gas are cooled and thus these help to stabilize the complex ions by reducing their internal energies. If one wishes to carry out gas-phase experiments, that is, to manipulate mass-selected ions inside the mass spectrometer, ion-trap analyzers offer the broadest arsenal of experiments including unimolecular fragmentations as well as bimolecular reactions with sufficiently volatile neutral reagents. Consequently, the choice of analyzer is also an important point. Mass analyzers use static or dynamic electric or magnetic fields to separate the ions either in time or in space. Sector-field mass analyzers use magnetic (B) and electrostatic (E) sectors to separate the ions

METHODS AND MINDSETS Ion sources and mass analyzers

The key to a mass spectrometric study of a supermolecule is its intact ionization and transfer into the high vacuum inside the mass spectrometer. Many different ionization methods—each with its particular scope—exist, most of which are, however, unsuitable for the ionization of weak complexes: electron and chemical ionization (EI, CI), for example, require volatile samples, and fast atom bombardment (FAB) usually applies polar protic and thus quite competitive matrices. Today, most supermolecules are, therefore, studied with ESI or its variant, coldspray ionization (CSI). MALDI also works in many instances, although the choice of less common, unpolar matrices is sometimes required together with special ion-labeling strategies7 to keep the complexes intact. Less commonly used, laser-induced liquid bead ion desorption (LILBID) and resonance-enhanced multiphoton ionization (REMPI) are useful techniques, because LILBID broadens the range of sample solvents that are compatible with ionization and because the supersonic jet expansion in REMPI cools the complexes to very low internal energies and thus prevents dissociation.6

Nebulizer gas stream

ESI capillary tip

Taylor cone

Charged droplets

Jet ++ ++ + + + ++ ++ +

Desolvated ions +

++ +++++

+ +

Desolvation gas stream

2

I High voltage

Figure 1

The principle of electrospray ionization (ESI). (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry via impulse (B) and kinetic energy (E) selection. They are usually coupled to EI/CI or FAB ion sources and therefore not often used to analyze supermolecules. Quadrupole mass analyzers are common low-cost analyzers that are frequently coupled to ESI ion sources. These analyzers can be used to select ions at a single m/z ratio or to scan a whole m/z range of up to 4000 with an accuracy of about 100 ppm. Their resolution is quite limited to a range below 4000. Although collision-induced fragmentation is possible in triple-quadrupole instruments, their utility for gas-phase experiments is quite limited. Instead, quadrupole ion traps, which usually combine higher resolution and a broader mass range with the ability to store ions over time, offer a variety of different experiments that can be conducted with mass-selected ions. Linear time-of-flight (TOF) and reflectron time-of-flight (ReTOF) mass analyzers are often coupled to MALDI and ESI ion sources. Linear TOF analyzers have a resolution in the range of 8000 and a large m/z range (>300 000). This comes with a lower mass accuracy of about 200 ppm. ReTOF instruments with a much better mass accuracy of 5–10 ppm provide higher resolution (15 000), but have a narrower mass range ( A2H@B21C7
[email protected] If one examines the corresponding exchange rates of the three assemblies, a quite interesting mismatch of thermodynamic and kinetic behavior is observed: the order of exchange rates is completely reversed: A2-H@DB24C8 (∼10−2 s timescale) > A2-H@B21C7 (∼10 s timescale)  A1-H@DB24C8 (multiminute timescale). While the most stable pseudorotaxane A1-H@DB24C8 forms most slowly, the least stable one A2-H@DB24C8 is the fastest to assemble. An important conclusion from these data is that errors must occur during the assembly reaction—not only for the simple fourcomponent self-sorting system but also for more complex architectures built from the same building blocks. With the microreactor technique in hand, MS should be able to examine the generation of errors and their correction steps. If one follows the formation of the two self-sorted pseudorotaxane by MS, the microreactor technique described above can be used for the shorter reaction times. In this case, the two axles would be combined in syringe 1 and the two crown ethers, in syringe 2. For times longer than a minute or so, just mixing all four components in a flask and recording mass spectra of the mixture after different reaction times is sufficient. These experiments clearly demonstrate the initial formation of two pseudorotaxanes (Figure 16): A1-H@DB24C8 and A2-H@DB24C8. While the first one is already one of the two final products, the latter one represents the quickly forming mismatched pseudorotaxane, which over time disassembles and reassembles

to yield the two expected self-sorted products. The error correction step can thus easily be observed.

6.2

A sequence-specific [3]pseudorotaxane

A more complex architecture52, 54 becomes available when two binding sites are combined in one building block. The orthogonality of the two binding motifs ensures that all building blocks are incorporated in exactly the intended positions. The [3]pseudorotaxane shown in Figure 17 is an example of this approach. The central phenylene group of the ditopic axle makes sure that self-sorting occurs. Consequently, the sequence of the two wheels is clear: DB24C8 moves to the position next to the anthracene, while B21C7 occupies the more remote ammonium ion. The sequence of wheels is thus well-defined and the pseudorotaxane in Figure 17 is the only possible product. The first question to be answered is how evidence can be gathered for the sequence of wheels along the axle by ESI-MS. After the final product has been characterized with respect to its structure, the error correction steps during its formation can be investigated. The ESI mass spectrum of the [3]pseudorotaxane shows only one intense signal corresponding to the desired product. Its structure can indeed be determined by fragmentation of the mass-selected singly charged pseudorotaxane ion in an IRMPD experiment (Box 6). The results of this experiment are shown in Figure 17(b). The pseudorotaxane ion at m/z 1364 is isolated first and then subjected to different IRMPD laser intensities. At first, the loss of neutral B21C7 is observed at quite low laser power (10% of the maximum of 25 W). At higher laser power, still no DB24C8 is lost, likely because the free ammonium binding site is now blocked by formation of a strong ion pair with the remaining PF6 − counterion. Consequently, the second step is the deprotonation of this binding site and loss of HPF6 . Only after this step, the larger crown can leave the axle. For

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry

[PR3-H]+ m /z 550

[PR1-H]+ m /z 642

[PR2-H]+ m /z 746

Reaction time 22 s

O O H2 N +

O

+

[C8-Na] m /z 471

O

5 min

O PF6

−

PF6 OH

− +

O

O

O O

O

O

O

O

O

H2N +

+

19

O

PR1-H·PF6 10 min

Fast

Ratio: 1 : 1 : 1: 1

HO

20 min O O O

O

40 min

O

O

+

− PF6 H2N +

O

O

+

−

H2N +

O

O

O O

PF6 O

O

+

O

15 h (a) 500

550

600

650

700

m /z

750

Medium

O

O H2 O N + O

O O

O O

− PF6

O H2 O N + O O O

+

O O

(b)

PR2-H·PF6

PF6

O O

−

Slow

OH

H2 O N + O

O O

PF6

HO

−

O O O

O

+

−

H2N +

+

O

O

O O

PF6

O

PR2-H·PF6

PR3-H·PF6

Figure 16 (a) ESI mass spectra obtained during the self-sorting process. [PR1-H]+ is the mismatched pseudorotaxane—an error, which is corrected with the further progress of the self-sorting reaction. (b) The schemes show the self-sorting pathways. Structures in gray represent the mismatched assembly formed during the course of the reaction. (Reproduced from Ref. 53.  American Chemical Society, 2010.) − 2 PF6

H2 N +

O

OH

O

O

IRMPD of m /z 1364

O

O

O

O

ps-rot+HPF6+H+ m /z 1364

ESI-MS

H2 N +

m /z 1008

Isolation % Laser intensity

O O

O O

O O

O

-B21C7

O

10

-HPF6

m /z 861 20 O

O O O H2 N + O O O

O O

O

−

2 PF6 O H2 O N OH + O O O

m /z 413

90 400

(a)

-DB24C8

(b)

600

800

1000

1200

1400

m /z

Figure 17 (a) [3]Pseudorotaxane generated from a ditopic axle with two different ammonium binding sites and the two crown ethers DB24C8 and B21C7. (b) ESI mass spectrum (first row), IRMPD experiment performed with the mass-selected singly charged [3]pseudorotaxane confirming the sequence (rows 2–4). (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

20

Techniques

Box 6: IRMPD Experiments to Unravel the Sequence of Wheels Incorporated in the SelfSorted [3]Pseudorotaxane A. Sample Preparation: Dissolve axle (250 µmol l−1 ) and the two crown ethers (1 : 1 : 1) in dichloromethane and let the mixture equilibrate over night for a complete self-sorting.

IRMPD, MS 3 Magnet IR laser beam

Ion generation

Accumulation Ion transfer

Hexapole

ESI source

FTICR cell

Sample inlet

Pumps

Pump

Pump

B. Procedure: Generate ions by positive-ESI → accumulate a sufficiently high number of pseudorotaxane ions in hexapole → introduce ion package into the FTICR analyzer cell → mass-select the desired pseudorotaxane ions → irradiate with IR laser (10.6 µm wavelength; 500–1000 ms) → detect product ions → repeat experiment at different laser power settings → determine sequence of crown ethers from fragmentation sequence. C. Necessary Controls: Confirm sequence in MS3 experiments by reselecting the fragmentation intermediate after loss of B21C7 and after HPF6 loss and subjecting them to the same fragmentation experiment. The result should be in line with the sequence deduced from the MS/MS experiments. D. Remarks: Charge state may have significant influence on fragmentation. For example, fragmentation of doubly charged pseudorotaxane ion governed by charge repulsion and thus not structure indicative.

[PR5-2H·PF6]+ Reaction m /z 915 time 22 s

[PR4-2H·PF6]+ m /z 1008 [PR6-2H·PF6]+ m /z 1456

Ratio: 1 : 1 : 1 H2 N + − PF6

H2 N + − PF6

Slow

O O O H2 O N + O O − O PF6

Fast

O O O H2 O N O + O − O O PF6

H2 N + − PF6

20 min

PR4a-2H·2PF6

− PF6− PF6 O O O O O H2 O O H2 O N N + O O O + O

OH

+ B21C7

O O

Medium

40 min

60 min [PR7-2H·PF6]+ m /z 1364

[PR5-H]+ m /z 769 1000

1200

m /z

+ B21C7

Fast

PR6-2H·2PF6

OH

PR4b-2H·2PF6 + B21C7

15 h

OH

O O

+ DB24C8 O O PF − 6 H2 O H2 O N N + O + O − PF6 O O

800

OH

PR5-2H·2PF6 + DB24C8

+ DB24C8 + B21C7

[PR4-H]+ m /z 861 10 min

(a)

OH

H2 N + − PF6

Slow

− PF6 O O O O H O H2 O O N2 O N + + O O O O − O O O PF6

OH

PR7-2H·2PF6

1400 (b)

Figure 18 A streamlined formation pathway for the [3]pseudorotaxane. (a) Mass spectra recorded during the formation of PR72H·2PF6. (b) Formation pathway of PR7-2H·2PF6 derived from the mass spectra shown. (Reproduced from Ref. 53.  American Chemical Society, 2010.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry

O O − PF6 H O N2 O OH +

O

O O

− O O PF6 O H2 O N + O O

O O − PF6 O H2 O OH N + O O

O O

O O O H2 O N + O O

O O H O N2 O

O O

O

O O

21

−

2 PF6

OH

+

O

O

HO −

−

O N N

O O

O

O NH2 O O O +

2 PF6

2 PF6 O

O O O H 2N + O O O O

O O

N N

O +

O

NH2

O O O H2N + O O O O

O O

O

HO

HO

HO

HO O +

O O

O O NH2 O O − 4 PF6

O N N

O H 2N O O

O +

O

O O

O O

O H2N O

O +

O

O O

N N

O

O +

O O

O O O H 2N + O O O O

O NH2 O O −

4 PF6

O O O O + H N O 2 O O

N N

O O O O H 2N + O O O O

O O O O H2N + O O O O

N N

HO

Figure 19 means.

Other pseudorotaxane assemblies, whose structures and formation pathways can be investigated by mass spectrometric

the pseudorataxane, these MS/MS experiments clearly confirm the expected sequence of crown ethers on the ditopic axle.55 After the characterization of the structure, we can address the formation pathways. From the kinetics observed for the simple four-component system above, one would expect that the formation of the [3]pseudorotaxane is quite straightforward. The fastest step should be threading the larger DB24C8 onto the narrow part of the ditopic axle. The second fastest step is then the slippage of that crown ether over the central phenylene group, while the slowest step is the final binding of the smaller crown B21C7 onto the free narrow part. This formation pathway is a “streamlined” process in the sense that each intermediate forms faster than it is consumed and thus accumulates and makes the subsequent step efficient. Consequently, no intense signals for mismatched assemblies are observed in the mass spectra recorded during assembly formation. This is indeed the case: Figure 18 shows the results of this experiment. The same approach can be applied to even more complex assemblies such as those shown in Figure 19.53, 54 In all

these structures, (tandem) MS provides structural evidence and yields information on the sometimes quite complex assembly and self-sorting pathways.

7

REACTIVITY OF SUPERMOLECULES IN THE GAS PHASE

The reactivity in the gas phase is significantly different from that in solution so that it will be described in its own chapter. One reason for these differences is the absence of competing solvents, but certainly more important is the fact that dynamic reactions are completely suppressed. Once a supermolecule is ionized and transferred into the high vacuum inside the mass spectrometer, it is an isolated particle that cannot exchange building blocks with other complexes around. Consequently, the gas phase offers unique conditions to study intramolecular processes within supermolecules. Although they may proceed in solution, their reversible formation necessarily leads to a superposition with intermolecular processes. They are at

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

22

Techniques

least difficult to separate by solution tools, and thus, the possibility to study only the intramolecular processes in the gas phase is one of the great advantages offered by MS.

7.1

“Neighbor-group assistance” in the fragmentation of metallosupramolecular polygons and cages

Supramolecular squares, such as the example shown in Figure 20, self-assemble from simple precursors (4,4 bipyridine and (dppp)Pt(II) triflate) and can be ionized in different charge states from +2 to +7 by ESI through the consecutive losses of counterions.56, 57 Consequently, the influence of the charge state on the fragmentation patterns can be examined by mass-selection of one particular charge state and a subsequent IRMPD experiments. In general, the higher the charge state the more abundant the chargeseparating fragmentation reactions, while losses of neutral fragments prevail only for the lower charge states that do not suffer much from charge repulsion. For the square ions, this is indeed observed. While the doubly charged square predominantly undergoes losses of neutral ligands, the triply charged analog exhibits a quite specific chargeseparating fragmentation.

The triply charged square fragments almost exclusively into a singly charged 1 : 1 and a doubly charged 3 : 3 fragments (Figure 21a). This is remarkable, because several other charge-separation fragmentation processes would be expected to compete. Especially, the separation into two 2 : 2 complexes, one singly and one doubly charged, is expected to be as energy demanding as the observed dissociation reaction. The first potential reason for this highly specific decomposition could be the more favorable distribution of charges over the fragments: a 3 : 3 complex would certainly more easily be able to stabilize two charges as compared to its 2 : 2 alternative. However, the fragmentation of the +5 charge state of the square is even more specific, in that it exclusively dissociates into a singly charged 1 : 1 and a quadruply charged 3 : 3 fragments. This is certainly not the most favorable way to distribute the five charges over the fragments and thus rules out the charge distribution as the only explanation for the selectivity observed in the dissociation reactions. An alternative explanation invokes a “backside-attack mechanism” as shown in Figure 21(b). The first step in the dissociation is the cleavage of one Pt–N bond, and it yields a ring-opened 4 : 4 complex. In the second step, another Pt–N bond cleavage generates the fragments. A 3+

calc.

m /z 1265.8

4+

calc.

m /z 912.1

Ph Ph P M N P Ph N Ph

exp. 702

911

915

5+

1264

1269

8

+

P N MP Ph N Ph

M = Pt(II) − 8 CF3SO3

exp.

699

Ph Ph

Ph N Ph P M N P Ph Ph

N Ph Ph N MP P Ph Ph

m /z 699.9 558

560 6+

m /z 558.4 7+ 2+

2+

m /z 457.4

m /z 1442.6

400

600

800

1000

1200 m /z

1400

m /z 1973.2

1600

1800

2000

Figure 20 A typical ESI-FTICR mass spectrum of the metallosupramolecular square shown in the upper right corner. The deviations observed in the isotope patterns of charge states +4 and +6 indicate a minor fragmentation into 2 : 2 complexes of ligand and metal. (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry

Pt PtOTf

+

+

m/z 756

23

Pt 5 TfO 3 +

Pt TfO Pt

m/z 912

Pt

m/z 1266

Pt

Pt Pt

Pt

3 TfO

4 TfO 2 +

+

m/z 1974

m/z 1365

Pt

m/z 1443 m/z 1896 40 ms 30 ms 20 ms 10 ms 0 ms 800

1000

1200

(a)

1400

1600

1800

m/z 4 TfO Pt

Pt

Pt

5 TfO 3+

Pt b f e

Pt

Pt

m/z 1443

+

+

Pt TfO

m/z 912

Pt Pt

2 TfO 3 TfO 2+ + +

Pt 5 TfO 3+

d

+

Pt

Cleavage at position (f)

5 TfO 3+

Pt 2

Pt

c

a

Pt

5 TfO 3+

Pt Pt m/z 1266

Pt

Pt

g

Pt Pt

m/z 912

Pt

Pt

Pt m/z 1974

Pt Cleavage at position (d)

(b)

Figure 21 (a) IRMPD mass spectra of the mass-selected triply charged square after different irradiation times. The singly charged 2 : 2 complex at m/z 1974 can be identified as a consecutive fragment from the doubly charged 3 : 3 complex. (b) Postulated “neighbor-group assistance” giving rise to 3 : 3 and 1 : 1 fragments. A similar backside attack is not possible for a fragmentation into two 2 : 2 complexes because of geometric reasons. (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.)

backside attack of the uncoordinated pyridine nitrogen atom on the second to last metal center leads to the formation of a new Pt–N bond either before or during the expulsion of the 1 : 1 fragment. Such a backside attack is impossible for the cleavage into two half squares. Consequently, the first pathway is energetically more favorable. This scenario is in line with the widely accepted mechanism for ligand exchange on square-planar d8 metal centers which proceeds through a trigonal bipyramidal intermediate. It represents the supramolecular analog of a neighbor-group effect. Mixing the tripy ligand shown in Figure 22 with the appropriate amount of (dppp)Pd(II) triflate in solution leads

N N

(dppp)Pd(CF3SO3)2

cage1 +

N

triply

cage2

Figure 22 Equilibrium between the two possible structures formed from the tridentate ligand and the palladium corner. (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

24

Techniques

3+

m /z 1993 Slow

m /z 667 2+

2+ Fast

2+

-tripy

m /z 1458

BSA

2+ 2+

+

m /z 2465

m /z 2247 + +

m /z 2656

BSA MS3

m /z 667 -tripy

+

MS3

+

+

m /z 3064 m /z 1866

m /z 1866 Laser power 50% (1 s)

2+ +

m /z 1485

500

m /z 2274 -tripy

100% (1 s)

1000

1500

-tripy

2000

2500

3000

m /z

Figure 23 Results of MS/MS and MS3 IRMPD experiments (Box 6) performed with triply charged cage1 and doubly charged cage2 (inset). The very low signal at m/z 2656 indicates that the corresponding ion is formed much slower compared to the time it takes to undergo rearrangement and ligand loss. This fragment can only be formed through the dissociation of two bonds and loss of one corner. The subsequent step must now proceed through a backside attack that involves a net cleavage of only one Pd–N bond and should therefore be more favorable than the first step. (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry to an equilibrating mixture of two cages: an M6 L4 bowl (cage1) and a smaller M3 L2 cage (cage2). When subjecting the triply charged cage1 to an IRMPD experiment, as discussed above, again a backside attack is observed (Figure 23).58 Indeed, two subsequent backsideattack steps easily rationalize how cage1 is converted into cage2 in the gas phase. It is again the high specificity of the fragmentation reactions that provides evidence for two cage-contraction steps. The important hint thus comes again from those fragments that are not observed in the mass spectra. These examples demonstrate how different the reactions of metallosupramolecular assemblies can be in solution and the gas phase. The “backside-attack mechanism” can now be used to predict and to explain the fragmentation pattern of complex supramolecular architectures.

7.2

25

Multivalency in the gas phase: host–guest chemistry of dendrimers

Dendrimers are “tree-like” architectures, which consist of a hyper-branched core and several outer shells, which are iteratively generated by a divergent or convergent synthetic approach (Figure 24). They exhibit multiple copies of the same terminal groups on their many branches. Consequently, they are interesting hosts when these terminal groups are binding sites for guest molecules. One example, a third-generation POPAM dendrimer,59 is shown in Figure 24. Each branch is terminated by an adamantyl urea group, two of which bind urea guest molecules by hydrogen bonding. When the guest is equipped with a strong acid, a proton transfer from the acid to the adjacent

O HO2C NH NH O H N H N O H N

H N

O

NH

O

HN

HN N

N

H2O3P

O H N

N

N

H N

H N

N

N

H N HO2C

N H

N H

N

N D

O

N H

N

N

HN

NH NH

O

HN

NH

O OC12H25

O C1

N

N H

U2

N H

N H

O

O

N H

O

N

O N H

N H

O

O

N

U1

O

HN

O

N H

N H

HN

NH

N

N H

O

N H

O H2O3P

OC12H25

O P1

O N H

O N H

N H O

HO3S

O

N H

O N H

OC12H25 OC12H25 OC12H25

O

OC12H25 OC12H25

S1

HN

HN

NH

O HO2C

OC12H25

O C2

O N H + NH

OC12H25 OC12H25

O −

O

P OH

N H N H

O

O

O N H

OC12H25

O

H2O3P P2

N H

O O

HO3S N H

OC12H25 OC12H25

S2

OC12H25 OC12H25 OC12H25 OC12H25 OC12H25

Figure 24 Third-generation POPAM dendrimer D and the different guests used in this study. Inset: the supposed binding motif. (Reproduced from Ref. 6.  John Wiley & Sons, Inc., 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

26

Techniques

5V

55 V 1450

1500

1550

1600

1650

1700

1750

1800

1850

1900

m /z

Figure 25 Tandem mass spectrometric experiment of mass-selected host–guest complex [D•(U2)4 ]3+ . With increasing collision energies, more and more guest molecules dissociate from the host. (Reproduced from Ref. 59.  Wiley-VCH, 2004.)

tertiary amine at the center of the binding pocket provides an additional electrostatic interaction (inset in Figure 24). When D is mixed with six equivalents of guest U2 and a small amount of acetic acid, all possible complexes of both components, from [D•(U2)0 ] to [D•(U2)8 ], appear in different charge states. A tandem mass spectrometric experiment with one of these complexes is shown in Figure 25. The host–guest complex [D•(U2)4 ]3+ was massselected and the collision energy was raised in small steps. With higher collision energies, the progressive loss of neutral guest molecules (blue rectangles in Figure 24) is observed. Competition experiments with different guests (Figure 24) support the proposed binding motif. [D•(U1)2 (U2)2 ]3+ exhibits preferential loss of the carboxylic acid U1 and thus indicates the binding of phosphonic acid U2 to be stronger—as expected from the different acidities. The binding of guests P2 and S2, where the urea unit is missing, also underlines the contribution of ion-pair interaction to the overall binding strength. The carboxylic acid C2, however, is too weak to bind without support from the urea hydrogen bonding. Competition experiments with the P1/P2 and S1/S2 pairs confirm the urea-containing guests to be bound more strongly. The postulated binding motif thus agrees well with the experimental results.

7.3

Tracking molecular mobility: H/D exchange reactions in the gas phase

The exchange of labile hydrogen atoms in the gas phase against deuterium is a tool to examine the gas-phase conformation of, for example, peptides and proteins, because those hydrogen atoms that are either buried deeply inside the ion or which are involved in hydrogen bonding usually do not undergo any exchange. The mechanism for gas-phase H/D exchange reactions between a substrate cation and a deuteration reagent such as ND3 , CH3 OD, D2 O, and CH3 COOD consists of three consecutive steps. First, the two reaction partners form an encounter complex and a proton is transferred from the substrate to the reagent. Then, an isotope scrambling takes place and a new complex is formed, which can dissociate into the deuterated substrate and the protonated reagent. These reactions are of course reversible, so that a backreaction is always possible. Studies with a large range of substrates and deuteration reagents revealed the rate of H/D exchange to be inversely proportional to the difference in proton affinities of the two reaction partners. Differences higher than about 85 kJ mol−1 are usually prohibitive for the exchange to take place.60–62 Surprisingly, this rule does not apply to substrates such as amino acids and peptides.63–65 H/D exchanges have been

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry

27

''Onium mechanism'' D

D

D

N D

OH

D + N D D O

+ N D

H

O

H + H N H

D

O

H N H

OH

H N H

R

R

H

H

D N D

D OH

H N H

O

+

R

OH R

(a) ''Relay mechanism'' D O D

D O H

O

H + H N H

OH R

H O D

D + O

H N H

OH R

H O D

D + O

H N H

OH R

O

D + H N H

OH R

(b)

Figure 26 Two H/D exchange mechanisms that involve a second functional group as an auxiliary to reduce the exchange barrier: (a) “onium mechanism”; (b) “relay mechanism.”

achieved for proton-affinity differences up to 200 kJ mol−1 for these molecules.66 Two different mechanisms were postulated to explain these findings (Figure 26). The exchange with more basic ND3 follows the “onium mechanism” in which a proton is transferred to the ammonia molecule.67 Less basic reagents such as D2 O rather follow the “relay mechanism” in which two protons are transferred at the same time.68 In principle, one can say that the exchange mechanism, and therefore the exchange rate, is very sensitive to structural variations in the substrate and that a second functional group in spatial proximity to the exchanged protons is required for some systems. For example, the presence of basic groups in the amino-acid side chains, as in histidine or lysine, accelerates the exchange. The next consequential step is to find whether it is possible to block the H/D exchange reaction by involving the participating groups in hydrogen-bonding interactions. This would allow the mass spectrometrist to distinguish hydrogen-bond hydrogens from non-hydrogenbonded ones. Beauchamp and coworkers were able to block the exchange on ammonium groups by using crown ethers as protective groups.69 18-crown-6 binds to the positively charged ammonium group, so that no H/D exchange is expected anymore. Experiments with glycine, for example, showed that only the acid OH can be exchanged. In the control experiment, using unprotected glycine, all four labile hydrogen atoms can be easily replaced. The participation of the second functional group is confirmed by H/D exchange on protonated ethylene diamine and its 18-crown-6 complex. Free ethylene diamine ions show a fast replacement of all five exchangeable hydrogens with ND3 as the deuteration reagent. Instead, their complexes with 18-crown-6 do not undergo any significant

exchange—not even at the free amino group. The three protons at the positively charged ammonium group are directly protected by hydrogen bonding to the crown. For the exchange on the second amino group, another functional group is required, which, however, is blocked by the crown ether and thus does not support the exchange at the amino group as well. More complex supermolecules have been investigated: Vainiotalo and coworkers demonstrated the cone formation of a tetratosylated resorcinarene to be intact in the gas phase.70 This conformation is fixed by four intramolecular hydrogen bonds at the wider rim of the resorcinarene. All four nontosylated resorcinarene OH groups are thus involved in hydrogen bonding and do not undergo isotopic exchange. Even the tumbling of ammonium ions inside the cavity of a resorcinarene could be investigated.71 Depending on whether the number of ammonium hydrogen atoms matches the number of hydrogenbonding sites of the host, exchange either occurs or is suppressed. The finding that 18-crown-6 protects an ammonium group against H/D exchange can be used to track its mobility on a given host molecule. Oligolysine chains bear multiple amino groups in their side chains. By ESI, it is easily possible to protonate some of them and generate crownether complexes of different charge states and stoichiometries. If the crown ethers on the oligolysine chain remain fixed at a certain position, one would expect the corresponding ammonium hydrogens not to be available for H/D exchange. Only a partial exchange should then be observed. If, however, the crown ethers are able to move from one side chain to another one, all ammonium and amine hydrogen atoms should be exchangeable. Box 7 provides the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

28

Techniques

Box 7: H/D Exchange Reactions on Oligolysine/18-Crown-6 Complexes in the Gas Phase A. Sample Preparation: Dissolve peptide (50 µmol L−1 ) and crown ether (1 : 2 to 1 : 5) in methanol/formic acid (1%). B. Procedure: Generate ions (positive ESI) → accumulate sufficiently high number of ions in hexapole collision cell → block entry of additional ions with entrance hexapole → introduce methanol-OD into hexapole continuously by opening pulsed valve for defined reaction delay → transfer ions into analyzer cell → detect ions → repeat experiment at different reaction times → evaluate progress of H/D exchange reaction over time. Detection Magnet

H/D exchange Hexapole collision cell

Ion transfer

Ion generation

Entrance hexapole

ESI source

FTICR cell

Sample inlet

CH3-OD Pump

Pumps

Pump

C. Necessary Controls: Repeat experiment with uncomplexed oligolysine chain: Do all labile hydrogen atoms exchange? Check simple model diamines for consistency. Check reproducibility of pulsed valve time control. D. Remarks: H/D exchange too is inefficient in ICR cell. Reaction in hexapole more efficient due to higher methanolOD pressure. Advantage of experiment without prior mass selection: data for all charge states and stoichiometries can be recorded simultaneously.

[(18C6)7•Lys15 + 7H]7+ m /z 542.5

[Lys15 +4H]4+ m /z 485.8

full H /D-X

d 51

[(18C6)5 • Lys15 + 6H]6+ m /z 544.6 [(18C6)3 • Lys15 + 5H]5+ m /z 547.6 d 43

CH 3-OD pulse times 0.00 ms

0.00 ms

1.38 ms

1.40 ms

1.40 ms

5.00 ms

d 52

9 H blocked

d 38

d 53

15 H blocked 2.00 ms

2000 ms

d 33

d 54

21 H blocked 500 ms 487 (a)

491 m /z

10 000 ms

495

545 (b)

550 m /z

555

Figure 27 H/D exchange experiment with [Lys15 +4H]4+ (a) and [(18-C-6)7 ·Lys15 +7H]7+ , [(18-C-6)6 ·Lys15 +6H]6+ , and [(18-C6)3 ·Lys15 +5H]5+ (b). (Reproduced from Ref. 72.  Nature Publishing Group, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry details of the experiment.72 After ion generation, the H/D exchange is performed with methanol-OD as the exchange reagent. The reaction time can precisely be controlled by a pulsed valve through which the methanol-OD is transferred to the hexapole collision cell of the FTICR instrument. The progress of the exchange reaction can then be followed over time. Figure 27 provides the results for free Lys15 in its +4 charge state and for three different representative complexes with different charge states and stoichiometries. The exchange proceeds at high rates and is almost finished for the free oligolysine ions after about 500 ms. Irrespective of the nature of the hydrogen atom, all labile hydrogens can be exchanged. All three complex ions also undergo a still quite fast exchange (Figure 27b). In particular, more hydrogen atoms can clearly be exchanged than would be expected for positionally fixed crown ethers. For [(18-C-6)7 ·Lys15 +7H]7+ as an example, one would expect 21 H atoms to be protected by crown ethers so that only 33 hydrogen atoms remain for the exchange. The exchange reaction has, however, already proceeded beyond this threshold after two seconds, providing evidence for the mobility of the crowns. Without going into detail here, two more important conclusions can be drawn from the exchange behavior: (i) the H/D exchange on model complexes of 18-crown6 with 1,12-diaminododecane results in the conclusion that the crown ethers move together with a proton from an ammonium site to an amine. (ii) The exchange also provides insight into the structural feature. Differences in the exchange behavior of acid-terminated complex [(18-C6)6 Lys15 -COOH+6H]6+ and its amide-terminated analog [(18-C-6)6 Lys15 -CONH2 ·6H]6+ show the acid complex to be zwitterionic, while the amide complex is not. The H/D exchange in the gas phase thus provides profound insight into the properties of supermolecules. A similar movement of crown ethers can be observed along the periphery of POPAM dendrimers.73

8

THE BEST FIT MODEL OF ALKALI METAL ION BINDING TO CROWN ETHERS: VALID IN THE GAS PHASE?

Many methods to determine thermochemical data of supermolecules are solution based.74 To examine solvent effects on the thermochemistry of supermolecules, it is necessary to gather gas-phase data and to compare them to the results obtained from solution experiments. Although still challenging, mass spectrometers offer a number of suitable experiments with which the thermochemistry of

29

noncovalent complexes can be examined under solventfree conditions.75, 76 In the following discussion, we focus on threshold collision-induced dissociation (TCID) experiments. In these experiments, a mass-selected ion is subjected to collisions with a stationary collision gas at precisely controlled kinetic energies. The threshold energy, at which a fragmentation reaction begins to appear, is a measure of the barrier for this reaction. Simple bond cleavages normally do not have a reverse activation barrier, and thus this experiment is also suitable for the determination of bond-dissociation energies in such cases. The TCID method was systematically used to examine crown ether–alkali-metal-ion (Li+ –Cs+ ) complexation energies (Table 1).77 For comparison with the tetra- to hexadentate crown ethers, binding data were also determined for smaller ethers, that is, monodentate dimethylether (DME) and bidentate dimethoxyethane (DXE). In solution, each crown ether binds most strongly to that alkali cation which fits best into the crown’s cavity. For example, K+ is most strongly bound to 18-crown-6, while Na+ prefers 15-crown-5. With the gas-phase data in hand, the validity of this best fit model can be tested under environment-free conditions. The data given in Table 1 show some clear trends: 1.

the bond-dissociation energies (BDE) between DME and the alkali metal ions decrease from the first ligand to the fourth one (BDE: first > second > third > fourth). Owing to the electrostatic nature of the complexes, the high charge density of the bare metal ion decreases when one or more dipoles like the ethers are bound to it, and consequently, the observed trend nicely agrees with expectation. The same holds for the first and second bond energies of the DXE ligands. 2. The alkali-metal-ion polarizability increases from Li+ to Cs+ . Thus, the binding energy of any ligand L in the MLn + complexes decreases with increasing alkali metal ion size: Li+ > Na+ > K+ > Rb+ > Cs+ . 3. The binding energy of 12-crown-4 is lower than that of two DXE molecules, although the number of donor atoms is identical in both complexes. Likewise, two DXE bind more weakly than four DME molecules (BDE: 4 DME > 2 DXE > 12-crown-4). This can be explained by an eclipsed conformation of the CH2 groups in the crown ethers and the DXE ligands when these are bound to an alkali-metal ion. This eclipsed conformation is enthalpically disfavored because of the steric strain existing in this conformation. Furthermore, the ligand is fixed in this conformation on binding so that the entropy is likely also unfavorable. This effect becomes larger when more CH2 CH2 O units are used in the ligand.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

30

Techniques

Table 1 Binding energies (in eV) of dimethylether (DME), dimethoxyethane (DXE), and three crown ethers to alkali metal ions. All data are taken from reference 77. Ligand

Li+

Na+

K+

Rb+

Cs+

1st DME 2nd DME 3rd DME 4th DME 1st DXE 2nd DXE 12-crown-4 15-crown-5 18-crown-6 1st and 2nd DME 3rd and 4th DME All DME Both DXE

1.71 1.25 0.92 0.70 2.50 1.44 3.85 n.d.b n.d.b 2.96 1.62 4.58 3.94

0.95 0.85 0.72 0.63 1.64 1.20 2.61 3.05 3.07 1.80 1.35 3.15 2.84

0.76 0.71 0.59 0.52 1.23 0.92 1.96 2.12 2.43 1.47 1.11 2.58 2.15

0.64 0.57 0.38 0.40a 0.97 0.51 0.96 1.18 1.98 1.21 0.89 1.99 1.48

0.59 0.49 0.41 0.37a 0.59 0.56 0.88 1.04 1.74 1.08 0.78 1.86 1.15

a Estimated b n.d.:

4.

value. not determined.

For all alkali metal ions, the binding energy of the metal ion–crown ether complexes increases if larger crown ethers are used. The more the donor atoms are incorporated in the crown ether, the higher the binding energy becomes. In view of the best fit model, this result is quite astonishing. It appears as if the best fit model is valid only in solution but does not accurately describe gas-phase binding. Consequently, the solution-phase best fit needs to be interpreted as a solvent effect. To understand this, we need to take a closer look at the TCID binding data: The bindingenergy differences between the crown ethers reveal that the best fit indeed plays a role in the gas phase. For Na+ , the binding energy increases from 12-crown-4 to 15-crown-5 by a large step of 0.44 eV, but from 15crown-5 to 18-crown-6, the increase is only by a very minor amount of 0.02 eV. These different stepsizes can be explained with a best fit of Na+ into 15-crown-5. In the case of K+ , the difference in BDE from 12-crown4 to 15-crown-5 is 0.16 eV and from 15-crown-5 to 18-crown-6 it increases up to 0.31 eV. A small step is followed by a larger step indicating a better fit of K+ in the cavity of 18-crown-6.

Consequently, the increase in binding strength with increasing number of donor atoms is modulated by the best fit, which thus plays only a minor role in the gas phase, where the ligands are dissociated from the metal ion without any replacement. The gas-phase data thus contain the total binding energy. In solution, a different reaction is examined: On dissociation of the crown, solvent molecules coordinate to the metal ion. Therefore, a large fraction of the total binding energy is compensated for and the small effects originating from the best fit are more clearly observable.

9

CONCLUSION

We hope that we have been able to demonstrate the many possibilities that MS offers to supramolecular chemistry. Beyond the analytical characterization, the large arsenal of tandem mass spectrometric experiments allows the chemist to address questions related to the secondary structures of supermolecules, to their reactivity, and to their thermochemistry. Although we have tried to separate these three aspects—structure, reactivity, and energetics—they are of course closely interrelated. The relative branching ratios in fragmentation reactions always depend on the potential-energy surface of the system under study. The parts of this surface that are accessible is determined by the structure of an ion, which defines the starting point for a reaction. Consequently, these three aspects need always to be thought together in order to achieve a straightforward interpretation of the results from MS experiments. The second point that should be made is that the situation of isolated particles in the high vacuum—in particular the ill-definition of temperature and the nonBoltzmann distribution of internal energies—may appear to be somewhat uncommon for many chemists who are used to do chemistry in solution. Nevertheless, if one considers the great new insights into, for example, the intramolecular reactivity of supermolecules, it is certainly worthwhile even for solution-phase supramolecular chemists to invest effort and time into mass spectrometric analyses of their supermolecules.

ACKNOWLEDGMENTS A special thanks goes to the engaged Schalley and coworkers who investigated supermolecules by MS and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Mass spectrometry and gas-phase chemistry accomplished their many beautiful results. Funding of our research by the Deutsche Forschungsgemeinschaft, the Freie Universit¨at Berlin, and the Fonds der Chemischen Industrie (FCI) is gratefully acknowledged. E. V. D. and S. R. thank the Studienstiftung des Deutschen Volkes and the FCI for Ph.D. fellowships.

31

23. H. Z. Zhang, E. S. Paulsen, K. A. Walker, et al., J. Am. Chem. Soc., 2003, 125, 9284–9285. 24. H. Z. Zhang, T. A. Ferrell, M. C. Asplund, and D. V. Dearden, Int. J. Mass Spectrom., 2007, 265, 187–196. 25. P. Ghosh, G. Federwisch, M. Kogej, et al., Org. Biomol. Chem., 2005, 3, 2691–2700. 26. C. A. Schalley, J. Hoernschemeyer, X.-Y. Li, et al., Int. J. Mass Spectrom., 2003, 228, 373–388.

REFERENCES

27. M. Sawada, Mass. Spectrom. Rev., 1997, 16, 73–90.

1. F. V¨ogtle, Supramolekulare Chemie, Teubner, Stuttgart/ Germany, 1992.

28. R. G. Cooks, J. S. Patrick, T. Kotiaho, and S. A. McLuckey, Mass. Spectrom. Rev., 1994, 13, 287–339.

2. J.-M. Lehn, Supramolecular Chemistry, Verlag Chemie, Weinheim/Germany, 1995.

29. I. H. Chu, D. V. Dearden, J. S. Bradshaw, et al., J. Am. Chem. Soc., 1993, 115, 4318–4320.

3. J. W. Steed and J. L. Atwood, Supramolecular Chemistry, John Wiley & Sons, New York, 2000.

30. A. Filippi, A. Giardini, S. Piccirillo, and M. Speranza, Int. J. Mass Spectrom., 2000, 198, 137–163.

4. For a review on the application of high-pressure mass spectrometry for the evaluation of thermochemical data, see: M. Meot-Ner (Mautner), Chem. Rev., 2005, 105, 213–284.

31. J. H. Guo, J. Y. Wu, G. Siuzdak, and M. G. Finn, Angew. Chem. Int. Ed., 1999, 38, 1755–1758.

5. J. H. Gross, Mass Spectrometry – A Textbook, Springer, Heidelberg/Germany, 2004. 6. C. A. Schalley and A. Springer, Mass Spectrometry and GasPhase Chemistry of Non-Covalent Complexes, John Wiley & Sons, New York, 2009.

32. J. Ramirez, F. He, and C. B. Lebrilla, J. Am. Chem. Soc., 1998, 120, 7387–7388. 33. S. Ahn, J. Ramirez, G. Grigorean, and C. B. Lebrilla, J. Am. Soc. Mass Spectrom., 2001, 12, 278–287. 34. G. Grigorean, J. Ramirez, S. H. Ahn, and C. B. Lebrilla, Anal. Chem., 2000, 72, 4275–4281.

7. K. A. Jolliffe, M. C. Calama, R. Fokkens, et al., Angew. Chem. Int. Ed., 1998, 37, 1247–1251.

35. W. A. Tao, D. X. Zhang, S. Wang, et al., Anal. Chem., 1999, 71, 4427–4429.

8. J. B. Fenn, M. Mann, C. K. Meng, et al., Mass Spectrom. Rev., 1990, 9, 37–70.

36. D. X. Zhang, W. A. Tao, and R. G. Cooks, Int. J. Mass Spectrom., 2001, 204, 159–169.

9. P. Kebarle and L. Tang, Anal. Chem., 1993, 65, A972–A986. 10. J. B. Fenn, Angew. Chem. Int. Ed., 2003, 42, 3871–3894.

37. M. Sawada, Y. Takai, H. Yamada, et al., J. Chem. Soc. Perkin Trans. 2, 1998, 2, 701–710.

11. S. Sakamoto, M. Fujita, K. Kim, and K. Yamaguchi, Tetrahedron, 2000, 56, 955–964.

38. D. V. Dearden, C. Dejsupa, Y. J. Liang, et al., J. Am. Chem. Soc., 1997, 119, 353–359.

12. J. W. Larson and T. B. McMahon, J. Am. Chem. Soc., 1982, 104, 6255–6261.

39. Y. J. Liang, J. S. Bradshaw, R. M. Izatt, et al., Int. J. Mass Spectrom., 1999, 187, 977–988.

13. C. Spickermann, T. Felder, C. A. Schalley, and B. Kirchner, Chem. Eur. J., 2008, 14, 1216–1227.

40. S. C. Nanita and R. G. Cooks, Angew. Chem. Int. Ed., 2006, 45, 554–569.

14. H. Wang, E. N. Kitova, and J. S. Klassen, J. Am. Chem. Soc., 2003, 125, 13630–13631.

41. C. A. Schalley and P. Weis, Int. J. Mass. Spectrom., 2002, 221, 9–19.

15. K. V´ekey, J. Mass Spectrom., 1996, 31, 445–463. 16. R. C. Dunbar, Mass Spectrom. Rev., 1992, 11, 309–339.

42. R. G. Cooks, D. X. Zhang, K. J. Koch, et al., Anal. Chem., 2001, 73, 3646–3655.

17. J. Laskin and C. Lifshitz, J. Mass Spectrom., 2001, 36, 459–478.

43. R. R. Julian, R. Hodyss, B. Kinnear, et al., J. Phys. Chem. B, 2002, 106, 1219–1228.

18. L. Sleno and D. A. Volmer, J. Mass Spectrom., 2004, 39, 1091–1112.

44. M. Kohtani, J. E. Schneider, T. C. Jones, and M. F. Jarrold, J. Am. Chem. Soc., 2004, 126, 16981–16987.

19. J. Laskin and J. H. Futrell, Mass Spectrom. Rev., 2005, 24, 135–167.

45. A. E. Counterman and D. E. Clemmer, J. Phys. Chem. B, 2001, 105, 8092–8096.

20. C. A. Schalley, R. K. Castellano, M. S. Brody, et al., J. Am. Chem. Soc., 1999, 121, 4568–4579.

46. Z. Takats, S. C. Nanita, G. Schlosser, et al., Anal. Chem., 2003, 75, 6147–6154.

21. C. A. Schalley, T. Mart´ın, U. Obst, and J. Rebek Jr., J. Am. Chem. Soc., 1999, 121, 2133–2138. ˚ 22. N. K. Beyeh, M. Kogej, A. Ahman, et al., Angew. Chem. Int. Ed., 2006, 45, 5214–5218.

47. U. Mazurek, O. Geller, C. Lifshitz, et al., J. Phys. Chem. A, 2005, 109, 2107–2112. 48. C. Piguet, G. Hopfgartner, B. Bocquet, et al., J. Am. Chem. Soc., 1994, 116, 9092–9102.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

32

Techniques

49. Y.-R. Zheng and P. J. Stang, J. Am. Chem. Soc., 2009, 131, 3487–3489.

64. S. Campbell, M. T. Rodgers, E. M. Marzluff, and J. L. Beauchamp, J. Am. Chem. Soc., 1994, 116, 9765–9766.

50. M. Albrecht, S. Mirtschin, M. de Groot, et al., J. Am. Chem. Soc., 2005, 127, 10371–10387.

65. M. E. Hemling, J. J. Conboy, M. F. Bean, et al., J. Am. Soc. Mass Spectrom., 1994, 5, 434–442.

51. J. Griep-Raming, S. Meyer, T. Bruhn, and J. O. Metzger, Angew. Chem. Int. Ed., 2002, 41, 2738–2742.

66. X. H. Cheng and C. Fenselau, Int. J. Mass Spectrom. Ion Processes, 1992, 122, 109–119.

52. W. Jiang, H. D. F. Winkler, and C. A. Schalley, J. Am. Chem. Soc., 2008, 130, 13852–13853.

67. S. Campbell, M. T. Rodgers, E. M. Marzluff, and J. L. Beauchamp, J. Am. Chem. Soc., 1995, 117, 12840–12854.

53. W. Jiang, A. Sch¨afer, P. C. Mohr, and C. A. Schalley, J. Am. Chem. Soc., 2010, 132, 2309–2320.

68. T. Wyttenbach and M. T. Bowers, J. Am. Soc. Mass Spectrom., 1999, 10, 9–14.

54. W. Jiang and C. A. Schalley, Proc. Natl. Acad. Sci. U.S.A., 2009, 106, 10425–10429.

69. S.-W. Lee, H.-N. Lee, H. S. Kim, and J. L. Beauchamp, J. Am. Chem. Soc., 1998, 120, 5800–5805.

55. W. Jiang and C. A. Schalley, J. Mass Spectrom., 2010, 45, 788–798.

70. E. Ventola, K. Rissanen, and P. Vainiotalo, Chem. Eur. J., 2004, 10, 6152–6162.

56. C. A. Schalley, T. M¨uller, P. Linnartz, et al., Chem. Eur. J., 2002, 8, 3538–3551.

71. E. Kalenius, D. Moiani, E. Dalcanale, and P. Vainiotalo, Chem. Commun., 2007, 3865–3867.

57. M. Engeser, A. Rang, M. Ferrer, et al., Int. J. Mass Spectrom., 2006, 255, 185–194.

72. D. P. Weimann, H. D. F. Winkler, J. A. Falenski, et al., Nature Chem., 2009, 1, 573–577.

58. B. Brusilowskij, S. Neubacher, and C. A. Schalley, Chem. Commun., 2009, 785–787.

73. H. D. F. Winkler, D. P. Weimann, A. Springer, and C. A. Schalley, Angew. Chem. Int. Ed., 2009, 48, 7246–7250.

59. M. A. C. Broeren, J. L. J. van Dongen, M. Pittelkow, et al., Angew. Chem. Int. Ed., 2004, 43, 3557–3562.

74. C. A. Schalley (ed.), Analytical Methods in Supramolecular Chemistry, Wiley-VCH, Weinheim/Germany, 2007.

60. J. L. Brauman, in Kinetics of Ion-Molecule Reactions, ed. P. Ausloos, Plenum Press, New York, 1979, pp. 153–164.

75. J. M. Daniel, S. D. Friess, S. Rajagopalan, et al., Int. J. Mass Spectrom., 2002, 216, 1–27.

61. S. G. Lias, J. Phys. Chem., 1984, 88, 4401–4407.

76. K. M. Ervin, Chem. Rev., 2001, 101, 391–444.

62. P. Ausloos and S. G. Lias, J. Am. Chem. Soc., 1981, 103, 3641–3647.

77. P. B. Armentrout, Int. J. Mass Spectrom., 1999, 193, 227–240.

63. E. Gard, D. Willard, J. Bregar, et al., Org. Mass Spectrom., 1993, 28, 1632–1639.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc027

Supramolecular Electrochemistry Amar H. Flood1 and Angel E. Kaifer2 1 2

Indiana University, Bloomington, IN, USA University of Miami, Coral Gables, FL, USA

1 Introduction 2 Quantities Measured in Voltammetric Experiments 3 Voltammetric Experiments 4 Supramolecular Effects on Half-Wave Potentials 5 Diffusion in Supramolecular Systems 6 Host–Guest Complexation Under Redox Control: Thermodynamic Limits 7 Supramolecular Switches 8 Mechanically Interlocked Molecular Switches 9 Conclusions Acknowledgments References Further Reading

1

1 2 4 6 7 7 13 14 20 20 20 21

INTRODUCTION

The simplest class of chemical reactions is electron transfer (ET). Clearly, the easiest way to change the nature of a chemical substance is to add or remove one or more electrons. In spite of this simplicity, a reactant that undergoes ET yields a product with a different overall charge and, quite often, different polarity and electronic distribution. These are important attributes that determine the kind of noncovalent interactions in which any chemical species may

engage. Therefore, it is not surprising that ET reactions represent one of the most common means to alter supramolecular structure on demand. Although ET processes can be driven by chemical or photochemical (see Photochemically Driven Molecular Devices and Machines, Nanotechnology) methods, they can be easily carried out and investigated in electrochemical cells. Therefore, electrochemical methods are usually employed to understand and control the effect that ET reactions have on supramolecular systems. The idea of controlling supramolecular structure on demand via ET reactions was introduced about 30 years ago. Since then, supramolecular chemists have explored a large number of systems containing molecular residues that can exchange electrons with ease. Here, we refer to these as redox centers (Figure 1). Among the most common ones, quinones and viologens (4,4
-bipyridinium derivatives) are easily reducible, while ferrocene and tetrathiafulvalene (TTF) undergo oxidation at very accessible potentials. Many metal complexes also exhibit fast, readily accessible oxidation and reduction processes. In many instances, molecules containing redox centers are targeted and prepared simply because of the synthetic accessibility and/or structural properties of some of these redox centers, without any goal of using their ET properties to control supramolecular structure. Of course, the presence of redox units in a molecule immediately opens the possibility of using electrochemical methods as an additional group of techniques for characterization of the resulting structure in addition to nuclear magnetic resonance (NMR) spectroscopy (see NMR Spectroscopy in Solution, Techniques), mass spectrometry (see Mass Spectrometry and Gas-Phase Chemistry of Supermolecules: A Primer, Techniques), and crystallography. Supramolecular chemistry has now progressed to a point where very sophisticated and complex

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

2

Techniques O

O Quinone + N

S

S

S

e W

Ferrocene

molecular architectures are synthetically accessible (see Rotaxanes—Self-Assembled Links, Self-Processes). It is then natural to start pondering over questions related to the effect that supramolecular architecture exerts on the kinetics and thermodynamics of ET reactions of the components. At the core of all these issues, we find the main subject of this chapter, that is, electrochemical techniques and their applications in supramolecular chemistry. Given the large variety of electrochemical techniques available and due to space limitations, our discussion focuses on voltammetric techniques on account of their wide availability and extensive use in supramolecular chemistry. This choice means that we will not address the large body of literature on the development of potentiometric selective electrodes (see Ion-Selective Electrodes With Ionophore-Doped Sensing Membranes, Supramolecular Devices). By the same token, we will not be able to do justice to the quickly growing popularity of electrochemical impedance spectroscopic methods. Our intention is to illustrate how, using a tutorial style, voltammetric techniques can be applied to supramolecular compounds and mechanically interlocked molecules (MIMs) with examples taken from the recent literature that best highlight the concepts used in data interpretation.

2.1

R

A

Solution

∆E becoming more negative LUMO

Fe Anions

Figure 1 Some typical redox-active centers used often in supramolecular systems.

2

−

∆E

Tetrathiafulvalene + N

Viologen

Working electrode

Potentiostat

S

QUANTITIES MEASURED IN VOLTAMMETRIC EXPERIMENTS Physical apparatus and processes

Voltammetry is usually performed in an electrochemical cell (Figure 2a) fitted with three electrodes, which are referred to as the working, reference, and auxiliary (or counter) electrodes. The three-electrode configuration allows one to set (or scan) the potential difference (
E) between the working and the reference electrodes while at the same time measuring the current (i) passing through the closed circuit defined by the working and auxiliary electrodes, the electrolyte solution, and the potentiostat, which is the electronic apparatus used to control and carry out these

Fermi level

Cations Electrolyte solution (a)

∆ E becoming more positive

HOMO

(b)

Figure 2 Representations of electrochemistry experiments. (a) Schematic representation of a three-electrode electrochemical cell showing the working (W), reference (R), and auxiliary (A) electrodes and the current flow through the working and auxiliary electrodes. The situation shown assumes that a reduction takes place at the working electrode. (b) Relevant energy levels to rationalize ET across an electrochemical interface.

experiments. The use of the third auxiliary electrode to close the current loop with the working electrode is critical for accurate measurements. In its absence, the passage of current through the reference electrode would cause its polarization, leading to a shift in its potential and a moving and, thus, unreliable reference point to measure the potential difference. The reference electrode is usually connected to the potentiostat at a point of high internal resistance to assist in the diversion of current to other branches of the circuit. Maintaining a very low current level through the reference electrode avoids its polarization and insures a continuous, stable electrode potential against which all other potentials in the cell can be measured in a reliable way. Let us examine the effect that an applied potential may have on the working electrode and its interface with the electrolyte solution. Assume that we are applying an increasingly negative potential to the working electrode against the constant potential value of the reference electrode. In effective terms, this raises the Fermi level of the electrons in the working electrode (Figure 2b), eventually facilitating their transfer to energetically accessible molecular orbitals, typically the lowest unoccupied molecular orbital (LUMO), on any suitable species on the solution side of the interface. Therefore, negative potentials drive the reduction of solution-phase redox centers within a supramolecular compound. By the same token, applying an increasingly positive potential to the working electrode gradually lowers the Fermi level of the electrons (Figure 2b), until the transfer of electrons from suitable molecular orbitals (highest occupied molecular orbital, HOMO) on an appropriate redox-active supramolecular

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Supramolecular electrochemistry compound becomes energetically possible, resulting in the oxidation of the corresponding solute. These heterogeneous ET processes (electrode-to-LUMO or HOMO-to-electrode) give rise to a current flowing through the working electrode/solution interface, the so-called faradaic current. The potentiostat circuitry can measure this interfacial current, while ensuring that the same level of reverse current will flow at the auxiliary electrode/solution interface. For typical supramolecular systems, oxidation at the working electrode usually occurs for a small volume: that is, only a thin layer of solution adjacent to the electrode surface is affected. At the auxiliary electrode, the potential will change as required to allow any suitable solution species to become reduced, for example, solvent or the supramolecular compound. Ion migration will transport the current across the electrolyte solution, which is typically selected for solubility and to avoid interferences with the supramolecular system. Electron flow completes the current loop through the electronic apparatus. A key concept in electrochemistry is represented in Figure 2(b). It shows how the electronic Fermi level of the working electrode can be controlled as the potential difference (
E) against a reference electrode. Not surprisingly, a working electrode can be viewed as a tunable redox reactant. As we make its potential more positive, we increase its oxidizing power. Conversely, biasing its potential in the negative direction is a way of increasing its reducing power.

given Eappl value. Redox couples that meet this requirement are referred to as Nernstian or reversible couples, and their ability to adapt quickly to applied potential changes is the result of fast electrochemical kinetics, that is, fast heterogeneous (interfacial) ET processes. The common redox centers mentioned above (Figure 1) fulfill this criterion although interesting deviations can occur when they are encapsulated inside supramolecular compounds.1
For any Ox/Red redox couple, the formal potential (E 0 ) is intimately related to the standard potential (E 0 ) given in thermodynamic tables. The latter would require the use of activities (a) for all chemical species involved in the corresponding form of the Nernst equation: RT aOx ln nF aRed

E = E0 +

E = E0 +

RT RT fOx [Ox] + ln ln nF fRed nF [Red]

Consider a generalized redox couple Ox + ne− = Red, where Ox and Red are oxidized and reduced species, respectively, and where n represents the number of electrons exchanged. In this case, there is a formal potential
E 0 associated with the redox couple that is measured in a voltammetric experiment. At potentials on either side of
E 0 , the equilibrium concentrations of oxidized and reduced species are established at the working electrode according to the Nernst equation (1). In this form of the equation, the applied potential (Eappl ) is defined with respect to the
formal potential E 0 (rather than the reference electrode).


Eappl = E 0 +

RT [Ox]O ln nF [Red]O

(1)

Here, [Ox]o and [Red]o represent the equilibrium concentrations of the species Ox and Red at the working electrode surface and all other symbols have their usual meaning. If the interfacial ET processes are fast enough (reversible in electrochemical jargon), the concentration levels at the electrode/solution interface predicted by the Nernst equation will be reached instantaneously after the application of a

(3)

where fOx and fRed are the corresponding activity coefficients. Let us define the formal potential as the sum of the first two terms in (3).


Quantitative Nernst relationships

(2)

For convenience, the use of concentrations is much preferred in practice. When the activities are replaced by concentrations, we have

E0 = E0 +

2.2

3

RT fOx ln nF fRed

(4)

Since the activity coefficients are medium-dependent, the formal potential is constant for a fixed medium composi
tion. Thus, E 0 is the potential typically determined from
voltammetric experiments. If we use this definition of E 0 in (3), we can obtain


E = E0 +

RT [Ox] ln nF [Red]

(5)

which is similar to (1). In (5), the concentrations of species Ox and Red are assumed to be uniform throughout the solution in contact with the electrode, and E represents the equilibrium potential of the corresponding electrode. Thus, (5) is associated with potentiometric experiments in which equilibrium potentials are measured and reflects the concentrations present in the cell. Uniform concentrations are usually present, for instance, at the beginning of a voltammetric experiment. However, when a given potential (Eappl ) is applied to the working electrode during a voltammetric experiment, (1) describes how the corresponding concentrations at the electrode surface will change. Equation (1) can be seen as a voltammetric reading of the Nernst equation for reversible redox couples. Both readings of the Nernst equation are perfectly valid as long as the ET kinetics of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Techniques

redox couple is sufficiently fast. Thus, (1) and (5) are the Nernst relationships associated with the typical experiments conducted in the laboratory.

2.3

3

Current responses

In voltammetric experiments, the potential of the working electrode is changed with time according to a well-defined program, while the current is measured as a function of the applied potential. The current is simply the result of electrons (or any other charged particles, such as ions) flowing through a conducting medium. In electrochemical experiments, the current is always given by the following equation: i = nF

carrying out the experiments in quiet, unstirred solutions eliminates mass transport by convection.

dN dt

(6)

where dN is the number of moles or reactant converted in the electrochemical process. It follows that the current is a direct measure of the instantaneous reaction rate (dN/dt), and electrochemical techniques offer as a unique attribute the possibility of measuring instantaneous reaction rates as an integral part of the experiment. Currents given by (6) arise from faradaic processes, that is, interfacial oxidations or reductions of the supramolecular species of interest. Unfortunately, in voltammetric experiments one must also deal with currents caused by nonfaradaic processes. Most notably, any electrode/solution interface behaves like a capacitor and, every time the electrode potential is changed (as it is done continuously or almost continuously in voltammetric experiments), nonfaradaic currents flow to charge the electrode capacitance to the new potential value. These nonfaradaic currents—also called background, charging, or capacitive currents—cannot be completely filtered out or separated from the faradaic current components. However, there are methods that allow their minimization, thus maximizing our ability to detect and measure even low levels of faradaic currents. The magnitude of the faradaic current passing through a cell may depend not only on the rate of the electrochemical reaction but also on mass transport rates. Quite often, the overall rate of a fast electrochemical reaction depends on how quickly the reactant molecules or ions may reach the working electrode surface, in which case the faradaic current is limited by the corresponding mass transport rates. Voltammetric experiments are usually carried out under conditions in which the only operating mass transport mechanism affecting the reactant is diffusion. Other mass transport modes for the reactant are minimized by the choice of experimental conditions. For instance, migration is commonly minimized by the use of a supporting electrolyte, and

VOLTAMMETRIC EXPERIMENTS

While voltammetric experiments admit many variations, the general principle is always the same. The potential of the working electrode is changed as a function of time according to a well-defined E(t) function and the current is measured as a function of potential. Potential and time are intimately related according to the chosen E(t) function, and the voltammogram is just the current versus potential plot produced by the experiment. Often the E(t) function is designated as the excitation function, as it represents a way to temporarily perturb the equilibrium in the electrochemical cell. By the same token, the i(E) or i(t) curves represent the system response. Let us consider some of the important details in the most popular type of voltammetric experiment: cyclic voltammetry (CV).

3.1

Cyclic voltammetry

In CV, the electrode potential varies linearly with time. Figure 3(a) shows a typical excitation function used in this method. The initial potential Ei is selected in such a way that there is no faradaic current flowing at the electrode when Ei is applied. In CV experiments, the most common situation is that only one of the redox partners (either Ox or Red in the couple of interest) is initially present in the solution. If we assume that the oxidized form Ox is present, then we must select a sufficiently positive Ei value in order to prevent the reduction of Ox
at the starting point of the experiment, that is, Ei  E 0 . During the scan (Figure 3a, red trace), the potential moves in the negative direction, eventually driving the reduction of Ox. After the current peak corresponding to the reduction (−) (Epc, i pc)

4

Es

Current (µA)

4

E 0′ Ei

Forward scan

2 0 −2 −4

Reverse scan

(Epa, i pa)

t 0.0

(+) (a)

(b)

−0.4 Voltage (V)

−0.8

Figure 3 Cyclic voltammetry. (a) Excitation function in a typical cyclic voltammetric experiment with two potential segments. (b) A typical cyclic voltammogram for a reversible redox couple.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Supramolecular electrochemistry process Ox + e− → Red has been observed (Figure 3b, red trace), the current decays slowly. This behavior occurs as a result of the gradually increasing thickness of the diffusion layer: that is, the region adjacent to the electrode surface where the concentration of Ox has been diminished by electrochemical reduction to Red. This region of slowly decreasing current is usually referred to as the diffusional tail. Eventually, the scan direction is reversed at the socalled switching potential Es , and the potential starts to move in the positive direction. During this return sweep (Figure 3a, blue trace), oxidation of the reduced species (Red) that has accumulated near the electrode surface takes place, producing faradaic current in the opposite direction. Therefore, in a typical cyclic voltammogram (Figure 3b), we observe a peak of cathodic (reduction) current during the forward portion of the potential scan and an anodic (oxidation) current peak in the reverse scan. During the forward scan, the excitation function can be expressed by the simple linear equation E(t) = Ei –vt

(7)

where v is the scan rate (usually expressed in V s−1 or mV s−1 ). By the same token, the potential during the reverse scan is given by E(t) = Es + v
t, where the reverse scan rate (v
) does not have to be equal to the forward value (v), although they usually are identical. For a reversible (electrochemically fast) redox couple, the mid-point between the anodic and cathodic peak potentials Epa and Epc is the so-called half-wave potential E1/2 , which
is related to the formal potential E 0 for the Ox/Red couple by the simple expression


E1/2 = E 0 −

RT DOx ln 2 nF DRed

(8)

where DOx and DRed are the diffusion coefficients of the oxidized and reduced forms Ox and Red. Since the ratio DOx /DRed is typically very close to unity, the half-wave potential is generally accepted to be an excellent approximation for the corresponding formal potential. Therefore, CV offers an easy way to estimate the formal potential
(E 0 ) of a redox couple, assuming that the electrochemical process is fast enough. How can we assess the kinetics of the ET events at the electrode/solution interface? Fully reversible redox couples should yield cyclic voltammograms in which the potential difference (
Ep ) between the anodic and cathodic peak potentials is ∼57.0/n mV at 25 ◦ C. This small theoretical value is not easily reachable because it requires the switching potential (Es ) to be very removed from the peak potential observed in the forward scan. In practice, reversible redox couples yield
Ep values in the range 57–63 mV (for

5

n = 1 at 25 ◦ C). Larger
Ep values should be taken as an indication of slower electrochemical kinetics or substantial uncompensated resistance in the electrochemical cell. The peak maximum (ip ) observed in the current response of the CV for the forward peak is given at 25 ◦ C by the Randles–Sevcik equation ip = (2.69 × 105 )n2/3 ACOx (vDOx )1/2

(9)

where A is the geometric active area of the working electrode (in cm2 ), COx is the concentration of electroactive species (in mmol l−1 ), v is the scan rate in V s−1 , and DOx is the diffusion coefficient in square centimeters per second. The use of these units leads to ip values in microamperes and is required because (9) contains evaluated constants. The functional current dependence on the square root of the scan rate, equivalent to a t −1/2 dependence, reflects the fact that the current is under diffusion control. This equation also shows that the peak current is directly proportional to the concentration of species Ox, which provides the foundation for the analytical uses of this voltammetric technique. CV is a versatile technique that, among other things, affords an easy way to check the stability of electrogenerated species. For instance, if the reduction of Ox leads to a reduced species Red that decomposes quickly or reacts easily with the solvent or other solution components, the reverse anodic peak (corresponding to Red − e− → Ox) will disappear or exhibit decreased current levels compared to the forward cathodic peak. As exemplified later, this type of response can be used to identify if a host–guest complex gets more or less stable following a redox change, or for MIMs, if they switch between different states (see below and Molecular Devices: Molecular Machinery, Supramolecular Devices). In the absence of these interesting chemical changes, that is, if Red is stable in the timescale of the CV experiment, both peak currents should be identical within error margins. We must point out here that in this discussion we have focused on the case in which only the oxidized species (Ox) is initially present in the solution and the forward potential scan moves in the negative direction. The case in which only the reduced species (Red) is present in the solution at the start of the experiment would require a forward scan toward more positive potentials. All arguments presented here would be identical, as long as oxidation is replaced by reduction (and vice versa). A practical issue relates to the confusion that may arise at times because of the various ways in which voltammetric data can be plotted (positive potentials increasing toward the right or the left, cathodic or anodic currents plotted upward). However, a careful inspection of the x- and y-axes in the CV plot should dispel any uncertainties, as reductions

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

6

Techniques

are always triggered by potentials moving in the negative direction and oxidations are driven by increasingly positive potentials.

3.2

Square wave voltammetry

In CV experiments, the potential of the working electrode is always changing. As explained before, the electrode capacitance leads to the presence of background nonfaradaic currents (ibck ) at all times. Their magnitude is given by the simple equation ibck = vC, where C stands for the capacitance of the working electrode and v is the scan rate. As we increase the scan rate, all currents increase in magnitude, but background currents grow fasterthan faradaic currents (ip α ν 1/2 ). Therefore, the minimum current levels at which CV experiments can be performed, about 0.1–1.0 mM, cannot be lowered by running faster scan experiments. If material or solubility limitations impose lower concentrations of electroactive species, one possibility is to use square wave voltammetry (SWV), a technique that allows substantial removal of nonfaradaic currents. The E(t) excitation function of SWV is shown in detail in Figure 4(a). The overall potential function can be thought of as the superimposition of a square wave function with a stepwise function, both having the same period τ . After each cycle, the potential is shifted by a small amount
Es , which is known as the step potential. The pulse amplitude
EPu describes the magnitude of the square wave function. The effective scan rate is simply given by the ratio
Es /τ . The current is only measured at the end of each positive or negative pulse (points 1 and 2 in Figure 4a), where nonfaradaic currents are minimized because the potential remains stable for the duration of the pulse. The difference current (i1 –i2 ) is plotted versus the initial potential of each cycle to produce a square wave voltammogram (Figure 4b) in which each faradaic process gives rise to a single peak, whose peak potential ∆E s

(−)

10

E½

Current (µA)

1 ∆E Pu

Ei

5

0 2 (a)

t

−0.2

t

(b)

−0.4

−0.6

−0.8

Voltage (V)

Figure 4 Square wave voltammetry. (a) Excitation function used in square wave voltammetry. Points 1 and 2 (shown only in the first cycle) are those at which current measurements are made. (b) A typical square wave voltammogram for a reversible couple.

corresponds exactly to the half-wave potential (E1/2 ) of the redox couple. Typical step potentials in SWV experiments are in the range 2–5 mV, while pulse amplitudes vary from 25 to 50 mV. Scan rates in the range 50–100 mV s−1 are common, but faster scan rates are indeed possible. Usually, the experiment is run as a forward scan only (either in the cathodic or anodic direction) and the reverse scan is not recorded. Compared to CV, SWV peaks do not exhibit a diffusional tail. In fact, the current should return to baseline levels after each peak. This is one of the factors that leads in SWV to improved resolution of peaks characterized by close half-wave potentials. Due to space limitations, we do not discuss in detail the potential excitation function used in differential pulse voltammetry (DPV), which is a technique developed before SWV and uses the same principles for removal of background currents and yields voltammograms in many ways similar to those obtained with SWV. DPV is, however, an inherently slower technique, as scan rates are usually limited to a few millivolts per second. However, neither of these techniques provides a direct means to quantify any changes in the stability of a supramolecular complex or the motions in a molecular machine.

4

SUPRAMOLECULAR EFFECTS ON HALF-WAVE POTENTIALS

As mentioned above, half-wave potential (E1/2 ) values can be obtained directly in voltammetric experiments and constitute an excellent approximation to formal potential
(E 0 ) values—see (8). Supramolecular effects may change the formal potential for oxidation (or reduction) of a redox center. For instance, complexation of redox-active guests can result in measurable shifts of the guest’s formal potential, particularly when the redox center is located close to the primary binding site. The resulting potential shift can be usually interpreted in terms of differential stabilizations of the two forms of the redox couple. Let us consider a guest G that forms an inclusion complex with a host H and shows an accessible one-electron oxidation. A shift of the formal potential for the guest’s oxidation (G+ /G) to more positive values in the presence of the host reflects an increased thermodynamic hindrance to oxidation. This outcome means that the stability of the nonoxidized (reduced) guest is enhanced by its interactions with the host relative to the encapsulated oxidized form G+ . In other words, the H·G complex is more stable than the H·G+ complex. If the potential shift were in the opposite direction, the host H would stabilize the oxidized guest G+ more than for G. By the same token, shifts in the values of the formal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Supramolecular electrochemistry potential for the guest’s reduction G/G− can be interpreted in similar ways. As discussed in Section 2 (Figure 2b), the value of a
formal (E 0 ) potential depends primarily on the energy of the relevant molecular orbital (HOMO for an oxidation and LUMO for a reduction). However, the effects of binding described in the previous paragraph reflect the energetic influence of the surrounding environment on the stabilities of the two forms of a redox couple. Supramolecular effects, such as the formation of a host–guest complex, may change the value of a formal potential in a way similar to the welldocumented effects of solvation. Finally, we must point out here that, while a redoxactive guest is the most common situation, there are also examples in which the host is the redox-active partner. Similar arguments concerning differential stabilities of the complexes can be used to rationalize any shifts in the formal potentials of the host upon guest additions.

5

DIFFUSION IN SUPRAMOLECULAR SYSTEMS

The voltammetric behavior of any molecules capable of fast heterogeneous ET is usually controlled by diffusion from bulk solution to the electrode/solution interface. In principle, this means that characteristic currents in voltammetric experiments depend on the value of the diffusion coefficient DOx or DRed . For instance, in CV with diffusing electroactive species, peak currents vary linearly with the square root of the D value (9). Supramolecular interactions usually lead to pronounced changes in the diffusivity, as the formation of a host–guest complex, or any other from of molecular association, changes the effective molecular weight, hydrodynamic radius, and diffusion coefficient of the diffusing species. Taking again the case of a redox-active guest, the formation of a host–guest complex will lead to a noticeable decrease in the observed current levels observed for any redox couples associated with ET reactions on the guest. Voltammetric and other electrochemical techniques can be used to determine D values of redox-active species. For instance, in CV experiments the slope of a plot of peak current versus scan rate depends on (D)1/2 according to (9). However, in the last few years the popularity of NMR methods to determine D values has increased almost exponentially.2 NMR spectroscopy offers some advantages for this purpose, as the determination relies on the rate of signal intensity decay as a function of applied field gradient and does not require previous knowledge of the concentration of the diffusing species (see Diffusion Ordered NMR Spectroscopy (DOSY), Techniques). Unlike with electrochemical methods, the diffusing species do not have to be

7

redox active. However, care must be exercised when comparing D values obtained with electrochemical and NMR methods, as differences resulting from solution composition often are far from negligible. This situation arises from the fact that the viscosity and density of deuterated solvents, which are commonly used in NMR spectroscopy, differ from the values for the isotopically unenriched solvents used in electrochemical experiments. Furthermore, any ion pairing in the electrochemical experiment is enhanced on account of the supporting electrolyte that is usually present at a 100-fold excess, whereas in the NMR samples there is usually no added electrolyte (see Ion-Pair Receptors, Molecular Recognition). However, it is possible to account for these differences in a quantitative way.3

6

HOST–GUEST COMPLEXATION UNDER REDOX CONTROL: THERMODYNAMIC LIMITS

The effect of host–guest complexation on the observable CV response, and vice versa the ability for oxidation and reduction to change the stability of the complex, is best understood with a “square scheme” (Scheme 1). The equilibria running vertically are chemical (C) steps associated with complexation. Those equilibria running horizontally are the electrochemical (E) steps. The square scheme is a thermodynamic cycle such that when the two redox potentials (E1 and E2 ) are determined in a CV experiment and Ka is measured independently (e.g., using an NMR titration, see Binding Constants and Their Measurement, Techniques), it is trivial to determine the stability of the oxidized or reduced complex, Karedox . This concept will be illustrated with a variety of examples. The CVs for titration of host into guest will be simulated4, 5 to

E1 = −0.6 V G−

G

Ka (M−1) 105

+H

+H

Karedox 1.2 × 1010

H·G−

H·G

E2 = −0.3 V

Scheme 1 Square scheme representative of electrochemical (horizontal) steps and chemical (vertical) steps involved in the one-electron reduction of a guest G in the presence of a suitable host H. The specific K and E values correspond to Figure 5.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

8

Techniques

allow each case to be demonstrated. All the values used for the half-wave potentials (E1/2 ) and the stabilities (Ka ) are handpicked to illustrate these different situations. In all simulations, the concentration of the redox-active guest is 1 mM and the diffusion coefficients of the complexes are 70% of DRed . For the sake of simplicity, fast kinetics for the chemical and electrochemical steps will mostly be used, in order to place the system under thermodynamic control. Two alternative cases will be examined first, wherein either (Section 6.1, example 1) the complex is more stable when the guest is reduced, or (Section 6.2, example 2) the complex gets destabilized upon guest reduction. The latter case is taken to an extreme when examining example 3 (Section 6.3), where reduction drives decomplexation to generate a supramolecular switch. Finally, (Section 6.4, example 4) kinetic effects are examined in this switching system.

6.1

Example 1: Host–guest complex is more stable when guest is reduced

One of the simplest cases (Scheme 1) is considered first where titration of the host into a solution of the guest generates smooth changes in the CVs (Figure 5). Initially, the guest is reduced at E1 = −0.6 V (Figure 5, red trace) and, when the guest is bound by the host, the reduction becomes easier by 300 mV, showing a redox process centered at E2 = −0.3 V (Figure 5, blue trace). The H·G complex has a high stability Ka = 1 × 105 M−1 (
Ga = −28 kJ mol−1 ). The formalism of the square scheme (Scheme 1) and the relationship
G = −nF E

(10)

−0.63

−0.33

Current (µA)

20 10 0 −10 −20

−0.57

−0.27 0.0

−0.2

−0.4

−0.6

−0.8

−1.0

−1.2

Voltage (V)

Figure 5 Host–guest complex gets stable upon reduction: simulated CV titration corresponding to the square scheme shown in Scheme 1 recorded at a scan rate of 0.1 V s−1 starting with the guest G (red CV) and upon addition of host (blue CV; 0, 0.25, 0.5, 0.75, 1, 2, and 5 equivalents).

allow the stability of the reduced complex (Karedox ) to be calculated. The difference between E1 and E2 , for the n = 1 process, dictates from (10) that the stability of the reduced complex is enhanced by 29 kJ mol−1 (300 mV) such that
Gredox = −57 kJ mol−1 and Karedox = a 10 −1 1.2 × 10 M . The large Ka and Karedox values ensure that titration of the host into the guest will directly form H·G and that reduction of the guest enhances the stability of the H·G− complex. Therefore, during the CV titration, the cathodic and anodic peaks for reduction of the uncomplexed guest centered at −0.6 V decrease concomitant with an increase in the redox peaks for the complex H·G seen at −0.3 V. This is exactly what is observed (Figure 5). Experimentally, such a square scheme (Scheme 1) can be deduced from a real CV titration. The fact that the guest’s reduction gets easier is indicated by a shift toward more positive potentials. For oxidations, the shift would be toward more negative potentials. This redox-shift behavior definitively indicates that the complex gets more stable after the ET. Therefore, it is fair to use the formal redox potentials observed from the CVs as the true thermodynamic
values in the square scheme: that is, E 0 . The H·G stability can be independently measured using typical methods, allowing the stability of the reduced (or oxidized) complex (i.e., Karedox ) to be calculated from the three measured parameters (Ka , E1 , E2 ). Real examples of this behavior can be seen in the reduction of ligands upon formation of coordination complexes. One example is in the self-assembly of a Cu(I)based pseudorotaxane6 (Figure 6), where a tetrazine-based ligand (colored red) is reduced (red trace in CV). The complex with the copper-macrocycle host is very stable (
Ga = −44 kJ mol−1 ) and, during the titration, the ligand’s reduction (blue trace) shifts to more positive potentials by 225 mV = 22 kJ mol−1 . Thus, the reduced form of the pseudorotaxane has a binding constant of Karedox = 400 × 109 M−1 (
Ga = −66 kJ mol−1 ). Other examples of this behavior include the guest cobaltocenium and cyclodextrin (CD) hosts (Scheme 2 and see Cyclodextrins: From Nature to Nanotechnology, Molecular Recognition). The cationic cobaltocenium does not bind to β-CD. Once reduced, however, the binding is turned on.7 Under the thermodynamic limit, this CV would deviate from that shown in Figure 5 because only one of the two complexes is stable. This case is made more complicated by the fact that the β-CD complex of reduced cobaltocenium does not undergo heterogeneous one-electron oxidation in the experimental timescale.7 While this would be an interesting extension of Example 1 (Section 6.1), such enhanced binding following redox chemistry is not as common as the opposite: that is, redox chemistry that destabilizes the host–guest complex.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Supramolecular electrochemistry

+ PF6−

N

O

N C N N

Re

N

+

O

O

C Cl C O O

O N C N N

N

O

O

O

O

O

O

Re C Cl C O O O

O

O

O

−205 mV

30

N

N N

Cu

60

Current (µA)

N N

−420 mV

+ PF6−

NN Cu

N N

0

−30 −45 mV

Guest HO

−255 mV

−60

HO

Host

9

−0.8 −0.4 0.0 Potential (V) (vs Ag/AgCl)

0.4

Complex

Figure 6 Complex formation between a copper-macrocyclic “host” and a redox-active bispyridyl-tetrazine “guest”. The set of CVs (20 V s−1 ) on the right correspond to the titration of up to 1.05 equivalents of the host to the guest (1.0 mM). (Reproduced with permission from Ref. 6.  Wiley-VCH, 2010.)

Co +

Co

E1/2 = −1.15 V vs SCE

β-CD

Very weak complex

β-CD

Ka 2000 M−1

Co

Scheme 2 Relevant chemical and electrochemical equilibria for the reduction of cobaltocenium in the presence of β-cyclodextrin (β-CD).

6.2

Example 2: Complex is destabilized when guest is reduced

Destabilization of complexes upon redox stimulation tends to occur for a very practical reason. The redox center is usually involved in the specific noncovalent interaction that holds the complex together. Therefore, a fundamental change in its redox state removes some or all of the attractive interaction, making the complex unstable with respect to complexation: that is, a net attraction is transformed into a net repulsion. At an intellectual level, this behavior has led to the creation of some exciting MIMs for use as artificial molecular machines (see Molecular Devices: Molecular Machinery, Supramolecular Devices and Photochemically Driven Molecular Devices and Machines, Nanotechnology). The first case examined here is the simplest and helps to reinforce the thermodynamic relationships in the square scheme. Here we assume a very large stability of the H·G complex (Ka = 1010 M−1 ) and that reduction of the guest

destabilizes the complex. Therefore, during the titration and formation of the H·G complex, the guest’s reduction is seen to occur at more negative potentials (Figure 7, blue trace). In the case where the reduction is shifted 300 mV ∼29 kJ mol−1 (Figure 7), the new association constant is lowered to Karedox ∼ 106 M−1 . During the titration, the peaks for the reduction of the guest (red trace) decrease in intensity concomitant with an increase in those for the complex (blue trace). The large association constants for the H·G complex and for its reduced form H·G− lead to tight binding even at 1 mM. For Karedox = 106 M−1 , the degree of complexation at 1 equivalent is 96%. This is an important point: as long as the stability of the reduced complex H·G− is strong enough to retain tight binding, a simple monotonic change between species G and H·G will be observed together with isocurrent points, which have the same meaning as isosbestic points in spectroscopic titrations. A real example in which both forms of the redox guest are bound strongly by the host is afforded by −0.63

20 Current (µA)

+ e−

−0.93

10 0 −10 −20

−0.87

−0.57 0.0

−0.2

−0.4

−0.6

−0.8

−1.0

−1.2

Voltage (V)

Figure 7 Complex gets modestly destabilized upon reduction: simulated CV titration under conditions when redox chemistry weakens the host–guest binding from 1010 to 106 M−1 : 0 equivalent (red CV) to 5 equivalent (blue trace) at 0.1 V s−1 .

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

10

Techniques

N+

N+ − e−

Fe

K ~1012 M−1

Fe +

E1/2 = 0.43 V (vs Ag/AgCl)

CB7

CB7

+

K ~1010 M−1

+

N

N − e−

Fe

+

E1/2 = 0.54 V (vs Ag/AgCl)

Fe

Scheme 3 Relevant chemical and electrochemical equilibria for the oxidation of ferrocenemethyltrimethylammonium in the presence of the cucurbit[7]uril (CB7) host.

some ferrocene-containing guests, such as ferrocenylmethyltrimethylammonium (FcN+ (CH3 )3 ), and the cucurbit[7]uril (CB7) host (see Cucurbituril Receptors and Drug Delivery, Molecular Recognition).8 As illustrated in Scheme 3, the initial complex is extremely stable (Ka ∼ 1012 M−1 ). The half-wave potential for the oxidation of the guest shifts to more positive values by about 110 mV in the presence of 1.0 equivalent of CB7. This means that one-electron oxidation of the guest decreases the stability of the complex, although its stability is still quite high (Karedox ∼ 1010 M−1 ). We can also examine the case (Figure 8) where a 100 mV shift changes the affinity from Ka = 105 M−1 to a moderate value of Karedox = 2 × 103 M−1 . In this instance, the speciation curve (Figure 8a) shows that, upon reduction to G− , only 50% is complexed as H·G− (blue trace) upon addition of 1 equivalent of guest with the other half present −0.63

% 100

1 Equivalent H·G

−

H·G 50

Current (µA)

20

−0.73

10 0 −10 −0.57

−20

−0.67

0 (a)

0 1 2 3 4 5 [H] (mM) (b)

0.0

−0.4 −0.8 Voltage (V)

Figure 8 Complex shows some dethreading upon reduction. (a) Simulated speciation curves for the binding affinity of the host–guest complex before (K = 105 M−1 ) and after reduction (K redox = 2 × 103 M−1 ). (b) Simulated CV titration (0, 0.25, 0.5, 0.75, 1, 2, and 5 equivalents) of host into redox-active guest under conditions where the complex is destabilized to a small extent upon reduction to G− (0.1 V s−1 ).

as G− . Thus, the CV titration data (Figure 8b) show that host–guest saturation does occur until after the addition of an excess of the host (5 equivalents = 94% complexed). During the titration, the peaks appear to shift their positions (red to blue trace). This effect arises from the overlapping of the original peaks decreasing in intensity relative to the new ones increasing.

6.3

Example 3: Supramolecular switching: guest reduction drives decomplexation

This example of host–guest complexes is more interesting for the purposes of switching and for the ultimate construction of molecular machines. In all the cases examined above, the host–guest complex remains intact for both the unreduced and reduced states; only the position of equilibrium shifts. Thus, at the end of the titration, whether or not that occurs with ∼1 equivalent (Figures 5–7) or with a large excess of added host (Figure 8), the observed CV represents the properties of the complex’s redox couple, H·G + e− = H·G− , that is, the amount of G or G− in solution is negligible. In contrast to this straightforward situation, complexes can be weakened so substantially that the product of the ET step is the unbound form of the reduced guest: that is, G− rather than the reduced complex H·G− . These are the cases explored here. In this first situation (Figure 9a), we simulate the effect of a 500 mV shift in the guest’s reduction from −0.6 V to −1.1 V for the guest in H·G. This generates a driving force of 48 kJ mol−1 , which essentially eliminates the high binding affinity of the complex (from 100 000 M−1 to Karedox = 0.0003 M−1 ). As illustrated in the CV titration (Figure 9b), again, new cathodic and anodic peaks grow during the addition of up to 5 equivalents of host (blue trace) concomitant with the loss in the parent redox process (red trace). However, there are key differences to all the prior examples. First, loss of the isocurrent points is consistent with multiple species being present in solution: that is, we now have species G− being formed from H·G− with the possible involvement of the decomplexed species H + G (upper left position in the square scheme in Figure 9a). Second, the emergence of the new cathodic peak at −0.86 V is different in character to that for the new anodic peak at −0.67 V. The shape of the CV titration reflects the mechanism of supramolecular switching. In the CV, the cathodic reduction peak at −0.86 V is shifted 230 mV compared to where it started and it simply decreases in peak intensity. By contrast, the anodic oxidation peak displays a shifting in its position from −0.57 to −0.67 V (only 100 mV). Neither of these new peak positions corresponds to the value of −1.1 V that was employed in the simulation. This situation

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Supramolecular electrochemistry −0.6 V

H+G 100 000 M−1

Ka

H + G−

K aredox

H·G

H·G−

−1.1 V

(a)

0.0003 M−1

−0.63

−0.86

20

I (µA)

10 0 −10 −0.67

−20

−0.57

(b)

−0.89

6 3

30

0

0

−3

−0.64 0.0

(c)

60

−0.83

−0.4

−0.69 −0.8

−30

−1.2

Voltage (V)

Figure 9 Supramolecular switching under thermodynamic control. (a) Square scheme for the reduction of G in the presence of host H. The selected thermodynamic parameters correspond to a strong decrease in complex stability upon reduction of the guest. (b) Simulated CV titration corresponding to increasing concentrations of host, from 0 equivalent (red trace) to 5 equivalent (blue trace). (c) Simulated CVs showing the dependence of peak potentials on scan rate: 0.01 V s−1 (magenta) and 10 V s−1 (green).

arises from the mechanism depicted in the square scheme (Figure 9a): Equal concentrations of H·G and H·G− can never be attained at the electrode surface at Eap = −1.1 V on account of the thermodynamically unstable character of H·G− with respect to rapid and complete decomplexation into H + G− . The same logic applies for the oxidation peak in the reverse scan. Recall that the exact position of each peak potential depends on the concentrations of each electroactive species as shown from the Nernst equation (5). Therefore, the cathodic peak (Ep,c = −0.86 V) for reduction of the H·G to ultimately generate G− is almost at the numerical average of the two reductions on account of rapid chemical kinetics from H·G− → H + G− . This behavior is not observed for the anodic peak in the return sweep, and the explanation for it provides a more general account of the phenomena expressed in the square scheme (Figure 9a).

11

First, two pathways can be accessed to transform between the unreduced starting material H·G and the reduced product G− : Pathway 1 is E2 then Karedox and pathway 2 is Ka then E1 . Second, the two pathways always have different rates. Consequently, the peak positions cannot always be correlated with the formal redox potential E2 because the concentrations of electroactive species at the electrode surface are almost never at equilibrium. Another consequence is the observation that peak positions change with scan rate (Figure 9c), whereas in all the former cases, the positions were largely invariant with the scan rate. Even though the rates of the reactions are fast enough to achieve thermodynamic control, the presence of two pathways cause systematic changes in the concentrations which shift the peak positions around. In such cases, the titration data and the scan-rate dependence need to be simulated in order to obtain estimates of E2 and, consequently, of Karedox . A classic representation of this behavior is provided from the oxidation-induced decomplexation9 (Figure 10) of a pseudorotaxane. This example considers oxidations instead of reductions and illustrates the generality of the approach used here for the interpretation of CVs. In this case, the guest is a thread-like molecule T-1 based on the strong electron donor TTF (colored green). It forms a stable charge-transfer complex when inside the cyclobis(paraquatp-phenylene) (CBPQT4+ , colored blue) ring on account of the electron acceptor properties of the two paraquat units. A binding constant of 50 000 M−1 (RT, CH3 CN) ensures 95% complexation under the conditions in which the CV was recorded (0.5 mM in a 1 : 1 ratio). The oxidation CV is different from the picture in Figure 9(b) because there is an additional oxidation of the TTF thread rather than a single redox process. The CV of the pseudorotaxane T-1 ⊂ CBPQT4+ (Figure 10, blue trace) is analyzed by comparison with the TTF thread alone (green trace). The first oxidation peak at +0.66 V is shifted significantly to more positive potentials compared to T-1 (Epa = +0.38 V). This +280 mV shift is indicative of destabilization of the complex upon formation of the oxidized monocation guest, TTF+ ; that is, T-1+ ⊂ CBPQT4+ is destabilized electrostatically by 27 kJ mol−1 such that Karedox ∼ 1 M−1 . The subsequent redox process to form the dication TTF2+ at E1/2 = +0.71 V is not shifted when compared to the free thread. Finally, the reduction wave at +0.44 V for the reformation of the neutral thread in the reverse scan is shifted slightly from the free thread. The behavior of the first redox process (with an oxidation peak at +0.74 V and reduction at +0.44 V) is identical to the behavior observed in Figure 9(b) (blue trace), even though a titration was not performed. Moreover, the authors state that the peak positions change with scan rate such that they get closer together at slower scan rates, which is exactly the same behavior seen in Figure 9(c). Thus, the behavior of the first

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

12

Techniques OH

TTF

(E ½ = 0.35 V vs SCE)

−

O O

O

O

S

T-1+

O

O

S

S

O

+N

−e

thread

O

N

OH

T-1

O

S

+0.74

O

O

S

S

S

S

O O O

+0.38

O

O

+

HO

HO

+0.74

5 × 104 M−1

+CBPQT 4+

+0.31

N+

−CBPQT 4+

+0.66

+N N

OH + N

+ O

O

S

O S

O O

S O

+0.44

O

N+

O − e−

+N

O O S

O

HO (Epa = 0.66 V vs SCE)

+

S

O

O S

N

OH + N

O

O S

S

+N

N+

O HO

T-1+ ⊂ CBPQT4+

Pseudorotaxane T-1 ⊂ CBPQT4+

Figure 10 Relevant processes corresponding to the oxidation-induced dissociation of a pseudorotaxane that contains a TTF station as the main component of the thread and the tetracationic cyclophane CBPQT4+ as the macrocycle. The CVs corresponding to the free thread (green trace) and pseudorotaxanes (blue trace) are shown in the center of the figure. (Reproduced from Ref. 9.  Wiley-VCH, 1997.)

redox process indicates that the oxidized thread T-1+ is unstable inside the guest and decomplexation occurs. The fact that the guest’s second oxidation is identical to the free thread is further confirmation of this dethreading behavior.

6.4

Example 4: Kinetic control of supramolecular switching

Under the right conditions, the rate of decomplexation from the stimulated complex to the decomplexed forms,

that is, H·G− → H + G− , can be observed using CV. For this to happen, the timescales of the switching must lie between 100 ms and 10 s for a standard room temperature experiment. Here we resimulate the previous example (Figure 9) but with a unimolecular decomplexation process where the rate constant is decreased from k = 3 × 109 s−1 to k = 3 s−1 . As can be seen from the CV titration (Figure 11a) simulated at 0.1 V s−1 , there is the expected decrease of the parent peaks (red trace): however, only one peak for the complex H·G emerges at −1.13 V. There is

−1.13

Current (µA)

Current (µA)

20 10 0 −10

−0.57 0.0

−0.4

−1.2

(b)

200 100

0

−20 −0.8

−1.13

10

−10

−20 (a)

20

−0.63

0 −100

−0.57 −1.07 0.0 −0.5 −1.0

Voltage (V)

Figure 11 Supramolecular switching under kinetic control. (a) Simulated CV titrations of host (0, 0.25, 0.5, 0.75, 1, 2, 4, 5 equivalents) into redox-active guest at 0.1 V s−1 . (b) Different simulated CVs showing the scan-rate dependence of the relative peak intensities. The two traces shown correspond to 0.01 V s−1 (magenta) and 10 V s−1 (green). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc028

Supramolecular electrochemistry Log K = 4

SUPRAMOLECULAR SWITCHES

Now that it is possible to measure (or estimate with simulations) the host–guest binding constants of the redox center in two oxidation states, one has the basis for supramolecular switches and molecular machines. In such situations, the CV is able to both stimulate the switching by oxidation or reduction and to analyze the outcome. A switch can be as simple as a two state system: for example, H·G + e− → H·G− . However, it is the ensuing chemical step, that is, H·G− → H + G− , which allows the stimulated state (H·G− ) to potentially do some work (
G). What is unique to systems held together by noncovalent interactions is the ability to achieve motion during this chemical step. A supramolecular switch that leads to the destabilization of the host–guest complex pushes the position of equilibrium from one side of an association reaction to the other, in which the CVs simulated above in Figures 9 and 11 are exemplary. All that is required to make a redox-stimulated supramolecular switch is (i) confirmation that the redox change weakened (or strengthened) the complex by an appreciable amount such that (ii) conditions can be selected (see below) to harness the change in stability. A typical switch effects at least a 90% population inversion, that is, from 10 : 1 to 1 : 10. For host–guest complexes, this requirement usually translates into a problem of selecting a concentration where the population of the starting state, for example, H·G, is 10-fold different from the switched state, for example, H + G− . While a ratio of the equilibrium constants Ka /Karedox of 10 might seem appropriate, recall that host–guest equilibria are also very sensitive to concentration (see The Thermodynamics of Molecular Recognition, Concepts). For instance, consider how dilution of a simple 1 : 1 mixture with Ka = 104 M−1 (Figure 12a) from 10 mM to 10 µM shifts the position of equilibrium from 95 to 8%. Bearing this concentration dependence in

log K = 4

1 0.1

50

50 log K = 2

0.01 0 (a) 0

1

2

0 (b) 0

log K = 6

100

1

2

[G] = 0.01 mM

[G] = 1 mM

log K = 6

100

log K = 4

50

50

log K = 4

0 (c) 0

7

[G] = 1 mM

[G]

10 mM 100

100 % H·G

no reoxidation peak in the return sweep; instead the peak at −0.57 V for G− → G is retained at a smaller intensity. These features are observed because at this slow scan rate the reduced complex H·G− has had enough time to switch back into its components H + G− , thus allowing G− to become oxidized to G at −0.57 V. However, when the scan rate is increased (Figure 11b) from 0.01 V s−1 (magenta trace) to 10 V s−1 (green trace), the relative intensity of the peak at −1.07 V, which corresponds to the oxidation of the reduced complex H·G− − e− → H·G, grows in. In contrast to the case under thermodynamic control (Figure 9), the slow step in the square scheme can be frozen out and, thus, there is no change in the peak positions with scan rate.

13

1

2

0 (d) 0

1

2

Equivalents of H added

Figure 12 Simulated speciation curves showing the fraction of complex formed under various conditions as noted in the panels. (a) Dilution effects with K = 1 × 104 M−1 . (b and c) Effects of the K value. (d) Effect of the K value at lower concentrations of guest and host.

mind, when the ratio of equilibrium constants is greater than 100-fold, a supramolecular switch may be more able to achieve the 10-fold population inversion. For example, with an initial binding constant of Ka = 10 000 M−1 (log K = 4) that weakens to Karedox = 100 M−1 (log K = 2, Figure 12b), the population of the complexed state at a 1 : 1 molar ratio of host and guest inverts by a factor of 9 (= 72%/8%). Again, the concentration used to show switching should match the overall binding affinities. Thus, if the overall binding affinity of the H·G and H·G− complexes are both enhanced by two orders of magnitude to log K = 6 and 4, respectively, there is a negligible population inversion of 1.3 (= 97%/72%) at 1 mM (Figure 12c). If the concentration of the system is correspondingly diluted by the same two orders of magnitude (Figure 12d), the 10-fold population inversion can be reinstated. Often, a particular concentration is required for the purposes of understanding the switch’s operation. To achieve good quality CV data, 1 mM concentrations are typical. However, the practical limits start at 5 mM where the currents become saturated by mass transfer. At the lower limit of 0.1 mM, nonfaradaic processes tend to dominate. For fundamental investigations, this concentration range is an ideal match to NMR (>1 mM) and UV–vis (300 nm; G stands for giant and S for small). The use of LUVs is recommended for supramolecular chemists entering the field, because they do not require specialized equipment, provide a rapid overview on various activities, and

offer numerous possibilities to dissect most of the relevant characteristics of ion channels, pores, and carriers. Highly reproducible standard procedures to prepare LUVs of defined size and membrane composition exist and various tutorials can be found in the literature.25 For routine use, LUVs prepared by freeze–thaw–extrusion techniques are usually preferable,16 whereas advanced dialytic detergent removal methods (rotating chamber) can be best when highest LUV quality is required. The nature, purity, and age of the phospholipids used influence LUV quality significantly.

3.1

Fluorescence spectroscopy with labeled vesicles

Among the different methods to characterize supramolecular transmembrane transport systems, fluorescence methods are most straightforward to implement and feature a high sensitivity. This enables the rapid screening of numerous activities under varying conditions. To characterize the transmembrane activity of synthetic transport systems by fluorescence spectroscopy, vesicles that are labeled with one or more fluorescent probes need to be prepared. Labeled vesicles can be equipped with internalized (PI ) and membrane-bound (PM ) probes during their preparation or the fluorescent probe can be added externally to the preformed vesicles (Figures 5 and 6). Examples of internal probes include HPTS, 8-aminonaphthalene-1,3,6trisulfonate (ANTS), CF, lucigenin, and calcein (Figure 7). These probes can be complementarily used to dissect certain specific transmembrane transport activities. For example, the pH-sensitive dye HPTS can report on proton transport,20, 26 lucigenin is sensitive toward halide anions27 and Addition of transporter

Vesicles destroyed

Symport

Internal External

Anion antiport

PM PI

X1−

X2− Cation antiport

PE M1+

(b)

(a)

Figure 5

I MAX (or v0(MAX)) Fractional activity Y I MIN (or v0(MIN)) I0 Time (s)

M2+ Ion transporter

Inside

I∞

Fractional activity Y

Membrane bound

X– + M+

(Calibration) It Emission/absorption intensity

P = fluorescent probe

Outside (c)

1 0.5

EC50 (nM)

0 Concentration transporter

(a) General configuration, (b) typical result, and (c) data analysis of vesicle flux experiments.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes

(a)

Intervesicular transfer

(b)

It

Addition of transporter

7

Addition of more vesicles

It

Time (s)

Application of gradient

Application of gradient

(c) All-or-none behavior originates from irreversible partitioning

Addition of transporter

Time (s)

It

(d)

Addition of transporter

(e) Ion transport is faster than detection limit Time (s)

Figure 6 (a) Kinetics in vesicle transport experiments can show either continuous (dashed) or all-or-none (solid) behavior, originating (b) from the ability (dashed) or inability (solid, c) of the synthetic transport system to transfer intervesicularly into newly added or previously unoccupied vesicles. (d) Inversion of sequence of addition from transport system to vesicles with ion gradients (dashed) to addition of ion gradients to vesicles with transport system (solid) shows that kinetics commonly report on the formation of the active system rather than direct ion transport which is far beyond detection limits (e). − O3S

OH N +

PI

SO3 −

O3S −

Probe efflux Anion antiport

HPTS (pyranine) X−

+ H3N

OH − Cation antiport

O3S −

+ N

− SO 3

HO

CF − −

N

OOC

N

N O

−

COO

−

O

+

M

Ion transporter

Inside

COO

OOC HO

H

−

− OOC

− SO3

+ +

O COO

ANTS

DPX

PI +

O

Outside

N

+

Lucigenin

COO

−

Calcein

Figure 7 Selected examples for internal probes for functional and structural studies on ion transport in membrane with details for the HPTS assay.

calcein toward calcium, and all probes can report on their own export. To quantitatively evaluate any activity exerted by a synthetic molecule, the fluorescence of the probe is recorded in a time-dependent manner during addition of the molecule to be investigated. A typical series of experiments is shown in Figure 5(b). Commonly, a constant initial fluorescence

I0 is recorded at the beginning of each experiment. Next, addition of various concentrations of the transport system leads to a change in fluorescence, which may show either an increase or a decrease depending on the type of experiment and the specific conditions. Lastly, a detergent like Triton X-100 is added at the end of each experiment to obtain the emission of the free fluorophore I∞ . Traces of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

8

Techniques

time-dependent fluorescence intensity It are normalized by subtraction of I0 and division by (I∞ − I0 ) such that the normalized fluorescence changes from 0 to 1 during the experiment. If necessary, the relevant range of intensities is then further narrowed down to IMAX and IMIN , the maximal and minimal values obtained without and with excess transporter before vesicle lysis. These values are set to 0 and 1 and define the fractional activity Y of the synthetic transport system. Alternatively, the activity may be quantified by determination of the initial rate v0 . From a plot of Y against the concentration, the minimal and maximal detectable activities Y0 and Y∞ as well as the EC50 , and Hill coefficient n can be determined via fitting with the Hill equation (Figure 5c, Section 4.2). Fluorescence kinetics for transport in vesicles can show either continuous or all-or-none behavior. Continuous behavior refers to a gradual change in fluorescence intensity that reaches completion with time (Figure 6a, dashed line). All-or-none behavior is characterized by an initial burst that, however, stops halfway and never reaches completion (Figure 6a, solid line). All-or-none occurs when an overly hydrophobic transport system partitions into some vesicles only and cannot transfer into the vesicles that remain untouched by the initial attack. Continuous behavior thus occurs with more hydrophilic transporters that are capable of reversible partitioning and intervesicular transfer. Validity of this interpretation can be demonstrated by intravesicular transfer experiments (Figure 6b).28 In this assay, new fluorescently labeled vesicles are added at the end of an experiment. More hydrophilic synthetic transport systems with continuous behavior will jump to the new membranes and cause an additional gradual increase in fluorescence (Figure 6b, dotted). Too hydrophobic all-or-none transport systems will be incapable of intervesicular transfer (Figure 6c). The gradual fluorescence change in transport experiments with labeled vesicles either directly reports on the kinetics of transmembrane ion transport or refers to the kinetics of formation of the functional system. To dissect these two possibilities, the sequence of addition can be reversed in transport assay in which specific gradients can be applied. For example, the sequence of addition in the HPTS assay (Section 3.1.1) can be reversed from application of pH gradient before transporter addition to transporter addition before application of pH gradient.29 If both curves are identical, ion transport is the rate-limiting step, but in most cases it has been observed that ion transport is much faster. If this is the case, addition of the transport system after application of the pH gradient will cause a relatively slow, continuous change in fluorescence (Figure 6d, dotted line) while application of the pH gradient after incubation with the transporter leads to a jump in fluorescence (Figure 6e, solid). Such a burst of ultrafast ion transport far beyond the time resolution of standard fluorescence spectrometers is

observed because the gradient will find a transport system ready for work, that is, self-assembly of the transport system has already occurred without being seen by the fluorescent probe. The remaining part of the kinetic trace is ideally independent of the sequence of addition, since the usual complex cocktail of processes should have reached the same stage after the same time.

3.1.1 HPTS and lucigenin assay The HPTS (or pyranine) assay is the ideal assay to characterize new synthetic transport systems because it is the least selective assay and produces a signal for most transport mechanisms (Figure 7).26 HPTS is a pH-sensitive fluorescent dye with a pKa of around 7.3. The emission spectrum of HPTS is rather insensitive toward changes in pH owing to rapid and complete photodissociation in the excited state. The excitation spectrum exhibits two maxima at 404 nm for the protonated and at 454 nm for the unprotonated form, with an isosbestic point at 416 nm. This permits the ratiometric (i.e., fluorophoreconcentration-independent) detection of pH changes in double-channel fluorescence measurements. For the HPTS assay, LUVs are loaded with HPTS and exposed to a pH gradient. A transporter added to the system may catalyze the collapse of this gradient by transporting either proton or OH− . The principle of the HPTS assay in detecting ion transport activity relies on the presumption that membrane transport is commonly electroneutral. This implies that translocation of positive or negative charges across the membrane is compensated by an additional transport of ions of the same charge but in opposite direction, for example, H+ /M+ antiport or OH− /X− antiport, or alternatively by transport of ions of opposite charge but in same direction, for example, H+ /X− symport or OH− /M+ symport (Figure 7). Unidirectional symport is commonly thought to be less favorable because it leads to an osmotic imbalance. The HPTS assay will also report the efflux of HPTS itself as pH gradient collapses, but this event can be readily identified by comparison of kinetics from assays dedicated to the detection of large pores and more dramatic damage (Section 3.1.2). Failure to detect activity may be traced back to a very high selectivity of the synthetic transport system. For example, a highly selective K+ carrier would not transport a proton or hydroxide ion which would be necessary to detect its activity by the HPTS assay (see Section 4.4 for a modified version of the HPTS assay). Even less intuitively, the activity of a proton transporter may not fully be detected by the HPTS assay, because proton transport will stop owing to the opposing membrane potential that has built up (Section 4.5). The use of the HPTS assay to detect ion selectivity will be described later on (Section 4.5),

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes as will be HPTS assays adapted to the detection of pores (Section 3.1.2), fusion (Section 3.1.3), endovesiculation (Section 3.1.4), photosynthesis (Section 4.9), and catalysis (Section 4.8). The lucigenin assay reports specifically on the transport of halide anions.27 Lucigenin is a fluorescent dye, which is moderately quenched by relatively high (10–100 mM) concentrations of halide anions (Cl− , Br− , and I− ), and also of SCN− , presumably by an aborted electron transfer mechanism.30 Asymmetric vesicles can be prepared that contain halides either inside or outside of the vesicle. Addition of the anion transporter leads to equilibration of the halide concentrations by either X− /Y− antiport or M+ /X− symport. The transport can then be monitored by either a decrease (initial high outside concentration of halide) or an increase (initial high inside concentration of halide) in the fluorescence of lucigenin.12, 27 The lucigenin assay has so far exclusively been used to determine Cl− transport but may be similarly useful for Br− and I− or even SCN− transport. Cross-reactivity patterns with exchanged internal and external ions may provide interesting information on the selectivity and transport mechanism of the membrane transporter. As with the HPTS assay, the lucigenin assay usually does not discriminate between anion transport and probe efflux. However, this difference can be made by comparison with dedicated probes for this purpose (Section 3.1.2). The lucigenin assay reports exclusively on anion transport, whereas the calcein assay can report on cations, particularly calcium. Cation binding to the imidodiacetate ligands causes the required change in fluorescence. At higher concentrations, calcein also undergoes self-quenching and is used like CF to monitor probe efflux through large pores (Section 3.1.2).

3.1.2 CF and HPTS/DPX assay The CF and the HPTS/DPX assay are probably the most popular assays to detect the activity of pores (Figure 7). Both, CF and HPTS/DPX assays rely on high local concentrations inside the vesicle such that the fluorescence of the probes is quenched. In the CF assay, self-quenching applies, while in the HPTS/DPX assay HPTS is entrapped together with the quencher DPX.16 Addition of pores results in translocation of the fluorescent probe or quencher and thereby leads to dilution. Consequently, quenching no longer applies and pore activity is signaled as fluorescence recovery. Both assays require a sufficiently large inner diameter of the pore such that HPTS, DPX, or CF can pass through. HPTS/DPX assay detects both cation- and anionselective pores, whereas CF is exclusively transported by anion transporters. The HPTS/DPX assay is preferably used in place of the previous ANTS/DPX assay31 because the fluorescence

9

intensity is much better. However, the ANTS/DPX assay remains preferable for the determination of pH profiles, because neither ANTS nor DPX is sensitive toward changes in pH in the physiologically relevant range of around pH 7. Calcein is sometimes used in place of CF, and both probes respond to parameters other than dilution (pH, cations). HPTS esters are routinely used as fluorogenic probes for esterase activity.32, 33 The transformative efflux of intravesicular HPTS acetate has been used to study synthetic catalytic pores (Section 4.8).32

3.1.3 Assays with membrane-bound probes Membrane-bound probes, which are incorporated directly during vesicle preparation, are commonly fluorescently labeled phospholipids. These probes are mainly used to address more complex changes in membrane structure, thereby providing insights into the structure and mechanism of the transporters. Depending on the specific desired application, various fluorescently labeled phospholipids are available. Popular examples include boron dipyrromethane (BODIPY) probes as FRET acceptors,34, 35 DOXYL probes as quenchers for parallax analysis,36 7nitro-benzofurazan (NBD) probes such as NBD-POPE for flip-flop assay,10, 37, 38 and so on (Figure 8). NBD probes are often used to assay flip-flop.10, 37, 38 Flip-flop refers to the reversible transversal diffusion of lipids from one leaflet to the other leaflet of a lipid bilayer membrane. In intact membranes, this transversal diffusion is very slow (t1/2 on the order of hours to days). However, it can be accelerated by biological or synthetic flippases, which are a special class of membrane transporters related to ion carriers. Alternatively, micellar pores are synthetic ion channels and pores with flippase activity and can thus be identified with flip-flop assay (Figure 2; interfacial location of the transporter, as second distinctive characteristic of micellar pores, can be identified by fluorescence depth quenching experiments with DOXYL probes). In the NBD assay, asymmetric vesicles are produced by addition of dithionite to the labeled vesicle solution, which reduces the nitro group of the outer layers of NBD chromophore to an amine group (Figure 8a). This external NBD reduction transforms the highly fluorescent push-pull fluorophore into a nonfluorescent chromophore. Because the lipid bilayer is impermeable toward dithionite, the NBDs at the inner surface are not affected. Reduction of the fluorescence intensity by 50% after dithionite addition supports the unilamellarity of vesicles and spherical bilayer membranes. Subsequent size exclusion chromatography gives vesicles in which only the inner leaflet of the membrane carries fluorescent lipids. Incubation of the asymmetric vesicle with varying transporter concentrations for varying time intervals yields partial equilibration of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

10

Techniques

NH2

PM NO2 N O N

N O N

Sodium dithionite

NH

NH

Non-fluorescent

O _ O P O O O O

O

O N

N F B N F

O DOXYL

H2N

N +

NH2

Safranin O

BODIPY

Mcellar pore / flippase

Sodium dithionite

N

PE

Sodium dithionite

O

Sodium dithionite

It

1

Micellar pore / flippase

Sodium dithionite No flip-flop

0.5 0.25

Flip-flop

NBD-POPE Time (s)

Figure 8 Selected examples for external and membrane-bound probes for functional and structural studies on ion transport in membrane with details for the NBD flip-flop assay.

the inner and outer leaflet, the extent of which can be detected by addition of dithionite and fluorescence monitoring. Variations in the assay are required to address the kinetics of flip-flop, NBD reduction by dithionite, and the potential transport of dithionite by the added transport system. BODIPY dyes are used in membrane fusion assays (Figures 2d and 8). To characterize fusion, the mixing of the lipid as well as the mixing (and the leakage) of the content of vesicles has to be measured in the presence of the fusogen. For content mixing, the encounter of internal quenchers and fluorophores can be used (e.g., HPTS or ANTS and DPX; see Section 3.1.2). For lipid mixing, vesicles separately labeled with FRET donors and FRET acceptors can be used, and fusion is detected as increasing FRET for increasing fusogenic activity.18

3.1.4 Assays with external probes Assays with externally added probes include the potentialsensitive dye safranin O (Figure 8). This cationic dye only loosely associates with an unpolarized vesicle membrane but binds more efficiently when an inside negative membrane potential is applied. This translocation into a more hydrophobic environment is accompanied with an increase in fluorescence and thereby reports on the extent of the applied membrane potential (Section 4.4).36

A modified version of the HPTS assay with external HPTS is used to measure endovesiculation (Figures 2c and 7).19 In this assay, HPTS is added to unlabeled LUVs. In the presence of an endovesiculator, external HPTS will be transported into the inner water pools of the produced multilamellar vesicles. This internalized HPTS will be insensitive to externally added quencher DPX and can be used as a measure for endovesiculation.

3.2

Absorption spectroscopy

Whereas the sensitivity of fluorescent probes is most attractive for biological applications, membrane-based sensing systems ultimately call for colorimetric probes, with which high levels of important analytes such as glucose or cholesterol can be seen with the “naked eye” as color changes from yellow to red. The PV (pyrocatechol violet)/CBA (4-carboxyphenylboronic acid) assay has been introduced recently for this purpose (Figure 9a).39 In this assay, LUVs are loaded with PV to yield yellow vesicles. Then, CBA is added extravesicularly. PV efflux through active pores is followed by spontaneous reaction of the catechol with CBA to afford the red boronate ester. Other colorimetric probes also exist that respond to changes in pH or transition-metal coordination.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes

H R N

N N H

O

R

OH

N H N

K+

N N

H N H

O

H N H

N

N

N

O

O N

N

N O

N HO OH B O

N

O

N H N H

R

H N H

11

N

N R

2–O

K

+

3PO

NH N

O

NH2

HO OH

H Cs+

5′-GMP

G-quartet

OH –O

O SO3–

–

SO3 HO OH PV (a)

O O B

CBA

K+

Cs+

–

O O

Yellow

Red

(b)

CD active

CD silent

Figure 9 Selected examples for (a) colorimetric and (b) CD probes that respond to covalent capture and ion-templated self-assembly, respectively.

3.3

Circular dichroism spectroscopy

Probes that report the activity of ion channels and pores as changes in their circular dichroism (CD) spectra have been developed on the basis of the ion selectivity of the templated assembly of GMP into G-quartets (Figure 9b).40 To detect the activity of pores with CD spectroscopy, vesicles are loaded with G-quartets. GMP efflux through active pores is reported as CD silencing due to G-quartet disassembly. This simple technique is not applicable to ion channels that are too small or too selective to mediate the efflux of GMP. To detect the activity of ion channels with CD, the potassium selectivity of G-quartets can be used. For example, vesicles are loaded with GMP in the presence of potassium ions at concentrations above the dissociation constant (KD ) of G-quartets. In the presence of cation transporters, external cation exchange from potassium to cesium results in CD silencing as a result of G-quartet disassembly within the vesicle in response to potassium ion efflux. Reversal of the direction of cation antiport with Cs-loaded vesicles and external potassium is even more attractive because the response to ion channel activity is chirogenic. This is one of the few methods where transport across and intactness of spherical membranes are simultaneously reported without additional effort (Section 4.1). G-quartet based CD probes have been adapted to the detection of osmotic pressure in vesicles. In this

case, vesicle shrinking under hyper-osmotic pressure and vesicle swelling under hypo-osmotic pressure are detected as intravesicular G-quartet assembly and disassembly, respectively.

3.4

NMR spectroscopy

NMR spectroscopy complements the use of fluorescence, absorption, and CD spectroscopy in investigating supramolecular membrane transport systems. Naturally appealing to the supramolecular chemist, the usefulness of NMR assays should not be overestimated. NMR is less straightforward to implement, and its sensitivity is limited. Nevertheless, it can be used to address questions on specific ion selectivities. In a conventional setup, a paramagnetic compound that cannot cross the membrane is added externally to the vesicle solution. This leads either to line broadening or to a frequency shift of the external ions such that they can be distinguished from internal ions. If both signals can be seen, concerns about vesicle destruction or transport of the shift reagent become redundant. In principle, transport of any ion for which NMR-active isotopes and a suitable shift reagent exist can be investigated and in fact the use of NMR to observe transport of Na+ , Cl− , or Br− has been suggested early on. Nonetheless, its routine use was mainly limited to 23 Na NMR spectroscopy, in which external dysprosium triphosphate is used as a paramagnetic shift reagent to separate the chemical

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

12

Techniques

shifts of intra- and intervesicular Na+ .41 Sodium flux is detected by line-width analysis or peak integration. More recently, Smith and Davis independently introduced the use of 35 Cl NMR with Co2+ as shift reagent,42, 43 and Davis and coworkers demonstrated that transport of hydrogencarbonate could be followed by using 13 C NMR with Mn2+ as an external shift reagent.11

3.5

Miscellaneous

Because of their advanced level of development, high sensitivity, and broad applicability, fluorescence spectroscopy with labeled LUVs and planar bilayer conductance experiments are the two techniques of choice to study synthetic transport systems. The broad applicability of the former also includes ion carriers, but it is extremely difficult to differentiate a carrier from a channel or pore mechanism by LUV experiments. However, the breadth and depth accessible with fluorogenic vesicles in a reliable userfriendly manner are unmatched by any other technique. Planar bilayer conductance experiments are restricted to ion channels and pores and are commonly accepted as substantial evidence for their existence. Extremely informative, these fragile single-molecule experiments can be very difficult to execute and interpret. Another example for alternative techniques to analyze synthetic transport systems in LUVs is ion-selective electrodes. Conductance experiments in supported lipid bilayer membranes may be mentioned as well. Although these methods are less frequently used, they may be added to the repertoire of the supramolecular chemist.

4

FUNCTIONAL STUDIES

The overall activity of a transporter is influenced by numerous parameters, which include buffer and membrane composition, membrane polarization, and osmotic stress, to name only a few. The comparison of the intrinsic activity of different transporters on an absolute scale is nearly impossible for this reason. This is not further problematic because absolute activities are probably the least interesting aspect of synthetic transport systems and arguably deserve little priority. What really matters is responsiveness to specific chemical or physical stimuli. This includes sensitivity toward membrane composition, membrane potential, pH, anions, cations, molecular recognition, molecular transformation (catalysis), or light. These stimuli-responsive, multifunctional, or “smart” transport systems are attractive for use in biological, medicinal, and materials sciences. Standard techniques to identify such unique characteristics rather than absolute activities or mechanistic details are outlined in this section.

4.1

Designing experiments

The study of transport processes in membranes is difficult because the relation between origins and phenotype is often complex. The design of meaningful experiments that can be interpreted with reasonable confidence is thus of highest importance. The key problem usually is to assess whether the characteristics found are significant. To demonstrate significance, experiments designed to yield dichotomic behavior are ideal.10 Isolated or parallel trends in complex systems can originate from less or completely unrelated processes, whereas the inclusion of negative or positive control compounds that show opposite trends is of central importance for meaningful data interpretation. For instance, increases in activity, ion selectivity, Hill coefficient, gating charge, and so on, in response to changed conditions for all studied compounds can originate from changes in delivery to or partitioning into the membrane, not to speak of simply overlooked (but sometimes hard to identify) technical errors. The identification of controls that show dichotomic behavior under identical conditions can suffice to demonstrate that at least one of the two opposing trends is significant. Before embarking into the search for significant characteristics, technical questions such as stirring, temperature control, intervesicular transfer (Section 3.1), delivery to the membrane, and so on should be under control. Stirring of the vesicular suspension at a strictly controlled constant temperature during an entire transport experiment is absolutely necessary to avoid errors. The yield of delivery to the membrane determines the relevant concentration of the transporter in the membrane and thus often controls the apparent activity measured as EC50 (Section 4.2). For delivery, a transporter is usually added as a 20 µl drop of 100 times more concentrated “stock” solution in a polar organic solvent to 2 ml vesicle suspension. The transfer of the transporter from this drop of solvent to the vesicular membrane is one of the least understood processes in transport experiments. From a technical viewpoint, it is highly important to find a solvent that provides efficient transfer to the membrane to ensure that a potentially efficient transporter is not simply overlooked. A potential approach to shed some light on this complex issue comprises an exhaustive solvent screening with the ultimate goal to elucidate which solvents help the transporter to cross the aqueous phase and reach the membrane. Having established a rough guideline, solvent mixtures may be even more beneficial. Additionally, nondestructive detergents have been recently applied to address delivery problems. For example, the highly viscous Triton X-100 (and other members of the triton family) efficiently lyses membranes above the critical micelle concentration (cmc ∼100 µM) but is an excellent delivery agent when reaching the membrane in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes inactive, monomeric form.44 Used at 8 mM as additive in a stock solution, injection of 20 µl to 2 ml vesicular suspension will produce micelles that solubilize transporters for a short time but fall apart before reaching the membrane. Other promising detergent additives for delivery include Span 80, which is a (beneficial) mixture of differently alkylated sorbitols, alkyl glucosides, cholate, and so on. Moreover, it is important to clarify basic questions before initiating the search for significant characteristics. For example, it is important to know if vesicles remain intact or get damaged during transport. This question can be addressed either with assays that demonstrate intactness (operational CD or NMR probes, see Sections 3.3 and 3.4), with meaningful comparisons of assays with different readouts (better EC50 s in HPTS assay than in CF assay),8 or with dedicated experiments such as internal trapping.16 In the latter assay, transport activity is measured and compared among three different vesicle experiments. These are, first, addition of all transporter in a single step to vesicles loaded with a fluorescent probe (Figure 10c); second, stepwise addition of transporter in small portions well below the EC50 (Figure 10b); and third, stepwise addition of transporter to vesicles loaded with a fluorescent probe plus a highly efficient inactivator (Figure 10a). Much higher activity in the absence of an internal inactivator (Figure 10b and c) than in its presence (Figure 10a) suggests that the transporter has been irreversibly trapped in the vesicle without release of the probe. This indirectly demonstrates that transport occurs across the membrane of intact vesicles, because vesicle destruction would lead to inactivator dilution below the concentration at which the inactivator becomes inefficient. Whereas internal trapping assays demonstrate that vesicles remain intact during transport, they do not exclude that lipids are actively involved.

PI

13

This question can be addressed with flip-flop and related assays (Section 3.1.3).

4.2

Stoichiometry: Hill analysis and undetectable active structures

Hill analysis is the most important technique to characterize synthetic transport systems.45–48 For Hill analysis, the dependence of the fractional activity Y (Section 3.1) on the concentration cM of the monomer used to self-organize or self-assemble into active transport systems is measured (Figure 11). The obtained dose–response curve (or Hill plot, or cM profile) is analyzed by nonlinear regression using the Hill equation Y = Y∞ + (Y0 − Y∞ )/[1 + (cM /EC50 )n ]

(1)

where Y0 is the minimal activity observed without transporter and Y∞ the maximal at saturation with excess transporter. The Hill analysis delivers the EC50 and the Hill coefficient n. Already introduced before (Figures 3 and 5), the EC50 is the effective monomer concentration needed to reach 50% activity. The smaller the better, the EC50 is a convenient empirical value to compare the activity of different transporters. Many different processes can contribute to EC50 , including delivery efficiency to the membrane (Section 4.1) and partitioning, self-assembly in solution, at the interface, or in the membrane, reorientation in the membrane, intervesicular transfer, and so on. Dissection of the different contributions to EC50 is possible, at least in part, by systematic modification of conditions, delivery additives, vesicles (membrane composition), and compounds (hydrophobicity). The determination of EC50 as such is very easy in the U-tube and with fluorescent vesicles. In planar bilayer conductance experiments, EC50 measurements can be less straightforward, particularly in single-channel

PI

(c) (a)

It

(b)

PI

PI

(c) (a) (b)

Time (s)

Figure 10 Internal trapping assays can be used to prove that transport really occurs across the membrane of intact vesicles. Transporters are added in small portions well below EC50 to labeled vesicles (a) with or (b) without internal inactivators. Without inactivators, activity will gradually increase and reach the value obtained when all transporters are added at once (b vs c). With inactivators, transporters are continuously trapped intravesicularly, and no activity is observed even at high total concentrations (a vs c). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

14

Techniques

EC50

Long

0.5

0 (a)

Monomer concentration cM

lifetime Relative energy

Y

n>1 Active structure

EC50

cM

n>1

Inactive precursor n≤1 Denature

n≤1

(b)

Short

Active structure

Stabilize (c)

Stability

Figure 11 Hill plots are dose–response curves that describe the dependence of activity on monomer concentration. Hill analysis can differentiate between (a) unstable supramolecular active structures (n > 1; known stoichiometry, undetectable suprastructure) and (b) stable supramolecular or unimolecular active structures (n ≤ 1; unknown stoichiometry, detectable suprastructure). Single channel lifetimes (τ ) differentiate between labile and inert active structures, whereas both open probabilities Po and Hill coefficients indicate thermodynamic stabilities.

experiments. Contributions from Po are usually the most important; also the conductance g matters much. On first view, the Hill coefficient n reports on the cooperativity of the transport process.45–48 Measured below EC50 to avoid artifacts from assay saturation, Hill plots with n > 1 show an upward curvature (Figure 11a), Hill plots with n < 1 show a downward curvature (Figure 11b), and Hill plots with n = 1 are linear. With n > 1, Hill coefficients correspond to the number of monomers (or stable dimers, trimers, etc.) in the active supermolecule, and EC50 s can represent its KD . With n ≤ 1, the situation is more complex. To fully understand and correctly use all results from Hill analysis, it is crucial to always remember that the Hill equation applies only when the monomer concentration cM is much higher than the concentration of the aggregated suprastructure. In the case of n > 1, this means not only that the stoichiometry of the active supermolecule is known, but also that the active supramolecular structure is unstable and therefore, a minority population. That is, endergonic selfassembly implies that the active supermolecule is formed only by a small fraction of the total initial monomer concentration. Therefore, at concentrations relevant for function (around EC50 ), the active supermolecule exists only in the presence of excess monomer, and classical, unselective methods such as IR or NMR become irrelevant because they will only report on the inactive monomers. Eventually, the claimed or required structural support by NMR, X-ray, CD, or other conventional techniques for n > 1 systems is thus intrinsically wrong, often seriously misleading. Applying higher concentrations would shift

the equilibrium toward higher fraction of aggregates but would then introduce the problem of forming inactive supramolecular polymers. The challenging n > 1 situation with undetectable active structures resembles the familiar scene where only a few do the work while many stand around and watch (Figure 11a and c, energetically uphill formation of active structure). As it will be hard to see the workers in a zoom-free photo of this scene, n > 1 systems are undetectable by routine techniques. As they are the most common, efficient, and useful ones, this is very important to understand and remember: with n > 1, few working, active structures are thermodynamically unstable; supermolecules have known stoichiometry (n or a multiple of n) and are intrinsically undetectable in routine structural studies. The complementary n ≤ 1 systems are either unimolecular or are formed by stable supermolecules that erroneously appear as cM in the inapplicable Hill plots (Figure 11b and c). To stay within the above picture, this situation reflects the perfect scene where everybody present really contributes to the work. The n ≤ 1 systems have unknown stoichiometry but are compatible with routine structural analysis. They are less desirable in practice because they tend to suffer from poor delivery and precipitation. Precipitation during delivery to the membrane is expressed in n < 1 and incomplete Hill plots at high concentration; original kinetic traces might show bursts of high noise due to scattering. Biological model pores such as gramicidin A, melittin, hemolysin A, and so on, all act as “invisible” n > 1 transporters. The less desirable n ≤ 1 systems can be destabilized into the more attractive n > 1 systems. Increasing the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes charge repulsion within stable tetrameric pore has been shown to shift the Hill coefficient from n = 1 to n = 4.48 Chemical or thermal denaturation can be generally used to destabilize active supermolecules and artificially increase monomer concentrations in Hill analysis.45 For example, the absorption and CD spectra of a synthetic self-assembled pore were concentration-independent in the absence of denaturants such as urea and guanidinium chloride, whereas their presence produced the expected n = 4 concentration dependence. This technique to convert exergonic into endergonic self-assembly provides general access to the stoichiometry of stable, active supermolecules. The complementary conversion of n > 1 into n ≤ 1 systems has been achieved as well. Covalent capture is an obvious possibility48 ; recent more subtle strategies include stabilizing interaction between macrodipoles and membrane potentials in polarized membranes.10

4.3

Pore diameter: Hille analysis and size exclusion

Approximation of the inner diameter of synthetic ion channels and pores can be most easily carried out by loading LUVs with fluorescent probes of different sizes and comparing the activities.8 The HPTS assay, which reports on proton transport, is compatible with all diameters. Translocation of HPTS/ANTS or DPX requires inner diameters of ˚ and the CF assay requires diameters larger than about 5 A, ˚ EC50 (HPTS) < EC50 (CF) demonstrates that about 10 A. ion transport is more efficient than dye export or vesicle destruction. However, valid interpretations require appropriate caution and controls, as EC50 (HPTS) < EC50 (lucigenin, CF) can support cation selectivity for large pores also,28 EC50 (HPTS, lucigenin) < EC50 (CF) can originate from chloride selectivity, and so on (Section 4.5). Moreover, U-tube experiments readily confirm that probes as large as CF, DPX, or safranin O can also be translocated via a carrier mechanism (Section 1).16 In the case of synthetic pores that have been equipped with internal binding sites for molecular recognition, the size dependence of pore blockage can be consulted as well.1, 2, 49–51 To identify larger pores or other membrane defects, fluorescently labeled polymers such as CF-dextrans51 or dedicated enzyme coupled flux assays are available.52 More quantitative insights can be obtained in planar bilayer conductance experiments. With the Hille equation (2)24 1/g = lρ/[π(d/2)2 ] + ρ/d

(2)

single-channel conductances g can be directly related to the inner diameter d of a channel.51, 53 Whereas the

15

resistivity of the recording solution ρ is an unproblematic experimental value, the length l of the channel or pore is an assumption that introduces substantial uncertainty as long as details of the active structure are unknown. The assumption of cylinders filled with electrolytes holds well for pores, whereas correction factors are needed with small channels to compensate for intrinsic underestimates from restricted mobility in their confined interior.51, 53

4.4

Voltage sensitivity

The dependence of the activity of synthetic ion channels and pores on the membrane potential can be grouped into ohmic and non-ohmic behavior. Ohmic ion channels and pores follow Ohm’s law; that is, fractional activity Y (e.g., transmembrane current I ) is proportional to the applied membrane potential or voltage V , and the proportionality factor is the resistance or, more commonly, its inverse conductance g (Figure 12a, dotted line). Non-ohmic channels violate Ohm’s law (Figure 12b, solid). Their activity Y exhibits exponential dependence described by the gating charge zg (3):
 Y ≡ I = g0 exp zg eV /kT V

(3)

The gating charge zg is formally the charge that has to be translocated across the bilayer to form the active structure (e = elementary charge; k = Boltzmann constant; T = temperature).10 The gating charge is commonly given as an absolute value, no matter whether activation or deactivation is observed with an increasing inside negative membrane potential. Inactivation at high membrane potentials is quite common and characteristic for highly symmetric, ohmic ion channels and pores like β-barrels. Common gating charges are, for example, 1.5 for the bee toxin melittin, 0.85 for synthetic rigid rod ion channels, or about 0.55 for ion channels based on macrocyclic peptide mimetics. The creation of voltage-gated ion channels has raised much interest among supramolecular chemists, because of their importance for biology and materials science.1, 2, 10 The connection with the former is obvious since voltagegated ion channels are abundant in natural systems. In materials science, there is a striking functional similarity to artificial photosystems and field-effect transistors, in which the current between source and drain is regulated by the voltage applied on the gate and which are also characterized by an I −V curve.54 Planar bilayer conductance experiments and fluorescence kinetics in polarized vesicles provide access to I −V curves and gating charges. In single-channel conductance experiments, zg is classically determined from changes in the open probability Po with applied voltage, but correct

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

16

Techniques Transporter addition Addition of valinomycin

(a)

zg > 0 (Non-ohmic )

K+

K+ Na+

V PE

V (mV)

V PI

−

X1

M1+

X2− M2+

Lysis

I∞ Emission intensity

Y

zg = 0 (Ohmic )

PI

I0 I0

PE

I∞

(c)

Time (s)

Ion transporter

PE (b)

Inside

Outside

Figure 12 (a) Synthetic transport systems are ohmic (gating charge zg = 0) or non-ohmic (i.e., voltage-gated, zg = 0). The dependence of activity (current) on membrane potential (voltage) can be determined in planar bilayer conductance experiments or by double-channel kinetics (c) in doubly labeled LUVs (b) that are polarized with a valinomycin-coupled potassium gradient, loaded with an internal probe PI for activity (e.g., HPTS) and decorated with external probes PE to measure depolarization (e.g., Safranin O; see Figures 5–9 for structures and data analysis).

reproduction of zg s was also demonstrated for macroscopic I–V curves. This is feasible because voltage-dependent changes in the overall current are mainly caused by changes in the open probability Po (i.e., number and stability of open channels) while changes in single-channel conductance (i.e., rectification) are generally much smaller.21 Polarized vesicles are convenient and reliable systems to record I−V curves and gating charges. (Figure 12b).10, 21 To prepare polarized vesicles, LUVs are initially loaded with K+ and the outside buffer is osmotically balanced with Na+ . Addition of the K+ carrier valinomycin leads to a dissipation of the potassium gradient and thereby to a reduction of the chemical potential. However, with increasing translocation of K+ , the vesicles become more and more negatively charged inside, leading to a counterbalancing Nernst potential. The chemical potential and the Nernst potential equilibrate depending on the initial magnitude of the potassium ion concentration gradient, leading, according to the Nernst equation, to a membrane potential V . The produced potential can be monitored through the fluorescence of safranin O (λexc = 522 nm; λem = 581 nm). Generally, a full assay is carried out as follows: Vesicles with HPTS and high K+ concentrations inside are added into a cuvette with a low K+ concentration buffer and safranin O. The fluorescence is monitored in triplechannel kinetics reporting on the intravesicular pH (ratiometric HPTS measurement to monitor transport activity, see Figure 7 and following section) and on the membrane potential. Addition of valinomycin at intermediate concentrations leads to rapid buildup of membrane potential without immediate collapse, which is signaled by an emission increase of the externally added probe safranin O

(Figure 12c). A base pulse is applied at constant membrane potential just before the addition of the transporter to generate a pH gradient. The activity of the added synthetic transport system is then monitored through the fluorescence changes in the HPTS channel and the safranin O channel (Figure 12c). Normalization as described in Section 3.1, and sometimes background subtraction, gives the fractional activity for a specific transport system concentration and a certain membrane potential. Changes in activity at different membrane potentials can then be used to determine the gating charge zg (Figure 12a). Comparability of gating charge zg from HPTS or ANTS/DPX assay in polarized vesicles and multichannel or single-channel (Po –V plots) planar bilayer conductance experiments has been confirmed.21 The voltage dependence of molecular recognition and catalysis will be described later on (Sections 4.7 and 4.8).

4.5

Ion Selectivity

Synthetic transport systems prefer to transport either anions or cations, and the preferred ions are transported according to a specific selectivity sequence or topology (Figure 13).8, 22, 24 Both anion/cation selectivity and selectivity sequences/topologies can be determined with planar bilayer conductance experiments and with fluorogenic vesicles. In planar bilayer conductance experiments, salt concentration gradients between the cis and trans chamber are used to measure ion selectivity. With salt gradients, a current is flowing in the absence of an applied voltage. The potential needed to stop this current from flowing is

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes

17

Reversal potential Vr

X− 0

(a)

OH– / X−

Cl− Br− I− F−

Emission intensity

0

Emission intensity

Anion selectivity

I (pA)

M+

V (mV) Time

Time

(b)

PI H+

M+

(c)

Inside

(d)

> I− > Cl− > F − > Cl− > I− > F − > Br − > I− > F − > Br − > F − > I− > F − > Br − > I− > Cl− > Br − > I− (anti-Hofmeister )

Time

VII IV

I I− Br − Cl− F − r−−1(Å−1)

I II III IV V VI VII VIII IX X (e) XI

Cs+ Rb+ Rb+ K+ K+ K+ Na+ Na+ Na+ Na+ Li+

> > > > > > > > > > >

Rb+ Cs+ K+ Rb+ Rb+ Na+ K+ K+ K+ Li+ Na+

> > > > > > > > > > >

K+ K+ Cs+ Cs+ Na+ Rb+ Rb+ Rb+ Li+ K+ K+

> > > > > > > > > > >

Na+ Na+ Na+ Na+ Cs+ Cs+ Cs+ Li+ Rb+ Rb+ Rb+

> Li+ > Li+ > Li+ > Li+ > Li+ > Li+ > Li+ > Cs+ > Cs+ > Cs+ > Cs+

selectivity

I− > Br − > Cl− > F − (Hofmeister ) Br − Br − Cl− Cl− Cl− F−

Time

Eisenman selectivity sequences / topologies

selectivity

II III IV V VI VII

K+ Rb+ Cs+ Na+ Li+

Outside

Halide selectivity sequences / topologies I

X−

Emission intensity

Emission intensity

Cation selectivity

XI IV I

Cs+ K+ Na+ Li+ r+−1(Å−1)

Figure 13 Determination of ion selectivity (a) from cis–trans ion gradients in planar bilayer conductance experiments and (b, c) by external ion exchange in the HPTS assay; Description of the results in (d) anion and (e) cation selectivity sequences or topologies, the latter showing selectivities as a function of reciprocal ion radii or ion dehydration energies.

called the reversal potential Vr (Figure 13a). The sign of Vr identifies preference for anions or cations in the presence of unidirectional salt (MX) gradients, or preference among anions or cations in the presence of antiparallel anion (X1 /X2 ) or cation (M1 /M2 ) gradients. Selectivities can be quantified by conversion of reversal potentials Vr into the permeability ratios P(M) /P(X) , P(M1) /P(M2) , or P(X1) /P(X2) using the Goldman–Hodgkin–Katz (GHK) equation.20, 24 In LUVs, ion selectivities can be determined with the HPTS assay by external cation and anion exchange. Sensitivity of transport activity toward anion but not cation exchange implies anion selectivity (Figure 13b)8, 20 ; responsiveness to cation but not anion exchange implies cation selectivity (Figure 13c). Furthermore, sensitivity toward external ion exchange suggests that dissipation of the pH gradient does not occur via HPTS export or vesicle destruction. The same principle of external or internal anion exchange can be applied to determine anion selectivities with the lucigenin assay.8

The identification of proton selectivity requires special attention because it can be difficult to detect and study in both planar bilayers and LUVs. In the HPTS assay, the apparent activity of proton transporters decreases with increasing H+ > M+ selectivity, because M+ antiport becomes more and more rate-limiting with increased selectivity. To solve this problem, valinomycin can be added.34 Recovered H+ transport activity in presence of the potassium carrier demonstrates H+ > K+ selectivity. Analogously, the proton carrier carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) has been used to confirm the M+ > H+ selectivity of, for example, amphotericin B. In the U-tube, ion selectivities are easily determined by comparison of transport efficiencies for detectable series of cations or anions. Different picrate salts are best known to determine cation selectivity sequences. The comparison of velocities with safranin O or DPX against CF or HPTS can provide direct evidence for anion/cation selectivity of molecular recognition and translocation.16

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

18

Techniques Eisenman IV, which resembles the potassium selectivity in neurons. The Eisenman theory was subsequently extended to anion selectivity sequences by Diamond and Wright (Figure 13d). There, topology sequences for anions have been restricted to halide anions because the variety of different shape and constitution among biologically relevant anions is more complex than among cations.57 This leads to possible mismatches of size and dehydration energy and deviations from expected sequences. For example, acetate and fluoride have similar dehydration energies but very different sizes. Complete dependence on dehydration energies, which would be halide topology I according to Diamond and Wright, is also referred to as Hofmeister or lyotropic series for historical reasons. In Hofmeister’s classical experiment on protein precipitation, hydrophobic anions have been termed chaotropes (structure breakers) and hydrophilic anions kosmotropes (structure makers). Selectivity for chloride as found in topologies IV, V, and VI is highly relevant for medicinal reasons; for example, synthetic chloride transporters have been suggested as a treatment to alleviate the symptoms of cystic fibrosis or macular degeneration.4 The dependence of the transport activity on the mole fraction of a binary mixture of “fast” and “slow” ions is a classical test for a cooperative multi-ion transport mechanism.8, 10, 27, 58 Commonly, a negative deviation from a simple linear additivity is considered an anomalous mole fraction effect (AMFE) and confirms the existence of multiple ion binding sites along the conduction pathway (Figure 14a).8, 10, 35 The conventional explanation refers to the presumption of single-file ion channels, in which occupation of more than one binding site is necessary for the ions to move really fast (Figure 14c and d). However, this

Anion/cation selectivities and ion selectivity sequences from planar bilayer conductance and LUV experiments are comparable, at least qualitatively and to a certain extent.22 Anion/cation selectivities for large pores are usually weaker than for small channels. They can be estimated in LUVs by comparing results from appropriate assays with different selectivities such as the anion-selective CF assay and the nonselective ANTS/DPX assays. The HPTS or lucigenin assays are not applicable in this case because of unselective probe export. Ion selectivities depend significantly and in an understandable manner on conditions. For instance, pH-gated inversion of anion/cation selectivity with synthetic pores can be designed rationally.55 To rationalize ion selectivity sequences, numerous theories exist.20, 22, 24, 56–58 Most popular for cation selectivity sequences is the Eisenman theory (Figure 13d). There, selectivity is based on the interplay between the energy gained from binding to the binding sites of synthetic transport systems and the dehydration penalty, which is necessary to translocate an ion from the aqueous phase into the hydrophobic interior of the lipid bilayer membrane. Eisenman topologies can be determined by plotting ion selectivities as expressed through permeability ratios, conductances, or fractional activities versus the reciprocal radius of the cation (Figure 13e) or the dehydration energy of the cation.56 Eisenman derived in total 11 topologies, covering all possible cases from the sequence completely determined by dehydration penalty (Eisenman I) to the one completely determined by binding (Eisenman XI). The former one is the least interesting because no sophisticated transport system is required, for example, transport catalyzed by tetraalkyl ammonium ions follows the Eisenman I topology. Important for bioinspired systems is Negative AMFE

Activity Y

Positive AMFE

Fast ion

(c)

Fast Y ion

Y

0.5

(d)

Slow ion (a)

IC50

Slow ion Mole fraction

(c)

YMAX

1

(b)

Mole fraction

(e)

Concentration of permeant ion

(d)

Figure 14 (a) Negative and (b) positive AMFEs, (c) schematic cooperative multiion transport and (d) noncooperative ion transport mechanisms, and (e) saturation with the permeant ion (for details on IC50 , see Figure 16). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes explanation is speculative and controversial, and obviously does not apply to positive AMFE (Figure 14b). Cooperative inhibition is a possible but less likely explanation for positive AMFE. More likely, ultrastrong binding of the more active (“fast”) ion saturates interfacial sites before all ions have been exchanged. Analogously, negative AMFE can originate from “early saturation” with strongly binding but slowly moving ions, inhibiting the transport of rapidly moving ions. The existence of different explanations underscores that AMFE of synthetic transport systems is not fully understood at this point. Their coincidence with significant ion recognition and selectivity demonstrates that the phenomenon is important and deserves clarification. Determination of the concentration dependence is likely to clarify the origin of AMFEs. The dependence of the activity of synthetic transport systems on ionic strength in general is an important, challenging, and underexplored topic. With increasing ion affinity, synthetic transport systems saturate more easily at increasing ionic strength (Figure 14e).24 This blockage by the permeant ion is characterized by the maximal activity YMAX and the IC50 : that is, the ion concentration needed to reach 50% of YMAX (see Section 4.7 for Hill analysis of blockage and ligand gating). The IC50 is an indicator for the dissociation constant of the ion-transporter complex (i.e., the lower the IC50 , the better the ion affinity). Systems saturation with the permeant ion is most easily determined in planar bilayer conductance experiments, where maximal activity is expressed as maximal conductance gMAX of the system. In LUVs, changes in ionic strength are more bothersome to execute. Clearly different from this more “transient” saturation at high ionic strength is “permanent” ion pairing at low ion concentrations. Such permanent, that is, thermodynamically very stable, ion pairing has been observed with multiply charged synthetic transport systems and multivalent counterions. Pertinent examples include phosphate scavenging in arginine-rich pores and magnesium binding to aspartaterich pores.55 This led to significantly reduced single-pore conductance levels (2), increased single-pore lifetimes (cf. Section 2.2 and Figure 11), and inversion of anion/cation selectivity. In this case, the tightly bound counterion is not the electrolyte, and the process should be considered in the context of blockage and ligand gating (Section 4.7) rather than ion selectivity. Ion pairing with hydrophobic counterions is one of the most useful strategies to activate hydrophilic polyions [CPPs, (cell-penetrating peptides), DNA, RNA, and biological K channels] to act as carriers in bulk and lipid bilayer membranes for sensing, voltage gating, and cellular uptake applications.16, 44 The more “transient” saturation with permeant ions relevant to this section is incompatible with the Hille equation [(2) of Section 4.3], which relates the conductance

19

with the inner diameter of a channels or pore. That is because the former assumes reduced conductance with increased occupation of binding sites, while the latter treats ion channels as electrolyte-filled cylinders. An elegant solution in theoretical terms, and presumably also in terms of ion channel design, is cooperative multi-ion hopping in a single file, an ideal mechanism to transiently accelerate inhibitory off-rates by ion–ion charge repulsion and thus to combine selectivity with speed (Figure 13c).10, 24, 29, 58 This more “transient” saturation with permeant ions often originates from weak ion pairing with charged sites of the transport system. This process is related to protonation or deprotonation of the same charge sites, that is, pH gating of the transport system. Reduced internal charge repulsion within the transport system due to ion pairing or de-/protonation usually causes the active structure to yield to permanent lateral membrane pressure and collapse (Section 4.6).

4.6

pH sensitivity

Many biological and synthetic transport systems have a hydrophobic external and charged internal surface. These ubiquitous internal acids or bases frequently account for pH sensitivity. For this reason, it is advisable to record pH profiles as early as possible during the characterization of the novel molecule and continue in-depth studies at optimized pH.31, 59 The best technique to determine pH profiles is the ANTS/DPX assay, since both probe and quencher are not very pH-sensitive (Section 3.1.2). However, most assays are applicable as long as systematic corrections are applied. Determination of pH dependences in planar bilayer conductance experiments is straightforward. The pH profiles of multiply charged transport systems are usually bell-shaped.31, 59 With charged acids such as ammonium cations, highest activity at pHMAX occurs below the intrinsic pKa of the monomeric acid (Figure 15a), whereas pHMAX with charged bases such as carboxylates occurs above the intrinsic pKa of the monomeric acid (Figure 15b). pHMAX describes the difference between pHMAX and pKa . The pH50 describes the window of activity between the effective pH50 describing the pH-gated opening and closing of the system. Both pH50 and pHMAX increase with the number of charges in the transport system. The bell-shaped pH profile can be convincingly understood and simulated with proximity effects that change the pKa ’s to minimize repulsion between proximal charges. Bell-shaped pH profiles over the range of gradual charging of the transport system show that intermediate charge repulsion gives highest activity. The produced internal pressure is ideal to counterbalance external lateral membrane pressure on the transport system and stabilize internal space (Figure 15c).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

20

Techniques Charged bases

Charged acids pHMAX

Y

1

Y 1

∆pHMAX

0.5

pKa

pKa

pHMAX ∆pHMAX

0.5

∆pH50

∆pH50

0

0

(b)

pH

(a)

pH

Internal charge repulsion

External membrane pressure

Closed

Open Closed

(c)

pH

Figure 15 Bell-shaped pH profiles of transport systems with multiple charged (a) acids or (b) bases with pHMAX < intrinsic pKa and pHMAX > pKa , respectively, and (c) their convincing explanation with the ICR-EMP model.

Internal underpressure with undercharged systems causes collapse of their internal space, while internal overpressure with overcharged systems causes their ejection from the membrane (ICR-EMP, internal charge repulsion-external membrane pressure model).59 Ion pairing at high ionic strength can cause inactivation by the analogous internal discharging (Section 4.5).

4.7

Ligand gating and blockage

The possible response of synthetic transport systems to chemical stimulation by molecular recognition can be either activation (ligand gating, opening) or inactivation (blockage, closing). In the context of host–guest chemistry, the synthetic transport system is the host (receptor), and the ligand is the guest. Ligand gating and blockage are the basis of the applications of synthetic transport systems to sensing and catalysis (Section 4.8).23, 60 Ligand gating and blockage are characterized by Hill analysis, in which fractional activity Y is plotted against the inactivator concentration and analyzed with the modified Hill equation (4) (Figure 16a and b):

Y = Y∞ + (Y0 − Y∞ )/(1 + cINACTIVATOR /IC50 )n

(4)

or against the activator concentration and analyzed with the modified Hill equation (5) (Figure 16c and d) Y = Y∞ + (Y0 − Y∞ )/(1 + cACTIVATOR /EC50 )n

(5)

Effective concentrations (EC50 s) and inhibitory concentrations (IC50 s) obtained for activator and inactivator can provide an approximation for KD of the host–guest complex, although many other additional components can have strong influences. Systematic overestimates may, for example, be obtained in the case of stoichiometric binding, in which binding may be much more efficient than it appears from Hill analysis.46 The Hill coefficient n obtained from Hill analysis of ligand gating or blockage can reflect the binding stoichiometry of ligand or blocker to the transport system, although significant influence of other parameters can often not be ignored (compare Section 4.2). The outcome of Hill analysis depends on numerous experimental parameters such as pH, ionic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes Concentration reaction IC50 (P)

IC50 (S)

Y

Concentration reaction EC50 (S)

21

IW > 0

EC50 (P)

pKD

IW = 0

1 0.5

(e)

V

0 (a)

(c) IC50 (P)

Y

IC50 (S)

EC50 (P)

EC50 (S)

IW ≈ 0

IW > 0

1 0.5 0

Concentration (S+ P)

(b)

(d) Concentration (S + P)

No inclusion complex

(f)

Inclusion complex

KD, kon, koff

Current (pA)

Addition of blocker

(g)

t1′

Open pore

t2′

Pore-blocker complex Closed

−

t1

t2

t3

Time (ms)

Figure 16 (a, b) Inactivation and (c, d) activation of synthetic transport systems, characterized by Hill analysis (Section 4.7), can be used to detect reactions in fluorogenic vesicles as long as effective (EC50 ) or inhibitory concentrations (IC50 ) of substrate (S) and product (P) differ sufficiently (Section 4.8). Pore opening (a and d) and pore closing (b and c) during the reaction is indicated by an arrow. (e) Woodhull analysis of the voltage dependence of blockage reveals (f) the depth of molecular recognition. (g) Detection of molecular recognition in planar bilayer conductance experiments, in which binding of a guest reduces the conductance of the porous host, and statistical analysis of the kinetics of the new conductance level of single host–guest complexes reveals dissociation constant as well as formation and dissociation kinetics.

strength, self-assembly, voltage, topological matching, ion selectivity, and so on. Ligand gating can be difficult to study in single-channel conductance experiments, and high and low affinity blockage can pass undetected. With intermediate affinity, new conductance levels for single host–guest complexes can appear (Figure 16g). In this case, single blockers can be detected entering and leaving the pore in a stochastic manner, and the resulting traces can be analyzed statistically to give thermodynamic and kinetic data for complex formation.61, 62 Important examples for ligand gating and blockage include counterion activation and inactivation of polyions such as CPPs, DNA, RNA, voltage-gated potassium channels, or synthetic pores for ion transport, sensing, and cellular uptake.16, 44 Activation by intrinsic components of the membrane as a special case of ligand gating is interesting for targeted pore formation with antimicrobials (activation of nisin by lipid II; activation of natural antibiotics and

synthetic mimics by anionic lipids) and antifungals (activation of amphotericin B, nystatin, etc., by ergosterol).63 The classical modes of pore blockage are the formation of pseudorotaxanes with polymer blockers and the formation of inclusion complexes with small-molecule blockers.60 The voltage dependence of blockage described in the Woodhull equation (6) identifies the existence of inclusion complexes.50 −logKD = −logKD (0 mV) + (lw zGUEST FV)/(2.303 l RT) (6) As charge of the guest (zGUEST ) and the length l of the pore are known, the dependence of the dissociation constant KD on the applied voltage V is applied in this equation to measure the Woodhull distance lw . Describing the distance from pore entrance to the active site, inclusion complexes have voltage-dependent lw > 0, whereas binding at the surface or in solution is revealed as voltage-insensitive lw = 0 (Figure 16e and f).50 Classical Woodhull analysis

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

22

Techniques

is carried out with planar bilayer conductance experiments, but it has been demonstrated that results obtained with polarized LUVs are at least qualitatively comparable.50

4.8

channels and pores have been observed routinely at the single-molecule level. Initially, these techniques were developed for structural studies of biological systems. Best known is the substituted-cysteine accessibility method (SCAM), which measures the “jump” in single-channel conductance when small, charged methanethiosulfonates react with cysteines in the ion-conducting pathway.23 Other realized examples include thermal cis–trans isomerization, photodeprotection, disulfide exchange and polymerization, and nucleophilic substitutions.23, 61, 64 Detection of reactions on the single-molecule level is particularly attractive to study single reactive intermediates (Figure 17b).61 For this purpose, the conductance levels for the poresubstrate and pore–product conjugate (or, in future applications toward catalysis, pore–substrate, and pore–product complexes) are identified first. Any new conductance level appearing during a reaction between the pore–substrate and pore–product levels originates from a reactive intermediate and can be used to explore their stability with regard to conditions such as pH. Reactions can be very generally detected with synthetic transport systems (indicated by reaction arrows in Figure 16a–d). The only condition is that the substrate and/or product activate or inactivate the transport systems, and that the IC50 or EC50 of substrate and product are sufficiently different. Then, changing ability of the reaction mixture to inactivate or activate the transport system can be easily followed as described in Section 4.7. Baseline discrimination is preferable to detecting reactions as “on/off” or “off/on” events. This technique has been used to produce fluorometric or colorimetric assays for enzymes as different as acetate kinase, aldolase, apyrase, cholesterol oxidase, citrate lyase, DNA exonuclease, DNA polymerase, elastase, esterase, ficin, galactosyltransferase, galactosidase, heparinase, hexokinase, hyaluronidase, invertase, lactate

Catalysis and sensing

The main applications of synthetic transport systems that open and close in response to chemical stimulation are catalysis and sensing. Sensing applications made much progress over the last few years, whereas the more challenging catalytic systems remained nearly unexplored. The only reported assay to study catalysis that takes place on the pathway across vesicle membranes uses entrapped HPTS acetate (Ac-HPTS) as fluorogenic substrate (Figure 17a).32, 33 Esterolysis during substrate efflux is detectable following HPTS fluorescence. This assay is compatible with Michaelis–Menten kinetics to extract KM and kcat . From there, catalytic efficiency and proficiency can be determined as in solution to report catalysis with the most relevant ground-state and transition-state stabilization. The assay holds for the determination of salt-rate profiles which are needed to extract n , a value similar to the Hill coefficient reporting on the number of operational ion pairs, the maximal efficiencies (kcat /KM )MAX , and from there the maximal transition-state stabilization.59 Combination with safranin O as external potential-sensitive probe (Section 4.4) in valinomycin-polarized vesicles is possible as well (Figure 17a). The voltage dependence of Michaelis–Menten kinetics can be used to determine longrange steering effects of membrane potentials to drive the substrate into the catalytic pore and to accelerate product release on the other side of the membrane.32 Catalytic pores have so far been beyond the reach of single-channel conductance measurements. However, covalent modifications within engineered biological ion

Stimulation PE V

+ V

−O

3S

O O

−O

3S

O

O−

Pore-substrate

OH

Pore-product Pore-intermediate 2

PI PE

−O

3S

SO3−

−O

3S

Substrate

SO3−

Products

Current (pA)

K+

t2

t1

Inside (a)

Closed

Outside Catalytic pore

Pore-intermediate 1

(b)

Time (ms)

Figure 17 Techniques to study (a) catalytic pores in unpolarized or polarized vesicles with entrapped fluorogenic substrates and (b) covalent modifications of ion channels and pores in single-channel conductance experiments. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes

inspired techniques is to attach the antigen of choice to a common blocker. Binding of the bulky and hydrophilic antibody then produces a soluble complex that unplugs the pore (Figure 18b). As common as antibodies but much easier to prepare, DNA aptamers have been widely considered as advantageous alternative to immunosensors. Recently, a technique that uses DNA aptamers as counterion-activated ion carriers in LUVs to both generate and transduce the analyte-specific signal has been introduced. The third technique to sense with transport systems is stochastic sensing.23, 61, 62 In this approach, the unique lifetime as well as conductance of single pore–analyte complexes is thought to produce a unique fingerprint that will reveal its presence in complex matrices. Today, it is quite clear that stochastic sensing is not applicable to complex matrices because any specific signatures are hidden behind very big noise. However, stochastic “sensing” remains a very attractive technique to address specific problems in relatively pure mixtures. The proposal to sequence single genes with pores has attracted much attention.62 Whereas it remains to be shown that the velocity of a single ss-DNA

oxidase, papain, phosphatase, phosphofructokinase, pronase, RNase, phytase, subtilisin, transaminase, triose phosphate isomerase (TIM), or tyrosinase.23, 60 This general detectability of the activity of enzymes demonstrated that synthetic transport systems can be used as multianalyte biosensors. Realized examples include cholesterol, citrate, glucose, lactate, lactose, phytate, polyphenol, and sucrose sensors. In these sensing systems, the synthetic transporters act as signal transducers, whereas the enzymes generate the signal by selectively recognizing and converting the analyte of interest in a complex matrix (Figure 18a). To generalize the use of synthetic transport systems as biosensors, signal amplifiers have been introduced. Signal amplifiers are bifunctional molecules that, on the one hand, covalently capture the product of the enzymatic signal generation and, on the other hand, activate or inactivate the signal transducer.23, 44, 60 Besides biosensing, the second technique developed to sense with transport systems is immunosensing.23 Several possibilities to attach either antigen or antibody to pore transducers have been reported, and one of the most

Substrate

Analyte

Productamplifier conjugate

Product

Signal generators

23

Signal amplifiers

enzymes (a)

Signal transducers

Signal detectors

Synthetic pores CPPs DNA

Fluorescence color, CD current

Antibody Blocker (b)

Antigen Open

Nucleotides

Current (pA)

Gene

Pore-C Pore-T Pore-A Pore-G

G, C, A, T

(c)

Exonuclease

Closed Time (ms)

Figure 18 (a) Biosensing, (b) immunosensing, and (c) stochastic sensing as leading techniques for sensing applications of synthetic (and bioengineered) transport systems. (a) In biosensing, enzymes are used to generate an analyte-specific signal that can be further amplified by bifunctional in-/activators for transduction by the transporter and detection with the techniques described in Section 3. (b) For immunosensing, antibodies (or aptamers) are coupled with the transporters, whereas (c) stochastic sensing aims to use the characteristic fingerprint from lifetime and conductances of analyte–pore complexes for, for example, gene sequencing. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

24

Techniques

rushing through a pore can be slowed down to reach singlebase resolution, the use of exonucleases is promising to bypass this intrinsic challenge (Figure 18c). The technique envisions attaching exonuclease to the pore and recording the produced nucleotide monophosphates according to the characteristic lifetime and conductance of their complex formed transiently in the pore. In this case, singlenucleotide discrimination and detectability of exonuclease activity are confirmed.

4.9

compensated by passive anion antiport or cation symport (electroneutral photosynthesis). The leading techniques to monitor artificial photosynthesis in lipid bilayer membranes are the HPTS assay and the Hurst assay (Figure 19).9, 65–67 In both assays, an external electron donor with appropriate redox potential, usually a tertiary amine such as ETDA, is combined with an internal electron acceptor with appropriate redox potential. To convert photonic into chemical energy, reduction of the internal acceptor by the external donor must be “uphill,” that is, thermodynamically unfavorable. In the HPTS assay, the internal acceptor is a quinone, which consumes two protons during its reduction with two electrons to a hydroquinone (Figure 19a).65, 66 The resulting increase in internal pH in response to irradiation with light is detected by the pH-sensitive HPTS (Section 3).

Photosynthesis

The central transport in artificial photosynthesis is transmembrane electron transfer in response to irradiation with light. This transmembrane charge separation either polarizes the membrane (electrogenic photosynthesis) or is − O3S

O3S −

OH

− O3S

SO3 −

O

EDTA+.

O Quinone

e−

Low pH

PI A I

DE

− O3S

− O 3S

−

OH

EDTA

O Photosystem

OH Hydroquinone O 3S

Inside

SO3

Outside

High pH

(a)

+FCCP

FCCP

Electrogenic

Light

Electroneutral

N

Co3+

lt

N N

N

+FCCP

H+

N

Dark control

N EDTA+. Co3+(bipy)3

(c)

Fractional activity Y

e− EDTA

Co2+(bipy)3

Photosystem (b)

Irradiation time (min)

Inside

Outside

(d)

1 0.5

EC50 (µM)

0 Concentration photosystem monomer

Figure 19 The HPTS assay and the Hurst assay to detect artificial photosynthesis. (a) In the HPTS assay, internal quinone reduction with light produces a proton gradient, which is detected by the pH-sensitive HPTS. (b) In the Hurst assay, internal cobalt reduction with light is detected by a change in color. (c) Electrogenic and electroneutral photosynthesis can be discriminated in the Hurst assay by the sensitivity toward the proton carrier FCCP. (d) Hill analysis of dose response curves delivers the basic information on the photosystem. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes The HPTS assay is elegant because artificial photosynthesis produces a pH gradient as in biology. However, the assay requires high selectivity of the photosystems because the minor leaks already present will cause the pH gradient to collapse. This also means that the HPTS works only with electrogenic photosystems. However, the pH gradient produced with quinone reduction has been coupled with ATP synthase to convert light into ATP by artificial photosystems in vesicles. In the Hurst assay, internal Co3+ (bipy)3 is used as electron acceptor, and the change in color during cobalt reduction with light is used to monitor photosynthesis (Figure 19b).9, 67 Operating exclusively with relatively large molecules, the Hurst assay is more robust than the HPTS assay. With the Hurst assay, electrogenic and electroneutral photosynthesis can be discriminated with the addition of external FCCP to enable proton-coupled electron pumping. Increasing activity in the presence of FCCP thus identifies electrogenic photosynthesis while insensitivity to FCCP is found for electroneutral photosynthesis (Figure 19c). As for other synthetic transport systems, artificial photosystems are best characterized in dose–response curves, with Hill analysis informing on activity (EC50 ), thermodynamic stability, and supramolecular nature of the photosystem (n, Figure 19d, Section 4.2).

5

EPILOGUE

This chapter aims to give an introductory summary to standard techniques used by and recommended for supramolecular chemists for the study of transport across bilayer membranes. Emphasis is on fluorescence techniques because they are most easily accessible for the nonexpert and can be used to address the most pertinent questions. The complementary information accessible with the more specialized planar bilayer conductance experiments is described as well. Emphasis is on the creation and identification of functional relevance, including topics such as ion selectivity, voltage gating, ligand gating, blockages, catalysis, sensing, and photosynthesis. The text is highly simplified. Meaningful design, execution, and interpretation of transport experiments are often an enriching and stimulating as well as very challenging adventure that requires the highest attention. Transport across bilayer membranes occurs in heterogeneous multiphasic environment and depends on numerous parameters. Multiple regions such as the aqueous phase, the highly charged head group region, and the nonpolar region of lipid tails exist and physicochemical properties change at each interfacial region. Particular attention is required to determine the importance of the not directly related delivery and partitioning to the final activity, with dichotomic

25

approaches being arguably the most relevant techniques to tackle the problem (Section 4.1). In general, failure to detect activity does not necessarily mean that the synthetic transport system does not exist, but interesting activities with high selectivities can be the most difficult ones to detect. Many more functional characteristics beyond the discussed selection exist. For instance, the influence of the composition of the membrane on activity has been almost completely ignored. This includes important topics such as membrane fluidity, phase transition, thickness (and hydrophobic matching), surface potentials 0 (and Gouy–Chapman theory),20, 22 partitioning, heterogeneity (“rafts”), or swelling and shrinking in response to stress. Mechanistic and structural studies are however not covered in this chapter. This was because the primary goal in research focusing on the creation of significant function is obviously to obtain experimental evidence for the desired significant function. Compared to any unique responsiveness of any synthetic transport system to physical or chemical stimulation, mechanisms of transport are simply less important, particularly considering that the different mechanisms (carrier, channel, pore, detergent) are often interchangeable depending on experimental conditions.16 To reiterate the fundamentals, ion channel activity is demonstrated with single-channel conductances; ion carriers work in the U-tube (decreasing activity with decreasing membrane fluidity does not prove carriers, as this can also be a simple partitioning effect).20 Structural studies are not included because they are usually obsolete and often misleading. Most significant synthetic transport systems are minority populations. Beware of overimplications from low-sensitivity, low-selectivity, and static techniques such as NMR, IR, X-ray, or transmission electron microscopy (TEM) data; they have caused much damage to the field! In this context, the most relevant structural data come from functional studies such as Hill analysis. Relevant structural studies operate under conditions relevant for function (micromolar to nanomolar concentrations, in membranes) and can specifically address a specific question in a selective manner. For example, FRET from a labeled transport system into an acceptor in the membrane can correctly inform on the role of partitioning and positioning in key processes such as activity, voltage gating, ligand gating,35 or blockage.35 Fluorescence depth quenching with DOXYL-labeled membranes is the most powerful tool to determine specific location, orientation, and repositioning in the membrane during these processes.20, 34, 36 Other selective and sensitive methods such as CD spectroscopy or isothermal titration calorimetry (ITC) are routinely used in meaningful structural studies, for example, to complement insights on interactions with the membranes, self-assembly

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

26

Techniques

of active structures, and the formation of host–guest complexes during ligand gating and blockage. The heart of the matter with synthetic transporters is to design and identify “smart systems” that respond to chemical or physical stimulation in a unique and significant manner. The techniques available to identify such significant functions can never be used without creative thinking and applying the grand principles of chemistry in a unique context. As a result, the techniques used are under constant development, never routine, calling for constant improvization and inspired innovation, depending on the system studied and the questions asked. This makes the use of transmembrane transport detection techniques very demanding and at the same time very entertaining and satisfactory.

16. T. Takeuchi, V. Bagnacani, F. Sansone, and S. Matile, ChemBioChem, 2009, 10, 2793. 17. A. Richard, V. Marchi-Artzner, M.-N. Lalloz, et al., Proc. Natl. Acad. Sci. U.S.A., 2004, 101, 15279. 18. M. Ma, Y. Gong, and D. Bong, J. Am. Chem. Soc., 2009, 131, 16919. 19. H. Matsuo, J. Chevallier, F. Vilbois, et al., Science, 2004, 303, 531. 20. S. Matile and N. Sakai, The characterization of synthetic ion channels and pores, in Analytical Methods in Supramolecular Chemistry, ed. C. A. Schalley, John Wiley & Sons, Weinheim, 2007, pp. 391–418. 21. N. Sakai and S. Matile, Chem. Biodiv., 2004, 1, 28. 22. N. Sakai and S. Matile, J. Phys. Org. Chem., 2006, 19, 452. 23. S. Matile, H. Tanaka, and S. Litvinchuk, Top. Curr. Chem., 2007, 277, 219. 24. B. Hille, Ion Channels of Excitable Membranes, 3rd edn, Sinauer Associates, Sunderland, MA, 2001.

ACKNOWLEDGMENTS We thank the University of Geneva and the Swiss NSF for financial support.

REFERENCES 1. A. L. Sisson, M. R. Shah, S. Bhosale, and S. Matile, Chem. Soc. Rev., 2006, 35, 1269. 2. S. Matile, A. Som, and N. Sorde, Tetrahedron, 2004, 60, 6405.

25. D. D. Lasic, Trends Biotechnol., 1998, 16, 307. 26. K. Kano and J. H. Fendler, Biochim. Biophys. Acta, 1978, 509, 289. 27. B. A. McNally, A. V. Koulov, B. D. Smith, et al., Chem. Commun., 2005, 1087. 28. G. Das and S. Matile, Proc. Natl. Acad. Sci. U.S.A., 2002, 99, 5183. 29. V. Gorteau, G. Bollot, J. Mareda, and S. Matile, Org. Biomol. Chem., 2007, 5, 3000. 30. K. D. Legg and D. M. Hercules, J. Phys. Chem., 1970, 74, 2114.

3. T. M. Fyles, Chem. Soc. Rev., 2007, 36, 335.

31. B. Baumeister, A. Som, G. Das, et al., Helv. Chim. Acta, 2002, 85, 2740.

4. A. P. Davis, D. N. Sheppard, and B. D. Smith, Chem. Soc. Rev., 2007, 36, 348.

32. N. Sakai, N. Sord´e, and S. Matile, J. Am. Chem. Soc., 2003, 125, 7776.

5. G. W. Gokel and N. Barkey, New J. Chem., 2009, 33, 947.

33. J.-L. Reymond, V. S. Fluxa, and N. Maillard, Chem. Commun., 2009, 34.

6. S. K. Berezin and J. T. Davis, J. Am. Chem. Soc., 2009, 131, 2458.

34. L. A. Weiss, N. Sakai, B. Ghebremariam, et al., J. Am. Chem. Soc., 1997, 119, 12142.

7. X. Li, B. Shen, X. Q. Yao, and D. Yang, J. Am. Chem. Soc., 2009, 131, 13676.

35. V. Gorteau, F. Perret, G. Bollot, et al., J. Am. Chem. Soc., 2004, 126, 13592.

8. R. E. Dawson, A. Hennig, D. P. Weimann, et al., Nature Chem., 2010, DOI: 10.1038/NCHEM.657.

36. N. Sakai, D. Gerard, and S. Matile, J. Am. Chem. Soc., 2001, 123, 2517.

9. A. Perez-Velasco, V. Gorteau, and S. Matile, Angew. Chem. Int. Ed., 2008, 47, 921.

37. K. Matsuzaki, O. Murase, N. Fujii, and K. Miyajima, Biochemistry, 1996, 35, 11361.

10. A. Hennig, L. Fischer, G. Guichard, and S. Matile, J. Am. Chem. Soc., 2009, 131, 16889.

38. B. D. Smith and T. N. Lambert, Chem. Commun., 2003, 2261.

11. J. T. Davis, P. A. Gale, O. A. Okunola, et al., Nature Chem., 2009, 1, 138.

39. S. M. Butterfield, A. Hennig, and S. Matile, Org. Biomol. Chem., 2009, 7, 1784.

12. B. A. McNally, E. J. O’Neil, A. Nguyen, and B. D. Smith, J. Am. Chem. Soc., 2008, 130, 17274.

40. A. Hennig and S. Matile, Chirality, 2008, 20, 932.

13. C. P. Wilson and S. J. Webb, Chem. Commun., 2008, 4007.

41. M. J. Pregel, L. Jullien, and J.-M. Lehn, Angew. Chem. Int. Ed., 1992, 31, 1637.

14. A. Satake, M. Yamamura, M. Oda, and Y. Kobuke, J. Am. Chem. Soc., 2008, 130, 6314.

42. J. M. Mahoney, A. V. Koulov, and B. D. Smith, Org. Biomol. Chem., 2003, 1, 27.

15. L. Ma, M. Melegari, M. Colombini, and J. T. Davis, J. Am. Chem. Soc., 2008, 130, 2938.

43. V. Sidorov, F. W. Kotch, Y.-F. Lam, et al., J. Am. Chem. Soc., 2003, 125, 2840.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Transport experiments in membranes 44. S. M. Butterfield, T. Miyatake, and S. Matile, Angew. Chem. Int. Ed., 2009, 48, 325. 45. S. Bhosale and S. Matile, Chirality, 2006, 18, 849.

27

56. G. Eisenman and R. Horn, J. Membr. Biol., 1983, 76, 197. 57. E. M. Wright and J. M. Diamond, Physiol. Rev., 1977, 57, 109.

46. F. Mora, D.-H. Tran, N. Oudry, et al., Chem. Eur. J., 2008, 14, 1947.

58. B. Hille and W. Schwarz, J. Gen. Physiol., 1978, 72, 409.

47. K. A. Connors, Binding Constants: The Measurement of Molecular Complex Stability, 1st edn, John Wiley & Sons, New York, 1987.

60. N. Sakai, J. Mareda, and S. Matile, Acc. Chem. Res., 2008, 41, 1354.

48. W. H. Chen, X. B. Shao, and S. L. Regen, J. Am. Chem. Soc., 2005, 127, 12727. 49. A. Som, N. Sakai, and S. Matile, Bioorg. Med. Chem., 2003, 11, 1363. 50. Y. Baudry, D. Pasini, M. Nishihara, et al., Chem. Commun., 2005, 4798. 51. V. Gorteau, G. Bollot, J. Mareda, et al., Bioorg. Med. Chem., 2005, 13, 5171. 52. J. S´anchez-Quesada, H. S. Kim, and M. R. Ghadiri, Angew. Chem. Int. Ed., 2001, 40, 2503. 53. N. Sakai, Y. Kamikawa, M. Nishii, et al., J. Am. Chem. Soc., 2006, 128, 2218. 54. R. S. K. Kishore, O. Kel, N. Banerji, et al., J. Am. Chem. Soc., 2009, 131, 11106.

59. A. Som and S. Matile, Chem. Biodiv., 2005, 2, 717.

61. S. H. Shin, T. Luchian, S. Cheley, et al., Angew. Chem. Int. Ed., 2002, 41, 3707. 62. J. Clarke, H. C. Wu, L. Jayasinghe, et al., Nat. Nanotechnol., 2009, 4, 265. 63. H. E. Hasper, N. E. Kramer, J. L. Smith, et al., Science, 2006, 313, 1636. 64. S. Majd, E. C. Yusko, A. D. MacBriar, et al., J. Am. Chem. Soc., 2009, 131, 16119. 65. D. Gust, T. A. Moore, and A. L. Moore, Acc. Chem. Res., 2001, 34, 40. 66. S. Bhosale, A. L. Sisson, P. Talukdar, et al., Science, 2006, 313, 84. 67. L. Zhu, R. F. Kairutdinov, J. L. Cape, and J. K. Hurst, J. Am. Chem. Soc., 2006, 128, 825.

55. N. Sakai, N. Sorde, G. Das, et al., Org. Biomol. Chem., 2003, 1, 1226.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc031

Vesicles in Supramolecular Chemistry Bart Jan Ravoo Westf¨alische Wilhelms-Universit¨at M¨unster, M¨unster, Germany

1 Introduction 2 Vesicles: Dynamic Supramolecular Structures 3 Conventional and Nonconventional Building Blocks for Vesicles 4 Stimuli-Responsive Vesicles 5 Molecular Recognition of Vesicles 6 Conclusion and Outlook References

1

1 1 4 8 10 14 14

INTRODUCTION

A bilayer of phospholipid molecule is all that separates “in” from “out” in living organisms. It was shown by Bangham and Horne in 1964 that phospholipid bilayer membranes can be easily formed in vitro,1 and it was reported by Kunitake and Okahata in 1977 that the formation of bilayers is not restricted to biological phospholipids.2 Vesicles (Lat. vesicula = small bubble) have been a versatile topic in supramolecular chemistry ever since. On the one hand, vesicles are of interest as highly dynamic supramolecular structures that mimic the remarkable properties of biological membranes (see Supramolecular Chemistry of Membranes, Supramolecular Aspects of Chemical Biology). On the other, vesicles are of interest as self-assembled responsive capsules that may be applied in drug delivery, as nanoreactors and nanosensors, or in the design of soft materials. This chapter reviews the state of the art of vesicles in supramolecular chemistry. The first section of this

chapter contains a general introduction to the structure and dynamics of vesicles as well as an overview of the most important methods to prepare and characterize vesicles. The second section describes the wide scope of molecular building blocks that can assemble into vesicles. Vesicles can be assembled from small, large, and giant amphiphiles, as well as from supramolecular amphiphiles. Vesicles composed of unusual building blocks as well as vesicles in organic solvents are also addressed. The third section of this chapter summarizes the literature on stimulus-responsive vesicles. The final section of this chapter reviews the most important recent developments regarding molecular recognition of vesicles according to the typical noncovalent interaction motifs: metal–ligand coordination, hydrogen bonding, and host–guest inclusion. The chapter closes with a brief outlook. This chapter is intended to provide insight to the fascinating supramolecular chemistry of vesicles by highlighting a selected number of recent publications, without providing a comprehensive review of the literature. Supramolecular ionophores and ion channels that transport ions and small molecules through membranes are covered in Transport Experiments in Membranes, Techniques and Membrane Transport, Supramolecular Aspects of Chemical Biology.

2

2.1

VESICLES: DYNAMIC SUPRAMOLECULAR STRUCTURES Structure and dynamics

Vesicles are dynamic supramolecular structures which consist of a molecular layer that encapsulates a small amount of solvent. The term “liposome” is generally reserved for vesicles composed of natural phospholipids, while

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

2

Techniques

the term “vesicle” includes those composed of synthetic amphiphiles, phospholipids, or any other components. “Polymersomes” are vesicles composed of polymers. Bilayer vesicles are closely related to liposomes and biological membranes. Most molecules that form bilayer vesicles in water are amphiphilic: they have a hydrophobic as well as a hydrophilic part. The hydrophilic part (“head group”) of the molecule interacts favorably with the surrounding water, while the hydrophobic part (“tail”) minimizes its exposure to water (see Introduction to Surfactant Self-Assembly, Concepts). Hence, the amphiphiles arrange in a bilayer and the formation of vesicles is driven primarily by hydrophobic interaction. Typically, the head group is polar and/or charged and contains phosphate, sulfate, ammonium, amino acid, peptide, carbohydrate, or oligo(ethylene glycol) groups. Typically, the “tail” is apolar and uncharged. The tail is usually composed of long hydrocarbon chains, which may be saturated or unsaturated; linear, cyclic, or branched; aromatic or aliphatic; or fluorinated. In accordance with the concept of the packing parameter,3 the amphiphile must have an approximately cylindrical shape, so that the molecules arrange into a bilayer, which may be slightly curved so that it can close into a spherical vesicle. If the “head” is substantially bulkier than the “tail,” the amphiphiles will tend to form micelles, not vesicles. If, on the other hand, the “tail” is larger than the “head,” the amphiphiles will assemble into inverted phases. It should be noted that the packing parameter cannot be defined exclusively on geometric considerations: attractive and repulsive interactions of head groups should also be taken into account. With the advent of polymersomes and vesicles of other “nonconventional” (i.e., not phospholipid-like) amphiphiles, numerous examples of monolayer vesicles in water have been reported. Typically, monolayer vesicles are prepared from small molecules with a hydrophobic core and two hydrophilic head groups (bolaform amphiphiles, see below) or from triblock-copolymers with two hydrophilic terminal blocks. The molecule must have a cylindrical or rectangular shape, so that it can arrange into a monolayer. Many amphiphiles also associate in organic solvents due to favorable electrostatic, dipolar, and/or hydrogenbonding interactions between the head groups. In this way, the formation of reverse vesicles (or inverse vesicles) is possible. In a reverse vesicle, the bilayer (or monolayer) is held together by polar interactions of the head groups (or core) in the interior of the molecular layer, while the apolar tails are exposed to the organic solvent. The number of reports on reverse vesicles is small, but steadily growing. An efficient method for the preparation of inverse liposomes was reported only recently.4 Irrespective of their composition, it is useful to differentiate between small unilamellar vesicles (SUVs, 1 µm

> 1 µm

k diff

k ff

Figure 1 Top: Small unilamellar vesicles (SUVs), large unilamellar vesicles (LUVs), giant unilamellar vesicles (GUVs), and multilamellar vesicles (MLVs). Bottom: Dynamic processes in bilayer membranes involve exchange (kex ), lateral diffusion (kdiff ), and flip-flop (kff ).

large unilamellar vesicles (LUVs, 100–1000 nm), giant unilamellar vesicles (GUVs, >1 µm), and multilamellar vesicles (MLVs). These different types of vesicles are illustrated in the top panel of Figure 1. SUVs, LUVs, and GUVs have a unilamellar membrane composed of a single molecular bilayer (or monolayer). SUVs and LUVs are the most widely studied types of vesicles. GUVs are of interest since their size is comparable to biological cells.5 MLVs have an onion-like structure and consist of many concentric bilayer (or monolayer) membranes. It can easily be calculated that the smallest SUVs (about 50 nm) of small amphiphiles contain about 10 000 molecules, whereas LUVs contain about a 100 000 molecules, and GUVs and MLVs contain many millions of molecules. The number of molecules in one vesicle can only be given as an approximate average, because it is impossible to prepare vesicles of an exactly defined size (Section 2.2). Vesicles are flexible and dynamic colloids. First of all, vesicles display Brownian motion in solution and the diffusion rate of a vesicle correlates inversely with its hydrodynamic radius (SUV < LUV < GUV ≈ MLV). Second, the membrane of a vesicle is often flexible, so that the vesicle can easily change shape provided that total surface area remains constant. It should be noted that membranes of polymersomes can be very rigid. Third, the molecules

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry

3

Figure 2 Schematic representation of vesicle fusion. Fusion of vesicles involves aggregation of vesicles, formation of intervesicular complexes and contact of the outer bilayer leaflets, formation of a bilayer stalk, opening of a fusion pore, and merger of the vesicles.

can diffuse laterally in the membrane. The lateral diffusion rate (kdiff ) strongly depends on the nature of the molecule as well as on temperature: shorter hydrophobic tails and higher temperatures generally increase the lateral diffusion. Vesicles can be characterized by a critical phase transition temperature Tc : above Tc , the molecules diffuse freely and the membrane is a two-dimensional fluid, while below Tc the molecules diffuse very slowly and the membrane is in a gel-like state. It should be emphasized that a vesicle membrane is not simply a two-dimensional fluid, since certain components may cluster into domains. In addition, many polymersomes have a solid-like bilayer (or monolayer) membrane, with no significant lateral mobility of the membrane components. Fourth, molecules in bilayer membranes can “flip-flop” from the inside leaflet to the outside leaflet of the membrane. Phospholipid flip-flop is pivotal in the assembly and maintenance of asymmetric biological membranes. The rate of flip-flop (kff ) strongly depends on the nature of the head group and of the tail as well as temperature: less polar, weakly solvated head groups, short tails, and high temperatures generally increase the transverse mobility. The rate of flip-flop in polymersomes can be insignificantly small. Fifth, molecules can temporarily leave the vesicle membrane, diffuse through the aqueous surroundings, and eventually readsorb in a vesicle. The rate of this exchange process (kex ) correlates with the water solubility of the molecule, which is generally low (nanomolar or less), and orders of magnitude lower than for amphiphiles that form micelles (millimolar or more), but it should not be neglected. The most important dynamic processes in vesicles are illustrated in the lower panel of Figure 1. Vesicles interact with each other. Aggregation, fusion, and fission are the most common processes (Figure 2). Aggregation means that vesicles aggregate into large clusters of vesicles, while each vesicle stays intact, so that the contents of the vesicles do not mix or leak out. Fusion means that vesicles merge into large vesicles, so that their membrane components and the contents of the vesicles mix (sometimes with leakage). Fission is the reverse of fusion: small vesicles are formed from large ones. Aggregation and fusion occur slowly but spontaneously in any

vesicle solution and eventually lead to flocculation and precipitation of the vesicles as hydrated multilayers. However, aggregation and fusion can also be triggered by a number of stimuli, including molecular recognition (Section 5).

2.2

Preparation and characterization

With a few notable exceptions (Sections 3.3 and 3.4), vesicles are metastable in aqueous solution: energy is required to dissolve the molecular components in water and induce the formation of vesicles, and eventually the vesicles flocculate and precipitate as hydrated multilayers. The energy required to obtain vesicles is usually provided by heat, stirring, ultrasonication, extrusion, or combinations of these. The preparation of vesicles is largely inspired by the protocols established for the preparation on liposomes.6 Typically, a lipid (or a synthetic amphiphile, or any other vesicle component) is first dissolved in an organic solvent such as chloroform. The organic solution is evaporated to dryness, leaving a thin film of lipid (or amphiphiles, or other components) arranged in multilayers on the wall of the vial. An aqueous solution is added, and the multilayers swell with water and slowly detach from the wall of the vial, resulting in a dispersion of MLVs. The formation of MLVs can be accelerated by heating, shaking, stirring, and/or freeze-drying. MLV can be processed to LUVs or SUVs by various methods. The most important methods include ultrasonication and extrusion. For ultrasonication, a vial with the MLV solution is placed in an ultrasound bath for 15–60 min. During ultrasonication, MLVs break up and form LUVs, then SUVs. Generally, vesicles become smaller with prolonged ultrasonication, but it is difficult to control the average vesicle size. Extrusion can be performed on a small scale (milliliters) by hand or on a larger scale (>100 ml) using a press. The MLV solution is pressed through polycarbonate membranes with a defined pore size. Upon extrusion, MLVs break up as a result of the shear force exerted on them, and LUVs or SUVs of a rather narrow size range are formed, depending on the pore size of the membrane.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

4

Techniques

Another important method for the preparation of vesicles is the so-called detergent-aided method. This method was originally developed for the preparation of liposomes, with the specific advantage that also membrane proteins and other sensitive components can be included in the vesicles because of the extremely mild conditions of preparation. In short, the vesicle components are dissolved in a concentrated detergent solution. In this solution, all components are dissolved in micelles. This micellar solution is either diluted below its critical micelle concentration or dialyzed to remove the detergent, while the vesicle components assemble into vesicles. Typically, LUVs with a rather broad size distribution result. The detergent-aided method is also very well suited to the preparation of polymersomes. Vesicles can also be prepared by injection of a concentrated solution of amphiphiles (or any other component) in an organic solvent such as diethyl ether, tetrahydrofuran (THF) or dimethylformamide (DMF) into an aqueous solution. The organic solvent is removed by evaporation or by dialysis. This method generally provides LUVs with a rather broad size distribution. The injection method is suitable to the preparation of polymersomes, but it is less suitable for the preparation of vesicles containing proteins, since they tend to denature in the organic solvent. In some cases, vesicles can also be obtained simply by dissolving and mixing the required components in aqueous solution, that is, purely by self-assembly. For example, the so-called catanionic vesicles self-assemble when cationic and anionic amphiphiles (each in the form of a micellar solution) are mixed in an equimolar ratio.7 Also supramolecular amphiphiles (Section 3.3) form vesicles by self-assembly when two (or more) components are mixed in the appropriate molar ratio. GUVs are usually obtained by electroformation.8 To this end, a multilayer film of amphiphiles is deposited on an electrode surface. The multilayer is swelled with water under the influence of an AC or DC electric field. As a result, GUVs detach from the electrode surface. It is very difficult to tailor the size of GUVs by electroformation. Recently, a microfluidic preparation of GUVs was presented, which provides monodisperse, cell-sized, and unilamellar vesicles.9 This method is an important breakthrough in the preparation of GUVs. Vesicles can be characterized with the help of a range of physicochemical methods. Microscopy is an essential tool in all investigations of vesicles. MLVs and GUVs can be observed in real time by optical and fluorescence microscopy. LUVs and SUVs are best observed by transmission electron microscopy (TEM), ideally by stain-free methods such as cryogenic transmission electron microscopy (cryo-TEM), which can easily resolve the individual membrane secluding each vesicle. Depending on the stability of vesicles in the absence of solvent, vesicles can

also be investigated by atomic force microscopy (AFM) and scanning electron microscopy (SEM). The size distribution of vesicles can be determined from histograms collected from microscopy images. Alternatively, dynamic light scattering (DLS) is a powerful method to measure the average size and size distribution of vesicles. The surface potential of vesicles can be determined by the ζ -potential measurement and by capillary electrophoresis. The critical phase transition temperature (Tc ) is typically determined by differential scanning calorimetry (DSC). Fluorescence spectroscopy is a particularly powerful method to investigate the arrangement of probe molecules in a vesicle membrane, since the fluorescence spectrum is highly sensitive to the polarity of the microenvironment in the bilayer, the formation of fluorescence resonance energy transfer (FRET) pairs, and the formation of excimers. The application of many of these methods will be illustrated in the following sections.

3

3.1

CONVENTIONAL AND NONCONVENTIONAL BUILDING BLOCKS FOR VESICLES Small amphiphiles

The first report on bilayer vesicles formed from synthetic amphiphiles dates from 1977, when Kunitake and Okahata described the formation of vesicles from di-n-dodecyl dimethyl ammonium bromide in aqueous solution.2 In the 1980s, it was shown by many groups that a wide range of amphiphilic molecules can form vesicles in water. In a sense, these amphiphiles are all very similar to phospholipids: they generally have two hydrophobic tails and a hydrophilic head group, so that the molecule has a cylindrical shape and packs efficiently into a bilayer sheet, which closes into a vesicle. On the other hand, the structural variety of synthetic amphiphiles provided vesicles with a range of functions that clearly surpass the properties of liposomes. Among others, synthetic vesicles can be made light sensitive, pH sensitive, and polymerized. The pioneering work on vesicles is summarized in reviews by Kunitake,10 Ringsdorf et al.,11 and Engberts and Hoekstra.12 Small amphiphiles need not necessarily have a phospholipid-like structure with two tails and one head group. For example, bolaform (or: bipolar) amphiphiles are amphiphilic molecules that contain two head groups separated by an extended hydrophobic chain. Bolaform amphiphiles form monolayer vesicles in which each amphiphile extends across the monolayer membrane, exposing both head groups to water and sheltering the hydrophobic chain from water.13 However, also these types of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry vesicle-forming amphiphiles are in fact inspired from nature: many extremophilic bacteria have membranes that contain a high percentage of bolaform amphiphiles.14 Monolayer membranes of bolaform amphiphiles are much more robust than bilayer membranes and contribute to the stability of extremophiles in acidic, alkaline, or hot environments. The innovative design of small amphiphiles continues to give rise to functional vesicles. The first example of vesicles composed of amphiphilic fullerenes was reported in 1999: it was shown by TEM that a dimethylammonium derivative of C60 forms SUVs in water upon ultrasonication and extrusion.15 Very recently, highly stable vesicles were also prepared from fluorinated amphiphilic C60 (Figure 3).16 In addition, vesicles can be prepared from amphiphilic triangular17 and rectangular oligo(phenylene ethylene)s18 as well as linear oligo(phenylene vinylene).19 Vesicles of such πconjugated amphiphiles are stabilized by π –π interactions. In the latter case, energy transfer between donors and acceptors in vesicles composed of a mixture of electron-rich and electron-poor oligo(phenylene vinylene)s could be observed by fluorescence spectroscopy and microscopy. Vesicles can also be prepared from amphiphilic perylenes.20 If the perylene vesicles are loaded with pyrene, the fluorescence of the vesicles is a sensitive pH indicator due to the pH-dependent

K+ Fluorous chains Anionic site Fullerene (a)

Fluorous-fullerene amphiphile

Bilayer vesicle

(b)

Figure 3 Vesicles composed of fluorinated fullerenes. (a) Model of an amphiphilic fluorinated fullerene anion which forms bilayer vesicles in water. (b) SEM image of a vesiclecovered substrate viewed with 80◦ tilting of the sample stage. (Reproduced from Ref. 16.  Wiley-VCH, 2010.)

5

FRET of the perylene in the vesicle bilayer and the pyrene in the vesicle interior.

3.2

Large and giant amphiphiles

A major innovation in the area of vesicles was triggered by Eisenberg and coworkers, who demonstrated in 1995 that also very large amphiphilic molecules can form vesicles.21 In a pioneering report in Science, it was shown that polystyrene-b-poly(acrylic acid) can form bilayer vesicles in water. These “polymersomes” were prepared by slow addition of water to a DMF solution of the block copolymer, followed by dialysis to remove the remaining DMF. The hydrophobic polystyrene forms the interior of the bilayer membrane, while the hydrophilic poly(acrylic acid) is exposed to water. It has been shown since that many block copolymers can form vesicles. Important advantages of polymersomes include their high kinetic stability and their very low membrane permeability (which increases with the length of the hydrophobic block). Although block copolymers that merely contain a hydrophobic block connected to a hydrophilic block can still be considered rather straightforward high molecular weight analogs of conventional small amphiphiles, the field of polymersomes has tremendously benefited from the design of more complex block copolymer architectures using new polymerization methods (such as atom transfer radical polymerization, ATRP) and highly efficient conjugation protocols (such as click chemistry). For example, ABA block copolymers can form monolayer vesicles. In fact, ABA block copolymers are the macromolecular equivalent of bolaform amphiphiles. ABC block copolymers can form bilayer vesicles if A and B (but not C), or B and C (but not A), are of similar polarity, but they can also form monolayer vesicles if A and C (but not B) are of similar polarity. Figure 4 outlines the most important block copolymer architectures for polymersomes. Block copolymers can also contain biopolymer segments, such as polypeptides or polysaccharides conjugated to synthetic segments (“biohybrid copolymers”).22, 23 It was shown that even amphiphilic homopolymers can form vesicles.24 Such amphiphilic homopolymers are the equivalent of polymerized bolaform amphiphiles. Also, the formation of vesicles composed of amphiphilic dendrimers (“dendrimersomes”) has recently been described.25 The blossoming field of polymersomes has been the subject of several reviews.26, 27 The versatility of polymersomes was significantly advanced by Nolte and coworkers, who expanded the scope from large to giant biohybrid amphiphiles.28 The key innovation in their work is the conjugation of very large hydrophilic proteins to hydrophobic synthetic polymers. These biohybrid block copolymers differ from other

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

6

Techniques

Figure 4

AB copolymers

ABA copolymers

BAB copolymers

ABABA copolymers

ABC copolymers

ABCA copolymers

Block copolymers for polymersomes. (Reproduced from Ref. 27.  Royal Society of Chemistry, 2009.)

protein–polymer conjugates in the sense that the protein to polymer ratio is predefined and the position of the conjugation site is precisely known. In a particularly elegant experiment, giant biohybrid amphiphiles self-assembled by cofactor reconstitution of polystyrene-modified heme and apo-horse radish peroxidase (HRP) as well as apomyoglobin. The biohybrid amphiphiles were obtained by adding a THF solution of the heme cofactor-appended polystyrene to an aqueous solution of the apoenzyme. TEM revealed the formation of LUVs with diameters of 80–400 nm. The activity of the HRP and myoglobin enzymes is retained in the polymersomes. An interesting novelty was reported by a German team in 2009 when they described vesicles composed of amphiphilic nanoparticles.29 It was demonstrated with TEM and fluorescence microscopy that CdSe/CdS core–shell nanoparticles with a brush-like coating of poly(ethylene oxide) form LUVs and GUVs. The vesicle wall is composed of a single layer of nanoparticles. Nanoparticle vesicles constitute a new class of organic–inorganic hybrid vesicles.

3.3

Supramolecular amphiphiles and self-assembled vesicles

Vesicle-forming amphiphiles must not be held together by covalent interactions exclusively: it is easily conceivable that an amphiphile is formed by noncovalent interaction of two (or more) components. Hence, although the individual components cannot form vesicles, vesicles selfassemble upon mixing of the components in the appropriate molar ratio. In this respect, the first example of

self-assembly of vesicles is the preparation of catanionic vesicles by Kaler and coworkers.7 It was found that, when a micellar solution of sodium dodecylbenzene sulfonate is mixed with a micellar solution of hexadecyl trimethylammonium tosylate, vesicles composed of a 1 : 1 mixture of anionic and cationic surfactant are spontaneously formed. It can be argued that the surfactants from tight ion pairs, which assemble into vesicles. It was found that catanionic vesicles can be assembled from a rather large variety of simple surfactants.30 In contrast to most vesicles, catanionic vesicles are thermodynamically stable supramolecular structures. A remarkable example of a ternary complex that selfassembles into vesicles was reported by Kim and coworkers.31 It was shown that vesicles are formed spontaneously in a mixture of cucurbit[6]uril, n-alkyl viologen, and dihydroxynaphtalene (Figure 5). Viologen and dihydroxynaphthalene form a stable charge transfer complex in the cavity of the cucurbituril host. The ternary complex is amphiphilic due to the presence of the long alkyl chain on the viologen. If the alkyl substituent is n-dodecyl, SUVs are formed; if the alkyl substituent is n-hexadecyl, LUVs are formed. The vesicles can be imaged by SEM, demonstrating the robustness of the supramolecular structure. In a comparable approach, it was recently shown by a Chinese team that also a ternary inclusion complex of β-cyclodextrin, 1-naphtylammonium chloride, and sodium bis(2-ethyl-1-hexyl)sulfosuccinate forms vesicles in aqueous solution.32 Polymersomes can also be assembled from large supramolecular amphiphiles. Recently, an elegant method for the preparation of vesicles through self-assembly of supramolecular graft copolymers was developed.33

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry

+

OH

+

N–R +

CH3 –N

HO MV2+ : R = −CH3 C1VC122+ : R = −(CH2)11CH3 C1VC16

2+

: R = −(CH2)15CH3

+

CH3 –N

≡

+

N–R OH

HO

(A)

1: R = − CH3 2: R = − (CH2)11CH3 3: R = − (CH2)15CH3

(b)

(a)

500 nm

(c)

500 nm

1 µm

Vesicles from unusual building blocks and vesicles in organic solvents

DHNp O N N-CH2 H H N N-CH2 8 O

CB[8] ≡

3.4

7

2 µm

(B)

Figure 5 Vesicles composed of a ternary complex of cucurbit[6]uril, n-alkyl viologen, and dihydroxynaphtalene. (A) Formation of a charge transfer complex of n-alkyl viologen and dihydroxynaphtalene in the cavity of cucurbit[6]uril. (B) TEM and SEM images of complex 2 and 3. (Reproduced from Ref. 31.  Wiley-VCH, 2002.)

Two types of polymersomes were prepared through selfassembly. For the first supramolecular graft copolymer, hydrophobic poly(4-vinylpyridine) was used as main chain and hydrophilic poly(N-vinylpyrrolidone) with carboxylic end groups was grafted onto the main chain through ionic interaction between the carboxylic groups and pyridine groups. For the second supramolecular graft copolymer, hydrophobic poly(4-acrylamidobenzoic acid) was used as main chains and poly(N-vinylpyrrolidone) with amino end groups was grafted onto the main chains through ionic interaction between the amino groups and carboxylic groups. Both vesicles were sensitive to pH. It was recently reported that vesicles are also formed spontaneously when polystyrene with pendant hydrophilic Au nanoparticles is mixed with polystyrene-coated Fe3 O4 nanoparticles.34 Nanoparticle vesicles are assembled when a homogeneous solution containing both nanoparticles in THF is slowly diluted with water under ultrasonication.

In recent years, it has become clear that vesicles can also be assembled from building blocks that are in no way reminiscent of phospholipids. In fact, the building block must not even be amphiphilic. This section highlights a number of examples of truly “nonconventional” building blocks for the assembly of vesicles. An Israeli–Indian team reported the formation of allpeptide vesicles in Ghosh et al.35 It was shown by TEM, SEM, and fluorescence microscopy that a trimer of ditryptophan assembles into vesicles in a mixture of water and methanol. The peptide forms a π-stacked network across the surface of the vesicle, similar to the cytoskeleton in biological cells. The formation of vesicles can also be induced by electrostatic interaction or “electrostatic self-assembly.” Among others, it was shown that LUVs are obtained when a polycation is mixed with a polyanion-b-poly(ethylene glycol).36 The vesicle size can be controlled by changing the polymer concentration. Vesicles are also obtained when a polyanionb-poly(ethylene glycol) is mixed with a cationic azobenzene surfactant.37 In that case, the vesicles are photoresponsive, that is, they dissemble if the azobenzene is isomerized from the trans form to the cis form. Finally, vesicles can also be prepared by mixing a cationic dendrimer and an anionic sulfonate dye.38 Araki and coworkers reported the formation of highly stable LUVs composed of two-dimensional hydrogen-bonded sheet structures of guanosine.39 The building block for these LUVs is a guanosine substituted with a phenylsilyl unit and an oligo(ethylene glycol unit). This molecule is not an amphiphile! Nevertheless, it assembles into vesicles. It is shown that the membrane of the vesicles is stabilized by a two-dimensional hydrogen-bonding network of guanosine, while the oligo(ethylene glycol) units are exposed to water. Vesicles can also be composed of inorganic building blocks. A recent report shows that polyoxometallates can form stable vesicles in a mixture of water and acetone.40 In this case, vesicles assembled from a hybrid of two anionic polyoxometallate clusters linked by a bifunctional organic ligand in the presence of tetra-n-butylammonium counterion. Furthermore, vesicles can also be assembled from porphyrins in a mixture of chloroform and methanol.41 By using TEM, AFM, and DLS, it was shown that the porphyrins arrange into reverse bilayer vesicles. The vesicles can even be observed in vacuum with SEM (Figure 6). The porphyrin bilayer interior is stabilized by π-stacking as well as hydrogen bonding of carboxylic acid residues,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

8

Techniques

1 µm

(a)

1 µm

(b)

0.2 0.4 0.6 0.8

µm

(c)

Figure 6 SEM (a and b) and AFM (c) images of vesicles of a Mn3+ –porphyrin complex in a mixture of chloroform and methanol. (Reproduced from Ref. 41.  Royal Society of Chemistry, 2009.)

while the alkyl substituents are exposed to the organic solvent. Schmuck and coworkers reported the formation of vesicles in dimethylsulfoxide (DMSO).42 The vesicles are composed of a monolayer of a self-complementary biszwitterion, which interacts through hydrogen bonding in the interior of the membrane as well as through dimerization of the zwitterion at the vesicle surface. Recently, there has been a fascinating report on reverse vesicles formed by hydrogen-bonded tri-ammonium cyclophanes and hexa-ammonium capsules.43 Using a range of methods, it was shown that these cationic macrocycles and cages can form reverse monolayer vesicles in chloroform and dichloromethane due to a combination of polar interactions inside the monolayer and exposure of the alkyl substituents to the solvent. Finally, it was recently demonstrated that polybutadieneb-poly(ethylene oxide) can also form vesicles in ionic liquids.44 This is the first report describing the formation of polymersomes (or any other type of vesicle) in ionic liquid.

4

STIMULI-RESPONSIVE VESICLES

By nature of their dynamic supramolecular structure, many vesicles can respond to a change in their environment. The design of stimuli-responsive vesicles that become permeable, disassemble, or change shape in response to

an external trigger is a promising strategy in the development of drug delivery systems as well as adaptive soft materials. Stimuli-responsive polymersomes have been the subject of a recent review.45 A number of recent examples of responsive vesicles are highlighted in this section. Many vesicles change size, shape, or permeability in response to a pH change. An early example includes the work of Klok and Lecommandoux, who showed that polymersomes composed of a polybutadiene-b-polypeptide shrink at low pH and swell at high pH.46 The change in size is fully reversible and attributed to a pH-dependent helix–random coil transition of the peptide segments. Eisenberg and coworkers recently described “breathing” vesicles of a triblock copolymer that display a reversible pH-induced volume change by a factor of 7.47 Another report showed that cyclodextrin vesicles (CDVs) (Section 5.3) can be decorated with peptides functionalized with an adamantane anchor.48 It was found that a (LeuGlu)4 octapeptide can induce a pH-dependent shape transformation of the vesicles: at pH 7.4, the peptide merely binds to the vesicle surface, whereas at pH 5.0 it forms a β-sheet and transforms the vesicles into a nanotube (Figure 7). It was shown that the vesicles release their contents as a result of this shape transformation. It should be emphasized that the pH range of this shape transformation matches the decrease in pH that occurs upon endosomal uptake by cells. Hence, these experiments suggest that the peptide-decorated CDVs may be a useful vehicle for intracellular delivery of drugs or antigens that are encapsulated inside the vesicle or bound on the surface of the vesicle. Vesicles can also respond to a temperature change. A spectacular example of temperature-responsive vesicles was reported by a Korean team, who showed that polymersomes composed of the so-called rigid rod amphiphiles [a rigid aromatic segment substituted with branched oligo(ethylene oxide) at one end] are porous at room temperature but impermeable at 65 ◦ C.49 The gating of the pores can be explained by the fact that the oligo(ethylene oxide) dendritic exterior exhibits a lower critical solution temperature (LCST) behavior in water. It should be noted that for many applications it would be desirable to design vesicles that are impermeable at room temperature but are porous upon heating. Another outstanding example of a stimuli-responsive polymersome was recently reported by van Hest and coworkers.50 It was demonstrated that polymersomes composed of a mixture of poly(ethylene glycol)-b-polystyrene and poly(ethylene glycol)-b-poly(styrene boronic acid) become permeable in the presence of D-glucose or D-fructose, which bind to the poly(styrene boronic acid) segments and hence make them hydrophilic instead of hydrophobic (Figure 8).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry

9

O HO R

O

S

C12H25

≡

O 7

R = −(CH2CH2O)n H, with n = 1 - 3.

1

CDV

O

O NH

NH

O NH

OH

O

O

NH

NH

NH2

O

O OH

O

O NH

NH

O

O O

NH

OH

O

OH

2

pH 5.0 pH 7.4

CDV + 2

Figure 7 Molecular structures of amphiphilic β-cyclodextrin derivative 1, which self-assembles into CDVs, and adamantane-modified octapeptide 2, which binds to the CDVs by host–guest interaction. Peptide 2 adapts a random coil conformation at pH 7.4 while bound to the vesicles. Upon acidification to pH 5.0, peptide 2 forms a β-sheet which induces a morphological change from a vesicle to a nanotube, with concomitant release of contents.

Vesicles can also be photoresponsive. Das and coworkers reported the formation of vesicles from donor–acceptor substituted butadiene amphiphiles.51 Small vesicles associate into large vesicles, which eventually cluster into a hydrogel. The hydrogel dissolves upon photoirradiation. The authors propose that the aggregation of the vesicles is due to π -stacking, which is weakened by photoisomerization of the butadiene amphiphiles. Zhang and coworkers described photoresponsive vesicles composed of supramolecular amphiphiles.52 The amphiphile is composed of an inclusion complex of α-cyclodextrin and azobenzene. Vesicles are not formed when the azobenzene is in the trans state and included in the cyclodextrin, but are formed only when the azobenzene is in the cis state, so

that the azobenzenes can form a bilayer. The formation of vesicles is fully photoreversible. A sensor is a system that displays a readily detectable response in the presence of a specific analyte. Indeed, stimulus-responsive vesicles have been tailor-made to function as highly specific sensors. The overwhelming majority of vesicle-based sensors are based on a very simple type of amphiphile: polydiacetylenes. These polymeric amphiphiles are easily formed from simple diacetylene amphiphiles by in situ photopolymerization of vesicles. If the vesicles are additionally equipped with ligand or receptor groups, the absorbance and fluorescence of the conjugated polymer backbone is highly sensitive to the presence of metal ions, anions, and small as well as large

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

10

Techniques O O

O

22

Br

O

O

m

O

B

O

O 22

O

O

O

H+

Br

O

OH−

m

HO

B

22

O

Br m

O HO

OH

O

HO B OH OH

OH

(a)

Stimuli-responsive block copolymers

O

O

22

O

Br m

O B O O H

Stimuli

Enzyme Inert block copolymers

Stimuli

(b)

Figure 8 (a) Molecular structure of poly(ethylene glycol)-b-poly(styrene boronic acid) and its equilibrium with carbohydrates in water at pH > 7. In the absence of carbohydrate, the block copolymer is amphiphilic and forms polymersomes. In the presence of carbohydrate, the block copolymer is hydrophilic and does not form polymersomes. (b) Schematic representation of the formation of permeable nanoreactors using the carbohydrate response of the block copolymers. The encapsulated enzyme catalyzes the hydrolysis of esters. (Reproduced from Ref. 50.  Wiley-VCH, 2009.)

biomolecules. The application of polydiacetylene vesicles as colorimetric sensors has been reviewed by Jelinek and Kolusheva.53

5

MOLECULAR RECOGNITION OF VESICLES

The interaction of vesicles with molecules in the surrounding solution is a particularly fascinating topic. Small and large molecules can bind to the bilayer membrane and mediate the interactions between vesicles. In this respect, synthetic vesicles are versatile model systems for the protein- and carbohydrate-mediated recognition, adhesion, and fusion of membranes that occur during endocytosis, viral infection, cell adhesion, and the growth of tissue from individual cells. Specific molecular interactions at membrane surfaces are influenced by the local environment at the membrane surface, which is a hydrophobic–hydrophilic interface. The confinement and organization of molecules in a membrane results in high local concentrations (up to 1 M), which are very unusual in homogeneous solutions. A closely related issue is the topic of multivalency (see Multivalency, Concepts). Molecular recognition at biological membranes, such as proteins binding to carbohydrate

residues of glycolipids, is usually mediated by multivalent (n : n) instead of monovalent (1 : 1) interactions. Multivalent interactions are not only stronger but usually more selective than monovalent interactions. Moreover, they can be attenuated or amplified by competing ligands. Multivalent interactions benefit from the high local concentration of interacting molecules at the membrane–water interface. It should be emphasized here that a biological membrane is not simply a two-dimensional fluid, since certain lipids and proteins may be present in domains.54 In biological membranes, multivalent interactions can be strengthened by receptor clustering or the formation of “lipid rafts” in fluid membranes. Hence, receptor clustering and raft formation are intimately linked not only to membrane fluidity but also to the efficiency of multivalent interactions. In synthetic vesicles, interacting molecules can be diluted or concentrated by mixing with inert amphiphiles, and lateral diffusion in the membrane is a function of temperature and chain length of the membrane components. A fluid bilayer can behave as an adaptable matrix in which interacting species will find optimal multivalent binding modes, exclusively accessible by clustering of binding units on the membrane surface. Molecular recognition of liposomes and vesicles has been the subject of a recent tutorial review.55

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry

5.1

Metal coordination of vesicles

In the last decade, several groups have demonstrated that metal ions can be coordinated to vesicles in a highly selective manner if suitable amphiphilic ligands are embedded in the membrane. In certain cases, coordination of metal ions to vesicles results in the formation of unusual complexes and remarkable changes in the lateral organization of the membrane. Three representative examples are highlighted below. Doyle and coworkers showed that ethylenediamine ligands inserted in liposomes bind selectively and cooperatively to Cu2+ in (common) CuL2 as well as (unusual) CuL4 complexes.56 The ligand was equipped with a cholesterol anchor and a fluorescent dansyl group that is quenched upon complexation of Cu2+ , so that metal complexation at the vesicle surface can be quantified with fluorescence spectroscopy. The metal–ligand complexation constants are much higher for the membrane-bound ligand than for the free ligand in solution. This effect is based on the fact that membrane-bound ligands are confined in a less polar environment. Increasing the ligand concentration on the vesicle membrane resulted in formation of CuL4 complexes in the membrane. These complexes do not form in solution. In a comparable approach, Smith and Jiang investigated the coordination of a Zn2+ – dipicolylamine (DPA) receptor complex to a fluorescent ligand at the surface of liposomes.57 The fluorescence of the ligand is quenched upon coordination to the unsaturated Zn2+ – DPA receptor complex. Titration of the vesicles with embedded Zn2+ – DPA receptor with ligand revealed a mixture of 1 : 1 and 1 : 2 binding modes. At higher receptor density, the dominant species is the 2 : 1 complex. Cationic lipids enhance the complexation by inducing clustering of the receptor. On the other hand, Sasaki and coworkers inserted a pyrene-modified crown ether into liposomes.58 The lateral distribution of the crown ether in the membrane depends on the presence of metal ions, and the relative intensity of

11

monomer and excimer emission of the pyrene group reflects the extent of clustering and phase separation in the vesicle membrane.59 When incorporated into liposomes, the crown ether clustered into domains, as evidenced by the strong pyrene excimer emission. Recognition and binding of Pb2+ ions at the membrane surface resulted in a distribution of the crown ether lipids in the membrane, revealed by a decrease in excimer emission. It was proposed that electrostatic repulsion of the bound Pb2+ induced distribution of the crown ether complex in the membrane. Lehn and coworkers investigated vesicle aggregation and fusion mediated by the coordination of Eu3+ to an amphiphilic diketone.60 The addition of Eu(NO3 )3 induced an increased average size of the vesicles which could be detected by cryo-fracture electron microscopy. The interaction of vesicles induced by Ni2+ and Co2+ binding to an amphiphilic dipyridine was also reported by Lehn and coworkers.61 The addition of Ni2+ and Co2+ ions to a solution of LUVs containing 3 mol% of amphiphilic dipyridine led to fusion of LUVs and formation of MLVs (Figure 9). To investigate the fusion process, LUVs were fluorescently labeled with rhodamine. In 2005, Webb and coworkers showed that it is possible to construct networks of vesicles linked by multiple metal–ligand coordination.62 Vesicle fusion depends on the ability of ligands to cluster in a membrane, just like the adhesion and fusion of biological membranes depends of the clustering of cell adhesion molecules in domains. Vesicle fusion was induced by the interaction of Cu(iminodiacetate) (IDA) complexes and poly(L-histidine), which functions as a supramolecular glue. Changes in the lipid distribution in the vesicles could be directly visualized through pyrene group fluorescence and vesicle adhesion was monitored by the increasing turbidity of the vesicle solution. An even more sophisticated approach to the metal–ligand coordination-mediated adhesion of vesicles mimics the formation of biological lipid rafts.63 The investigation was

Figure 9 Formation of giant vesicles by fusion of LUVs equipped with an amphiphilic dipyridine ligand and filled with rhodamine sulfonate (50 µM) in presence of NiCl2 (0.1 µM) observed by fluorescence microscopy. The time between the first panel (upper left) and the last one (lower right) was 7 s. Scale bar: 10 µm. (Reproduced with permission from Ref. 61.  National Academy of Sciences, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

12

Techniques contact area, just as the adhesion of biological membranes depends of the clustering of cell adhesion molecules. Very recently, a Japanese team reported an elegant example of molecular recognition of vesicles.64 It was shown by agglutination assays that vesicles equipped with boronic acid specifically bind to vesicles containing diol lipid such as phosphatidylinositol. It was demonstrated that the interaction of boronic acid vesicles and inositol vesicles is the result of selective coordination of the boronic acid and the carbohydrate diol. 20 µm

(a)

20 µm (b)

Figure 10 Fluorescence micrographs of vesicles containing 5 mol% of amphiphilic histidine (red) mixed with vesicles containing 5 mol% of Cu(IDA) (blue) in (a) DMPC (dimyristoyl phosphatidylcholine) and (b) DMPC/cholesterol. (Reproduced from Ref. 63.  American Chemical Society, 2006.)

based on a complementary pair of synthetic cell adhesion molecules, a fluorinated amphiphilic Cu(IDA) complex, and an amphiphilic histidine. The fluorinated pyrene anchor attached to the Cu(IDA) complex induces receptor clustering by fluorine–fluorine interactions. The ratio of monomer and excimer emission of the pyrene group reflects the extent of clustering in the vesicle membrane59 and allows visualization the vesicles by fluorescence microscopy (Figure 10). Mixing vesicles containing 5 mol% of Cu(IDA) complex with vesicles containing 5 mol% of histidine ligand led to a strong increase of turbidity caused by the formation of vesicle aggregates. No vesicle fusion occurred. The formation of vesicle aggregates is dependent on the degree of phase separation of the Cu(IDA) complex. Thus, adhesion of the vesicles is based on receptor clustering in the membrane and multivalent metal–ligand coordination in the intervesicular

5.2

Hydrogen bonding of vesicles

Vesicles can also interact via hydrogen bonds. It should be emphasized that bilayer membranes present a hydrophobic interface which strengthens hydrogen bonds that would not be effective in a homogeneous aqueous solution. Nevertheless, hydrogen bonding can give rise to significant binding at membranes only if multiple hydrogen bonds align in a multivalent arrangement. Hydrogen bonding plays a key role in the molecular recognition of carbohydrates by membrane proteins. An early example of a biologically inspired hydrogen-bonding motif to aggregate vesicles was described by Zasadzinski and coworkers.65 Liposomes equipped with biotin units cluster into aggregates upon the addition of streptavidin. Each streptavidin can bind up to four biotins. The liposomes remained intact, even though biotin and streptavidin form a very strong noncovalent complex (Ka ∼ 1015 M−1 ) held together by multiple tight hydrogen bonds. The aggregated liposomes could be redispersed by the addition of excess biotin. More recently, the streptavidin–biotin interaction was used to immobilize liposomes on surfaces in a microarray format.66 A well-known example of supramolecular interaction via three hydrogen bonds is the complementary barbituric acid (BAR)–2,4,6-triaminopyrimidine (TAP) pair. This pair was conjugated to an amphiphilic molecule and inserted into LUVs by Lehn and coworkers.67 It could be shown via a FRET assay using amphiphilic acceptor and donor dyes68 that there is an exchange of lipids between the BAR and the TAP LUVs as a result of vesicle aggregation and fusion. The complementary surface charge of the (negatively charged) TAP and (positively charged) BAR vesicles is critical to the intervesicular interaction. The aggregation of TAP and BAR vesicles leads to fusion into GUVs. More recently, Bong and coworkers investigated the induction of vesicle fusion mediated by a biologically inspired motif based on multiple hydrogen bonding.69 Vancomycin binds to D-Ala-D-Ala peptides via five hydrogen bonds. Their biomimetic system consists of a modified vancomycin glycopeptides and a lipidated D-Ala-D-Ala

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry dipeptide which acts as a membrane-bound fusogen. Mixing LUVs that expose the complementary binding partners leads to a strong increase in size which is monitored by DLS. Fusion of vesicles could be verified by a FRET assay by using acceptor and donor dyes68 embedded in vesicles exposing complementary binding partners. The same team recently described the hydrogen-bonding interaction of vesicles equipped with trimers of cyanuric acid and melamine.70 In water, these interactions alone are negligible, but the interaction is enhanced by multivalent binding of cyanuric acid and melamine trimers at the vesicle surface. Hydrogen bonding leads to aggregation and fusion of the vesicles. It could be argued that the ultimate example of multivalent and selective hydrogen bonding is displayed by nucleic acids. Chemists have equipped peptide nucleic acids (PNAs) and deoxyribonucleic acids (DNAs) with hydrophobic anchors and investigated their hybridization with complementary strands of DNA. Marques and Schneider prepared amphiphiles with PNA “head groups” that can insert into liposomes in aqueous solution.71 Binding of DNA to the PNA liposomes was measured using capillary electrophoresis. A French team recently tagged single-stranded DNA to liposomes and showed with fluorescence spectroscopy that complementary DNA can bind selectively to the tagged liposomes.72 H¨oo¨ k and coworkers reported fusion of liposomes induced by DNA hybridization.73 To this end, DNA strands were conjugated to cholesterol and inserted into liposomes. The length and sequences of the DNA strands were designed such that hybridization occurs in a zipper-like way and brings only vesicles with complementary DNA sequences in close contact. The lipid rearrangements and vesicle fusion that resulted as a consequence of the bilayer contact were investigated by FRET.68

5.3

Host–guest chemistry of vesicles

Amphiphilic macrocyclic host molecules have been investigated for many years. Among others, it is known that amphiphilic crown ethers,74 cryptands,75 calixarenes,76 cyclodextrins,77 and curcubiturils78 can form bilayer vesicles in aqueous solution. However, the host–guest chemistry of such host vesicles remained largely unexplored for many years. Darcy and Ravoo prepared bilayer vesicles composed entirely of amphiphilic cyclodextrin host molecules.77 These vesicles have a membrane that displays a high density of embedded host molecules that bind hydrophobic guest molecules such as t-butylbenzyl and adamantane derivatives. The characteristic size-selective inclusion behavior of the cyclodextrins is maintained, even when the host molecules are embedded in a hydrophobic

13

membrane. For example, adamantane carboxylate binds preferentially to β-CDVs (Ka = 7000 M−1 ), weakly to γ -CDVs (Ka = 3000 M−1 ), and very poorly to α-CDVs (Ka < 100 M−1 ).79 The molecular recognition of CDVs was taken one step further with the investigation of intra- and intervesicular interaction mediated by orthogonal host–guest and metal–ligand complexation.80 The ligand can bind to β-cyclodextrin by hydrophobic inclusion of the adamantyl group in the cavity of the cyclodextrin as well as to metal ions such as Cu2+ and Ni2+ by coordination of the ethylenediamine group. The addition of CuL2 led to selective intravesicular binding with no significant aggregation or size changes of the CDVs, while the addition of even micromolar concentrations of NiL3 led to intervesicular interaction with a rapid aggregation of the vesicles and an increasing turbidity of the solution. These observations were explained on the basis of the complexation constants of Ni2+ and Cu2+ with the ethylenediamine ligand: the complexation constants of Ni2+ are many orders of magnitude lower than those of Cu2+ . The Cu2+ complex exists exclusively as a divalent species that binds with both adamantanes to only one vesicle. Hence, all complexes are saturated at the vesicle surface and no intervesicular interaction occurs. The Ni2+ complex is available as a dynamic mixture of trivalent, divalent, and monovalent species and a small amount of free ligand. Hence, multiple free coordination sites are available on the vesicle surface, which leads to efficient intervesicular complexation in the contact area as soon as two vesicles collide. Furthermore, CDVs were decorated with carbohydrates such as maltose and lactose through host–guest interaction of the cyclodextrins with adamantyl glycosides (Figure 11).81 It was shown in agglutination assays that such carbohydrate-decorated vesicles bind specifically to the lectins concanavalin A and peanut agglutinin, depending on the type and density of carbohydrate on the vesicle surface. In this way, an artificial glycocalix was constructed entirely by self-assembly. In this glycocalix, three carbohydrates (cyclodextrin, maltose, and lactose) operate simultaneously yet independently. Curcubiturils are another class of host molecules that have been assembled into bilayer vesicles in water. Kim and coworkers synthesized an amphiphilic cucurbit[6]uril, which forms vesicles and forms host–guest complexes at the vesicle surface.78 It is possible to decorate the surface of the host vesicles with guest molecules. Exposure of cucurbituril vesicles to a fluorescent spermidine derivative led to fluorescent vesicles. Exposure of the cucurbituril vesicles to α-mannose-substituted spermidine led to vesicles coated with α-mannose, which binds specifically to the lectin concanavalin A. Concanavalin A does not bind when the vesicles are coated with a galactose spermidine

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

14

Techniques

Figure 11 Artificial glycocalix by self-assembly. Agglutination of CDVs can be induced by concanavalin A (green) in the presence of maltose–adamantane conjugate (yellow–red linker). Similarly, agglutination of CDVs can be induced by peanut agglutinin (not shown) in the presence of lactose–adamantane conjugate (blue–red linker).

conjugate. These experiments demonstrate how synthetic host membranes can interact with proteins via multivalent interactions mediated by carbohydrates.

6

CONCLUSION AND OUTLOOK

Vesicles are dynamic supramolecular structures that are investigated by many chemists around the world. Research on synthetic vesicles has been inspired by the remarkable properties of liposomes and biological membranes, and the relevance of vesicles as biomimetic model system for biological membranes and their interactions is evident. Not surprisingly, the large majority of synthetic vesicles are composed of “phospholipid-like” amphiphilic molecules. With the advent of polymersomes, the size of amphiphiles has expanded from small to large to giant. Supramolecular amphiphiles can also be assembled from non-amphiphilic building blocks. In the last decade, many reports have also demonstrated that vesicles can be prepared from a wide range of molecules that are not amphiphilic. In order words, it has become evident that vesicles must not be based on hydrophobic interaction, but instead on a wide range of other noncovalent interactions (π –π interactions, electrostatic interactions, hydrogen bonding, host–guest interaction) that can give rise to the assembly of vesicles. It is expected that this trend will continue in the near future. A key challenge for the years to come will be the development of methods to prepare vesicles of a tailor-made, monodisperse size, irrespective of their composition. On the basis of pioneering reports, it is likely that microfluidic and nanofluidic reactors will resolve this longstanding issue. In addition, it can be foreseen that supramolecular chemistry will play a key role in the development of stimuli-responsive vesicles. Stimuli-responsive vesicles will be important elements of sensors, nanoreactors, and drug

delivery systems. Stimuli-responsive vesicles will also be pivotal to the development of sophisticated membrane mimics, artificial cells, and synthetic tissue. In this respect, stimuli-responsive interaction of vesicles with other vesicles as well as other materials and also the preparation of vesicles inside vesicles will be important challenges. Both areas have recently seen exciting developments.

REFERENCES 1. A. D. Bangham and R. W. Horne, J. Mol. Biol., 1964, 8, 660. 2. T. Kunitake and Y. Okahata, J. Am. Chem. Soc., 1977, 99, 3860. 3. J. N. Israelachvili, D. J. Mitchell, and B. W. Ninham, J. Chem. Soc., Faraday Trans. 2, 1976, 72, 1525. 4. S. H. Tung, H. Y. Lee, and S. R. Raghavan, J. Am. Chem. Soc., 2008, 130, 8813. 5. M. I. Angelova and F. M. Menger, Acc. Chem. Res., 1998, 31, 789. 6. V. Torchilin and V. Weissig, eds., Liposomes: A Practical Approach, Oxford University Press, Oxford, 2003. 7. E. W. Kaler, A. K. Murthy, B. E. Rodriguez, and J. A. N. Zasadzinski, Science, 1989, 245, 1371. 8. M. I. Angelova and D. S. Dimitrov, Faraday Discuss., 1986, 81, 303. 9. S. Ota, S. Yoshizawa, and S. Takeuchi, Angew. Chem., Int. Ed. Engl., 2009, 48, 6533. 10. T. Kunitake, Angew. Chem., Int. Ed. Engl., 1992, 31, 709. 11. H. Ringsdorf, B. Schlarb, and J. Venzmer, Angew. Chem., Int. Ed. Engl., 1988, 27, 113. 12. J. B. F. N. Engberts and D. Hoekstra, Biochim. Biophys. Acta, 1995, 1241, 323. 13. T. Benvegnu, M. Brard, and D. Plusquellec, Curr. Opin. Colloid Interface Sci., 2004, 8, 469. 14. P. L. G. Chong, Chem. Phys. Lipids, 2010, 163, 253.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Vesicles in supramolecular chemistry

15

15. A. M. Cassell, C. L. Asplund, and J. M. Tour, Angew. Chem., Int. Ed. Engl., 1999, 38, 2403.

43. X. N. Xu, L. Wang, and Z. T. Li, Chem. Commun., 2009, 6634.

16. T. Homma, K. Harano, H. Isobe, and E. Nakamura, Angew. Chem., Int. Ed. Engl., 2010, 49, 1665.

44. Y. He, Z. Li, P. Simone, and T. P. Lodge, J. Am. Chem. Soc., 2006, 128, 2745.

17. S. H. Seo, J. Y. Chang, and G. N. Tew, Angew. Chem., Int. Ed. Engl., 2006, 45, 7526.

45. M. H. Li and P. Keller, Soft Matter, 2009, 5, 927.

18. F. Garcia, G. Fernandez, and L. Sanchez, Chem.—Eur. J., 2009, 15, 6740. 19. F. J. M. Hoeben, I. G. Shklyarevskiy, M. J. Pouderoijen, et al. Angew. Chem., Int. Engl., 2006, 45, 1232. 20. X. Zhang, S. Rehm, M. M. Safont-Sempere, and F. W¨urthner, Nat. Chem., 2009, 1, 623.

46. F. Ch´ecot, S. Lecommandoux, Y. Gnanou, and H. A. Klok, Angew. Chem., Int. Ed. Engl., 2002, 43, 1339. 47. S. Y. Yu, T. Azzam, I. Rouiller, and A. Eisenberg, J. Am. Chem. Soc., 2009, 131, 10557. 48. F. Versluis, I. Tomatsu, S. Kehr, et al. J. Am. Chem. Soc. 2009, 131, 13186.

21. L. Zhang and A. Eisenberg, Science, 1995, 268, 1728.

49. J. K. Kim, E. Lee, Y. Lim, and M. Lee, Angew. Chem., Int. Ed. Engl., 2008, 47, 4662.

22. A. J. Dirks, R. J. M. Nolte, and J. J. L. M. Cornelissen, Adv. Mater. 2008, 20, 3953.

50. K. T. Kim, J. J. L. M. Cornelissen, R. J. M. Nolte, and J. C. M. van Hest, Adv. Mater., 2009, 21, 2787.

23. A. Bertin, F. Hermes, and H. Schlaad, Adv. Polym. Sci., 2010, 224, 167.

51. N. S. S. Kumar, S. Varghese, G. Narayan, and S. Das, Angew. Chem., Int. Ed. Engl., 2006, 45, 6317.

24. E. N. Savariar, S. V. Aathimanikandan, and S. Thayumanavan, J. Am. Chem. Soc., 2006, 128, 16224.

52. Y. Wang, N. Ma, Z. Wang, and X. Zhang, Angew. Chem., Int. Ed. Engl., 2007, 46, 2823.

25. V. Percec, D. A. Wilson, P. Leowanawat, et al. Science, 2010, 328, 1009.

53. R. Jelinek and S. Kolusheva, Top. Curr. Chem., 2007, 277, 155.

26. D. E. Discher and A. Eisenberg, Science, 2002, 297, 967.

54. W. H. Binder, V. Barragan, and F. M. Menger, Angew. Chem., Int. Ed. Engl., 2003, 42, 5802.

27. C. LoPresti, H. Lomas, M. Massignani, et al. J. Mater. Chem., 2009, 19, 3576. 28. I. C. Reynhout, J. J. L. M. Cornelissen, and R. J. M. Nolte, Acc. Chem. Res., 2009, 42, 681. 29. M. S. Nikolic, C. Olsson, A. Salcher, et al. Angew. Chem., Int. Ed. Engl., 2009, 48, 2752.

55. J. Voskuhl and B. J. Ravoo, Chem. Soc. Rev., 2009, 38, 495. 56. E. L. Doyle, C. A. Hunter, H. C. Phillips, et al. J. Am. Chem. Soc., 2003, 125, 4593. 57. H. Jiang and B. D. Smith, Chem. Commun., 2006, 1407.

and

58. D. Y. Sasaki, T. A. Waggoner, J. A. Last, and T. M. Alam, Langmuir, 2002, 18, 3714.

31. Y. J. Jeon, P. K. Bharadwaj, S. W. Choi, et al. Angew. Chem., Int. Ed. Engl., 2002, 42, 4474.

59. H. J. Galla and W. Hartmann, Chem. Phys. Lipids, 1980, 27, 199.

32. B. Jing, X. Chen, X. D. Wang, et al. Chem.—Eur. J., 2007, 13, 9137.

60. V. Marchi-Artzner, M. J. Brienne, et al. Chem.-Eur. J., 2004, 10, 2342.

33. J. Qian and F. Wu, Chem. Mater., 2009, 21, 758.

61. A. Richard, V. Marchi-Artzner, M. N. Lalloz, et al. Proc. Natl. Acad. Sci. U.S.A., 2004, 101, 15279.

30. E. W. Kaler, K. L. Herrington, A. K. Murthy, J. A. N. Zasadzinski, J. Chem. Phys., 1992, 96, 6698.

34. X. Zhang, Y. Yang, J. Tian, and H. Zhao, Chem. Commun., 2009, 3807. 35. S. Ghosh, M. Reches, E. Gazit, and S. Verma, Angew. Chem., Int. Ed. Engl., 2007, 46, 2002. 36. Y. Anruka, A. Kishimura, M. Oba, et al. J. Am. Chem. Soc., 2010, 132, 1631. 37. Y. Wang, P. Han, H. Xu, et al. Langmuir, 2010, 26, 709. 38. F. Gr¨ohn, K. Klein, and K. Koynov, Macromol. Rapid Commun., 2010, 31, 75. 39. I. Yoshikawa, J. Sawayama, and K. Araki, Angew. Chem., Int. Ed. Engl., 2008, 47, 1038. 40. C. P. Pradeep, M. F. Misdrahi, F. Y. Li, et al. Angew. Chem., Int. Ed. Engl., 2009, 48, 8309. 41. L. Wang, H. Liu, 2009, 1353.

and

J. Hao,

Chem.

Commun.,

42. C. Schmuck, T. Rehm, K. Klein, and F. Gr¨ohn, Angew. Chem., Int. Ed. Engl., 2007, 46, 1693.

T. Gulik-Krzywicki,

62. S. J. Webb, L. Trembleau, R. J. Mart, and X. Wang, Org. Biomol. Chem., 2005, 3, 3615. 63. R. J. Mart, K. P. Liem, X. Wang, and S. J. Webb, J. Am. Chem. Soc., 2006, 128, 14462. 64. A. Kashiwada, M. Tsuboi, and K. Matsuda, Chem. Commun., 2009, 695. 65. S. Chirovolu, S. Walker, J. Israelachvili, et al. Science, 1994, 264, 1753. 66. D. Stamou, C. Duschl, E. Delamarche, and H. Vogel, Angew. Chem., Int. Ed. Engl., 2003, 42, 5580. 67. V. Marchi-Artzner, T. Gulik-Krzywicki, M. A. GuedeauBoudeville, et al. Chem. Phys. Chem., 2001, 2, 367. 68. D. Hoekstra and N. D¨uzg¨unes, Methods Enzymol., 1993, 220, 15. 69. Y. Gong, M. Ma, Y. Luo, and D. Bong, J. Am. Chem. Soc., 2008, 130, 6196. 70. M. Ma, Y. Gong, and D. Bong, J. Am. Chem. Soc., 2009, 131, 16919.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

16

Techniques

71. B. F. Marques and J. W. Schneider, Langmuir, 2005, 21, 2488.

77. B. J. Ravoo and R. Darcy, Angew. Chem., Int. Ed. Engl., 2000, 39, 4324.

72. A. Gissot, C. di Primo, I. Bestel, et al. Chem. Commun., 2008, 5550.

78. H.-K. Lee, K. M. Park, Y. J. Jeon, et al. J. Am. Chem. Soc., 2005, 127, 5006.

73. G. Stengel, R. Zahn, and F. H¨oo¨ k, J. Am. Chem. Soc., 2007, 129, 9584.

79. P. Falvey, C. W. Lim, R. Darcy, et al. Chem.-Eur. J., 2005, 11, 1171.

74. K. Monserrat, M. Graetzel, and P. Tundo, J. Am. Chem. Soc., 1980, 102, 5527.

80. C. W. Lim, O. Crespo-Biel, M. C. A. Stuart, et al. Proc. Natl. Acad. Sci. U.S.A., 2007, 104, 6986.

75. P. Ghosh, T. Khan, and P. K. Bharadwaj, Chem. Commun., 1996, 189.

81. J. Voskuhl, M. C. A. Stuart, and B. J. Ravoo, Chem.-Eur. J., 2010, 16, 2790.

76. Y. Tanaka, M. Miyachi, and Y. Kobuke, Angew. Chem., Int. Ed. Engl., 1999, 38, 504.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc032

Rheology Christophe Chassenieux1 and Laurent Bouteiller2,3 1

Universit´e du Maine, Le Mans Cedex 09, France UPMC Univ Paris 06, Paris, France 3 CNRS, Paris, France 2

1 Introduction 2 What Information can Rheology Provide to the Supramolecular Chemist? 3 Practical Description of the Techniques 4 Conclusion References

1 1 8 11 11

in a deeper presentation of rheology should consult well-established textbooks.1, 2 The second section of this chapter is based on a wealth of examples taken from the supramolecular chemistry literature and illustrates the wide range of information that can be derived from rheological studies. The third section describes the practical aspects of the main experimental approaches.

2 1

INTRODUCTION

The purpose of rheology is extremely simple: it consists in the study of the deformation or the flow of a sample under a given stress, in order to derive relevant information concerning the inner structure and dynamics of the sample. However, because of the diversity of samples, which range from dilute solutions to gels and soft solids, a single type of rheometer cannot be expected to be adequate for every possible system. Moreover, with a given rheometer, several experimental approaches can be used (small or large deformations, steady shear or transient behavior, etc.) to focus on a particular aspect of the system. The versatility of the rheology toolbox may, therefore, seem to be excessively complex to the supramolecular chemist. In this chapter, it is our aim to convince the reader that minimal knowledge of rheology is sufficient to access a rich source of information. As far as possible, we have limited ourselves to an illustrative and qualitative approach. Readers interested

WHAT INFORMATION CAN RHEOLOGY PROVIDE TO THE SUPRAMOLECULAR CHEMIST?

Rheology or viscosimetry is the obvious choice when quantitative values are sought to describe the resistance to flow (the viscosity) of solutions or melts. However, rheology can also yield useful information about the structure of the assemblies (their size or cross-link density), about their dynamics, and even about their self-assembly mechanism. Of course, a rheological experiment essentially probes a sample at the macroscopic scale: it is, therefore, mostly sensitive to the formation of assemblies of large dimensions.

2.1

Probing dynamics

2.1.1 Fast versus slow dynamics Frequency-dependent experiments can yield valuable information about the dynamics of a self-assembled system. The basics and the practical aspects of the technique are detailed in Section 3.3. Briefly speaking, the experiment consists of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Techniques

O N H

(a)

N H

N H

10

1

0.1

0.01

0.001 0.001

0.01

0.1

1 w (rad s−1)

(b)

10

to the 100-nm-thick crystalline fibers formed by triphenylene derivative 1, the very small diameter of the bis-urea nanotubes allows the rearrangement of the structure within seconds, at room temperature. To study the influence of molecular scale dynamics on macroscopic properties, Craig et al. have designed a particularly interesting organometallic system (Figure 3a)8 in which poly(4-vinylpyridine) (PVP) chains dissolved in dimethyl sulfoxide (DMSO) are reversibly cross-linked through coordination with a bifunctional Pd(II)–pincer complex. The independent control of the dissociation rate

O C6H13 O C11H22OH

+

+

105

+

+

+

+ + −

O

+

104

+ +

+

103

−

+

+

O C6H13 O C6H13 (a)

102

500 nm

(b)

100

Figure 2 (a) Structure of bis-urea 2 and (b) variation of G
and G

with oscillation frequency of a viscoelastic solution in dodecane ([2] = 12 mM). (Reproduced from Ref. 4.  Elsevier, 2007.)

(•) G ′ ( ) G ″ (Pa)

C6H13 O

N H

100

106 C6H13

O

(+) h (Pa s)

sandwiching the sample between two plates (or cylinders), applying a given oscillatory strain (or stress) to one of the plates, and finally measuring the induced movement of the other plate. The amplitude of the oscillatory strain is chosen to be small enough so that the sample is not perturbed by the analysis. The induced movement is decomposed into an in- and an out-phase component. The in-phase component results from the elasticity of the sample, and, therefore, the magnitude of the in-phase component is called the elastic modulus (G
). Conversely, the out-phase component reveals the damping of the oscillations by the viscosity of the sample, and the magnitude is called the loss modulus (G

). Both G
and G

depend on the frequency of the oscillations and represent the mechanical signature of the material. At a first qualitative level, the relative magnitude of G
and G

is a quick indication of the mainly elastic or mainly viscous behavior of the sample. Figure 1(c) shows the typical signature of the elastic gel formed by an organogelator: G
> G

over the whole frequency range of the experiment. This signature means that a network spans the whole sample and is able to restore the mechanical energy of the oscillation. In this particular case, microscopy and diffraction experiments show that the network is formed by selfassembly of the triphenylene derivative 1 into crystalline fibers, which are further entangled and bundled together (Figure 1a and b).3 The fact that G
> G

means that the network does not relax under the applied stress (even at low frequencies), that is, the fibers do not break or disentangle. In contrast, bis-urea 2 (Figure 2) forms viscoelastic solutions: the self-assembled objects present in the sample are robust enough to withstand the mechanical solicitation at high frequencies, but given enough time (at frequencies lower than 0.1 rad s−1 ), these objects rearrange to relax the applied stress.4 Small-angle neutron scattering studies have shown that, in this case, the self-assembled objects are very long and rigid nanotubes of diameter 2.6 nm.5–7 In contrast

G ′ ( ), G ″ ( ) (Pa)

2

0.01 (c)

0.1

1

10

−

+

+

+

+

100

w (rad s−1)

Figure 1 (a) Structure of triphenylene organogelator 1, (b) AFM of a gel of 1 in ethanol/water, and (c) variation of G
and G

with oscillation frequency of the same gel ([1] = 4.3 mM). (Reproduced from Ref. 3.  Royal Society of Chemistry, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Rheology

3

at which the cross-links are dissociated, rather than the fraction of time for which they are dissociated, that is the critical determinant of the mechanical response. Moreover, at low cross-linker content, the relaxation rate determined by rheology (i.e., the frequency at which G
= G

) is in very good agreement with the dissociation rate constant determined by NMR on a model compound. The good agreement between the bulk (13 s−1 ) and the molecular (17 s−1 ) rate constants is consistent with a relaxation involving only the dissociation of cross-links, rather than a combination of dissociation and reptation of the chains.

−

2.1.2 Stress relaxation mechanism OTf

R 2N +

−

H

N M

N

R2 N

O

OTf

NR2 H N

+

M N NR2

O

(a) 1.00E + 05

G ′, G ″ (Pa)

1.00E + 04 1.00E + 03 1.00E + 02 1.00E + 01 1.00E + 00 0.01 (b)

0.1

1

10

100

Frequency (1 s−1)

Figure 3 (a) Schematic representation of reversible cross-links between PVP through coordination with bis(Pd(II)-pincer) complex 3a (R = CH3 ) or 3b (R = Et) and (b) dynamic moduli for networks of 5% (by functional group) 3a (red, G
, G

) or 3b (blue, G
, G

) and PVP in DMSO at 100 mg ml−1 . (Reproduced from Ref. 8.  American Chemical Society, 2005.)

constant (kd ) relative to the equilibrium constant (Keq ) is accomplished through simple steric effects at the square planar Pd(II) complexes: the exchange rate is about 80 times faster with dimethylamino ligands than with diethylamino ligands, although both complexes have very similar thermodynamic stabilities. Figure 3(b) shows that the mechanical spectra of the two systems are markedly different, hinting at the strong influence of the dissociation kinetics compared to thermodynamics. This was confirmed quantitatively by the observation that a master curve is obtained when the frequency of the experiment is normalized by the dissociation rate constant of the complex.8 It is, therefore, the rate

As hinted in the previous example, dynamic experiments can yield information about the relaxation pathway of the system. In particular, the concentration dependence of the relaxation time can be used to discriminate between several potential mechanisms. In the case of supramolecular polymer solutions, three main mechanisms of stress relaxation have been proposed to account for the very diverse experimental results. The model developed by Cates et al.9 assumes that a supramolecular chain can spontaneously break anywhere along its backbone and that it has to diffuse by reptation (i.e., by a snake-like thermal motion) so as to release the stress present at entanglement points with other chains. Because the diffusion of the chain is slowed down by the presence of the neighboring chains, the relaxation time (τ ) is an increasing function of the concentration (τ ∼ c1.5 ).9 In an other approach, Shikata et al. proposed that the entanglement release occurs by a crossing-through mechanism at the entanglement point (Figure 4). This concerted chain interchange reaction, also called phantom crossing, is only a function of the local structure of the system, and, therefore, the relaxation time is not expected to depend on the concentration (τ ∼ c0 ).10 Phantom crossing model

Tentative cross-linking

Entanglement formation

Crossing-through reaction

Figure 4 Schematic representation of the phantom crossing model for entanglement release. (Reproduced from Ref. 10.  American Chemical Society, 2004.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

4

Techniques

Finally, Tanaka recently proposed an extension of the phantom crossing model, whereby the neighboring chains are considered to exert topological forces, so that the chain essentially behaves as if it was a macrocycle. An increase in concentration amplifies the constraints on the chains in contact and therefore accelerates the entanglement release (τ ∼ c−α , with 0 ≤ α ≤ 2).11 Accordingly, the experimentally determined concentration dependence of the relaxation time can be used to pinpoint the most relevant relaxation mechanisms. In fact, all three behaviors for τ (decreasing, constant, or increasing function of the concentration) have been observed experimentally.

a lower viscosity (shear thinning). It is then of interest to continue the experiment, but with a very low stress in order to follow the possible recovery process of the supramolecular assembly. Usually, organogels made of thick crystalline fibers do not recover their initial viscosity after the application of a large stress because the broken fibers cannot heal at a temperature well below their melting point. However, there are interesting examples where the supramolecular interactions are dynamic enough to allow self-healing, which may lead to thixotropic behavior.12, 13 For instance, Figure 5 shows a hydrogen-bonded assembly in decane whose viscosity decreases by three orders of magnitudes under an applied stress of 5 Pa. If the stress is subsequently reduced to 1 Pa, the viscosity returns to its initial value in about 10 min.14 In this case, the system has been shown by small-angle neutron scattering to form long fibrils with a cross section of 10 nm2 , which corresponds to about 10 individual hydrogen-bonded filaments.15 The limited thickness of the fibrils probably allows a sufficient mobility of the monomers which can reestablish broken hydrogen bonds.

2.1.3 Recovery after stress Instead of performing the rheological experiment in the linear regime, it is possible to probe the flow behavior of the sample under large strains. This can conveniently be done, for instance, by applying a continuous rotation. If the applied stress is large enough, the supramolecular sample will necessarily be destructured and will present O

O

N

N

H19C9

C9H19 N

H

O

O

N

N N

H O

H O H

O

H N H N

N

O

N H

O

O

H N

O O

O H

N

O

O

N

O

N

O N

C12H25

O

H

N

H19C9 O

C12H25

N

N O

H O

H

C9H19 O

AA9

AA9

n

BB

BB

n

(a) 1000 000

h (mPa s)

100 000 10 000 5 Pa 1 Pa

1000 100 0 (b)

500

1000

1500

2000

Time (s)

Figure 5 (a) Structure of the hydrogen-bonded filaments and (b) recovery curve of a 5 mM solution in decane at 25 ◦ C. (Reprinted with permission from Ref. 14.  American Physical Society, 2007.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Rheology

2.2

Probing the size and structure

2.2.1 Semiquantitative molar mass measurement The viscosity of dilute solutions is often used to qualitatively characterize the size of the self-assembled aggregates. This is possible first because at low concentrations, the interactions between the aggregates can often be neglected, so that the viscosity can be considered to directly report on the hydrodynamic volume of the aggregates. Furthermore, it has been clearly demonstrated that the dilute solution viscosity does not depend on the dynamics of formation of the aggregates.16 For example, simple relative viscosity measurements (ηrel = η/η0 , where η is the viscosity of the solution and η0 is the viscosity of the solvent) have been used to prove that bis-urea 4a, bearing both hydrogen-bonding units and hydrophobic parts, self-assembles into high-molar-mass structures in water (Figure 6).17 The fact that reference compounds 4b and c, which lack the hydrogen-bonding units or the hydrophobic part, yield a much lower viscosity implies that both interactions are necessary for the stability of this assembly in water or acetonitrile.17 However, to probe more subtle effects, it is necessary to eliminate the concentration dependence of the viscosity. This is usually done by considering the reduced specific viscosity (ηred = ηspe /C = (η − η0 )/η0 C, where C is the concentration of the solution), which is then extrapolated to zero concentration to yield the intrinsic viscosity ([η]). The intrinsic viscosity is more informative than the relative viscosity because it is quantitatively related to the molar mass (M) of macromolecules through the Mark–Houwink equation [η] = KM a , where K and a are empirical constants that depend on the nature of the

polymer, its architecture, the solvent, and the temperature, but not on the molar mass. The importance of removing concentration effects before interpreting viscosity data is particularly obvious in the example shown in Figure 7. Here, macromolecular guests were complexed with mono-, di-, or tritopic hosts.18 The reduced specific viscosity data yield linear plots, which can be extrapolated to zero concentration. As expected, the obtained intrinsic viscosities for the stoichiometric complexes increase with the number of arms, and thus confirm the increasing hydrodynamic volume. However, a simple comparison of specific (or relative) viscosities at concentrations higher than 2.5 g dl−1 would yield erroneous conclusions because of the different concentration dependencies of the systems considered. Of course, a practical difficulty with this approach is that the extrapolation to zero concentration must be performed in a concentration range where the supramolecular system has a fixed structure and molar mass, that is, the supramolecular assembly needs to be sufficiently stable. In general, if the concentration of a solution is increased, there is a point at which the aggregates will start to overlap and get entangled. At this overlap concentration (C ∗ ), the viscosity of the solution increases significantly, which makes C ∗ easily detectable (Figure 8).4, 19 This overlap concentration is reached when the local concentration inside a sphere containing a single aggregate is equal to the macroscopic concentration. Therefore, C ∗ is directly linked to the size of the aggregate C∗ =

O O

(CH2)m

O

N R

n

(a)

Hydrophobic

M 4 3 3 πRg NA

O N R

N R

Hydrogen bonding

N R

(CH2)m

O

4a : n = 7.2 4b : n = 7.2 4c : n = 6

m=9 m=9 m=0

h/h0

2

Water

Acetonitrile

Hydrophobic interactions

50 nm

(c)

R=H R = CH3 R=H

Hydrogen-bonding and hydrophobic interactions

4a 4b 4c

(b)

O

n

3

1

5

(d)

Figure 6 (a) Amphiphilic bis-urea 4a and reference compounds 4b and c; (b) relative viscosity of 16 mM solutions in water or acetonitrile at 25 ◦ C; (c) cryo-TEM; and (d) schematic structure of bis-urea 4a filaments in water. (Reproduced from Ref. 17.  American Chemical Society, 2007.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

6

Techniques

O

O

O

O

O

O

O

O 2 PF6−

CH2 O

O

O

O

O

O

O

O

+

CH2 O

O

O

O

O O 3

CH2

2

O

O O

+

N

N

CH3

O

O

O

O

O

O CH2

n

O

O

O

O

O

N

O

(a)

Reduced viscosity (dl g−1)

0.9

a b d

0.6

g

0.3

...

0 0

1

2

Concentration (g

(b)

3 dl−1

4

)

Figure 7 (a) Structure of tritopic host and polymeric guest and (b) reduced viscosity plot showing the extrapolated intrinsic viscosities for solutions of stoichiometric mixtures of polymeric guest with tritopic (δ), ditopic (γ ), monotopic (β), or no (α) host (chloroform, room temperature). (Reproduced from Ref. 18.  American Chemical Society, 2005.)

1.E + 03

where M and Rg are the molecular weight and the radius of gyration of the aggregates, respectively, and NA is Avogadro’s number.19, 20 If some hypothesis can be made about the shape of the aggregates, then the molecular weight dependence of the radius of gyration can be accounted for, and an estimation of the molecular weight can be derived from the value of C ∗ .20

1.E + 02

2.2.2 Cross-linking density

1.E + 05

h /h0

1.E + 04

1.E + 01

1.E + 00

1.E − 01 0.001

C* 0.01

0.1 Concentration (g l−1)

1

10

Figure 8 Concentration dependence of the relative zero-shear viscosity for solutions of bis-urea 2 in dodecane (25 ◦ C). (Reproduced from Ref. 4.  Elsevier, 2007.)

Another quantitative information that can be derived from rheological measurement is the cross-linking density of gels. Indeed, according to the classical theory of rubber elasticity,1 the plateau modulus (derived from frequencydependent experiments) is proportional to the density of elastically active chains. These chains can result from chemical cross-linking, supramolecular interactions, or simple entanglements. Therefore, one usually has to make some assumptions concerning the proportion of entangled chains to deduce an estimate of the cross-linking density due to supramolecular interactions. For instance, the elastic modulus of hydrogels based on hydrophobically modified polyethylene glycol (PEG) and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Rheology β-cyclodextrins has been used to monitor the effect of the PEG functionality on the cross-linking density.21

2.3

Probing the self-assembly mechanism

2.3.1 Competition between chains and cycles The formation of linear supramolecular chains from ditopic monomers is usually in competition with the formation of oligomeric cyclics. Because of the large size difference between these species, viscosimetry is a sensitive tool for determining the influence of concentration22 or temperature23 on the ring-chain equilibrium. For example, Figure 9 shows the evolution of the specific viscosity of several similar supramolecular polymers.22 The low-concentration part of the plot displays a linear relationship between specific viscosity and concentration, which is characteristic for noninteracting assemblies of constant size (ηspe ≈ [η]C, see Section 2.2). These results indicate the presence of cyclic dimers at low concentrations, as confirmed by NMR. At higher concentration, a sharp rise in viscosity is due to the formation of chains of increasing size. Therefore, the plot shows that the methyl substituents on monomer UPy4 introduce a conformational bias, which promotes cyclic dimer formation, as compared to nonsubstituted monomer UPy1.22

polymer, over a useful concentration range (Figure 10).26, 27 Indeed, if the monomer concentration and the equilibrium constant of the chain stopper are large enough, nearly all chain ends are occupied by a chain stopper, which means that the length of the filament is inversely proportional to the chain stopper fraction and independent of the monomer concentration. As a consequence, it becomes possible to independently vary the length and the concentration of the chains, so that scaling exponents for the chain length and concentration dependence of rheological properties can be obtained and compared with theoretical values to derive some information on chain flexibility and dynamics.27

(a)

2.3.2 Using chain stoppers to probe the assembly

C13H27 H O N O

N

N H

Increasing concentration

Ct

(b)

Chain stoppers, that is, monotopic monomers, have been widely used to reduce the chain length of supramolecular polymers and thus the viscosity of their solutions.24, 25 More interestingly, chain stoppers can be used to block the concentration dependence of the length of the supramolecular

7

Chain stopper

Monomer

Figure 10 Schematic representation of the influence of concentration on the chain length of a supramolecular polymer without (a) or with (b) a chain stopper at a fixed stopper-to-monomer ratio. (Reproduced from Ref. 26.  American Chemical Society, 2005.)

C13H27 O N H

R

N H

H N H

N N

O

10

1 2 4 5 6

UPy2 : R =

(±)

hsp

UPy1 : R =

UPy4 : R =

1

0.1

UPy5 : R = 10 (a) UPy6 : R =

(b)

100 Concentration (mM)

Figure 9 (a) Structure of ditopic monomers UPy1–UPy6 and (b) specific viscosity versus concentration of their solutions in chloroform at 20 ◦ C. (Reproduced from Ref. 22.  American Chemical Society, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

8

2.4

Techniques

Sensitivity of a system to shear

Shear is an interesting stimulus, which can be used to trigger a specific response from a supramolecular material. This section is devoted to the description of the main responses of supramolecular assemblies to shear forces.

2.4.1 Shear thinning Shear-thinning behavior (the decrease of viscosity with shear rate) is a characteristic feature of supramolecular assemblies of large size, like supramolecular polymers or self-assembled fibrillar networks. Indeed, intermolecular interactions can usually be disrupted if the shear rate is high enough.

reduction in turbulent flow as a result of the interactions of the long structures with the small vortices developed during the flow. For this reason, the effect is very attractive in pumping processes because a significant amount of energy can be saved. Supramolecular polymer based on self-assembled bis-urea 2 (Figure 2) is also an efficient hydrodynamic drag reducer for hydrocarbons, which shows that the transient nature of the chain is not an obstacle.31 In contrast, the reversibility of the association can be expected to avoid mechanical degradation, which typically occurs with classical polymers, due to irreversible scission of the backbone.

3

PRACTICAL DESCRIPTION OF THE TECHNIQUES

2.4.2 Shear thickening In contrast, shear thickening is less frequently encountered. An increase in the viscosity with the shear rate is usually explained by one of the two following mechanisms. The first mechanism ascribes thickening to the increased tension along stretched chains, whereas the second attributes the thickening to a reorganization of the structure, which leads to an increase in the number of elastically active chains. For example, in the case of the reversibly cross-linked supramolecular polymer networks shown in Figure 3, shear thickening has been attributed to a shear-induced transformation of elastically inactive intrachain cross-linkers to elastically active interchain cross-linkers.28

2.4.3 Shear banding At high shear rates, an initially homogeneous sample may separate into regions with different shear rates, this is called shear banding. For example, bis-urea 2 supramolecular solutions display complex shear banding effects because of a strong coupling between the shear flow and the alignment of the supramolecular chains owing to their large persistence length.29, 30 Alignment of chains in the flow direction may facilitate their association into longer chains. The elongated chains would have a stronger tendency to align, and this would provide a positive feedback mechanism, which could explain the extreme shear thinning that can eventually lead to shear banding.

2.4.4 Behavior in turbulent flow Hydrodynamic drag reduction is an unusual effect involving macromolecules in a turbulent flow: for very dilute solutions, the resistance to flow can be reduced compared to that of the pure solvent. Indeed, the presence of high-molecularweight polymer chains can produce large levels of drag

3.1

Viscosimetry

By definition, viscosity of a material (η) represents its resistance to flow, though other kinds of viscosities have been defined for solutions to describe in various ways how the solubilization of a solute into a solvent at a given concentration (C) impacts the viscosity of the latter (η0 ). The relative viscosity (ηrel ) is the ratio of the viscosity of the solution (η) with respect to the viscosity of the solvent; its value is useful to estimate the viscosity enhancement of a solvent on solute addition. The specific viscosity measures the sole contribution of the solute to the viscosity enhancement with respect to the solvent and is given by ηspe = (η − η0 )/η0 . However, to compare different solutes, it is more convenient to consider the reduced viscosity which takes into account the solute concentration: ηred = (η − η0 )/(Cη0 ). Finally, the inherent viscosity is given by ηinh = 1/C ln(η/η0 ). The extrapolation to zero concentration of both ηred and ηinh allows one to compute the intrinsic viscosity of the solute (Figure 7): [η] = lim (ηred ) = lim (ηinh ) C→0

C→0

The concentration dependence of the reduced viscosity is also informative. According to the Huggins development ηred = [η] + kH [η]2 C, and the value of the Huggins coefficient (kH ) is an indication of the ability for a given gelator to self-assemble (the higher the kH , the more strongly the gelator will self-assemble). From a practical point of view, a capillary viscometer is usually used to perform viscosity measurements as long as diluted solutions are considered. The experiment consists in measuring the time (t) needed for a given volume (V ) of solution to flow down a straight circular capillary of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Rheology

3.2

radius r and length h at a volume flow rate Q and to compare it with the flow time of the solvent (t0 ) measured in the same conditions. According to the Poiseuille law, Q = (πr 4 P )/(8ηh), where P is the pressure applied
V to the liquid and since the flow time is given by t = 0 dV /Q it follows that t/t0 = η/η0 . Experimentally, t needs to be measured at several concentrations to derive ηred , ηinh , and finally [η]. Viscosimetry is a cheap technique and is very convenient for comparing gelators in terms of their ability to selfassemble in solution. Nevertheless, it should be restricted to Newtonian liquids (that is, liquids for which the viscosity does not depend on the shear rate (Section 2.4)), because from the capillary wall to its center the shear rate is not constant. For non-Newtonian fluids, proper viscosity calculations require corrections of the raw data once the dependency of their viscosity with the shear rate is known (Section 3.2). Unfortunately, for a given viscosimeter, the range of accessible shear rates is rather limited and it is often difficult to run measurements at very low shear rates where diluted solutions are usually expected to present a Newtonian plateau. An interesting alternative is to use a rolling ball viscometer filled with a solution, where the time needed for a calibrated ball to roll over a fixed capillary distance is measured and related to the viscosity of the solution through the Stokes law. Since the inclination angle of the capillary is variable, various shear rates can be achieved.25 In conclusion, capillary viscosimeters are tools which give an easy access to the viscosity of dilute solutions but which ideally require knowledge of the rheological profile of the solutions (that is, the relation between viscosity and shear rate). To determine such a profile, measurements are needed over a wider range of shear rates, which can be achieved with a rheometer equipped with the proper tools as detailed in the next section.

9

Flow measurements

Rheometers allow flow rate measurements over a much extended range of shear rates and stresses when compared to viscometers. Two kinds of experimental configuration can be discriminated: (i) stress-imposed rheometers, which apply a given torque to the tools and which measure the shear rate, and (ii) strain-imposed rheometers, which apply a given rotation rate to the tools and which measure the torque. Whatever is the nature of the rheometer, depending on the magnitude of the viscosity, different tools may be used (Figure 11): Couette Geometry The sample is loaded between two coaxial and concentric cylinders. The rotation of the inner cylinder at an angular rate ω induces the application of a torque (M) to keep the outer cup fixed. For a very small gap between the cylinders, the velocity profile within the gap is linear. It follows that viscosity is given by η = KM/ω where K is an instrumental constant depending on the lengths and the radii of the cylinders. The so-called Couette geometry is suitable for liquids displaying low viscosities and allows measurements at high shear rates. In contrast, it should be prohibited for elastic liquids, which tend to climb along the rotating cylinder (Weissenberg effect). Cone and Plate Geometry The solution is sandwiched between a rotating plate and a fixed cone of radius R and with a very small angle β. The plate can rotate at a given rate and accordingly the torque M is measured. Since the angle is small, the flow between the cone and the plate almost occurs between two parallel plates, the shear rate is then given by γ˙ ≈ r/β, where r is the distance to the center axis. The shear rate and stress are homogeneous within the gap and proportional to the

w w

w

b

R

h R

Figure 11 Schemes of the different geometries that can be used for flow or viscoelastic rheological measurements. From left to right: double cylinders or Couette geometry, cone and plate geometry, and parallel plates geometry. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

10

Techniques

torque in the latter case. It follows that the viscosity is given by η = KM/ω, where K is an instrumental constant depending on R and β. This geometry is suitable for liquids displaying a wide range of viscosities, but the cone radius must be adapted to the magnitude of the viscosity. R must increase when the viscosity decreases and one should pay attention to the inertia of the tools which precludes measurements at high shear rates. Parallel Plates Geometry The geometry is the same as the previous one, but the cone has been replaced by a plate. The main difference, in the latter case, arises from the nonhomogeneous strain field which can be corrected. The main advantage of this geometry is the possibility of easily tuning the gap between the plates, which allows the modification of the shear rate range to be investigated without changing the rotation rate of the plate. In all three cases, a flow measurement is achieved by varying the shear rate over a wide range of values and by measuring the corresponding viscosity. Of course, for each applied shear rate, one should wait in order to reach a constant value of the viscosity with time before increasing further the shear rate. The rheogram, which is then obtained, reflects the steady state of the sample. Most of the time, the samples are not Newtonian over the whole shear rate range investigated and this is why the viscosity is often extrapolated to zero shear (that is, on the Newtonian plateau). An often critical question is “can we see when we shear?” To address such a question, rheology is coupled with another technique which allows an observation of the sample in direct or reciprocal space. It should be said that most of the time this coupling is restricted to flow measurements and is useful to “visualize” the nonuniformity of flow within the gap of the geometry and/or the alignment of the samples under flow, both of which can occur in the nonlinear regime of flow measurements. In that sense, rheology can be coupled with scattering experiments (light, X-ray, or neutron32 ) allowing the measurements of birefringence,7 dichroism, or the local structure of the sample.12 At higher length scales, magnetic resonance imaging (MRI)33 or ultrasound34, 35 can also be used to determine the local velocity profile within the gap.

3.3

Viscoelastic measurements

The tools are exactly the same as the ones described in the previous section, the main difference with flow measurements lies in the fact that we now apply a strain or stress that oscillates sinusoidally with time instead of being

a linear or a logarithmic function of it. The response of the sample is then measured in terms of stress or strain over a given frequency range (in the following, we apply a strain). A sinusoidally oscillating shear strain γ = γ 0 sin(ωt) induces an oscillating shear stress shifted by δ : σ = σ 0 sin(ωt + δ). For pure elastic solids, δ = 0◦ , whereas for liquids, δ = 90◦ . For viscoelastic fluids, δ stands between these two limits; the shear stress is given by σ = σ 0 [cos δ sin(ωt) + sin δ cos(ωt)] and two moduli are defined according to G
(ω) = σγ 0 cos δ and G

(ω) = σγ 0 sin δ. G
is the elastic 0 0 modulus and represents the energy that is stored within the sample during its deformation and that can be recovered. G

is the loss modulus that corresponds to the viscous response of the sample and represents a quantity proportional to the dissipated energy during the deformation of the sample. For elastic solids, G
 G

and its magnitude can be used for the characterization of the elastic networks in terms of correlation length and topology. For viscous liquids, G

> G
and G
∝ ω2 while G

∝ ω (Figure 3b) and their magnitude can be used to compute the viscosity of the solution. For viscoelastic fluids, if the measurements are run over the proper frequency range, one can measure a transition from an elastic-like behavior at high frequency to a liquid-like behavior at low frequency (Figure 3b). The crossing point between G
and G

corresponds to the relaxation time of a sample; that is, the time needed for the latter to relax a macroscopic strain. As mentioned previously, the concentration dependence of this relaxation time allows conclusions to be made about the mechanism of self-assembly of the gelators. Furthermore, one can clearly see that the response of a sample depends not only on the frequency (ω) but also on the amplitude of the strain (γ 0 ). To keep things as simple as possible, most of the times, one must establish the linear viscoelastic response regime of the sample by measuring both G
and G

at a given frequency over a wide strain range. Below a critical value of the latter (γ c ), G
and G

remain almost constant and depend only on the frequency, whereas above γ c one leaves the linear viscoelastic response regime and experimental results are far more difficult to interpret. This is why viscoelastic measurements reported in the literature are usually restricted to the linear viscoelastic response regime. We note that the large-amplitude oscillatory strain (LAOS) measurements have also been used to understand the rheological properties of self-assemblies in the nonlinear viscoelastic regime of associating polymers,36 but to the best of our knowledge not yet for gelators. A strong analogy may be drawn between flow and viscoelastic measurements. The Newtonian behavior at low shear rates obtained in the former case is the equivalent of the linear viscoelastic response regime for the latter, whereas shear-thinning or

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Rheology shear-thickening properties may be viewed as nonlinear properties of the sample, which means that its structure is affected by shear whatever is the kind (flow or viscoelastic) of measurements performed.

3.4

Choice of the tools—sample preparation

As already mentioned, the geometry should fit with the magnitude of the viscosities and of the moduli. Sometimes, at high shear rates, some slippage of the sample can occur. In this case, it is possible to adapt the surface state of the geometries (anodized, sanded, or scarred) or their critical surface energy through their chemical nature (polymer, metal, ceramic, etc.) to promote adhesion of the samples. The slippage results in a variation of the viscosity with the shear rate according to η ∝ 1/γ˙ , which is also the signature of shear thinning. One way to discriminate between both scenarios consists in obtaining rheograms with various gaps to see if they superimpose (shear thinning) or not (wall slippage), and in that sense parallel plate geometries are perfect tools. One should remember that the geometries should be perfectly thermostated (most of the time with Peltier setups) since the temperature has a huge influence on the viscosities and the self-assembling properties of the gelators. Furthermore, solvent evaporation during the measurements must be prevented (especially with parallel plates and cone and plate geometries) with appropriate solvent traps. Regarding the small values of the gaps, the presence of large-scale inhomogeneities and of air bubbles may also affect the measurements. The samples must then be homogeneous down to the micrometer scale, and may be degassed prior to measurement.

3.5

The main advantage of this technique is that the “rheograms” are obtained faster and on a wider frequency range than with a rheometer, but measurements are restricted to the linear viscoelastic response regime. The major difficulty arises from the finding of proper tracers which should not interact with the gelators, which should also be dispersed easily within the solvent, and finally which should have a size bigger than the correlation length of the samples.

4

CONCLUSION

Rheology, through the use of various instruments and various experimental procedures, can be used to characterize a wide range of samples going from dilute solutions to gels and to soft solids. This technique is often used in the field of supramolecular chemistry in a rather qualitative manner, in order to probe, for example, if a particular assembly is larger than a reference system. However, rheology can be a much richer source of information. We have shown through selected examples from the supramolecular chemistry literature that rheology can yield useful information not only about the size of the assemblies but also about their structure, their dynamics, and their self-assembly mechanism.

REFERENCES 1. C. W. Macosko, Rheology: Principles, Measurements and Applications, Wiley-VCH, New York, 1994. 2. H. A. Barnes, J. F. Hutton, and K. Walters, An Introduction to Rheology, Elsevier Science Publishers, Amsterdam, 1989. 3. A. Kotlewski, B. Norder, W. F. Jager, et al., Soft Matter, 2009, 5, 4905.

Microrheology

This technique is pretty new37 and has already been successfully applied to measure the viscoelastic properties of gels based on the gelators.38 The principle of this technique is based on the tracking of submicrometer tracers embedded within a solution. The mean-squared displacement of the tracers is given by

r 2 (τ ) ∝ τ α , where α = 1 for liquids and α < 1 for viscoelastic fluids. It appears that

r 2 (τ ) is also related to the viscoelastic modules through the generalized Stokes–Einstein relation: G
(ω) + iG

(ω) =

11

kT π aiω

˜r 2 (ω)

where k is the Boltzmann constant, T the temperature, a the radius of the tracer, and

˜r 2 (ω) the Laplace transform of

r 2 (τ ).

4. G. Ducouret, C. Chassenieux, S. Martins, et al., J. Colloid Interface Sci., 2007, 310, 624. 5. L. Bouteiller, O. Colombani, F. Lortie, and P. Terech, J. Am. Chem. Soc., 2005, 127, 8893. 6. T. Pinault, B. Isare, and L. Bouteiller, Chem. Phys. Chem., 2006, 7, 816. 7. T. Shikata, T. Nishida, B. Isare, et al., J. Phys. Chem. B, 2008, 112, 8459. 8. W. C. Yount, D. M. Loveless, and S. L. Craig, J. Am. Chem. Soc., 2005, 127, 14488. 9. M. E. Cates and S. J. Candau, J. Phys.: Condens. Matter, 1990, 2, 6869. 10. T. Shikata, D. Ogata, and K. Hanabusa, J. Phys. Chem. B, 2004, 108, 508. 11. F. Tanaka, Langmuir, 2010, 26, 5374. 12. M. Lescanne, P. Grondin, A. d’Al´eo, et al., Langmuir, 2004, 20, 3032.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

12

Techniques

13. X. Huang, S. R. Raghavan, P. Terech, and R. G. Weiss, J. Am. Chem. Soc., 2006, 128, 15341.

26. F. Lortie, S. Boileau, L. Bouteiller, et al., Macromolecules, 2005, 38, 5283.

14. E. Buhler, S. J. Candau, E. Kolomiets, and J.-M. Lehn, Phys. Rev. E, 2007, 76, 061804.

27. W. Knoben, N. A. M. Besseling, L. Bouteiller, and M. A. Cohen-Stuart, Phys. Chem. Chem. Phys., 2005, 7, 2390.

15. E. Kolomiets, E. Buhler, S. J. Candau, and J.-M. Lehn, Macromolecules, 2006, 39, 1173. 16. W. C. Yount, H. Juwarker, and S. L. Craig, J. Am. Chem. Soc., 2003, 125, 15302. 17. E. Obert, M. Bellot, L. Bouteiller, et al., J. Am. Chem. Soc., 2007, 129, 15601. 18. F. Huang, D. S. Nagvekar, C. Slebodnick, Gibson, J. Am. Chem. Soc., 2005, 127, 484.

and

H. W.

19. T. Park and S. C. Zimmerman, J. Am. Chem. Soc., 2006, 128, 11582. 20. M. Bellot and L. Bouteiller, Langmuir, 2008, 24, 14176. 21. F. van de Manakker, L. M. J. Kroon-Batenburg, T. Vermonden, et al., Soft Matter, 2010, 6, 187. 22. A. T. ten Cate, H. Kooijman, A. L. Spek, et al., J. Am. Chem. Soc., 2004, 126, 3801. 23. B. J. B. Folmer, R. P. Sijbesma, and E. W. Meijer, J. Am. Chem. Soc., 2001, 123, 2093.

28. D. Xu, J. L. Hawk, D. M. Loveless, et al., Macromolecules, 2010, 43, 3556. 29. J. van der Gucht, M. Lemmers, W. Knoben, et al., Phys. Rev. Lett., 2006, 97, 108301. 30. W. Knoben, N. A. M. Besseling, and M. A. Cohen-Stuart, J. Chem. Phys., 2007, 126, 024907. 31. E. Sabadini, K. R. Francisco, and L. Bouteiller, Langmuir, 2010, 26, 1482. 32. M. W. Liberatore, F. Nettesheim, P. A. Vasquez, et al., J. Rheol., 2009, 53, 441. 33. C. J. Elkins and M. T. Alley, Exp. Fluids, 2007, 43, 823. 34. L. B´ecu, S. Manneville, and A. Colin, Phys. Rev. Lett., 2004, 93, 018301. 35. S. Manneville, L. B´ecu, P. Grondin, and A. Colin, Colloids. Surf. A: Physicochem. Eng. Aspects, 2005, 270–271, 195. 36. J. Wang, L. Benyahia, C. Chassenieux, et al., Polymer, 2010, 51, 1964.

24. T. Pinault, C. Cannizzo, B. Andrioletti, et al., Langmuir, 2009, 25, 8404.

37. T. A. Waigh, Rep. Prog. Phys., 2005, 68, 685.

25. T. Pinault, B. Andrioletti, and L. Bouteiller, Beilstein J. Org. Chem., 2010, 6, 869.

38. J. Van der Gucht, N. A. M. Besseling, W. Knoben, et al., Phys. Rev. E, 2003, 67, 051106.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc033

Langmuir–Blodgett Films Jeremy J. Ramsden Cranfield University, Bedfordshire, UK

1 Introduction 2 Langmuir Films 3 Film Transfer 4 Transferred Film Metrology 5 Posttransfer Processing 6 Relationships to Other Technologies 7 Applications 8 Conclusions Acknowledgments References

1

529 529 534 537 538 538 539 540 540 540

INTRODUCTION

Langmuir–Blodgett (LB) films represent an early example of nanotechnology. Sometimes, their origin is traced back to the Japanese art of suminagashi , used to produce decorated paper from the twelfth century onward (but originating much earlier in China) and, in Europe, to Benjamin Franklin, who made some observations on the spreading of oil on Clapham Common in the late eighteenth century, but a more definite start was made by Agnes Pockels, who essentially created the first Langmuir film in her domestic kitchen. She wrote up a description and sent it to Lord Rayleigh, who in turn sent to Nature, where it duly appeared.1 This turned out to be the first of a long series of articles (the initial Nature paper was followed by three others in the same journal2–4 ). The theme was later taken up by Irving Langmuir, working in the research

laboratories of the General Electric Company during and after the First World War,5 who had the resources to develop a more sophisticated apparatus; the floating monolayers that are the essential precursors to LB films are usually called Langmuir films nowadays. The next important step was the development of the method of sequential transfer of floating monolayers onto a solid substrate, pioneered by Katharine Blodgett (working with Langmuir at General Electric); these assemblies are nowadays called LB films. Unsurprisingly, given its development over more than a century, there is a very considerable literature on the subject. It is not the purpose of this chapter to recapitulate the contents of the many existing excellent reviews (see Roberts6 and references therein for excellent coverage up to 1985, focusing especially on applications; Binks7 goes into more molecular details; somewhat different, and slightly later, viewpoints are given by Kuhn8 and Peterson9 ), but to present a brief summary of salient points that have perhaps not been brought together before, as well as presenting some of the more recent developments. Floating monolayers (sometimes called “insoluble monolayers” in the literature even though they are not completely insoluble) are covered briefly because they are of course essential precursors to LB films, but the monolayers have their own extensive literature, which is not appraised here.

2

LANGMUIR FILMS

Langmuir films are floating monolayers of amphiphilic molecules. In their compressed state, they are nanoplates according to standard terminology.10 If the molecules are floating on water, they should have a hydrophilic moiety (often called the head) and a hydrophobic moiety (often called the tail). The free energy of interactions between

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

2

Techniques

entities (denoted by 1) immersed in a fluid (denoted by 2) is related to the surface tension of the interface γ 12 by11 G121 = −2γ 12

(1)

By definition, taking the fluid to be water, if G121 < 0, then the entity is hydrophobic and if G121 > 0, then the entity is hydrophilic. The interfacial surface tension depends on the single-substance surface tensions of the interacting entities, γ 1 and γ 2 . These single-substance tensions have an apolar (Lifshitz–van der Waals) and a polar (Lewis acid/base) component: + γ ab γ i = γ LW i i ,

i = 1 or 2

The former are combined as
 2  LW LW = γ − γ γ LW 12 1 2

(2)

(3)

and similarly for the latter, but the expression is more complicated because γ ab has an electron-donor subcomponent γ
(also called the dativit`a12 ) and an electron-acceptor subcomponent γ ⊕ (also called the recettivit`a); hence, there are cross terms:
    
⊕
⊕ ⊕

⊕ ab γ 12 = 2 γ1 γ1 + γ2 γ2 − γ1 γ2 − γ1 γ2 (4) corresponding, respectively (from left to right on the righthand side of the equation), to the polar cohesive interaction energy between the electron donors and acceptors of the entity, the polar cohesive interaction energy between the electron donors and acceptors of the water, the polar adhesive interaction energy between the electron acceptors of the entity and the electron donors of the water, and the polar adhesive interaction energy between the electron donors of the entity and the electron acceptors of the water. The term G121 is the interfacial free energy per unit area for an infinite planar interface between the entities (subscript 1) and water (subscript 2) at the “distance of closest approach” 0 , equal to approximately 0.16 nm.13 Many single-substance surface tensions, from which G121 may be calculated from equations (1–4), have been tabulated (for a useful compilation of many values, see Ref. 14), otherwise, if tabulated values are not available for the substance of interest, they can be measured (for a discussion of methods, see Ref. 14) or, provided that at least the chemical structure of the substance is known, they can be estimated from the structure according to an approach developed for proteins (equations (44) and (45) in Ref. 15). There are, in fact, no absolute criteria for determining whether a molecule is a good Langmuir film former. It

probably suffices that there is merely a difference in the hydrophobicity or hydrophilicity between the two moieties. There must also be attractive lateral interaction between the molecules. Since the hydrophilic (or polar) moiety is de facto hydrated, it will repel its congeners; therefore, one usually relies on LW interactions between the hydrophobic (apolar) moieties, which increase with increasing size of the moiety. These considerations make it clear why the classic film formers are the long-chain fatty acids and their salts, such as stearic acid, CH3 (CH2 )n=16 COOH. Empirically, it has been found that n must be equal to at least 12 in order for the molecule to be a good film former. The geometry of the molecule is also important. An ideal film former is a cylinder (the central image in Figure 1). That is not to say that other shapes cannot form monolayers, but there may be difficulties in transferring them to solid substrates. A completely apolar molecule has also been found to form stable Langmuir monolayers.16 Many substances, especially metal oxides, show great hysteresis between their advancing and receding contact angles with water. Hence, initially perfectly symmetrical spherical nanoparticles of, for example, silica can be carefully spread on the surface of a Langmuir trough (Section 2.1), provided they are sufficiently hydrophobic not to immediately go into solution; the part that is immersed will become hydrated and hydrophilic, providing the necessary polar/apolar asymmetry (for a demonstration of the spreading of nanoparticles, see Ref. 17). The above considerations apply equally well, mutatis mutandis, to molecules floating on a nonaqueous liquid, although such systems have been of negligible importance in the development of the field. It was earlier recognized18 that if the –COOH moiety is complexed with a multivalent cation such as Hg2+ or Cd2+ (incorporated into the system by dissolving a salt such as cadmium perchlorate into the subphase at a low concentration) much more rigid films can be created; a divalent cation cross-links two adjacent amphiphiles, a trivalent one, three, and so forth.

(a)

(b)

(c)

Figure 1 Idealized amphiphilic molecules. The polar moiety is shown as a circle. The ideal film former is cylindrical (b). Molecules that are conical (a) or obconical (c) in shape tend to form nonplanar structures such as cubic phases or micelles.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

Langmuir–Blodgett films

2.1

The Langmuir trough

The trough is shown in cross-section in Figure 2. Langmuir himself had great difficulties in finding suitable materials from which to make the container. The main difficulty is to ensure that it can be kept scrupulously clean. Modern materials such as polytetrafluoroethylene (PTFE) are good in this respect, although not always as dimensionally stable as would be desired. However, troughs carefully machined from solid blocks of PTFE are usually satisfactory. The trough is filled with the liquid, most commonly ultrapure water, to be used as the subphase, with volume V . After the surface has been cleaned from adventitious dust and so on by moving the barrier over it (in Figure 2 this would be done by sweeping the barrier from right to left, sucking up any impurities from the extreme lefthand corner, and then sweeping it back again), amphiphiles are placed on the freshly swept surface (of area A) of the subphase. Since, at this stage, the amphiphiles should be dispersed as uniformly as possible in the plane of the surface of the subphase, the amphiphiles are typically dissolved in a solvent at a rather dilute concentration (one millimolar or less). By virtue of their amphiphilic nature, the molecules can be tricky to dissolve and since there is no true alkahest it may be necessary to employ a mixture of solvents. Very often the solvent or solvents are immiscible with the subphase, but this is not an absolute requirement. High volatility is desirable in order not to have to wait too long before all the solvent molecules have evaporated (although the temporary presence of “molecular lubricant” might be required to ensure correct ordering on compression). The amphiphile solution is often dispensed from a needle placed very close to the subphase surface. The procedure is known as “spreading.” We denote the total quantity of added amphiphile molecules as M. Chloroform is often used to spread fatty acids on water despite its relatively high boiling point.

m

The moving barrier is a key component of the Langmuir trough. The design shown in Figure 2 is perhaps the most common; a variant is to have two barriers; they will typically be made of the same material as that of the vessel containing the subphase. An alternative design is to use a continuous flexible material enclosing a variable area with a constant perimeter. If very large numbers of layers are required, the hydrodynamic compression technique of Nitsch and Kurthen,19 which uses no moving barriers at all, is very attractive.

2.2

Monolayer metrology

The most basic requirement is to determine the surface pressure π . This is perhaps most commonly achieved using a Wilhelmy plate (Figure 3; for more discussion, see Ref. 7). The plate should be designed such that the principal downward force is γ p, where γ is the surface tension of the subphase surface and p the perimeter of the plate where it intersects the surface (i.e., twice the width plus the thickness). The secondary force is the pull of gravity on the rest of the plate; it depends on the proportion of its depth that is immersed, which should not change during an experiment. The downward forces are compensated by a load of one kind or another, the value of which necessary to achieve zero deflexion being recorded. It is customary to define the surface pressure as the difference between the measured surface tension and that of pure water. There is no guarantee that the surface pressure is uniform over the entire area A. Highly cohesive, rigid films give rise to severe pressure gradients,20, 21 which can be problematical when it is desired to transfer the monolayer to a substrate (Section 3) at constant pressure. It is also of interest to microscopically observe the floating monolayer. The only practical methods are those based on (usually visible) light, since there must be no contact. Snell’s law relating the angles of reflexion θ 1 and refraction θ 2 at the boundary between two media 1 (e.g., F

b

s

c

Figure 2 Cross-section of a Langmuir trough. c, container; s, subphase; m, monolayer of amphiphilic molecules; and b, moving barrier.

3

Figure 3

A Wilhelmy plate partly dipping into the subphase.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

4

Techniques

air) and 2 (e.g., water) is

c

n1 θ 1 = n2 θ 2

(5)

where the n are the refractive indices; when the reflected and transmitted rays are perpendicular to each other, we have θ 1 = θ B (the Brewster angle) = arctan

n2 n1

p

s

l

(6) g 0

and the reflected light is linearly polarized (Brewster’s law). Any optical arrangement making use of this phenomenon is particularly sensitive to the presence of additional material (with a polarizability different from that of either bulk medium) at the interface; Brewster angle microscopy (BAM, see Brewster Angle Microscopy, Techniques) captures images using light incident at the Brewster angle. The most common arrangement is to use p-polarized incident light, which is not reflected at all at the Brewster angle; therefore, any perturbation because of the presence of amphiphiles at the interface appears against a zero background. If the floating monolayer is photoluminescent and can be excited at a wavelength at which the subphase is not, imaging on the basis of the fluorescent emission also provides information about the nature of the molecular arrangement within the monolayer. If the amphiphiles have no intrinsic fluorescence, they can be doped with a small fraction of fluorophores. This procedure is, indeed, often adopted. However, from our existing knowledge that the pressure–area isotherms of novel compounds synthesized in the laboratory are often dramatically distorted if inadequate care has been taken over their purification, extreme caution is required when doping with fluorophores; the resulting mixed film, even if the fluorophores are only present at a fraction of a percentage, may behave significantly differently from the pure film. The metrologies available for in situ observation of the floating monolayer are rather limited. Much more information, especially about the structure of complex monolayers, has been obtained from examination following their transfer onto a substrate, in which case a much wider range of metrologies is available (Section 4). This assumes, of course, that structural details are faithfully transferred.

2.3

Pressure–area isotherms

After the amphiphiles have been spread on the Langmuir trough with its barriers opened to the fullest extent (and any solvent has evaporated), they are then slowly compressed to yield an isotherm prototypically shown in Figure 4. The

A

Figure 4 Sketch of a prototypical pressure–area isotherm. The different “phases” are marked: g, “gaseous,” molecules are separated from one another or grouped into small isolated rafts; l, “liquid,” molecules are more condensed but still highly compressible; s, “solid,” molecules are densely packed; and c, “collapsed,” the solid plate is ruptured and fragments ride over each other in the manner of tectonic plates.

isotherm can be characterized—very approximately—as having four successive stages, “gaseous” (essentially zero pressure; no interaction between molecules), “liquid” (denser than gaseous but still highly compressible), “solid” (virtually incompressible), and collapsed. Although the stages are sometimes called phases and considered to be analogous to the eponymous three-dimensional phases, this is not a very exact terminology. If the criterion for a phase is a distinct molecular symmetry, in many cases hundreds of phases characterize the isotherm (mixtures of different molecules will generally give more complicated isotherms).9 Furthermore, the “gaseous” stage is really more like a three-dimensional vacuum, the “liquid” stage more like a three-dimensional gas, and the “solid” stage can have either liquid or solid character; below a melting temperature Tm , the monolayer is rigid and the molecules have very limited mobility, whereas above Tm , they are rather more mobile (such that, for example, any defects such as pinholes are quickly annealed) and there may be no discernible transition between the “liquid” and “solid” stages. At this stage, the molecules can be considered to be in a eutactic environment, typically having an exact orientation, although the solid “phase” may contain domains of different orientation (for a simulation of this phenomenon, see Ref. 17). The ability to choose with ease the two-dimensional density (i.e., the number of molecules packed into unit area) is one of the great advantages of the Langmuir trough technique. Phenomena that depend on a critical (perhaps in the sense of a crossover between two types of behavior) density can scarcely be investigated by any other techniques.22 The collapse pressure π c and collapse area Ac can be systematically determined according to an extrapolation

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

Langmuir–Blodgett films 0.0002

5

0.263

0.00015 0.460

0.0001

0.615 0.527

5E−5

1.283

0.924 1.156

0

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Frequency (Hz)

Figure 5 Fourier transform of the derivative (with respect to time) of the interfacial pressure obtained through compressing (at a rate of 9.1 mm2 /s) a 1 : 1 behenic acid:cholesterol mixture spread on water.24 (Reproduced from Ref. 24.  Collegium Basilea, 2008.)

procedure.17 The nature of the collapsed state, which might be compared to “pack” or “pressure” ice in the Antarctic, or even tectonic plates in collision, and hence used to model those phenomena, depends strongly on the rigidity of the film. Very stiff materials, such as lead behenate, may fragment into a multiplicity of polygonal nanoplatelets; more flexible materials may simply buckle or crumple. Considerable insight into the lateral cohesion of the floating monolayer can be obtained by examining the hysteresis of the isotherm. The greater the cohesion, the greater the hysteresis. To date, this is a rather underexploited approach to characterize partially compressed monolayers (i.e., at a pressure lower than π c ; collapse is usually an irreversible phenomenon). However, monolayer viscoelasticity (compression and shear moduli) has recently been determined by such an approach.23 Especially, for monolayers consisting of more than one substance, there is an immense amount of information contained in the pressure–area isotherm. For example, a mixed behenic acid/cholesterol film gives what seems at first sight to be a very noisy isotherm, but in fact it contains characteristic oscillation frequencies, as can readily be seen from the Fourier transform of the derivative (with respect to time; the monolayer was compressed at a uniform rate, hence in effect making area and time equivalent) of the isotherm (Figure 5). The origin of these oscillations has been traced to spatial structuring.24

2.4

Useful monolayer relations

The most basic relation (conservation of total amphiphile mass M; i.e., what was added) is M = A + cb V

(7)

where  is the surface concentration and cb is the bulk concentration, usually very low but since V might be quite large the total amount of dissolved amphiphile might be appreciable; amphiphiles with short alkyl chain lengths (n < 12) may have significant solubility. This equation implies that a plot of M versus A at a particular surface pressure π, ˜ constructed from a series of isotherms with different M, yields  (from which the partial molecular area of the amphiphile can be deduced) and cb .25 A further plot of  versus π˜ may reveal conformational transitions. It is a general rule that the longer the apolar moiety (cf. the central molecule in Figure 1), the more stable and rigid the film. Similarly, if a metal salt is present in the subphase, the “softer” the metal (in the Lewis acid/base sense26, 27 ) the more stable and rigid the film. When spreading particles of radius r, the immersion depth di is given by28 di = r(1 + cos θ a )

(8)

where θ a is the advancing contact angle in the liquid in which they are immersed.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

6

Techniques

3

FILM TRANSFER

n

n

The basic discovery was that by moving a solid substrate through the floating monolayer, provided certain conditions were fulfilled, the film could be quantitatively transferred to the substrate29 ; that is, the transfer ratio T, defined by T=

A substrate area

(9)

where A is the change in floating monolayer area, usually at constant surface pressure, is equal to unity. If T is negative, it may be inferred that material from the surface of the substrate has been transferred to the surface of the subphase. The primary factors directly influencing T are the interfacial energetic nature of the substrate, which can be either hydrophilic or hydrophobic; the direction of travel of the substrate (up or down); the speed v of travel; and the surface pressure. Atmospheric conditions are also important, especially relative humidity, but not directly; the interval between successive transfers also appears to play a role— both these secondary influences operate through changing the interfacial energetic nature of the substrate. A further important factor for multilayer deposition is the strength of adhesion between successive layers. The first three primary factors determine the shape of the meniscus formed between the substrate and the surface of the subphase. Attempts have been made to analyze the process in minute detail and relate the factors (their parameters) to T30–32 ; here, only salient features of practical importance are mentioned. Figure 6 serves to visualize the process. If the substrate is hydrophilic (assuming that the subphase is water) and is moving up (i.e., the initial position is immersed in the subphase), the meniscus will have a contact angle 90◦ with the subphase, as shown in the right-hand diagram of Figure 6. Given that wetting is a mesoscale phenomenon,35 it is presumably permissible to neglect any effect of the floating monolayer on the shape of the meniscus. Hence, on lowering S˜ from a starting position in the air, the apolar moieties will be automatically brought into contact with the substrate and deposition, again with T = 1, should result. The motion of the substrate also favors the necessary form of the meniscus. The deposited ˜ structure is then SAP. It may then be moved upward to ˜ deposit the next layer with orientation SAPPA, and so forth. This is the most common deposition mode for classical LB films and is usually denoted “Y-type.” If the film is not very stable, as is commonly the case with unsaturated phospholipids with moderate chain length, it may happen that when the structure SPA moves downward to deposit the next layer, the first one is removed instead unless v is very great (which may be impracticable). Vincent Schaefer, working with Langmuir, developed the horizontal transfer method to overcome this difficulty36 (cf. suminagashi ); in effect, if the substrate moves perfectly parallel to the floating monolayer, v is infinite; practically speaking, it is sufficient that it is very great. Note that only one side of the substrate is coated. This method is really only practicable when coating a single monolayer or a

7

2 4 6 µm

X Z

2.000 µm/div 20.000 nm/div

09021326.002

Figure 8 Atomic force micrograph of a bilayer of barium stearate deposited at 25 mN m−1 on freshly cleaved Muscovite mica (one upstroke and one downstroke) and stored in the subphase for about 30 min before imaging. It appears that elongated patches of the bilayer have detached themselves from the surface and stacked themselves onto the top of undisturbed areas (see also Figure 9).

bilayer (SPAAP), not only because the substrate needs to be mounted differently for LS compared with LB transfer and the creation of a sufficiently versatile substrate holder might be difficult but also because the coating asymmetry between two faces of the substrate introduces an ambiguity into the system. The LS technique is also required for transferring extremely rigid films. The meniscus may be mechanically forced to adopt the shape contrary to its equilibrium one, namely by moving a hydrophobic substrate sufficiently rapidly upward or a hydrophilic substrate sufficiently rapidly downward. In these cases, no transfer should take place (T = 0). By these and other astucious means, modes of deposition other ˜ than “Y-type” can be achieved: “X-type” is SAPAP. . .) and “Z-type” is SPAPA. . .). A decisive factor is often the surface energy of the immediately previously deposited film.37 Even at present, the mechanism of non-“Y”-type film formation is disputed, film instability also having been proposed as part of the mechanism.38, 39 Spontaneous film reorganization seems to be quite a common phenomenon.39 It has been observed that even an SPAAP bilayer reorganizes such that patches of 0, 2, 4, and 6 layers are all readily observable (Figures 8 and 9). Schwartz et al., who have observed similar phenomena with odd numbers of layers,40, 41 ruled out the entropic effects that seem to drive thermal crystal roughening as a possible mechanism but pointed out the paradox of unfavorable

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

8

Techniques nm 21.8 16.3 10.9 5.4 µm

0.0 0

3

6

9

12

15

Figure 9 Profile of the barium stearate deposit (Figure 8). The zero is of course arbitrary; it is presumed that the height of 5.4 nm corresponds to the uncoated mica surface. A bilayer is about 5 nm thick; hence, what we are seeing is that patches of the bilayer have detached themselves from the surface and stacked themselves onto the top of undisturbed areas. The stacking seems to be invariably associated with an immediately adjacent patch of removal, but not all the removed material stacks onto an immediately adjacent bilayer, implying considerable surface mobility.

interaction between the exposed alkyl chains and water at the edges of the reconstructed stacks. It is customary to carry out deposition at constant surface pressure; it is usual for the substrate to be dipped in and out of the subphase close to the surface pressure transducer. Originally, constant surface pressure was maintained by observing the output signal from the Wilhelmy plate or other surface pressure transducer (Section 2.2) and manually advancing the moving barrier (Figure 2) to keep π constant; nowadays, it is common to link a motor driving the barrier with the surface pressure transducer via a feedback circuit. Because of the soft mechanical element in the feedback loop (the partially compressed floating monolayer), this is not, in fact, a very easy problem to solve, especially if film transfer is rapid or if the monolayer is exceptionally rigid. Success of the whole operation might, furthermore, be vitiated by surface pressure gradients. At the very least, the inevitable distance between the moving barriers and the pressure transducer will introduce delay into the feedback loop and hence the possibility of chaotic operation. Especially, if it is intended to deposit a large number of layers, the trough will have to be very large and, hence, the moving barrier will initially be very distant from the surface pressure transducer, enhancing the likelihood of surface pressure gradients. The alternative would be to use a small trough and periodically replenish the monolayer material. Many trough designs place the substrate dipping mechanism (and pressure transducer) near one end of a long rectangular trough. Usually, it is only desired to coat one face of the substrate [which is, in any case, the result from Langmuir–Schaefer (LS) transfer]. Nevertheless, unless special precautions are taken with the opposite face (which should be superhydrophilic and superhydrophobic, if transfer is going to be entirely suppressed), transfer will take place on both faces. Unless, however, the floating monolayer is rather fluid, there is some ambiguity regarding the supply of material to the “back” face (facing away from

the moving barrier). The flow induced through needing to supply the back face might introduce inhomogeneity.42 One way to remove this ambiguity is to place the dipping mechanism at the center of the trough and use two symmetrical moving barriers (or the constant perimeter barrier). (For a discussion and another solution to this problem, see Ref. 43.) The whole procedure can, at least in principle, be carried out more-or-less automatically, including replenishment of the monolayer material and maintaining the level of fluid in the subphase constant, provided that the trough is liberally endowed with sensors and actuators. Composite films may be deposited from multiple troughs, on each one of which a different monolayer is floating. The vast majority of reported work has been carried out using a single substance; that is, every transferred monolayer is chemically identical. There is enormous unexplored scope for creating structures of complex functionality by using two or more different substances. One of the few examples of such a complex structure is provided by the extensive series of experiments undertaken by Kuhn and M¨obius44 on elucidating the nature of energy transfer from a layer of emitting fluorophores across intervening inert spacer layers of variable thickness to a layer of receptor molecules. It is doubtful whether this work, which required great precision because of the quantitative inferences drawn, could be undertaken with any other experimental technique. There is obvious scope for using the LB technique to assemble superlattices. Similarly underexploited is the investigation of films, each monolayer of which is composed of two or more substances present in comparable quantities. A great many processes, such as selective adsorption from a bulk mixture or catalysis, depend on the controlled heterogeneity of a surface, and one way of achieving such heterogeneity is by creating a mixed Langmuir film that forms stable domains (Figure 1021 ) and transferring it to a solid

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

Langmuir–Blodgett films

20 nm

20 µm x 20 µm

0 nm

Figure 10 Atomic force micrograph of a mixed behenic acid/pentanoic acid film transferred using the LB–LS technique onto Muscovite mica at a surface pressure of 15 mN/m.21 (Reproduced by permission of the Collegium Basilea.)

substrate. At present, however, the fabrication of such films is a wholly empirical process. Moreover, systems forming fine stable domains appear to be rare. One notices, however, that natural cell membranes (lipid bilayers) contain a great variety of different lipid molecules. As already noted, these mixed films show very complex compression behavior (Figure 5).24

4

TRANSFERRED FILM METROLOGY

A much vaster range of techniques is available for examining films deposited on a solid substrate compared with those usable with the floating monolayer (cf. Section 2.2).45 X-ray diffraction, especially with a grazing angle of incidence in order to minimize interference from the substrate, can yield the usual structural information (see XRay Diffraction: Addressing Structural Complexity in Supramolecular Chemistry, Techniques); the experiment must be carried out in vacuum. (Grazing incidence X-ray diffraction is possible with a single floating monolayer, with extreme difficulty and laboriousness and making use of a synchrotron source.) Neutron diffraction can also be very informative but is of course not so readily available. A more exotic technique is to use a positron beam.46 These techniques need a minimum of several monolayers in order to yield useful data. The surface can be probed with atomic force microscopy (see Atomic Force Microscopy (AFM), Techniques) and related techniques such as scanning

9

near-field optical microscopy (see Scanning Near-Field Optical Microscopy (SNOM), Techniques) and scanning tunneling microscopy (see Scanning Tunneling Microscopy (STM), Techniques); these techniques have the advantage that films can be examined in the presence of liquid (e.g., water), if it is necessary for their stability. Somewhat less convenient, because most LB films are not electrically conductive and hence need to be coated with a few nanometers of a suitable metal, is scanning electron microscopy (see Scanning Electron Microscopy, Techniques), which also requires a vacuum. Techniques based on reflexion can deliver very precise information regarding film structure and have the advantage that the measurements can be carried out in situ, in the presence of liquid if necessary. The classic technique is scanning angle reflectometry, although it is rather slow. Ellipsometry is based on the same principles but has the disadvantage that the optical invariants associated with the measured parameters are less sensitively linked to the structural parameters of the film47 compared with reflectometry. Reflectometry can, moreover, nowadays benefit from the availability of planar optical waveguides, which have thousands of total internal reflexions per millimeter, to yield data with a greatly improved signal-to-noise ratio ideally suited to the structural characterization of thin films.48 Essentially any transparent dielectric material can be made into a waveguide and used as a substrate for an LB film. The most usual arrangement for measuring the propagation constants of the waveguide is to incorporate a grating coupler into it.49 Since the grating is a region of periodically modulated refractive index, either chemical or topographical heterogeneity is necessarily introduced. However, the chemical heterogeneity can be buried, and since very weak coupling is sufficient, the topographical modulation can be so shallow (e.g., a depth of 5 nm with a grating constant of 500 nm) that it has a negligible influence on the transferred film. With the help of the grating, the propagation constants (that is, the effective refractive indices N) of the waveguide with its adlayer, the transferred film, can be determined. The highest sensitivities ∂N/∂x, where x is an adlayer parameter such as its thickness d or its refractive indices, are obtainable with the thinnest, monomode waveguides that can only support the zeroth modes of both the transverse electric (TE) and transverse magnetic (TM) polarizations,49 allowing two adlayer parameters to be determined. Nevertheless, an LB film is typically uniaxial with different ordinary (no ) and extraordinary refractive indices (ne ), requiring the measurement of at least three waveguide parameters to enable no , ne , and d to be determined. One may use a thicker waveguide and measure the zeroth and first modes with lesser sensitivities, one may determine d independently33 (or, indeed, any one of the three parameters), or one may use two wavelengths.50

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

10

Techniques

C

nC < nF nA

A

C A

z

F

nF

S

nS < nF

dF F ∆zF,S

(a)

S

(b) E or H

x

N / dA /10−4 nm−1

4 3

TM0

TE0

2

TE1

TM1

1 0

0

(c)

200

400

600

dF /nm

Figure 11 The principle of optical waveguide lightmode spectroscopy (OWLS). (a) The electromagnetic field distribution (shaded) of a zeroth guided mode in a four-layer waveguide comprising the high refractive index core (F), a lower refractive index support (S), a transferred Langmuir–Blodgett film (A), and the cover medium (C), typically air or water. (b) The same waveguide structure, but showing the guided mode in the ray-optic representation. All reflexions are total internal, but there is a phase difference  between the ˆ = 1. The sum of all the phase differences as ˆ i , and |R| reflected and incident light (i.e., the Fresnel coefficient of reflexion Rˆ = |R|e the ray undergoes one cycle must equal an integral multiple of 2π; hence, the guided modes are discrete. The penetration depth z equals the distance at which the electromagnetic field intensity has decayed to 1/e of its peak value. (c) Waveguide sensitivities (with respect to film thickness) of the first four modes as a function of the waveguide thickness.

Figure 11 shows the principle of the waveguide measurement, which is nowadays usually referred to as optical waveguide lightmode spectroscopy (OWLS). The technique has been extensively reviewed elsewhere (see Ref. 51 and references therein, which also covers scanning angle reflectometry and ellipsometry as well as some other techniques such as surface plasmon resonance, useful if the Langmuir film is transferred to a noble metal substrate; Ref. 52 is a more recent survey). OWLS, in particular, as the most sensitive of these optical techniques, is able to yield useful data on a single monolayer.

5

POSTTRANSFER PROCESSING

Skeleton films were invented and extensively investigated by Blodgett.53 The idea is to create a mixed film and subsequently solvent-extract one of the components. There has recently been a revival in interest as a possible way of creating thin films of controlled porosity.54 If the deposited film incorporates a metal (e.g., by having had metal ions in the subphase to complex with a

carboxylate group of the amphiphile), it may be pyrolyzed to produce an inorganic compound (such as a metal oxide, if the pyrolysis is carried out in an oxygen-containing atmosphere).55, 56 This is a useful way of readily making ultrathin coatings from a huge variety of metal oxides, more conveniently than by physical vapor deposition. It is very easy to incorporate a polymerizable group (e.g., diacetylene) in the amphiphile. If polymerization can be accomplished using ultraviolet light, the process is very convenient. It is useful for creating tougher films. However, it seems that the polymerization often leads to extra defects being incorporated into the film.

6

RELATIONSHIPS TO OTHER TECHNOLOGIES

A thin-film assembly technique with some superficial resemblance to the LB methodology is that of the selfassembled monolayer (SAM)57 : the precursor molecules have a similar asymmetry to those of the Langmuir precursors (Figure 1), but the polar moiety should have strong

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

Langmuir–Blodgett films chemical affinity for the substrate, to which it binds spontaneously and the LW interactions serve to order the molecules to form a dense, close-packed monolayer. A variant of the SAM concept is the spontaneous formation of lipid bilayers from lipid vesicles colliding with a substrate.58 A multilayer technique also requiring repeated dipping of a substrate into liquid is alternating polyelectrolyte deposition (APED).59 A polyanion and a polycation are selected (they may be long-chain organic polymers, nanoparticles, or nanorods), and the substrate must also be electrostatically charged. It is first dipped into a solution of the polyelectrolyte of opposite charge, with which it is completely covered within a few seconds and its charge is reversed, after which it is withdrawn, rinsed in pure solvent, and then dipped into the other polyelectrolyte, leading again to charge reversal, and so forth, in principle at libitum. The successive layers penetrate into one another to some extent60 ; hence, the resulting film is less laminar than an LB structure. LB films share some features with nematic liquid crystals61 ; in the form, the molecules are much more strongly constrained within two dimensions than in the latter.

7

APPLICATIONS

The attraction of the LB approach is the very precise molecular structuring achievable with relatively low-cost equipment operating under mild conditions (room temperature and pressure). In other words, one has the precision of a technique such as molecular beam epitaxy (MBE) with much simpler infrastructure. Disadvantages are the extreme sensitivity of the film quality to dust and vibration and the difficulty of adapting the fabrication procedure to largescale, high-throughput production. The LB technique was used by Frewing to create very precise films in fundamental studies of boundary lubrication.62 One of the early hopes was that the technique could be used to create ultrathin insulating layers for field effect transistors and other electronic devices. This application is, however, bedeviled by the presence of very small defects (pinholes) in the films. These defects can be obviated by using a material whose transition temperature is well below room temperature (at which, it is assumed, fabrication takes place), such as the phospholipids found in nature as the main amphiphilic components of the ubiquitous bilayer lipid membrane that surrounds cells and their internal organelles, but these molecules are not very robust and would not have the longevity required in typical electronics applications. Another attempted application has been the creation of planar optical waveguides, but it turned out to

11

be very difficult to create films with sufficient structural perfection to have the requisite low absorption coefficient. Hence, at present, the main applications remain in the research environment. LB films are particularly useful for the reconstitution of natural cell membranes on planar substrates for various kinds of biologically oriented investigation.63 Although most reported work has used pure, synthetic lipids, the technique can equally well be used with natural cell membrane extracts containing proteins (J.J. Ramsden and V. Mirsky, unpublished observations). Langmuir monolayers have been used to create biomimetic structures (see, e.g., Nandi and Vollhardt’s review of monolayers made from chiral molecules64 ). Another line of research has been to assemble complex optical structures with exquisite precision; examples are the work of Kuhn and M¨obius on resonant energy transfer involving dye molecules44 ; and second-harmonic generation,65, 66 the principal requirement of which is a noncentrosymmetric structure, readily achievable using the LB technique. The LB technique has been indispensable in molecular electronics research, based on the Aviram–Ratner concept. In this case, deposition is on a metal substrate, which serves as one of the electrodes.67, 68 Multilayers made from iron-containing fatty acids have been used not only for fundamental investigations69 but also as candidates for ferromagnetic nonvolatile memory applications. There has also been recent interest in the creation of ferroelectric LB multilayer films.70 The prototypical chemical sensor involves concentrating the analyte (the environmental concentration of which constitutes the measurand) in the vicinity of the transducer capable of converting chemical presence into an electrical or optical signal. The sensitivity of most transducers decays (exponentially in the case of evanescent fieldbased optical waveguide sensors) from the surface of the transducer away into the bulk environment; hence, ultrathin films capable of highly concentrating an analyte in the immediate vicinity of the transducer surface are very desirable. Ultrathin films also have the advantage of not introducing any delay into the response due to mass transport limitations. It follows that LB oligolayers with strong selective affinity for the analyte are excellent candidates for the capture layers used to coat transducers. An important class of analyte is constituted by the small organic molecules used as medicinal drugs; since, in order to penetrate into the living cell (as is usually required for their action), they must first interact with the cell membrane. Hence, phospholipid bilayers mimicking that membrane are obvious candidates for capture layers and their efficacity as such has been demonstrated71, 72 ; captured concentrations three or four orders of magnitude higher than in the bulk can readily be achieved. In the case of optical transducers

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

12

Techniques

based on total internal reflexion, because of the detailed structural information regarding the LB films obtainable from analyzing the optical data, film-coated substrates are also very useful for pharmacological studies and molecular biological work aimed at elucidating the mode of action of toxins.73 LB overlays have also been successfully placed on side-polished optical fibers in order to create sensors for vapors such as hydrochloric acid and ammonia.74–76 Of particular relevance to the core topic of this chapter is the ease with which Langmuir films can be created from many unconventional amphiphiles (and even nonamphiphiles77 ) and deposited on substrates using the LB technique (for some examples of large, complicated organic and metal-organic molecules transferred as LB films, see Refs. 78, 79). The LB technique constitutes a particular approach to creating supramolecular structures with strict laminar order. The LB technique has also turned out to be valuable for creating controlled striations in the thin film coating a substrate. Traditionally, it was always emphasized that the dipping motion should be extremely uniform, and it was customary to employ the best available precision screw drives for the lifting mechanism. The discovery of physisorption instabilities during Langmuir wetting80 has transformed the inevitable (to some degree) fluctuations into a technique for producing aligned stripes with precision in the nanoscale.24, 81

had fallen to rather less than half of what it was in the 1990s, which was a decade of consistently strong activity, from which it might be concluded that we are currently in an “LB trough.” Nevertheless, there are signs of renewed growth; the expansion of nanotechnology has given fresh impetus to LB films, and it would appear that there is particularly strong and hitherto unexploited potential for investigating and developing composite films incorporating a greater variety of objects than the single amphiphiles of conventional architecture that have hitherto received by far the most attention. Convenience of manufacture remains a challenge for large-scale introduction of commercial devices, but there is still enormous unexploited potential for the technique as a research tool. A grand challenge for the field is to tackle the “inverse design problem,” namely, how to design a structure to have a particular specified functionality.

ACKNOWLEDGMENTS The author thanks Gergely Antal for his skillful assistance with the experiments reported in Figures 8 and 9.

REFERENCES 1. A. Pockels, Nature (London), 1891, 43, 437–439.

8

CONCLUSIONS

The LB technique offers a unique way of achieving a eutactic environment in two dimensions (at the floating monolayer stage), which can then be sequentially transformed into three-dimensional structure preserving the order. The achievable control is at least equal to that achievable using MBE, although the types of molecules available as precursors are very different. The infrastructure required and the apparatus itself is far simpler and cheaper than physical vapor deposition (PVD) setups. The drawback is the rather slow and cumbersome fabrication procedure, which does not lend itself well to large-scale, high-throughput production. Automation is possible but requires complicated precision robotics. The hydrodynamic compression trough is very useful here. What are the perspectives for LB films? Surveying the history of the field since the first paper of Pockels over a century ago, it is clear that there have been several revivals followed by a languishing of interest. If one looks at the publishing and citation activity (a rough guide to which can be obtained by simply counting the number of returns obtained through searching for the term “LB y” in Google Scholar, where y is the year), by 2009, the level

2. A. Pockels, Nature (London), 1892, 46, 418–419. 3. A. Pockels, Nature (London), 1893, 48, 152–154. 4. A. Pockels, Nature (London), 1894, 50, 223–224. 5. I. Langmuir, Liquids. J. Am. Chem. Soc., 1917, 39, 1848–1906. 6. G. G. Roberts, Adv. Phys., 1985, 34, 475–512. 7. B. P. Binks, Adv. Colloid Interface Sci., 1991, 34, 343–432. 8. H. Kuhn, Thin Solid Films, 1989, 178, 1–16. 9. I. R. Peterson, J. Phys. D, 1990, 23, 379–395. 10. ISO TS 27687 : 2008, Nanotechnologies–Terminology and Definitions for Nano-Objects, British Standards Institution, London, 2008. 11. C. J. van Oss, Colloids Surf., B, 1995, 5, 91–110. 12. M. G. Cacace, E. M. Landau, and J. J. Ramsden, Q. Rev. Biophys., 1997, 30, 241–278. 13. C. J. van Oss and R. J. Good, Colloids Surf., 1984, 8, 373–381. 14. C. J. van Oss, Forces Interfaciales En Milieux Aqueux, Masson, Paris, 1996. 15. J. J. Ramsden, Adsorption kinetics of proteins. In Encyclopedia of Surface and Colloid Science, ed. A. Hubbard, Dekker, New York, 2002, pp. 240–261.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

Langmuir–Blodgett films

13

16. M. Li, A. A. Acero, Z. Huang, and S. A. Rice, Nature, 1994, 367, 151–153.

46. Cs. Szeles, A. V´ertes, E. Kiss, et al., J. Colloid Interface Sci., 1995, 176, 358–361.

17. G. Tolnai, A. Agod, M. Kabai-Faix, et al., J. Phys. Chem. B, 2003, 107, 11109–11116.

47. E. K. Mann, Langmuir, 2001, 17, 5872–5881.

18. I. Langmuir and V. J. Schaefer, J. Am. Chem. Soc., 1937, 59, 2400–2414. 19. W. Nitsch and C. Kurthen, Thin Solid Films, 1989, 178, 145–155. 20. M. M´at´e, J. H. Fendler, J. J. Ramsden, and J. Szalma, Langmuir, 1998, 14, 6501–6504. 21. S. Alexandre, C. Lafontaine, and J.-M. Valleton, J. Biol. Phys. Chem., 2001, 1, 21–23. 22. W. Meier and J. J. Ramsden, J. phys. Chem., 1996, 100, 1435–1438. 23. P. Cicuta and E. M. Terentjev, Eur. Phys. J. E, 2005, 16, 147–158. 24. S. Alexandre and J.-M. Valleton, J. Biol. Phys. Chem., 2008, 8, 33–36. 25. G. Schwarz, Ber. Bunsen-Phys. Chem., 1996, 100, 999–1003.

48. J. D. Swalen, M. Tacke, R. Santo, and J. Fischer, Opt. Commun., 1978, 18, 387–390. 49. K. Tiefenthaler and W. Lukosz, J. Opt. Soc. Am. B, 1989, 6, 209–220. 50. R. Horvath, G. Fricsovszky, and E. Papp, Biosens. Bioelectron., 2003, 18, 415–428. 51. J. J. Ramsden, Q. Rev. Biophys., 1994, 27, 41–105. 52. J. J. Ramsden, High resolution molecular microscopy, in Proteins at Solid-Liquid Interfaces, ed. Ph. Dejardin, SpringerVerlag, Heidelberg, 2006, pp. 23–49. 53. K. B. Blodgett, J. Phys. Chem., 1937, 41, 975–986. 54. K. P. Girard, C. M. Hanley, T. K. Vanderlick, and J. A. Quinn, Ind. Eng. Chem. Res., 2001, 40, 4283–4287. 55. D. T. Amm, D. J. Johnson, T. Lausen, and S. K. Gupta, Appl. Phys. Lett., 1992, 61, 522–524.

26. R. G. Pearson, J. Am. Chem. Soc., 1963, 85, 3533–3539.

56. T. Laursen, T. J. Johnson, D. T. Amm, and J. E. Haysom, Mater. Sci. Eng., 1994, A174, L13–L15.

27. R. G. Parr and R. G. Pearson, J. Am. Chem. Soc., 1983, 105, 7512–7516.

57. W. C. Bigelow, D. L. Pickett, and W. A. Zisman, J. Colloid Sci., 1946, 1, 513–538.

28. Z. H´orv¨olgyi, G. Medveczky, and M. Zr´ınyi, Colloid Polym. Sci., 1993, 271, 396–403.

58. G. Cs´ucs and J. J. Ramsden, Biochim. Biophys. Acta, 1998, 1369, 61–70.

29. K. B. Blodgett, J. Am. Chem. Soc., 1935, 57, 1007–1022.

59. A. Yu. Grosberg, T. T. Nguyen, and B. I. Shklovskii, Rev. Mod. Phys., 2002, 74, 329–345.

30. S. Egusa, N. Gemma, and M. Azuma, J. Phys. Chem., 1990, 94, 2512–2518. 31. J. G. Petrov 2490–2496.

and

P. G. Petrov,

Langmuir,

1998,

14,

32. L. Y. Zhang and M. P. Srinivasan, Colloids Surf. A, 2001, 193, 15–33.

60. (a) J. J. Ramsden, Yu. A. Lvov, and G. Decher, Thin Solid Films, 1995, 254, 246–251; Errata: (b) J. J. Ramsden, Yu. A. Lvov, and G. Decher, Thin Solid Films, 1995, 261, 343–344. 61. G. W. Gray, Proc. R. Soc. London, Ser. A, 1985, 402, 1–36.

33. J. J. Ramsden, Philos. Mag. B, 1999, 79, 381–386.

62. J. J. Frewing, Proc. R. Soc. London, Ser. A, 1942, 181, 23–42.

34. P. Wiggins, J. Biol. Phys. Chem., 2010, 10, 152–157. 35. P. G. de Gennes, Rev. Mod. Phys., 1985, 57, 827–863.

63. O. Michielin, J. J. Ramsden, and G. Vergeres, Biochim. Biophys. Acta, 1998, 1375, 110–116.

36. I. Langmuir and V. J. Schaefer, J. Am. Chem. Soc., 1938, 60, 1351–1360.

64. N. Nandi and D. Vollhardt, Chem. Rev., 2003, 103, 4033–4075.

37. R. Popovitz-Biro, K. Hill, E. Shavit, et al., J. Am. Chem. Soc., 1990, 112, 2498–2506.

65. W. Hickel, B. Menges, O. Althoff, et al., Thin Solid Films, 1994, 244, 966–970.

38. E. P. Honig, Langmuir, 1989, 5, 882–883.

66. G. J. Ashwell, R. C. Hargreaves, Nature, 1992, 357, 393–395.

39. A. Angelova, F. Penacorada, B. Stiller, et al., J. Phys. Chem., 1994, 98, 6790–6796. 40. D. K. Schwartz, J. Garnaes, R. Viswanathan, J. A. N. Zasadzinski, Science, 1992, 257, 508–511.

and

41. D. K. Schwartz, R. Viswanathan, and J. A. N. Zasadzinski, J. Phys. Chem., 1992, 96, 10444–10447. 42. O. Albrecht, H. Matsuda, K. Eguchi, and T. Nakagiri, Thin Solid Films, 1992, 221, 276–280. 43. B. R. Malcolm, Thin Solid Films, 1989, 178, 17–25.

C. E. Baldwin,

et al.,

67. A. S. Martin, J. R. Sambles, and G. J. Ashwell, Phys. Rev. Lett., 1993, 70, 218–221. 68. G. J. Ashwell, B. J. Robinson, M. A. Amiri, et al., J. Mater. Chem., 2005, 15, 4203–4205. 69. N. Rozlosnik, D. L. Nagy, E. Giesse, et al., Mol. Cryst. Liq. Cryst., 1995, 262, 357–360. 70. S. Ducharme, T. J. Reece, C. M. Othon, R. K. Rannow, IEEE Trans. Device Mater. Reliab., 2005, 5, 720–735. 71. J. J. Ramsden, J. Phys. Chem., 1993, 97, 447–448.

44. D. Moebius and H. Kuhn, J. Appl. Phys., 1988, 64, 5138–5141.

72. J. J. Ramsden, Experientia, 1993, 49, 688–692.

45. M. Vandevyvera, Thin Solid Films, 1988, 159, 243–251.

73. J. J. Ramsden, Chimia, 1999, 53, 67–71.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

14

Techniques

74. D. Flannery, S. W. James, R. P. Tatam, and G. J. Ashwell, Opt. Lett., 1997, 22, 567–569.

78. M. J. Cook, J. McMurdo, D. A. Miles, et al., J. Mater. Chem., 1994, 4, 1205–1213.

75. D. Flannery, S. W. James, R. P. Tatam, and G. J. Ashwell, Appl. Opt., 1999, 38, 7370–7374.

79. H. Gong, C. Wang, M. Liu, M. Fan, J. Mater. Chem., 2001, 11, 3049–3052.

76. S. W. James, N. Rees, R. P. Tatam, and G. J. Ashwell, Opt. Lett., 2002, 9, 686–688.

80. K. Spratte, L. F. Chi, and H. Riegler, Europhys. Lett., 1994, 25, 211–217.

77. Gy. Tolnai, F. Csempesz, M. Kabai-Faix, et al., Langmuir, 2001, 17, 2683–2687.

81. L. Li, P. Gao, K. C. Schuermann, et al., J. Am. Chem. Soc., 2010, 132, 8807–8809.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc034

Affinity Capillary Electrophoresis as a Tool to Characterize Intermolecular Interactions Steffen Kiessig, Alexandra Stettler, Samuel Fuhrimann, and Maria Anna Schwarz Solvias AG, Basel, Switzerland

1 Introduction 2 High-Performance Capillary Electrophoresis (HPCE) 3 ACE 4 Conclusion Nomenclature References

1

1 2 3 16 17 17

INTRODUCTION

Noncovalent interactions are an important part of many biochemical and chemical reactions and form the basis of living systems. The assessment of such interactions is of fundamental interest in describing biochemical and physiological processes. Most biochemical functions involve, at the molecular level, noncovalent bonds. For example, the storage and replication of genetic information and base stacking in the DNA double helix are dependent on hydrogen bonding, π -stacking, and electrostatic forces. All reactions in cells are catalyzed by enzymes often requiring a reversible binding of the substrate to the enzyme. For artificial noncovalently bound structures, molecular recognition is the basis for chiral catalysis, sensors, and separation principles in chromatography, to name just a few. Furthermore, in drug development, the study of interactions with proteins,

oligonucleotides, and artificial drug delivery substances is of great interest. It is not surprising that great efforts have been made in the last two decades to develop methods that are able to qualify and quantify noncovalent interactions of different origin. Methods for the measurement of binding parameters of a ligand to a receptor can be classified into two categories: mixture based (Fourier transform infrared spectroscopy (FTIR), Raman, nuclear magnetic resonance (NMR), UV, densimetric techniques, potentiometric titrations, isothermal calorimetry, etc.) and separation based (ultrafiltration–centrifugation, chromatography, and electrophoresis). All these methods rely on the alteration of molecular or dynamic parameters during the titration by forming a receptor–ligand molecule. While with mixturebased methods selective detection systems are required for the evaluation of the formed complex, in separation-based methods the selectivity is given by dynamic parameters (retention, migration, etc.). Thus, with the utilization of flexible detection systems (UV, fluorescence, MS, and amperometry) for the receptor–ligand complex, a wide range of interactions (electrostatic, hydrogen bonding, and hydrophobic) are accessible for the characterization of reversible bonds. In contrast, with mixture-based studies, the applications are limited by the detection method. Certainly, these factors have contributed to the fast development of affinity capillary electrophoresis (ACE) as the method of choice for different interacting systems. Note that the term ACE used here includes affinity studies in general, without considering the measurement conditions (eluting profiles) as described in the following sections.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

2

Techniques

The general advantages of ACE are not only the speed and simplicity of measurement, but also the sensitivity and the possibility of quantifying reversible interactions by the calculation of conventional binding constants. In principle, measurements with more than one receptor, as well as the analysis of complex equilibria (associated equilibria) and equilibria of higher orders, are feasible. Even though only calculations of conventional binding constants (e.g., if one or more than one ligand binds) are often practicable by ACE (stoichiometry parameters have to be assessed by reference methods), with an intelligent measurement strategy additional information regarding the selectivity or the number of bound ligands can be obtained. Species formed during the separation process in the presence of varying ligand concentrations can be characterized by changing their own ionic mobility or by studying the characteristics of the detection signal. This fact is of particular interest if new and unknown interacting systems are studied. ACE has the potential to become a powerful tool for studying interactions of small molecules (drugs, inorganic cations, intercalators) and biological macromolecules (proteins, DNA, enzymes, peptides).1, 2 ACE is an analytical approach in which the electrophoretic migration behavior or peak area of the receptor or the receptor–ligand molecules is evaluated to quantify and identify specific binding. In general, two different modes of capillary electrophoresis (CE) are used—mobility change analysis (here termed as ACE-µ, but also named in the literature as electrophoretic mobility shift assay (EMSA)) and concentration change evaluation. ACE-µ can be used to study low-affinity complexes with fast kinetics. Moreover, the evaluation of peak areas of pre-equilibrated samples can also be meaningful for high-affinity systems with slow complexation kinetics.

2

HIGH-PERFORMANCE CAPILLARY ELECTROPHORESIS (HPCE)

High-performance capillary electrophoresis (HPCE) is an instrumental analytical technique where the electrophoretic separation is performed in narrow-bore capillaries or, as recently introduced, in microchannels of planar systems with an internal diameter between 10 and 100 µm.3, 4 The use of capillaries has numerous advantages, particularly with respect to detrimental effects of Joule heating. The high-electric resistance of the capillaries enables applications of high-electric fields (up to 30 kV) with only minimal heat generation, which is furthermore efficiently dissipated because of the large surface area-to-volume ratio of the capillary. The use of high-electric fields results in short

Injection Inlet

Figure 1

+

Detection −

+

EOF −

Outlet

Scheme of HPCE, normal polarity mode.

analysis times, high efficiency, and resolution. Moreover, a number of separation modes are available in CE to vary the selectivity, which makes the technique applicable to a wide range of analytes. The instrumentation of HPCE is uncomplicated (see the schematic drawing in Figure 1). Briefly, both ends of the narrow-bore fused silica capillary are immersed into reservoirs containing a buffer solution that also fills the capillary. The reservoirs also contain electrodes that provide electrical contact between the high-voltage power supply and the capillary. The sample is loaded onto the capillary by replacing one of the buffer reservoirs by a sample reservoir and applying external pressure (hydrodynamic injection) or an electric field (electrokinetic injection). After the injection, the reservoir is replaced, the electrical field is applied, and the separation starts. The detection is usually performed at the opposite end of the capillary (normal polarity mode). UV/vis detection is by far the most common detection technique in HPCE. Other techniques include fluorescence, amperometry, conductivity, and mass spectrometry. Modern HPCE instruments are fully automated and thereby allow easy operations and precise quantitative analyses. An important feature of HPCE is the so-called electroosmotic flow (EOF). The EOF is the bulk flow of liquid in the capillary, which causes movement of most species (regardless of their charge and hydrodynamic radius) in the same direction, and thereby allows a simultaneous separation of cations and anions. The EOF originates from the dissociation of the silanol groups of the capillary wall made from fused silica. The negatively charged capillary wall attracts ions of opposite charge and a solution double layer is formed. The applied electric field causes movement of the cations of the diffuse part of this double layer and attracts them to the cathode. Owing to the solvation of the cations, water molecules are dragged with them and cause the movement of the bulk solution—the EOF. The magnitude of the EOF (expressed by the mobility) is affected by a number of parameters such as the pH of the buffer, its ionic strength, temperature, and the presence of various additives. Suppression or even reversal of the EOF can be achieved through capillary wall modification, either permanent or dynamic. Permanent modification of the capillary wall is achieved by covalently bonded or physically adhered phases. The most common approach for these permanent

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions wall modifications is silylation followed by deactivation with suitable functional groups (polyacrylamide, polyethylene, glycol, or polysaccharides). Nowadays, different capillaries with a stable permanent coating are commercially available. However, a common uncoated capillary is employed in the dynamic modification approach, which is based on the addition of an appropriate modifier to the background electrolyte. This modifier interacts with the capillary wall and in this way affects the EOF. The potential disadvantage of this approach is the long-term stability and relatively long equilibration time needed to obtain a reproducible surface.

3

ACE

Applications of affinity measurements are divided into two groups according to the goal of the affinity measurement. Affinity interactions with the objective of enhancing separation selectivity serve mainly for the separation of mixtures of substances with very similar or identical electrophoretical behavior (e.g., chiral separation or MEKC—micellar electrokinetic chromatography—for the separation of neutral molecules). The analytes/receptors migrating in an electrical field undergo an interaction with the selectors dissolved in the background buffer, which changes their electrophoretical behavior. This effect can be used for (i) enhancing the separation selectivity or, as discussed here, (ii) simply for the identification and quantification of specific binding. Till now, a variety of CE-based methods for studying interactions have been established. Electrophoresis-based methods use various experimental approaches and are sometimes summarily termed ACE. However, no unifying definition of ACE exists in the literature and some authors refer ACE to one specific method for studying interactions. Throughout this chapter, however, the term ACE is used in its general meaning, which means all CE-based methods studying interactions. S mp

mp +L

3.1

Theory and modes of ACE

ACE, including affinity capillary electrochromatography (also termed electrokinetic affinity chromatography or capillary electroaffinity chromatography), with a variety of experimental approaches, is a well-established method for the study of binding interactions. Owing to the chemical and physicochemical nature of the interactions and formed complexes, various methods are available. At present, there are seven ACE modes developed for capillaries/channels, excluding partial-filling techniques. To avoid a confusion with the term ACE (used here for affinity measurements in general), ACE-µ (ACE used as mobility charge analysis) has been defined for one of these specific methods (Table 1). Mobility shift assay (ACE-µ, Figure 2(a) and Scheme 1) is the favorite method in HPCE for the investigation of simple 1 : 1 complexes. The mobility of the injected receptor (fixed amount) is monitored when the ligand is dissolved in buffer at varying concentrations. Both the ligand and the receptor can be injected as sample or added to the background buffer. In many cases, the detectability of the compound with the installed detection system decides whether ligand or receptor will be injected. The buffer additive should not decrease the sensitivity of sample detection. Here, we use the descriptors S for solute (injected sample; in the literature S is sometimes described as the receptor R) and L for ligand (solved in buffer solution). Note that the description is general, and S and L could be any type of molecule such as proteins, enzymes, and inorganic ions. According to the equilibrium S + L
SL, the shift of the solute mobility is modified by increasing Lconcentrations in the background buffer (Figure 2a). ACEµ is suitable for measuring weak to intermediate affinities but is especially advantageous for the evaluation of weak binding. A specific example is presented later.

mp·L S

(a)

∆mp + L ∆mp·L

3

mp

mp·L

(b)

Figure 2 The principle of (a) ACE-µ for weak binding and (b) for tight binding systems by evaluating the peak area (S, reference molecule; P (protein), solute; L, ligand). (Reproduced from Ref. 5.  Springer, 1998.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

4

Techniques

Table 1

Overview of ACE modes with measured parameters, conditions, and procedures of measurement.

Method

Sample

Buffer

Parameters

Requirements/comments

Affinity capillary electrophoresis (ACE-µ) as mobility shift

[solute = S]

[ligand = L]

µS , µS , µSL , K

Weak–intermediate interactions µS
= µ 1 : 1/1 : 2 interaction Approximation [L] = [L0 ] MEKC

Pseudostationary phase: For example, micelles, liposomes, dendrimers Mobile phase: For example, cyclodextrins, peptides Hummel–Dreyer method (HD) Vacancy affinity capillary electrophoresis (VACE)

Chiral separations [S]

[L]

AL , µSL or µL , Ki , ni

µS = µSL or µL = µSL

Empty buffer

[S] + [L]

µS , µS , µSL , KB , n

Weak–intermediate interactions µS
= µSL , 1 : 1 interaction Approximation [L] = [L0 ] Weak soluble ligands

Vacancy peak (VP) analysis

Empty buffer

[S] + [L]

AL , µS , µSL , KBi , ni

Preferably µS = µSL Weak–intermediate interactions Statistical interpretation

Frontal analysis (FA)

[S] + [L]

Empty

h, KBi , ni

µS = µSL and µS
= µL Intermediate–strong interactions Suitable for drug–protein studies

Frontal analysis continuous capillary electrophoresis (FACCE)

[S] + [L] continuous sampling

Empty

h, KBi , ni

µS = µSL and µS
= µL Lower detection limits Weak–intermediate interactions

Pre-equilibration CE

[S] + [L]

Empty

Strong interactions

Partial-filling ACE

[S]

[L]-plug

AS , AL , ASL , µS , µSL , KBi , ni µS /µreference

The Hummel–Dreyer (HD) method uses an identical experimental setup to ACE-µ. The capillary is filled with a buffer containing [L], to be studied at varying concentrations. When a small amount of sample is injected, a typical elution profile is monitored (Scheme 1). In contrast to ACE-µ, the peak area of [L] (A) is evaluated for the determination of the concentration of the bound ligand. A requirement for the application of the HD method is µS = µSL , so that the correct peak area of [L] is recorded. This approach offers the possibility of checking the proposed stoichiometry model. VACE (vacancy affinity capillary electrophoresis) and VP (vacancy peak) analysis also use an identical experimental setup. However, the electropherograms are treated differently. In VACE, the shift in migration time of negative peaks is evaluated to measure the extent of binding (analogous to ACE-µ). The capillary is filled with buffer containing [S] and [L]. The concentration of one interacting partner is fixed and the concentration of the other is varied. Measuring A as well as µ of negative peaks could

See ACE-µ

be demanding on microchips, because the reproducibility of the detection signal and the detection sensitivity is not sufficient for such an approach. The experimental setup for frontal analysis (FA) is quite different to the methods described above. The capillary is filled with buffer and a large sample plug of equilibrated [S] and [L]. As before, various cases have to be considered depending on the relation between the ionic mobility of [L], [S], and [SL]. Ideally, µS = µSL (e.g., protein molecules interacting with small ligands) and µL differs significantly from these. On applying voltage, the analyte zones (consisting of [L], [S], and [SL]) migrate according to their ionic mobility and form a typical elution profile containing two plateaus. One plateau is given by the concentration of free solute [S] and the complex [SL]; the second plateau is related to the free dissolved [L]. It is not surprising that the limit of detection of this method is significantly lower than by methods using narrower solute zones. FA is a method that has a potential for miniaturization.6

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions

the binding isotherm of such an interaction is commonly expressed by the following equation (see also (3)):

mS, eff = f ([L]buffer) mS mSL Injection S = constant

mS, eff = f ([L]buffer) mS mSL

[L]buffer + [S]buffer VACE

[SL] + [S] mSL = mS [L]buffer

S [L]buffer − [SL] A ~ [SL]

HD

[L]buffer + [S]buffer VP

[SL] + [S] mSL = µS

mL

[L]buffer − [SL] A ~ [SL]

[SL] + [S] [L]

h ~ [L]

S+L

FA

[SL] + [S] [L]

h ~ [L] FACCE

Schematic elution profiles of the ACE methods.

Analysis of binding using pre-equilibrated samples containing both interaction partners is only suitable for the characterization of strong interactions (Figure 2b). In addition, the complex formed also has to be stable in the electric field applied during the separation. Essential conditions for this application (pre-equilibrated samples without additives in the buffer) are as follows: (i) a large binding constant (e.g., hybridization of oligonucleotides, enzyme–substrate binding); (ii) a sufficient mobility difference between the substrate and the complex; and (iii) constant UV absorption of the peaks to be quantified, especially in the presence of more equilibrium reactions.

3.2

x=

[L]buffer ACE-m

mL

mL

Kj [L] [SL]
= nj cS 1 + Kj [L] m

mS, eff ~ [SL] mL mS, eff ~ [SL]

Scheme 1

5

j =1

where x is the molar fraction of bound ligand per solute or the concentration of ligand bound by 1 mol of solute; [SL] and [L] are the equilibrium concentrations of bound and free ligand, respectively; cS is the total (analytical) concentration of the solute; nj is the number of binding sites of class j ; and Kj is the corresponding association constant. The more the types of binding sites present on the solute, the more complicated is the calculation of binding parameters. Therefore, the common first approximation is a 1 : 1 association.7 The simplified form of (1) can then be linearized and the relevant association constant is thereby calculated. If there is a deviation from linearity observed using this simplification, multiple equilibria have to be considered and nonlinear models according to (1) should be used for the calculation of the binding parameters. As shown above, at present there are eight affinity electrophoresis modes developed for capillaries and channels. Most of these modes allow, in addition to the determination of the binding constant, the determination of the number of ligand molecules that bind to the different types of binding sites. Methods for the calculation of binding parameters can be divided into three groups according to the way of acquiring the parameters. The binding parameters can be extracted from (i) mobility changes (µ), (ii) peak area (A) of the species, or (iii) plateau (h) of the elution profile.8

3.2.1 Mobility shift assay (ACE-µ and VACE) Mobility shift assay is the favorite method in capillary zone electrophoresis for the investigation of simple 1 : 1 equilibria. The determination of the ionic mobility of the solute in ACE-µ is carried out under equilibrium conditions. By varying the ligand concentration, a shift in the ionic mobility (µ) starting from µS approaching µSL for the equilibrium S + L
SL is observed. Under the assumption of a simple 1 : 1 equilibrium, the mobility shift can be derived from the ratio between free dissolved [S] and the concentration of formed complex [SL] (2–8). For higher order equilibria, (8) must be extended to all complexes (differing in number of ligands) formed.

Mathematical description

In general, interactions are characterized by association constants K (frequently also called binding constants) and by the number of ligands that bind to the same class of binding sites present on the solute. Mathematically,

(1)

µ = xS µS + xSL µSL

(2)

The mole fraction of the species from (2) is defined by xS =

[S] [S] + [SL]

and

xSL =

[SL] [S] + [SL]

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

(3)

6

Techniques

Using the definition of the association constant K for a 1 : 1 complex and of the k’ (analogous to the capacity factor in (12)), the ratio of bound to free substrate molecules can be derived as follows: [SL] [S][L] [SL] = KB [L] k = [S]

KB =

(4) (5)

The measured mobility can, thus, be expressed by the following two equations:   [S] [S] µ + 1− µSL µ= [S] + [SL] S [S] + [SL] 1 [SL] = µ + µ 1 + [SL]/[S] S [S] + [SL] SL

(6)

KB [L] 1 µS + µ 1 + KB [L] 1 + KB [L] SL

(7)

µ=

Nonlinear regression analysis of a plot of µ against [L] (8) provides the association constant, and the electrophoretic mobility of the formed complex SL derived from µmax . [L] can be approximated in the equation by the total ligand concentration (under the condition that the ligand concentration is much higher than the concentration of the solute or that the association constant is small ). µ = f ([L]) =

µS + KB [L]µSL 1 + KB [L]

µ − µS = KB [L] = k  µSL − µ

or (8)

Instead of a nonlinear fitting, (8) is often linearized. The linearized plot is known as a Scatchard or Benesi–Hildebrand plot.9 In addition, the calculation of the free standard reaction enthalpy (R G) can be helpful if association constants are compared using different techniques. R G0 = −RT ln KB

(9)

The fitting of a more complex reaction, for example, for higher order equilibria, with more than one bound ligand by an equation valid for 1 : 1 fitting reflects neither the real interaction between solute and ligand nor their real effect on the net mobility of solute. Biochemical interactions involving proteins and oligonucleotides often exhibit a higher stoichiometry, particularly if the ligand molecules are small. In a few cases, with a complete titration curve (determination of µSL by a plateau) and a known charge for the ligand (e.g., metal ions), an estimation of the number of bound ligands can be made.

In contrast, if the charge of the ligand or solute molecule is unknown, then the shift of ionic mobility cannot be directly obtained from the absolute charge, and (8) has to be extended with consequences for the association constant. With an order higher than two and with a nearly anticooperative (higher order of ligand is often hindered due to steric reasons) or noncooperative binding behavior, fittings of the corresponding equation are almost impossible: µS + µSL KB1 [L] + µSL2 KB1 KB2 [L]2 + µSL3 KB1 KB2 KB3 [L]3 µ= 1 + KB1 [L] + KB1 KB2 [L]2 + KB1 KB2 KB3 [L]3

(10)

For a highly cooperative binding (the first binding step favors the following association and it can be assumed that the solute exists only in the form of the complex with the highest stoichiometry), the following equation can be used as a simplification of (10): µ=

µS + KB × [L]n × µSLn 1 + KB × [L]n

(11)

In this case, KB is the apparent overall association constant (overall stability constant). Bowser and Chen10 have discussed in detail the consistency of theory and experimental data. In their study, a 1 : 2 stoichiometry has been investigated and cooperative or anticooperative reaction types have been distinguished.

3.2.2 Peak-area-changes assays (HD, VP, pre-equilibrated CE) In free solution methods using peak-area-changes of the solute for the calculation of binding parameters, two different approaches are used, depending on the stability of the complex formed by the interaction. In the case of strong interactions, a direct separation of free and bound solute is possible and the method is commonly called pre-equilibrium CE.11 The solute is preequilibrated with different concentrations of the ligand and then injected into the channel filled with buffer. On applying high voltage, the free solute is separated from the free ligand and the complex. Peak areas are used for the determination of the equilibrium concentration of free and bound ligand and the association constant is calculated according to (1). Calibration measurements are required to relate the concentration to the peak area. This method is applicable to interactions yielding sufficiently stable complexes with slow dissociation kinetics. Preequilibrium electrophoresis on chips is preferentially used in immunoassays where the interaction is strong enough for a direct separation.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions In the second approach, the equilibrium is established during the separation similar to mobility shift assays. Peakarea-change assays based on this approach are applicable to weaker interactions with fast kinetics and have been established only in capillaries so far. Two methods are available, the HD method with the experimental setup identical to mobility shift assay and the VP method with the same setup as in VACE.12 The peak-area-changes of the ligand, which is added to the background electrolyte in varying concentration, are evaluated.

3.2.3 Elution-profile-changes assays (FA, FACCE) Methods evaluating elution-profile-change for the characterization of an equilibrium work with pre-equilibrated samples, which are injected in a large plug (FA13 ) or continuously (frontal analysis continuous capillary electrophoresis (FACCE)14 ) to the buffer. It is assumed that the mobility of the complex is close to the mobility of the solute and that the mobility of the ligand differs from it sufficiently. Free ligand leaks out from the sample plug because of its different mobility and makes its own plateau. The concentration of free ligand is extracted from the height of the free ligand plateau by means of calibration. Elution-profile-change assays have been employed only in capillaries so far.

3.3

Applications of affinity measurements for the study of interactions

In this section, various applications are introduced and explained in detail. The receptor or solute molecules (defined here as analyte that is injected as sample) differ not only in terms of their origin (biological and artificial molecules) but also in their abilities to bind to different kinds of ligands. Therefore, electrostatic and hydrophobic interactions are the main forces involved in reversible binding. In most cases, ACE-µ is applied and corresponding evaluation of association strengths are explained.

3.3.1 Pharmaceuticals micelles/metal ions MEKC is usually used as a separation technique in which the basic properties of micellar liquid chromatography and CE are combined. MEKC was first described by Terabe15 in 1984 for the separation of nonionic aromatic compounds and is a powerful separation technique for lipophilic and nonionic species. By addition of surfactants to the background electrolyte, new options for solving electrophoretic separation problems are opened, but it is also possible to apply this technique to study the affinities of drug molecules to surface-active compounds. The term micellar affinity capillary electrophoresis (MACE) is used

7

to describe the study of such interactions employing the same phenomena as present in MEKC. Of main interest is not the achievement of optimal separation or high detection sensitivity, but the study of the effect of the type of micelle forming compounds and its concentration (above the critical micelle concentration (CMC)) on the ionic mobility of the solutes/drug and, therefore, the partition behavior. This is affected by the reversible interactions of the solute with the micelle. MACE and ACE (affinity to nonmicellar buffer additives) are based on the same principle. In absorption studies, the application of colloidal systems, which show specific and unspecific interactions with mainly lipophilic substances is of main interest. An obvious application is the study of lipophilic and poorly absorbable drugs that are administered orally or transdermally.16 Such interactions with surface-active agents may either cause a diminution of the bioavailability by trapping the drug in the micelle or, on the other hand, lead to an improved solubility (prevention of the precipitation of the drug) and facilitated transfer of the solute across lipid membranes (e.g., the intestinal wall in the gastrointestinal tract) and therefore to an improved bioavailability.17, 18 The interactions between a lipophilic or hydrophilic drug and micellar phases are caused by weak physicochemical forces such as hydrophobic (unspecific) and electrostatic effects (specific: dipole –dipole, dipole-induced dipole) and steric effects, whereas the hydrophobic binding to the micellar systems is dominant. An indirect indication for the presence of interactions between the micellar phase and drugs is given by molecular and dynamic parameters of the drug and the micelles (ionic mobility, diffusion coefficient, hydrodynamic radius, apparent molecular mass), which are altered by the solubilization of lipophilic substances in a characteristic manner. A variety of very different methods may be employed for the characterization of solubilization and distribution equilibria between surface-active compounds (e.g., sodium dodecyl sulfate (SDS), dodecyldimethylammonium, palmitoylcholine) and drugs. 13 C and 1 H NMR studies may be used to gain structural information on the alteration of the micellar phase during a solubilization process.19–21 By following the Brownian motion of molecules, diffusion coefficients, as well as the hydrodynamic radii and charge of the micellar complexes, can be obtained. The molecular weight of colloidal particles can be obtained by exploiting scattering effects such as extended light scattering.22 In one of the few publications dealing with equilibrium constants and distribution coefficients, it was demonstrated that electronic absorption spectroscopy and fluorescence measurements are suitable for the determination of the association constants of drugs/solutes to the micellar phases.23–25 With micellar affinity electrophoresis, the distribution behavior of drugs may be studied in an easy way simply

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

8

Techniques

0.0002

m (cm2 V−1 s−1)

0.0001

Etliefrine HCL

0.0000

Chloramphenicol

−0.0001

Propranolole HCl Ibuprofen Na

−0.0002 −0.0003 −0.0004

Salicalic acid 0

5

10

15 20 c bile acid (mM)

25

30

35

Figure 3 Effective electrophoretic ionic mobilities µ of different drugs influenced by bile acid concentration (buffer: 0–30 mM glycodeoxycholic acid, Na), 20 mM phosphate, pH 7.4, detection 220 nm). Note, here and in the following applications µ means the effective ionic mobility with consideration of the EOF.

by observing the effect of the micellar composition and concentration on the ionic mobility of the analyte. By examining the alteration of the electrophoretic properties of the solubilized drugs, a more precise characterization is then possible by studying the micellar phase itself or the complete system.26, 27 This is of particular interest in the early stages of bioavailability studies for the acquisition of important parameters to characterize the interactions between drug and different micellar partners. In the following section, the interaction of selected artificial drugs with bile acids is described. Bile salts or mixtures thereof have been used in drug formulations because of their physiological acceptance. This is a difference to MEKC where mostly more common amphiphilic substances (e.g., SDS) are employed. Bile salts are naturally present in the duodenum and jejunum mainly in the form of glycine and taurine derivatives. Their physiological importance lies in their ability to lower the surface tension of water, to emulsify fats, and thus to promote enzymatic attack. These properties make these compounds suitable for the improvement of the solubility of lipophilic drugs. MACE has so far been used to study the effect of the lipophilicity, polarity, and charge of pure and mixed (incorporating phospholipids and fatty acids) bile salt micelles not only on a range of lipophilic drugs but also on hydrophilic drugs.28–30 Different di- and trihydroxy bile salts and drugs that differ in lipophilicity, basicity, and structure have been compared in order to examine solvatochromic equilibria.30, 31 The principle exploited in the determination of thermodynamic equilibrium constants is the indirect measurement of the capacity factor affected by the tenside concentration (in this case, the bile acid concentration). A pronounced shift in the migration times and thus effective ionic mobility

for some drugs is observed because of their intermediate affinity to the bile salt phase (Figure 3). The measured ionic mobility of the drugs, µ, is a function of the ionic mobility of the solute on its own, the strength of the interaction with the buffer components as well as their mobility in the electric field (12). The maximum shift of µ is limited by the mobility of the ligand in the background, in this case by the mobility of the bile salt micelle. Strongly lipophilic drugs such as propranolole are shifted from the cationic to the anionic side. Etilefrine as representatives of hydrophilic drugs display only a relatively small change in mobility, despite the fact that etilefrine is a cation. It has been shown9 that besides the type of bile salt (lipophilicity, aggregation number of micelles), the pH value and the addition of metal ions may also have a strong effect on the equilibrium (Figure 4). As expected with increasing bile acid concentration in the separation buffer, the ionic mobility is shifted to more negative values due to increasing partition into the anionic micelles. An interesting phenomenon is observed when Ca 2+ is also present in the CE buffer (Figure 4, the triangles). Contrary to the assumption that by the addition of Ca2+ the measured ionic mobility of the solute, partially distributed in the bile acid micelle, will be shifted in the direction of the cationic side, a further increase of the ionic mobility is observed. This effect can be explained only by the formation of ternary systems in which the solute is present in a higher concentration than in the simple bile acid micelles. If there are no mathematical models suitable for the estimation of binding constants or partition coefficients, the changes in the ratio of free dissolved solute as well as the effects on the micellar phase can be estimated. On the other hand, the reason for the increasing interactions at lower pH values is the higher aggregation of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions

9

Decrease of free dissolved drug

Effective ionic mobility (cm2 V−1 s−1)

0.0002 0.0001 D2

0.0000 −0.0001

D1

−0.0002 −0.0003 + 1 mM Ca2+ solved in the CZE buffer

−0.0004

D1 D2

−0.0005 0

5

10

15

20

TDCA (mM)

Figure 4 Effective ionic mobility of drugs D1 and D2 in dependence on the concentration of the bile acid taurodeoxycholic acid (TDCA), 10 mM sample; sigmoid fit. In the presence of Ca2+ , the ionic mobility is shifted to higher negative values.

the bile salt micelles (formation of secondary micelles with helix-like structures) and the increased potential difference between the micelle and the cationic drugs (electrostatic interactions are influenced). Mathematical description Instead of using (8), the following equation has to be applied for the calculation of the partition coefficient KP (for explanation of the symbols see Nomenclature section, µ corresponds to the measured ionic mobility of the drug). By fitting the experimentally obtained data (µ), KP can be calculated with the knowledge of the mobility of the solute (drug) µS and the concentration of the surfactant (also termed here as L). The partial molar volume ν of the micelle, the CMC, if unknown, and the mobility of micelle can be set as variables, calculated during the fitting process. kP =

µS − µ ν(cL − CMC) = KP µ − µmc 1 − ν(cL − CMC)

nS,mc cS,mc VS,mc = nS,aq cS,aq VS,aq cS,mc KP : partition coefficient : KP = cS,aq

kP : capacity factor : kP =

(12) (13) (14)

For (12), it is assumed that the volume of the micellar phase is proportional to the tenside concentration and that the partial molar volume ν remains constant. It is also assumed that the ionic mobility of the micellar phase does not change on taking up a solute (µmc = constant). In contrast to highperformance liquid chromatography (HPLC), substances which have an infinitely high kP value, that is, which are completely dissolved in the micellar phase, can be detected. In this case, the sample molecule migrates with the mobility of the micelle.

For the illustrated drugs (solute), KP is about 10 for propranolole and nearly 0 for chloramphenicol and ibuprofen; the partial molar volume ν of the micelle has been calculated as 0.02 l mmol−1 at a pH of 7.4. Note that for salicylic acid only a slight change of the ionic mobility is visible due to its very similar mobility compared to that of the micelle itself (Figure 3).

3.3.2 DNA–metal ions Studies of the interactions between small molecules and nucleotides, oligonucleotides or ss (single-stranded)-DNA are important to elucidate the functional mechanisms of DNA and are critical for understanding the control of expression of the phenotype from the genotype. In recent years, great efforts have been made to develop and evaluate analytical methods to investigate and describe interactions between DNA or DNA fragments32–37 and small molecules or ions. Even though the most important interactions with mono- and divalent cations including transition ions have been studied, the investigations and the resulting conclusions are nonsystematic and remain somewhat controversial. Furthermore, it is often difficult to correlate structures of metal ion–DNA or metal ion–oligonucleotide complexes measured in the solid state with investigations in aqueous solution in an equilibrated state. Additional complications arise from measurements carried out in inappropriate chemical environments. In particular, buffer components such as TRIS (tris-(hydroxymethyl)aminomethane) or phosphate that are known to coordinate to metal ions have often been used. There are only a few electrophoretic studies describing the characterization and quantification of DNA–metal ion interactions. Apparent equilibrium constants have been

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

10

Techniques

determined for the interactions of Ag+ , Mg2+ , Ca2+ , and Fe2+ /Fe3+ , and double-stranded DNA molecules (calfthymus DNA).38 With the exception of investigations with Fe2+ or Fe3+ cations in all of these studies, the concentration (UV-absorbance) of the formed DNA–metal ion complex has been measured. However, the evaluation of peak area or height of the formed complex can easily lead to inaccurate results. The UV absorption spectra of the metal ion–DNA complexes change with progressive coordination of metal ions and the binding between the metal ions and the oligonucleotide is not strong enough for the migration of nondissociated metal complexes in an electric field. In the following section, the application of the mobility shift assay is described to show the possibility to characterize the affinity properties between various oligonucleotides interacting with different kinds of metal ions.39, 40 In the case of metal ion–oligonucleotide interactions, the reaction kinetics are fast, binding is moderate, and the ionic mobility of the metal-complex is significantly different to the mobilities of the oligonucleotides. To describe the interactions of oligonucleotides with metal cations, theoretically it has to be assumed that binding of higher orders, as well as the formation of ternary systems, can occur. The reaction scheme (Scheme 2) has to be considered when the mobility shift function is discussed. Needless to say, such complex systems cannot be fully described mathematically. However, a scheme of possible interactions may be used to explain unexpected phenomena in the migration behavior in the presence of all interacting partners. The oligonucleotide-free parent solution of metal salt in buffer (BUFF) will contain aquated metal ions and metal buffer complexes M (BUFF) with the total metal ion concentration [M]tot = [M(aq)] + [M(BUFF)] (Scheme 2, (DNA)(BUFF) + BUFF (2)

+ M (4)

DNA

(DNA)(BUFF)(M) + BUFF (3)

+ M (1) DNA K M(DNA)

(DNA)(M)

K

M(DNA) M(DNA)(BUFF)

+ BUFF (5)

M

M(BUFF)

(DNA)(BUFF)(M)

+ DNA (6)

+ (DNA)(BUFF) (7)

Scheme 2 Equilibria between the analyte (DNA (S)) with two different ligands (BUFF (L) and M (L)).

(5)). Also DNA–buffer interactions (2) must be taken into account for the interpretation of the measured mobility shift by forming ternary complexes (4). As Stellwagen et al.41 have shown, buffers used to maintain neutral physiological pH values such as TRIS and zwitterionic “Good” buffers interact with DNA oligonucleotides. On addition of the oligonucleotide (DNA) to the metal-containing buffer, the complexes M (DNA) and M (BUFF)(DNA) will be formed. For more detailed insight into the calculation of overall binding constants, the authors recommend the literature reported in Ref. 39. Two different potential binding sites exist for metal ions at the oligonucleotides: the phosphate group and the N7 of the purine bases adenine and guanine (Figure 5). The higher electronegativity of the N7 of guanine is responsible for stronger interactions with cations than other nitrogen donors within the heterocycles. Hard cations prefer to bind to the phosphate group of the backbone, while softer cations preferentially interact with the nitrogen donor of the purine bases. Binding of the metal ions may be direct or indirect through water molecules. Systematic studies have indicated that the choice of the buffer solution is critical and that the commonly used TRIS and 3-(N-morpholino)propanesulfonic acid (MOPS) buffers bind transition metal ions by introducing additional equilibria into the solution phase (Scheme 2). The observed binding affinities (slope of µ) of oligonucleotides for group 2 and transition metal ions may be rationalized in terms of a two-site binding model involving phosphate and nitrogen donors in the nucleotide and quantified using limiting mobility values (maximum shift of µ). Findings of various studies are that the binding strength and stoichiometry are affected not only by the kind of metal ion (Figure 6) but also by the sequence of the nucleotides and the 3 -end bases (intramolecular interactions lead to the formation of macrochelates, Figure 7). Furthermore, it can be deduced that hard and soft cations have different preferences for interacting with the different binding sites present in the oligonucleotide. Much more interesting is the finding of higher aggregates (formation of a quadruplex in the presence of transition metal ions) visible by a typical ionic mobility of the formed complex.39 Mathematical description For a simple 1 : 1 interaction as is the case for Mg2+ and Ca2+ , (8) is sufficient to calculate the binding constant or, in the case of a 1 : n interaction, (11) can be applied (n will be calculated during the fitting process). However, for Ni2+ a mathematical model describing an anticooperative behavior in which two Ni2+ ions bind to one oligonucleotide, the following equation (15) seems to be more promising to describe the behavior of changed ionic mobility during

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions O

O

N HO H H O

O

O

H

P

O−

H

H

N

NH

N

O

N

O H H

O H H

O

H

O

NH2

c

H

H

H

P

O−

O b

O

H O

H

O

H

H

N

O−

P O

N

7

NH2

H

N

O

H

N

P O− O

OH

H

N3 NH2

O H

O

O P a O

N

N1

9

H H

N

O

O H H

H

N

H

OH

H

H

3′-end

NH2

H

N

O H

N

H

NH2

NH

N O

H

O H

N

H

N

O

H

O

N

P O−

O

O

NH2

H

O

O

N HO

O

O

NH 3

NH

5′-end

11

H

OH

H

H

K(Ni, TGCA)>K(Ni, TGAC)

Figure 5

Chemical structure of tetramers d(TGCA) and d(TGAC). Ni2+

Ionic mobility of TTT G (cm2 V−1 s−1)

TTTG −1.5 × 10−4

−2.0 × 10−4

Mg2+ Ca2+

−2.5 × 10−4

−3.0 × 10−4 0

2

4

6

8

10 × 10−3

Concentration of cations (M)

Figure 6 Ionic mobility shifts of the tetranucleotide TTTG affected by various metal ion concentration of Mg2+ , Ca2+ , and Ni2+ . In the presence of Ni2+ , the mobility reaches a plateau value faster than that with Ca2+ or Mg2+ and that the final mobility is about twice as high with Ni2+ as with Ca2+ or Mg2+ . This is compatible with (i) the formation of nickel complexes that are more stable and (ii) the binding of two Ni2+ ions. The determined stability data confirm that the KB1 for Ni2+ is about an order of magnitude greater than for Ca2+ or Mg2+ . The KB values for Ni2+ lead to an order of stability G  T ≈ C > A for the variable base at the 3 -end.

complexation. µ=

µS + µSL KB1 [L] + µSL2 KB1 KB2 [L]2 1 + µSL KB1 [L] + µSL2 KB1 KB2 [L]2

(15)

3.3.3 Artificial receptors–peptides It is not surprising that the development of synthetic molecules which can interact noncovalently with biological

systems is of great interest for the fundamental understanding of molecular interactions. While the development of small synthetic molecules that bind with high affinity to biomacromolecules (proteins, DNA, oligonucleotides) has been successful, the development of synthetic receptors interacting with small biologically relevant molecules in water with similar high affinities has proven to be more difficult.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

Techniques

Ionic mobility of the oligonucleotides (cm2 V−1 s−1)

12

−1.0× 10−4

TCAG TACG TGAC TGCA

−1.5× 10−4

−2.0× 10−4 CCCC −2.5× 10−4

−3.0× 10−4

−3.5 ×10−4 0

2

4

6 2+

c of Ni

8

10 × 10−3

(M)

Figure 7 Ionic mobility shifts of various tetranucleotides in dependence on the Ni2+ concentration. For the tetranucleotides with a terminal 3 -G, the sequence specificity for Ni2+ binding is CG-3 > TG-3 > AG-3 and that TTGG-3 > TGGT-3 > GGTT-3 as examples.40 As expected, the dominant influence is from a 3 -terminal G base, consistent with the formation of a macrochelate.

In the following section, the determination of binding affinities between a water-soluble synthetic diketopiperazine receptor and a biomolecular counterpart is described. The binding affinities were measured by microchip-ACE-µ and compared with results provided by isothermal titration calorimetry (ITC). Diketopiperazine receptors consist of a diketopiperazine serving as a rigid, structure-directing template and of two peptidic side chains: the “receptor arms.” While two-armed receptors bind tripeptides (arginine-rich peptides) with high sequence selectivity and binding affinities between R G = −6 and −4 kcal mol−1 in water, the strength of binding between a one-armed receptor and arginine-rich peptides is significantly lower. These results have been established with ACE-µ by the evaluation of the ionic mobility of the receptor molecule, which is dependent on the concentration of the peptides (Figure 8), and also with ITC where the reaction enthalpy of every titration point was measured.42 In contrast to ACE-µ measurements on capillary with separation times of about 15 min, the time required for 1 measurement on the chip is 20 s. A TRIS/boric acid buffer and salicylic acid and ibuprofen as inert reference molecules have provided ideal conditions. With increased peptide concentrations in the running buffer, the effective ionic mobility of the analyte molecule is shifted to lower values. This reflects the change of the net receptor charge due to counterion binding by the positively charged peptide molecule. The starting point of effective ionic mobility that is identical to the intrinsic effective ionic mobility of the receptor is different for the two receptors.

In the case of the two-armed receptor, the calculated binding constants provided by the two independent methods are in fair agreement (for KB in the range of 500–2000 M−1 with a Gibbs free energy R G of about −5 to −10 kcal mol−1 ).42 Both methods have shown that an RRR (for syntax, see Figure 9) peptide binds stronger than RRS or RSR. In contrast, the ionic mobility of the onearmed receptor is influenced by the peptides dissolved in the buffer, but the reaction enthalpies in the calorimetric experiments are too small for a reliable analysis. Two facts could be responsible for the failure. First, the R G is small; second, the reaction is entropy driven (−R G ≈ T R S, large R S at a small R H ). Further ACE-µ investigations revealed that interactions between the one-armed receptor and serine containing peptides are entropy driven. Less conclusive is the number of bound ligands determined by ACE-µ. The possibility of 1 : 2 interactions could not be excluded (Figure 9).

3.3.4 Protein–protein/drugs The interactions of proteins with various ligands are important processes in living organisms.43, 44 On the one hand, these interactions are part of the endogenous regulation pathways. On the other hand, they are often the target of drugs as extraneous substances to fight pathological symptoms interfering with proteins in different ways. The analysis of protein–ligand interactions is therefore a preferred subject of biochemical research. A variety of studied systems will be described in this section.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

Ionic mobility of the receptors (cm2 V−1 s−1)

ACE—Tool to characterize intermolecular interactions

−0.00028

1:1 1:2

One-armed receptor dRG = −2.90 kcal mol−1

−0.00030

13

1:1 −0.00032

1:2 Two-armed receptor dRG = −3.97 kcal mol−1

−0.00034 −0.00036 −0.00038 −0.00040 −0.00042

0.0

0.5

1.0 1.5 c Tripeptide RRS (mM)

2.0

Figure 8 Altered behavior of the ionic mobility with increasing concentrations of RSR (for syntax, see Figure 9) in the running buffer and fitted curves with a 1 : 1 and 1 : 2 stoichiometry. Two-armed receptor Peptide

(a)

Peptide One-armed receptor (b)

Figure 9 Scheme of a noncovalent interaction between peptides and (a) a two-armed receptor with a 1 : 1 stoichiometry and (b) a one-armed receptor with a 1 : 2 stoichiometry. The peptides were the following: RRR: Ac-Arg-Arg-Arg-NHPr, RRS: Ac-Arg-ArgSer-NHPr, and RSR: Ac-Arg-Ser-Arg-NHPr.

Protein–protein interactions are characterized almost exclusively by noncovalent binding of the partners, for example, by electrostatic and hydrophobic interactions. Noncovalent binding has an important property: reversibility. Binding and dissociation result in the pathway regulation switching processes on and off, like signaling processes upon binding of the target to its specific receptor.45 During the last few years, the entire palette of ACE methodology has been applied to study protein–ligand interactions. This section mainly deals with proteins as binding partners of other proteins. However, some aspects of other types of ligands important for the discussion of the investigation of protein–ligand interactions in regulation pathways have to be provided as well. The interactions of proteins with (small molecule) drugs are not discussed here but have been nicely summarized in recently published reviews.46–50

The binding of proteins to oligonucleotides is of great interest due to their role in DNA replication and gene expression.51–53 ACE was proven as a useful tool for the investigation of these interactions in vivo.50, 54, 55 The experimental setups use ACE-µ56, 57 in the case of lowaffinity systems, and pre-equilibration CE58 in the case of high-affinity systems such as antibody–DNA interactions. Nucleotides under physiological conditions yield a charge difference from the complexed to the uncomplexed form. This effect mostly enables a good separation of the complex from its forming species (pre-equilibration CE) or a measurable shift in mobility of the protein on complexation (ACE-µ). Aptamers are a special type of oligonucleotides that bind to protein targets with high affinity.59 The aptamer approach has been used to provide specific binding partners for targets60 and also as therapeutic agents.61 Aptamers may also act as a platform for coupled functional groups like anticancer drugs directed against a specific cell type.62 The binding constants of aptamers to their targets are mostly in the range of 10−9 M. On one hand, this requires the use of sensitive detection methods in ACE. As a general rule, the applied concentrations of analyte and ligand should be in the same range as the binding constant. In CE, the needed sensitivity may be obtained by the application of laser-induced fluorescence (LIF) detection. This means in most cases one reactant (either analyte or ligand) has to be labeled with a fluorescent tag since the commercially available analytical systems for LIF do not provide the use of native fluorescence of proteins. However, DNA and aptamers can be labeled easily and reliably by different chemical approaches enabling the detection in the required concentration range. On the other hand, high binding constants provide the chance to use the approach of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

Techniques

pre-equilibrated samples (analyzing free and bound species as separated peaks). Recently, the ACE analysis of thrombin was described. Obubufano et al. described the application of microchip ACE in the field55 exploiting the high-affinity binding between an aptamer and thrombin for both the detection of thrombin at low concentrations in blood as complex matrix and the test of substances that interfere with the interaction (competitive assay). Owing to the fast separations on microchips, the use of the approach for high throughput is feasible, for example, in screening of potential drugs interfering with the investigated interaction reaction. Huang et al. described a similar approach in the capillary: the description of the thrombin–antithrombin interaction by means of an aptamer probe63 providing low detection limits for thrombin and antithrombin, respectively, and real-time monitoring of the interaction reaction. In this case, the aptamer acts as a molecular sample that enables the labeling of another molecule and therewith the monitoring of the further binding reaction. Another well-established field in ACE is the analysis of interactions with antibodies. Owing to the binding constants of complexes with antibodies (range: 10−9 M), the approach of pre-equilibrated samples was applied again. Several examples of antigen–antibody interactions using ACE were reported. These investigations focus rather on the measurement of antigen concentrations than on the estimation of binding constants. Human serum albumin,64 bovine serum albumin,65 green fluorescent protein,66 human growth hormone,67, 68 and insulin69 served as antigens. In this context, further competitive70–73 and noncompetitive assays74–77 applying the principle of ACE were described. Utilizing the interaction of antibodies, the use of ACE even for the analysis of more complex systems like the binding to entire viruses is possible.78, 79 Okun et al. performed binding studies of neutralizing antibodies to the common cold virus.80 Although viruses are large heterogeneous analytes, a separation of free virus and virus–antibody complex was achieved (Figure 10). Furthermore, information about the stoichiometry and the stability of the antibody–virus complex for different viruses and antibodies was obtained. The stoichiometry of virus–receptor binding was determined using the changed mobility of the virus on incubation with increasing concentrations of receptor fragments.81 Depending on different receptor fragments, changes in the stoichiometry were observed. Having a suitable detection system present for the protein–protein interaction under investigation is also very important in ACE. An interesting example was discussed by Pedersen et al.82 Pre-equilibrated samples were used in order to demonstrate the binding of free Gc-globulin

HRV2 IS

0 µM

20

0.06 µM

Absorbance, 205 nm

14

0.21 µM

0.62 µM 0 1.53 µM

Free mAb 8F5

−20

3.06 µM 2

4

6

8

10

Migration time (min)

Figure 10 Formation of complexes between human rhinovirus HRV2 and neutralizing monoclonal antibody mAb 8F5 analyzed by CE. A fixed concentration of HRV2 (15 nM) was incubated at room temperature with an increasing concentration of mAb 8F5 prior to the CE analysis. o-Phthalic acid was used as internal standard (IS). (Reproduced from Ref. 80.  American Chemical Society, 2000.)

and the Gc-globulin/G-actin complex. A separation of the Gc-globulin isoforms is obtained both with and without complexation with G-actin demonstrating equal contributions of all isoforms to the interaction with G-actin (Figure 11). First UV detection was applied in the study. However, owing to its low sensitivity, high concentrations of the binding partners had to be used (e.g., in Figure 11: 27 µM of Gc-globulin with increasing amounts of G-actin). The applied concentrations are much higher than the provided binding constant that is in the range of 10−9 M. The obtained binding isotherm was not evaluable although the interaction itself was proven for certain. The application of much lower protein concentrations required a more sensitive detection system. Fluorescence detection was enabled by labeling the Gc-globulin covalently with carboxyfluorescein. In the case of labeling, the influence of the reaction on the protein function as well as on the obtained binding constants and complex stoichiometry must be analyzed critically due to the introduction of the fluorescence tag to the molecule. Significant influences regarding distribution of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions

2.4 × 105

15

Gc-globulin

2.3 × 105 M

2.2 × 105 2.1 × 105 2.0 × 105

A

1.9 × 10

5

1.8 × 105 1.7 × 105 B

1.6 × 105

A 200 nm (a.u.)

1.5 × 105 1.4 × 105 1.3 × 105 C

1.2 × 105 Gc-globulin – actin complex

1.1 × 10

5

1.0 × 105 9.0 × 104 D

8.0 × 104 7.0 × 104 6.0 × 104 5.0 × 104

E

4.0 × 104 3.0 × 104 Actin

2.0 × 104 1.0 × 104

F

0 300

400

500

600

700

800

900

Time (s)

Figure 11 Electropherograms showing the interaction between Gc-globulin and G-actin using UV detection and pre-equilibrium CE. The samples were pressure injected with tinj = 5 s. The concentration of Gc-globulin was kept constant at 27 mM, whereas the concentration of G-actin was varied between 4.5 and 54 mM. The Gc-globulin–G-actin molar ratios are as follows: (A) 1 : 0.17, (B) 1 : 0.22, (C) 1 : 0.33, (D) 1 : 0.67, (E) 1 : 1, (F) 1 : 2. (Reproduced from Ref. 82.  Wiley-VCH, 2008.)

attached tags, adsorption behavior, and molecular functionality have been reported depending on the labeling chemistry.83–85 However, a prediction of the influence is often difficult. Hence, a case-by-case evaluation of the application is required. In the present example, only a marginal influence on the separation was detected. The obtained binding data fit nicely with results from orthogonal methods using both labeled and unlabeled Gc-globulin supporting further only a minimal influence of the fluorescence labeling on the obtained results. Protein–protein interactions with rather weak to moderate binding, for example, leading to slight conformational changes, are much harder to detect by analytical methods, because they are more likely to be disturbed by the examining method itself. In that respect, ACE offers the advantage of maintaining nearly native conditions. These conditions

may be tuned by the composition and conditions of the background electrolyte. For high-affinity interactions using pre-equilibrated samples, these advantages are applicable without restrictions.86 However, the analysis of weak to moderate protein–protein interactions by ACE-µ is mainly hampered by the difficulties arising from high protein contents in the background electrolyte. On one hand, the protein (acting as ligand) is necessary in the background electrolyte in increasing amount in order to perform the ACE-µ experiment obtaining a mobility shift for the protein (analyte) under investigation. In discussing weak to moderate interactions, binding constants in the range 10−3 –10−6 M are assumed. On the other hand, protein concentrations in the same range dissolved in the background electrolyte may introduce tremendous changes to the CE separation system causing very often extreme losses in separation efficiency

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

16

Techniques

or until total breakdown of the separation. Hence, there are only a very few reports dealing with ACE investigations of weak to moderate protein–protein interactions. There are approaches to circumvent the discussed problems of ACE-µ for weak-to-moderate protein–protein interactions in case the addition of the protein ligand to the background electrolyte is not feasible. One way might be the combination of both approaches of ACE: measuring the mobility shift of the analyte protein using pre-equilibrated samples. ACE was used to analyze the binding of HIV proteins to possible target proteins. ACE using equilibriummixture analysis was able to detect the protein–protein interaction by changes in the mobility.87, 88 Subsequent studies revealed that some viral proteins share a binding motif with interferon type I, which was used to explain the regulatory effect on the proliferation of lymphocytes. ACE was employed to demonstrate the binding of interferon type I to the cellular receptor protein of the viral protein.89 The same approach was successfully applied to the interaction of HIV capsid protein p24 with a cyclophilin fusion protein, although the dissociation constant was in the range of 10−5 M (Figure 12). The mobility shift of the fusion protein with increasing amounts of p24 in the pre-equilibrated sample was used to determine the dissociation constant.90 The other way is to mimic the binding of the protein ligand by using a binding peptide instead of the entire protein. The binding peptide describes the binding domain of the ligand that is involved in the protein–protein interaction. This approach was successfully applied to

Absorbance (200 nm) (m/a.u.)

EOF

1

C

•

B

•

12 8 4

A

•

0 2

3

4

5 t M (min)

6

7

CONCLUSION

D

2 IS

4

While the applicability of ACE in professional circles is still challenged by technical limitations (commercialization of suitable CE systems, powerful detection systems, and intelligent analysis software) and by diffusion processes, it is now used quite broadly in analytical chemistry and biological sciences. As is apparent from above, any kind of interaction partner can be investigated; however, the following requirements have to be fulfilled:

20 16

the binding of protein ligands to cyclophilins, which are members of the enzyme family of peptidyl-prolyl cis/trans isomerases.91 Here, the potential binding sites of the protein ligand were identified by screening a cellulosebound peptide library. Afterward, the identified peptides were synthesized and characterized by ACE-µ by their binding constants. It is clear that binding constants obtained by this approach provide only an estimated value since analyzing the binding peptide instead of the entire ligand protein can simulate the binding only under similar, but not, identical conditions. However, it may provide information for protein–protein interactions that are not accessible by ACE-µ. It has to be noted that the approaches for weak-tomoderate protein–protein interactions may be used for competitive assays as well. Here, information about the position of interaction as well as about the degree of influence on the interaction by competing agents is available. For the well-characterized model system cyclophilin—p24, the binding constant is not evaluable anymore if cyclophilin is treated with its specific inhibitor cyclosporine A. Hence, ACE was able to demonstrate that p24 uses the same binding site as cyclosporine A at the cyclophilin molecule and that the binding of cyclosporine A is much stronger than the binding of p24.91

8

Figure 12 Monitoring the protein–protein interaction of p24 (1) and rDmCyp20-GFP (2) using ACE. Samples of rDmCyp20GFP (6 µM) were incubated with increasing concentrations of p24: (A) 0 µM p24, (B) 6 µM p24, (C) 12.5 µM p24, and (D) 22 µM p24. EOF indicates the position of the electroosmotic flow marker DMSO. The position of the internal standard Ac-Ala-Ala is marked with IS. (Reproduced from Ref. 90.  Wiley-VCH, 2001.)

• • • • •

no interactions between the channel surface and the molecules under observation; low conductivity of the molecules (buffer as well as analyte); no interference of the background buffer with the detection signal; sufficient limit of detection of the solute for its detection at suitable concentrations of ligands; significant differences between the ionic mobility of the ligand and the substrate (needed for mobility shift assay); detectability of either the ligand or substrate; solubility in the CE buffer system (with and without ligand); and at least one binding partner has to be charged.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

ACE—Tool to characterize intermolecular interactions The methodical limitations concern analyte recovery (in the case of peak-area-changes assays), peak identification, and in understanding the interplay between binding rates and separation parameters. Perhaps the most limiting factor for using the technique outside specialist laboratories is the fact that ACE is not one, but a suite of different techniques united by a capillary electrophoretic separation step. Furthermore, in the case of unknown stoichiometry, only absolute dimensions of the binding constants are accessible. ACE continues to be a versatile technique to assess the extent of interactions between a myriad of species including proteins, DNA, sugars, antibiotics, and other biological compounds. Recent developments in the use of ACE have expanded and developed new ACE methodologies that have broadened the tools available to the researcher working in the area of molecular recognition. Some of these techniques are of the high-throughput type and require even less quantity of material than that was previously necessary in earlier ACE studies. The future is bright for ACE. For example, recent work detailing the integration of ACE onto a microfluidic format92 is exciting and promises to be an area much researched in the future.

17

6. T. Le Saux, H. Hisamoto, and S. Terabe, J. Chromatogr., A, 2006, 1104, 352. 7. Y. Tanaka and S. Terabe, J. Chromatogr., B, 2002, 768, 81. 8. M. H. Busch, L. B. Carels, H. F. M. Boelens, et al. J. Chromatogr., A, 1997, 777, 311. 9. R. H. Neubert and H. H. R¨uttinger, Affinity Capillary Electrophoresis in Pharmaceutics and Biopharmaceutics, Marcel Dekker, Inc., New York, Basel, 2003. 10. M. T. Bowser and D. D. Y. Chen, Anal. Chem., 1998, 70, 3261. 11. K. L. Rundlett and D. W. Armstrong, Electrophoresis, 2001, 22, 1419. 12. M. H. Busch, J. C. Kraak, and H. Poppe, J. Chromatogr., A, 1997, 777, 329. 13. J. Ostergaard and N. H. Heegard, Electrophoresis, 2003, 24, 2903. 14. J. Y. Gao, P. L. Dubin, and B. B. Muhoberac, Anal. Chem., 1997, 69, 2945. 15. S. Terabe, K. Otsuka, K. Ichikawa, et al. Anal. Chem., 1984, 56, 111. 16. A. J. Hoogstraate, P. W. Wertz, C. A. Squiere, et al. Eur. J. Pharm. Sci., 1997, 5, 189. 17. M. A. Hammad and B. W. Mueller, Eur. J. Pharm. Biopharm., 1998, 46, 361. 18. D. Khossravi, J. Int. Pharm., 1997, 155, 179.

NOMENCLATURE KB KP kP µeff

association/binding constant partition coefficient capacity factor effective mobility

19. M. G. Casarotto and D. J. Craik, J. Phys. Chem., 1992, 96, 3146. 20. C. P. Borges, S. Honda, H. Imasato, et al. Spectrochim. Acta, Part A, 1995, 51, 2575. 21. V. E. Yushmanov, J. R. Perussi, H. Imasato, and M. Tabak, Biochim. Biophys. Acta-Biomembrans, 1994, 1189, 74. 22. M. Janich, J. Graener, and R. Neubert, J. Phys. Chem. B, 1998, 102, 5957.

Indexes

23. W. Caetano and M. Tabak, J. Colloid Interface Sci., 2000, 225, 69.

D L Mc S

24. P. M. Nassar, L. E. Almeida, and M. Tabak, Biochim. Biophys. Acta-Biomembrans, 1997, 1328, 140.

drug ligand micelle solute

25. C. P. F. Borges, I. E. Borissevitch, and M. Tabak, J. Lumin., 1995, 65, 105. 26. A. Fuertoes-Matei, J. Li, and K. C. Waldron, J. Chromatogr., B, 1997, 695, 39.

REFERENCES

27. P. G. Muijselaar, H. A. Claessens, and C. A. Cramers, J. Chromatogr., A, 1997, 765, 295.

1. N. H. Heegaard, M. H. Nissen, and D. D. Y. Chen, Electrophoresis, 2002, 23, 815.

28. M. A. Schwarz, R. H. Neubert, and H. H. Ruettinger, J. Chromatogr., A, 1996, 745, 135.

2. C. Jiang and D. W. Armstrong, Electrophoresis, 2010, 31, 17.

29. M. A. Schwarz, K. Raith, G. Dongowski, Neubert, J. Chromatogr., A, 1998, 809, 219.

3. J. L. Felhofer, L. Blanes, and C. D. Garcia, Electrophoresis, 2010, 31, 2469.

30. M. A. Schwarz, R. H. Neubert, and G. Dongowski, Pharm. Res., 1996, 13, 1174.

4. N. T. Tran, I. Ayed, A. Pallandre, and M. Taverna, Electrophoresis, 2010, 31, 147.

31. M. A. Schwarz, R. H. Neubert, and H. H. Ruettinger, J. Chromatogr., A, 1997, 781, 377.

5. Y. H. Chu and C. C. Cheng, Cell. Mol. Life Sci., 1998, 54, 663.

32. E. Sletten and N. A. Froystein, Chem. Soc. Rev., 2005, 34, 875.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

and

R. H.

18

Techniques

33. R. Ahmad, H. Arakawa, and H. A. Tajmir-Riahi, Biophys. J., 2003, 84, 2460.

61. P. R. Bouchard, R. M. Hutabarat, and K. M. Thompson, Annu. Rev. Pharmacol. Toxicol., 2010, 50, 237.

34. H. Arakawa, J. F. Neault, and H. A. Tajmir-Riahi, Biophys. J., 2001, 81, 1580.

62. A. S. Barbas, J. Mi, B. M. Clary, and R. R. White, Future Oncol., 2010, 6, 1117.

35. N. G. A. Abrescia, T. Huynh-Dinh, and J. A. Subirana, J. Biol. Inorg. Chem., 2002, 7, 195.

63. C. C. Huang, Z. Cao, H. T. Chang, and W. Tan, Anal. Chem., 2004, 76, 6973.

36. A. A. Ouameur and H.-A. Tajmir-Riahi, J. Biol. Chem., 2004, 279, 42041.

64. Y. H. Chu, W. J. Lees, A. Stassinopoulos, and C. T. Walsh, Biochemistry, 1994, 33, 10616.

37. M. de la Fuente, A. Hernanz, and R. Navarro, J. Biol. Inorg. Chem., 2004, 9, 973.

65. N. H. Chiem and D. J. Harrison, Electrophoresis, 1998, 19, 3040.

38. A. A. Ouameur, H. Arakawa, R. Ahmad, et al. DNA Cell Biol., 2005, 24, 394.

66. G. M. Korf, J. P. Landers, Biochem., 1997, 251, 210.

39. A. R. Stettler, V. Chaurin, E. C. Constable, et al. J. Biol. Inorg. Chem., 2007, 12, 194.

67. K. Shimura and B. L. Karger, Anal. Chem., 1994, 66, 9.

and

D. J. O’Kane,

Anal.

68. R. G. Nielsen, E. C. Rickard, P. F. Santa, et al. J. Chromatogr., 1991, 539, 177.

40. A. R. Stettler, V. Chaurin, E. C. Constable, et al. Electrophoresis, 2008, 29, 3342.

69. L. Tao and R. T. Kennedy, Electrophoresis, 1997, 18, 112.

41. N. C. Stellwagen, A. Bossi, C. Gelfi, and P. G. Righetti, Anal. Biochem., 2000, 287, 167.

70. W. Thormann, M. Lanz, J. Caslavska, et al. Electrophoresis, 1998, 19, 57.

42. A. R. Stettler, P. Krattiger, H. Wennemers, and M. A. Schwarz, Electrophoresis, 2007, 28, 1832.

71. M. J. Schmerr and A. Jenny, Electrophoresis, 1998, 19, 409.

43. A. J. Wilson, Chem. Soc. Rev., 2009, 38, 3289.

72. L. Ye, C. Le, J. Z. Xing, et al. J. Chromatogr., B, 1998, 714, 59.

44. C. Choudhary and M. Mann, Nat. Rev. Mol. Cell Biol., 2010, 11, 427. 45. J. Eriksen, T. N. Jørgensen, and U. Gether, J. Neurochem., 2010, 113, 27.

73. L. Tao, C. A. Aspinwell, and R. T. Kennedy, Electrophoresis, 1998, 19, 403. 74. N. H. Heegaard, J. Chromatogr., A, 1994, 680, 405.

46. X. Liu, F. Dahdouh, M. Salgado, and F. A. Gomez, J. Pharm. Sci., 2009, 98, 394.

75. N. M. Schultz and R. T. Kennedy, Anal. Chem., 1993, 65, 3161.

47. N. A. Guzman, T. Blanc, and T. M. Phillips, Electrophoresis, 2008, 29, 3259.

76. M. H. A. Busch, H. F. M. Boelens, J. C. Kraak, et al. J. Chromatogr., A, 1996, 744, 195.

48. N. H. Heegaard, Electrophoresis, 2009, 30, 229.

77. J. P. Ou, Q. G. Wang, T. M. Cheung, et al. J. Chromatogr., B, 1999, 727, 63.

49. W. E. Pierceall, L. Zhang, and D. E. Hughes, Methods Mol. Biol., 2004, 261, 187. 50. N. H. Heegaard, Electrophoresis, 2003, 24, 3879.

78. L. Kremser, D. Blaas, and E. Kenndler, Electrophoresis, 2009, 30, 133.

51. S. K. Joshi, K. Hashimoto, and P. A. Koni, Genesis, 2002, 33, 48.

79. L. Kremser, D. Blaas, and E. Kenndler, Electrophoresis, 2004, 25, 2282.

52. Y. L. Deribe, T. Pawson, and I. Dikic, Nat. Struct. Mol. Biol., 2010, 17, 666.

80. V. M. Okun, B. Ronacher, D. Blaas, and E. Kenndler, Anal. Chem., 2000, 72, 4634.

53. J. Simicevic and B. Deplancke, Mol. Biosyst., 2010, 6, 462.

81. V. M. Okun, R. Moser, B. Ronacher, et al. J. Biol. Chem., 2001, 276, 1057.

54. M. Berezovski and S. N. Krylov, J. Am. Chem. Soc., 2003, 125, 13451.

82. J. T. Pedersen, J. Ostergaard, G. Houen, Heegaard, Electrophoresis, 2008, 29, 1723.

55. A. Obubuafo, S. Balamurugan, H. Shadpour, et al. Electrophoresis, 2008, 29, 3436.

83. W. J. Underberg and J. C. Waterval, Electrophoresis, 2002, 23, 3922.

56. M. F. Fraga, E. Ballestar, and M. Esteller, J. Chromatogr., B, 2003, 789, 431.

84. I. S. Krull, R. Strong, Z. Sosic, et al. J. Chromatogr., B, 1997, 699, 173.

57. G. J. Foulds and F. A. Etzkorn, J. Chromatogr., A, 1999, 862, 231.

85. L. Kremser, T. Konecsni, D. Blaas, and E. Kenndler, Anal. Chem., 2004, 76, 4175.

58. N. H. Heegaard, D. T. Olsen, and K. L. P. Larsen, J. Chromatogr., A, 1996, 744, 285.

86. J. Ostergaard and N. H. Heegaard, Electrophoresis, 2006, 27, 2590.

59. T. Mairal, V. C. Ozalp, P. Lozano S´anchez, et al. Anal. Bioanal. Chem., 2008, 390, 989.

87. Y. Xiao, W. Wu, M. P. Dierich, and Y. H. Chen, Int. Arch. Allergy Immunol., 2000, 121, 253.

60. Y. Kim, C. Liu, and W. Tan, Biomark. Med., 2009, 3, 193.

88. Y. H. Chen, Y. Xiao, W. Wu, et al. Immunobiology, 2000, 201, 317.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

and

N. H.

ACE—Tool to characterize intermolecular interactions

19

89. T. Yu, Y. Xiao, Y. Bai, et al. Immunol. Lett., 2000, 73, 19.

91. S. Kiessig and F. Thunecke, J. Chromatogr., A, 2002, 982, 275.

90. S. Kiessig, J. Reissmann, C. Rascher, et al. Electrophoresis, 2001, 22, 1428.

92. M. Vlckova, A. R. Stettler, and M. A. Schwarz, J. Liq. Chromatogr. Relat. Technol., 2006, 29, 1047.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc035

Ion Chromatography and Membrane Separations Using Macrocyclic Ligands John D. Lamb and Na Li Brigham Young University, Provo, UT, USA

1 2 3 4

Introduction Selective Binding by Macrocyclic Ligands Ion Chromatography Application of Macrocycles to Ion Chromatography 5 Liquid Membranes 6 Conclusions References

1

1 2 7 9 17 23 23

INTRODUCTION

One of the most intriguing characteristics of macrocyclic ligands is their binding selectivity for specific guest species. From the inception of studies of synthetic macrocycles in the late 1960s and early 1970s, researchers from various chemical fields were investigating this selectivity and how it might be applied to separations science. Indeed, the 1987 Nobel Prize for chemistry was awarded to Cram, Lehn, and Pedersen for their development and use of molecules with structure-specific interactions of high selectivity.1 Investigation into macrocycle selectivity and its applications has gained increasing momentum during the past 40 years; and from 2004 to 2009, more than 800 papers relating to macrocyclic chemistry have been published each year.2 Separations and other practical analytical chemistry applications for these compounds include liquid membrane separations,

chromatography of various types, capillary electrophoresis (CE), luminescent sensors, and ion selective electrodes,3 among others. And beyond these, other applications have been made in macrocycle–metal complexes, such as magnetic resonance imaging (MRI) contrast agent sensors4 and catalysts.5 This review chapter focuses specifically on the application of macrocyclic ligands to ion chromatography (IC) and liquid membrane separations. It begins with a brief description of the intrinsic selectivities of the macrocyclic hosts used to date in these applications. Since the majority of applications involve the separation of metal cations, the discussion focuses primarily on these species, with some attention being given to the separation of anions and neutral species. The chapter then proceeds to describe the basic elements and chemical features of IC and how macrocycles such as crown ethers, cryptands, resorcinarenes, and cyclodextrins (CDs) have been incorporated into both the mobile and stationary phases of IC to effect novel separations of both cations and anions. The development of capacity gradient chromatography of anions, unique to macrocycle-based IC, is described. The application to both analytical separations and preconcentration/matrix elimination is included. The final section introduces the various types of liquid membranes, including bulk liquid membranes (BLMs), solid supported liquid membranes (SLMs), emulsion liquid membranes (ELMs), and polymer inclusion membranes (PLMs). It then proceeds to describe the various studies done to characterize the use of macrocyclic ligands as selective carriers of cations, anions, and neutral species across these membranes for separations purposes. The chapter does not endeavor to branch out into the expansive field of biomembranes and biomembrane mimics.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

2

Techniques

2

SELECTIVE BINDING BY MACROCYCLIC LIGANDS

S-donor macrocycles are typically sulfur-substituted crown ethers. The mixed donor macrocycles include those with N/S, S/O, N/O, and N/S/O combinations. Calixarenes and resorcinarenes are most often O-donors, but may contain various functional groups and thus be classified here as mixed donor macrocycles. The macrocycles most widely used in IC and liquid membrane separations have been the crown ethers, nitrogen-substituted crown ethers, cyclen, cyclam, cryptands, calixarenes, resorcinarenes, and CDs.

The separations chemist relies on the selectivity of chemical systems to achieve the desired separations outcome. As chemists search the literature to find selective molecules for such applications among known macrocyclic structures, it is important that they be attuned to the structural and system parameters that can affect this selectivity, especially if the macrocycle must be modified structurally to meet the needs of the desired separation. The selectivity of macrocyclic ligands is determined by various characteristics of the host, guest, and reaction medium. In this section, we discuss parameters that affect selectivity in some detail inasmuch as this selectivity lies at the heart of separations that may be achieved. We begin with a discussion of host characteristics.

2.1

2.2

2.2.1 Size-fit and conformational changes It is important for the designer of separations systems, based on macrocycle selectivity, to be familiar with the ligand design features that influence that selectivity. Macrocyclic hosts typically bind guests by sequestering them in the host cavity. For example, the binding of alkali and alkaline earth metal ions to crown ethers involves a guest–ion interaction with the crown cavity, and is considered to be electrostatic in nature. It follows, therefore, that one important contributor to selectivity is the relative sizes of guest ion and host cavity, since this influences the distances and orientation between the ligand dipoles and ion charge. In general, this kind of host–guest interaction has been likened to that of a “key” and a “lock”6 ; and in the particular case of metal cations and crown ethers, when the ionic radius of the metal matches the cavity size of the crown ether, this particular contribution to complex stability is

Structures of macrocycles used in ion chromatography and membrane separations

Macrocyclic ligands used in IC and liquid membrane separations can be divided into four classes according to ligand heteroatom: O-donor macrocycles, N-donor macrocycles, S-donor macrocycles, and mixed donor macrocycles. Figure 1 shows the structures of the representative members of these groups. Typically, the O-donor macrocycles are crown ethers and their substituted analogs, as well as CDs. The typical N-donor macrocycles are, in increasing ring size and number of nitrogen atoms: tacn (L1 ), cyclen (L2 ), and cyclam (L3 ) and their analogs. The R1 N

O O

O

n

Crown ethers

S

NH HN

S

S N

N

R2

O

NH HN

Macrocyclic ligand characteristics that influence selectivity

R3

Tacn (L1)

NH HN

NH HN

Cyclen (L2)

Cyclam (L3)

O S

S

S

O

S-donor crown ether

N/O donor (2.2.2-cryptand)

O

O

S

H N

N/S-donor macrocycle

O O O OH HO

HO

S N R

S

O

N

OH

O

R N

N

N O O N

n = 1, 12-crown-4 n = 2, 15-crown-5 n = 3, 18-crown-6 n = 4, 21-crown-7 n = 5, 24-crown-8 n = 7, 30-crown-10 O

O

HO HO

S

O

OH HO OHHO OH

HO O OH

OH

O S

S

OH

OH O HO

O

OH O OH

O OH

OH OH OH HO

H R

H

H R

R

H R

HO

OH O

OH

OH OHO

O

O HO

N/O-donor crown

Figure 1

O/S-donor crown

Calixarene

Resorcinarene

Cyclodextrin (n = 1, 2, 3)

Basic structures of macrocyclic ligand types used in IC and liquid membrane separations.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations

3

Table 1 Radii of mono- and divalent cations and their stability constants with 18crown-6 in methanol at 25 ◦ C.8 Cation Li+ Na+ K+ Rb+ Cs+ a

˚ a Radius (A) 0.76 1.02 1.38 1.52 1.67

Log K b

4.36 ± 0.02 6.06 ± 0.03 5.32 ± 0.11 4.79 ± 0.05

˚ a Radius (A)

Cation Be2+ Mg2+ Ca2+ Sr2+ Ba2+

Log K b b

0.72 1.00 1.18 1.49

3.86 ± 0.02 >5.5 7.04 ± 0.08

Ionic radii of metals when coordination number is 6. measurable heat.

b No

[2.2.2]b

10.0

[2.2.1]b

9.0

8.0 [2.1.1]b

[3.2.2]b

7.0

log K

maximized. Experimental results not only confirmed this point early in macrocyclic chemistry research but also made it clear that other contributions could be equally important, as is described below. For example, Lamb et al.7 summarized the binding stability of 18-crown-6 with alkali and alkaline metal ions in water. K+ and Ba2+ form the most stable metal ion complexes in their respective groups, and this can be explained in part because their ionic radii closely match the cavity size of 18-crown-6.7 From the X-ray crystallographic data, the cavity radius of 18-crown˚ The sizes 6 is determined to be between 1.34 and 1.43 A. + 2+ of K and Ba closely match the size of this ligand cavity (Table 1). This size-fit selectivity among metal cations is also applicable to cryptands. Zhang et al.9 summarized size–match relationship between cryptands and alkali and alkaline metal ions. The smaller cryptand [2.1.1], with a cavity size of ˚ formed the most stable complex with Li+ (ionic 0.8 A, ˚ among the alkali metal cations. Cryptand radius: 0.76 A) ˚ matches the size of Na+ [2.2.1], with a cavity size 1.1 A, ˚ and showed the highest selectivity (ionic radius 1.02 A), + for Na . Large cryptands like [3.3.2] and [3.3.3] form more stable complexes with larger cations such as Rb+ and Cs+ . Figure 2 shows the dramatic correspondence between cation selectivity and ligand cavity size among the alkali metal cations for a sequence of cryptands.7, 10 Although this size-fit principle is only one of the several factors that come into play, it remains a significant contributor to all host–guest interactions to the degree that it influences the energy associated with the proximity of attracting charges (sometimes in the form of dipoles or H-bonds) between species. As macrocyclic chemistry developed, the influence of ligand flexibility and the associated conformational energies was appreciated. For example, large crown ethers like 30crown-10 can form complexes with metals in a variety of conformations, and in general, the greater the flexibility of the ligand, the less appropriate it becomes to consider the cavity as preorganized to accommodate the host. X-ray diffraction techniques have vividly illustrated this principle,

[3.3.3]c

6.0

5.0

4.0

3.0

2.0 0

a

a 75

100

Li+

+

125

150 +

+

Na K Rb Metal ion radius (pm)

a 175 Cs+

Figure 2 Selectivity of cryptands among alkali metal cations (a, value reported monoaza-18-crown6 > 1,10-diaza-18-crown-6. Thaler et al.15 compared the stability constants of Ag+ with several crown ethers to corresponding monoaza- and diaza- analogs in CH3 OH and propylene carbonate (PC) solutions. When oxygen atoms were replaced by nitrogen atoms, the cavity sizes changed only slightly since the van der Waals radii of oxygen and nitrogen are quite similar. For the four classes of crown ethers studied (12-crown-4, 15-crown-5, 18-crown-6, and 21-crown-7), they reported that the replacement of oxygen atoms by nitrogen atoms caused the stability constants of the Ag+ complex to increase in the following order: crown ether < monoaza-crown < diaza-crown. Furthermore, Kodama et al.16 reported that when the six oxygen atoms of 18-crown-6 were totally replaced by nitrogen, the stability constants of the resulting macrocycle with transition metal ions are significantly higher than with alkali and alkaline earth metal ions. For instance, the stability constant of hexaaza-18-crown-6 with Ca2+ is 2.5 ± 0.2, while that with Co2+ is 18.9 ± 0.2. The relationship of ligand selectivity to donor atom type can be understood from hard–soft acid–base theory.

X

R O

O

O O

X

O

N

N O

N

O

R′ X

N O

1,7-N2 18C6 (6) 1,4-N2 18C6 (8) 1,7-(CH3 N)2 18C6 (7) 1,4-(CH3 N)2 18C6 (9)

Log β

R

R′ X

Figure 5 Structures of macrocycle isomers with nitrogen donors at different positions. (Reproduced from Ref. 14.  Wiley-VCH, 1998.)

O-donor macrocycles can bind alkali, alkaline, and rare earth metals predominantly by electrostatic forces. By contrast, N-donor macrocycles can bind softer Lewis acids like transition metals, and in this particular case, the interactions are predominantly coordinate in nature. Thiasubstituted ligands are much softer than O- and N-donor macrocycles and can selectively bind softer Lewis acids like Ag+ and Hg2+ . In addition to donor atom type and number, the position of the donor atoms has a noticeable effect on the stability constants with metal cations. Solov’ev et al.14 studied the stability constants of K+ with two isomers of diaza18-crown-6 in which nitrogen atoms occupied different positions. The structures of the two ligands are shown in Figure 5 and the stability constants and thermodynamic parameters are shown in Table 2. Clearly, not only the number of heteroatoms but also their positions in the ring, has a significant effect on ligand selectivity.

2.2.3 Substitution effects on complexation selectivity Often in separations applications of macrocycles, it is necessary to add substituents to the ligand to adjust its

Table 2 Stability constant (log β) and thermodynamic data (kJ mol−1 ) for diazacrown-K+ complexes in methanol at 298 K.14 (Structures from Figure 5). M:L

5

−G

−H

+T S

1:1 1:2 1:1 1:2 1:1 1:2

4.17 (0.42) 5.97 (0.44) 2.6 (0.4) 5.5 (0.2) 4.92 (0.58) 8.60 (0.73)

23.8 34.1 14.8 31.4 28.1 49.1

(2.4) (2.5) (2.3) (1.1) (3.3) (4.2)

1.4 (0.1) 6.1 (0.6) 13 (6) 11 (1) 28.5 (1.8) 44.6 (1.3)

22.4 (2.4) 28.0 (2.6) 2 (6) 20 (1) −0.4 (3.8) 4.5 (4.4)

1:1 1:2

4.71 (0.54) 8.44 (0.54)

26.9 (3.1) 48.2 (3.1)

19.8 (0.8) 37.1 (0.2)

7.1 (3.2) 11.1 (3.1)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

6

Techniques

solubility or phase partitioning. In such cases, however, the user must be aware of potential alterations these substituents might make to the very ligand selectivity that is being applied to the separations objective. As an example, Monsef et al.17 studied the stability constants of aza-18-crown-6 and dibenzopyridino-18-crown-6 with three different metal ions, Tl+ , Pb2+ , and Cd2+ , using direct current and differential pulse polarographic techniques (DPP). The stability constants were determined in some binary mixed solutions of dimethylformamide (DMF) and other solvents such as methanol, n-propanol, nitromethane, and acetonitrile at 22 ◦ C. In all of the solvent systems they studied, aza-18-crown-6 showed stronger binding than dibenzopyridino-18-crown-6 for 1 : 1 complexes. Aza-18crown-6 and dibenzopyridino-18-crown-6 have the same ligand donor atoms and arrangement, but dibenzopyridino18-crown-6 has two phenyl rings as well as the pyridine ring. These electron-withdrawing benzyl groups reduce the electron-donating character of the oxygen atoms on the 18crown-6 ring. Furthermore, they reduce the flexibility of the crown ether ring that can weaken the binding with metal ions if this preorganized arrangement does not suit. Finally, the nitrogen atom in the 18-crown-6 ring is part of the pyridine unit, so the electron-donating power of the nitrogen atom is reduced by the resonance inherent in the pyridine ring. Thus, when designing ligands for use in separations, all such factors must be considered as the hydrophilicity of the ligand is adjusted by substitution. Another example of this principle is illustrated in the work of Shchori and Jagur-Grodzinski et al.7, 18 who studied the effect of substituent groups on the selectivity of dibenzo-18-crown-6 for Na+ in DMF. When the electronwithdrawing group −NO2 was added to the phenyl ring, the stability constant with Na+ decreased from 2.69 to 1.99. Yet, when the electron-donating group −NH2 was added, the stability constant with Na+ increased from 2.69 to 2.76. In addition, the positions of substitutions can also have a noticeable effect on the selectivity of host–guest association. As an example, Chi et al.19 studied the stability constants of alkali metals with two isomers of diaza-18-crown-6 derivatives. In one isomer, two 2,6-difluorobenzyl groups substituted two protons on two nitrogen atoms on the crown ether ring; in the other isomer, 3,5-difluorobenzyl was the substituent group. For the four alkali metal ions Na+ , K+ , Rb+ , and Cs+ that they studied, 2,6-difluorobenzyl substituted aza-crown ether showed higher binding stability than the 3,5-difluorobenzyl substituted azacrown.

2.3

Ion effect on macrocycle selectivity

Since the binding of macrocyclic ligands to alkali and alkaline metals is electrostatic in nature, the binding between

them has no real stereochemical requirements. Such metal ions can bind to macrocyclic ligands with various coordination numbers and conformations. For example, Li+ can form a 1 : 2 sandwich complex with the smallest crown ether, 12-crown-4. Li+ is coordinated to eight oxygen atoms of the two 12-crown-4 molecules.12 With the larger 15crown-5, Li+ is six-coordinated with five oxygen atoms of 15-crown-5 and one Cl− from the counter ion in the inner sphere. In this case, Li+ fits well in the crown ether cavity, and only a slight displacement from the crown ring was observed for this 1 : 1 complex.20 For the much larger 18-crown-6, 1 : 2 complexes have been reported. Watson et al.21 reported that two Li+ cations can be encapsulated in the ring, and are coplanar with the six oxygen atoms. Also, 18-crown-6 can dramatically distort to coordinate with two Li+ ions.22 Steed6 has reviewed the coordination chemistry of alkali metal cations with crown ether ligands in detail. Size fit was described above as one of the primary reasons for the selectivity of macrocyclic ligands among cations. But cations of like size may exhibit significantly different binding constants with metal cations of different charge. For example, Ba2+ exhibits stronger binding to 18-crown-6 than singly charged K+ of very similar size. However, the role of cation charge in selectivity is complex, because charge affects not only the energy of association with the ligand but also the energy that must be expended in the desolvation of the cation, as described below. For example, Ca2+ is similar in size to Na+ ; yet the experimental results show that 18-crown-6 prefers Na+ over Ca2+ . This selectivity can be manifested in both the enthalpy and the entropy of the complexation reaction as described by Lamb et al.8 Cation polarizability is also a factor in selectivity. Kodama et al.16 studied the complexation of aza-18-crown6 with alkali, alkaline earth, and some d-block metals. This ligand showed higher selectivity toward the more polarizable, soft d-block metal ions than the hard, nonpolarizable alkali and alkaline earth metal cations.

2.4

Solvent

Selectivity of the macrocyclic ligand is significantly affected by the reaction medium, specifically the nature of the solvent. The stability constants of alkali metals with 18-crown-6 have been reported in several solvents, such as water, methanol, acetonitrile, benzonitrile, and PC. Table 3 lists example stability constants to illustrate the wide range of stabilities that can be achieved.8, 23–25 The effect on selectivity is thus illustrated in Figure 6. Specifically, in all of the solvents shown, the relative binding strengths, which are at the root of selectivity, are often maintained even though the strength of complexation can vary dramatically

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations

7

Table 3 Stability constants of 18-crown-6 with alkali metals in several solvents.8, 23–25

Li+ Na+ K+ Rb+ Cs+ a

H2 O23

CH3 CN24

0.80 ± 0.10 2.03 ± 0.10 1.56 ± 0.02 0.99 ± 0.07

4.75 ± 0.11 5.76 ± 0.13 4.89 ± 0.09 4.36 ± 0.08

Benzonitrile25

log K

5

3 2

Water Acetonitrile Benzonitrile Propylene carbonate Methanol

1 1.0 1.1 Na+

1.2

5.26 ± 0.06 6.12 ± 0.08 5.34 ± 0.07 4.50 ± 0.04

CH3 OH8 4.36 ± 0.02 6.06 + 0.03 5.32 ± 0.11 4.79 ± 0.05

PC is propylene carbonate.

6

4

4.74 ± 0.02 4.89 ± 0.09 6.11 ± 0.11 5.84 ± 0.08

PC24,a

1.3

1.4 K+

1.5 1.6 1.7 Rb+ Cs+

Cation radius

Figure 6 Selectivity of 18-crown-6 to alkali metals in different solvents (from Refs. 8, 23–25).

with solvent. In this case, not only is the order of selectivity maintained but the degree of selectivity of 18-crown-6 for K+ over other cations in the same group is also largely conserved. The difference in absolute binding strengths can be understood in terms of the smaller ion desolvation energy for the nonaqueous solvents than for water. In this sense, the dielectric constant of the solvent plays an important role. Water has a high dielectric constant compared to many nonaqueous solvents, and this parameter is often a good predictor of the relative energy required to desolvate cations. Moreover, the energy of desolvation of the ligand must be taken into account. This energy is especially high for O-donor and N-donor ligands that can hydrogen bond to water or to other hydrogen-bonding solvents. When multiple factors come into play, it is not always easy to extrapolate selectivity in one solvent to predict that in another. Katsuta et al.26 studied the stability of complexes formed by alkali metals with dibenzo-18-crown-6 (DB18C6) and dibenzo-24-crown-8 (DB24C8) in different solvents. They found that, as would be predicted, the complexes were much less stable in water than in any of the nonaqueous solvents they chose such as methanol, DMF, PC, and acetonitrile. However, the degree of selectivity for K+ over Na+ in water and methanol is reduced for DB18C6

and actually reversed for DB24C8 in PC and acetonitrile. They rationalize these results in terms of the transfer activity coefficients of the respective species, alkali metal ions, ligands, and, in particular, the complexes formed. In effect, the solvation of any resulting complex is partly governed by the conformation of the ligand surrounding the metal ion, which in turn affects the degree of association of solvent with the complexed metal ion. This effect is more pronounced as ligand flexibility increases. The same is true when mixed stoichiometries are present or when the cation is not well sequestered in the ligand cavity. This latter point is underscored in the work of Arnett and Moriarity27 on the complexation of dicyclohexano-18-crown-6 in various solvents. In strongly solvating solvents, a greater affinity is observed for larger cations. The net result for the separations chemist is that, in general, complex stability is higher in nonaqueous solvents, and because there are so many parameters affecting selectivity, direct measurement is really required when a new solvent system is employed.

3 3.1

ION CHROMATOGRAPHY Principles of suppressed ion chromatography

In the early 1970s, new discoveries led to the development of the analytical method now called “ion chromatography.” This name applies specifically to modern high performance liquid chromatographic determination of ionic species, and is distinct from other chromatographic ion separations that are not carried out in high performance mode. While some applications of non-HPLC (high performance liquid chromatography) employing macrocyclic ligands have been studied, this chapter focuses specifically on the analytical method defined narrowly as IC. A sensitive, universal detector for inorganic cations and anions in chromatography was the principle unsolved problem that delayed the development of IC. Although these ions could be separated efficiently by ion-exchange, sensitive direct detection could not be achieved by any spectrophotometric detector because most common ions do not

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

8

Techniques

absorb in the visible or UV ranges.28 Conductance is a universal property of ions in solution and is directly dependent on concentration. Yet, while the most direct method to detect inorganic ions would be to use a conductivity detector, the background conductance of the ionic mobile phase (eluent) constituents makes the sensitive detection of inorganic ions difficult. Specifically, the analyst is faced with measuring a very small signal atop a very large background. In 1975, Small et al.29 published a method that laid the foundation for modern IC. A conventional liquid chromatographic system consists of four parts: a pump that pumps the flowing mobile phase, a sample injector, an ionexchange column, and a detector. The new instrumental setup that Small et al. reported contained a second column that was added between the separator column and the detector known as a suppressor. Take anion separation as an example: when the analyte KCl passes through the anion separator column, Cl− ions exchange with the anionexchange sites on the resin, typically quaternary amines. The mobile phase NaOH then elutes the analyte from this column in the form of NaCl as shown in (1). The newly added suppressor column is a strong cation-exchanger, so when eluent and analyte pass through this column, reactions (2) and (3) occur. The background conductance of the eluent NaOH is converted to H2 O, which has a much lower conductance than NaOH. In addition, after the suppressor column, the detected signal is the conductance of HCl, which has much higher conductance than NaCl. (equivalent conductances H+ 34.96, Na+ 5.01 mS·m2 ·mol−1 ).30 In this way, not only is the background conductance of eluent reduced but also the detection sensitivity is significantly improved. Resin+ Cl− + OH− → Resin+ OH− + Cl−

(1)

Resin− H+ + NaOH → Resin− Na+ + H2 O −

+

−

+

+

(2) −

Resin H + NaCl → Resin Na + H + Cl

(3)

Over more than 30 years of development, this suppressor concept has been modified and improved significantly and the previously used packed column suppressors have evolved into membrane-based self-regenerating suppressors. Dionex Corporation markets different suppressors for cation and anion determination based on specific ion-exchange membranes. For the anion suppressor, two cation-exchange membranes divide the unit into three compartments, that is, two regenerant chambers and one eluent chamber. Water in the regenerant channels is electrolyzed when an electrical potential is applied. The resulting H+ ions in one regenerant chamber pass through the cation-exchange membrane to neutralize the OH− ions in the eluent, yielding a reaction similar to (2). The counter

ions of the eluent pass through the other cation-exchange membrane into the other regenerant chamber and flow to waste. In contrast to anion separations, cation separations are typically carried out using sulfonate-based stationary phases for alkali and alkaline earth metal cations, or chelating groups such as iminodiacetate for transition metal ions. In such cases, conductimetric detection can be achieved with a suppressor that contains an anion-exchange membrane to neutralize, in this case, an acidic eluent by the introduction of hydroxide ion. IC sometimes employs methods of detection other than conductivity. For transition metal cations, UV–vis detection is used after postseparator reaction with a chelating agent. Electrochemical detectors such as pulsed amperometric detectors have also been employed, especially for organic ions. We focus here mainly on conductimetric detection since the vast majority of published work with macrocycles uses this method.31, 32 In recent years, it has become clear that it is possible to carry out some aspects of IC without a suppressor. Nonsuppressed and indirect conductimetric detection methods have played a minor role in IC over the past 30 years. The stationary and the mobile phases of nonsuppressed IC are quite different from those with suppressed IC. First, sensitive detection in nonsuppressed IC is achieved by carefully selecting the eluent composition. The mobile phases used are usually aromatic carboxylates33 such as benzoate and phthalate, which have much lower conductance than the analytes of interest due to their large size and concomitant low conductivity. Second, there is a marked difference in the stationary phases used. Specifically, agglomerated ion-exchange resins are the main type of anion stationary phase for suppressed IC.34, 35 For anion separation, these resins are comprised of sulfonated polymer cores such as polysterene-divinylbenzene (PS-DVB) coated with a monolayer of substituted latex particles coated on the surface.36 These latex particles have various compositions and functionalities (usually quaternary amines), which can regulate the separation selectivity. By contrast, the stationary phases made for nonsuppressed IC are based primarily on the approach taken by Fritz’s group,37 which involves the functionalization of polymer beads by reacting them with sulfuric acid or aminating reagents. Third, suppressed IC uses a conductivity detector since it is universal in its response to ions. In nonsuppressed IC, however, while conductivity detectors are sometimes used,38 other detection methods like spectrophotometric39 and electrochemical40 detection are more common. In such cases, indirect detection of an analyte may be achieved when a detected eluent ion is “missing” as an analyte ion moves through, taking its place and yielding a negative peak.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations

3.2

New perspectives in IC

As for most current chromatographic techniques, simple, fast, and sensitive separations are the aim of IC. There are three principle areas of research for the improvement of IC methods, including the design of new ion-exchange stationary phases, the development of more effective mobile phases for specific applications, and the development of new detection techniques. With reference to stationary phase design, in a 2002 review Sarzanini35 described the underlying principles and development strategies of stationary phases used in several IC columns. Shortly after that, Weiss and Jensen34 summarized the various modern stationary phases used in IC. They also introduced the detailed chemical exchange processes involved in the common stationary phases in anion- and cation-exchange columns. Furthermore, Paull et al.41 have detailed the specific stationary phases designed for cation and anion separations in the analysis of complex sample matrices, a common and difficult problem for the analyst. One of the most intriguing recent developments in the stationary phases of IC is the design of monolithic stationary phases for ion separations. A monolithic stationary phase consists of a single piece of porous stationary phase material that has about 15% higher porosity than packed columns consisting of individual beads. It is also less compressible than packed beads.42 As a result, low back pressures and high mass transfer efficiencies can be maintained with monolithic columns even while using high flow rates. For example, Lucy’s group42 used didodecyldimethylammonium bromide (DDAB) to coat a reversed-phase monolithic column and thereby generate an ion-exchanger to separate seven anions. As is noted below, this strategy of adsorbing a hydrophobically substituted ion exchange molecule to an inert column substrate has also been a common approach to applying macrocycles to IC. The column prepared by Lucy could separate anions in ultra-short time (30 s) because the sturdy, noncompressible monolithic design allows eluent flow rates as high as 10 ml/min. In another example, Haddad’s group prepared a monolithic column that was in situ polymerized by butyl methacrylate, ethylene dimethacrylate, and 2-acrylamido-2-methyl1-propanesulfonic acid within fused-silica capillaries; it was then coated by Dionex AS10 or AS18 quaternary ammonium functionalized latex particles.43 In this case, seven anions could be separated in less than 2 min using high flow rates without sacrificing separation efficiency. Several review papers about the application of monolithic columns for fast IC separations have been published.44, 45 Most recently, in 2009, Nordborg and Hilder45 reviewed the advances in polymer monoliths for IC between the years 2003 and 2008. Materials and methods to coat the surface

9

of monoliths and copolymerization of the functional groups were introduced. Developments have also occurred in the area of eluent composition. Bicarbonate was the mainstay IC eluent for anions for many years, largely because the preferable eluent, NaOH, can react with CO2 in the atmosphere forming carbonate, which results in unpredictable concentration changes of NaOH and unstable baselines. The development of the air-isolated electrolytic eluent generator for hydroxide eluents now keeps the eluent in a closed system impeding the reaction of NaOH with CO2 so that hydroxide is becoming the eluent of choice for anion IC.46 Beyond these considerations, various eluents are commonly used for other applications, depending on the specific separation objectives.47, 48 For example, the chelating agent EDTA can be added to the mobile phase to retain Cu2+ , Zn2+ , and Pb2+ as complex anions in the presence of inorganic anions such as Cl− , NO2 − , Br− , and NO3 − .48 In similar fashion, macrocyclic ligands, particularly crown ethers, can also be used as the mobile phase in IC, as described below. In the area of detection, several interesting innovations have been developed over the years. However, most of these are targeted at specific applications, leaving conductivity detection as the prime detection method for inorganic ions, both cations and anions. Among recent developments, we find hyphenated detection methods that can not only achieve sensitive detection but also expand the analyte range of IC.36 For example, inductively coupled plasma spectrometry-mass spectrometry (ICP-MS) is one of the attractive detection methods for IC because of its low detection limits and wide dynamic linear range. Divjak et al.49 reported the detection of halogen and oxyhalogen anions, sulfate, phosphate, selenite, selenate, and arsenate by ICP-MS. With a 50 µl injection sample, the detection limit for BrO3 − , and Br− is 1.0 µg l−1 using this method.

4

APPLICATION OF MACROCYCLES TO ION CHROMATOGRAPHY

Since the early days of macrocyclic ligand studies, these molecules have been incorporated into separation systems including chromatography. In the 1970s, Cram’s group pioneered the application of an optically pure macrocycle for chiral chromatographic separations by liquid chromatography. (R ,R)-tetranaphthyl-22-crown-6 was attached to macroreticular cross-linked polystyrene p-divinylbenzene resin as the stationary phase for chiral separations of amino acids and ester salts.50 Since then, macrocyclic ligands have been introduced into IC either as mobile phase or as stationary phase components. In this section, we introduce several typical types of macrocyclic ligands that have been applied to IC.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

10

4.1

Techniques

Macrocyclic ligands in the stationary phase

Macrocyclic ligands can be applied to a stationary phase in three different ways: adsorbed to the stationary phase, covalently bonded to the stationary phase, or polymerized into the stationary phase.

4.1.1 Crown ethers

3 6 1

10 µs

The most obvious and first explored application of crown ethers and similar macrocycles to IC is in the separation of metal cations, specifically of the alkali and alkaline earth series. Kimura’s group51 pioneered the adsorption of hydrophobic crown ethers onto silica-based columns to effect cation separations, a technique labeled dynamic coating. This was accomplished by passing the coating solution of hydrophobically substituted crown ethers in a mixture of methanol and water through an octadecylsilylsilica (ODS)-packed column, so that the crown ether functional group was strongly retained on the ODS particles by hydrophobic effects.51 The structures of the crown ethers (10) used are shown in Figure 7. These columns were then applied to the separation of alkali and alkaline metal ions in water. Unlike conventional cation-exchange materials that separate these ions according to hydrated ionic size and charge, these crown ether-coated columns displayed selectivity for certain alkali metals related to the previously known selectivities of crown ethers measured in homogeneous solution (described above). For instance, the 18-crown-6 and 15-crown-5 based columns showed the highest selectivity for K+ , and the elution order was Li+ < Na+ < Cs+ < Rb+ < K+ . Smaller 12-crown4 derivatives showed higher selectivity for Na+ with the elution order Li+ < Cs+ < Rb+ < K+ < Na+ . Clearly, the choice of an aliphatic substituent to add hydrophobicity (which minimizes any influence on the electron density of the oxygen donors) did little to alter the inherent selectivity of 18-crown-6, even in such a two-phase system. Yet this minimal substitution made a difference to the selectivity

of 15-crown-5, an effect which is enhanced by the greater hydration energy of Na+ over K+ . This energy must be added to partially dehydrate the metal ion as it passes from the aqueous mobile phase to be retained in the much lower dielectric environment of the stationary phase. Using a more robust chromatographic substrate, Lamb and coworkers52, 53 adsorbed n-tetradecyl-18-crown-6 (TD18C6) onto nonpolar polystyrene-divinylbenzene beads. Compared with the silica substrate, this column has the distinct advantage of being stable in basic eluents. Thus, both anion and cation separations can be accomplished on the same column based on the cation–macrocycle interaction. Cations can be separated due to the selectivity of the macrocycle among cations, whereas anion separations result from the affinities of anions for the positively charged cation–macrocycle complex. The capacity of the column can be modified by changing the cation in the mobile phase, the temperature, and the concentration of organic modifier (e.g., methanol). By choosing an eluent cation with a low affinity for the stationary phase macrocycle, the capacity of the column can be adjusted downward. In addition, the complexation reactions between cations and macrocycles are generally quite exothermic, so increasing temperature leads to weaker complexation, thus decreasing the capacity of the column for anions. And since the stability of the cation–macrocycle complex is greatly affected by the solvent, nonaqueous solvents can be used to adjust the column capacity for anions upwards. The length of the aliphatic side-chain on the macrocycle was found to have a strong influence on the stability of the adsorbed column. Specifically, a column prepared with decyl-18-crown-6 was not as stable to mixed organic/water eluents as that prepared with the more hydrophobic TD18C6. An example separation of anions with the TD18C6 column is shown in Figure 8.52

O

5

7

8

2 4

11 10

O

O

9

12

14

13

O

n

0

n = 1, 2, 3 10

Figure 7 Structures of crown ethers coated on ODS column packing. (Reproduced from Ref. 51.  American Chemical Society, 1986.)

10

20

30 Minutes

40

50

Figure 8 Anion separation by a n-tetradecyl-18-crown-6 (TD18C6) column, 1 = F− , 2 = CH3 CO2 − , 3 = Cl− , 4 = NO2 − , 5 = Br− , 6 = SO4 2− , 7 = NO3 − , 8 = C2 O4 2− , 9 = CrO4 − , 10 = phthalate, 11 = I− , 12 = PO4 3− , 13 = citrate, 14 = SCN− , eluent : 50 mM aqueous KOH-acetonitrile (80 : 20, v/v). (Reproduced from Ref. 52.  Elsevier, 1993.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations The TD18C6 column was also used to separate three alkaline earth metal cations, five alkali metal cations, and the ammonium ion.53 Another approach to applying macrocycle selectivity to cation separations is illustrated in the work of Xu et al.54 who dynamically coated a C30 -substituted silica column with both dodecylsulfate and 18-crown-6 adsorbents to separate mixtures of cations containing H+ , NH4 + , alkali, and alkaline earth metal ions. They compared the separation efficiency of this column to that coated only with the common ion-exchanger dodecylsulfate. Without the adsorbed 18-crown-6, NH4 + and K+ eluted together because of their similar ionic radii and charge, but with the added 18-crown6 coating, the column exhibited high selectivity for K+ , so that NH4 + and K+ were well separated. The selectivity here is due in part to the different mode of complexation of 18crown-6 for these two ions: K+ is sequestered in the cavity, whereas NH4 + sits atop the cavity forming hydrogen bonds to alternate ligand oxygen atoms.7 In this chromatographic example, the macrocycle served not as the primary column active site but as a secondary feature that adjusted the selectivity of the column. Although a crown ether-coated column can effectively separate cations and its anion capacity can be adjusted by controlling the amount of adsorbed crown ether, it is susceptible to loss of the macrocyclic active site. When water eluents are used, this is generally not a problem, but if mixed mobile phase solvents are used (e.g., water/methanol), the coated crown ether can be washed off the column resulting in a loss of capacity.34 To avoid this problem, macrocycles can also be covalently bonded to the stationary phase in IC. For example, the stationary phase of the Dionex IonPac CS15 cation separator column contains carboxylic acid groups, phosphonate groups, and 18-crown-6 ether groups permanently bonded to the polystyrene-divinylbenzene macroporous beads.3 Compared to the CS12 cation separator column, which is similar to the CS15 column except that it has no crown ether on the stationary phase, the CS15 column shows better resolution for ammonium and sodium ions, which are typically difficult to quantify together because of their similar selectivity by sulfonic acid or carboxylic acid cation-exchangers.55 In addition, in a 4000-to-1 concentration ratio of sodium to ammonium ion and a 10 000-to-1 ratio of ammonium to sodium ions, ammonium and sodium ions can be well separated isocratically (i.e., with a single eluent) using a CS15 column.55 Copolymerization of polymeric crown ether with silica gel or other support materials is another way to incorporate crown ether into stationary phases. Blasius and coworkers thoroughly studied methods for polymerizing cyclic polyethers with various polymeric matrices and applied

11

O

Si

O

C H2

H C H

O

O O

O O

n (a)

11 CH3 CH3 O O Si(CH2)3NHCOC(CH2C)n x O CH2 C O x

NH2 O

O

O

O O x = an end group of initiator or poly crown ether (b)

12

Figure 9 (a) A polymeric crown ether stationary phase. (Reproduced from Ref. 56.  IUPAC, 1982.) (b) structure of the polyether modified silica stationary phase. (Reproduced from Ref. 57.  American Chemical Society, 1983.)

them in ion separations.56 An example polymer structure (11) is shown in Figure 9(a).56 Nakajima et al.57 also explored stationary phases in which the macrocycle is covalently bonded to silica. Figure 9(b) shows the structure of the polyether modified silica. These stationary phases have several obvious advantages over adsorbed systems: first, by adding crown ether to the stationary phase, these columns have specific selectivity for certain ions; second, they are compatible with organic solvents like methanol that cannot be used in stationary phases made by adsorbing crown ether to support material. This provides another degree of freedom to the analyst, namely, changing the percentage of organic solvents in the mobile phase. As we discussed previously, the stability constants of crown ethers and the selectivities among metal cations can change with solvent. Third, they have shown high capacity and chemical and thermal stability. Unfortunately, the approaches described above have some inherent disadvantages. First, for columns of the type in Figure 9(a), the flow rate is usually low (0.05–0.1 ml min−1 ) because of the low mechanical rigidity, needed to withstand high pressure, of polymeric crown ether resins. This results in long analysis times. Second, columns like those based on 12 suffer from poor hydrolytic stability due to the potential for acid or base hydrolysis; furthermore, the underlying silica structure can be ionized and thus tends to concentrate eluent as well as analyte ions, resulting in a loss of selectivity. Third, the separation efficiencies of both types of column material have not been

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

12

Techniques

found to be high, likely caused by irregularities in particle size and shape.

NH4 Conductivity

To use crown ethers in the mobile phase is more convenient than in the stationary phase since it avoids the cumbersome work of design, synthesis, and column packing, and the possibility that the macrocyclic analogs used will not display the same selectivity as expected from the parent compound. The main reason for including crown ethers in the mobile phase is their cation selectivity.58–60 By selectively binding one of the analytes, a higher resolution for two analytes having similar selectivities with a traditional ion-exchanger can be achieved. For example, traditional cation-exchangers containing sulfonic or carboxylic functional groups have similar selectivities for NH4 + and Na+ , and thus the resolution between these two peaks is usually low, making quantification difficult. Indeed, in real-world samples, NH4 + and Na+ are typically quite disproportionate, either one or the other having a much higher concentration. By adding a crown ether to the mobile phase, this problem can be overcome.58 Figure 10 compares the separation of cations with and without 18-crown-6 in the eluent using a standard cation-exchange separator column. The stability constants (1 : 1 complex) of 18-crown-6 with NH4 + and Na+ in aqueous solution at 25 ◦ C are 1.23 and 0.80, respectively. After adding crown ether, the retention time of NH4 + increased dramatically, enhancing the resolution between NH4 + and Na+ peaks. In this same study, Bruzzoniti and coworkers58 reported the separation of a large number of cations: NH4 + , alkali (Li+ , Na+ , K+ ), and alkaline earth (Ca2+ , Mg2+ , Ba2+ , Sr2+ ) metal cations. Both monovalent and divalent cations were separated in the same analytical run. The influence of 18-crown-6 on the retention of cations was in the following order: K+  Ba2+ > NH4 + > Sr2+ , as shown in Figure 10. This order is in keeping with the relative stability constants of 18-crown-6 with these cations and illustrates the advantage of using the unsubstituted macrocycle in a well-characterized medium like water: namely, one can be fairly confident that the thermodynamic selectivity will be carried over into the separation. As discussed in the early part of this chapter, macrocyclic complexes are generally more stable in nonaqueous solvents than in aqueous solution. Therefore, it is expected that adding 18-crown-6 to a nonaqueous mobile phase would increase the retention time of cations to be separated. Fritz’s group added 18-crown-6 to a nonaqueous IC mobile phase to study the retention of alkali metal cations and ammonium ion on a sulfonic acid cation-exchange resin.59 The retention factors of all the ions increased with increasing concentration of 18-crown-6 in acetonitrile eluent containing 1 mM methanesulfonic acid. Most notably,

Sr Ba

Li

0

5

(a)

10

15

Minutes

Ca, NH4 Mg Li

Conductivity

4.1.2 Crown ethers in the mobile phase

Ca Mg

K

Na

Na Sr K

0 (b)

5

10

15

20 25 Minutes

Ba

30

35

40

45

Figure 10 Separation of cations (a) without and (b) with the addition of 18-crown-6 ether in the eluent. (Ion charges implicit) Column: IonPac CS12 cation exchange. (Reproduced from Ref. 58.  Wiley-VCH, 2008.)

the separation factor of Li+ /Na+ increased from 2.6 without 18-crown-6 to 3.8 when including 18-crown-6 in the mobile phase, which made the separation of 1 ppm of Li+ from 500 ppm Na+ possible. One interesting study incorporated a crown ether in both the mobile and stationary phases. Specifically, the Dionex Ionpac CS15 column contains covalently bonded 18-crown-6 in the stationary phase and can give good resolution at a 4000 : 1 concentration ratio of Na+ to NH4 + .55 Lamb’s group found that the addition of 18-crown-6 to a mobile phase can improve peak resolution between NH4 + and Na+ even further, so that accurate analysis can be performed at a concentration ratio of as much as 60 000-to-1 using this same column. Figure 11 shows the separation of NH4 + and Na+ at a concentration ratio of 1 : 60 000.60 This kind of resolution is very helpful in the analysis of biological or environmental samples, where sodium ion is commonly present at very high concentrations and can make the analysis of other ions like ammonium ion extremely difficult. Possible concerns over the use of crown ethers in the mobile phase are effects on the suppressor, cost, and toxicity. The ligand indeed has an affect on suppressor performance after days of use, but this can be overcome easily

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations 0.05 Na+

stationary phase, the resulting cation complex serves as a positively charged functional site for anion exchange. Since [2,2,2] complexes with Li+ , Na+ , and K+ cations to significantly varying degrees, the capacity of the column is adjusted by changing the cation present in the mobile phase. In chromatography, gradients are applied in order to shorten the retention times of highly retained species. This is typically achieved by gradually increasing the strength of the eluent. On the other hand, a capacity gradient can be achieved using macrocycle-based columns by switching from one cation with a higher affinity for the macrocycle to another with a lower affinity. In this way, the capacity of the column drops rapidly and late-eluting anions emerge much sooner. Figure 12 compares the elution of a series

Ca2+

Conductivity (mS)

0.04

0.03

NH4+

0.02

0.01

0 0

5

10 Time (min)

15

20

1–9

Figure 11 Separation of Na+ (600 mg l−1 ) and NH4 + (0.010 mg l−1 ) at the concentration ratio of 60 000 : 1; Column: CG15 and CS15; Eluent: 10.80 mM H2 SO4 and 10.0 mM 18-crown-6. (Reproduced from Ref. 60.  Elsevier, 2003.)

with a simple rinse.60 In terms of toxicity and cost, this ligand is less toxic than some solvents and additives commonly used in chromatography and CE60 and no more costly.

14

0

10

20

(a)

30

40

50

Time (min) 1

NaOH

4

2

3

Conductivity

Crown ethers have undergone limited investigation as components in ion chromatographic detection systems. One example is the work of Jane and Shih,61 who coated a piezoelectric quartz crystal with dibenzo-16-crown-5oxydodecanoic acid. The detector was used for cation and anion detection after separation on a diaza-18-crown-6based separator column with nonionic eluents. The frequency response of this detector for both cations and anions, due to cation complexation and anion association with the resulting complex, was as reproducible and sensitive as standard conductimetric detection, but peak broadening resulted from a relatively large cell volume.

LiOH

10–13

4.1.3 Crown ethers in detection systems

5 6

7 8

0

10

20

(b)

9

30

40

50

Time (min) 7 1

Capacity gradient 8

4

9

2 3 5

6

11 12 13 14

10

4.2

13

Cryptands in ion chromatography 0

Modified cryptand structures can be used in the stationary phases of IC either by adsorption to or covalent bonding to hydrophobic resins. Lamb and coworkers introduced the use of decyl-[2,2,2] (D222) in IC stationary phases by adsorbing D222 onto polystyrene-based resin beads and thereby developed the concept of “capacity gradient elution” of anions.62 When an eluent containing cations like Li+ , Na+ , or K+ passes through the D222-coated

(c)

10

20

30

40

50

Time (min)

Figure 12 Separation of 14 common anions under isocratic and capacity gradient conditions: (a) with 20 mM LiOH as eluent (isocratic); (b) with 20 mM NaOH as eluent (isocratic); (c) with a 20 mM NaOH to 20 mM LiOH gradient. Peaks: 1 = F− , 2 = CH3 CO2 − , 3 = Cl− , 4 = NO2 − , 5 = Br− , 6 = NO3 − , 7 = SO4 2− , 8 = C2 O4 2− , 9 = CrO4 − , 10 = I− , 11 = PO4 3− , 12 = phthalate, 13 = citrate, 14 = SCN− . (Reproduced with permission from Ref. 62.  Century Publishing, Inc, 1990.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Techniques 8 −

F MCA, NO2− DCA Cl −

O

−

N

O O

O O

BCA

TCA

3−

PO4

NO3

DBA

MBDCA MCDBA TBA

0 0

5

10

15 Minutes

20

25

Figure 14 Separation of haloacetic acids and inorganic anions. Column: Cryptand A15 µ (150 mm × 3 mm); eluent gradient: −8 to 3 min, 10 mM NaOH; t = 3 min, 10 mM LiOH. Monochlorocacetic (MCA), monobromoacetic (MBA), dichloroacetic (DCA), bromochloroacetic (BCA), dibromoacetic (DBA), trichloroacetic (TCA), monobromodichloroacetic (MBDCA), monochlorodibromoacetic (MCDBA), and tribromoacetic (TBA) acids. (Reproduced from Ref. 64.  Elsevier, 2008.) 25

20

15

10

5

2.5

5

0 7.5

Figure 15 Separation of heparins using a cryptand-based column. (Reproduced from Ref. 66.  Elsevier, 2008.)

O M+

−

Br

0 O

2−

MBA

SO4

Signal/µs

of anions from a D222 column under both isocratic and capacity gradient conditions. One of the primary advantages of capacity gradients over traditional eluent gradients is that with capacity gradients the ionic strength of the eluent need not change, that is, only the identity of the eluent metal cation need be changed, whereas with traditional gradients changes in eluent ionic strength can cause major disturbances in the chromatographic baseline. As mentioned previously, columns prepared by adsorbing macrocyclic ligands to resin beads have obvious drawbacks such as limited eluent choice and potential loss of capacity because the macrocyclic active sites can gradually bleed off. Pohl, Woodruff, and coworkers63 succeeded in overcoming these problems by covalently bonding the [2,2,2] ligand to polystyrene chromatographic beads (13). The baseline separation of polarizable and nonpolarizable anions in the same run by capacity gradient was demonstrated. Figure 13 shows the monomer of cryptand [2,2,2] used to graft to the polystyrene stationary phase. In one application of this method, Sarzanini and coworkers reported the separation of haloacetic acids (HAAs), a class of disinfection by-products.64 The separation of HAAs was traditionally achieved through conventional anion-exchange columns with a fixed capacity using an eluent gradient, from weak to strong eluents. With the cryptand column, a NaOH–LiOH step gradient was used to adjust the column capacity so that the separation of nine HAAs was achieved in 25 min with good resolution (Figure 14). The cryptand column has also been used for quantification of low molecular weight heparins (LMWHs) in bio-samples. Heparin is a highly negatively charged, sulfated glucosaminoglycan, which is an anticoagulant used to prevent thromboembolic diseases, and during kidney dialysis and cardiac surgery. LMWHs are negatively

µs

14

N 13

Cryptand Cryptand Cryptand

Figure 13 The monomer of cryptand [2,2,2] used to graft to an ion chromatographic stationary phase. (Reproduced from Ref. 63.  Elsevier, 2002.)

charged fractionation or depolymerization products of heparin with molecular weight of less than 8 kDa on average.65 Abballe et al.66 illustrated the separation of LMWHs by the cryptand column using a NaOH–LiOH step gradient. LMWHs were eluted as a unique peak, making their quantification easy and fast (Figure 15). In recent years, analysts have investigated the use of cryptand columns for preconcentrating trace amounts of anions in complex matrices.67–69 The general principle of preconcentration and matrix elimination is to use a concentrator column to trap and enrich the trace ions of interest

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations while eliminating all or most of the matrix ions that would interfere in the chromatographic analysis. Following this pretreatment step, chromatography would be carried out as usual.70 This approach is amenable to automation and therefore desirable in the laboratory. Haddad et al.70 have reviewed the instrumentation requirements for preconcentration, including use of the cryptand-based column. Many recent studies have focused on determining trace amounts of perchlorate in drinking water because perchlorate ion can interfere with iodine uptake in humans, resulting in thyroid dysfunction.69 Pohl and coworkers68 have used a cryptand [2.2.2] column to concentrate perchlorate and remove drinking water matrix anions from a large sample by a weak base eluent. Figure 16(b) shows the separation of 10 µg l−1 of perchlorate from 1000 mg l−1 matrix ions using this concentration method followed by a standard anion chromatographic separation and suppressed conductivity detection. This is to be compared to Figure 16(a), in which no concentrator column was used and the perchlorate peak is not properly separated from the huge background matrix ion peak. The US Environmental Protection Agency (EPA) has published accepted methods for perchlorate analysis in drinking water in which these approaches have been carefully optimized69 to allow for reliable determination down to o- nitroaniline. While most liquid membrane separations involve hydrophobic membranes, it is possible to use an aqueous membrane to separate species between organic layers. For example, Armstrong et al.89 reported the use of CDs as membrane carriers used to separate hydrophobic isomers. They found that CD carriers can enhance, inhibit, or reverse the bulk membrane selectivity among different guests. For

(b)

(c) Na

10

Metal ions transported (moles × 104)

Na Li K, Rb, Cs

5

0

K (d)

K (e)

(f)

Na

10

Na

Na 5 Li K

0 0

20

40

60

Li, K Rb, Cs

K, Rb, Cs 0

20 40 Time (h)

60

0

20

40

60

Figure 24 Amounts of metal cations (mol × 104 ) transported into the receiving phase versus time (h) for competitive BLM transport of alkali metal cations (0.20 M in each) by 0.010 M 2 in (a) chloroform, (b) dichloromethane, (c) carbon tetrachloride, (d) 1,2dichloroethane, (e) 1,1,1-trichloroethane, and (f) o-dichlorobenzene. (Reproduced from Ref. 87.  Elsevier, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations example, CD carriers greatly enhance the transport of pnitroaniline over the o- isomer. In the blank membranes containing no CD carrier, trans-stilbene is favored over cis-; however, the transport of cis-isomer is greatly enhanced when CD carriers were included in the membrane. While much of the research using macrocycles in liquid membranes has focused on cation separations, anion separations are also an important field of study. Resorcinarenes provide a promising framework for constructing molecular anion carriers, given that the upper and lower rims can be modified with many functional groups. As shown in Figure 25, Lamb and coworkers have modified the upper rim of resorcinarene with aza-18-crown-690 (27) or bis(pyridylmethyl)amine (BPA)91 (28) and studied their anion separation capabilities in PIMs or BLMs. The focus was on the separation of ReO4 − , a nonradioactive surrogate for pertechnetate, from other anions. Anion transport was studied using the Li+ , Na+ , K+ , and Pb2+ salts of anions in the source phase. The anions were cotransported through the membrane with the complexed cation in order to maintain charge neutrality. The monomer undecyl-aza18-crown-6 (ACU) was compared to the carrier performance of ACR (aza-crown resorcinarene) to examine the O O O

O O

O

O

O

O

O

N

N

O

O

O

O

O

N

O

O O

C11H23

C11H23

C11H23

C11H23

27

(a)

N

O

O

O O

O

O N

O O

O

O

NN

N

N

N

N

N N

N

N

N O O

O

O O O

C11H23

(b)

O O

C11H23

C11H23

C11H23

28

Figure 25 Structures of the resorcinarene derivatives. (a) azacrown resorcinarene (ACR); (b) bis(pyridylmethyl)amine resorci(isoG1)10 Cs+ BPh4 − narene (BPAR). (Reproduced from Refs. 90. and 91.  Elsevier, 2008 and 2006.)

21

effect of the preorganized anion-attractive cavity in the ACR molecule.90 The selectivity of the carrier to anions depended on the type of counter cation. For example, K+ or Na+ or a mixture of these two cations facilitated greater transport of ClO4 − and ReO4 − , while Pb2+ promoted the transport of NO3 − . As described above, PIMs consist of a polymer network, a plasticizer liquid, and a carrier. Typically, CTA is used as the polymer network and o-nitrophenyloctyl ether is used as plasticizer. However, depending on the solubility of the macrocyclic carrier, it may be desirable to use an alternate plasticizer with different properties. Lamb’s group studied the effect on separations of several alternate plasticizers such as ethyl benzoate (EB), 2-ethoxyethyl ester benzoic acid (EEB), dibutylphthalate (DBPT), ethyl phthalyl ethyl glycolate (EPEG), 2-nitrophenyl octyl ether (NPOE), and tris(2-butoxyethyl) phosphate (TBEP) using the resorcinarene-based carrier BPAR (shown in Figure 25). The nitrogen donors of BPA moiety make BPAR good coordination ligands for transition metals.91 The transition metal (Cu2+ , Zn2+ , Fe3+ ) complexes of BPAR were used as the carriers for anion transport through the membranes. The polarity and viscosity of the plasticizers play an important role in anion transport. More polar plasticizers stabilized anions in the solvents, thus promoting the partitioning of anions into the membrane, while higher viscosity inhibited the diffusion of anions and decreased transport. For example, halide transport decreased with decreasing polarity of the plasticizer, except in the case of the fairly polar EPEG, which has the highest viscosity. Among cation separations using macrocyclic membrane carriers, an interesting potential application is the separation of Cs+ from nuclear waste. 137 Cs is responsible for over 40% of the short-term radioactivity from many nuclear waste tank materials. In one novel study, self-assembly to form a macrocyclic structure was employed to effect selective transport of Cs+ from a simulated nuclear waste mixture. Specifically, the Davis group showed that guanosine derivatives like the molecule 5 -(tert-butyldimethylsilyl)2 ,3 -O-isopropylidene isoguanosine (isoG 1) (Figure 26a) can self-assemble to form a hydrogen bonded pentamer (Figure 26b).92 Cai et al.93 first reported the single crystal structure of the (isoG 1)10 Cs+ BPh4 − complex (Figure 27). In this complex, the Cs+ ion is sandwiched between two hydrogen bonded self-assembled isoG 1 pentamers (29). Lamb’s group subsequently made PIMs and BLMs with isoG1 as carrier to examine its selectivity among alkali and alkaline metals. Excellent flux of Cs+ and selectivity over other alkali metal cations was observed.94 In addition, an interesting carrier concentration effect was noticed. Specifically, if the self-assembled pentamer was precomplexed with Cs+ , it performed as an effective Cs+ carrier at all concentrations; but if only isoG 1 monomer was added, with the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

22

Techniques H

N

N Si O O

N

N

O

H

N

H O

O

(a) R H N H N N R

N 6

N

R

H

N

H

N

H

M+

H

O

H

N

N

H

O

O O

H N

N

H

N N H

N

H H

N N

H

N

O

N

R

M+

N N R

H N

H

H

N

N

N

O

H 1N 3 2 N O

Si O

N N

N R N

isoG 1, R = O

isoG pentamer

O 29

(b)

Figure 26 (a) Structure of isoG1. (b) Hydrogen bonding in the self-assembled pentamer. (Reproduced from Ref. 93.  Wiley-VCH, 2000 and Ref. 94.  Springer, 2001.)

(a)

(b)

(c)

Figure 27 Crystal structures of (isoG1)10 Cs+ BPh4 − from top view, space-filling top view, and side view. (Reproduced from Ref. 93.  Wiley-VCH, 2000.)

expectation that it would spontaneously form the cyclic pentamer carrier, the expected flux of Cs+ was obtained only below a certain critical concentration. Above that concentration, no flux was obtained, implying that cyclic pentamer was not forming but rather some other conglomerate that

did not serve as a cation carrier. The selectivity of Cs+ over Na+ in the PIM reached nearly 10 000 : 1 with this carrier. Crown ethers have been modified with many substituent groups to provide specific separation functionalities in liquid membranes. In designing lipophilic carriers for membrane transport of saccharides, Smith’s group bonded 3-(chloromethyl)benzo-15-crown-6 to a boronic acid group to design a crown boronic acid (30), which is the first example of an artificial heterotopic sodium saccharide cotransporter.95 (Figure 28) Arylboronic acid was chosen for its strong Lewis acidity and lipophilicity. Transport of p-nitrophenyl β-D-glucopyranoside through the membrane was found to be pH dependent. At pH 11.0, flux was about three times faster than at pH 6.3. Also, transport of the glucoside was about five times faster with the carrier than without. The glucoside was transported together with Na+ to maintain electrical neutrality. Therefore, to increase the flux of glucoside, both the concentrations of glucoside and Na+ can be increased. Given that counterions must be extracted along with ions of interest into the membrane, efforts have been made to provide macrocyclic structures with the ability to associate directly with ions of both charges. Another example of a heterotopic receptor (31) from Smith’s group is shown in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations

O

O

O

O

O + Na O O O

O O

HO

O OpNP

HO HO HO

+

Na+, OH− CH3

O

23

HO OH B

C O O N CH3 CH3 O CH3

−2H2O

HO HO O OH C −B O O O O OpNP N CH3 O CH3 CH3 CH3

O

30

Figure 28 Cooperative transport of Na+ and glucoside by a crown boronic acid carrier. (Reproduced from Ref. 95.  American Chemical Society, 1995.)

REFERENCES

t -Bu

O

O NH

NH A−

O

O

M+

N O

N O

31

Figure 29 Cooperative inclusion of a cation and an anion in a heterotopic receptor. (Reproduced from Ref. 96.  American Chemical Society, 2004.)

1. L. K. James, Nobel Laureates in Chemistry 1901–1992, Merck & Co., Inc., U.S., 1993, pp. 708–714. 2. Data from SciFinder Scholar, a database of chemical abstracts service (CAS) of American Chemical Society. 3. J. D. Lamb, J. S. Gardner, and K. Gloe, Macrocyclic Chemistry: Current Trends and Future Perspectives, ed. K. Gloe, Springer, Netherlands, 2005, pp. 349–363. 4. J. L. Major, R. M. Boiteau, and T. J. Meade, Inorg. Chem., 2008, 47, 10788. 5. O. P. Lam, P. L. Feng, F. W. Heinemann, et al., J. Am. Chem. Soc., 2008, 130, 2806. 6. J. W. Steed, Coord. Chem. Rev., 2001, 215, 171.

96

Figure 29. This receptor contained an aza crown ring that was designed to include cations, whereas anions may be bound by hydrogen bonding to the amine group to maintain electrical neutrality. This receptor has similar selectivity to that of the monotopic receptor—dicyclohexano-18-crown6 among alkali metal ions (K+ > Na+ > Li+ ); however, the flux obtained with this ditopic receptor was up to 10 times higher than with the monotopic receptor.

7. J. D. Lamb, R. M. Izatt, J. J. Christensen, and D. J. Eatough, Coordination Chemistry of Macrocyclic Compounds, ed. G. A. Melson, Plenum Press, New York and London, 1979, pp. 145–217. 8. J. D. Lamb, R. M. Izatt, C. S. Swain, and J. J. Christensen, J. Am. Chem. Soc., 1980, 102, 475. 9. X. X. Zhang, R. M. Izatt, J. S. Bradshaw, and Krakowiak, Coord. Chem. Rev., 1998, 174, 179.

K. E.

10. J. M. Lehn and J. P. Sauvage, J. Am. Chem. Soc., 1975, 97, 6700. 11. B. Masci and P. Thu´ery, Cryst. Eng. Comm., 2007, 9, 582.

6

CONCLUSIONS

The host–guest selectivity of macrocyclic ligands as measured in homogeneous solution can translate effectively into multiphase separations systems such as IC and liquid membranes, even when macrocyclic structures must be modified to accommodate system demands. Separations scientists have applied this selectivity in novel ways to these two methodologies to effect separations that have potential or realized practical uses, both in analytical chemistry and preparative separations. To date, only a fraction of the macrocyclic structures that exhibit such potential have been studied, and to the degree that this line of research is pursued vigorously, many further innovations can be expected.

12. H. Hope, M. M. Olmstead, P. P. Power, and X. J. Xu, J. Am. Chem. Soc., 1984, 106, 819. 13. R. Ungaro, A. Casnati, F. Ugozzoli, et al., Angew. Chem. Int. Ed., 1994, 33, 1506. 14. V. P. Solov ev, N. N. Strakhova, V. P. Kazachenko, et al., Eur. J. Org. Chem., 1998, 7, 1379. 15. A. Thaler, B. G. Cox, and H. Schneider, Inorg. Chim. Acta, 2003, 351, 123. 16. M. Kodama, E. Kimura, and S. Yamaguchi, Dalton Trans., 1980, 12, 2536. 17. Z. Monsef, G. Rounaghi, and A. Sarafraz, J. Inclusion Phenom. Macrocyclic Chem., 2001, 39, 321. 18. E. Shchori and J. Jagur-Grodzinski, Isr. J. Chem., 1973, 11, 243.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

24

Techniques

20. P. C. Stark, M. Huff, E. A. Babaian, et al., J. Inclusion Phenom. Macrocyclic Chem., 1987, 5, 683.

48. P. Hajos, G. Revesz, O. Horvath, et al., J. Chromatogr. Sci., 1996, 34, 291. ˇ and W. Goessler, J. Chromatogr. A, 49. B. Divjak, M. NoviS, 1999, 862, 39.

21. K. A. Watson, S. Fortier, M. P. Murchie, et al., Can. J. Chem., 1990, 68, 1201.

50. G. Dotsevi, Y. Sogah, and D. J. Cram, J. Am. Chem. Soc., 1976, 98, 3038.

22. A. S. Antsyshkina, G. G. Sadikov, M. A. Porai-Koshits, et al., Russ. J. Coord. Chem., 1994, 20, 274.

51. K. Kimura, H. Harino, E. Hayata, and T. Shono, Anal. Chem., 1986, 58, 2233.

23. R. M. Izatt, R. E. Terry, B. L. Haymore, et al., J. Am. Chem. Soc., 1976, 98, 7620.

52. J. D. Lamb, R. G. Smith, and J. Jagodzinski, J. Chromatogr. A, 1993, 640, 33.

24. K. Ohtsu, T. Kawashima, and K. Ozutsumi, J. Chem. Soc. Faraday Trans., 1995, 91, 4375.

53. B. R. Edwards, A. P. Giauque, and J. D. Lamb, J. Chromatogr. A, 1995, 706, 69.

25. A. F. D. de Namor, M. L. Zapata-Ormachea, O. Jafou, and N. Al Rawi, J. Phys. Chem. B, 1997, 101, 6772.

54. Q. Xu, M. Mori, K. Tanaka, et al., J. Chromatogr. A, 2004, 1023, 239.

26. S. Katsuta, H. Tachibana, and Y. Takeda, J. Solution Chem., 2002, 31, 499.

55. M. A. Rey, C. A. Pohl, J. J. Jagodzinski, et al., J. Chromatogr. A, 1998, 804, 201.

27. E. M. Arnett and T. C. Moriarity, J. Am. Chem. Soc., 1971, 93, 4908.

56. E. Blasius and K. P. Janzen, Pure Appl. Chem., 1982, 54, 2115.

28. P. R. Haddad, Anal. Chem., 2001, 73, 266A.

57. M. Nakajima, K. Kimura, and T. Shono, Anal. Chem., 1983, 55, 463.

19. K. W. Chi, K. T. Shim, H. Huh, et al., Bull. Korean Chem. Soc., 2005, 26, 393.

29. H. Small, T. S. Stevens, and W. C. Bauman, Anal. Chem., 1975, 47, 1801. 30. P. Atkins and J. De Paula, Atkins’ Physical Chemistry, 7th edn, Oxford University Press, New York, 2002, p. 834. 31. H. Small, Ion Chromatography, Plenum Press, New York and London, 1989. 32. P. R. Haddad and P. Jandik, Ion Chromatography, ed. J. G. Tarter, Marcel Dekker, Inc., New York and Basel, 1987, pp. 87–156. 33. P. R. Haddad and R. C. Foley, Anal. Chem., 1989, 61, 1435. 34. J. Weiss and D. Jensen, Anal. Bioanal. Chem., 2003, 375, 81. 35. C. Sarzanini, J. Chromatogr. A, 2002, 956, 3. 36. P. R. Haddad, P. N. Nesterenko, and J. Chromatogr. A, 2008, 1184, 456.

W. Buchberger,

37. D. T. Gjerde, J. S. Fritz, and G. Schmuckler, J. Chromatogr., 1979, 186, 509. 38. K. J. B. A. Karim, J. Y. Jin, and T. Takeuchi, J. Chromatogr. A, 2003, 995, 153. 39. F. G. P. Mullins, Analyst, 1987, 112, 665. 40. K. Hayakawa, T. Sawada, K. Shimbo, and M. Miyazaki, Anal. Chem., 1987, 59, 2241. 41. B. Paull and P. N. Nesterenko, Analyst, 2005, 130, 134. 42. P. Hatsis and C. A. Lucy, Anal. Chem., 2003, 75, 995. 43. P. Zakaria, J. P. Hutchinson, N. Avdalovic, et al., Anal. Chem., 2005, 77, 417. 44. S. D. Chambers, K. M. Glenn, and C. A. Lucy, J. Sep. Sci., 2007, 30, 1628. 45. A. Nordborg and E. F. Hilder, Anal. Bioanal. Chem., 2009, 394, 71.

58. M. C. Bruzzoniti, R. M. De Carlo, and M. Fungi, J. Sep. Sci., 2008, 31, 3182. 59. P. J. Dumont and J. S. Fritz, J. Chromatogr. A, 1995, 706, 149. 60. D. A. Richens, D. Simpson, S. Peterson, et al., J. Chromatogr. A, 2003, 1016, 155. 61. Y. S. Jane and J. S. Shih, Analyst, 1995, 120, 517. 62. J. D. Lamb, P. A. Drake, and K. E. Woolley, Advances in Ion Chromatography, eds, P. Jandik and R. M. Cassidy, Century International, Franklin, MA, 1990, Vol. 2, p. 215. 63. A. Woodruff, C. A. Pohl, A. Bordunov, and N. Avdalovic, J. Chromatogr. A, 2002, 956, 35. 64. M. C. Bruzzoniti, R. M. De Carlo, K. Horvath, et al., J. Chromatogr. A, 2008, 1187, 188. 65. J. F. K. Limtiaco, C. J. Jones, and C. K. Larive, Anal. Chem., 2009, 81, 10116. 66. F. Abballe, A. Lombardi, I. Maccone, et al., J. Pharm. Biomed. Anal, 2008, 48, 467. 67. J. D. Lamb, D. Simpson, B. D. Jensen, et al., J. Chromatogr. A, 2006, 1118, 100. 68. R. Lin, B. De Borba, K. Srinivasan, et al., Anal. Chim. Acta, 2006, 567, 135. 69. H. P. Wagner, B. V. Pepich, C. A. Pohl, et al., J. Chromatogr. A, 2006, 1118, 85. 70. P. R. Haddad, P. Doble, and M. Macka, J. Chromatogr. A, 1999, 856(1–2), 145–177. 71. G. Arena, A. Contino, E. Longo, et al., J. Supramol. Chem., 2002, 2, 521.

46. P. R. Haddad, Anal. Bioanal. Chem., 2004, 379, 341.

72. J. Wang, R. G. Harrison, and J. D. Lamb, J. Chromatogr. Sci., 2009, 47, 510.

47. J. G. Dorsey, W. T. Cooper, B. A. Siles, et al., Anal. Chem., 1998, 70, 591R.

73. N. Thuaud, B. S´ebille, A. Deratani, et al., Chromatographia, 1993, 36, 373.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Ion chromatography and membrane separations

25

74. M. W¨ossner and K. Ballschmiter, Fresenius J. Anal. Chem., 2000, 366, 346.

86. R. A. Bartsch, E. G. Jeon, W. Walkowiak, and W. Apostoluk, J. Membr. Sci., 1999, 159, 123.

75. R. H. Pullen, J. J. Brennan, and G. Patonay, J. Chromatogr. A., 1995, 691, 187.

87. R. A. Bartsch, N. K. Dalley, V. S. Talanov, et al., Tetrahedron, 2005, 61, 8351.

76. Y. Kitamaki and T. Takeuchi, Anal. Sci., 2004, 20, 1399.

88. J. Anzai, Y. Kobayashi, A. Ueno, and T. Osa, J. Chem. Soc. Chem. Commun., 1985, 15, 1023.

77. Y. Kitamaki, J. Y. Jin, and T. Takeuchi, J. Pharm. Biomed. Anal., 2003, 30, 1751. 78. H. Strathmann, AIChE J., 2001, 47, 1077. 79. M. F. San Rom´an, E. Bringas, R. Iba˜nez, and I. Ortiz, J. Chem. Technol. Biotechnol., 2010, 85, 2. 80. R. M. Izatt, G. C. Lindh, R. L. Bruening, et al., Pure Appl. Chem., 1986, 58, 1453. 81. J. J. Christensen, J. D. Lamb, S. R. Izatt, et al., J. Am. Chem. Soc., 1978, 100, 3219. 82. G. Arena, A. Contino, A. Magri, et al., Supramol. Chem., 1998, 10, 5. 83. T. G. Levitskaia, J. D. Lamb, K. L. Fox, and B. A. Moyer, Radiochim. Acta, 2002, 90, 43. 84. W. Walkowiak and C. A. Kozlowski, Desalination, 2009, 240, 186. 85. L. Mutihac, Curr. Drug Discovery Technol., 2008, 5, 98.

89. D. W. Armstrong and H. L. Jin, Anal. Chem., 1987, 59, 2237. 90. J. D. Lamb, C. A. Morris, J. N. West, et al., J. Membr. Sci., 2008, 321, 15. 91. J. S. Gardner, Q. P. Peterson, J. O. Walker, et al., J. Membr. Sci., 2006, 277, 165. 92. J. T. Davis, S. Tirumala, and A. L. Marlow, J. Am. Chem. Soc., 1997, 119, 5271. 93. M. Cai, A. L. Marlow, J. C. Fettinger, et al., Angew. Chem. Int. Ed., 2000, 39, 1283. 94. S. C. Lee, J. D. Lamb, M. Cai, and J. T. Davis, J. Inclusion Phenom. Macrocyclic Chem., 2001, 40, 51. 95. J. T. Bien, M. Y. Shang, and B. D. Smith, J. Org. Chem., 1995, 60, 2147. 96. J. M. Mahoney, G. U. Nawaratna, A. M. Beatty, et al., Inorg. Chem., 2004, 43, 5902.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc037

Dynamic Light Scattering in Supramolecular Materials Chemistry J¨org Braun1 , Kasper Renggli1 , Julia Razumovitch1 , and Corinne Vebert2 1 2

University of Basel, Basel, Switzerland University of Geneva, Geneva, Switzerland

1 Introduction to Dynamic Light Scattering 2 Self-Processes 3 Supramolecular Reactivity and Devices 4 Soft Matter and Nanotechnology 5 Supramolecular Aspects of Chemical Biology 6 Conclusions References

1 1.1

1 3 5 7 10 11 11

INTRODUCTION TO DYNAMIC LIGHT SCATTERING Principle of dynamic light scattering

Light scattering (LS) is a powerful technique for characterizing the structure, structural formation, and interaction of supramolecular systems (see Figure 1 for a schematic representation). Studying the scattering of light by structures with sizes in the submicrometer range allows the determination of critical characteristics such as shape or internal structure. The absolute value of the intensity of the scattered light is monitored in a static light scattering (SLS) mode, whereas instantaneous variations in intensity are recorded by dynamic light scattering (DLS). A combination of these

two LS modes yields complementary thermodynamic and hydrodynamic information such as the molecular weight, size, and shape of the system under investigation.1 In this chapter, we focus on the use of DLS in supramolecular materials science. The detailed theory behind DLS, which is not the topic of this review, is reported in several excellent publications.1–3 Briefly, electromagnetic waves—or photons—interact with the local electron cloud of the analyte. This interaction results in an energy transfer from the electromagnetic wave to the electrons, inducing fluctuations, depending on the polarizability of the analyte. Since any accelerated particle emits light, the electrons reemit photons. This process, extending from the interaction between the incident photon to the reemission of light is called LS, and the analysis of the characteristics of the scattered light yields information about the system under investigation. When a particle is subject to Brownian motion and irradiated, two frequencies of equal intensity are generated in addition to the frequency that would normally be scattered, inducing a positive and a negative Doppler shift proportional to the particle velocity. The interference between the nonshifted wave (photon reemission) and the two waves due to Brownian motion yields infinitesimal variations in intensity. Detection of these is the basic principle of DLS, which is therefore particularly suited to the study of properties of solutions. The scattered intensity is acquired as a function of time and is then self-correlated. This yields the relaxation time due to the Brownian motion and leads to the characterization of the particle size through hydrodynamic models of the diffusion coefficients.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

2

Techniques Cell Laser Bath −

Figure 1 setup.

1.2

Correlator

Amplifierdiscriminator

oto Ph iplier lt mu

Computer storage

Schematic representation of a typical light scattering

Limitations to dynamic light scattering

The intensity of the scattered light monitored over the course of an LS measurement depends on both the concentration and angle of detection. Therefore, to accurately quantify the size of a particle, rather than merely observing relative changes that depend on the composition of the surrounding medium, measurements must be carried out at several concentrations and angles. In the case of small particles S (S = laser wavelength/20, in nanometer), however, these can be represented as single scattering centers; here, the intensity of the scattered light does not exhibit angular dependence. The shape of the particle is assumed to be spherical, and hydrodynamic models of the diffusion coefficient yield the equivalent hydrodynamic size of a sphere of the particle under investigation. The most common model uses the Stokes–Einstein equation, which assumes no intermolecular interactions between small, spherical particles. Concentration-dependent measurements therefore allow extrapolation to infinite dilution for accurate size quantification by DLS. However, few macromolecules are of a size that matches this size criterion. When particles cannot be described as point scattering centers, the scattered light from two different parts along the same molecule interferes constructively or destructively, leading to an angle dependence of the intensity of the scattered light. As in cases of shape anisotropy, for which the autocorrelation function must be corrected, the determination of the size of the molecules is hampered by DLS angle-dependent measurements.

1.3

Alternatives to dynamic light scattering

There are several other techniques complementary to DLS that allow the determination of the diffusion coefficient of supramolecular systems in dilute solution. These include Taylor dispersion analysis (TDA),4 hydrodynamic chromatography (HDC),5 and size exclusion chromatography

(SEC),6 which can all be used if the supramolecular system has a low polydispersity. Mes and coworkers7 compared TDA, DLS, HDC, and SEC and showed that all four methods can be used effectively to determine diffusion coefficients of systems with low polydispersities by measuring a series of styrene acrylonitrile (SAN) copolymers. Although these are polymeric systems, it is possible to apply the findings to supramolecular ensembles. The characterization of samples of low polydispersity was achieved best with TDA and DLS, since they both allow the rapid and absolute determination of the diffusion coefficient. However, TDA has the disadvantage that it is subject to interference due to the presence of low-molecular-mass chromophoric compounds. DLS, on the other hand, is influenced much more by the polydispersity of the sample than TDA. Furthermore, the use of DLS enables direct measurements of the Z-average diffusion coefficient of a polydisperse sample but requires a relatively large amount of the sample and is concentration dependent. Unlike TDA, DLS is especially suited for the analysis of high-molecular-mass systems, such as supramolecular systems, and is not disturbed by the presence of low-molecular-mass impurities. When analyzing polydisperse samples, the question arises as to whether an average diffusion coefficient still has any physical meaning, as the diffusion coefficient distribution for those samples spans an enormous range. Therefore, a diffusion coefficient distribution should be given for polydisperse systems. The chromatographic methods of HDC and SEC have, in theory, the ability to determine a relative diffusion coefficient distribution. The results of Mes indicate that this is indeed the case for SEC. HDC, however, can only be accurately used for samples with low polydispersities.7 Both methods require calibration with well-known standards, and the determination of diffusion coefficients is then based on the assumption that systems with comparable sizes coelute, which, however, might not always be the case. To fully characterize supramolecular systems, it is sometimes advisable to consider complementary techniques. Additional hydrodynamic methods include sedimentation velocity and viscosimetry, both of which provide information on the coarse structural properties of the system. The use of small-angle X-ray scattering (SAXS)8 and smallangle neutron scattering (SANS)9 provides additional information on a much shorter wavelength scale (>0.2 nm) compared with DLS since measurements are performed at very low angles (typically 0.1–10◦ ). Both techniques are capable of achieving structural characterization of partially or fully ordered systems with sizes up to 150 nm and characteristic internal repeat distances between 5 and 25 nm. USAXS (ultra-small angle X-ray scattering) and USANS (ultra-small angle neutron scattering) can resolve

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

DLS in supramolecular materials chemistry

−2.1

Intensity

Light

SAXS

a−1

0.0001

0.001

0.01

−4.0 0.1

K (1/A)

Figure 2 Graphical representation of the scattered light intensity versus the scattering angle.

even larger dimensions. The disadvantage compared to DLS is intrinsic, since far more energy and higher safety arrangements are needed. By combining both the techniques, large supramolecular aggregates can be fully characterized through the determination of the properties of the ensemble (DLS) and internal structure (SANS or SAXS). Schaefer and coworkers10 investigated the combined use of DLS and SAXS for the characterization of colloidal silica. In Figure 2, the power law for the scattered light intensity indicates that the aggregates are fractal objects. At the shorter wavelength regime accessible to SAXS, the power law indicates that the monomer units that make up the aggregate remain intact. Further, local structural changes can be monitored by “point probes,” using methods such as circular dichroism (CD)11 and fluorescence techniques.12

2

SELF-PROCESSES

Self-processes trigger the spontaneous organization of molecules into a higher ordered structure. Since the resulting supramolecular assemblies are composed of several molecules, few conventional techniques can be used to analyze their properties. In this context, DLS is a powerful tool for characterizing the properties of the ensemble. Self-processes are inherent in the self-assembly of copolymers. These are composed of chemically or physically incompatible units along the same macromolecule, such as polar/hydrophobic monomers or rigid/flexible polymer segments, respectively. The most widely investigated self-assembling copolymers are amphiphilic linear block copolymers. The chemical incompatibility between covalently linked hydrophobic and hydrophilic polymer segments drives the organization of the macromolecules into

3

supramolecular structures. The characteristics of the resulting self-assembly, such as size or morphology, depend on several macromolecular properties such as molecular weight, polydispersity, and hydrophobic to hydrophilic volume fraction. In dilute aqueous solution, a delicate balance between steric, attractive, and repulsive forces drives the organization of the molecules into nanosized objects of various shapes such as vesicles or spherical- and rodlike core–shell micelles. Owing to the constant progress in polymer chemistry, a plethora of self-assembling copolymers can be synthesized according to various chemistry routes,13 and several excellent review publications describe the self-assembly of block copolymers.13–18 Aside from linear copolymers, sophisticated copolymer macromolecular architectures are also achievable.13, 16, 19–21 Mai et al.22 synthesized hyperbranched multiarm copolyethers of 3ethyl-3-(hydroxymethyl)oxetane-propylene oxide (PEHOstar-PPO) of different molar ratios of PPO arms to PEHO cores. DLS in combination with transmission electron microscopy (TEM) enabled the determination of PEHOstar-PPO aggregates that self-assembled into large spherical micelles of defined sizes. As the core-to-arm ratio increases, the micelle size decreases. Combining DLS, 1 H NMR (proton nuclear magnetic resonance), and FTIR (Fourier transform infrared) spectroscopy, they showed the presence of large compound micelles resulting from the aggregation of single micelles interacting through hydrogen bonds. Sheikh et al.23 produced self-processed nanoparticles from poly(ε-caprolactone) (PCL) grafted poly(vinyl alcohol) (PVA) copolymer to immobilize both hydrophobic and hydrophilic molecules. The sizes of the core–shell-type nanoparticles, that is, micelles, were analyzed by DLS and TEM. Interestingly, they were able to show that if the particles are loaded with small molecules, for example, drugs, the size of the self-assembled particles increased. However, the pH or solvent composition itself influences the properties of self-organized structures by affecting the intermolecular interactions between the molecules building those systems, held together by noncovalent interactions. In several systems, intermolecular interactions were, therefore, frozen by covalent binding of the self-assembled molecules.24–26 However, Gohy et al.27–29 demonstrated that covalent binding between the hydrophobic and hydrophilic blocks is not mandatory to selfassemble copolymer structures. They showed, using poly(styrene) and poly(ethylene) polymer segments— either covalently linked (PS-b-PEO) or held by a bis(2,2 :6 ,2 -terpyridine)ruthenium(II) complex (PS-[Ru]PEO)—that both resulting copolymers self-assemble into similar morphologies. An alternative to the connection between the polymer segments via metal ions to achieve an effective noncovalent self-assembly is molecular recognition among specially designed ligands and receptors. The

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

4

Techniques

most important molecular receptors studied in supramolecular chemistry are cyclodextrins, which are appealing and versatile structures for the design of inclusion complexes with hydrophobic guests.30–34 Munteanu et al.35, 36 synthesized cyclodextrin-containing compounds to act as molecular receptors as well as supramolecular building blocks. DLS was used to investigate the host–guest interactions and the self-assembly properties of the resulting complexes. The shrinkage of polymer networks held by supramolecular interactions was controlled owing to the dual-complexing ability of the cyclodextrin-containing compound. Other classes of host molecules are crown ethers37–40 and macrocyclic polyphenols.41, 42 The latter have been successfully used for the design of a variety of receptor molecules showing complexation properties. Recent progress in this field is further exemplified by the contribution of “click” chemistry, a powerful strategy that enables the modular assembly of new molecular entities.43–48 Interest in hybrid materials based on block copolymers of natural polymers (peptides and proteins) and synthetic polymers has increased in recent years.49–51 Following from the fact that the properties of the block copolymers can be controlled by both the primary structure of the peptide and the size and properties of the synthetic polymer, these tailor-made block copolymers can self-assemble into novel supramolecular structures, which possess functional properties suitable for several applications such as drug delivery.52, 53 In this context, elastin-like polypeptides are popular due to their ability to mimic the main extracellular matrix protein that is responsible for the reversible elongation and contraction of various tissues in vertebrates.54, 55 The composition of those block copolymers is dominated by small hydrophobic amino acids, such as alanine (A), glycine (G), proline (P), and valine (V). Pechar et al.56 synthesized short polypeptides, (VPGVG)4 and (VPAVG)4 , which were coupled via their N-termini with activated semitelechelic poly(ethylene glycol) O-(N-Fmoc-2-aminoethyl)O  -(2-carboxyethyl)undeca(ethylene glycol) (Fmoc-PEGCOOH) to yield monodisperse Fmoc-PEG-peptide diblock copolymer. Both the presence of the terminal hydrophobic Fmoc group and the hydrophilic poly(ethylene glycol) (PEG) chain in the copolymers were shown to play a crucial role in their self-associative behavior, leading to reversible formation of supramolecular thermoresponsive assemblies. The associative behavior of the peptides and their PEG derivatives was demonstrated by DLS. Similarly, Koga et al.57 investigated the self-assembly of PEG covalently linked to an oligopeptide. DLS revealed that the copolymers form spherical structures of 100–200 nm, the oligopeptide adopting an α-helical structure as proven by CD. Using DLS, Yang et al.58 studied the kinetics of the self-processes underlying the formation of a hybrid hydrogel system

resulting from the organization of hydrophilic synthetic N(2-hydroxypropyl)methacrylamide (HPMA) polymer backbones modified by oppositely charged peptide grafts. The two distinct pentaheptad peptides acted as physical crosslinkers through the formation of antiparallel coiled-coil heterodimers. A pair of random coil peptides was also designed to reveal the role of the coiled-coil grafts. Incubation of peptide-grafted HPMA copolymers, addition of a competing peptide or a denaturant to the self-assembled hydrogels resulted in partial disassembly or collapse of the structure. These results correlated with the changes in the secondary structure of the peptide grafts as measured by CD spectroscopy. Owing to its unique structure and high mechanical performance, spider silk remains an intriguing natural polymer, which still focuses research efforts. Zhou et al.59 investigated a peptide that mimics the repetitive GGX motif of the spidroin silk protein. DLS, supported by AFM (atomic force microscopy) and TEM, showed that, in water, the peptide self-assembled into discrete and stiff nanorods. Adding MgCl2 or CaCl2 , the formation of a mass of nanorod bundles was observed. Evidence that the metal ions and peptide interaction was obtained when ethylenediaminetetraacetic acid (EDTA) was added, which induced the disruption of the bundles. Polymers are the higher molecular weight analogs of surfactants and lipids, which undergo self-processes. Shukla et al.60 presented a detailed analysis of the dynamic properties of entangled semiflexible, thread-like micelles selfprocessed by cationic surfactants such as cetylpyridinium chloride or cetyltrimethylammonium bromide subsequent to the addition of sodium salicylate in aqueous solution. DLS was performed in combination with rheological measurements in order to investigate the dynamic properties of the system. The LS results were consistent with the theoretical model of dynamical coupling between the concentration fluctuations and the dissipative properties of the system. Imura et al.61 investigated the self-process mechanism of “natural” glycolipid biosurfactants using DLS. These can be integral components of tissues, while some are secreted by cells into the growth medium. Some are valuable for their curing properties such as antibiotic, antifungal, or even anticancer agents.62–65 It was observed that mannosylerythritol lipids self-assemble into large unilamellar vesicles and, at higher concentrations, into sponge-like structures with water-channel radii of 50 nm, which is relatively large compared with those obtained with “synthetic” surfactant systems.66 In this regard, unique and complex molecular structures arecomposed of several chiral centers that are known to be molecularly engineered by microorganisms probably evolved into these sophisticated glycolipid biosurfactants self-assemblies. Similar to self-assembling synthetic and natural polymers, hard systems, such as silica-containing systems,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

DLS in supramolecular materials chemistry undergo self-processing. Schiller et al.67 prepared mesoporous silica particulate structures by combining cooperative self-assembly in inverse miniemulsion with a sol–gel process, which is commonly used for the fabrication of materials from a solution of chemical precursors of an integrated network (or gel ) of either discrete particles or polymers. Over the course of the synthesis, cetyltrimethylammonium bromide acted as a lyotropic template, whereas an amphiphilic block copolymer stabilized the droplets. Depending on the reaction time, either porous particles, a few hundred nanometers in size, capsules, or porous flakes were observed and analyzed by DLS. By varying the amount of block copolymer, the morphology and mesostructural ordering, as well as the porosity, can be tuned. Naik and Sokolov68 described the synthesis of micrometer-sized nanoporous and mesoporous silica spheres using disodium trioxosilicate (as an economical silica source) and cetyltrimethylammonium chloride as the structure-directing agent. The mechanism of the sphere formation was studied by DLS and further supported by AFM. Two stages were found: in a first step, 20–30-nm size primary particles were formed, which then fused together to form micrometer-sized mesoporous spheres. Using DLS, Aerts et al.69 carefully analyzed the synthesis of the silicalite-1 zeolite, which is synthesized from tetrapropylammonium hydroxide, tetraethylorthosilicate, and water. The kinetics of the reaction revealed two diffusive processes and polydispersity in size and shape arising from a slower process. The faster process corresponds to collective particle diffusion. The self-diffusion coefficient provided a way to estimate the equivalent hydrodynamic radius. These observations revealed a complex, polydisperse mixture of particles present at the onset of the zeolite formation, which should be considered when modeling the zeolite assembly process.

3 3.1

SUPRAMOLECULAR REACTIVITY AND DEVICES Supramolecular reactivity

In contrast to the interactions between at least two molecules being referred to as molecular reactivity, supramolecular reactivity refers to interactions of an ensemble of molecules, which are held together by noncovalent interactions (mainly London forces). Therefore, several stimuli influence the properties of supramolecular systems and their reactivity. As DLS is particularly suited to the characterization of growth or reduction in size of a reacting assembly, some examples of supramolecular systems for which this technique was used to analyze the supramolecular reactivity of the ensemble are given in the following.

5

A most sophisticated design of constituent molecules is required to achieve precisely controlled and well-defined supramolecular systems that can respond to external stimuli such as pH or temperature. Since the information determining structure formation and interaction of the resulting assembly is encoded in their molecular configuration, selfassembling copolymers, in particular, address this major goal. Such amphiphilic block copolymers are composed of at least a permanently hydrophilic block and a “smart” block, that is, either hydrophilic or hydrophobic, with inherently tunable properties for stimuli responsiveness. Lowe et al.70 described the characteristics of a doubly responsive diblock copolymer composed of N-isopropylacrylamide (NIPAM) and 4-vinylbenzoic acid (VBZ) monomers. They demonstrated that the diblock copolymer can form normal and inverse micelles in aqueous environments by taking advantage of the stimuli-responsive characteristics of both building blocks to pH and temperature. For this purpose, 1 H NMR spectrometry was used to assign the hydration of the charged block of the polymer depending on the pH. In combination with DLS, it was shown that, when raising the temperature to 40 ◦ C (above the lowest critical solution temperature of the NIPAM block71 ) while maintaining the pH of the solution at 12, supramolecular selfassembly results in nanosized species that are composed of a hydrophobic NIPAM core stabilized by a hydrophilic VBZ corona. Conversely, lowering the solution pH to 2 at ambient temperature resulted in the formation of aggregates in which the VBZ block forming the core is now hydrophobic, stabilized by the hydrophilic NIPAM block. However, NMR (nuclear magnetic resonance) provides an indirect proof of supramolecular aggregation, whereas DLS yields the size and size distribution of the supramolecular system. Figure 3 clearly shows the dimensions of the two different supramolecular structures, which are schematized in Figure 4. Cao et al.72 synthesized poly(N-isopropylacrylamide-covinyl laurate) copolymers. The copolymerization of PVL (poly(vinyl laurate)) and PNIPAM (poly(N-isopropylacrylamide)) resulted in a reduced and broadened lower critical solution temperature (LCST) of the copolymer solutions compared with a solution of PNIPAM alone, which facilitates the formation of hydrophobic microdomains far below the LCST, causing a pronounced aggregation in solutions. Time-dependent DLS measurements demonstrated that the temperature-induced transition of the copolymers is divided into three stages: pre-LCST aggregation, coil–globule transition at the LCST, and post-LCST aggregation. The direct synthesis of functional, self-assembling copolymers in aqueous media under mild conditions without protection/deprotection chemistry steps, however, remains a current challenge that focuses research efforts on developing stimuli-responsive delivery systems such as micelles

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

6

Techniques

30

Mean (vol.%)

25 20

pH 12 T = 25 °C

pH 2 T = 25 °C

15 D h = 15.8 nm

10 D h = 66.7 nm

5 0 1

10

100

1000

Hydrodynamic diameter, D h (nm)

Figure 3 Hydrodynamic size distribution of self-assembled poly(NIPAM-block -VBZ) block copolymers at pH 12 and T = 25 and 50 ◦ C.70

and vesicles. Owing to their intrinsic sensitivity to the properties of the surroundings and their ability to undergo specific interactions such as receptor–ligand recognition, natural blocks such as oligonucleotides,73 polypeptides, and polysaccharides are being investigated as constitutive segments of self-assembling block copolymers. Zhao et al.74 synthesized pH-responsive block copolymers composed of PNIPAM and poly(L-lysine) (PLL). DLS was combined with SEM (scanning electron microscope) to analyze the effect of the length of PLL, pH, and temperature on the LCST of PNIPAM as well as on the resulting properties of the self-assembly of these PLL-based copolymers. They showed that PNIPAM-b-PLL copolymers self-assembled into micelle-like aggregates, where PNIPAM is hydrophobic at acidic pH and at high temperatures. In contrast, when the self-assembly of the PLL-based copolymers is induced in basic media and at low temperatures, they self-assemble into structures in which the PLL is the hydrophobic block. Since, with polymers, a relatively long part of the macromolecule needs to be influenced in order to change

the supramolecular reactivity, Lee et al.75 synthesized short amphiphilic calixarene molecules with a small hydrophilic part consisting of amine ethers or amine alcohols. Combining DLS and TEM, they were able to demonstrate the formation of vesicles and observed different self-association behaviors by merely changing a short part of the hydrophilic segment, which was introduced in the last step of the synthesis. The calixarene molecules with a small hydrophilic part assembled into well-defined and tunable vesicles that decrease significantly in diameter with increasing hydrophilic chain length. Further increasing the chain length induced the collapse of the vesicles into spherical micelles. The vesicles were also observed transforming into small globular micelles at lower pH, which can be used to trigger the release of an encapsulated hydrophilic guest molecule in a device to transport hydrophilic molecules and release them with such a pH trigger. Dendrimers have attracted increasing attention in recent years because of their unique structure, interesting properties, as well as their potential applications in medicine, catalysis, gene therapy, and nanoreactors.76–83 Poly(amidoamine) (PAMAM) dendrimers are monodisperse, highly branched polyelectrolytes with ammonium functional groups at the outer rim (primary amine) and at the branch points (tertiary amine). Wang et al.84 showed the spontaneous supramolecular complexation of amine-terminated PAMAM dendrimers with an anionic surfactant, sodium dodecyl sulfate (SDS). They observed a strong electrostatic interaction between the SDS and the amine termination of the PAMAM dendrimers when the pH was below 7.4. The binding of SDS induced a physical hydrophobic modification of the outer surface of the PAMAM dendrimer, which promoted the formation of PAMAM/SDS supramolecular complexes. Using DLS, they were able to show a decrease in the size of the supramolecular aggregates with decreased pH and SDS concentration. These results were further confirmed by isothermal titration calorimetry (ITC) measurements, which—in addition to the size and aggregation behavior identified by DLS—yielded the thermodynamic properties of the system, such as binding energies between the SDS and the dendrimer.

VBZ Low pH

High T

Low T

High pH NIPAM

Figure 4 media.

Proposed structure formation of poly(NIPAM-block -VBZ) diblock copolymer in normal and inverse micelles in aqueous

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

DLS in supramolecular materials chemistry In addition to pH, solvent composition influences supramolecular assembly as well as reactivity. Polyoxometalates (POMs) belong to a distinct class of oligomeric, structurally well-defined multinuclear complexes with interesting optic, electric, and magnetic properties relevant to many fields.85, 86 Rare earth-substituted, especially lanthanidesubstituted, POMs can offer unique functionality to POMs, such as excellent luminescence properties. These POMs have found applications in various fields including catalysis, medicine, or material sciences.87–92 Mishra et al.93 studied the self-assembly and formation of “blackberry” type supramolecular structures by a type of yttrium-containing POM (K15 Na6 (H3 O)9 [(PY2 W10 O38 )4 (W3 O14 )]·9H2 O or {P4 Y8 W43 }) macroanions using DLS. {P4 Y8 W43 } macroions were found to form hollow, spherical, single-layer “blackberry” structures in water and water–acetone mixed solvents. The blackberry size lent itself to accurately control either by changing the acetone content in water–acetone mixed solvents or by changing the solution pH in aqueous solution. The blackberry size increased with decreasing pH (lower charge density) or higher acetone content in the mixed solvent (lower dielectric constant), whereas the blackberry size changed in response to the change of external conditions.

3.2

Supramolecular devices

The preparation of nanoscale architectures is critical to the advances in the development of supramolecular devices, that is, ultraminiature machine components, which are functional at the nanoscale. In the following section, examples of such supramolecular devices characterized by DLS are given. Release of compounds from delivery systems is currently a major field of research,94–97 especially in medical applications, for which the release of active compounds is needed at desired locations. Mueller et al.98 used DLS to assess the successful embedding of a hydrophobic substrate into the hydrophobic membrane of block-copolymer vesicular structures, without affecting the morphology of the self-assembly. Poly(butadiene)-b-poly(ethylene oxide) self-assembles into unilamellar vesicles in dilute aqueous solution. A constant hydrophobic shell thickness was measured by DLS when the average hydrodynamic radius and size distribution depended on the nature of the encapsulated hydrophobic substrate. Stoikov et al.99 synthesized novel p-tert-butyl thiacalix[4]arenes functionalized with hydrazide groups at the lower rim of the cone and the partial cone in the 1,3-alternate conformations. The affinity of the selfassembly to p-(Al3+ , Pb2+ ) and d-(Fe3+ , Co3+ , Ni2+ , Cu2+ , Ag+ , Cd2+ ) block elements was investigated by

7

DLS. They showed that the p-tert-butyl thiacalix[4]arenes functionalized with hydrazide groups are effective extractants of soft metal cations. The complex stoichiometry depended on the receptor configuration. All the p-tert-butyl thiacalix[4]arene derivatives with hydrazide fragments were able to form nanoscale aggregates but did not show selfassociation abilities without complex formation with metal cations. Another application of supramolecular devices is in the field of diagnostics. Magnetic resonance imaging (MRI) has become one of the most efficient diagnostic techniques in recent years. Contrast results from local differences in spin relaxation time along the longitudinal (T1 ) and transverse (T2 ) planes of the main magnetic field applied to a specimen. Imaging agents need to be used in order to improve contrast. In current medical diagnostics, the most frequently used contrast agents are T1 agents, and among these, due to its high number of unpaired electrons, Gd3+ , in its complex Gd-diethylenetriaminepentaacetate (Gd-DTPA), is the most utilized. By assembling complexes with macromolecules100–102 or supramolecular aggregates,103–105 the rotation of the complex is slow enough to enhance the relaxation value, which increases the contrast. Vaccaro et al.106 studied in detail a supramolecular system with potential application as a tumor-specific MRI contrast agent using DLS. The system was composed of mixed aggregates formed by an anionic monomer containing the DTPAGlu, a derivative of DTPA and an uncharged monomer containing the bioactive peptide C-terminal cholecystokinin octapeptide (CCK8) amide, capable of aiming the assembly at tumor cells. Mixed aggregates, formed either by the monomer or by the DTPAGlu derivative associated with the chelating agent or in its complex form, have been investigated. A number of features of the aggregation behavior were revealed by DLS and, in particular, the effect of electrostatic interactions on intra- and interaggregate electrostatic repulsions on the aggregation mechanism. The high number of gadolinium complexes, each of them characterized by a high relaxation value (r1p = 21.0 mM−1 s−1 ), allows very effective paramagnetic contrast agents to be obtained compared with micelles formed by first-generation monomers.104 Micelles and bilayer structures were detected at physiological pH, whereas on decreasing solution pH or increasing ionic strength, the formation of bilayer structures was favored.

4

SOFT MATTER AND NANOTECHNOLOGY

DLS is not only a powerful technique for investigating supramolecular systems but also a very versatile method

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

8

Techniques

in other fields such as nanotechnology and soft matter. DLS has advantages over other methods, such as SANS and SAXS in characterizing soft matter, in that the DLS equipment is relatively inexpensive, does not emit harmful radiation, and does not require special solvents. Working on soft matter systems usually raises important issues that can be answered by the use of DLS. One such issue is the analysis of polydispersity. Another is the investigation of charged systems and polyelectrolyte solutions such as DNA. In this section, we cover some of these aspects to demonstrate how broad the field of applicability of DLS is. Particle size and particle size distribution (PSD) of dispersed systems affect key properties such as surface area, reactivity, opacity, packing density, and rheological properties. Quantifiable changes in particle size or PSD provide valuable indications of aggregation or dissociation phenomena, which can help predict the stability and a range of macroscopic properties of colloidal systems. When analyzing polydisperse samples, one may wonder whether an average diffusion coefficient still has any physical meaning, as the diffusion coefficient distribution for those samples spans an enormous range. Therefore, for polydisperse samples, a diffusion coefficient distribution should be given. Mes and coworkers7 compared different methods to analyze the polydispersity of a polymer sample by measuring a series of practically representative SAN copolymers. They showed that, by the use of DLS, it is possible to directly obtain the Z-average diffusion coefficient of a polydisperse polymer sample. DLS is therefore specially suited to the analysis of high-molecular-mass polymers and is not disturbed by the presence of low-molecular-mass impurities. In a more recent study, K¨atzel et al.107, 108 investigated a polydisperse fractal system of pyrogenic silica. The interpretation of DLS data for pyrogenic silica is of great relevance for technical applications but is hampered by the fractal structure of the aggregates. They determined the radius of gyration and the fractal dimension from the scattered intensity patterns of SLS and X-ray scattering. The hydrodynamic aggregate radii were obtained from multiangle DLS. Their DLS results showed a strong angular dependence caused by influences of the rotational diffusion. A normalized plot of the scattered intensity by all nonporous silica grades led to a master curve. Such behavior had been predicted by numerical simulations.108 This shows that the polydispersity of the primary particles in the silica grades does not influence the comparability of structural and diffusional analysis of pyrogenic silica, as both mean values show good correlation. This chapter is a good example of depolarized DLS measurements used to directly obtain information on rotational diffusion of polydisperse fractal systems. Working with more complex systems usually present additional factors that must be considered in the correlation

function when analyzing the data. An especially interesting topic is the use of DLS with nonspherical or noncompact objects, in which internal modes of particle mobility, such as polymer segment motion, contribute to the correlation function and thereby to the particle sizing. These internal modes are not only important for particle sizing but can also provide further interesting information concerning the sample characteristics. In their publication, Galinsky and Burchard109–111 showed a method of separating contributions stemming from internal modes and particle self-diffusion to the decay rate of the correlation function by using the length scale from the LS experiment. They investigated potato starch, which had been degraded in an alcohol suspension by adding different amounts of concentrated HCl. The degraded starch was studied in an aqueous solution of 0.5 M NaOH as a mimic of randomly hyperbranched polymeric nanoparticles. All samples contained the large, highly branched amylopectin, which rules out the properties of the investigated starch solutions.112 Their results indicated angular dependence of the apparent diffusion coefficient when the particles possess some internal flexibility. The branching units probably influence the relaxation processes. The motions of the branching points are coupled with the spring–bead relaxations of a linear chain and can have a noticeable effect on the dynamics of the short chains connecting the two branching units. Furthermore, they can alter the common Zimm–Rouse spectra.113 The effect was only noticeable if the branching density was high; that is, the interconnecting chains between the two branching points were short. This example illustrates that DLS is not just a tool for particle sizing by measuring the diffusion coefficient but also yields detailed information about the complex internal dynamics of systems deviating from the simple case of solid spheres, such as flexible linear or branched polymer samples, or rod-like colloidal particles. Soft matter often deals with charged systems and these systems, in particular, provide a critical challenge for LS experiments. Characterizing charged particles in solution implies that the particle interactions caused by Coulomb forces must be suppressed. The electrostatic interactions, either repulsive or attractive, affect chain conformation and influence the diffusion coefficient. An additional factor exists in solutions of finite concentrations of charged molecules and may increase the diffusion speed—repulsive electrostatic force. Because of this effect, data analysis yields a final result for the particle size, which is below the real value. In these cases, the presence of salts in the buffer can screen the intermolecular interactions between the charged molecules. Measurements in solutions with gradient salt concentrations can therefore be helpful in finding the optimum salt concentration to screen electrostatic interactions and therefore slow down the diffusion speed and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

DLS in supramolecular materials chemistry reveal the actual particle sizes. The study of these interactions involving polyions, counterions, and coions has attracted much interest, and there has been considerable theoretical114, 115 and experimental116, 117 work in this area. Borsali et al. described the effect of charged interactions on DLS results for the interesting case of DNA in aqueous solution. In this study they investigated the structure and dynamics of DNA solutions in the dilute and semidilute range of concentrations at different temperatures, with and without the addition of salt, combining SANS with SLS and DLS. High-molecular-weight calf thymus DNA was dissolved in TE buffer, and the approximate length of DNA was determined by agarose gel electrophoresis to be ∼400 bp. Any residual NaCl in solution was removed by dialysis of the DNA solution against double-distilled water. The other solutions at different salt concentrations (5, 10, 100, 250, 500, and 1000 mM NaCl) were prepared from the stock, salt-free system by adjusting the salt content. The Coulomb interparticle interactions were screened by the addition of salt, which made possible a determination of the single particle relaxation process from the polyelectrolyte slow mode. This slow mode is a second-relaxation process found in many DLS experiments performed with polyelectrolyte samples, and it is still not generally understood.118 The DLS experiments carried out with DNA at different salt concentrations showed that, in addition to the two well-documented modes (cooperative and slow), there is an ultraslow mode, suggesting the existence of large aggregates, which can precipitate if too much salt is added to the particle solution (salting out). These aggregates have been directly observed by other authors119 using cryoelectron microscopy. In the study of Borsali and coworkers, a special focus on their investigation of hydrophobic polyelectrolytes is of particular relevance, since it may precisely demonstrate the existence of the above-mentioned aggregates. Zimbone and coauthors120 described the combination of DLS and SLS as well as SEM to characterize the conformation of DNA molecules over a wide range of molecular weights. They presented the data from DLS measurements of DNA macromolecules with chain lengths varying between 700 and 115 000 bp. It was shown that, in standard buffer solutions, DNA behaves as a worm-like chain for lengths below 10 000 bp. However, the data discussed by the authors are not in good agreement with the results obtained with fluorescence spectroscopy. This fact can be explained by the attachment of the fluorescent probe to the DNA molecule, affecting the diffusivity of the macromolecule, which is thus not comparable with those obtained by LS measurements performed with nonmodified DNA molecules. One major problem in DLS measurements is the contamination of the sample solution by dust particles. This must be avoided at all costs due to their strong scattering

9

intensity. In most cases, this can be resolved by simple filtration and sometimes followed by a centrifugation step. However, with certain experiments with rather large samples or with samples that adsorb to the filter membrane, it is important to find an alternative route to remove dust. Ruf described a different approach to overcome this problem. This method treats the dust particle analytically in order to allow particle sizing even for nonfiltered samples.121 The suggested procedure was tested with octaethylene glycol dodecyl monoether, a surfactant that forms small micelles in aqueous solution. After filtration and centrifugation, a second population of particles with sizes approximately one order of magnitude larger than the micelles remained in the solution. It can be assumed that these particles are by-products of the synthesis of the surfactants. Since the amount of dust present in the system could be adjusted by the total duration of centrifugation, these larger particles served as model dust. Dust contribution to DLS signals was taken into account by extending the usual scheme by two additional parameters, an offset in the amplitude autocorrelation function and an offset in the intensity autocorrelation function. The normalized intensity and the normalized field autocorrelation function are then related by a quadratic equation, which simplifies the Siegert relation in the case when dust particles are not present. Thus, by a relatively simple technique, it was possible to deal with the unwanted dust contributions, which is mainly interesting if physical removal of the dust is not possible. Ali et al.122 studied the interaction between an amphiphilic drug, amitriptyline hydrochloride (AMT), and neutral polymers, poly(vinylpyrrolidone) (PVP) and PEG, using DLS. They showed that AMT interacted more strongly with PVP than with PEG. This was indicated by a large decrease of the aggregate size with an increase in the AMT concentration, which is a sign of a sequential collapse of the polymer conformation. The partial negative charge of the oxygen atom of the amide group present along the polymer backbone and the cationic head group of the drug were found to be responsible for the contraction of the drug–polymer complex. The results obtained from this study indicate a way to use DLS as a tool for the design of a drug delivery system. In another study, Destremaut and coworkers123 established an on-line DLS setup to measure colloidal sizes in a pressure-driven microfluidic flow system. Theoretical arguments were given to underline the difficulties inherent in performing such measurements under flow in a microchannel, due to the Poiseuille flow that induces a shear-dependent decorrelation term.124 They constructed a specific DLS setup around a microfluidic poly(dimethylsiloxane) (PDMS)-based chip. This setup enables an estimation of the size of Brownian scattering centers flowing in microchannels, thus validating the theoretical estimations experimentally. The formation of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

10

Techniques

electrostatic coacervates in a solution of oppositely charged nanoparticles and block copolymers in the microfluidic channel were also studied. Hence, this miniaturized device dedicated to DLS measurements is a promising tool to study dynamics on the subsecond timescale. This tool might be particularly interesting for investigating nanoparticle synthesis, phase transition, or, more generally, fast dynamics in soft matter systems.

5

SUPRAMOLECULAR ASPECTS OF CHEMICAL BIOLOGY

Even if DLS does not provide the visualization of the particles under investigation, it appears to be an extremely valuable method for the characterization of biological systems in combination with microscopy techniques such as TEM or SEM, analytical ultracentrifugation, SAXS, and SANS. This noninvasive technique is fast, accurate, and simple to perform, and the sample can be recovered for subsequent analyses. DLS is an established technique for protein characterization. For example, to find optimum protein crystallization conditions, DLS is used as a prescreening tool to identify critical parameters of the process such as the hydrodynamic size, polydispersity, and aggregation factors. DLS is also used as a method to characterize the hydrodynamic properties of globular proteins in electrolyte solutions. However, applications of DLS to characterize charged protein solutions, just as any other charged molecules in solution, could face certain difficulties in subsequent data treatment.125, 126 Proteins can be considered as a complex arrangement of polypeptide chains, which organize in a unique threedimensional structure. This structure is stabilized by a fine balance between hydrogen bonds and electrostatic and hydrophobic interactions providing a large degree of flexibility to the system. The hydrodynamic size of proteins is thus defined by the molecular weight and arrangement of the polypeptide chains. However, several factors can influence this functional structure, and pH, temperature, or ionic strength might affect the molecular conformation along with a subsequent change in shape and size. Thus, monitoring the hydrodynamic size of a protein is one way of observing its stability under native or denaturing conditions. Modern devices have appeared, in which DLS analysis is combined with microplate reading or chromatography. In contrast to conventional biomedical techniques, which need powerful near-infrared lasers (830 nm), these newly developed techniques use He–Ne lasers (red light, 633 nm). In most cases, these commercial systems for particle sizing operate at a single detection angle (90◦ ) of the scattered light by the protein solution. The optimal use of these

modern devices is therefore hampered by the limitations of DLS (Section 1.2), and attention must be paid to the refractive index of plastic plates that might differ for different kinds of plastics, since this could significantly influence the results. Markossian and coauthors127 studied the thermal denaturation and aggregation of rabbit muscle glyceraldehyde3-phosphate dehydrogenase (GAPDH) using differential scanning calorimetry (DSC), DLS, and analytical ultracentrifugation. The DLS data supported a previously proposed mechanism of protein aggregation that involves a first stage of the formation of the start aggregates followed by their subsequent coalescence.128 A method of size estimation of the start aggregates has been developed based on the construction of graphs of LS intensity versus hydrodynamic radius. The results directly indicate that the initial stage of thermal aggregation of the proteins is the stage of start aggregate formation. The hydrodynamic radius of the start aggregates remained constant in a certain temperature range and was independent of the protein concentration. Petta et al.129 presented a DLS study of the phase separation of the ocular lens emerging on cold cataract development. The intensity of the autocorrelation functions of the lens protein was analyzed with the aid of two methods, providing information on the population and dynamics of the scattering elements associated with cold cataracts. The authors demonstrated that the process arises from the matching of light wavelengths with the spatial dimensions of fluctuations. There are two molecular mechanisms that explain the increase in the intensity of scattered light by cataractous protein lenses. The first is aggregation through the formation of high-molecular-weight aggregates, and the second is the phase separation of the lens cytoplasm into protein-rich and protein-poor phases. However, there is still a lack of systematic methods other than DLS data analysis in cataract diagnostic, resulting in the absence of a reliable parameter for the cataract index. On the basis of the meticulous data analysis described in this chapter, the authors demonstrated the robustness of the potential use of DLS as a noninvasive technique for cataract diagnosis. There are several types of experiments for which DLS is already well established as a technique.130–133 The increase in size and dynamics of aggregate formation can be investigated by DLS by monitoring the kinetics of the size distribution, which increases when both the protein aggregates and the monomers are present in the experimental probe at the same time. Since the scattered intensity is proportional to the molecular radius to the power of six, the results of particle sizing are shifted toward the size of the aggregate, which can thus be detected at very low concentration. The thermal stability of proteins was measured in this way by DLS.134–137 In this case, the scattering intensity, and therefore the size of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

DLS in supramolecular materials chemistry particles, is monitored as a function of temperature. Heatinduced melting of protein causes molecule denaturation and subsequent aggregation, which increases the scattering intensity. Interactions of proteins in solution can be investigated by DLS as well.138–140 Assembly and fibrillation of amyloid proteins are believed to play a critical role in various human diseases, including Alzheimer’s, Parkinson’s, Huntington’s, and type II diabetes. Liu et al.141 studied the changes in secondary, tertiary, and quaternary structures of amyloid proteins. DLS, TEM, and terahertz absorption spectroscopy demonstrated a transformation of the native insulin structure from the α-helix into the β-sheet structure prior to aggregation. As shown by thioflavin T fluorescence binding, the bulk formation of mature aggregates is subsequent to a midpoint transition of the aggregation of insulin. These findings are expected to provide a new way of considering amyloid fibrillation and threat related diseases. Owen et al.142 have used DLS to study the self-assembly of a microbial siderophore, which is one in the A–E marinobactin series that facilitates Fe(III) acquisition by the bacteria through coordination of Fe(III) with the marinobactin head group. Depending on the concentration of Fe(III), different morphologies, from micelles to vesicles, self-assembled from the siderophore. The different sizes and shapes of the systems were easily assessed by DLS and supported by SANS. Wilson143 described DLS as a useful tool to study protein crystallogenesis, which is the process of the conversion of individual protein molecules into a macroscopic crystal in solution. It was shown that DLS and SLS could serve as tools for monitoring the thermodynamics and kinetics of protein crystal growth. However to distinguish between protein assemblies and monomers, the peak of scattering intensities should be resolved, with this normally happening when the particle size of one population is, on average, two times larger than the other. In most cases, the assembly of two proteins of the same size does not yield aggregates with a size that is twice that of the monomers because of the changes of molecular conformation. For example, for globular proteins, a twofold increase in the particle size means a sixfold increase of the molecular weight. Since an increase in the polydispersity index reflects changes in PSD, monitoring the polydispersity index instead of the hydrodynamic radius can be more powerful in the case of such an analysis.

6

CONCLUSIONS

Our initial aim for this chapter was to review the basic principle of LS with reference to excellent publications on the theoretical background. Aside from demonstrating the broad field of applicability of DLS, we wish to highlight the major limitations of the technique to prevent misuse.

11

Prior knowledge of the size, polydispersity, or existing interactions in the system under investigation as well as careful sample preparation to avoid the presence of dust are essential for accurate particle sizing. The use of DLS, in the wide field of supramolecular materials chemistry, was then exemplified through nonexhaustive yet representative examples. Prior to the description of the use of DLS to monitor the reactivity of supramolecular systems, we referenced conventional self-processed systems at first. With the self-assembly of copolymers being an active field of research, we have exemplified the use of DLS in characterizing the properties of supramolecular structures resulting from the assembly of synthetic and hybrid copolymers as well as supramolecular and natural polymers. Owing to their biological function and current use in the medical field, the self-assembly of low-molecular-weight analogs of amphiphilic block copolymers, that is, surfactants and lipids, was also cited. However, DLS is not only a powerful technique for characterizing soft matter but is also relevant for the investigation of hard systems as shown by the characterization of silica-based ensembles. In this context, DLS appears to be valuable in following the kinetics of chemical reactions and assessing the reactivity of the supramolecular systems. The changes in size in self-assembled structures from either synthetic or hybrid copolymers in response to changes in temperature, pH, or solvent composition of the surroundings, or on complex formation can be assessed by DLS. The use of DLS further demonstrates the functional supramolecular assembly of devices with a high potential for application as ultraminiaturized machine components that function on the nanoscale, such as for target release or extraction. DLS was also shown to be particularly suited to characterize mechanisms involved in biological systems, such as protein crystallization and aggregation, being the latter often responsible for human diseases, for which the achievement of a comprehensive understanding of the underlying biological mechanism would enable the diagnosis and development of successful treatments.

REFERENCES 1. W. Sch¨artl, Light Scattering from Polymer Solutions and Nanoparticle Dispersions, Springer, Berlin, Heidelberg, 2007. 2. K. S. Schmitz, An Introduction to Dynamic Light Scattering by Macromolecules, Academic Press, 1990. 3. W. Brown, Dynamic Light Scattering: The Method and Some Applications, Oxford University Press, Oxford, 1993. 4. G. Taylor, Proc. R. Soc. London, Ser. A: Math. Phys. Sci., 1953, 219(1137), 186–203.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

12

Techniques 5. J. Bos and R. Tijssen, in Journal of Chromatography Library, ed. E. R. Adlard, Elsevier, 1995, vol. 56, pp. 95–126. 6. A. C. van Asten, R. J. van Dam, W. T. Kok, et al., J. Chromatogr., A, 1995, 703(1–2), 245–263. 7. E. P. C. Mes, W. T. Kok, H. Poppe, and R. Tijssen, J. Polym. Sci., Part B: Polym. Phys., 1999, 37(6), 593–603. 8. O. Glatter and O. Kratky, Small Angle X-Ray Scattering, Academic Press, London, 1982. 9. B. Jacrot, Rep. Prog. Phys., 1976, 39(10), 911–953.

10. D. W. Schaefer, J. E. Martin, P. Wiltzius, and D. S. Cannell, Phys. Rev. Lett., 1984, 52(26), 2371. 11. H. M. Keizer and R. P. Sijbesma, Chem. Soc. Rev., 2005, 34(3), 226–234. 12. N. Bruns, K. Pustelny, L. M. Bergeron, et al., Angew. Chem. Int. Ed., 2009, 48(31), 5666–5669. 13. S. Forster and T. Plantenberg, Angew. Chem. Int. Ed., 2002, 41(5), 689–714. 14. G. Riess, Prog. Polym. Sci., 2003, 28(7), 1107–1170. 15. I. W. Hamley, Angew. Chem. Int. Ed., 2003, 42(15), 1692–1712. 16. M. Antonietti and S. Forster, Adv. Mater., 2003, 15(16), 1323–1333. 17. T. P. Lodge, Macromol. Chem. Phys., 2003, 204(2), 265–273. 18. S. B. Darling, Prog. Polym. Sci., 2007, 32(10), 1152–1204. 19. H. A. Klok and S. Lecommandoux, Adv. Mater., 2001, 13(16), 1217–1229. 20. A. W. Bosman, R. Vestberg, A. Heumann, et al., J. Am. Chem. Soc., 2003, 125(3), 715–728. 21. S. Aoshima and S. Kanaoka, Chem. Rev., 2009, 109(11), 5245–5287. 22. Y. Y. Mai, Y. F. Zhou, and D. Y. Yan, Macromolecules, 2005, 38(21), 8679–8686. 23. F. A. Sheikh, N. A. M. Barakat, M. A. Kanjwal, et al., J. Mater. Sci.: Mater. Med., 2009, 20(3), 821–831. 24. R. Kishore, A. Jofre, J. B. Hutchison, et al., Micro- and Nanotechnology: Materials, Processes, Packaging, and Systems III , Bellingham, 2007, p. 6415. 25. K. L. Wooley, J. Polym. Sci., Part A: Polym. Chem., 2000, 38(9), 1397–1407. 26. Q. Zhang, E. E. Remsen, and K. L. Wooley, J. Am. Chem. Soc., 2000, 122(15), 3642–3651. 27. J. F. Gohy, B. G. G. Lohmeijer, S. K. Varshney, and U. S. Schubert, Macromolecules, 2002, 35(19), 7427–7435. 28. J. F. Gohy, H. Hofmeier, A. Alexeev, and U. S. Schubert, Macromol. Chem. Phys., 2003, 204(12), 1524–1530. 29. O. Regev, J. F. Gohy, B. G. G. Lohmeijer, et al., Colloid Polym. Sci., 2004, 282(4), 407–411. 30. R. Villalonga, R. Cao, and A. Fragoso, Chem. Rev., 2007, 107(7), 3088–3116. 31. C. J. Easton, Pure Appl. Chem., 2005, 77(11), 1865–1871. 32. J. Storsberg, H. Ritter, H. Pielartzik, and L. Groenendaal, Adv. Mater., 2000, 12(8), 567–569.

33. S. Choi and H. Ritter, E-Polymers, 2007, 45. 34. A. Harada, J. Polym. Sci., Part A: Polym. Chem., 2006, 44(17), 5113–5119. 35. M. Munteanu, S. Choi, and H. Ritter, J. Inclusion Phenom. Macrocycl. Chem., 2008, 62(3–4), 197–202. 36. M. Munteanu, S. Choi, and H. Ritter, Macromolecules, 2009, 42(12), 3887–3891. 37. O. A. Fedorova, E. Y. Chernikova, Y. V. Fedorov, et al., J. Phys. Chem. B, 2009, 113(30), 10149–10158. 38. M. Lee, D. V. Schoonover, A. P. Gies, et al., Macromolecules, 2009, 42(17), 6483–6494. 39. S. J. Li, F. H. Huang, C. Slebodnick, et al., Chin. J. Chem., 2009, 27(9), 1777–1781. 40. A. Konig and J. Adams, Macromol. Symp., 2008, 275, 197–203. 41. C. D. Gutsche, Calixarenes revisited, in Monographs in Supramolecular Chemistry, ed. J. F. Stoddart, RSC, London, 1998. 42. C. D. Gutsche, Calixarenes, in Stoddart, J. F., Monographs in Supramolecular Chemistry, ed. J. F. Stoddart, RSC, London, 1989. 43. V. V. Rostovtsev, L. G. Green, V. V. Fokin, and K. B. Sharpless, Angew. Chem. Int. Ed., 2002, 41(14), 2596–2599. 44. P. Mobian, J. P. Collin, and J. P. Sauvage, Tetrahedron Lett., 2006, 47(28), 4907–4909. 45. S. Chittaboina, F. Xie, and Q. Wang, Tetrahedron Lett., 2005, 46(13), 2331–2336. 46. M. Ortega-Munoz, J. Morales-Sanfrutos, F. PerezBalderas, et al., Org. Biomol. Chem., 2007, 5(14), 2291–2301. 47. T. R. Chan, R. Hilgraf, K. B. Sharpless, and V. V. Fokin, Org. Lett., 2004, 6(17), 2853–2855. 48. F. Perez-Balderas, M. Ortega-Munoz, J. MoralesSanfrutos, et al., Org. Lett., 2003, 5(11), 1951–1954. 49. J. Kopecek, Eur. J. Pharm. Sci., 2003, 20(1), 1–16. 50. M. Pechar, P. Kopeckova, L. Joss, and J. Kopecek, Macromol. Biosci., 2002, 2(5), 199–206. 51. C. Wang, R. J. Stewart, and J. Kopecek, Nature, 1999, 397(6718), 417–420. 52. M. R. Dreher, D. Raucher, N. Balu, et al., J. Controlled Release, 2003, 91(1–2), 31–43. 53. R. Herrero-Vanrell, A. C. Rincon, I. T. Molina-Martinez, et al., Invest. Ophthalmol. Vis. Sci., 2003, 44, 4441. 54. S. A. Jensen, B. Vrhovski, and A. S. Weiss, J. Biol. Chem., 2000, 275(37), 28449–28454. 55. M. Miao, C. M. Bellingham, R. J. Stahl, et al., J. Biol. Chem., 2003, 278(49), 48553–48562. 56. M. Pechar, J. Brus, L. Kostka, et al., Macromol. Biosci., 2007, 7(1), 56–69. 57. T. Koga, K. I. Kitamura, and N. Higashi, Science and Engineering Review of Doshisha University, 2006, 47(3), 185–191. 58. J. Yang, K. Wu, C. Konak, and J. Kopecek, Biomacromolecules, 2008, 9(2), 510–517.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

DLS in supramolecular materials chemistry 59. Q. H. Zhou, J. Lin, F. Yuan, et al., Curr. Nanosci., 2009, 5(4), 457–464. 60. A. Shukla, R. Fuchs, and H. Rehage, Langmuir, 2006, 22(7), 3000–3006. 61. T. Imura, N. Ohta, K. Inoue, et al., Chem.—Eur. J., 2006, 12(9), 2434–2440.

13

86. C. L. Hill, Chem. Rev., 1998, 98(1), 1–2. 87. M. T. Pope and A. M¨uller, Chem. Rev., 1998, 98(1), 239–272. 88. V. A. Grigoriev, D. Cheng, C. L. Hill, and I. A. Weinstock, J. Am. Chem. Soc., 2001, 123(22), 5292–5307. 89. H. C. Aspinall, J. L. M. Dwyer, N. Greeves, Organometallics, 1998, 17(9), 1884–1888.

et al.,

62. J. I. Arutchelvi, S. Bhaduri, P. V. Uppara, and M. Doble, J. Ind. Microbiol. Biotechnol., 2008, 35(12), 1559–1570.

90. H. C. Aspinall, Chem. Rev., 2002, 102(6), 1807–1850.

63. D. Kitamoto, H. Isoda, and T. Nakahara, J. Biosci. Bioeng., 2002, 94(3), 187–201.

91. S. Kobayashi and M. Kawamura, J. Am. Chem. Soc., 1998, 120(23), 5840–5841.

64. S. C. Lin, J. Chem. Technol. Biotechnol., 1996, 66(2), 109–120.

92. X. L. Wang, Y. H. Wang, C. W. Hu, and E. B. Wang, Mater. Lett., 2002, 56(3), 305–311.

65. S. S. Cameotra and R. S. Makkar, Curr. Opin. Microbiol., 2004, 7(3), 262–266.

93. P. P. Mishra, J. Jing, L. C. Francesconi, and T. B. Liu, Langmuir, 2008, 24(17), 9308–9313.

66. L. Porcar, W. A. Hamilton, and P. D. Butler, Langmuir, 2003, 19(26), 10779–10794.

94. Y. C. Wang, L. Y. Tang, T. M. Sun, et al., Biomacromolecules, 2008, 9(1), 388–395.

67. R. Schiller, C. K. Weiss, J. Geserick, et al., Chem. Mater., 2009, 21(21), 5088–5098.

95. F. Wang, Y. C. Wang, L. F. Yan, and J. Wang, Polymer, 2009, 50(21), 5048–5054.

68. S. P. Naik and I. Sokolov, Solid State Commun., 2007, 144(10–11), 437–440.

96. F. Ahmed and D. E. Discher, J. Controlled Release, 2004, 96(1), 37–53.

69. A. Aerts, L. R. A. Follens, M. Haouas, et al., Chem. Mater., 2007, 19(14), 3448–3454.

97. W. X. Chen, X. D. Fan, Y. Huang, et al., React. Funct. Polym., 2009, 69(2), 97–104.

70. A. B. Lowe, M. Torres, and R. Wang, J. Polym. Sci., Part A: Polym. Chem., 2007, 45(24), 5864–5871.

98. W. Mueller, K. Koynov, K. Fischer, molecules, 2009, 42(1), 357–361.

71. A. J. Convertine, B. S. Lokitz, Y. Vasileva, et al., Macromolecules, 2006, 39(5), 1724–1730.

99. I. I. Stoikov, E. A. Yushkova, I. Zharov, et al., Tetrahedron, 2009, 65(34), 7109–7114.

72. Z. Q. Cao, W. G. Liu, P. Gao, et al., Polymer, 2005, 46(14), 5268–5277.

100. S. Laus, A. Sour, R. Ruloff, et al., Chem.—Eur. J., 2005, 11(10), 3064–3076.

73. N. Cottenye, F. Teixeira, A. Ponche, et al., Macromol. Biosci., 2008, 8(12), 1161–1172.

101. A. Dirksen, S. Langereis, B. F. M. de Waal, et al., Chem. Commun., 2005, (22), 2811–2813.

74. C. W. Zhao, X. L. Zhuang, C. L. He, et al., Macromol. Rapid Commun., 2008, 29(22), 1810–1816.

102. E. Toth, R. D. Bolskar, A. Borel, et al., J. Am. Chem. Soc., 2005, 127(2), 799–805.

75. M. Lee, S. J. Lee, and L. H. Jiang, J. Am. Chem. Soc., 2004, 126(40), 12724–12725. 76. Y. H. Kim and O. W. Webster, J. Am. Chem. Soc., 1990, 112(11), 4592–4593. 77. G. R. Newkome, C. N. Moorefield, G. R. Baker, et al., Angew. Chem. Int. Ed. Engl., 1991, 30(9), 1176–1178. 78. A. M. Naylor, W. A. Goddard, G. E. Kiefer, and D. A. Tomalia, J. Am. Chem. Soc., 1989, 111(6), 2339–2341. 79. J. Haensler and F. C. Szoka, Bioconjug. Chem., 1993, 4(5), 372–379.

et al.,

Macro-

103. E. Nakamura, K. Makino, T. Okano, et al., J. Controlled Release, 2006, 114(3), 325–333. 104. A. Accardo, D. Tesauro, P. Roscigno, et al., J. Am. Chem. Soc., 2004, 126(10), 3097–3107. 105. C. Glogard, G. Stensrud, R. Hovland, et al., Int. J. Pharm., 2002, 233(1–2), 131–140. 106. M. Vaccaro, A. Accardo, G. D’Errico, et al., Biophys. J., 2007, 93(5), 1736–1746. 107. U. K¨atzel, M. Vorbau, M. Stintz, et al., Part. Part. Syst. Char., 2008, 25(1), 19–30.

80. D. A. Tomalia, Prog. Polym. Sci., 2005, 30(3–4), 294–324.

108. U. K¨atzel, R. Bedrich, M. Stintz, et al., Part. Part. Syst. Char., 2008, 25(1), 9–18.

81. W. Chen, D. A. Tomalia, and J. L. Thomas, molecules, 2000, 33(25), 9169–9172.

Macro-

109. G. Galinsky and W. Burchard, Macromolecules, 1995, 28(7), 2363–2370.

82. L. Sun and R. M. Crooks, J. Phys. Chem. B, 2002, 106(23), 5864–5872.

110. G. Galinsky and W. Burchard, Macromolecules, 1996, 29(5), 1498–1506.

83. D. Cakara and M. Borkovec, Croat. Chem. Acta, 2007, 80(3–4), 421–428.

111. G. Galinsky and W. Burchard, Macromolecules, 1997, 30(22), 6966–6973.

84. C. Wang, E. Wyn-Jones, J. Sidhu, and K. C. Tam, Langmuir, 2007, 23(4), 1635–1639.

112. J. D. Fox and J. F. Robyt, Carbohydr. Res., 1992, 227, 163–170.

85. S. Q. Liu, D. G. Kurth, B. Bredenkotter, and D. Volkmer, J. Am. Chem. Soc., 2002, 124(41), 12279–12287.

113. B. H. Zimm and R. W. Kilb, J. Polym. Sci., Part B: Polym. Phys., 1996, 34(8), 1367–1390.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

14

Techniques

114. J. Skolnick and M. Fixman, Macromolecules, 1977, 10(5), 944–948.

130. C. Follmer, F. V. Pereira, N. P. Da Silveira, C. R. Carlini, Biophys. Chem., 2004, 111(1), 79–87.

115. M. Le Bret, J. Chem. Phys., 1982, 76, 6243. 116. M. Drifford and J. P. Dalbiez, J. Phys. Chem., 1984, 88(22), 5368–5375.

131. F. A. Bettelheim, R. Ansari, Q. F. Cheng, and J. S. Zigler Jr, Biochem. Biophys. Res. Commun., 1999, 261(2), 292–297.

117. S. F¨orster, M. Schmidt, and M. Antonietti, Polymer, 1990, 31(5), 781–792.

132. J. Schuler, J. Frank, W. Saenger, and Y. Georgalis, Biophys. J., 1999, 77(2), 1117–1125.

118. M. Sedl´ak, J. Chem. Phys., 1996, 105, 10123.

133. V. Militello, C. Casarino, A. Emanuele, et al., Biophys. Chem., 2004, 107(2), 175–187.

119. P. Wissenburg, T. Odijk, P. Cirkel, and M. Mandel, Macromolecules, 1995, 28(7), 2315–2328. 120. M. Zimbone, P. Baeri, M. L. Barcellona, et al., Int. J. Biol. Macromol., 2009, 45(3), 242–247. 121. H. Ruf, Langmuir, 2002, 18(10), 3804–3814. 122. M. S. Ali, G. Ghosh, and D. Kabir ud, Colloids Surf. B,: Biointerfaces, 2010, 75(2), 590–594. 123. F. Destremaut, J.-B. Salmon, L. Qi, and J.-P. Chapel, Lab Chip, 2009, 9(22), 3289–3296. 124. R. T. Foister and T. G. M. V. D. Ven, J. Fluid Mech., 1980, 96(1), 105–132. 125. B. I. Ipe, A. Shukla, H. Lu, et al., ChemPhysChem, 2006, 7(5), 1112–1118. 126. J. Rubin, A. San Miguel, A. S. Bommarius, and S. H. Behrens, J. Phys. Chem. B, 2010, 114(12), 4383–4387. 127. K. A. Markossian, H. A. Khanova, S. Y. Kleimenov, et al., Biochemistry, 2006, 45(44), 13375–13384. 128. H. A. Khanova, K. A. Markossian, B. I. Kurganov, et al., Biochemistry, 2005, 44(47), 15480–15487. 129. V. Petta, N. Pharmakakis, G. N. Papatheodorou, and S. N. Yannopoulos, Phys. Rev. E: Stat. Nonlin. Soft Matter Phys., 2008, 77, 061904.

and

134. K. A. Markossian, I. K. Yudin, and B. I. Kurganov, Int. J. Mol. Sci., 2009, 10(3), 1314–1345. 135. B. I. Kurganov, Biochemistry (Moscow), 2002, 67(4), 409–422. 136. B. I. Kurganov, E. R. Rafikova, and E. N. Dobrov, Biochemistry (Moscow), 2002, 67(5), 525–533. 137. M. Vedadi, F. H. Niesen, A. Allali-Hassani, et al., Proc. Natl. Acad. Sci. U.S.A., 2006, 103(43), 15835–15840. 138. A. D. Hanlon, M. I. Larkin, and R. M. Reddick, Biophys. J., 2010, 98(2), 297–304. 139. M. Latina, L. T. Chylack Jr, P. Fagerholm, et al., Invest. Ophthalmol. Vis. Sci., 1987, 28(1), 175–183. 140. W. F. Tan, L. K. Koopal, and W. Norde, Environ. Sci. Technol., 2009, 43(3), 591–596. 141. R. Liu, M. X. He, R. X. Su, et al., Biochem. Biophys. Res. Commun., 2010, 391(1), 862–867. 142. T. Owen, R. Pynn, J. S. Martinez, and A. Butler, Langmuir, 2005, 21(26), 12109–12114. 143. W. W. Wilson, J. Struct. Biol., 2003, 142(1), 56–65.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc039

Brewster Angle Microscopy Keith J. Stine University of Missouri—St. Louis, St. Louis, MO, USA

1 Introduction: The Role of Brewster Angle Microscopy in Supramolecular Chemistry of Monolayers 1 2 Principles and General Applications of Brewster Angle Microscopy 2 3 Application of Brewster Angle Microscopy to the Study of Monolayers of Macrocyclic Compounds 11 4 Application of Brewster Angle Microscopy to the Study of Monolayers of Chiral Compounds 20 5 Application of Brewster Angle Microscopy to the Study of Monolayers in which Hydrogen-Bond Complex Formation Occurs at the Water Surface 24 6 Emerging Developments in the Application of Brewster Angle Microscopy 26 7 Conclusion 27 References 28

1

INTRODUCTION: THE ROLE OF BREWSTER ANGLE MICROSCOPY IN SUPRAMOLECULAR CHEMISTRY OF MONOLAYERS

The study of monolayers of amphiphilic compounds at the water–air interface has spanned many decades. Much of the early work on monolayers has been summarized in the monograph by Gaines.1 The early work on monolayers that elucidated the basic thermodynamic behavior of singlechain amphiphiles, lipids, sterols, and other molecules set a

foundation of the understanding of basic monolayer behavior that could be applied as the intense interest in studying the behavior of supermolecules at the water–air interface arose over the past two decades. Spreading monolayers at the water–air interface provides a two-dimensional environment in which the packing and interactions of supermolecules can be studied and in which the mean molecular separation between molecules can be systematically varied and their interaction with different species present in the subphase studied. While measurements of surface pressure–mean molecular area (π –A) isotherms, surface potential–mean molecular area isotherms (V–A), and other surface-pressure-related phenomena such as collapse and relaxation at constant area are necessary to characterize a monolayer system, the revitalization of monolayer science that began in the 1980s was driven by the emergence of new spectroscopic and imaging methods that could be applied to monolayers in situ at the water–air interface. Amongst these new methods have been fluorescence microscopy, infrared reflection absorption spectroscopy (IRRAS), grazing incidence X-ray diffraction (GIXD), and Brewster angle microscopy (BAM). The revelation that monolayers exhibited a rich and varied microstructure of coexisting domains in the micronsto-millimeter size range was first revealed by the application of fluorescence microscopy to monolayers of phospholipids2, 3 and subsequent application to monolayers of fatty acids and their esters.4 The fluorescence microscopy technique requires the introduction of a small percentage of a fluorophore-labeled lipid or amphiphile to provide contrast between coexisting phases. The orientation of the fluorescent probe has also been exploited successfully in many cases to image regions of varying amphiphile tilt orientation relative to the interface.5 The method of fluorescence microscopy has limitations with respect to monolayer studies of supermolecules; for example, it is unlikely that

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

2

Techniques

fluorescent probe lipids would partition in a meaningful way in a monolayer of a macrocyclic amphiphile. Fluorescence microscopy does not lend itself to studies of the change in monolayer morphology upon interaction with compounds such as guest molecules injected into the subphase. The method is also not suitable for studies of monolayers formed by adsorption from the subphase, or of monolayers of molecules with structures very different from the classical hydrocarbon chain amphiphiles and lipids, or of spread monolayers of polymers or biomacromolecules. However, the method is still broadly used for studies of phospholipid monolayers in particular as it is less expensive than the present state-of-the-art Brewster angle microscopes and provides superior lateral resolution compared to all except the latest state-of-the-art BAM instruments. BAM was introduced in the early 1990s by the groups of H¨onig and Meunier.6, 7 The method is applicable to adsorbed or spread monolayers of any type of surface-active species. It can be used to image changes in microstructure due to collapse, multilayer formation, or complexation with species from the subphase beneath the monolayer. BAM is a broadly useful tool for studying supramolecular phenomena at the water–air interface. The information provided by BAM is also useful in characterizing monolayers that will then be transferred to solid supports and become Langmuir–Blodgett films (see Langmuir–Blodgett Films, Techniques). In this chapter, we first present the basic principles of BAM. The full mathematical derivations for describing the optical reflectivity observed in BAM are not presented. It is assumed that the reader will most likely make use of a commercially available BAM and thus detailed instrument plans have not been presented. In this chapter, examples are presented from the application of BAM to the study of monolayers of macrocyclic compounds, monolayers in which supramolecular complex formation occurs, monolayers of chiral amphiphiles, and others. Each general class of compounds studied by BAM provides a useful insight into how the technique can be applied in different situations. The chapter does not review the extensive BAM studies conducted on single-chain fatty acids, fatty esters, and alcohols related to phase diagram determination. The chapter also does not review the numerous studies of monolayers of membrane lipids carried out using BAM. However, examples from these two fields of study are used to illustrate specific applications of BAM. The general aspects of monolayer preparation and phase behavior are not reviewed here in detail and readers with no background in monolayers should consult other reviews first. Given that there are many hundreds of papers reporting use of BAM, referencing is not comprehensive. Some new, emerging applications of BAM are described and some general insights summarized at the end.

2

2.1

PRINCIPLES AND GENERAL APPLICATIONS OF BREWSTER ANGLE MICROSCOPY Basic principle of Brewster angle microscopy

The basic principle of BAM is that for p-polarized light incident on the interface between air and water; there is a specific angle at which the reflectivity displays a sharp minimum.8 If the air–water interface was truly perfect and flat, then at the Brewster angle, the reflectivity would be zero; however, the presence of capillary waves and a density transition zone gives the result of a very low but nonzero reflectivity. The reflectivity detected in BAM, represented by R = IR /I0 , where I0 is the incident light intensity and IR is the reflected light intensity, is typically near 10−6 in the vicinity of the Brewster angle. The Brewster angle, θ B , is determined by the condition θ B = tan−1 (n1 /n0 ), where n1 is the refractive index of the subphase and n0 = 1.00 is the refractive index of air. The refractive index of the aqueous subphase will depend weakly on temperature, the wavelength of the incident light, and the presence of other components in the subphase. Using a reported refractive index9 of 1.33 211 for water at T = 20 ◦ C and λ = 632.8 nm, one obtains a value of θ B = 53.10◦ . Laser beams are collimated well enough to provide illumination at a precise angle and minimal reflectivity. Many BAMs use a He–Ne laser at λ = 632.8 nm, but other wavelengths such as the 514.5 nm line from argon ion lasers, and others, have also been used. When an organic monolayer or thin film is present on the water surface, its effect is to introduce a thin layer of different refractive index, typically with n ∼ 1.45; this change in the refractive index results in violation of the Brewster angle condition and an enhancement in reflectivity. The increase in reflectivity due to the presence of an organic layer depends upon its thickness, refractive index, and whether it is isotropic and is described by a single value of the refractive index or whether the film is anisotropically ordered and must be described by a 4 × 4 dielectric tensor.10 The optical theory behind BAM is closely associated with that used to describe the ellipsometry of multilayered thin films on nonmetallic substrates.11

2.1.1 Reflectivity of the water–air interface The reflectivity is calculated using the Fresnel reflection coefficient rp , which is defined as Erp /Eip , where Erp and Eip are the complex amplitudes of the electric vectors of the reflected and incident p-polarized light respectively.11 Values may also be calculated for rs = Ers /Eis , the reflection

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy coefficient for s-polarized light. The expressions for these reflection coefficients are given by the following equations in which θ 0 is the angle of the incident beam measured from the normal direction and θ 1 is the angle of the refracted beam relative to the normal direction: tan(θ 0 − θ 1 ) tan(θ 0 + θ 1 ) − sin(θ 0 − θ 1 ) rs = sin(θ 0 + θ 1 )

rp =

(1) (2)

The reflectivity Rp = |rp |2 for p-polarized light and Rs = |rs |2 for s-polarized light. The starting point for the derivation of (1) and (2) makes use of Maxwell’s equations and requires conservation of the electric and magnetic components of the incident, reflected, and refracted beams. The angles θ 0 and θ 1 are related by Snell’s law. When θ 0 + θ 1 = 90◦ , meaning that the angle of incidence and angle of refraction are perpendicular, (1) becomes zero as the denominator approaches infinity. Using the condition θ 0 + θ 1 = 90◦ together with Snell’s law, n0 sin θ 0 = n1 sin θ 1 , will result in the condition for the Brewster angle, tan θ B = n1 /n0 . Figure 1 shows the variation of reflectivity with angle for illumination of the water–air interface by both p-polarized and s-polarized radiation calculated using (1) and (2).

2.1.2 Reflectivity of an isotropic monolayer The treatment of reflectivity from a single isotropic layer on top of a subphase in ambient can be treated by describing the system in terms of the refractive index of 100

Reflectivity (R)

10−1

s -polarized

10−2 10−3 10−4

the ambient n0 , the refractive index n1 , and thickness d1 of the layer, and the refractive index now denoted as n2 for the subphase. The reflected beam is now the sum of the waves reflected from the ambient/layer interface and from the layer/substrate interface.11 A monolayer at the water–air interface is subject to the thin-layer approximation in that d  λ, where λ is the wavelength of the incident light. The thin-layer approximation is described further in Section 2.9 and is the basis for quantitative applications of BAM to estimations of film thickness.

2.1.3 Reflectivity of an anisotropic monolayer at the water–air interface When significant anisotropy arises due to a structural feature of the monolayer such as long-range tilting of the molecular orientation or an anisotropic lattice structure, the monolayer can no longer be described by a single value of refractive index as the interaction with the incident light now depends on the orientation of the molecules with respect to the plane of incidence. The consequences for reflectivity in BAM measurements have been considered in a number of publications for tilted phases of hydrocarbon chain bearing amphiphiles such as fatty acids or phospholipids; these are described in more detail in Section 2.5. The calculations are complicated and are not presented here. The reflectivity of the p-polarized light varies with the magnitude of the tilt angle and its direction and is calculated in terms of two refractive index values, no and ne , the ordinary and extraordinary values, as well as the tilt angle and the azimuthal angle relative to the plane of the incident light. Addition of an analyzer to the BAM introduces a strong dependence of Rp on the analyzer angle that allows for good contrast in images based on anisotropy, whereas the variation in reflected intensity without an analyzer due to anisotropy is much weaker.9 Calculations of how the reflectivity varies with orientation and analyzer angle have been presented using a 4 × 4 matrix approach10, 11 ; in this study, anisotropy is also reported at the glass–air interface in a Langmuir–Blodgett film in which the molecules form an orthorhombic lattice.11

p -polarized

10−5 10−6

3

2.2 20

40

60

General features of the Brewster angle microscopy experiment

80

Incident angle (q0)

Figure 1 Reflectivity as a function of incident angle for both s-polarized (blue line) and p-polarized light (red line) as a function of angle at the water–air interface for a refractive index of n = 1.332 for water (20 ◦ C, λ = 632.8 nm) and n = 1.000 for air. The curves are calculated using (1) and (2) in the text together with Snell’s law.

The Brewster angle microscope is positioned over the monolayer trough such that the incident laser beam hits the part of the water surface where the monolayer is present at all stages during the experiment. The surface pressure (π) and (if used) surface potential (V ) transducers also need to be positioned in this segment of the trough. The trough itself may be either commercial or home built; in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

4

Techniques

either case, the space to accommodate the BAM must be considered. The BAM and trough should be on a heavy secure table such as a large balance table for best image quality and reduction of vibrations; a vibration isolation table can be used if available. Positioning the trough onto an XY translation stage allows observation of different regions of the monolayer, which is of interest as monolayers are often heterogeneous. Without the ability to laterally translate either the trough or the BAM, one must rely on viewing whatever regions of the monolayer drift beneath the laser spot and this may not be fully representative of the range of structures present. The refracted beam will hit the trough bottom, which is generally Teflon and will scatter light that can contribute to the background brightness in the image. This refracted beam should be blocked, and this can be done with a piece of black glass resting on the bottom of the trough at a small inclined angle. The alignment of the BAM is critical in terms of both the angle of incidence and the adjustment of the polarizer. Most BAMs can adjust the incident angle with a resolution of 0.001◦ . The water surface should be illuminated with p-polarized light oriented in a plane perpendicular to the surface. The refractive index of water varies slightly over the temperature range of interest for most monolayer studies (10–40 ◦ C). The refractive index of water for λ = 632.8 nm has been reported as n = 1.33 282 at 10 ◦ C (θ B = 53.12◦ ) and n = 1.32 972 at 40 ◦ C (θ B = 53.06◦ ). The refractive index variation with subphase composition is also significant if experiments are carried out on buffers or electrolyte solutions of various concentrations. Optimization can be made by viewing the image from just the aqueous subphase and making small adjustments in the angle of incidence until the lowest reflectivity is obtained. Recent commercial BAMs carry out these alignments automatically via software control. Figure 2 shows the basic components of a Brewster angle microscope. BAM is useful for assessing the cleanliness of the water surface prior to the spreading of a monolayer. If Incident beam

CCD camera Analyzer

Polarizer

Subphase

q0

impurities are present, they often appear as fluid streak-like domains caused by greasy dirt and the water surface appears as if a monolayer was already present. Dust and particulates on the water surface show up as brightly reflecting spots that indicate a need to clean the surface. Repeated surface aspiration and cleaning of the trough should result in a featureless water surface under BAM that is then ready for spreading of the monolayer. A fringe pattern may be visible due to the inhomogeneous profile of the laser beam. These observations compliment traditional tests of water surface purity in which the barrier is run forward to a high compression ratio, and it is checked if the surface pressure increases tangibly according to some criterion such as registering no surface pressure change for a 5 : 1 compression. Running the barrier forward across a clean water surface should also result in the observation of no features by BAM. The images from the BAM experiment can be recorded and stored as video files in a computer. The setup will either have a stand-alone video monitor, or the BAM images can be viewed in real time on the computer screen depending on what software is available. The new commercial BAMs come with software for direct image acquisition and processing. These newer BAMs are especially powerful compared to older models such as the BAM-1plus in that they correct for the loss of focus at the edges of the images caused by imaging at an angle. In earlier designs, the distortion of the aspect ratio in the image due to imaging at an angle was corrected by tilting the CCD chip inside the camera. A home-built design that removed the problem of the images being in good focus only along a strip across the middle of the image was introduced; this required the use of a specially designed and custom-fabricated objective.12 The incorporation of the objective in a home-built BAM provided real-time imaging with a resolution of 1 µm and images in focus across the entire image plane. A highly compact design for a home-built BAM in which the entire instrument is oriented vertically has also been introduced as it is especially suitable for positioning in a trough used for Langmuir–Blodgett deposition.13 A low-cost, compact design constructed from standard optical components has also been reported.14

Objective

2.3

q1

Figure 2 The basic geometry and components of a Brewster angle microscopy experiment.

Selection of instrumentation for Brewster angle microscopy

A number of companies offer commercial BAMs that can either be coupled with a commercial trough or used with a home-made trough. It is possible to build a basic BAM given sufficient technical expertise in optical instrument design; however, most researchers use commercially

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy available BAMs that are now quite sophisticated. The quality of images produced by BAM instruments has improved significantly for the most recent (and most expensive) models. Nanofilm Technologie GmbH (NFT) made BAM available commercially with the BAM-1 and the BAM-1plus in the early 1990s. The BAM-1plus corrects for distortion of the image aspect ratio but not for the image focusing issue and has a lateral resolution of ∼4 µm. Images reported using this instrument and similar home-built designs are generally out of focus on the two ends of the image and are sometimes cropped to show only the central portion which is in better focus. The BAM-2plus from NFT is a more powerful BAM introduced in the later 1990s. It is superior to the BAM-1plus in that the lateral resolution is now 2 µm. The instrument also provides images that are in focus within the entire field of view. The BAM-2plus keeps the images in focus by scanning the orientation of a mirror so that parallel strips of the image come into focus sequentially and are then reconstructed. The BAM-1plus and BAM-2plus are equipped with analyzers for imaging anisotropy in monolayers. A simpler Mini-BAM instrument from NFT without an analyzer and a wide field of view of 4 mm × 6 mm is also available. A number of more powerful instruments are now available, such as the nanofilm ultrabam (NFT, now called Accurion), which provide fully focused images in real time and are capable of performing quantitative analysis of reflectivity in BAM images. The nanofilm ep3bam provides high-resolution images and is able to be upgraded to function as an imaging ellipsometer for use on other substrates. Another current BAM that is available is the Optrel Bam 3000 from KSV Instruments, which provides 2 µm resolution, an adjustable analyzer, and a scanning option for building focused images. The Optrel and Accurion instruments allow adjustment of the Brewster angle over a wide range allowing in principle for imaging on surfaces other than water such as glass slides.

2.4

General features of monolayer phase behavior observed by Brewster angle microscopy

The interpretation of BAM images involves the use of knowledge from fields including the basic phase behavior of monolayers, the range of basic domain forms seen in monolayer systems, nucleation and domain growth behavior in monolayers, and binary phase behavior if mixed monolayers are being studied. It is important for the user to understand how to interpret surface pressure–area (π –A) isotherms that should be measured simultaneously with the BAM experiment. A keen eye for making and comparing observations under different monolayer conditions is necessary. While it is often relatively easy to observe some interesting looking images upon starting a monolayer project

5

using BAM, going beyond qualitative description should be the goal. While it is not possible to predict monolayer domain shapes and their behavior from molecular structures, what is seen using BAM generally can be interpreted using known patterns and principles. The monolayer is first spread at high molecular area where the surface pressure is low and then the monolayer is compressed to record the π –A isotherm. At such high areas and very low surface pressures, one generally observes coexistence of the two-dimensional gas phase (G) with either a two-dimensional liquid-like phase often referred to as the liquid-expanded (LE) phase or with a form of condensed phase, which may be of varying degree of order from that of a liquid-condensed (LC) phase, as found in fatty acid and ester monolayers to solid-like phases found when much stronger intermolecular forces are present.5 The gas phase is of low molecular density and is dark under BAM. An LE phase appears as bright circular islands in a dark background or sometimes as a two-dimensional “foam” structure that resembles a network of cells.12 Upon compression of monolayers with coexisting gas and LE phases, the fraction of the surface covered by the gas phase shrinks until a uniform bright field is seen; ideally, this occurs just as the surface pressure begins to rise tangibly. In contrast, monolayers showing coexistence of a gas phase with an LC or solid-like phase appear differently under BAM. If the second phase is LC or more solid-like, then generally bright domains are seen that are not round in shape and can have angular boundaries with the gas phase. These domains may have anisotropic ordering (Section 2.5), which can be confirmed by rotating the analyzer away from 0◦ and looking for variations in the reflectivity from domains or regions within domains. Domains of weakly ordered LC phases often appear bright and rounded in shape. It is also possible to observe the coexistence of gas, LE, and LC phases at very low surface pressures and below a certain temperature as monolayers possess a triple point and when out of equilibrium the phase rule can be violated.4, 15 When a solid-like and a gas phase are found coexisting upon spreading, the π –A isotherm remains flat and then rises. The gas phase is squeezed out but may persist between domains if they have difficulty in merging. The merging of solid-like domains upon compression is not always a smooth process as they are pushed into each other, sometimes fragment, and reorganize. When the LE and gas phase are present upon spreading, compression will first bring the monolayer into the LE phase, which should appear as an all bright uniform field of view. If there is a kink or plateau in the surface pressure isotherm, then the emergence of domains of a condensed phase should become visible by BAM. The emerging phase will be of higher reflectivity, and could also potentially possess anisotropic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

6

Techniques

ordering within domains visible by BAM. If the condensed phase has a tilted orientation of hydrocarbon chains, then anisotropy will be generally be visible if the analyzer angle is set well away from 0◦ . Under some conditions and especially for more weakly ordered LC phases, the domains that form may be rounded shapes. In more highly ordered phases, as found in many of the examples cited in this chapter, many other domain shapes can be observed. These are often observed as growth forms and appear as highly branched dendritic or fractal-like structures. It is possible that these nonequilibrium growth forms will relax in shape over time but often they are not observed to do so; for many monolayers, the equilibrium domain shapes are never observed. The observation of dendritic growth forms that do not relax over tens of minutes or longer indicates the formation of a highly ordered phase. Special precautions, such as the use of enclosed troughs, and against effects including subphase evaporation or contamination would need to be taken to carry out observations of monolayers over periods of many hours. Further compression will generally bring the monolayer fully into the condensed phase. In observations of mixed monolayers by BAM, a major question of interest is the miscibility of the components. BAM observation of morphologically distinct regions, especially when they resemble what is seen in the pure monolayers of the components, is a sign of immiscibility. When the components are miscible, generally a uniform morphology of a somewhat intermediate nature is observed. The phase behavior can be complex, as immiscibility can be partial or complete and can depend upon temperature, monolayer composition, and surface pressure. For example, BAM was used to study mixed monolayers of 7-(2-anthyrl)heptanoic acid (2A7) and myristic acid at a series of mole ratios and the monolayers of the pure species.16 The π –A isotherm of 2A7 was steep and indicated the formation of a solid-like phase consistent with BAM observation of domains resembling irregular fragments in coexistence with the gas phase. The π –A isotherm of myristic acid showed a plateau identified as the well-known phase transition from the LE phase into the coexistence of the LE and LC phase.4 Rounded condensed-phase domains with internal anisotropy were seen as expected for a tilted LC phase. It was noted that addition of small amounts of 2A7 to the monolayer prevented the formation of these domains indicating some miscibility at low mole fraction of 2A7. Equimolar mixed monolayers at near-zero surface pressure showed coexistence of foam-like regions characteristic of myristic acid and regions rich in 2A7, which in contrast appeared darker as this is a shorter molecule than myristic acid. Upon compression of the equimolar mixed monolayer, alternating dark and bright stripes formed. Upon collapse, bright bands formed across the darker stripes, which suggested that the darker stripes were rich in 2A7.

BAM contrast inversion due to relative thickness changes in coexisting monolayer regions of a phase-separated binary mixture was observed during compression of mixed monolayers of a triaroylbenzene derivative (C8METAB, see structure in Figure 7) consisting of a 1,3,5-tris(4hydroxy-benzoyl)benzene core ether linked to methyl octanoate side chains, and methyl stearate, and is shown in Figure 3.17 Mixed monolayers of C8METAB and methyl stearate of mole ratio 1 : 2 showed round domains rich in methyl stearate at higher molecular areas. These domains were brighter than the surroundings and were internally anisotropic, as seen in condensed-phase domains of methyl stearate or similar fatty esters.18 The methyl stearate domains became darker than the surrounding field as the compression proceeded. It was concluded that compression forces the C8METAB molecules to reorient from a flat to a standing orientation on the water surface now resulting in a greater reflectivity for the C8METAB regions than for the methyl stearate domains. Collapse was then visible in the C8METAB-rich regions, as evidenced by the emergence of grainy bright pattern.

2.5

Brewster angle microscopy observation of domain anisotropy

When the refractive index of a monolayer varies with direction in the plane of the monolayer, anisotropy is present such as is the case for phases of amphiphiles with hydrocarbon chains that are tilted. In addition to a magnitude of tilt, the tilt will have a direction defined by the azimuthal angle. Fatty acids, esters, and alcohols exhibit complex phase diagrams with a variety of vertical and tilted condensed phases, which have been studied extensively and reviewed.19, 20 Within the domains of a tilted phase, there can be regions with the same azimuthal angle that are large enough to be imaged by BAM. When the analyzer of the BAM is set at an angle other than 0◦ , regions of different azimuthal angle will have different reflectivity and there will be contrast between them in the BAM image. In tilted condensed phases, there are examples of discontinuous changes in the azimuthal direction and hence reflectivity between segments as well as examples of continuous variation in reflectivity across a segment as the azimuthal tilt direction gradually varies. There are other reasons, such as an asymmetric lattice structure, that can give anisotropy in the refractive index in a monolayer, but anisotropy due to molecular tilt is by far the most widely studied source and is also observed in the condensed phases of monolayers of membrane lipids as well as synthetic amphiphiles.21, 22 The observations of these domain segments of varied tilt angle had been achieved using polarized fluorescence microscopy prior to the introduction of BAM.6

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy

7

x

(a)

(b)

(c)

(d)

Figure 3 Brewster angle microscopy images of monolayers of mixtures of monolayers of the triaroylbenzene derivative C8METAB and methyl stearate showing both an example of contrast inversion and also of collapse in a binary mixed monolayer; in this experiment, the angle of the analyzer is set at 30◦ . A small “X” near a domain in panel A locates a condensed-phase domain showing some internal anisotropy. Reproduced from Ref. 17.  Elsevier, 2008, and the caption reads as follows: Brewster angle micrographs of mixed monolayers of methyl stearate and C8METAB of mole ratio 1 : 2 C8METAB/methyl stearate on water at 295 K during compression. (A) 69 A˚ 2 per molecule, (B) 53 A˚ 2 per molecule, (C) 37 A˚ 2 per molecule, (D) 22 A˚ 2 per molecule. The scale of the images is 900 µm × 700 µm.

Monolayers of methyl eicosanoate were examined in a tilted condensed phase L2  in which the tilt direction was toward next-nearest neighboring molecules.23 In this phase, six-armed “star” defect patterns are seen in which the domain is divided into six regions, each resembling one-sixth of a pie shape. Each segment or “slice” of the pie has a different reflectivity indicating a different azimuthal direction. These domains undergo a “blooming” transition upon cooling that originates from the center of the domain and during which the azimuthal directions change but the segmentation is maintained. An analysis of the reflectivity variation was presented that depended on nine parameters including the dielectric constant of water, elements ε xx and εyy of the molecular dielectric tensor, monolayer thickness, angle of incidence of the laser beam, the analyzer angle, and a small tilt of the polarization angle away from the p-orientation that improved image contrast. These were processed in a detailed optical model to obtain reflectivity variations that could be compared with those observed and used to determine the best fitting values of the monolayer thickness and tilt angle. Monolayers of ethyl palmitate and ethyl stearate are attractive systems in which to study internal domain anisotropy by BAM.18 Monolayers of these compounds have two-phase coexistence regions between the LE and LC

phases and the domains of the tilted condensed phase are round and 100–200 µm in size.21 In these condensed-phase domains, the most regular structure that could be observed was the round domain divided into six equal segments like a sliced pie with each segment having a different reflectivity, as seen in Figure 4. Analysis of the reflectivity differences was consistent with a structure in which the azimuthal direction was parallel to the domain periphery and jumped by 60◦ at each segment boundary. In these images, the analyzer angle was set at 60◦ and, when rotated to −60◦ , the relative brightness of the segments in the domain reversed, with the brightest segment becoming the darkest and vice versa; the observation of this inversion in contrast confirms the presence of anisotropy. This experiment is a good illustration of the value of recording BAM images at different analyzer angles including pairs at positive and negative angles of the analyzer of the same magnitude. In these monolayers, not all domains were organized into neat six segmented pies; others had corner-notched segments of different brightness or segments with zigzag boundaries and some domains appeared to have one orientation within them. Dendritic growth of the condensedphase domains could be observed in monolayers of ethyl stearate when they were compressed more rapidly, but in these monolayers the dendrites were transient forms that

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

8

Techniques

(a)

(b)

(c)

(d)

(e)

(f)

Figure 4 Liquid-condensed-phase domains in monolayers of ethyl stearate showing anisotropy, the monolayer is in the LE phase and liquid-condensed phase coexistence region. Reproduced from Ref. 18.  American Chemical Society, 1996, and the figure caption reads as follows: The most regular texture in ethyl stearate monolayers is a subdivision of the domain into six parts of uniform orientation: (a) analyzer angle of about 60◦ ; (b) analyzer angle of about −60◦ ; (c) parallel polarizer and analyzer; (d) crossed polarizer and analyzer. [In half the domains of this type, the arrangement of the segments is as in parts a and b. The other domains of this type are inverted images of those of parts a and b; (e) analyzer angle of about 60◦ ; (f) analyzer angle of about −60◦ . The bar represents 100 µm.

relaxed into circular domains within minutes. These growth forms were observed to have anisotropic ordering with different branches having different reflectivity. The contrast in reflectivity due to segments of different orientation diminished as the magnitude of the tilt angle decreased. Compression of these monolayers ultimately resulted in a transition to a vertically oriented phase in which the anisotropy disappeared. A subsequent study of monolayers of palmityl acetate included both BAM and GIXD data and showed a transition within the domains from the azimuthal tilt directions being oriented perpendicular to parallel to the segment boundaries.22

Anisotropy within condensed-phase domains can also be found in which the azimuthal direction of tilt is varying continuously such that a gradation or pattern of gradually changing reflectivity is observed within a domain. Monolayers of the phospholipid dimyristoylphosphatidylethanolamine (DMPE) were found to exhibit cardioid (kidney)-shaped condensed-phase domains within which the azimuthal direction bended gradually within the domain.24 Analysis of the variation in the gray levels in the images with the analyzer set at 60◦ allowed determination of the magnitude of the tilt and the azimuthal direction. The analysis makes use of an expression for the reflectivity as a function of the angle of incidence, angle of the analyzer, and dielectric anisotropy in the monolayer represented by ε⊥ and ε a = ε  − ε ⊥ that was obtained as a second-order approximation and is less onerous than the full optical formalism. A novel extension of BAM useful for the determination of tilt angles in condensed phases was achieved by adding a photon counter and an autocorrelator to a homebuilt BAM in a method the authors called Brewster angle autocorrelation spectroscopy.25 The experiment makes use of the fluctuation in the intensity of the reflected light as domains of different azimuthal tilt direction drift under the laser spot when the analyzer is set at an angle of 80◦ . A derived equation relates the magnitude of the tilt angle to the limiting value of g2 =< (I − < I >)2 > / < I >2 where the brackets denote time averages of reflected intensities. The method was applied to determine the tilt angle variation near a phase transition in octadecanol between tilted and untilted condensed phases. The obtained variation in the tilt angle with surface pressure as the monolayer crossed the phase transition agreed very well with that determined by GIXD experiments.26

2.6

Multilayer formation

BAMs is an ideal method for observing films at the water surface, which exhibit coexisting regions of discrete numbers of molecular layers such as monolayer together with trilayer and thicker. Such a situation is commonly encountered when studying compressed monolayers of thermotropic liquid crystals (see Self-Organization and SelfAssembly in Liquid-Crystalline Materials, Soft Matter), which are generally rod-shaped molecules, or bent “banana” shapes, or disc shaped in the case of discotic liquid crystals. The rod-shaped molecules exhibit a plethora of phases in three dimensions that include the isotropic phase, nematic phase, and a wide variety of vertically oriented, tilted, or hexatically ordered smectic phases as a function of decreasing temperature.27 Studies of liquid crystals at the water–air

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy

9

here a range of coexisting multilayer structures was also seen.30

2.7

Figure 5 BAM image of multilayer regions coexisting in spread and compressed films of the liquid crystal 8CB. Reproduced from Ref. 31.  American Physical Society, 2007, and the figure caption reads as follows: Brewster angle microscope image of the coexistence of 8CB multilayers. The background is 8CB trilayer. The layer reflectivity increases with thickness, so that different gray levels correspond to different thicknesses. The rippled variation in color within one domain is due to variations in illumination. The black scale bar is 1.0 mm.

interface provide a unique environment in which to observe orientational ordering and domain dynamics. One of the most well-known thermotropic liquid crystal molecules is 4-octyl-4 -cyanobiphenyl (8CB), which has also been the subject of a number of monolayer studies using BAM.28–31 8CB forms a smectic-A phase in which the molecules are organized in layers and are of average vertical orientation, although there is no long-range positional ordering within a given layer. The formation of layers is also seen at the water surface for 8CB. The surface ˚ 2 per molecule and pressure of 8CB begins rising near 50 A then enters a long plateau at ∼5 mN m−1 , which extends ˚ 2 per molecule. BAM imaging in the down to near 10 A long plateau shows coexisting rounded domains of different reflectivity,31 as seen in Figure 5. Inside the plateau region, the film is found to consist of a monolayer coexisting with domains that are composed of the monolayer plus an integer number of bilayers above it. The thicknesses of the different coexisting domains were examined using quantitative assessment of the reflectivity within domains, which required calibration of the response of the gray scale in the images against the incident light intensity.29 In 8CB, a second rise and second plateau in surface pressure was seen at very low areas, and

Monolayer collapse

BAM is also well suited for determining when a monolayer collapses under compression and forms three-dimensional structures. Collapse during continuous monolayer compression is usually quite obvious by BAM in that distinctly brighter spots or other shapes such as long bright jagged or wavy lines appear in the monolayer. The appearance of such features should coincide with features in the π –A isotherm such as a sudden drop in surface pressure followed by a plateau or by a roll over onto a plateau, which will not always be exactly flat. Such changes should occur near molecular areas that seem reasonable as possible two-dimensional packing limits as estimated from the dimensions of the molecular profile. It is possible to observe the collapse and further three-dimensional aggregation either under slow continuous compression or by pausing at a surface pressure at which the monolayer is metastable and observing structures emerge as a function of time. A study of the collapse in stearic acid monolayers was an early application of BAM.32 Stearic acid monolayers compressed to 30 mN m−1 at 20 ◦ C on subphases of pH = 2, 3, and 5 exhibited significant surface pressure relaxation over a period of 150 min. During the surface pressure relaxation, BAM showed the nucleation and growth of crystallites whose form depended on the pH, varying from bright fragments at pH = 2, to structures which clearly showed internal anisotropy at pH = 3, to smaller dendritic forms at pH = 5. A subsequent study further examined the collapse in a series of fatty acid monolayers and also for methyl arachidate on a variety of subphases.33 The collapse of stearic acid on a pH 3.0 subphase was examined by stopping the compression at 40 mN m−1 , which is in the surface pressure “spike” after which the surface pressure quickly drops into a plateau near ∼18 mN m−1 . In this case, BAM showed the nucleation of many small bright dots, which grew and overlapped if compression was resumed into the plateau region, as seen in Figure 6. In contrast, the stearic acid monolayer compressed on a pH 8.0 subphase containing 10−4 M CaCl2 showed a surface pressure rising to just over 60 mM m−1 and then rolling over into a plateau without a spike. BAM observation at 63 mN m−1 showed chains of many small bright microcrystals. It is important to realize that monolayers at surface pressures above the equilibrium spreading pressure (ESP) are thermodynamically metastable and should ultimately transform to three-dimensional aggregates in equilibrium with a monolayer whose surface pressure equals the ESP

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

10

Techniques

Surface pressure (mN m−1)

60 50 40 (b)

30 20 10

10 (a)

20

BAM enables a number of additional characterizations of a monolayer (or thin film, in the case of multilayer and other features) such as the recording of reflectivity isotherms that can be used to estimate changes in the film thickness as a function of surface pressure. Quantitative BAM also enables estimation of differences in film thickness between regions within an image. In the simplest approach, the reflectivity R at the Brewster angle is noted to be equal to Cd 2 , where C is a constant and d is the monolayer thickness. A more developed equation that makes use of the thin film approximation35 is


30

R=

Area (Å2 mol−1)

πd λ

2

  

n21 − n22 − 1 + 1+

n22 n21

n22

  

(3)

(c)

Figure 6 BAM images of a monolayer of stearic acid in the collapse regime, along with the surface pressure (π)–mean molecular area (A) isotherm at 20 ◦ C on a pH 3 subphase. Reproduced from Ref. 33.  American Chemical Society, 1996, and the figure caption reads as follows: Surface pressure/area isotherm (a) of a stearic acid monolayer on acidified aqueous subphase (pH 3.0) at 20 ◦ C and BAM images of (b) 3D nuclei (bright “dots”) observed at a surface pressure of 40 mN m−1 within the steep part of the isotherm (i.e., the 3D nucleation occurs at π < π c = 50 mN m−1 ) and (c) overlap and growth of the 3D nuclei within the “plateau” region of the π /A isotherm (after the pressure “spike” at π c ). The bar represents 200 µm.

of the molecule at hand for the given subphase conditions. The possible mechanisms of monolayer collapse, especially for lipid monolayers, have been reviewed and, in addition to formation of microcrystals, include ejection of material into the subphase, fracture, and a folding mechanism in which the layer bulges up and then lays over resembling what would occur on pushing on a rug from two ends.34

2.8

Quantitative analysis of Brewster angle microscopy images

While many studies of monolayers using BAM descriptively report the microstructures observed, a more quantitative analysis of the BAM images can yield information on thickness changes in monolayers and has been referred to as quantitative BAM. Quantitative BAM can be applied using a simple optical model for films at the water–air interface in which there is no or negligible anisotropy within the film. The measurement requires that a calibration be performed to calibrate gray-scale intensities from pixels in the image with calculated values of reflectivity, and the conduct of such a calibration has been reported.35–37 Quantitative

In (3), d is the film thickness, λ is the wavelength of the laser radiation, n1 is the refractive index of the monolayer, and n2 is the refractive index of the subphase. The value of n2 is known and a value or reasonable range of values for the refractive index of the film material is then used in the calculation. The next requirement is to calibrate the gray scale of the CCD camera with reflectivity. In the calibration procedure, the pure water surface is used. Camera settings such as shutter time, gain settings, and the entire optical configuration must be kept unchanged throughout the calibration and for any experiments that make use of the calibration. The BAM images are recorded from the bare water surface as the incident angle is varied by a few tenths of a degree on either side of the Brewster angle. The range of variation should not exceed saturation of the camera response. The gray scale should be averaged over a region or image. A plot of gray scale intensity versus incident angle will be parabolic in shape. The theoretical reflectivity is then calculated as a function of incident angle using (3). Finally, the calculated reflectivity values for each incident angle are plotted versus the gray scale values obtained at those angles. It will now be possible, subject to the abovestated assumptions, to relate gray-scale intensities from images with reflectivity and use (3) to estimate the film thickness. Quantitative BAM has been applied to studies of lung surfactant films at the water surface.35 It has been applied to obtain reflectivity versus surface pressure isotherms for monolayers of monopalmitin, monoolein, and monolaurin.36 In studies of monolayers of dipalmitoylphosphatidyl˚ glycerol, the method was used to conclude that a 4 A increase in thickness occurred during compression across the LC + LE coexistence region due to a conformational change.37

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy

3

APPLICATION OF BREWSTER ANGLE MICROSCOPY TO THE STUDY OF MONOLAYERS OF MACROCYCLIC COMPOUNDS

BAM has been applied in studies of amphiphilic derivatives of macrocycles. As these compounds are capable of forming host–guest complexes, the study of their monolayers is often motivated by an interest in their transfer as Langmuir–Blodgett films for use as sensors or by an interest in studying their recognition behavior in a membrane-like environment. The dispersion of amphiphilic macrocycles in mixed monolayers with fatty acids or lipids has proven to be of interest. A number of significant issues arise in studies of amphiphilic macrocycles related to their collapse behavior, instability toward aggregation, response to recognition of guest molecules in the subphase, and miscibility behavior in mixed monolayers that can be at least partly addressed by BAM. In all the cases, the interpretation of the BAM images is aided by comparison with the π –A isotherm data and considerations of the dimensions of the molecule. The structures of a selection of the supermolecules mentioned in this chapter are shown in Figure 7.

3.1

Examples involving monolayers of amphiphilic calixarenes

There have been a number of studies of calixarene monolayers incorporating the use of BAM addressing a range of phenomena from behavior and stability of monolayers of the calixarenes (see Calixarenes in Molecular Recognition, Molecular Recognition) themselves at the water surface, to their complexation of guest molecules, and their behavior in mixed monolayers. One of the earliest studies examined the formation of complexes between para-tertbutylcalix[8]arene and the fullerenes C60 and C70 spread together at the water–air interface.51 The 1 : 1 complexes of the calixarene and fullerene were prepared by refluxing in benzene and then taking the recovered solid and dissolving it in chloroform for spreading onto the water surface presumably as the intact complex. The π –A isotherms for the calix[8]arene showed a rise starting near ˚ 2 per molecule and then steadily increasing to close 180 A to 80 mN m−1 (which is greater than the surface tension of water) with no sign of a collapse plateau. The surface pressure of the monolayer of the calix[8]arene + C60 ˚ 2 per molecule started to rise at higher areas near 230 A and started to plateau just above 60 mN m−1 . In contrast, the surface pressure for the calix[8]arene + C70 monolayer started rising at a similar area but during the ascent shifted to lower areas than seen for the pure calix[8]arene

11

monolayer. BAM observations showed differences amongst ˚ 2 per these three monolayers. The calix[8]arene at 301.6 A molecule showed bright islands and dark areas indicating a two-dimensional solid-like phase coexisting with a twodimensional gas phase. Assignment of the phase coexisting with the darker gas phase in monolayers at higher area and low surface pressures near zero as solid-like rather than LE or liquid crystalline can be decided by observing the shape of the domains. If the boundaries of the bright domains smoothly vary and especially if any domains are circular or elliptical, then a LE or possibly liquid-crystalline phase is present. If the boundaries appear jagged or have sharp turns or corners, then the phase is assigned as being solid-like. The domains of solid-like phases do not merge easily upon compression, while those of the LE or liquid-crystalline type merge. In this image, the presence of extra bright, mountain-shaped clusters of molecules was also observable, and their brightness and bordering by optical interference patterns, known as Newton’s rings, indicates that these are due to three-dimensional aggregates of the calix[8]arenes. The calix[8]arene + C60 monolayer was reported to appear similar at low surface pressure and high surface areas as the monolayer of the calix[8]arene alone. The monolayer of the calix[8]arene and C70 showed different features, and illustrates the important distinction between multilayer formation and anisotropic ordering in a monolayer that can be made by informed application of BAM. In the BAM image, regions representing three different intensities of brightness were seen, and it was concluded that these were monolayer, bilayer, and trilayer regions. In this case, it is reported that there is no variation in the brightness of these image regions as the analyzer is rotated and thus the authors conclude that they are observing the presence of regions of different thickness. In this study, a noteworthy dynamic effect is also reported, In the case of the calix[8]arene + C70 monolayer, a large region that was fairly uniform suddenly fragmented like the shattering of a sheet of ice upon a small mechanical vibration of the trough. This indicates that the monolayer was of a remarkably brittle, two-dimensional structure. It was noted that the solid-like domains of these monolayers did not merge easily upon compression but tended to crash into each other and form some clusters and ultimately bilayers. These observations indicate that BAM applied to monolayers should be viewed as more than the static capturing of images at fixed conditions and as a method for observing dynamic changes in real time. The behavior of monolayers of a para-tertbutylcalix[8]arene in which the phenolic hydroxyls were ether linked to 3-hydroxypropionic acid groups was studied using π –A and V − A isotherms, BAM, and infrared spectroscopy of Langmuir–Blodgett films.52 The calix[8]arene can undergo conformational changes on compression that alter the nature of the hydrogen bonding present, which

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

12

Techniques

R6

R5

A

R7

R8

OR2

OR1

R3O

OR4

O

D

O N H

OCH2CH3 OCH2CH3 P

S NH2

O

Calix[4]arene derivatives OCH3

O N C16H33

C16H33 N O

O

O

O

O

OR

OR

O

O

O

O

O

O

P

O

N H

C

O

B

O

Chiral crown ether

Diazacrown ether

OR

O O

CH3O OR

O

O

OCH3 CH3O

O

HO O OH

O

OCH3 6-position

OH O

OH

3-position HO

O

HO O OH O HO

O

O OHOHO

Cryptophane (anti form)

2-position OH OH

O

(CH2)n O

O

OH HO

OCH3 OCH3 (CH2)n

(CH2)n

C8METAB

RO

O

b-cyclodextrin

NH2

O OH

N

OH OH O OH

OH OH O

N

O

C11H23HN

N

NHC11H23

2C11H23-melamine

O OH

Figure 7 Structures of a selection of the molecules described in this chapter. (A) calix[4]arene derivatives: CALIX1 (Ref. 38), R2 = R4 = H, R1 = R3 = group A, R5 = R6 = R7 = R8 = p-tert-butyl; CALIX 2 (Ref. 39), R1 = R2 = R3 = R4 = n-C12 H25 , R5 = R7 = group B, R6 = R8 = H; CALIX3 (Ref. 39), R1 = R3 = n-C12 H25 , R2 = R4 = CH3 , R5 = R7 = group B, R6 = R8 = H; CALIX4 (Ref. 40): R1 = R3 = group C, R2 = R4 = H, R5 = R6 = R7 = R8 = p-tert-butyl; CALIX5 (Ref. 41), R1 = R2 = R3 = R4 = H, R5 = R6 = R7 = R8 = dodecanoyl (O=C-(C11 H23 )); CALIX6 (Ref. 41) R1 = R3 = group D, R2 = R4 = H, R5 = R6 = R7 = R8 = dodecanoyl (O=C-(C11 H23 )); CALIX7 (Ref. 41) R1 = R3 = PO(OEt)2 , R2 = R4 = H, R5 = R6 = R7 = R8 = dodecanoyl (O=C-(C11 H23 )); CALIX8 (Ref. 42), R1 = R3 = H, R2 = R4 = N-acetyl-pivaloyloxymethyl-6-aminopenicillanic acid, R5 = R6 = R7 = R8 = p-tert-butyl; CALIX9 (Ref. 42), R1 = R3 = H, R2 = R4 = benzylpenicillin ethyl ester, R5 = R6 = R7 = R8 = p-tertbutyl; CALIX10 (Ref. 42), R1 = R3 = H, R2 = R4 = benzylpenicillin propyl ester, R5 = R6 = R7 = R8 = p-tert-butyl; Cyclodextrin derivatives: CD1 (Ref. 43): the 2-positions and 3-positions are all ether linked to n-C6 H13 , the 6-positions are all substituted with –NH3 + ; CD2 (Ref. 44), the 2-positions and 3-positions are all ester linked to hexanoyl groups (O=C-C5 H11 ), the 6-positions are all unmodified; CD3 (Ref. 44), the 6-positions are all modified by tert-butyldimethylsilyl groups, the macrocycle is α-cyclodextrin (six glucose units instead of seven for the β-cyclodextrin as shown); CD4 (Ref. 44), the 6-positions are all modified by tert-butyldimethylsilyl groups; CD5 (Ref. 44), the 6-positions are all modified by tert-butyldimethylsilyl groups, the macrocycle is γ -cyclodextrin (eight glucose units instead of seven for the β-cyclodextrin as shown); Chiral Crown Ethers (Ref. 45), R = benzyl, p-phenylbenzyl, nC12 H25 , or n-C16 H33 ; Diazacrown Ether (ACE-16 in Ref. 46); C8METAB (Ref. 17), R = (CH2 )7 CO2 CH3 ; Cryptophanes (Ref. 47): only anti -conformation is shown, n = 3 or n = 5 for anti -cryptophanes, n = 9 or n = 10 for mixed 1 : 1 anti - + syn-cryptophanes; 2C11 H23 -melamine (Refs 48–50). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy can be intramolecular, intermolecular, or directed to water molecules. This study was carried out using the NFT IElli2000 BAM, which provides especially well-focused images, as seen in Figure 8. ˚ 2 per The surface pressure rose starting near 320 A molecule, displayed some changes in slope along its rise, ˚ 2 per and then a plateau starting near 46 mN m−1 and 200 A 2 ˚ per molecule. molecule that extended down to near 90 A Compression–expansion cycles revealed hysteresis in the surface pressure. The magnitude of the compressibility modulus, Cs−1 = −A(∂ /∂A)T was consistent with the presence of a condensed phase. The BAM observations were consistent with the earlier assignment51 of the phases present at low pressure as solid-like and gas. During the first compression, large, fairly uniform solid domains with straight edges and some smaller solid domains are seen that merge on compression, although not easily; some scattered, but small and brighter features are also seen in the images. The subsequent expansion shows the bright domains breaking apart and also shows the emergence of many small, branched domains. During the second compression, these small, branched domains remain visible and are also visible on a second expansion in the midst of the brighter islands. Compression to lower areas generated bright striations indicative of monolayer collapse. The emergence of the smaller, branched domains upon the first expansion was interpreted as being consistent with infrared spectra of the monolayers transferred as Langmuir–Blodgett films and molecular mechanics calculations that suggested that on compression the hydrogen bonding shifted from being

between calixarenes to being between calixarenes and water molecules, thus resulting in a less cohesive film. Large difference in the V −A isotherms between the first and second compression also suggested a significant conformational change. Monolayers of compounds that exhibit hysteresis of surface pressure during compression–expansion cycles generally also exhibit surface pressure relaxation at fixed area as a function of time. During surface pressure relaxation, BAM is often useful for observing morphological changes due to the structural changes or nucleation and domain growth that is associated with the surface pressure relaxation. A study of monolayers of a para-tertbutylcalix[4]arene derivative 1,3 modified on the lower rim hydroxyls with benzylamidoethoxy groups (CALIX1, see Figure 7) was reported using surface pressure and BAM.38 The monolayers were studied over a temperature range of 5–25 ◦ C. The surface pressure rise on compression was seen to start at ˚ 2 per molecule and was notably linear except at 125–128 A ◦ 5 C where there was an inflection region near 25 mN m−1 indicating a phase transition. The surface pressure plateaus varied with temperature. The phase transition observed at 5 ◦ C manifested itself under BAM as the rapid appearance of many irregular, small, and bright domains. Thus, in studies using BAM, one should be alert to isotherm features including small kinks and changes in slope as conditions near it should be especially observant for changes visible by BAM. The isotherm feature at 5 ◦ C also showed a small maximum and reversal indicating a supersaturation kinetic effect upon compression. Such overshoots in

1st c

1st c

1st c

(i) < 500 Å2; 0 mN m−1

(ii) 410 Å2; 0 mN m−1

(iii) 320 Å2; 2 mN m−1

1st e

2nd c

(vi) 425 Å2; 0 mN m−1

(vii) 330 Å2; 0 mN m−1

13

2nd e

(viii) 330 Å2; 0 mN m−1

1st c

(iv) 275 Å2; 30 mN m−1

1st e

(v) 400 Å2; 0 mN m−1

2nd e

(ix) 340 Å2; 0 mN m−1

(x) 125 Å2; 46 mN m−1

Figure 8 BAM images of monolayers of a para-tert-butyl calix[8]arene derivative described in Ref. 52, showing rigid fractured ice-sheet-like domains merging during the first compression, hysteresis in the domain morphology on expansion and recompression, and then collapse. Reproduced from Ref. 52.  American Chemical Society, 2005, and the figure caption reads: Brewster angle microscope (BAM) images of C8A monolayer at the air–water interface during the two successive cycles: first compression (1st c), first expansion (1st e), second compression (2nd c), second expansion (2nd e), processes and collapse of the film, images i–iv, images v and vi, image vii, images viii and ix, and image x, respectively. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

14

Techniques

surface pressure in isotherms often precede rapid domain growth as the two-dimensional supersaturation is relieved. In this study, experiments were carried out in which the surface pressure relaxation and BAM images were observed as a function of time over periods up to 800 s. Compres˚ 2 at 20 ◦ C corresponding to an initial surface sion to 104 A pressure of 25.6 mN m−1 showed the emergence of highly regular dendritic domains after 50 s. The small magnitude of the surface pressure relaxation suggested that the density difference between the emerging ordered phase in the dendritic domains and the initial phase was small. Compression ˚ 2 , respectively to lower surface areas of 95.6 and 83.5 A also showed the emergence and growth of domains of a lancet-like shape that appeared at later times to indicate overgrowth of a second layer. It was thus concluded that during compression this monolayer is far from equilibrium. The limiting molecular areas were found to be consistent with a conformation in which the benzylamido groups were oriented parallel to each other and at an angle to the calixarene cone axis. Calix[4]arenes modified at the upper rim with two phosphonate groups and on the lower rim hydroxyls with either four dodecyl chains (CALIX2, see Figure 7) or two dodecyl chains and two methyl groups (CALIX3, see Figure 7) were studied as monolayers using BAM and GIXD.39 The monolayers were studied on subphases of pure water, and on those containing Th(NO3 )4 , Eu (NO3 )3 , CdCl2 , and NaCl at a series of temperatures from 14 to 39 ◦ C. CALIX2 clearly displayed plateaus in its surface pressure isotherms indicating a first-order phase transition between an LE phase and a condensed phase. This phase-transition plateau resembled that seen for single-chain amphiphiles, which is unusual in the reported studies of monolayers of macrocyclic compounds. The plateau surface pressure was reduced and the mean molecular area at its onset was increased on the Th4+ subphase versus pure water. In contrast, CALIX3 did not show a surface pressure plateau and its isotherms were essentially the same at 15 and 25 ◦ C and varied little between the pure water and Th4+ -containing subphases. BAM in the case of these calixarenes showed the formation of bright domains only in the two-phase coexistence region and not immediately upon spreading. CALIX2 showed the formation of domains described as lancet or sickle shaped and with an irregular border. The observation of uniform reflectivity indicated that there was no anisotropy of molecular orientation within these domains. The domains of CALIX2 were similar on the NaCl-containing subphase, but were considerably more irregular in shape on the Cd2+ -, Eu3+ -, and Th4+ -containing subphases. BAM thus has shown here the effect of cation binding on the shape of the condensed-phase domains of this compound. BAM observations of CALIX3 showed roughly rectangular

domains. The observation of domain segments of different reflectivities indicated the presence of anisotropy in the molecular orientation. The presence of cations in the subphase increased the observance of more rounded domains with frayed peripheries and sharply defined regions of different reflectivities indicating anisotropic ordering. The “filigree” structure of the domains in which they resemble bundles of filaments viewed from the top was interpreted as indicating the presence of a second molecular layer within these domains. The GIXD data showed that only CALIX2 formed a two-dimensional lattice and that the lattice parameters were not affected by the presence of Th4+ in the subphase. BAM proved useful in evaluating the effectiveness of using metal complex formation based upon chloride bridging between Pd(II) centers as a strategy to stabilize monolayers of a 1,3-(distal) p-tert-butylcalix[4]arene derivative bearing methionine groups amide linked to short aminoethoxy groups (CALIX4, see Figure 7).40 The two methionines, tethered to the lower rim of the calix[4]arene (denoted as L), are capable of forming an intramolecular complex with Pd(II) with the two sulfur and two nitrogen donor atoms from each methionine coordinated to a single Pd(II), forming the mononuclear species designated as PdLCl2 , which required two chloride counterions. Each of the two methionines can also separately form a complex with Pd(II) through a nitrogen and sulfur donor atom and with two chloride ions serving as the additional ligands forming the binuclear species designated as Pd2 LCl4 . Monolayers of these two species were compared with the Pd2 LCl4 species capable of chloride bridging between neighboring molecules intended to provide stabilization of the monolayer. The π –A isotherms of the calix[4]arene amphiphile with no Pd(II) denoted as L, and of the PdLCl2 and Pd2 LCl4 species showed rising surface pressure ascend˚2 ing fairly steeply that gave limiting areas of 162 ± 3 A 2 2 ˚ per molecule, and 152 ± 3 A ˚ per per molecule, 145 ± 3 A molecule for L, PdLCl2 , and Pd2 LCl4 respectively. These limiting areas were found to be consistent with the dimensions of the upper rim of the calix[4]arene cone. Compression–expansion cycles revealed significant hysteresis for all three species, especially when compressed into the upper plateau region where collapse occurred. Area relaxation at constant surface pressures ranging from 33 to 52 mN m−1 showed significant decreases in area of as much as 25% over a 90-min period for the L and PdLCl2 species; in contrast, the Pd2 LCl4 species showed very limited area relaxation of at most 5%, indicating that the stabilization strategy was largely successful. BAM provided complimentary information confirming the success of the stabilization strategy via chloride ligand bridging between binuclear complexes. BAM images of L at 0 mN m−1 show a twodimensional, hexagonal network pattern on spreading where

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy the brighter walls of the hexagons were of higher molecular density. Upon compression, first the network merged and a few bright islands were seen and then, at higher surface pressure, a pattern of many bright “worm-like” domains and darker regions appeared, which was interpreted as representing multilayers or aggregates. The same morphology was not recovered on expansion for monolayers of L. The monolayers of PdLCl2 showed regions of lower reflectivity than those of L with numerous holes. On expansion, PdLCl2 showed formation of a hexagonal network pattern. This study is a clear example of how hysteresis in the surface pressure upon a compression–expansion cycle is generally correlated with different BAM observations during compression and expansion. The most significant observation was that for Pd2 LCl4 : the as-spread monolayers displayed extensive block-like regions of uniform reflectivity that coalesced into a bright uniform film during compression, which during expansion maintained much of its uniformity except for the opening of some dark rifts in the monolayer. The observation of large regions of uniform brightness by BAM, but with straight borders and corners, is a clear indication of a cohesive condensed phase, attributed to the chloridebridging effect. Miscibility studies of calixarenes in mixed monolayers with other species involving use of BAM have been reported for calixarenes mixed with cholesterol41 and phospholipids.42, 53 Studies of miscibility in binary mixed monolayers can be quite complicated as the presence or absence of phase separation of the components can be a function of composition, temperature, and surface pressure. In principle, the mapping of an entire twodimensional binary phase diagram is possible, and as such would require extensive experimentation. Miscibility studies are often motivated by an interest in the behavior of the oriented calixarene in a membrane-like environment or dispersed with a simple amphiphile for use as a sensor after transfer as a Langmuir–Blodgett film. A study of three amphiphilic calix[4]arenes in mixed monolayers with cholesterol was reported.41 The derivatives were para-dodecanoylcalix[4]arene (CALIX5, see Figure 7), 25,27-bis-diethoxyphosphoryltetradecanoylcalix[4]arene (CALIX6), and 25,27-bis-dihydroxyphosphoryloxytetradodecanoyl-calix[4]arene (CALIX7). These derivatives showed steeply rising π –A isotherms with limiting ˚ 2 per molecule respectively areas of 120, 115, and 116 A for CALIX5, CALIX6, and CALIX7. The collapse pressures are reported as 15.2, 23.4, and 35.5 mN m−1 respectively for CALIX5, CALIX6, and CALIX7. The collapse pressure did not vary with composition for mixed monolayers of CALIX5 and CALIX6 with cholesterol indicating immiscibility of CALIX5 and CALIX6 with cholesterol. Cholesterol was also immiscible with CALIX7 and partly destabilized the monolayer. BAM images of

15

monolayers of CALIX6 showed smooth curved borders between the brighter and darker regions at low surface pressures and a uniform brightness on compression that indicated formation of a LE-like phase for this monolayer. Examination of mixed films of these three compounds with cholesterol at 1 : 1 mole ratios showed visual evidence for immiscibility, especially for mixtures of CALIX7 with cholesterol. Monolayers of CALIX5 and CALIX6 mixed with cholesterol showed LE-like domains but with many small holes. However, in the case of mixtures of CALIX7 with cholesterol, BAM showed a remarkable emergence of long fibrillar aggregates during intermediate stages of the compression that were uncharacteristic of either the pure monolayer of CALIX7 or of cholesterol monolayers. The miscibility of two of these compounds, CALIX6 and CALIX7, with the phospholipids dipalmitoylphosphatidylcholine (DPPC), dipalmitoylphosphatidic acid (DPPA), dipalmitoylphosphatidylserine (DPPS), and dipalmitoylphosphatidyethanolamine (DPPE) was studied.53 The dependence of the collapse pressure on composition was used to determine that CALIX5 was immiscible with these phospholipids except at mole fractions either below 20% or above 80%. The collapse pressure varied proportionately with composition for mixtures of the phospholipids with CALIX6, and it was concluded that CALIX6 was miscible with these four phospholipids. BAM images for CALIX5 mixed with DPPA in a 1 : 1 ratio were reported and showed formation of a granulated film before compression, different from either pure monolayer and suggestive of immiscibility. BAM images at other points during compression or for the other lipids mixed with CALIX5 were not reported in this study. In another study, interest in the miscibility of the calixarene derivative with phospholipid in a mixed monolayer was motivated by the prospective antibacterial activity of the para-tertbutylcalix[4]arene derivatives, to which two penicillin units were appended. 6-Aminopenicillanic acid was linked by amide bonds to 1,3-bis(O-acetyl)-p-tertbutylcalix[4]arene to form derivative CALIX8 (Figure 7).42 In the second derivative (CALIX9), benzylpenicillin ethyl ester was ether linked to the 1,3 positions and in the third derivative (CALIX10) benzylpenicillin propyl ester was ether linked to the 1,3 positions. Mixed monolayers of these compounds with DMPE were studied by π –A and V − A isotherms and by BAM, as seen in Figure 9. The limiting areas of these compounds were found to be ˚ 2 per molecule, respectively. The π –A 175, 180, and 181 A isotherms showed the most commonly reported dependence for amphiphilic macrocycles of near-zero surface pressure at high areas followed by a steady rise and then an inflection upon entering a collapse plateau. Analysis of π –A isotherms measured as a function of composition

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

16

Techniques

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

Figure 9 BAM images of mixed monolayers of the three calixarenes described in Ref. 42 with the phospholipid DMPE. The domains formed in mixed monolayers with DMPE are seen to depend upon the structure of the calixarene derivative (panels G–I below). Reproduced from Ref. 42.  American Chemical Society, 2007, and the figure caption reads: BAM images of calixarene, DMPE, and DMPE/calixarene monolayers. Line A–C: pure Calix II; line D–F: pure DMPE; line G–I: mixed films. (A) ∼ 0 mN m−1 ; A = 240 A˚ 2 per molecule; (B) = 6 mN m−1 ; A = 185 .6 A˚ 2 per molecule; (C) collapse point; (D) ∼ 0 mN m−1 , A = 90 A˚ 2 per molecule; (E) = 6 mN m−1 ; A = 57 .8 A˚ 2 per molecule; (F) collapse point; (G) DMPE/Calix I; (H) DMPE/Calix II; (I) DMPE/Calix III. The G, H, and I images were taken at 12, 13, and 14 mN m−1 and 110.9, 82.5, and 87.5 A˚ 2 per molecule, respectively. The composition of all the mixed films was xDMPE = 0 .75 . Scale: the width of the snapshots corresponds to 400 µm.

carried out by plotting mean molecular area versus mole fraction of DMPE at 10, 15, and 20 mN m−1 showed positive deviations from ideality. BAM showed that these three calixarenes showed uniformly bright images once compressed to rising surface pressures. Observation of a two-dimensional LE and gas pattern for CALIX9 at high areas was noted. The observation that these calixarenes alone form uniform one-phase monolayers along the entire rise in surface pressure aids interpretation of BAM images of mixed monolayers, as domain formation in the mixed monolayer becomes easier to interpret. In the case of mixed monolayers of DMPE and CALIX8 for xDMPE = 0.75, domains form that closely resemble those formed by the pure lipid. In the case of mixed monolayers of CALIX9 or CALIX10 with DMPE, such domains also form but are considerably smaller, a result indicating a degree of miscibility between the calixarene and DMPE, suggesting that they would be membrane active.

3.2

Examples involving monolayers of amphiphilic cyclodextrins

Cyclodextrins (CDs) are cyclic oligosaccharides composed of glucose units joined by α-1,4-glycosidic bonds (see Cyclodextrins: From Nature to Nanotechnology, Molecular Recognition). The most commonly studied CDs are α-CD composed of six glucose units, β-CD composed of seven glucose units, and γ -CD composed of eight glucose units. The structure presents a hydrophobic cavity for binding of guests with primary hydroxyl groups from the 6-positions equal to the number of glucose units presented on the primary face of the open CD “bucket” and secondary hydroxyl groups from the 2- and 3-positions equal to twice the number of glucose units presented on the wider secondary face of the open CD “bucket.” CDs can form inclusion complexes with a variety of organic guest molecules and thus are of potential interest for chemical sensor development and studies of host–guest inclusion complex formation.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy There have been numerous papers examining the monolayer behavior of CD derivatives bearing alkyl chains and other constituents, and many of these studies have employed BAM. BAM is highly useful for determining miscibility of components in mixed monolayers and this has been applied to mixed monolayers of two related CD derivatives.43 Mixed monolayers of per-6-ammonium-per-6-deoxy-per-O2 ,O3 hexyl-β-cyclodextrin (CD1, see Figure 7) and the related derivative per-O2 ,O3 -hexanoyl-β-cyclodextrin (CD2) were examined. It was found that CD1 had a higher collapse pressure ( c = 45.3 mN m−1 ) than CD2 ( c = 37.4 mN m−1 ). In studies of mixed monolayers, comparisons are often made of the observed molecular area at a chosen surface pressure in the π –A isotherms versus that predicted from the molecular areas of the two pure components at the chosen surface pressure and assuming ideal mixing via the equation A12 = x1 A1 + x2 A2 , where x1 and x2 are the mole fractions of CD1 and CD2 in the monolayer respectively. Positive deviations from ideality are generally taken as indicating possible immiscibility between monolayer components. The excess free energy of mixing at a given temperature can also be calculated using the π –A isotherm data on a series of mixtures. In this particular study, a peak in the deviation of the molecular area from the ideal value was consistently observed near a mole fraction of compound CD1 of 0.60. BAM was reported to show phase separation by the emergence of a network-like structure with high-reflectivity “cages” surrounding regions of lower reflectivity. It is interesting to note that this was observed upon decompression, and indicates how BAM can follow hysteresis-like effects in monolayers where a different microstructure is seen on expansion than is observed during compression. Monolayers of the per-6-O-(tert-butyldimethylsilyl) derivatives of α-, β-, and γ -CDs (CD3, CD4, CD5, see Figure 7) were examined by BAM and π –A isotherms.44 In this study, BAM provided the key information needed to understand the problem encountered with reproducibility of the π –A isotherms for these monolayers. The isotherms of these compounds were exhibiting significant variation of ˚ 2 molecule in the “lift-off area,” the mean molec±30 A ular area at which the surface pressure first starts to rise. The application of BAM demonstrates how surface pressure isotherms alone cannot reliably indicate correct monolayer formation. The derivatives were spread from five different spreading solvents—chloroform, chloroform/ethanol (4 : 1 v/v), ethanol (which is not a spreading solvent), hexane, and hexane/isopropanol (7 : 3 v/v). BAM observations showed the presence of small crystalline aggregates a few microns in size even at high molecular areas just after spreading and well above the lift-off area. The formation of these small crystallites was attributed to supersaturation of the

17

amphiphiles during the solvent evaporation. Variation in their number for different experiments could account for the isotherm variability. BAM showed that the aggregates moved closer together and increased in number upon compression and increased greatly in number upon collapse. Spreading from the solvent hexane/isopropanol (7 : 3 v/v) resulted in the desired homogeneous monolayer, which was observed by BAM to remain homogeneous throughout compression until the collapse pressure was reached. Atomic force microscopy (AFM) (see Atomic Force Microscopy (AFM), Techniques) on derivatives transferred onto mica modified by an initial monolayer of cadmium stearate was generally consistent with the BAM results, showing less aggregation in the transferred films originally spread from hexane/isopropanol. It was determined that the monolayers of the CD derivatives spread from hexane/isopropanol were more suitable for use in sensor construction due to more consistent CD orientation and much less prevalence of aggregates. This study demonstrates the power of BAM to guide Langmuir–Blodgett film development, to assess reasons for irreproducibility in monolayer behavior, and to make an optimal selection of spreading solvent. Studies of unmodified native α-cyclodextrin at the water–air interface alone and together with sodium dodecyl sulfate (SDS) have recently been reported and have incorporated the use of BAM.54 Pure α-cyclodextrin was found to adsorb significantly and cover most of the interface and could be seen by BAM as a film with many small gaps, dark areas, and regions of differing reflectivity. The film appeared to be more fluid at 303 K and αcyclodextrin concentration of 2.5 mM than it did at 283 K and a concentration of 35 mM. Using isothermal titration calorimetry, it was found that α-cyclodextrin formed a complex with SDS of 2 : 1 stoichiometry. BAM observation of the water–air interface for the mixture of 35 mM α-cyclodextrin and 17.5 mM SDS at 286 K showed formation of a rigid continuous film, while upon lowering the concentrations to 5.0 mM α-cyclodextrin and 2.5 mM SDS the film was still fairly continuous and rigid but displayed some gaps. This study shows how BAM can reveal new phenomena in studies where a macrocycle is studied adsorbed at the water surface, and that the inclusion of a guest to form a stoichiometric complex can alter the properties of the adsorbed monolayer. These studies should be quite straightforward as the only requirement is the observation of the water surface of the solution at a controlled temperature. It could also prove of interest to sweep the surface clean and study the dynamics of the appearance of the adsorbed film. Measurement of surface pressure versus time for the adsorbing film could augment the experiment to assess if thermodynamic equilibrium has been reached. In this study, the transfer of the α-cyclodextrin films onto mica for subsequent imaging by AFM confirmed the existence of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

18

Techniques

dimer-like arrangements of α-cyclodextrin forming tubular structures. BAM is also uniquely suited to study the influence of the interaction of a macrocycle in the subphase with a spread monolayer on the monolayer morphology and structure.55 Monolayers of cholesterol, dimyristoylphosphatidycholine (DMPC), dimyristoylphosphatidylglycerol (DMPG), 3/1 DMPC/cholesterol mixtures, and 3/1 DMPG/cholesterol mixtures were examined using surface pressure isotherms, BAM, and polarization-modulation infrared reflection absorption spectroscopy (PM-IRRAS). These studies used a BAM-2plus and report making use of a feature of the I-Elli2000 software that enables easy reflectivity calibration from the CCD image gray scale and estimation of film thickness in regions of the image assuming fixed refractive index values for the water and for the film. This study contains film thickness estimations for different regions reported to have relative errors close to 5%; the estimate of these thicknesses makes use of the proportionality of the reflectivity to (nd )2 , where n is the film refractive index and d is the film thickness. Using this quantitative BAM approach, it was found that, after collapse, the ˚ thickcholesterol monolayer showed regions of 18 ± 1 A ˚ ness in coexistence with regions of 54 ± 3 A thickness, suggesting the formation of trilayer regions upon collapse of the cholesterol monolayer. The π –A isotherms of cholesterol monolayers spread on saturated β-cyclodextrin solution rose gradually starting from much higher molecular areas and attained only about half the surface pressure of cholesterol on water, with dramatic hysteresis seen on expansion. The BAM images of cholesterol compressed on the β-cyclodextrin solution showed regions of greater and more variable reflectivity, suggesting thicknesses of 40 ± 2, ˚ While the effect of sub60 ± 3, 80 ± 4, and 100 ± 5 A. phase β-cyclodextrin on the surface pressure isotherms of the two pure lipids was slight, BAM revealed the presence of β-cyclodextrin domains beneath these monolayers at low–moderate surface pressures. The isotherms of 3/1 DMPC and cholesterol on the β-cyclodextrin subphase showed a significant surface pressure arising at much higher molecular areas and BAM showed bright domains that persisted until a pressure of 43 mN m−1 was reached. The disappearance of these bright domains upon compression suggested that the β-cyclodextrin–cholesterol complex was being squeezed out of the interface. Similar behavior was found for the DMPG + cholesterol mixed monolayers. It is of particular significance that the lipid DMPC and DMPC/cholesterol mixed monolayers were chosen for these experiments as DMPC and the mixed monolayer are normally homogeneous; therefore, any brighter domains observed must be due to the interaction with the β-cyclodextrin in the subphase. Using the BAM imaging results together with PM-IRRAS data and calculations, a

model was presented in which the CDs orient parallel to the plane of the interface (the axes through the cavity being perpendicular to the interface) in a head-to-head and tail-to-tail fashion. The association with cholesterol was concluded to be through interaction with the CD cavity, while that with phospholipid was through hydrogen bonding to the lipid head group. The observation by BAM of thicknesses that ˚ was instrumental in arriving were close to multiples of 20 A ˚ is about twice the length of two CDs at this model as 20 A stacked on top of each other.

3.3

Other studies using BAM and amphiphilic macrocycles

Amphiphilic derivatives of crown ethers (see Crown and Lariat Ethers, Molecular Recognition) and cryptophanes (see Cyclotriveratrylene and Cryptophanes, Molecular Recognition) have been a part of monolayer studies involving the use of BAM and provide additional examples of how BAM can be applied to supramolecular chemistry in monolayers. Chiral amphiphilic crown ethers (Figure 7)45 were spread as monolayers and their recognition of the enantiomers of the amino acids valine, alanine, tryptophan, and phenylglycine introduced into the subphase was evaluated. BAM showed that the chiral crown ether with two benzyl groups collapsed by forming multilayer islands, while closely related derivatives bearing alkyl chains instead collapsed via a folding mechanism at the water surface. The monolayers were spread on acidic subphases containing the L or D enantiomers of each of the four amino acids of concentration 0.0325 mM. Chiral discrimination effects were noted by observation of modest but significant differences in isotherm parameters such as compressibility modulus, collapse pressure and area, and surface potential. However, no changes of interest were noted in features observed by BAM. Thus, it is possible that a monolayer system can show interesting and quantitatively significant recognition phenomena of subphase species but no noteworthy changes under BAM. Larger-scale reorganization of the monolayer in response to recognition events is needed to see visually significant changes using BAM. BAM is an especially useful method for helping in characterization of the behavior of molecules at the water–air interface that do not exhibit stable monolayer behavior but are instead subject to relaxation effects, which are generally due to slow aggregation. Monolayers of the compound N,N  -dihexadecyl-4,13-diaza-18crown-6 (ACE-16, see Figure 7) were examined alone and in mixtures with palmitic acid; these exhibited significant surface pressure relaxation effects.46 In this study, BAM is used to distinguish three scenarios—dendritic domain growth during relaxation, monolayer collapse, and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy separation of components due to immiscibility for the case of mixed monolayers of ACE-16 and palmitic acid. The π –A isotherm for ACE-16 showed a plateau starting at ˚ 2 per molecule that extended 30 mN m−1 and near 80 A down to areas much less than any realistic profile of the ˚ 2 per molecule for the molecule that would be close to 43 A crown ether parallel to the surface and the alkyl chains perpendicular. When the monolayer was compressed to 32 mN m−1 and then held at fixed area, a relaxation of the surface pressure to 25 mN m−1 was observed over a period of 40 min. BAM showed the growth of dendritic domains during this period after a brief 2-min induction period; from their appearance, it was likely that these domains were two dimensional. In contrast, compression to 50 mN m−1 was followed by relaxation of the surface pressure to close to 40 mN m−1 with BAM showing formation of aggregates assigned as most likely being due to monolayer collapse. The compression of 1 : 1 mixed monolayers of ACE-16 and palmitic acid to either 23 or 37 mN−1 followed by holding at fixed area showed significant surface pressure relaxation. In the case of mixed monolayers of ACE-16 and palmitic acid, BAM showed the emergence of many small point-like domains that were assigned as a possible demixing of the palmitic acid from the ACE-16. The binding of a cation to a crown ether amphiphile can significantly alter the domain microstructure observed by BAM. In the case of monolayers of benzodithia-15crown-5-styryl dye,56 BAM showed large islands on 1 mM AgClO4 indicating coexistence of a condensed phase with a gas phase. In contrast, on 1 mM NaCl subphase, the appearance of circular domains of the gas phase surrounded by brighter regions suggested that in this case the coexistence was between a LE phase and a gas phase. On pure water, the domains were more irregular but clearly showed coexistence of a darker two-dimensional gas phase with a brighter pattern. The assignment of LC phase, as being present at low surface pressures, is generally done if one sees other than round or fluid boundaries between the brighter and darker regions by BAM; if these boundaries appear rounded and smooth, or if a two-dimensional foam-like arrangement is observed, then the phase is generally assigned as LE. Cryptophanes are examples of a molecular structure that one might not expect to form monolayers at the surface of water given their roughly spherical profile. However, their structure possesses well-defined regions that are either polar or nonpolar; as the upper and lower cyclotriverarylene caps are nonpolar, the two belts of linking ether bonds are polar and the connecting short alkyl chains are nonpolar. In this study, a series of four cryptophane compounds (Figure 7) with varying lengths of the alkyl chain linker (C3, C5, C9, and C10) and pure anti -stereochemistry for C3 and C5 and mixed anti - and syn-stereochemistry for

19

C9 and C10 were spread on the water surface.47 The π –A isotherms measured as compression–expansion cycles were reported over two ranges, termed “long-range” from 250 to ˚ 2 per molecule and “short-range” from 250 to 145 A ˚2 65 A per molecule. The long-range isotherms displayed a rise followed by a plateau and then another rise in surface pressure. The long-range isotherms showed large hysteresis, while the short-range isotherms showed significantly less hysteresis. The surface pressure plateau, which resembles that associated with first-order phase transitions with a coexistence region, was found to represent a range of gradual formation of three-dimensional aggregates. This is an important distinction to be aware of when interpreting π –A isotherms that can be greatly aided by the use of BAM. While dramatic surface pressure hysteresis and considerations of molecular dimensions suggested this conclusion, BAM can definitively confirm whether during a surface pressure plateau one is observing a firstorder phase transition or formation of three-dimensional aggregates signifying monolayer collapse. In this study, the C5-cryptophane developed a grainy pattern in the plateau region, while the C9-cryptophane developed bright domains of variable sizes that were oriented parallel to the barrier. It was concluded that the cryptophanes studied only formed monolayers up to modest surface pressures of 7–8 mN m−1 , except 12 mM m−1 for the C3 cryptophane. Ellipsometry measurements, which do not provide images, gave monolayer thicknesses consistent with the dimensions of these cryptophanes. It is important in monolayer studies to correctly distinguish collapse from actual two-dimensional phase transitions. A study of amphiphilic derivatives of a molecular clip and a molecular tweezer provided an additional example of how BAM can aid the study of host–guest complexation in monolayers.57 The molecular clip and tweezer derivatives had either a 1,4-diacetoxybenzene derivative or a 1,4diacetoxynapthalene spacer. The guest molecule chosen was 1,2,4,5-tetracyanobenzene (TCNB) and was spread as a 1 : 1 mixture in chloroform with the host molecule. The two molecular clips and the narrower molecular tweezer showed fluid-like domains during compression as observed by BAM, while the wider molecular tweezer derivative showed solid-like domains with sharper edges. The presence of TCNB did not influence the π –A isotherms of the two molecular clip derivatives. The mean molecular areas from the π –A isotherms were consistent with orientation of the sidewalls of the clips or tweezers perpendicular to the water surface. Comparison of the BAM images of monolayers of the narrower molecular tweezer derivative with and without TCNB showed significant differences. The pure tweezer at low surface pressure showed shapeless wavy aggregates, while the complex showed a network of smaller, bright, fractal-shaped aggregates. At higher pressures, the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

20

Techniques

monolayer of pure tweezer compound showed LE-like domains with smooth bright bands or smooth boundaries, while the monolayer of the complex showed smaller fractal-structured domains. The monolayer of the narrower tweezer compound complexed with TCNB also showed anisotropy upon rotation of the analyzer such that contrast inversion could be observed in a selected small group of domains. The observation clearly indicates that the domains of the molecular complex of the tweezer compound with TCNB have anisotropic orientations.

4

APPLICATION OF BREWSTER ANGLE MICROSCOPY TO THE STUDY OF MONOLAYERS OF CHIRAL COMPOUNDS

In contrast to the issues of interest in applying BAM to the study of monolayers of macrocyclic amphiphiles, studies of chiral monolayers using BAM are primarily concerned with how molecular chirality is manifested in terms of the mesoscopic chiral features of condensed-phase domains (see Chirality, Concepts). Many of the studies of chiral monolayers focus on chiral discrimination effects, which may be broadly classified as being either homochiral (interaction between like enantiomers is more favorable than that between opposite enantiomers) or heterochiral (interaction between opposite enantiomers is more favorable than that between like enantiomers), by comparing the domain morphology and isotherm behavior of monolayers of pure enantiomers with those of racemic mixtures.58, 59 Of particular interest in studying monolayers of racemic mixtures is whether visual evidence for separation of enantiomers in two dimensions can be observed for the case of homochiral discrimination. In the case of homochiral discrimination, the π –A isotherms of the pure enantiomers appear more condensed than those of the racemic mixture, while the opposite is observed for the case of heterochiral discrimination. The π –A isotherms of the two pure enantiomers should be equivalent, and this must be verified in these studies. Monolayers of diastereomers of molecules with more than one chiral center have also been examined in a smaller number of studies, and these systems can be quite complex. Methods frequently applied to study these systems in conjunction with BAM in studies of monolayers of chiral amphiphiles are IRRAS to assess molecular ordering, and GIXD for the determination of two-dimensional lattice structures. In this section, attention is focused on chiral discrimination in synthetic amphiphiles. Chiral domain formation has been observed in phospholipid monolayers, dating back to the pioneering work of McConnell2 using fluorescence

microscopy in the 1980s; many of the earlier studies of chiral discrimination in monolayers were carried out using fluorescence microscopy. In studies of monolayers of chiral amphiphiles with a single chiral center, it has often been the case that the domains of the individual enantiomers exhibit curved structures with the sense of the curvature being unique to the handedness of the chiral center. If such systems have been found to exhibit homochiral discrimination, as usually assessed by observation of more condensed π –A isotherm behavior for the pure enantiomers than for the racemic mixtures, interesting chiral segregation effects can often be observed. These studies are generally carried out for chain lengths and temperatures for which a clearly defined phase transition from a LE phase to a two-phase coexistence with the emerging condensed phase is observed. In BAM observation of racemic monolayers, the simultaneous appearance of domains with both senses of curvature has been seen. Monolayers of chiral amphiphiles that show heterochiral discrimination do not show such chiral segregation and show other complex domain shapes. In the case of homochiral discrimination, observation of curved domains is not ubiquitous, and in many cases branched growth forms are observed instead. BAM studies applied to any supramolecular system containing chiral centers should be cognizant of the possibility of observing domains with chiral features. Such curvature can be explained on the basis of anisotropy of the variable known as the line tension, which is the two-dimensional equivalent of surface tension, and exists at the interface between the condensed-phase domains and the surrounding LE phase. A long-range twisting of the molecular orientation of tilted chiral molecules in the condensed phase can also explain the bending of domains. Many of the systems studied have been amphiphilic amino acid derivatives, and amide–amide hydrogen bonding promoting that long-range ordering in these monolayers is a major force resulting in the formation of large curved domains. One of the earliest studies on the use of BAM examined monolayers of the racemic mixture and of the pure L- and 60 D-enantiomers of N-dodecylgluconamide. This system ◦ presented π –A isotherms at 10 C that were more condensed for the racemic mixture than for the pure enantiomers, which were equivalent; however, at 25 ◦ C, the opposite was true, although the isotherms crossed at a certain point. Significant surface pressure relaxation at fixed area was observed, which differed in extent between the enantiomers and the racemic mixture and was much greater at 25 ◦ C than at 10 ◦ C, as expected on the basis of the short chain length. BAM observation during compression of either of the pure enantiomers at 25 ◦ C showed extensive growth of two-dimensional dendritic domains. Differences could not be found between the dendritic forms of the two enantiomers. The monolayers of the racemic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy mixture gave an indistinct grainy pattern on compression. The domains seen in monolayers of these pure enantiomers were nonequilibrium growth forms that did not manifest chiral differences. Chiral segregation was not observed in the racemic mixture, at least on the length scales probed by BAM. It must be kept in mind in BAM studies that structures on length scales below the lateral resolution will not be seen by the method. Chiral single-chain molecules capable of very strong lateral hydrogen bonding are more likely to exhibit this sort of extensive dendritic growth on compression. In a following study, monolayers of the enantiomers of N-dodecylmannonamide and of the racemic mixture were examined and very different results were obtained than those found for the enantiomers of Ndodecylgluconamide.61 In this case, the surface pressure isotherms clearly show homochiral discrimination. BAM observation of the monolayers of the pure enantiomers showed formation of feather-like dendrites with curved main growth axes and with curved side branches. The side arms were observed to curve exclusively counterclockwise for the L-enantiomer and clockwise for the D-enantiomer.

(a)

21

BAM allowed observation of the growth of these curved side arms in real time. BAM observation of the monolayers of the racemic mixture showed dendritic growth with the side arms curved in either direction in equal proportions. Chiral discrimination effects have been observed using BAM in monolayers of a range of single-chain Nacylamino acid derivatives,62 some of which had previously been imaged using fluorescence microscopy. The domain morphologies observed by BAM and by fluorescence microscopy were in agreement. BAM observation of monolayers of N-stearoylserine methyl ester (SSME) enantiomers showed distinct curved domains resembling spirals that curved clockwise for the D-enantiomer and counterclockwise for the L-enantiomer, as seen in Figure 10. Racemic monolayers of SSME showed a number of twinned domains showing both senses of curvature as well as larger flower-like domains with curved segments growing around their peripheries either in a clockwise or counterclockwise sense. These observations were taken as evidence for chiral segregation in the racemic monolayer.

(b)

(c)

Figure 10 Images of domains of monolayers of the D-enantiomer, L-enantiomer, and racemic mixture of N-stearoylserine methyl ester clearing showing chiral discrimination effects. The domains of the enantiomers display unique curvature, and the domain of the racemic mixture shows features with both senses of curvature and hence evidence for chiral segregation. Reproduced from Ref. 62.  American Chemical Society, 2003, and the figure caption reads as follows: Chiral discrimination of the condensed-phase domains of N-stearoyl serine methyl ester monolayers spread on pH 3 water. (a) D-enantiomer (b) L-enantiomer (c) and (d) 1 : 1 DL racemate. Image size 80 × 80 µm. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

22

Techniques

The π –A isotherms for these monolayers indicated homochiral discrimination. Similar curved spiral domains were seen in monolayers of N-palmitoylaspartic acid. Very striking, curved, condensed-phase domains were observed by BAM in monolayers of N-α-palmitoylthreonine, curving with opposite sense for the two pure enantiomers and showing twin-like structures with one arm curving in each sense for the racemic mixture paper.63 Further studies of chirality-related effects using BAM pointed out the complexities involved in considering how the structure of the observed domains relates to the molecular chirality but is also strongly and often dominantly affected by domain nucleation and growth rates.64 The presence of Zn2+ in the subphase was found to have dramatic effects. Monolayers of N-hexadecanoyl-L-alanine spread at 298 K showed large islands of the condensed phase upon spreading that resembled floating ice packs. These experiments were carried out with a 4 mm × 6 mm field of view (provided by the NFT Mini-BAM instrument); some of these domains were a few millimeter wide. The phases upon spreading were the condensed phase and the gas phase, and

the π –A isotherm did not exhibit a plateau at this temperature. Upon compression, the condensed-phase domains covered more of the surface and the gas was squeezed out as they merged, in accordance with the thermodynamic lever rule for two-phase coexistence. The domains seen in these monolayers were sensitive to temperature and a change from 298 to 303 K altered the morphology entirely such that the condensed phase appeared as hook shapes that were chiral and whose two arms defined axes of dendritic growth as the compression continued. This observation underscores the need for temperature control during monolayer experiments and BAM observations especially since the π –A isotherm was essentially the same at these two temperatures. Monolayers of N-hexadecanoyl-D,L-alanine showed a plateau in the π –A isotherm and the condensedphase domains that formed were highly branched fractal shapes and could be as large as 10 mm in size, as seen in Figure 11. Visual evidence of chiral phase separation could not be found in this racemic monolayer despite the differences in the π –A isotherms and IRRAS data showing a homochiral

60 Π (mN m−1)

50 40 30 E

20

D

A-C

10 0 0.0

0.2

0.4

0.6

A (nm2 mol−1)

0.8 (a)

(b)

(c)

(d)

(e)

Figure 11 Reproduced from Ref. 64.  American Chemical Society, 2005, and the figure caption reads as follows: /A isotherm and corresponding BAM images (A–E) of N-hexadecanoyl DL-alanine methyl ester on a pure aqueous subphase (pH 6, T = 298 K). The images were recorded at the points indicated on the /A plot. They represent an area of W × H = 6 mm × 4 mm. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy preference based upon greater ordering of the hydrocarbon chains. This illustrates that the domains seen here are nonequilibrium growth forms and have grown so rapidly that the two-dimensional diffusion required for opposing enantiomers to form separate domains was not possible. The behavior of N-hexadecanoyl-L-alanine on a subphase containing Zn2+ was strikingly different as domains of the condensed phase showed chiral S-shaped domains, torusshaped domains, and seahorse-shaped domains. Monolayers of N-hexadecanoyl-L-alanine methyl ester and of Nhexadecanoyl-D,L-alanine methyl ester were also studied and the condensed-phase domains observed were primarily dendritic growth forms. A consideration often overlooked in BAM studies is the possible influence of the compression rate on the domain structures. In the case of N-acylamino acid monolayers, the associations due to amide–amide hydrogen bonding are very strong and promote rapid domain growth and also make it unlikely that relaxation to an equilibrium domain shape can occur on any realistic experimental timeframe. Domain shape relaxation kinetics are noted to be dependent on the strength of intermolecular forces; for example, dendritic condensed-phase domains formed in a phospholipid monolayer required 5 h to relax to equilibrium shapes and, for the phospholipid DMPE, compression rates ˚ 2 per molecule per minute were needed as slow as 0.2 A to observe equilibrium domain shapes.21 Examination of the variation of domain structure with time after their formation or with compression rates are not commonly reported; however, it is advisable to consider examining these variables when carrying out BAM experiments. The significance of the amide–amide hydrogen bonding in monolayers of the amino acid derivative Nstearoylvaline was assessed by comparing the behavior of monolayers of N-stearoyl-L-valine with N-stearoylN-methyl-L-valine.65 While dendritic growth forms were observed by BAM upon compression for monolayers of N-stearoyl-L-valine, methylation of the amide nitrogen that removes the prospects for amide–amide hydrogen bonding resulted in the observation of condensed-phase domains that were irregular, mostly rounded shapes. In the case of monolayers of the compound N-tetradecylγ ,δ-dihydroxy-pentanoic acid amide (TDHPAA), chiral discrimination effects in the domain structures were observed by BAM.62 The small, spear-like crystals showed a longer and shorter branch at one end and were seen to be mirror images of each other when the two enantiomers were compared. The racemic mixture formed a symmetric, almost straight and narrow crystallite for its condensed phase. Monolayers of the R-enantiomer of 3-palmitoyl-snglycerol were found to behave differently than monolayers of 3-palmitoyl-rac-glycerol under BAM observation,

23

although in a more subtle manner that required observation of domain anisotropy making use of the analyzer of a BAM.66 The condensed-phase domains of the Renantiomer were almost round and were divided neatly into seven segments that appeared with different reflectivities, indicating that the azimuthal tilt direction is different in each pie-piece-shaped segment. The molecules within these domains were all tilted by the same angle and the tilt was directly radially; however, the azimuthal direction of the tilt direction jumped discontinuously at the borders between the domain segments. At 20 ◦ C, the Renantiomer and the racemic mixture behaved identically under BAM observation and also exhibited identical surface pressure isotherms. At a lower temperature of 5 ◦ C, a phase transition was observed for the racemic mixture that was not observed for the R-enantiomer in that BAM showed changes in the azimuthal orientations upon compression inside the racemic domains. A kink in the surface pressure isotherm for the racemic compound was seen that was not seen in the isotherm of the enantiomer. GIXD data found that the enantiomer was ordered in an oblique lattice, while that for the racemic mixture was rectangular. Observing this difference in domain behavior and detecting the phase transition using BAM required attention to the effect of rotating the analyzer. A derivative of 1-stearoyl-rac-glycerol bearing a hydroxyl group on the 12position showed very different behavior and BAM images of monolayers of this compound at 6 ◦ C showed domains with several large curved arms that were sometimes closed into rings and were of either sense of curvature.67 The tendency to form curved domains diminished with increasing temperature. Thus, it is important to conduct studies at a series of temperatures as condensed-phase domain structures can often vary significantly. Monolayers of ethyl 4-fluoro-2,3-dihydroxystearate diastereomers were examined recently and provide an example of a more complex chiral system as these molecules contain three chiral centers.68 It was not possible to produce all of the possible diastereomers in pure form in this case. Four synthetic products were studied as monolayers: (a) a 69 : 31 ratio mixture of the (R,R,S)/(R,R,R) enantiomers referred to as RDIA, (b) a 67 : 33 ratio mixture of the (S,S,R)/(S,S,S) enantiomers referred to as SDIA, (c) a 60 : 40 ratio mixture of the two enantiomers pairs (R,R,S)/(S,S,R) and (R,R,R)/(S,S,S) referred to as RAC, and (d) a pure enantiomer whose absolute configuration was not determined referred to as ENAN. Surface pressure isotherms at 20 ◦ C showed that the ENAN compound gave a highly condensed isotherm, while RAC was the most expanded and showed signs of a phase transition near 15 mN m−1 . Monolayers of SDIA and RDIA were intermediate in their behavior and fairly close to each other in surface pressure. BAM showed different morphologies

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

24

Techniques

for all four of these monolayer systems. The RAC monolayer formed rounded domains with fringes. Both SDIA and RDIA monolayers formed long, curved needle structures. The domains of ENAN were much smaller, elongated bead shapes and were present upon spreading. Their appearance immediately on spreading was consistent with the surface pressure isotherm, which suggested finding the gas and condensed phases at low pressure. Additional information was obtained from tapping mode AFM study of these monolayers transferred as Langmuir–Blodgett films and from molecular modeling. The structural differences between systems are a challenge to unravel, but BAM can certainly provide some directly relevant information concerning two-dimensional organization at the water surface.

5

5.1

APPLICATION OF BREWSTER ANGLE MICROSCOPY TO THE STUDY OF MONOLAYERS IN WHICH HYDROGEN-BOND COMPLEX FORMATION OCCURS AT THE WATER SURFACE Molecular recognition by melamine amphiphiles

Amphiphilic derivatives of melamine of the form 2,4-di(nalklyamino)-1,3,5-triazine (2Cn H2n+1 -melamine) present three hydrogen-bond donors and two hydrogen-bond acceptors when spread at the water–air interface. The recognition of small soluble species with complimentary hydrogen bonding by these melamine amphiphiles has been studied in monolayers using surface pressure isotherms BAM and GIXD.48, 69–71 Binding to both faces of the melamine by the recognized molecule could result in formation of linear hydrogen-bonding networks at the water–air interface. The recognition of thymine by monolayers of 2C11 H23 -melamine (Figure 7) was examined.69 Monolayers of C11 H23 -melamine alone showed surface pressure isotherms with a plateau, indicating a first-order phase transition from a LE phase into a coexistence with a condensed phase. The monolayers were studied over the temperature range 10.2–31.9 ◦ C, and the surface pressure of the plateau increased regularly with temperature. The monolayers collapsed at the end of the plateau at 31.9 ◦ C. The observation of the first-order phase transition leads to the expectation of observing well-defined, condensed-phase domains by BAM in the coexistence region and such was the case in this study. In the case of amphiphiles with a head group and hydrocarbon chain structure, the temperature range

for observing the first-order phase-transition plateau can be adjusted into the experimentally accessible temperature range by adjusting the hydrocarbon chain lengths with longer chain lengths, generally increasing the temperature at which the transition plateaus will be observed. The authors previously studied monolayers of 2C10 H21 -melamine and 2C12 H25 -melamine alone using surface pressure isotherms, equation of state calculations, BAM and GIXD.71 For monolayers of 2C11 H23 -melamine alone, BAM showed the appearance of numerous compact domains of similar size in the coexistence regions that grew larger under compression. The monolayers of 2C11 H23 -melamine were then studied on subphases containing thymine (0.05 or 0.10 mM) and striking differences were observed. The surface pressure isotherms depended upon compression rates such that, the slower the rate, the greater was the reduction in the pressure of the plateau, which could be lost altogether at a low enough compression rate. Monolayers of 2C11 H23 melamine on thymine subphases are not in equilibrium ˚2 under compression except at the slowest used rate of 1 A per molecule per minute. This indicates that time is required for the completion of the recognition of the dissolved thymine by the 2C11 H23 -melamine monolayer. BAM observation of 2C11 H23 -melamine on 0.1 mM thymine subphase ˚ 2 per molecule per minute at 27 ◦ C to and compressed at 1 A 2 ˚ per molecule over a 30-min period showed the forma80 A tion of large, bright, symmetric, dumbbell-shaped domains as large as 400 µm. The interiors of these domains had a subtle texture of smoothly varying brightness that indicated the presence of anisotropy most likely associated with a small molecular tilt varying in direction within the domains. The formation of domains on the 2C11 H23 -melamine could also be followed by stopping compressions carried out at slightly higher rates and then following surface pressure relaxation and BAM observation. Compression stopped at 13.5 mN m−1 , showing formation of narrower dumbbells during relaxation, while stopping at 15.6 mN m−1 showed formation of many small irregularly branched domains that merged into a network. GIXD indicated a centered rectangular lattice with a small molecular tilt toward next nearest neighbors. The same lattice was seen on the aqueous thymine subphase but with different dimensions. These studies show the promise of BAM for studying recognition processes with hydrogen bonding between the amphiphile and a subphase species. It is also clear that such a system can display very rich behavior as function of the physical and chemical parameters and that attention must be paid to questions of appropriate compression rates and approaches to equilibrium. The BAM observations can clearly show dramatic differences in domain structures due to molecular recognition at the water surface. A further study compared the recognition of uracil, which can hydrogen-bond to one face of the 2C11 H23 -melamine

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy molecules, with the recognition of barbituric acid, which can bind to both faces by 2C11 H23 -melamine monolayers.48 The presence of 0.5 mM uracil in the subphase had a similar effect as that seen in the study using thymine in that its complete recognition by the melamine amphiphile resulted in disappearance of the surface pressure plateau and the formation of very different domain structures, as seen in Figure 12. On the uracil-containing subphases, many round domains were seen with a gradually varying inner texture, indicating a varying molecular tilt direction within the domains. Over time, the number of domains decreased and the larger domains expanded in size significantly at the expense of the smaller ones, a process generally known as Ostwald ripening and seen in many situations involving nucleation and growth of domains. The isotherms of the 2C11 H23 melamine monolayers were shifted to lower areas on a 0.01 mM barbituric acid subphase, indicating the formation of a more condensed structure. BAM observation of 2C11 H23 -melamine monolayers on barbituric acid containing subphase showed the formation of large, homogeneously bright regions that could merge and become uniform. This observation was viewed as being consistent with formation of an extensive hydrogen-bonded network between melamine and barbituric acid. As an interesting reminder of how modest changes in molecular structure can result in completely different domain structures, 2C11 H23 melamine was modified to 2C12 H25 -O-(CH2 )3 -melamine. On aqueous uracil subphases, the domains formed for this modified melamine amphiphile showed a distinctly varying inner texture indicating greater anisotropy. Their appearance became more frayed at the peripheries upon increasing the uracil concentration from 3 to 5 mM. On 0.1 mM aqueous barbituric acid subphases, this modified melamine amphiphile showed starfish-like domains with an irregular number and size of arms in which each arm appeared with a different brightness, indicating different azimuthal orientations within each arm. The weakening of the hydrogenbonding network was cited as a possible reason for these different observations. A subsequent study developed a kinetic model of the recognition process that could fit the surface pressure relaxation data.70

5.2

Highly ordered monolayers formed by adsorption from the subphase

A particularly strong advantage of the BAM method is that it can be used to directly image monolayers formed by adsorption from the subphase. Phase transitions as a function of surface coverage, related to the subphase concentration, adsorption time, and temperature can be observed for such adsorbed monolayers. In one series of studies, the

25

(a)

(b)

(c)

Figure 12 Brewster angle microscopy images of the condensedphase domains of the melamine amphiphile from the studies described in Section 5.1 shown on different subphases illustrating the effect of molecular recognition by lateral hydrogen bonding on domain structure. Reproduced from Ref. 70.  American Chemical Society, 2005, and the figure caption reads as follows: BAM images of characteristic condensed-phase domains of C11H23-melamine monolayers spread on water (a), 0.5 mM uracil subphase (b), and 0.1 mM thymine subphase (c). Compression rate per molecule, 0.04 nm2 min −1 ; T, 25 ◦ C; image size, 400 µm × 400 µm.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

26

Techniques

behavior of adsorbed monolayers of molecules containing amide groups and thus capable of strong lateral hydrogen bonding was examined by BAM and surface pressure measurements. The molecule N-dodecyl-γ -hydroxybutyric acid (DHBAA) was investigated, both as adsorbed and spread monolayers.49 The chain length here is chosen to be long enough to allow formation of spread monolayers but short enough to allow sufficient solubility such that solutions can be prepared from which adsorbed monolayer formation can be studied. BAM observation of the adsorption process was carried out by sweeping the surface of the DHBAA solution clean and then measuring the change in surface pressure as a function of time for close to 1 h while simultaneously observing the interface with BAM. The measurements were carried out at a series of temperatures from 5 to 30 ◦ C and at 10 ◦ C and adsorption was followed at a series of DHBAA concentrations from 0.01 to 0.01 mM. The measurement of surface pressure versus time during adsorption at 10 ◦ C showed a sharp inflection point near 16 mN m−1 at all five concentrations studied, and the time required to observe the inflection decreased with increasing concentration. These observations defined a critical surface pressure π c and critical time tc for the transition; at tc , the surface pressure that had been rising steadily abruptly changed to increasing with a very gradual slope. The transition was not observed above 15 ◦ C. At 15 ◦ C, BAM showed the formation of highly ordered condensed-phase domains with four growth directions that grew in size until covering most of the field of view. In this experiment, measurement of the integrated BAM reflectivity as a function of time was performed. The reflectivity remained flat for about 10 min after the surface pressure inflection and then steadily climbed, although with up and down fluctuations. After the induction time following the phase transition, there were condensedphase domains sufficient in size and number to increase the reflectivity. The integrated reflectivity represents the light reflected from the entire illuminated spot on the water surface. As the monolayer domains are slowly drifting about on the water surface, their exact numbers under the laser spot will fluctuate with time, and this accounts for the somewhat jagged appearance of the increase in reflectivity versus time. DHBAA was then used to form spread monolayers and π –A isotherms were measured by compression. Given the short chain length of the species, high compression rates were used to avoid dissolution. In the compression isotherm, surface pressure plateaus were observed, indicating a first-order phase transition and a two-phase coexistence region. The surface pressure at onset of the phase transition in the spread monolayers increased with temperature, the molecular area at onset decreased, and the width of the plateau region also decreased, signifying the coexistence of two phases. The onset surface pressures for spread monolayers were close to those for the surface pressures

observed at the inflection points in the adsorbed monolayers. The morphology of the condensed-phase domains formed upon compression of the spread monolayers was similar in characteristic features to that seen formed in the adsorbed monolayers. A related study compared spread and adsorbed monolayers of the molecules N-(γ -hydroxypropyl)tridecanoic acid amide (HTRAA) and N-(γ -hydroxypropyl)tetradecanoic acid amide (HTEAA).50 HTEAA is water insoluble and more suitable for GIXD. This study included the application of GIXD, which revealed an oblique lattice structure independent of whether the monolayers were formed by spreading or adsorption. The dendritic appearance of the condensed-phase domains under BAM was attributed to the strong hydrogen bonding present between these molecules. These studies suggest that the prospects for studying adsorbed monolayers of supermolecules and their complexes directly as adsorbed monolayers should be considered and that BAM is well suited for this purpose.

6

6.1

EMERGING DEVELOPMENTS IN THE APPLICATION OF BREWSTER ANGLE MICROSCOPY Time-resolved Brewster angle microscopy

A recent extension of BAM involves equipping the setup with a pulsed Nd-YAG laser that can deliver a 5 ns pulse of 355 nm wavelength and 60 mJ energy radiation to the water surface from a direction perpendicular to the water surface.72, 73 The same spot as subjected to the pulse is then observed by BAM using a 532 nm laser illumination at the Brewster angle using a CCD camera that can record an image every 40 ms and with a 1 ms shutter time. This pulse and then probe (image) version of BAM was applied to monolayers of a spiropyran derivative that underwent a conversion to a merocyanine form upon illumination but relaxed thermally back to the spiropyran form.73 The technique made observing a number of photochemically induced changes possible using BAM. Collapsed aggregates of the spiropyran form were observed to transiently spread into larger, thinner domains of the merocyanine form. The merocyanine form has a more polar head group and a different alkyl chain orientation that makes it able to spread better on the water surface. The spreading was transient due to conversion back to the spiropyran over a few hundred microseconds. A transient wave structure induced by the laser pulse was observed and attributed to jumping the monolayer from a gas phase for the spiropyran to an LE and gas state. The modification of BAM to include pulsed laser illumination followed by high-speed imaging should find

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc040

Brewster angle microscopy application to additional photophysical and photochemical studies in monolayers.

6.2

Light-scattering microscopy coupled with Brewster angle microscopy

The method of light-scattering microscopy (LSM) was introduced and can provide information that is new and complimentary to that provided by BAM. The LSM method has not yet been widely applied to monolayers of supramolecular compounds and seems well suited to provide information especially for those systems subject to aggregation effects. LSM is an application in which the BAM setup is modified by adding an intensified CCD camera perpendicular to the water surface and a microscope objective that can collect light that is scattered normal to the surface, while the reflected light is detected and imaged by BAM. The microscope and camera perpendicular could also be used for fluorescence microcopy together with BAM and LSM to give a powerful combination of techniques. Particles or aggregates of 100 nm or greater in size should be able to be imaged by LSM using the same low-power lasers such as a 10 mW He–Ne laser as used for BAM illumination. This method could be applied to a number of problems, and revealed some new, remarkable monolayer phenomena.73 LSM was applied to observe the formation of biominerals beneath phospholipid monolayers. The formation of calcium oxalate crystallites underneath phospholipid monolayers in the LE + LC region was readily observed. Streptavidin crystals growing beneath monolayers of the synthetic lipid 1,2-dioleyl-rac-glycero-3-(8-(3,6-dioxy) octyl-1-amino-diacetic acid (referred to as Cu-DO-IDA) were simultaneously imaged using BAM and LSM. The protein streptavidin coordinates to this lipid through two histidines, forming a complex with a Cu2+ ion from the subphase that also complexes to the lipid head group. LSM was able to discern vacancies within these two-dimensional streptavidin crystals that grew beneath the Cu-DO-IDA monolayers. LSM revealed a remarkable topographic instability previously unknown for phospholipid monolayers. LSM images of monolayers of DPPC compressed through the coexistence region of the LE and LC phases and into the LC phase showed corrugated patterns of scattering originating around where the boundaries of the LC domains had been located.74 These monolayers were transferred to mica for AFM analysis in a tapping mode, and regions that represented budding, a bulging of lipid up away from the surface, were found. These buds proliferated rapidly during collapse but, in the LC phase below collapse, represented a small fraction of five-membered ring structures > monosubstituted

benzenes > disubstituted benzenes > trisubstituted benzenes, which correlate with the guest’s cross section (Figure 18). Among the disubstituted benzenes, the 1,4substitution pattern is strongly preferred over the 1,3- and 1,2-patterns. The following example nicely highlights the high shape selectivity among isomeric guests. When empty 6 was heated for four days in 98% HPLC-grade o-xylene, a mixture of 6
o-xylene, 6
m-xylene, and 6
p-xylene formed, of which the latter hemicarceplex dominated.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands In >99% m-xylene, a 2 : 1-mixture of 6
m-xylene and 6
p-xylene was produced. Thus, the ability to complex isomeric xylenes increases in the order o-xylene < m-xylene  p-xylene. Increasing the interhemispheric bridges by one atom, as in hemicarcerand 7 and 9 (Figures 2 and 18), strongly increases structural recognition in complexation studies and substantially narrows the range of guests that form stable isolatable hemicarceplexes 7
guest and 9
guest.36, 38 For example, 9 formed stable hemicarceplexes with structurally related norbornane derivatives 50–57 and some trisubstituted benzenes, but, remarkably, failed to complex 2-chloro- or 2-bromonorbornane, which are only slightly larger than 50–57, or with many mono- and disubstituted benzenes, even though they fitted comfortably into CPK models of 9.38 Molecular mechanical calculations suggest that constrictive binding is small and, more importantly, does not involve French door gating for the latter complexes (Figure 9). Thus, guests enter the inner phase, but also depart rapidly.71 However, once French door gating starts contributing to the lowering of constrictive binding, 9
guest can be isolated without decomplexation and constrictive binding becomes strongly guest-size-dependent to the extent that a small structural change, such as the substitution of the OH in 53 for a Cl, increases constrictive binding enough to prevent complexation. In addition to the linker groups, the intrahemispheric spanners also influence the molecular recognition properties of a hemicarcerand. Recently, Cram et al. reported the syntheses and binding properties of 13 (MM), 58 (EE), 59 (PP), 60 (EM), 61 (PM), and 62 (PE) (Figure 20).41, 88 These hosts have either methylene (M), ethylene (E), or propylene (P) spanners in one cavitand. In the crystal structure of cavitand 63 with propylene spanners, two spanners are outward and two are inward (Figure 20b).41 As a consequence, the P bowl is more rectangular shaped and deviates substantially from C4 R

R R

13: A = CH2; B = CH2 (MM) 58: A = (CH2)2; B = (CH2)2 (EE) 59: A = A = (CH2)3; B = (CH2)3 (PP) 60: A = (CH2)2; B = CH2 (EM) 61: A = (CH2)3; B = CH2 (PM) 62: A = (CH2)3 B = CH2 (PE) A Br Br

O OB BO O B B OO O O O O OO

(a)

Figure 20

13, 58 –62

O

OO

R

R = C5H11 R R

symmetry. However, if bonded rim-to-rim with relatively rigid E or M bowls as in 61 or 62, the P bowls possess perfect C4 symmetry and assume a bo-su conformation, with the four bridges outward and the four spanners upward (Figure 20c). Thus, P bowls reorganize substantially upon being incorporated into hemicarcerands. CPK models of hemicarcerands 13, 58–62 provide the order PP > PM > MM > PE > EM > EE in maximum portal size. Because P bowls are flexible, the order of portal adaptability to guest shape for complexation–decomplexation is PP > PM > PE > MM > EM > EE. For these hosts, the inner cavity decreases in the order PP > PE > EE > PM > EM > MM. These hosts show high structural recognition in complexation. However, unlike hemicarcerand 6, which preferentially binds 1,4-disubstituted benzenes,35, 58 they prefer 1,2-disubstituted benzene guests to 1,3- and 1,4-disubstituted isomers. For example, heating empty EE for four days in 3-ClC6 H4 COCH3 gave a 1 : 1 mixture of EE
3-ClC6 H4 COCH3 and empty EE. Under the same conditions, a 2 : 1 mixture of EE
2-ClC6 H4 COCH3 and empty EE was formed in 4-ClC6 H4 COCH3 as solvent. Thus, the relative rates of complexation of the three isomeric guests must be 1,2-isomer  1,3-isomer  1,4 isomer, which explains the host’s ability to scavenge trace amounts of the 1,2-isomer in neat 4-ClC6 H4 COCH3 . Another example is the formation of a 2 : 1 mixture of EE
(CH3 )3 CPh and EE
PhCH(CH3 )CH2 CH3 after heating EE in (CH3 )3 CPh, which contained 2% PhCH(CH3 )CH2 CH3 as impurity, indicating that EE complexes PhCH(CH3 )CH2 CH3 around 25 times faster than its isomer (CH3 )3 CPh. The selectivity of hemicarcerand EE for 1,2-disubstituted benzenes contrasts that of MM, which encapsulates preferentially paradisubstituted benzenes. The cavity of MM is narrower with a longer polar axis than that of EE and therefore better suited for taller guests, whereas the shortest but most spherical of the three isomers prefers the more spherical

R

OO O O O O OO A A A AO O O O

R

15

R

(b)

A A O O O

Br Br A OO

R R

63, A = (CH2)3

R

(c)

(a) Hemicarcerands 13, 58–62. Conformation of P bowl in X-ray structure of (b) 63 and (c) in hemicarcerands 61 and 62.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

16

Molecular recognition

cavity of EE. Such high structural recognition in combination with ease to tailor the dimension of the inner phase and the host’s portals makes hemicarcerands ideal building blocks for hydrocarbon storage, separation, and purification applications.

8

WATER-SOLUBLE HEMICARCERANDS

In the previous examples, constrictive binding primarily controls selectivity in complexation studies. Water-soluble hemicarcerand 64 and 65 made possible a detailed analysis of differences in intrinsic binding among different hemicarceplexes (Figure 21).89, 90 In water, the hydrophobic effect, which is typically stronger than solvophobic effects in common organic solvents, drives complexation of nonpolar guests. Consequently, water-soluble hemicarceplexes are intrinsically more stable and their stabilities can be compared under equilibrium conditions. Stable one-to-one hemicarceplexes of octaacid 64 with 14 guests were prepared in D2 O at pH 9.89 Complexation was complete in a few minutes at room temperature, except for naphthalene, where dissolution of the lipophilic solid in D2 O was the rate-limiting step. Among the common guests 1,4-(CH3 )2 C6 H4 and 1,4-(CH3 O)2 C6 H4 that were studied for 64 and the structurally related nonpolar 13, complexes of 64 are stable at room temperature in D2 O, whereas those of 13 decomplex rapidly at 25 ◦ C in CDCl3 . The four salts (CH3 )4 N+ Br− , Ph(CH3 )3 N+ Br− , PhCH2 (CH3 )3 N+ Br− , and 3-CH3 C6 H4 CO2 − Na+ failed to complex 64 in D2 O buffer even though CPK models of hemicarceplexes can be assembled. Probably, D2 O solvates their charges better than it does the interior of 64. It appears that the enthalpic solvation energies of the ions CH3 CH3

H3C

O HO

HO

HO

HO O

OOO O O O

O O O

O

CH3

CH3

OO O O O

O O O OO

O O O

CH3 CH3

CH3

CH3 CH3

H3C

O OH

O O OH OO O O

64

Figure 21

by water inhibit complexation, though the release of many inner-phase and guest-solvating water molecules would provide an entropic driving force for complexation. Piatnitski et al. carried out a detailed thermodynamic analysis of the binding properties of the water-soluble hemicarcerand 65.90 This host lacks one of the linkers of 64, which facilitates guest exchange. Thus, 65 displays thermodynamic selectivity in its binding properties, which differs from many other hemicarcerands, for which constrictive binding controls selectivity. In water, 65 binds small hydrophobic guests with affinities that reach K = 108 M−1 , which is higher than those measured for other receptors with hydrophobic cavities, such as cyclodextrins and cyclophanes.91, 92 Guest size, hydrophobicity, and charge are important factors in determining binding strength. An enthalpy–entropy compensation plot for binding of small hydrophobic guests provided a slope α = 0.75 (Figure 22a). α varies from 0 to 1 and is a measure of to what extent the enthalpic gain is compensated by an entropic loss. Thus, it reflects the amount of host reorganization upon binding. Flexible enzymes reorganize substantially upon substrate binding and have α = 1. The α value obtained for 67 is smaller than that for β-cyclodextrin (α = 0.9) and comparable to that of cyclophanes (α = 0.78) and indicates that 65 is relatively inflexible. Furthermore, the T
S intercept at
H = 0, T
S0 = 4.2 kcal mol−1 is much higher than that for β-cyclodextrin and cyclophanes and has been taken as a measure for guest desolvation upon binding, which is an important driving force for complexation.93 Among aromatic guests with methyl or methoxy groups, meta and para substitution patterns are preferred over the ortho pattern, which can be rationalized with the ability of both methyl groups to undergo CH–π interactions if they are either para or meta (Figure 22b). Consistent with

O OH

HO

OH

HO

O OO O H

O O O

O

O

CH3

OO O O O

O H OO

CH3

O O O

CH3 CH3

O OH

O O OH OO O O

CH3

65

Water-soluble hemicarcerands 64 and 65.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

OH

OH

Carcerands and hemicarcerands

17

5 O O CH3

OCH3

T∆S/ kcal mol−1

OCH3 OCH3

0

CH3

CH3

OCH3

CH3 H3CO OCH3

−5 H3CO H3CO

−10 −20

−16

OOC

H2C

CH3

− − CH2 COO OOC

CH3

H3CO

OCH3

OCH3 OCH3

−12 −8 ∆H/kcal mol−1

(a)

OCH3 CH3

OCH3

OCH3

−

OH CH2OH

CH3

H2C

−4

CH2

0

COO− −OOC

H2 C

H2 C

CH3

COO−

CH3 −

OOC

(b)

C H2

CH3

− CH2 COO

65 p -xylene

−

OOC

CH3

C H2

− CH2 COO

65 m -xylene

−

OOC

C H2

C H2

COO−

65 o -xylene

Figure 22 (a) Enthalpy–entropy plot for hemicarceplexes 65
guest, and (b) illustration of interactions between xylene isomers and hemicarcerand 65. (Reproduced from Ref. 90.  Wiley-VCH, 2000.)

this model are the measured binding enthalpies (−
H ), which decrease in the order meta > para  ortho for xylenes and dimethoxybenzenes and the complexationinduced chemical shifts of the methyl protons, which are much larger for the meta and para isomers.

9

CHIRAL RECOGNITION PROPERTIES OF ASYMMETRIC HEMICARCERANDS

We have seen earlier that many hemicarcerands are chiral due to the twisting of the host’s cavitands. However, interconversion between the two enantiomeric twistomers is typically fast at ambient conditions, thus preventing separation of twistomers. Introducing one or more chiral bridging units locks the hemicarcerand in one twisted conformation. Examples are asymmetric hemicarcerands (S)4 -66, (SS)4 -67, (S)-68, and (SS )-69, which have one and four bisoxymethylene-1,1 -binaphthyl or threonide bridges (Figure 23a).94–96 In X-ray structures of (SS)4 67
guest (guest = CH3 C(O)N(CH3 )2 and (CH3 )2 SO), the host’s cavitands are twisted along the polar axis by approximately 15◦ , which is slightly less than in the

strongly twisted 14
CH3 C(O)N(CH3 )2 (24◦ ).95 These hosts typically display moderate thermodynamic chiral selectivity as a consequence of different interactions between enantiomers and the surface of the host’s inner cavity (intrinsic binding energy). For example, heating (SS)4 67
guest in Ph2 O containing excess racemic 2-butanol produces the two diastereomeric hemicarceplexes in a 2 : 1 ratio. This ratio amounts to an intrinsic binding energy difference of 300 cal mol−1 , which is comparable to the selectivity observed in chiral recognition studies of chiral cryptophanes and self-assembled molecular capsules.6, 97–99 Interestingly, both diastereomeric complexes have substantially different retention factors in thin-layer chromatography. The high sensitivity of the surface-adsorption properties of (SS)4 -67 likely results from the host adapting its shape to the configuration of the guest. In complexation studies of (SS )-69, with only one threonide bridge, chiral recognition factors were smaller, for example, 1.4 for 2butanol, and did not lead to changes in surface-adsorption properties of diastereomeric complexes. Under the same conditions, the chiral recognition factor for PhS(O)CH3 was only 1.6. Perhaps the most remarkable chiral selectivity is observed for binaphthyl bridged hosts (S)4 -66 and (S)-68. Compared

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

18

Molecular recognition

R

R

R

O O

O

O

R

R

O

O OO

O

A

A

A

A

O O O

O

O

O O

O

O OO

O

O

(S)

H2C

O OO O

O

H2C

A

(CH2)4

O

66, 68: A =

O

O

(H2C)4

O

R

OO O

O

O

O (H2C)4

R

R

H2C (S) O

67, 69: A =

H2C (S) O R

R

R

R

R

R

66, 67

(a)

H3C

O H3C

O

O O

O

O

R = (CH2)2C6H5

(R )-70

O

O

O O

O

S O

CH3

O

CH3

S OO

R

H3C

S O

CH3

O

R 68, 69

O

O

(S )-70

O O H3C

O

O S

O O

O

CH3

OO

(S )-68

O

O H3C O S O O

O

O

O O

(b)

O H3C

O H3C O S O O

O

(S )-68 (R )-70

O

O H3C

O

(S )-68 (S )-70

Figure 23 (a) Chiral hemicarcerands 66–69. (b) Proposed mechanism for the high chiral selectivity in the complexation of p-CH3 C6 H4 S(O)CH3 by host (S)-68.

to the fairly rigid threonide bridge, the bisoxymethylene-1, 1 -binaphthyl is more flexible and easily responds to differences in the degree of complementarity between host and guest by changing its naphthyl-to-naphthyl dihedral angle. When (S)-68 was heated in the presence of excess racemic p-CH3 C6 H4 S(O)CH3 , only (S)-68
(R)-pCH3 C6 H4 S(O)CH3 formed. The chiral selectivity factor must be greater 20 : 1 and the free energy difference
G > 2.4 kcal mol−1 for both diastereomeric complexes.96 The fast decomplexation rate of (S)-68
(R)-p-CH3 C6 H4

S(O)CH3 suggests that chiral recognition involves an equilibration between diastereomeric complexes rather than a kinetic resolution. Under the same conditions, racemic C6 H5 S(O)CH3 gave diastereomeric complexes (S)-68
(R)-C6 H5 S(O)CH3 /(S)-68
(S)-C6 H5 S(O)CH3 in 1.6 : 1 ratio. Yoon and Cram explained the high selectivity in the complexation of p-CH3 C6 H4 S(O)CH3 with inability of (S)-p-CH3 C6 H4 S(O)CH3 to adapt an orientation inside (S)-68 that maximizes host–guest interactions. In CPK models, both guests can be easily pushed through

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands one of the larger openings in the host shell. The resulting complexes have structures in which the guest lies roughly in the equatorial plane of the host. This orientation results in a minimal number of stabilizing host–guest contacts. However, only (R)-p-CH3 C6 H4 S(O)CH3 is able to rotate about 90◦ around an equatorial axis of (S)-68 to approximately align itself with the polar axis of the host such that both methyl groups can form favorable CH–π interactions with the host’s cavitands (Figure 23b). Apart from the cavity asymmetry, the chiral bridging units also create asymmetry in the shape of the host’s portals, which gives rise to the kinetic chiral selectivity in complexation or decomplexation of these hemicarcerands or hemicarceplexes. Kinetic and thermodynamic selectivity may differ substantially as in the case of (S)4 66.94 For example, complexation studies with racemic BrCH2 CH2 CHBrCH3 or BrCH2 CHBrCH2 CH3 gave an equilibrated mixture of diastereomeric complexes in a ratio 2 : 1 for both guests. However, for BrCH2 CH2 CHBrCH3 , the thermodynamically less stable diastereomer dissociated five times faster than the more stable isomer. In the case of BrCH2 CHBrCH2 CH3 , the kinetic stability of the two diastereomeric complexes was reversed and the dissociation rate of the more stable isomer was ninefold larger than that of the less stable complex. Thus, whereas the difference in intrinsic binding in each diastereomeric pair is only
G0 = 0.3 kcal mol−1 , the difference in constrictive binding, which is the

G‡ value for the complexation diastereomeric TSs, is

G‡ = 1.6 kcal mol−1 for BrCH2 CHBrCH2 CH3 and

G‡ = 0.7 kcal mol−1 for BrCH2 CH2 CHBrCH3 . Cram suggested that differences in steric repulsions in the diastereomeric TSs probably give rise to the observed chiral selectivities. Thus, the host is able to discriminate between the steric requirements of a CH3 group versus Br atom or a CH2 CH3 versus a CH2 Br group, whose volumes and surface areas differ by 44c, such that it is frozen at room temperature for 44c. In the ground state, the equatorially located substituents of 44c point toward an equatorial opening in the host shell, through which they can easily protrude in order to satisfy their space demand. Obviously, the guest experiences large steric interactions as it spins and the octyl groups have to pass by the rigid bridging units.

10.2

OPr PrO OPr PrO

Bond rotation of amides and ring-flip of cyclohexanes

Confining a molecule inside a molecular container not only affects its rotational and vibrational degrees of freedom but also conformational changes of the guest, which are easily tractable by variable NMR spectroscopy and therefore ideal in order to study the effect of confinement on a TS. Cram and coworkers studied the effect of incarceration on the cis–trans isomerization of (CH3 )2 NCHO and (CH3 )2 NCOCH3 inside 4.33 For (CH3 )2 NCHO, the C–N rotational barrier decreased in

the order liquid phase > inner phase > vacuum and was approximately 1 kcal mol−1 lower inside 4 than for the free amide in nitrobenzene. For (CH3 )2 NCOCH3 , the order was inner phase > liquid phase > vacuum and the barrier approximately 2 kcal mol−1 higher inside 4 as compared to the free amide. On the basis of examinations of molecular models of 4
(CH3 )2 NCHO and 4
(CH3 )2 NCOCH3 , Cram explained these trends with the different ratio of free and occupied space in the inner phase. In 4
(CH3 )2 NCHO, the guest is loosely held inside the container, whereas it is strongly compressed against the host walls in 4
(CH3 )2 NCOCH3 and even more so in the TS for bond rotation [4
(CH3 )2 NCOCH3 ]‡ . Thus, the rigid container resists being deformed more than the solvent cage resists being moved to accommodate the TS. Depending on the mix of free and occupied space, the inner phase may be more like vacuum, liquid, or even solid. Large effects due to the rigidity of the container were also observed for the ring inversion of 1,4-thioxane and 1,4-dioxane inside carcerand 4 (Figure 26a). Chapman and Sherman measured 1.8 and 1.6 kcal mol−1 higher ring-flip barriers inside 4 as compared to the liquid phase.105 The increase in barrier height was substantially more than inside the self-assembled hydrogen-bonding capsule 752
C6 H12 , in which the ring inversion barrier of cyclohexane-d11 increases by only 0.3 kcal mol−1 (Figure 26b).106 In the latter case, Rebek and coworkers argued that steric interactions in the TS are unlikely the reason for the modest increase, since the TS is more planar than cyclohexane and should fit better into the “jelly doughnut”-shaped capsule. They suggested that the loss of favorable (C–H/D)–π contact stabilizes the ground state relative to the TS inside the container. The origin for the increased barriers inside 4 is not fully clear and may result from similar ground state effects, such as stabilizing host–guest contacts that are lost in the TS, or from steric constraints in the TS. However,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands R

R

R

R

Guests: O

O O

O O

O O

O

O

O

O

O

O O O

O

O

21

O

∆∆Gring flip = 1.6 kcal mol−1

O Guest O O O

O

S

O S

O ∆∆Gring flip = 1.8 kcal mol−1

R

R

R 4 Guest

(a)

H N

H N

N

N

C O

O C

2

+ O N C NH

O C N HN N HN C O

(b)

R

75

N C NH O

752 C6H12

∆∆Gring flip = 0.3 kcal mol−1

Figure 26 (a) Ring-flip dynamics of 1,4-thioxane and 1,4-dioxane inside carcerand 4. (b) Structure and assembly of self-assembly capsule 752
C6 H12 and ring-flip dynamics of encapsulated cyclohexane.

G‡ring flip =
G‡ring flip (encapsulated)−
G‡ring flip (free).

one can conclude that the effects are pronounced inside the carcerand, which is a result of its rigidity. Host rigidity translates into high sensitivity to small structural perturbations of the guest.

11

HEMICARCERANDS AS MOLECULAR REACTION FLASKS

The application of molecular containers as “molecular reaction flasks” has been a very exciting and rewarding venture in host–guest chemistry. In recent years, the exploration of reactions and reactivity inside covalent or self-assembled molecular capsules has produced very spectacular and unexpected discoveries.15, 16 For example, it has been demonstrated that molecular capsules may allow the taming of otherwise fleeting reactive intermediates,107 alter the regiochemistry of reactions,22, 28, 29 give rise to new

forms of rate accelerations in pericyclic reactions,21–23 and in some instances show enzyme-like behavior in ester and acetal hydrolysis reactions.24 Hemicarcerands were the first molecular containers in which chemical reactions involving encapsulated reactants have been investigated. In this section, some of the advances in inner-phase chemistry are reviewed. Innerphase reactions may take place either entirely inside the carcerand, where they are influenced by the shape and size of the inner phase, or at the electrostatic inner surface of the hosts with its unusual high inner-phase polarizability.108, 109 Typically, these reactions involve one or two encapsulated reactants, in which case the host takes over the role of the solvent cage in equivalent condensed phase reactions. Proper solvation is particularly important in reactions involving zwitterionic intermediates or ion pairs. The absence of polar solvent molecules and the hydrophobicity and reduced deformability of the inner

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

22

Molecular recognition

phase will be particularly felt in these types of reactions. A second kind of inner-phase reaction is best described as “through-shell” reactions. They involve both encapsulated and bulk-phase reactants, the latter being transferred through the host shell somewhere along the reaction path. Bond formation or breaking of through-shell reactions may take place inside one of the openings in the host shell. Thus, outcomes often depend on orientation and rotational mobility of the encapsulated reactant as well as the size, shape, and flexibility of the portals relative to those of the bulk-phase reactant.102, 110

11.1

Through-shell reactions

11.1.1 Proton transfer reactions Proton transfer between incarcerated bases and bulk-phase acids is a simple through-shell reaction and provides insight into the effect of incarceration on the guest’s acidity or basicity. Consistent with observations in many enzyme-catalyzed reactions, the hydrophobicity of the inner phase should alter the pKa of the incarcerated guest.111 Cram and coworkers studied proton transfers between a strong bulk-phase acid and incarcerated amines 31
pyridine, 31
(CH3 CH2 )2 NH, and 31
CH3 (CH2 )3 NH2 (Figure 27).33 Despite a large enough opening in the shell of 31, attempts to protonate incarcerated pyridine with CF3 COOD in CDCl3 failed. Cram suggested that the reduced basicity may result from ineffective solvation of the pyridinium ion by the rigid host, the inability to form a contact ion pair in the inner phase, and the larger size of the pyridinium ion compared to pyridine. For 31
(CH3 CH2 )2 NH, instantaneous decomplexation of 31
(CH3 CH2 )2 ND2 + accompanied through-shell proton transfer. The ability to protonate 31
(CH3 CH2 )2 NH with CF3 COOD in CDCl3 results from the location of the R

R

R

nitrogen of (CH3 CH2 )2 NH in the equatorial region close to the portals. After protonation, the counterion pulled the guest out of the inner phase. Addition of excess CF3 COOD to 31
CH3 (CH2 )3 NH2 led to a 2 : 1 mixture of 31
CH3 (CH2 )3 ND3 + and 31
CH3 (CH2 )3 NH2 , which remained constant over time although slow decomplexation took place. Complete protonation of 31
CH3 (CH2 )3 NH2 required 100 equivalents of CF3 COOD. Excess CD3 COOD only H/D-exchanged the amine protons. These results show that the acidity of incarcerated CH3 (CH2 )3 NH3 + is comparable to that of CF3 COOH in CDCl3 . Furthermore, the strong upfield-shifted amine protons of 31
CH3 (CH2 )3 NH2 imply guest alignment along the polar axis of 31. In this orientation, through-shell protonation most likely occurs through the holes in the polar caps of 31.

11.1.2 Electron transfer reactions Electron transfer reactions are well suited to be studied between an incarcerated guest and a bulk-phase reducing or oxidizing agent, since electron transfer processes do not require direct contact between donor and acceptor complexes but may take place over long distance through electron tunneling.112 An oxidation–reduction cycle for different ortho- and para-hydroquinones could be carried out in the interior of 6 (Figure 28).113 Oxidation with Ce(NH4 )2 (NO2 )6 –silica gel–CCl4 or Tl(O2 CCF3 )3 –CCl4 led to the parent incarcerated quinones in essentially quantitative yields. Reduction back to the hydroquinones was possible with SmI2 /CH3 OH. The same reagent reduced nitrobenzene to N-hydroxyl aniline. Surprisingly, aniline, which is the product in the liquid phase, is not formed. The latter result—the high yields and the instability of free o-quinones—suggests that all reduction/oxidation took place inside 6 rather than

R O

O O

O

O

O

O

O

O

O H

O H

CF3 CH3

NH2

N O

O

O

H

O O O

O

O

O O

R

R

R

31 pyridine

H CH3

N

O CF3 H O

CH3

R

R = (CH2)2C6H5

Figure 27 Structure of hemicarceplex 31
pyridine and proposed proton transfer mechanism for 31
CH3 (CH2 )3 NH2 and 31
(CH3 CH2 )2 NH. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands R

O OO H

R

R

O H N

N

O

R

OH

OO O

H

O OO

O

CH3 (H)

N

N

OH

N

N

O CH3 (H)

H

H

Fe2+ H

O

OH

N

N H O

H

O O

OH

OO O

CH3 (H) R

R

R

23

CH3 (H)

R

36 ferrocene

R = (CH2)2C6H5

Figure 28 Through-shell oxidation–reduction cycles of ortho- and para-hydroquinones and nitrobenzene inside 6 and electrochemical oxidation of 36
Fc.

by a dissociation (bulk-phase reaction)—association mechanism. It also demonstrates that electrons are transferred readily through the host shell in and out of the inner phase. A second example in which through-shell electron transfer was examined quantitatively is the electrochemical oxidation of Fc incarcerated inside hemicarcerand 36. Electron transfer was strongly hindered kinetically and thermodynamically compared to free Fc (Figure 28).114 The half-way potential for the oxidation was more positive, due to the hydrophobicity of the inner phase, and the electron transfer rate was reduced 10-fold. The latter may result partially from the higher mass of 36
Fc compared to Fc and also from a reduction of the electronic coupling between the Fc center and the electrode surface which is affected by the ˚ Whether the increase in distance from 3.5 to about 9 A. hemicarcerand’s aromatic structure mediates the electron coupling is not clear.

11.1.3 Nucleophilic substitutions and isotopic exchanges The alkylation studies of Kurdistani et al. provide much insight into the interplay between guest reactivity, orientation, and bulk-phase reagent size.110 Different phenols were alkylated in the inner phase of 6. Two factors determined reactivity: (i) portal size and (ii) preferred guest orientation relative to the equatorially located portals. Alkylation with NaH/CH3 I in THF of 4-HOC6 H4 CH3 (pcresol) or 4-HOC6 H4 OH (p-hydroquinone) was impossible. Under the same conditions, 2-HOC6 H4 CH3 (o-cresol), 3HOC6 H4 CH3 (m-cresol), and 3-HOC6 H4 OH (resorcinol) were quantitatively methylated. 2-HOC6 H4 OH (catechol) gave a mixture of mono- and dimethylated carceplexes. As

discussed in Section 10.1, the preferred inner-phase orientation of 1,4-disubstituted benzene guests suggests that the OH group of 4-HOC6 H4 CH3 is located in a protected polar cap of the host. In ortho- or meta-disubstituted benzenes, one substituent resides inside a shielded polar cap, whereas the second substituent is located near a portal. Therefore, these reactions must occur in the entryways through a linear TS, which is partially “solvated” by the alkoxy units that align the host’s portals (Figure 29). Since this “pseudo solvent cage” has limited flexibility, larger alkylating agents failed to react. Likewise, in D2 O-saturated CDCl3 no H/D exchange of OH groups was possible when the guest was 4HOC6 H4 CH3 , 2-HOC6 H4 OH, or 4-HOC6 H4 OH.110 In the presence of diazobicyclo[5.4.0]undec-7-ene, 4-HOC6 H4 OH exchanged its hydroxyl protons, but not the rotationally more fixed 4-HOC6 H4 CH3 . In THF–NaH at 25 ◦ C followed by D2 O quench, the hydroxyl protons of 2-HOC6 H4 OH, which are more exposed to the equatorial-located portals, exchange, but not the protected hydroxyl protons of 4HOC6 H4 OH and 4-HOC6 H4 CH3 .

Transition state model

R

H d− O− C I d HH

Shielded OH group R

OH

Figure 29 Transition-state model for alkylation of orthodisubstituted phenolates and shielded OH group in paradisubstituted phenols.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

24

Molecular recognition

11.1.4 Nucleophilic additions Inner-phase guest orientation and mobility also control reactivity in through-shell borane and methyllithium additions to benzaldehyde 73, benzocyclobutenone 76, and benzocyclobutadione 77 inside hemicarcerand 6.102 BH3 ·THF reduced all three incarcerated guests to benzyl alcohol, benzocyclobutanol, and 7-hydroxybenzocyclobutanone (78), respectively (Figure 30). Guest reactivity differed from that in the liquid phase and increased in the order 77 ≈ 76 > 73. Furthermore, incarcerated 77 added only 1 equivalent of BH3 ·THF. An aqueous workup was required in order to reduce the second carbonyl group to 7
79. Hydrolysis of 7
79 gave incarcerated cis-benzocyclobutenediol 80. Crystal structures helped to interpret the different reactivity of the guests. In hemicarceplexes 6
73 and 6
76, the guest’s carbonyl groups are located inside a host’s cavitand and reduction requires reorientation of the guest. The additional conformational energy adds to the activation energy and is higher for 73 as compared to 76. In hemicarceplex 6
77, one carbonyl is shielded; the other is perfectly positioned for through-shell reaction inside an entryway. After addition to the exposed C=O, coordination of the boron of 81 to a host’s ether oxygen hinders guest

rotation and prevents exposure of the second C=O until 81 is hydrolyzed. Guest orientation also explains outcomes of CH3 Li additions to incarcerated 73, 76, and 77 (Figure 31). Again, guest reactivity decreased in the order 77  76 > 73. Compound 77 added 1 equivalent of CH3 Li already at −78 ◦ C to yield 82 and Moore rearrangement product 83. No doubleaddition took place. Hemicarceplex 6
76 required room temperature for reaction completion. Under the same conditions, 6
73 reacted sluggishly and incompletely. Very interesting are the formation of host cleavage products 86 and 25 in these reactions. The former results from the cleavage of one of the host’s spanners initiated by nucleophilic attack of lithium alcoholate 87 at the acetal carbon (Figure 32). On the other hand, at 0 ◦ C incarcerated lithium alcoholates 84 and/or 85 cleaved one of the O–(CH2 )4 –O linkers of 6 via β-eliminations. Bulk-phase lithium alcoholates are not basic enough to induce this reaction. The incarcerated counterparts must be several orders of magnitude more reactive. Three factors contribute to these rate accelerations: (i) the absence of aggregation of R–OLi in the inner phase; (ii) the poor ability of 6 to “solvate” R–OLi, which increases its Ground state guest orientation

H

BH3.thf

Reactive guest orientation

O

H O

O

O 6 73

77 O

81

OH 80

OBH2

H2O O

O

OH

O

78 OH O BH 79

(a)

O

6 76

BH3.thf

H2O O

O O

(b)

O

6 77

Figure 30 (a) Borane reductions of 77 inside 6. (b) Ground state and “reactive” orientations of guests 73, 76, and 77 in inner-phase borane reductions and CH3 Li additions. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands

CH3Li

CH3

CH3 OLi

OLi C

O

84

25

O

CH3

O

O 77

O

85 OLi CH3

CH3 OH

H2O

O

82

O 83

H2O

O R

CH3 OLi

HO O

O

87

86 + 86 89

(H2C)4

O HO

CH3 OH + 88

Figure 31

O

O (CH2)4

O

O OO

CH3

R

OO O

O

O O

(H2C)4

O

76

R

R

O

O

(CH2)4 O

OO O

O

H2O

O 89

R

R = (CH2)2C6H5

CH3

R

R

R

86

Through-shell CH3 Li additions to incarcerated 76 and 77. R

R

R

R

R

R

R

R

H2O

O O O O O H H O Li+ R

O

O O

OO O O

Li+ O O O O

O O

O O

OO O

86 + 86 89

O

O R

6 87

Figure 32

Proposed mechanism of formation of 86 and 86
89.

basicity; (iii) lithium coordination to an oxygen lone pair of the cleaved C–O bond positions the alkoxide O in close proximity to the β-H of the bridge and provides charge compensation during the concerted syn elimination. These examples and those discussed in the previous sections show that through-shell and inner-phase chemistry clearly differs from “conventional” chemistry in the bulk phase with respect to reactivity and selectivity. Inner-phase and through-shell reactions show the following characteristic features: 1.

Guest functional groups that reside inside a host’s polar cap are less reactive than those exposed to an equatorial portal, which have the potential for high through-shell reactivity.

2.

3.

4.

The reactivity of bulk-phase reactants is largely influenced by their size and shape relative to that of the host’s equatorial portals. If functional groups are protected in the guest’s most favorable orientation, reactivity depends on the innerphase rotational mobility of the guest. The basicity and nucleophilicity of incarcerated lithium alcoholates exceed those of bulk-phase alcoholates by several orders of magnitude, resulting in efficient innermolecular elimination or nucleophilic transacetalization and formation of hemicarcerands with one extended portal. In these inner-phase reactions, small structural changes of the guest have a sound effect on the reaction mode.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

26

Molecular recognition N N C R R O

∆

N

O S

O S O >130°

+

>180°

O S O +

N

C R R 74 3-Sulfolene −

(a)

N N

C R R

C + N 2 R R

N N+ C R R

Figure 34

N N N

N

90 (b)

N H

71

N

N H

CH3 91

H C

N

Transition state model

Figure 33 (a) Products and mechanism of diazirine thermolysis. (b) Aryldiazirines 71, 90, and 91 and transition-state model for the inner-phase phenyldiazirine fragmentation.

11.2

Intramolecular thermal reactions

Intramolecular reactions are well suited to compare the inner phase with other bulk phases in terms of reaction rate and to highlight special characteristics of the inner phase as a reaction environment. For example, the ring inversions of cyclic alkanes, which were discussed in Section 10.2, are slowed upon incarceration, likely due to steric interactions in the TS or selective stabilization of the ground state. In intramolecular reactions involving bond formation and or breaking, TS stabilization by the hemicarcerand is possible if bond breaking/formation takes place in close proximity of the host’s aryl units. This was demonstrated in an investigation of the thermal fragmentation of aryldiazirines inside hemicarcerands.109, 115 The thermolysis of diazirines is a common method to produce carbenes and its mechanism has been studied in detail (Figure 33).116, 117 Compared to the bulk phase, inner-phase fragmentation of 90 is 15-fold accelerated, that of 71 slightly faster (1.2-fold), and that of 91 2.4-fold slower.115 Furthermore, all inner-phase TSs are stabilized enthalpically by 2–3 kcal mol−1 , which, in the case of 71 and 91, is partially or fully compensated by unfavorable entropic contributions to
G‡ . The unfavorable
T
S ‡ term likely results from loss of vibrational degrees of freedom as the guest expands upon reaching the TS, leading to a tighter hemicarceplex. The favorable enthalpic stabilization is interesting and was explained with the high polarizability of the inner phase.108, 109 The stretched C–N

Thermal extrusion of SO2 from 74
3-sulfolene.

bonds of the TS are more polarizable than those of the ground state. Thus, the TS will be more strongly stabilized through dispersion interactions,118 especially since bond breaking takes place in close proximity to the highly polarizable aryl units of a cavitand (Figure 33b). If the extrusion reaction is reversible, incarceration may strongly increase the thermal stability of the encapsulated reactant. For example, Reinhoudt and coworkers studied the extrusion of SO2 and butadiene from 3-sulfolene incarcerated inside 74 by mass spectrometry (Figures 25 and 34).119 In the gas phase, the extrusion of SO2 and butadiene from free 3-sulfolene readily takes place at 100–130 ◦ C. Substantially higher temperatures were required for carceplex 74
3-sulfolene. SO2 was detected only above 170–180 ◦ C. At lower temperature, only the intact carceplex was observed. Above 180 ◦ C, also empty 74 was detected but not a SO2 carceplex or a butadiene carceplex. Since 74 is stable at such high temperatures, guest escape due to the thermal destruction of 74 can be excluded. Hence, the detected SO2 and butadiene must result from 74
3sulfolene and must escape from the inner phase through one of the larger side portals. Reinhoudt explained the unusually high thermal stability of incarcerated 3-sulfolene with a fast recombination in the inner phase (Figure 34). Below 180 ◦ C, a thermal equilibrium among 3-sulfolene, SO2 , and butadiene is established. Above 180 ◦ C, this equilibrium is pulled toward the extrusion products via their escape from the inner phase. This example shows how confinement changes the rates of bimolecular reactions by providing a very high local concentration of both reactants.29, 120

11.3

Inner-phase stabilization of reactive intermediates

The possibility to photolyze incarcerated guest molecules presents a pathway to generate and protect highly strained and reactive molecules inside carcerands.107 This allows NMR spectroscopic characterization of otherwise fleeting species, which complements matrix isolation spectroscopy, ultrafast spectroscopy, or flow techniques. The concept

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands Host–guest reaction products

RI Precursor

hn

RI

Dimer Bulk phase reactant

P

Figure 35

Reactive intermediate stabilization by incarceration.

of reactive intermediate stabilization by incarceration was introduced by Cram, Tanner, and Thomas with “the taming of cyclobutadiene” and is outlined in Figure 35.17 Photolysis of a suitable stable photochemical precursor yields the reactive intermediate in the inner phase. Once generated, the surrounding host prevents destructive reactions, such as dimerization or trapping, with bulk-phase reactants that are too large to pass through an opening in the host shell. Difficult to prevent are innermolecular reactions with the surrounding host, which may take place with incarcerated carbenes, nitrenes, radicals, and arynes, thus limiting their lifetime. In the following sections, several examples are discussed.

11.3.1 Cyclobutadiene The “taming of cyclobutadiene” inside 31 is the first example of an inner-phase stabilization of a reactive intermediate and nicely demonstrates the power of this approach (Figure 36).17 Cyclobutadiene 92 is the prototypical example to verify the theory of aromaticity.121, 122 It is severely angle-strained in addition to being antiaromatic.

1/2

1/2

CO2 ∆

O

254 nm

O 92

93

>300 nm

∆

O O 94

O O

O

∆

O 95

97

CHO CHO 96

O2

Figure 36

Photochemical generation of 92 reactions inside 31.

27

Cram and coworkers generated 92 inside 31 by irradiating α-pyrone hemicarceplex 31
93. Irradiation above 300 nm converted 31
93 to photopyrone 31
94, which, as a solid, rearranged to 31
95 at 90 ◦ C. At higher temperature, 31
95 reverted quantitatively back to 31
93. Controlled irradiation of 31
93 with unfiltered UV light produced cyclobutadiene nearly quantitatively. Prolonged photolysis gave acetylene. In the absence of oxygen, cyclobutadiene was stable up to 60 ◦ C and could be characterized by 1 H NMR spectroscopy. Its lifetime is controlled by the barrier for passage through the larger opening inside 31. If a solution of 77
92 was heated in a sealed tube at high temperatures, the guest escaped the protective shelter and dimerized. Also, oxygen, which easily passes through the host shell, trapped the guest as malealdehyde 96, presumably via an intermediate dioxetane 97.

11.3.2 Anti-Bredt bridgehead olefins The approach of inner-phase stabilization of reactive intermediates works particularly well for species with highly strained multiple bonds, such as cyclobutadiene, which have a lower tendency to react with the surrounding host. Other examples are anti -Bredt bridgehead olefins,123 which have a trans-cycloalkene and are unstable if their olefinic strain OS ≥ 21 kcal mol−1 .124 For example, bicyclo[2.2.2]oct-1ene 98 and (Z)-bicyclo[3.2.1]oct-1-ene 99 have OS = 46.4 and 21.9 kcal mol−1 , respectively. Both are fleeting in solution, due in part to their high tendency to dimerize or rearrange, but were stabilized recently at room temperature inside hemicarcerand 6.19 For the inner-phase synthesis of 98 and 99, Jones’ carbene route was chosen, in which carbene 100 rearranged to 98 (major) and 99 (minor).125 Innerphase photolysis of diazirine 101 gave a complex product mixture composed of hemicarceplexes 6
98, 6
99, and 6
103 and small amounts of carbene–hemicarcerand insertion products. Mechanistic studies suggest that photochemically excited 101* directly rearranges to 98 and 99 without participation of carbene 100 (Figure 37). Both incarcerated anti -Bredt olefins are stable at room temperature in the absence of oxygen and oxidize to ketoaldehydes 104 and 105 in aerated solution. A thermal retro-Diels–Alder reaction of 98, which in Jones’ seminal pyrolysis studies had served as indirect proof for formation of 98,125 could also be induced inside 6. At 62 ◦ C, 98 slowly rearranged to triene 106, which escaped the inner phase and was identified in the bulk by its characteristic 1 H NMR spectrum.

11.3.3 o-Benzyne ortho-Benzyne, which has a highly strained triple bond, is another important reactive intermediate in some nucleophilic aromatic substitutions and was recently stabilized

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

28

Molecular recognition

HC N 103 2

106 60 °C O2

98

hn

H C N N 101

O CHO

O O

104

101*

O O2

CHO O

99

O

105

Fast O2

105

102

H C 100

Figure 37

Inner-phase photochemistry of incarcerated 101. O + CO

107a

Figure 38

O

107b

77

107

Resonance structures of o-benzyne. 400 nm

in the inner phase of hemicarcerand 6 (Figure 38).18, 126 oBenzyne is also of interest due to its unusual structural and electronic properties.127, 128 Chapman first matrix-isolated o-benzyne by photolyzing benzocyclobutenedione 77 at 8 K (Figure 39).129 The same route led to the successful innerphase synthesis of o-benzyne.18 Irradiation of hemicarceplex 6
77 at 400 nm gave hemicarceplex 6
109, which upon further photolysis at 280 nm decarbonylated to yield 6
107. Because of the high π bond strain of 50 kcal mol−1 ,130 o-benzyne underwent a Diels–Alder reaction with the surrounding 6, which was fast above −75 ◦ C (Figure 39b).126 This reaction is very selective and 107 adds exclusively across the 1,4-position of an aryl unit of 6. The MM3* minimum-energy conformer of 6
107 shows strong preorganization of the reactive triple bond for the observed Diels–Alder reaction ˚ between the reactwith distances of 4.53 and 4.05 A ing carbons of the host and guest.131 This high preorganization is reflected in the moderately negative activation entropy
S ‡ (298 K) = −10.7 cal mol−1 K−1 .126 Interestingly, the measured
H ‡ is slightly higher than the ‡ calculated
Hcalc for the addition of 107 to benzene.131 Thus, the increased reactivity of 6 must be compensated by steric interactions originating from a repulsion between H(1) and aryl unit A (Figure 39b). This suggests that an incarcerated 3,6-disubstituted o-benzyne may not be

C C 108

(a)

280 nm

O

400 nm − CO

O

O 109

Aryl unit A R

R

R

R

OR

O

H1

O O O

O

R O O

R O

O

O

O

O

O HR H 1

O

O

O

O

O (b)

O O

O O

O

6 107

110

Figure 39 (a) Photochemistry of 77 in argon at 8 K and inside hemicarcerand 6 and (b) Diels–Alder addition of incarcerated obenzyne to yield 110.

able to react with the host and may be stable at higher temperatures. At −75 ◦ C, the lifetime of 6
107 was long enough to record a 1 H NMR spectrum. The protons of 107 resonated at δ 4.99 and δ 4.31. Under the assumption that they feel the same shielding by the surrounding host as the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands

in 17 and 30% yield at 77 and 15.5 K, respectively. Yields of 5-methyl-cyclohepta-1,2,4,6-tetraene (MeCHTE) 120 in the related inner phase p-tolylcarbene rearrangement were even higher. Photolysis of 6
91 and perdeuterated 118
91 at 77 K afforded 6
120 and 118
120 in 41 and 67% yield, respectively.140 Likely steric interactions in the TSs for the p-tolylcarbene–host reactions increase the lifetime of incarcerated 3 119 beyond that of 3 111, increasing its probability for photochemical ring expansion. Both incarcerated allenes 113 and 120 persisted for months at room temperature in the absence of oxygen and could be characterized by 1 H NMR spectroscopy. Upon exposure of incarcerated CHTE to oxygen, rapid autoxidation to benzene took place. Under these conditions, oxygen diffused into the hemicarcerand, added to the central allene carbon and produced a spirocyclic dioxirane 121, which subsequently rapidly decarboxylated to benzene. The constrictively stabilized CHTE and MeCHTE allowed the measurement of several barriers of phenyland tolylcarbene rearrangements, which previously were only available from high-level calculations, and comparison between experiment and theory.135, 136, 141 For example, in an attempt to measure the enantiomerization barrier of CHTE, 71 was photolyzed inside chiral hemicarcerand 69 and produced diastereomeric hemicarceplexes 69
(+)CHTE and 69
(–)-CHTE in a 2 : 3 ratio.139 In the asymmetric host environment, guest protons H2 experienced different host-induced shielding, allowing differentiation by 1 H NMR spectroscopy. The absence of coalescence at 100 ◦ C gave a lower limit of 19.6 kcal mol−1 for the enantiomerization barrier, which agrees with all current calculations.135, 136 For the corresponding MeCHTE hemicarceplexes, exchange rate constants could be extracted from line-shape analysis of high-temperature NMR spectra.142 Furthermore, photolysis of 69
91 produced69
(+)MeCHTE and 69
(–)-MeCHTE in the ratio = 1 : 1.15 (de = 7%), which slowly equilibrated into the thermodynamic ratio of 1 : 1.8 and allowed measurement of exchange rate constants. The experimental enantiomerization free energy, which was computed from these rate constants, agreed very well with the computed enantiomerization barrier.

protons of benzene, the chemical shifts of “free” o-benzyne were estimated at δ 7.0 and δ 7.6, which are in excellent agreement with the calculated shifts.128 Much less upfield shifted are the guest 13 C signals, which provide more insight into the electronic properties of o-benzyne. The measured chemical shift for the quaternary carbon of 107 at δ 181.33 is within the experimental error of the average of the three chemical shift tensor principal values δ193 ± 15 of matrix-isolated, 13 C-enriched 107 at 20 K in argon.132 The 13 C NMR spectrum of incarcerated o-benzyne also provided direct 13 C– 13 C coupling constants. Comparison with the 13 C– 13 C coupling constants of model compounds suggested a cumulenic character of o-benzyne (Figure 38), which, however, contradicts most recent results of ab initio calculations.128, 133 These calculations predict that obenzyne is aromatic according to its geometric, energetic, and magnetic properties and that the in-plane π -bond induces a small amount of bond localization resulting in acetylenic character.

11.3.4 Phenylcarbene rearrangement The phenylcarbene (PC) rearrangement was recently studied inside hemicarcerands 6 and 69. In the gas phase, PC 111 ring-expands to cyclohepta-1,2,4,6-tetraene (CHTE) 113 involving bicyclo[4.1.0]hepta-1,3,5-triene 112 as intermediate (Figure 40).134–136 CHTE, which is a bent and twisted allene with 40 kcal mol−1 strain energy,136 is the local minimum on this part of the potential energy surface and enantiomerizes via the planar cyclohepta-1,3,5trienylidene 114 as TS.135, 136 Ring expansion can also be triggered photochemically via excitation of triplet PC 3 111 generated from phenyldiazomethane or phenyldiazirine (Figure 41).137–139 However, photolysis of 6
71 at 77 K produced only insertion products 115 (85% yield) and 116 (4.5%). Both formed via insertion of transient 111 into an inwardpointing acetal C–H and linker α-C–H bonds of 6 respectively. The trick to rearrange transient 3 111 to CHTE was the deuteration of 6. In the partially deuterated 117
71, a kinetic isotope effect of kH /kD = 9.8 slowed the carbene insertions and increased the lifetime of 3 111,100 such that photochemical rearrangement to CHTE was possible

H

H H

C

H

H

111

Figure 40

112

29

114

H

H

113

The Baron mechanism of the phenylcarbene rearrangement.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

30

Molecular recognition

N

N CH

O O

hn

1

ISC

111

3

hn

111

71

113

CH3OH

HCl R

R R

−CO2

O2

121

R H

H

Cl

OCH3

CH3OH OO

O O O O OO

O

H O

O OO

R

OO OO

O

O

R

O

R

O OO O

R

R

O O OO X X O O

115

Figure 41

C

CH3

CH3

91

119

O X X O OO O

H

X O

OO O X O

O

A

O

N CH

R

A

A N

R

R

R

CH3

R

A

O O X X O OO O

R

R

117: X = CD2; A = (CH2)4

120

118: X = CD2; A = (CD2)4

Chemistry of 71 and 91 inside 6 and 69 and structures of 117–118.

This example illustrates nicely how kinetic experiments in confined space allow mapping of potential energy surfaces of important organic chemical processes involving highly reactive intermediates, which is difficult to achieve with other techniques such as laser flash photolysis, collision-induced dissociation (CID), or matrix isolation.

11.3.5 Carbenes Most carbenes R–C–R , in which R and R are H, alkyl, vinyl, or aryl, are too reactive to be observable inside a hemicarcerand.143 For example, incarcerated arylcarbenes rapidly insert into C–H or C–O bonds of the hemicarcerand or add to one of the cavitand’s aryl units even at very low temperature.100, 138, 140 However, stability and reactivity of carbenes can be tailored especially with heteroatom substituents that stabilize the carbene’s singlet state through electron donation (push effect).144 In fact, many diaminocarbenes are stable and isolable at room temperature.145, 146 In cases where intrinsic stabilization (edonation) is not sufficient, extrinsic effects (incarceration) may render an otherwise fleeting singlet carbene stable under normal conditions.

Fluorophenoxycarbene 122 is such a species and was recently room-temperature-stabilized by incarceration (Figure 42).147 In 122, the O- and F-substituents stabilize the singlet state by ∼60 kcal mol−1 compared to singlet methylene.148 This stabilization is, however, not large enough to render free 122 persistent at room temperature. On the contrary, if generated photochemically from N N O C F hn

−N2

O

C

F

H2O/cat H+

O

O C

H

O

H F C F

Slow

123

122

124

125 HF

+

O

H C

H+ F

H2 O

H OH OC F

122

126

Figure 42 Photochemistry of 123 inside hemicarcerand 6 and mechanism of the acid-catalyzed trapping of carbene 122 with water.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands diazirine 123, 122 dimerizes instantaneously, reacts with moisture, or is trapped in the presence of alkenes.149 Liu et al. generated incarcerated 122 by irradiation of fluorophenoxydiazirine hemicarceplex 6
123 at low temperature (Figure 42).147 Incarcerated 122 persisted for weeks at room temperature. The 13 C and 19 F NMR spectra of 6
123 provided interesting insight into the electronic properties of 122. The carbenic carbon resonated at δ 285.7 ppm, which compares well with chemical shifts of other persistent heteroatom-substituted carbenes.146 The strongly downfield-shifted fluorine, the unusually large 19 F– 13 C coupling constant, and the considerable upfield shift of the ipso carbon of 122 relative to that of 123 point toward strong participation of both O and F atoms in the carbene stabilization through push–push effects. In fact, the push effect of the O substituent is mostly responsible for the stability of 123 and its low tendency to react with the surrounding host. This was concluded from the high reactivity of fluorophenyl carbene 127, which lacks the O substituent. Attempts to generate and observe through NMR spectroscopy 127 inside the same hemicarcerand via photolysis of 6
128 failed. Low-temperature UV–vis spectroscopy suggests that incarcerated Ph-C-F rapidly adds to one of the aryl units of 6 below −100 ◦ C.150 C

N N C

F

127

F

128

In the presence of trace amounts of acid, incarcerated 122 slowly reacted with water in the bulk phase to yield phenylformate hemicarceplex 6
124 and phenyl difluoromethyl ether hemicarceplex 6
125 (Figure 42). The requirement of acid catalysis in the inner-phase watertrapping reaction is surprising since catalysis is not required for free 122. This suggests that the water trapping of 122 is initiated by protonation and that water is not acidic enough in the inner phase to protonate 122, contrary to residual

water in an organic solvent. The hydrophobicity of the inner phase and lack of solvation of the hypothetical ion pair [122H]+ [OH]− are likely reasons for the absence of this acid–base reaction similar to the examples discussed in Section 11.1.1.151 This shows that incarceration not only prevents dimerization of 122 but also slows trapping reactions with water by many orders of magnitude.

11.3.6 Phenylnitrene Very recently, phenylnitrene (PN) and its intramolecular rearrangement have been investigated inside hemicarcerand 6 (Figure 43).152, 153 PN is an important reactive intermediate for organic synthesis and photoaffinity labeling of biomacromolecules.154 It is isoelectronic with PC and, above −100 ◦ C, undergoes similar intramolecular rearrangements to the highly strained cyclic ketenimine 130 which can be trapped with amines or other nucleophiles.155, 156 Below −100 ◦ C, 1 PN intersystem-crosses to triplet 3 PN, which is known to dimerize rapidly in solution (Figure 43). Though at first glance PN and PC show many similarities in their chemistry, their reactivity differs dramatically, which has been subject of extensive investigations over the past decades, and reflects itself in the inner-phase chemistry of both species. For example, in solution, 1 PN ring-expands rapidly at room temperature to 130 in the subnanosecond time scale, whereas ring expansion of 1 PC can only be observed at elevated temperatures in the gas phase due to the substantially higher activation energy and the higher intermolecular reactivity of 1 PC.135, 157, 158 Consequently, 1 PC does not rearrange to CHTE if generated inside 6.100, 139 Ring expansion cannot compete with much faster insertions into hemicarcerand bonds. The situation is different for 1 PN, in which case intramolecular pathways (intersystem crossing and ring expansion) are much faster than reactions with the hemicarcerand. Thus, photolysis of incarcerated phenylazide at −86 ◦ C, at which temperature ring expansion is faster than intersystem crossing, produced O

N3 hn

N 1

PN 130

129 N

3PN

H2O

R

131

hn

132

N

R

OO

O O O O

OO O O OO OO HN H O O

O O

O OO O

133

CH insertion

134

Figure 43

RR

NH

OO N

31

R

R

R

R

134

Inner-phase photochemistry of phenylazide 129 inside hemicarcerand 6.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

32

Molecular recognition

6
130, whose 13 C NMR spectrum could be recorded for the first time[1].152 At this temperature, incarcerated 130 slowly decayed within 5 h to triplet 3 PN by ring contraction and intersystem crossing, allowing for a precise determination of the activation parameters of this process. If the photolysis was carried out in THF/water 8 : 1, water trapped 130 in the inner phase to produce lactam 131. The inner-phase photolysis studies of 6
129 uncovered a novel photochemical reaction of 130.153 Under the photolysis conditions, 130 underwent efficient photochemical electrocyclization to the anti -Bredt imine 132, which at −5 ◦ C thermally rearranged via a 1,5-shift to the slightly more stable 1-azaspiro[2.4]hepta-1,4,6-triene 133. The reactivity and lifetime of triplet 3 PN and triplet 3 PC differed remarkably in the inner phase of 6. Whereas 3 PC reacts with 6 already at 15 K and persists probably only a few minutes at this temperature,100, 139 the lifetime of incarcerated 3 PN is 13.6 min at −3 ◦ C.153 Both 3 PC and 3 PN preferentially insert into inward-pointing acetal C–H bonds of 6 to produce 115 and 134, respectively (Figures 41 and 43). The difference in reactivity toward C–H insertion between triplet carbenes and an isoelectronic triplet nitrenes is well known and is believed to proceed through a nitrogen rehybridization in the rate-limiting H-abstraction step of the nitrene. Rehybridization is not needed in the carbene reaction.156 C–H insertion reactions involving free PN are essentially impossible to study in solution by laser flash photolysis since C–H insertion cannot compete with dimerization, which is orders of magnitude faster.160 Thus, confining 3 PN inside the molecular container[2], which eliminates dimerization, provided an elegant way to explore this important type of chemistry and allowed for the first time an accurate measurement of the activation parameters for a C–H insertion reaction involving 3 PN[3].

11.4

Photoelectron and triplet energy transfer

The concept of single-molecule incarceration inside a hemicarcerand, which provides an insulating multi-angstromthick wall around the guest, has also helped to better understand triplet excited state quenching by photoelectron transfer (PET) and energy transfer (ET). Both are important photophysical processes 163 and play a central role in biological photosynthesis,164 visual transduction,164 organic photochemistry,165 semiconductor photocatalysis, and imaging.166–169 The idea behind through-shell PET and ET is to generate a triplet excited state inside a hemicarcerand and to measure rate constants and efficiencies of quenching the excited state by bulk-phase quenchers that are not covalently connected and are prevented from coming closer than

˚ to the incarcerated guest. Equations (3) approximately 7 A and (4) schematically describe energy and electron transfer quenching of an incarcerated triplet state Host
G(T1 ) with a bulk-phase quencher Q170 : Host
G(T1 ) + Q(S0 )
Host
G(T1 ) · · · Q(S0 ) −→ Host
G(S0 ) · · · Q(T1 )
Host
G(S0 ) + Q(T1 ) (3) Host
G(T1 ) + Q
Host
G(T1 ) · · · Q −→ Host
G− · · · Q+
Host
G− + Q+

(4)

ET is a weakly coupled nonadiabatic process and proceeds by a Dexter electron exchange mechanism. Its rate constant kET can be approximated by the Golden Rule171–173 :
kET =

2π h

 × |ν|2 × FCWDS

(5)

FCWDS is the Franck–Condon weighted density of states and ν the electronic coupling matrix element. In a semiclassical treatment, this equation can be separated into a preexponential factor A and an exponential term that relates kET to the driving force
G and nuclear reorganization energies of reactant λv and solvent λs :  kET = A × exp

−(λs +
G + λv )2 4λs kB T

 (6)

The dependence of the rate constant for photoinduced electron transfer, kPET , on the driving force and reorganization energies is similar. Equation 6 predicts a parabolic dependence of log kET on the driving force. At −
G = (λs + λv ), kET is largest and decreases at smaller (normal region) and more exothermic driving force (inverted region). The experimental observation of the Marcusinverted region for electron or energy transfer between noncovalently linked triplet excited state/quencher pairs has been very difficult mainly because ET and PET at high driving force are much faster than the rate of diffusional encounter. In the hemicarceplex/quencher encounter complexes (Host
G(T1 )· · ·Q(S0 ) and Host
G(T1 ) · · ·Q), the ˚ excited state and quencher are separated by about 7 A. Because of their strong distance dependence,172, 174 ET and PET are substantially slower than diffusion, which has made observation of the inverted region possible in these systems.170, 175, 176 In their seminal work on through-space triplet ET, Deshayes and coworkers studied acetophenone hemicarceplex 14
Ac and probed through-shell triplet ET chemically via isomerization of cis-piperylene to transpiperylene (Figure 44).177

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands

hn

108

ET H3 C

33

O PYR

cis -Perylene 107

FLA

ACR

14 Ac

trans -Perylene

Figure 44 Photosensitized isomerization of cis-piperylene catalyzed by 14
Ac.

Triplet ET was 2.7-fold slower for 14
Ac compared to free acetophenone. This corresponds to an almost diffusioncontrolled rate for 14
Ac. Since triplet energy is transferred through an electron exchange mechanism, which necessitates a close contact between donor and acceptor, sufficient overlap of highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) in the donor–acceptor pair must exist. Whether the intervening hemicarcerand plays a role in this through-space ET is uncertain. In a subsequent investigation, Farr´an and Deshayes measured triplet ET rates between incarcerated biacetyl (14
biacetyl) and various bulk-phase acceptors.175 Hemicarcerand 14 retarded triplet ET, which suggests a reduced electron coupling between donor and acceptor as a result of their larger separation. Also, log kET and
G showed a hyperbolic relationship as predicted by the Golden Rule. Interesting is the extremely slow triplet ET rate to O2 . Oxygen is typically a very efficient quencher. Farr´an and Deshayes concluded that the quenching rate drops off drastically if oxygen is prevented from making direct contact with the donor. Parola et al., who independently measured triplet ET rates from 14
biacetyl to quenchers used by Deshayes in addition to several others, agreed that the difference between kET of free and incarcerated biacetyl results from different electronic exchange matrix elements v.170 However, they were careful in taking the observed parabolic-like relationship as firm evidence for inverted behavior especially since their data were strongly scattered. A parabolic-like correlation may simply reflect nonhomogeneity of the quenchers, as a consequence of their different sizes, which leads to different donor–acceptor distances and/or orientations and hence to different values for v. This may also be the reason for the failure to observe the inverted region in electron transfer quenching experiments between 14
biacetyl and bulk-phase aromatic amine donors. In a subsequent investigation, Deshayes and Piotrowiak provided clear support for the parabolic Marcus relationship and explained earlier data in a quantitative manner by

kTT [mol−1 s−1]

ANT PNAP

DPO

BANT

106

NAP

RET

DPH DPB DBA

105

TPP

104 −10

PIP

−5

0

5

10

15

20

25

30

35

∆G0 [kcal−1 mol−1]

Figure 45 Rate constant versus driving force −
G of triplet ET from 14
biacetyl to aryl (•) and alkene (◦) acceptors and theoretical curves generated using the semiclassical Marcus–Jortner formalism of triplet energy transfer. (Reproduced from Ref. 176.  American Chemical Society, 1998.)

taking into account the different internal nuclear reorganization energies λv of the acceptors.176 According to MO calculations, λv varies by more than 20 kcal mol−1 among the different acceptors. Thus, two acceptors with nearly identical driving forces and transfer rates may belong to different regions of the Marcus parabola. Examples are dibromoanthracene (DBA) and diphenylbutadiene (DPB) (Figure 45). Both were assigned to the correct region of a Marcus parabola based on their activation energy of transfer, which is negative for the former (typical for the inverted region) but positive for the latter (normal region behavior). Deshayes and Piotrowiak identified four groups of acceptors: (i) rigid aromatics that display small λv ; (ii) acyclic olefins that twist around the double bond upon triplet excitation and therefore have large λv ; (iii) cyclic olefins with even larger λv ; and (iv) O2 , which has essentially no λv . Each acceptor group has its own log kET versus
G correlation (Figure 45). The remaining scattering in the experimental data may result from differences in size and shape of the acceptors, leading to different effective electronic couplings in the corresponding encounter complexes and possibly also to different encounter frequencies.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

34

Molecular recognition

As pointed out earlier, electronic coupling between donor and acceptor is strongly reduced, leading to a reduced energy and electron transfer rate, if both are separated by an intervening hemicarcerand. Deshayes and Piotrowiak addressed the dependence of electronic coupling between incarcerated biacetyl and the bulk-phase quencher on the hemicarcerand size.178 The electronic coupling between the incarcerated donor and bulk-phase acceptor can be described by a superexchange mechanism and viewed as a sequence of guest–hemicarcerand and hemicarcerand–solute interactions. The total electronic coupling matrix element ν total is therefore the product of matrix elements for the guest–hemicarcerand ν GH and the hemicarcerand–acceptor interaction ν HA : ν total ∝ ν GH × ν HA

(7)

For hemicarceplexes 6
biacetyl, 31
biacetyl, and 14
biacetyl, which vary in size and linker characteristics, kET increased with decreasing host size in the order kET (6) < kET (14)  kET (31). The same trend was observed for the average electronic coupling matrix elements: |ν(6)| = 0.20 cm−1 , |ν(14)| = 0.26 cm−1 , and |ν(31)| = 0.66 cm−1 . Since ν HA should be independent of the hemicarcerand size, the spread in ν total reflects differences in the guest–hemicarcerand electronic coupling ν HA among the hosts. These trends are consistent with predictions, according to which the time-averaged guest–hemicarcerand interaction should be cavity size dependent and should increase with decreasing cavity size. One can also conclude that the o-xylylene linkers in 14 do not cause special enhancement of electronic coupling. These investigations not only unravel the role of the hemicarcerand in the mechanism of through-shell triplet ET, but also improve our understanding of solvent-mediated electron transfer,179–185 in which a solvent molecule, separating donor and acceptor, provides the pathway for electronic coupling. Since the thickness of a hemicarcerand is comparable to that of common organic solvents, the measured electronic coupling matrix elements are good estimates for the magnitude of solvent-mediated contributions to electronic coupling in triplet excitation transfer.

12

CONCLUSIONS AND OUTLOOK

The conceptual idea and realization of molecular container compounds has opened a new and intellectually challenging research field: the chemistry of and within molecular container compounds and their complexes. The molecular architecture of hemicarcerands, which features a relatively rigid frame with smaller openings, through which guests

have to pass in order to enter or leave the inner phase, leads to unique molecular recognition properties. Binding selectivity depends primarily on the size and shape complementarity between the guest’s cross section and the dimension of the host’s gates and inner phase. Such high structural recognition, in combination with the ease to tailor the dimension of the inner phase and the host’s portals, makes hemicarcerands ideal building blocks for hydrocarbon storage, separation, and purification applications or as recognition sites in molecular sensors.186 Another future application of hemicarcerands that relies on their ability to fully embrace a guest molecule and to control guest egress is drug delivery.187, 188 The hydrophobic inner phase makes water-soluble hemicarcerands ideal for solubilizing and delivering highly water-insoluble drug molecules. The recent development of gated and dynamic hemicarcerands, which spontaneously release guests in response to photoirradiation or a change in the environment, shows great promise for such delivery applications. However, the functionality of these systems has to be demonstrated first in aqueous medium under physiological conditions. Moreover, hemicarcerands have become interesting new tools for physical organic chemists to study reaction mechanisms and long-distance phenomena. Molecular containers made possible the investigation of highly strained and reactive molecules under normal working conditions by generating them in the protective inner phase. They also allowed the investigation of electronic interactions between encapsulated and bulk-phase molecules through the intervening hemicarcerand and have provided experimental support for theoretical models of long-distance spin–orbit coupling,189 as well as electron and energy transfer. It is anticipated that this field of research will further grow and the recent development of multicavitand nanocapsules will make possible the investigation of chemical reactivity of macromolecular guests that are of interest to material and biological sciences.

NOTES [1] An elegant alternative way to protect the strained cyclic keteneimine and arylnitrene is to incorporate the aryl unit of the nitrene into the host structure, such that the reactive nitrene group points into the inner cavity.159 [2] Recently, phenylnitrene has also been generated inside a deep cavitand.161 [3] An alternative method to suppress dimerization, is to generate the arylnitrene inside a polymeric matrix.162

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands

35

ACKNOWLEDGMENTS

24. M. D. Pluth, R. G. Bergman, and K. N. Raymond, Science, 2007, 316, 85.

The author thanks the National Science Foundation (Grants CHE-0518351 & CHE-0957611) for financial support of his research.

25. C. J. Hastings, M. D. Pluth, R. G. Bergman, and K. N. Raymond, J. Am. Chem. Soc., 2010, 132, 6938.

REFERENCES 1. D. J. Cram and J. M. Cram, Container Molecules and Their Guests, The Royal Siciety of Chemistry, Cambridge, 1994. 2. D. J. Cram, Nature, 1992, 356, 29. 3. A. Jasat and J. C. Sherman, Chem. Rev. (Washington, DC), 1999, 99, 931. 4. R. Warmuth and J. Yoon, Acc. Chem. Res., 2001, 34, 95. 5. E. Maverick and D. J. Cram, Compr. Supramol. Chem., 1996, 2, 367. 6. For cryptophane molecular containers, see T. Brotin and J.-P. Dutasta, Chem. Rev. (Washington, DC), 2009, 109, 88. 7. M. Fujita, M. Tominaga, A. Hori, and B. Therrien, Acc. Chem. Res., 2005, 38, 369. 8. J. L. Atwood, L. J. Barbour, and A. Jerga, Perspect. Supramol. Chem., 2003, 7, 153. 9. J. Rebek, Acc. Chem. Res., 2009, 42, 1660.

26. M. Yoshizawa, S. Miyagi, M. Kawano, et al., J. Am. Chem. Soc., 2004, 126, 9172. 27. L. S. Kaanumalle, C. L. D. Gibb, B. C. Gibb, and V. Ramamurthy, J. Am. Chem. Soc., 2005, 127, 3674. 28. M. Yoshizawa, Y. Takeyama, T. Kusukawa, and M. Fujita, Angew. Chem. Int. Ed., 2002, 41, 1347. 29. J. Chen and J. Rebek Jr., Org. Lett., 2002, 4, 327. 30. D. J. Cram, S. Karbach, Y. H. Kim, et al., J. Am. Chem. Soc., 1985, 107, 2575. 31. D. J. Cram, S. Karbach, Y. H. Kim, et al., J. Am. Chem. Soc., 1988, 110, 2554. 32. J. A. Bryant, M. T. Blanda, M. Vincenti, and D. J. Cram, J. Am. Chem. Soc., 1991, 113, 2167. 33. J. C. Sherman, C. B. Knobler, and D. J. Cram, J. Am. Chem. Soc., 1991, 113, 2194. 34. J. Jung, H. Ihm, and K. Paek, Bull. Korean Chem. Soc., 1996, 17, 553. 35. T. A. Robbins, C. B. Knobler, D. R. Bellew, and D. J. Cram, J. Am. Chem. Soc., 1994, 116, 111. 36. Y.-S. Byun, T. A. Robbins, C. B. Knobler, and D. J. Cram, Chem. Commun. (Cambridge, UK), 1995, 1947.

10. F. Hof, S. L. Craig, C. Nuckolls, and J. Rebek Jr., Angew. Chem. Int. Ed., 2002, 41, 1488.

37. D. J. Cram, R. Jaeger, and K. Deshayes, J. Am. Chem. Soc., 1993, 115, 10111.

11. S. J. Dalgarno, N. P. Power, and J. L. Atwood, Coord. Chem. Rev., 2008, 252, 825.

38. Y.-S. Byun, O. Vadhat, M. T. Blanda, et al., Chem. Commun. (Cambridge, UK), 1995, 1825.

12. B. C. Gibb, Org. Nanostruct., 2008, 291.

39. C. N. Eid, C. B. Knobler, D. A. Gronbeck, and D. J. Cram Jr., J. Am. Chem. Soc., 1994, 116, 8506.

13. J. I. van der Vlugt, T. S. Koblenz, J. Wassenaarm, and J. N. H. Reek, in Molecular Encapsulation, eds. U. H. Brinker and J.-L. Mieusset, John Wiley & Sons, Ltd, Chichester, 2010, p. 145. 14. R. Warmuth, in Molecular Encapsulation, eds. U. H. Brinker and J.-L. Mieusset, John Wiley & Sons, Ltd, Chichester, 2010, 227. 15. M. D. Pluth, R. G. Bergman, and K. N. Raymond, Acc. Chem. Res., 2009, 42, 1650. 16. T. S. Koblenz, J. Wassenaar, and J. N. H. Reek, Chem. Soc. Rev., 2008, 37, 247. 17. D. J. Cram, M. E. Tanner, and R. Thomas, Angew. Chem. Int. Ed., 1991, 30, 1024.

40. C. von dem Bussche-Heunnefeld, D. Buehring, C. B. Knobler, and D. J. Cram, Chem. Commun. (Cambridge, UK), 1995, 1085. 41. R. C. Helgeson, K. Paek, C. B. Knobler, et al., J. Am. Chem. Soc., 1996, 118, 5590. 42. D. J. Cram, M. T. Blanda, K. Paek, and C. B. Knobler, J. Am. Chem. Soc., 1992, 114, 7765. 43. M. E. Tanner, C. B. Knobler, and D. J. Cram, J. Am. Chem. Soc., 1990, 112, 1659. 44. D. A. Makeiff and J. C. Sherman, Chem.—Eur. J., 2003, 9, 3253.

18. R. Warmuth, Angew. Chem. Int. Ed. Engl., 1997, 36, 1347.

45. X. J. Liu and R. Warmuth, J. Am. Chem. Soc., 2006, 128, 14120.

19. P. Roach and R. Warmuth, Angew. Chem. Int. Ed., 2003, 42, 3039.

46. E. S. Barrett, J. L. Irwin, A. J. Edwards, and M. S. Sherburn, J. Am. Chem. Soc., 2004, 126, 16747.

20. M. Ziegler, J. Brumaghim, and K. N. Raymond, Angew. Chem. Int. Ed., 2000, 39, 4119.

47. D. A. Makeiff and J. C. Sherman, J. Am. Chem. Soc., 2005, 127, 12363.

21. J. Kang, J. Santamaria, G. Hilmersson, and J. Rebek Jr., J. Am. Chem. Soc., 1998, 120, 7389.

48. X. J. Liu, Y. Liu, G. Li, and R. Warmuth, Angew. Chem. Int. Ed., 2006, 45, 901.

22. M. Yoshizawa, M. Tamura, and M. Fujita, Science, 2006, 312, 251.

49. Y. Liu, X. Liu, and R. Warmuth, Chem.—Eur. J., 2007, 13, 8953.

23. D. Fiedler, H. van Halbeek, R. G. Bergman, and K. N. Raymond, J. Am. Chem. Soc., 2006, 128, 10240.

50. S. Mecozzi and J. Rebek Jr., Chem.—Eur. J., 1998, 4, 1016.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

36

Molecular recognition

51. K. Nakamura and K. N. Houk, J. Am. Chem. Soc., 1995, 117, 1853.

80. M. I. Page and W. P. Jencks, Proc. Natl. Acad. Sci. U.S.A., 1971, 68, 1678.

52. J. Sherman, Chem. Commun. (Cambridge, UK), 2003, 1617.

81. X. J. Liu, Y. Liu, and R. Warmuth, Supramol. Chem., 2008, 20, 41.

53. R. G. Chapman, N. Chopra, E. D. Cochien, and J. C. Sherman, J. Am. Chem. Soc., 1994, 116, 369. 54. J. R. Fraser, B. Borecka, J. Trotter, and J. C. Sherman, J. Org. Chem., 1995, 60, 1207. 55. R. G. Chapman and J. C. Sherman, J. Org. Chem., 1998, 63, 4103. 56. R. G. Chapman, G. Olovsson, J. Trotter, and J. C. Sherman, J. Am. Chem. Soc., 1998, 120, 6252. 57. K. Nakamura, C. Sheu, A. E. Keating, et al., J. Am. Chem. Soc., 1997, 119, 4321. 58. D. A. Makeiff, D. J. Pope, and J. C. Sherman, J. Am. Chem. Soc., 2000, 122, 1337. 59. M. L. C. Quan and D. J. Cram, J. Am. Chem. Soc., 1991, 113, 2754. 60. S. Ro, S. J. Rowan, A. R. Pease, et al., Org. Lett., 2000, 2, 2411. 61. D. J. Cram, Angew. Chem., 1988, 100, 1041. 62. D. J. Cram, M. E. Tanner, and C. B. Knobler, J. Am. Chem. Soc., 1991, 113, 7717. 63. F. M. Raymo, K. N. Houk, and J. F. Stoddart, J. Am. Chem. Soc., 1998, 120, 9318. 64. J. Yoon and D. J. Cram, Chem. Commun. (Cambridge, UK), 1997, 1505. 65. R. G. Chapman and J. C. Sherman, J. Am. Chem. Soc., 1999, 121, 1962. 66. K. Paek, H. Ihm, S. Yun, et al., J. Org. Chem., 2001, 66, 5736. 67. K. N. Houk, K. Nakamura, C. Sheu, and A. E. Keating, Science, 1996, 273, 627. 68. D. J. Lacks, J. Chem. Phys., 1995, 103, 5085. 69. Y. Liu and R. Warmuth, Org. Lett., 2007, 9, 2883. 70. T. Felder and C. A. Schalley, Angew. Chem. Int. Ed., 2003, 42, 2258. 71. C. Sheu and K. N. Houk, J. Am. Chem. Soc., 1996, 118, 8056. 72. J. Yoon, C. B. Knobler, E. F. Maverick, and D. J. Cram, Chem. Commun. (Cambridge, UK), 1997, 1303. 73. E. L. Piatnitski and K. D. Deshayes, Angew. Chem. Int. Ed., 1998, 37, 970. 74. R. C. Helgeson, A. E. Hayden, and K. N. Houk, J. Org. Chem., 2010, 75, 570. 75. T. Gottschalk, B. Jaun, and F. Diederich, Angew. Chem. Int. Ed., 2007, 46, 260. 76. M. Mastalerz, Angew. Chem. Int. Ed., 2010, 49, 5042. 77. S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, et al., Angew. Chem. Int. Ed., 2002, 41, 899.

82. N. Nishimura and K. Kobayashi, Angew. Chem. Int. Ed., 2008, 47, 6255. 83. N. Nishimura, K. Yoza, and K. Kobayashi, J. Am. Chem. Soc., 2010, 132, 777. 84. C. Naumann, S. Place, and J. C. Sherman, J. Am. Chem. Soc., 2002, 124, 16. 85. J. Sun, B. O. Patrick, and J. C. Sherman, Tetrahedron, 2009, 65, 7296. 86. Y. Liu and R. Warmuth, Angew. Chem. Int. Ed., 2005, 44, 7107. 87. N. Guex and M. C. Peitsch, Electrophoresis, 1997, 18, 2714. 88. R. C. Helgeson, C. B. Knobler, and D. J. Cram, J. Am. Chem. Soc., 1997, 119, 3229. 89. J. Yoon and D. J. Cram, Chem. Commun. (Cambridge, UK), 1997, 497. 90. E. L. Piatnitski, R. A. Flowers, Chem.—Eur. J., 2000, 6, 999.

and

K. Deshayes

II,

91. M. V. Rekharsky and Y. Inoue, Chem. Rev. (Washington, DC), 1998, 98, 1875. 92. S. B. Ferguson, E. M. Seward, F. Diederich, et al., J. Org. Chem., 1988, 53, 5593. 93. W. L. Jorgensen, T. B. Nguyen, E. M. Sanford, et al., J. Am. Chem. Soc., 1992, 114, 4003. 94. J. K. Judice and D. J. Cram, J. Am. Chem. Soc., 1991, 113, 2790. 95. B. S. Park, C. B. Knobler, C. N. Eid Jr., et al., Chem. Commun. (Cambridge, UK), 1998, 55. 96. J. Yoon and D. J. Cram, J. Am. Chem. Soc., 1997, 119, 11796. 97. A. Scarso, A. Shivanyuk, O. Hayashida, and J. Rebek Jr., J. Am. Chem. Soc., 2003, 125, 6239. 98. C. Nuckolls, F. Hof, T. Martin, and J. Rebek Jr., J. Am. Chem. Soc., 1999, 121, 10281. 99. D. Fiedler, D. H. Leung, R. G. Bergman, and Raymond, J. Am. Chem. Soc., 2004, 126, 3674.

K. N.

100. C. Kemmis and R. Warmuth, J. Supramol. Chem., 2003, 1, 253. 101. D. Place, J. Brown, and K. Deshayes, Tetrahedron Lett., 1998, 39, 5915. 102. R. Warmuth, E. F. Maverick, C. B. Knobler, and D. J. Cram, J. Org. Chem., 2003, 68, 2077. 103. P. Timmerman, W. Verboom, F. C. J. M. van Veggel, et al., Angew. Chem. Int. Ed. Engl., 1994, 33, 2345. 104. N. Chopra, R. G. Chapman, Y.-F. Chuang, et al., J. Chem. Soc., Faraday Trans., 1995, 91, 4127.

78. J.-M. Lehn, Chem. Soc. Rev., 2007, 36, 151.

105. R. G. Chapman and J. C. Sherman, J. Org. Chem., 2000, 65, 513.

79. P. T. Corbett, J. Leclaire, L. Vial, et al., Chem. Rev. (Washington, DC), 2006, 106, 3652.

106. B. M. O’Leary, R. M. Grotzfeld, and J. Rebek Jr., J. Am. Chem. Soc., 1997, 119, 11701.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Carcerands and hemicarcerands 107. R. Warmuth, Eur. J. Org. Chem., 2001, 423. 108. C. Marquez and W. M. Nau, Angew. Chem. Int. Ed., 2001, 40, 4387. 109. S. S. Carrera, J.-L. Kerdelhue, K. J. Langenwalter, et al., Eur. J. Org. Chem., 2005, 2239. 110. S. K. Kurdistani, R. C. Helgeson, and D. J. Cram, J. Am. Chem. Soc., 1995, 117, 1659. 111. A. Warshel, P. K. Sharma, M. Kato, et al., Chem. Rev. (Washington, DC), 2006, 106, 3210. 112. H. B. Gray and J. R. Winkler, Proc. Natl. Acad. Sci. U.S.A., 2005, 102, 3534. 113. T. A. Robbins and D. J. Cram, J. Am. Chem. Soc., 1993, 115, 12199. 114. S. Mendoza, P. D. Davidov, and A. E. Kaifer, Chem.— Eur. J., 1998, 4, 864. 115. R. Warmuth, J.-L. Kerdelhue, S. S. Carrera, et al., Angew. Chem. Int. Ed., 2002, 41, 96. 116. M. T. H. Liu, Chem. Soc. Rev., 1982, 11, 127. 117. M. T. H. Liu, Y. K. Choe, M. Kimura, et al., J. Org. Chem., 2003, 68, 7471. 118. S. M. Ngola and D. A. Dougherty, J. Org. Chem., 1996, 61, 4355. 119. A. M. A. van Wageningen, P. Timmerman, J. P. M. van Duynhoven, et al., Chem.—Eur. J., 1997, 3, 639. 120. J. Kang, G. Hilmersson, J. Santamaria, and J. Rebek Jr., J. Am. Chem. Soc., 1998, 120, 3650. 121. A. A. Deniz, K. S. Peters, and G. J. Snyder, Science, 1999, 286, 1119. 122. G. Maier, Angew. Chem. Int. Ed., 1988, 27, 309. 123. H. Hopf, Classics in Hydrocarbon Chemistry, VCH, Weinheim, 2000. 124. W. F. Maier and P. V. R. Schleyer, J. Am. Chem. Soc., 1981, 103, 1891. 125. A. D. Wolf and M. Jones Jr., J. Am. Chem. Soc., 1973, 95, 8209. 126. R. Warmuth, Chem. Commun. (Cambridge, UK), 1998, 59. 127. H. H. Wenk, M. Winkler, and W. Sander, Angew. Chem. Int. Ed. Engl., 2003, 42, 502. 128. H. J. Jiao, P. v. R. Schleyer, B. R. Beno, et al., Angew. Chem. Int. Ed. Engl., 1997, 36, 2761. 129. O. L. Chapman, K. Mattes, C. L. McIntosh, et al., J. Am. Chem. Soc., 1973, 95, 6134. 130. R. P. Johnson and K. J. Daoust, J. Am. Chem. Soc., 1995, 117, 362. 131. B. R. Beno, C. Sheu, K. N. Houk, et al., Chem. Commun. (Cambridge, UK), 1998, 301. 132. A. M. Orendt, J. C. Facelli, J. G. Radziszewski, et al., J. Am. Chem. Soc., 1996, 118, 846. 133. S. G. Kukolich, M. C. McCarthy, and P. Thaddeus, J. Phys. Chem. A, 2004, 108, 2645. 134. P. P. Gaspar, J. P. Hsu, S. Chari, and M. Jones Jr., Tetrahedron, 1985, 41, 1479. 135. S. Matzinger, T. Bally, E. V. Patterson, and R. J. McMahon, J. Am. Chem. Soc., 1996, 118, 1535.

37

136. P. R. Schreiner, W. L. Karney, P. v. R. Schleyer, et al., J. Org. Chem., 1996, 61, 7030. 137. R. J. McMahon, C. J. Abelt, O. L. Chapman, et al., J. Am. Chem. Soc., 1987, 109, 2456. 138. R. Warmuth and M. A. Marvel, Angew. Chem. Int. Ed., 2000, 39, 1117. 139. R. Warmuth and M. A. Marvel, Chem.—Eur. J., 2001, 7, 1209. 140. J.-L. Kerdelhue, K. J. Langenwalter, and R. Warmuth, J. Am. Chem. Soc., 2003, 125, 973. 141. C. M. Geise and C. M. Hadad, J. Org. Chem., 2002, 67, 2532. 142. R. Warmuth, J. Am. Chem. Soc., 2001, 123, 6955. 143. M. Jones Jr. and R. A. Moss, in Reactive Intermediate Chemistry, eds. R. A. Moss, M. S. Platz, and M. J. Jones, Wiley-Interscience, Hoboken, NJ, 2004, p. 273. 144. R. A. Moss, Acc. Chem. Res., 1989, 22, 15. 145. R. A. Moss, in Carbene Chemistry: From Fleeting Intermediates to Powerful Reagents, ed. G. Bertrand, Fontis MediaMarcel Dekker, Lausanne, 2002, p. 57. 146. D. Bourissou, O. Guerret, F. P. Gabbai, and G. Bertrand, Chem. Rev. (Washington, DC), 2000, 100, 39. 147. X. Liu, G. Chu, R. A. Moss, et al., Angew. Chem. Int. Ed., 2005, 44, 1994. 148. N. G. Rondan, K. N. Houk, and R. A. Moss, J. Am. Chem. Soc., 1980, 102, 1770. 149. R. A. Moss, G. Kmiecik-Lawrynowicz, and K. KroghJespersen, J. Org. Chem., 1986, 51, 2168. 150. Z. Lu, R. A. Moss, R. R. Sauers, and R. Warmuth, Org. Lett., 2009, 11, 3866. 151. D. A. Makeiff, K. Vishnumurthy, and J. Am. Chem. Soc., 2003, 125, 9558.

J. C. Sherman,

152. R. Warmuth and S. Makowiec, J. Am. Chem. Soc., 2005, 127, 1084. 153. R. Warmuth and S. Makowiec, J. Am. Chem. Soc., 2007, 129, 1233. 154. H. Bayley, Photogenerated Reagents in Biochemistry and MolecularBiology, Elsevier, Amsterdam, 1983. 155. N. P. Gritsan and M. S. Platz, Chem. Rev. (Washington, DC), 2006, 106, 3844. 156. W. T. Borden, N. P. Gritsan, C. M. Hadad, et al., Acc. Chem. Res., 2000, 33, 765. 157. N. P. Gritsan, Z. Zhu, C. M. Hadad, and M. S. Platz, J. Am. Chem. Soc., 1999, 121, 1202. 158. M. S. Platz, Acc. Chem. Res., 1995, 28, 487. 159. G. Bucher, C. Toenshoff, and A. Nicolaides, J. Am. Chem. Soc., 2005, 127, 6883. 160. T. Y. Liang and G. B. Schuster, J. Am. Chem. Soc., 1987, 109, 7803. 161. G. Wagner, V. B. Arion, L. Brecker, et al., Org. Lett., 2009, 11, 3056. 162. A. Reiser and L. Leyshon, J. Am. Chem. Soc., 1970, 92, 7487.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

38

Molecular recognition

163. P. Piotrowiak, K. Deshayes, Z. S. Romanova, et al., Pure Appl. Chem., 2003, 75, 1061.

177. A. Farran, K. Deshayes, C. Matthews, and I. Balanescu, J. Am. Chem. Soc., 1995, 117, 9614.

164. E. Kohen, R. Santus, and J. G. Hirschberg, Photobiology, Academic Press, San Diego, CA, 1995.

178. Z. S. Romanova, K. Deshayes, and P. Piotrowiak, J. Am. Chem. Soc., 2001, 123, 11029.

165. G. J. Kavarnos, Fundamentals of Photoinduced Electron Transfer, VCH Publisher, New York, 1993.

179. N. E. Miller, M. C. Wander, and R. J. Cave, J. Phys. Chem. A, 1999, 103, 1084.

166. M. Graetzel, J. Photochem. Photobiol., A, 2004, 164, 3.

180. E. W. Castner Jr., D. Kennedy, and R. J. Cave, J. Phys. Chem. A, 2000, 104, 2869.

167. A. Hagfeldt and M. Graetzel, Chem. Rev. (Washington, DC), 1995, 95, 49. 168. B. O’Regan and M. Graetzel, Nature, 1991, 353, 737. 169. D. Eaton, in Photoinduced Electron Transfer I, ed. J. Mattay, Springer, Berlin, Heidelberg, 1990, vol. 156, p. 199. 170. A. J. Parola, F. Pina, E. Ferreira, et al., J. Am. Chem. Soc., 1996, 118, 11610. 171. G. L. Closs, M. D. Johnson, J. R. Miller, and P. Piotrowiak, J. Am. Chem. Soc., 1989, 111, 3751. 172. G. L. Closs, P. Piotrowiak, J. M. MacInnis, and G. R. Fleming, J. Am. Chem. Soc., 1988, 110, 2652.

181. R. J. Cave, M. D. Newton, K. Kumar, and M. B. Zimmt, J. Phys. Chem., 1995, 99, 17501. 182. K. Kumar, Z. Lin, D. H. Waldeck, and M. B. Zimmt, J. Am. Chem. Soc., 1996, 118, 243. 183. R. W. Kaplan, A. M. Napper, D. H. Waldeck, and M. B. Zimmt, J. Am. Chem. Soc., 2000, 122, 12039. 184. H. Han and M. B. Zimmt, J. Am. Chem. Soc., 1998, 120, 8001. 185. A. M. Napper, I. Read, R. Kaplan, et al., J. Phys. Chem. A, 2002, 106, 5288.

173. J. Jortner, J. Chem. Phys., 1976, 64, 4860.

186. B.-H. Huisman, D. M. Rudkevich, A. Farran, et al., Eur. J. Org. Chem., 2000, 269.

174. N. Koga, K. Sameshima, and K. Morokuma, J. Phys. Chem., 1993, 97, 13117.

187. C. L. D. Gibb and B. C. Gibb, J. Am. Chem. Soc., 2004, 126, 11408.

175. A. Farran and K. D. Deshayes, J. Phys. Chem., 1996, 100, 3305.

188. T. V. Nguyen, H. Yoshida, and M. S. Sherburn, Chem. Commun. (Cambridge, UK), 2010, 46, 5921.

176. I. Place, A. Farran, K. Deshayes, and P. Piotrowiak, J. Am. Chem. Soc., 1998, 120, 12626.

189. Z. S. Romanova, K. Deshayes, and P. Piotrowiak, J. Am. Chem. Soc., 2001, 123, 2444.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc054

Cyclodextrins: From Nature to Nanotechnology Stephen F. Lincoln and Duc-Truc Pham University of Adelaide, Adelaide, South Australia, Australia

1 2 3 4 5 6 7

Introduction Cyclodextrin Modification Cyclodextrin Enantioselectivity Cyclodextrin Catalysts Metallocyclodextrins Cyclodextrin Rotaxanes and Catenanes Cyclodextrin Molecular Devices and Nanomachines 8 Conclusion References

1 1.1

1 3 5 8 11 15 18 25 26

INTRODUCTION Cyclodextrin characteristics

Cyclodextrins (CDs) are naturally occurring homochiral macrocycles composed of α-1,4-linked D-glucopyranose in the 4 C1 conformation, and as a class constitute one of the most widely used molecular components in supramolecular chemistry. The CDs were first reported by Villiers in 1891 and were produced from starch by the action of CD transferases produced by bacteria exemplified by Bacillus macerans and Bacillus circulans. Genetically engineered CD transferases of higher activity are now widely used in industrial CD production which amounts to thousands of tonnes annually. The smallest of these “native” CDs are α-, β-, and γ -CD, which consist of 6, 7, and 8 α-1,4linked D-glucopyranose units, respectively, and are the most

available and studied CDs (Figure 1).1, 2 Larger CDs have been characterized but are less studied.3–5 The D-glucopyranose units are labeled alphabetically from A in a clockwise direction when viewed from the narrower end of the truncated CD annulus delineated by a circle of primary hydroxyl groups on the C6 carbons. The wider end of the annulus is delineated by a circle of secondary hydroxyl groups on the C2 and C3 carbons. Together, these hydroxyl groups account for the solubilities of α-, β-, and γ -CDs in water, which are 145, 18.5, and 232 g dm−3 at 298.2 K, respectively.1 The annular interiors are hydrophobic in nature and tend to complex the hydrophobic parts of guest species to form water-soluble host–guest or inclusion complexes. Such complexation may show substantial size discrimination and enantioselectivity between guests and these characteristics have led to a vast range of studies of both native and modified CD systems. In this brief review, the intent is to build a basic understanding of native and modified CDs and their host–guest complexation chemistry and through this to reach the research frontiers where fascinating developments are occurring.

1.2

Cyclodextrin host–guest complexation

A very important property of CDs is their ability to partially or fully complex a wide range of guest species, X, within their annuli to form host–guest complexes in water as shown in Figure 2. There are a variety of interactions driving the complexation process, which include the relatively weak interactions between the hydrophobic guest and water and similarly weak interactions between water and the hydrophobic CD annular interior. By comparison, the interactions between water and the hydrophilic regions

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

2

Molecular recognition

7.8 Å

9.5 Å

7.9 – 8.0 Å

5.7 Å

A ring HO 3 2 OH OH 4 O 1 O O OH B ring 5 O HO 6 HO HO O OH n O OH HO O HO O OH OH HO O C ring HO O O E ring O HO OH HO D ring (a) F-H ring

(b)

(c)

a-CD n = 1 (C36H60O30) b -CD n = 2 (C42H70O35) g -CD n = 3 (C48H80O40)

(d)

a-CD

b-CD

g-CD

Figure 1 (a) The general formula for α-, β-, and γ -CDs. (b) The dimensions and simplified shapes of their annuli which have volumes ˚ 3 , respectively.1 (c) Views from above the wider end of the CD annuli and (d) side views where carbon and of 174, 262, and 472 A oxygen atoms are shown in gray and red, respectively, and hydrogen atoms are omitted.

CD

+

Hydro phobic guest X

K11 CD

Hydro phobic guest X

= H2O

Figure 2 The CD complexation of an either completely or partially hydrophobic guest, X, at the left to form a 1 : 1 host–guest complex, CD·X, at the right characterized by a complexation constant K11 = [CD·X]/([CD][X]) in water.

of the CD and the guest and the secondary interactions between the hydrophobic guest and the annular interior are usually stronger. The overall stereochemical fit of the guest into the CD annulus also has an important effect on the complexation process. The balance of these interactions determines the stability of the host–guest complex which translates into substantial variations of H ◦ and S ◦ of complexation as the identities of the CD and the guest change.6

While there is no necessary relationship between H ◦ and S ◦ for CD complexation, a linear-free energy relationship between TS ◦ and H ◦ according to (1) is observed for wide ranges of guests complexed by α-, β-, and γ -CDs. The slope of this linear relationship, α, indicates to what extent the enthalpic gain, H ◦ , caused by variations in the guest for a particular CD is counteracted by entropic loss, S ◦ , according to (2). When H ◦ is zero, the corresponding value of TS ◦ = TS0◦ = G◦ according to (3),

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology which indicates that the CD host–guest complex is stable in the absence of enthalpic stabilization. Under these conditions, TS0◦ = 8 (0.79), 11 (0.80), and 15 (0.97) kJ mol−1 for α-, β-, and γ -CDs, respectively, where the numbers in brackets are the corresponding α values. Thus, without enthalpic stabilization, these complexes have significantly negative G◦ values as a consequence of release of water from the CD annuli, dehydration of the peripheral CD hydroxy groups, and the guest upon complexation generating a substantial entropy gain.6 ◦

TS = αH

◦

◦

TS = αH ◦

◦

(1) ◦ ◦

(3)

In addition to the host–guest complexes composed of one CD and one guest as shown for the complex CD·X characterized by K11 in Figure 2, other CD to guest ratios arise in complexes as exemplified by CD·X2 , CD2 ·X, and CD2 ·X2 characterized by the sequential complexation constants K12 , K21 , K22 , and K22 , respectively, as shown in Figure 3. Hydrophobic guests are generally complexed more strongly than hydrophilic guests and some of the latter complex very weakly if at all. There is a tendency for those guests which best fit the CD annuli to form the more stable CD host–guest complexes, and because of this and their hydrophobic nature, aromatic guests are particularly widely employed in CD complexation studies. When CDs are modified through substitution of one or more hydroxyl groups, the range of possibilities for host–guest complexation is greatly extended and a variety of examples of the resulting complexes are considered in the following sections. The structures of CD host–guest complexes determined by X-ray crystallography show the guest to reside either completely or partially within the CD annuli.7 However, the structure in the solid state is not necessarily identical to that in solution in which most CD studies have been conducted using a range of spectroscopic and other techniques.6 Nuclear magnetic resonance (NMR) provides the most direct evidence of complexation within the CD annulus

CD

−X

CD··X

K21 + CD

−X

CD2·X

K12 +X −X

K22′ +X −X

in solution through 2D 1 H NMR rotating frame Overhauser enhancement spectroscopy, ROESY, and nuclear Overhauser enhancement spectroscopy, NOESY.8 These techniques detect through-space NOE (nuclear Overhauser enhancement) interactions occurring between the H3, H5, and H6 protons lining the inside of the CD annulus and a ˚ or smaller distance proton of a guest species within a 4 A which can only occur if the guest is wholly or partially within the annulus. Such NOE interactions are indicated by cross-peaks, the intensity of which is inversely proportional to the interaction distance raised to the power of six and directly to the number of protons in that environment.

(2)

G = H − TS

K11 +X

3

CD··X2

K22 + CD

− CD

CD2·X2

Figure 3 Multiple complexation equilibria for CD host–guest complexes of different stoichiometries.

2

CYCLODEXTRIN MODIFICATION

The arrays of primary hydroxyl groups on the C6 carbons and secondary hydroxyl groups on the C2 and C3 carbons at the narrow and wide ends of the native CDs, respectively, provide the opportunity for single substitution through to complete substitution.2, 9 Such modifications may be chosen to alter CD annular size, shape, charge, and polarity, and thereby better accommodate and orientate chosen guests within the CD annulus. In addition, CDs may be linked together10 or attached to polymeric backbones11, 12 or surfaces13, 14 in a variety of ways to present a great array of supramolecular chemistry. In this section, only a very general picture of the multitudinous methods for modification of CDs is presented. This may be supplemented by referring to the methods used to prepare the CD systems considered in succeeding sections.

2.1

Monosubstituted cyclodextrins

Monosubstitution is the most studied of CD modifications and is generally carried out through well-established pathways; the most common of which appear in Figure 4.2, 9 Direct substitution may be achieved through alkylation, acylation, and sulfonation as shown in Figure 4(a). Alternatively, and because of their ease of preparation, CD sulfonates and CD halides may be used as convenient intermediates for a range of nucleophilic displacement reactions as shown in Figure 4(b). Amino groups may also be attached in this way and provide convenient routes to further modification as seen in Figure 4(c). The modification of native CDs may be achieved through substitution of either a C2, C3, or C6 hydroxyl group and a wide range of methodologies generally employing the processes outlined in Figure 4, or similar ones, to achieve selectivity in substitution between these sites.9 These methodologies frequently exploit the knowledge that the C6 hydroxyl groups are the most basic and usually the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

4

Molecular recognition

2.2

CD-OR (a)

CD-OH

CD-OCOR CD-OSO2R CD-OSO2R

(b)

CD-Nu

CD-OH CD-Y CD-OSO2R CD-NHR

(c)

Linked cyclodextrin dimers

CD-NRCOR′

CD-Y

Figure 4 Some pathways to modified CDs through substitution at one or more primary or secondary hydroxyl groups where R is an alkyl or aryl group, Nu is a nucleophile, and Y is a halide.

most nucleophilic while the C2 and C3 hydroxyl groups are more acidic and the latter are sterically the most difficult to access. While direct substitution at C6A retains the configuration of the D-glucopyranose ring substitution at C2A and C3A usually causes inversion at these carbons as shown for the preparation of (2A S,3A S)-3A -amino-3A -deoxy-β-CD in Figure 5.15 The substitution of the A D-glucopyranose ring of β-CD, 1, with a tosyl group at C2 is preceded by interaction with dibutyltin oxide, Bu2 SnO, which activates the C2 carbon toward reaction with tosylchloride, TsCl, to give the tosylated product 2A -O-(4-methybenzenesulfonyl)-βCD, 2. Under basic conditions in the presence of ammonium bicarbonate in water, the manno-2A , 3A -epoxide-β-CD, 3, is formed and with heating transforms into the (2A S,3A S)3A -amino-3A -deoxy-β-CD, 4, with inversion at C2 and C3. Often, traces of di- and trisubstituted CD at either C2, C3, or C6 are synthesized along with the monosubstituted CD but purification by chromatography or recrystallization usually gives the pure major product. Such multiple substitutions are employed to give CDs with particularly desirable properties.2, 9 Selective substitution at either all C2, C3, or C6 or at all of these carbons simultaneously may also be achieved to give multiply substituted CDs in which the size of the annuli is greatly extended.2, 9

ring G O HO

HO 3 2 4 O 6 5

1

2.3

ring B

Cyclodextrin polymers

A wide range of polymers incorporating CDs has been prepared by adaptation of the methods employed for monosubstitution. A widely employed method is to attach a CD to a polymer with reactive groups as exemplified by the random substitution of polyacrylic acid by α-CD and β-CD in Figure 7(a).18 The substitution of polyacrylate 12 with either α-CD or β-CD through reaction of their 6A -(amino)-6A -deoxy derivatives, 13, in the presence of N,N  -dicyclohexyl carbodiimide (DCC) in water gives the randomly substituted polyacrylate 14 isolated from NaOH solution as the sodium salt. The extent of substitution may be varied at will over a substantial range by changing the reactant ratios. Similar substitutions of polyvinyl polymers have also been reported.19 An alternative approach is to build the polymer chain and incorporate the CD substituent as shown in Figure 7(b).20 In this synthesis, β-CD substituted at the C6A carbon with an aliphatic diamine, 15, is reacted with glycidylmethacrylate, 16, to form a monovinyl β-CD monomer, 17. Subsequent copolymerization with N-isopropylacrylamide, 18, yields the water-soluble β-CD substituted polyacrylamide copolymer 19 with molecular weights up to 104 . An extensive

1. Bu2SnO DMF, 100 °C, 2 h

OH 1 O

Substitution with a bifunctional substituent may either form a bridge across a CD end16 or link two CDs together in a dimer as shown in Figure 6.17 The reaction of the succinate diester 5 with either 6A -amino-(6A -deoxy-β-CD) 6 or 3A amino-((2A S,3A S)-3A -deoxy-β-CD) 7 or both produces three succinamide linked β-CD dimers N,N -bis(6A -deoxyβ-CD-6A -yl)succinamide, 8, N,N -bis((2A S,3A S)-3A -deoxyβ-CD-3A -yl)succinamide, 9, and N-((2A S,3A S)-3A -deoxyβ-CD-3A -yl)-N  -(6A -deoxy-β-CD-6A -yl) succinamide, 11, through 6A -(3-(4-nitrophenoxycarbonyl)-propionamido)(6A -deoxy-β-CD), 10, in 70–85% yield. A substantial range of linked CD dimers has been studied.10

2. Et3N, TsCl DMF, RT, 10 h

2 OSO2 HO 3 4 1 O ring G O O ring B 5 6 2 HO

NH4HCO3 O 4 3 ring G O 6 HO

Figure 5

2 O

5 3

1 O ring B

HO 6 H2N OH 3 2 1 4 5 O ring B 60 °C, 3 h O O ring G 4

NH4OH 25%

The inversions at C2A and C3A of the A ring of β-CD through substitution.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

C6A NH2

HN

H N O O

H N C6A

Pyridine

+

NH

C6A

5

b-CD RT/48 h

O O

8 6

5

O2N

C3A

NO2

NH2

H N

C3A

O O

H N C3A

Pyridine

+

b-CD RT/48 h

6, pyridine RT/48 h

9 7 H N C6A

HN O O

H N

C3A

O O

H N C6A

7, pyridine 10

O2N

RT/48 h 11

Figure 6

Reaction sequences for the syntheses of linked β-CD dimers 8, 9, and 11.

=

+

CO2H 12

=

=

=

C6A NH2

a-CD or b-CD

CO2Na NH

NaO2C O 1. DCC, 60 °C/NMP 2. NaOH

C6A

a-CD or b-CD

14

13

(a)

=

=

O O

O O C6A NH

NH2 n

b-CD

16

HO A

C6 NH

O glycidylmethacrylate

(b)

15

NH

n b-CD

HN

17

O

O

O

HO 18

HN

N -isopropylacrylamide

n

C6A NH

DMF, 70 °C

DMF, 60 °C

n = 2 or 6

HN O

19

b-CD

Figure 7 (a) The substitution of polyacrylic acid 12 with either α- or β-CDs to give the randomly substituted sodium polyacrylate 14. (b) The formation of the β-CD substituted copolymer 19 through reaction of 15 with 16 to give 17 which is then reacted with 18.

range of CD-substituted polymers and their properties and biomedical applications have been reported.21

3

CYCLODEXTRIN ENANTIOSELECTIVITY

As a consequence of the homochirality of native and modified CDs, there exists a possibility of chiral discrimination between enantiomeric guests, or enantioselectivity, in

the formation of diastereomeric host–guest complexes.22 This CD enantioselectivity is of major importance in a range of racemate resolution technologies as exemplified by thin layer, high-performance liquid, and gas–liquid phase chromatography and capillary electrophoresis.23–26 In such usage, the CD may either be attached to a surface to form a stationary phase or may be part of a moving phase. Selective diastereomeric CD host–guest complex precipitation is also used in racemate resolution.7, 27 To gain a basic insight into the nature of the chiral interactions involved, several examples of enantioselectivity between guests by CDs and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

6

Molecular recognition β-CD diastereomeric complexes. However, the extent of enantioselectivity tends to decrease when the guest enantiomers are modified to complex more strongly as the additional weak interactions induced diminish the chiral complementarity between the guest and the β-CD annulus. For many chiral guests, the native CDs show small selectivities in complexing enantiomers and extensive thermodynamic studies show that for β-CD a range of stereochemical factors influence the mode of host–guest complexation.32 Generally, the small extent of enantioselectivity is attributable to small differences in H ◦ for the formation of the diastereomeric host–guest complexes being offset by counteracting S ◦ differences such that differences in G◦ are small. However, exceptions to this generalization do occur as exemplified by the strong enantioselectivity shown by β-CD in complexing helical 1,12dimethylbenzo[c]phenanthrene-5,8-dicarboxylate for which the complexation constants, K11M and K11P for the left- and right-handed helical enantiomers, 22 and 23 in Figure 8(b), are 1.87 × 104 dm3 mol−1 (H ◦ = −51.1 kJ mol−1 , S = −90 J mol−1 K−1 ) and 2.2 × 103 dm3 mol−1 (H ◦ = −35.1 kJ mol−1 , S ◦ = −53.2 J mol−1 K−1 ), respectively, such that the difference in free energy for complexation between the enantiomers 22(M−) and 23(P −)G◦ = 5.2 J mol−1 K−1 in aqueous solution at 298.2 K.33 Both 2D 1 H ROESY NMR and molecular modeling studies indicate that the carboxylate groups of 22 and 23 are close to the ring of secondary hydroxyl groups of β-CD in the host–guest complex and that 23 penetrates more deeply into the β-CD annulus. However, this deeper penetration into the cavity is enthalpically unfavorable but entropically favorable because it requires a greater dehydration of the carboxylate groups of 23 than is the case for the lesser penetration of 22. Thus, the enantioselectivity of β-CD is dominated by the difference in enthalpy due to the deeper penetration of 23 into the βCD annulus. Enantioselectivity between 22 and 23 is also shown by γ -CD for which K11M = 3.1 × 103 dm3 mol−1 (H ◦ = −30.2 kJ mol−1 , S ◦ = −34.4 J mol−1 K−1 ) and

other aspects of the chiral nature of CDs are discussed below.

3.1

Chiral discrimination by native cyclodextrins

The liquid chromatographic separation of the R- and Senantiomers of tryptophan and other aminoacids and several of their derivatives by α-CD bonded to silica gel28, 29 prompted a detailed study of enantioselectivity complexation of tryptophan by α-CD.30, 31 Both 1 H and 13 C NMR spectroscopies show that coupling constants, chemical shifts, and nuclear relaxation times change more for R-tryptophan upon complexation by α-CD than for Stryptophan in D2 O at 298.2 K. For both enantiomers, 1 H NOESY NMR studies show the indole ring to be adjacent to the α-CD secondary hydroxyl groups with the benzene ring being deeper in the annulus in the host–guest complex (Figure 8a). Molecular modeling shows this orientation to be ∼4 kJ mol−1 more stable than the reverse orientation. It also shows that the chirality of R-tryptophan allows it to form twice as many hydrogen bonds with α-CD as does S-tryptophan. Such hydrogen bonding stabilizes the α-CD·R-tryptophan complex by 12.7 kJ mol−1 more than it does the α-CD·S-tryptophan complex. It appears that for the α-CD diastereomeric host–guest complexes to show significant differences in stability the chiral guests must either fit snugly into the annulus or there should be a highly localized interaction with the annular interior, and that the guest stereogenic center should interact strongly with one of the α-CD C2 and C3 hydroxyl groups.28–31 Thermodynamic studies of complexation of 43 enantiomer pairs by β-CD show that there is a fine balance between the orientation of the guests within the annulus and the strength of complexation in determining enantioselectivity.32 Thus, guests with low symmetry nonpolar groups which complex in the β-CD annulus and those with larger distances between the stereogenic center and their most hydrophobic group are more likely to show enantioselectivity in their NH a-CD

+

(a)

(b)

CO2− 20

−O C 2

22

NH

KR or KS

CO2−

CO2−

NH3+

−

21

O2C 23

NH3+

CO2−

Figure 8 (a) Complexation of R- and S-tryptophan zwitterion, 20, to form diastereomeric host–guest complexes 21 characterized by KR and KS , respectively. (b) The M-, 22, and P -, 23, enantiomers of helical 1,12-dimethylbenzo[c]phenanthrene-5,8-dicarboxylate. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

7

K11P = 6.9 × 102 dm3 mol−1 (H ◦ = −16.0 kJ mol−1 , S ◦ = 0.45 J mol−1 K−1 ) and G◦ = 3.7 J mol−1 K−1 . The lower stabilities of the γ -CD host–guest complexes are attributable to a deeper penetration of the carboxylate groups into the γ -CD annulus causing a more extensive and endothermic dehydration of 22 and 23. It appears that the difference in stability of the diastereomeric host–guest complexes in the two systems is due to a chiral helical structure assumed by CDs in water.

Table 1 Host–guest complexation constants, K11 , for the hosts β-CD, D-24 and L-24 and the guests D-26, L-26, D-27, and L-27 in the formation of 28 and 29 in water at pH 7.0 at 298.2 K.

3.2

enantioselectivity is small to moderate and tends to favor the complexation of D-26 and D-27 by β-CD, L-26 and D-27 by D-24, and D-26 and L-27 by L-24. Hexa-coordinated tris-bidentate metal complexes exist as - and -enantiomers which may form diastereomeric host–guest complexes with native and modified CDs. This is exemplified by the - and -tris-(1,10-phenantholine)ruthenium(II) complexes - and -[Ru(phen)3 ]2+ , 30 and 31, which complex with heptakis(6-carboxymethylthio-6-deoxy)-β-CD, 32, to form host–guest complex, 33, and its -[Ru(phen)3 ]2+ analog as shown in Figure 10.36 The complexation is largely dependent on the electrostatic attraction between dicationic 30 and 31 and heptaanionic 32. The host–guest complexation constants K11 and K11 = 1.25 × 103 and 5.90 × 102 dm3 mol−1 indicate an enantioselectivity of 2.12 by 32 in favor of 30 over 31 as determined by 1 H NMR spectroscopy in D2 O at pD 7 and 298.2 K. It appears from 2D 1 H ROESY NMR studies that the origin of the discrimination is that 30 penetrates more deeply into the annulus of 32 than does 31. Under the same conditions, the enantioselectivity between - and -[Rh(phen)3 ]2+ is 1.43 in favor of the -enantiomer. When 32 is replaced by its γ -CD analog, the enantioselectivity is 1.28 and 1.13 in favor of the -enantiomer of the Ru(II) and Rh(II) complexes, respectively. This enantioselectivity is also detected by capillary zone electrophoresis where the two enantiomers are cleanly separated with the -enantiomer showing the longer retention time. These data indicate that subtle

Chiral discrimination by modified cyclodextrins

The introduction of molecular asymmetry into the CD structure by a single substitution renders all of the glucopyranose units inequivalent, which may affect the extent of enantioselectivity in the complexation of chiral guests. In addition, the substituent may interact directly with the guest, and thereby enhance enantioselectivity as appears to be the case with 6A -amino-6A -deoxy-β-CD in which the amino group is protonated in water at pH 6.9 and engenders a stronger interaction with negatively charged guests than is the case for β-CD.34 Nevertheless, enantioselectivity remains quite small due to the enthalpy–entropy offset discussed above. Sometimes there may be a competition between the substituent self-complexing inside the CD annulus and intermolecular host–guest complexation of chiral guests. This is exemplified by the 6A -N-(N  -formyl-D-phenylalanyl)6A -deoxy-amino-β-CD and its L-analog, D-24, L-24, D-25, and L-25, which undergo self-complexation in competition with intermolecular host–guest complexation of the D- and L-enantiomers of N-dansylalanine, 26, and N-dansylphenylalanine, 27, to form the diastereomeric host–guest complexes, 28 and 29, respectively, as shown in Figure 9.35 From the host–guest complexation constants, K11 , in Table 1, it is apparent that K11 for the β-CD complexes is either larger than or similar to those for the D-24 and L-24 complexes with the exception of the L-24·D-27 complex 29 which is more stable. Generally,

K11 (dm3 mol−1 )

Guest Host = β-CD

D-24

179 ± 13 114 ± 13 197 ± 20 153 ± 14

42 ± 13 54 ± 10 160 ± 36 83 ± 28

D-26 L-26 D-27 L-27

L-24

C6A NH

HN HN

O NH 25

O

113 ± 18 95 ± 17 139 ± 24 231 ± 45

O

O

O

O C6A NH

C6A NH b -CD

+ 24

N

N −O

2C

SO2 NH

R 26 R = CH3; 27, R = CH2Ph

−O

2C

NH

SO2

b-CD

R 28 R = CH3; 29, R = CH2Ph

Figure 9 The competing intramolecular equilibrium between 24 and 25, in which the substituent may be in either the D- or L-form, and the intermolecular equilibria between 24, 26, 27, 28, and 29 where 26 and 27 may be in either the D- or L-form. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

8

Molecular recognition

2+

+

N N

b -CD

b -CD C6

A

C6

S

A

CO2−

32 per-CO2− - b-CD

31 Λ-[Ru(phen)3]2+

N N

S

CO2− 30 ∆-[Ru(phen)3]2+

N RuII

N

N

N

5−

S

N

N

N

C6A

K11∆ / K11Λ

RuII

;

RuII

CO2−

7−

S

N

N

N

N

C6A

N

N N

CO2−

2+

33

Figure 10 The complexation of - and -[Ru(phen)3 ]2+ , 30 and 31 (viewed down the C3 axes), by 32 for which only two of the seven—SCH2 CO− 2 substituents are shown. Only the most stable diastereomeric host–guest complex, 33, is shown. +

O C6A

b-CD

N

O

O O Eu(OH2)5

H N +

2

C6A N

1

HN

NH2

H2N

O O Eu(OH2)3 O−

O−

3 4

34

0

O

O

O

O

5 35

36

Figure 11 The complexation by the metallocyclodextrin 34 of either D- or L-tryptophan anion 35 to form the diastereomeric host–guest complexes 36.

changes in enantioselectivity are induced by CD annular and metal complex size variation. In contrast, α-, β-, and γ CDs show little tendency to complex either [Ru(phen)3 ]2+ or [Rh(phen)3 ]2+ largely because of the lack of electrostatic attraction between these CDs and the metal complexes. Alternatively, the metal center may be coordinated by a CD substituent as shown for pentaaquo(6A -[bis(carboxylatomethyl) amino]-6A -deoxy-β-CD)europium(III), 34, which enantioselectively complexes the D- and L-tryptophan anions, 35, to form the diastereomeric host–guest complexes 36 in D2 O at pD 10 as shown in Figure 11.37 The 1 H NMR doublet assigned to the tryptophan H2 proton resolves into two doublets with the upfield doublet being assigned to D-35, and a partial resolution of the H3 triplet into two triplets also occurs while the resolutions of the H1, H4, and H5 resonances are smaller. Separate resonances for free and complexed 35 are not observed consistent with exchange between these states being in the fast exchange limit of the 600 MHz 1 H NMR timescale with 34 acting as a chiral shift reagent through its complex 36. Crosspeaks in the 2D 1 H ROESY NMR spectrum arising from dipolar interactions between the β-CD H3 and H5 annular protons of 36 and the H2 proton of complexed D-35 show that they are in close proximity in 36, whereas analogous cross-peaks show that the β-CD H3 and H5 annular protons of 36 and the H4 proton of complexed L-35 are in close proximity and indicate that D-35 and L-35 are differently oriented in 36.

Chiral discrimination also extends to the formation of the helical polymer 37 in water when the 4-tertbutoxyaminocinnamoylamino substituent at the C3A carbon of the modified α-CD monomer is complexed by a second monomer and so on to form a polymer composed of at least 15 such units in water as shown in Figure 12.38 Negative and positive Cotton circular dichroic effects at 327 and 288 nm, respectively, are consistent with the polymer assuming a left-handed anticonfiguration and a slanted complexation of the substituents in the α-CD annuli.

4

CYCLODEXTRIN CATALYSTS

The complexation of hydrophobic substrates in the annuli of CDs where they may react catalytically with either the hydroxyl groups defining the CD ends or catalytic groups substituted on either end render them potential enzyme mimics.39 A wide range of such CD-based catalysts has been studied. They generally exhibit kinetic characteristics similar to the Michaelis–Menten scheme typifying enzymes which include saturation, nonproductive substrate binding, and competitive inhibition although often at pH values distant from a physiological pH. While other CD catalysts also exhibit these characteristics, they do not have biochemical analogs. Four examples of CD-based catalysts are now discussed.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

a-CD a-CD

O

O C3 C3

Figure 12

A

A

O O

NH

NH

O

a-CD

O O

C3A

a-CD

O O O

NH

O C3A

37

NH

O

A left-hand helical rotating polymer, 37, formed by 3A -(4-tert-butoxyaminocinnamoylamino)-3A -deoxy-α-CD. CH3

O−

O− O

CH3 O

a-CD or b -CD

38

9

CH3

O

O

O O−

CH3 O

O−

O O−

O K11 = 1/KM

+

kcat

NO2 NO2 39

+ OH−, − H+

+

NO2

NO2 40

41

42

− CH3CO2−

43

38

Figure 13 The sequence in which a secondary alkoxy group of 38 catalyzes the hydrolysis of 39–43 through the Michaelis-type complex 40.

4.1

A cyclodextrin hydrolase mimic

An early example of hydrolase mimicry is the hydrolysis of 3-nitrophenyl acetate catalyzed by α-CD and βCD at pH 10.6 in 5% acetonitrile water as shown in Figure 13.40, 41 Under these conditions, a small proportion of the CDs carry a deprotonated secondary hydroxyl, or alkoxy, group, 38 (pKa = 12.1), which upon complexation of 3-nitrophenylacetate, 39, makes a nucleophilic attack at the guest carboxylate carbon in 40 which resembles a Michaelis complex. This results in the hydrolysis of 39 to 43 and the attachment of an acetate group to the CD of 41 and 42. For the system to be truly catalytic, this acetate group should subsequently hydrolyze to regenerate the catalyst 38 as shown in Figure 13. The K11 characterizing the formation of 40 is equivalent to the reciprocal of the Michaelis constant KM and kcat is the rate constant for the hydrolysis in 40 which is compared with that for the hydrolysis of 39, kuncat , under the same conditions but in the absence of the CD. It is seen from Table 2 that kcat /kuncat = 300 and 96, respectively, for 3-nitrophenyl acetate in the presence of α- and β-CDs. For 4-nitrophenyl acetate, the corresponding kcat /kuncat = 3.4 and 9.1 consistent with the magnitude of the catalysis being dependent on the nature of the guest stereochemistry and probably the positioning of the guest with respect to the catalytic center in the Michaelis complex. This is further emphasized by the data for 3- and 4-tert-butylphenylacetate for which kcat /kuncat = 260 and 1.1, respectively, in Table 2. It is also evident that the magnitude of KM has little relationship to

catalytic effectiveness. (A similar catalysis by a metallocyclodextrin (metalloCD) is discussed in Section 5.1).

4.2

A cyclodextrin enantioselective hydrolase mimic

The more effective CD-based catalysts are usually substituted CDs in which the substituent has an important role in the catalysis. Thus, the enantioselective aldol condensation of acetone and 4-nitrobenzaldehyde, 45, to the corresponding aldol, 49, by a β-CD substituted at the C6A carbon with a 1,2-diaminocyclohexane, 44, appears to proceed through the sequence shown in Figure 14.42 The reaction occurs in 5% v/v acetone/water at pH 4.80 and 298.2 K under which conditions the secondary amine is protonated (pKa ∼ 5.7). The KM = 6.31 × 10−3 mol dm−3 and pertains to 46 prior to the formation of the enamine in 47 which is thought to be rate-determining and characterized by kcat = 1.05 × 10−4 s−1 . The subsequent formation of the new carbon–carbon bond in 48 is followed by regeneration of the catalyst 44 and release of aldol 49 in the enantiomer ratio R-49/S49 = 97/3. The enantiomeric excess of R-49 is attributed to the combined chiralities of the diamino substituent and the β-CD annulus positioning 45 in 47 such that the enamine attachment to the aldehyde carbon of 45 in 47 dominantly occurs from one side. Thus, while the rate of formation of 49 is only accelerated 6.45-fold by 44 over the reaction in the presence of the small molecule 1,2-diaminocyclohexane precursor to 44, the enantioselectivity in the reaction is high, and although the reaction pH is low, it resembles the action of an aldolase.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

10

Molecular recognition Table 2

Constants for the hydrolysis of phenyl acetates in the presence of α- and β-CDs.a 104 kuncat (s−1 )b

Substrate

104 kcat (s−1 )

102 KM (mol dm−3 )

kcat /kuncat

3-Nitrophenylacetate 4-Nitrophenylacetate 3-tert-Butylphenylacetate 4-tert-Butylphenylacetate

46.4 69.4 4.90 6.07

Catalyst = α-CD 4250 1.9 243 1.2 1290 0.20 6.7 0.65

300 3.4 260 1.1

3-Nitrophenylacetate 4-Nitrophenylacetate 3-tert-Butylphenylacetate

46.4 69.4 —

Catalyst = β-CD 4440 0.80 634 0.61 1220 0.013

96 9.1 250

a In b

0.5% acetonitrile–water at pH 10.6 and 298.2 K. In the absence of CD.

H + N C6A H A NH2 C5 b -CD

O K11 = 1/KM

+ NO2

44

H N + C6A H A NH2 C5 OO

H N C5A H OH

H NH C5A O

+ H2O

− OH+

kcat

H + N C6A O H C5A NH2 HO

H + N C6A

H + N C6A

b-CD

−

− OH

5%v/v acetone/ water pH 4.80

NO2

45

NO2

46

+

NO2

NO2 47

49

44

48

Figure 14 The enantioselective aldol condensation of 4-nitrobenzaldehyde, 45, with acetone catalyzed by 44 to give the aldol R-49 in 97% enantiomeric excess. The loss of hydroxide from 46 and the loss of a proton from 47 are likely to be highly synchronized.

4.3

A cyclodextrin oxidation catalyst

of benzilic alcohols to aldehydes and anilines to nitrobenzenes.44

The oxidation of an alcohol catalyzed by a modified CD is illustrated in Figure 15, where the catalyst is β-CD substituted at two C6 carbons with dihydroxyacetone through ester bonds 50.43 In the presence of H2 O2 , which appears to add to the ketone function to form a hydroperoxide adduct as shown in 51, benzylic alcohol, 52, forms the Michaelistype complex 53 with a KM = 2.0 × 10−3 mol dm−3 at pH 7.25 and 298.2 K in aqueous solution. Subsequently, oxidation occurs in 53 in which 52 is oxidized to benzaldehyde, 54, through an overall transfer of an electron pair with kcat = 2.69 × 10−7 s−1 (kcat /kuncat = 1690) and regeneration of the catalyst. The α-CD analog of 50 is a similarly effective catalyst and both catalyze oxidations for a range O O C6A

O

O HO O

C6 b -CD

50

Figure 15

O

O C6A

O

HO

+ H2O2

H 52

O +

O HO C6

K11 = 1/KM

4.4

An organometallocyclodextrin hydrogenation catalyst

CDs may be used to change the characteristics of other catalysts as exemplified by the attachment of CDs to catalytic organometallic complexes to give organometallocyclodextrins. In such catalysts, the metal center has a low oxidation state and acts as a soft acid which interacts with soft base centers in organic molecules. Potentially, the CD component in such a catalyst may result in selectivity of complexation of reactant species and thereby selectivity in catalysis. O HO O O O O O O HO C6A C6 C6A kcat H HO

O

O O C6 O +

+ 2H2O

− 52

51

53

50

54

The catalyzed oxidation of benzylic alcohol, 52, to benzaldehyde, 54, in the presence of catalyst 50 and H2 O2 .

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

Ph P PhI Rh N P Ph Ph

S C6A

+

+

56

Ph P PhI Rh N P Ph Ph

+

S Organic phase C6A Aqueous phase

b CD

bCD

55

57

Ph P PhI Rh N P Ph Ph

S C6A

+

Ph P PhI Rh N P Ph Ph

S C6A

11

+

+

59

Organic phase Aqueous phase

H2

bCD

bCD

58

55

Figure 16 The hydrogenation of 56 at the organic/aqueous phase interface through complexation in the β-CD annulus of catalyst 55 to give 59.

An interesting example of this is provided by the watersoluble Rh(I) metalloCD 55 catalyzing the hydrogenation of alkene 56 in a two-phase system in which the organic phase is N,N-dimethylformamide and the aqueous phase is 30% N,N-dimethylformamide and 70% water as shown in Figure 16.45 It appears that the catalysis occurs at the phase interface where the organometallic Rh(I) component of 55 is also soluble in the organic phase together with 56, and the β-CD component is soluble in the aqueous phase. An equilibrium exists between 55 and 56 and the host–guest complex 57 in which 56 resides largely in the β-CD annulus. This brings 56 into close proximity to the Rh(I) center to which it attaches through its alkene bond and hydrogenation occurs in a sequence of catalytic steps including oxidative addition and reductive elimination to give the hydrogenated product 59 in a second complex 58. Subsequent release of 59 regenerates the catalyst 55. When the phenyl group of 56 is replaced by n-C6 H13 to give alkene 60 it is found that 55 preferentially hydrogenates 56 over 60 in a ratio of 68/32 at 295.2 K probably because of preferential complexation of 56 in the β-CD annulus of 55. In contrast, the catalyst in which a phenyl group replaces β-CD in 55 shows no discrimination in hydrogenation of the two alkenes. In the presence of carbon monoxide and hydrogen catalyst 55 also hydroformulates alkenes.

5

METALLOCYCLODEXTRINS

MetalloCDs are CDs bearing one or more groups which coordinate metal ions. By applying the principles of Lewis acid–base theory, the nature of the coordinating group may be varied to selectively coordinate a wide range of metal ions.46, 47 The simplest metalloCDs are those formed when either CD hydroxyl groups or deprotonated hydroxyl groups coordinate a metal ion, but they tend to be less stable in solution by comparison with metalloCDs incorporating multidentate coordinating groups and consequently are less

studied. Here, we discuss six metalloCDs which are biological mimics, energy transfer systems, and organometallic complexes and which exemplify the great potential to design a range of fascinating systems.

5.1

Metallocyclodextrins as biological mimics

Apart from their intrinsic interest as metal complexes, a major interest in metalloCDs arises from the metal ion being coordinated adjacent to the hydrophobic CD annulus and thereby resembling the active site of a metalloenzyme. The formation of such a metalloCD in water is illustrated by the diamine in 6A -(3-aminopropylamino)-6A -deoxy-βCD, 61, coordinating Cu(II) from [Cu(OH2 )6 ]2+ with a Kcoord = 2.2 × 107 dm3 mol−1 to give the metalloCD 62 in which only two molecules of water are shown coordinated to Cu(II) in Figure 17 although there could be up to four.48, 49 One coordinated water molecule deprotonates with a pKa = 7.84 to give the hydroxo species 63. While both 62 and 63 may complex the substrate 4-tert-butyl-2nitrophenyl phosphate, 64, in their β-CD annuli it is only 65 which is catalytically active (coordinated water usually shows little activity as a nucleophile). The formation of 65, in which it is possible that one of the phosphate oxygens is coordinated by Cu(II), is characterized by a Michaelis constant KM = 1.2 × 10−3 mol dm−3 at pH 7 and 298.2 K. Under these conditions, the hydroxo group of 65 makes a nucleophilic attack on the phosphorus of 64 with a rate constant kcat = 2.3 × 10−2 s−1 at pH 7.0 and forms the intermediate 66 which dissociates to the hydrolysis products dimethylphosphate and 4-tert-butyl-2-nitrophenol and its conjugate base to regenerate the catalyst 61. Thus, the rate of hydrolysis is accelerated in the presence of 64 as compared to that in its absence, kuncat = 3.2 × 10−7 s−1 , such that kcat /kuncat is 7.2 × 104 at 298.2 K. The catalysis by 63 exhibits the Michaelis–Menton kinetic profile characterizing enzymatic catalysis and more than

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

12

Molecular recognition

+

OH H2O H2N

C6A

HN

CuII

H2N

bCD

C6A

HN

+

b CD

O

O

P

+

OCH3 OCH3 P H2O CuII O NO2 H2N O HN HO

OCH3 OCH3

NO2

64

− 64

63

61

KM

65 k cat

+ [CuII(OH2)6]22+

pKa − H+

Kcoord

+ H+ 2+

OH2

H2N

O H2O

CuII

H2O

C6A

HN

+

OH

O

OCH3 OCH3 + HO P O

CuII

H2O

NO2

HN HO

H2N

bCD

OCH3

P

CuII

H2N

A

+HO

C6

HN

b CD

H2O

62

Figure 17

+

OCH3

NO2

63

66

The catalysis of the hydrolysis of 4-tert-butyl-2-nitrophenyl phosphate 64 by 63. +

2+

C6A b-CD

A

C6A′ b -CD

N Cu OH2 H2O N

S

S

S

S

C6 b-CD

pKa − H+

N Cu OH2 HO N

C6A′ b-CD

+ H+

68

67 KM

O −

O HN HO

− 69 + H2O

kcat

+

+

2H2O

C6A b-CD

+

S

S

69 − H 2O

C6A′

N Cu O HO N

NO2 O

HN

Figure 18

O HN

NO2

NO2

O

b -CD

70

The catalyzed hydrolysis of 4-nitrophenyl indol-3-ylpropionate 69 by 68.

ten 64 are catalytically hydrolyzed by each 63 representing a turnover rate >10. Similar studies have been reported for the hydrolysis of carboxylate esters with Zn(II) as the metal ion in an imitation of carboxypeptidase.50 Generally, the catalytic activity of such metalloCD enzyme mimics shows a substantial variation with change in the coordination site and metal ion. MetalloCDs are also formed by linked CD dimers in which the linkers incorporate metal coordinating groups.39, 51 Such a metalloCD is represented by 67 in Figure 18 where two β-CDs are linked through sulfur substituted at the 6A carbons and the Cu(II) coordinating

group is the bidentate 2,2 -bipyridyl unit. One of the two waters molecules coordinated to Cu(II) in 67 has a pKa = 7.15. The second water molecule is displaced as Cu(II) coordinates the carbonyl oxygen of 4-nitrophenyl indol-3ylpropionate, 69, as it is complexed in both β-CD annuli of 70 with a KM = 1.4 × 10−5 mol dm−3 , such that the carbonyl group is adjacent to the catalytic center. The nucleophilic attack of the hydroxo ligand on the carbonyl carbon of 69 results in hydrolysis to indol-3-ylpropanoate and 4-nitrophenol with a high catalyst turnover characterized by kcat = 2.05 × 10−4 s−1 . Thus, in water at pH 7 and 310.2 K, the observed rate constant, kobs = 5.5 × 10−4 s−1 ,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology compares with 3.0 × 10−8 s−1 observed in the absence of catalyst 68; a 18 300-fold acceleration of the rate of hydrolysis of 69 when [67] = 10−4 mol dm−3 .

5.2

Metallocyclodextrins as myoglobin mimics

A particularly sophisticated example of a metalloCD is the myoglobin mimic 71. Here, [5,10,15,20-tetrakis(psulfonatophenyl)porphinato]iron(II), Fe(II)TPPS, is bound  within the linked CD dimer 2A ,2A -O-[3,5-pyridinediylbis(methylene)]bis-per-O-methyl-β-CD in which all hydroxyl groups are methylated, except those on the C2A  and C2A carbons, to form the supramolecular complex 71 with a formation constant Kform > 107 mol dm−3 as shown in Figure 19.52 This supramolecular complex binds dioxygen with a change in Fe(II) coordination from five to six and a K = 3.4 × 103 dm3 mol−1 , the ratio of the rate constants kon = 1.3 × 107 dm3 mol−1 s−1 and koff = 3.8 × 103 s−1 at pH 7.0 and 298.2 K in complex 72, and thereby mimics oxymyoglobin. Alternatively, 71 may bind carbon monoxide, but more strongly with K = 5.0 × 107 dm3 mol−1 , kon = 2.4 × 106 mol dm−1 s−1 , and koff = 4.8 × 10−2 s−1 . In another reaction of 71, nitric oxide is oxidized to nitrate and Fe(II) is oxidized to Fe(III). An intriguing experiment has been reported for an analog of 71 in which the −OCH2 PyCH2 O–bridge joining the C2A  and C2A carbons of the linked permethylated β-CDs is replaced by a — SCH2 PyCH2 S — bridge joining the C3A  and C3A carbons of the two linked β-CDs in which all hydroxyl groups are methylated except those on the C3A  and C3A carbons.53 When the oxy form of this analog was injected into a Wistar rat, it was found that the CO form was excreted in urine in accord with CO binding more strongly than O2 , and with no ill effects on the test animal. This promises an opportunity to both sequester and monitor CO in mammals.

5.3

Energy transfer in metallocyclodextrins

When several CDs bearing coordinating groups simultaneously form a metal complex, a variety of metalloCD structures may result. An example of this appears in Figure 20 where a 6-coordinate Ru(II) metalloCD, 73, forms in aqueous solution as a result of bidentate coordination by the bipyridyl nitrogens of three 6A -mono[4-methyl-(4 -methyl2,2 -bypyridyl)]-per-O-methylated-β-CDs in which all of the C2, C3, and C6 (except C6A ) hydroxyl groups are methylated.54 Each of the -per-O-methylated-β-CD, TMβCD, annuli of 73 subsequently complex the adamantyl units of three 4 -((1-adamantyl)-2,2 : 6 , 2 -terpyridyl)(2,2,2terpyridyl)osmium(II) complexes, 74, to form the assembly 75. When 75 is irradiated at 324 nm, this energy is absorbed by the Ru(II) unit and is transferred with k = 6.4 × 1010 s−1 to the three complexed Os(II) complexes which then luminesce at 730 nm on a picoseconds timescale. Energy transfer also occurs in the dimeric assembly 77 formed in dimethylformamide when the fullerene C60 is complexed by two Re(I) metalloCDs 76 shown in Figure 21.55 When triscarbonyl(6A -(4-pyridylmethyl) amino-6A -deoxy-β-CD)(2,2 -pipyridyl)rhenium(I), 76, is excited at 340 nm, it luminesces at 570 nm with a lifetime τ = 98 ns and a quantum yield φ = 7 × 10−3 . However, when 76 is complexed in the dimer 77 τ = 12 ns and φ = 1.4 × 10−3 ; and reductions consistent with either energy or electron transfer between the Re(I) and C60 units of 77.

5.4

Organometallocyclodextrins

Organometallic complexes have been less studied than conventional coordination complexes as substituents in metalloCDs.56 Of these, ferrocene and its derivatives have been the most studied as both a guest in a range of CD host–guest complexes and as metalloCD substituents as exemplified by 78–80 in Figure 22.57 In 78, ferrocene

SO3− O

TMb-CD C2A O

SO3−

−O S 3

O C2A′ TMb-CD

TMb-CD C2A O

C2A′ TMb-CD

N N N FeII N N

N SO3− + O2

kon koff

N N FeII N N

−O S 3

O 71 Fe(II) TPPS

Figure 19

72 SO3−

13

Fe(II) TPPS

SO3−

O

SO3−

Dioxygen binding by the metalloCD myoglobin mimic 72, where TMβ-CD is the per-O-methylated CD.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

14

Molecular recognition

N N 75

73

C6A TMb-CD

O

TMb-CD C6A O

N

N

N

+3

RuII N

N

O

N

N

N

N

OsII

N N O C6A

N N N

N

N

OsII

N N

N

C6A

N

N

C6A

N RuII N

N

O

N

OsII N

N

N

O C6A Energy transfer k = 6.4 × 1010 s−1

74

324 nm

TMb-CD

N N

730 nm

N

OsII

N

N N

Figure 20 The complexation of the adamantyl groups of three Os(II) complexes, 74, by the TMβ-CDs of 73 to form the multimetal centered assembly 75 in which rapid energy transfer from the Ru(II) center to the Os(II) centers occurs followed by luminescence at 730 nm.

2+

+ N b-CD OC

HN

ReI N

2 OC N

N

C6A NH

CO C6A

OC

+

b-CD

N

OC 76

N

ReI CO CO

b-CD

CO HN

ReI N

OC

N

C6A

C 60

N 77

C60

Figure 21 The complexation of the fullerene C60 by two Re(I) metalloCDs, 76, to form the dimeric assembly 77 in which either photoinduced energy or electron transfer from the metal complex substituents of 77 to C60 occurs.

Fe b-CD

O 78 C6A

HO + O

HO

81

Fe

;

Fe b-CD

b-CD A

82 C6

O

O

79 C6A

O NH

Fe

O

b-CD

80 C6A

NH

Figure 22 The self-complexation of the ferrocenyl substituents of 78–80 and the displacement of the ferrocenyl substituent from the β-CD annulus of 78 by 2-methyl-2-adamantol, 81, to form the host–guest complex 82.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

and the name catenane is derived from Latin for chain, catena. The macrocyclic nature and size of CDs render them ideal components for rotaxanes and catenanes; the examples of which are discussed below.

is attached to β-CD at C6A through an ester link to a cyclopentadienyl ring while in 79 and 80 an amide link is used. In 20% ethylene glycol aqueous solution, circular dichroic and 1 H NMR studies show 78–80 to self-complex largely because of the hydrophobic nature of the ferrocenyl substituent. However, in the presence of 2-methyl-2-adamantanol, 81, the ferrocenyl substituent is largely displaced from the β-CD annulus of 78 to form the host–guest complex 82 because of the hydrophobic nature of 81 and its good fit to the β-CD annulus. The ferrocenyl substituent of 79 is less readily displaced by 81, which is attributed to the more rigid nature of its amide link to βCD as compared with that of the ester link of 78. There is no detectable displacement of the aliphatic substituted ferrocenyl substituent of 80 by 81 and this is attributed to the hydrophobic nature of the aliphatic substituent and the tighter fit of the ferrocenyl substituent of 80 to the β-CD annulus. The metalloCD 80 forms aggregates in solution and shows surfactant behavior as a consequence of its aliphatic substituent.

6

6.1

Rotaxanes and catenanes are unusual supramolecular assemblies held together by mechanical restraint.58 A rotaxane consists of a macrocycle threaded onto a linear molecule in a similar manner to the mounting of a wheel on an axle. Accordingly, the name rotaxane is derived from Latin for wheel and axle, rota and axis, respectively. A catenane consists of macrocycles joined as links in a chain

H 2N

Cyclodextrin rotaxanes

The first CD rotaxanes were reported by Ogino in 1981 as exemplified in Figure 23.59 The rotaxane is formed in dimethylsulfoxide by threading β-CD onto 1,12diaminododecane 83 and two inert octahedral Co(III) complexes are then attached to either end to prevent dethreading as shown in Figure 23, which also illustrates the nomenclature for rotaxanes and the principle steps through which many CD rotaxanes are formed. A labile equilibrium exists between β-CD and 1,12-diaminododecane and the [2]-pseudorotaxane 84, [2]-[1,12-diaminododecane]-[βCD]-[pseudorotaxane]. Thus, the nomenclature has the form [number of entities]-[threading species]-[CD]-[type of rotaxane] where “pseudo” indicates that the CD may readily dethread. When cis-[Co(en)2 Cl2 ]+ is added, one end of 83 displaces a chloro ligand from cis-[Co(en)2 Cl2 ]+ to form cis-[Co(en)2 (NH2 (CH2 )12 NH2 )Cl]2+ , 85, and the analogous [2]-pseudo rotaxane, 86, which coexist in a labile equilibrium. Subsequently, a second cis-[Co(en)2 Cl2 ]+ adds to 85 and 86 to form 87 and 88, respectively, where the latter is a [2]-rotaxane, [2]-[µ-(1,12-diaminododecane)bis(cischloro)(bis(1, 2-diaminoethane))cobalt(III)]-[β-CD]-[rotaxane] in which the Co(III) complex end groups are too large to allow β-CD to dethread. The Co(III) complexes at each end of the axle are of either  or  chirality such that the axles formed are a mixture of the combinations

CYCLODEXTRIN ROTAXANES AND CATENANES

b-CD

+ b-CD

NH2 83

NH2

H 2N

− b- CD

84 [2]-Pseudorotaxane + cis- [Co(en)2Cl2]+

+ cis- [Co(en)2Cl2]+

2+ H 2N

NH2

H 2N

NH2

H 2N

Co

NH2 Cl

+ b- CD

2+ H 2N

− b- CD H2N

H2N

85 NH2

Figure 23

NH2

H 2N

Cl

86 [2]-Pseudorotaxane

+ cis- [Co(en)2Cl2]+

4+ H N Cl 2 H2N Co NH2

NH2 Co NH2

NH2 Cl

b-CD

NH2 Co

+ cis-[Co(en)2Cl2]+

H 2N

87

H 2N

NH2

15

H 2N H2N

NH2 Co NH2

NH2

Cl

4+

Cl H2N

b-CD

H2N Co NH2 H 2N

NH2

88 [2]-Rotaxane

The sequence for the formation of the [2]-rotaxane 88.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

16

Molecular recognition

+ a-CD

H2 N

NH2

H 2N

− a-CD

89

NH2

a-CD

90 [2]-Pseudorotaxane

NO2

NO2

NH2

− a-CD

91

O2N

HN

SO3Na

O2N

NO2

92 [2]-Pseudorotaxane

O2 N

NH

NO2

O2N NO2

O2 N

NO2 SO3Na NO2

HN NH

NO2

a-CD

93 O2 N

O2N 94 [2]-Rotaxane NO2

Figure 24

O2 N

NO2

HN

NH 2

a-CD

NO2

O2N

NO2

NO2 + a-CD

HN

SO3Na

O 2N

NO2

NO2 O2 N

O2N

SO3Na

O2N

O2N

NO2

The sequence for the formation of the [2]-rotaxane 94.

of , , , and  chiralities and the [2]-rotaxane isolated in 7% yield exists as four diastereomers as the βCD may either be oriented as shown in 87 or may possess the opposite orientation. The 1,10-diaminodecane, 1,14diaminodotetradecane, and α-CD analogs of 88 have also been prepared.60, 61 Organic end groups may also be employed as shown in Figure 24 for the formation of the [2]-rotaxane 94 in water.62 Here, the α-CD and E-4,4 -diaminostilbene 89 are in equilibrium with the [2]-pseudorotaxane, 90, and 2,4,6-trinitrobenzenesulfonate reacts with an axle amino group to form axle 91 which with α-CD is in equilibrium with the [2]-pseudorotaxane 92. The substitution of the second amino group in 91 then forms the axle 93 whose trinitrophenyl end groups are too large to allow threading of α-CD. A similar substitution in 92 prevents dethreading of α-CD in the [2]rotaxane, [(E)-4,4 -bis(2,4,6-trinitrophenylamino)stilbene][α-CD]-[rotaxane] 94 which was obtained in 10% yield in water:acetone 3 : 2 v/v solution. While the threading of a CD onto an axle followed by attachment of blocking groups is a common method for preparing rotaxanes, alternative approaches have been developed. An innovative method is the use of photocyclodimerization of 2-anthracene carboxyl substituents on the 6A carbons of α-CD in the presence of γ -CD in aqueous solution at 298.2 K as shown in Figure 25.63 The sequential preassembly of the substituents of two 6A -(2-anthracenecarbonyl)-6A -deoxy-α-CDs, 95, in a

head-to-tail orientation in the annulus of γ -CD in the 1 : 2 host–guest complex 97 is favored by the steric hindrance arising from the α-CD entities. The sequential complexation constant for 96 is K11 = 270 dm3 mol−1 and for 97 is K12 = 21 700 dm3 mol−1 where the latter species may be viewed as a [3]-pseudorotaxane with two axles. The large K12 magnitude as compared with that of K11 is attributable to a combination of π –π interactions between the anthracenyl substituents and the closer host–guest fit in 97. Irradiation of 97 at >320 nm causes photocyclodimerization to give a bulky axle which is prevented from dethreading by the steric hindrance of the α-CD end groups in the anti and syn-head-to-tail [2]-rotaxanes, 98 and 99, in 60 and 35% yields, respectively. Only 3 and 2%, respectively, of the anti - and syn-head-to-head axle, which cannot form [2]-rotaxanes are produced. Lengthening of the axle can lead to the threading of several CDs to form polypseudorotaxanes which become polyrotaxanes when bulky end groups are attached at either end of the axle.11, 12 An interesting example of such a polyrotaxane is shown in Figure 26 where the polyethyleneglycol-αCD polyrotaxane 100 is reacted with epoxide 101 in 10% NaOH in water to form two or three links between the primary and secondary hydroxyls of the ends of adjacent α-CDs to make a molecular tube in the [2]-rotaxane 102.64 Treatment with 25% NaOH in water hydrolyzes the dinitrophenyl end groups from the axle to release the molecular tube 103.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

17

C6A O O

g -CD

+

a-CD

95

a-CD

g -CD

K11

O C6A

O O

AO

C6

98 ant i-head-to-tail, 60%

a-CD

C6AO O a-CD

K12

+1

96

−1

h n > 320 nm

a-CD

O C6A

O

O C6A O

O C6A

99 syn -head-to-tail, 35%

a -CD

C6A O

a-CD

O

g-CD

O a-CD

97

Figure 25 Sequential complexation of 95 to form 96 and 97 and the photocyclodimerization of 95 to the anti - and syn-head-to-tail axles in the [2]-rotaxanes 98 and 99. O 2N

NH

O

O

O

O

O

O

O

O

O

O

O

O

NO2 100

OH

O 2N

NH NO2

O OH

O

102

OH

OH O

O

O

O

OH

O

O

OH

NO2

O 2N

Cl O 101 NaOH (10%)

OH

HN

OH O

O

O

O HN

OH

OH

NO2

O2N

NaOH (25%)

OH

OH

103

OH

OH

OH

OH

OH

OH

OH

OH

Figure 26 The reaction of the polyrotaxane 100 with epoxide 101 to form the molecular tube in the [2]-rotaxane 102 and release of the molecular tube 103.

6.2

Cyclodextrin catenanes

The formation of catenanes by CDs is less explored than the formation of rotaxanes. Part of an extensive study of the formation of several catenanes in aqueous solution is shown in Figure 27.65 When the bitolyl derivative diamine chain link precursor 104, in which n is either 3

or 4, threads heptakis(2,6-di-O-methyl)-β-CD, DMβ-CD, to form the [2]-pseudorotaxane 105 subsequent reaction with teraphthaloyldichloride, 106, gives the [2]- and [3]catenanes 107 and 108 in yields of 3.0 and 0.8% when n is 3 and 2.4 and 0.3% when n is 4, respectively. The isomeric catenanes 109 and 110 are obtained as a 40 : 60 isomeric mixture in 1.1% yield when n is 3, and as a 50 : 50

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

18

Molecular recognition

DMb-CD

O

O

O

O

NH2

n

O

O

O

O

NH2

105

Cl

n

C

C

O

O

106

O

NH

NH2

n

n

O

NH2

+

104

O

n

Cl

O

n

HN

C O

C O

NH

O

n

O

C O or

+ C O O 107

NH

O

n

O

O

HN

108

n

O C

O C

NH

O

O

109 and 110

n

Figure 27 The formation of the catenanes 107 and 108 and of 109 and 110 in which two DMβ-CDs are present in either the head-to-head or the head-to-tail configurations shown by the dashed DMβ-CDs (n = 3 or 4).

isomeric mixture in 0.4% yield when n is 4. The separate macrocycles threading DMβ-CD in 107 and 108 are also obtained.

7

CYCLODEXTRIN MOLECULAR DEVICES AND NANOMACHINES

A molecular device is a supramolecular assembly which can be stimulated to perform operations resembling those of everyday macroscopic devices such as shuttles, switches, and hinges. These are explored herein in a sequence of increasing complexity leading to large supramolecular assemblies which combine several of these abilities and which are sometimes called nanomachines and potentially have a range of practical applications. There is increasing interest in constructing such nanomachines on surfaces as is illustrated in the later parts of this section.

7.1

Cyclodextrin rotaxane-based devices

When the length of the axle in a [2]-rotaxane is significantly longer than that of the CD, the possibility that the CD

may be stimulated to shuttle along the axle arises. This is exemplified by the [2]-rotaxane shown in Figure 28 in which the redox properties of the tetrathiafulvalene unit allow control of the position of α-CD through a two-step electron transfer process.66 In 111, the α-CD complexes the hydrophobic and uncharged thiafulvalene unit in aqueous solution at 298.2 K. Upon addition of either D2 O2 or Fe(II), the thiafulvalene is oxidized to either its radical cation or dication and loses its hydrophobicity such that the α-CD shuttles to complex the 1,2,3-triazole unit of the axle as shown in 112 and 113 through UV–vis, circular dichroic, and 1 H NMR studies. Electrochemical studies show that the process is reversible. Shuttling is not always reversible as shown for the α-CD [2]-rotaxanes in Figure 29, where the tetracationic axle is composed of an azobenzene group linked at either side to a viologen group linked through a propylene group to a 2,4-dinitrobenzene end group.67 In the initially formed [2]-rotaxanes, 114 and 115, α-CD shuttles backward and forward on the azobenzene entity at temperatures up to 373.2 K in dimethylsulfoxide with a rate constant, k(363.2K) = 94 s−1 and G‡ = 80 kJ mol−1 . Upon heating above 373.2 K, α-CD shuttles irreversibly to positions over the propylene units as shown in 116. Alternatively when

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

HO2C O

O

S

O

CO2H

S

O

O

S a-CD S

HO2C

N N

O N

111 −

+e

CO2H

− e−

HO2C O

O HO2C

S + S

O

CO2H

S

O

O

S

112 + e−

N O N N a-CD CO2H

− e−

HO2C O

O HO2C

S + S

O

19

CO2H

S + S

O

O

113

N O N N a-CD CO2H

Figure 28 The sequence of reversible redox-driven shuttling of α-CD between tetrathiafulvalene and 1,2,3-triazole axle sites in response to changes in the axle oxidation state in the [2]-rotaxanes 111–113.

O2 N

NHa-CD +N

NO2

NO2

HN

N+ CH2

N N

CH2 +N

N+

CH2

+N

N+

CH2

+N

O2 N

116 > 373.2 K

NH

O2N

+N

NO2

N+ CH2

N N

HN

a-CD

114

NO2 O2 N

≤ 373.2 K

NH

O2N

a-CD +N

NO2

N+ CH2

115

NO2

HN N N

UV light

N+

O2 N

Visible light

N N O 2N

NO2 a-CD 117 O 2N

HN

NH +

N

+N

N+

N+

NO2

O2 N

Visible light

NH NO2 a-CD

+N

N+ CH2

HN N N

CH2 +N

N+

NO2

O2 N

116

Figure 29 Multimode molecular shuttling of α-CD in the [2]-rotaxanes 114–117. The alternative positions for α-CD in 116 and 117 are shown by the dashed outline of the α-CD annulus. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

20

Molecular recognition

114 and 115 are irradiated with UV light, the azobenzene unit photoisomerizes from the E to the Z isomer and α-CD shuttles to the propylene unit as shown in 117. Irradiation with visible light partially reverses the isomerization back

to 114 and 115 but the dominant effect is the production of 116. In contrast, no photoisomerization of 116 and 115 occurs in water and this is attributed to strong hydration of the cationic viologen units preventing passage of α-CD to

O

O −O

HO 118

119

O HO

120

O

O

O

−O

a-CD

a-CD

a-CD

O− 335 nm

Na2CO3

O

N

O O

N

O

280 nm

O

N

O

O

O− O

O

SO3Na SO3Na NaO3S

Figure 30

NaO3S

SO3Na NaO3S

NH2

NH2

NH2

The pH-dependent photoisomerization and shuttling processes of the [2]-rotaxanes 118–120.

CH3O

121

N N

O NH

N NH

CH3O

a-CD

a-CD

HN N

HN

HN

O O 350 nm

N CH3O

OCH3

254 nm

O NH

N a-CD

122

O

a-CD

N HN

O

HN

350 nm

CH3O

N HN

123

a-CD HN

NH

O

O

HN

N OCH3

6

E, Z 1

2

254 nm

cis, cis or Z, Z

NH

O

OCH3

N

OCH3

N

E, E 2

OCH3

HN

HN

trans, cis or E, Z

N

N N

O

N HN

trans, trans or E, E

OCH3

Muscle extension

Muscle contraction

Isomer ratio at photostationary states at: 350 nm & 254 nm

O

NH

a-CD

NH N

O

N

OCH3 N

Z, Z 1

0

CH3O

Figure 31 The extended, 121, half contracted, 122, and contracted, 123, states of the photochemically controlled Janus [2]-rotaxane molecular muscle. The relative proportions of the three states at the 350 and 254 nm photostationary states in D2 O are shown on the left. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology the alternative propylene unit and such that the photoisomerization is sterically hindered by α-CD remaining over the azobenzene entity. Similar shuttling processes and their solvent dependence have been reported.68, 69 Another example of photoisomerization of a double bond and solvent conditions controlling CD shuttling is represented by the [2]-rotaxane 118 in Figure 30 in which the axle contains a stilbene unit.70 UV–vis, fluorescence, and 1 H NMR studies show that in water no photoisomerization occurs consistent with hydrogen bonding between the α-CD hydroxy groups and the carboxylic acid substituents of the isophthalic acid unit of the axle increasing the barrier to photoisomerization. However, upon raising the pH to 10.5 with Na2 CO3 , a reversible photoisomerization of the stilbene unit between the E and Z forms in 119 and 120 occurs and drives the to-and-fro shuttling motion of αCD. A similar use of stilbene photoisomerization to control shuttling in has been reported by Anderson.71 The incorporation of two isomerization centers increases the sophistication of the molecular devices as shown in the system resembling a molecular muscle depicted in Figure 31.72 It consists of two interpenetrating [2]rotaxanes where one provides the axle for the C6A substituted α-CD of the other and vice versa in a hermaphroditic

N H3C A C6 124Z

O

− HO

+ HO

H3C N

125Z

125Z ′

126

126

ON

300 nm 254 nm

OFF

− HO DG ° = −4.2 (kJ mol−1) 2Z:2E = 5.6 : 1

O

A

O

C6 124E ′

H3C N

126

DG ° = −3.4 (kJ mol−1) 2Z ′:2E ′ = 4.2 : 1

dDG ° = 0.8 kJ mol−1

+ HO

126

H 3C N

O

A

C6 125E ′

C6A 125E b-CD

b-CD

b-CD OH DG ° = −11.3 (kJ mol−1) 1Z ′:1E ′ = 100 : 1

N

b-CD OH

C6A

Photochemical switch

H3 C O

O

H3C N C6A

b-CD

dDG ° = −8.8 kJ mol−1

H3C N

Cyclodextrin intra- and intermolecular devices and nanomachines

The incorporation of both intra- and intermolecular interactions into molecular devices presents an increased range of design possibilities. Thus, both amide and alkene isomerizations and intermolecular host–guest complexation of 1-adamantol are exploited in the operation of the molecular device based on E-6A -deoxy-6A -(N-methylcinnamido)β-CD shown in Figure 32.73 The thermal E to Z isomerization about the amide bond controls the equilibrium between isomeric 124E and 124Z in which 1-adamantol,

C6 124Z ′

DG ° = −2.5 (kJ mol−1) 1Z:1E = 2.6 : 1

C6 124E

7.2

A

b-CD OH

A

or Janus complex. Each [2]-rotaxane incorporates a stilbene unit such that when both are in the E isomeric form the molecular muscle is in its extended form, 121, when one is in the Z form the muscle is partially contracted, 122, and when both are in the Z form, 123, the muscle is fully contracted. For the photostationary state under 350 nm radiation the ratio 121 : 122 : 123 is 2 : 2 : 1 in D2 O, whereas at the 254 nm stationary state the ratio 121 : 122 : 123 is 6 : 1 : 0. Thus, coupling of the two photoisomerizations increases the transformational control achievable in a molecular assembly.

O

H3C N

O

21

b-CD OH

b-CD

Figure 32 The 124E , 124Z , 124E
, and 124Z
equilibrium system in which intermolecular host–guest complexation of 1-adamantol competes with cinnamate self-complexation in 124Z
and E to Z isomerization about the amide bond occurs while the E stereochemistry about the cinnamate alkene bond is retained. Photoisomerization by irradiating at 300 nm switches the stereochemistry about the cinnamate alkene bond to Z to produce the 125E , 125Z , 125E
, and 125Z
system in which no competitive cinnamate self-complexation occurs. The cinnamate alkene photoisomerization is reversed by irradiation at 254 nm to complete an on–off photoswitching process. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

22

Molecular recognition

126, occupies the β-CD annuli of both in water. An equilibrium exists between 124E and 1-adamantol and 124E
in which the cinnamate group is unable to enter the vacant β-CD annulus. A similar equilibrium exists between 124Z and 1-adamantol and 124Z
in which the cinnamate group now self-complexes in the β-CD annulus because of the more favorable Z stereochemistry about the amide bond. Both of the last two equilibria may be driven to the right by extraction of 1-adamantol with hexane. Photoisomerization about the cinnamate alkene bond through irradiation at 300 nm converts 124E , 124Z , 124E
, and 124Z
to 125E , 125Z , 125E
, and 125Z
, respectively. Although both 125E and 125Z form intermolecular host–guest complexes with 1-adamantol, the cinnamate self-complexes in neither 125E
nor 125Z
because of the unfavorable Z stereochemistry about the cinnamate alkene bond. Irradiation at 254 nm photoisomerizes the stereochemistry about the alkene bond from Z to E and completes the photochemical switch between the two systems. The differences in the free energy changes, δG◦ = −11.3 + 2.5 = −8.8 kJ mol−1 and δG◦ = −3.4 + 4.2 = 0.8 kJ mol−1 , associated with the host–guest complexation of 1-adamantol in the first and second systems, respectively, are directly related to the different constraining effects of

HO

CH2 HO

CH O HO

CH CH2 CH CH2

HN

CH2

O HO

O

C6A

CH

O HO

UV-light viscosity increases

127 CH2

O 128

HO

O CH2

HO

O HN (CH2)12 NH

N N HN O

O HO

C6A

(a)

O HO

CH2

CH2

CH2

HO

129

CH2 CH

O HO

O

HN C3A

CH2 CH

O HO

O

CH

UV-light viscosity decreases

O 130

Visible-light viscosity increases

CH2 CH

O 128

CH2

HO

HO

CH O HO

b-CD

CH O CH2

N N

HN (CH2)12

O NH

CH2

CH2 CH

O HN C3A

O HO

CH2 CH

CH HO

CH2 CH

CH2

CH2 CH

O 127

CH2

b-CD

HN (CH2)12

N N

O NH

CH

CH

CH O

CH2

CH2

CH b-CD O

CH CH2

O

CH2

CH

CH2 CH

CH

CH2 CH

O HO

CH

Visible-light viscosity decreases

CH O HO

CH2

CH2

HO

CH

CH

O

O

O NH

CH2

HO

CH

b-CD

CH

HO

HN (CH2)12

N N

O

HO

CH2

the E and Z stereo chemistries about the cinnamide alkene bond between phenyl group and the amide function. The formation of hydrogels through CDs attached to a polymer backbone complexing hydrophobic substituents on a second polymer backbone to form cross-links between polymer strands in water is attracting considerable attention.74 Innovative examples of these macroscopic assemblies arise where the photoisomerization of an azobenzenebased substituent enables the viscosity of an aqueous polymer solution to be varied at will as shown in Figure 33.75 Example Figure 33(a) is based on a polyacrylate with 2.2% of the carboxylic acid groups randomly substituted with β-CD attached through the C6A carbon, 127, which complexes the E-azobenzene substituent of a second polyacrylate with 2.7% of the carboxylic acid groups randomly substituted with azobenzene attached through a dodecyl tether, 128, to form a viscous solution consistent with the complexation of 128 by 127 being characterized by K = 1.2 × 104 dm3 mol−1 . On irradiation with UV light, the E-azobenzene photoisomerizes to Z-azobenzene, 129, and the viscosity of the solution increases from 2.5 × 102 Pa s−1 two-fold consistent with the “locking” shown for the cross-link formed with 129.

O

O HO

CH

CH O (b) 130

CH2

O 129

Figure 33 The opposite effects of irradiation with UV- and visible light in systems (a) and (b) are attributed to the E to Z and vice versa photoisomerization of the azobenzene substituents of polymers 128 and 129 impacting on the mode of complexation by the β-CD substituents of polymers 127 and 130. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology The reverse situation applies when the same experiment is carried out on polymer 130 where the 1.6% substituted polyacrylate has β-CD attached through the C3A carbon as shown in Figure 33(b). Under the same conditions, the solution of polymers 130 and 128 is much less viscous at 6.5 × 10−1 Pa s−1 because of the weaker complexation of the E-azobenzene substituent by the β-CD substituent as indicated by a lower K = 1.4 × 102 dm3 mol−1 . Upon photoisomerization to form the Z-azobenzene substituent, the viscosity drops by an order of magnitude consistent with this substituent of 129 complexing even less effectively in the β-CD substituents of 130. (It should be noted that the viscosity of 128 alone under the same conditions is 8.4 × 10−2 Pa s−1 ). For both systems (Figure 33a and b) the photocontrolled viscosity switching is repetitively reversible. Thus, both systems may be viewed as nanomachines which demonstrate the impact of variations in CD complexation and azobenzene substituent photodimerization in controlling the characteristics of macroscopic systems. Another type of macroscopic assembly formed by highly modified CDs is represented by CD substituted with long hydrophobic substituents.76–78 This is exemplified in Figure 34 where the highly substituted β-CD 131 forms a spherical bilayer vesicle through hydrophobic interactions between the long hydrophobic substituents in water at pH 7.4.78 This resembles the lipid bilayer membranes of biological cells. A further resemblance occurs through Vesicle interior

the complexation of the adamantyl group of the octapeptide 132 in the β-CD annuli on the vesicle exterior which mimics the selective docking sites on mammalian cell exteriors. The K11 for a single 131 on the exterior of the vesicle complexing 132 is 3.5 × 104 dm3 mol−1 in water, and 132 adopts a random coil configuration when complexed. Upon changing the pH to 5.0, the vesicles change to nanotubes and 131 forms a β-sheet on the nanotube external surface. This transformation is reversed by changing the pH back to 7.4. When the vesicles are formed in the presence of tetrasodium 1,3,6,8-pyrenetetrasulfonate at pH 7.4 this dye is encapsulated and upon changing the pH to 5.0 it is released.

7.3

Cyclodextrin surface mounted nanomachines

Increasingly, CD-based molecular devices are being attached to surfaces to both increase their versatility and to extend their physical dimensions to the extent that they are often referred to as nanomachines. The first example shown in Figure 35 is based on the silica MCM-41 nanoparticle to which is attached an array of azobenzene derivatives in the E form, or “stalks,” adjacent to the many pores which characterize the nanoparticle as represented for a single pore by 134.79 In water, the dye Rhodamine B, 133, freely exchanges in and out of the pore. However, when β-CD subsequently complex the stalks egress of Rhodamine B Vesicle exterior

CD bilayer

R7

C2A

(H25C12S)7 C6A

N H C6A (SC12H25)7

R7

A

131

(H25C12S)7 C6A

C2

C6 (SC H ) 12 25 7

131 R7

C2A

131

CO2H

CO2H

C2A R7

N H

C6A (SC H ) 12 25 7

131

CO2H

C2A R7

O CO2H

132

O NH

NH2

O CO2H

3

132

O

O NH

NH2

O

3

N H

O

O

131

NH

O NH

(H25C12S)7 C6A b-CD

N H

O

O N H

A

NH

C2A R7

b-CD

O

O

O

b-CD

131

23

N H 3

NH

NH2

O CO2H

132

(R = −(CH2CH2O)nH, n = 1 = 1– –3)

Figure 34 The vesicle bilayer formed by the highly modified β-CD 131 in water at pH 7.4 and the complexation of the adamantyl groups of the peptides, 132, by 131 on the external surface of the vesicle. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

24

Molecular recognition

Step 2: Cap pore entrance

N

N

b-CD

O

NH

HN

Azobenzene derivatives anchored to nanoparticle surface

Si O O O

Step 1: Load pore

N

N

N

O

O

HN

N

O NH

Si O O O

Si O O

N

HN

NH

HN

NH

O

N

O CO2H

N

Si O O

O N

Pore in MCM-41 silca nanoparticle

N O

CO2H HO2C N

O

133

N

N 135

134 N

N

N

N

N

CO2H O O HN

O

HN

HN

NH

N

Si O O

O

N

O HO2C

O

Si O O N

Step 3: 351 nm irradiation b-CD

Azobenzene E to Z photoisomerization

+2

b-CD dethreads Rhodamine B released

136

Figure 35 Scheme for the entrapment of Rhodamine B, 133, in a pore of a MCM-41 nanoparticle, 134, through capping the pore by complexation of β-CD on azobenzene derivatives attached to the nanoparticle surface, 135 (Steps 1 and 2). The Rhodamine B is subsequently released by uncapping the pore through photoisomerization of the azobenzene unit from the E to Z form and decomplexation of β-CD from 136 (Step 3).

from the pore of 135 is precluded and the system is effectively blocked. Irradiation at 351 nm causes the azobenzene derivative stalks to photoisomerize to the Z form and the β-CD to decomplex in response so that the pore is now opened and Rhodamine B exits from the pore 136 as shown in Figure 35. It is envisaged that such modified nanoparticles might act as photochemically controlled drug

delivery systems when Rhodamine B is replaced by a drug molecule. A second example of a nanomachine involves the attachment of multiple C6 sulfur substituent modified β-CD to a gold surface as shown in 138 in Figure 36.80 When the gold surface is coated in this manner, it becomes a surface on which molecular printing is achieved by selectively

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

Fe+

Fe H N

Fe+ Fe Fe

O

O

Fe+

NH O Fe

HN

NH

N

N N

N

O

O

O HN

HN

NH HN O

O

N O

N

N

Fe+

N

N N

OH N

O

O

Fe

NH

O NH

N

NH

25

O

Fe+ Fe

Fe

Fe

Fe

NH

+

OHN Fe+ Fe+

139

O N

N

NH

Fe Oxidation

− b-CD

137

Fe

NH

NH

HN

N H

O

O O

N HO

Fe

Fe

N

N

N

O

C6A

= b -CD

HN

HN O

HN O

Fe

Fe Attachment to b-CD through C6A and to the Au surface through S

=

Reduction

NH Fe

O

NH

O

N

N

+ b -CD

Fe

138

= Au surface S

Figure 36 Scheme for attachment of a ferrocenyl dendrimer, 137, to a gold surface functionalized with β-CD, 138, and subsequent release upon oxidation of the ferrocenyl groups to ferrocinium groups, 139, which do not complex significantly in β-CD.

complexing guest molecules in the β-CD annuli.13 Thus, the water-soluble eight ferrocenyl terminated dendrimer/βCD assembly 137 docks, or prints, on the modified gold surface to form a stable surface assembly 138. However, when the ferrocenyl groups are electrochemically oxidized to ferrocium groups in 139 their positive charge greatly decreases the stability of their β-CD complexation such that the multicharged dendrimer is released from the surface. Recently, the versatility of the use of β-CD modified gold surfaces has been demonstrated by the assembly of a printboard for antibody recognition and lymphocyte cell counting.81 Clearly, surface modifying nanomachines hold great promise for practical application.

8

CONCLUSION

This brief examination of CD chemistry commenced with the naturally occurring CDs and proceeded through an increasingly sophisticated sequence of supramolecular assemblies, some of which resemble metalloproteins,

biological cells, and muscle components. Others have no counterpart in nature, but exhibit sophisticated characteristics in their own right. This is exemplified by rotaxanes, polymers, and light activated molecular devices and hydrogels. Increasingly, CDs are being attached to surfaces to generate versatile behaviors as shown by controlled guest release from silica nanospheres and molecular printboards. The great breadth of this supramolecular array is underpinned by synthetic chemistry which has been refined to the stage where almost any desired CD modification is achievable. In conjunction with the current powerful understanding of supramolecular chemistry, increases in the sophistication of CD supramolecular assemblies are likely to be limited only by imagination. Exactly where this will lead is not readily predictable, but it is likely that “smart” polymer, molecular device, and surface chemistry will be much to the fore. CDs are widely deployed in the agrochemical, cosmetic, food, and pharmaceutical industries and it seems inevitable that these and other uses will grow in both extent and sophistication as CD supramolecular chemistry progresses.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

26

Molecular recognition 30. K. B. Lipkowitz, S. Raghothama, and J.-A. Yang, J. Am. Chem. Soc.; 1992, 114, 1554.

REFERENCES 1. J. Szejtli, Chem. Rev., 1998, 98, 1743.

31. K. B. Lipkowitz, Chem. Rev., 1998, 98, 1829.

2. C. J. Easton and S. F. Lincoln, Modified Cyclodextrins: Scaffolds and Templates for Supramolecular Chemistry, Imperial College Press, London, 1999.

32. M. V. Rekharsky and Y. Inoue, J. Am. Chem. Soc., 2000, 122, 4418.

3. W. Saenger, J. Jacob, K. Gessler, et al., Chem. Rev., 1998, 98, 1787.

33. K. Kano, H. Kamo, S. Negi, et al., J. Chem. Soc., Perkin Trans. 2., 1999, 15.

4. T. Endo and H. Ueda, FABAD J. Pharm. Sci., 2004, 29, 27.

34. M. V. Rekharsky and Y. Inoue, J. Am. Chem. Soc., 2002, 124, 813.

5. H. Taira, H. Nagase, T. Endo, and H. Ueda, J. Inclusion. Phenom. Macrocyclic. Chem., 2006, 56, 23.

35. K. Takahashi, H. Narita, M. Oh-Hashi, et al., J. Inclusion. Phenom. Macrocyclic Chem., 2004, 50, 121.

6. M. V. Rekharsky and Y. Inoue, Chem. Rev., 1998, 98, 1875. 7. K. Harata, Chem. Rev., 1998, 98, 1803.

36. K. Kano and H. Hasegawa, J. Am. Chem. Soc., 2001, 123, 10616.

8. H.-J. Schneider, F. Hacket, and V. R¨udiger, Chem. Rev., 1998, 98, 1755.

37. D.-T. Pham, P. Clements, C. J. Easton, and S. F. Lincoln, Tetrahedron Assym., 2008, 19, 167.

9. A. R. Khan, P. Forgo, K. J. Stine, and V. T. D’Souza, Chem. Rev., 1998, 98, 1977.

38. M. Miyauchi, Y. Takashima, H. Yamaguchi, and A. Harada, J. Am. Chem. Soc., 2005, 127, 2984.

10. Y. Liu and Y. Chen, Acc. Chem. Res., 2006, 39, 681.

39. R. Breslow and S. D. Dong, Chem. Rev., 1998, 98, 1997.

11. G. Wenz, B.-H. Han, and A. M¨uller, Chem. Rev., 2006, 106, 782.

40. R. L. van Etten, J. F. Sebastian, G. A. Clowes, M. L. Bender, J. Am. Chem. Soc., 1967, 89, 3242.

and

12. A. Harada, A. Hashidzume, H. Yamaguchi, Y. Takashima, Chem. Rev., 2009, 109, 5974.

and

41. R. L. van Etten, G. A. Clowes, J. F. Sebastian, M. L. Bender, J. Am. Chem. Soc., 1967, 89, 3253.

and

13. O. Crespo-Biel, B. Jan Ravoo, J. Huskens, D. N. Reinhoudt, Dalton Trans., 2006, 2737.

and

42. S. Hu, J. Li, J. Xiang, et al., J. Am. Chem. Soc., 2010, 132, 7216.

14. J. Liu, W. Ong, E. Rom´an, et al., Langmuir, 2000, 16, 3000.

43. L. G. Marinescu and M. Bols, Angew. Chem. Int. Ed., 2006, 45, 4590.

15. T. Murakami, K. Harata, and S. Morimoto, Tetrahedron Lett., 1987, 28, 321. 16. C. Rousseau, B. Christensen, T. Petersen, and M. Bols, Org. Biomol. Chem.., 2004, 2, 3476. 17. C. J. Easton, S. J. van Eyk, S. F. Lincoln, et al., Aust. J. Chem., 1997, 50, 9. 18. X. Guo, A. A. Abdala, B. L. May, et al., Macromolecules, 2005, 38, 3037. 19. G. Crini, G. Torri, M. Guerrini, et al., Eur. Polym. J., 1997, 33, 1143. 20. Y.-Y. Liu, X.-D. Fan, and L. Gao, Macromol. Biosci., 2003, 3, 715. 21. F. van de Manakker, T. Vermonden, C. F. van Nostrum, and W. E. Hennink, Biomacromolecules, 2009, 10, 3157.

44. L. Marinescu, M. Mølbach, C. Rousseau, and M. Bols, J. Am. Chem. Soc., 2005, 127, 17578. 45. M. T. Reetz and S. R. Waldvogel, Angew. Chem. Int. Ed., 1997, 36, 865. 46. J. M. Haider and Z. Pikramenou, Chem. Soc. Rev., 2005, 34, 120. 47. F. Bellia, D. La Mendola, C. Pedone, et al., Chem. Soc. Rev., 2009, 38, 2756. 48. S. E. Brown, J. H. Coates, C. J. Easton, and S. F. Lincoln, J. Chem. Soc., Faraday Trans., 1994, 90, 739. 49. L. Barr, C. J. Easton, K. Lee, et al., Tetrahedron Lett., 2002, 43, 7797. 50. D. H. Kim and S. S. Lee, Biorg. Med. Chem., 2000, 8, 647.

22. C. J. Easton and S. F. Lincoln, Chem. Soc. Rev., 1996, 25, 163.

51. B. Zhang and R. Breslow, J. Am. Chem. Soc., 1997, 119, 1676; 1998, 120, 5854.

23. R. Bhushan and S. Joshi, Biomed. Chromatogr., 1993, 7, 235.

52. K. Kano, Y. Itoh, H. Kitagishi, et al., J. Am. Chem. Soc., 2008, 130, 8006.

24. S. Li and W. C. Purdy, Chem. Rev., 1992, 92, 1457.

53. H. Kitagishi, S. Negi, A. Kiriyama, et al., Angew. Chem. Int. Ed., 2010, 49, 1312.

25. V. Schurig and H.-P. Nowotny, Angew. Chem. Int. Ed., 1990, 29, 939. 26. H. Nishi and S. Terabe, J. Chromatogr., 1995, 694, 245.

54. J. M. Haider, R. M. Williams, L. De Cola, and Z. Pikramenou, Angew. Chem. Int. Ed., 2003, 42, 1830.

27. J. A. Hamilton and L. Chen, J. Am. Chem. Soc., 1988, 110, 5833.

55. A McNally, R. J. Forster, N. R. Russell, and T. E. Keyes, Dalton Trans., 2006, 1729.

28. D. W. Armstrong, T. J. Ward, R. D. Armstrong, T. E. Beesley, Science, 1986, 232, 1132.

and

56. F. Hapiot, S. Tilloy, and E. Monflier, Chem. Rev., 2006, 106, 767.

29. D. W. Armstrong, X. Yang, S. M. Han, and R. A. Menges, Anal. Chem., 1987, 59, 2594.

57. Y. Han, K. Cheng, K. A. Simon, et al., J. Am. Chem. Soc., 2006, 128, 13913.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cyclodextrins: from nature to nanotechnology

27

58. S. A. Nepogodiev and J. F. Stoddart, Chem. Rev., 1998, 98, 1959.

71. C. A. Stanier, S. J. Alderman, T. D. Claridge, and H. L. Anderson, Angew. Chem. Int. Ed., 2002, 41, 1769.

59. H. Ogino, J. Am. Chem. Soc., 1981, 103, 1303. 60. H. Ogino and K. Ohata, Inorg. Chem., 1984, 23, 3312.

72. R. E. Dawson, S. F. Lincoln, and C. J. Easton, Chem. Commun., 2008, 3980.

61. K. Takaizumi, T. Wakabayashi, and H. Ogino, J. Sol. Chem., 1996, 25, 947.

73. R. J. Coulston, H. Onagi, S. F. Lincoln, and C. J. Easton, J. Am. Chem. Soc., 2006, 128, 14750.

62. C. J. Easton, S. F. Lincoln, A. G. Meyer, and H. Onagi, J. Chem. Soc., Perkins Trans. 1, 1999, 2501.

74. L. Li, X. Guo, L. Fu, et al., Langmuir, 2008, 24, 8290.

63. C. Yang, T. Mori, Y. Origane, et al., J. Am. Chem. Soc., 2008, 130, 8574.

75. I. Tomatsu, A. Hashidzume, and A. Harada, J. Am. Chem. Soc., 2006, 128, 2226.

64. A. Harada, J. Li, and M. Kamachi, Nature, 1993, 364, 516.

76. B. J. Ravoo and R. Darcy, Angew. Chem. Int. Ed., 2000, 39, 4324.

65. D. Armspach, P. R. Ashton, R. Ballardini, et al., Chem. Eur. J., 1995, 1, 33.

77. R. Auz´ely-Veltry, F. Djeda¨ıni-Pilard, S. D´esert, et al., Langmuir, 2000, 16, 3727.

66. Y. L Zhao, W. R. Dichtel, A. Trabolsi, et al., J. Am. Chem. Soc., 2008, 130, 11294.

78. F. Versluis, I. Tomatsu, S. Kehr, et al., J. Am. Chem. Soc., 2009, 131, 13186.

67. H. Murakami, A. Kawabuchi, R. Matsumoto, et al., J. Am. Chem. Soc., 2005, 127, 15891.

79. D. P. Ferris, Y.-L. Zhao, N. M. Khashab, et al., J. Am. Chem. Soc., 2009, 131, 1686.

68. Y. Kawaguchi and A. Harada, Org. Lett., 2000, 2, 1353.

80. C. A. Nijhuis, J. Huskens, and D. N. Reinhoudt, J. Am. Chem. Soc., 2004, 126, 12267.

69. Y. Kawaguchi and A. Harada, J. Am. Chem. Soc., 2000, 122, 3797. 70. Q-C. Wang, D.-H. Qu, J. Ren, et al., Angew. Chem. Int. Ed., 2004, 43, 2661.

81. M. J. W. Ludden, X. Li, J. Greve, et al., J. Am. Chem. Soc., 2008, 130, 6964.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc055

Cucurbituril Receptors and Drug Delivery Anthony Ivan Day and J. Grant Collins UNSW@ADFA, Canberra, ACT, Australia

1 Introduction 2 Cucurbituril as a Drug Delivery Vehicle 3 Drug Types that can Benefit from Cucurbituril Encapsulation 4 Preparation of Association Complexes Drug@Cucurbituril 5 Drug Release Mechanisms 6 Conclusion References

1

1 2 10 12 14 16 16

INTRODUCTION

Cucurbit[n]uril (Figure 1) is a relatively new family of macrocyclic cagelike molecules with a broad range of potential applications including drug delivery, the prime focus of the following discussion.1 Cucurbit[n]uril is abbreviated in this chapter as Q[n] but is also often abbreviated as CB[n] in cited literature. The Q[n] possess a cavity and are shaped like open-ended empty barrels (Figure 2). Chemically, these macrocycles behave as molecular hosts and a variety of smaller molecules can be encapsulated within their cavities. Physically, the Q[n] are a macrocyclic framework of repeating five and eight membered rings of principally C–N bonds forming a relatively rigid and chemically robust structure. The portals are rimmed by carbonyl oxygens, which give them their electronegative property. The factors that favor encapsulation are the hydrophobic nature of the cavity, van der Waals contact forces, cation

attraction (ion–dipole interaction), and/or hydrogen bonding at the electronegative openings or portals to the cavity. The relative dimensions of guest molecules compared to the portal openings can be a determining factor (see Section 3), but a range of cavity sizes has been synthesized.2–5 The host–guest nature of Q[n] and its relatively rigid framework are two aspects of this macrocycle that facilitates its exploitation as a drug delivery vehicle. The Q[n] are often compared to the cyclodextrins (CD) in terms of cavity sizes, shape, electropotential surfaces, and hydrophilic or hydrophobic characteristics.1, 6–8 CDs have had a history in drug formulations since 1953 and after a relatively slow beginning have eventually found reasonably broad application.9–13 In comparison, the first connection for Q[n] to drug delivery was only made a decade ago in patents as a potential application.14, 15 The host–guestbinding characteristics of both CD and Q[n], which appear to be similar are in reality significantly different, especially in binding strength, driving forces to binding, compatibility to cations, and hydrophobic differences.7, 8, 16 The Q[n] generally have higher binding constants, a much higher binding ratio, a high affinity for cations, and hydrophobic groups, and the cavity of the Q[n] is accessible from two equal sized portals.1 In addition, the Q[n] are structurally much more robust molecules. Since the discovery of the simplest Q[n] family (Figure 1a), some efforts have also been directed toward expanding their utility through the introduction of substitution (Figure 1b). Substitution of Q[n] is possible at the two carbons of the junction of the cis-fused imidazolone rings of the glycoluril moieties. There are two main approaches to this, either by direct reaction on these carbons for a specific Q[n] or by the introduction of a substituted glycoluril during the synthetic process. Direct reaction has been achieved through oxidation of the methine carbon to give perhydroxylated Q[n] (R1 = R2 = OH), which is most successful

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

2

Molecular recognition cucurbituril analogs, and an example of a benzofuran substituent introduced as a linking group between two adjacent glycoluril N groups has been reported.29, 30

O N

N CH2

N

N CH2

R1

R2

O

2 n

Figure 1 Structural representation of cucurbit[n]uril, Q[n] where n = 5–10, (a) R1 = R2 = H, (b) R1 = R2 = substituent, or R1 or R2 are different substituent groups.

O N

O N

R1

N R2

N

CUCURBITURIL AS A DRUG DELIVERY VEHICLE

N

H

N

H N

O

CH2

N

CH2

O

s

u

Figure 3 Structural representation of partially substituted Q[n], s = substituted glycoluril moiety, u = regular glycoluril moiety, and n = s + u. The number of s units in a substituted Q[n] is variable and their position relative to each other is variable.

where n = 5–6.17, 18 These are further elaborated through a variety of alkyl derivatives of the OH groups.17 The synthetic approach of introducing a substituted glycoluril has to date, been most successfully developed for partially substituted Q[n] where only some of the introduced glycoluril moieties carry substitution (Figure 3).18–25 This type of substitution has been possible for Q[5] and Q[6] and is likely to be extended to the higher homologs. Fully substituted Q[n] (Figure 1b) are limited to the smaller homolog Q[5] or very low yields of fully substituted Q[6].21, 27, 28 Alternative methods to the introduction of substitution include

2.1

Toxicology and pharmacokinetics

As a drug delivery vehicle, the Q[n] need to be relatively harmless biologically or at least have low toxicity. Preliminary toxicology and pharmacokinetics studies have shown the Q[n] to fit the category of relatively low toxicity or nontoxic depending on the method of entry to the body. Toxicology has been evaluated for acute effects in mice orally and intravenously and in cell cultures of both animal and human cell lines.31–33 A single dose feed of Q[7] and Q[8] as a mixture in equal proportions has demonstrated no toxicity up to 600 mg kg−1 . The lack of toxicity is consistent with a separate study where only 3.6% of the total Q[n] was adsorb into the blood stream from the alimentary canal as measured using 14 C labeled Q[7] or Q[8].31 The evaluation of intravenous administration was limited to Q[7] as this Q[n] has sufficient aqueous solubility to reach a practical limit. Acute toxicity was found to be >250 mg kg−1 , which is relatively low. A maximum tolerated dose was used as a measure of toxicity, where an animal’s weight loss was observed to not fall below 10% following dosing. All animals begin to gain weight within five to eight days after dosing, indicating no extended effects. Intravenous administration studies with 14 C labeled Q[7] have shown that clearance into the urine has a mean half-life of clearance at 12.8 h. Q[7] does not cross the blood brain barrier, and accumulation in the liver and spleen is very low relative to the kidneys. Q[7]’s chemical and thermal stability and its relatively quick clearance into the urine following a higher activity in the kidneys suggest that Q[7] is excreted without modification.

Q[5]

Q[6]

Q[7]

Q[8]

Q[10]

4.4

5.8

7.3

8.8

∼11.6 9.1 Å

2.4

Figure 2

3.9

5.4

6.9

∼10 Å

Barrel representation Q[5–10] with the dimensions of the portal, the cavity, and the depth as indicated.1

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

Cucurbituril receptors and drug delivery The relatively low toxicity of Q[7] in the intravenous trials has also been supported by in vitro cell culture studies. Q[7] and Q[8] have been shown to enter muscle mouse embryo cells,34 and Q[7] is internalized by murine macrophage (RAW264.7).33 Cells of human kidney (HEK293), human hepatocyte (HepG2), and murine macrophage (RAW264.7) were found to tolerate Q[5] and Q[7] up to 1 mM without significant effects,33 while Q[7] resulted in an IC50 at 0.53 mM over a 48 h period (1 mM could be tolerated for 3 h without effect) for Chinese hamster ovary (CHO-K1) cells.32 Metabolic function tests of CHO-K1 over 48 h established no cytotoxic activity up to 0.5 mM (∼600 mg of Q[7] kg−1 of cells). The low aqueous solubility of Q[8] 20 µM precluded the possibility of a toxic limit being established.32 A number of Q[n] derivatives have been synthesized in order to modify Q[n] solubility, manipulate molecular guest-binding properties, and provide functionality for further structural developments.5, 6 Most of these derivatives have not been evaluated for biological compatibility, although some have been suggested as drug delivery vehicles.6 There is an immerging potential for structural variants of Q[n] from small changes of substituents and functional groups to polymeric forms including nanocapsules.19–37 As each of these find a place in drug delivery, their individual toxicology and pharmacokinetics will require evaluation.

2.2

Attributes of Q[n] as a delivery vehicle

The Q[n] in its simplest form or with substituents and functional groups have potential as an aid to drug delivery by providing one or more of the following benefits: • • • • •

increasing bioavailability, providing bioprotection, improve chemical stability, provide a method for slow release, and facilitate drug targeting.

The unique barrel-shaped cavity of Q[n] with its two openings that are slightly smaller in diameter than the internal cavity provides a site for drugs or parts of drugs to be bound or encapsulated without chemical modification. Supramolecular forces act on the drugs to hold them in place; the physical nature of the Q[n] molecular cage screens hydrophobic parts of drugs and protects sensitive functionality. The Q[n] are predicted to become another valuable tool in the box for the facilitation of drug efficiency and efficacy. A number of drugs have been evaluated for Q[n] encapsulation with a variety of structural variations (Figure 4). Significant binding features and biological findings are discussed throughout the following sections.

3

2.2.1 Bioavailability One of the main factors governing bioavailability of a drug is its aqueous solubility and the ultimate requirement for the drug to be above required minimum plasma concentrations to be effective, whether the drug is administered orally or intravenously. The increasing challenges for drug applications today are the prevalence of drugs being developed with poor aqueous solubility.38–40 The Q[6–8] have been shown to facilitate the solubility of poorly soluble cytotoxic drugs such as camptothecin (CPT) and the benzimidazoles—albendazole (ABZ) and MEABZ.41–44 The increase in solubility for ABZ and MEABZ was up to 2000- and 3000-fold, respectively.41, 44 It has also been found that the solubility of ABZ can be increased to an even greater extent (2400-fold) with the partially substituted Q[n], α,δ-(CH3 )4 Q[6] (Figure 3, R1 = R2 = CH3 , the two s units are separated by two u units).44 This indicates that Q[n] assisted drug solubility has potential yet to be explored as new substituted Q[n] are developed. It should be noted that to increase the solubility of a drug using Q[n], it is not always necessary for the whole or a major part of the drug to be cavity encapsulation. The ABZ@Q[6] (1900-fold increase) and ABZ@α,δ(CH3 )Q[6] only show encapsulation of the methyl carbamate group by 1 H NMR, but the close portal association of the remaining hydrophobic portion of the molecule is apparently sufficient to allow a suitable polarization of the overall complex or perhaps disrupting of π –π stacking and hence increased solubility. In general, solubility is achieved by encapsulating the hydrophobic parts of the drug molecule and creating a polar complex of drug@Q[n]. In some cases, the two individual components have considerably lower aqueous solubility when compared to the final complex.44 The low aqueous solubility of Q[6]45 would appear to be an impediment to an intravenous administration especially once the drug has been released from Q[6], but Q[6] is highly soluble in physiological saline due to its affinity for Na+ ions. Some of the known substituted Q[6] have significant natural water solubility.46 However, Q[8] has very low aqueous solubility, which is not improved significantly by the presence of metal ions such as Na+ .46 While Q[8] may not be suitable for intravenous use, there should be no impediment to oral applications. Examples of relatively high solubility of drug@Q[8] complexes are known; yet, the Q[8] alone has low solubility. Future derivatives of Q[8] are bound to change its solubility status.

2.2.2 Bioprotection Bioprotection for drugs primarily refers to maintaining the maximum effectiveness of a drug by protecting it from unintended chemical reactions within the body. The

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

4

Molecular recognition

O

+

NH2 N H

N

S

OH O

N OR

O

N O

Albendazole (ABZ) R = CH3 MEABZ R = (CH2)2OCH3

Camptothecin lactone form (CPT)

O

OH

O OH OH

H N

O

O OCH3 O

HN

OH

O

H O

F

5-Fluorouracil

Doxorubicin

OH NH2 O

O

NHNH2

OCH3 O

O S NH O HN

N H N Isoniazid O N

Glibenclamide

Cl H 2N

NH2

O

O

N

OH

Pyrazinamide

N H

Atenolol

NH2

NO2 N

Memantine

Figure 4

S

N H

O

N H

Ranitidine

Drugs discussed in the text.

objective is to maximize the opportunity for the drug to complete its intended task with a sustained and desired concentration for the time required. In contrast, there is a natural detoxification by the body of drugs through direct excretion or chemical modification to aid excretion. As a consequence, a balance must be found between the pharmacokinetics and the pharmacological objective. Drugs encapsulated in Q[n] have shown some promise in this regard especially in the area of cytotoxic dinuclear platinum drugs such as CT008, CT033, and CT233 derived from alkyl polyamines (Figure 4). Platinum-based cyctotoxic drugs are highly reactive to thiol-containing plasma proteins and are degraded to nonactive metabolites.47, 48 As a result, most of the platinum drugs are deactivated on administration. Studies have demonstrated protection of the

Table 1 Half-life (t1/2 ), in minutes, of the reaction of CT008, CT033, and CT233 with cysteine at 37 ◦ C.

Free Encapsulated in Q[7] Encapsulated in Q[8]

t1/2 CT033 (min)

t1/2 CT008 (min)

5 15 10

110 1000 1700

t1/2 CT233 (min) 5 40 45

The t1/2 is defined as the time taken for the free intact platinum complex to reduce in concentration to 50%.

platinum metal center of the three dinuclear platinum drugs CT008, CT033, and CT233 (Table 1). Encapsulation in both Q[7] and Q[8] provides substantial protection. It should be noted that protection was achieved

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

Cucurbituril receptors and drug delivery O

OH OAc

HO OH

H N

N H

OH

5

F O Dexamethasone acetate

Ethambutol

Overall charge NH3 Cl Pt

NH3 Z

NH2

NH2 Pt

NH3

Cl

NH3

NH2 O Pt NH2 O

H N

Z = (CH2)2 Z = NH2+ Z = N(CH3)2+ Z = CH2NH2+ NH3

O

(CH2)3NH2Pt NH2 (CH2)3 BBR3464 Z =

NH3

4+

O S

H N

N

N H

N O R′

2+ 3+ 3+ 3+

O

Oxaliplatin R

CT008 CT033 CT233 BBR3571

OR″

Prilocaine

Lansoprazole R = R′ = H, R″ = CH2CF3 Omeprazole R = OCH3, R′ = R″ = CH3 O O X RNH

RO NR′

Procaine X = O, R = H, R′ = ethyl tetracaine X = O, R = butyl, R′ = methyl Procainamide X = N, R = H, R′ = ethyl

Figure 4

N

N H

NR′2

Dibucaine R = butyl, R′ = ethyl

Continued .

even though the platinum metal centers are not completely inside the cavity of the Q[n] but rather reside just inside the portal (Figure 5). The degree of protection was found to be dependent on the ability of the metal center to enter the portal; hence, with a larger cavity as in Q[8], the linking ligand is able to fold drawing in the metal closer. This was clearly demonstrated with CT008 and CT233. The modest protection of CT033 was due to the limited ability of Q[n] to protect both platinum centers at the same time as a result of strong competing ion–dipole affinities between either Pt monocationic head or the protonated ligand amine and the portals. These competing affinities result in exposing at least one platinum center (Figure 6). Where maximum protection can be achieved, the degradation is significantly decreased. Compared to the free drug CT008 degrades 15 times slower, and the more reactive CT233 degrades nine times slower than its free state (Table 1).

In relation to the protective properties of Q[n], the inhibition of enzymatic hydrolysis of peptide-based drugs has been shown in isolated peptides. Therefore, a potential preservative effect is indicated for peptide substrates.49

2.2.3 Improved chemical stability Improved chemical stability refers to a reduction or prevention of chemical change of a pharmaceutical guest following encapsulation. This could include, for example, aerial oxidation, hydrolysis of aqueous preparations, thermal degradation, and sensitivity to light. The stability of association complexes as a function of the ease or difficulty at which they dissociate is discussed in Section 5 “the mechanism of release.” The importance of chemical stability is reflected in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

6

Molecular recognition

a

a c

b

b

d

Figure 5

Cl

c d

d c b

a

CT008 encapsulated in Q[7] and Q[8].

+

NH3 Pt NH2

+

NH2

NH3

+

NH2

NH2 Pt

Cl

NH3

Q[n] encapsulation.50 The separation of individual drug molecules by encapsulation could prevent conversion to alternative polymorphs in the solid state or aggregation in solution. The aggregation of laser dyes in solution can be avoided by Q[7] encapsulation, which suggests that this could also be applicable to hydrophobic drugs.53

Shuttle

Figure 6 Q[n] shuttling over the CT033 ligand showing the competing affinities between Pt monocationic head and the protonated amine.

concerns for extended shelf lives of drugs, the ease of handling during the physical and perhaps chemical stresses of formulation such as compatibility with excipients and solvents including moisture. Amorphous materials and crystal polymorphs can also provide challenges to manufacture and medicinal availability. Two forms of demonstrated improved stability through Q[n] drug encapsulation have been reported. These include thermal stability evaluation and a resistance to hydrolysis. Drugs such as atenolol, glibenclamide, memantine, paracetamol, pyrazinamide, and isoniazid (Figure 4) encapsulated in Q[7] as solid association complexes have all been shown to have higher thermal stability.50, 51 This has been consistently demonstrated using differential scanning calorimetry. However, a thorough evaluation of thermal stability using high-performance liquid chromatography (HPLC) analysis to determine minor degradation products has yet to be performed. A dramatic improvement in stability to hydrolysis has been shown in the case of ranitidine (Figure 4). Formulations of ranitidine hydrochloride are known to have instability toward humidity and pH. Under test conditions of 50 ◦ C at pH 1.5, the stability of ranitidine is extended to beyond two weeks in the presence of a slight excess of Q[7] with no decomposition compared to four days without Q[7] (50% decomposition).52 The stabilization of drug polymorphic states is an area of stability that has been proposed as being applicable to

2.2.4 Slow release or controlled release Slow release is aimed at sustained plasma concentrations over time. The alternative of frequent administrations leads to concentration spikes or requires constant concentration maintenance, through continuous administration. The development of slow release and controlled release systems incorporating Q[n] has so far been focused primarily on the development of methods to active ingredient release and not necessarily specific to administrative application. Biocompatible gels are recognized as one example where the potential exists for the controlled delivery of drugs.54 5-Fluorouracil has been shown to release slowly from Q[6]mediated alginate hydrogel beads.55 The alginate beads that formed in the absence of 5-fluorouracil are spherical with a diameter of 2.5 mm. Bead preparation with 5-fluorouracil (Figure 4) loading gave beads ranging from 3 to 4 mm in diameter with loading capacities of 3.87–6.13 wt%, respectively. It was found in vitro that slow release at pH 6.8, occurred with an optimal loading of 5.94 wt% (release of t1/2 = 2.7 h). The 5-fluorouracil is held in the network structure of the hydrogel and not encapsulated in the cavity of Q[6]. Encapsulation of drugs in the cavities of Q[n] is a more direct approach to slow release. However, the rate of release is dependent on a number of factors as discussed in mechanisms to release. Q[7] and Q[8] cavity bound drugs in solution are usually released in seconds, as exchange of a drug between the cavities or into a cell is facile. High binding constants are not necessarily a reflection of the ease of exchange as demonstrated by the saturation transfer experiment between Q[8] and Q[7] for the drug ABZ, which indicated a transfer of one molecule

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

Cucurbituril receptors and drug delivery

2+

2+

N N

N N

N

Ru

N

N

Figure 7

N Ru N

N N

N

CH3

7

(CH2)n n = 2, 7, and 10 CH3

Dinuclear ruthenium drugs (Rubbn ).

per 3 s.41 This shows that in the presence of a suitable receptor or through consumption, the drug would readily be released. However, slow release is possible through encapsulation when there are mechanical restrictions to the exit of a guest from a cavity. This is the case for the new drugs dinuclear ruthenium (Rubbn ) complexes (Figure 7). The Rubbn complexes are cytotoxic and have excellent antibiotic activity against bacteria.56 As a preliminary study of encapsulation, the Rubbn complexes are found to be encapsulated slowly and are released slowly over several hours from the cavity of Q[10].57 There are three reported examples of controlled release using Q[6] as central components in nanostructures. Two of these require internal cellular triggers such as pH changes58 or redox reactions59 to release their drug “cargo,” while the third responds to an external stimuli of an oscillating magnetic field that induces a local heating trigger.60 The mechanism of controlled release for each of these nanostructures is discussed in Sections 2.2.5 and/or 6.

2.2.5 Drug targeting Drug targeting is a desirable goal where ideally a drug can be delivered specifically to a set of diseased cells or invading cells (microorganisms) and the drug “off loaded” to only these cells. In the supramolecular context a molecular structure loaded with a drug acts as a vehicle that is carried in the blood stream and interstitial fluids, and attached to the vehicle is a cell specific (ideally) or selective, molecular tag. Certain cell types can be selectively targeted, such as cancer cells that over express a particular receptor or cell types with specific lectins. Drug targeting has the potential to limit unnecessary damage and reduce side effects. A model of targeted drug delivery has been demonstrated using Q[6]-based carbohydrate clusters, which were synthesized from (allyloxy)12 Q[6] (Figure 1, where R1 = R2 = allyloxy) to attach sugar units of choice to the “equator” of Q[6] (Scheme 1).61 It was found by in vitro

experiments with HepG2 hepatocellullar carcinoma cell with over expressed galactose receptors as a potential target that a Q[6]-based galactose cluster was translocated intracellularly. The experiment was extended to a drugcarrying model, where an encapsulated spermine (drug model) was also covalently bonded to a fluorescent dye molecule. The translocation of the dye indicated the validity of the model. It is suggested that the Q[6]-based galactose cluster carrying the drug model was a galactose receptormediated endocytosis.61 Extending this type of targeting, where the cavity can be utilized for drug delivery, could find greater application for the higher homologs given the larger cavities. An alternative targeted drug delivery system, also reported, incorporates derivatized Q[6] into a polymer nanocapsule or vesicle (Scheme 2).59 Again contrary to the concept of encapsulating a drug in the cavity of a Q[n], the drug was encapsulated in the polymer nanocapsule’s core and the cavities of the Q[6]s are used as a multifunctional platform for imaging probes and targeting groups. The structure of the nanocapsules, which are vesicle-like can be prepared at a uniform average diameter of 190 nm with a membrane depth of ∼6 nm. It is suggested that the structure is a layering of 5-6 Q[6] with interdigitation of the disulfide “tails” forming a hollow core. These nanocapsules are robust structures that are not easily ruptured unlike conventional vesicles. However, the disulfide “tails” are sensitive to reduction, which provides a potential mechanism for release of the core contents into the cytoplasm of the cell (see Section 5). Targeted delivery of doxorubicin was demonstrated in vitro using a disulfide-tailed Q[6] nanocapsule with a surface elaboration of a folate–spermidine conjugate as the targeting group. The targeting group is attached by encapsulation of a spermidine moiety in the Q[6] cavity and the doxorubicin entrapped in the vesicle-like structure. Selective intracellular uptake was observed in the cancer cell line HeLa.59 In addition, doxorubicin unloading was observed over a period of 2 h incubation at 37 ◦ C (see Section 5).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

8

Molecular recognition

O H OR

N

N

R1

H

CH2

RO

N

N

H OR

CH2

O

O SH

H

R2

OR H

n

n=6

= sugar

R1 = R2 = O

Scheme 1 An example of a Q[6]-based carbohydrate cluster and the general reaction scheme to its synthesis. The average number of sugar groups attached equals 11.2. n=6

O

O R1 = R2 =

N

N CH2

R1

O

S

R2 N

N CH2 O

S S

N H S

S

O

OH X

OH

n Q[6]

HO

HO

OH

HO

OH

S

S S

OH S S

HO

Targeting group S S

S S SS

OH

S S S

OH

S

S

S S

S

HO

OH HO

Drug-encapsulated internally

Drug release Disulfide reduction

Scheme 2 The general scheme to the amphiphilic Q[6], the formation of nanocapsules and reductive rupture of the disulfide bonds, and collapse of the nanostructure to release the drug “cargo.”

This was supported by doxorubicin accumulation in the cell nuclei and by a comparative cell viability study, where the folate-targeted nanocapsule performed ∼1.9-fold better than equivalent concentrations (10 µM) of free doxorubicin. However, free doxorubicin—doxorubicin loaded disulfidetailed Q[6] nanocapsules without targeting groups—and doxorubicin load regular Q[6] nanocapsule with targeting groups all have comparable activity. Perhaps, this indicates that disulfide-tailed Q[6] nanocapsules are translocated by

endocytosis and easily spill their contents through cytoplasm reduction but are not as readily translocated as a targeted nanocapsule. The doxorubicin load regular Q[6] nanocapsule with targeting groups would appear to be translocated readily, and doxorubicin leakage is responsible for cytotoxicity. Therefore, the combination of targeting and reductive rupture provides the maximum cytotoxicity. The nanocapsules at 60 µM have been shown to be relatively nontoxic to the HeLa cells when incubated over a period of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

Cucurbituril receptors and drug delivery two days. Since the nanocapsules are robust self-assembled structures, their toxicity needs to be considered initially as a single entity, similarly to a polymer but as break down occurs further evaluation is required. Robust nanocapsules of the type discussed present a promising potential future for highly active drugs as the loading capacity of a system of this type is relatively low and in the case discussed was found to be 1.5 wt%. Q[n] encapsulated drugs with targeting groups are likely to have a higher loading capacity but the controlled release of the drug requires further development.

2.2.6 Encapsulation in Q[n] and drug activity It is important that the pharmacological activity of a drug not to be adversely affected when administered by molecular delivery vehicles such as Q[n] or any supramolecular structure utilizing Q[n]. Much of the pharmacological activity measured to date, where Q[n] are involved have been in the form of in vitro biological tests. In a number of examples, the importance of the delivery system is in increasing the aqueous solubility of a hydrophobic drug, and while most show no significant change in activity, there are some that have slightly lower activity. As the aqueous solubilities of the hydrophobic drugs are increased and plasma concentrations would also be higher, in vivo these drugs may prove equally active in spite of small differences found in vitro. In vitro examination of ABZ@Q[7], on the cancer cells, human T -cell acute lymphoblastic leukemia cells (CEM), ovarian cancer cells (1A9), and human colorectal cancer cells (HT-29) showed no significant change in the cytotoxic activity of ABZ or MEABZ as a consequence of encapsulation.41 This was also true for MEABZ@Q[7] and MEABZ@Q[8] for the cell lines HT-29 and PC-3 (human prostate cancer cells).44 The encapsulated cytotoxic drug CPT also showed comparable activity to the free drug.42 Although there are exceptions, as the activities of CPT encapsulated in Q[7] or Q[8] highlights, there are slight variations in activity for different cell lines, for example, human nonsmall lung cells (A549) are slightly better where as a moderate decrease in activity was found for the human leukemia cell line (K562) (Table 2).42 The antituberculosis drug ethambutol when encapsulated in Q[7] also showed no significant decrease in pharmacological activity.33 A series of platinum base cyctotoxic drugs encapsulated in Q[n] have also been examined for in vitro biological activity. The dinuclear polyamine platinum drugs, diPt and BBR3571, encapsulated in Q[7] showed no significant difference in cytotoxic activity between the free drug and the encapsulation complex for L1210 and L1210/DDP cell lines (murine leukemia cells and cisplatin-resistant

9

Table 2 Activity comparison (IC50 at µM) for camptothecin (CPT) free, encapsulated in Q[7] and Q[8] for the cell lines—human nonsmall lung cells (A549), human leukemia cells (K562), and murine macrophage cells (P388D1). Cell Line

CPT Free

CPT@Q[7]

A549 K562 P388D1

7.76 0.43 2.47

6.36 0.93 2.38

CPT@Q[8] 6.78 1.13 2.98

cells, respectively).62, 63 In contrast, the oxaliplatin@Q[7] complex revealed a significant decrease in cytotoxic activity of varying degrees across five cell lines (A549 human nonsmall cell lung, SKOV-3 human ovarian, SKMEL2 human melanoma, XF-498 human CNS, and HCT-15 human colon).64 Similarly a series of platinum phenanthroline intercalating cytotoxic drugs with ancillary ligands of ethylene diamine or diaminocyclohexane as (1R,2R) and (1S,2S) led to mixed results for in vitro biological tests. Across the range of cavity sizes Q[6] to Q[8], the results were variable in activity, both positively and negatively. Perhaps, the surprising result was that the [Pt(5-Cl-phen)(S,S-dach)]2+ @Q[6] complex had a 2.6fold increased activity, [Pt(5-Cl-phen)(S,S-dach)]2+ @Q[7] >400-fold decrease; and only a 4.4-fold decrease for the larger Q[n], [Pt(5-Cl-phen)(S,S-dach)]2+ @Q[8].65 Where there have been decreases or even increases in activity in vitro, the exact reason for this change has not been established. It is not known whether this is simply a reflection of high binding affinities and therefore a decrease in available free drug and hence a decrease in activity or some other mechanism related to cellular uptake. The latter may also be related to whether the drug@Q[n] complex is entering the cell or the drug alone (see Section 2.1). Clearly, a factor that can affect in vitro biological activity is the ease of release of the drug, as demonstrated by encapsulation of BBR3464 (Figure 4). This highly active cytotoxic drug is encapsulated in a ratio of 1 : 2 (BBR3464: Q[n]) for n = 7 and 8 and a ratio of 1 : 1 for Q[10]. The consequence is that the activity is moderated by the ease of release from the cavity. Almost all activities are lost through encapsulation with Q[7] with activity returning as the cavity size increases and/or the ratio decreases (Table 3).62 High binding affinity could also help to explain why oxaliplatin@Q[7] has decreased activity and the [Pt(5-Clphen)(S,S-dach)]2+ @Q[n] complexes have variable results according to the Q[n] used. However, without a more detailed study to establish the mechanism of cellular membrane transport (drug release prior to cell uptake or drug @Q[n] and released after uptake), a proper analysis of variations in activity is not possible. Given the embryo cells and macrophage drug@Q[n] may occur at least in some cases.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

10

Molecular recognition

Table 3 In vitro evaluation of BBR3464 with the murine leukemia cell line (L1210) and the cisplatin-resistant cell line (L1210/DDP) measured as IC50 the required concentration to induce 50% inhibition of growth. Drug

Q[n]

IC50 (µM) L1210

BBR3464 BBR3464 BBR3464 BBR3464

Nil 10 8 7

57 nM 0.7 6.6 >37.5

L1210/DDP 24.5 nM 0.2 1.4 >37.5

Molybdocene dichloride (Cp2 MoCl2 ) is another cytotoxic metal complex that has been evaluated for its in vitro biological activity following Q[7] encapsulation. Cp2 MoCl2 @Q[7] was found to have improved activity when compared to free Cp2 MoCl2 with the 2008 cell line and the MCF-7 cell line. It is not known whether the improvement is related to solubility or membrane permeability.66 The only in vivo drug@Q[n] study performed to date is that involving the dinuclear platinum drug BBR3571 (Figure 4), which is similar in structure to CT033 except that it has the unsymmetrical spermidine linking ligand. The complex BBR3571@Q[7] such as CT033@Q[7] has less than ideal protection when encapsulated in Q[n] (see Section 2.2.2). At physiological pH, the amine in the spermidine linking ligand is protonated and this drives one of the platinum centers away from the portal. In spite of the limited protection provided by Q[7], in vivo studies in mice determined that the maximum tolerated dose was 1.7 times less toxic (BBR3571@Q[7] compared to equivalent levels of free drug).62 Whether this is a consequence of a reduction in toxic metabolites or not, this was not determined. An examination of the cytotoxic activity in vivo on subcutaneous tumors in mice (2008 ovarian carcinoma cell line) at equivalent drug doses (BBR3571), comparing the free drug to the drug@Q[7] showed identical activity against the tumor.62 An alternative to drug activity maintenance is that of drug reversal. Q[7] has been proposed as drug moderator or reversing agent by the competitive binding of drugs that are administered for enzyme modulation. This relies on a balanced competitive binding of the drug relative to the enzyme.16, 67

can be achieved that include bioavailability, bioprotection, improved chemical stability, slow release or controlled release, and drug targeting. In addition, the Q[n] are biocompatible and with careful evaluation can be used to improve or at least maintain the drugs pharmacological activity. A range of drugs have been evaluated for biological activity following encapsulation, and there is now sufficient evidence to suggest that drug activity can be maintained with the correct choice of Q[n] but each combination would require evaluation. The type and the structure of the drugs suitable for encapsulation are also becoming more predictable, although this process is yet to be finalized. The structural features required for encapsulation are discussed in Section 3.

2.3

Drug@Q[n] and administration

There is limited information to date with regard to which routes are applicable to drug@Q[n] administration or more specifically which Q[n] best serves each route. The solubility of Q[6] in saline solutions, reasonable solubility of Q[7] in water, and high solubility in saline suggest that these two Q[n] are applicable to intravenous administration especially as Q[7] is known to be readily excreted in urine.31 The increased aqueous solubility of substituted Q[n] suggests that substitution would render all relevant Q[n] homologs suitable in the future to intravenous administration. The high thermal and chemical stability of all the Q[n] means that they would also be suitable for oral delivery of drugs as they would easily tolerate the pH changes of the alimentary canal.68 The low solubility of Q[8] in nonacid conditions may favor its application in oral delivery. Formulation studies of Q[6] and excipients indicate that Q[n] can be readily compressed into stable and durable tablets for oral administration.69 Topical use for a water-dispersible ultraviolet (UV) shield (sunscreen agent) through supramolecular formulation with Q[n] by ball mill grinding has been patented. This is described as environment friendly with high UV absorbance, nontoxic, and a product removable with water.70 Other medicinal applications or routes of entry to the body are yet to be reported.

3

DRUG TYPES THAT CAN BENEFIT FROM CUCURBITURIL ENCAPSULATION

2.2.7 Conclusion 3.1 Considerable potential has been realized for Q[n] as a drug delivery vehicle through either cavity and/or portal encapsulation. There are a number of benefits that

Q[n] and drug matching

The whole of a drug or part of a drug can be bound to Q[n] through encapsulation within the cavity, or at either

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

Cucurbituril receptors and drug delivery carbonyl portal, or can involve both sites of the Q[n]. There are a number of basic considerations required to determine the suitability of Q[n] as a vehicle for delivery. These include: • • • •

the the the the

molecular shape and size, polarity or charge, presence of hydrogen-bonding donors, and presence of hydrophobic moieties.

Maximum association stability is achieved if a number of the criteria are satisfied.

3.1.1 Size and shape Obviously, the size and shape of a drug molecule preclude it from being encapsulated in the cavity or to partially enter the portal if the molecule’s physical dimensions are incompatible with a specific Q[n]. In some cases, only parts of a molecule need to be encapsulated to affect a desired outcome. The molecular size choices fall into a range for ˚ (Figure 2).1 a molecular width or diameter of ∼2.4–10 A ˚ The proposed depth of a Q[n] is ∼9.1 A (including van der Waals radii), although theoretical calculations indicate that ˚ 1, 6 the extent of the electron influence is actually ∼14 A. The length of the molecule is then only limited by how much of it is to be encapsulated. The equal access to two portal openings allows a molecule to thread to an undefined length. The width or diameter is dictated by the size of the portal opening, which can flex slightly to facilitate access to the larger diameter cavity. The Q[n] as relatively rigid structures can flex slightly into an ellipsoid shape, or the carbonyl portals can flex outward to accept a slightly larger width or diameter.71 The shape of a molecule is not only an important determinant to fitting a Q[n] cavity but also an added feature to increased binding strength. Maximum occupation of the space within the cavity increases the van der Waals contacts and therefore favors binding. Adamantyl ammonium ion or carborane are molecules that are roughly spherical in shape, and these molecules fit the cavity of Q[7] very well.72, 73 The binding affinity of the adamantyl ammonium ion in Q[7] is high (Ka = 4 × 1012 M−1 ) and only slightly lower in Q[8] (Ka = 8 × 108 M−1 ).7 While the shape and size in this example is important, there are other driving forces such as the charge and hydrophilicity.

oxygen atoms that form the rim of the portal. In contrast, anions are unfavorable to binding.74, 75 There are numerous examples of cationic assisted bindings. Dipole–ion interactions between a cationic functional group and the portal greatly increase the stability of a drug@Q[n] association complex. The most common cationic group for drug stability is an ammonium ion.41–44, 50, 51, 76–78 An ammonium ion-binding interaction is at its best when the charge carried is localized and not too diffuse.79 A protonated amine has a higher affinity for the portal than a quaternary ammonium ion.79, 80 Diffuse cations of NR4 + , SR3 + , or PR4 + are often encapsulated within the cavity rather than at the portal. Cationic groups of coordination metal complexes such as platinum or ruthenium cytotoxic drugs also form effective ion–dipole portal interactions.57, 62, 65, 80–82 Examples of affinity selection between cations that are carried on the same molecule are particularly prominent in the platinumbased drugs CT233, CT033, and BBR3571 (Sections 2.2.2 and 2.2.6). A protonated amine dominates the portal position over the Pt monocationic head in the latter two examples, whereas the Pt monocationic head is preferred over the NR4 + , which can be accommodate within the cavity.79, 80 The strong portal association of a protonated amine results in a stabilization of the cation and an increase in the pKa .44, 76, 77 Dipole–dipole interactions with the portal are primarily in the form of hydrogen bonding. This can be found in the neutral cytotoxic drug oxaliplatin, through the HN of the ligand, or the NH of amide groups of the organic drugs, atenolol or glibenclamide.50, 64 The nitrovinylamino group of ranitidine provides supporting binding strength, through hydrogen bonding at the opposite portal of Q[7] to a protonated amine, as this drug spans across the cavity (Figure 8).52 Hydrogen bonding is also evident for the encapsulation of guests carrying carboxylic acid groups. In general, the carboxylic acid group sits at one portal and the remaining parts of the molecule thread through the cavity to the opposite portal.75

3.1.3 Hydrophobic moieties Many drugs are difficult to administer because of their low water solubility. The Q[n] have polar portals that interact

NO2 NH+

3.1.2 Polarity and charged functional groups An electropositive or cationic functional group is of particular importance in the binding of a molecule or ion to Q[n]. There is a high affinity for functional groups with positive character driven by the electronegative carbonyl

11

S O

N H

N H

Figure 8 The ammonium salt of ranitidine encapsulated in Q[7] and stabilized by ion–dipole interaction at one portal and hydrogen bonding at the other.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

12

Molecular recognition

with water and ions and possess a hydrophobic cavity compatible to hydrophobic moieties. A number of the drugs previously discussed (Sections 2.2.1 and 3) have both polar groups and hydrophobic moieties. It is this combination that favors a stable association complex with Q[n]. An obvious example is the improved solubility of the antiinflammatory steroid drug dexamethasone acetate (up to 17-fold increase) utilizing the α,δ-(CH3 )4 Q[6] (Figure 3; R1 = R2 = CH3 , s = 2).83 This partially substituted Q[6] with aqueous solubility (1.4 mM) ∼75 times higher than unsubstituted Q[6] has an ellipsoid cavity capable of accommodating slightly wider guests such as dexamethasone acetate. σ -Carborane, a potential therapeutic agent (boron neutron capture therapy), is normally water insoluble but as a molecular guest in Q[7], it is water soluble.72, 84

3.2

Drug modification for improved Q[n] binding

Prodrugs are drugs that carry functional groups that are cleaved by a biological process such as an enzyme to release the active component. These functional groups are often applied to drugs to improve their water solubility.85 The prodrug approach to improved bioavailability or toxicity reduction, foreseeably lends itself to an added potential when combined with the noncovalent process of encapsulation by Q[n]. Relatively minor changes to a drug could impart an advantage to encapsulation and stability by exploiting the favorable binding features required for better Q[n] encapsulation and therefore provide greater benefits to the activity of the drug component relatively simply and reversibly. Prodrugs via esters are relatively common, and an ideal ester functional group that could lead to a prodrug@Q[n] combination has, in principle, been demonstrated where choline or phosphonium analogs provide the cation as an ester, for the dipole–ion interacting function.74, 86 Adding cationic groups to a hydrophobic core has been shown to be very effective for increasing binding with Q[n].7, 8 In some circumstances, minor changes to a drug to improve encapsulation may not even require that the drugs be prodrugs. The cytotoxic dinuclear platinum drug CT233 is a case in point. In order to achieve the best protection for the platinum metal centers, the central amine of CT033 was modified to a quaternary ammonium ion to give CT233 so that it could be accommodated within the cavity.79, 80 Through charge dispersion, the dominant ion–dipole drivers become the two Pt monocationic head centers, which act to fold the linking ligand within the cavity (Figure 4). Chain folding is favored because of improved van der Waals contacts and isolation of the relatively hydrophobic chain from the aqueous environment

Table 4 Cytotoxic IC50 in the murine leukemia line L1210 cancer cell line and its cisplatin-resistant cell line L1210/DDP. Complex (µM) CT033 CT233

L1210

L1210/DDP

0.13 8.6

0.12 3.2

RF 0.9 0.4

The resistance factor (RF) is defined as IC50 resistance/IC50 sensitive.

outside the cavity. This may also be facilitated by the small electronegative potential within the cavity.6 The cytotoxic activity of CT233 relative to the parent platinum complex CT033 was found to be lower, but the resistance factor was excellent (Table 4). While some activity has been lost, the in vitro cysteine degradation model showed a threefold increase in protection of the platinum metal centers in CT233 (see Section 2.2.2). This protection may translate into a significant advantage in vivo given that platinum drug concentrations in plasma fall rapidly without protection. As a second example of drug modification to gain advantage, the organic cytotoxic drug ABZ was modified by replacing the methyl carbamate with a methoxyethyl carbamate to give MEABZ that facilitates aqueous solubility of the carbamate group but has little effect on the overall solubility of MEABZ. In conjunction with this change and encapsulation of the remaining hydrophobic portion of the molecule in Q[8], the solubility was improved 1.3-fold compared to ABZ@Q[7] but 3.5-fold relative to ABZ@Q[8]. As a bonus, the cytotoxic activity of MEABZ was also found to be up to 10 times more active than ABZ.44 The potential for improvement through optimization of encapsulation as demonstrated for CT233 and MEABZ opens the possibilities of combining the concept of minor modification or prodrugs with Q[n] encapsulation.

4

4.1

PREPARATION OF ASSOCIATION COMPLEXES DRUG@CUCURBITURIL General procedure

4.1.1 Association complex—cavity or portal bound After evaluating the suitability of a drug for encapsulation according to size, shape, polarity, and hydrophilicity (see Section 3), one of the following methods can be used to prepare an association complex. Most examples involve the preparation of the association complex in water. Even though Q[6], Q[8], and Q[10] have low water solubility, a polar drug or a drug with a polar group can often increase the solubility of the association complex

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc056

Cucurbituril receptors and drug delivery Chemical shift ∆ppm − +

− −

0 −0.3 to − 0.4 0

− +

+ +

+

+

−

The low solubility of Q[8] means that this is generally suspended in water and a solution of the drug added. Association complexes that are formed with high stability generally do not require filtration but this can be included. Solid products are obtained from aqueous solutions by lyophilization or evaporation.

∼1.0–1.5

+

•

Poorly Soluble Drugs and a Binding Ratio of 1 : 1. A typical preparation involves mixing solid drug and solid Q[n] in H2 O. Mixtures are homogenized and sonicated. Following this process, mixtures are left to stand overnight. Filtration through a 0.5 µm PETE syringe filter gives clear solutions of drug@Q[n] complexes. Solid products are then obtained by lyophilization.

•

pH < 6. Given that anions are unfavorable to encapsulation and that encapsulation is improved when cations can be formed such as amine protonation, the pH can be important in the formation of stable association complexes with Q[n]. The above preparative methods are also applicable to drugs that can benefit from acidic preparative media. As a consequence of the stabilization of cations and hydrogen-bonding groups, following encapsulation, the pH of medium can be increased without dissociation of the drug (see Section 5). The degree to which the pH can be increased is dependent on the degree of stabilization and is specific to each case.

•

Cosolvent Processing. There are few examples of the use of cosolvents in the preparation of Q[n] association complexes, but the principle is sound. The problem exists for the unsubstituted Q[n] that there are few organic solvents that can dissolve both the Q[n] and an organic drug. While trifluoroacetic acid (TFA) has been used to encapsulate σ -carborane in Q[7] with water titrated into the mixture and eventual evaporation of TFA, it would not be broadly applicable.72 With the development of substituted Q[n] that have a broader solvent range, this approach becomes more appropriate.

•

Ball Mill Grinding. A relatively recent approach to improving bioavailability of insoluble drugs is ball mill grinding, which disrupts the crystal structure and creates ultrafine particles with better dissolution rates. Cogrinding of drugs in the presence of excipients or molecular hosts further improves bioavailability through nanoparticle formation.87, 88 The potential for ball mill cogrinding of Q[7] or Q[8] and poorly soluble or insoluble drugs has been reported with the highly water insoluble molecules such as [60]fullerene and a series of σ -phenylphenols.69, 89, 90 The synthesis of Q[n] association complexes by ball milling is faster

+

+ −

13

− −

Relative chemical shift scale

Figure 9 Proton resonance relative chemical shift differences compared to the free guest in the same solvent.

drug@Q[n] beyond the solubility of each individual component. The solubility of Q[6] can be assisted through the use of saline. Given that most drug@Q[n] association complexes will be soluble to 1 mM, 1 H NMR spectroscopy provides a very powerful tool for analysis. Shifts in proton resonances for the free drug relative to the drug@Q[n] can provide information as to the location of each proton in the cavity or in the vicinity of the portal of a bound drug. Shifts upfield of 1–1.5 ppm generally indicates a location at the geometric center of the Q[n] cavity. A shift upfield Cs+ which does not correlate particularly well with that observed for monesin.24 A key design feature of 11 is its ability to mimic the monesin intramolecular hydrogen bond, a structural feature that is also integral to the design of 12 and related carboxylic acid derivatives.25 O

O

O

HO O

O

O

O

O O

O O

O OH

NH2 HO

11

12

NH

O N H HO

N

O O

O

O Calcimycin (13)

Monesin tends to bind most strongly to monovalent metal ions and hence is sometimes referred to as a “monovalent polyether antibiotic.” There also exist a class of “divalent polyether antibiotics” such as lasalocid A and calcimycin (13) that binds to both monovalent and divalent metal ions. Lasalocid A exhibits a preference for the doubly charged species. The order of complexation strength Ba2+ > Cs+ > Rb+ ≈ K+ > Na+ ≈ Ca2+ ≈ Mg2+ > Li+ reflects a combination of ion size and charge density issues.26 Calcimycin is produced by fermentation of Streptomyces chartreusensis and acts as an antibiotic against gram positive bacteria and fungi. It allows divalent cations to cross cell membranes with selectivity in the order Mn2+  Ca2+ > Mg2+  Sr2+ > Ba2+ .27

1.4

Nitrogen podands

In addition to the classic bidentate chelate ligand ethylene diamine, a tremendous variety of linear polyamine ligands such as spermine, spermidine, putrescine, and cadaverine are known, many of which have biochemical roles as their somewhat evocative names suggest (Figure 12). Both spermidine and spermine, which is formed from it, are involved in cellular metabolism in eukaryotic cells. Spermine is an essential growth factor in some bacteria and exists as a polycation at physiological pH because the propylene and

H2N H2N

NH2 Putrescine NH2 Cadaverine

H2N H2N H2N

Figure 12

H N

NH2 Ethylene diamine

N H

NH2 Spermidine

N H

NH2 Spermine

Examples of polyamine podands.

butylene spacers reduce electrostatic repulsion on protonation, increasing basicity. As a result, protonated compounds of this type can also act as simple anion receptors as well as ligands for cations. For example, crystals of spermine phosphate were first described as early as 1678 by Anton van Leeuwenhoek, who obtained them from human semen. The proton binding ability of polyamines has been reviewed.28 Spermine is associated with nucleic acids and is thought to stabilize helical structure, particularly in viruses. Putrescine and cadaverine are produced by the breakdown of amino acids, particularly in dead organisms, and both are toxic in large doses. They cause the foul smell of rotting flesh and are also implicated in the smell of bad breath. Early work on complexes of these kinds of ligands clearly showed the additional stabilization imparted by the macrocyclic effect. The Zn(II) complexes 14 and 15, both of which form complexes using four chelating donor atoms, are compared. However, the macrocyclic complex 14 is about 104 times more stable than the podand analog 15 as a consequence of the additional preorganization of the macrocyclic effect.29 Despite this relative instability, however, polyamine podands bind reasonably effectively to a range of transition metals. Spermine, for example, forms a mononuclear complex with copper(II) sulfate containing both sixand seven-membered chelate rings, Figure 13(a).30 In contrast, a binuclear bis(bidentate) complex is known with palladium(II) chloride in which the coordinating chloride ligands occupy two of the four sites on each square planar palladium(II) center to which the spermine is bound by six-membered chelate rings, Figure 13(b).31 Interestingly, a related complex is known for lithium iodide even though alkali metal salts are not generally complementary to amine donor ligands in comparison to oxygen donors. The 2+

2+

NH HN Zn

NH HN

NH HN

NH2 NH2

14

Zn

15

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands

9

(a)

(a)

(b)

(c)

(b)

(d)

Figure 14 Helicates derived from oligopyridyl ligands: (a) mononuclear single helicate,43 (b) binuclear double helicate based on four-coordinate metal ions,44 (c) binuclear double helicate based on six-coordinate metal ions,45 and (d) circular helicate templated by a chloride anion.36

(c)

Figure 13 X-ray structures of metal complexes of spermine (H atoms omitted): (a) mononuclear copper(II) sulfate aquo complex,30 (b) dinuclear palladium(II) chloride complex,31 and (c) coordination polymeric lithium iodide complex.32

structure is a 1-D coordination polymer based on tetrahedral Li+ ions, Figure 13(c).32 Perhaps one of the most important classes of N-donor podand-type ligand are the polypyridyls.33 The combination of rigidity, excellent stability with a range of particularly lower oxidation state transition metals, and synthetic versatility has led to a cornucopia of metallosupramolecualr polypyridyl complexes. Broadly speaking, the three most important classes of compound are (i) helicates34 (including circular helicates35, 36 ), (ii) arrays such as grids, racks, and ladders,37, 38 and (iii) coordination polymers.39 Work particularly by the groups of Constable,40 and Lehn37, 41 among others in the late 1980s and early 1990s popularized helicates (helical metal complexes) as abiotic single, double, and triple helical supramolecular frameworks with interesting topology and as intermediates in topologically complex synthesis such as the preparation of a molecular trefoil knot.42 The simplest type of helicate is a mononuclear single helix exemplified by the silver(I) hexafluorophosphate complex of 2,2 :6 ,2 :6 ,2 :6 ,2 -quinquepyridine which is near-planar but twisted into a shallow helical conformation as a result of unfavorable steric interactions between the pyridyl terminii of the podand, Figure 14(a).43

Oligopyridyl ligands with greater terminal steric bulk or additional pyridyl rings as in 2,2 :6 ,2 :6 ,2 :6 ,2 :6 , 2 -sexipyridine display a more pronounced helicity. In the case of metal ions with a tendency toward tetrahedral or octahedral coordination geometries, the [4 + 4] or [6 + 6] double helicates result, as in the dicopper(I) perchlorate complex of tetramethyl-2,2 :6 ,2 :6 ,2 quaterpyridine (16) in the case of 4-coordinate copper(I), Figure 14(b) or a range of complexes of 6-coordinate metals such as Cd2+ , Fe2+ , Co2+ , Ni2+ , and Cu2+ with 2,2 :6 ,2 :6 ,2 :6 ,2 :6 ,2 -sexipyridine (17), Figure 14(c).44, 45 Interestingly, while the hexadenate sexipyridine can form a double helicate with the Jahn–Teller distorted copper(II), quaterpyridine does not because the preference for a distorted octahedral geometry of the metal ion is inconsistent with the helix-forming requirements of the ligand. As a result, reduction of the mononuclear copper(II) quaterpyridine complex results in redox-reversible helicate formation. One-electron oxidation of the compound gives a mixed-valence Cu(I)–Cu(II) helicate, which on further oxidation decomposes to give the mononuclear Cu(II) species (Scheme 2). The Cu(II)/Cu(I) redox interconversion in helicates with various podands has been extensively studied and different electrochemical behavior is observed according to the structural features such as denticity, rigidity, and steric hindrance of the helicand.46 In one case, the mixed-valence Cu(II)/Cu(I) helicate complex is stabilized by specific metal–metal interactions and can be isolated and structurally characterized by X-ray crystallography.47

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

10

Molecular recognition

Dinuclear helicates are far from being the limit in terms of the number of metal centers, and the reaction of copper(I) salts with ligands of type 18 gives tri-, tetra-, and pentanuclear double helicates based on tetrahedral copper(I).41 Helicates containing even more metal ions are known, often based on very simple ligands. For example, the pyrazole derivative 19 forms penta- and heptanuclear helicates with Zn(II) and Cd(II) salts.49 A particularly unique form of helicate is the circular helicate discovered by Lehn’s group in 1996, Figure 14(d). The complex is based on a podand with three bipyridyl binding domains and selfassembles around a chloride anion to give a pentanuclear complex with Fe(II).50 The analogous hexafluorophosphate salt is hexanuclear while a variation in the ligand spacer gives a tetranuclear species.35 Pyridyl ligands are not the only podands to form helicates. Extensive work from Raymond’s group in Berkley has been based on deprotonated 1,2-dihydroxy benzene-derived podands. Depending on the spacer unit between two of these binding domains, a large number of triple helices and hollow tetrahedral coordination shells have been prepared by self-assembly. The remarkable chemistry of the hollow shell compounds is covered in Self-Assembly of Coordination Cages and Spheres, Self-Processes and Reactivity in Nanoscale Vessels, Supramolecular Reactivity. For example, a mixture of three such ligands in a 3 : 2 ratio with Ga3+ gives the selective formation of three individual, homoleptic triple helices (Scheme 3).51

R R N

N N

N N R

N N

N N

R

R = H, CH3 16

17

+2 e−

[CuII(16)]2+

2 Mononuclear

N

[CuI(16)]+

Fast

2 Mononuclear

−2 e−

[CuI2(16)2]2+ Double helix

+ e−

Slow + e−

[CuII2(16)2]4+ Double helix

Scheme 2 state.48

− e−

[CuICuII(16)2]3+ Double helix

− e−

Helix formation as a function of copper oxidation

N

N

O

O n

N

N

N

N 18 H N N N

N

N H

N N

N N

N H

N

N N

N N N

20 19

OH OH OH O

OH

OH O

NH + 3

3

+ 3

O

CH3OH

OH

O

O OH

O

O

NH

+ O

O Ga O

OH

NH

NH

+

NH OH

O

NH

NH OH

O

O NH

Ga(acac)3

NH

O

Ga

O

NH

O

Ga

6−

O

NH

Ga

6−

O

OH O

NH O

3

Ga O

O Ga

3

OH

Scheme 3

6−

O

Positive cooperativity in the self-assembly of Ga(III) triple helicates.51

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

O

3

Podands

Figure 15

Decanuclear quadruple helicate.52

One of the most striking helicates is the decanuclear quadruple helicate formed from the reaction of silver(I) triflate and ligand 20, Figure 15. The compound is formed at the same time as a 20 metal atom grid complex comprising two separated 2 × 5 arrays of Ag(I) ions.52 Amid a growing range of spectacular structural studies, it is also noteworthy that helicates of lanthanoid metal cations have been used as model systems in a full thermodynamic analysis of the self-assembly process in the form of the extended site binding model.53, 54 Further details of helicates and related complexes are given in Self-Assembly of Coordination Chains and Helices, Self-Processes.

1.5

Tripodal ligands

Among the most enduringly popular podand ligands are tripodal species based on tris(aminoethyl)amine (tren, 21).

A huge range of pyridyl-based tripodal ligands have also been realized. While they are relatively flexible molecules, metal complexation results in a rigid array of chelate rings which can be used, for example, in face capping an octahedral metal center. Ligand of type 21 finds application in the extraction of metal ions; for example, compounds 21 (R = benzyl or napthylmethyl) both extract perrhenate in protonated form and are significantly more effective than compound 22 and in the case of the benzyl derivative, even more effective than cryptand analogs.55 Dipicolylamine (23) has been used very effectively in a range of tren-related tetrapodal architectures to produce dimetallic complexes with interesting sensing and catalytic functionality and the subject has been reviewed.56 For example, compound 24 is an effective fluorescent sensor for dianionic phosphate derivatives, particularly peptides with phosphotyrosine residues.57 The complex works by a threecomponent self-assembly mechanism with the phosphate derivative enhancing binding of the second zinc ion and preventing PET (photo-induced electron transfer) quenching from the uncoordinated amine nitrogen atom in the mono-zinc precursor. Compound 25 also acts as a sensor for phosphates, exhibiting a red-shift on addition of pyrophosphate. Unlike compound 24, however, the compound functions as a single molecular unit. Tripodal receptors based on an arene core have proved particularly popular, in part due to the degree of preorganization afforded by steric hindrance around the arene core. Early work by Lehn’s group showed that the tripodal receptor 26 in conjunction with planar ligand 27 gives a discrete capped trinuclear complex on reaction with copper(I).58

H N RHN

N

N NHR NHR

N

N N N

N

R = H, CH2Ph, CH2C10H7 21

22

4 NO3− N

N Zn2+ N N

Zn2+ N N 24

11

N N

23

NO2

N O− Zn2+ Zn2+ N N

N

25

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

12

Molecular recognition

N N

N

N

N N

26

N

Figure 17 X-ray crystal structure of the Fe(II) complex of a 2,2 -bipyridyl derived tris(imidazolium) receptor including Br− .60

N N

N N

N

27

Discrete mononuclear complexes stable in solution over long periods are afforded by the novel “coelenterand” (hollow stomach) ligand 28.59 This coelenterand exhibits two different coordination modes, in which the metal ion is either inside or outside the aryl “stomach” (Figure 16). There is a relatively conventional tripodal chelating mode (analogous to face-capping ligands such as triphosphines, tris(pyrazolyl)borates, and tris(pyrazolyl)methanes). The more unusual encapsulating mode is exhibited by the complex produced by reaction of 28 with [RuCl2 (DMSO)4 ] which gives a fully encapsulating complex that can be crystallized in the presence of [ZnCl4 ]− . In the X-ray crystal structure, the Ru(II) is bound both to the pyrazole nitrogen atoms and the carbon atoms of the arene ring, exhibiting a short Ru–ring centroid distance of ˚ (compared to normal values in organometallic 1.58 A

N

N

N N

N N N

N

N N M N N

Encapsulating

N N

M N N N N

Tripodal chelating

28

Figure 16

Coordination modes of the coelenterands.59

˚ suggesting that the complexes of about 1.67–1.70 A), arene–π interaction is enhanced by the chelation of the ligand. This kind of concept has been extended to imidazoliumbased ligands, in which three 2,2 -bipyridyl substituents bind an Fe(II) center resulting in the formation of an anion-binding pocket capable of complexing Cl− and Br− in acetonitrile solution with log K > 7. The X-ray crystal structure of the metallocryptand shows that halides are bound solely by CH· · ·anion interactions, Figure 17.60 Deprotonation of imidazolium ligands gives the highly topical N-heterocyclic carbene (NHC) class of ligands which exhibit strong σ -donor character and represent interesting alternatives to phosphines.61 Carbenes are beginning to be used extensively in catalysis and coordination chemistry and a number of multidentate derivatives of the podand type have been developed. Ligands 29 and 30 represent dipodal carbenes, both of which form mononuclear palladium(II) complexes with catalytic activity in crosscoupling reactions.62 The tren-based ligand 31 forms both a 2 : 3 complex with copper(I) in which the copper centers are linear, two-coordinate and a mononuclear tripodal tris-chelate.63 In contrast, the dipodal ligand 32 is not geometrically disposed to chelate a single metal center and forms a 2 : 2 metallomacrocycle with palladium(II) chloride.64 Tripodal benzene-derived ligands have also been used as siderophores (strong complexants for Fe3+ ) which are of significant biochemical and medical interest. The chemistry and biology of siderophores, which feature a number of interesting natural podands such as mycobactin, have recently been comprehensively reviewed.65 One early example is the tricatecholate mesitylene derivative 33 which

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands

CH3 N O

N Cy

N

CH3 N C N

CH3 O N C

N C

N

CH3

C CH3 N N

30

N C N

C N N

N C N Bu

N C N Bu

Cy

29

13

32

N

31

N C N CH3

CH3

is structurally related to the naturally occurring enterobactin but without its ester linkages (which are sensitive to hydrolysis). This ligand is a remarkably strong binder of iron(III) with a binding constant of 1046 , although this is still some million times lower than that for the enterobactin natural analog. The mesityl spacer group in 33 is smaller than enterobactin’s lactam ring perhaps resulting in a more strained complex geometry. Tests in Escherichia coli culture show that the complex is able to help the bacteria accumulate the very insoluble iron(III) and hence promote growth.66 Unsurprisingly, the hexadeprotonated nature of the bound ligand makes iron complexation highly pH dependent, and under mildly acidic conditions (pH 5) there is significant competition for Fe3+ from EDTA4− (ethylenediaminetetracetic acid) which is a stronger acid in its protonated form. The chemistry of EDTA, one of the best known podands, is covered in the next section.

OH OH O

HN OH O HO

N H NH O HO 33

1.6

OH

EDTA-type ligands

One of the best known podand-type ligands is EDTA which, in its tetraanionic form, is an extremely common chelate

[M·indicator]n + + EDTA4 − → [M(EDTA)](4–n)− + Free indicator Color 1 HO2C

N

HO2C (a)

Color 2

N

CO2H

H4EDTA

H N

CO2H

O

O

−

O

H N O

N N H

O

O

N H

The Murexide anion used as an indicator in EDTA titrations

(b)

Scheme 4 (a) Metal ion analysis using indicator displacement by the strongly binding podand EDTA4− . (b) Color changes during the titration of Ni2+ with EDTA4− and murexide under basic conditions. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

14

Molecular recognition

ligand used as a strong complexant in analytical chemistry, for example, in the analysis of Ca2+ and Mg2+ in urine samples. Metal ion analysis is commonly carried out using the indicator displacement assay technique during complexometric titration as shown in Scheme 4, in which a colored indicator such as the murexide ion in the determination of calcium is displaced from the metal center by complexation with EDTA4− . The flexible nature of EDTA4− means that it binds a wide variety of metal cations with selectivity depending mainly on the metal ion charge (highly charged ions are bound more strongly) and size. Hard metal ions are favored because of the carboxylate donors, and hence softer metals such as Ag(I) are bound relatively weakly, although notably the binding of the univalent alkali metal ions that cannot interact with the amine nitrogen atoms is very weak indeed, Table 2. While EDTA commonly binds octahedral metal ions as a hexadentate chelate as in the classic NH4 [Co(EDTA)]·2H2 O published in 195967 (Figure 18a), larger metals such as Mn2+ can adopt a seven-coordinate structure involving coordinated water, as in [Mg(H2 O)6 ][Mn(EDTA)(H2 O)]·H2 O (Figure 18b).68 The success of EDTA chelates has sparked extensive research on a wide range of analogs including the closely related trimethylenediamine tetraacetate (tmdta), which as its Fe(III) complex is used in the bleaching of photographic films and paper, to more unusual analogs such as 34

(a)

Figure 18 X-ray molecular structure of EDTA complexes (a) NH4 [Co(EDTA)]·2H2 O67 and (b) [Mg(H2 O)6 ] [Mn(EDTA)(H2 O)]·H2 O.68

which are designed to lower the overall ligand charge. This podand shows selectivity for Cd(II) and Pb(II) over Zn(II). The ligand forms significantly more stable complexes with Cd(II) and Pb(II) than EDTA and hence has potential applications in the extraction of these metals.69 EDTA chemistry has also inspired a range of lariat-type ligands for lanthanoid metal ions that have found extensive applications in biomedical imaging, for example, as MRI (magnetic resonance imaging) contrast agents or as luminescent anion and pH sensors.70 One such complex, 35, has been shown to be an effective lactate and citrate sensor in diluted microliter samples of human serum, urine, or prostate fluids allowing a simple, fast method of detecting prostate cancer.71 Similarly, paramagnetic lanthanoid complexes of the EDTA analog 36 find clinical applications as contrast agents in MRI.72

Table 2 Stability constants (log K) in aqueous solution for metal complexes of EDTA4− .7 Mg2+ Ca2+ Sr2+ Ba2+ Mn2+ Fe2+ Co2+ Ni2+ Cu2+

8.7 10.7 8.6 7.8 13.8 14.3 16.3 18.6 18.8

Zn2+ Cd2+ Hg2+ Pb2+ Al3+ Fe3+ Y3+ Cr3+ Ce3+

La3+ Lu3+ Sc3+ Ga3+ In3+ Th4+ Ag+ Li+ Na+

16.7 16.6 21.9 18.0 16.3 25.1 18.2 24.0 15.9

15.7 20.0 23.1 20.5 24.9 23.2 7.3 2.8 1.7

O

N O−

N

N N

O

H 2O H

O

Since podands benefit from multiple interactions with their guest cations, significant research has been carried

O O

N Eu3+ N N

N −

O

O

35

HO2C

N

N

N

CO2H

N S

HO2C CO H CO2H 2

NH −

CH3 34

Octopus-type podands

CH3 HN

N

1.7

−

O N

(b)

O O

36

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands OR OR

CH3OC2H4O OC2H4OCH3 CH3OC2H4O Si CH3OC2H4O

O

O

S O

S

OR

S

Si

CH3OC2H4O CH3OC2H4O Si CH3OC2H4O

S

O

S O

O RO

RO R = Alkyl varying numbers of ethyleneoxide units

S

OC2H4OCH3 Si OC H OCH 2 4 3 OC2H4OCH3

S CH3OC2H4O Si OC2H4OCH3 CH3OC2H4O CH3OC2H4O Si OC2H4OCH3 OC2H4OCH3

37

out on systems with multiple arms, of which the “octopus” podands are perhaps the best representative example. Named for their multiple (although not necessarily eight) tentacles, octopus podands such as 37 and 38 have six arms radiating out from a hexasubstituted aryl core.73, 74 Compounds of type 37 are extremely effective alkali metal ion extraction agents; even more so than the crown ethers despite their water insolubility. Their broad 1 H NMR spectra suggest restricted conformational motion about the hexasubstituted core. The siloxane analogs 38 can, interestingly, be polymerized into hybrid silicas.

1.8

OC2H4OCH3

S

S

S

OC2H4OCH3

S

S RO

15

Salt-binding ligands

A key application area in podand chemistry is in the binding and extraction of metal salts. Since metal ions necessarily come with an accompanying counter-anion, significant effort has been devoted to the design of systems capable of binding anions and metal cations simultaneously. By including both anion- and cation-specific recognition functionalities, the selectivity and, in some cases, catalytic specificity can be dramatically enhanced. This section looks at systems in which the metal ion is relatively labile and the focus is on binding both the metal and the anion. We resume the topic in Section 2.4 below which covers more inert metal complexes specifically designed to bind anions. Some of the earliest simultaneous receptors are “cascade complexes” which originally date back to the 1970s,75 in which a podand such as a Schiff’s base binds metal cations which in turn coordinate to the counter-anions. Cascade receptors have been extensively used as models for enzyme active sites.76 Copper complexes are especially popular in this regard because of interest in dicopper hemocyanin respiratory proteins that are responsible for oxygen transport in molluscs and some arthropods, and studies on the coppercontaining enzyme tyrosinase. For example, podands of

38

type 39 exhibit modest antiferromagnetic coupling between ˚ apart. While the two copper(II) centers, which are ∼3 A this situation does not accurately reflect the properties of hemocyanins, the presence of different environments for the two copper atoms (one is distorted trigonal bipyramidal while the other is square pyramidal) may be related to the different modes of bonding proposed for the two copper atoms in tyrosinase.77 Manganese(II) cascade complexes of another unsymmetrical ligand 40 represent functional models for manganese catalase, an enzyme that catalyzes the disproportionation of hydrogen peroxide into dioxygen and water. The complex is again asymmetrical as shown by the X-ray structure of [Mn2 (40–H)(CH3 CO2 )2 (NCS)] (41), in which one Mn(II) center is distorted square pyramidal while the other is distorted octahedral.76 This bimetallic Schiff’s base motif underlies a large number of bimetallic cascade podand complexes, as in sterically hindered divanadyl complex 42 which binds alkoxide ions in a cascade fashion, while the copper analog binds oxide in a 2 : 1 sandwich fashion, allowing the oxide ligand to simultaneously interact with four copper(II) centers in two separate complexes.78 Compounds 43 and 44 represent cascade complexes that bind hydrogen phosphate anion, HPO4 2− , in water. The tetrahedral copper(II) center coordinates the phosphate anions which are further stabilized by charge-assisted hydrogen bonding to the tripodal peripheral groups. Complex 43 is protonated at neutral pH and binds HPO4 2− with Ka = 2.5 × 104 M−1 in water. The more preorganized 44 binds phosphate more weakly at 1.5 × 104 M−1 but is more selective for this anion. Thermodynamic studies show that phosphate binding by 43 is entropically driven, while complexation by 44 is enthalpy based as a result of decreased solvent organization around the guanidinium groups in 44 compared to the more exposed ammonium groups in 43.79

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

16

Molecular recognition

R1

CH3

Br Br

NCS O N X

O CH3

O

CH3

(CH2)n

Cu

Cu N

CH3

N

N

N

OH N

N(CH3)2 R2N

2

R R1 = Br, CH3 R2 = H, CH3 X = Br, OAc n = 1, 2 39

R = CH3, Et

NCS N O

Mn O N O CH3 CH 3

40

+ + NH3 H3N − − NH3 O O P O − O Cu

O O N R R

R

i Pr

N H +

N

42

NH

43

44

N N O N U O O O O

O

HN

O

O 45

O

2

O O

O

O

O

O

R

− + R = t Bu N H O H N − O S O O O O 47

O

NH

O

+

HN O

O

R

O

O NH

N Ni

N O N U O O O O

R

+ − N NH O HO − O P HN − NH HN O N Cu N N N

H

N

+ N

R

R = t Bu

41

+

N N

N OO ON V V O O− O

R

N Mn

O O

46

A ditopic, Schiff’s base podand that simultaneously binds uranyl ions while hydrogen bonding to anions such as dihydrogen phosphate via pendant arms is known (45).80 The compound forms an interesting 1 : 2 host:guest complex with H2 PO4 − , with one phosphate strongly bound to the metal center and the other hydrogen bonded to the first. Addition of crown ether moieties as in 46 results in binding of K+ as well as phosphate. The salen-based receptor 47 is an example of a selective extractant for metal sulfates. The ligand forms a complex involving deprotonated phenolic hydroxyl groups and protonated morpholine residues adapted for hydrogen bonding to the anion. The ligand readily extracts CuSO4 into chloroform solution, for example, with essentially 100% efficiency.81

ANION-BINDING PODANDS

Anion binding has emerged over the last 20 years as a highly active research area, not least as exemplified by an entire issue of Chemical Society Reviews dedicated to the topic in 2010.82 Anions are ubiquitous in biology with up to 75% of enzyme substrates and cofactors being anionic. They also play a role in disease with cystic fibrosis caused by a defective chloride transport protein.7, 83 Anions can also have a large environmental impact, for example, perchlorate pollution in the Colorado river and the radioactive anion 99 TcO4 − which can leach from nuclear waste.7 There is a wealth of literature on anion recognition with interest in both macrocyclic and podand systems.7, 83–85 Podands have a great deal of potential in anion binding and selected acyclic anion receptors have been the subject of a recent review.86 As with cation binding, the preorganization of a podand for anion binding can be tuned quite readily by altering the rigidity or tailoring steric or other noncovalent interactions to finely tune the system. The development of anion-binding podands over the past 20 years highlights the huge diversity and diversification

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands of the podand concept and the majority of this chapter is devoted to anion-binding podands. Anion-binding podand receptors are dominated by NH hydrogen bond acceptors such as amide,87, 88 pyrrole,89 indole,90 and urea/thiourea87 derivatives.83 Charged systems containing guanidinium91, 92 and imidazolium93 functional groups are also commonly used. Podands provide not only receptors which are often synthetically simpler than macrocyclic systems87 but the flexibility in binding group choice and variable level of preorganization, which in turn allows a great deal of flexibility in receptor design to suit almost any application. Crucially, as in cation-binding podands, anion-binding systems offer flexibility generally leading to rapid binding and decomplexation kinetics7, 94 coupled with binding constants that are suitable for analytical and sensing applications. It is perhaps fair to say that anion binding is the major focus of work in the podand field at present and the following sections give an overview of some representative anion-binding podands that cover a range of structural and functional features, highlighting the wealth of variety that is possible by tuning the preorganization of a receptor, changing the binding moiety, and introducing reporter groups which can turn receptors into sensors.

2.1

Cholapods—preorganized anion-binding podands

2.1.1 Concept and properties Podand systems are intrinsically more flexible than macrocyclic systems; however, this reduction in preorganization can also lead to lower binding affinities for anions. Rigid, aryl end groups can increase binding affinity1 ; however, to truly enhance binding strength, a rigid receptor design with a highly preorganized binding cavity is required. This concept is typified best in anion-binding receptors by the cholapods, pioneered by Anthony Davis of Bristol University, United Kingdom.95 The cholapods are derivatives of the bile acid and cholic acid, and are based on a steroidal, fused alicyclic ring system. This provides a highly rigid scaffold on which binding functionality can be added. The cholic acid core allows for variation in design by both regioand stereo-control.96 The conformation of the core cholic acid scaffold is curved with three hydroxyl groups on the α surface (Figure 19). The equatorial 3α-OH group is the least hindered, while the 12α-OH is less hindered than the 7αOH because of unfavorable 1,3-diaxial interactions between the 7α-OH and CH2 groups. The hydroxyl groups can themselves act as hydrogen-bond donors to bind anions; however, it is possible to convert all or individual hydroxyl functional groups to amine or amide moieties to further

17

b Surface 12

OR

7 OH

5

OH

3

OH

Figure 19

O 17

a Surface

Basic structure of cholic acid derivatives.

enhance the hydrogen-bonding capabilities or to provide a versatile receptor design. The large extended structure, with separated functionality of the cholapods, is an advantage when trying to bind often large anions.96 Neutral anion receptors have many potential advantages over charged systems. Primarily they may provide more selective binding by ensuring directionality, as anisotropic hydrogen bonding is the primary binding interaction rather than nondirectional electrostatic attraction. In addition, neutral systems do not have counterions which can compete to bind to the receptor. Binding constants measured are therefore absolute affinities and not relative to the counterion affinity. Neutral podand cholic acid derivatives have been studied extensively.96, 97 A range of neutral receptors have been produced, incorporating hydrogen-bond donor NH functionality to the α surface of the cholic acid in the form of amides, sulfonamides, carbamates, ureas, and thioureas. Examples of first generation receptors of this type are compounds 48 and 49.98 In 48, free rotation is possible around the C3 –N bond; however, further preorganization (in addition to that provided by the core) is achieved by the restricted rotation of the carbamate-O/NH groups and the preferential Z,Z-conformation across the carbamate moieties. Binding affinity was measured by 1 H NMR spectroscopic titrations in CDCl3 , with 49 showing large affinity, for chloride (92 000 M−1 as the tetraethylammonium TEA salt), two orders of magnitude higher than its closest rival bromide. Binding constants are smaller for 48, O O O Ts

N

H

H

O

OCH3

ON

NH

48 O N N H Ts Ts

N

H H

OCH3

Ts

49

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

18

Molecular recognition

with fluoride binding being the strongest, Ka = 15 400 M−1 (fluoride was not measured for 49). Further refinement of the receptor design has involved the addition of electron-withdrawing substituents on the binding arms, increasing the acidity of the hydrogen-bond donor and hence increasing the potential strength of the hydrogen bonds to guest anions. Receptors 50 and 51 exhibit a p-nitrophenylsulfonyl group, and the use of nitro or trifluoromethyl groups on the carbamates leads to a 10-fold increase in the binding constant (3.4 × 107 M−1 for chloride (as the TEA salt) for 51 compared to 92 000 M−1 for 48).99 O O O O S

OC20H41

O

mutations in the cystic fibrosis transmembrane conductance regulator protein, which acts both as a transmembrane chloride channel and as a regulator of other ion channels.7 Neutral cholapods have been shown to be effective anionophores, transporting nitrate, hydrogen carbonate, and notably chloride, particularly in the case of compound 52. The anion transport was measured using a dye technique in which the fluorescence of lucigenin encapsulated in a vesicle is quenched by the inward flow of chloride ions.104 The charged guanidinium derivatives such as compound 53 proved capable of extracting N-acyl amino acids from an aqueous phosphate buffer into chloroform, often with good efficiencies (measured using NMR spectroscopy) and good enantioselectivies.105, 106

O N

H

H

O

ON H N

R O R

50 R = NO2 51 R = CF3

N N+

By increasing the number of hydrogen-bond donors to five using urea or thiourea moieties, exceptionally high binding constants can be achieved. For example, compound 52 shows a chloride (as the TEA salt) association constant of 1.03 × 1011 M−1 , the highest binding affinity measured for a neutral organic anion receptor, and highlights the level of binding strength that can be achieved through highly preorganized receptors, with convergent binding cavities and multiple, strong hydrogen-bond donors.100 O N S NH SN O H O S N H H N H

OC20H41 NO2

NO2 NO2

O

H N

NO2

52

OCH3

O

H

H

ON NH

CF3

H CF3

53

Membrane transport experiments using compound 53 and a U-tube apparatus showed that the compound can transport N-acetyl-DL-phenylaniline with 70% enantiomeric excess, with the anions bound by hydrogen bonds from the guanidinium and carbamate NH groups. Ammonium-based cholapods have also been investigated as “smart transfer agents” as well as membrane transporters. The use of ammonium groups provides not only a positive charge but also hydrogen-bonding groups. Receptors of type 54 transfer anions from the aqueous to the organic phase.107 The extent of anion transfer is dependent on the lipophilicity of the anion (given by the Hofmeister series, Table 37 ). It was hoped that anion recognition would allow anti-Hofmeister behavior in which less lipophilic anions are extracted more readily. However, while the lipophilic anions were extracted less preferentially, the order of extraction in the Hofmeister series remained intact.

It is also possible to produce charged cholapods, typically using guanidinium or ammonium groups but also imidazolium101 and triazolium102 moieties. These charged systems have been employed as anion receptors, “smart transfer agents,” and membrane transport anionophores.

O N H N+

H H

H

H N H

N

ON H

OR O

54 R = CH3 55 R = C20H41

2.1.2 Membrane transport by cholapods Transmembrane transport of anions is a highly active area of research with many implications for biological and medicinal chemistry,103 most notably in the case of the genetic disease cystic fibrosis which arises through

While compound 54 is ineffective at membrane transport of anions, the addition of a long alkyl chain as in 55 does result in the transport of anions through a chloroform

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands Table 3

19

The Hofmeister series.

Weakly hydrated (hydrophobic)

Strongly hydrated (hydrophilic)

Anions: Organic anions > ClO4 − > I− > SCN− > NO3 − > ClO3 − > Br− > Cl−  F− , IO3 − > CH3 CO2 − , CO3 2− > HPO4 2− , SO4 2− > citrate3− Cations: N(CH3 )4 + > NH4 + > Cs+ > Rb+ > K+ > Na+ > H+ > Ca2+ > Mg2+ , Al3+

liquid membrane (as part of a U-tube set-up) with a small selectivity for chloride. A bicyclic guanidinium core has been utilized by de Mendoza and coworkers for transport of uronic acid salts. Receptors 56 and 57 contains a bicyclic guanidinium group connected to two modified deoxycholic acid motifs.108 The binding of D-glucuronate, 58, (as the tetrabutylammonium, TBA, salt) was evaluated using 1 H NMR spectroscopic titrations in acetonitrile. The receptors with the most flexible linker group showed the lowest binding constants, consistent with reduced preorganization. The glucuronate is bound to the guanidinium groups via the carboxylate moiety with the hydroxyl groups of the deoxycholic acid derivatives providing additional hydrogen bonds to the carbohydrate alcohol groups. The largest contribution to the binding strength is from the ion pairing of the carboxylate and the guanidinium groups.

the guanidinium moiety. The cholic acid groups can then encapsulate the anion by OH hydrogen bonds to the inner surface, presenting a lipophilic outer surface, allowing the complex to pass through the membrane.

2.1.3 Anion sensing by cholapods The addition of fluorescent moieties can provide a reporter group which, in principle, can result in a selective receptor with high anion binding that can signal this binding through changes in the fluorescence emission. The advantage of a podand-type receptor is that binding and release are generally fast in comparison to rigidly preorganized macrocyclic systems. Fang and coworkers110, 111 have developed cholapods functionalized at the C-24 position, which provides both additional hydrogen-bonding functionality and an anthracenyl reporter group, compounds 60 and 61.

N

O

N + N S

H

H

n

S

NH

n

HO

OH

HO HO HO 58

CO2− O OR

O O

O

NH

OH

O

HN S

NH

HN

60 56 n = 1 57 n = 2

O

OH

HO

NH

OH OH

coworkers109

Regan and have synthesized a range of charged anion transporters. An interesting example is that of 59, which is known as a molecular umbrella. This dicationic species containing ammonium and guanidinium functionalities can bind adenosine triphosphate (ATP) through the interaction of the phosphate residues of the anion with

S

HN S HN

NH NH 61

O O OH OH OH

HN

N

N H

O H3N + 59

NH2 HN +NH2

OH

OH OH

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

20

Molecular recognition

Compound 60 shows strong anion affinity in acetonitrile, particularly for carboxylates with an acetate binding constant of 7.69 × 104 M−1 . The compound functions as a PET sensor, with fluorescence quenching observed on addition of anions. Compound 61 is also a PET sensor and shows remarkably strong anion binding to dicarboxylates in highly competitive solvents (1 : 1 v/v methanol:water); for example, L-glutamate is bound with a binding constant of 5.57 × 106 M−1 . As with 60, fluorescence quenching is observed on addition of anions; however, only a maximum of 20% reduction in intensity is achieved.

2.2

Induced-fit anion binding

2.2.1 General considerations The principal advantage of highly preorganized receptors, such as the cholapods, is their specific and strong anion binding. However, this can also be a disadvantage in some circumstances, for example, in sensing. The small changes in conformation resulting from anion binding mean that the communication of the binding event to the reporter group, such as a fluorophore or chromophore, must be done via a change in the electronic distribution of the receptor. This may prove inefficient in comparison to processes requiring significant structural or conformational rearrangement. An alternative approach is to use conformationally flexible receptors which can change their shape or relative disposition of binding sites or chromophores on anion binding and in doing so, they lead to a change in the physical property of the molecule. Flexible systems also have faster complexation/decomplexation kinetics, which is also an advantage in sensing applications. Induced-fit binding can also lead to anion sensors capable of a significant degree of discrimination between anionic guests, because the induced conformation of the host is dependent on the size and geometry of the anion bound; therefore, each host/guest geometry is unique and can potentially affect a reporter group in a distinct way. The term discrimination in this context is distinct from binding selectivity (as measured by the magnitude of the binding constant) since it refers to the response of the receptor system to the guest-binding event. The distinction is particularly well exemplified in colorimetric sensor arrays in which each individual receptor is only very poorly selective, but the array as a whole can be highly discriminating in its ability to recognize particular guests based on the pattern of their response.112, 113 A possible disadvantage of an induced-fit receptor is the intrinsically lower binding constants compared to those of the preorganized systems. Typically, an unfavorable entropic contribution is expected when binding an anion as the system becomes more ordered. In macrocyclic systems, with less conformational freedom, this entropy cost

is paid during the synthesis of the molecule. However, the situation can be complicated by the restrictions on small conformation motion imposed by the binding.114 For flexible systems this is not the case, and the reorganizational energy represents an unfavorable contribution to the overall binding free energy. The conformational flexibility can potentially mean that a large range of anions are bound with similar binding constants, resulting in reduced thermodynamic binding selectivity. However, as the sensing method is dependent on conformation rather than binding strength, induced-fit sensors can still discriminate for particular anions, even though that anion may not have the highest affinity for the receptor. By using well-designed molecular architectures, it is also possible to increase the preorganization of a receptor (and hence increase the binding constant) while still allowing a degree of flexibility. The trialkylbenzene motif has been extensively used for this purpose because of its ability to balance these two competing attributes and is discussed in the following section.

2.2.2 The trialkylbenzene motif The hexasubstituted benzene moiety, typically consisting of ethyl substituents in the 1, 3, and 5 positions and binding arms in the 2, 4, and 6 positions, has proved to be a highly versatile motif in both cation and anion binding. It has been combined with a wide range of binding groups to create both neutral and charged receptors and sensors. The moiety provides a balance between flexibility and preorganization; steric interactions between adjacent substituents on the hexasubstituted benzene ring favor an alternating three-up, three-down arrangement of substituents by approximately 15 kJ mol−1 (so-called “steric gearing”).115, 116 This arrangement creates a convergent binding cavity, although the stability of the three-up, threedown conformation can be influenced by both electrostatic repulsion and steric interactions between binding arms. In cationic receptors, the C3 symmetric “three-up” conformation is by no means the only conformation observed and can be in equilibrium with the “two-up, one-down” conformation also present in solution (Figure 20).117 In some cases, additional conformational flexibility is also possible by involving the relative orientation of binding functional groups with respect to the cavity. Hence, in and out conformations of the binding groups are also possible; for example, in the hypothetical molecule in Figure 21, conformation (a) has all binding arms (R) in the in conformation, forming a convergent binding cavity, while (b) has all R groups in the out conformation and has a divergent cavity. Both in and out conformations have been shown to be present at room temperature in solution.117

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands R

HN +

NH

−

O

O

21

O

NH

R

R

R

+

H+ N

CO2−

HN

62

R

63

R Two-up, one-down

Three-up

Figure 20

HN

NH

R

R

CO2−

H N

H N

R

Trialkybenzene motif and its conformations. R

R

R R R

R (a)

(b)

Figure 21 In (a) and out (b) conformations of the trialkylbenzene motif.

The trialkylbenzene motif in anion binding was pioneered by Ansyln and coworkers118 and proved to be a highly successful platform for a range of anion sensors incorporated into an indicator displacement assay (IDA). An IDA system consists of a receptor such as 62 and a fluorophore or chromophore guest molecule such as the fluorescent indicator 63, which weakly binds to the receptor in solution. When an anion that has a higher affinity than the guest dye is added to the system, it displaces the indicator from the binding cavity. The change in the microenvironment of the fluorophore causes a change in the UV–vis absorption and/or fluorescence emission of the indicator. In the case of 62, the receptor proved to be highly selective for the tricarboxylate citrate, above even other carboxylates, as well as other salts and sugars.115, 119 The molecular structure of the 62·citrate complex determined by X-ray crystallography can be seen in Figure 22. The assay is able to detect the concentration of citrate in water (a highly competitive solvent) through increases in the absorbance and emission intensities. The concentration of citrate was also determined in orange juice and other soft drinks. Several other sensing arrays, for example, compound 64 with pyrocatechol violet (65), have been developed by the Ansyln group. The binding of gallate-like anions can occur through the reaction of the phenol group with the boronate ester and the binding of the carboxylate to the guanidinium group.120 This results in the displacement of the indicator from the binding cavity, resulting in

Figure 22 Molecular structure of 15·citrate. Second complex in the asymmetric unit is omitted for clarity.115

an observable color change. This sensing system was used to accurately determine the age of Scotch whiskey, as gallate-like anions leach into the whiskey over time from the storage barrels. The same receptor with alizarin complexone (66) was able to determine the concentration of tartrate in wine and fruit juice.121 Several other assays designed to sense the biologically important anions glucose6-phosphate122 and inositol triphosphate (IP3 ) have also been developed.123 O OH HN +

65

NH NH

SO3−

OH B− OH N+ H 64

OH OH

H N+

HN HN

−

O2 C −

O2 C

O

H N

66

HO OH

O

Combining several different anion receptors into a microarray can allow for the discrimination of anions present in a solution and the receptors are often known as electronic tongues or noses. Here, a series of receptors, each with often subtly different peak anion selectivity, can discriminate between various anions through their differential colorimetric, fluorescence, or electrochemical responses. An example of this sensing array concept has been developed by Ansyln and coworkers who used a differential receptor system to discriminate between the nucleotides

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

22

Molecular recognition + NH2 O

Polymer resin

HN

O

* O

NH n + NH3

HN

R

H N

N H

N H

O

N H

O

R

H N R

+ NH3 O O

HO = Fluorescein indicator = NH +

NH2

R

O NH

N NH H

O

R

H N R

O

CO2H

+

NH3 O

67 Image program

CCD

Microspheres arranged in cavities

Flows in

Flows out 100 µm Light source

Figure 23 An array-sensing system with receptor with varying “R” groups and indicator is illustrated. (Reproduced from Ref. 112.  American Chemical Society, 2001.)

ATP, adenosine monophosphate (AMP), and guanidine triphosphate (GTP). The trialkylbenzene motif was used as the basis for the receptors, with each receptor immobilized onto a polymer resin by attachment to one of the binding arms (for example, 67 in Figure 23). The two remaining arms were functionalized with short peptides, synthesized using combinatorial synthesis. Thirty polymer beads were then used to generate the sensing array. The sensing method was based on an IDA, with fluorescein used as the chromophore. Analyte solutions were allowed to flow over the beads and the change in absorbance of the fluorescein as it was displaced from the receptor by the analyte was monitored using a CCD (charged coupled device) camera. Principal component analysis (PCA) was then successfully used to determine which analytes were present in the mixture.112, 113 Steed and coworkers have used the trialkylbenzene motif in designing receptors such as 68 and 69 with pyridinium derivatives providing charge-assisted CH· · ·X− hydrogen bonds and/or NH· · ·X− hydrogen bonds, as can be seen in the molecular structure (determined by X-ray crystallography, Figure 24) of the 69·Br− complex. This shows the host in the three-up conformation with both in and out conformations of the binding arms present. 1 H NMR spectroscopic titrations show that compound 68 binds chloride strongly with a binding constant of 82 000 M−1 in CD3 CN/DMSO (v/v 95/5). The high affinity shows the complementarity between the binding cavity and the chloride.117

R N+

R

N+ N+ 68 R = H 69 R = NH2

R

Figure 24 Receptor design for 68 and 69 along with the X-ray structure of 69·Br− .117

Variable temperature (VT) NMR spectroscopy in acetone-d6 revealed the presence of the two-up, one-down conformer, evidenced by the splitting and up-field shifting of the ethyl CH3 resonance as the proton moves into the shielding area of the pyridinium groups for the “down” ethyl group. In and out binding arm fluctuations were also observed. VT NMR spectroscopic experiments with varying amounts of added chloride revealed that binding to 1 equivalent of Cl− switches the system to the symmetric “three-up” conformation, while in the presence of substoichiometric amounts of Cl− the Cl− /PF6 − exchange equilibrium can be observed on the NMR spectroscopic timescale. While such anion exchange equilibria are sometimes slow in macrocyclic systems,124 they are generally fast in podands highlighting the degree of preorganization afforded by the steric crowding in 69.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands A similar receptor (70), which also utilizes NH and CH charge-assisted hydrogen bonding and shows strong binding to chloride, has been designed by Fabbrizzi and coworkers.125 HN N+

N+

N H

N+

HN

70

The incorporation of halogen atoms onto a pyridyl ring can also allow halogen bonding, for example, compound 71.126 The halogen bonding occurs between the positive region of electrostatic potential of the halogen, which is enhanced by the electron-withdrawing cationic pyridinium motif, and the negatively charged anion. When in the threeup conformation, compound 71 is capable of forming a convergent binding cavity with very strong binding to phosphate, PO4 3− (as the sodium salt), with a binding constant of log Ka = 5.6 in water (at pH 12.1). A related receptor, compound 72, derived from a para-iodotetrafluorophenyl binding group, also binds anions strongly using halogen bonds.127 Interestingly, receptor 72 displays a different anion selectivity with chloride (Ka = 1.9 × 104 M−1 in acetone) having a larger binding constant than oxo-anions.

It is likely that the smaller cavity size in 72 is a poor match to oxo-anions and is more complementary to chloride. The binding arm 1,4-diazabicyclo[2.2.2]octane (DABCO, 73) is useful as its derivatives can be both mono- and dicationic, leading to a tri- (74) and hexacationic (75) species and allows for the investigation of the effect of charge on the guest interaction. The strongest binding was found for complementary anions, for example, a tricationic host and a trianionic guest. An interesting example of this charge matching is in the selective binding and precipitation of ferricyanide, an Fe(III) species, over ferrocyanide, an Fe(II) species, with the tricationic host.128 Further development in the trialkylbenzene motif has led to receptor 76 by Schmuck and coworkers, which exhibits exceptionally high binding of tricarboxylates in water. The cationic guanidinium groups create a tricationic receptor which binds carboxylates in a 1 : 1 stoichiometry largely by electrostatic interactions. A binding constant of >105 M−1 was measured for citrate, by UV–vis and fluorescence spectroscopy, with excellent selectivity over monoanionic species.129 NH2

H2N+

NH

O HN O

NH

F N+

I

F O

I

O

F F

NH2

O

O

I

I 71

72 F F N+

N

N

N

N+

N+ N+ N

N+ N+

N+

N+ 73

74

HN H2N

O +NH 2

76

N+ 75

N + N

F

F

N

NH

O2N

O

+NH 2

HN

I

N+

O

HN

F

O F

H N

O

F

N+

O N H

F I

23

NO2

N + N

H H H

N

N +

NO2

77

As an alternative to the more usual hydrogen-bonding groups such as urea, amides, and guanidinium, it is possible to use imidazolium groups to create CH+ · · ·X− hydrogen bonds as in receptor 77.130 The charged nature of this system means that the main interaction is electrostatic. To further enhance the strength of this form of hydrogen bond, an electron withdrawing nitro group can be attached to the imidazolium ring at the C-4 position, enhanced by

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

24

Molecular recognition

the electron-withdrawing nature of the nitro group. Ab initio calculations and 1 H NMR spectroscopic titrations were used to determine the binding constant for halide anions, with both methods concurring to give a selectivity series of Cl− > Br− > I− . The binding constant for Cl− in DMSO-d6 (dimethyl sulfoxide) is 4800 M−1 with a 10-fold selectivity over Br− . In addition, the compound also binds dihydrogen phosphate with a binding constant of 2500 M−1 owing to the strong basicity of dihydrogen phosphate increasing its binding affinity. The principle of using anion binding to shift the equilibrium of conformations of a host has been used by Duan and coworkers to synthesize the chloride-selective sensor 78.131 In this system, anions are bound by chargeassisted hydrogen bonding to the benzimidazolium group with a binding constant of 3.9 × 103 M−1 in DMSO. The 78·3BPh4 complex shows only monomer emission from the naphthyl reporter groups. However, on addition of chloride an excimer band is observed in the emission spectrum (Figure 25). The anion-induced conformational

change brings two naphthyl groups into close contact and allows for excimer formation. Only chloride is able to induce a conformation where this can occur. Building on previous work, Duan and coworkers have developed the ditopic and the tripodal ferrocenyl derivatives 79 and 80.132 Of the anions tested, both receptors bind chloride strongly with moderate selectivity. Differential pulse voltammograms (DPV) show a cathodic shift of 50 mV in the E1/2 value of 79. Smaller cathodic shifts were observed for other anions. No response was observed with the tripodal derivative 80, despite anion 1 H NMR spectroscopic titrations confirming that the receptor does bind anions strongly, in the region of 103 M−1 . Fe N+ N N N+

Fe 79 N+

N+

N

Fe

N N+

N

Fe

N+

N N+

N N N+

78

80

2.2.3 Anion-induced excimer formation in calixarene-based podands

10

Intensity

Fe

5

+

Cl−

+

0 300

400 500 Wavelength (nm)

Figure 25 Emission spectrum of 78 in the absence (dotted line) and presence (solid line) of TBA-Cl. (Reproduced from Ref. 131.  Royal Society of Chemistry, 2005.)

The design of induced-fit anion sensors in not limited to the trialkylbenzene motif. Calixarenes with pendant binding arms have also been explored with pyrene derivatives typically used as the fluorophore. Pyrene excimer emission has been used as a fluorescent reporter which requires close contact between adjacent pyrene molecules for its formation. They are an ideal candidate for induced-fit sensing as conformational changes can dramatically affect intramolecular pyrene–pyrene distances in both the ground and excited states. Kim and coworkers133 have synthesized receptor 81, with two pendant pyrenyl binding arms. The system can shift

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands 200 180 160 I (arbitrary units)

from the dominance of a dynamic excimer (dimer formation in the excited state) in the free host to a static dimer (ground state dimer formation) on addition of fluoride. The binding of the fluoride in the receptor holds the pyrenyl groups in close proximity and allows a ground state dimer to form. This results in a 73-nm red-shift in the excitation spectrum and a 12-nm blue-shift in the excimer emission.

25

140 120 100 80 60 40 20

HN

NH

O

O O

400

450

500 550 l (nm)

600

650

O O

Figure 26 Emission spectrum of 82 on increasing amounts of TBA-Cl. (Reproduced from Ref. 134.  American Chemical Society, 2008.)

O

81

N+ NH

by intramolecular charge transfer (ICT) to the pyridinium moieties resulting in the observation of both pyrene excimer and monomer emission in the presence of excess chloride. This example shows the importance of conformation in this class of receptor. The sensor discriminates the anion which induces a conformation allowing excimer formation to occur and has the appropriate electronic effect on the system, rather than the one that is the most strongly bound.

N H

N+ N+

H N

HN N+

82

Steed and coworkers134 have also synthesized a calix[4] arene-derived receptor 82. The compound is locked into a 1,3-alternate conformation through steric interactions between the mesitylene rings. This creates a ditopic receptor with pyridinium binding groups and a pyrenyl reporter. The receptor binds 2 equivalents of dicarboxylates strongly (K11 > 100 000 M−1 in acetonitrile) with the dicarboxylate capable of spanning across the two binding arms. However, no significant change in the fluorescence emission was observed for this anion. The binding of chloride is an order of magnitude lower than that of dicarboxylates. However, chloride binding alters the conformation of the receptor in such a way as to allow ground state interactions between the pyrenyl groups and promote the formation of an excimer on excitation (Figure 26). Chloride binding also has an electronic effect in that it prevents fluorescence quenching

2.2.4 Induced-fit molecular clip sensors Molecular clips were first developed as hosts for molecular guests in the 1970s by Chen and Whitlock.135 The concept has been further developed by Zimmerman,136 Rebek,137 and Harmata.138 Typically, molecular clips are simple receptors consisting of two binding domains, which when binding molecular guests have generally consisted of aromatic groups tethered by a linker/spacer, which bind the guest through π –π and ion–dipole interactions (Section 3). Preorganization can depend on the rigidity of the receptor and its ability to maintain a convergent binding site. The use of molecular clip receptors in induced-fit anion sensing has been relatively unexplored, even though the design of these receptors appears to be well suited to the task. In general, these systems consist of biaryl units with the addition of anion-binding groups. The relatively rigid nature of the coupled aromatic systems means that the conformational freedom is largely restricted to rotation about the inter-aryl bond. This provides a degree of preorganization to the system. Receptor 83 shows a 2.4-fold increase in emission intensity after the addition of 2.5 equivalents of fluoride, which is attributed to a concept known as conformational restriction, in which one fluoride anion is bound by

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

26

Molecular recognition

both urea moieties, rigidifying the receptor and reducing nonradiative decay. Further addition of F− resulted in a decrease in the intensity. The binding of the second fluoride (one F− bound to each urea group) allows more conformational flexibility and hence increases nonradiative decay.139

(87), for example, which has been studied extensively for pharmaceutical use.83 OCH3

NH

n -Bu NH S NH

NH

86

NH

HN

NH n -Bu

HN

S

HN S HN 83

N H+ HN

S

n -Bu

n -Bu 84

A similar receptor design based on a 2,2 -binaphthalene derivative (84) shows a red-shift in the UV–vis spectrum on addition of fluoride, consistent with planarizing of the binaphthyl groups and conformational restriction, although for this receptor, only small changes were observed in the fluorescence emission.140 Lin and coworkers have developed receptor 85 which consists of naphthyl groups at the end of urea-binding moieties.141 On binding ortho-phthalate, there is a significant increase in emission at 460 nm, which is not observed with meta or para isomers of phthalate. The emission is from the excited state of the product of a photochemical reaction between the naphthyl groups, and it is only ortho-phthalate which induced the conformation necessary for this photochemical reaction to occur.

N H

H N

R1 R1 NH HN+

N H

87

Two primary mechanisms have been suggested for the biological activity of prodigiosins. The first involves the ability of prodigiosins to mediate into-cell transport of HCl, while the alternative involves the coordination of copper and subsequent modification of DNA.83, 142 The transport of HCl into a cell is necessarily intimately linked with anion binding and transport, and so efforts have been made to understand the anion recognition behavior of prodigiosins. Sessler and coworkers have investigated a series of prodigiosin analog compounds 88 and 89–92.83, 143, 144 Both types of compounds bind chloride strongly when protonated. Binding constants in the order of 106 M−1 were observed for 88 and 105 M−1 for 89 measured by isothermal titration calorimetry (ITC) in acetonitrile. These data show that even very simple acyclic pyrrole-derived sensors can bind anions strongly due to the relatively rigid nature of the skeleton, creating a preorganized binding cavity with NH hydrogen-bond donors and electrostatic attraction.

O H N

N H+ HN

NH

N H+ HN

EtO

R4

O 88

R2

R3

89 R1 = Et, R2 = R4 = CH3, R3 = H

O 85

90 R1 = Et, R2 = R3 = R4 = H 91 R1 = Et, R2 = H, = R3 = R4 = OCH3 92 R1 = R2 = R4 = CH3, R3 = H

2.3

Small molecule anion receptors and sensors

2.3.1 Prodigiosins and their analogs Prodigiosins are a class of compounds isolated from the microorganism of Serratia and Streptomyces genus. These naturally occurring pigments are dark red and colonies of the gram-negative bacteria often resemble droplets of blood and have been put forward as the scientific explanation to many apparent miracles.142 More recently, it has been shown that prodigiosins have immunosuppressive and anticancer properties and the compounds are a promising lead in new drug therapies. The structure of prodigiosins consists of a tripyrrolic skeleton (86) as in prodigiosin 25-C

The molecular structure of compound 88 was determined by X-ray crystallography and confirms that chloride is bound by two NH hydrogen bonds and electrostatic interactions in an essentially planar conformation (Figure 27). The molecular structure of 89·Cl− determined by X-ray crystallography shows the chloride bound by three NH hydrogen bonds with the receptor in a slightly twisted conformation (Figure 27). Further modification to the prodigiosin analogs 90–92 also proved to be effective at binding chloride with binding constants in the range of 104 –105 M−1 . The synthesis of tetrapyrrolic receptors such as 93, creating an additional hydrogen-bond donor site, also helps bind chloride effectively (∼105 M−1 ).83, 144

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands

27

are a wide variety of structural motifs which are discussed in the following section.

2.3.2 Small molecule podand receptors

Figure 27 Molecular structure of 88·Cl− and 89·Cl− determined by X-ray diffraction.143

Isophthalamide anion receptors, for example, compound 96, were originally synthesized by Crabtree and coworkers as a binding motif with little rigidity or preorganization. However, reasonable binding affinities with halides were observed in the range of 103 –104 M−1 in dichloromethane. The molecular structure determined by X-ray diffraction shows a distinctly nonplanar binding mode.146, 147

NH

NH HN+

HN

O NH

(a)

Membrane transport studies were also conducted on naturally occurring prodigiosins and their analogs, compounds 88 and 89, and suggest that the anticancer properties of the receptors follow the trend for membrane transport efficiencies, rather than the anion-binding strengths. This result supports the theory of symport transport of H+ /Cl− as the source of the anticancer properties of this class of compound and that large chloride binding constants impair the release of the chloride, that is, the kinetics of the systems are more important than the thermodynamics.143

H N

N H N

HN

N HN O

O

94

O

HN O

B O O

HN

93

N

NH O B O O

95

Prodigiosin mimics based on amidopyrroles synthesized by Gale and coworkers also showed membrane transport of HCl.145 The amidopyyrole derivative 94 with a pendant imidazole group proved to be the most effective membrane transporter. Interestingly compound 95 showed relatively weak anion binding even when protonated, that is, 397 M−1 in acetonitrile, but proved effective at transporting HCl, highlighting the importance of kinetic lability in this process. Prodigiosins have shown that small molecules consisting of a relatively rigid skeletal framework and convergent hydrogen-bond donors can be highly effective at anion binding. Research in this area is highly active and there

96

97

Smith and coworkers synthesized isophthalamide derivatives functionalized with boronate esters, 97.148 The carbonyl group is able to coordinate to the boron atom and increases the preorganization while also increasing the dipole moment, allowing stronger ion–dipole interactions. The NH residue has a greater positive charge and so has a stronger interaction with the acetate. Consequently, compound 97 has an order of magnitude higher binding constant than 96. Gale and coworkers have designed isophthalamide and 2,6-dicarboxyamidopyridine derivatives.86 The addition of two indole motifs provides the molecule with a total of four NH hydrogen-bond donors.149 Compound 98 is highly selective for fluoride (Ka of 1360 M−1 in DMSO) even compared to other basic anions such as acetate (Ka of 250 M−1 in DMSO). It was postulated that the twisted binding conformation observed in the molecular structure (measured by X-ray diffraction, Figure 28) is also found in solution and is more stable than the binding modes possible with other anions.

NH

HN

NH

HN

O

O

98

Figure 28 Compound 98 with the molecular structure of 52·F− determined by X-ray crystallography.149

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

28

Molecular recognition

R

NH O

H N

R N H

HN

N H

102 R = H 103 R = Br

O

Figure 29 Compounds 102 and 103 with the molecular structure of 102·Cl− determined X-ray crystallography.152

99

Modification of the core motif allows for 2,5-diamidopyrrole derivatives, 99, having a additional NH hydrogenbond donor compared with isophthalamide receptors.150 A binding constant of 2500 M−1 for benzoate in acetonitrile is observed with an asymmetric binding mode apparent in the molecular structure measured by X-ray crystallography. ortho-Phenylaminediamine derivatives have also been investigated as anion receptors containing four NH hydrogen-bond donors in the form of urea groups or amide and pyrrole moieties.151 Compounds 100 and 101 are selective for carboxylates with compound 100 showing significantly higher binding constants for carboxylates than 101. It is suggested that the more open binding cleft in 100 is more structurally complementary to carboxylates than 101. The presence of chloro groups in an analogous compound to 100 showed an increase in binding constant from 3210 M−1 for acetate to 8079 M−1 . It is suggested that the increased acidity of the hydrogen atoms in the central ring allows for CH· · ·O hydrogen bonding, effectively preorganizing the receptor into a planar conformation. Indeed, using thiourea derivatives reduces the acetate binding constant by an order of magnitude, as the large size of the sulfur atom results in steric hindrance between the sulfur and phenylene hydrogen atoms, distorting the binding cleft.

acetone). The addition of bromine in the 3 and 8 positions (103) leads to a marked increase in binding strength through their electron-withdrawing effect. Figure 29 shows the molecular structure of the 102·Cl− complex determined using X-ray crystallography. The indolocarbazole derivative 104 is able to bind anions in water in a helical conformation creating a tubular cavity in which multiple NH hydrogen bonds bind the anion. The binding conformation was confirmed using NOESY (nuclear Overhauser effect spectroscopy) NMR experiments with chloride bound strongest (Ka of 65 M−1 in water). While this binding constant is small, it is measured in a highly competitive medium. The use of less competitive solvents on related compound shows strong anion binding.153 CO2−

−O C 2

N H

CO2−

N H H N

H N

−O C 2

104

O

O

O NH HN

NH HN NH

O

HN HN

NH 100

101

It has been shown above that intramolecular interactions can enhance the preorganization of small molecule acyclic receptors.152 As an alternative, increasing structural rigidity using aryl rings can also provide preorganization. This is typified well by the indolocarbazoles. The rigidity of compound 102 is provided by five fused aromatic rings and it contains two convergent NH hydrogen-bond donors. Compound 102 is an effective anion receptor with benzoate and dihydrogen phosphate bound strongly (log Ka of 5.3 and 4.9, respectively, measured by UV–vis titration in

As final examples of acyclic receptors, compounds 105 and 106 are recently published examples of compounds with the potential to extract sulfate from solution through crystallization.154 Sulfate is generally a difficult anion to bind due to its delocalized charge and low basicity. It has important environmental impact particularly as a component of nuclear waste. The receptors designed by Gale and coworkers are designed to provide NH hydrogen bonds from a variety of moieties such as ureas, amide, and pyrrolic groups. The molecular structure of 59·SO4 − determined by X-ray crystallography shows that each SO4 − oxygen is bound by two NH hydrogen bonds, with a binding constant in DMSO/10% H2 O of >104 measured by 1 H NMR spectroscopic titration (Figure 30). Crystallization of 59·SO4 − from DMSO/10% H2 O occurs in as little as 20 min.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands

29

O NH

N H

N H

HN

O

O NH

HN R

R 105 R =

106 R = NH

Figure 30

NH

Compounds 105 and 106 and the molecular structure of 105·SO4 2− determined by X-ray diffraction.154

2.3.3 Small molecule podand anion sensors Small molecule anion sensors have been an active field of research for the last 20 years. Fluorescent sensors have received significant attention and have been reviewed extensively.155–157 They offer advantages such as high sensitivity and simple instrumentation. The incorporation of a fluorophore such as the molecular clip, indolocarbazoles, and indoles described previously not only provide a reporter group but can also provide a structural element to the receptor. +

NH3

+

H3N

N+ H

lone pair, preventing PET, and therefore luminescence is observed. Thermodynamically, this can also be rationalized as the nitrogen lone pair energy is greater than that of the anthracenyl HOMO (highest occupied molecular orbital); therefore, electron transfer is possible. The interaction with HPO4 2− stabilizes the lone pair, so its energy is less than that of the anthracenyl HOMO; therefore, electron transfer is disfavored. This effect is also given the name chelationenhanced fluorescence (CHEF). +

+ + NH3 NH3 +

H3N + H3N +

NH

HN H N

H N

NH 108 107

A luminescent sensor for anions (107) was synthesized by Czarnik and coworkers and is possibly the first example of its type.158 In this system, an anthracenyl reporter and a tertiary amine receptor are linked by a methylene bridge. This sensor works via PET. Under the conditions of the experiment—an aqueous solution at pH 6—all amine groups except the benzylic amine (which is less basic) are protonated. The excitation of an electron to an excited state by a photon leaves an electron hole, into which the benzylic amine lone pair donates an electron. The excited electron cannot radiatively decay back down to the ground state, thereby giving fluorescence, and must relax via a nonradiative method. The fluorescence is therefore quenched.157 Addition of HPO4 2− to 107 leads to complexation, forming three NH· · ·O− hydrogen bonds by the primary amines and one N· · ·HO hydrogen bond to the benzylic hydrogen with a log Ka of 0.82. It is likely that a partial or full proton transfer occurs leading to protonation of the nitrogen

Citrate and sulfate can also give rise to the CHEF effect in 107, although their binding constants are lower. Sulfate by itself does not have any acidic hydrogens, but leads to water dissociation, thereby leading to amine protonation. Similarly, the receptor 108 binds pyrophosphate in an analogous way to the above, with a Ka sufficient to allow micromolar fluorescent sensing. The mechanism of sensing is again a CHEF effect, found on pyrophosphate complexation.159 Gunnlaugsson and coworkers have synthesized a series of 4-amino-1,8-naphthalimide-based receptors 109 and 110.160 These receptors are quenched by acetate, H2 PO4 − , and F− through PET with the greatest quenching observed with F− in DMSO. Usually for 4-amino-1,8-naphthalimide receptors, PET quenching is only observed when binding functionality is in the 4 position. However, in this instance the position of the binding groups does not affect the PET, with quenching found in both the 4-amino and the imide positions.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

30

Molecular recognition

N

O

O H N

HN

H N O

CF3

109 H N

H N O

O

N

CF3

O

NH2 110

The Gunnlaugsson group has also synthesized the simple anion receptor 111 in which the addition of acetate, H2 PO4 − , and F− leads to fluorescence quenching by PET. Chloride, however, results in an increase in fluorescence emission. This was attributed to twisting of the molecule on binding chloride which reduces the efficiency of the pre-existing intramolecular PET process; hence, emission is increased.161

pyrophosphate. A structureless emission band at 475 nm is then observed with quenching of the monomer emission. This is assigned to an excimer emission. Furthermore, significant excimer emission is only seen for pyrophosphate, showing good selectivity over other anions, for example, HPO4 2− , H2 PO4 − , and Cl− .162 Colorimetric sensors are sensors in which binding of an anion results in a visible color change. They are a particularly attractive form of anion sensor as qualitative results can be achieved through “naked eye” sensing which does not require any equipment. There have been many examples of small molecule colorimetric sensors156, 163 with a selection discussed below to give a flavor of the concepts which can be used when designing receptors. Hong and coworkers have combined two chormophores—azophenol and p-nitrophenyl into a receptor design, compound 113.164 Red-shifts were seen in the absorption bands of the receptor on addition of the highly basic anion dihydrogen phosphate to chloroform, resulting in a visible color change from light yellow to violet. Less basic anions such as Br− , Cl− , and HSO4 − do not cause significant color changes; however significantly, anions of similar basicity to those of H2 PO4 − , F− , and acetate do not cause red-shifts to the same extent due to their different binding geometries. NO2

N

CF3

O

N

N H

N H

N

111 H2N +

H N

H2N

S

NH OH HN NH

PPi

O P O

O− O−

P O

O−

O−

N

O2N

NH H + N NH NH H + N N H 112

In addition to intramolecular excimer formation as a means of anion sensing, anion-induced self-assembly systems can also be used to good effect. The pyrenefunctionalized guanidinium receptor 112 demonstrates a fluorescence emission at 376 nm in its monomeric state. On addition of pyrophosphate (PPi ), a 2 : 1 stoichiometry self-assembly system is formed between the receptor and

S

HN 113

NO2

Gunnlaugsson and coworkers have synthesized a range of colorimetric and fluorescent anion sensors derived from a 1,8-naphthalimide chromophore.165 For example, compound 114 proved to bind anions strongly in DMSO by UV–vis spectroscopic titrations (H2 PO4 − , log β = 4.0; F− , log β = 4.4; and acetate, log β = 4.95). A color change from yellow to purple was observed on the addition of acetate, H2 PO4 − , and F− (Figure 31) due to the effect of hydrogen bonding of the anion to the thiourea group on the ICT. Interestingly, this receptor is effective in aqueous solvent mixtures as well as buffered aqueous systems with similar color changes observed. Dipyrrolyl quinoxaline (DPQ)-derived receptors have been developed as colorimetric anion receptors.166 The receptors proved effective at binding F− (Ka = 118 000 M−1

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands O2N

O

N

NO2

O

O N H

31

N H

121

HN S (a)

O

CF3

NH

O

N

N H

(b)

O

O N H

N O

N H

LH–

LH2

114

L2−

122

Figure 31 (a) Compound 114. (b) Compound 114 without acetate (left) and on the addition of acetate (right). (Reproduced from Ref. 165.  American Chemical Society, 2005.)

for F− in CHCl3 for 115). A color change from yellow to blue is observed on addition of F− to the receptors in CHCl3 and DMSO. It is suggested that the color change is due to disruption of the conjugation of the molecule by puckering of the pyrrole moieties on binding F− . DPQ derivatives also have a dual function as fluorescent anion sensors with quenching of fluorescence observed with receptors 115–117. R2

N

N R1

R1 R1

NH HN

R1

115 R1 = R2 = H 116 R1 = H, R2 = NO2 117 R1 = F, R2 = H

Extended DPQ chromophores, compounds 118–120, have been investigated by Anzenbacher and coworkers.167 The effect of using 5,8-aryl substituents on the fluorescence was twofold with a red-shift in the emission maximum and an increase in the quantum yield. These modifications of the DPQ design also lead to an increase in binding affinity toward anions. Fluoride is bound strongly by all hosts (e.g., 118 Ka = 51 300 M−1 in acetonitrile); however, pyrophosphate was bound very strongly (Ka = 93 700 M−1 in acetonitrile). The addition of fluoride or pyrophosphate leads to the appearance of a new absorbance band at 500–550 nm and a decrease in the band at 400–450 nm. Besides a colorimetric response, fluorescence quenching is also observed.

Figure 32 Compound 122 in the presence of TBA-F (L = 122). (Reproduced from Ref. 169.  American Chemical Society, 2005.)

most common methods of producing a colorimetric sensor. The sensing of fluoride itself is a highly topical subject and has been recently reviewed.168 The urea derivative 121 with two electron-withdrawing nitro-substituents deprotonates a single urea NH proton on the addition of fluoride leading to a color change.169 This deprotonation was confirmed by 1 H NMR and crystallographic methods. The drive for the deprotonation is ascribed to the intrinsic acidity of the urea NH, enhanced by electron-withdrawing substituents and the high stability of the HF2 − anion formed after deprotonation. Fabbrizzi and coworkers have synthesized a naphthalimide-substituted urea capable of double deprotonation.169, 170 The addition of TBA-F to 122 in DMSO leads to a yellow to red color change after the addition of a few equivalents of anion, and on further addition, a second deprotonation step occurs leading to a blue coloration. This process can be monitored by UV–vis spectroscopy with the emergence of a new band at 540 nm for the single deprotonated species and a decrease in the free host band at 400 nm. With further addition of F− , a new band at 600 nm forms corresponding to the doubly deprotonated species, with a decrease in the band intensity at 540 nm (Figure 32). Isosbestic points are observed for all new bands showing a clear transition between species. Carboxylates such as acetate also lead to a similar effect. The use of electron-withdrawing substituents on pyrrole 2,5-diamides, for example, 123 synthesized by Gale and coworkers,171 also leads to deprotonation of a urea NH with a concurrent color change from yellow to blue due to O2N

R

NO2

R N

N

118 R = H 119 R = OCH3 120 R = N(CH3)2

NH

O2N

H N

O

HN O

NO2

NH HN

The phenomenon of urea deprotonation has been used extensively in the detection of fluoride and is one of the

123

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

32

Molecular recognition

charge transfer from the deprotonated nitrogen atom to the nitrophenyl moiety.

2.4

Pt2+ N

O N H

Metal and Lewis acid-derived podand receptors

4. 5.

a substitutionally inert metal is used in a structural role; a Lewis acidic metal ion forms part of the binding site; self-assembled complexes of substitutionally labile metals involving thermodynamic anion templation; anion-binding solid-state coordination polymers; metal-based redox, colorimetric, or luminescence reporter groups.

There is often a large degree of overlap between classes; for example, it is possible for a metal to perform both a structural role and a reporter group role in the same molecule.

2.4.1 Metals as structural elements By using substitutionally inert metals such a Ru(II), a low spin d 6 metal which when bonded to hard donors results in a large crystal field stabilization energy (CFSE), the metal can perform a structural role akin to that of the organic cholic acid or trialkylbenzene motifs. The receptors 124–126 developed by Loeb and coworkers consist of a tetra-substituted Pt(II) complex.173–175 Pyridyl amide derivatives are used as ligands to the metal and as anion-binding groups. This type of receptor can exist in many conformations, for example, a cone, partial cone, 1,2-alternate, and 1,3-alternate, analogous to that of calixarenes, with a 1,2-alternate conformation found in the solid state with PF6 − .173 The use of isoquinoline as a progression from pyridine resulted in an interesting anion selectivity. When bound to halides, complex 125 binds two anions strongly in DMSO, measured using 1 H NMR spectroscopic titrations. A 1,3-alternate conformation is found in the solid state, with the receptor behaving in a ditopic manner. The anion is bound with NH and CH hydrogen bonds as well as an electrostatic contribution from the Pt(II). In contrast, with H2 PO4 − and SO4 2− , a cone conformation is observed in the solid state with a 1 : 1 host:guest stoichiometry (Figure 33).174 The use of pyrrolylpyridine ligands in complex 126 allows for competition between binding groups by allowing a choice between NH and CH hydrogen-bond donors.

4

124

The use of metals as part of anion receptors is well established and the versatility of metal ion coordination chemistry has led to their varied use in receptor design, and we covered some examples in Section 1.7. In general, there are five major classes of metallic receptors:172 1. 2. 3.

N H

N

Pt2+ O

N H

N H

4

125 Pt2+ N H N 4

126

Figure 33 Compounds 124, 125, and 126 and the molecular structure of 125·SO4 + determined by X-ray crystallography.174

For 126 with anions such as Cl− , HSO4 − , and NO3 − , predominant downfield shifts were observed for the CH proton in DMSO, while for more basic anions such as acetate, NH downfield shifts were observed with acetate binding in a 1 : 2 host:guest stoichiometry. Binding in nitromethane shows significant downfield shifts in the NH proton only. It is suggested that the strong hydrogen-bond acceptor properties of the DMSO compete with the anion to bind with the NH, with only basic anions able to compete. The poor hydrogen-bond acceptor ability of nitromethane means that there is less competition and therefore the NH can interact with the anion.175 Steed and coworkers have synthesized a series of (arene)ruthenium(II) complexes such as complexes 127 and the analogous complex formed with the ligand 129.176 Receptors 127 and 128 have a low number of equivalent protons in the NMR spectrum suggestive of low symmetry in solution. Anions are bound strongly in acetonitrile with a N H N

Ru+ N N Cl

HN

R

N

NH

129

R

127 R = H 128 R = NO2

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands 1 : 1 and 2 : 1 host:guest stoichiometry observed. Interestingly, on addition of strongly bound anions such as chloride, the methylene protons collapse into a singlet suggestive of a more symmetric species and this is attributed to the loss of the ligated chloride to form a 16-electron species which would provide the required symmetry. Inclusion of a carbazole fluorophore, as in the complex derived from ligand 129, can create a fluorescent sensor, whereby the binding of chloride results in charge transfer from the chloride to a Ru-pyridyl-centered orbital which quenches the fluorescence emission. Complexes 127–129 are on the border line of substitutionally stable and labile complexes with prolonged exposure to high equivalents of chloride leading to the displacement of a ligand and direct Ru–Cl complexation. This is on the timescale of hours to days, and the ligand exchange is slow on an experimental timescale. Gale and coworkers have designed receptor 130 which is an interesting example of tuning preorganization, with Pt forming an integral structural role in preorganizing the receptor. Receptor 130 binds anions weakly in a DMSOd6 —0.5% water solution with H2 PO4 − having an affinity constant of 90 M−1 due to flexibility around the aryl–aryl bond of the bipyridine. However, when receptor 130 is reacted with PtCl2 (DMSO)2 , to form the Pt complex, 131, the H2 PO4 − binding constant is increased to 3644 M−1 as the binding groups are forced into a syn arrangement forming a preorganized, convergent binding cavity. The receptor also proved effective as a colorimetric sensor for F− , through deprotonation of a urea NH group. A color change from yellow to purple is observed after deprotonation.177 Cl

Cl Pt

N

N

O

N

O NH

HN

N

O

N

NH HN

N N

H N

N O

O N N

N

O N H

N H

H N

H N

Ot Bu

Ru2+ N

N N

Ot Bu O

O 133

Stable transition metal pyrazole complexes using Mn (134), Re (135), and Mo (136) have been synthesized by P´erez and coworkers.179–181 To create a convergent binding cavity, the metal fragment was functionalized with CO groups which prefer to adopt a fac arrangement of the CO groups to maximize back bonding. The rhenium complex proved to be substitutionally inert due to its d 6 configuration; however, a pyrazole ligand was displaced from the Mn and Mo complexes in the presence of anions. Hydrogen-bond formation between anions and the NH groups of the pyrazole is observed by 1 H NMR spectroscopic titrations as well as in the solid state. The solid-state structures reveal an unfavorable deformation in the N–Re–N bond angles of 135 when binding anions when compared to the tetrahedral B and Zn derivatives. Interestingly, HSO4 − leads to protonation of the pyrazole unit and binding of a sulfate anion. Fluoride, on the other hand, deprotonates NH groups of a pyrazole. CO N N (OC)3M+

H

N N

CO

Mo+ N N N N N N H H H

H

131

Beer and coworkers have developed Ru(Bpy)3 2+ derivatives which are intrinsically chiral due to the helicity of the Ru(Bpy)3 2+ but can also allow the incorporation of additional chiral functionality, for example, 132 and 133.178 Each receptor was isolated in an enantiomerically pure form and binds chiral anions such as N-Cbz-Glu and lactate in DMSO determined via 1 H NMR spectroscopic titrations. However, in all cases enantiomeric selectivity could not be achieved or was low.

H CH3 Ph

132

N N H 130

CH3

Ru2+ N

HN

NH HN

H N H

O NH

Ph

O N

33

134 M = Mn 135 M = Re

136

The P´erez group has also synthesized a range of d 6 transition metal-derived anion receptors based on rhenium and ruthenium, for example, complexes 137182 and 138.183 Compound 137 contains a bidentate pyrazolylamidino ligand which provides one NH hydrogen-bond donor and an additional pyrazolyl NH donor. The complex binds chloride particularly strongly because of its rigid nature, with

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

34

Molecular recognition

a binding constant of 8725 M−1 in CD3 CN measured by 1 H NMR spectroscopic titrations. CO OC OC

Re+

N N

N N

N H N NH

Ru+

Cl

N

Zn+

NH

Cl

H

N N H N N H

N

Cl−

NH 137

138

Receptor 138 utilized biimidazole as the anion-binding group. In this system, the metal center effectively preorganizes the biimidazole to bind anions by preventing rotation around the aryl–aryl bond and by also preventing selfassociation (Figure 34). The receptor binds anions strongly in CD3 CN, for example, Ka = 5920 M−1 for HSO4 − .

139

Figure 35 Compound 139 and the crystal structure of 139 determined by X-ray diffraction.184

2.4.2 Labile metal-derived anion receptors N

The use of labile metal atoms, that is, those in which ligand association and disassociation are fast on an experimental timescale, has also been investigated, where the receptor– anion complex is essentially self-assembled in solution as the thermodynamic product. Halcrow has designed a receptor involving the d 10 Zn2+ metal and a pyrazole derivative (139).184 This receptor is analogous to the systems P´erez described previously; however, the lability of these ligands with the Zn(II) metal means the complex, while being thermodynamically stable, is not kinetically stable in solution. The solid-state structure shows the assembled complex with a tetrahedral Zn2+ and three 3(5)t-butylpyrazole groups (Figure 35). The pyrazole derivative hydrogen bonds to the chloride of an adjacent complex in the crystal, forming a hydrogenbonding polymer. Stable complexes of this type utilizing

O N H

N H 140

Figure 36 Compound 140 and molecular structure of [Ag(140)2 (NO3 )(HOCH3 )] complex. Only one complex in the asymmetric unit is shown for clarity.186

covalent bonding can be synthesized using a triprotonated trispyrazolylborate dication and has been shown to bind chloride in the solid state.185 Metal complexes of pyridylurea ligands and silver salts have been investigated by Steed and coworkers.186 In this work, the ligand self-assembles with a Ag+ cation to form an [Ag(140)2 ]+ species. This is able to bind anions such as nitrate strongly in a 1 : 1 and 1 : 2 host:guest stoichiometry, measured by 1 H NMR spectroscopic titration (K11 = 30 200 M−1 , K12 = 2900 M−1 in CD3 CN), as well as a 1 : 3 stoichiometry due to ligation of nitrate to the silver center (K13 = 550 M−1 ). A solid-state structure of the [Ag(140)2 (NO3 )] was determined by X-ray crystallography and is shown in Figure 36. The structure shows a convergent binding site with the nitrate bound by NH and CH hydrogen bonds.

2.4.3 Lewis acid-metal-based receptors

Figure 34 Molecular structure of 138·NO3 − determined by Xray diffraction.183

Metals that act as Lewis acids can increase the effectiveness of a receptor by directly interacting with an anion, polarizing a hydrogen-bond donor, hence increasing its acidity, providing a structural element, or more usually a combination of both. Direct coordination of anions to metal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands complexes is exemplified by cascade and related complexes as discussed in Section 1.6 and there is a history of metal centers such as tin187 and mercury188 being used in anion coordination chemistry. Rudkevich and coworkers have synthesized a range of cascade-type anion receptors containing a UO2 fragment as a Lewis acid center. Compounds such as 141 bind anions strongly, particularly H2 PO4 − , measured via a conductance method (Ka > 105 M−1 in CH3 CN:DMSO v/v 99 : 1) and a selectivity over Cl− of 102 . The molecular structure of 141·2H2 PO4 − shows bond formation between the dihydrogen phosphate O and the UO2 center, that is, a Lewis acid interaction. 1 H NMR spectroscopic titrations and the molecular structure also confirm hydrogen-bond formation between the amide functionality and the anion.80

N UO2 O O

N

O

Beer and coworkers have also described five mechanisms of sensing using redox groups1 : 1. 2. 3. 4. 5.

a through-space interaction with the anion-binding site in close proximity to the redox center; direct coordination of the anion to the redox center; a through-bond interaction through conjugation between binding site and redox center; an anion-induced conformational change which leads to a perturbation in the redox center; an interference mechanism whereby the interaction between several redox active centers is affected by anion binding.

Beer190 has synthesized a large library of electrochemical sensors. The first cobaltocene sensor devised was compound 143 and its binding properties were investigated using Br− .191 Binding is via electrostatic interactions and the cobaltocene Cp2 Co+ /Cp2 Co redox couple displays a cathodic shift, that is, the anion increases the cobaltocene reduction potential.

O O

NHR RHN O

O

O O

O

O

O

Co+

Co+

141 R = 4-CH3C6H4

Receptor 142 is capable of self-organizing in the presence of NaX (where X is a halide) into dimeric structure held together by Lewis acid interactions between two crown ether moieties and the Na. This allows a binding cleft to be formed between the urea groups, which is able to bind anions, for example, chloride as shown in the crystal structure determined by X-ray crystallography (Figure 37).189

O

O 143

It is also possible to add hydrogen-bonding functionality to this class of sensors, for example, 144192 and 145.193, 194 1 H NMR spectroscopic titrations of compounds 144 and 145 reveal the highest binding affinity of H2 PO4 − . As would be expected, the formation of macrocyclic systems OCH3

2.4.4 Metals as sensing units

O

Electrochemical-based anion receptors have been a versatile means of sensing since the pioneering work of Beer and coworkers in the late 1980s, principally using cyclic voltammetry. A range of redox active moieties, including cobaltocene, ferrocene, and Ru(bpy)3 2+ derivatives, have been incorporated into these receptors.

N H H N

Co+

O

O 144

H

H Co+

N

O

Co+

OCH3 OCH3

N

N

OCH3

145

142 O

O

O

O O

Figure 37

35

O

N H

N H

Compound 97 and the crystal structure of the 142·NaCl complex, determined by X-ray crystallography.189

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

N H

O

Co+

36

Molecular recognition

Co+

Co+ Co+

Co+ HN

NH

O

O

O

HO

O

O

O O

OH

HN

NH

O

O

O

OH OH

S

S

O O

CH3 CH3 146

CH3

CH3

Co+

Co+

O O

147

HN

NH

OH O

O

O

O

S

CH3

CH3

HO O O S O

OH O O S O

HO O

S

O

O NH NH

O

CH3

CH3

= Co+

148

149

orientating the amide groups for carboxylate hydrogen-bond formation. Ferrocene derivatives have also been studied for electrochemical sensing. However, as the sensor is neutral there is no intrinsic electrostatic interaction unless the ferrocene is oxidized to ferrocinium, in which case the receptor becomes cationic and binding is increased.83 Receptors 150 and 151 contain a mixture of hydrogenbond acceptor and donor groups.197 These receptors are able to selectively bind dihydrogen phosphate in acetonitrile because of its binding group complementarity and high basicity. Large cathodic shifts in the redox potential were observed in the presence of excess H2 PO4 − (150, 120 mV, 151, 240 mV in acetonitrile), although 151 exhibits irreversible oxidation. Receptor 150 was able to sense H2 PO4 − even in the presence of competing anions such as Cl−

analogous to 101 increases the binding constants significantly, that is a 10-fold increase. Calixarene-cobaltocene sensors, for example, compounds 146–149 have also been developed.195, 196 An interesting aspect of these sensors is that varying the topology of the receptor allows for the selective sensing of a specific anion. For example, the topology of 146 favors Cl− determined via NMR spectroscopic titrations. Further functionalization of the lower rim with tosylate allows tuning of the binding properties. For example, when the tosyl groups were para to the cobaltocenium (147), H2 PO4 − binding was favored because of the tosyl groups forcing the cobaltocenium groups together. However, when ortho, 148, the tosyl groups force the cobaltocenium groups slightly apart, favoring Cl− . Finally, compound 149 is preorganized for carboxylate binding, because of the bridging cobaltocenium O

Fe

O

O N H H N

N N

NH2 NH2

Fe

N H

N

NH2 Fe

N H

NH

O 150

151

152

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc058

Podands and HSO4 − . Tripodal and calixarene derivatives have been synthesized. All display large cathodic shifts in the redox potentials. The receptor 152 is interesting as it is difunctional.197 Typically in neutral hydrogen-bonding receptors, H2 PO4 − binds more strongly than HSO4 − as its higher basicity forms stronger hydrogen bonds. In this receptor, however, two binding modes are possible; the first operates for nonacidic guests and the receptors act as the hydrogenbond donor from the amide. The second mode applies to acidic guests in which proton transfer occurs, allowing electrostatic and hydrogen-bonding interactions with the guest. In the case of H2 PO4 − , it does not fit well into either category and therefore is not bound strongly. HSO4 − , however, fits well into the second binding mode and is bound strongly, giving rise to this unusual selectivity. The electrochemical behavior of the receptor on addition of HSO4 − showed a new oxidation peak, cathodically shifted by 220 mV from the free receptor, showing that anion binding greatly increases the ease of oxidation, although the receptor showed irreversible behavior. Gale and coworkers198 have appended the ferrocene moiety onto the 2,5-diamidopyrrole core described previously. Receptor 153 showed a large anodic shift in the redox potential of −130 mV with F− in dichloromethane. This compares to only −75 mV for chloride, and this selectivity corresponds well with binding constants measured by NMR techniques (Ka = 170 M−1 for H2 PO4 − and 90% 3b: X = Br, >90% 3c: X = NO2, >90% 3d: X = CONH2, ~30%

N

2a-d (35 equivalents) pH 9, 4 °C, 2 h

(b)

1 Y139 on TMV coat protein

3a-d

N

OR O

CH3

300 nm HO

H2NOR

N N

4a-e (8 mM)

5a: R = Bn

3a pH 6, rt, 2 h

5b: R = alyl

HN O H

H N

5c: R =

5a-e

N H

O

4

NH H S

(c) (a) N O

O 1

O N H

(d)

OH

HN

1: TMV Glu 97, 106

R1

2

HNR R (7a-g) EDC/HOBt Conditions i or ii pH 7.4

O N H

O

3

HN

N R2

HN 8a-g

+

O N H

N

O

HN 9

Figure 2 Exterior and interior surface modification of TMV: (a) Schematic of TMV detailing the location of chemical modifications. Yellow—signifies the position of Tyr139 for exterior modification. Red (Glu97) and blue (Glu106)—signify the position of the glutamic acid residues targeted for interior functionalization. (b and c) Exterior modification of TMV. (b) Tyr139 was functionalized by reaction with diazonium salts to form the corresponding azo adducts. (c) The resulting ketone displayed by 3a was readily targeted for further conjugation by reaction with alkoxyamines. (d) Interior modification of TMV. Glu97 and Glu106 were targeted for functionalization via the formation of amide bonds after chemical activation by a combination of EDC and HOBt. (Reproduced with permission from Ref. 4.  American Chemical Society, 2005.)

The interior surface of TMV coat proteins was altered by targeting Glu97 and 106 for the formation of amide bonds to be modified by additional chemical reactions. As seen in Figure 2(d), activated glutamic acid residues of intact wild-type (WT) TMV were reacted with various amine substrates, including bulky dye molecules, resulting in 21–86% of the coat protein monomers being singly or doubly modified with no change in the assembled structure of the virus. The preservation of the ability of the interior-functionalized coat proteins to self-assemble was successfully tested by disassembling intact interiormodified TMV and reassembling the modified coat protein monomers into higher ordered structures upon introduction into specific pH and ionic strength buffers. These findings suggest that the addition of small molecules to the interior of the virus particles does not provide enough steric hindrance to prevent interior-modified coat protein monomers from self-assembling into coat protein oligomers. Furthermore, dual-modified virus particles were synthesized by chemically conjugating rhodamine to the interior of exterior surface azo-modified TMV. In summary, the advantage of using TMV as a biotemplate for

nanomaterial synthesis has been expanded by the chemical modification of the interior and/or exterior of the virus. In addition to developing a strategy to chemically modify the surface of TMV with small molecules, Francis and coworkers have further exploited the ability of TMV to display evenly spaced chemical functionalities on the capsid surface of the virus to synthetically construct a light-harvesting system.22 The rodlike assemblies of genetically modified TMV coat proteins, containing an inserted cysteine residue displayed on the exterior of the capsid for thiol functionality, were used as a biological scaffold for the spatially selective arrangement of chromophores. Coat proteins modified with dye molecules maintained their self-assembly properties by forming higher order protein oligomers under the appropriate conditions. By anchoring acceptor and donor chromophores at distinct intervals spaced along the surface of the capsid, controlled by the reaction of thiol reactive dye molecules with the precisely spaced Cys residues, the phenomenon of fluorescence resonance energy transfer (FRET) was observed. Rods of coat protein only displaying acceptor and donor dyes exhibited

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

Viruses as self-assembled templates a minimal transfer of energy. By incorporating an additional dye molecule into the system as an intermediary chromophore, the ability of the system to transfer energy from acceptor to donor increased to >90%. In summary, Francis and coworkers have demonstrated the exploitation of the rigid structure of TMV with precisely positioned sites for the chemical conjugation of a broad range of chromophores to construct artificial light-harvesting systems. Expanding on the work by Francis and coworkers to target Tyr residues on the exterior of TMV for chemical conjugation, Wang and coworkers have utilized coppercatalyzed azide-alkyne 1,3-dipolar cycloaddition, or “click chemistry,” in combination with the diazonium coupling

to modify the surface of the virus as shown in Figure 3.5 Implementing click chemistry to further functionalize the surface of the azo adduct of TMV may broaden the range of potential chemical structures of starting materials, which can be incorporated onto the surface of the virus. Using this methodology, it was possible to covalently attach small molecules, peptides, and polymers to the exterior surface of TMV. Most notably, this approach allows for the display of an azide functionality so that a second click chemistry reaction can be carried out, which significantly improves the pool of potential molecules that can be utilized to derivatize TMV. As an application of this approach, the authors demonstrated that TMV modified via click chemistry to display moieties intended to either inhibit or

TM V

TM V

pH = 9.0, 4 °C, 2 h

OH

OH 2130

+

2130

5

N2

N N 2

H3O+

NaNO2

pH = 7.8 R1-N3 (a – f) 4 °C, 24 h CuSO4, NaAsc

H2N 1 (g – i)

R2

TM V

TM V

CuSO4, NaAsc

OH (for 3f)

2130

OH

2130

N N

N N N N N

R2 3g–3i

N N N

3a – 3f

N N N

HO O

OH

O

HO

N3

O

N3 a

R1

b

O

O

H N

N H

O

O

H N

N H

O

O

N3

N H c

HO HN

O N3

O

O− P

O

N+

+

O

O

3

d

N3 f

N3 n

O

COOH

O O

N H

O

N O

g

NH

O H3N

O

e O

N3

HO

H N

HN h

O

O

i

O

Figure 3 Schematic illustrating the functionalization of TMV by click chemistry. (Reproduced with permission from Ref. 5.  Wiley-VCH, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

6

Self-processes

promote cellular growth exhibited their predicted properties on cellular metabolism. In conclusion, the disadvantages of targeting tyrosine residues on TMV exclusively with diazonium salts to incorporate uniformly spaced functionalities have been overcome by combining this approach with copper-catalyzed azide-alkyne 1,3-dipolar cycloaddition to expand the range of small molecules, peptides, or polymers available, which can be covalently coupled to the surface of TMV. Wang and coworkers have also exploited the precise, rodlike structure of TMV and employed chemical modifications to synthesize conductive polyaniline (PANi) nanowires using the virus particles as a biotemplate.23 As depicted in Figure 4, the combination of TMV, polyaniline, poly(sulfonated styrene), and ammonium persulfate resulted in the formation of biotemplated nanofibers, the length of which was dependent upon the length of the virus. Adjusting the pH influenced the arrangement of polymerized polyaniline on the virus to assemble either nonconductive, branched oligoaniline or conductive PANi nanowires. Initial preparation of the TMV-templated PANi nanowires at neutral pH resulted in the arrangement of viral particles end-to-end to form nanofibers longer than the 300 nm length of the TMV. The electronic properties of these longer TMV/PANi nanowires were probed to yield a conductivity measurement of ∼1 × 10−5 −1 cm−1 . In conclusion, Wang and coworkers have demonstrated the construction of water-soluble, conductive PANi nanowires of controllable length using native TMV as a biotemplate in conjunction with a facile ordered assembly and polymerization process. These conductive PANi nanofibers have the potential to be implemented in optical, electronic, or sensing applications.

PS ,A SS P e, ilin An 4 pH An

TMV

pH

ilin 6.

e,

AP

Conductive short rod

S

5

PANi/TMV long fiber

Aniline, PSS, APS pH 4

Conductive composite long fiber

Figure 4 Schematic illustrating the assembly of TMVtemplated polyaniline fibers. (Reproduced with permission from Ref. 23.  American Chemical Society, 2007.)

2.3

Metal or metal oxide formation with TMV templates

The exterior coat protein surface of TMV displays various charged amino acid side chains at uniform intervals, which could serve as potential nucleation sites for the controlled layering of inorganic materials to synthesize biotemplated nanomaterials. Exploiting the evenly spaced functionalities on the surface of TMV, in conjunction with the robust stability of the virus, Mann and coworkers were the first to utilize TMV as a template for the synthesis of metallic nanoparticles.24 By incubating native TMV with metal salt solutions and hydrogen sulfide, an evenly dispersed coating of CdS or PbS nanocrystals were observed on the exterior of the virus, yielding hollow nanotube complexes of metallic nanoparticles. Additionally, at higher pH, the viral-templated deposition of iron oxide on the exterior of the TMV was achieved. In both instances, metallic nanoparticle synthesis was restricted to the exterior surface of the TMV because of the preferential electrostatic interaction between the metal salt precursor and the charged functionalities displayed on the exterior surface of the virus. Building upon this initial study, Mann and coworkers further explored various conditions for the preferential synthesis of metallic nanoparticles on the exterior or interior of the TMV.25 By adjusting the pH of the reaction conditions and relying on the difference in pKa s of the exposed amino acid functionalities within the interior RNAbinding channel or the exterior of the virus surface, the interior or exterior electrostatic charge was chemically and selectively tuned to discriminate between interior or exterior viral-templated metal nanoparticle deposition. Utilizing this strategy, at acidic pH where amino acid side chains on the surface of TMV are positively charged, the formation of discrete platinum and gold nanoparticles on the exterior surface of native TMV could be achieved from reaction with negatively charged metal salt precursor as seen in Figure 5(a). Similarly, at neutral or slightly alkaline pH, silver nanoparticles were observed in the interior of native TMV owing to the preferential interaction between the negatively charged inner channel of the virus and silver cation precursor as visualized in Figure 5(b). Kern and coworkers have also explored controllable, uniform metal deposition by employing TMV as a biotemplate. In their studies, the addressable amino acid side chains that function as metal nucleation sites are first activated with a metal complex, followed by incubation with the desired metal for metallization and its subsequent chemical reduction to form metal nanoparticles.11, 12 By employing this approach of electroless deposition, discrete nickel, silver, and cobalt nanoclusters were selectively formed on the interior or exterior of the virus by controlling the pH and the activation complex used in the reaction. In addition

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

Viruses as self-assembled templates

50 nm (a)

(b)

Figure 5 TEM images of metallized TMV: (a) TEM image of TMV selectively coated with discrete gold nanoparticles. The viral-templated nanoparticles with diameters of 8.6 ± 3 nm were formed by incubating TMV with HAuCl4 at pH 2.9 for 30 min by stirring in the dark followed by reduction with hydrazine at pH 2.3. (b) TEM image negatively stained with uranyl acetate of silver nanoparticles selectively formed within the RNA-binding channel of the TMV. To form silver nanoparticles of diameters measuring 6.5. By subsequently lowering the pH, the structure of the capsid was altered as the pores in the viral protein shell closed and the crystallization of either paratungstate or decavanadate was achieved within the interior of the viral particle. The resulting metallic nanoparticles were only observed in the presence of CCMV and exhibited the same monodisperse diameter as the viral particles seen in Figure 10. The nucleation of the metallic crystals was presumed to be induced by the electrostatic attraction between the negatively charged precursor metal ions and the positively charged amino acid side chains displayed within the interior of the viral capsid at the Nterminus of each coat protein. In another study, Douglas and coworkers employed the same strategy to synthesize titania nanoparticles as well as amorphous titania.43 Using anionic titanium salts in conjunction with the pH-dependent gating mechanism of CCMV, a composite of β-TiO2 was formed, which possessed the same diameter of the viral capsid and demonstrated photocatalytic activity in the presence of methylene blue. To further expand the range of materials able to be synthesized within the interior of CCMV, Douglas and coworkers engineered a CCMV mutant, which highlighted the importance of electrostatic interactions in facilitating mineralization of inorganic ions when using a viral template.6 In this work, the N-terminus of each coat protein subunit was genetically altered to eliminate the positive electrostatic charge displayed within the cavity of CCMV by replacing six arginine residues and three lysine residues with negatively charged glutamic acid residues at the N-terminus of each coat protein. The genetic modification of the coat proteins exhibited no deleterious effects on their ability to self-assemble, permitting the mutated CCMV virus particles to retain their structural integrity. The newly established electrostatic charge within the interior of the mutated CCMV capsids, as compared to native CCMV capsids, permitted the nucleation of positively charged iron(II) precursor ions enclosed within the capsid, facilitating the formation of iron oxide nanoparticles by exploiting the pH-dependent gating mechanism of CCMV. Collectively, the work by Trevor Douglas, Mark Young and coworkers demonstrates the ability to utilize an additional icosahedral virus for the encapsulation of reagents to spatially control inorganic nanomaterial synthesis within the cavity of a viral template. More importantly, this series of work illustrates an elegant means to exploit the in-depth understanding of the physicochemical properties as well as highly controlled dimensions of viral nanoparticles for the

(a)

100 nm

(b)

(c)

Figure 10 Formation of mineralized paratungstate within CCMV: (a) Schematic depicting the strategy for mineralizing within the interior of CCMV. In step I, the genetic material is removed from the virus and the empty virus capsids are subsequently isolated and purified by ultracentrifugation in sucrose gradients. In the next step, step II, the interior of the empty viral capsids are selectively mineralized by exploiting the pHdependent gating mechanism of CCMV by incubating empty virons with molecular tungstate ions (WO4 2− ) at pH 6.5. (b) TEM of unstained CCMV capsids containing mineralized paratungstate. The electron-rich core of the mineralized viral cages is observed with the size of the paratungstate nanoparticles being restricted to dimensions of CCMV capsids. (c) Negatively stained TEM image of paratungstate-mineralized CCMV capsids. Observed is the uncompromised protein shell of the virus encapsulating the paratungstate nanoparticles. Scale bar is the same as (b). (Reproduced with permission from Ref. 1.  Nature Publishing Group, 1998.)

synthesis of inorganic nanomaterials with precise sizes and functionalities. Intact CCMV or its coat proteins have also been employed as viral building blocks for the self-assembly of molecular materials by Nolte, Cornelissen and coworkers. For example, the coat proteins of CCMV were utilized to form DNA-templated tubular assemblies of viral coat proteins.44 As illustrated in Figure 11, the templates consisted of a single-strand DNA (ssDNA) of repeating thymine units of various lengths (Tq, where q is the number of thymine bases present). To form the negatively

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

Viruses as self-assembled templates

13

H H N N N R1

N H N

+

H

G1

H H N

Tq

Guest (G)

N

Tq -G

N N

R2

H N H HO

−

(a)

O q-1 O P O O

Tq

O

O

O H

N

H N

H

O

G2

H

N

R2

N N N

O

O N O

H N

CN R2

H

N H

G3

R1, R2 =

O

O

O

O

O X

X = H (R1), CH3 (R2)

Tq-G

Tq

T = 1 capsid (b)

CP pH 7.5

Tubular assemblies + capsids

Figure 11 Schematic depiction of the fabrication and self-assembly of CCMV-templated tubular structures: (a) ssDNA comprised of various lengths of repeating thymine groups, designated Tq where q depicts the number of thymine units equaling 5, 10, 15, 20, 25, 30, or 40, interacted with one of three different organic guest molecules, each possessing a diaminotriazine group, to form a rigid, helical, negatively charged DNA-hybrid complex (Tq –G). (b) When the Tq –G complex was incubated with CCMV coat proteins, self-assembled tubular structures were observed. When only Tq was incubated with CCMV coat proteins, intact viral capsids were observed, suggesting that the Tq –G complex is necessary to template the formation of the tubes. Since the exact interactions between the Tq –G complexes and the viral coat proteins to form the tubular structures are yet to be elucidated, their schematic arrangement depicted in (b) is only speculative. (Reproduced with permission from Ref. 44.  Wiley-VCH, 2010.)

charged template for assembly of the tubular structures, one of three organic molecules tested (designated G for guest molecules), each displaying a diaminotriazine functionality, was stably bound to the ssDNA via hydrogen bonding and hydrophobic interactions yielding a helical DNA-hybrid complex. The DNA–guest molecule complexes, when in the presence of CCMV coat protein at pH 7.5 and low salt concentration, were found to template the self-assembly of tubular structures. Nolte, Cornelissen and coworkers postulate that the formation of the observed tubular structures is due to the interaction of the positively charged N-terminus of CCMV coat proteins with the negatively charged phosphate groups exposed in the rigid ssDNAhybrid complexes. As another example of implementing viruses as the basis for construction of hierarchical biological assemblies, Nolte, Cornelissen and coworkers have used dendrons to

link individual CCMV particles together to fabricate a highly ordered assembly.45 To accomplish this, the authors exploited the electrostatic interactions between dendrons and CCMV. Dendrons were fabricated with various generations of polyamine functionalities, which are positively charged at physiological conditions, in addition to a photocleavable o-nitrobenzyl linker between the polyamine functionalities and the base of the dendron. These polycationic moieties electrostatically interacted with the negatively charged exterior protein shell of CCMV to link together the virus particles and create a hierarchical assembly. Importantly, the size of the assembly was found to be dependent upon the concentration and generation of the dendrons present. To disassemble the linked virus particles, the highly ordered assembly of CCMV and dendrons was exposed to UV light to photochemically degrade the dendrons which resulted in the reversibility of the assembled virus particles.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

14

Self-processes

The ability to photochemically release the viruses from their self-assembled configuration allows for spatial and temporal control of the disassembly process. In addition, the general versatility of this strategy was demonstrated by using dendrons to link and assemble the protein cage magnetoferritin, which may be advantageous for its magnetic properties. As an application for this hierarchical biological assembly, the polycationic nature of the dendrimers may also facilitate the uptake of CCMV or other protein cages encapsulating therapeutic entities in vivo.

4

P9 DNA

P6

P8

P7

(a)

FILAMENTOUS VIRUSES

While most of the plant viruses discussed in the previous two sections are rod or icosahedral shaped, bacterial viruses come in various shapes including spheres, satellites, and filaments. Owing to the rapid turnover cycle (i.e., infection, replication, and recovery with fast-growing bacterial hosts) and the seminal work of phage display by Smith,46 a combinatorial affinity screening (aka biopanning) procedure using the filamentous M13 bacteriophage has revolutionized the field of biotemplate-based materials synthesis. While such display libraries have been proposed for other viruses (e.g., TMV), M13 phage display-based biopanning remains the most potent system with infinite possibilities. In this section, we describe the basic characteristics of M13 followed by a series of recent endeavors largely driven by Belcher and coworkers.

4.1

P3

Characteristics

M13 phage is a filamentous bacteriophage belonging to group Ff of the class I of filamentous bacteriophage, which infect Escherichia coli. M13 is composed of a single strand of circular DNA containing about 6400 nucleotides encapsulated by a coat protein shell. The diameter of the phage is 6 nm, while the length of the phage is dependent upon the length of its DNA and measures ∼1 µm for the native strain of M13. As shown in Figure 12(a), there are five different coat proteins that comprise the protein shell of M13. The predominant form of coat protein that composes M13 is the gene-8 major coat protein, termed pVIII. pVIII exists with ∼2700 copies covering the length of M13. The remaining forms of coat protein present on the phage particles are minor coat proteins and only about five copies of each are present. Protein-7 and protein-9 (pVII and pIX) exist at one end of the phage while protein-3 and protein6 (pIII and pVI) are found at the other end of the phage. While all forms of the coat protein are necessary to maintain the structural integrity of M13, pIII also has a role in the recognition and infection of host cells.47

(b)

Figure 12 Structure of M13 filamentous phage: (a) Schematic diagram of the structure of M13 denoting the location and population of each coat protein subunit encapsulating a single, circular strand of DNA. (b) Portion of repeating array of pVIII coat protein subunits as determined by fiber X-ray studies. Individual pVIII coat protein molecules are represented as ribbons of α-helices. (Reproduced with permission from Ref. 47.  Elsevier, 2001.)

Each coat protein subunit is arranged as an α-helix measuring 1 nm by 7 nm. As seen in Figure 12(b), the coat protein subunit α-helices exist at a small angle in relationship to the axis of the virus because they are constrained by interactions with neighboring α-helices. The N-terminus of the coat protein capsid is comprised of acidic amino acid residues that are exposed to the solvent surface of the phage. The C-terminus is mainly comprised of basic amino acid residues exposed to the interior of the phage, which interact with the negatively charged DNA core. The interior of the capsid structure includes a sequence of 19 apolar amino acids, which provide the hydrophobic interactions necessary for individual capsid subunits to interact with one another.48, 49 The ability of the M13 bacteriophage to incorporate foreign DNA into its genetic core and subsequently display foreign peptides makes M13 an advantageous viral template. Although the DNA of the native strain of M13 has about 6400 nucleotides, foreign DNA with a length of up to 12 000 nucleotides can be successfully incorporated.48 When the foreign DNA is spliced into a gene encoding for the synthesis of the viral capsid, the phage is able to display foreign peptides or proteins on its exterior surface. The fusion of foreign peptides to native phage coat proteins has been achieved for the pIII, pVIII, and pVI forms of phage coat proteins. As illustrated in Figure 13, exploiting this technique of phage display has facilitated the formation of phage display libraries that can be implemented to discover peptides or proteins with high binding

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

Viruses as self-assembled templates gIX

gVIII

Phage genome

gIII

880 nm

pIX

Elution Target

Phage library

Phage display

Repeat interaction pVIII 6.6 nm

pIII

(a)

Biopanning

Phage

No Identifying Yes specific binding Consensus peptide peptides sequence

(b)

× 106 Bacterial amplification

DNA analysis

Figure 13 Schematic depiction of phage display for determining peptide sequences with affinity for binding specific targets. (Reproduced with permission from Ref. 15.  Elsevier, 2006.)

affinity to small molecules or inorganic materials of interest. First, the phage library is allowed to interact with the target. Phage bound to the target are isolated by elution and subsequently amplified by bacterial amplification. This strategy is repeated several times to enhance the phage library with phage-displaying peptides that bind to the target with the greatest affinity. Unique to this system is the link between the selected peptide or protein displayed on the phage and the DNA sequence encoding for the peptide found within the core of the phage. Analysis of the phage DNA will yield the peptide sequence(s) that exhibit the greatest binding affinity toward the target.15, 50

4.2

nanocrystals with preferential alignment guided by the peptide template. By lowering the temperature of the system, water dispersed from the viral particles and singlecrystalline quantum-dot-layered nanowire structures were synthesized. Additionally, Belcher and coworkers were also able to demonstrate the versatility of M13 phage to display more than one genetically engineered peptide sequence to nucleate the growth of two different quantum dots to facilitate the construction of multicomponent materials. Expanding on these findings, Belcher and coworkers have synthesized annealed ZnS and CdS nanowires by increasing the temperature of the system to disintegrate the virus, leaving behind annealed nanowires.9 Using the same approach of exploiting genetically engineered M13 phage as a template for nucleation, the synthesis of CoPt and FePt nanowires was achieved following nucleation of CoPt and FePt nanoparticles and then thermal denaturation of the virus to anneal the nanowires, as shown in Figure 14 for the formation of CoPt nanowires. For each distinct nanowire fabricated, the crystals comprising the nanowires retained their preferred crystallographic alignment, which is proposed to be due to the stability instilled by the high-affinity binding peptides displayed symmetrically along the coat protein. In another attempt to demonstrate the versatility of applying their genetically engineered M13 phage system for nanomaterial synthesis, Belcher and coworkers have

Application of phage display: biopanning

(222) (001) (112)

By employing the strategy of combinatorial phage display, Belcher and coworkers have selected a sequence of peptides that selectively bind with high affinity to semiconductor single crystals based on their identity or crystal structure.51 A library of random peptides for phage display was incorporated into the pIII coat protein of M13 phage and were probed for their selective binding affinity toward five different semiconductors. For each single-crystal semiconductor studied, a single discriminatory peptide sequence was determined which differentiated among the range of semiconductors examined on the basis of their composition or crystalline face structure. This work has yielded the basis for binding organic peptides, arranged as a spatially controlled template on the surface of the M13 phage virus, to inorganic semiconductors for the synthesis of nanomaterials using a bottom-up approach. Belcher and coworkers have also exploited the M13 phage as a biotemplate to synthesize quantum dot nanowires.52 Specifically, genetically engineered M13 phage were employed to display a previously determined peptide sequence in the pVIII M13 coat protein that specifically facilitates the nucleation and growth of ZnS or CdS

15

(200)

200 nm

100 nm (a)

(111) (110)

(b)

Figure 14 Phage-templated CoPt wires: (a) TEM image of CoPt nanoparticles bound to genetically engineered phage. Phage displaying a peptide sequence for the high-affinity binding of CoPt were mixed with CoCl2 and H2 PtCl6 in a 1 : 1 ratio followed by the addition of NaBH4 to initiate the reduction of metal ions to form metallic nanoparticles. (Inset) STEM image of the CoPt nanoparticles bound to phage. Scale bar for STEM image is 100 nm. (b) Low resolution TEM image of CoPt nanowire formed by annealing CoPt nanoparticles templated on phage. By heating the assembly of CoPt nanoparticles on the phage to 350 ◦ C, the biological phage component was destroyed, leaving behind annealed CoPt nanowires. (Inset) Electron diffraction of the annealed CoPt nanowires. Observed in the electron diffraction pattern is the crystalline property of the system as well as the unique (110) and (001) crystal planes. (Reproduced with permission from Ref. 9.  American Association for the Advancement of Science, 2004.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

16

Self-processes

recently utilized M13 phage as a biological scaffold to construct a photochemical catalyst.53 Genetically mutated M13 phage displaying a peptide sequence exhibiting selective binding for iridium oxide in conjunction with the chemical attachment of zinc porphyrins to the phage coat protein to act as photosensitizers functioned as photocatalytic nanoarchitecture capable of oxidizing water. Collectively, the work achieved by Belcher and coworkers reveals the feasibility of exploiting the organized structure of M13 phage as a biotemplate for the bottom-up synthesis of inorganic nanomaterials. While genetically modified M13 phage has been exploited for its ability to selectively template the synthesis of metallic and semiconductor nanowires and nanoparticles, Schaak and coworkers have used the novel approach of utilizing native M13 phage particles to form metal nanowires.54 The electrostatic charge exhibited by the surfaceexposed anionic carboxylate functionalities on native M13 phage was shown to interact with cationic metal species to facilitate the phage-templated synthesis of rubidium, ruthenium, and palladium nanowires. The formation of the metallic nanowires was limited to the pVIII coat protein region of the phage permitting the pIII coat protein region located at the tip of the phage to be modified for the binding of an additional material. The pIII tip of the phage was genetically modified to display a 12 amino acid peptide sequence with selective binding affinity to Fe3 O4 . Constructing phage particles with modified pIII coat proteins allowed for the synthesis of a multicomponent system, which successfully responded to the presence of a magnetic field and displayed catalytic activity. While variations of M13 phage have successfully been employed as templates for the synthesis of metallic nanomaterials, Lee and coworkers have exploited genetically modified M13 phage for the biological application of tissue regeneration.55 M13 bacteriophage were genetically altered to display the peptide sequence RGD, which promotes cell adhesion integrin binding, or the peptide sequence IKVAV, for promoting the adhesion of neural cells and the growth of neurites. The addition of the RGD or IKVAV peptides to the coat protein of the phage capsid resulted in increased interaction between the phage and hippocampal neural progenitor cells and had no deleterious effects on the viability of the cells. Cells grown and differentiated into neurites in the presence of genetically altered phage matrices aligned their growth with the orientation of the phage and exhibited an increased length of neurites in comparison to cells not exposed to genetically engineered phage matrices. Building on the findings in this fundamental study, Lee and coworkers have used a shearing strategy to construct liquid crystalline films of genetically modified phage to align the growth of a series of cell types.56 Genetically altered phage displaying the integrin binding peptide RGD

were self-assembled into 2D matrices by shearing a solution of RGD-displaying phage on a substrate. The linearly aligned phage matrices exhibited no adverse effects on cell viability. The RGD-displaying 2D phage films were also observed to guide the direction of growth of differentiated cells such that over 80% of the elongated cells were aligned within a 20◦ angle of the oriented phage after four days of incubation. Collectively, the work by Lee and coworkers demonstrates the ability of genetically modified phage to easily self-assemble into aligned matrices. These genetically altered phage films are able to interact with cells to facilitate the controlled directional growth of the cells. The results of these studies could find applications in medical therapies for tissue engineering or as a mock system for the study of cell-signaling pathways.

5

CONCLUSIONS

In this work, we described the fundamental characteristics, chemical modification schemes, and the evolution of three virus shape categories into unique directions in bottom-up assembly approaches for novel nanomaterials development. A growing pool of chemical toolkits, such as click chemistry, often combined with a small genetic modification of rod-shaped TMV viruses have led to exciting systems with precisely spaced chemical functionality. TMV’s simple structure and precise dimensions along with its robust assembly scheme have also been exploited for the fabrication of light-harvesting assemblies and conducting nanowires. Moreover, the precise dimensions and/or dynamic environmentally responsive properties of viral assemblies have enabled synthesis of metal or metal oxide nanoparticles (e.g., TiO2 within CCMV capsids) and coatings with precisely controlled dimensions, in addition to increased surface area and functionality such as photocatalytic activity. Potent affinity selection (i.e., biopanning) based on M13 phage display combined with larger scale fabrication strategies is increasingly enlisted in a wide range of biomedical and catalytic applications. We envision that these exciting developments, in conjunction with a growing number and better manipulation of chemical functionalities with an in-depth understanding of the viral nanotemplates, will lead to a wider array of novel materials and more enhanced device performances that can be readily deployed in real applications in the near future.

ACKNOWLEDGMENTS We gratefully acknowledge partial financial support from the U.S. Department of Education GAANN fellowship (J. M. R.) and U.S. National Science Foundation Grant No. CBET-0941538.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

Viruses as self-assembled templates

REFERENCES 1. T. Douglas and M. Young, Nature, 1998, 393, 152. 2. M. Knez, A. M. Bittner, F. Boes, et al., Nano Lett., 2003, 3, 1079.

17

30. H. Yi, G. W. Rubloff, and J. N. Culver, Langmuir, 2007, 23, 2663. 31. W. S. Tan, C. L. Lewis, N. E. Horelik, et al., Langmuir, 2008, 24, 12483.

3. H. Yi, S. Nisar, S.-Y. Lee, et al., Nano Lett., 2005, 5, 1931.

32. F. Sainsbury, M. C. Ca˜nizares, and G. P. Lomonossoff, Annu. Rev. Phytopathol., 2010, 48, 437.

4. T. L. Schlick, Z. B. Ding, E. W. Kovacs, and M. B. Francis, J. Am. Chem. Soc., 2005, 127, 3718.

33. G. P. Lomonossoff and J. E. Johnson, Prog. Biophys. Mol. Biol., 1991, 55, 107.

5. M. A. Bruckman, G. Kaur, L. A. Lee, et al., ChemBioChem, 2008, 9, 519.

34. T. Lin, J. E. Johnson, F. A. M. Karl Maramorosch, and J. S. Aaron, Advances in Virus Research, Academic Press, New York, NY, 2003, vol. 62, pp. 167.

6. T. Douglas, E. Strable, D. Willits, et al., Advanced Materials, 2002, 14, 415. 7. S.-Y. Lee, E. Royston, J. N. Culver, and M. T. Harris, Nanotechnology, 2005, 16, S435. 8. J. K. Scott and G. P. Smith, Science, 1990, 249, 386.

35. J. A. Speir, S. Munshi, G. Wang, et al. Structure (London), 1995, 3, 63. 36. Q. Wang, T. Lin, L. Tang, et al., Angew. Chem. Int. Ed., 2002, 41, 459.

9. C. Mao, D. J. Solis, B. D. Reiss, et al., Science, 2004, 303, 213.

37. Q. Wang, E. Kaltgrad, T. Lin, et al., Chem. Biol., 2002, 9, 805.

10. R. D. Wells and R. M. Wartell, Biochemistry of Nucleic Acids, University Park Press, Baltimore, MD, 1974, vol. 6.

38. S. Meunier, E. Strable, and M. G. Finn, Chem. Biol., 2004, 11, 319.

11. K. Burton, Biochemistry of Nucleic Acids, Butterworths; University Park Press, London, Baltimore, MD, 1974.

39. E. Strable, J. E. Johnson, and M. G. Finn, Nano Lett., 2004, 4, 1385.

12. G. Sumbali and R. S. Mehrotra, Principles of Microbiology, Tata McGraw-Hill, New Delhi, 2009.

40. N. F. Steinmetz, G. P. Lomonossoff, and D. J. Evans, Langmuir, 2006, 22, 3488.

13. T. Douglas and M. Young, Science, 2006, 312, 873.

41. N. Steinmetz, G. Lomonossoff, and D. Evans, Small, 2006, 2, 530.

14. M. Young, W. Debbie, M. Uchida, and T. Douglas, Annu. Rev. Phytopathol., 2008, 46, 361. 15. A. Merzlyak and S.-W. Lee, Curr. Opin. Chem. Biol., 2006, 10, 246. 16. N. F. Steinmetz and D. J. Evans, Org. Biomol. Chem., 2007, 5, 2891. 17. J. N. Culver, Annu. Rev. Phytopathol., 2002, 40, 287. 18. P. J. G. Butler, Philos. Trans. R. Soc. London, B Biol. Sci., 1999, 354, 537.

42. N. F. Steinmetz, S. N. Shah, J. E. Barclay, et al., Small, 2009, 5, 813. 43. M. T. Klem, M. Young, and T. Douglas, J. Mater. Chem., 2008, 18, 3821. 44. A. dela Escosura, P. G. A. Janssen, A. P. H. J. Schenning, et al., Angew. Chem. Int. Ed., 2010, 49, 5335. 45. M. A. Kostiainen, O. Kasyutich, J. J. L. M. Cornelissen, and R. J. M. Nolte, Nat. Chem., 2010, 2, 394.

19. K. Namba and G. Stubbs, Science, 1986, 231, 1401.

46. G. P. Smith, Science, 1985, 228, 1315.

20. A. Klug, Philos. Trans. R. Soc. London, B Biol. Sci., 1999, 354, 531.

47. S. S. Sidhu, Biomol. Eng., 2001, 18, 57.

21. T. L. Schlick, Z. Ding, E. W. Kovacs, and M. B. Francis, J. Am. Chem. Soc., 2005, 127, 3718.

49. D. A. Marvin, R. D. Hale, C. Nave, and M. H. Citterich, J. Mol. Biol., 1994, 235, 260.

22. R. A. Miller, A. D. Presley, and M. B. Francis, J. Am. Chem. Soc., 2007, 129, 3104.

50. G. P. Smith and V. A. Petrenko, Chem. Rev., 1997, 97, 391.

48. D. A. Marvin, Curr. Opin. Struct. Biol., 1998, 8, 150.

23. Z. Niu, J. Liu, L. A. Lee, et al., Nano Lett., 2007, 7, 3729.

51. S. R. Whaley, D. S. English, E. L. Hu, et al., Nature, 2000, 405, 665.

24. W. Shenton, T. Douglas, M. Young, et al., Adv. Mater., 1999, 11, 253.

52. C. Mao, C. E. Flynn, A. Hayhurst, et al., Proc. Natl. Acad. Sci. U.S.A., 2003, 100, 6946.

25. E. Dujardin, C. Peet, G. Stubbs, et al., Nano Lett., 2003, 3, 413.

53. Y. S. Nam, A. P. Magyar, D. Lee, et al., Nat. Nano, 2010, 5, 340.

26. S. Balci, A. M. Bittner, K. Hahn, et al., Electrochim. Acta, 2006, 51, 6251.

54. K. N. Avery, J. E. Schaak, and R. E. Schaak, Chem. Mater., 2009, 21, 2176.

27. A. K. Manocchi, N. E. Horelik, B. Lee, and H. Yi, Langmuir, 2009, 26, 3670.

55. A. Merzlyak, S. Indrakanti, and S.-W. Lee, Nano Lett., 2009, 9, 846.

28. C. Yang, A. K. Manocchi, B. Lee, and H. Yi, Appl. Catal., B Environ., 2010, 93, 282.

56. W.-J. Chung, A. Merzlyak, S. Y. Yoo, and S.-W. Lee, Langmuir, 2010, 26, 9885.

29. E. Royston, A. Ghosh, P. Kofinas, et al., Langmuir, 2007, 24, 906. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc087

Peptide Self-Assembly Silvia Cavalli1,2,3 , Hana Robson Marsden4 , Fernando Albericio1,2,5 , and Alexander Kros4 1

CIBER-BBN, Networking Centre on Bioengineering, Biomaterials, and Nanomedicine, Barcelona, Spain Institute for Research in Biomedicine, Barcelona, Spain 3 University of Udine, Udine, Italy 4 University of Leiden, Leiden, The Netherlands 5 University of Barcelona, Barcelona, Spain 2

1 2 3 4

Introduction Amphiphilic Peptides (amino acids only) Self-Replicating Peptides Other Classes of Amphiphilic Peptide for the Construction of Nanostructures 5 Conclusions References

1

1 2 8 9 14 14

INTRODUCTION

One of the main challenges in supramolecular chemistry, the chemistry of complex noncovalent structures at the nanoscale, involves the issue of forming homogeneous and structurally well-defined architectures with tuneable properties to cover a wide range of possible applications. Therefore, an accurate design and good understanding of the rules governing the molecular assembly of specific monomeric building blocks are key features for the successful engineering of “smart” supramolecular architectures with predictable properties and functions.1–4 Among others, peptides are a particularly attractive class of molecules, which can be used as molecular

building blocks since their structural folding and stability have already been studied in detail.5–9 Nature provides about 20 amino acids as constituents for peptides and proteins. In general, hydrogen bonding, and electrostatic and hydrophobic interactions together have to be taken into account when designing peptide sequences for selfassembly.10–12 The choice of specific side groups in the sequence plays an important role, as side groups are responsible of imparting the character of the amino acid (e.g., polar, nonpolar, aliphatic, positively or negatively charged, or aromatic). Tuning pH or temperature as well as changing environmental conditions (e.g., ionic strength or presence of counterions in the medium) also influences the self-assembling process. Certain amino acid residues have unique functions in driving the folding. For example, proline (P), due to its conformational rigidity, has often been found in turns such as hairpins. It is also commonly found at the beginning of α-helices or at the edge of β-sheet strands.13, 14 Another important residues for stabilizing the assembly is cysteine (C), which provides a reactive thiol for disulfide cross-linking.15 Although the current research efforts have already led to an enhanced understanding of the criteria that govern the self-assembly processes of amphiphilic peptides, owing to their complexity, more investigation is required to gain a better insight into the way aggregation and peptide secondary structure influence each other, when dispersed in an aqueous medium or deposited at interfaces. Amino acids and peptide building blocks can be seen as information carriers, which introduce structural “smartness” in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

2

Self-processes

nanostructures, especially due to their ability to respond to external parameters (i.e., changes in solvent, pH, and temperature, or sensitivity to electronic or photonic energy, and to the presence of chelating metals),13, 16–18 which is of large interest especially when the responsiveness is reversible. Most commonly, short oligopeptides are used as they can be either biomimetic or de novo designed with a wide variety of amino acid sequences. The availability of these peptide-based building blocks has been mainly fueled by the advent of straightforward and fast synthetic methodologies, mostly based on solid-phase protocols that offer easy access to a wide variety of oligopeptides with virtually any amino acid sequence of about 5–50 residues.19–21 Moreover, the possibility to incorporate nonnatural amino acids or functional moieties in the peptide sequence is extremely valuable for the introduction of an increased level of functionality in the assemblies.22 In addition, the intrinsic chiral nature of amino acids, with glycine as the only exception, can lead to the expression of handedness to a higher hierarchical level.23 Finally, the use of biologically relevant peptide sequences can generate new nanomaterials with potential applications in the field of biotechnology and bioengineering.24, 25 In this chapter, the main categories of peptide amphiphiles (PAs)26–28 are discussed by highlighting a number of relevant examples of peptide-based nanostructures, which reveal some of the rules governing their self-assembly into supramolecular architectures and demonstrate the importance of amphiphilic peptides as molecular building blocks for nanoscience.29

2

AMPHIPHILIC PEPTIDES (AMINO ACIDS ONLY)

Amphiphilic peptides, which are composed only of amino acids, are organized in amphipathic sequences comprised of both hydrophobic and hydrophilic domains. Each PA has a critical aggregation concentration (CAC) depending on the balance between the hydrophobic and the hydrophilic parts. The well-ordered nanostructures can be formed through self-assembly when the concentrations are above their CAC with their hydrophobic segment buried inside the core and their hydrophilic segment exposed to water.30–32 The length of both hydrophobic and hydrophilic parts as well as the geometrical constraints affect the types of nanostructures that may form on dispersion of amphiphilic peptides in aqueous media. Thus, peptide sequence and length are both critically important in the design of such molecules to tune the formation of micelles, nanovesicles, nanotubes, nanofibrillar networks, or membrane sheets. When designing amphiphilic peptides, the balance between the hydrophilic

and hydrophobic domains is a critical feature,26 as larger hydrophobic segments reduce the solubility of the molecule in aqueous solutions. However, small hydrophobic domains increase the solubility but decrease the tendency to aggregate. Typically, three to nine nonpolar amino acid residues (e.g., G, A, V, I, L, P, and F) constitute the hydrophobic segment of the PAs, whereas the hydrophilic headgroup is composed of polar amino acid residues with a positive charge (H, K, and R), a negative charge (D and E), or a combination of both. In general, PAs have been designed as N–C type of sequences bringing the hydrophobic moiety on the N-terminal and the hydrophilic domain on the C-end. The charge on the N-terminal is usually blocked by an acetyl group, while the C carboxylic group is either converted to an amide or left as such. In a few cases, the hydrophilic headgroup is positioned at the N-terminus, as in the case of KA6 -NH2 .33 The following examples demonstrate a clear relationship between molecular geometry and shape of the selfassembling nanoobjects formed. However, one has to take into account the influence of other factors such as electrostatic interactions relating to solution pH and ionic strength. Ac-V6 D-OH is a typical example of a PA forming nanotubes in which the valine (V) tails pack backto-back to form the bilayer walls of tubular structures through hydrophobic interactions and the charged groups face toward the water phase.34 In another example, the shorter peptide I3 K formed nanotubes with diameters of about 10 nm and lengths over 5 mm.35 Also in this case, the self-assembly was driven by the hydrophobic affinity between isoleucine I3 tails that packed in the core and the K residues projected outward facing the water molecules. These peptide bilayers then further assembled into twisted ribbons. The fusion of the helical ribbons results in the formation of stable nanotubes. Ac-GAVILRR-NH2 is another interesting example of the effect that molecular geometry has on nanostructure formation. This PA was designed to adopt a cone-shape molecular structure by progressively increasing side chain size of the hydrophobic amino acid residues along the tail GAVIL adjacent to a large cationic headgroup composed of two arginine residues (R). Upon dispersion in aqueous solutions, the formation of both vesicular and doughnut-shaped nanostructures was observed at concentrations above its CAC.36 It was proposed that short spherical micelles merged together side by side and that subsequently bent and fused to form the nanodoughnut structures. The bending was the result of the tension originating from the interaction of the coneshaped peptide side chains.37 In the following sections, self-assembled architectures based on the two main peptide secondary structures (αhelices and β-sheets) are discussed. This is followed by some examples of fiber formation constituted by dipeptide

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

Peptide self-assembly

nonpolar residues Val/Ile/Phe-Leu and Leu/Val-Tyr with (i, i + 4) spacing have been shown relevant for α-helix formation and stabilization.41, 42 Stabilization by interactions between residues with similar polarity is rather common. However, α-helical stabilization through interactions between polar and nonpolar residues has also been demonstrated, as in the case of Lys or Arg with Val or Ile with (i, i + 4) spacing.43 Moreover, cation-π interactions, such as Lys or Arg with Phe to Tyr, have been shown to increase α-helix stability. Such interactions rely on the delocalized electrons of the aromatic side chains that interact with cations.44, 45 Recently, the quadrupole interaction has been added to the toolbox of peptide chemists.46 Zheng and Gao47 showed that the electrostatic interactions between aromatic amino acid side chains having opposite quadrupole moments are sufficient to introduce specificity into peptide binding interactions. Finally, salt bridges between two oppositely charged residues (e.g., Glu-Lys) have also been found to improve α-helix stability.48 The strength and contributions of surface salt bridges to protein stability rely on the interaction distance and geometry, solvent exposure, and influence of adjacent charged residues.49 In the majority of the cases, the interactions discussed above are combined in a synergic way to produce a stable α-helix structure. The so-called protein folding problem, that of anticipating the folded structure of a protein when given the amino acid sequence, has not yet been solved. However, the coiled-coil motif is one of a relatively small number of folding constructs for which the relationships between the primary amino acid sequence and the final folded structure are relatively well understood. The α-helical coiled-coil motif50–52 is composed of α-helical peptides bound in a highly specific manner. The specificity of peptide binding arises from the precise arrangement of chemical functionality on the helical surface: the majority of coiled-coil forming peptides are characterized by a heptad repeat, denoted abcdefg, with apolar amino acids at most of the a and d positions, resulting in an amphipathic helix (Figure 1). The

systems based on π –π interactions. Another interesting category, the “Lego” peptides, for which charge complementary aggregation takes place is also discussed. Finally, some examples of self-replicating systems based on both α-helical coiled coils and β-sheet forming peptides are reviewed.

2.1 α-Helices and the coiled-coil motif Among the types of local structure in proteins, the αhelix is the most regular and the most predictable from the sequence, as well as the most prevalent. In an α-helix, the amino acids are arranged in a right-handed helical structure, in which every backbone N–H group contributes to a hydrogen bond with the backbone C=O group of the amino acid four residues earlier and where each amino acid residue corresponds to a 100◦ turn in the helix such that the helix ˚ along the has 3.6 residues per turn and a translation of 1.5 A helical axis. The pitch of the α-helix, the vertical distance ˚ (the between two consecutive turns of the helix, is 5.4 A product of 1.5 and 3.6).38 The side chains of the α-helix, which can either be charged, hydrophilic, or hydrophobic, are positioned along the axis with a negative charge at the C-terminus and a positive pole at the N-end. This unique conformation enables the display of the side chains on the surface of the helix for molecular recognition. Hydrogen bonding, hydrophobic interactions, polar/nonpolar interactions as well as salt bridges play important roles in stabilizing α-helices. In addition, local and long-range hydrogen bonding between side chains and the main chain is critical in maintaining helical structure and stability. In general, residues such as Pro, Asp, and Glu are common at the N-terminus of native α-helices, while His, Lys, and Arg are common at C-termini.39, 40 Hydrophobic interactions between nonpolar residues are significant and govern both inter- and intrachain interactions in this protein structure. For instance, specific interactions between the intrachain Ionic interactions

g

e′

c d

b′ N

f

b

e

Hydro- d′ phobic interactions

g1

d1

f′ a

f1

c1

a′

e′1

N

d2

a2

a′1 c′

3

g2

d3

a3 e′2

g3

d4

a′4 e′3

g4

C

a′5 e′4

C

b′1

g′

f′1 (a)

Ionic interactions

(b)

Figure 1 Schematic representation of a parallel dimeric coiled coil. The helical wheel diagram in (a) top view down the axis of the α-helices from N-terminus to C-terminus. Panel (b) provides a side view. The residues are labeled as a–g in one helix and a –g in the other. (Reproduced from Ref. 52.  Wiley-VCH, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

4

Self-processes

residues in positions a, d, e, and g of the heptad strongly influence the function, formation, and stability of coiled coils. The hydrophobic residues occupy positions at the interface of the two helices (a, d and a , d ). The packing of the hydrophobic a, d face against that of another coiled-coil forming peptide produces the majority of the binding energy. The apolar face of the helix is not parallel to the helical axis, but winds around the helix once every ∼15 nm, such that the packing of the hydrophobic strips against one another leads to the coiling of individual α-helical “coils.” Positions e, g and e , g , which are solvent exposed, are generally occupied by polar residues that impart specificity to the binding partners through electrostatic interactions. Thus, altering the buried residues at positions a and d as well as the residues in the e and g positions, which are associated with the formation of salt bridges, has a tremendous impact on packing and oligomerization.53–56 Many aspects of coiled-coil binding are determined by the amino acid sequence: the oligomerization state (two or more peptides), size (∼2–200-nm long), direction of binding (parallel or antiparallel), homoor heterobinding, stability, and rigidity. As an example, the natural coiled-coil GCN4 assembles into a dimer when Ile is in the a position and Leu is in the d position. However, when reversing the residues, the formation of a tetramer was observed.57 Owing to the noncovalent character of the interactions among these peptides, association is sensitive to changes in the environment, for example, pH, temperature, ionic strength, and metal ions, which affect the electrostatic or hydrophobic interactions. This versatility arising from a simple helix has stimulated many advances in peptide selfassembly. Fairman and coworkers58 obtained quantitative thermodynamic data for the oligomerization of a modified GCN4 coiled coil. A stable trimeric coiled coil was designed by placing Val at each a position and Leu at each d position of the heptad repeating unit (coil-Va Ld ). Guanidinium chloride denaturation curves were collected at several peptide concentrations and analysis of data provided a free energy of stabilization of −18.4 kcal mol−1 for the trimer. The heat capacity, Cp , was 8.6 cal deg−1 mol−1 per residue. This value corresponded well to that expected in the case of a coiled coil with a well-defined tertiary structure. Inspired by α-helices in nature and their ability to self-assemble into oligomeric coiled coils, a number of researchers have tailored helical peptides in order to generate fibrous structures.59 As an example, Woolfson and coworkers60 described the design of a single-peptide, based on an α-helical coiled-coil motif, named MagicWand, which self-assembled into extended and thickened nanoto mesoscale fibers of high stability and order. The peptide had a heptad sequence repeat, abcdefg, with isoleucine and leucine residues at the a and d sites to promote dimerization.

In addition, to direct the assembly of the peptides into fibers, the terminal quarters of the peptide were cationic amino acid residues (Lys) and the central half were anionic (Glu). The resulting +, −, −, + arrangement gave the peptide its name. Electrostatic and cation–π interactions were shown to play an important role in driving fiber formation, stability, and thickening. The same group engineered a second-generation selfassembling fiber (SAF) system based on two complementary coiled-coil forming peptides.61 The SAF peptides had Ile at a positions and Leu at d positions in order to promote dimeric binding. The helical interface was connected further with interhelical charge–charge interactions between successive g and e sites. Complementary charge residues were introduced to the surfaces of the interacting coiled coils, and a buried Asn pair was also incorporated in the hydrophobic core in the a position to specify the sticky-ended heterodimer designed to propagate the SAFs. Furthermore, to make the resulting coiled coils as soluble as possible and control the aggregation, charged residues were also placed at the b, c, and f positions, resulting in supercharged and highly hydrophilic surfaces with random charged patches. Fibers could therefore be thinned down to ∼10 nm and made much more flexible such that they wrapped around one another forming a network. Exploiting coiled-coil systems, other types of supramolecular architectures have been constructed as in the case of the work described by Ryandov.62 In this example, two coiled coils were designed in order to build nanoreactors. A noncovalent supradendrimer (SD) framework self-assembled from a single-peptide sequence, QEIARLEQEIARLEYEIARLE-NH2 (SD-1), the design of which was based on a coiled-coil dimer. The hydrophobic face of SD-1 was conceived to help the formation of a homodimer, while the polar sites were converted into two polar faces in order to create a network of electrostatic interactions and help associations between dimers rather than within a single dimer. For this reason, Arg residues were incorporated at position c and Glu residues at positions g and e. In such a construct, the individual dimers interacted with the adjacent dimers by forming salt bridges at their polar surface, thereby organizing into large rod-like supramolecular structures. In order to introduce larger cavities in the architecture, an additional peptide SD-2, presenting a complementary coiled-coil sequence, was designed that enabled the assembly of SD-1/SD-2 heterodimers. The final supramolecular organization (SD-1,2) was obtained by mixing SD-1 and SD-2. The hollow space generated in SD1,2 resulted in constrained environment, used to template the conversion of ionic metals into colloids. In this respect, the N-terminal Gln at position f of SD-1 was changed to Cys in order to decorate the cavities with thiol groups that could help stabilize the metallic particles. Using these

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

Peptide self-assembly

5

SH Ag HS SH

(a)

200 nm (b)

50 nm (c)

Figure 2 Synthesis of silver nanoparticles within SD-1,2. (a) A growing silver nanoparticle (Ag) hosted by an SD-1,2 cavity is depicted as a gray circle. (b) A network of cavities filled with silver nanoparticles (gray circles). Arrows indicate the confined space of the cavity within which the particle is to grow. Cysteine residues forming encapsulating thiol (SH) clusters are shown as circles marked with crosses. (c) An electron micrograph of a spherical 1.2-mm wide spread of silver nanoparticles. (d) High magnification of an edge portion of the image shown in part (c). (Reproduced from Ref. 62.  Wiley-VCH, 2007.)

nanoreactors, silver nanoparticles of 5.2 ± 0.5 nm were prepared by citrate reduction of silver nitrate (Figure 2).

2.2 β-Sheet forming peptides In the β-sheet structure, the polypeptide chains are nearly fully extended and regular hydrogen bonds form between the peptide backbone amide protons and carbonyl oxygen groups of adjacent chains. The adjacent β-strands can lie either in a parallel or in an antiparallel manner. In both cases, the β-strands have conformations pointing alternate amino acid side chains to opposite faces of the sheet. The peptide chains in antiparallel β-sheets (a repeat period of ˚ per residue pair) are slightly more extended than those 7.0 A ˚ per residue in parallel β-sheets (a repeat period of 6.5 A pair). The repeat period refers to the average distance along the chain axis of each dipeptide unit. Contributions from electrostatic and hydrophobic forces between amino acid side chains on the same face of the sheet often help to stabilize the structure.38

β-Sheet peptides mostly self-assemble into fibrous structures, which often aggregate further into higher order architectures. Systematic studies have been carried out by several research groups to disclose the rules associated with forming well-defined higher order β-sheet assemblies.63 In this section, some examples of β-sheet forming PAs, composed of amino acids only, are discussed. Aggeli and coworkers thoroughly investigated the spontaneous self-assembly of β-sheet peptides into tapes, ribbons, fibrils, and fibers.64, 65 In one example, they have demonstrated that polyelectrolyte β-sheet complexes (PECs) could be formed on mixing aqueous solutions of cationic and anionic peptides, resulting in the spontaneous self-assembly of fibril networks and the production of hydrogels in appropriate pH windows.66 In general, the two peptides, as well as having the tendency to form antiparallel β-sheets, were designed in such a way that their oppositely charged amino acid side chains arranged in a complementary manner and were present in a sufficient number to stabilize the resulting hydrogels. Amphiphilic peptides have also been used to create stimuli-responsive hydrogels.13 As an example, Rajagopal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

6

Self-processes

(+) Folding trigger

Self-assembly

Lateral association

(−) Folding trigger

Ordered facial association

Noncovalent crosslink Hydrogel Nanoscale structure

Figure 3 Environmentally triggered folding, self-assembly, and noncovalent fibril cross-linking leading to hydrogel formation. Crosslinks are formed by the irregular facial self-assembly of hairpins. (Reproduced from Ref. 67.  Springer, 2006.)

et al.67 prepared a class of peptides, the ability of which to self-assemble into hydrogels was dependent on their folded state. As shown in Figure 3, under unfolding conditions soluble peptides were freely flowing in aqueous solutions. When the folded peptides were triggered by external stimuli (i.e., by adjusting the pH of the solution from acidic values to pH 9 under low ionic strength conditions), they adopted a β-hairpin conformation and self-assembled into a highly cross-linked network of fibrils affording mechanically rigid hydrogels.14, 67 The group of Ghadiri has studied the design, principles, and the preparation strategies to achieve synthetic organic nanotubes, with special emphasis on noncovalent processes such as self-assembly and self-organization.68, 69 As an example, hollow β-sheet tubular structures could be formed by the stacking of cyclic peptide rings. The cyclic peptides, containing an even number of alternating D and L-amino acids, in which the amide bonds were perpendicular to the plane of the ring, aggregated into microcrystalline tubes via a pH-controlled assembly strategy.70 The sequence of octapeptide cyclo[–(L-Gln-D-Ala-L-Glu-D-Ala)2 –] was chosen to improve the solubility in basic aqueous solutions. Controlled acidification promoted hydrogen-bond interactions, thus allowing the self-assembly into hollow tubes composed of ring-shaped subunits stacked through antiparallel β-sheets. These structures exhibited a hydrophobic exterior and a hydrophilic interior and could insert into bilayer membranes, introducing pores. Varying the number of amino acid residues, and consequently the ring size, modulated the porosity. Thus, the small 8-residue rings were shown to be able to transport only small ions through membranes, while the larger 10-membered rings could also transport larger molecules such as glucose and glutamate. Well-defined assemblies of β-sheet structures have also been achieved at the air–water interface.71 Rapaport et al.72

have investigated a family of amphiphilic peptides comprised of alternating hydrophilic and hydrophobic amino acids (Y and Z) with the generic sequence X–Y– (Z–Y)n –X, where the N- and C-terminal residues (X) carried charged ammonium and carboxylate groups, respectively. Variations in amino acid sequence and in the number of dyads (n) participating in hydrogen-bond formation were expected to tune the intermolecular interactions. Peptides Pro-Glu-(Phe-Glu)n -Pro (n = 4, 5, or 7) formed two-dimensional self-assembled β-sheet monolayers at the air–water interface, as confirmed by in situ grazing-incidence X-ray diffraction (GIXD) investigations. The alternating hydrophobic phenylalanine (Phe) and the hydrophilic glutamic acid (Glu) residues caused the peptide chains to orient as β-pleated sheets parallel to the water surface. However, the flexibility of the peptide backbone and the repetitive nature of the amino acid sequence induced dislocation defects (Figure 4a) that limited long-range order to one-dimension (1D), in the direction normal to the peptide backbone (direction a). The two-dimensional (2D) self-assembled architectures could be induced by introducing proline (Pro) residues at the peptide termini. Proline was chosen to prevent the formation of disordered β-sheets along the a direction (Figure 4a) based on three characteristic features: its tertiary amide that prevented participation as a donor in the hydrogen-bond formation; its restricted dihedral angle ( ∼ −60◦ ), significantly different from that of β-sheet peptides ( ∼ −120 to −150◦ ), therefore making its inclusion in the β-sheet ribbon sterically unfavorable; and finally its cyclic side chain, which determined geometric constraints, thus minimizing disorder at the ribbon edge. In addition, attractive electrostatic interactions between the chain termini were expected to juxtapose the β-sheet ribbons along the b direction (Figure 4b), facilitating the formation of the two-dimensional order.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

Peptide self-assembly

7

~6.9 Å 1D

b

~4.7 Å

a

(a)

(c)

a 2D

b H2N +

H N

O N H

O

H N

O N H

O

H N

O

H N

N H

O

O N H

O

H N

O N

O−

O O

(b)

HO

O

HO

O

HO

O

HO

O

HO

O

Figure 4 Schematic diagrams of β-strand assemblies at the air–water interface (rods and open dots represent peptide backbones and hydrophobic amino acids, respectively). View down the normal to the β-sheet of (a) one-dimensional order and (b) two-dimensional order induced by distinct chain termini. (c) Schematic representation of the peptide Pro-Glu-(Phe-Glu)4 -Pro in the β-pleated conformation and the targeted β-sheet crystalline assembly at the air–water interface. (Reproduced from Ref. 72.  American Chemical Society, 2000.)

A schematic representation of the peptide Pro-Glu-(PheGlu)4 -Pro in the β-pleated conformation and the targeted β-sheet crystalline assembly at the air–water interface is shown in Figure 4(c). Well-defined β-sheet arrays at the interface were also observed for larger peptides able to form triple-stranded β-sheets.73 Furthermore, it was shown by the same group that parallel β-sheet assemblies could form as well using two amphiphilic peptides that differed only by inverting the position of Glu and Lys residues along the strand (PA : CH3 CO-Pro-Cys-Phe-Ser-Phe-Lys-Phe-GluPro-NH2 and peptide PB : CH3 CO-Pro-Cys-Phe-Ser-PheGlu-Phe-Lys-Pro-NH2 ).74 In this way, the two cross-strand pair interactions, formed between the oppositely charged Glu and Lys residues, supported the hydrogen-bonded parallel arrangement of the neighboring peptides. A special class of β-sheet folding (poly)peptides is constituted by amyloid peptides. The amyloid aggregation is intrinsically polymorphic, both at the intermediate and final fibrillar levels.75 Amyloid fibrils are implicated in a number of neurodegenerative and systemic diseases.76 Fibrillar structures isolated from tissues of patients display a significant structural polymorphism,77 which seems to have implications in different functional activity and cytotoxicity.78, 79 Amyloid fibrils exhibit exceptional strength and stability and are highly resistant to degradation, which makes these assemblies attractive biomaterials. However, to generate homogeneous material the precise control of the fibril polymorphism is necessary. In a recent work, Pellarin et al.80 used computer simulations of a simplified model of an amyloid polypeptide to elucidate the process of fibril morphology differentiation demonstrating that amyloid polymorphism is under kinetic control.

2.3

“Lego” peptides

Zhang81 introduced the concept of “Peptide Lego,” based on ionic self-complementary peptides. These peptide building blocks contain two distinct surfaces, one being hydrophilic and the other hydrophobic, similar to the “Lego bricks” that have both pins and holes positioned in a wellordered manner, allowing precise assembly into a predetermined organization. In aqueous solutions, the hydrophobic side shields itself from water driving the self-assembly of the peptides, comparable to spontaneous protein folding as observed in nature. The structural feature of these “Lego bricks” is based on complementary ionic bonds with regular repeats on the hydrophilic surface due to the alternation of positively and negatively charged amino acid residues at specific intervals. Lysine (Lys) and arginine (Arg) are typically used as the positively charged residues, whereas glutamic (Glu) and aspartic (Asp) acids are used to generate a negative charge. The complementary ionic sites have been classified into several moduli (i.e., modulus I, II, III, IV, etc.), depending on the alternation of the charges (Figure 5). For example, in the case of modulus I, the (+) positively charged and (−) negatively charged amino acid residues are alternated by one (+ − + − + − +− and − + − + −+ in the complementary peptide). Consecutively, for modulus II, III, IV, and so on, charges are alternated by two, three, four, and so on (++ −− ++ −−, + + + −−−, and + + + + −−−, etc.). The charge orientation can also be designed with a mixed order to yield entirely different molecules (mixed moduli). The first member of the “Peptide Lego” was serendipitously discovered from a segment in a left-handed Z-DNA binding protein in yeast, named Zuotin.82 The self-complementary 16 residue [(Ala-GluAla-Glu-Ala-Lys-Ala-Lys)2 ] peptide, originally found in a

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc088

8

Self-processes

3 (a)

(b)

(c)

Figure 5 Examples of modulus II. Molecular models of the extended β-strand structures of individual molecules are shown for (a) ARARADADARARADAD (RAD16-II, R, arginine; A, alanine; and D, aspartate) and (b) EAEAKAKAEAEAKAKA (EAK16-II, A, alanine; E, glutamate; and K, lysine). The distance ˚ between the charged side chains along the backbone is ∼6.8 A; the methyl groups of Ala are found on one side of the sheet and the charged residues are on the other side. Conventional β-sheet hydrogen-bond formation between the oxygens and hydrogens on the nitrogens of the peptide backbones are perpendicular to the page. (c) A proposed staggered assembly of molecular models for EAK16. The complementary ionic bonds and hydrophobic alanines are shown. Although an antiparallel β-sheet is illustrated, a parallel β-sheet model is also possible. (Reproduced with permission from Ref. 82. Zhang, S.; Holmes, T.; Lockshin, C.; Rich, A. Proc. Natl. Acad. Sci. USA 1993, 90, 3334-3338 ' 1993 National Academy of Sciences, U.S.A.)

region of alternating hydrophobic and hydrophilic residues in Zuotin, interacted strongly with itself to form a stable structure promoted by the hydrated salt ions.82 These “Lego” molecules represent a class of self-assembling βsheet peptides that spontaneously undergoes association in aqueous solution into well-ordered nanofibers.

2.4

SELF-REPLICATING PEPTIDES

The self-assembly of peptides has been used as a driving force for chemical catalysis in amide bonds formation, resulting in self-replication.89 The discovery that peptides have replicative properties has led to the design of a wide range of autocatalytic systems mostly based on the coiledcoil motif and exploiting native chemical ligation.90–92 In self-replication, the product (i.e., peptide) acts as a template to preorganize precursors (i.e., peptide fragments) based on its own sequence in order to catalyze the formation of the product itself (Scheme 1). As a result, a new product is formed that is identical to the template. In coiled-coil self-replication, the ability of the template to form a ternary complex with the two peptide fragments is mediated by interhelical hydrophobic interactions which, in turn, may promote the chemical ligation process. After dissociation, the newly synthesized peptide could then act as a template for a new set of fragments, resulting in exponential product growth. Improvements in the design of self-replicating peptides have been directed toward the enhancement of the autocatalytic efficiency.93–95 The use of self-replicating peptides has also been suggested in the creation of novel materials for biomedical applications.96 Ghadiri and coworkers97 highlighted a synthetic chemical approach, exploiting native chemical ligation, toward the rational de novo design of complex self-organized molecular systems. The work showed the design and implementation of a network of nine α-helical peptides directing each other’s synthesis through a complex network of various autocatalytic and cross-catalytic cycles, including two-, three-, and four-member pathways. More recently, β-sheet forming peptides have also been proven able to act as templates for autocatalysis. For this purpose, Rubinov et al.98 developed an autocatalytic process based on an antiparallel β-sheet forming peptide. The peptide served as a template for the association of two fragments, which reacted through native chemical

Dipeptide hydrogels and π –π interactions

Large aromatic groups have been shown to be able to induce the assembly of dipeptides driven by π –π interactions to form low-molecular weight gelators (LMWGs).83 The majority of dipeptides conjugated to Fmoc or naphthalene groups form hydrogels at low pH ( 1500) because ni − is a powerful hydrogen bond acceptor and back-electron transfer is slow due to its spin-forbidden nature. Because the absorption maximum of ni -HBR− is located at a shorter wavelength than that of succ-HBR− , the sum of the absorptions shifts to the blue. The rate at which the absorption shifts (kshift ) is related to the rate of change in the relative populations of succ-HBR− and ni -HBR− , that is, the time taken for lightinduced shuttling of the macrocycle (∼1 µs). After charge recombination, which takes circa 100 µs, the macrocycle shuttles back to its original position. Cycling of this process at 104 times per second generates ∼10−15 W of mechanical power for each shuttle.

As shown above, the detailed molecular design allows for supermolecules and noncovalent molecular assemblies to carry out motion using PET as a trigger. Within this volume, there is a special chapter devoted to molecular machines, some of which are driven by light and utilize PET as stimulus. The reader is therefore encouraged to seek further examples of this intriguing research.

5

PET IN NUCLEIC ACIDS

In a DNA double helix, π-aromatic compounds, that is, purine and pyrimidine bases, form well-ordered structures based on π-stacking between the bases and hydrogen bonding with the bases of the complementary DNA strands. Charge transport and electrical conduction in π-stacked aromatic arrays of DNA strands were proposed in the early 1960s,121 that is, in addition to their principal role in the storage and replication of genetic information. In the last two decades, the charge transfer in DNA has attracted substantial attention of the scientists due to the potential application in the fields of the bioscience and nanotechnology. The oxidation and reduction processes are an essential part of many biological processes that involve DNA. In the biological environment, DNA is always attacked by various naturally occurring oxidizing and reducing agents. The oxidative DNA damage is known to cause mutation, apoptosis, and cancer.122, 123 The charge transfer in DNA is the key issue that explains the oxidative damage at remote sites in DNA. Likewise, the repair mechanism of the damaged DNA substructures also takes place in living cells. For example, DNA photolyase enzymes repair T-T cyclobutane dimer lesions by negative electron (excess electron) transfer in DNA.124–126 Thus, the understanding of the charge transfer in DNA is essential for the understanding of the photoinduced damage or repair of DNA. In the field of the nanotechnology, DNA is expected to take an important role as a molecular wire,127, 128 because DNA can be regarded as an artificial copolymer that may be synthesized with a great degree of control over sequence (adenine A, cytosine C, guanine G, and thymine T) and the chain length and width due to a well-defined ˚ and width of ∼2 nm. The unique unit length of 3.4 A potential of DNA to act as a nanoscale electrical wire appeals to various scientists. Furthermore, DNA is useful in forming artificial structures in addition to duplex or quadruplex structures.129, 130 The nano sheets generated by the programmed sequences are one of the examples of such artificial structures.131–137 This suggests that DNA can be employed as a scaffold of various chromophore arrays for various functions.71, 138–141 Because of the tremendous potential of DNA in this wide field and numerous potential

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc090

16

Supramolecular devices

applications associated with charge dynamics in DNA, we summarize our understandings of the charge transfer in DNA.

5.1

Hole transfer in DNA

In general, two types of the charge carriers have to be considered in the charge transport mechanisms in DNA. One of the possible carriers is a hole, which is transferred in the DNA via the consecutive oxidation of nucleobase by the neighboring radical cation of the nucleobase, which is generated by the hole injection to DNA. The other carrier is electron for the excess electron transfer, in which the radical anion of the nucleobase causes reduction of the neighboring nucleobase. The oxidation and reduction potentials have been estimated by the electrochemical method.142 The oxidation potentials of the purine bases, that is, G (1.47 V vs normal hydrogen electrode (NHE)) and A (1.94 V), are lower than those of the pyrimidine bases (2.12 and 2.09 V for C and T, respectively). On the other hand, the reduction potentials of the pyrimidine bases, T (−2.12 V) and C (−2.21 V), are higher than others ( E0

N

N

E < E0

S

N

N N

Figure 50

O

−2e−

N

S

O

N N

N

An electrochemically driven molecular shuttle on an Au surface.77

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

N

34

Supramolecular devices

NH HN

HN

NH

O

O

H N

Ph N H

Ph

O

O

O

O

O

−

−e

N H

Ph

O

H N O

O

NH

O

H N

O

O

t-Bu

N

7

O

HN

NH

Ph

t-Bu

N

7

O

+e−

t-Bu

t-Bu

O

An electrochemically driven molecular shuttle based on an amide rotaxane.78

Figure 51

O

O NH

O

H N

N H

Ph

HN

NH

O Ph

O

O

HN

O

Ph O

O

HN

NH

Ph

O

H N

N H

O

N

7

O

N O

O

O

11 N

NH

O

Ph

O

O N H N

Ph

O

O

E1/2 succ= −0.68V

E1/2 ndi= −0.53V

O

O NH

HN

HN

NH

O O

Ph

O

H N

N H

Ph

O

H N

N H

N

7

O

O

O

HN

O O

O

O

NH

N O

O

Ph

11 N

O NH

N H N

Ph

O

O

Ph

Ph

Ph

E1/2 ndi= −1.01V E1/2 succ= −1.21V

O

O

Ph

H N

N H

O

O

Ph O

O

HN

Ph

O

H N N O

O

O N

O

N H

HN 2−

O

11 N

O NH

NH

2−

O

O

O

HN

NH Ph

Ph

O

O

Ph

Ph

O

Ph

O

NH

N H N

Ph

O

O

Figure 52

An electrochemically driven, three-state molecular shuttle based on an amide rotaxane.79

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Ph

Ph

Molecular devices: molecular machinery of monolayer interfaces, implying that directional molecular mechanical motion in supramolecular systems may be translated into controlled motion of bulk liquids in appropriate surface-modified containers.

3.3

Other electrochemically driven molecular machines

Leigh et al. developed electrochemically switchable hydrogen-bonded molecular shuttles (Figure 51) based on a series of [2]rotaxanes containing succinamide and naphthalimide hydrogen-bonding stations for a benzylic amide macrocycle.78 The tetraamide macrocycle resides on the succindiamide moiety due to the favorable hydrogenbonding interactions. The shuttling of the tetraamide macrocycle is induced by reduction of the naphthalimide moiety to the corresponding radical anion, which attracts the tetraamide macrocycle. The reverse process can be realized by the oxidation of the naphthalimide radical anion. Later, the same group developed an amide [2]rotaxane comprising a succindiamide unit and a naphthalene diimide unit acting as a three-state redox-active molecular shuttle (Figure 52) that displays switching both in solution and on surface.79 Three oxidation states of the naphthalene diimide unit can be accessed electrochemically in solution, each one with a different binding affinity for the tetraamide macrocycle, which can be monitored by cyclic voltammetry experiments. It is worth mentioning that the reduction potential of the naphthalene diimide unit was sufficiently low (–0.68 V) to make the rotaxane compatible with operation in self-assembled monolayers on gold. This is the first time that shuttling between two of the states has been demonstrated for a self-assembled monolayer of amide-based molecular shuttles. The shuttling process is reversible and fast (occurs on a millisecond time scale), and provides an inspiration for developing molecular machines on conductive surfaces.

4

PHOTO-DRIVEN MOLECULAR MACHINES

Light is an ideal source of energy to drive artificial molecular machines.80, 81 First, it can be used to monitor the state of the machine using various spectroscopy techniques. Second, the response is very fast, even on nano- or picosecond scales. Third, photo-driven processes in such systems are usually reversible, resulting in photo-driven molecular machines having an autonomous operation that is very important for the construction of molecular devices. And last, photo-driven molecular machines are clean and convenient compared with chemically driven machines and

35

electrochemically driven molecular machines, as the former require chemicals to be added in each forward/reverse step, and, with each cycle, the amount of the driver chemicals increases, thus resulting in a change in the overall equilibria; while the later ones require a supporting electrolyte, which can be considered to be a contamination of the local environment. However, most light-driven molecular machines are often limited by the photostationary states, in which case they can not achieve 100% of one or the other state. The phototstationary state of a reversible photochemical reaction is the equilibrium chemical composition under a specific kind of electromagnetic irradiation (usually a single wavelength of visible or UV light). It is a property of particular importance in photochromic compounds, often used as a measure of their practical efficiency and usually quoted as a ratio or percentage. The position of the photostationary state is primarily a function of the irradiation parameters, the absorbance spectra of the chemical species, and the quantum yields of the reactions. In general, photo-driven molecular machines comprise two parts: (i) a switching component, based on noncovalent interactions within the recognition units, which is responsible for executing mechanical movement, and (ii) a light-harvesting unit, which utilizes light to control the competitive interactions between different recognition sites.81 There are several important photoresponse systems (Figure 53) that have been widely studied and applied in the construction of photo-driven molecular machines: the photoisomerization of azobenzenes and stilbenes, the reversible electrocyclization of diarylethenes, the photochromic reactions of fulgides, the interconversion of spiropyrans with merocyanines (and the associated spirooxazine/merocyanine system), as well as chiroptical switching in overcrowded alkenes.8 Among such photoswitching processes and moieties, the photoisomerization of azobenzenes and stilbenes is most widely studied for mechanical molecular machines because they are simple and easy to introduce into designed molecules, which then function as light-operated molecular machines. The movements in photo-driven molecular machines are usually accompanied by marked changes in both physical and chemical properties, such as color, charge, and stereochemistry, and a wide range of applications based on their switching properties have been both proposed and, in some cases, realized, from liquid crystal display technology to variable-tint spectacles and optical recording media.8

4.1

Azobenzene-based molecular machines

Azobenzene is a good building block for photo-driven molecular machines not only because trans-azobenzene can

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

36

Supramolecular devices Stilbene isomerization

Azobenzene isomerization

hn

N N

hn N N

hn′/∆

hn′/∆

(a)

(b)

Diarylethene electrocyclization

Fulgide electrocycliztion

F F

F F

F

F

F

hn

F

F

F

F

O

O

hn

F

O

O

O

hn′ S

O

hn′ S

S

O

O

S

(c)

(d)

Spiropyran-merocyanine interconversion NO2

hn N O R

(e)

NO2

O

N R

hn′/∆

Examples of molecules displaying photoisomerism.8

Figure 53

O O

O

O O

O

O

Figure 54

Visible-light

N

O K

O

N N

O O O

O

O O +

UV-light

K

N O

O O

O O

Light-controlled complexation of the two crown ether units with potassium cation.6

be easily isomerized by UV light to cis-azobenzene and cisazobenzene can be easily driven back to the trans-one by heating or irradiation by visible light but also due to its synthetic accessibility. Shinkai et al. reported photoresponsive crown ethers (Figure 54) comprising two 15-crown-5 molecules linked by an azobenzene unit, which exhibited a butterfly-like motion.6 They found that the rate of the thermal isomerization was suppressed by added alkali metal cations and different alkali metal cations could be extracted efficiently by the different conformations of the molecules with high selectivity. These studies demonstrate that in principle ion extraction and ion transport through a liquid membrane can be controlled by light. Although this is not a mechanical molecular machine, it is considered as the prototype of early molecular machines and

provides inspiration for the construction of new molecular machines. Nakashima et al. reported the first light-driven molecular shuttle based on a rotaxane (Figure 55) comprising an α-CD linked mechanically with an axle containing an azobenzene unit located at the center, two paraquat units at each side of the azobenzene unit, and 2,4-dinitrophenyl stoppers.82 The location of α-CD could be changed between the azobenzene and ethylene moieties by lights of different wavelengths or by the heat both in water and in DMSO. Tian et al. developed a photo-driven [2]rotaxane molecular shuttle (Figure 56) with dual fluorescent units.83 The [2]rotaxane comprises an α-CD bound mechanically with an axle containing an azobenzene unit and two different fluorescent naphthalimide units. The two fluorescent naphthalimide units exhibit different fluorescent signals

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

37

NO2 O2N

N

N CH2CH2

O

O2N

N N

Visible light or heat

O

CH2CH2

UV light

N

N

NO2

= a-CD

NO2 O2N

N

N CH2CH2

O N N

O2N

N

N CH2CH2

O

NO2

Figure 55

The first light-driven molecular machine based on a rotaxane.82

depending on the movement of the α-CD along the axle. The input and output of the molecular shuttle are both optical, making it an ideal candidate for construction of molecular machines due to the fact that no additional chemicals are required as drivers. Willner et al. attached a CD-azobenzene molecular shuttle (Figure 57) onto a gold electrode surface using a ferrocene-functionalized β-CD (Fc-β-CD) threaded on a monolayer-immobilized long alkyl component containing an azobenzene unit and a bulky anthracene stopper.84 The Fc-β-CD prefers to reside on the trans-azobenzene component initially. The shuttling between two states is controlled by chronoamperometry. The system moves like a molecular train by the shuttling of Fc-β-CD between a trans-azobenzene station site and an alkyl-chain railway component. The assembly functions as a molecular optoelectronic system that records optical information and is read (or accessed) as an electrical signal.

4.2

Stilbene-based molecular machines

Stilbene is a molecule similar to azobenzene and can also be switched between its Z and E isomers by light. Anderson

et al. reported a unidirectional photo-induced shuttling in a rotaxane comprising an α-CD threaded onto a symmetric stilbene dumbbell (Figure 58).85 The α-CD ring resides on the stilbene unit initially and it slides to the biphenol unit under the irradiation of light at 340 nm. The reverse process could be realized by the irradiation of light at 265 nm, making the α-CD ring reside on the stilbene unit again. Interestingly, although the [2]rotaxane demonstrated here had a symmetrical thread, the displacement was shown to be unidirectional, with the narrower six-rim of the CD always closest to the Z olefin. They also studied the mechanism for the rotaxane photoisomerization, finding that the photoactive unit waited until thermal motion had relocated the macrocycle, then photoisomerized to block its return, illustrating how molecular machines could differ from their macroscopic analogs by utilizing random thermal motion. Tian et al. reported a lockable light-driven molecular shuttle (Figure 59) based on a [2]rotaxane comprising an α-CD bound mechanically with an axle to a stilbene unit.86 At neutral pH, owing to the hydrogen bonds formed by the carboxyl groups of the stoppers and the hydroxyl groups of the α-CD, the α-CD resided on the stilbene center even

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

N

CH2

NaO3S

O

N

O NaO3S

O

N H2N

NaO3S

O

CH2

N

N

CH2 N

O

= a-CD SO3Na

430 nm

360 nm

NaO3S

O

CH2 N H2N

O NaO3S

Figure 56 A light-driven rotaxane-type molecular shuttle that can be selectively addressed using two different wavelengths of light absorbed by the phthalimide chromophores.83

O

Supramolecular devices

N

38

if the system was irradiated by light and the trans-stilbene changed to the cis-stilbene. By the addition of a base such as Na2 CO3 , the hydrogen bonds were largely disrupted. Furthermore, as a result of the better size match of α-CD and the biphenyl, the α-CD slided to the biphenyl unit. The inputs and outputs are all photochemical and the capability of one of the states to be locked makes this a unique example of a light-driven molecular machine. The same group then combined the recognition features of α-CD with azobenzene and stilbene and developed a light-driven, [2]rotaxane-based half adder (Figure 60) based on a photochemically driven [2]rotaxane.87 In a half adder, the combination of an AND gate and an XOR gate can carry out elementary addition by using the XOR gate to generate the sum digit and the AND gate to generate the carry digit.88 The [2]rotaxane comprised an α-CD macrocycle locked onto a dumbbell with two different photoswitchable binding sites, an azobenzene site, and a stilbene site and was end-capped by two fluorescent units that could be addressed (or excited) independently by using light at different wavelengths (UV–vis). The AND gate could be realized by the output of the changes of absorption response to the two different inputs (irradiated at 380 nm and 313 nm). The XOR gate could be realized by the fluorescence output response to the same inputs. Lastly, the two signals were combined to construct a half adder, a type of a device widely used in mathematics and computing. It is worth mentioning that all of the processes are reversible, indicating that the half adder can be operated repeatedly. This is the first example that could mimic a half adder on a unimolecular level. Tian et al. also developed a molecular switch (Figure 61) based on a [3]rotaxane with three stable states that responded to multiple light inputs.89 This [3]rotaxane comprised two α-CD macrocycles as the ring component, an azobenzene unit and a stilbene unit as the recognition state, and two different fluorescent naphthalimide units as stoppers. As demonstrated above, the azobenzene group and the stilbene group can be photoisomerized by light of different wavelengths, so the location of the two α-CD macrocycles can be altered by the irradiation of light at different wavelengths.

4.3

Amide-based molecular machines

In 2001, Leigh et al. reported a photoinduction of fast, reversible translational motion in a molecular shuttle (Figure 62) based on a peptide [2]rotaxane with two binding sites, a succinamide unit and a naphthalene imide unit.90 The molecular shuttle was realized by the photoexcitation of the naphthalene imide unit the corresponding radical anion (∼1 µs; previous photo-driven shuttles generally function

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

39

Fc O

H N

O

N H

O S

N H

N

N

320 nm < l < 380 nm

= b -CD

l > 420 nm

Fc O S

N H

N N

O

H N

N H

O

Figure 57

A photoactive “molecular train.”84 O NaO

ONa

O

340 nm

ONa NaO

265 nm

O O

NaO

ONa

O

O

O

NaO

O ONa

= a-CD

Figure 58

A photo-driven molecular shuttle based on a stilbene/α-CD rotaxane.85

on the time scale of minutes to hours). Owing to the higher affinity of the amide macrocycle for the naphthalene imide cation radical, the tetraamide macrocycle shuttles from the original succinamide station to the naphthalene imide station. The reverse process was realized by charge recombination (∼100 µs) to revert the system to its original state. Leigh et al. also reported a chiroptical molecular switch (Figure 63) with a large amplitude displacement of a tetraamide macrocycle driven by light.91 The tetraamide macrocycle resides on the fumaramide unit of the dumbbell component initially. Upon photoisomerization of the

olefin station at 350 nm using benzophenone as the sensitizer, the macrocycle moves to the glycyl-l-leucine (GlyLeu) unit, locking the molecule in a co-conformation whose aromatic rings were held in a well-defined chiral environment, which could be detected by circular dichroism. The reversible process can be induced using irradiation by light at 400–670 nm using a catalytic amount of Br2 . This is a novel way to drive molecular shuttles that can be used in areas where chiral transmission from one chemical entity to another underpins a physical or chemical response.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

40

Supramolecular devices

O NaO3S

O O

NaO3S

H

H O

O 335 nm

N H2N

N

O

O

H2N

NaO3S

O

H

NaO3S

O

O O

O

O

HO

H

O OH

= a-CD

Na2CO3

NaO3S O N

NaO3S O

-

H2N

OOC 335 nm

N

O

H2N O

-

OOC

O NaO3S

280 nm

−OOC

NaO3S

Figure 59

COO−

A lockable light-driven molecular shuttle based on a stilbene/α-CD rotaxane.86

A molecular shuttle based on a bistable peptide [2]rotaxane (Figure 64) with remarkable positional discrimination was also reported by the same group.92 The position of the tetraamide macrocycle can be altered by irradiation of light and recovered by heat due to the light-induced conformation change between fumaramide (trans) and maleamide (cis) isomers.

4.4

O

Ruthenium(II) complex-based molecular machines

Mechanically interlocked molecules containing ruthenium(II) complexes have also been considered as good candidates for the realization of photo-driven molecular machines since the ruthenium(II) complexes can be easily involved in photo-induced ligand substitution reactions to generate relative motion of mechanically interlocked molecules.93 Sauvage et al. developed a photo-driven molecular machine prototype (Figure 65) based on a disassociative excited state of ruthenium(II)-containing [2]catenanes.94 They synthesized two [2]catenanes via ruthenium(II) template synthesis that differed only in the length of the linker on one of the two interlocked rings. The relative movement of the two interlocked rings in the two catenanes could be caused by the light-induced ligand substitution at the ruthenium(II) complexes upon irradiation

with light at λ > 400 nm in the presence of a chloride anion, while the reverse process could be realized by heat. The whole process was quantitative as evidenced by UV–vis measurements and by 1 H NMR spectroscopy. Interestingly, a marked ring-size effect was noticed; subtle structural factors in ring-size control the rate of the motion. A weak point of the system was the limited control over the shape of the photoproduct, since the decoordinated ring can occupy several positions. The same group applied this approach to photo-driven rotaxanes; however, the thermal reverse reaction was neither clean nor selective.95 The catenanes and rotaxanes described above both contain only one binding site for the ruthenium(II) complexes. In effect, the relative motion of the two macrocycles yields a two-state system: the coordinated state and the state characterized by a lack of coordination, in which the two macrocycles are loosely threaded. Sauvage et al. developed a molecular switch (Figure 66) based on lightinduced changes in the geometry of the complex.96 In a DMSO solution, this ruthenium(II) complex with terpyridine and phenanthroline ligands linked by a long alkyl chain ((CH2 )18 ) exhibits changes in geometry under light irradiation, while the original geometry is recovered by heating the solution. Although this is not a mechanically interlocked molecular machine, it may potentially be used in mechanically interlocked structures to perform mechanical work.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

SO3Na H 2N

Stopper B O

O N

N

NaO3S

41

N

N O

O

SO3Na SO3Na H 2N

O

O N

N

NaO3S

N

N O

O = a-CD

Stopper A

SO3Na

450 nm or ∆

313 nm 280 nm or ∆

380 nm

O N

SO3Na

280 nm or ∆

H 2N

N

N

O N

N

NaO3S

O

N O

SO3Na

O

O

N

N

O SO3Na

NaO3S

NH2 NaO3S

313 nm

450 nm

280 nm

380 nm

O N O NaO3S

SO3Na

O H 2N N O

N

N

NaO3S

Figure 60

A photo-driven molecular shuttle acts as a half adder.87

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

O

42

Supramolecular devices

SO3Na H2N

O

O N

N

NaO3S

N

N O

O

SO3Na 380 nm

280 nm 313 nm

450 nm or heat

= a-CD

O N N

N

O

N

N

NaO3S

O

N O O

H2N

O N

NaO3S

N O

NaO3S

O SO3Na NH2 NaO3S

Figure 61

A photo-driven molecular switch based on a [3]rotaxane with three states.89

O

O NH Ph Ph

HN O

O

H N

N H

O HN NH

t -Bu

N O O

O

t -Bu

Charge recombination

355 nm, DABCO

O

O NH HN

O Ph Ph

N H

− N

H N O O

NH H N O

Figure 62

O

t-Bu

O

t -Bu

A photo-driven molecular shuttle with fast and reversible translational motion.90

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

SO3Na

Molecular devices: molecular machinery

O

O HN

NH O Ph Ph

43

O

H N

N H

N H

8

NH

O HN

O

H N

Ph

O

Ph

O

O

O Either 312 nm, CH2Cl2 or CH3CN, 20 min, 62% Or 400 – 670 nm, cat. Br2, CH2Cl2, 2 min, >95% Or 130 °C, C2H2Cl4, 6 days, 95%.

Either 254 nm, CH2Cl2, 20 min, 56% Or 254 nm, CH3CN, 20 min, 49% Or 350 nm, benzophenone, CH2Cl2, 20 min, 70%

Ph

O

O

Ph O

NH

NH O N H

HN O N H

5

NH

O

H N

Ph

O

O HN

Ph O

O

Figure 63

A photo-driven chiroptical molecular switch based on a peptide rotaxane.91 O O NH O 9

N H Ph

Figure 64

Ph

O

HN

H N O

O

H N

N H

Ph Ph

254 nm

O HN

O NH

HN NH

Ph

∆

Ph

O

H O

O

N H

O NH

O

Ph

Ph

H N HN

N O

9

N H

O

O

A photo-driven molecular shuttle with remarkable positional discrimination.92

Balzani, Stoddart, and coworkers reported an autonomous artificial nanomotor (Figure 67) powered by sunlight by incorporating ruthenium(II) complexes into a bistable [2]rotaxane, which underwent shuttling in an intramolecular charge-separated state.97 Irradiation of the ruthenium–trisbipyridine complex (Figure 67, green structure) generated a highly reducing excited state. An intramolecular electron transfer then occurred between the excited metal center and the most easily reduced bipyridinium station (blue), on which the macrocycle resides preferentially. The result was destabilization of the macrocycle–station interactions so that the macrocycle preferred to relocate to the alternative bipyridinium unit (pink). Remarkably, in this system, the back electron-transfer process was slow enough

to allow shuttling of the ring toward the other station in approximately 10% of the molecules in the relaxed process. The investigated system also showed other quite interesting properties besides being powered by sunlight and operating as an autonomous motor: it worked in mild conditions, it was remarkably stable, and it could be driven at high frequency (kilohertz).

4.5

Other photo-driven molecular machines

Balzani, Stoddart, and coworkers developed the “firstgeneration” light-fueled “piston cylinder” molecular machine (Figure 68a) of a pseudorotaxane type in 1993.98 In this

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

44

Supramolecular devices 2+

O O

O

N

N

Ru N O

hn Et4NCl

O N

N

N

Ru

N

O

N

∆

N O

O

O O

O

O

O

O

O O O

O

O

O

O N

Cl Cl

N

N

O O

O

O

O

O

O 2+

O O

O

Ru N O

O

O O

O N

N

O

O

O

O

O

O

O

hn Et4NCl

N N

N N

∆

N

Cl Ru Cl N

N N

O O

O

O

O

O O

O O

O

O

O

Figure 65

O

O

N

O

O

O

Photo-driven molecular machines based on a ruthenium(II)-containing [2]catenane.94 (CH2)18

(CH2)18

O

O N

O

N

N

hn in pyridine Y = 100%

N

Ru

N

N

O

(1) 140 °C 2 h in DMSO (2) Reflux 2 h in pyridine

N

N Ru

N

N

N N

Figure 66

A molecular switch based on light-induced geometrical changes.96

molecular machine, they used 9-anthracenecarboxylic acid as a photosensitizer and triethanolamine as a “sacrificial” reductant to control the threading and dethreading processes of a cyclophane CBPQT4+ and a 1,5-bis(2-(2hydroxyethoxy)ethoxy)naphthalene (1/5BHEEN) unit by light. Later, they improved the molecular machine (Figure 68b) by incorporating a metal (Re or Ru)–pyridine complex into the cyclophane unit.99 This so-called secondgeneration “piston cylinder” molecular machine can also perform threading and dethreading motions like the first one, but no external photosensitizer and “sacrificial” reductant are needed, and the process can be repeated many times. Abraham et al. developed a photo-driven molecular shuttle (Figure 69) based on a [2]rotaxane with CBPQT4+

as the ring component and an axle with two binding sites, a diarylcycloheptatriene unit, and an anisole unit.100 The ring resides on the diarylcycloheptatriene station with additional “alongside” interactions with the anisole moiety in the resting state (ground state). Upon photo-induced dissociation and formation of the tropylium cation in the thread, the positive charge repels the tetracationic macrocycle, which then resides on the electron-rich anisole property. This process may be reversed by heating, which results in the recombination of the tropylium cation with the previously expelled methoxide anion. This is the first example of photo-switch utilizing a photo-heterolytic reaction to control the switching process.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

45

N O

O N

O

ON

O

N

O

N

N

N O

O

Ru2+

O

O

O

O

O

N

N

O

N Relax

hn 10% of the molecules

O

O

O

ON

O O N

O O

N

O

N

O

N

N

N

Ru3+ N

N N

Figure 67

An autonomous artificial nanomotor powered by sunlight.97

+

COO−

N

HO

O

N

N

N

HO

O

O

O

hn

O N

O

OH

O

O2

N

N

O

OH

N

(a) OC CO OC Re Cl N

HO

OC CO OC Re Cl

N

N

O N O

+

N

HO

N hn

N

N

O

O

O2

O N

O N

O

OH N

N

(b)

Figure 68

The first- (a) and second-generation (b) of light-fueled “piston cylinder” molecular machines.98, 99

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

O

OH

46

Supramolecular devices

O

O

O N

N

OCH3

N

N

O

N O O HN

HN

5

O

OCH3

∆

O

Figure 69

O

N

360 nm

O

ON

N

O

A photo-driven molecular shuttle with a folded molecular thread.100

MOLECULAR MACHINES DRIVEN BY INTEGRATED DRIVING FORCES

In some cases, particularly in complex systems, molecular machines are driven by cooperative effects of multiple driving forces to generate some specific functions. Some cases have been described above, for example, the photodriven molecular machines excited by light but recovered by heat. However, in this part, we mainly focus on a kind of molecular machine that performs work by multiple driving forces, some of which may contain more complex structures and can show complicated functions. Early in 1994, Kaifer, Stoddart, and coworkers reported the first bistable molecular shuttle (Figure 70) based on a [2]rotaxane driven by chemical and electrochemical stimuli.101 The [2]rotaxane comprised a CBPQT4+ ring threaded by an axle consisting of two binding sites, benzidine and biphenol units. Under redox reactions and the addition of acid or base, the CBPQT4+ ring moves back and forth along the axle. This molecular shuttle can be switched by two different mechanisms and is a good candidate for the construction of complex molecular machines. Leigh et al. developed a unidirectional rotation in a mechanically interlocked molecular motor (Figure 71) driven by integrative forces in 2003.102 A [2]catenane and a [3]catenane consisting of one or two tetraamide macrocycles interlocked with a macrocycle containing four binding sites were synthesized: (A) a secondary fumaric acid diamide moiety that shows the strongest affinity for the tetraamide macrocycle (Figure 71, green moiety); (B) a tertiary fumaric acid diamide moiety with a N-methyl substituent on each of the fumaroyl termini (red moiety), which shows diminished affinity for the tetraamide macrocycle

as a result of steric effects due to the N-methyl substituents. The third station (C) is a succinic amide-ester (orange moiety), which shows the lowest affinity for the amide macrocycle. Finally, a simple mono-amide moiety that shows the weakest affinity toward the amide macrocycle is also included. This amide moiety affects only the behavior of the [3]catenane. In the [2]catenane, the ring can be switched between the three different stations (A, B, and C), but the rotation is not unidirectional because the ring can go from A through B and C and back to A in each direction. However, in the [3]catenane, each ring is able to act as a barrier for the other one and each ring can only migrate in the direction where the other macrocycle does not reside. Thus, the rotation in the [3]catenane is unidirectional. Moreover, although neither of the two catenanes is three-dimensionally chiral, they are chiral in the two dimensions in which the rotation occurs. This is the first unidirectional, tristable molecular machine involving mechanically interlocked molecules and is of great importance for developing even more complex molecular machines. Balzani, Stoddart, and coworkers reported a dual-mode “co-conformational” molecular switch based on catenanes containing both bipyridinium and dialkylammonium recognition sites.103 Either one or two dialkylammonium and two bipyridinium recognition sites have been introduced into a π-electron-deficient ring component of three [2]catenanes. Two π-electron-rich macrocyclic polyethers (BPP34C10 or 1/5DN38C10) were used as their other ring components. One of the catenanes is shown in Figure 72. The BPP34C10 ring mainly resides on the bipyridinium unit and it slides to the dialkylammonium site only when two electrons are added to reduce bipyridinium units and one proton is added to protonate the dialkylammonium

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

47

+

N

Si O

O

O

HN

NH

O

O

O

N

−e−

N

Si O

O

O

NH

O

O

O H2N

O Si

O

O

O

O Si

N Pyridine

NH2 O

O

N

O

O

N

Figure 70

O

N

O

N

O

O

+e−

TFA

Si O

O

N

HN

N

N

O

O

O Si

N

A chemically and electrochemically switchable molecular shuttle.101

unit. This constructs a molecular “AND” gate with the addition of two electrons and one proton as two inputs and the relocation of the BPP34C10 as an output. These catenanes are a new kind of prototype for the construction of molecular machines, particularly with reference to the controlled unidirectional ring circumrotation in catenanes. Stoddart et al. reported a three-component molecular switch (Figure 73) driven by two external inputs.104 This molecular switch combines the recognitions of DB24C8/ secondary ammonium and DN38C10/paraquat derivative. The former can be switched by pH changes and the latter by electrochemical potentials. They found that the connection/disconnection of the two systems could be controlled independently by an acid/base and redox stimulation. Nakashima et al. developed a multi-mode-driven molecular shuttle (Figure 74) based on [2]rotaxanes using the

recognition of azobenzene by α-CD.105 The rotaxanes consist of α-CD as a ring, azobenzene as a photoactive moiety, a primary station, and viologen as a hydrophilic group and an energy barrier unit for the slipping of α-CD. Different from the light-driven molecular machine reported by the same group in Section 4.1, the shuttling of the [2]rotaxane could only occur in DMSO, but not in water. The α-CD shuttles within the azobenzene moiety at temperatures below 100 ◦ C, while between the azobenzene moiety and the propylene moieties at temperatures above 100 ◦ C. The α-CD can also shuttle between the azobenzene and the alkylene moieties via irridation by UV light and visible light. These studies provide a powerful technique for controlling and monitoring the molecular shuttling processes and afford the opportunity to design and develop nanoscale switching devices based on multi-mode-driven molecular shuttles.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

48

Supramolecular devices

HN

NH O

O N H

(CH2)4

O A

O

O

D

E/Z

(CH2)4

HN

NH

O (CH ) 2 12

O

CH3

O

CH3 N

O

O

NH

O

O

O

B

C

CH3

O

NH

(CH2)12

(CH2)12

NH

N

CH3

HN O O H N B O

(CH2)12

A

A

A

D C

B

C

O

(CH2)4

O N

HN

O

N

O O

O

E/Z

O NH

NH

N H

NH

(CH2)4

C

N H

O

HN

NH O

O

NH

N H

OA

O

O

A'

STATE I

B

B C

A'

STATE II B'

D

B

D

A C

C

C

B' A'

A'

STATE III B

D

D

B'

A' C

C B

D

C

Figure 71

6 6.1

Leigh’s molecular motors.102

THE APPLICATIONS OF MOLECULAR MACHINES Applications in solid-state molecular electronic devices

As demonstrated at the beginning of this chapter, one of the most appealing applications for molecular machines is in the construction of molecular devices. We have demonstrated many examples for various molecular machines that are good candidates for molecular devices, but the realization from the molecular systems operating in solutions to practical applications is still a challenge. One reason

is that useful molecular devices generally work in the solid state, at interfaces or on surfaces, while the molecular machines demonstrated above mostly function in solution.8 The transfer of molecular-machine technology into solid substrates is a key step in the development of many potential applications. Generally speaking, directly transmitting mechanical changes at the molecular level to the macroscopic world will require organization and cooperation of many molecular-level machines. For the molecular devices to work on the macroscopic level, integration of the nanoscale (and subnanoscale) movement of the molecular machines with their structural complexity as well as their connection to the substrates, and so on is required. Some

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

N

N +2e

O O O

O

O

O

O

49

O

+H+

O O

N

N N H

N

N O O O

O

O

O

O

N

N O O O

N

O

N N

O

O

N H2 O

N

+H+

O O O

O

O

O

O

O

O

N H N

O

O

O

N

O

O O O N

N

+2e

N H2

Figure 72

A chemically and electrochemically driven molecular switch behaves as an “AND” gate.103

examples of operational supramolecular machines that have been converted into molecular devices already exist. The earliest report on application of an active molecular component into solid-state molecular electronic devices dates back to 1999 when Stoddart, Heath, and coworkers reported redox-active, degenerate, two-station, and Vshaped [2]rotaxanes106 (Figure 75) as a monolayer between electrodes made from titanium and aluminum oxide. These devices can be configured to generate AND and OR logic functions. Later, they reported the fabrication and transport properties of single-molecule-thick electrochemical junctions based on the same Langmuir–Blodgett monolayer of the compounds.107 Although the I–V characteristics of the junction were reversible in the forward bias mode, the junction resistance was irreversibly decreased upon application of a sufficient reverse bias, causing an irreversible decrease in current, which was similar to the solution-phase oxidation of the phenoxy groups. The devices provide a compelling argument that molecular switch devices may play an important role in unimolecular computing. A [2]catenane (Figure 37) was sandwiched between an ntype polycrystalline silicon bottom electrode and a metallic

top electrode in the solid state using the Langmuir–Blodgett technique, constructing an electrochemically reconfigurable molecular switch tunnel junction (MSTJ) as a secondgeneration, solid-state molecular electronic device.108 The device exhibits bistable current/voltage characteristics. The switch can be turned on at +2 V, turned off at −2 V, and read between 0.1 and 0.3 V and can be cycled many times. The monolayer of the redox-switchable [2]catenanes was also deposited on a gold surface.109 It was found that not only the morphologies of the films of the [2]catenanes were different in their different redox states but their current/voltage behaviors were also dependent on the redox states prior to their transfer onto gold. The above devices inspired scientists to develop solidstate MSTJ devices made from a [2]pseudorotaxane and [2]rotaxane (Figure 76) in which hydrophobic and hydrophilic regions were directly incorporated into the molecular structure to allow self-organization.110 The electrical properties were found to be highly dependent on the supramolecular structure, the presence of bistability within the (super)molecule, and the organization of the Langmuir–Blodgett film. The devices exhibited a

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Supramolecular devices

O

+ H+

N

O

N N

N N

O O

O

O

O

O

+

O

e−

O

O

O

N

N

N Ru

II

N

Figure 73

N N

N RuII

N

N

N

O

A three-component molecular switch.104

Ox

O O

O

O

O

O

O

O

−

+

H+

Red

+ H+

H2N

N

N

O

O

O

O

N

O

N

N RuII

N

O

O

O

O

N N N

N O

O

O

O

O

O

H2N O hn hn′

O

O O

O

O

O O O

O

+ O

O

O

H2N

O

O

O

O

O

N

O

O

O

O N O

N

RuII N

N

O

Ox

O

− H+

O

H2N O

O

Red

O

O

O

+

N

O

O

O

O

N

O

O

O

O

O

50

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

51

Shuttling O2N

NO2 O2N

N H

N

N

N N

N

N H

N

NO2

4ClO4 Heating over 100 °C

= a-CD Shuttling

O2N

NO2 O2N

N H

N

N

N N

N

N H

N

NO2

4ClO4 Heating up to 100 °C

O2N

NO2 O2N

N H

N

N

N N

N

N H

N

NO2

4ClO4 UV light Visible light

N N O2N

NO2 O2N

N H

N

N

N Visible light

N H

N

4ClO4 O2N

NO2 O2N

N H

NO2

N

N

N N

N

N

N H

NO2

4ClO4

Figure 74

A multi-mode-driven molecular shuttle based on an azobenzene rotaxane.105

(noncapacitive) hysteretic current–voltage response that switched the device between high- and low-conductivity states, although control devices exhibited no such response. This study represents a significant first step toward elucidating molecular structure/device property relationships for active molecular electronic devices. As demonstrated above, Stoddart et al. doped the molecular switches in Langmuir–Blodgett and self-assembled monolayers. In solid-state MSTJs, the same group also immobilized the molecular switches within a solid-state polymer electrolyte,111 and the resulting microfabricated, planar, three-terminal equivalent of a standard electrochemical cell was used for electrical addressing. They found that certain steps within the molecular-mechanical switching cycle greatly slowed down by the polymer environment, but the overall mechanism remained unchanged from that

observed in other environments. The colorimetric retention times of the devices could be controlled over a dynamic range of 103 to 104 s by varying the molecular structure of the switch. These results suggest that electrochemical measurements should go far toward quantifying how the kinetic parameters that describe the switching cycle are influenced by the physical environment of the molecular switches. Goddard III et al. studied the mechanism of the above bistable rotaxane molecular switch, finding that the two states have a 40 ∼ 60 unit difference in conductance.112 They determined that ring-on-DNP was the ON state, while ring-on-TTF was OFF. The change in the delocalization of the molecular orbitals affected by the ring movement played a key role in the switching mechanism. Later, various studies utilizing molecular dynamic simulation113, 114 and density functional theory115 were carried out to probe the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Supramolecular devices

O

relationship between the physical environment and the switching properties on different surface such as on Au (111) surfaces113, 116 and at the air/water interface.114 It was found that the ring-on-DNP co-conformation was only the most conductive if the cyclophane participated directly in electron tunneling by providing low-lying LUMO orbitals. The molecular switches exhibit a folded conformation, while the added intra- and intermolecular interactions do not change the relative stabilities of the two co-conformations. These studies demonstrate the relationship between the structures and the properties of the molecular switches, which is critical for the construction of more complicated molecular electronic devices used in molecular computing. Stoddart, Heath, and coworkers developed a 160-kilobit molecular electronic memory circuit fabricated at a density of 1011 bits cm−1 (pitch 33 nm; memory cell size 0.0011 µm2 ) using a monolayer of bistable [2]rotaxane (Figure 77) with both hydrophobic and hydrophilic regions as the data storage elements in 2007.117 This memory circuit has achieved the dimensions of a dynamic random access memory circuit projected to be available by 2020. This proves that it is feasible and promising to use molecules as nanoscale components to create miniaturized electric circuits and develop molecular computing.

O O N O O

HO

O

N O

O O NO O

O O O N O

HO N

N O O O O

O

N N

HO

N N

O

O

O

Figure 75

O

Chemical structures of the compounds used in the molecular devices.106, 107

O

ON O O

O

O

O

O

O

O

O

O

O

O

O

O

O

O N

O

O

O

O

O

O

52

6.2

Application in drug delivery

The switching properties of molecular machines make them excellent candidates for drug delivery for the release of the drug can be easily controlled (or switched) by the change in the states of the molecular machines. Stoddart, Zink, and coworkers developed an operational supramolecular nanovalve (Figure 78) based on a [2]pseudorotaxane in 2004.118 A macroscopic valve is a machine furnished with a controllable component that regulates the flow of gases or liquids from a reservoir, while nanovalves developed by Stoddart, Zink, and coworkers are based on interlocked molecules that are capable of functioning as gatekeepers for the controlled release of dye molecules from mesoporous silica substrates (MCM-41).119 The nanovalve consists of a tethered DNP-containing derivative as the gatepost, a CBPQT4+ ring that recognizes DNP units, and tris(2,2
-phenylpyridyl)iridium(III), Ir(ppy)3 , a luminescent probe molecule that can be easily detected by emission spectra. The operation of the nanovalve involves four stages: (i) preparing the container, (ii) filling it, (iii) closing the valve, and (iv) opening the valve to release its contents. It can be easily concluded from the formal demonstration that the valve can be opened and closed by the reduction and oxidation of CBPQT4+ , and the trapped molecules are released during this process.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

53

t Bu

N

N

O

O

OCH3

O

O

OCH3

O

O

OCH3

O Et

O

t

O

S

S

S

CH3S

S

S

H2 N C

H2 O C

H2 O C O

Bu N

N

tBu

N

N O

Et

O

O

S

S

S

S

S

N CH3S t

OH

O

Bu N

Figure 76

O

O

N

Chemical structures of two [2]rotaxanes with both hydrophobic and hydrophilic regions.110

This nanovalve can be reconfigured with built-in photosensitizers, such as tethered 9-anthracenecarboxylic acid and tethered [Ru(bpy)2 (bpy(CH2 OH)2 )]2+ (bpy = 2, 2
bipyridine), to introduce light to trigger the reduction of CBPQT4+ to release the luminescent probe molecules.119 This control can be expressed in both a regional and temporal manner by using light as an ON/OFF stimulus for the supramolecular nanovalves. This nanovalve is a true molecular machine that can perform practical work. The nanovalve demonstrated above is an irreversible one that acts like a cork in bottle, that is, the trapped molecules can only be released from the system but cannot be trapped again due to the pseudorotaxane properties of the building blocks.120 Later, the same group reported a reversible molecular valve (Figure 79) based on a [2]rotaxane comprising a CBPQT4+ ring threaded onto an axle with two binding sites, a TTF unit, and a DNP unit; this has been demonstrated many times in Section 3.121 The trapped molecules can be released and trapped again via oxidation and reduction of the TTF unit, as long as the CBPQT4+ ring moves back and forth on the axle. This operational valve is also a true molecular machine consisting of a solid framework with movable parts to accomplish a specific task and may be used in the controlled release of drugs in the future.

For further investigation of the nanovalves, a series of nanovalves was constructed on the basis of mesoporous silica nanoparticles using two bistable [2]rotaxanes with different spacer lengths between their recognition sites as the gatekeepers and Ir(ppy)3 and rhodamine B as trapped molecules.122 It was found that the nanovalves were more efficient when the bistable [2]rotaxane-based gatekeepers were anchored deep within the pores than when they were attached closer to the pores’ orifices. Furthermore, the nanovalves were less leaky when the lengths of the linkers between the surface and the rotaxane molecules were shorter. These properties are similar to that of the macroscopic valve in that the effectiveness of the valve in controlling flow is highly dependent on the fitting and matching of its components—the valve will leak when it is too loose and will not open when it is too tight.119 The following can be expressed as a general trend: small guest molecules require short linkers and large guest molecules require long linkers. This study is fundamental and essential to the future design of drug delivery systems that can release drugs with different structural dimensions. Pseudorotaxanes based on dialkylammonium threads encircled by DB24C8 rings were tethered to an MCM-41 surface to yield pH-driven supramolecular nanovalves.123 The components of this system dissociated upon deprotonation of the threads induced by addition of base,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Hydropholic stopper

O O

O O O

Figure 77

Hydrophobic stopper

Chemical structure of the [2]rotaxane used in the crossbar memory.117

N S S

N

S S O

O

O

N

N

O

O

O

O

O

O

O

O

O

O

H2 O C

O

O

OCH3

OCH3

Supramolecular devices OCH3

54

releasing the trapped coumarin 460 molecules. A variety of bases was used for activation; the steric size of the activating bases greatly affected the rate of release. This study can be used in targeted, acid-/base-controlled drug delivery. Pseudorotaxanes based on diaminoalkanes threads encircled by CB[6] rings were also tethered to an MCM41 surface to yield pH-driven supramolecular nanovalves (Figure 80) that could employ biocompatible components and operate in water.124 The nanovalves are highly pH dependent due to highly pH-dependent 1 : 1 complexes of the CB[6] and diaminoalkanes. This study approaches biotechnological and medical applications in a futuristic way, and in vivo applications of CB[6]-based nanovalves may be realized using the natural variations in pH that exist within healthy and diseased cells in living systems in the future. For practical applications of nanovalves as drug delivery materials, stimuli that exist in vivo must be used to release the cargo molecules; enzymes are appropriate choices. Stoddart, Zink, and coworkers developed enzymeresponsive snap-top covered silica nanocontainers, which perform drug delivery functions in vivo.125 They prepared two different snap-top systems, one with an ester-linked stopper and the other with an amide-linked stopper based on the α-CD/ethylene glycol chains recognition motif. The ester-linked system can be selectively activated by porcine liver esterase to release cargo molecules (rhodamine B), while the amide-linked one was left intact. It is expected that the controllable release of drug in vivo can be applied to cure diseases more efficiently in the future and the divergent synthetic approach described here will allow the snap-top motif to be very easily adapted to accommodate many other applications.

6.3

Applications in macromolecules and polymers

A new type of macromolecules, mechanically bonded macromolecules, has been developed by incorporating interlocked structures into traditional macromolecules.14 Owing to the introduction of mechanical bonds, these macromolecules usually show good physical and mechanical properties, suggesting their potential applications in materials science, nanotechnology, and medicine. By incorporating the acid/base-controllable [c2]daisy chain (Figure 3) into polymers, Stoddart et al. developed poly[c2]daisy chains (Figure 81)126 via highly efficient Huisgen-type 1,3-dipolar cycloaddition frequently utilized in “click chemistry.” The poly[c2]daisy chains not only display good polymeric properties, such as low diffusion coefficient and a large SEC molecular weight, but

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

O

O

O

O

O

O

+ N

N

O

O

+

O

O

NH

+N

NH N

N

O

N

+N

N+

O

Ir

O Si O BnO

+

+ N

+

N+ O

55

O N

O

O

O

or N

O

N +

Si O BnO

O

O

−

Cl

O OH

Reduction

Nanovalves closed

Figure 78

Nanovalves open

The irreversible nanovalve.119 (Reproduced from Ref. 119.  Wiley-VCH, 2007.)

O

O

O

O

O

+ N

+ N

S

S

S

S

N +

N +

O

Step 1

O

+ N

+ N

S

S

S

S

N +

O

O

O

O

O

Loading or recharging

N +

R4+

O

O

O

O

O

O

O

O

O

O

O

O

OH

∆ CH3CN

Step 4

Release of guest molecules

Close valve

Step 2

O O NH

NCO

Si O

Open valve

O

O

Step 3 Si O

(a)

O

O

(b)

Figure 79 A reversible nanovalve. (Reproduced with permission from Ref. 121, T. D. Nguyen, H.-R. Tseng, P. C. Celestre, A. H. Flood, Y. Liu, J. F. Stoddart and J. I. Zink, Proc. Natl. Acad. Sci. USA, 2005, 102, 10029. Copyright (2005) National Academy of Sciences.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

56

Supramolecular devices

(a)

(b)

(c)

Loading

Raise pH

Capping

or

Comtrolled release

+

H3N

or

N N N

H3N

Si O

OO

H2N NN N

NH

+ NH2

NH O

NN N

NN N

+ NH2

NH

or

H2N +

Si O O

O

Si O

Si

O O

NH O

O

O

Cucurbit[6]uril

Si O

O O

Si O

O

Rhodamine B

Figure 80 pH-Responsive supramolecular nanovalves based on cucurbit[6]uril pseudorotaxanes.124 (Reproduced from Ref. 124.  Wiley-VCH, 2008.)

O O O N O H2 O O O H2 O N O O O O

N O N N

N

N

N

O

N N N

O

O

N

O O

O

O

6PF6

n CF3COOH

P1-t -Bu

O O N H

O

O O O O

N N

N

O

N O

O

N N N

O

O

O O

N O

O

N

O

O

N

H N

O O

4PF6

Figure 81

Acid–base actuation of [c2]daisy chains.126

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

n

Molecular devices: molecular machinery

Br

Br EtO

EtO O

O

O

x

57

O

y

O

O

O

O

O

O

Oxidation

N N N

N N

O

OCH3

O

Reduction

S

OCH3

O

O O

O O

O

N

N S

N N

O

O

N

N

O O

O

O

N

S

S

S

N

S

O

N O

N

O O

Figure 82

N N N

N

N

O

S

O

O

O

S

O O

An electrochemically driven poly[2]catenane.128

also quantitative and fully reversible switching behavior in solution. The switching rates for both contraction and extension processes on the polymer are faster than its monomeric precursor. This poly[c2]daisy chains are considered to be robust polymeric switching materials and will be good candidates for smart materials and functional nanodevices. A [c2]daisy chain dimer polymer with extension and contraction properties was reported by Grubbs et al. via ring-closing olefin metathesis.127 This polymer is similar to natural fibers because it can experience both extension and contraction states. It is worthy of mention that the radius of the polymer was observed to increase by 48% after expansion by adding a base. These good mechanical properties make it an exceptional precursor for functional materials. Stoddart et al. also incorporated a bistable [2]catenane (Figure 39) into macromolecules to form a side-chain poly[2]catenane (Figure 82).128 This poly[2]catenane could also behave as a molecular switch that can be addressed in an “on” or “off” state electrochemically by the virtue of the redox properties of the TTF moiety. Furthermore, the switching properties remain even in spherical aggregates. This study provides a good mechanically interlocked switchable polymeric scaffold for the construction of solidstate molecular electronic devices.

7

MOLECULAR MOTORS, ROTORS, AND PROPELLERS

In the previous sections, our main focus was on mechanical molecular switches based on mechanically interlocked structures: pseudorotaxanes, rotaxanes, and catenanes. Molecular motors, rotors, and propellers based on single rotor molecules, as important molecular machines, have also attracted great attention during the last two decades.129 Molecular motors can be defined as molecules that are able to convert any energy input into controlled motion. Inspired by the unidirectional rotary motion of F1 ATPase,130 much effort has been focused on systems that allow controlled molecular rotation and translation. Early in 1967, a molecular propeller-like compound131 was reported by Akkerman et al. In 1981, Mislow et al. synthesized a molecular gear system.132 In these systems, the two brakes or gears move cooperatively, and the relative movements of the two gears can be induced by the hit of solvent molecules. However, because the hit of the solvent molecules are all spontaneous, these cannot be considered as molecular motors due to the random movements of the rotor in different directions. As a unidirectional molecular rotor, it should meet the following requirements: The relative stable different parts

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

58

Supramolecular devices OCH3

OCH3

N

N

Hg2+

2+

add Hg

N

OCH3

EDTA

N CH3O

Figure 83

A molecular brake.134

of the system can rotate unidirectionally under external stimuli; the mechanical repeatability of the rotor should be similar to macroscopic motors; the system can perform practical work by converting the input energy. Moreover, it should work in the solid or on the surface instead of in solution, making it act as a molecular device.

7.1

Triptycene-/helicene-based molecular rotors

In 1994, Kelly et al. synthesized a molecule that could operate as a reversible molecular brake (Figure 83),133, 134 in which the wheel—represented as a three-toothed gear and constructed as a triptycene—spins rapidly at 30 ◦ C in the absence of Hg2+ (or other metal ions), while after the addition of metal ions the rotation of the triptycene is blocked and the brake works. Later, the same group developed a molecular system that could achieve a unidirectional rotary motion of 120◦ around a single bond (Figure 84).135 This molecule consists of an amino-substituted triptycene unit linked by a single bond

with a disubstituted [4]helicene, involving the continuation of the friction-braking action to prevent anticlockwise rotation. The amino group in the molecule should be armed as an isocyanate by the addition of phosgene. The reaction between the isocyanate group and the hydroxy group does not occur in its low-energy conformations owing to the long distance that makes it difficult for them to interact with each other; however, if the triptycene rotates about 60◦ to a conformationally excited state, the urethane forms, trapping the molecule in an energetically excited conformation around the single bond. Followed by cleavage of the urethane group, a unidirectionally clockwise 120◦ rotation of the triptycene is realized. Although this is not a continuous and fast rotation, the design principles here may prove relevant for a better understanding of biological and synthetic molecular motors producing unidirectional rotary motion. A molecular motor (Figure 85)136 was then developed to accomplish repeated unidirectional 120◦ rotation by introducing an amino group on each blade of the triptycene and a 4-(dimethylamino)pyridine unit to selectively deliver phosgene (or its equivalent) to the amine in the “firing CH3

CH3

Cl

O

O N C

O

NH2

CH3

Cl

Rotate

Et3N

O

N

O

(CH2)3OH

(CH2)3OH

(CH2)3OH

C O

Urethane formation

CH3

H2N O

Figure 84

Rotate over Eact

HN O

(CH2)3OH

CH3

CH3 [H2O] (Cleave urethane)

O

O (CH2)3

O

NH

(CH2)3O

O

Sequence of events in the 120◦ molecular rotor.135 (Reproduced from Ref. 135.  Nature Publishing Group, 1999.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery H2N

H2 N

H 2N

CH3

CH3

CH3 O

NH2 Cl

A

H 2N

Cl

N

O

NH2

A O

N

O

O

N OH C

O

OH (CH3)2N

N(CH3)2

CH3 A

As above

NH2

NH2 H 2N

A N

A

NH2 H 2N

N

O

N

O

OH (CH3)2N

OH (CH3)2N

Sequence of events in the molecular rotor.136 (Reproduced from Ref. 136.  American Chemical Society, 2007.)

position.” The molecular motor would experience three above-mentioned 120◦ clockwise rotations to complete a 360◦ unidirectional rotation.

7.2

N

O

H2N

(CH3)2N

Figure 85

H 2N

CH3

O

O N C

A

H2N

CH3

O

CH3

N

N(CH3)2

H 2N

H N

H 2N

A O

NH O

OH (CH3)2N

CH3

H 2N

O N

O

NH2

CH3

H 2N

A

H 2N

O

OH (CH3)2N

OH (CH3)2N

Chiral helical alkene-based molecular rotors

In 1991, Feringa et al. synthesized a chiroptical molecular switch (Figure 86)137 based on “pseudoenantiomeric” forms, P and M (indicating the opposite helicity, P denotes the right-handed helicity and M stands for the left-handed

helicity), of a chiral helical compound, which is a sterically overcrowded olefin. The conversion between different enantiomers and diastereomers of the molecule can be induced by light and heat, which can be detected by circular dichroism (CD). An improved stereoselective molecular switch (Figure 87) was obtained when donor and acceptor substituents were introduced into the thioxanthene lower half.138 Owing to the introduction of the nitroarene acceptor moiety and the dimethylaminoarene donor moiety, the switching process takes place in the visible wavelength

300 nm

O S M cis

Figure 86

Cl

H N

NH2Cl

A

H 2N

N

O

59

O

250 nm

S P trans

A chiroptical molecular switch.137

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

60

Supramolecular devices S

S

365 nm

NO2

(H3C)2N

435 nm

NO2

(H3C)2N S

S M cis

Figure 87

P trans

A chiroptical molecular switching process based on donor–acceptor-substituted dissymmetric alkenes.138–140

range. It is worth mentioning that this kind of chiroptical molecular switches can be used as dopant for liquid crystalline (LC) phase in the modulation of mesophases and physical properties of LC materials. For instance, when a host–guest system consisting of the chiroptical molecular switch (1 wt%) and a nematic LC material 4
-(phenyloxy)4-biphenylcarbonitrile was alternately irradiated with 435 and 365 nm light (5 min irradiation time), the LC phase changes from the nematic to the cholesteric phase, and vice versa, which is perfectly stable during eight switching cycles.139, 140 A light-driven unidirectional molecular motor (Figure 88)141–143 was then developed by the same group based on a similar overcrowded olefin structure by just

CH3ax

adjusting the energy barriers of the helicity inversion steps by structural modifications. Under the stimuli of appropriate wavelength (UV light) and temperature, the rotor completes a full 360◦ rotation process via two photochemical trans–cis isomerizations and two thermal irreversible helix inversions. However, a dominant disadvantage of the above first generation of unidirectional, light-driven molecular rotors is that the upper part of the molecules is too close to the lower part, that is, the gap of the fjord region is too small. Thermal isomerization of the molecule will meet a considerable steric hindrance, making it take several hours to complete the rotation process and limit the speed of the rotation. In the following studies, Feringa et al. proposed two solutions:

CH3 eq ≥ 280 nm ≥ 380 nm

CH3 ax

CH3eq

(P,P)-trans

(M,M)-cis

60 °C

20 °C

CH3 eq CH3 eq

≥ 280 nm

CH3 ax

≥ 380 nm

CH3 ax

(M,M)-trans

Figure 88

(P,P)-cis

Photchemical and thermal isomerization processes of the molecular motor.141

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

S

S

CH3 ax

CH3eq

hn

CH3O

365 nm

CH3O

S

S

(2′R )-(M )-trans

60 °C

(2′R )-(P )-cis

∆

60 °C

S

∆

S

CH3eq

CH3ax

hn

CH3O

365 nm

CH3O

S

S

(2′R )-(P )-trans

Figure 89 rotors.146

(2′R )-(M )-cis

Photchemical and thermal isomerization processes of the second generation of unidirectional, light-driven molecular

Step 1 > 280 nm

−40 °C

Stable (2R,2′R )-(P,P )-cis

Step 4

Unstable (2R,2′R )-(M,M )-trans

20 °C

Step 2

−10 °C

> 280 nm

−40 °C Step 3

Unstable (2R,2′R )-(M,M )-cis

Figure 90

61

Stable (2R,2′R )-(P,P )-trans

Photchemical and thermal isomerization processes of the smallest light-driven molecular rotor.144

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

62

Supramolecular devices

(i) reducing the Gibbs free energy of activation of the thermal helix inversion step by lowering the energy of the transition state through release of steric hindrance in the fjord region and (ii) increasing the energy of the ground state, for instance, by introducing larger substituents at the stereogenic center.144, 145 The same group found that the introduction of a single stereogenic center is sufficient to control unidirectional motion and developed the second generation of unidirectional light-driven molecular rotors. A new type of molecular motor containing a chiral 2-methyl-2,3-dihydrothiopyran upper part as the rotor and a (thio)xanthene lower part as the stator (Figure 89) was synthesized.146, 147 Under a similar stimulus, the rotor undergoes a repetitive 360◦ rotations in four distinct steps relative to the stator with lower-energy barriers. The molecular motors can also be attached to a gold surface via two thiol-functionalized “legs” on the stator,148–150 making it a useful prototype for solid-state molecular devices. In 2003, Feringa et al. reported the smallest light-driven molecular motor (Figure 90)144 by greatly increasing the speed of rotation. It contains only 28 carbon and 24 hydrogen atoms by changing the six-membered ring into a cyclopentane moiety. The energy differences between the (pseudo-)diaxial and (pseudo-)diequatorial conformations of the methyl substituents in both isomers can direct the relative rotation of the upper and lower parts of the molecule, decreasing the half-life of the thermal isomerization to only 18 s at room temperature, while the fastest speed for this process reported before is 2400 s. The structure of the second-generation, light-driven molecular motors can be further tuned by introduction of bulky substituents at the stereogenic center of the molecule rotors (Figure 91a).145 This is due to a more strained structure with elongated carbon–carbon double bonds and a higher energy level of the ground state relative to the transition state for the rate-limiting thermal isomerization step. By introduction of a tert-butyl unit to the rotor moiety, the half-time of the rotation can reach up to 5.74 × 10−3 s,

t Bu

(a)

Figure 91

(b)

Chemical structures of two molecular motors.145, 151

which exhibits the potential to be developed as naonomachines. Moreover, this kind of motor is very effective in inducing helical organization in a liquid-crystal film. By using a phenyl-substituted molecular motor (Figure 91b) as a dopant (1 wt%) into the liquid-crystal film, the nanosized motor can rotate microscopic-scale objects by harvesting light energy. In 2010, Feringa et al. synthesized a lockable, lightdriven molecular rotary motor152 featuring a selfcomplexing [1]pseudorotaxane based on the DB24C8/ secondary ammonium ions recognition motif. The molecular motor was locked in an acidic environment due to the formation of [1]pseudorotaxane, which stabilized the location of the rotor and the motor. When a base was added, the molecular motor was unlocked and the rotary motion could be realized again. This is an important step toward the construction of controllable molecular motors at a singlemolecule level. Recently, a light-driven molecular motor (Figure 92)153 with the ability to reverse the rotary direction was designed and synthesized. In this system, the unidirectionality was controlled by the relative motion at a single stereogenic center, while it was relatively easy to invert the direction of the rotation from clockwise to anticlockwise by changing the configuration at this stereogenic center via a basecatalyzed epimerization. The ability to change directionality is an essential step toward mechanical molecular systems with adaptive functional behavior.

7.3

Ferrocence-based molecular pliers

In 2003, Aida et al. reported a light-driven chiral molecular scissor (Figure 93),154, 155 which can perform an open and close motion. The molecular scissor consists of three different components: two phenyl groups as the blade moieties, 1,1
,3,3
-tetrasubstituted as the pivot part, and two phenylene groups that are strapped by an azobenzene unit through ethylene linkages as the handle parts. The open and close motion of the blade parts can be induced by the pivotal motion of the connected ferrocence unit, which is caused by the photoisomerizable azobenzene unit. A pivotal motion at the ferrocene unit can be clearly observed by a change in the CD spectral profile of the enantiomers in response to ultraviolet and visible lights due to the existence of the chiral ferrocene unit. An alternative method to operate the molecular scissors156 was to change the oxidation state of the ferrocene unit, making the open–close motion possible only by UV light. The molecular scissors can be further developed as molecular pliers (Figure 94),157 which can twist guest molecules by incorporating a zinc porphyrin unit at each cyclopentaduenyl ring of the ferrocene module. Owing

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

S (3′S)-(M )

(3′S)-(P )

Step 3

hn

S

CON(CH3)2

Step 2

S

S

∆

Base

Step 5

Step 5′

Base

Epimerization changes the direction of rotation

CON(CH3)2

S

S

∆

(3′R )-(M )

Step 4′

CON(CH3)2

Step 3′

hn

(CH3)2NOC

Anticlockwise rotation

S

S

Step 2′

(3′R )-(P )

∆

(3′R )-(M )

S

(3′R )-(P )

hn

Step 1′

(CH3)2NOC

S

CON(CH3)2

S

S

A light-driven molecular motor with direction-reversing properties.153 (Reproduced from Ref. 153.  Nature Publishing Group, 2011.)

(CH3)2NOC

Figure 92

∆

(3′S)-(P )

(3′S)-(M ) Clockwise rotation

S

hn

Step 1

S

S

S

Step 4

(CH3)2NOC

Molecular devices: molecular machinery

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

63

64

Supramolecular devices was locked by the formation of a double intramolecular Zn–N coordination between the zinc porphyrin and aniline unites. After the addition of a cis-1,2-bispyridylethylene, the intramolecular lock was cleaved and an intermolecular 1 : 1 complex was formed, which could be detected by CD spectroscopy. Moreover, the self-locking operation of the molecule can be executed by the photo-induced isomerization of the host/cis-guest complex into separated host and trans-guest.

7.4 hν h ν′

Figure 93 Molecular structures of the molecular scissor and cartoon representation of its open and close motion induced by photoisomerization of the azobenzene unit.154 (Reproduced from Ref. 154.  American Chemical Society, 2003.)

to the good coordination abilities of zinc porphyrin to nitrogenous bases, the molecular pliers can bind a bidentate guest 4-(isoquinolin-4-yl)isoquinoline, forming a stable host–guest complex. The twisting of the guest was realized by the movements of the two zinc porphyrin units, which were triggered by the light-induced conformation change of the azobenzene handle parts as mentioned above. A self-locking molecule (Figure 95)158 was then synthesized by changing the azobenzene unit to two aniline units attached to each cyclopentadienyl ring of the ferrocene module. In apolar solvents such as benzene, the molecule

Fullerene or carborane-wheeled single molecular nanomachines

Tour et al. developed a series of surface-rolling molecules (Figure 96),159–161 namely, nanocars and nanotrucks, which are composed of three components: spherical fullerene wheels, freely rotating alkynyl axles, and a molecular chassis. The use of spherical wheels and freely rotating axles enables the rolling motion of the molecules on gold surface, not stick-slip or sliding translation. These studies underscore the ability to control the directionality of motion in molecular-sized nanostructures via precise molecular design and synthesis. A light-activated unidirectional molecular motor162 was synthesized; this contained an oligo(phenylene ethynylene) chassis and an axle system with four carboranes incorporated with Feringa’s molecular motor as the chassis. This kind of molecular motor indeed rotates upon irradiation with 365 nm light as evidenced by kinetics studies in solution, providing us with a motorized nanocar (Figure 97). Nanoworms163, 164 were constructed by changing the chassis from Feringa’s motor to an azobenzene unit. Both carborane-wheeled and fullerene-wheeled nanoworms were synthesized; the carborane-wheeled nanoworms showed

Ar N N

Zn N

Ar

N N Zn N N N

N

N

Ar

UV N

Ar

Fe

Ar

N

N

Zn

N N

N

Ar

N N

Fe

N Vis

N Zn

N N

N

N

Ar Ar

Figure 94 Schematic representation of photoisomerization of a 1 : 1 complex of molecular pedal host with a rotary guest.157 (Reproduced from Ref. 157.  Royal Society of Chemistry, 2007.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery H2 N

65

NH2 N N Zn N

N N

N N Zn N N

N

UV

N N N Zn N

Vis

N N N Zn

N

N

N

NH2

N H2 Internally double-locked

Externally locked

Figure 95 Schematic representation of photoisomerization between the host/cis-guest complex into separated host and trans-guest.158 (Reproduced from Ref. 158.  American Chemical Society, 2006.)

preliminary solution-based photoisomerization from trans to cis forms, compared to slight changes in fullerene-wheeled nanoworms, because the rapid intramolecular energy transfer to the fullerene wheels quenched the photoexcited state of the motor (Figure 98). Micrometer movement of dye-labeled carborane nanocars165, 166 was observed on a glass surface at room temperature via single-molecule fluorescence imaging. The transition of the nanocars was coupled with a rotational motion and was consistent with a wheel-like rolling mechanism. Further studies have suggested that the speed of the nanocars was dictated by the strength of the interactions between the nanocar wheels and the surface, which means that changing the surface should have a large impact on the mobility of the nanocars (Figure 99). A “nanodragster” (Figure 100)167 was synthesized, which consists a p-carborane small-wheeled short front axle and a fullerene large-wheeled rear axle. This kind of molecular vehicles should help establish the role of p-carborane wheels in directional rolling motion.

8

CONCLUSIONS “. . . . . . the language we call chemistry took less than half a century to make the revolutionary transition from molecules to supermolecules, on the back of having taken well over a century to make the painstaking journey from atoms to molecules. . . . . .” J. Fraser Stoddart (2009).168

In summary, we have demonstrated different types of mechanical molecular machines in detail and the applications of these molecular machines in solid-state molecular devices, drug delivery, and mechanically bonded macromolecules. Just like J. Fraser Stoddart said, supramolecular chemistry is still in its infancy.168 As an interesting part of supramolecular chemistry, although molecular machines are young, they have attracted much attention not only due to their good topology and interesting properties but also because they have great potential in many fields referred to in this chapter. The development of molecular machines and particularly their practical applications is a multidisciplinary endeavor that demands input from scientists with various backgrounds.8 Input from biologists is required to uncover the workings of complex motor proteins or help in assembling and addressing molecular devices capable of drug delivery with spatial and temporal control. The assistance from mathematicians and physicists can provide an insight into guiding construction of more sophisticated logic and sensor systems, while help from materials and surface scientists can develop methods that enable deposition and organization of artificial molecular machinery for molecular computing. Polymer scientists can be helpful in interfacing artificial molecular machines with “smart” macromolecules, while engineers can bring artificial molecular machines into real life. Here, the chemists play a central role in designing and synthesizing molecular machines, developing the supramolecular frameworks,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

66

Supramolecular devices C10H21O

C12H25

C12H25

OC10H21

C10H21O N

N

N

N

OC10H21

C10H21O

OC10H21

C10H21O

C12H25

OC10H21

C12H25

Nanotrucks OC10H21

OC10H21

C10H21O

C10H21O

C10H21O

C10H21O

OC10H21

OC10H21

OC10H21

C10H21O

OC10H21

C10H21O Nanocars

Figure 96

Chemical structures of nanotrucks and nanocars.159–161

S H3C

S

Figure 97 motor.162

Chemical structure of the nanocar-shaped molecular

which are essential for the development of artificial molecular machines. The discipline of chemistry of molecular machines, as a combination of the “bottom–up” approach and supramolecular chemistry, will inevitably lead to broad applications in many fields in the future. Perhaps most importantly, the chemical “bottom–up” approach embodied in the preparation and studies of artificial molecular machines, which is already enabling the preparation of nanoscale objects, will be on the forefront of efforts to translate the properties designed and encoded at the molecular level into mechanical work at the macroscopic level. For the science of molecular machines, this is just the beginning of a long but exciting journey. It is clear, however, that new, interesting, and important results are round the corner.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Chemical structures of (a) carborane-wheeled nanoworm and (b) fullerene-wheeled nanoworm.163, 164 Figure 98

(a)

N N

(b)

C10H21O

OC10H21

C10H21O

C10H21O

OC10H21

N

N

OC10H21

OC10H21 C10H21O

C10H21O

OC10H21

C10H21O

OC10H21

Molecular devices: molecular machinery

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

67

68

Supramolecular devices

CH3O

OCH3

NH HN S

N

CO2 O N

Figure 99

Chemical structure of a dye-labeled, four-wheeled nanocar.165 OC10H21

OC10H21

C10H21O

C10H21O

C10H21O

C10H21O

OC10H21

OC10H21

CH3

Figure 100

CH3

Chemical structure of a nanodragster.167

ACKNOWLEDGMENTS We thank Professor Harry W. Gibson for his enormous help in the preparation of this article. The work was supported by the National Natural Science Foundation of

China (20604020, 20774086, 20834004, and 91027006), National Basic Research Program (2009CB930104), the Fundamental Research Funds for the Central Universities (2010QNA3008), and Zhejiang Provincial Natural Science Foundation of China (R4100009).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

REFERENCES 1. R. Ballardini, V. Balzani, A. Credi, et al., Acc. Chem. Res., 2001, 34, 445. 2. I. Rayment, H. M. Holden, M. Whittaker, et al., Science, 1993, 261, 58. 3. H. Noji, R. Yasuda, M. Yoshida, and K. Kazuhiko Jr, Nature, 1997, 386, 299. 4. J. F. Stoddart, Chem. Soc. Rev., 2009, 38, 1802. 5. R. P. Feynman, Eng. Sci., 1960, 23, 22. 6. S. Shinkai, T. Nakaji, T. Ogawa, et al., J. Am. Chem. Soc., 1981, 103, 111. 7. H. Tian, Angew. Chem. Int. Ed., 2010, 49, 4710. 8. D. A. Leigh, F. Zerbetto, and E. R. Kay, Angew. Chem. Int. Ed., 2007, 46, 72. 9. V. Balzani, A. Credi, F. M. Raymo, and J. F. Stoddart, Angew. Chem. Int. Ed., 2000, 39, 3348. 10. P. R. Ashton, R. Ballardini, V. Balzani, et al., J. Am. Chem. Soc., 1998, 120, 11932. 11. A. M. Elizarov, S.-H. Chiu, and J. F. Stoddart, J. Org. Chem., 2002, 67, 9175. 12. J. D. Badji, V. Balzani, A. Credi, et al., Science, 2004, 303, 1845. 13. J. D. Badji, C. M. Ronconi, J. F. Stoddart, et al., J. Am. Chem. Soc., 2006, 128, 1489. 14. L. Fang, M. A. Olson, D. Ben´ıtez, et al., Chem. Soc. Rev., 2010, 39, 17. 15. J. Wu, K. C.-F. Leung, D. Ben´ıtez, et al., Angew. Chem. Int. Ed., 2008, 47, 7470. 16. F. Coutrot and E. Busseron, Chem. Eur. J., 2008, 14, 4784. 17. F. Coutrot, C. Romuald, and E. Busseron, Org. Lett., 2008, 10, 3741. 18. F. Huang, K. A. Switek, and H. W. Gibson, Chem. Commun., 2005, 3655.

69

28. M. Montalti and L. Prodi, Chem. Commun., 1998, 1461. 29. P. R. Ashton, R. Ballardini, V. Balzani, et al., J. Am. Chem. Soc., 1997, 119, 10641. 30. W. Zhou, J. Li, X. He, et al., Chem. Eur. J., 2008, 14, 754. 31. C.-F. Lin, C.-C. Lai, Y.-H. Liu, et al., Chem. Eur. J., 2007, 13, 4350. 32. P. L. Anelli, N. Spencer, and J. F. Stoddart, J. Am. Chem. Soc., 1991, 113, 5131. 33. P. R. Ashton, M. Blower, D. Philip, et al., New J. Chem., 1993, 17, 689. 34. V. Balzani, A. Credi, S. J. Langford, et al., J. Am. Chem. Soc., 2000, 122, 3542. 35. D. B. Amabilino, C. O. Dietrich-Buchecker, A. Livoreil, et al., J. Am. Chem. Soc., 1996, 118, 3905. 36. A. S. Lane, D. A. Leigh, and A. Murphy, J. Am. Chem. Soc., 1997, 119, 11092. 37. W. Clegg, C. Gimenez-Saiz, D. A. Leigh, et al., J. Am. Chem. Soc., 1999, 121, 4124. 38. M. Asakawa, G. Brancato, M. Fanti, et al., J. Am. Chem. Soc., 2002, 124, 2939. 39. T. D. Ros, D. M. Guldi, A. F. Morales, et al., Org. Lett., 2003, 5, 689. 40. C. M. Keaveney and D. A. Leigh, Angew. Chem. Int. Ed., 2004, 43, 1222. 41. D. S. Marlin, D. G. Cabrera, D. A. Leigh, and A. M. Z. Slawin, Angew. Chem. Int. Ed., 2006, 45, 77. 42. D. A. Leigh and A. R. Thomson, Org. Lett., 2006, 8, 5377. 43. J. D. Crowley, D. A. Leigh, P. J. Lusby, et al., J. Am. Chem. Soc., 2007, 129, 15085. 44. Y.-L. Huang, W.-C. Hung, C.-C. Lai, et al., Angew. Chem. Int. Ed., 2007, 46, 6629. 45. A. Harada, Acc. Chem. Res., 2001, 34, 456. 46. H. Shigekawa, K. Miyake, J. Sumaoka, et al., J. Am. Chem. Soc., 2000, 122, 5411.

19. T. Han and C.-F. Chen, Org. Lett., 2007, 9, 4207.

47. Y. Kawaguchi and A. Harada, Org. Lett., 2000, 2, 1353.

20. M.-L. Yen, W.-S. Li, C.-C. Lai, et al., Org. Lett., 2006, 8, 3223.

48. J. W. Lee, S. Samal, N. Selvapalam, et al., Acc. Chem. Res., 2003, 36, 621.

21. A. Credi, V. Balzani, S. J. Langford, and J. F. Stoddart, J. Am. Chem. Soc., 1997, 119, 2679.

49. S. I. Jun, J. W. Lee, S. Sakamoto, et al., Tetrahedron Lett., 2000, 41, 471.

22. S. A. Vignon, T. Jarrosson, T. Iijima, et al., J. Am. Chem. Soc., 2004, 126, 9884.

50. V. Sindelar, S. Silvi, and A. E. Kaifer, Chem. Commun., 2006, 2185.

23. T. Iijima, S. A. Vignon, H.-R. Tseng, et al., Chem. Eur. J., 2004, 10, 6375.

51. H. Zhang, Q. Wang, M. Liu, et al., Org. Lett., 2009, 11, 3234.

24. M. Han, H.-Y. Zhang, L.-X. Yang, et al., Org. Lett., 2008, 10, 5557.

52. S. Chakrabarti, P. Mukhopadhyay, S. Lin, and L. Isaacs, Org. Lett., 2007, 9, 2349.

25. C. He, Z. Shi, Q. Zhou, et al., J. Org. Chem., 2008, 73, 5872.

53. M. C. Jim´enez, C. Dietrich-Buchecker, and J.-P. Sauvage, Angew. Chem. Int. Ed., 2000, 39, 3284.

26. H. W. Gibson, H. Wang, C. Slebodnick, et al., J. Org. Chem., 2007, 72, 3381.

54. L. Jiang, J. Okano, A. Orita, and J. Otera, Angew. Chem. Int. Ed., 2004, 43, 2121.

27. T.-C. Lin, C.-C. Lai, and S.-H. Chiu, Org. Lett., 2009, 11, 613.

55. A. Livoreil, C. O. Dietrich-Buchecker, and J.-P. Sauvage, J. Am. Chem. Soc., 1994, 116, 9399.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

70

Supramolecular devices

56. J.-P. Collin, P. Gavi˜na, and J.-P. Sauvage, New J. Chem., 1997, 21, 525.

83. D.-H. Qu, Q.-C. Wang, J. Ren, and H. Tian, Org. Lett., 2004, 6, 2085.

57. N. Armaroli, V. Balzani, J.-P. Collin, et al., J. Am. Chem. Soc., 1999, 121, 4397.

84. I. Willner, V. Pardo-Yissar, E. Katz, and K. T. Ranjit, J. Electroanal. Chem., 2001, 497, 172.

58. I. Poleschak, J.-M. Kern, and J.-P. Sauvage, Chem. Commun., 2004, 474.

85. C. A. Stanier, S. J. Alderman, T. D. W. Claridge, and H. L. Anderson, Angew. Chem. Int. Ed., 2002, 41, 1769.

59. B. Korybut-Daszkiewicz, A. Wieckowska, R. Bilewicz, et al., Angew. Chem. Int. Ed., 2004, 43, 1668.

86. Q.-C. Wang, D.-H. Qu, J. Ren, et al., Angew. Chem. Int. Ed., 2004, 43, 2661.

60. P. R. Ashton, V, Balzani, J. Becher, et al., J. Am. Chem. Soc., 1999, 121, 3951.

87. D.-H. Qu, Q.-C. Wang, and H. Tian, Angew. Chem. Int. Ed., 2005, 44, 5296.

61. M. Asakawa, P. R. Ashton, V. Balzani, et al., Angew. Chem. Int. Ed., 1998, 37, 333.

88. A. P. de Silva and N. D. McClenaghan, J. Am. Chem. Soc., 2000, 122, 3965.

62. H.-R. Tseng, S. A. Vignon, and J. F. Stoddart, Angew. Chem. Int. Ed., 2003, 42, 1491.

89. D.-H. Qu, Q.-C. Wang, X. Ma, and H. Tian, Chem. Eur. J., 2005, 11, 5929.

63. S. Nygaard, K. C.-F. Leung, I. Aprahamian, et al., J. Am. Chem. Soc., 2007, 129, 960.

90. A. M. Brouwer, C. Frochot, F. G, Gatti, et al., Science, 2001, 291, 2124.

64. Y. Liu, A. H. Flood, P. A. Bonvallet, et al., J. Am. Chem. Soc., 2005, 127, 9745.

91. G. Bottari, D. A. Leigh, and E. M. P´erez, J. Am. Chem. Soc., 2003, 125, 13360.

65. J. M. Spruell, W. F. Paxton, J.-C. Olsen, et al., J. Am. Chem. Soc., 2009, 131, 11571.

92. A. Altieri, G. Bottari, F. Dehez, et al., Angew. Chem. Int. Ed., 2003, 42, 2296.

66. A. H. Flood, J. F. Stoddart, D. W. Steuerman, J. R. Heath. Science, 2004, 306, 2055.

and

93. S. Bonnet and J.-P. Collin, Chem. Soc. Rev., 2008, 37, 1207.

67. W.-Q. Deng, A. H. Flood, J. F. Stoddart, and W. A. Goddard III, J. Am. Chem. Soc., 2005, 127, 15994.

94. P. Mobian, J.-M. Kern, and J.-P. Sauvage, Angew. Chem. Int. Ed., 2004, 43, 2392.

68. T. Ikeda, S. Saha, I. Aprahamian, et al., Chem. Asian J., 2007, 2, 76. 69. D. Cao, M. Amelia, L. M. Klivansky, et al., J. Am. Chem. Soc., 2010, 132, 1110. 70. K. Moon, J. Grindstaff, D. Sobransingh, and A. E. Kaifer, Angew. Chem. Int. Ed., 2004, 43, 5496. 71. W. Wang and A. E. Kaifer, Angew. Chem. Int. Ed., 2006, 45, 7042. 72. W. S. Jeon, E. Kim, Y. H. Ko, et al., Angew. Chem. Int. Ed., 2005, 44, 87. 73. J. W. Lee, I. Hwang, W. S. Jeon, et al., Chem. Asian J., 2008, 3, 1277. 74. D. Sobransingh and A. E. Kaifer, Org. Lett., 2006, 8, 3247. 75. P. R. Ashton, R. Ballardini, V. Balzani, et al., Chem. Eur. J., 1997, 3, 152. 76. Y. Liu, A. H. Flood, and J. F. Stoddart, J. Am. Chem. Soc., 2004, 126, 9150. 77. E. Katz, O. Lioubashevsky, and I. Willner, J. Am. Chem. Soc., 2004, 126, 15520. 78. A. Altieri, F. G. Gatti, E. R. Kay, et al., J. Am. Chem. Soc., 2003, 125, 8644. 79. G. Fioravanti, N. Haraszkiewicz, E. R. Kay, et al., J. Am. Chem. Soc., 2008, 130, 2593. 80. V. Balzani, A. Credi, and M. Venturi, Chem. Soc. Rev., 2009, 38, 1542.

95. J.-P. Collin, D. Jouvenot, M. Koizumu, and J.-P. Sauvage, Eur. J. Inorg. Chem., 2005, 1850. 96. S. Bonnet, J.-P. Collin, and J.-P. Sauvage, Chem. Commun., 2005, 3195. 97. V. Balzani, M. Clemente-Le´on, A. Credi, et al., Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 1178. 98. R. Ballardini, V. Balzani, M. T. Gandolfi, et al., Angew. Chem. Int. Ed., 1993, 32, 1301. 99. P. R. Ashton, V. Balzani, O. Kocian, et al., J. Am. Chem. Soc., 1998, 120, 11190. 100. W. Abraham, L. Grubert, U. W. Grummt, and K. Buck, Chem. Eur. J., 2004, 10, 3562. 101. R. A. Bissell, E. C´ordova, A. E. Kaifer, and J. F. Stoddart, Nature, 1994, 369, 133. 102. D. A. Leigh, J. K. Y. Wong, F. Dehez, and F. Zerbetto, Nature, 2003, 424, 174. 103. P. R. Ashton, V. Baldoni, V. Balzani, et al., Chem. Eur. J., 2001, 7, 3482. 104. R. Ballardini, V. Balzani, M. Clemente-Le´on, et al., J. Am. Chem. Soc., 2002, 124, 12786. 105. H. Murakami, A. Kawabuchi, R. Matsumoto, et al., J. Am. Chem. Soc., 2005, 127, 15891. 106. C. P. Collier, E. W. Wong, M. Belohradsk´y, et al., Science, 1999, 285, 391. 107. E. W. Wong, C. P. Collier, M. Belohradsk´y, et al., J. Am. Chem. Soc., 2000, 122, 5831.

81. S. Saha and J. F. Stoddart, Chem. Soc. Rev., 2007, 36, 72.

108. C. P. Collier, G. Mattersteig, E. W. Wong, et al., Science, 2000, 289, 1172.

82. H. Murakami, A. Kawabuchi, K. Kotoo, et al., J. Am. Chem. Soc., 1997, 119, 7605.

109. M. Asakawa, M. Higuchi, G. Mattersteig, et al., Adv. Mater., 2000, 12, 1099.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular devices: molecular machinery

71

110. C. P. Collier, J. O. Jeppesen, Y. Luo, et al., J. Am. Chem. Soc., 2001, 123, 12632.

137. B. L. Feringa, W. F. Jager, and B. de Lang, J. Am. Chem. Soc., 1991, 113, 5468.

111. D. W. Steuerman, H.-R. Tseng, A. J. Peters, et al., Angew. Chem. Int. Ed., 2004, 43, 6486.

138. W. F. Jager, J. C. de Jong, B. de Lang, et al., Angew. Chem. Int. Ed., 1995, 34, 348.

112. W.-Q. Deng, R. P. Muller, and W. A. Goddard III, J. Am. Chem. Soc., 2004, 126, 13562.

139. B. L. Feringa, N. P. M. Huck, and H. A. van Doren, J. Am. Chem. Soc., 1995, 117, 9929.

113. S. S. Jang, Y. H. Jang, Y.-H. Kim, et al., J. Am. Chem. Soc., 2005, 127, 1563.

140. B. L. Feringa, N. P. M. Huck, and A. M. Schoevaars, Adv. Mater., 1996, 8, 681.

114. S. S. Jang, Y. H. Jang, Y.-H. Kim, et al., J. Am. Chem. Soc., 2005, 127, 14804.

141. N. Koumura, R. W. J. Zijlstra, R. A. van Delden, et al., Nature, 1999, 401, 152.

115. Y. J. Jang, S. Hwang, Y.-H. Kim, et al., J. Am. Chem. Soc., 2004, 126, 12636.

142. B. L. Feringa, R. A. van Delden, N. Koumura, E. M. Geetsema, Chem. Rev. 2000, 100, 1789.

116. Y. H. Jang, S. S. Jang, and W. A. Goddard III, J. Am. Chem. Soc., 2005, 127, 4959.

143. B. L. Feringa, Acc. Chem. Res., 2001, 34, 504.

117. J. E. Green, J. W. Choi, A. Boukai, et al., Nature, 2007, 445, 414. 118. R. Hernandez, H.-R. Tseng, J. W. Wong, et al., J. Am. Chem. Soc., 2004, 126, 3370. 119. S. Saha, K. C.-F. Leung, T. D. Nguyen, et al., Adv. Funct. Mater., 2007, 17, 685. 120. T. D. Nguyen, K. C.-F. Leung, M. Liong, et al., Adv. Funct. Mater., 2007, 17, 2101. 121. T. D. Nguyen, H.-R. Tseng, P. C. Celestre, et al., Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 10029. 122. T. D. Nguyen, Y. Liu, S. Saha, et al., J. Am. Chem. Soc., 2007, 129, 626. 123. T. D. Nguyen, K. C.-F. Leung, M. Liong, et al., Org. Lett., 2006, 8, 3363. 124. S. Angelos, K. Patel, J. F. Stoddart, and J. I. Zink, Angew. Chem. Int. Ed., 2008, 47, 2222. 125. K. Patel, S. Angelos, W. R. Dichtel, et al., J. Am. Chem. Soc., 2008, 130, 2382. 126. L. Fang, M. Hmadeh, J. Wu, et al., J. Am. Chem. Soc., 2009, 131, 7126. 127. P. G. Clark, M. W. Day, and R. H. Grubbs, J. Am. Chem. Soc., 2009, 131, 13631.

and

144. M. K. J. ter Wiel, R. A. van Delden, A. Meetsma, and B. L. Feringa, J. Am. Chem. Soc., 2003, 125, 15076. 145. J. Vicar´ıo, M. Walko, A. Meetsma, and B. L. Feringa, J. Am. Chem. Soc., 2006, 128, 5127. 146. N. Koumura, E. M. Geertsema, A. Meetsma, B. L. Feringa, J. Am. Chem. Soc., 2000, 122, 12005.

and

147. N. Koumura, E. M. Geertsema, M. B. van Gelder, et al., J. Am. Chem. Soc., 2002, 124, 5037. 148. R. A. van Delden, M. K. J. ter Wiel, M. M. Pollard, et al., Nature, 2005, 437, 1337. 149. W. R. Browne and B. L. Feringa, Nat. Nanotech., 2006, 1, 25. 150. G. T. Carroll, M. M. Pollard, R. van B. L. Feringa, Chem. Sci., 2010, 1, 97.

Delden,

and

151. J. Vicario, N. Katsonis, B. S. Ramon, et al., Nature, 2006, 440, 163. 152. D. Qu and B. L. Feringa, Angew. Chem. Int. Ed., 2010, 49, 1107. 153. N. Ruangsupapichat, M. M. Pollard, S. R. Harutyunyan, and B. L. Feringa, Nat. Chem., 2011, 3, 53. 154. T. Muraoka, K. Kinbara, Y. Kobayashi, and T. Aida, J. Am. Chem. Soc., 2003, 125, 5612. 155. K. Kinbara and T. Aida, Chem. Rev., 2005, 105, 1377.

128. M. A. Olson, A. B. Braunschweig, L. Fang, et al., Angew. Chem. Int. Ed., 2009, 48, 1792.

156. T. Muraoka, K. Kinbara, and T. Aida, Nature, 2006, 440, 512.

129. C. P. Mandl and B. K¨onig, Angew. Chem. Int. Ed., 2004, 43, 1622.

157. T. Muraoka, K. Kinbara, and T. Aida, Chem. Commun., 2007, 1441.

130. T. Elston, H. Wang, and G. Oster, Nature, 1998, 391, 510.

158. T. Muraoka, K. Kinbara, and T. Aida, J. Am. Chem. Soc., 2006, 128, 11600.

131. O. S. Akkerman and J. Coops, Rec. Trav. Chim. Pays-Bas, 1967, 86, 755.

159. Y. Shirai, A. J. Osgood, Y. Zhao, et al., Nano. Lett., 2005, 5, 2330.

132. F. Cozzi, A. Guenzi, C. A. Johnson, et al., J. Am. Chem. Soc., 1981, 103, 957.

160. Y. Shirai, A. J. Osgood, Y. Zhao, et al., J. Am. Chem. Soc., 2006, 128, 4854.

133. T. R. Kelly, M. C. Bowyer, K. V. Bhaskar, et al., J. Am. Chem. Soc., 1994, 116, 3657.

161. Y. Shirai, J.-F. Morin, T. Sasaki, et al., Chem. Soc. Rev., 2006, 35, 1043.

134. T. R. Kelly, Acc. Chem. Res., 2001, 34, 514.

162. J.-F. Morin, Y. Shirai, and J. M. Tour, Org. Lett., 2006, 8, 1713.

135. T. R. Kelly, H. de Silva, and R. A. Silva, Nature, 1999, 401, 150. 136. T. R. Kelly, X. Cai, F. Damkaci, et al., J. Am. Chem. Soc., 2007, 129, 376.

163. T. Sasaki, and J. M. Tour, Org. Lett., 2008, 10, 897. 164. Y. Shirai, T. Sadaki, J. M. Guerrero, et al., ACS Nano, 2008, 2, 97.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

72

Supramolecular devices

165. S. Khatua, J. M. Guerrero, K. Claytor, et al., ACS Nano, 2009, 3, 351.

167. G. Vives, J. Kang, K. F. Kelly, and J. M. Tour, Org. Lett., 2009, 11, 5602.

166. G. Vives, and J. M. Tour, Acc. Chem. Res., 2009, 42, 473.

168. J. F. Stoddart, Nat. Chem., 2009, 1, 14.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc092

Molecular Logic Gates A. Prasanna de Silva Queen’s University, Belfast, UK

1 Introduction 2 Illustrative Examples of Molecular Logic Gates 3 Conclusion Acknowledgments References

1

1 2 8 8 9

INTRODUCTION

The phenomena of supramolecular chemistry are particularly suitable for use in molecular computation since the dynamic and reversible interaction of a guest species with a host molecule can be viewed easily as an input applied to a device. The output from the device would be any conveniently observable property which signals the interaction. Indeed, the first experimental demonstration of molecular logic-based computation was based on a quintessential supramolecular interaction between an alkali cation and a crown ether macrocycle.1 However, it is now clear that most phenomena of chemistry and biochemistry can be viewed as general computational operations2 provided that resettability and speed are downgraded in importance. The latter two points arise since (i) irreversibly bound, but modular, systems have been involved in the conceptual development of supramolecular science and (ii) ion–molecule interactions

occur over millisecond timescales or longer. An important early goal of molecular computation is the molecularlevel emulation of Boolean logic gates, which are the workhorses of modern semiconductor-based computers3, 4 and which have a long history in mathematics.5, 6 These molecular logic gates form the focus of our attention in this review. Logic itself grew as a subject from the work of Aristotle, but its involvement with mathematics had to wait for George Boole.5 He saw logic patterns in the common use of language, which he interpreted algebraically. This led to the discovery of various logic functions such as YES, NOT, AND, OR, NOR, NAND, INHIBIT, and XOR (Figure 1), which processed the truth or falsity of one or two statements in characteristic ways.6 However, it was the use of logic functions in electronic devices in general and computers in particular that really got the public’s attention. When logic functions are manifested as material objects, they are logic gates. These are important components of the electronic engineer’s toolbox.3, 4 In the context of logic, it is important to bear in mind the relationship between language, mathematics, electronic engineering, chemistry, and biology. Language is where logic was first appreciated. Mathematics is where logic was formalized in terms of functions. Electronic engineering is where logic functions took on material form in terms of semiconductor structures. Chemistry is where logic functions take on material form in terms of molecular structures. Biology is where molecular logic devices can operate within very small but vital spaces where the electronic engineering devices cannot go. The overall aim of molecular logic must be to demonstrate its applicability and usefulness in as many areas as possible.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

2

Supramolecular devices

Input 0 1

Input1 Input2 0 0 1 1

0 1 0 1

Table 1

Output 0 1

1 0

YES

NOT

0 1 1 1

1 0 0 0

AND

OR

NOR

1 1 1 0

0 1 0 0

0 1 1 0

NAND INHIBIT XOR

Figure 1 Electronic engineering symbols and truth tables for YES, NOT, AND, OR, NOR, NAND, INHIBIT, and XOR logic gates.

2

ILLUSTRATIVE EXAMPLES OF MOLECULAR LOGIC GATES

There are already hundreds of molecular logic gates that could be featured in a general introductory review of this kind. Many of these have already been covered in the review literature.2, 7–19 For the present purpose, we have allowed our choice to be dictated by the diversity of logic types and mechanisms. However, fluorescent cases will dominate the discussion because of their relative ease of observation even down to the single-molecule level.

2.1

Input H+

Output Fluorescencea

0 (10−10 M) 1 (10−4 M)

0 (0.002) 1 (0.41)

a Quantum

Output 0 0 0 1

Truth table for 1.

YES logic

This is perhaps the simplest logic gate of all, once nonswitching cases are put aside. Nonswitching cases are gates where inputs have no effect on the true–false output. A YES gate simply passes whatever bit (0 or 1) comes its way with no processing (Figure 1). Indeed, electronic engineers consider this case to be utterly trivial since any point along a conducting wire satisfies this criterion. Specifically, a YES gate produces an output of 1 or 0 when the input is 1 or 0, respectively. Molecule 120, 21 employs a general design. In this case, the fluorophore (anthracene) and the receptor (amine) have to communicate across the short σ -bond spacer. Photoinduced electron transfer (PET) can perform this communication if the thermodynamics are favorable.22, 23 PET usually outruns fluorescence under such conditions. Experimentally, this means that the fluorescence decay rates are slower when compared with the rates of PET. However, PET thermodynamics can be altered to become unfavorable

yields in methanol : water (1 : 4, v/v), λexc 366 nm, λem 398, 420, 444 nm.

when the receptor binds the input species. Fluorescence is therefore switched “on” from the “off” state when the concentration of H+ is raised from “low” to “high” across a threshold defined by the binding constant of 1 and H+ (related to its pKa value). This corresponds to YES logic (Figure 1) with H+ input and fluorescence output as shown in the truth table (Table 1). “High” or “low” levels of a signal are represented by the symbols 1 and 0 in this and the following truth tables. Cases such as 1 can be seen to abound among classical pH indicators,24 whether they produce changes in fluorescence or absorbance. OH N

OH

1

Electrochemical properties such as redox potential can also serve as inputs to molecular YES logic gates since redox indicators also have a rather long history.24 Complex 225 is poorly emissive at 610 nm (when excited at 453 nm) as a result of PET from the ruthenium-based lumophore to the benzoquinone moiety across the dimethylene spacer. If a potential of −0.6 V (vs standard calomel electrode SCE) is applied to it in wet acetonitrile, the benzoquinone unit is fully reduced to the hydroquinone form once two coulombs are passed per equivalent of 2. This succeeds because the reduction potential of 2 is −0.44 V. Upon reduction, a luminescence enhancement factor of 6 is seen. The system is stable in both the benzoquinone and hydroquinone forms in the absence of applied potentials, which means that these cases show evidence of memory unlike ion-driven cases such as 1. However, the hydroquinone form can be reset to 2 by application of two coulombs at a potential of +1.1 V. Solvent polarity can also serve as an input and 3 produces a fluorescence output according to YES logic. Changing the solvent from hexane to the more polar methanol : water (1 : 4, v/v) produces a jump in the fluorescence quantum yield. In line with older data on simple coumarins,26 3 shows a nπ –π π* state inversion with increasing polarity and the ππ* state becomes the lower energy form.27

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

Molecular logic gates

2+

Table 2

3

Truth table for 5.

O

N

N Ru(II) N N

O

0 (15◦ C) 1 (60◦ C)

1 (0.56) 0 (0.24)

a

2

CH3

O

O

CH3O 3

Even physical properties such as temperature can serve as input to molecular YES logic devices producing fluorescence output. These are a subset of luminescent molecular thermometers.28, 29 Very sensitive thermometers of the YES logic type are acrylamide polymers labeled with benzofurazan-containing fluorophore, 4.30 The fluorescence of the benzofurazan unit drops with increasing solvent polarity, whereas the acrylamide polymer molecules form a low-polarity globular structure at high temperatures. So, the combined system switches “on” fluorescence as the temperature increases beyond a quite sharp threshold.

x

* CO2

y

CONH

N

N O N SO2N(CH3)2 4

2.2

Output Fluorescencea

Quantum yields in hexane, λexc 380 nm, λem 436 nm.

CH3

2

CH3

Input Temperature

NOT logic

The NOT logic gate, or the inverter, is popular with electronic engineers. As far as the output is concerned, it is a clear case of perpetual opposition to the inward signal. NOT logic is also key in human language as a negation of associated statements. It can be a lawyer’s delight. As Figure 1 shows, a NOT gate produces an output of 1 or 0 when the input is 0 or 1, respectively. The Arrhenius relationship between rate constants and temperature is well known to chemistry undergraduates.31

When such a process is made to compete with fluorescence, it is clear that the observable emission intensity can decrease strongly with increasing temperature in suitable instances such as 5.32 The process of interest in 5 is a photodissociation of the N1–C5 bond, which is largely reversible owing to intramolecular radical recombination within the solvent cage. Since the temperature is smoothly signaled by the fluorescence intensity, 5 can also serve as a luminescent thermometer,28, 29 although less sensitive than cases such as 4.30 Molecule 5 is also a NOT logic gate (Figure 1) with temperature input and fluorescence output as given in the truth table (Table 2).

N

N

5

2.3

AND logic

While emulating Boolean YES and NOT operations at the molecular level are well and good, two-input logic needs molecular counterparts before the field can be considered to be worthwhile in any significant way. AND logic is a particularly important target since it is one of the commonest gates in the electronic engineer’s box of tricks. AND logic arises when an output of 1 is produced only when two inputs are simultaneously 1 (Figure 1). It is also a common way of tightly associating two or more statements in language. For instance, Google
throws in an AND logic operator between any two search terms separated by a space so as to sharpen the selectivity of the search. We think it is important to note the very human nature of AND logic as exemplified by the slogan “Unity is strength” and many variants thereof. While observations of this kind are necessarily imprecise and anecdotal, they are important to note for several reasons. If molecular logic is to prosper, it is important to recognize the pervasiveness of Boolean logic in everyday life and language. As noted in the introduction, Boolean logic grew out of language analysis and this point needs to be emphasized. It is through the recognition of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

4

Supramolecular devices Table 3

Truth table for 6.

Input1 H+ 0 0 1 1

(0 M) (0 M) (10−3 M) (10−3 M)

0 1 0 1

Input2 Na+

Output Fluorescencea

(0 M) (10−2 M) (0 M) (10−2 M)

0 0 0 1

(0.012) (0.013) (0.020) (0.068)

a Quantum

yields in methanol : 2-propanol (1 : 1, v/v), λexc 387 nm, λem 446 nm.

logic in common phenomena that chemists can apply similar recognitions to chemical effects. Much of the molecular logic literature has arisen in this way. Consider molecule 6 which is designed on the basis of the PET process mentioned previously. However, 6 contains two receptors, each of which can launch a PET process to the fluorophore.1 Thus, fluorescence is produced as the output only when both chemical inputs (H+ and Na+ ) are present at high enough concentrations in order to block the two receptors—the amine and the benzocrown ether, respectively. This is the AND logic condition, which is summarized in the truth table (Table 3). We note that if even one of these two receptor sites is unoccupied by its corresponding input species, PET will still occur from it and no fluorescence will be seen.

O N

Oligonucleotides can match proteins at the logic game. Gate 735 carries a fluorescein fluorophore at the 3 -end. Input1 is the complementary oligonucleotide 8. Input2 is the minor groove-binder 9. Compound 9 can only bind when the duplex is formed between 7 and 8. Then the fluorescein unit of the short oligonucleotide 7 is rather close to 9. If 9 is excited at 350 nm, we have efficient EET23, 33 from 9 to fluorescein and a strong fluorescence emerges at 520 nm. Since the fluorescein unit absorbs poorly at 350 nm, strong emission at 520 nm cannot be arranged in any other simple and legitimate way. O O 5′-GCCAGAACCCAGTAGT-3′- NHC(=S)NH

O

7 OH

3′-CGGTCTTGGGTCATCA-5′ 8

OH CH3

N N

N N

NH

N H

O

9

O O

O−

O

CN 6

Biomolecules can produce AND logic gates too. Cytochrome c possesses a strongly colored heme unit at its active site. Any attempt to observe fluorescence from a tryptophan unit close by will lead to failure. This is because the emission will be heavily quenched owing to electronic energy transfer (EET)23, 33 from the tryptophan to the heme.However, this situation only applies to the native state of the protein. If the protein is denatured and its tertiary structure is disrupted, the tryptophan unit gets far enough away from the heme so as to give out a good fluorescence signal. Konermann34 carefully chooses urea and H+ inputs at such concentrations (4.8 M and 1.3 × 10−3 M, respectively) so that neither can achieve the denaturing action by itself. However, when applied together, they reversibly denature the protein.

Now we move to an example of an AND gate involving a photochromic system. Compound 10 is such a gate where a light dose is employed as input1 and H+ serves as input2 . trans-10 isomerizes to its cis-form when it receives a dose of light at 365 nm. The cis-form cyclizes in the presence of sufficient H+ to give the fluorophore 11 which emits at 515 nm.36–38 The 515-nm emission emerges only under these rather specific conditions of a light dose at 365 nm and a high concentration of H+ . HO

OH

O 10

OH +

O

11

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

Molecular logic gates

2.4

OR logic

5

OR logic when the complexation event is taken as the output.

OR logic is also one of the commonest gates at the electronic engineer’s disposal. It is also a common way of building in flexibility (by OR-ing two or more statements) in language. OR logic arises (Figure 1) when an output of 1 is produced if any one of two inputs achieves a state of 1. Compound 1239 can launch a PET process to the pyrazoline fluorophore from the aromatic amino acid receptor.40 Ca2+ or Mg2+ can be supplied at high enough concentrations to bind to the receptor so that PET is stopped and fluorescence is produced. The truth table (Table 4) shows similar fluorescence enhancement factors due to the cation-induced conformational changes in the receptor.

N

N N N

N

N 13

O

O 14 CO2− N

CO2− CO2−

O CH3O2S

N N 12

15 CN

While a conveniently observable signal is best for use as an output from a molecular device, it is not essential. Here is an example which employs NMR line shifts, among other experiments, to deduce a complexation event. Fujita’s host, self-assembled from six (diaminoethane)Pd2+ and four exotridentate 13 ligands,41 gives the mixed complex 14·15·host if it is presented with the small aromatic guests 14 and 15 as inputs. If only 14 is presented, it will form the complex 142 ·host. Similarly, the complex 152 ·host is produced if only 15 is presented to the host. In other words, the host will form a complex of some kind whether one or the other or both the guests are offered to it. This is, therefore,

Table 4

Ferrocene derivative 16 has an electron-withdrawing dimesitylboranyl substituent, which leads to an easily accessible reduction potential.42 Upon encountering the Lewis base CN− (input1 ), the substituent changes to an electronreleasing cyanoborate so that the reduction potential jumps to a much larger negative figure. Tetrazolium violet 17 has an intermediate reduction potential so that a mixture of 16 and 17 in the presence of CN− leads to the reduced form of 17, the deep violet color (or the absorbance at 500 nm) of which is a convenient output. The tetrazolium dyes are well known as sensors of reducing environments in cells.43 F− can be used as input2 to give the same final result because of the similar anion-induced change in potential. The fact that CN− and F− produce the same absorbance output is evidence of OR logic.

Truth table for 12. Ph

Input1 Ca2+

Input2 Mg2+

O

Output Fluorescencea

B O

0 0 1 1

(0 M) (0 M) (10−3 M) (10−3 M)

a Quantum

0 1 0 1

(0 M) (0.5 M) (0 M) (0.5 M)

0 1 1 1

(0.0042) (0.24) (0.28) (0.28)

yields in water at pH 7.3, λexc 389 nm, λem 490 nm.

Ph

Fe(II)

(CH3)5 16

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

6

Supramolecular devices

N

N + N N

17

2.5

NOR logic

NOR logic is considered a universal gate by electronic engineers since it is possible to wire NOR gates into arrays to reproduce any Boolean logic operation that is desired.3, 4 However, molecular NOR gates which are presently available cannot claim this universality since “wiring” is anything but straightforward (see below). NOR logic (Figure 1) requires that the output of 1 is produced only when both inputs are in state 0. Compound 18 contains an anthracene fluorophore and a methylene spacer, as well as a 2,2 -bipyridyl receptor which can bind either H+ or Zn2+ .44 Either of these two binding events makes the 2,2 -bipyridyl unit a better electron acceptor because of its positive charge and its enforced planarity.45, 46 Therefore, a PET process becomes possible from the anthracene fluorophore to the bound 2,2 bipyridyl receptor. The fluorescence output is therefore quenched in the presence of either input or both. The truth table (Table 5) shows how this leads to NOR logic. N

N

2.6 CH3

18

Simple as this case of 18 may be, it represents a way of overcoming the difficulties of integrating molecular logic gates. The integration aspect is clear when we recognize that an all-electronic NOR gate can be constructed by Table 5

Truth table for 18.a

Input1 H+ 0 0 1 1 a

(10−7 (10−7 (10−2 (10−2

M) M) M) M)

0 1 0 1

Input2 Zn2+

Output Fluorescenceb

(0 M) (10−3 M) (0 M) (10−3 M)

1 0 0 0

(1.0) (0.13) (0.13) (0.12)

10−5 M in methanol : water (1 : 1, v/v). quantum yields, λexc 368 nm, λem 405, 425, 440 nm.

b Relative

feeding the output of an OR gate as the input of a NOT gate. This can be clearly observed in the electronic symbol for NOR logic in Figure 1, where symbols for NOT and OR are linked in series. However, molecular logic gates tend to have inputs and outputs differing qualitatively and/or quantitatively. Therefore, physical integration tends to be very difficult. On the other hand, a functional integration of NOT and OR logic operations to achieve the same goal is possible. Functional integration is an important concept in molecular logic,44 where the input–output relationship of a given molecular behavior is analyzed in terms of a combination of simpler logic operations such as NOT, AND, and OR. While this idea is commonly used in the analysis of electronic logic arrays,3, 4 its application to molecular behavior requires careful attention. For instance, it is particularly beneficial to apply this method to molecular behaviors driven by three or more inputs. However, for an introductory review of this kind, the method is best described with a simpler case. The case of 18 involved consideration of chemical-induced fluorescence quenching as a NOT logic operation. The fact that H+ and Zn2+ produced the same end result (of fluorescence quenching) involved an OR logic. Their combination, although involving no physical gate-to-gate linking, gave NOR logic when the output was reasonably taken as the fluorescence intensity (or quantum yield). Such ideas of integrating several logic functions within a single molecular structure47 are now quite widespread.

NAND logic

NAND logic is also in the universal gate category, according to electronic engineers.3, 4 The current inability of molecular-level counterparts to inherit this mantle has been mentioned previously in Section 2.5. NAND logic means that the output achieves the state 0 only when the inputs are both simultaneously 1. The crown ether 19, which helped to launch the entire subject of supramolecular chemistry,48 has a significant fluorescence emission in the ultraviolet region. This is changed only slightly when 19 is presented with K+ and counteranions, provided that the latter is not I− .49 This lack of a fluorescence change is in contrast to the clear evidence from many quarters concerning the crown ether-K+ interaction. However, if I− is employed as the counteranion, a strong quenching of fluorescence is found. Although encircled by the crown ether, the K+ ion is exposed along the axial directions of the macrocycle. The anion approaches and ion pairs with the K+ in this way. I− , being very oxidizable, can launch a PET process to the dialkoxybenzene fluorophore close by. This is one mechanism for fluorescence quenching. Another

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

Molecular logic gates mechanism would be the heavy atom effect due to the iodine nucleus.50 The overall phenomenon is a case of NAND logic since both input1 (K+ ) and input2 (I− ) are required for the fluorescence output to become “low.” O O

O

O

O O 19

2.7

INHIBIT logic

of 2 equivalents of H+ diprotonates the cyclam within 21 so that it binds to two cyanide groups of the dianionic 22. This brings 22 close to the naphthalene units so that EET occurs to quench the naphthalene emission at 335 nm. This is the starting state for the XOR logic gate. Addition of more acid (input1 ) protonates 22 so that it is no longer bound to 21. Thus, EET stops and the 335-nm emission is resurrected. Addition of a tertiary amine (input2 ) restores the neutral cyclam within 21, which also causes detachment of 22. Again, the 335-nm emission occurs. Addition of equal amounts of acid and tertiary amine causes no net effect due to neutralization, and hence the starting state (with its quenched emission) is maintained.

Unlike the gates discussed so far, INHIBIT logic is noncommutative. Figure 1 shows how input1 , when held at logic state 1, ensures that the output will be 0. Input2 has no such power of veto. In the Boolean analysis,4 there is another INHIBIT logic gate where the disabling ability is in the hands of input2 instead, but this distinction is not critical for the development of molecular INHIBIT logic operations. Previous paragraphs have discussed how PET processes quench fluorescence. However, 2051 is a case where the PET process gives rise to a charge transfer (CT) emission under certain conditions. This happens when the Noxide receptor1 binds H+ ,52, 53 since 20 then becomes a donor–acceptor system. Compound 20 also contains a benzocrown ether receptor2 , which binds K+ . PET is therefore stopped and takes the CT emission down with it. So this is an H+ , K+ -driven INHIBIT gate where K+ is the disabling input. O

O

O

O O

O O

O

O

N+

O−

N N O

O

O

O

N N O O

O

O

O

O O

O

O

O

21

O 2−

N

O

Ru[II](CN)4 N 22

20

2.8

O

O

O O

7

XOR logic

XOR logic is valued by electronic engineers because it compares the logic state of two input signals. If they are the same, an output 0 is sent. If they are different, an output of 1 emerges (Figure 1). A system composed of 21 and 22 can give rise to XOR logic behavior.54 Dendrimer 21 has a cyclam receptor as its core and a set of naphthalene fluorophores as the shell. PET effects are unimportant in this case owing to the long spacers between the fluorophores and the receptor. Addition

2.9

Half-adder

Although many relatively complex logic systems are now available,2, 7–19 we will only consider one arithmetic processor—the half-adder—in this review. Many interesting molecular-scale arithmetic systems have come to light in recent years.55–71 Since unknown molecular processes in the brain have allowed humans to be numerate, it was important to demonstrate numeracy with synthetic molecules. The simplest semiconductor-based arithmetic device is the half-adder, which has two inputs and two output channels.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc093

8

Supramolecular devices

Addition of the two binary inputs uses AND logic to generate the carry digit and XOR to produce the sum digit.3, 4 This is the concept underlying essentially all the intrinsically molecular-scale arithmetic systems available today. We focus on a simple and clear case of this kind.71 The inputs are Ca2+ and H+ , while the outputs are transmittance at 390 nm for the sum digit and fluorescence quantum yield for the carry digit. PET-based “fluorophore-spacer1 receptor1 -spacer2 -receptor2 ” 2371 is the AND gate. Compound 23 is operated in parallel with the XOR gate 24 simply by using them as an equimolar mixture in the same solution.71 Compatibility of the two gates was checked carefully so that each would maintain its properties to a significant extent in the presence of the other. This simple system of two components shows the emergent behavior of numeracy. The experimental truth table for the half-adder is given in Table 6. − − CO2− CO2− CO2 CO2

N

N O

O

CH3

Table 6 Truth table for the systems 23 and 24.a Input1 H+ 0 0 1 1

(10−9.5 M) (10−9.5 M) (10−6 M) (10−6 M)

Input2 Ca2+ 0 1 0 1

( 0

0.5

Voltage V / volts

−1

O

0

Voltage / V

−0.5

4

1.5

5

2 (c)

(f)

−2×10−9 −1.5 Voltage / volts

(h)

1×10−8

2×10

−8

3×10−8

4×10−8

5×10−8

−0.002 −3

−0.001

0

0.001

0.002

0.003

0.004

−0.005 −2

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

6 1

−2×10−8 −1.5

−1×10−8

0

0.5

Electron flow for v>0

1

Electron flow for V > 0

Bias V / volts

0

Top Au pad 2

Au Bottom electrode

0

−0.5

N

N

N

N

−

0

O

O

Au

−0.5

0

C

C

Bias V / volts

Bottom Au electrode

Top Au pad 1

−1

−1

−2

N

C

Electron flow for v > 0

N+

5×10−7

2×10−8

4×10−8

6×10

8×10

−9

1×10−8

1.2×10−8

−0.04 −1.5

−0.02

0

0.02

0.04

0.06

−2 −3

−1

0

1

2

3

4

5

−1

S

S

−1

S

S

−2

Cell 1 Cell 2 Cell 3 Cell 4 Cell 5 Cell 6

−1.5

N

I

−

O

−0.5

Au

O

O

Au

−1

d.

0

NO2

NO2

Voltage / V

0

N

NO2

C C

O

1

1

0.5

1

1.5

Cycle 9

Cycle 5

Cycle 4

Cycle 3

Cycle 2

Cycle 4 Cycle 5 Cycle 6

Cycle 3

Cycle 2

Cycle 1

3

2

Cycle 1

1.5

2

Electron flow for v >0

Electron flow for V > 0

0.5

Voltage / V

N

S

0

N

Bias / volts

−0.5

Bottom Au electrode

N

Top Au pad

Figure 6 The rectification of MOM sandwiches consisting of three elements: (i) a macroscopic bottom Au or Al electrode, (ii) an 0.3-mm2 top Al or “cold Au” electrode pad, and, between them: (iii) (a) an LB monolayer of 1127 ; (b) an LB monolayer of 1460 ; (c) an LB monolayer of 1554 ; (d) an LB monolayer of 1655 ; (e) an LS monolayer of 1756 (the repeated scans are offset for clarity); (f) an LB monolayer of 1857 ; (g) an LS monolayer of 1957 ; (h) an LS monolayer of 20.59

(g)

Current / Amperes

Current / A Current I / microAmperes

Current / milliAmperes Current / Amperes Current / Amperes

10 Supramolecular devices

Single-molecule electronics

11

Table 1 Summary data for 11 unimolecular rectifiers 11–21 studied at the University of Alabama. table All compounds were measured at room temperature in air between Au electrodes inside a Faraday cage (11 was also measured earlier between Al electrodes at 300 K,25 also between 105 K and 370 K51 ; 21 was also studied at 4.2 K). The column “# pads” lists how many independent typical MOM pads were discussed in each publication as rectifying (out of hundreds measured). LB transfer pressure = film pressure at which LB monolayers were transferred (close to collapse pressure), IVT = maximum of the optical intervalence transfer or intramolecular charge transfer band. RR = rectification ratio (6). “Survives cycling” means RR does not decrease with cycling of I –V measurements. “U, A, or S” refers to unimolecular, asymmetry, or Schottky rectification. “1-level or 2-level?” describes whether the I –V curves satisfy the AR 2-level (LUMO + HOMO) model or a 1-level model (LUMO only). #

11 12 13 14 15 16 17 18 19 20 21

Type

LB transfer pressure (mN/m)

D+ -π -A− D+ -π -A− D+ -π -A− D+ -π -A− D+ I− D-σ -A D-σ -A D-σ -A D-σ -A D-σ -A D-σ -A

20 20 20 28 22 22 23 32 35 35 20

# Pads

16 — 3 3 24 1 9 4 1 — —

IVT (nm)

530 530 530 504 480 720 — — 595 1220 —

LB or LS

LB LS LS LB LB LB LB LB LB LS LS

Using the atom positions of the crystal structure in the semiempirical computer program INDO, the computed ground-state static electric dipole moment was a whopping 26.2 Debyes (8.725 × 10−26 cm). P-3CNQ crystallizes in the centrosymmetric space group P 21 /c.69 The molecule (Figure 8) is not planar: the least-squares plane for the picolinium ring forms an angle of 30.13◦ with the least-squares plane of the six-membered central ring of the 3CNQ part. For a single crystal of 22, an interionic charge transfer band with peak at 12 400 cm−1 (or λmax = 806 nm) and an ICT band (or intravalence band, IVT) at 18 600 cm−1 (or λmax = 538 nm) were found, and unambiguously confirmed by polarized spectroscopy.71 Ashwell exploited this chemistry to prepare many zwitterionic analogs of 22 that held promise as nonlinear optical materials with high second-order polarizability β. Indeed, 11 forms Z-type LB multilayers with a very high, if resonant, χ (2) zzz = 180 pm V−1 .72 The blue narrow absorption band in the solution may be due to an IVT band; when several molecules are packed in a multilayer, an additional ICT band may also occur. The synthetic yield of 11 was vastly increased when it was carried out in a mixture of dry dimethyl sulfoxide and pyridine25 : the late M. V. Lakshmikantham found25 that the transformation (Figure 9), promoted by base, occurred via an electron-transfer process, and involved two equivalents of the ion-radical salt Li+ TCNQ•− . The lepidinium (i.e., N-alkyl-4-metylquinolinium) salt (e.g., lepidinium bromide or tosylate) loses H• to the first Li+ TCNQ•− , and was converted to a radical cation salt, which next coupled with

RR

Survives cycling?

2–27 5 5 3–64 8–60 2 10 2–5 30 13 6–60

No Yes Yes No No — Yes No Yes No Yes

U, A, or S?

U, U, U, U, A A A U, U, U U

A A A A

A A

1- or 2-level?

2? 1? 1? 2? 1 1 1 2? 2? 2? 2

Ref.

27 58 58 59 54 55 56 57 57 59 40

more free Li+ TCNQ•− , to give a zwitterionic adduct intermediate; finally, a spontaneous loss of HCN led to the product 11.25 Adding an acidic proton to 11 will create the cation 23 that is no longer blue.25 It is believed25, 71, 72 that the ground state of 11 is very polar and zwittewrionic, T-D+ -π-A− , while its first excited state T-D-π-A is less polar. (Here, “T” is the alkyl “tail”). In 11, the quinolinium and tricyanoquinodimethanide rings may be twisted by some angle θ (θ = 30◦ for the similar zwitterion 2269 ) because of steric hindrance. The neutral excited state T-D-π -A might be more planar (θ ≈ 0◦ ); if θ ≈ 90◦ then IVT = 0, and the blue color must disappear. Thus 11,25–27, 51 is a ground-state zwitterion D+ -π-A− , connected by a twisted “pi” bridge (not a “sigma” bridge): it is sparingly soluble in polar solvents; its ground-state static electric dipole moment is µGS = 43 ± 8 Debyes at infinite dilution in CH2 Cl2 .25 It has an hypsochromic IVT optical absorption band in solution between 600 and 900 nm,25, 65 which fluoresces in the near-IR.63 From the hypsochromic shift, the calculated excited-state dipole moment is µES = 3–9 Debyes.63 Since the molecule exhibits a hypsochromic shift in its absorption spectrum, the ground state was definitely D+ -π A− , and the first electronically excited state was D0 -π-A0 . Thanks to a finite twist angle θ (0 < θ < 90◦ ) between the quinolinium ring and the tricyanoquinodimethanide ring, an IVT band between the D+ and A− ends of the molecule is observed, polarized perpendicular to the monolayer, at λmax = 535 nm60 : this corresponds to the λmax = 538 nm band observed for the IVT band of P3CNQ.71 There was also an intermolecular aggregate peak

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc094

12

Supramolecular devices D

A X

X

R N

N

CN CH2NH2

NC

NC

X X 11: X = H, R = n -C16H33 12: X = H, R = CH3C(O)S-C14H28 13: X = H, R = CH3C(O)S-C16H32 14: X = F, R = n -C16H33 25: X = H, R = CH3C(O)S-C8H16 26: X = H, R = CH3C(O)S-C10H20

Q− Q+ N

e−

e−

3C

3C

N

Q− Q+

Q− Q+ N

3CNQ0 Q0

Q+

CN

3CNQ0 Q0

3C

CN

24 3CNQ0 Q0

Q

X

NC

3CNQ0 Q0

N

X

CN

CN

23

3C

X

R N

CN

NC

X

22, P-3CNQ

CN

R N

CH3

S

S

S

Au

Au

27

28

S

Figure 7 R-Q-3CNQ structures 11–14, 25, and 26 (drawn as zwitterions), zwitterionic P-3CNQ structure 22. For Q-3CNQ, 23 shows the zwitterionic (“benzenoid”) state, and 24 shows the less polar quinonoid state. Structure 27 shows that, if the molecules are tilted ca. 45◦ from the normal to an Au substrate, as found for 11, then the zwitterions are stabilized by intermolecular interactions (the direction of forward electron flow is shown by the arrow). If, however, the tilt angle is close to 0◦ , then the intermolecular dipoles repel each other, and a SAM on Au, shown in 28, will have a quinonoid structure of 24 (the direction of forward electron flow is shown by the arrow).70

at λmax = 570 nm25 polarized in the plane of an LB multilayer,60 which was earlier mistaken25 to be the IVT peak: this peak is far from the λmax = 800 nm peak observed for P-3CNQ.71 Molecule 11 forms amphiphilic Pockels–Langmuir monolayers at the air–water interface, with a collapse pres˚ 2 at 20 ◦ C25 ; sure of 34 mN m−1 and collapse areas of 50 A these monolayers transfer on the upstroke only, with transfer ratios around 100% onto hydrophilic glass, quartz, or aluminum,25, 73 or onto fresh hydrophilic Au,26, 27 but transfer poorly on the downstroke onto graphite, with a transfer ratio of only about 50%.73 The LB monolayer thickness of ˚ by X-ray diffraction, spectroscopic ellip11 was 23–25 A sometry, surface plasmon resonance, and XPS.25, 27, 64 With

˚ and a calculated molecular a monolayer thickness of 23 A ˚ the molecule on Al or Au is tilted by an length of 33 A, estimated 46◦ from the surface normal.25 The XPS spectrum of one monolayer of 11 on Au shows two N(1s) peaks.63 An angle-resolved XPS spectrum shows that the N atoms of the CN group are closer to the Au substrate, than the quinolinium N.63 The ionicity of 11 in solution is confirmed by its dipole moment; its ionicity as an LB multilayer and monolayer was confirmed by XPS and optical spectroscopy.63–65 The valence-band portion of the XPS spectrum agrees roughly with the density of molecular energy states.63 The contact angle of a drop of water is 92◦ above a monolayer of 11 deposited on fresh hydrophilic Au (this exposes the nonpolar tail to water).64 The orientation

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc094

Single-molecule electronics

13

14

6)

7(1

11

1.9

C2

6)

12

13 9.9 (3)

1 .0

14

7)

7) 8( 1 1.2 12

(3)

7) 2(1 9 .8 11

7) 6(1 1.6

(3) 9.9 13

7) (1 75 1.

117.27(18)

)

12

C3

(3

14 4.4 (3)

C4

1.0

9(1

179

5.0

.

24 35(

11

11 3.7

12

) (1 7 32 4.

11

8. 0

2)

)

C11

)

2(2

14

3 .3

)

137.7(3)

24

4 .2

8(1 126 .82 (17 )

7( 0.5

12

13

1)

13

(1

7.6

C6

12

16 3(

2(

N14

C10

8.4

144.0

1.5

0.8

8.9

116.78(17) 12

14

15) .82( 115

17

0.8

117.00(17)

C9

14

)

146.5

6) 5(1 1.6 12

(16

C12

119.73(17)

C13

Figure 8

.23

C17

136.5

C8

C7

8) 6(1 0.7 12

.82

C15

145.9

121

18

17

C16

5.4

)

N5

13

7.9

137.2(3

18)

6) (1 38 4. 12

13

C18

117.87(17)

2) 7(1 1.5 12

9)

)

.75(

N21 7.3

19.63(19

C19

121

13

8) 4(1 1.5 12

5 .0

136.5

12

) (15 .66 117

C20

6.8

C22

N1

ORTEP plot of P-3CNQ, picolinium tricyanoquinodimethanide, 22.69 NC

CN

NC

CN

CH2

CH3 +

+ N − TsO R

H

Li

+

+ − N TsO R

+

NC − CN

Li

+

NC − CN NC − CN

NC

NC

CN

CH2 + − N TsO R

CN + Li

+ − N R

+

TsO

NC − CN

NC − CN –HCN

CN HCN

+ N − TsO R

Figure 9

11

Reaction mechanism for the synthesis of the zwitterion 11.25

of 11 is confirmed by a grazing-angle Fourier transform infrared (FTIR) study of 11 on Al25 and on Au.64 LB monolayers and multilayers of 11 were sandwiched between macroscopic Al electrodes,25 and later, using the

“cold-gold” technique,26, 27, 61 between Au electrodes.26, 27 Between Al electrodes (with their inevitable patchy and defect-ridden covering of oxide), the monolayer has a dramatically asymmetric current. For 11, RR = 264 at

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc094

14

Supramolecular devices 0

1

−0.5

0.8

Current / mA

−1.5 −2

0.4

−2.5

0.2

−3 0 −0.2 −3

log10(I/mA) versus V

−1 0.6

−3.5 −2

−1

0

1

2

3

−4

Bias measured / volts

Figure 10

I –V curve and log10 I versus V (first cycle) for 11.27

˚ 2 , the current at 1.5 V.25 Assuming a molecular area of 50 A 1.5 V corresponds to 0.33 electrons molecule−1 s−1 .25 The RRs vary from pad to pad, as does the current, because these are all two-probe measurements; all electrical resistances (Al or Au, Ga/In or Ag paste, wires, etc.) are in series. As high potentials were scanned repeatedly, the I –V curves become less asymmetric; the RRs decrease gradually with repeated cycling of the bias across the monolayer. In the range 105 K < T < 390 K, the rectification onset for 11 between Al electrodes is independent of temperature.51 With oxide-free Au electrodes, the current through the “Au | LB monolayer of 11 | Au” pads increases dramatically, as expected, but the asymmetry persists: the highest current across a pad of area 0.283 mm2 is 8.20 mA, or 2.9 A cm−2 or 90 400 electrons molecule−1 s−1 ; the highest RR is 27 (Figure 10).26, 27 The pad resistance is 268 ; ˚ 2 , the pad should congiven the area per molecule of 50 A tain 5.66 × 1011 molecules. The measured resistance is still rather high; the Landauer quantum of resistance of 12.9 k molecule−1 , divided by the number of molecules, yields an estimated minimum resistance of 2.28 × 10−8  per pad, 10 orders of magnitude smaller. The best RR is 27.53 at 2.2 V.27 Figure 6(a) shows how the RR decreases from cycles 1 to 6. For some cells, the current increases until breakdown occurs; in some cells this happens at 5.0 V, that is, the cells suffer dielectric breakdown only at a field close to 2 GV m−1 .27 A small minority of monolayers of 11 sandwiched between Al electrodes rectifies in the reverse direction.52 Ashwell and coworkers confirmed that Z-type 30-layer films of 11 rectify between Au electrodes74 ; as expected, these currents are 3 orders of magnitude smaller74 than those reported for the monolayer.27 Ashwell and coworkers

studied several zwitterionic systems by STS, starting with 1175 ; for 11, and for several other zwitterionic systems,43, 76, 77 the addition of acid protonates the 3CNQ ring makes the solution colorless, and stops the rectification, while addition of base restored the blue color and the rectification.75 At 8 K, a monolayer film of 11 reverses its rectification direction.62 A tetrafluoro analog of 11, that is, molecule 14, also rectifies, Figure 6(b).60 The unwelcome gradual decreases in the electrical conductivity and in the RR of an LB monolayer of 11 (from an initial value of 2725, 27 to close to 1 upon repeated cycling) led to combining the LB and SAM techniques, by measuring thioacetyl variants of 11, which could bind strongly to Au electrodes.53, 58, 76 These variants were synthesized53, 58 with the aim of preparing molecules that can (i) form good Langmuir (or Pockels–Langmuir) monolayers at the air–water interface, then (ii) bind covalently to an Au substrate after either LB or LS transfer: the good ordering, afforded by the LB technique, should combine with a sturdy chemical bond to the Au substrate (SAM formation) after LB transfer. The variant of 11 with an undecyl “tail” followed by a thioacetyl termination (“C11 thioacetyl”) gave disappointing results53 : the Pockels–Langmuir film collapsed at relatively low surface pressures, compared to 11, and yielded disordered LB monolayers, with competition between strong physisorption by the dicyanomethanide end of the molecule and Au-to-thiolate chemisorption. The monolayer rectified in either direction, depending on where in the LB monolayer, that is, on which molecule (“right side up” or “upside down”) the STM tip was probing.53 Longer variants (“C14 thioacetyl” 12 and “C16 thioacetyl” 13 derivatives) did much better58 : rectification was observed in standard I –V measurements (not by STM).58 Theoretical calculations found that in the gas phase, the ground state of Q-3CNQ is not zwitterionic (i.e., the 3CNQ ring is quinonoid, as in 24, not benzenoid, as in 23), and has a small twist angle of 9–11◦ ;78 it was supposed that the zwitterionic state would be stabilized in solutions, in films, or in the solid state.78, 79 SAMs of 25 and 26 on Au do rectify, but have quinonoid, rather than benzenoid bond lengths in the 3CNQ ring70 : it seems that the benzenoid character demonstrated by us for the 3CNQ ring in 11 depends crucially on intermolecular charge–charge interactions that stabilize the zwitterions (molecules tilted on the substrate, so that the benzene ring overlap is staggered); in contrast, under certain conditions (e.g., when molecules are not staggered) the ring may become quinonoid (Figure 7).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc094

Single-molecule electronics Mattern has noticed that for all systems shown in Table 1, the “forward” direction of enhanced current is reversed, compared to the AR mechanism80 ! More puzzles! Other researchers have seen rectification in monolayers or in single molecules physisorbed onto metal substyrates or studied by STM; in particular, the groups of Yu and Ashwell have made important contributions, already reviewed previously elsewhere.6 Recent work not reviewed previously was a C60 dyad rectifier reported by the group of Patnaik,81 and fruitful collaboration by the groups of Yu and Tao, whereby rectification was measured for a single molecule covalently bound to two Au electrodes.82 In addition, a recent report of FET behavior35 by a single 1,4,-benzenedithiol molecule chemisorbed to two Au nanoelectrodes (source and drain electrodes) and affected by the electric field on an oxide-covered Si semiconducting gate electrode.

15

4. D. J´erˆome, A. Mazaud, M. Ribault, and K. Bechgaard, J. Phys. Lett., 1980, 41, L95–L97. 5. C. K. Chiang, C. R. Fincher, Jr., Y. W. Park, A. J. et al., Phys. Rev. Lett., 1977, 39, 1098–1101; erratum: Phys. Rev. Lett., 1978, 40, 1472. 6. R. M. Metzger, J. Mater. Chem., 2008, 18, 4364–4396. 7. G. E. Moore, Electronics, 1965, 38(8), 114. 8. ITRS-2007, International technology roadmap for semiconductors, 2007 version (http://www.itrs.net/reports) Section on Emerging Research Materials (sub-22 nm). 9. G. Hoffmann, L. Libioulle and R. Berndt, Phys. Rev., 2002, B65, 212107. 10. R. M. Metzger and C. A. Panetta, New J. Chem., 1991, 15, 209–221. 11. R. M. Metzger, The quest for D-σ -A unimolecular rectifiers and related topics in molecular electronics, in Molecular and Biomolecular Electronics, ed. Robert R. Birge, Am. Chem. Soc. Adv. in Chem. Ser. 240, American Chemical Society, Washington, DC, 1994, pp. 81–129. 12. R. M. Metzger, Mater. Sci. Eng., 1995, C3, 277–285.

6

CONCLUSION

Unimolecular rectification is now an accomplished experimental fact. Its utilization in practical devices may not be too far in the future. However, it is important to fully understand and control the ground and first excited states of molecules that exhibit rectification.

ACKNOWLEDGMENTS This work was achieved by the diligence and insight of so many colleagues, students, and postdoctoral fellows, to whom I owe an immense debt of gratitude, and facilitated by grants from the United States National Science Foundation. I mourn the death of three good friends and wonderful scientists: Ian Robert Peterson (1945–2005), M. V. Lakshmikantham (1931–2006), and Michael P. Cava (1926–2010).

13. R. M. Metzger, J. Mater. Chem., 1999, 9, 2027–2036. 14. R. M. Metzger, J. Mater. Chem., 2000, 10, 55–62. 15. R. M. Metzger, J. Solid Struct. Chem., 2002, 168, 696–711. 16. R. M. Metzger, Unimolecular rectifiers, in Molecular Nanoelectronics, eds. M. A. Reed and T. Lee, American Scientific Publishers, Stevenson Ranch, CA, 2003, pp. 19–38. 17. R. M. Metzger, Chem. Rev., 2003, 103, 3803–3834. 18. R. M. Metzger, Molecular rectifiers, in Encyclopedia of Supramolecular Chemistry, eds. J. L. Atwood, and J. W. Steele, Marcel Dekker, New York, NY, 2004, vol. 2, pp. 1525–1537. 19. R. M. Metzger, Chem. Rec., 2004, 4, 291–304. 20. R. M. Metzger, Unimolecular electronic devices, in Intelligent Materials, eds. M. Shahinpoor and H. J. Schneider, Royal Society of Chemistry Publishing, Cambridge, UK, 2008, pp. 205–230. 21. R. M. Metzger, Unimolecular electronics: results and prospects, in Nano and Molecular Electronics Handbook , ed. S. E. Lyshevski, CRC Press, Taylor and Francis Group, Boca Raton, FL, 2007, pp. 3–1 to 3–25. 22. A. Aviram and M. A. Ratner, Chem. Phys. Lett., 1974, 29, 277–283.

REFERENCES

23. G. J. Ashwell, J. R. Sambles, A. S. Martin, et al., J. Chem. Soc. Chem. Commun., 1990, 1374–1376.

1. R. M. Metzger, Prospects for truly unimolecular devices, in Lower-Dimensional Systems and Molecular Electronics, eds. R. M. Metzger, P. Day, and G. Papavassiliou, NATO ASI Series B248, Kluwer, Dordrecht, The Netherlands, 1991, pp. 659–666.

24. A. S. Martin, J. R. Sambles, and G. J. Ashwell, Phys. Rev. Lett., 1993, 70, 218–221.

2. J. Ferraris, D. O. Cowan, V. Walatka, Jr, and J. H. Perlstein, J. Am. Chem. Soc., 1973, 95, 948–949.

26. T. Xu, I. R. Peterson, M. V. Lakshmikantham, and R. M. Metzger, Angew. Chem. Int. Ed., 2001, 40, 1749–1752.

3. D. O. Cowan, J. A. Fortkort, and R. M. Metzger, Design constraints for organic metals and superconductors, in Lower-Dimensional Systems and Molecular Electronics, eds. R. M. Metzger, P. Day, and G. C. Papavassiliou, NATO ASI Ser. B248, Plenum Press, New York, 1991, pp. 1–22.

25. R. M. Metzger, B. Chen, U. H¨opfner, et al., J. Am. Chem. Soc., 1997, 119, 10455–10466.

27. R. M. Metzger, T. Xu, and I. R. Peterson, J. Phys. Chem., 2001, B105, 7280–7290. 28. M. A. Reed, C. Zhou, C. J. Muller, et al., Science, 1997, 278, 252–253.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc094

16

Supramolecular devices

29. L. A. Bumm, J. J. Arnold, M. T. Cygan, et al., Science, 1996, 271, 1705–1707.

58. A. Jaiswal, D. Rajagopal, M. V. Lakshmikantham, et al., Phys. Chem. Chem. Phys., 2007, 9, 4007–4017.

30. J. Chen, M. A. Reed, A. M. Rawlett, and J. M. Tour, Science, 1999, 286, 1550–1552.

59. W. J. Shumate, Ph. D. dissertation, University of Alabama, 2005.

31. S. J. Tans, M. H. Devoret, H. Dai, et al., Nature, 1997, 386, 474–477. 32. C. P. Collier, G. Mattersteig, E. W. Wong, et al., Science, 2000, 289, 1172–1175. 33. J. Park, A. N. Pasupathy, J. I. Goldsmith, et al., Nature, 2002, 417, 722–725.

60. A. Honciuc, A. Otsuka, Y.-H. Wang, et al., J. Phys. Chem., 2006, B110, 15085–15093.

34. B. Xu and N. J. Tao, Science, 2003, 301, 1221–1223. 35. H. Song, Y. Kim, Y. H. Jang, et al., “Observation of Molecular Orbital Gating”, Nature, 2009, 462, 1039. 36. W. Schottky, Naturwissenschaften, 1938, 26, 843. 37. R. Landauer, IBM J. Res. Dev., 1957, 1, 223–231. 38. D. V. Averin and K. K. Likharev, J. Low Temp. Phys., 1986, 62, 345–372. 39. W. Wang, T. Lee, and M. A. Reed, J. Phys. Chem., 2004, B108, 18398–18407. 40. A. Honciuc, R. M. Metzger, A. Gong, and C. W. Spangler, J. Am. Chem. Soc., 2007, 129, 8310–8319. 41. U. Mazur and K. W. Hipps, J. Phys. Chem., 1995, 99, 6684–6688. 42. U. Mazur and K. W. Hipps, J. Phys. Chem., 1999, B103, 9721–9727. 43. G. J. Ashwell, W. D. Tyrrell, and A. J. Whittam, J. Am. Chem. Soc., 2005, 126, 7102–7110. 44. J. Sagiv, J. Am. Chem. Soc., 1980, 102, 92–98. 45. C. D. Bain, E. B. Troughton, Y. T. Tao, et al., J. Am. Chem. Soc., 1989, 111, 321–335. 46. N. J. Geddes, J. R. Sambles, D. J. Jarvis, et al., Appl. Phys. Lett., 1990, 56, 1916–1918. 47. N. J. Geddes, J. R. Sambles, D. J. Jarvis, et al., J. Appl. Phys., 1992, 71, 756–768. 48. C. Krzeminski, C. Delerue, G. Allan, et al., Phys. Rev., 2001, B64, 085405. 49. M. L. Chabinyc, X. Chen, R. E. Holmlin, et al., J. Am. Chem. Soc., 2002, 124, 11731–11736. 50. V. Mujica, M. A. Ratner, and A. Nitzan, Chem. Phys., 2002, 281, 147–150. 51. B. Chen and R. M. Metzger, J. Phys. Chem., 1999, B103, 4447–4451. 52. D. Vuillaume, B. Chen, and R. M. Metzger, Langmuir, 1999, 15, 4011–4017. 53. A. Jaiswal, R. R. Amaresh, M. V. Lakshmikantham, et al., Langmuir, 2003, 19, 9043–9050. 54. J. W. Baldwin, R. R. Amaresh, I. R. Peterson, et al., J. Phys. Chem., 2002, B106, 12158–12164. 55. R. M. Metzger, J. W. Baldwin, W. J. Shumate, et al., J. Phys. Chem., 2003, B107, 1021–1027. 56. A. Honciuc, A. Jaiswal, A. Gong, et al., J. Phys. Chem., 2005, B109, 857–871. 57. W. J. Shumate, D. L. Mattern, A. Jaiswal, et al., J. Phys. Chem., 2006, B110, 11146–11159.

61. N. Okazaki and J. R. Sambles, in Extended Abstracts of the International Symposium on Organic Molecular Electronics, Nagoya, Japan, 2000, p. 66. 62. N. Okazaki, J. R. Sambles, M. J. Jory, and G. J. Ashwell, Appl. Phys. Lett., 2002, 81, 2300–2302. 63. J. W. Baldwin, B. Chen, S. C. Street, et al., J. Phys. Chem., 1999, B103, 4269–4277. 64. T. Xu, T. A. Morris, G. J. Szulczewski, et al., J. Phys. Chem., 2002, B106, 10374–10381. 65. F. Terenziani, A. Painelli, A. Girlando, and R. M. Metzger, J. Phys. Chem., 2004, B108, 10743–10750. 66. O. Kwon, M. L. McKee, and R. M. Metzger, Chem. Phys. Lett., 1999, 313, 321–331. 67. W. A. Little, Phys. Rev., 1964, 134, A1416–A1424. 68. L. B. Coleman, M. J. Cohen, D. J. Sandman, et al., Solid State Commun., 1973, 12, 1125–1132. 69. R. M. Metzger, N. E. Heimer, and G. J. Ashwell, Mol. Cryst. Liq. Cryst., 1984, 107, 133–149. 70. A. Girlando, C. Sissa, F. Terenziani, et al., Chem. Phys. Chem., 2007, 8, 2195–2201. 71. S. Akhtar, J. Tanaka, R. M. Metzger, and G. J. Ashwell, Mol. Cryst. Liq. Cryst., 1986, 139, 353–364. 72. G. J. Ashwell, in Organic Materials for Nonlinear Optics, eds. G. J. Ashwell and D. Bloor, Royal Society of Chemistry, Cambridge, 1993, pp. 31–39. 73. R. M. Metzger, H. Tachibana, X. Wu, et al., Synth. Metals, 1997, 85, 1359–1360. 74. G. J. Ashwell and G. A. N. Paxton, Aust. J. Chem., 2002, 55, 199–204. 75. G. J. Ashwell, W. D. Tyrrell, and A. J. Whittam, J. Mater. Chem., 2003, 13, 2855–2857. 76. G. J. Ashwell, A. Chwialkowska, and L. R. Herrmann High, J. Mater. Chem., 2004, 14, 2848–2851. 77. G. J. Ashwell and M. Berry, J. Mater. Chem., 2005, 15, 108–110. 78. A. Broo and M. C. Zerner, Chem. Phys., 1995, 196, 423–436. 79. M. Pickholz and M. C. dos Santos, (Theochem), 1998, 432, 89–96.

J. Mol. Struct.

80. R. M. Metzger and D. L. Mattern, Unimolecular electronic devices, in Unimolecular and Supramolecular Electronics, ed. R. M. Metzger, Springer Topics in Current Chemistry, Springer, Berlin Heidelberg, New York, in press. 81. S. S. Gayathri and A. Patnaik, Chem. Commun., 2006 1977–1979. 82. I. D´ıez-P´erez, J. Hihath, Y. Lee, et al., Nature Chem., 2009, 1, 635–641.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc094

Molecular Redox Sensors James H. R. Tucker University of Birmingham, Birmingham, UK

1 Introduction 2 Receptors in Solution 3 Receptors Immobilized at a Surface 4 Conclusion References

1

1 2 9 13 13

INTRODUCTION

The field of electrochemical sensing is a vast area1 and although many aspects of it do not conventionally fall under the realm of supramolecular chemistry, the connection of electrochemistry to supramolecular chemistry is well established as a step to many different applications, including sensing.2, 3 This review covers the developments and the progress in the field of supramolecular voltammetric/amperometric sensing using redox-active sensors, where current is measured as a function of applied potential using various techniques. Other types of electrochemical sensing, also having connections to supramolecular chemistry (e.g., potentiometric sensing with ion selective electrodes), are not covered here and the reader is referred elsewhere to a recent general review.1 A number of methods for sensing or analyzing charged or neutral species in solution using voltammetric techniques have been identified.1, 4, 5 However, in the electrochemical sensing of nonredox-active species, supramolecular chemistry can play a particularly important role in this area with

the use of redox-active receptors that can respond electrochemically to the complexation of a nonredox-active guest. This process is shown schematically in Scheme 1. The complexation by a redox-active host, H, of a charged or neutral guest, G, imparts a change in its electrode potential, which enables the binding process to be read-out by an electrochemical technique, usually cyclic voltammetry. In the design of such a molecular redox sensor, the objectives are twofold; namely, guest selectivity and a significantly large redox response to complexation. The second aspect can be conveniently illustrated by the following square scheme (Scheme 2) and (1). Essentially, the change in the electrode potential of H on complexation with G is related to the change in the host–guest binding constant upon its oxidation or reduction. In (1), E[H] and E[H – G] are the formal electrode potentials of H and the complex, [H–G], respectively (each is usually approximated from the average of the anodic and cathodic peak potentials from cyclic voltammetry), K + and K are the binding constants in the oxidized and reduced form of the receptor, respectively, F is the Faraday constant, R is the universal gas constant, T is the temperature, and n is the number of electrons transferred (in Scheme 2, n = 1). The change in electrode potential upon complexation (i.e., the redox response),
E, is often quoted in the literature, where
E = E[H – G] − E[H] . K+ = exp[−nF (E[H – G] − E[H] )/RT ] K

(1)

Therefore, a successful sensor will be the one that is selective for a particular target and binds it considerably more strongly either in its oxidized or reduced form. The suitability of (1) for estimating the magnitude of the redox-switched binding enhancement from potential shifts has been examined in detail for crown-ether systems that bind cations.6 One finding is that the strength

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

2

Supramolecular devices is generally restricted to sensor systems where the redoxactive unit is an integral part of the receptor. In other words, examples of sensing where the redox-active group is a separate species, such as in impedance measurements as discussed in a recent review,17 or in a competitive binding assay,18 or as a target analyte,19 will not be covered here.

Binding site

Redox-active center

H

G

[H –G]

Scheme 1 The complexation of a guest, G, by a redox-active receptor, H.

+

H

e−

K

G

e−

EH

H+

[H – G]

+

G

K+

2

RECEPTORS IN SOLUTION

In this section, freely diffusing receptors in solution are discussed, in which the redox-active center or reporter group is, or is an integral part of, an inorganic (i.e., metalcontaining) or an organic (i.e., metal-free) group.

E [H –G]

2.1

Metal-containing reporter groups

[H – G]+

2.1.1 Metallocenes Scheme 2

Square scheme for a redox-switchable system.

of the host–guest binding interaction appears to dictate whether cyclic voltammograms exhibit one-wave or twowave behavior when adding aliquots of guest to the redoxactive host. An equation related to (1) is sometimes used7 when monitoring the shift in the observed potential as a function of substrate concentration in weakly binding systems. The area of supramolecular electrochemistry and its relevance to sensing and redox-switched binding has already been the subject of a book3 and a number of reviews.8–12 In a review by Gale, Beer, and Chen,9 the different mechanisms by which complexation can impart a redox response, for example, via through-bond or through-space interactions, were discussed. In general, the proximity of the receptor site to the redox center and the charge density of the guest are important factors in determining the magnitude of the redox response and the effectiveness of the electrochemical “read-out.” However, other factors related to the conditions used for the experiment, for example, the supporting electrolyte13 and the solvent,14 can also influence the magnitude of the redox response to complexation. The concepts behind molecular redox sensors have been previously summarized in a recent article in which an overview of the range of receptors designed for the binding and sensing of charged and neutral guest species was presented.15 For consistency, this review follows a similar format but gives an update of recent developments in the field, for example, recent activity on the binding of neutral species, chiral sensing, and surface sensing. The review does not include recent advances made in electrochemical approaches to biomolecule (e.g., DNA16 ) sensing and

Ferrocene has been by far the most used redox-active group in a molecular redox sensor due to its stability, its ease of functionalization, and its well-understood and reversible redox chemistry. A whole host of ferrocene-containing compounds are now known that bind cations, anions, and neutral molecules. Indeed some of the first examples of redox-active receptors, for which the complexation of other species was demonstrated, were ferrocenyl crown-ether compounds, as reported in the late 1970s and in 1980. Initial progress in this area has been reviewed extensively.10, 20 Representative examples of such compounds are 1,21 2,22 3,23 4,24 and 525 (Figure 1), which all bind and electrochemically respond to various s-block cations in organic solvents. Related receptors also bind charged ammonium26 and amino acid27 salts through H-bonding interactions with the crown-ether rings. Although studies with crown-etherbased systems have continued to attract recent interest,28 the complexation and sensing of hard cations using other binding motifs is becoming more common. For example, Bryce et al. found that compound 6 (Figure 1) responded electrochemically to Mg2+ solely over other mono- and divalent cations in acetonitrile.29 Likewise, compound 7, with two nonequivalent ferrocene centers, was also found to bind and sense Mg2+ in dichloromethane where only one ferrocene unit underwent a redox response to complexation.30 The histidine conjugate 8 was found to recognize sodium ions electrochemically, with the ferrocene unit undergoing an unusual negative shift in its redox couple, indicating that the ferrocene unit in the bound complex was easier to oxidize, probably due to structural changes upon complexation at the syn carbonyl groups.31 Apart from other similar unusual examples,28, 32 the complexation of s-block metal cations normally imparts positive (anodic) shifts in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

Molecular redox sensors

3

O N O Fe

N

O Fe

O

Fe

O O

O O

N

N

O

O

O

O

O

O

O

1

2

3 O N

O O O

Fe

N

O

O

N Fe

Fe

Fe

HO

O

O

N

O 4

O

O

6

5 O

N N

N

Fe Fe

Fe

O

N

N H

O

8

7

Figure 1

O

Ferrocene receptors for s-block metal cations.

ferrocene-centered redox couples, as a result of the close proximity of the charged cation making ferrocene oxidation more difficult. As expected, similar effects are also found with ferrocenyl receptors that bind and sense lanthanide cations.33, 34 To allow the sensing of d-block and main group metal cations, softer donor atoms have been incorporated into binding moieties in ferrocene-containing receptors (e.g., compounds 9–13; Figure 2). This area has attracted

considerable interest over the past few years, most notably from the group of T´arraga and Molina, who have recently reviewed much of their work on ferrocene derivatives that sense cations and anions,35 with examples of receptors that are photo-active as well as redox-active, allowing a read-out of binding through multisignaling pathways. As found with the crown-ether derivatives, the complexation of these metal cations generally imparts positive shifts in the ferrocene-based redox couples of each receptor. S N

N Fe

HN

X S

HN

Fe

N 10a X = O 10b X = S

9

N

N

N

N H Fe

N N

O

N

N

Fe

Fe

11

Figure 2

N

O

H N

12

N

O O Fe

N 13

Ferrocene receptors for d-block metal cations.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

4

Supramolecular devices

It is important that the factors responsible for generating a significant electrochemical response to complexation are fully understood if effective sensors are to be designed. In this respect, a detailed study by Plenio and Mart´ınez-M´anez14 of transition metal complexation by a ferrocenyl molecule containing a cyclam ligand, 9, has provided further information on the importance of the Femetal distance and the solvent in governing the redox response to complexation. In particular, a linear dependence of the shift in the Fc+ /Fc couple upon complexation,
E, versus the intermetallic distance, 1/r, was found for the complexes [M(FcCyclam)]2+ (M = Co, Ni, Cu, and Zn), which was clear evidence for a primarily coulombic through-space Fe–M2+ interaction. Other studies on the binding of hard metal cations by crown-ether derivatives 1 and 2 have found linear relationships between the shift in potential and either21 the charge/(radius)2 or22 the charge/radius of the bound cation, respectively. It was also shown by Plenio and Mart´ınez-M´anez how the solvent can have a dramatic effect on the shift in potential, in that the complexation of Zn2+ by 9 imparted shifts ranging from +280 mV in THF/CH3 CN (THF, tetrahydrofuran, 30 : 2 v/v) to +409 mV in CH3 CN alone.14 The expected trend toward larger shifts for solvents with lower dielectric constants was only observed for solvent mixtures of relatively high polarity. For solvent mixtures of relatively low polarity, the shifts were lower than expected due to ion pairing effects. Lippolis et al. found that the magnitude of the electrochemical response to the binding of various d-block and main group cations in acetonitrile changed upon varying the donor atoms in related mixed-donor macrocycles 10a and 10b.36 The donor trend preference of N > O > S for Zn2+ was adjudged the reason for the largest redox response with this cation, while Cu2+ gave the largest redox response with ligand 10b. Interestingly, it was mentioned that a selective response to Hg2+ and Pb2+ had been observed previously for an analogous receptor where the three sulfur atoms in 10b were replaced with oxygen atoms. There is considerable recent interest in the sensing of potentially toxic cations such as mercury, cadmium, and lead, and sometimes recognition in an aqueous environment has been achieved. For example, bis-ferrocene receptor 11 was found to form a 1 : 1 complex with Hg2+ in acetonitrile/water (ratio 3 : 7), which resulted in positive shifts in the closely spaced redox couples of each ferrocene unit.37 No response was found with various other s-block and d-block cations. Similarly, Delgado et al. found that receptor 12 could respond electrochemically to range of cations in water including Cd2+ and Pb2+ .38 However, it gave a selective response to Cu2+ when exposed to an equimolar mix of five cations, in agreement with the highest stability constant found with this cation as predicted by the

Irving–Williams stability order. Supporting data indicated a 2 : 2 (host/guest) complex stoichiometry in solution with a 1 : 2 stoichiometry in the presence of excess Cu2+ . Recently, rhodamine-based ferrocene derivatives have been used for cation sensing, for example, system 13 that responds to Cr3+ , which is an ion involved in various biochemical processes at the cellular level.39 An unusual anodic shift of −140 mV was observed in the Fc/Fc+ redox potential upon complexation of this cation in ethanol. There are now numerous reports of metallocene derivatives that bind and sense anions. Early work in this area was carried out by Beer, who first reported a redox-active anion receptor in 1989, and much of the work in this area has now been reviewed.35, 40 In contrast to what is found with most cations, the Fc/Fc+ redox couples undergo appreciable negative (cathodic) shifts upon complexation with inorganic anions in organic solvents. These negative shifts are consistent with a stabilization of the oxidized forms of each receptor by the negative charge of each guest. Representative examples of metallocene-containing anion receptors are depicted in Figure 3. The complexes are commonly stabilized by cation–anion electrostatic interactions as well as H-bonding and Lewis acid–base interactions. Oxoanions have long been popular targets for anion sensing, not least because of their biological relevance. Beer and Mart´ınez-M´anez reported a number of ferrocene-containing receptors that selectively bind and sense oxoanions at a particular pH value. For example, at pH 4 in THF/water (70 : 30 v/v), ligand 14 selectively forms a 1 : 1 complex with hydrogen sulfate (HSO4 − ) in the presence of phosphate, with the Fc+ /Fc redox couple undergoing a shift of −54 mV upon complexation.41 Under these conditions, all four secondary amines of the receptor are protonated. The biological anion ATP could also shift the redox couple of 14 cathodically by up to −100 mV, depending on the pH of the medium. More recently, pyrophosphate, HP2 O7 3− , has been targeted. For example, compound 15 formed a 2 : 1 (host/guest) complex with this anion in organic solvents and underwent a cathodic shift of −100 mV upon complex formation, while other anions, including H2 PO4 − , had no effect.42 Studies indicated the presence of hydrogen-bonding interactions involving the proton within the triazole ring of the receptor. Other receptors have utilized (C–H)+ · · ·X− H-bonds to bind guest species, for example, the imidazolium macrocycle 16 which was found to bind and electrochemically respond to a range of anions in acetonitrile.43 In addition to inorganic anions, organic anions, including chiral species,44 can be targeted. An early example of a carboxylate anion receptor was the cobaltocenium-containing calixarene derivative 17,45 which bound these guests selectivity over halide ions, due to two rigidly held amide units on the upper rim of the calixarene that create an ideal site

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

Molecular redox sensors N Fe

N H

N H

H N

H N

Fe

15

14

N

N

N

OTs OTs

H N

19

Figure 3

CF3 CF3

18

16 OH

+

BF4−

O O

O

17

Fe

N H

Fe

OH

O

N

O

OH

N H H N

Co N

+

O

Fe

N

N Fe

5

B

B OH Fe

20

Fe

21

Metallocene receptors for inorganic and organic anions.

for bidentate coordination. Recently, the bis-amide derivative 18 was found to interact with dicarboxylates such as the di-anion of 1,3-phenylenediacetic acid by reaction with the two trifluoroacetyl carbonyl carbons of the receptor.46 The introduction of two negative charges proximate to the ferrocene center induced a large negative shift in the ferrocene-centered oxidation potential upon the formation of the 1 : 1 covalent adduct. Very simple receptors have also been developed for anion sensing. One such example is commercially available 19, which can sense dihydrogen phosphate and ATP anions in organic media.47 The large negative shifts of the Fc/Fc+ redox couple upon complexation (e.g.,
E = −470 mV in CH2 Cl2 upon addition of H2 PO4 − ) were a result of strong ion pairing interactions between the anion and the double positive charge of the quaternary ammonium group and the oxidized ferrocenium ion. Another commercially available compound is ferrocene boronic acid, 20,48 one of a number of known redox-active boron-based receptors,48–54 which was found by Shinkai to recognize fluoride selectively over a range of other halide ions and oxoanions in water. Negative shifts in the Fc/Fc+ redox couple of about −200 mV were observed upon the addition of excess amounts of fluoride, which formed a tetrahedral adduct with the Lewis acidic boron center. More recently, a boron receptor containing two mesityl groups, 21, was found to bind and electrochemically recognize either F− or CN− in acetonitrile through the formation of adducts of tetrahedral geometry at the boron center. In contrast, a related boronate

ester derivative with weaker Lewis acidity was found to bind and respond to fluoride only.49 In contrast to the recognition of cations or anions, the electrochemical sensing of neutral molecules by redoxactive receptors is relatively rare, with few systems undergoing a significant electrochemical response to complexation. Some ferrocene compounds that electrochemically respond to neutral molecules and charge-neutral species are shown in Figure 4. Metallocene 22 was found to bind neutral carboxylic acids in organic solvents through complementary hydrogen bonds with their amide (N–H· · ·O) and pyridine (O–H· · ·N) groups.55 Cyclic voltammetry studies in chlorinated solvents revealed negative shifts in its Fc/Fc+ redox couple, with the redox response to complexation of the diacid glutaric acid being relatively large (−90 mV), which was consistent with the guest being bridged between the Cp rings of 22 and its cobaltocenium analog in solution. In subsequent studies, 22 and the 1,3 regioisomer 23 were found to bind cyclic organic molecules such as barbital in organic solvents, with 22 generally giving the larger redox response in interactions with guests that could bind in a wedge-like manner across the ferrocene unit.56 As a part of the recent interest in the electrochemical sensing by redox-active receptors of biological molecules such as amino acids27, 57, 58 and DNA bases,59–61 receptor 24 was found to bind the zwitterionic amino acids, glycine and glutamine, in acetonitrile–water (55 : 45), giving cathodic shifts in the Fc/Fc+ redox couple upon complexation.58

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

6

Supramolecular devices O

Fe

O

N H

N

HO

O

O NH

O

O HN

HN

Fe

O

N

N

NH O

O

N H

N

HO Glutaric acid

23

22

O

N(CH3)2 OH B OH Fe

OH

O Fe O

B

25

24

OH

OH

OH

O

Barbital

OH

HO

N

B

N

Fe N

B HO

Fe

26 OH

Bn 27

Figure 4

Ferrocene receptors for neutral molecules and charge-neutral species.

The data supported a 1 : 1 complex stoichiometry in which the guests were encapsulated through H-bonds to both the ammonium and carboxylate groups. Charge-neutral bile salts can be electrochemically recognized by ferrocene-containing receptors62 and heteroditopic crown-ether derivatives containing both cation- and anion-binding sites allow anion recognition to be tuned by the presence of cations in the formation of overall charge–neutral complexes.63, 64 Using a similar concept, receptor 25 developed by Aldridge and Fallis was found to give a unique electrochemical response to HF (hydrogen fluoride), a gas associated with the hydrolysis of certain chemical warfare agents (CWAs).51 The combination of protonation of the amine group and fluoride coordination at the boronic acid center resulted in a net negative shift of −80 mV. Exposure to other acids such as HCl only resulted in protonation of the amine, resulting in a positive shift in the redox couple. An enantiopure ferrocene-containing boronic acid related to 25 was originally shown by Shinkai52 to bind and electrochemically respond to sugars in water at neutral pH. In a more recent study,53 diboronic acid receptor 26 was found to bind and sense D-glucose. A positive shift in the Fc/Fc+ redox couple observed due to diol complexation (through boronate ester formation) at the boron center strengthening the intramolecular amine–boronic acid interaction, making the ferrocene center harder to oxidize. In a rare example

of chiral sensing by a supramolecular redox-active receptor, the enantiopure imine-based receptor 27 was found by Tucker to bind either (S)- or (R)-Binol, with each enantiomer giving a significantly different redox response and binding affinity in dichloromethane (Scheme 3), allowing the enantiomeric excess of mixtures containing different ratios of (S)/(R)-Binol to be read-out electrochemically.54

2.1.2 Other metal-containing reporter groups Apart from metallocenes, a number of other redox-active groups consisting of (or incorporating) metal centers have been incorporated into supramolecular receptors. Many of these systems electrochemically respond to cations but an increasing number of species that respond to anions and neutral molecules are now known, and representative examples are displayed in Figure 5 (compounds 28–33). A number of the cation binders are organometallic crown ether and metallo-crown or metallo-thiacrown derivatives, with these systems representing some of the earliest examples of redox-active receptors,20 for example, compound 28 which was found to respond electrochemically to Na+ and K+ cations in acetonitrile solution.65 A series of selfassembled [12]metallocrown-3 complexes, two of which are compounds 29 and 30, were found by Severin to bind halide salts of small group 1 metals strongly in organic solvents, with affinities similar to those of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

Molecular redox sensors

N

B O

Fe

N

(R )-Binol

O

Bn

Bn

Fe

B HO

N

(S )-Binol

OH

Bn

Fe

7

B O

O

27 Smaller redox response ∆E = +44 mV

Larger redox response ∆E = +95 mV

Scheme 3

Chiral sensing of (R)- and (S)-Binol by ferrocene receptor 27.

O S

O

S

O

O O

O

O

O

3

29 M = Rh 30 M = Ir

O

M

O

N

O

N

Mo

M

N

M

28

2+

O O

O N H

N N

N H

N

N N N

N Ru

Ru N

N

H N

N

N

H N

N

N Ru

N

N

N O

N

N

O

O 31

32 O

2+

N H

Cu

H N

N H

O

O O

N N

H N

NH N

33

Figure 5

O O

2

O

O

O

4+

NH2

Redox-active receptors with metal-based (nonmetallocene) reporter groups.

cryptands.66 X-ray crystal structures revealed that the metal cation was bound in the center of the cavity, coordinated by three oxygen atoms of the metallo-crown and the halide counterion. No detectable complexation of KCl was observed due to the small size of each cavity. These receptors were found to respond electrochemically to complexation, with shifts in oxidation peak potential of more than +300 mV observed upon addition of LiCl and NaCl to 29. The LiBF4 complex of 30 was also found to sense fluoride anions in organic solvents; a cathodic shift of −203 mV was observed upon the addition of excess

NBu4 F, as a result of the fluoride anion forming an unusual molecular Li–F bond. Transition metal bipyridine or terpyridine complexes incorporating Fe(II),51 Ru(II),68, 69 and Co(II)70 centers have been used as redox reporter groups for a range of guest species. The bimetallic receptor 31 was found to respond to chloride anions in CH3 CN, as evidenced by a cathodic shift of −110 mV in the ligand-centered reduction couple of the two bis-substituted bipyridine groups.67 Schmittel recently reported the multisignaling azacrown system 32 which was able to respond to Pb2+ cations through a positive shift of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc096

8

Supramolecular devices

+180 mV in the Ru(II)/Ru(III) redox couple in acetonitrile. A range of other s- and d-block cations gave responses of b > c).

K+ complex will favor Donnan failure and will thereby lower (i.e., worsen) the upper detection limit of the ISE (Figure 22).

4.2

Ionophore-based ion-selective electrodes for pH

Because of the key role of water for all living organisms, the measurement of pH is arguably the most important application of potentiometry. This is underscored by the fact that the pH is not defined on the basis of the hydrogen ion concentration but by its activity, which is readily accessible only by potentiometry.85, 86 In the majority of cases, such measurements are performed with commercially available glass electrodes.23 The use of glass as the sensing membrane has a number of disadvantages,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

20

Supramolecular devices

though. Proteins and other biological materials readily adsorb to glass, which requires frequent electrode cleaning. Even the speciality glasses used for pH glass electrodes have a very high electrical resistance and are, therefore, difficult to be miniaturized. Moreover, the use of thin glass membranes as dictated by the high electrical resistance also makes glass electrodes prone to breakage, which is particularly undesirable in the food industry and in in vivo measurements. This has lead to a significant interest in the development of ionophore-based ISEs for pH. From a host–guest chemistry point of view, H+ ionophores are as simple as it gets. While a remarkable variety of H+ ionophores have been tested,4 simple trialkylamines and pyridine derivatives that bind the hydrogen ion in a monodentate fashion are suitable for most applications. Ionophore-based pH electrodes illustrate in exemplary fashion that finding a compound that forms a very stable complex with the target ion is not sufficient for the preparation of a good ISE. Tridodecylamine (H+ -I),87 for example, binds H+ strongly and permits pH measurements up to pH 11, where the hydrogen ion activity is so low that the alkali metal cations often used to prepare solutions of such high

pH are starting to interfere. However, the strong binding of the hydrogen ion to this ionophore also promotes Donnan failure (Section 4.1 and Figure 22) at low pH, that is, the coextraction of hydrogen ions and counteranions from the aqueous sample into the sensing membrane and the concomitant loss of the linear response to pH. This limits the use of ISEs based on H+ -I to samples of pH 4.5 and higher. For measurements at lower pH, ionophores that form less stable H+ complexes are needed, such as H+ -II, which is suitable for the pH range from 0 to 8.88 Evidently, the smaller binding constants limit the selectivity at higher pH. The membrane matrixes doped with these ionophores have a strong effect on selectivity. The useful pH range is reduced at higher pH by polymers or plasticizers with functional groups that interact with interfering cations, and polymers. Plasticizers and ionic sites that stabilize counteranions in the sensing membrane facilitate Donnan failure at the lower end of the pH range. Evidently, inert matrixes are preferred as they provide the widest measurement ranges.89 Since perfluorocarbons are the least polar and least polarizable condensed phases known, they are, expected to offer

O

N

O

C18H37

N H+-II

H+-I

F F F

F F

FF

FF

FF

FF

FF

F

FF

F F

F N

F F F F F F F F

F F F F F F F F F H+-III

FF

FF

F

FF

F F F

FF

F

F F F

F F

FF

FF

FF

FF

FF

F

FF

F

F N

F F F F F F F F

F

FF

FF

F

FF

F F F

FF

F

F F F F F F F F F

H+-IV

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

for K+ , Na+ , and Li+ will be discussed in this section to the extent that they represent different types of ionophore design. However, it must be emphasized that much more work has been performed in this field, and ISEs have been developed for over two dozen inorganic cations. In particular, decades of research have been made to improve the design of ISEs for the clinically relevant ions Ca2+ and Mg2+ ,90 and numerous ISE have been reported for many heavy metal ions.4, 91 The cyclic peptide valinomycin binds to K+ through 6 of its 12 carbonyl groups. Sensors based on this ionophore exhibit remarkable selectivity, such as a 104.5 -fold selectivity over Na+ .17 This selectivity permits analysis in blood and urine, a purpose for which ISEs based on this ionophore continue to be used widely in large throughput (“mainframe”) clinical chemistry analyzers. Disadvantages of valinomycin include, however, a limited solubility in polymeric matrixes of lower polarity and a relatively low hydrophobicity.92 Most synthetic alternatives to valinomycin take advantage of ether oxygens that bind to K+ . Earlier examples of monocyclic crown ethers have given way to more selective bis-15-crown-5 ethers that bind the K+ in a sandwichlike fashion. K+ -II,93 which provides for a 103.7 -fold selectivity over Na+ , is an excellent example. The individual 15-crown-5 macrocycles are too small to bind the K+ in their center. As a result, the K+ ion comes to lie in between the two 15-crown-5 macrocyclic rings, providing for a total of 10 oxygen–ion interactions. Ions smaller than K+ cannot take advantage of interactions with all these ether oxygens, while ions larger than K+ would benefit from interactions with the additional ether oxygens in larger crown ethers. The selectivity of ISEs based on bis-15-crown-5 ethers is, however, limited by the large conformational freedom of these ionophores. Among several efforts to restrict the conformational freedom of the ionophore and thereby increase size selectivity, the development of bridged calix[4]arenecrown-5 compounds such as K+ -II, which was reported to have a 104.2 -fold selectivity over Na+ , is particularly noteworthy.94 Similar trends as for the K+ ionophores can also be seen for Na+ . Very early examples of ISEs based on

the weakest stabilization to interfering cations and counteranions. Indeed, H+ selective ISEs with fluorous membranes and fluorophilic ionophores are among the most selective of all.62, 67 The highest selectivity was observed for H+ -III, whose -(CH2 )5 - spacers shield the H+ -binding nitrogen from the three strongly electron withdrawing perfluoroalkyl groups. Surprisingly, the -(CH2 )3 - spacers of H+ -IV are already too short for sufficient shielding of the nitrogen center. While the selectivity for H+ over Na+ of ISEs based on H+ -III is larger than 13 orders of magnitude and cannot be determined accurately because it is not possible to prepare solutions with Na+ concentrations that are sufficiently high to cause interference, the H+ over Na+ selectivity of ISEs based on H+ -IV is only 109.6 -fold. Using the recently developed ISE theory, this can be explained quantitatively by a pKa of H+ -III in the fluorous phase of 15.4, which is 5.6 pKa units larger than for H+ -IV.67

4.3

Ion recognition based on ion–dipole interactions

4.3.1 Ion-selective electrodes for alkali metal cations Attractive interactions that can be used to design ionophores include ion–dipole interactions, hydrogen bonds, ligation to metal cations, and the formation of covalent bonds. Ligation of an anion to a metalloporphyrin or covalent binding of carbonate to trifluoroacetophenone ionophores are examples of guest recognition based on a monotopic interaction. However, more often than not, ISE ionophores bind ions through multitopic interaction. This is particularly true for the large class of cation receptors that interact with their guests by ion–dipole interactions. Some of the earliest and very successful examples include natural compounds such as the antibiotics valinomycin, monactin, and nonactin already mentioned above (Figure 3), while the majority of recent reports on ISEs with ionophores of this class describe the use of synthetic compounds. Because it would be well beyond the scope of this chapter to discuss this field comprehensively, examples of ionophore-based ISEs

C12H25 O

O O

O O

O

O O

O

O K+-II

CH2

CH2

O

O

O O

21

OCH(CH3)2 O

O O

O

CH2

CH2 OCH(CH3)2

O K+-III

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

22

Supramolecular devices H3C C7H15 H3C N

N

C7H15

H3C C12H25 O O

O O

O

O

O

O

O

H3C

N

O

O

O

O

C7H15

Na+-I

O O

O

O

CH2

CH2

CH2 OC2H5

O

Na+-IV

O

N

N

Li+-I

O O

O

antibiotics were soon followed by more selective synthetic ionophores binding Na+ with multiple carbonyl groups. A particular successful ionophore of that generation is Na+ -I (102.3 -fold selectivity over K+ ), which has been used successfully for intracellular measurements despite the high K+ concentrations in this environment.31 Many ionophores introduced subsequently were macrocyclic crown ethers with oxygens as the recognition sites, as in the case of the K+ ionophores too. With its smaller size, Na+ fits well into 15-crown-5. Not surprisingly, the most successful biscrown ether ionophores have two 12-crown-4 macrocyclic rings, between which the Na+ is bound with eight ion–ether interactions (e.g., the commercially successful Na+ -II with a 102.1 -fold selectivity over K+ ).95 As in the case of K+ selective bis-crown ethers too, the conformational freedom of the bis-12-crown-4 ethers is large, which limits the achievable selectivity, and as in the case of K+ ISEs too, extremely successful bridged calix[4]arenes have been proposed. Na+ -III with a 105.3 -fold selectivity over K+ is

O

O

O

OC2H5

O

Na+-III

O

O

Na+-II

O CH2

O

O O

O O

an outstanding example.96 An alternative approach relying largely on steric repulsion to exclude larger ions and prevent the formation of complexes between one ion and two ionophores is exemplified by Na+ -IV (103.0 -fold selectivity over K+ ).97 Interest in the analysis of Li+ , the smallest alkali metal ion, using ISEs stems predominantly from its use as a mood-stabilizing drug.4, 98 The most important interferent in clinical measurements is Na+ , which occurs in blood in a concentration of approximately 140 mM. Early work with diamide ionophores4 such as Li+ -I, which forms 1 : 2 complexes with Li+ , found selectivities (102.3 -fold selectivity over Na+ ) barely permitting the first successful measurements in blood.99, 100 In 1987, the first commercial clinical analyzer measuring Li+ with ISEs became available.101 While various approaches were taken to improve the selectivity over Na+ , the most successful one is arguably the one in which 1,4,8,11-tetraoxacycolotetradecane was decorated with bulky substituents hindering the formation

CH2CH(CH3)2 N CH2CH(CH3)2 O

O

O

O

O

CH2CH(CH3)2

N CH2CH(CH3)2

Li+-II

Li+-III

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

4.3.2 Ion-selective electrodes for NH4 +

of 1 : 2 complexes and inhibiting the binding of larger ions. The highest potentiometric selectivity was reported for Li+ -II (103.3 -fold over Na+ ),102 but similar selectivities were also found for several other compounds.103 Interestingly, a more than 104 -fold selectivity but a slow response was reported for the sterically even more hindered Li+ -III.103 A fair number of ionophore-based ISEs with selectivities for Rb+ and Cs+ , the two largest alkali metal cations, have also been described. In some of these cases, the potentiometric selectivities suggest that the ionophore binds the target ion selectively. However, in a number of cases such evidence is missing. Since Cs+ has the lowest free energy of hydration of all alkali metal cations, even an ionophore-free ion-exchanger membrane will exhibit selectivity for Cs+ over the other alkali metal cations. To show evidence for selective recognition of Cs+ by the ionophore, the selectivity of an ionophore-based ISE for Cs+ has be higher than for an ionophore-free ion-exchanger ISE based on a membrane containing the same membrane components but no ionophore (but including ionic sites), a control experiment that is unfortunately too often not reported.

Because of the key role of NH4 + in various biological processes, the direct measurement of this ion is important for clinical and environmental analyses. Ammonium concentrations in food also provide a measure of freshness. Moreover, various enzymes catalyze the deamination of numerous organic compounds, which makes NH4 + selective ionophores also of interest for enzyme-based ISEs (Section 3.1.7). For all these reasons, there has been a continued interest in the development of ionophore-based ISEs for NH4 + . For many years, almost all the work on NH4 + ISEs focused on the macrocyclic antibiotics nonactin (NH4 + -I) and monactin (NH4 + -II), see Figure 3.4, 104 While the type of sensing matrix has some effect on the selectivities of ISEs based on these ionophores, an approximately 10fold selectivity over K+ and an approximately 100-fold selectivity over Na+ is characteristic for these sensors. Only in recent years several interesting alternatives have been proposed. Significant improvements of the selectivity of either Na+ or K+ have been obtained by use of rigid crown ethers with bulky substituents that block complexation with larger ions (e.g.,NH4 + -III),105, 106 a

O O

O

O

N H NH O

O O

O

O

O

O

O

O O

O

O

N NH H H N

O

O HN HN

O

O

NH4+-VI

NH4+-IV

NH4+-III

O

O O

O

O

O O

O

(CH3)2Si

O

O

O (CH3)2Si

Si Si(CH3)2

O

Si(CH3)2 O

O

O

O

O

O

O O

23

O

O

O

+ -V

NH4

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

24

Supramolecular devices

strategy similar to the one described above for Li+ and Na+ , A 1000-fold selectivity over Na+ was also achieved with the macrobicyclic NH4 + -IV, which was explained by both hydrogen bonds to the ether oxygens and cation–π interactions.107 In comparison, the dendrimer NH4 + -V with four 15-crown-5 groups has only a low level of host preorganization, which makes the reported nearly thousand fold selectivity over Na+ rather surprising.108 A report on the macrobicyclic peptide ionophore NH4 + VI illustrates some of the often unexpected challenges of ionophore design. Molecular modeling was used for the design of this ionophore, and did result in a selectivity close to the one for monactin- and nonactin-based ISEs, but the poor solubility of NH4 + -VI proved to be an unexpected problem.109

4.3.3 Ion-selective electrodes for alkaline earth metal cations While there has been only a limited interest in ionophorebased ISEs for Be2+ , Sr2+ , and Ba2+ ,4 the development of Mg2+ and Ca2+ ISEs suitable for measurements in blood samples has been for several decades one of the most competitive topics in the field of ion-selective potentiometry. Success came much faster for Ca2+ than for the smaller and more hydrophilic Mg2+ , for which ionophore-based ISE were finally introduced in clinical analyzers in 1994.91 This can be at least partially explained by the fact that, as a result of the stronger hydration of Mg2+ , the use of a hypothetical ionophore that binds Mg2+ and Ca2+ equally strong would result in an ISE with selectivity for Ca2+ over Mg2+ . Therefore, binding of Mg2+ by the ionophore has not only to be selective over Ca2+ , but it also has to compensate for the difference in the free energies of ionophore-unassisted phase transfer from the aqueous into the sensing membrane. Analogous problems arise in the development of other strongly hydrophilic cations (e.g., Li+ ) and anions (e.g., sulfate and phosphate). Because of the intense interest in Mg2+ and Ca2+ ISEs, a lot of work was performed in which the structures of ionophores with different levels of preorganization were varied very methodically by use of substituents with different levels of bulkiness and electronegativity.4 Arguably, this work contributes to the most sophisticated examples in the field of supramolecular chemistry. Unfortunately, because a lot of this work was performed before methods for the determination of complex stabilities in ISE membranes became available, the majority of this work was directed by the experimentally observed potentiometric selectivities alone. Complex stabilities have only been determined much later for a few selected examples.

Ionophores that meet the demands of clinical chemistry and biochemical and physiological research had been developed by the end of the 1990s. More recent work has mostly focused on ionophores that are more readily available but have not led to major breakthroughs in selectivity. Since the history of Ca2+ and Mg2+ ISEs has been reviewed,4, 110 this chapter will only briefly highlight some of the ionophores that represent the current state of the art. Despite their apparently low level of host preorganization, the noncyclic amide ionophores Ca2+ -I (ETH 1001) and Ca2+ -II (ETH 129) continue to be the default ionophores for many real-life applications. While the former is suitable for measurements in blood serum,111 the latter has a better selectivity over Na+ and K+ and is more suitable for intracellular measurements.112 However, ionophores with higher selectivities have recently been developed. Two of the most elaborate examples that only resulted from painstaking optimization include the diazacrown ethers Ca2+ -III and Ca2+ -IV.113 In the complexes of these two ionophores with Ca2+ , the metal ion is situated in the crown ring, and the two pendants bind to the Ca2+ from the top and bottom. O N

(CH2)11 O

N

O H3C

O

H3C

O

O O O

O N

N

O (CH2)11

O

Ca2+-I

O H37C18

N H

Ca2+-II

O

O O

O N

N

O

O O

N H

C18H37

O

O

n Ca2+-III: n = 2 Ca2+-IV: n = 1

The structures of the most successful Mg2+ ionophores have a strong resemblance to those of Ca2+ ionophores. While early work with diamide ionophores for Mg2+ was also performed, the most selective noncyclic ionophores for Mg2+ are the triamides Mg2+ -I and Mg2+ -II.114, 115 A systematic optimization similar to the one for Ca2+ -III

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

O

O

O

N

N H O

O

(CH2)n N H

(CH2)n

25

N

O

N

N H

(CH2)n

Mg2+-I : n = 4 Mg2+-II : n = 5

O

O

O

N H

O

N

O N

O

O N H

O

Mg2+-III

and Ca2+ -IV resulted in ISEs based on Mg2+ -III, which exhibits a 300-fold selectivity for Mg2+ over Ca2+ .113

4.3.4 Ion-selective electrodes for transition metal and lanthanide cations Even though the number of transition metal and lanthanide cations is very large, the day seems to be soon approaching when an ionophore-based ISE will have been reported for the detection of every one of these cations. Arguably, with notable exceptions, the overall level of sophistication of the ionophore design used for these sensors has not yet reached the level that it has in the case of the ISEs for alkali and alkaline earth metal cations discussed above. Many early investigations on ISEs for heavy metal cations (e.g., Cu2+ , Ag+ , Zn2+ , Cd2+ , Hg2+ , Pb2+ , and UO2 2+ )4 were inspired by ionophores that were already readily available from other type of research, such as fundamental studies in inorganic chemistry or the development of chelating reagents for solvent extraction. The same is often true too for ISEs reported to have selectivity for specific lanthanide cations, a field that has gained increased popularity over the past ten years. In a fair number of cases, the ionophores have not been specifically designed for use in ISEs, and suffer from insufficient hydrophobicity. Also, many of these ISEs were described before the theory was developed to permit nano- and picomolar detection limits (Section 3.1.3 and Figure 10). As a result, the lower detection limits observed in most of these studies were not low enough to permit analysis of reallife environmental and biological samples, which limited the enthusiasm for fine tuning of the selectivity to meet

the needs of actual applications by ionophore optimization as described in Sections 4.3.1 and 4.3.3 earlier. Similarly, numerous older studies are hard to interpret quantitatively on the basis of modern ISE theory because either selectivities have not been determined adequately, control experiments with ionophore-free ion-exchanger selectivities have not been reported, or ionophore-site ratios have not been optimized for optimum selectivity. The key type of interactions used in almost all the ionophores for transition metal and lanthanide cations are ion–dipole interactions. However, since these cations vary widely in their hardness as Lewis acids,116 the character of the cation–ligand interactions ranges from mostly electrostatic for hard Lewis acids to mostly covalent for soft Lewis acids. For the latter, ligands with functional groups that are soft Lewis bases, such as amines, pyridines, thioethers, and thiocarbonyls are common. Another feature peculiar to ISEs for this group of analytes is the fairly larger number of electrically charged ligands. This is undoubtedly the result of the use of host compounds first developed for solvent extraction, where many chelating reagents are electrically neutral when fully deprotonated but bind metal cations as mono- or polyanionic species. Unfortunately, the efficient use of electrically charged ionophores is not as easy as that of electrically neutral ionophores.52, 53 As a result, the full impact of the ionic site theory needed to put electrically charged ionophores to optimum use (Sections 4.2 and 5.2) has yet to happen. Numerous recent studies using such ionophores still do not take full advantage of the selectivity improving variation of the molar ratio of ionic sites and ionophore.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

26

Supramolecular devices

However, combining the common ionophore design strategies of host rigidity to enhance ion size selectivity and bulky substituents to control complex stoichiometry with the inherent selectivity provided by soft coordinating groups permits exceptional selectivities. Therefore, in spite of the lesser level of fine tuning that has been performed, heavy metal ion ISEs are among the most selective ISE reported to date. Their comprehensive discussion is well beyond the scope of this chapter, and only a few will be mentioned here. From a historical perspective particularly noteworthy is Pb2+ -I, which was used in 1997 to demonstrate for the first time that picomolar detection limits can be achieved with ISEs with a metal ion-buffered inner filling solution.9 Other examples of extremely selective ISEs include those based on the ionophores Ag+ -I117 and Ag+ -II.16 While the former gives the most selective Ag+ ISEs based on nonfluorous membranes reported to date, the latter is a fluorophilic ionophore that provides selectivities over many ions by an additional two orders of magnitude (e.g., pot log KAg,J for K+ : -11.6; Pb2+ : -10.2; Cu2+ : -13.0; Cd2+ : -13.2). Evidently, this is another example of the extremely high selectivities that result when the ionophores are very selective and the transfer of interfering ions into the sensing membrane is disfavored by the extremely low polarity and polarizability of fluorous phases (for other examples, see H+ sensors in Section 4.2 and the carbonate sensors in Section 4.5). Such high selectivities are also the key to low detection limits, and indeed Ag+ -II has been used to demonstrate a detection limit for Ag+ of 4.1 ppt, which is the lowest detection limit obtained so far for an ionophorebased Ag+ ISE. C12H25

N

C12H25 N

S O

S S

O S C12H25

N

S S N

S

S

Rf10

Rf10

Rf10 = C10F21 C12H25

Pb2+-I

Ag+-I

Ag+-II

As the work with Pb2+ -I first demonstrated, accurately determined (so-called “unbiased”) selectivity coefficients45 can be much larger than what had been reported in the earlier literature. This is particularly true for many reports on the selectivity of transition metal cation ISEs made before 1997, which need to be interpreted carefully. Because of this and because of the new possibilities that nano- and picomolar detection limits offer, it appears that despite a lot of work that has been done in the field of

transition metal and lanthanide cation ISEs, this field has still a large potential to grow and perform much better than it was the case in the past.

4.4

Anion recognition based on hydrogen bonds

While cation–dipole interactions dominate in the recognition of inorganic cations, anion–dipole interactions have been rarely taken advantage of in the design of ionophores for ISEs. Instead, a number of different interaction types have been used to develop ionophores for anions. These include hydrogen bonds, coordination to metal centers, and covalent bonds. Among the earliest examples of electrically neutral ionophores for ISEs that bind inorganic anions through multiple hydrogen bonds were ionophores with two thiourea bonds, such as SO4 2− -I and Cl− -I.118, 119 More recent examples include increasingly complex structures designed to improve host–guest preorganization, such as ureafunctionalized calix[4]arenes (e.g., CO3 2 -I).120 Not surprisingly, binding strengths increased when the host preorganization was increased and when the hydrogenbond donor strength was increased. The latter can be achieved, for example, by switching from a urea to a thiourea group or by using more strongly electron withdrawing substituents attached to the hydrogen-bond donating group.121 However, this concept has its limitations, as an increased hydrogen-bond donor strength typically also comes along with an increased acidity. Loss of a proton reduces the number of hydrogens available for hydrogen bonding and, in the case of an electrically neutral ionophore, results in a negatively charged compound that for electrostatic reasons is even less likely to bind an anion.63 One of the biggest complications in the use of a slightly acidic ionophore for anions is that such deprotonation does not only depend on the pH but also the anion and cation concentrations in the sample. To understand this, let us consider for simplicity an ISE membrane that contains an electrically neutral but acidic ionophore that binds SO4 2− with 1 : 1 stoichiometry and an equal amount of a tetraalkylammonium ions as ionic sites (Figure 23). When exposed to sulfate solutions of moderately low pH, all the ionophore will be in its electrically neutral state, and half of the ionophore will form a complex with SO4 2− while the other half is in its noncomplexed form (left hand side of Figure 23). If the sulfate concentration of the aqueous solution is kept constant but the pH is raised significantly by addition of KOH, the composition of the ion-selective membrane can be changed by deprotonation of the ionophore. Since deprotonation of the ionophore in the bulk of the ISE membrane and loss of the proton into the aqueous

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes t-Bu

27

t-Bu t-Bu t-Bu

O S

N

N

N

H

H O N H H S O O O

S

O S

N N

H

H

H

N H

S S

N

Cl− -I

SO42− -I

solution cannot happen in disregard of the electroneutrality requirement of the ISE membrane bulk, ionophore deprotonation will be accompanied by K+ transfer into the sensor membrane (top right of Figure 23). What is happening is ion exchange. It follows that ionophore deprotonation will not happen at the pH at which the ionophore would be deprotonated if it were present in the aqueous phase, but that instead the pH at which this ion exchange occurs depends on (i) the pK of the ionophore in the membrane phase, (ii) the K+ concentration in the aqueous sample, (iii) the free energy of K+ transfer from the aqueous into the membrane phase, and (iv) the stability of a possible complex in the membrane between the K+ and the negatively charged, deprotonated ionophore. Therefore, the pH at which ionophore deprotonation occurs changes (a) if NaOH is used to raise the pH instead of KOH, or (b) if the aqueous solution contains in addition to sulfate and KOH also another source of K+ that raises the K+ concentration

O

HO OH H N N H H N

N H

S

CO32−-I

in the aqueous sample. To make things even more complicated, ion exchange is not the only mechanism that can lead to ionophore deprotonation. Consider again the membrane of the above composition, this time exposed to a sulfate solution of a pH that is relatively high but not too high, so that all the ionophore is again in its electrically neutral state, and half of the ionophore is in the form of a complex with SO4 2− . Lowering the sulfate concentration in the aqueous solution at constant pH may now result in the transfer of H+ and SO4 2− (in a 2 : 1 ratio) from the sensing membrane into the sample (Figure 23b). In this case, coextraction rather than ion exchange occurs. In either case, the observation of the solution pH at which the ionophore in the sensing membrane is deprotonated is meaningless if this experiment is performed with sulfate-free aqueous solutions. Unfortunately, the complexity of such systems has been underestimated, and too many poorly designed experiments have been described in the ISE literature.

Ion exchange

SO42− K+ K+

(a)

Aqueous phase

LH·SO42−

R+

LH·SO42−

R+

LH LH

R+

[SO42−](aq) constant pH(aq) increased

SO42− K+

LH·SO42−

R+

LH·SO42−

R+

L−K+

R+

−K+

R+

L−

R+

L−

R+

−

L

R+

L−

R+

+

K

R+

L

Membrane phase

Coextraction

SO42− K+ K+

LH·SO42−

R+

LH·SO42−

R+

LH LH

R+ R+

pH(aq) constant [SO42−](aq)

SO42− H+

decreased H+

(b)

Figure 23

Ion exchange and coextraction in membranes with an ionophore that can be deprotonated.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

28

Supramolecular devices

O O

OBn P

OBn

X2− H N + N

O

H

H

N

N

N N

N H

O NH

HN

N H NH

HN

NH

HN

N

N

O SiPh2

SO42−-II

HSO3−-I

Very similar problems are often encountered for ionophores with positively charged, protonated groups that form hydrogen bonds to anionic analytes. Examples include protonated macrocyclic amines and ionophores with guanidinium groups (see also Ref. 4). The complexation chemistry of such compounds in the aqueous phase is among the early highlights of host–guest, but it has been difficult to use such receptors successfully in ionophore-doped ISEs. To minimize such problems, very basic ionophores are preferable, but even the properties of ISEs based on ionophores with guanidinium groups, such as SO4 2− -II,122 HSO3 − -I,123 and SO4 2− -III124 are often difficult to explain and control.

4.5

Anion receptors with metal centers

Among the different interaction types that have been taken advantage of to develop ionophores for anions, the formation of coordination bonds between the analyte anion and a metal center of the ionophore is arguably the most common one. It was also one of the first host–guest chemistry concepts used successfully in ISEs for anions. Among the first such sensors were tin-organic receptors for chloride-selective electrodes (e.g., Cl− -II)125 and nitrite-selective electrodes with lipophilic derivatives of vitamin B12 with a Co(III) center (NO2 − -I).126 From the point of view of real-life applications, one of the most successful ISEs for anions is the one with a manganese(III) tetraphenylporphyrin as ionophore (Cl− -III),127 which was successfully applied to measure chloride in blood serum and intracellular environments. In the mean time, ionophores with a very large number of different metal centers have been investigated; Mn(III), Co(III), Ga(III), Ge(III), Mo(V),

O SiPh2

SO42−-III

Ru(II), Rh(III), Pd(II), In(III), Sn(IV), Hf(III), Hg(II), and U(VI) are examples of many (see Refs 4 and 85). Strategies for the use of these ionophores in a manner that gives the highest possible potentiometric selectivities can be rather involved and have partially been developed only recently.53, 128, 129 Key challenges are the limited chemical stability of some of these ionophores, and the proper control of their stoichiometries with the target ions on one hand and interfering ions on the other hand. Typical examples of ionophores that have suffered from chemical instabilities are organotin compounds and ionophores with a UO2 center. This is particularly unfortunate because both ionophore classes have shown very good phosphate selectivity, which is difficult to achieve since phosphate has a very high hydration energy and requires particularly strong and selective binding for the ionophore to permit preferential transfer of phosphate into the sensing membrane. Several UO2 -salophenes (e.g., phosphate-I) have shown interesting phosphate selectivity due to strong phosphate binding to the U(VI) center, but these ionophores may decompose in the presence of phosphate due to formation of inorganic uranium(VI) phosphate. Buffering of both phosphate and uranyl ions in the inner filling of the electrode is an interesting approach to extend the sensor life time.130, 131 Ionophore decomposition was also observed for several organotin compounds. Very high phosphate selectivities could be achieved with an elegant design putting up to four Sn(IV) centers into one ionophore (phosphate-II),132 but sensor lifetimes were limited to less than one day due to ionophore decomposition. One approach that has been used to control stoichiometries is steric hindrance, as it is similarly discussed above for cation ionophores (Section 4.3). One of the nicest examples of this approach is the one of picket fence porphyrins.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

N Br N

Sn Cl

N Cl N

Co

Mn

N Py N

Cl–-II

29

N

N

Cl–-III

NO2–-I

Cl Sn N

N UO2 O O

Cl Sn

Phosphate-I

A common problem with metalloporphyrin-based ISEs is relatively high interference from OH− . Indeed, OH− binding to many metalloporphyrins is so strong that two metallophorphyrins may bind to one OH− in a sandwichlike 2 : 1 complex.129 A possible consequence of the high OH− affinity is a rather complicated potentiometric response mechanism, as it is illustrated schematically in Figure 24. At high concentrations of the target anion X− in the aqueous sample, the sensing membrane doped with a singly positively charged metalloporphyrin (as typical for metalloporphyrins with a M(III) metal center) and 33 mol% anionic sites contains ionophore complexes LX and free ionophore L+ in a ratio 2 : 1 (Figure 24, left). Here, the ISE will respond to the target anion with the theoretical (Nernstian response) of −59 mV per 10-fold increase in

L+X− Membrane

R−

L+X−

Cl Sn

Cl Sn

Phosphate-II

the activity of X− in the aqueous phase. When the ISE membrane is exposed to aqueous solutions of high pH and low X− activity, ion exchange occurs, and the sensing membrane contains a 1 : 1 ratio of LOH and [L2 OH]+ complexes (Figure 24, right). In this concentration range, the ISE does not respond to X− but only to pH changes. At intermediate concentrations of X− and OH− in the sample (Figure 24, center), the sensing membrane contains a 1 : 1 ratio of LX and [L2 OH]+ . Because there is no free ionophore in this concentration range, the X− concentration in the membrane is not buffered well, and unusual potentiometric responses such as apparently “twiceNernstian” response of −116 mV per 10-fold increase in X− at constant pH may be observed in this range.49, 122, 129, 133 This increase in response slope may appear attractive, but

L+X− L+OH−L+

R−

L+OH−

R−

L+OH−L+

+

L

Sample

Low pH high X− activity

High pH low X− activity

Figure 24 Formation of different ionophore complexes in an ionophore-doped membrane when exposed to sample solutions of differing hydroxide and target anion concentrations. Note the middle region in which target ion and hydroxide complexes of different stoichiometry coexist. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

30

Supramolecular devices

L+: 4, R–: 1

L+: 3, R+: 2 O

3×

O

L

L+: 3, R+: 4 O

O

1×

L

O

L

O

L

1:1 Stoichiometry

O 4×

L

O

O O

5×

1× O

2×

L

2×

R

4×

R

O

O

1:1 Stoichiometry

2:1 Stoichiometry

O

L

O O

8×

L

O

O

R

Sensor selectivity: determined by ionophore for the most weakly bound CO32–

Figure 25 Formation of different CO3 2− complexes of a positively charged ionophore (L+ ) in membranes with different charges and concentrations of ionic sites.114 Note that in membranes with more than one type of complexes, the potentiometric selectivity is determined by the most weakly bound ions.

unfortunately it is also accompanied by worse selectivities over other interfering ions than would be observed for ISEs with the left hand membrane composition.133 Moreover, while response times of electrode membranes with compositions represented by the left hand side and center of Figure 24 are fast, the ion exchange that occurs in the transition range betweens these two regions causes very slow response times. This all points to the fact that simultaneous formation of LX and [L2 OH]+ in an ISE membrane should be avoided. One very elegant approach to prevent the formation of [L2 OH]+ -type porphyrin complexes is to use porphyrins with very bulky peripheral substituents. Both In(III) and Ga(III) picket fence porphyrins have been successfully used in ISEs without observation of apparently “twice-Nernstian” responses (picket fence porphyrin).129 Alternatively, dimer formation was avoided by attachment of the ionophore to the backbone of the polymer matrix.64 t -Bu

t -Bu

C=O

O=C

t -Bu

HN

C=O

t -Bu

NH O=C

N N

NH M

N N

NH

M: In(III) or Ga(III) picket fence porphyrin

The apparently “twice-Nernstian” responses discussed above result from different complex stoichiometries for the target anion and OH− . However, the target anion itself may

also form complexes of different stoichiometries with the ionophore, an effect that can be controlled with the concentration and charge of the ionic sites. An example that was experimentally observed for a Mn(III) porphyrin is illustrated in Figure 25.134 In this case, 25 mol% of anionic site relative to the ionophore result in the strongest binding of carbonate since two ionophores per carbonate assist the transfer of the carbonate into the sensing phase. When 67 mol% of cationic sites are used instead, a significant fraction of the carbonate in the membrane is only bound by one ionophore molecule, and when 133 mol% cationic sites are used, two carbonate molecules have to share one ionophore. The latter is possible because in a Mn(III) porphyrin axial ligands can bind to the Mn(III) center both from above and from below the porphyrin plane. Figure 26 shows the membrane composition for the same ionophore and ionic site ratios if the sensing membranes are exposed to solutions of the interfering ion thiocyanate. A whole different set of complexes is observed. Whether negatively or positively charged ionic sites give the higher selectivity, and which mole ratio of ionophore and ionic sites gives the most selective ISE membrane depends on the stoichiometries and stabilities of the complexes of the target anion and the interfering anions. Because these are typically not known in advance, different membrane compositions need to be tested experimentally to determine the most selective membrane compositions. While a thorough treatment is often not trivial, the amount of experimental work is somewhat reduced by the knowledge of specific ratios of ionic sites and ionophore that give the highest selectivities for specific complex stoichiometries and charges of ionophores, target ion, and interfering ions. The numerical values for the ratio of ionic sites and ionophore that

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

L+: 3, R+: 2

L+: 4, R–: 1 3×

1×

NCS

L

L

2 × SCN 1:1 Stoichiometry

L+: 3, R+: 4

NCS

3 × SCN

NCS

L

L

1×

R

L

NCS

1:1 Stoichiometry 1×

1×

31

NCS

No available ionophore 2×

4×

R

R

Sensor selectivity: determined by ionophore for the most weakly bound SCN–

Figure 26 Formation of different SCN− complexes of a positively charged ionophore (L+ ) in membranes with different charges and concentrations of ionic sites.114 SCN− is chosen here as a representative interfering, and ionophore and site concentrations correspond to Figure 25. Membrane phase

Aqueous phase Sample contains only primary analyte X2−

X2–

X2– X2–

K

X2–

X2–

X2–

X2–

X2– X2–

R+ R+

X2–

+

X2–

+

X2–

+

+

Y–

Y–

R– R–

Analyte X2−

−2 log [L+ (mol/kg)]

R+ R+ R+ R+ R+ R+

−3 −4 Y− Interfering ion

−5 −6 −150

−100

−50

0

50

100

−zR[RzR]/Ltot (mol %)

Sample contains only interfering ion Y–

Y–

Y– Y– K Y– Y– Y– Y– Y

–

Y

Y– –

Y

–

R+ R+ R+ R+ R+ R+

Y–

Y– Y– Y–

Y– Y–

R+ R+

Y–

Y–

R– R–

+

+

Highest selectivity

Figure 27 Dependence of the free ionophore concentration on the charge sign and concentration of ionic sites. Top row and blue line: divalent target ion. Bottom row and red line: monovalent interfering ion. The highest potentiometric selectivity is obtained at the ratio of ionic sites and ionophore for which there is a high concentration of free ionophore when the sensor is exposed to target ions but a very low concentration of free ionophore when the sensor is exposed to interfering ions. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

32

Supramolecular devices

O F3C CF3

+

O

O O

2−

CO3

O

F3C

O O

O

O

CF3

O

C4H9 C4H9

Figure 28

C4H9

Covalent binding of CO3 2− to a trifluoroacetophenone ionophore.

give the highest selectivity are well known from theory and have been tabulated but typically cannot be predicted intuitively.53, 128, 135, 136 For example, for a singly positively charged ionophore L+ that binds both the target anion X2− and the interfering ion Y− with 1 : 1 stoichiometry, the optimum ratio is 62 mol% cationic sites relative to the ionophore. Why this ratio provides the best selectivity is illustrated by Figure 27. It shows with a blue line how the concentration of free ionophore in the bulk of the sensing membrane exposed to an aqueous solution of primary anion X2− depends on the concentration of the ionic sites in the sensing membrane. Note that positive values on the x-axis refer to cationic sites, and negative values to anionic sites. The red line gives the corresponding concentration of free ionophore in the sensing membrane if this membrane is exposed to solutions containing only the interfering anion Y− . The optimum concentration of 62 mol% cationic sites corresponds to a membrane that has a high free ionophore concentration when the sensor is exposed to solutions of the target anion but a very low concentration of free ionophore when the membrane is exposed to interfering ions only (note the logarithmic scale of the y-axis). The same tabulated values for ionic site–ionophore ratios can also be used to investigate and optimize the selectivities of ISE membranes doped with electrically neutral ionophores, but note that for those ionophores this selectivity optimization is often simpler because the number of conceivable complexes is often smaller.53, 128, 135, 136

4.6

C4H9

Ionophores that form covalent bonds to the target ion

Only a small number of ISEs based on an ionophore that binds the analyte of interest by formation of a covalent bond have been reported. Many of them are carbonate-selective electrodes, which are of great interest to clinical chemistry and environmental monitoring. Initially introduced as inert membrane matrixes, trifluoroacetyl-p-alkylbenzenes were found to give ISEs a good selectivity for carbonate. Further studies confirmed that carbonate binds to trifluoroacetophenone derivatives by nucleophilic addition to the carbonyl

group,137 resulting in the formation of a covalent bond (Figure 28).138–142 However, because the concentration of CO3 2− at physiological pH is low in comparison to that of CO3 2− , these ISEs suffered from interference of ions such as Cl− , Br− , and salicylate. The latter two are not typically present in the human body, but are components of common drugs. Therefore, significant effort has been spent to prepare improved carbonate-selective electrodes. Most of this work has been devoted to the development of trifluoroacetophenone derivatives with different substituents on the aromatic ring. Replacing of alkyl substituent in para position by electron acceptors such as amide, ester, and sulfone groups or adding strong electron withdrawing groups, such as NO2 or Br, in meta position enhanced selectivities by up to 1.5 orders of magnitude.143–145 Moreover, a neighboring group that stabilizes the carbonate–ionophore complex through intramolecular H bonding resulted in a more than 146 10-fold enhancement in CO2− Also, molec3 selectivity. ular tweezers-type ionophores have been also obtained by coupling two trifluoroacetobenzoyl groups in one receptor, providing 10- to 300-fold improved selectivity for CO3 2− over other anions.147 Because the carbonate selectivity provided by these ionophores is often not satisfactory enough, supplementary approaches have been used to reduce interference from highly lipophilic anions, such as salicylate and perchlorate. These include coating of the ionophore-doped sensing membrane with a hydrophilic, porous layers that slows down the access of large, lipophilic anions such as salicylate to the sensing membrane.148–151 A similar concept as for carbonate ionophores has also been used for sulfite sensors. In this case, the well-known addition of sulfite to benzaldehyde derivatives by reversible formation of a covalent bond has been taken advantage of (Figure 29).152 However, while used for ion-selective optodes, this approach has not been tested for ISEs. Optodes based on the addition of H2 O or alcohols to phenyl ketones have also been reported for the detection of humidity and alcohols in beverages, respectively, but because of the nonionic nature of these analytes, ISEs based on these ionophores are not feasible.4

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

OH

O

SO3− H +

C18H37O

HSO3−

4.7

H

C18H37O O

O

Figure 29

33

Covalent binding of HSO3 − to a benzaldehyde ionophore.

Ionophore-based ion-selective electrodes for organic ions

One approach to the detection of electrically neutral organic analytes is based on the use of enzymes that catalyze the formation of an ion that can be detected potentiometrically, as discussed in Section 3.1.7. The direct detection of electrically neutral organic species with ISEs is not possible since the transfer of these species across the interface of an aqueous sample into an ISE membrane does not involve the transfer of an electric charge across this interface. However, electrically neutral species can affect the phase boundary potential indirectly. Several studies have shown how alcohols, carboxylic acids, and oligoethers (including several surfactants of this type) cause a potentiometric response upon distribution from the aqueous sample into ionophore-free cation-exchanger membranes loaded with metal cations.4, 153, 154 This can be explained by specific binding of these electrically neutral, hydrophobic organic species in the ISE membrane to the metal cations, which lowers the activity of the metal cation in the sensing membrane and causes a change in the phase boundary potential at the sample–membrane interface (2). An analogous principle has also been described for the detection of phenols.155 While this approach does not involve the use of an ionophore, the interaction of the metal cation with these organic substrates falls undoubtedly within the field of host–guest chemistry. A much larger number of ISEs have been proposed for the detection of electrically charged organic analytes. Many of these ISEs fall into the category of ionophorefree ion-exchanger electrodes, and while they do often exhibit high selectivities for fairly hydrophobic organic ions over many inorganic ions, they are from the point of view of supramolecular chemistry not of interest. However, a number of ionophores have been used to prepare ISEs for the detection of organic ions. Some of these ionophores have structures that are quite similar to those of ionophores used in ISEs for inorganic ions. These include, for example, crown ethers and calixarenes used for the detection of protonated amines and amino acid esters, and metalloporphyrins for the detection of

anions.4 Recent examples include, for example, tin(IV) and zirconium (IV) porphyrins for the monotopic recognition of phthalate (phthalate)156 and citrate (citrate),157 respectively. COO− COO− Phthalate

−COO

COO− HO COO− Citrate

Other ionophores have been specifically designed for the selective recognition of organic analytes. Among the recent particularly unique examples are receptors with three trifluoroacetophenone groups for the recognition of phenylalanine (phenylalanine-I)158 and sophisticated hydrogenbonding hosts for nucleobases, such as the adenosine 5 monophosphate binding receptor AMP-I.159 A special feature of organic ions is that many of them are optically active. Since the complexes of a chiral ionophore with two enantiomers are diastereomers, their formation energies differ from one another and permit the preparation of enantioselective ISEs. While already recognized back in the 1970s,4 examples for this concept continue to be reported on a regular basis. The organic chemist will recognize in ionophores used for this purpose structural elements familiar from enantioselective catalysis. A typical recent example is the popular binaphthyl group of phenylglycineOCH3 -I, a receptor that has been used for the enantioselective potentiometric detection of phenylglycine methyl ester.160 In summary, considerable efforts have been spent to show the feasibility of ISEs for organic analytes. While some authors used ionophore-based ISEs to measure organic target analytes in relatively simple samples such as drug tablets that contained besides the analyte only hydrophilic inorganic ions, only very few reports exist on the use of such electrodes for measurements in more complex samples such as blood, saliva, or milk.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

34

Supramolecular devices

X

O O

O

CF3 O

N

CF3

F3C

O O O

O

O

O

O

H

N

O RO P O − O O−

O

N H

N

N

H N

H N

N

N O OR

N

O

OH OH AMP-I

Phenylalanine-I

O O O

O H3CO

O

OCH3

O

PhenylglycineOCH3-I

5

5.1

QUANTITATIVE THEORY OF ION-SELECTIVE ELECTRODES THAT BENEFITS THE HOST–GUEST CHEMIST Quantification of selectivity

(i.e., E ◦ ).   2.303RT pot z /z log Ki,j aj i j zi F 2.303RT 2.303RT pot ◦ =E + log(Ki,j ) + log aj zi F zj F ◦

EMF = E +

◦

Selectivity is not only a main characteristic of an ISE, but it can also be used to determine the stoichiometries and stabilities of ionophore complexes and to optimize the selectivity in view of specific applications (Section 4.3). In extension of the Nernst equation (1), the potentiometric pot selectivity coefficient, Ki,j , is given by the following equation3, 50, 136, 161 : ◦

EMF = E +

  2.303RT pot z /z log ai + Ki,j aj i j zi F

(7)

where ai is the concentration of the ion of interest—often referred to as the primary ion—and aj is the concentration of an ion j potentially interfering with the determination of ion i. In the absence of interfering ions, (7) simplifies to the Nernst equation (1). In the absence of primary ions, (7) simplifies to the Nernstian equation formulated for the interfering ion j , but with a different intercept

=E +

2.303RT log aj zj F

(8)

The different value for the intercept reflects the different free energies of transfer of the primary and interfering ions from the aqueous sample into the ISE membrane and the different free energies of complexation. Equation (8) correctly predicts the different response slope of the calibration curve for an interfering ion that differs in the charge number from the primary ion. Moreover, when the primary and interfering ion have the same charge number and the two ions form complexes of the same stoichiometry, (7) not only predicts the response of an ISE to solution of ions i or j , but it also applies to mixed solutions of the two ions. An example for a response to the primary ion i with a constant background concentration of ion j is shown in Figure 30. At low concentrations the ISE responds only to the interfering ions, and at high concentrations the ISE responds only to the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc097

Ionophore-doped sensing membranes

35

EI0

59.2 mV

aJ + = 10

−4

aJ + = 10−2

EMF (mV)

EMF (mV)

EJ0 EK0

59.2 mV

aJ + = 0 −7

−6

−5

−4 −3 log a I+

−2

−1

Figure 30 Potentiometric responses to a primary ion (I+ ) in a background with a constant concentration of an interfering ion (J+ ). Note that the lower detection limit for the primary ion worsens as the concentration of the interfering ion increases.

primary ion. The crossing point obtained from the extrapolation of the two linear sections of the response curve pot can be used to determine Ki,j . This method of selectivity determination is referred to as the fixed interference method (FIM),136, 161 while the complementary method of measuring the response to the interfering ion in a fixed background of primary ion is referred to as fixed primary ion method (FPIM). When the primary and interfering ions have different charges or form complexes of different stoichiometries, the potentiometric response in a relatively narrow region near the crossing point of the linear response sections differs slightly from what is predicted by (7). These effects have been discussed elsewhere50, 162 and are beyond the scope of this chapter, in particular, since even in those cases the extrapolation of the linear sections can still be used to pot determine Ki,j values. A different approach for the deterpot mination of Ki,j values is the separate solution method (SSM),136, 161 which is based on the measurement of the calibration curves for separate solutions that contain either only primary ions or only interfering ions (Figure 31). If they are carefully performed, the FIM, FPIM, and SSM pot methods all give within error the same values for Ki,j . This requires most importantly that only sections of the EMF response curve in which the ISE exhibits a Nernstian response to the primary or interfering ion are used. When pot experiments are performed carefully, differences in log Ki,j experimentally determined with different methods do not exceed 0.1–0.2 units, and larger differences are cause of concern. Evidently, an interfering ion characterized by a very pot small value of Ki,j affects the measured EMF only at very low activities of the primary ion. On the other hand, pot strongly interfering ions have very large values of Ki,j . pot There are no generally acceptable or unsatisfactory Ki,j pot values. What value of Ki,j can be tolerated depends on the

−6

−5

−4

−3 log a I+

−2

−1

0

Figure 31 Potentiometric response curves to the primary and interfering ions as measured with the separate solution method (SSM). As illustrated for ion K, the detection limit of the ISE may be significantly worse than for the primary ion and may make it difficult to determine unbiased selectivity coefficients.

specific application for which the ISE is used. For example, when Na+ with an expected activity in the 140 mM range is measured in blood with an expected H+ activity around pot 10−7.2 M (i.e., pH 7.2), KNa,H value can be relatively high without a cause of concern. On the other hand, a pH measurement in the same sample requires a very small value pot of KH,Na since the H+ activity is more than six orders of magnitude lower than the Na+ activity. The best ISEs developed over recent years are characpot terized by Ki,j values for different ions that cover many orders of magnitude. As a result, selectivity coefficients are typically reported for convenience on a logarithmic pot scale as log Ki,j values, which is also more informative when presented graphically. This approach distinguishes ISE publications from many reports on chemical sensing in the general, organic, and biological chemistry literature, where changes in an observed variable such as absorbance or fluorescence are not infrequently reported for an arbitrary concentration. For example, some authors may have observed the formation of a fluorescent complex between Ag+ and a new silver ionophore and measured the fluorescence intensity at a silver concentration of 10−6 M. They then proceed to measure the fluorescence intensity for 10−6 M solutions of other metal ions and report the fluorescence intensity for those ions in percent relative to the fluorescence intensity for Ag+ . Clearly, values such as Leu ∼ Val > unnatural amino acid tertleucine (Tle), which can be seen in Figure 20(b). CA 63 was performed using the entire dataset to give a 2D plot in order to identify diagnostic patterns present in the array.89 Anslyn et al. also summarized pattern-based peptide recognition in a review.90

peptides in water at neutral pH by the complexation of 51 and azopholxin 55 as an indicator (Figure 18c).87 The seminal report of Kubo et al. in 1995 was a milestone for colorimetric chiral recognition. They described (S)-BINOL-appended calixarene 56, which exhibited distinct color changes in EtOH by binding to the enantiomers of amino alcohol 57 and related amino acid 58 (Figure 19). This chiral sensing is derived from the differences of the binding constants for each enantiomer; the host molecules were appropriately designed for efficient binding with host (R)-isomers via hydrogen bonding.88 In 2006, Anslyn et al. prepared an enantioselective differential array comprised of optical sensors. The system employed Cu2+ complexes of bidentate N-donor ligands 59–61 as receptors and the catechol and salicylated chromophores (pyrocatechol violet 50, chromoxane cyanin R 62, and chrome azurole S 63) as indicators (Figure 20a). A series of IDAs was able to discern hydrophobic amino acids with high enantio- and chemoselectivity. All of the IDAs conformed to the relative chemoselective ordering for complex stability

6.3

Colorimetric sensors for saccharides

Colorimetric sensors for saccharides are of particular interest in a practical sense. If a system displaying a color change can be developed, it could be incorporated into a diagnostic test paper for D-glucose, similar to universal indicator paper used for pH. A pioneering study by Shinkai et al. described a colorimetric boronic-acid-based saccharide sensor 64 in aqueous solution. This sensor molecule 64 employs an intramolecular interaction between the tertiary amine and the boronic acid group to promote color changes on addition of saccharide (Figure 21a).91 Following the publication, a number of arylboronic acid

O

O N

OH X HO

X=

N

O

O

O

O

O 56 NH2

NH2 OH

OH O

(a)

57

58

(i)

(ii)

(b)

Figure 19 (a) Molecular structures of (S)-BINOL-appended calixarene 56 and guest species such as amino alcohol 57 and related amino acid 58 and (b) visible difference of 56 between enantiomers of (i) 57 and (ii) 58: (i) 56, 56 + (R)-57, and 56 + (S )-57 and (ii) 56 + (R)-58 and 56 + (S )-58 from left to right (Redrawn from Ref. 88.  Nature Publishing Group, 1996.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc098

20

Supramolecular devices OCH3 OCH3

CH3O

OCH3

OCH3

NH

NH

NH

NH

HN

NH OCH3

OCH3

OCH3 59

60

CO2H HO

Cl

CH3O 61

OCH3

CO2H O

CO2H HO

Cl

CO2H O

SO3Na

SO3Na (a)

62

Trp

63

Phe

Leu

Val

Tle

D

L

(b)

Figure 20 (a) Molecular structures of bidentate N-donor ligands 59–61 as receptors and the catechol and salicylated chromophores (chromoxane cyanin R 62 and chrome azurole S 63) as indicators and (b) colorimetric output for 59 (3.5 × 10−2 M), Cu(OTf)2 (2.35 × 10−4 M), 63 (3.5 × 10−5 M), and amino acid (2.0 × 10−4 M) in CH3 OH:H2 O (1 : 1) 5 × 10−2 M HEPES buffer at pH = 7.8 (Redrawn from Ref. 89.  American Chemical Society, 2006.)

sensors for sugars were reported.92, 93 Since these sensors are inherently selective for fructose or, by design, glucose, there was a demand for chemosensors that allowed the detection of other saccharides. In 2006, Rusin and Strongin et al. prepared a boronic-acid-functionalized rhodamine derivative 65 (Figure 21b), which displays an unprecedented degree of colorimetric and fluorimetric selectivity for ribose and ribose derivatives. For example, nucleotides and nucleosides such as adenosine can be detected selectively compared to fructose, glucose, and other common saccharides. Hypothetically, the secondary interactions with the rhodamine chromophore occur when 65 binds to specific sugars. Ribose is detected colorimetrically in this example at levels potentially relevant for diagnosing ribose5-phosphate isomerase deficiency. On the other hand, boronic acid 66 and its associated chromophoric species 67 produced by oxidation (Figure 21b) exhibited selective detection of fructose compared to ribose.94 Furthermore,

in 2007, Anslyn et al. reported boronic-acid-based peptide receptors 68 (Figure 21c), derived from a combinatorial library, for pattern-based saccharide sensing in neutral aqueous media. The binding of saccharides to these receptors was assessed colorimetrically using an indicator uptake protocol in the taste-chip platform. The chemosensor array is capable of classifying disaccharides and monosaccharides as well as discriminating between compounds within each saccharide group.95

6.4

Colorimetric sensors to visualize ordinary organic molecules

The development of protocols to detect a variety of organic molecules by color changes required consideration of the facile qualitative estimations of bioactive

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc098

Colorimetric sensors

HO

O

O B

B

HO

HO

O

N

HO

+ D-(+)-glucose −2H2O

21

N

OH

2H2O

N

N

N

SO3Na (a)

N

SO3Na

64

(HO)2B

B(OH)2

OHHO

HO O

N

OH HO

OH HO

OH

NH

CO2− B(OH)2

B(OH)2

65

B(OH)2

B(OH)2

66 O

OH HO

HO

O HO

OH

m

n

B(OH)2

n = 0, 1, 2... m = 0, 1, 2...

(b) R1

H N O

O N H

B(OH)2

B(OH)2 67

R3

H N R2

O

O N H

R5

H N R4

NHFmoc O

68 R1–5 = Side chains of natural amino acids and

(c)

+ N H2

− B(OH)3

Figure 21 (a) D-Glucose binding of a boronic-acid-based saccharide sensor 64, (b) molecular structures of a boronic-acidfunctionalized rhodamine derivative 65, boronic acid 66, and its associated chromophoric species 67, and (c) molecular structures of boronic-acid-based peptide receptors 68.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc098

22

Supramolecular devices

N

R

R

N

N OH HN

O HO

O O

O OH HO

N O

OH

O HO

HO O OH

N

N

Zn

R

O

OH O OH

HO

M N

OH

(a)

70 M = 2H or metal ions

R N R

N R

R

OH O OH

69

O

N

N

R

71 R = OSiCH3t Bu

OH OH O OH

OH OH O

O

Sn4+ Co3+ Cr3+

Mn3+ Fe3+ Co2+

O OH

Figure 22 (a) Molecular structure of a methyl-red-modified βcyclodextrin (MR-CD) 69.

molecules. In 1992, Ueno et al. reported a methyl-redmodified β-cyclodextrin (MR-CD, 69, Figure 22), which underwent a color change from yellow to red on addition of some organic compounds because of host–guest complexation. In an aqueous solution containing 10% ethylene glycol at pH 1.60, marked spectral changes occur with increasing concentration of 1-adamantanecarboxylic acid. The band at 455 nm decreases, while there is an increase in the peak at 510 nm. The spectral behavior demonstrates that MR-CD 69 releases its methyl red moiety from the cavity into the bulk water environment, and protonation occurs at the azo group, resulting in a structural change. Color changes were conducted by addition of various organic molecules such as L-borneol, L-camphor, D-camphor, L-fenchone, D-fenchone, L-menthol, D-menthol, geraniol, nerol, cyclohexanol, cycloctanol, 1-admantanol, and 1-adamantanecarboxylic acid. Based on the increase of absorbance at 510 nm relative to the original value as a sensitive parameter, adamantane derivatives showed a higher degree of sensitivity, which was correlated with the binding constant. Selectivities of L-isomers were higher than those of the D-isomers because of the discrimination of chiral isomers by chiral MR-CD.96 In 2000, Suslick et al. prepared array-based vapoursensing devices to detect and differentiate between chemically diverse analytes. The chemoselective response of a library of immobilized vapour-sensing metalloporphyrin dyes 70 (Figure 23a) permits the visual identification of a wide range of ligating species (alcohols, amines, ethers, phosphines, phosphites, thioethers, and thiols) and even weakly ligating molecules (arenes, halocarbons, and

Cu2+ Ru2+ Zn2+

Ag2+ 2H+

TPP array

DMF

Ethanol

Acetonitrile

Acetone

THF

(b) P(OC2H5)3

P(C4H9)3

C6H13SH

Pyridine

Hexylamine

CH2Cl2

CHCl3

(C3H7)2S

Benzene

Figure 23 (a) Structure of metal complexes of tetraphenylporphyrins 70 and 71 and (b) color-change profiles for a series of vapors; the degree of ligand softness (roughly their polarizability) increases from left to right and top to bottom; each analyte was delivered to the array as a nitrogen stream saturated with analyte vapor at 20 ◦ C for >15 min (Redrawn from Ref. 11.  Nature Publishing Group, 2000.)

ketones). In this system, solutions of various metal complexes of tetraphenylporphyrin 70 in either CH2 Cl2 or C6 H5 Cl were spotted onto reverse-phase silica thin layer chromatography (TLC) plates to produce the sensor array. A comparison of color changes at the saturation point for a wide range of analytes is shown in Figure 23(b), which shows the colors by simply subtracting the original control images from the final sample images. Each analyte is easily distinguished from the others, and there are family resemblances among chemically similar species (e.g., pyridine and n-hexylamine). Analyte distinction originated both

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc098

Colorimetric sensors in the metal-specific ligation affinities and in their specific unique color changes on ligation. An advantage of utilizing chemoselective sensors in an array is that unique patterns can be identified at low vapor concentrations (e.g.,p>luminescence on binding of Zn2+ .101 Complex 61 shows a decrease in luminescence, while complexes 62–64 show an increase, with the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

4+

17

4+

N N

N Ru

N

N N

N N

N

N

N

N

N

62

4+

N N

N Ru

N N

63

N

CO N Re N

OC Cl

O N

65

O

O

O O

Heavy atoms can also be detected using transition-metalbased systems, for example, Pb2+ causes luminescence enhancement in the Re(I) tricarbonyl complex 65. In this case, the Pb2+ is chelated into a large azacrown ether ring attached directly to a phenanthracene (phen) ligand, resulting in significant luminescent enhancement.102

9

TRANSITION-METAL-BASED ANION SENSORS

Anionic guests also perturb the emission properties of transition-metal-based lumophores in host–guest type complexes. Modification of classic ruthenium polypyridyl complexes with anion recognition groups has led to many first-rate examples of anion-sensing complexes for a large

N

N

N

Ru

N

N

difference in the behavior ascribed to the different pathways for MLCT excited-state decay.

OC

N

N

N

N N

N

Ru

N

N

N

Ru

N

4+ N

N

N

N

N

N

N

N

Ru

N

N

Ru

N

Ru 61

N

N

N

N

N

N

N

N

64

N

N

N

range of anions.103–109 Beer and coworkers have utilized the [Ru(bpy)3 ]2+ moiety to great effect in developing a series of Ru(II) complexes where one of the three bpy ligands has been modified, with amide links, to include acyclic-, macrocyclic-, and calixarene-based anion receptors (66–71).103, 108, 110 In these complexes, the binding of either Cl− or H2 PO4 − results in a blue shift, and increase in intensity, of the MLCT emission band. By inclusion of an acyclic ferrocene anion receptor (71), Beer and coworkers were able to selectively bind H2 PO4 − over Cl− and HSO4 − in acetonitrile.111 Alteration of the groups attached to one of the bpy ligands has led to the development of many responsive, luminescent Ru(II)-based anion sensing systems, including phosphate detection in aqueous environments,112 dinuclear Ru(II) anion-binding complexes,113 and selective anion-sensing complexes.114 Crystal structures of some of these Ru(II) complexes reveal the importance of hydrogen bonding in the anion-binding process, as well as the importance of cavity size in the macrocyclic receptors (Figure 5). Other groups have also utilized Ru(II) bipyridyl metallolumophores to great effect. For example, Baitalik and coworkers prepared mono- and dinuclear mixed ligand complexes (72 and 73) that feature bpy and 4,5-bis(benzimidazol-2-yl)imidazole that act as effective sensors for F− and AcO− , and to a lesser extent H2 PO4 − through luminescence quenching in acetonitrile (Figure 6).115 Thomas and coworkers have used the self-assembly of kinetically stable Ru(II) metallomacrocycles (74–76) to

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

18

Supramolecular devices

2+ OCH3 OCH3

O N N

N

2+ O

N

Ru N

66

N

N

N H

N

N

O

Ru N

N

H N

N

O

OCH3

Ru

H N

N

N H

N

N

N

2+

O

N H

OCH3 OCH3

67

N

H N

N OCH3 68

O

O

O

2+

OH RO O HO 2+ O

2+ O

HN

NH

O

O

N Ru N

N

N

Ru

N

N

Ru

N

N

N 69

N

N

O

70

O

(a)

(b)

(d)

H N

N

HN

N

N H

N

N

HN

N

N

N

Fe

Fe

71

(c)

(e)

Figure 5 Perspective views of crystal structures of Ru(II)-based anion receptors prepared by Beer and coworkers (c = 66 and d = 70). (a) and (b) demonstrate the effect that the macrocycle size has on the binding of Br− and Cl− , respectively, that is, the macrocycle is more accommodating for the smaller chloride anion. Solvent molecules, hydrogen atoms, and some counter anions have been omitted for clarity. Carbon, gray; nitrogen, blue; oxygen, red; Ru(II), scarlet; Cl− , green; and Br− , bronze. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

19

N H

N H

H N

N N

N

72

73

Figure 6 Schematic of the imidazole-based ligand used to prepare 72 and 73, and crystal structures of the resulting mono- and di-nuclear complexes. Carbon, gray; nitrogen, blue; and Ru(II), scarlet. N

M

N

N N

N N

N

N

N

Ru N

M = [Pd(en)]2+ (74) M = [ReCl(CO)3]+ (75) M = [Pt(en)]2+ (76)

N Ru

N

N N

N

M

N

4+

N

N Ir

N

4+

N

N N

N

N

Ir

N

N N

N

77

affect the sensing of anions where BF4 − , SO4 2− , and BPh4 − occupy the large central cavity and modulate the Ru(II) luminescent properties. In these complexes, the major attractive forces between the metallomacrocycle and anion are a combination of electrostatics and π –π interactions.116 While Ru(II) complexes have been frequently studied for use as luminescent anion sensors, they are by no means the sole transition metal ion applicable for this purpose. Re(I), Os(II), Pt(II), and Ir(III) have been utilized in anion sensing complexes to good effect. For example, the Ir(III) terpyridine complexes 77 and 78 show a quenching of luminescence on addition of halide anions, but interestingly the same quenching is not observed with other anions (NO3 − , SO4 2− , H2 PO4 − , PF6 − , and BF4 − ).117

10

TRANSITION METAL DNA-BINDING SYSTEMS

Nucleic acids are also a target for interaction with luminescent transition metal complexes.118 The anionic nature of DNA lends itself to targeting with cationic

78

N

N

complexes, in particular, Ru(II) complexes have been intensively researched as they show luminescent enhancement on interaction with DNA (brought about by protection from quenching). There are three methods by which metal complexes can bind to DNA: (i) groovebinders, (ii) metallointercalators, and (iii) metalloinsertors (Figure 7).119 Early work in this field was carried out by Barton and coworkers who recognized the stereo selectivity of [RuII (phen)3 ]2+ , where the  enantiomer has a greater affinity for the right-handed helix of DNA (Figure 8).120–122 Changing the nature of the ligands coordinated to the Ru(II) center can drastically alter the binding affinity that the complex has for DNA. For example, while [Ru(bpy)3 ]2+ binds relatively weakly through electrostatic interactions to one of the grooves, the replacement of one bpy ligand with another heteroaromatic ligand, 1,4,5,8,9,12hexaazatriphenylene (HAT) to give [RuII (bpy)2 (HAT)]2+ (79) results in a much stronger binding in a more intercalative manner.123, 124 By further increasing the size of the heteroaromatic ligand to dipyrido[3,2-a:2 ,3 -c]phenazine (dppz), the binding affinity is increased further. The complex [Ru(bpy)2 (dppz)]2+ (80) is nonluminescent in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

20

Supramolecular devices

(a)

(b)

Groove binder

Metallointercalator

(c)

Metalloinsertor

Figure 7 Binding modes of metal complexes with DNA: (a) groove binding, (b) intercalation, and (c) insertion. (Reproduced from Ref. 119.  Royal Society of Chemistry, 2007.) 2+

2+

N

N N

N

N

N

Ru N

Ru N

N

N

N

N

Λ

∆

Figure 8 Enantiomers of [RuII (phen)3 ]2+ ; the  enantiomer shows higher selectivity toward DNA.

water; however, on binding to DNA the luminescence is switched on.125–127 Such an effect is also seen with [Ru(bpy)2 (dpqa)]2+ (81); however, the quantum yield is significantly higher than that of [Ru(bpy)2 (dppz)]2+ .128 Other ligand types have also been used to enhance the binding of Ru(II) complexes to DNA. For instance, the inclusion of the fluorescent 4-amino-naphthalimide (NH2 nap) structure into the Ru(II) complex [Ru(bpy)2 (NH2 nap)]2+ (82; Figure 9) by Gunnlaugsson and coworkers

yielded a DNA-binding complex that also cleaved plasmid DNA on irradiation for 5 min.129 While there have been many attempts made to enhance the binding of metal complexes to DNA, brought about through elegant synthetic modifications, one of the most interesting areas of research involves targeting specific sequences of DNA. This involves the linking of a luminescent Ru(II) complex to an oligonucleotide that contains a sequence complementary to the target region, thus giving sequence-specific luminescent probes.130–135

11

Cellular imaging has become a topic of intensive research in recent years, and luminescent metal-based probes have been intensively utilized for such a process. The d 6 transition metal ions of ruthenium(II), rhenium(I), and iridium(III) have been particularly effective as luminescent intracellular imaging agents owing to their kinetic inertness (low rates of ligand exchange), attractive photophysical properties (large

2+

N N

2+

N

N

N

N

N

N

N

N

N

N 80

79 [Ru(bpy)2(HAT)]2+

N

N

N

N

N Ru

Ru N

N

N N

N

Ru N

2+

N

N N

CELLULAR IMAGING USING TRANSITION-METAL-BASED SYSTEMS

[Ru(bpy)2(dppz)]2+

H N O 81

[Ru(bpy)2(dpqa)]2+

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

21

2+

N N

N Ru

N

N N

O N 82 O NH2

Figure 9 Schematic view and crystal structure of 82. Counter anions, solvent molecules, and hydrogen atoms omitted for clarity. Carbon, gray; nitrogen, blue; oxygen, red; and Ru(II), scarlet.

Stokes shifts, long luminescence lifetimes, and enhanced photostability), and their ability to coordinate a diverse range of ligands (this allows for easy modulation/tuning of emission properties). The application of d 6 metal ions in cellular imaging has recently been reviewed by Coogan and coworkers and the reader is directed to this for a more in-depth discussion.136 The derivatization and subsequent bioconjugation of metallolumophores in order to allow for cell uptake and localization is an important area of research and typically involves the attachment of a lumophore to a biomolecule such as biotin (vitamin H), estradiol, or other small biomolecules, for instance, oligonucleotides, peptides, or proteins. Lo and coworkers have utilized a variety of bis- and trisbiotinylated, cyclometalated Ir(III) complexes (83–86) to great effect for cellular imaging.137, 138 These complexes show an increase in luminescence on binding to avidin

(a protein that biotin has a strong affinity for) and also localize in cytoplasm when incubated with HeLa cells. Bioconjugated rhenium(I) complexes have also been used in fluorescent imaging studies, such as complex 87 in which the Re(I) lumophore is linked to a small peptide. When 87 is incubated with human leukocytes at low temperature, it is seen to localize at the periphery of the cells and internalize on warming to room temperature (Figure 10).139 The modulation of the photophysical properties of Ru(II) polypyridyl complexes through interaction with the various biologically relevant molecules mentioned in the previous sections (pH, O2 , anions, cations, and DNA) has culminated in their use at the forefront of transition-metalbased intracellular imagine agents. Although there are many good examples that utilize the Ru(II) lumophores, many of which were reviewed by Coogan and coworkers, only a few select examples are discussed in order O

Biotin

Biotin

N

H N

N H H N

Biotin

N Ir N

N H

Biotin

N

H N

N H H N

Ir N N

84

O

O

H N

N H

N

N H

N

N H

83

N

N

O

N H

Biotin

N N

Ir

H N

N H

Biotin

Ir N

O

H N

N O

N H 85

H N

N Biotin

H N

N O

86

N H

Biotin

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

H N

Biotin

22

Supramolecular devices + O H

O

H N

N H

N H

O

O

H N

N H

O

OH O

N 87

N

N Re

OC

CO CO

(a)

(b)

(c)

Figure 10 Schematic of 87 and fluorescence microscopy images of human leukocytes incubated with (a) 1 nM 87, (b) 50 nM fluorescein-labeled fNLFNTK, and (c) 50 nM 87. (Reproduced from Ref. 139.  American Chemical Society, 2004.)

NH2 N N

S N +

NH2

N H

N Ru

N H

N

N

Nucleolus

N

Nucleus

88 (a)

(b)

Figure 11 Schematic view of 88 and (a) optical and (b) fluorescence microscope images of XX uptake into mammalian cells. (Reproduced from Ref. 140.  Royal Society of Chemistry, 2009.)

to give an idea of the current state of research in this area. Turro and coworkers incorporated, through a thiourea linkage, an intercalating phenanthridine moiety into a [RuII (bpy)2 (phen)]2+ lumophore for use as an RNA-binding complex (88). Binding to RNA increases the luminescence intensity ninefold as a result of the protection of the phenanthridine group. Intracellular RNA affinity was tested by incubation with mammalian breast cancer cells, and localization of the complex was observed at regions known to be rich in RNA (cytoplasm and in the nucleolus; Figure 11).140

Bioconjugation of Ru(II) complexes has also led to some novel intracellular imaging agents. Notably, the work of Lo and coworkers in which Ru(II) polypyridyl complexes were bioconjugated with estradiol groups.141 Compound 89 was used in live cell imaging studies and was observed to be readily internalized by HeLa cells with retention of emission and distribution within the cytoplasm. While this very brief overview of transition-metal-based sensing systems barely scratches the surface of this ever expanding field of research, it demonstrates how transitionmetal-based lumophores can be applied to many areas of sensing for biological and environmental targets. The

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

23

2+

N N

N Ru

O N H

H N

N

N N

HO 89

H H

H

HO

use of elegant supramolecular chemistry, in the form of host–guest interactions, has been paramount in this field along with the careful design of lumophores through coordination and organometallic chemistry. The use of transitionmetal-based sensing and imaging systems continues to expand with new, and often more effective, lumophores and receptors being developed every year.

12

SENSING USING LANTHANIDE LUMINESCENT CHEMOSENSORS

The previous section outlined some of the developments made within the field of luminescent sensing using dmetal ion coordination complexes. From that discussion, it is clear that metal-based complexes possess relatively long lifetimes, when compared to organic fluorophores (which emit within few nanoseconds). Moreover, these emit at long wavelengths, which, as discussed previously, are highly advantageous when designing sensors for biological applications or with the view of developing sensors that require inexpensive excitation and detection (detector) electronics. Another family of metal ions, the lanthanides, have also emerged as potential emitters for sensing applications.142, 143 Many of the lanthanides emit with linelike emission bands, at long wavelength, either in the visible or the NIR regions, which makes them ideal for biological applications. However, owing to forbidden transitions, their extinction coefficients are small and, consequently, it is often difficult to populate their excited states directly, unlike the examples discussed in the aforementioned sections within this chapter.144 Nevertheless, the population of the lanthanide excited states can normally be successfully achieved by using indirect population (sensitization) by a chromophore possessing the appropriate excited-state energy, that is, the so-called antenna effect.145 Because of this, the excited states of the lanthanides are

populated first by excitation of the singlet states of the antenna, followed by energy transfer to the lanthanides, which normally involves intersystem crossing to the triplet state of the antenna. Because of this, the lanthanides possess long, excited-state lifetimes, which are usually in the microsecond and millisecond ranges.146 There are two main methods in which lanthanide ions can be employed in sensing. Either the antenna can be structurally modified to allow for the incorporation of a receptor moiety of a similar nature to those discussed above, or the analyte can bind directly to the lanthanide ion. On both occasions, the photophysical properties of the lanthanides are modulated and, hence, a direct correlation can be established between the concentration of the analyte and the photophysical output of the lanthanides.147 One of the first examples of lanthanide luminescence sensing was developed by de Silva et al.148 This system was based on the use of a PET mechanism, which was pH sensitive and hence, on quenching of the terpy antenna, which also functioned as a polypyridyl coordinating ligand for 90, no Eu(III)-centered emission was observed. However, the Eu(III) emission, appearing as linelike emission bands, and occurring in the red part of the spectrum, owing to the deactivation of the excited-state 5 D0 to the ground state, 7 FJ (J = 0–6) was switched on on protonation of the two tertiary amines in 90, as this blocked the aforementioned PET quenching, and the sensitization from the terpy antenna was possible. Shortly after 90 was reported, compound 91, a charge neutral Eu(III) complex, based on the cyclen macrocycle, was developed by Parker et al. as a pH sensor.149 This structure has since been slightly modified as a pH sensor for incorporation into hydrogels150 by using 92, as well as the Tb(III) analog. The two Eu(III) and Tb(III) complexes were employed either separately151 or simultaneously152 in solution as molecular logic gate analogs, and for use in mimicking parallel processes, where it has been shown that the lanthanide emission arising from the Eu(III) complex is highly sensitive to pH, while the Tb(III)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

24

Supramolecular devices

N

N 4H2O

Eu O

O

O

O

O P

250

Intensity

200

O

Inputs

100 50

550 600 Wavelength [nm]

650

N

N (CH3)N

NCH3

92

Acidic media Basic media From acidic to basic media

500

N O

Eu

O

91

N

N O

O P

150

0 450

N O

N

N

H N

O

N Eu

O

90

(CH3)N

H N

N

N N

3+

O P O

N

Outputs

OH−

O2

0

0

1

1

1

0

1

1

0

1

1

0

1

0

0

1

1

0

0

0

545 nm 622 nm 700 nm

700

Figure 12 The changes in the Eu(III) and Tb(III) emission of complex 92 and the Tb(III) analog as a function of pH and O2 in water. The corresponding truth table formed from the changes in the two emissions. The UV–vis and the fluorescence emission spectra of both were effected similarly in the presence of these two “inputs,” and hence only the lanthanide emission was effected differently as a function of these two inputs. (Reproduced from Ref. 152.  Royal Society of Chemistry, 2009.)

analog is sensitive to pH and O2 (Figure 12). For these two examples, only the lanthanide emission was shown to be differently affected by pH and O2 (referred to in logic gate operations as “inputs,” while the emission would be the “output”). Many other examples of cyclen-based sensors have recently emerged for both group I and II cations, as well as transition metal cations. For example, 93,153 a Tb(III)-based sensor emitting in the green part of the spectrum (due to deactivation of 5 D4 to 7 FJ (J = 6, 5, 4, 3), with emission occurring at 490, 545, 586, and 622 nm, respectively), was developed with the view of sensing Na+ selectively in the presence of K+ under physiological pH conditions, but within the physiological pH range the emission was “switched off” and on binding of K+ to the diazacrown ether moiety resulted in large enhancements in the emission. The Eu(III) analog was also developed but did not result in the same changes; this was owing to wrong energy match between the Eu(III) excited state and the antennae. Similarly 94,154 a newer version of the same idea, was developed where K+ was selectively detected over Na+ , using an Eu(III)-based sensor. The antenna on this occasion was able to successfully populate the Eu(III) excited state, a process that was effected on binding to the analyte with concomitant changes in the Eu(III) emission. Sensor 95 is another Tb(III) complex, which was developed for binding

to transition metal ions such as Cu(II) and Hg(II) through a single iminodiacetate moiety.155 The Tb(III) emission was not significantly affected by pH, because it was switched on. But on binding to either Cu(II) or Hg(II), the emission was quenched. Sensor 96 is also based on the use of iminodiacetate moieties, but on this occasion it was used to bind Ca(II) selectivity, and as such employed for imaging of microcracks in bone structures, where free Ca(II) binding sites occur on cracking or damage to the bone matrix.156 This was analyzed by using bone samples that were polished and then scratched, followed by analysis using both steady-state emission studies as well as confocal fluorescence scanning microscopy. Sensor 97 is an elegant example of a picolyl-based Zn(II) targeting sensor developed by Pope et al.157 where the emission from the Eu(III) was significantly modulated on binding of Zn(II); prior to binding, the complex was highly emissive and fully coordinated by the cyclen structure as well as by one of the pyridyl moieties. But on binding to Zn(II), these pyridyl moieties take part in coordination to Zn(II), opening up a vacant site at the Eu(III) ion, which then coordinates a water and hence the Eu(III) emission is quenched. Sensor 98 was also developed for Zn(II) detection by Nagano et al.158 using a diethylenetriaminepentaacetic acid (DTPA)-based

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

25

O

N H N

O

N

N

O

N

O O

N

O

O

N

N

N

N

N

O

N

O

Tb

O O

O

N

N O

3+

O

O

Eu

O

N

N

O O

N

O

93

O

94

NaO2C

N H N

O N

N N

NaO2C

H N

O N

CO2Et

N

3+

N

CO2Et

N

O

Tb

O

3+

O

N

N N

N

O

NaO2C

N

N O

NaO2C

95

O

Eu

N

CO2Na

96

CO2Na

O

O O

O

N

N Eu

N

N

N

O

−I

O

N

N O

N

O O 97

O− O

N

O

−

O

N O−

Eu N N −O

O

N N H

O

chelate conjugated to a quinoline antenna. This sensor was shown to bind Zn(II) through the tetrakis(2pyridylmethyl)ethylenediamine (TPEN) part of the structure, which also functions as the antenna. Compound 99 is a recent example of the sensing of other d-metal ions by the Pope group.159 This system employed a long wavelength antenna and hence was able to excite the lanthanide excited state within the visible region. To achieve this, the authors employed NIR emitting lanthanides such as Nd(III) and Yb(III), and were able to sense ions such as Cu(II), Cd(II), and Hg(II). The use of NIR emitting probes is ideal for use in biological systems as these systems are transparent within the NIR region so the only luminescent signal arising from a sample containing such NIR emission would be that of the sensor. This was the idea behind sensor 100, which was developed for binding to DNA.160 This system, based on a Ru(II) polypyridyl complex, conjugated to either Yb(III) or Nd(III) cyclen complexes, bound to DNA through

N O N

98

N

N O

Ln N

N

N O

N O

N

H N

O 99

O

O

electrostatic and intercalation modes, which effected the MLCT emission arising from the Ru(II) complex. As the Ru(II) complex also functioned as an antenna for the lanthanides, the emission from these ions was also modulated, and hence it was possible to obtain binding constants from both the Ru(II)-based MLCT emission and, in the case of Nd(III), from the lanthanide emission. In contrast, the emission from Yb(III) did not change to the same extent as was seen for Nd(III) and, hence, this latter example can possibly be employed as a dual sensor/imaging agent where the binding information can be obtained from the MLCT emission and the location by observing the NIR emission. The last examples discussed in this chapter are examples of the binding of anions to cyclen-based lanthanide complexes, where the binding occurs directly onto the metal center. This gives rise to significant changes, particularly for Eu(III) complexes, in the hypersensitive J = 2 transition, which is sensitive to the local environment of the metal ion.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

26

Supramolecular devices N N

N O

3+

N N

N

N

N

N

O

Ln

O

5+

H N

O

O

Ru N

N

N

O

O

H N

N

N

N H

N H

O Eu

O

N N Ln1 N N

O

O

H O

O

N

O O

105

Ln1, Ln2, or Ln3 = Eu(III) or Tb(III)

+3 N O

Au

N Eu O

N

N

a: Ln = Tb(III) b: Ln = Eu(III)

SO3H

+3

O

N

N N

NCH3

N

N

Ln N

O

Ln2

104

O

N

N

O

N

O O

103

O H N

O

O N

+3

N

O

O O

O

O

N

N

N

O

N

+9

Ln3

N

H N

Eu

102

O

O

N

N

O

O

O

O

O

+6 O

O

N

N

101

N

N

N N Tb N N

N

N

100

N

OH

Tb

O

N

Tb

O

H N

N

N

N

N

O

O CF3

O N

N

N

Eu N

N

N O

N

106 107

One of the first examples of these types of sensors was 101, developed by Parker et al.161 Here, the complex consisted of seven coordination sites and, hence, the complex was coordinatively unsaturated; and, consequently, possessed two metal-bound water molecules that could be replaced by molecules with a higher affinity for the metal ion center than water. Moreover, this system possesses three chiral antennae and, hence, the binding of anions could also be monitored by using chiral emission or circular polarized emission (CPL). Compound 102 is an extension of this principle, where two cyclen complexes were conjugated together and used to sense biscarboxylates such as diaryl anions, but, unlike 101, the antennae were missing from this structure and,

R

N

N

SO3H

hence, excitation of the aryl anion resulted in sensitized emission from this ternary complex.162 The sensing of such biscarboxylates can also be achieved by using a singe cyclen complex, such as 103, on this occasion, coordinated to a single crown ether.163 The use of two crown ethers such as those in 104 has also been employed and here a radiometric sensing of biscarboxylates was possible as a mixed Eu(III) and Tb(III) complex could be formed from this structure.164 Simple cyclen complexes have also been developed, which lack the anteanne shown in the aforementioned examples. An example of this design is 105a, a Tb(III) complex that was shown to be able to form ternary complexes with salicylic acid, the active ingredient of aspirin.165

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing The Eu(III) complex of this system, 105b, was also formed166 and used for forming ternary complexes with β diketonates, which could then be removed by appropriate ligands such as anions. This resulted in concomitant changes in the emission of these Ln(III) ions; in the former, the emission was “switched on” due to the formation of a ternary complex possessing an antenna, that is, the analyte, while for the latter, the emission was “switched off” as the sensitizing antenna was removed from the terniary complex by a nonaromatic anion. An extension to this design is the Eu(III) complex 106, which was designed to allow for incorporation of the complex onto a gold surface: either gold nanoparticles167 or onto a flat gold surface.168 As in 105, a ternary complex could then be formed between the β-diketonate anteana and the complexes and these were then broken up by using displacement antenna. Hence, 105 and 106 are new types of sensors, which can be best described as enabling the sensing of anions in the form of displacement assays. Recently, the same design principle was used for the sensing of cations such as Fe(II) in the work of Kotova et al.,169 who made 107, a ternary complex between a cyclen-derived Eu(III) complex and the water-soluble antenna 4,7-diphenyl-1,10phenanthroline-disulfonate, which can form complexes with ions such as Fe(II), Zn(II), and Cu(II), where the antenna is displaced from the Eu(III) complex with concomitant quenching of the Eu(III) emission due to lack of population of the Eu(III) excited state by the antenna. The above account has given some insight into the use of lanthanide complexes for sensing. This is a very active area of research and is fast growing, as can be clearly seen from the few examples listed, which have, with a few exceptions, been based on the cyclen structure. As for many of the transition metal ion examples shown above, these have also been used for imaging proposes, particularly within cells, an area recently reviewed by one of the leaders in the field.170

27

ACKNOWLEDGMENTS The authors would like to thank the various PhD students that have received their training in luminescence sensing chemistry under the supervision of Prof. Gunnlaugsson. Their work has been in part highlighted in this chapter, and the works presented in the PhD thesis of many of these students has been used in part in the preparation of this chapter. Dr Kitchen would like to thank IRCSET for the awarding of an empower postdoctoral fellowship. Prof. Gunnlaugsson would like to thank IRCSET (Postdoc and PhD studentships), SFI (RFP 2006, 2008, and 2009 and PI2010 awards), and Enterprise Ireland and HEA PRTLI for financial support.

REFERENCES 1. A. P. de Silva, H. Q. N. Gunaratne, T. Gunnlaugsson, et al., Chem. Rev., 1997, 97, 1515. 2. J. Mater. Chem., 2005, 15, 2617–2976. 3. J. R. Lakowicz, Principles of Fluorescence Spectroscopy, 2nd edn, Plenum Publishing, New York, 1999. 4. J. F. Callan, A. P. de Silva, and D. C. Magri, Tetrahedron, 2005, 61, 8551. 5. A. P. de Silva, T. S. Moody, and G. D. Wright, Analyst, 2009, 134, 2385. 6. (a) E. V. Anslyn, J. Org. Chem., 2007, 72, 687; (b) S. L. Wiskur, H. Ati-Haddou, J. J. Lavigne, and E. V. Anslyn, Acc. Chem. Res., 2001, 34, 693. 7. (a) R. A. Bissell, A. P. de Silva, H. Q. N. Gunaratne, et al., Chem. Soc. Rev., 1992, 21, 187; (b) A. W. Czarnik, Acc. Chem. Res., 1994, 27, 302. 8. A. P. de Silva, B. McCaughan, B. O. F. McKinney, and M. Querol, Dalton Trans., 2003, 1902. 9. M. Cajlakovic, A. Bizzarri, C. Konrad, and H. Voraberger, Encyclopedia of Sensors, 2006, 7, 291. 10. K. Hanaoka, Chem. Pharm. Bull., 2010, 58, 1283.

13

CONCLUSION

This chapter has dealt with some of the many examples of supramolecular luminescent sensors. We have attempted to give a short, yet comprehensive, overview of the field with much of our discussion focusing on new samples. Owing to limited space, we can only give a snapshot of this very active area of research and, consequently, apologize to the many thousands of researchers that we were unable to include. The area of luminescent sensing and hence luminescent imaging is fast growing and central to supramolecular chemistry and we look forward to witnessing the growth in this field in the years to come.

11. A. Roda, M. Guardigli, R. Ziessel, et al., Microchem. J., 2007, 85, 5. 12. S. Faulkner, S. J. A. Pope, and B. P. Burton-Pye, Appl. Spectrosc. Rev., 2005, 40, 1. 13. G. Accorsi, A. Listorti, K. Yoosaf, and N. Armaroli, Chem. Soc. Rev., 2009, 38, 1690. 14. S. Shinoda and H. Tsukube, Analyst, 2011, 136, 431. 15. K. K.-W. Lo, S. P.-Y. Li, and K. Y. Zhang, New J. Chem., 2011, 35, 265. 16. S.-K. Kim and J. L. Sessler, Chem. Soc. Rev., 2010, 39, 3784. 17. R. A. Bissel, A. P. de Silva, H. Q. N. Gunaratne, et al., Top. Curr. Chem., 1993, 168, 223. 18. J. W. Steed and J. L. Atwood, Supramolecular Chemistry, John Wiley & Sons, Ltd, Chichester, 2000.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

28

Supramolecular devices

19. P. A. Gale and R. Quesada, Coord. Chem. Rev., 2006, 250, 3219. 20. J.-M. Lehn, Supramolecular Chemistry, VCH, Weinheim, 1995. 21. A. P. de Silva and A. D. D. Rupasinghe, J. Chem. Soc., Chem. Commun., 1985, 1669. 22. A. P. de Silva, H. Q. N. Gunaratne, T. Gunnlaugsson, and M. Nieuwenhuizen, Chem. Commun., 1996, 1967. 23. (a) T. Gunnlaugsson, M. Nieuwenhuyzen, L. Richard, and V. Thoss, J. Chem. Soc., Perkin Trans. 2, 2002, 141; (b) T. Gunnlaugsson, B. Bichell, and C. Nolan, Tetrahedron Lett., 2002, 43, 4989.

44. M. Tian and H. Ihmels, Chem. Commun., 2009, 3175. 45. Y. Zhou, X.-Y. You, Y. Fang, et al., Org. Biomol. Chem., 2010, 8, 4819. 46. T. Gunnlaugsson, T. C. Lee, and R. Parkesh, Tetrahedron, 2004, 60, 11239. 47. L. Fernandes, M. Boucher, J. Fernandez-Lodeiro, et al., Inorg. Chem. Commun., 2009, 12, 905. 48. X. Wang, W. Zheng, H. Lin, et al., Tetrahedron Lett., 2009, 50, 1536. 49. A. Basoglu, S. Parlayan, M. Ocak, et al., J. Fluoresc., 2009, 19, 655.

24. H. He, M. A. Mortellaro, M. J. P. Leiner, et al., J. Am. Chem. Soc., 2003, 125, 1468.

50. L. Fabbrizzi, M. Licchelli, G. Rabaioli, and A. Taglietti, Coord. Chem. Rev., 2000, 205, 85.

25. A. Minta, J. P. Y. Kao, and R. Y. Tsien, J. Biol. Chem., 1989, 264, 8171.

51. (a) R. Nishiyabu, Y. Kubo, T. D. James, and J. S. Fossey, Chem. Commun., 2011, 47, 1106; (b) S. Arimori, M. L. Bell, C. S. Oh, et al., Chem. Commun., 2001, 1836.

26. F. A. Khan, K. Parasurman, and K. K. Sadhu, Chem. Commun., 2009, 2399. 27. S. Rochat, Z. Grote, and K. Severin, Org. Biomol. Chem., 2009, 7, 1147. 28. Y. Ando, Y. Hiruta, D. Citterio, and K. Suzuki, Analyst, 2009, 134, 2314. 29. T. Gunnlaugsson, B. Bichell, and C. Nolan, Tetrahedron, 2004, 60, 5799. 30. A. Hamdi, S. H. Kim, R. Abidi, et al., Tetrahedron, 2009, 65, 2818. 31. P. Ashokkumar, V. T. Ramakrishnan, and P. Ramamurthy, Eur. J. Org. Chem., 2009, 5941. 32. (a) L. Fabbrizzi, M. Lichelli, P. Pallavicini, et al., Angew. Chem. Int. Ed. Engl., 1994, 33, 1975; (b) L. Fabbrizzi, M. Lichelli, P. Pallavicini, et al., Chem Eur. J., 1996, 2, 167. 33. L. Fabbrizzi, M. Lichelli, P. Pallavicini, and A. Taglietti, Inorg. Chem., 1996, 35, 1733. 34. (a) S. Kumar, P. Kaur, and S. Kaur, Tetrahedron Lett., 2002, 43, 1097; (b) S. Kaur and S. Kumar, Chem. Commun., 2002, 2840. 35. Y. Zheng, X. Cao, J. Orbulescu, et al., Anal. Chem., 2003, 75, 1706.

52. A. P. de Silva, H. Q. N. Gunaratne, T. Gunnlaugsson, et al., Appl. Chem., 1996, 68, 1443. 53. C. A. Hunter and R. J. Shannon, Chem. Commun., 1996, 1361. 54. M. D. Phillips, T. M. Fyles, N. P. Barwell, T. J. James, Chem. Commun., 2009, 6557.

and

55. M. E. Langmuir, J.-R. Yang, A. M. Moussa, et al., Tetrahedron Lett., 1995, 36, 3989. 56. A. P. de Silva, H. Q. N. Gunaratne, and T. Gunnlaugsson, Tetrahedron Lett., 1998, 39, 5077. 57. Y. Hong, J. W. Y. Lam, and B. Z. Tang, Chem. Commun., 2009, 4332. 58. B. Wang, J. Tan, and L. Zhu, Colloids Surf. B Biointerfaces, 2010, 79, 1. 59. (a) J. G. Vos and J. M. Kelly, J. Chem. Soc., Dalton Trans., 2006, 4869; (b) K. E. Erkkila, D. T. Odom, and J. K. Barton, Chem. Rev., 1999, 99, 2777; (c) B. Armitage, Chem. Rev., 1998, 98, 1171. 60. L. D. Van Vliet, T. Ellis, P. J. Foley, et al., J. Med. Chem., 2007, 50, 2326.

36. J. Kang, M. Choi, J. Y. Kwon, et al., J. Org. Chem., 2002, 67, 4384.

61. (a) E. B. Veale, D. O. Frimannsson, M. Lawler, and T. Gunnlaugsson, Org. Lett., 2009, 11, 4040; (b) E. B. Veale and T. Gunnlaugsson, J. Org. Chem., 2010, 75, 5513.

37. E. Tamanini, K. Flavin, M. Motevalli, et al., Inorg. Chem., 2010, 49, 3789.

62. T. Phillips, I. Haq, A. J. H. M. Meijer, et al., Biochemistry, 2004, 43, 13657.

38. Z. C. Xu, X. H. Qian, and J. N. Cui, Org. Lett., 2005, 7, 3029.

63. R. B. P. Elmes, M. Erby, S. M. Cloonan, et al., Chem. Commun., 2011, 47, 686.

39. R. Parkesh, T. C. Lee, and T. Gunnlaugsson, Org. Biomol. Chem., 2007, 5, 310. 40. R. Parkesh, T. C. Lee, and T. Gunnlaugsson, Tetrahedron Lett., 2009, 50, 4114.

64. (a) A. T. Byrne, A. E. O’Connor, M. Hall, et al., Br. J. Cancer, 2009, 101, 1565; (b) W. M. Gallagher, L. T. Allen, C. O’Shea, et al., Br. J. Cancer, 2005, 92, 1702.

41. B. A. Wong, S. Friedle, and S. J. Lippard, Inorg. Chem., 2009, 48, 7009.

65. S. O. McDonnell and D. F. O’Shea, Org. Lett., 2006, 8, 3493.

42. B. A. Wong, S. Friedle, and S. J. Lippard, J. Am. Chem. Soc., 2009, 131, 7142.

66. A. Palma, M. Tasior, D. O. Frimannsson, et al., Org. Lett., 2009, 11, 3638.

43. S. Cao, H. Li, T. Chen, and J. Chen, J. Solution Chem., 2009, 38, 1520.

67. S. S. Gayathri, M. Wielopolski, E. M. P´erez, et al., Angew. Chem. Int. Ed., 2009, 48, 815.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

29

68. J. Koll´ar, P. Hrdloviˇc, and S. Chmela, J. Photochem. Photobiol. A: Chem., 2008, 195, 64.

85. T. Gunnlaugsson, P. E. Kruger, P. Jensen, et al., Tetrahedron Lett., 2003, 44, 8909.

69. G. Loving and B. Imperiali, J. Am. Chem. Soc., 2008, 130, 1363.

86. S. Camiolo, P. Gale, M. B. Hursthouse, and M. E. Light, Org. Biomol. Chem., 2003, 1, 741.

70. K. Kudo, A. Momotake, Y. Kanna, et al., Chem. Commun., 2011, 47, 3867.

87. A. N. Swinburne, M. J. Paterson, A. Beeby, J. W. Steed, Org. Biomol. Chem., 2010, 8, 1010.

71. P. A. Gale and T. Gunnlaugsson, Chem. Soc. Rev., 2010, 39, 3581–4008.

88. S. V. Bhosale, M. B. Kalyankar, and S. J. Langford, Org. Lett., 2009, 11, 5418.

72. (a) Coord. Chem. Rev., 2006, 250, 240; (b) Topp. Curr. Chem., 2005, 255.

89. J. R. Hiscock, C. Caltagirone, M. E. Light, et al., Org. Biomol. Chem., 2009, 7, 1781.

73. (a) M. R. Duke, E. B. Veale, F. M. Pfeffer, et al., Chem. Soc. Rev., 2010, 39, 3936; (b) P. A. Gale, Chem. Soc. Rev., 2010, 39, 3746; (c) T. Gunnlaugsson, M. Glynn, G. M. Tocci (n´ee Hussey), et al., Coord. Chem. Rev., 2006, 250, 3094; (d) E. Galbraith and T. D. James, Chem. Soc. Rev., 2010, 39, 3831; (e) V. Amendola, L. Fabbrizzi, and L. Mosca, Chem. Soc. Rev., 2010, 39, 3889; (f) M. Cametti and K. Rissanen, Chem. Commun., 2009, 2809; (g) P. A. Gale, S. E. Garcia-Garrido, and J. Garric, Chem. Soc. Rev., 2008, 37, 151; (h) P. Padros and R. Quesada, Supramol. Chem., 2008, 20, 201; (i) J. W. Steed, Chem. Soc. Rev., 2010, 39, 3686; (j) A.F. Li, J.-H. Wang, F. Wang, and Y.-B. Jiang, Chem. Soc. Rev., 2010, 39, 3729; (k) C. R. Bondy and S. J. Loeb, Coord. Chem. Rev., 2003, 240, 77; (l) V. Amendola, D. Esteban-Gomez, L. Fabbrizzi, and M. Licchelli, Acc. Chem. Res., 2006, 39, 343; (m) B. T. Nguyen and E. V. Anslyn, Coord. Chem. Rev., 2006, 250, 3118.

90. K. Ghosh and A. R. Sarkar, Tetrahedron Lett., 2009, 50, 85.

74. J. L. Sessler, P. A. Gale, and W. S. Cho, Anion Receptor Chemistry, Royal Society of Chemistry, Cambridge, UK, 2006. 75. (a) R. Martinez-Manez and F. Sancenon, Coord. Chem. Rev., 2005, 250, 3118; (b) R. Martinez- Manez and F. Sancenon, J. Fluoresc., 2005, 15, 267; (c) C. Suksai and T. Tuntulani, Topp. Curr. Chem., 2005, 255, 163. 76. (a) C. L. Henderson, J. Li, R. L. Nation, et al., Chem. Commun., 2010, 46, 3197; (b) P. A. Gale, O. A. Okunola, P. Prados, et al., Nature Chem., 2009, 1, 138. 77. E. Galbraith and T. D. James, Chem. Soc. Rev., 2010, 39, 3831.

and

91. Y. Sun, C. Zhong, R. Gong, et al., J. Org. Chem., 2009, 74, 7943. 92. V. Balzani and S. Campagna, eds., Photochemistry and Photophysics of Coordinations Compounds I & II, Topics in Current Chemistry, 280 & 281, Springer, New York, 2007. 93. R. Grigg, J. M. Holmes, S. K. Jones, and W. D. J. Amilaprasadh Norbert, J. Chem. Soc., Chem. Commun., 1994, 185. 94. J. M. Price, W. Xu, J. N. Demas, and B. A. DeGraff, Anal. Chem., 1998, 70, 265. 95. M. Cattaneo, F. Fagalde, and N. E. Katz, Inorg. Chem., 2006, 45, 6884. 96. P. D. Beer, O. Kocian, R. J. Mortimer, and C. Ridgway, J. Chem. Soc., Chem. Commun., 1991, 1460. 97. D. I. Yooh, C. A. Berg-Brennan, H. Lu, and J. T. Hupp, Inorg. Chem., 1992, 31, 3192. 98. L. J. Charbonni`ere, R. F. Ziessel, C. A. Sams, A. Harriman, Inorg. Chem., 2003, 42, 3466.

and

99. D. B. MacQueen and K. S. Schanze, J. Am. Chem. Soc., 1991, 113, 6108. 100. H. D¨urr, R. Schwarz, I. Willner, et al., J. Chem. Soc., Chem. Commun., 1992, 1338. 101. F. Loiseau, R. Passalacqua, S. Campagna, et al., Photochem. Photobiol. Sci., 2002, 1, 982. 102. Y. Shen and B. P. Sullivan, Inorg. Chem., 1995, 34, 6235. 103. P. D. Beer, Chem. Commun., 1996, 689. 104. P. D. Beer, N. C. Fletcher, and T. Wear, Polyhedron, 1996, 15, 1339.

78. (a) P. A. Gale, J. R. Hiscock, S. J. Moore, et al., Chem. Asian, J., 2010, 5, 555; (b) R. M. Duke, J. E. O’Brien, T. McCabe, and T. Gunnlaugsson, Org. Biomol. Chem., 2008, 6, 4089; (c) G. W. Bates, P. A. Gale, and M. E. Light, Chem. Commun., 2007, 2121.

105. P. D. Beer, Acc. Chem. Res., 1998, 31, 71.

79. M. E. Huston, E. U. Akkaya, and A. W. Czarnik, J. Am. Chem. Soc., 1989, 111, 8735.

107. S.-S. Sun and A. J. Lees, Coord. Chem. Rev., 2002, 230, 171.

80. (a) E. B. Veale, M. G. Tocci, F. M. Pfeffer, et al., Org. Biomol. Chem., 2009, 7, 3447; (b) E. B. Veale and T. Gunnlaugsson, J. Org. Chem., 2008, 73, 8073.

108. P. D. Beer and E. J. Hayes, Coord. Chem. Rev., 2003, 240, 167.

81. T. Gunnlaugsson, P. E. Kruger, T. C. Lee, et al., Tetrahedron Lett., 2003, 44, 6575. 82. Y. Q. Gao and R. A. Marcus, J. Phys. Chem. A, 2002, 106, 1956.

106. P. D. Beer and J. Cadman, Coord. Chem. Rev., 2000, 205, 131.

109. F. Szemes, D. Hesek, Z. Chen, et al., Inorg. Chem., 1996, 35, 5868. 110. P. D. Beer and F. Szemes, J. Chem. Soc., Chem. Commun., 1995, 2245.

83. A. P. de Silva and T. E. Rice, Chem. Commun., 1996, 163.

111. P. D. Beer, V. Timoshenko, M. Maestri, et al., Chem. Commun., 1999, 1755.

84. R. M. Duke and T. Gunnlaugsson, Tetrahedron Lett., 2010, 51, 5402–5405.

112. P. D. Beer, A. R. Graydon, and L. R. Sutton, Polyhedron, 1996, 15, 2457.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

30

Supramolecular devices

113. P. D. Beer and J. Cadman, New J. Chem., 1999, 23, 347. 114. P. D. Beer, S. W. Dent, G. S. Hobbs, and T. J. Wear, Chem. Commun., 1997, 99. 115. D. Saha, S. Das, C. Bhaumik, et al., Inorg. Chem., 2010, 49, 2334. 116. P. de Wolf, P. Waywell, M. Hanson, et al., Chem. Eur. J., 2006, 12, 2188. 117. W. Goodall and J. A. G. Williams, J. Chem. Soc., Dalton Trans., 2000, 2893. 118. N. Hadjiliadis and E. Sletten, eds., Metal Complex - DNA Interactions, John Wiley & Sons, Ltd, Chichester, UK, 2009. 119. B. M. Zeglis, V. C. Pierre, and J. K. Barton, Chem. Commun., 2007, 4565. 120. J. K. Barton, A. T. Danishefsky, and J. Am. Chem. Soc., 1984, 106, 2172.

J. M. Goldberg,

121. C. V. Kumar, J. K. Barton, and N. J. Turro, J. Am. Chem. Soc., 1985, 107, 5518. 122. J. K. Barton, J. M. Goldberg, C. V. Kumar, and N. J. Turro, J. Am. Chem. Soc., 1986, 108, 2081.

140. N. A. O’Connor, N. Stevens, D. Samaroo, et al., Chem. Commun., 2009, 2640. 141. K. K.-W. Lo, T. K.-M. Lee, J. S.-Y. Lau, et al., Inorg. Chem., 2008, 47, 200. 142. (a) C. M. G. dos Santos, A. J. Harte, S. J. Quinn, and T. Gunnlaugsson, Coord. Chem. Rev., 2008, 252, 2512; (b) J. P. Leonard, C. B. Nolan, F. Stomeo, and T. Gunnlaugsson, Top. Curr. Chem., 2007, 281, 1. 143. (a) S. Shinoda and H. Tsukube, Analyst, 2011, 136, 431; (b) J.-C. G. B¨unzli, Chem. Rev., 2010, 110, 2729; (c) S. V. Eliseeva and J.-C. G. B¨unzli, Chem. Soc. Rev., 2010, 39, 189; (d) C. P. Montgomery, B. S. Murray, E. J. New, et al., Acc. Chem. Res., 2009, 42, 925; (e) A. Thibon and V. C. Pierre, Anal. Bioanal. Chem., 2009, 394, 107; (f) E. G. Moore, A. P. S. Samuel, and K. N. Raymond, Acc. Chem. Res., 2009, 42, 542; (g) S. Faulkner, L. S. Natrajan, W. S. Perry, and D. Sykes, Dalton Trans., 2009, 3890; (h) M. D. Ward, Coord. Chem. Rev., 2007, 251, 1663; (i) J.-C. G. B¨unzli, Acc. Chem. Res., 2006, 39, 53; (j) G. R. Motson, J. S. Fleming, and S. Brooker, Adv. Inorg. Chem., 2004, 55, 361.

123. A. B. Tossi and J. M. Kelly, Photochem. Photobiol., 1989, 49, 545.

144. (a) D. Parker and J. A. G. Williams, J. Chem. Soc. Perkin Trans. 2, 1995, 1305; (b) N. Sabbatini, M. Guardigli, and J.-M. Lehn, Coord. Chem. Rev., 1993, 123, 201.

124. F. de Buyl, A. Kirsch-De Mesmaeker, A. Tossi, and J. M. Kelly, J. Photochem. Photobiol. A, 1991, 60, 27.

145. T. Gunnlaugsson and J. P. Leonard, Chem. Commun., 2005, 3114.

125. J.-C. Chambron, J.-P. Sauvage, E. Amouyal, and P. Koffi, Nouv. J. Chim., 1985, 9, 527.

146. J.-C. G. Bunzli and G. R. Choppin, eds., Lanthanide Probes in Life, Chemical and Earth Sciences, Theory and Practice, Elsevier, New York, 1989.

126. Y. Jenkins, A. E. Friedman, N. J. Turro, and J. K. Barton, Biochemistry, 1992, 31, 10809. 127. R. M. Hartshorn and J. K. Barton, J. Am. Chem. Soc., 1992, 114, 5919. 128. K. A. O’Donoghue, J. M. Kelly, and P. E. Kruger, Dalton Trans., 2004, 13.

147. (a) D. Parker, R. S. Dickins, H. Puschmann, et al., Chem. Rev., 2002, 102, 1977; (b) H. Tsukube and S. Shinoda, Chem. Rev., 2002, 102, 2389; (c) H. Tsukube, S. Shinoda, and H. Tamiaki, Coord. Chem. Rev., 2002, 226, 227; (d) D. Parker, Coord. Chem. Rev., 2000, 205, 109.

129. G. J. Ryan, S. Quinn, and T. Gunnlaugsson, Inorg. Chem., 2008, 47, 401.

148. A. P. de Silva, H. Q. N. Gunaratne, and T. E. Rice, Angew. Chem. Int. Ed. Engl., 1996, 35, 2116.

130. W. Bannwarth, D. Schmidt, R. L. Stallard, et al., Helv. Chim. Acta, 1988, 71, 2085.

149. T. Gunnalugsson and D. Parker, Chem. Commun., 1998, 51.

131. J. Telser, A. Cruickshank, K. S. Schanze, and T. L. Netzel, J. Am. Chem. Soc., 1989, 111, 7221. 132. Y. Jenkins and J. K. Barton, J. Am. Chem. Soc., 1992, 114, 8736. 133. D. J. Hurley and Y. Tor, J. Am. Chem. Soc., 1998, 120, 2194. 134. D. Ossipov, P. I. Pradeepkumar, M. Holmer, and J. Chattopadhyaya, J. Am. Chem. Soc., 2001, 123, 3551. 135. G. N. Grimm, A. S. Boutorine, P. Lincoln, et al., ChemBioChem, 2002, 3, 324. 136. V. Fern´andez-Moreira, F. L. Thorp-Greenwood, and M. P. Coogan, Chem. Commun., 2010, 46, 186. 137. K. Y. Zhang and K. K.-W. Lo, Inorg. Chem., 2009, 48, 6011. 138. K. Y. Zhang, S. P.-Y. Li, N. Zhu, et al., Inorg. Chem., 2010, 49, 2530. 139. K. A. Stephenson, S. R. Banerjee, T. Besanger, et al., J. Am. Chem. Soc., 2004, 126, 8598.

150. T. Gunnlaugsson, C. P. McCoy, and F. Stomeo, Tetrahedron Lett., 2004, 45, 8403. 151. (a) T. Gunnlaugsson, D. A. Mac D´onaill, and D. Parker, Chem. Commun., 2000, 93; (b) T. Gunnlaugsson, D. A. Mac D´onaill, and D. Parker, J. Am. Chem. Soc., 2001, 123, 12866. 152. C. S. Bonnet and T. Gunnlaugsson, New J. Chem., 2009, 33, 1025. 153. T. Gunnlaugsson and J. P. Leonard, Chem. Commun., 2003, 2424. 154. E. A. Weitz and V. C. Pierre, Chem. Commun., 2011, 57, 541. 155. B. K. McMahon and T. Gunnlaugsson, Tetrahedron Lett., 2010, 51, 5406. 156. B. McMahon, P. Mauer, C. P. McCoy, et al., J. Am. Chem. Soc., 2009, 131, 17542. 157. S. J. A. Pope and R. H. Laye, Dalton Trans., 2006, 3108. 158. K. Hanaoka, K. Kikuchi, H. Kojima, et al., J. Am. Chem. Soc., 2004, 126, 12470.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Luminescent sensing

31

159. J. E. Jones, A. J. Amoroso, I. M. Dorin, et al., Chem. Commun., 2011, 47, 3374.

166. C. M. G. dos Santos, P. B. Fernandez, S. E. Plush, et al., Chem. Commun., 2007, 3389.

160. A. M. Nonat, S. J. Quinn, and T. Gunnlaugsson, Inorg. Chem., 2009, 48, 4646. 161. R. S. Dickins, T. Gunnlaugsson, D. Parker, and R. D. Peacock, Chem. Commun., 1998, 1643.

167. (a) J. Massue, S. E. Quinn, and T. Gunnlaugsson, J. Am. Chem. Soc., 2008, 130, 6900; (b) C. S. Bonnet, J. Massue, S. J. Quinn, and T. Gunnlaugsson, Org. Biomol. Chem., 2009, 7, 3074.

162. A. J. Harte, P. Jensen, S. E. Plush, et al., Inorg. Chem., 2006, 45, 9465.

168. N. S. Murray, S. P. Jarvis, and T. Gunnlaugsson, Chem. Commun., 2009, 4959.

163. S. E. Plush and T. Gunnlaugsson, Org. Lett., 2007, 9, 1919.

169. O. Kotova, S. Comby, and T. Gunnlaugsson, Chem. Commun., 2011, 47, in press.

164. S. E. Plush and T. Gunnlaugsson, Dalton Trans., 2008, 3801.

170. D. Parker, Aust. J. Chem., 2011, 64, 239.

165. T. Gunnlaugsson, A. J. Harte, J. P. Leonard, and M. Nieuwenhuyzen, Chem. Commun., 2002, 2134.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc099

Photoswitching Materials Chi-Chiu Ko1,2 and Vivian Wing-Wah Yam1,3 1

University of Hong Kong, Hong Kong, China City University of Hong Kong, Hong Kong, China 3 Institute of Molecular Functional Materials, University of Hong Kong, Hong Kong, China 2

1 Introduction 2 Photochromism of Selected Photochromic Families 3 Photochromic Ligands and Their Transition-Metal Complexes 4 Molecular Machines Photodriven by Photochromic Bridges 5 Photocontrollable Receptors 6 Conclusion Acknowledgments References

1

1 1 4 13 23 28 29 29

INTRODUCTION

Photoswitchable materials are mainly engineered from molecules that are capable of undergoing reversible changes in their intrinsic properties through the use of light as an external stimulus. As the absorption properties of most of these molecules change reversibly with photoexcitation, most of them are also photochromic materials. Many of these photoswitching processes operate through reversible chemical transformations, such as trans–cis isomerization, ionization, pericyclic ring-opening and ring-closing reaction, and intramolecular group transfer reaction.1, 2 Such changes have already led to a wide range of applications in consumer products such as toys, cosmetic products,

photochromic inks for security markings, ophthalmic and sunglass lenses, optical filters, optoelectronic devices, and optical memory. These photoswitching molecules also play important roles in the biological systems. For example, the trans–cis isomerization of rhodopsin is the crucial photoswitching process that is responsible for vision. Owing to these important applications, quite a number of families of photochromic molecules have been extensively studied and reported in the literature.1, 2 Through systematic investigations and subsequent rational designs on the photochromic frameworks using the state-of-the-art organic synthetic methodologies, the photochromic properties could be modified, tuned, and tailored for molecular switching. Since there are quite a number of extensive reviews on the properties of different families of photoswitchable and photochromic compounds,1, 2 this chapter focuses on selected recent works dealing with the functionalization of different families of photochromic molecules for various photoswitching functional properties.

2 2.1

PHOTOCHROMISM OF SELECTED PHOTOCHROMIC FAMILIES Stilbene and azo compounds

The photoswitching properties associated with azobenzene and stilbene derivatives (Figure 1) are attributed to the reversible trans–cis photoisomerization of N=N and C=C double bonds, respectively.1 As a result of the steric constraints in the cis isomer, which reduce the π electron delocalization, the absorption and triplet-state energies for the trans-isomer are generally lower than those of the ciscounterpart, thus leading to the photochromic behavior.3

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

2

Supramolecular devices

X X

h n1

X X

h n1

h n2

h n2 or ∆

R

Photochromism of stilbene (X=CH) and azobenzene

Figure 1 (X=N).

Spirocarbon X O nN

h n1 h n 2, ∆

X

n N

O

Figure 2 Photochromism of spiropyran (X=CH) and spirooxazine (X=N).

2.2

Spiropyran and spirooxazine

Spiropyran and spirooxazine are structurally related photochromic families. The photochromism of these families is attributed to the reversible photoinduced ring-opening reaction from the cleavage of the spiro-C–O bond, which results in the planarization of the molecule and the extension of the π -conjugation in the colored photomerocyanine conformer that leads to a shift in the absorption to the lower energy visible region (Figure 2).1

2.3

Benzopyran and naphthopyran (chromene)

The derivatives of benzopyran and naphthopyran are generally referred to as chromene (Figure 3).2 They are structurally related to spiropyrans except that substituents at the 2-position of the benzopyran do not form a heterocyclic ring system. Although the photochromic reactivity of this family is also based on the reversible ring-opening and ring-closing of the pyran moiety as a result of C–O bond cleavage and formation, its photochemistry is considerably different.

2.4

Diarylethene

The photochromism of these diarylethenes has been ascribed to the reversible photocyclization and photocycloreversion (Figure 4). The discovery of photochromic diarylethenes has stemmed from the photocyclization reaction of cis-stilbene to give dihydrophenanthrene, which

h n1

O

h n 2, ∆

O

Figure 3

Photochromism of a benzopyran.

S

Figure 4

S

R

R

S

S

R

Photochromism of diarylethenes.

returns quickly to stilbene in the dark in deaerated solutions.1, 4 When the acyclic alkenes were replaced with a cyclic structure, such as the most commonly used cyclopentenes, perfluorocyclopentenes, maleic anhydrides, and maleimides, the photocyclization and photocycloreversion reactions were found to become the only two major photoinduced reactions as trans–cis isomerization was inhibited.5 Owing to the important contribution by Irie, the family of diarylethenes has received growing attention in recent years.5 They have been shown to exhibit excellent fatigue resistance and thermal stability in both photochromic isomers if heterocyclic aryl groups with relatively low aromatic stabilization energies were used in the design of the diarylethenes.6 As the geometrical structural changes are relatively small in the photochromic reactions, most of these reversible photoreactions could also proceed in the crystalline state. Thus, diarylethenes also exhibit crystalline state photochromism.5, 7 Single crystals of the cocrystals with two or three different diarylethene compounds in various ratios exhibit interesting wavelength-dependent photochromism.8 For example, upon exposure to UV light, the cocrystals of three diarylethenes (dioxazolylethene, dithiazolylethene, and dithienylethene; Figure 5a) become purplish black in color, as all three different diarylethenes undergo cyclization. With different excitation wavelengths, the black color could be partially bleached to give a yellow crystal with >620-nm excitation or a red crystal with 435-nm and >690-nm excitation, or a blue crystal with 435-nm excitation (Figure 5b).8 Not only does the crystalline photochromism of the diarylethenes change the absorption properties reversibly, but also the interconversion between the open form and the closed form results in the alteration of the molecular packing and intermolecular interactions. In the crystal of 1, 2-bis(5-methyl-2-phenyl-4-thiazolyl)perfluorocyclopentene with sizes on 10–100 µm scale, the size contracts and expands reversibly by the photochromic reaction in the crystalline state (Figure 6a).9 In the rod-like crystal of the same compound, reversible bending is observed by the photochromic reaction (Figure 6b). Such a crystal bending can also induce the movement of a 90 times heavier gold microparticle over a distance of 30 µm, resembling the hitting of a tennis ball.9

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials

FF

F F S

F F

UV

S

Vis

FF

F F N S

F F N

UV

S

Vis

FF

F F S

F F

3

F F S >620 nm

FF

F F

N

370 nm

>690 nm

N S

435 nm

S 435 nm

FF

F F N O

F F N

UV

O

Vis

F F

FF F F

N

N O

O

(a)

(b)

Figure 5 (a) Crystalline photochromism of dioxazolylethene, dithiazolylethene, and dithienylethene in three-component cocrystals. (b) Photographs of a three-component cocrystal upon UV and subsequent visible light irradiation at different wavelengths. (Reproduced from Ref. 8.  American Chemical Society, 2007.)

F F N

F F N

365 nm >500 nm

S

F F

F F

N

F F 100 µm

N S

S

10 µm

100 µm

100 µm

100 µm

50 Tip displacement (µm)

S

F F

10 µm

(a)

40 30 20 10 0

0

10

20

30 40 50 Cycle number

(b)

60

70

80

Figure 6 (a) Reversible contraction and expansion of a photochromic rectangular single crystal of 1,2-bis(5-methyl-2-phenyl-4thiazolyl)perfluorocyclopentene with a thickness of 330 nm. (b) Reversible bending of a crystalline rod of the same compound that is prepared by sublimation. (Reproduced from Ref. 9.  Nature Publishing Group, 2007.)

2.5

Triarylmethanes

Triarylmethanes, in particular, those substituted with nitriles, hydroxy, and alkoxy groups, are found to readily undergo photolytic dissociation when exposed to UV light. Since the three aryl groups are bonded through an sp3 carbon, no conjugation exists between the aryl moieties. On photoionization, the cationic carbon becomes sp2 hybridized and the triarylmethane cation becomes planar. As a result, the triarylmethane cations absorb in the visible region. Such changes in the absorption properties on photodissociation and the thermally induced recombination reactions resulted in the photochromic behavior (Figure 7).1, 2 Acridane, xanthone, and thioxanthone

skeletons are also derived from triphenylmethanes with two phenyl rings bridged by nitrogen, oxygen, and sulfur heteroatoms, respectively.

N hn

CN

+

∆

N

+

CN−

N

N

Figure 7

Photochromism of a triarylmethane leuconitrile.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Supramolecular devices

3

PHOTOCHROMIC LIGANDS AND THEIR TRANSITION-METAL COMPLEXES

3.1

Z

Luminescence switching and sensitized photochromism

Ru

N N

N 2+ N

N

N N

Figure 8 Schematic drawing of the ruthenium(II) complex with styryl-bridged bipyridine ligand.

N

X=N; Z=N (AZO) X=N; Z=CH (PHAZO) X=C; Z=CH (STPY) X=C; Z=C-NO2 (NSP)

N

N

Reversible trans–cis photoisomerization reactions of C=C and N=N were demonstrated to occur on both the lowest excited singlet state surface and the triplet state surface.10 Pioneering works on the sensitized photochromism of the azobenzene and stilbene derivatives by intermolecular energy transfer using organic triplet sensitizers were reported as early as the 1960s.10, 11 In the early 1970s, diffusion-controlled intermolecular sensitization of these photochromic compounds by the triplet state of transitionmetal complexes, such as [Ru(bpy)3 ]2+ and [Re(CO)3 (bpy)Cl], was demonstrated by the groups of Whitten12 and Wrighton.13 The intramolecular sensitization of 4stilbazole by the triplet metal-to-ligand charge transfer (3 MLCT) excited state in [Ru(bpy)2 (4-stilbazole)2 ]2+ was subsequently reported.14 Later on, the sensitized triplet-state photochromism of the styryl moiety of the styryl-bridged bipyridine ligand in the rutheium(II) complex (Figure 8) by an intramolecular energy transfer from the 3 MLCT excited state of ruthenium(II) bipyridyl complexes has also been demonstrated.15 On the basis of the difference in the triplet-state energy between the trans and cis isomers of the stilbene and azo derivatives,16 Yam et al. have designed a number of photoswitchable luminescent rhenium(I) tricarbonyl diimine complexes through the incorporation of the stilbene- and azo-containing pyridine ligands (Figure 9a).17 Through rational design, the lower lying 3 IL state of the trans isomer of some of these complexes could intramolecularly quench the typical emissive 3 MLCT state, providing an additional nonradiative deactivation pathway. On the contrary, with the cis isomer formed on irradiation, the intramolecular energy transfer processes (quenching) would be disfavored or even blocked owing to its higher triplet-state energy,

N

X X

NPENB

(a)

Relative emission intensity

4

400 (b)

500 600 Wavelength (nm)

700

Figure 9 (a) Schematic drawing of stilbene- and azocontaining pyridine ligands. (b) Emission spectral traces of [Re(CO)3 (phen)(NSP)]+ on irradiation at λ = 330 nm in degassed CH2 Cl2 . (Reproduced from Ref. 17.  Royal Society of Chemistry, 1995.)

and hence resulting in the restoration of the typical 3 MLCT emission intensity on photoinduced trans–cis isomerization (Figure 9b). Besides, these ligands in the complexes also undergo photochromic trans–cis isomerization with excitation into the MLCT absorption band. An intramolecular energy transfer from the 3 MLCT excited state of the complexes was proposed for the sensitized photochromism of these ligands.17 Later on, the photomodulation of the luminescence intensity of rhenium(I) complexes with different stilbene-like pyridine ligands (Figure 10a), such as 1,3,5-tris(4-ethenylpyridyl)benzene,18 octadecyl-substituted styryl pyridine,19 aza-crown-substituted styryl pyridine,20 and 1,2-di-(4-pyridyl)ethylene,21 and their 3 MLCT-sensitized photoisomerization processes were also reported.22 The detailed mechanism (Figure 10b) of the photosensitization of the proposed intramolecular energy processes from the 3 MLCT to the 3 IL (stilbene- or azo-containing pyridine ligands) state was recently confirmed by direct observation with ultrafast time-resolved absorption, infrared, and resonance Raman spectroscopic studies on [Re(CO)3 (bpy) (stpy)]+ by Vlˇcek and coworkers.23 This mechanism has been further supported by quantum chemical calculations by Daniel and Bossert.24

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials

5

1MLCT(bpy)

N

~0.2 ps 3MLCT(bpy)

N

N

~3.5 ps

hn

N

3IL (stpy) t

N

12 ps

3IL (stpy) p

>1 ns

N N

O O

O [Re(t-stpy)(CO)3(bpy)]+

O

(a)

[Re(c-stpy)(CO)3(bpy)]+

(b)

Figure 10 (a) Schematic drawing of stilbene-like pyridine ligands. (b) Proposed excited-state behavior of [Re(CO)3 (bpy)(stpy)]+ . (Reproduced from Ref. 23.  American Chemical Society, 2005.)

h n2

O N

O OC

N Re

Quenching OC

Cl CO

h n1

h n2

O N

O OC Quenching

OC

N Re

Cl CO

h n1

Figure 11 Schematic representation of the photoswitching of the luminescence behavior of the rhenium(I) complex with bis(anthracene)-functionalized bipyridine.

The enhancement of the luminescence properties of the rhenium(I) complexes has also been demonstrated by Belser and coworkers with a bis(anthracene)-functionalized bipyridine ligand (Figure 11).25 This rhenium complex did not show any MLCT phosphorescence as the emissive excited state was effectively quenched by the anthracene moieties. With the disruption of π -conjugation through the reversible photoinduced [4+4] cycloaddition of the anthracene moieties, the lowest lying excited state of the cyclized dimer could not quench the MLCT emission, leading to the enhancement of the luminescence properties.

In addition to the photoswitching “on” of the luminescence with photochromic ligands that show negative photochromism, the switching “off” of the luminescence through the development and incorporation of ligands with positive photochromism, such as spiropyran, spirooxazine, and diarylethene, into different luminescent transition-metal complexes could also be achieved. Although organic spiropyran and spirooxazine have been extensively reported for many years,1 it was only in 1998 that the first design and synthesis of transition-metal complexes with spirooxazine-containing ligand were reported.26 The corresponding ruthenium bipyridyl complexes were found to significantly perturb the photochromic properties of the ligands. The photochromic ring-opening reactions of spiropyrans and spirooxazines have also been demonstrated to be sensitized by intermolecular energy transfer using organic triplet donors.27, 28 Intramolecular photosensitization of the spirooxazine using inorganic/organometallic photosensitizers has been achieved by the incorporation of spirooxazine-containing ligands into metal complexes with sufficient triplet-state energy for the intramolecular energy transfer process.29 A spirooxazinecontaining pyridine ligand (SOPY) was designed and incorporated into the rhenium(I) tricarbonyl diimine system, [Re(CO)3 (N–N)(SOPY)]+ (N–N = t Bu2 bpy, (CH3 )2 bpy, phen) (Figure 12).29 The observation of the formation of photomerocyanine with excitation into the MLCT absorption band in the complexes is indicative of an efficient intramolecular photosensitization reaction of the spirooxazine moiety by the MLCT excited state.29 This led to a

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

6

Supramolecular devices O C

O C Re

OC N

O C

O C

IL and MLCT excitation

N

OC

Re

N

h n or ∆

N

O

N

N

O

N

O

N

N

O O N

O Photomerocyanine form

Photosensitized photochromism of [Re(CO)3 (N–N)(SOPY)]+ . (Reproduced from Ref. 29.  American Chemical Society, OC OC N O X

OC

CO

Re

Cl N

X O

X O

N

N

N

N O

(a)

CO

OC Re Cl N N

O N

O N

X= CO [Re(CO)3(L1)Cl] X=CH2 [Re(CO)3(L3)Cl]

X=CO [Re(CO)3(L2)Cl] X=CH2 [Re(CO)3(L4)Cl]

1LC

Photochromic ring-opening

3LC

ISC

1MLCT

ISC

Normalized emission intensity

Figure 12 2000.)

3MLCT

560

Phosphorescence (b)

(c)

640 720 Wavelength (nm)

(d)

560 640 720 Wavelength (nm)

800

Figure 13 (a) Rhenium(I) tricarbonyl complexes with spirooxazine-containing bipyridine ligands. (b) Qualitative energy state diagram for the closed form of spirooxazine-containing bipyridyl rhenium(I) complexes. Normalized emission spectra of the closed forms (——) and the open forms (· · · · · ·) of (c) [Re(CO)3 (L1)Cl] and (d) [Re(CO)3 (L2)Cl] in 77 K EtOH–CH3 OH–CH2 Cl2 (4 : 1 : 1 v/v/v) glass. (Reproduced from Ref. 30.  Wiley-VCH, 2004.)

shift of the excitation wavelength from the UV region in the free ligand to the lower energy visible region in the complexes. The efficient energy transfer from the 3 MLCT to the 3 SOPY followed by the ring-opening reaction was further supported by the observation of the presence of intraligand (IL) phosphorescence, which resembles the free ligand phosphorescence and is insensitive to the change in the nature of the diimine ligand, instead of the typical MLCT phosphorescence. However, no photosensitization was observed in the subsequent report on a series of spirooxazine-containing bipyridine ligands and their rhenium(I) tricarbonyl complexes (Figure 13a).30 This has been attributed to the lower 3 MLCT energy (166–188 kJ mol−1 , estimated from the phosphorescence of these complexes) as compared to the triplet-state energies of spirooxazine (210–225 kJ mol−1 ) (Figure 13b).30 Moreover, the quantum efficiencies of the photochromic ring-opening reactions of these complexes were found to be in line with the energies of the MLCT excited states ([Re(CO)3 (L1)Cl] (0.15) < [Re(CO)3 (L2)Cl] (0.18) < [Re(CO)3 (L3)Cl] (0.48)

< [Re(CO)3 (L4)Cl] (0.65)), supportive of the quenching mechanism from the triplet state of spirooxazine to 3 MLCT, in which the efficiency of the quenching pathway increases as the energy of the 3 MLCT excited state decreases. With these spirooxazine-containing ligands, the emission wavelengths of the complexes could also be reversibly switched from the typical MLCT emission to the LC emission of the merocyanine by the photochromic reactions (Figure 13c and d). Similar photoswitching of the MLCT emission to the LC emission through the photochromic reaction of a nitro-spiropyrancontaining ruthenium(II) tris(bipyridyl) complex was also demonstrated.31 The switching “off” of the luminescence with spiropyrancontaining ligand has also been demonstrated by Gust, Moore, and coworkers with a spiropyran-functionalized porphyrin and its zinc complex (Figure 14).32 Since the excited state of photomerocyanine form was much lower lying in energy, on conversion to the photomerocyanine form, the typical fluorescence of the zinc porphyrin showed a significant drop in intensity due to the quenching of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials

7

NO2

N O

NO2 N

N N Zn N N

h n1

N N Zn

h n 2, ∆

+

−O

N

N

Figure 14

Photochromism of nitro-spiropyran-functionalized porphyrin.

the singlet emissive zinc porphyrin excited state by the photomerocyanine. On the other hand, the photoswitching of the luminescence behavior of transition-metal complexes has also been shown to be achieved through the incorporation of photochromic diarylethene-containing ligands.33 These diarylethene-containing ligands have received growing attention in recent years as they have been demonstrated to show excellent fatigue-resistance and thermal stability of both photochromic isomers. The luminescence of diarylethene-containing metal complexes could be switched off or the emission energy of the complexes can be shifted to the red through the photochromic reaction of the diarylethene moiety on the ligand. Pioneering study by Lehn and coworkers demonstrated luminescence modulation of the rhenium(I) and tungsten(0) complexes with pyridine-containing diarylethene ligands derived from the attachment of one or two pyridines to the thiophene moieties of the dithienylethenes.34 Branda and Norsten reported the incorporation of the same ligand into two ruthenium porphyrin units,35 to achieve a photomodulated phosphorescence in the NIR (near-infrared) region (Figure 15). Since the wavelengths of excitation for the highest phosphorescence intensity are within a narrow range of 400–480 nm, which have little effect on the photochemical interconversion between the open and closed forms of the diarylethene unit in either direction, the modulated phosphorescence signals could be used as nondestructive readout for the photochromic stage. This concept could be applied to the construction of the nondestructive photoswitchable binary devices. Apart from coordination of this pyridine-containing diarylethene ligand to various metal complexes, the pyridine groups of the ligand can also be methylated to form the bis(pyridinium)-containing diarylethene (Figure 16a).36 The photochromic reactions of the bis(pyridinium)containing diarylethene have also been demonstrated to

show different biological activity in the transparent nematode worms (Caenorhabditis elegans). The bis(pyridinium)containing diarylethene retains its photochromic reactivity, and the photoinduced ring-closing and ring-opening reactions could be repeated several times inside the body of the worms as confirmed by light microscopy (Figure 16b and c). Although both forms of the diarylethene are eventually toxic to the organisms, the open form shows significantly less paralysis as compared to the closed form. Thus, the worm treated with the open form could be paralyzed by UV excitation as a result of the formation of the closed form. The closed-form-immobilized worms can regain their activity by visible light excitation. This demonstrates the capability of switching the biological activity with different light excitations. Besides, the photocyclization of diarylethene-containing ligands could also be sensitized by the triplet excited state of the metal complexes in a dinuclear tris(bipyridyl)ruthenium(II) complex (Figure 17).37 The transfer of energy from the 3 MLCT state to the 3 IL state followed by photocyclization was proposed by Belser, De Cola and coworkers.37 The typical MLCT emission of this dinuclear ruthenium(II) complex could be switched off by the photocyclization reaction (Figure 17b). In contrast to the dinuclear ruthenium complex, the osmium analog could not be photosensitized as the 3 MLCT energy was lower than that of the triplet state of the dithienylethene.37 As a consequence, the reactive excited state of the dithienylethene could be quenched in the presence of the lower lying excited state in the osmium complex, rendering an exceptionally low quantum efficiency for the photocyclization even on UV excitation into its IL absorption. A series of photochromic bis(2-methylbenzothien-3yl)maleimide-coupled polypyridyl ligands (L5 and L6) and their ruthenium(II) complexes, [Ru(bpy)2 (L5)]2+ and [Ru(bpy)2 (L6)]2+ , were recently developed by Scandola

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

8

Supramolecular devices 3

S

N

×5

S

NN Ru

N

NN

NN

CO

×5 1

0

>580 nm

365 nm

2

300

400

F F

F F

S

NN N Ru OC N N

Emission intensity

(b)

F F

S NN Ru N N CO

N

Open form

600

0

700

200

400 600 Time (s)

800

Closed form

700 (a)

500 l (nm)

Emission intensity

OC

NN Ru

F F Absorbance

F F

F F

750

800

850

l (nm)

(c)

Figure 15 (a) Photochromism of diarylethene-bridged dinuclear ruthenium porphyrin. (b) Absorption spectral changes of the complex (6 µM) in benzene solution on photocyclization with 365 nm excitation. (c) Emission spectra of the open and closed forms of the complex in benzene solution. Inset: modulated emission signal of the complex with alternate irradiation at 365 nm (shaded areas) and >580 nm (unshaded areas). (Reproduced from Ref. 35.  Wiley-VCH, 2001.)

F F

F F

F F

F F

UV

F F

F F

Vis ⊕

H3C N

S

S

⊕

N CH 3

⊕

H3C

S N

S

⊕

N

CH3

(a)

(b)

(c)

Figure 16 (a) Photochromism of the bis(pyridinium)-containing diarylethene. Optical microscopy images of the worm with the bis(pyridinium)-containing diarylethene being interconverted between (b) the open form and (c) the closed form with 490- and 365-nm light excitation for 20 and 2 min, respectively. (Reproduced from Ref. 36.  American Chemical Society, 2009.)

and coworkers.38 The singlet and triplet photochromic pathways of the free ligands and the MLCT-sensitized photochromism (Figure 18) in the complexes were investigated in detail by ultrafast spectroscopic and computational studies. The design and studies involving the ligand itself directly forming part of the diarylethene framework, which should possess a greater perturbation and interaction between the metal center and the photochromic moiety, were also reported recently.33 Luminescence modulation with these ligands has also been demonstrated. In a zinc complex

of diarylethene–phthalocyanine hybrids, the characteristic luminescence of the zinc phthalocyanine moiety could be successfully switched off and on through the photocyclization and photocycloreversion processes (Figure 19).39 Other photoregulating luminescent properties of the metal complexes of phthalocyanine40 and azaporphyrin41 were subsequently developed. By the Suzuki cross-coupling reaction of the substituted thienylboronic acid and 5,6-dibromo-1,10-phenanthroline, Yam and coworkers have developed a series of photochromic 5,6-dithienyl-1,10-phenanthroline ligands, with

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

F F F F

S

S

15

N N

N MII

MII N

N

N

N

N M=Ru; Ru(µ-L)Ru M=Os; Os(µ-L)Os

N

N

N

e (× 104 M−1 cm−1)

F F

9

I (arbitrary units)

Photoswitching materials

10

500

5

600 700 Wavelength (nm)

800

N 0 300

(a)

(b)

400

500 600 Wavelength (nm)

700

800

Figure 17 (a) Dithienylethene-bridged dinuclear tris(bipyridyl)ruthenium(II) and osmium(II) complexes. (b) UV–vis spectral changes accompanying the photocyclization of the dinuclear ruthenium complex. Inset: the corresponding decrease in emission intensity of the dinuclear ruthenium complex. Conditions: CH3 CN, 293 K. (Reproduced from Ref. 37.  American Chemical Society, 2004.) N

~1 ps

N 30 ps O

N

O

Cyclization

Ru(S0)-*L5(S1)

*Ru(S1)-L5(S0) 60 ◦ C results in a continuous 360◦ unidirectional rotational motion of the top half of the molecule with respect to the bottom. Second generation of light-driven molecular motors with their speed of rotation controlled through the incorporation of

(CH3)ax (CH3)ax

(CH3)eq ≥ 280 nm ≥ 380 nm

Trans (right-handed helicity)

(CH3)eq

Cis (left-handed helicity) 20 °C

60 °C

(CH3)eq (CH3)eq

Trans (left-handed helicity)

≥ 380 nm

(CH3)ax

≥ 280 nm

(CH3)ax

Cis (right-handed helicity)

Figure 38 Photochemical and thermal isomerization processes of light-driven, unidirectional molecular rotor based on sterically crowded alkene. (Reproduced from Ref. 73.  Nature Publishing Group, 1999.)

heteroatoms, such as sulfur and oxygen, has subsequently been reported.74 Fine-tuning of the rotor function through the attachment of different substituents or groups on onehalf of the molecules has also been recently reported.75 Doping of a similar molecular rotor into the liquidcrystal film allows the control of the liquid-crystal texture rotation.76 In the presence of 365-nm light, the molecular rotor rotates in a unidirectional manner through the photochemical trans–cis isomerization and the helicity inversion process. This induces the helical reorganization in the liquid-crystal film. Under an optical microscope, the texture reorganizes in a clockwise rotational fashion, whereas in the absence of 365-nm light source, anticlockwise rotational texture reorganization as a result of the thermal isomerization process can be observed. Such a texture rotation can also be used to rotate a microscopic glass rod (5 × 28 µm) on the surface of the film (Figure 39).76 The photodriven unidirectional, mechanically interlocked molecular rotor has been reported by Leigh and coworkers.77 This rotor is a [3]catenane that is constructed by two relatively small isophthalamide macrocycles on a big ring with four hydrogen-bond-accepting units with different macrocycle-binding affinities: fumaramide (A), N-methylated fumaramide (B), succinic amide ester (C), and isolated amide (D) (Figure 40a). As the hydrogenbond-accepting ability follows the order A > B > C > D, the macrocycle-binding affinity follows the same order. Through sequential photochemical isomerization of fumaramide double bonds in the [3]catenane, (Figure 40b) unidirectional circumrotation of the isophthalamide macrocycles along a large, four-stationed ring can be achieved. On photoisomerization to their maleamide counterparts, the affinity of the amide station for the macrocycle drops significantly; hence the macrocycle moves to the next unoccupied station with the highest binding affinity. When A is selectively photoisomerized with 350 nm excitation, the isophthalamide macrocycle at A moves to C in a counterclockwise direction, as the clockwise pathway is blocked by the isophthalamide macrocycle stationed at B. With subsequent photoisomerization of B, the isophthalamide macrocycle at B moves to D in a counterclockwise direction again, as the clockwise pathway is inhibited by the macrocycle stationed at C. At 100 ◦ C, the thermal isomerization of maleamide–fumaramide at A and B occurs and the isophthalamide macrocycles return to the stations A and B but with the rings originally stationed at A and B being swapped (Figure 40b). Through the following sequential selective photochemical isomerization with 350- and 254nm excitations, the sequential counterclockwise rotations of the isophthalamide macrocycles occur. The subsequent thermal isomerization would regenerate the initial state of the system.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials

(a)

(b)

21

24 nm

(d)

0 nm

(c)

Figure 39 (a) Structure of the molecular rotor. (b) Polygonal texture of a liquid-crystal film doped with the molecular rotor (1% by weight). (c) Glass rod rotating on the liquid crystal during irradiation with ultraviolet light. Frames 1–4 (from left) were taken at 15 s intervals and show clockwise rotations of 28◦ (frame 2), 141◦ (frame 3), and 226◦ (frame 4) of the rod relative to the position in frame 1. (d) Surface structure of the liquid-crystal film (atomic force microscopy image; 15 µm2 ). (Reproduced from Ref. 76.  Nature Publishing Group, 2006.)

A

D

A′ O O

D O

(CH2)4

N H

N H

C

O

O

C HN

(a)

A′

CH3 F A B O E N C NHD O HN O H O N HO N N O CH3 (CH2)12

Z ,E

(CH2)12

(CH2)4

O

C

Z ,Z

E ,E

O

B

D

H N

O H NH N O

E /Z

E ,E

B′

D

A

A′

C

O

NH HN O

B

A′ B

D

D

C

B′

C A

B (b)

Z,E

D

B

Z,Z

C

E ,E

Figure 40 (a) Structure of the photodriven unidirectional, mechanically interlocked molecular rotor. (b) Stimuli-induced unidirectional rotation in a four-station [3]catenane. (Reproduced from Ref. 77.  Nature Publishing Group, 2003.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

22

Supramolecular devices

The photodriven rotary motion has also been demonstrated by the research group of Sauvage in a tris(bipyridyl)ruthenium(II) complex-templated [2]catenane,78 which consists of a bis(1,10-phenanthroline)-containing 63-membered ring and a 2,2 -bipyridine-containing 42-membered ring (Figure 41). When this catenane is irradiated with a 250W halogen lamp (λ > 300 nm) in the presence of 10-fold excess of Et4 NCl in CH2 Cl2 or in CH3 CN, the photoinduced ligand substitution of the bipyridine is observed. The

O

−

O

O

N Ru

N

O O

2+

O

O

2PF6−

O

O

n+ O

nPF6− O

O

O N

O

de-coordination and the corresponding ligand substitution are confirmed by 1 H NMR spectroscopy and mass spectrometry. On heating, the recomplexation occurs to regenerate the original catenane. The photocontrol of the frictionless rotary molecular motion of the cyclopentadienyl rings (Cp) in ferrocene has also been developed using the two Cp rings substituted with zinc porphyrin and aniline units (Figure 42).79 Since the amine group of the aniline unit of the Cp ring can

O

O

− N N

N O

hn

N

∆

N

Ru N

O

L

O N N

O

O O

O O

O

O

O

−

L

a: Et4NCl/CH2Cl2 b: CH3CN

O O

N

O

O

O

a: L = Cl; n = 0 b: L = CH3CN; n = 2

O

Figure 41 Schematic representation of the photoinduced and thermal motions taking place in the tris(bipyridyl)ruthenium(II) complextemplated [2]catenane. (Reproduced from Ref. 78.  Wiley-VCH, 2004.) H2 N

NH2

N N N Zn N

N N Zn N N

N

Self-locking N

NH2

Photoisomerization

N N Zn N N

N NZnN N

N

N NZn N N

N

N NZnN N

Trans-

N H2

NH2

Internally double-locked state

N

Photoisomerization

H2 N

N NH2 N N Zn N N

NH2

N N N Zn N N N N N N NZn N N

N NZn N N NH2

Cis-

N H2

Internally double-locked state

Figure 42 Molecular structure of internally double-locked state and schematic representation of the self-locking operation in response to the photoisomerization of 1,2-di-(4-pyridyl)ethylene. (Reproduced from Ref. 79.  American Chemical Society, 2006.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials form a weak coordination with the zinc porphyrin unit, the rotary motion of Cp rings is restricted as they are locked by double intramolecular Zn–N coordination (internally double-locked state). To release and control this coordination photochemically, 1,2-di-(4-pyridyl)ethylene, which is capable of undergoing trans-to-cis photoisomerization, is added to compete with the relatively weak intramolecular aniline coordination. The cis-1,2-di-(4-pyridyl)ethylene has a much stronger binding affinity than the trans isomer, as its geometry allows the formation of two Zn–pyridyl coordination bonds. With the formation of the bis(Zn–pyridyl) coordination with the cis isomer, the Zn–aniline coordination is released and Cp rings are rotated. The resulting configuration change of the molecule is supported by clear CD spectral changes. In the presence of trans-1,2di-(4-pyridyl)ethylene, the ferrocene molecule retains its internally double-locked configuration as the trans isomer can only form weak coordination to the Zn–porphyrin unit. Thus, the rotary motion of the Cp rings can be controlled through the trans–cis photoisomerization of 1,2di-(4-pyridyl)ethylene (Figure 42).

5

PHOTOCONTROLLABLE RECEPTORS

Molecules that can specifically recognize and bind to a compound (or a guest molecule) are normally referred to as receptors or host compounds. The selectivity and binding affinity between the receptor and the guest molecules are governed by the interactions between the guest and the cavity or the binding site of the receptor. Therefore, the binding stability constant, which is a measure of the binding affinity, shows strong dependence on the conformation and properties of the binding sites, such as size, shape, polarity, charge distribution, hydrophobicity, hydrophilicity, and others. As illustrated above, almost all photochromic moieties are capable of reversibly modulating the changes in the conformation and charge distribution by photoexcitation. Thus, the incorporation of these photochromic moieties into a host system, if suitably designed, can lead to the host system with photoswitchable binding affinity. These systems are designed mainly based on the photoinduced conformational changes so as to modify the size of the binding site or charge redistribution to alter the electrostatic interaction. The photocontrolled release and uptake of the guest molecules can be used to transport membranes.80

N N N N O

C

O

O

O

O

N

Photoswitchable binding affinity based on size/conformational changes

The crown ethers and related compounds are probably the most extensively investigated host system for the binding of

O C

C O N

O

O

O

O

C O N

N

Figure 43 Trans–cis isomerization of azobenzene-capped 1,10diaza-18-crown-6. (Reproduced from Ref. 81.  American Chemical Society, 1980.)

cations. Depending on the size, conformation, and rigidity of the crown cavity, these macrocyclic polyethers can selectively bind to specific cations with different binding affinities. The pioneering work by Shinkai and coworkers has demonstrated the use of azobenzene as a tool to change the conformation of 1,10-diaza-18-crown-6 (Figure 43) and the subsequent binding affinities of different metal ions on trans–cis photoisomerization.81, 82 In the presence of the azobenzene unit, it confers a conformational rigidity and leads to a smaller crown-ether cavity as compared to the unsubstituted 18-crown-6. Thus, in the ion-extraction study, the trans isomer preferentially binds to Li+ and Na+ ions but not the K+ ion, which is known to be the preferred ion for the 18-crown-6 unit. On conversion to the cis isomer, it shows an enhanced binding affinity toward the larger K+ and Rb+ ions. Further modification of the azobenzene unit to azopyridine, a photoswitchable receptor for heavy metal cation, particularly for Cu2+ , was subsequently reported.83 On the other hand, Yamashita et al. also reported a bis(anthracene)-functionalized crown ether (Figure 44), in which the sodium-ion binding affinity could be significantly altered by the intramolecular photochemical [4+4]cycloaddition of the anthracene moieties.84 The thermal backward ring-opening reaction rate is found to be lowered in the presence of the metal ion by the stabilization effect, following the order of Na+ > K+ > Li+ , which is in line with the binding affinity of the metal ion with 15crown-5. Later, the control of the binding affinity of lithium ion for the bis(cinnamate ester)-containing crown ethers using the intramolecular [2+2]photocycloaddition was also reported.85 In addition to photomodulation of the crown cavity, the conformational modification of the bis(crown ether) unit

O

O O

5.1

23

O

O

hn

O

∆

O

O O O

Figure 44 Reversible photocycloaddition reactions of bis(anthracene)-functionalized crown ether. (Reproduced from Ref. 84.  Elsevier, 1980.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

24

Supramolecular devices

O O

Photo-isomerization

O

O

K+

O N

N

hn ∆

N

O O O

O

O

O

O O O K+ O O O

O O K O

O

O

N

+

O

O O

O

O N N

O O

N N

O

O

O

O

O O

O

O

O Thermal isomerization

O O (b)

(a)

IN Aqueous phase

Liquid membrane phase

OUT Aqueous phase

Figure 45 (a) Photoswitable K+ -ion binding of butterfly crown ethers. (b) Schematic representation of light-driven K+ -ion transport. (Reproduced from Ref. 86.  American Chemical Society, 1981.)

has also been widely used in changing the selectivity of different metal ions. The bis(crown ether) units are known to bind metal ions with sizes slightly larger than their cavity in a sandwich fashion. The corresponding binding affinity is much higher than that for the mono(crown ether) unit. The bis(crown) azobenzenes or “butterfly crown ethers” (Figure 45a), in which only the cis configuration allows the sandwich binding mode, have been designed.86 In the metal ion-extraction study, the cis-azo-bis(benzo-15crown-5) extracts K+ ion 42.5 times more effectively than the trans conformer.86 The photocontrolled permeability of potassium ion transport using this azo-bis(benzo-15-crown5) in the liquid membrane (Figure 45b) has also been demonstrated.86, 87 Photoswitching of the bis(crown) configuration by the photochromic diarylethene has also been demonstrated with a series of bis(benzocrown)-substituted dithienylperfluorocyclopentenes reported by Takeshita and Irie.88 In the benzo-15-crown-5-containing diarylethene, the open form with two crown moieties effectively binds with the potassium and the rubidium ions in an intramolecular sandwich fashion (Figure 46). However, such a binding mode is restricted by the confined conformation in the closed form,

F2 C F2C C F2

S O S O

O K+

O O

(a)

F F

O O O

F F

S O O O

O O

O

O

(b)

and thus the binding abilities between the closed form and the potassium as well as rubidium ions decrease dramatically. Similar changes in binding ability of the diarylperfluorocyclopentene for benzo-18-crown-6 moiety and cesium ion have also been observed. Photocontrol of the binding affinity for small organic molecules, such as 4,4 -dipyridyl, cyclohexanol, methoxybenzene, toluene, nerol, and genraniol, has been achieved by the azobenzene-capped β-CD (Figure 47).89 Through the trans–cis photoisomerization, the depth and the size of the β-CD cavity are changed as a result of the conformational modification. In the binding study, the cis

O O

O

N N

O O

hn ∆

Figure 47

O O

Figure 46 Structural representations for (a) intramolecular sandwich ion-bound species of the open form and (b) the confined configuration in the closed form.

O O

S

O

N N O

F F

+

Photoswitchable binding mode of azobenzene-capped β-CD.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials

H N

O

O N

Figure 48

H N

S

S N H H N

N

H N

S

S O

hn2

O

Photocontrollable binding of porphyrin guest molecule in a diarylethene-tethered β-CD.

isomer is found to be capable of forming the inclusion complex with two small guest molecules, such as toluene, methoxybenzene, nerol, and genraniol, whereas the trans isomer can only accommodate one guest molecule. For the larger 4,4 -dipyridyl guest molecule, only the cis isomer can form the inclusion complex. This observation suggests that the β-CD cavity of the cis isomer is larger and deeper. Extension of this work to connect two β-CD units with azobenzene has also been reported by the same research group.90 However, no guest molecules have been found to show distinguishable selectivity for the cis and trans conformations. Recently, a photocontrollable release and uptake of porphyrin guest molecule was demonstrated by a diarylethene-tethered β-CD, in which each thiophene unit of the diarylethene was connected to a β-CD through the amide linkage (Figure 48).91 In the complexation study with meso-tetrakis(4-sulfonatophenyl)porphyrin (TSPP), the open form strongly binds to TSPP with both CD cavities in a cooperative fashion. On conversion to the closed form, the binding constant of TSPP drastically decreases by a factor of 35. The binding enthalpy on conversion to the closed form is also decreased by half, which is close to the binding enthalpies of TSPP and native β-CD. These observations are suggestive of a single β-CD–TSPP interaction, as the condensed thiophene unit in the closed form does not sterically allow the cooperative binding.

5.2

H N

hn1

25

Photoswitchable binding affinity based on the change of charge distribution

The considerable conformational variations associated with the photochromic reactivity and the corresponding designs

on the development of photoswitchable host molecules have been illustrated in the above examples. In addition to the conformational changes, the photochromic reactivity of some families can also induce substantial changes in the charge distribution of the molecules. For example, spirooxazines, spiropyrans (Figure 2), and chromenes (Figure 3) change from the neutral and relatively nonpolar molecule in the closed form to the highly polar zwitterionic form in the photochromic ring-opened isomers.1 A more significant change in the charge as well as the polarity of the molecule can result from the photochromic families, such as triarylmethane leuconitriles (Figure 7). Such changes can be readily used to vary the binding affinity of the charged guest molecules, in particular ions, as the host–guest stabilization is significantly modified because of the difference in the electrostatic interaction. Kimura and coworkers have developed a series of azacrown ether-containing spiropyrans (Figure 49),92 which show an enhancement of metal ion binding, in particular, for Li+ ion, on conversion to the photomerocyanine form. The increased metal ion binding affinity is attributed to the additional intramolecular ionic interaction between the phenolate anionic moiety of the photomerocyanine form and the metal ion, which stabilizes the metal ion–crown ether complexation. The intramolecular phenolate–Li+ interaction has further been supported by 7 Li-NMR spectroscopy.92, 93 Such strong cation–anion interactions can also induce photochromic forward isomerization from the spiropyran to the photomerocyanine form and slow down the thermal backward reaction after photogeneration of a high concentration of photomerocyanine.93 On visible light excitation, the photomerocyanine undergoes photoinduced NO2

N O

NO2

O O

UV (or dark)

N+

Vis

X−

N O

+ M+X−

n

−O

O

N M+ O

O

n

Figure 49 Photocontrol of metal ion binding by aza-crown ether-containing spiropyrans. (Reproduced from Ref. 92.  Royal Society of Chemistry, 1991.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

26

Supramolecular devices

N O

NO2

NO2

N O

N O

N S

N

NO2 O

(a)

N S

S

O

N O

O

O

n

O N

n

N S

n

N

O O

O2N

n

O N

NO2 NO2

N O O

O

N (b)

O

n N

M2+

O

N+

−

O

2+

N

n

O

M O

n

O

N O n

Figure 50 (a) Structures of aza-crown ethers and thiacrown ethers containing spiropyrans, spirooxazines, and spirothiopyrans. (b) Metal ion-induced formation of the photomerocyanine in the cryptand-containing spiropyrans.

ring-closing reaction to the spiropyran form, thus reducing the metal ion complexation capability. Through different wavelengths of excitations, the binding and release of metal ion can be controlled. Extension of this work to other aza-crown ethers and thiacrown ethers containing spiropyrans, spirooxazines, and spirothiopyrans (Figure 50a), based on similar design principles for the binding of different types of metal ions, has subsequently been reported.94–97 On the other hand, Inouye and coworkers also reported cryptand-containing Source phase

spiropyrans (Figure 50b). These cryptand-containing spiropyrans also show metal ion-induced formation of the photomerocyanine form. Moreover, they have high sensitivity and selectivity toward alkaline-earth metal ions. The selectivity of different alkaline-earth metal ions can also be varied with the size of the crytand moiety.98 Application studies of these crown ether-containing spiropyrans as metal ion transporters in liquid membranes (Figure 51a) have also been examined under various lighting conditions.99 The transportability of Li+ increases

NO2 O

Light 1

N

O N

Light 2

Receiving phase

NO2

N O

N O

O

N O

N

N

O N

O

+

NO2

UV O O−

Receiving phase

Source phase

Pic-

N+

O2N

+

Li

NO2

N

N

Visible

−O

O

+ N

NO2

Li

Pic+ N

O2N

Membrane phase

O−

(a) (b)

+ N

+ N

O2N

O −O N Li+ N

O

O

Pic-

NO2

O− + N

N Li+ N

−O

O

Pic-

NO2

Figure 51 (a) Apparatus for liquid membrane transport under photoirradiation conditions. (b) Plausible mechanisms for Li+ ion transport through liquid membrane with crown ether-containing spiropyrans as transporter under UV and vis excitations at the source and receiving phases, respectively. (Reproduced from Ref. 99.  American Chemical Society, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials with UV irradiation at the source phase and/or with visible light excitation at the receiving phase. Uphill transportation of the Li+ ion from the source phase to the receiving phase can be achieved when the receiving phase and the source phase are irradiated with visible light and UV light excitation, respectively. The mechanism of uphill transportation is illustrated in Figure 51(b). The UV light excitation at the source phase generates the photomerocyanine form, which strongly binds to the Li+ ion, to take up the Li+ ion. The corresponding Li+ –photomerocyanine complex is then carried into the liquid membrane phase. Visible light excitation at the receiving phase regenerates the spiropyran form to release the Li+ ion, as the binding affinity is significantly lowered. The metal ion transport selectivity for Li+ , Na+ , and K+ follows the same order as the metal ion complexation selectivity of the photomerocyanine.93 When these spiropyrans are further functionalized with triethoxylsilyl moiety and immobilized onto silica gel, photocontrollable metal ion separations can be achieved.100 With similar photoinduced cation-binding stabilization principle, a series of iminodiacetate-substituted nitrospiropyrans for photoswitchable binding and sensing of divalent metal ions have been very recently reported (Figure 52).101 These spiropyran chelates are water soluble and the esters of the chelates are capable of diffusing across the plasma membrane of living cells. Thus, the potential uses of these chelate-bearing spiropyrans for the optical control of metal ion binding in cells and in vitro sensing of Zn2+ , Mn2+ , Ni2+ , Eu3+ , and Gd3+ have also been examined.101 Both the absorption and the emission of the photomerocyanine forms are blue-shifted in the presence of the metal ions, and thus can provide distinguishable sensing signal for the metal ions. On the basis of the difference in the metal ion binding affinity between the photomerocyanine form and the spiropyran form as a result of the additional stabilization of the metal ion–phenolate interaction as well as the ready interconversion of the two forms with different wavelengths of excitations, the photomanipulation of the binding of metal ion in living cells can be achieved.101 Importantly, the

NO2 UV/2-photon

N O

NO2

Vis

N+

−O

N −O C 2

N

CO2−

−O C 2

Mn+ −O C 2

Mn+ Mn+ = Gd3+, Eu3+, Mn2+, Ni2+, Zn2+

Figure 52 Photoswitchable metal ion binding of an iminodiacetate-substituted nitro-spiropyran. (Reproduced from Ref. 101.  American Chemical Society, 2008.)

O −

O2C

N Ca2+

Ph Ph

d+ UV

CO2−

27

d− O

∆ N

Ca2+ CO2− CO2−

Figure 53 Photoswitchable calcium ion binding of an iminodiacetate-substituted photochromic naphthopyran. (Reproduced from Ref. 102.  American Chemical Society, 2008.)

spiropyran–photomerocyanine interconversion can also be probed by two-photon excitations, both of which are in the NIR region.101 Similar to the spiropyrans, the open forms of chromenes are also zwitterionic in nature and possess a phenolate moiety. Thus a water-soluble iminodiacetate-substituted photochromic chromene (Figure 53) that is capable of enhancing the metal ion binding affinity through the photochromic ring-opening reaction has also been designed and reported.102 To simulate the physiological environment, the metal ion binding study of the 5-iminodiacetate-substituted naphthopyran is conducted in aqueous solution that is buffered at pH 7.6.102 In the closed form, the naphthopyran chelator binds only very weakly as a 1 : 1 complex with Sr2+ , Ca2+ , and Mg2+ . The binding constant for 1 : 1 complexation for Ca2+ is the highest among these metal ions. On photoconversion to the open form, the binding constant for 1 : 1 complexation for Ca2+ increases by 77-fold. The design based on the electronic charge rearrangement of the photochromic moiety, which is generated by photochromic forward reaction for additional electrostatic stabilization of the metal ion complexation with the host pendant, has been illustrated in the above examples. In contrast, the charge redistribution in the photochromic reaction can also result in the destabilization of the metal ion complexation, leading to the photorelease of metal ions. A photochromic naphthopyran with its phenyl moiety functionalized with a crown ether is designed to achieve the photorelease of Pb2+ ions (Figure 54).103 For naphthopyran, the photochromic open form generated by UV excitation is zwitterionic in nature, with the cationic site close to the phenyl moieties. Therefore, on conversion to the open form, the Pb2+ ion bound to the crown-ether moiety experiences electrostatic repulsion exerted by the cationic carbon in the open form, leading to the significant drop in the Pb2+ ion binding stability. On visible light excitation, the closed form is regenerated and the stronger binding affinity for the Pb2+ ion is restored. The photoinduced expulsion of metal ions can also be achieved by the coupling of the crown-ether moiety into the family of the photochromic triarylmethane leuconitriles

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

28

Supramolecular devices

UV

O

+

Vis

O O Pb2+

Pb2+

O

O

O

O

+

−O

O

O

O

O

Figure 54 Photorelease and uptake of Pb2+ ion with crown-ether containing naphthopyran. (Reproduced from Ref. 103.  Royal Society of Chemistry, 1997.)

O O

M+ O O

N

O

O

O

O

O O

hn ∆

N

+

CN−

M+

+

N

N

CN

+

CN

N O

O N O O

Figure 55 1991.)

O

M+

O

+

hn ∆ O

O

O

N

N

O

O

O O

+

CN−

+

M+

O O O

Photorelease of metal ion in crowned triarylmethane leuconitrile. (Reproduced from Ref. 92.  Royal Society of Chemistry,

such as malachite green and crystal violet (Figure 55),104, 105 which undergo thermally reversible photoionization reaction to give the corresponding triarylmethane cation and a cyanide anion. The photoreleasing mechanism of the metal ion is also attributed to the electrostatic repulsion between the photochromic triphenylmethane cation form and the bound metal ion in the crown cavity. Thus, the effectiveness of repulsion is strongly dependent on the positions of the crown-ether moieties in the triarylmethane leuconitrile unit. As the charge in the triphenylmethyl cation of malachite green mainly delocalizes across the central carbon and the two aminophenyl units, the photorelease of the guest metal ion is very effective when the amine groups are functionalized to give aza-crown ether moieties.105 However, when the crown ether is functionalized on the phenyl moiety, the photorelease of the metal ion is not very effective. Similarly, the effective photorelease of the metal ion can also be achieved by the tris-aza-crowned crystal violet.106 Moreover, the tris-aza-crowned crystal violet shows high selectivity toward Cs+ ion binding. This has been explained by the intramolecular cooperative binding mode of the three aza-crown ether units with the relatively big Cs+ ion. This

cooperative complexation mode results in a serious molecular distortion; thus, anomalous absorption changes in the UV region in the presence of different concentrations of Cs+ ion have been observed.

6

CONCLUSION

The selected examples in this chapter demonstrate the success of the development of photoswitchable molecular materials through the rational designs based on the coupling of photochromic moieties into molecules with different functions, such as luminescence, nonlinear optics, liquid crystal, molecular machine, receptor, electrical conductor/semiconductor, and many others. It is important to emphasize that photoswitchable materials are not confined only to the selected areas in these examples. With the understanding of both the photoswitching processes and the functional properties at the molecular levels, materials that possess different varieties of photoswitchable functional properties can be readily designed and synthesized.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials To date, many photochromic families have been extensively reported and their properties have been well documented. It is, therefore, not difficult to design photoswitchable compounds with desirable photoresponsive behaviors, such as the excitation wavelengths for the switching and thermal reversibility. However, tedious organic synthesis may be required to develop a system with favorable properties. In the recent works on the development of ligands with photochromic moiety and their coordination compounds, it has been demonstrated that photochromism could be significantly perturbed with the incorporation of different transition-metal complexes. This provides a new strategy for the development of new classes of metal-containing photochromic materials with novel properties, such that through the incorporation of different metal centers, ready tuning of the excitation wavelength and photochromic behavior can be achieved. Moreover, the photochromic ligand can also act as a photoswitch to control the functional properties associated with the transitionmetal complexes. Yet, the correlations of the perturbed photochromism via complexation with different metal centers are still not fully established. Further investigations are required to establish a clear correlation between the photochromic properties of the ligands and the excited state, as well as the nature of the metal complexes, so that the photochromism of the metal-containing photochromic material could be tailored and widely utilized as the building blocks for the development of photoswitchable materials in various aspects. Clearly, it has been demonstrated from many examples in this chapter that a lot of photoswitching molecular materials with a wide variety of switchable functional properties have been produced. However, the fabrication of devices from these materials has remained a great challenge due to the lack of suitable processing methods as well as the difficulties in controlling the reversibility and stability of these photoswitchable molecules. The combination of these photoswitching molecular materials with solid matrices, such as sol–gels and polymers, may lead to processable photoswitching materials. However, the photoswitching properties and activities of these molecules are usually perturbed or modified when they are doped in the solid matrices. Innovative explorations of processing methods of these photoswitching materials are, therefore, essential for their potential device applications.

ACKNOWLEDGMENTS

29

REFERENCES 1. H. D¨urr and H. Bouas-Laurent, Photochromism: Molecules and Systems, Elsevier, Amsterdam, 1990. 2. J. C. Crano and R. J. Guglielmetti, Organic Photochromic and Thermochromic Compounds, vol. 1: Main Photochromic Families, Plenum Press, New York and London, 1999. 3. S. L. Murov, Handbook of Photochemistry, Marcel Dekker, New York, 1973. 4. R. M. Kellogg, M. B. Groen, and H. Wynberg, J. Org. Chem., 1967, 32, 3093. 5. M. Irie, Chem. Rev., 2000, 100, 1685. 6. S. Nakamura, and M. Irie, J. Org. Chem., 1988, 53, 6136. 7. S. Kobatake, T. Yamada, K. Uchida, et al., J. Am. Chem. Soc., 1999, 121, 2380. 8. S. Takami, L. Kuroki, and M. Irie, J. Am. Chem. Soc., 2007, 129, 7319. 9. S. Kobatake, S. Takami, H. Muto, et al., Nature, 2007, 446, 778. 10. G. S. Hammond, J. Saltiel, A. A. Lamola, et al., J. Am. Chem. Soc., 1964, 86, 3197. 11. E. K. C. Lee, H. O. Denschlag, and G. A. Haninger, Jr., J. Chem. Phys., 1968, 48, 4547. 12. P. P. Zarnegar and D. G. Whitten, J. Am. Chem. Soc., 1971, 93, 3776. 13. M. Wrighton and D. L. Morse, J. Am. Chem. Soc., 1974, 96, 998. 14. P. Zarnegar, C. R. Bock, and D. G. Whitten, J. Am. Chem. Soc., 1973, 95, 4367. 15. J. R. Shaw, R. T. Webb, and R. H. Schmehl, J. Am. Chem. Soc., 1990, 112, 1117. 16. S. L. Murov, Handbook of Photochemistry, Marcel Dekker, New York, 1973. 17. V. W.-W. Yam, V. C.-Y. Lau, and K.-K. Cheung, J. Chem. Soc., Chem. Commun., 1995, 259. 18. S.-S. Sun and A. J. Lees, Organometallics, 2002, 21, 39. 19. V. W.-W. Yam, Y. Yang, J. Zhang, et al., Organometallics, 2001, 20, 4911. 20. J. D. Lewis, R. N. Pertuz, and J. N. Moore, Chem. Commun., 2000, 1865. 21. O. S. Wenger, L. M. Henling, M. W. Day, et al., Inorg. Chem., 2004, 43, 2043. 22. K. S. Schanze, L. A. Lucia, M. Cooper, et al., J. Phys. Chem. A, 1998, 102, 5577. 23. M. Busby, P. Matousek, M. Towrie, and A. Vlˇcek Jr, J. Phys. Chem. A, 2005, 109, 3000. 24. J. Bossert and C. Daniel, Chem. Eur. J., 2006, 12, 4835.

We acknowledge support from the University of Hong Kong, the City University of Hong Kong, and the University Grants Committee Areas of Excellence Scheme (AoE/P-03/08).

25. A. Beyeler, P. Belser, and L. De Cola, Angew. Chem. Int. Ed., 1997, 36, 2779. 26. R. F. Khairutdinov, K. Giertz, J. K. Hurst, et al., J. Am. Chem. Soc., 1998, 120, 12707.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

30

Supramolecular devices

27. Ralph S. Becker and J. K. Roy, J. Phys. Chem., 1965, 69, 1435.

53. V. Aubert, V. Guerchais, E. Ishow, et al., Angew. Chem. Int. Ed., 2008, 47, 577.

28. J. Hobley and F. Wilkinson, J. Chem. Soc., Faraday Trans., 1996, 92, 1323.

54. K. A. Green, M. P. Cifuentes, T. C. Corkery, Angew. Chem. Int. Ed., 2009, 48, 7867.

29. V. W.-W. Yam, C.-C. Ko, L.-X. Wu, et al., Organometallics, 2000, 19, 1820.

55. R. Sakamoto, M. Murata, S. Kume, et al., Chem. Commun., 2005, 1215.

30. C.-C. Ko, L.-X. Wu, K. M.-C. Wong, et al., Chem. Eur. J., 2004, 10, 766.

56. T. R. Kelly, Molecular Machines, Springer, Berlin, 2005.

31. R. T. F. Jukes, B. Bozic, F. Hartl, et al., Inorg. Chem., 2006, 45, 8326.

58. H. Muakami, A. Kawabuchi, K. Kotoo, et al., J. Am. Chem. Soc., 1997, 119, 7605.

32. J. L. Bahr, G. Kodis, L. De la Garze, et al., J. Am. Chem. Soc., 2001, 123, 7124.

59. H. Murakami, A. Kawabuchi, R. Matsumoto, et al., J. Am. Chem. Soc., 2005, 127, 15891.

33. C.-C. Ko and V. W.-W. Yam, J. Mater. Chem., 2010, 20, 2063.

60. A. Coskun, D. C. Friedman, H. Li, et al., J. Am. Chem. Soc., 2009, 131, 2493.

34. A. Fern´andez-Acebes and J.-M. Lehn, Adv. Mater., 1998, 10, 1519.

61. C. A. Stanier, S. J. Alderman, T. D. W. Claridge, and H. L. Anderson, Angew. Chem. Int. Ed., 2002, 41, 1769.

35. T. B. Norsten and N. R. Branda, Adv. Mater., 2001, 13, 347.

62. Q.-C. Wang, D. H. Qu, J. Ren, et al., Angew. Chem. Int. Ed., 2004, 43, 2661.

36. U. Al-Atar, R. Fernandes, B. Johnsen, et al., J. Am. Chem. Soc., 2009, 131, 15966.

63. A. S. Lane, D. A. Leigh, and A. Murphy, J. Am. Chem. Soc., 1997, 119, 11092.

37. R. T. F. Jukes, V. Adamo, F. Hartl, et al., Inorg. Chem., 2004, 43, 2779.

64. E. M. Perez, D. T. F. Dryden, D. A. Leigh, et al., J. Am. Chem. Soc., 2004, 126, 12210.

38. M. T. Indelli, S. Carli, M. Ghirotti, et al., J. Am. Chem. Soc., 2008, 130, 7286.

65. W. Zhou, D. Chen, J. Li, et al., Org. Lett., 2007, 9, 3929.

39. B. Z. Chen, M. Z. Wang, Y. Q. Wu, and H. Tian, Chem. Commun., 2002, 1060. 40. Q. F. Luo, B. Z. Chen, M. Z. Wang, and H. Tian, Adv. Funct. Mater., 2003, 13, 233. 41. H. Tian, B. Z. Chen, H. Y. Tu, and K. M¨ullen, Adv. Mater., 2002, 14, 918. 42. V. W.-W. Yam, C.-C. Ko, and N. Zhu, J. Am. Chem. Soc., 2004, 126, 12734. 43. C.-C. Ko, W.-M. Kwok, V. W.-W. Yam, Phillips, Chem. Eur. J., 2006, 12, 5840.

and

D. L.

44. J. K.-W. Lee, C.-C. Ko, K. M.-C. Wong, et al., Organometallics, 2007, 26, 12. 45. T.-W. Ngan, C.-C. Ko, N. Zhu, and V. W.-W. Yam, Inorg. Chem., 2007, 46, 1144. 46. P. Belser, L. de Cola, F. Hartl, et al., Adv. Funct. Mater., 2006, 16, 195. 47. P. H.-M. Lee, C.-C. Ko, N. Zhu, and V. W.-W. Yam, J. Am. Chem. Soc., 2007, 129, 6058.

et al.,

57. S. Saha and J. F. Stoddart, Chem. Soc. Rev., 2007, 36, 77.

66. W. Abraham, K. Buck, M. Orda-Zgadzaj, et al., Chem. Commun., 2007, 3094. 67. K. Hirose, Y. Shiba, K. Ishibashi, et al., Chem. Eur. J., 2008, 14, 3427. 68. P. R. Ashton, R. Ballardini, V. Balzani, et al., Chem. Eur. J., 2000, 6, 3558. 69. V. Balzani, M. Clemente-Le´on, A. Credi, et al., Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 1178. 70. S. Saha, A. H. Flood, J. F. Stoddart, et al., J. Am. Chem. Soc., 2007, 129, 12159. 71. N. Armaroli, V. Balzani, J.-P. Collin, et al., J. Am. Chem. Soc., 1999, 121, 4397. 72. G. S. Kottas, L. I. Clarke, D. Horinek, and J. Michl, Chem. Rev., 2005, 105, 1281. 73. N. Koumura, R. W. J. Zijlstra, R. A. van Delden, et al., Nature, 1999, 401, 152. 74. N. Koumura, E. M. Greertsema, M. B. van Gelder, et al., J. Am. Chem. Soc., 2002, 124, 5037. 75. M. K. J. ter Wiel and B. L. Feringa, Tetrahedron, 2009, 65, 4332.

48. W. Tan, Q. Zhang, J. Zhang, and H. Tian, Org. Lett., 2009, 11, 161.

76. J. Vicario, N. Katsonis, B. S. Ramon, et al., Nature, 2006, 440, 163.

49. S. Fraysse, C. Coudret, and J.-P. Launay, Eur. J. Inorg. Chem., 2000, 1581.

77. D. A. Leigh, J. K. Y. Wong, F. Dehez, and F. Zerbetto, Nature, 2003, 424, 174.

50. Y. Tanaka, A. Inagaki, and M. Akita, Chem. Commun., 2007, 1169.

78. P. Mobian, J.-M. Kern, and J.-P. Sauvage, Angew. Chem. Int. Ed., 2004, 43, 2392.

51. R. Sakamoto, M. Murata, and H. Nishihara, Angew. Chem., Int. Ed., 2006, 45, 4793.

79. T. Muraoka, K. Kinbara, and T. Aida, J. Am. Chem. Soc., 2006, 128, 11600.

52. R. T. F. Jukes, B. Bozic, P. Belser, et al., Inorg. Chem., 2009, 48, 1711.

80. J. J. Grimaldi and J.-M. Lehn, J. Am. Chem. Soc., 1979, 101, 1333.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Photoswitching materials

31

81. S. Shinkai, T. Nakaji, Y. Nishida, et al., J. Am. Chem. Soc., 1980, 102, 5860.

94. M. Tanaka, T. Ikeda, Q. Xu, et al., J. Org. Chem., 2002, 67, 2223.

82. H. L. Ammon, S. K. Bhattacharjee, S. Shinkai, Y. Honda, J. Am. Chem. Soc., 1984, 106, 262.

and

95. M. Tanaka, K. Kamada, H. Ando, et al., J. Org. Chem., 2000, 65, 4342.

83. S. Shinkai, T. Kouno, Y. Kusano, and O. Manabe, J. Chem. Soc., Perkin Trans 1, 1982, 2741.

96. K. Kimura, T. Yamashita, M. Kaneshige, and M. Yokoyama, J. Chem. Soc., Chem. Commun., 1992, 969.

84. I. Yamashita, M. Fujii, T. Kaneda, et al., Tetrahedron Lett., 1980, 21, 541.

97. K. Kimura, T. Utsumi, T. Teranishi, et al., Angew. Chem. Int. Ed., 1997, 36, 2452.

85. S. Akabori, T. Kumagai, Y. Habata, and S. Sato, J. Chem. Soc., Chem. Commun., 1988, 661.

98. M. Inouye, Y. Noguchi, and K. Isagawa, Angew. Chem. Int. Ed., 1994, 33, 1163.

86. S. Shinkai, T. Nakaji, T. Ogawa, et al., J. Am. Chem. Soc., 1981, 103, 111.

99. H. Sakamoto, H. Takagaki, M. Nakamura, and K. Kimura, Anal. Chem., 2005, 77, 1999.

87. S. Shinkai, M. Miyazaki, and O. Manabe, J. Chem. Soc., Perkin Trans 1, 1987, 449.

100. K. Kimura and Y. Nakahara, Anal. Sci., 2009, 25, 9.

88. M. Takeshita and M. Irie, J. Org. Chem., 1998, 63, 6643.

101. T. Sakata, D. K. Jackson, S. Mao, and G. Marriott, J. Org. Chem., 2008, 73, 227.

89. A. Ueno, H. Yoshimura, R. Saka, and T. Osa, J. Am. Chem. Soc., 1979, 101, 2779.

102. S. Kumar, D. Hernandez, B. Hoa, et al., Org. Lett., 2008, 10, 3761.

90. T. Aoyagi, A. Ueno, M. Fukushima, and T. Osa, Macromol. Rapid Commun., 1998, 19, 103.

103. M. T. Stauffer, D. B. Knowles, C. Brennan, et al., Chem. Commun., 1997, 287.

91. A. Mulder, A. Jukovi´c, F. W. B. van Leeuwen, et al., Chem. Eur. J., 2004, 10, 1114.

104. K. Kimura, M. Kaneshige, and M. Yokoyama, J. Chem. Soc. Chem. Commun., 1994, 1103.

92. K. Kimura, T. Yamashita, and M. Yokoyama, J. Chem. Soc., Chem. Commun., 1991, 147.

105. K. Kimura, R. Mizutani, M. Yokoyama, et al., J. Am. Chem. Soc., 2000, 122, 5448.

93. K. Kimura, T. Yamashita, and M. Yokoyama, J. Chem. Soc., Perkin Trans. 2, 1992, 613.

106. R. M. Uda, M. Oue, and K. Kimura, Chem. Lett., 2001, 1236.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc100

Supramolecular Chemistry in Medicine Stephen Faulkner1 and Alan M. Kenwright2 1 2

University of Oxford, Oxford, UK University of Durham, Durham, UK

1 Introduction 2 Supramolecular 3 Supramolecular 4 Supramolecular 5 Supramolecular 6 Conclusions References

1

Chemistry in Medicine Approaches to Payload Delivery Chemistry in Drug Discovery Chemistry in Diagnosis

1 2 4 12 15 15 15

INTRODUCTION

Our objective is to provide an overview of the emerging role of supramolecular chemistry in medicine at the time of writing. Any such attempt is hazardous, since it is all too easy to include current research that turns out to be next year’s dead end and overlook research that turns out to be next year’s hot topic. Nevertheless, we hope that this chapter will provide some idea of the current role of supramolecular chemistry in medicine and perhaps encourage others to think about a wider range of medicine-related research in terms of supramolecular chemistry. Our aim is not to provide a detailed account of all the aspects of supramolecular chemistry that have a role in medicine (that is beyond the scope of this chapter, and in any case, many of them are covered in greater detail in other chapters, in particular, Self-Assembly of Nucleic Acids, Viruses as SelfAssembled Templates, Peptide Self-Assembly and Magnetic Resonance Imaging Contrast Agents, Supramolecular Devices) but rather to give a broad overview of the field

and then show how recent developments in supramolecular chemistry are affecting medicine and highlight some possible directions for further development by examples from the contemporary literature. For this reason, many of the references in this chapter are to recent reviews. Where reference is made to primary literature, it is for illustrative purposes and does not imply that the work cited is somehow definitive, or that other work in the same field that is not cited is negligible. In this context, there is considerable overlap between a number of fashionable areas of chemistry, namely, “SMC (supramolecular chemistry),” “nanomaterials,” “smart materials,” and “soft matter.” We are not too concerned with definitions and boundaries (which tend to be ill-defined in many cases) when dealing with their applications in medicine. If the reader feels at some stage that we have strayed beyond the proper definition of SMC in choosing an example, they should look at the underlying principle that is being exploited in the context of medicine. It is almost certain that the same principle occurs in another example that does fit the reader’s definition of SMC. Many of the examples cited are still at the “proof of concept” stage and may never see widespread use because of the difficulty/cost of large-scale production but are still of interest because the principles they demonstrate will underpin developments in coming years. The role of supramolecular chemistry in medicine can be viewed in two ways. The first involves the rather narrow view of the ways in which current understanding of supramolecular chemistry can be used to help synthesize new drugs and/or drug delivery systems (DDSs). This might be termed synthetic supramolecular chemistry in medicine. The second involves the rather wider view that many of the key interactions in biological systems are supramolecular. Examples include the interaction between base pairs in DNA, the formation of cell membranes, cell recognition by binding between glycoproteins and lecithins,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

2

Supramolecular devices

protein–protein binding, and specific protein-binding events with agonists and antagonists (lock and key hypothesis). Since it is typically one or more of these processes that the medicinal chemist wishes to influence, it is worth considering that whatever entity the medicinal chemist produces, from small molecule to large covalent polymer or nanoparticle, its interaction with the body will at some stage be primarily supramolecular. Therefore, the tools of supramolecular chemistry become the tools for controlling and optimizing that interaction. In this chapter, we look at examples from both points of view.

(a) Guest

Host

(i)

Flexible guest

Rigid host

Flexible guest

Flexible host

(ii)

2

SUPRAMOLECULAR CHEMISTRY IN MEDICINE (b)

The term supramolecular chemistry was coined nearly 40 years ago and has been used to describe “chemistry beyond the molecule,” or the understanding of how molecules interact with one another without forming strong chemical bonds. This quest for understanding has much in common with medicinal chemistry, where drug action and localization are usually mediated by intermolecular interactions and do not result in the formation of chemical bonds. Many of the underpinning hypotheses of medicinal chemistry presage those of supramolecular chemistry by almost a century. In this chapter, we aim to show how the two disciplines underpin and cross-fertilize one another and also draw attention to the ways in which supramolecular approaches to medicine are becoming ever more effective.

2.1

How the principles of supramolecular chemistry underpin medicinal chemistry?

We begin by briefly definining some of the key concepts of medicinal chemistry and attempting to show how they relate to the underpinning ideas of supramolecular chemistry. The lock and key hypothesis1 represents an early attempt to rationalize the behavior of drugs and enzyme substrates. Essentially, a small molecule plays the role of the key, while a receptor or enzyme cavity acts as the lock. Initially, it was believed that the best drug molecules were those with a perfect fit into the cavity (Figure 1a) and that inactivity could be ascribed to a poor (or nonexistent) fit. In time, it became clear that this picture represents an oversimplification and that interactions between the cavity and a drug molecule will induce a change in the shape of the cavity (Figure 1b), provided that the interaction between the two is sufficiently strong. Supramolecular chemistry has built up a very similar picture of the interactions between a generalized host–guest pair, although it is only relatively recently that the importance of induced fit and

Figure 1 Pictorial representation of the lock and key hypothesis, showing binding of a drug molecule with (a) a “perfect” fit and with (b) an “induced” fit.

conformational changes has become important in the more general arena of supramolecular chemistry.2 Receptors on cell surfaces mediate their communication with one another and can control their behavior through interactions with their natural substrates. Drug molecules can also interact with these receptors to influence the behavior of the system as a whole. Agonist molecules bind to the receptors and trigger a response, while antagonists bind and cause no direct response. Instead, the medicinal effect of antagonists arises from inhibition of receptor binding. In both cases, recognition of the receptor binding domain plays a key role in determining the physiological outcome and the thermodynamics and kinetics of binding. The DNA double helix is probably the most widely recognized supramolecular architecture. Its structure is represented in Figure 2. Two complementary strands interact with one another by hydrogen bonding between complementary base pairs (G-C and A-T), and the double-helical structure is stabilized by base-stacking interactions.3 Since the structure is stabilized by myriad small interactions, it can accommodate some mismatches between base pairs (e.g., through G-T pairing); however, since DNA encodes biological and biochemical function, repair mechanisms are essential to ensure that cell function is maintained. A wide range of such mechanisms exist, including direct repair, excision and repair of bases (or whole nucleotides), and cross-link repair.4 Further response can involve arresting the cell cycle progression to allow repair and transmission of damaged, or incompletely replicated, chromosomes or programmed cell death (apoptosis) in the case of seriously damaged sequences.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

Supramolecular chemistry in medicine

Major groove

Minor groove

Hydrogen Oxygen Nitrogen Carbon Phosphorus

T

A

C

G

Pyrimidines

Figure 2

3

Purines

Structure of DNA.

2.1.1 Surface recognition Interactions between proteins and between proteins and cell surfaces govern a vast array of processes, from electron transfer to cell adhesion. Such interactions are clearly supramolecular and are driven by changes in solvation at the protein surface and by spatially matched interactions between the two surfaces. Two protein surfaces that interact strongly with one another will have complementary domains where supramolecular interactions, and particularly electrostatic interactions, dipole–dipole interactions, and hydrogen-bonding domains, are all maximized when the two protein surfaces are attached to one another. Furthermore, the recognition process liberates solvent from both surfaces, providing a highly favorable entropic contribution. In real systems the flexibility of protein structures adds another dimension to the problem of surface recognition. Just as with the interactions between proteins and small molecules discussed above, the induced-fit hypothesis is an invaluable concept for understanding interactions between protein surfaces. The balance between induced fit and conformational selection remains a topic of considerable interest in biology and is potentially of considerable importance in the development of the next generation of pharmaceutical agents. Kovacs et al.5 have pointed out that game theory

models can be applied to understanding the interactions between two proteins; in binding between two proteins, the conformational arrangement of one protein serves to define the environment of the other (and vice versa). Induced fit can be thought of as being similar to a well-defined game in which the strategy of one player is defined by the last move made by the other player. In this case the available “moves” involve changes in conformation or folding (and unfolding) that can occur either parallelly or unilaterally. The analogy has been taken further to describe a “hawk-dove” game, in which flexible proteins are doves and rigid ones are hawks6 and in which the payoff is an overall decrease in free energy. In such a model, no binding is possible if both partners are rigid proteins (since the complementary interactions between them cannot be maximized without a change in shape). Induced fit corresponds to an encounter between a (more) rigid protein and a (more) flexible protein, in which the enthalpy gain of the overall process is offset by some loss of freedom of motion (i.e., entropy) in the more flexible protein and augmented by desolvation of the binding domains. An encounter between two flexible proteins leads to an essentially similar outcome, but both will lose freedom of motion: such a process can be regarded as conformational selection. Csermely, Pallotai, and Nussinov7 have pointed out that the induced-fit hypothesis also resembles an ultimatum game, in which the first player

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

4

Supramolecular devices

(the more rigid protein) proposes how to divide the energy gain and the second can either accept the proposed split (in which case binding will occur), or reject it. They also point out that induced fit is essentially a subset of an extended conformational selection model in which Chettaoui et al.8 have proposed that multiple binding cascades can be modeled by “games network theory” and have used the ideas of an interconnected network of different games and players to model the plasminogen activator network, which involves interactions between seven different proteins. Inhibition of protein–protein interactions offers a potential target for medicinal chemists. Hamilton et al. have illustrated this approach through the synthesis of a range of calixarene derivatives that bind to and inhibit chymotrypsin.9 The calixarene core in these molecules is appended with anionic peptide domains, which are spatially matched to positively charged domains on the chymotrypsin surface, resulting in strong electrostatic attraction. This approach has recently been extended to self-assembled systems that utilize a quadruplex DNA core.10 Equally, stabilizing protein–protein interactions has considerable therapeutic potential. For instance, p53 is an important cell cycle protein involved in programmed cell death (apoptosis), which associates to form a tetramer in vivo.11 Furthermore, p53 mutations are commonplace in a wide variety of cancers.12 Giralt et al. have investigated the development of protein surface recognition sequences that are rich in positively charged residues and bind to anionic domains in monomeric p5313 and concluded that such an approach offers considerable potential for targeting the cell cycle in cancerous cells.

2.2

Potential advantages of supramolecular assemblies in medicine

While it is possible to achieve much by way of curing disease in a test tube (or a cultured cell line), the situation in whole animals is much more complex. Clearly, it is never going to be possible to achieve differentiation between cell types in a system where the rate of localization is slower than the rate of excretion. Similarly, in the case of radiotherapy, localization must occur quickly enough to ensure that the bulk of the ionizing radiation is discharged at the target, and not while the drug is being transported through healthy tissue. Supramolecular chemistry has much to offer here too. We see in subsequent sections how dendrimers and cyclodextrins can be used to protect drug molecules and control their localization kinetics, ensuring that they are not metabolized before reaching their target. In such systems, thermodynamics also play a big part in that effective drug action depends on the existence of an equilibrium in which the drug can dissociate from its host.

We have also already seen how self-assembly is exploited in the interactions between receptors and drugs, so essentially the effectiveness of such delivery systems depends on the perturbation of equilibria. Occasionally, transport effects work in favor of the medicinal chemist. Maeda observed that passive targeting is possible in cases where tumor tissue is differentiated from normal tissue14 and postulated the EPR (enhanced permeability and retention) effect.15 In such circumstances, the slow transport of macromolecules works to advantage, and a polymer-bound drug (or nanoparticle-bound drug) will accumulate in tumor interstitial fluid because of increased vascular permeability of the platform and increased retention at the tumor site. In Maeda’s case, neocarzinostatin was bound to a styrene–maleic anhydride, and the bioconjugate was found to localize in tumors, giving tumor: blood concentration ratios in excess of 2500 : 1. This effect has subsequently been exploited extensively, utilizing systems in which localization at the tumor site is followed by enzymatic cleavage of the drug molecule from the polymer support.16

3

SUPRAMOLECULAR APPROACHES TO PAYLOAD DELIVERY

Efficient payload delivery is one of the key goals of contemporary medicinal chemistry. In this sense, the payload may be diagnostic (e.g., a marker for imaging) or therapeutic (a drug or gene). For simplicity, we begin by considering the case where the payload is a small drug molecule. Traditionally, the pharmacokinetics of small-molecule drugs have been described in terms of the ADME model, referring to the discrete rates of absorption (the process of the drug entering the blood circulation), distribution (the process of the drug being distributed through the body), metabolism (transformation of the drug into daughter metabolites), and excretion (elimination of the substances from the body). In some more recent descriptions, the term Liberation (release of the drug from its formulation) is also included, leading to the so-called LADME model.17 Pharmacokinetics is a major field of study in its own right, but the model outlined above allows a qualitative discussion for (and by) nonspecialists. However, we should not take the view that this is simply a matter of balancing a few straightforward kinetic equations. One major complicating factor is the question of biodistribution—where does the drug go in the body? How much arrives at the site where it is required at any particular time? This in turn depends on factors such as the location and/or route of administration (oral, transdermal, parenteral, etc.), blood flow (tissues with the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

Supramolecular chemistry in medicine highest blood flow receive the drug more readily), the ability of the drug to cross membranes (lipid-soluble nonionic drugs find it relatively easy to cross lipid membranes but have very low water solubility, so are likely to be present in the bloodstream in very low concentrations), and the tendency of the drug to bind to common proteins such as serum albumin. In the case of warfarin, for example, the bulk of the drug present in the bloodstream at any time is protein bound and inactive. This slows and limits the biodistribution but also slows the rate of clearance of the drug from the body, thereby increasing the time over which the drug is effective. Since protein binding is itself a supramolecular interaction, the scope for intervention using supramolecular chemistry is obvious. The question of what proportion of the drug arrives at the required site in an active form is vital since most drugs are capable of interacting with more than one site and producing unwanted side effects. Keeping the dose of drug as low as possible while still producing the desired therapeutic effect not only makes good economic sense but can also be vital to the well-being of the patient. In extreme cases, such as some forms of chemotherapy, current doses are calculated on the patient’s ability to tolerate the side effects rather than on optimal therapeutic dose. The incorporation of the drug into a supramolecular construct can affect any of the above-mentioned rates and significantly change the effective dosage of the drug in question. It is probably worth considering at this stage what would be the ideal situation in drug delivery. We would like a drug to be easily administered (preferably orally); to rapidly achieve a favorable biodistribution, crossing any necessary barriers such as membranes; to interact specifically only with the desired target in order to minimize both the necessary dose and any possible side effects; to have a slow rate of clearance before interaction with the target in order to maximize the potential for interaction, but for unwanted daughter metabolites postinteraction to have a relatively rapid rate of clearance. Since the molecular interactions involved in supramolecular systems are typically weak and reversible, they are very attractive candidates for the design and production of socalled DDSs, which would allow improved spatial and temporal biodistribution. In the following sections, we look at examples showing how SMC can assist in approaching this ideal.

3.1

Complexation, delivery, stabilization, and biocompatibility

Supramolecular chemistry offers a wide range of possibilities for influencing drug delivery. We begin by looking at some simple, discrete effects that can be achieved

5

using supramolecular chemistry, before going on to look at examples using these effects. Complexation is simply the process by which a drug can be included in a supramolecular construct of some kind. While this may not initially seem a very important process, it can have major effects on pharmacokinetics and biodistribution. For example, the inclusion of a lipophilic drug in a construct that has a hydrophilic surface will completely change the distribution of drug through the body. Similarly, the complexation of a drug that would not normally be able to permeate cell membranes into a construct that can will significantly alter the biodistribution. This leads to the second effect that is of interest, namely, delivery. Since we are dealing primarily with weak and reversible interactions in SMC, there is scope for changing and controlling the rate at which the complexed drug is released, either in a general, long-term way (controlled release leading to sustained dosage levels) or by specific interactions on arrival at the target (targeted delivery). Targeted delivery in turn allows for the prospect of increased stability of drugs or other payload, which would normally be rapidly metabolized and excreted, such as small peptides or even short sequences of DNA. These can exist with much greater stability within the supramolecular assembly and then perform their therapeutic role when released at the appropriate target. Finally, inclusion as part of a supramolecular complex can significantly alter the biocompatibility of the payload since some such complexes are capable of crossing the cell membrane to deliver the payload inside the cell even though the payload on its own would not be able to cross the membrane. In the following sections, we look at some of the current approaches to payload delivery involving SMC.

3.2

Cavitands and dendrimers

Cavitands are organic host molecules having structures that contain an enforced cavity capable of forming host–guest complexes with other molecules. Common examples include calixarenes, cyclodextrins, and cucurbiturils. Most of the interest within medicine has tended to focus on cyclodextrins largely because they are relatively easy to prepare in pure form and because of their excellent biocompatibility. Cucurbiturils have received much less attention, largely because they were much more difficult to obtain in pure form, but recent improvements in methods of purification, particularly for some of the intermediate sizes such as cucurbit[8]uril, mean that this may soon change. Cyclodextrins are composed of cyclic rings of glucose units linked head to tail and occur naturally as degradation products of starch and so are readily available in spite of their apparent structural complexity. The structures of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

6

Supramolecular devices OH O OH

OH O

O

O

HO

O OH HO

HO

O

O

OH

HO

O

O

HO

OH

O

OH HO

HO

HO

O

HO

OH

HO

O

O

O

OH

b−CD

OH O

HO

O

O

OH

OH

HO

OH

O

OH

OH

HO

OH

OH

HO

O

a−CD

HO

HO

O

HO

OH O

O

O

O

OH

HO

OH

O

O HO OH O HO

O

HO

O

OH HO

OH OH

O

OH

O

O

O

HO HO

OH O

OH

g−CD

HO

HO

OH O

HO

OH

O

HO

OH

OH

HO

O HO

O

O

OH

O

O O HO

Figure 3

Cyclodextrin structures.

common cyclodextrins are shown in Figure 3. The most commonly used form is the one containing seven glucose units (so-called β-form). The –OH groups on the glucose units provide a hyrdrophilic exterior, giving good water solubility, while the interior of the ring provides a hydrophobic environment suitable for binding small lipohilic molecules. They have been used for a long time in pharmaceutical formulations to mask unpleasant odors and flavors, increase water solubility, stabilize the drug, and so on.18 However, simple complexation of small-molecule drugs with cyclodextrin

can greatly increase drug solubility and drug activity. A good example of this effect is shown by the encapsulation of a transplatinum complex as a possible anticancer drug.19 More recently, interest has shifted to the use of cyclodextrins as reversible “cross-linkers” in larger supramolecular structures. In this context, there is particular interest in the interactions of cyclodextrins with polyethers such as polyethylene glycol (PEG) to form self-assembled structures such as rotaxanes. Examples include the preparation of polypseudorotaxanes of PEG derivatives of insulin and cyclodextrins for use as a sustained release system.20

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

Supramolecular chemistry in medicine

Cucurbit[5]uril

Figure 4

Cucurbit[6]uril

7

Cucurbit[7]uril

Structures of cucurbit[n]uril for n = 5, 6, and 7.

The formation of polypseudorotaxanes with polyethers has seen considerable interest21 because of the ability of the polypseudorotaxanes formed to interact to form supramolecular hydrogels. These have the interesting property that under conditions of high shear they have relatively low viscosity, making them injectable, but when the shear is removed they re-form as high-viscosity gels, which limits their biodistribution and leads to slow payload release. In this context, as in many others throughout this chapter, the ability to produce a wide range of derivatives, in this case of cyclodextrin, cleanly and in good yield is key to the technology. This is true not just of the ability to produce cyclodextrin derivatives but also of the ability to cleanly incorporate reactive tags into a wide range of biomolecular targets. It is worth noting that the ability to carry out such reactions has been greatly enhanced in recent years by the development of clean, selective reactions such as olefin metathesis and click chemistry.22 A wide range of cyclodextrin-based supramolecular assemblies are now available,23 and in particular, the attachment of cationic polymers such as polyethylenimine has facilitated transport across cell membranes, permitting targetted delivery of both siRNA24 and DNA.25 Cucurbiturils are cyclic structures consisting of a number of glycoluril units, each linked to the next by two methylene bridges. They are distinguished by reference to the number (n) of glycoluril units in the structure, where n is typically between 5 and 10, with a particular structure being referred to as cucurbit[n]uril or CB[n]. The form of the cyclic structure resembles a barrel that is open at both ends and that is decorated around the rims by (polar) carbonyl groups. This makes them excellent candidate hosts for the formation of host–guest complexes. The inside of the barrel cavity is hydrophobic, but the molecule overall is not, and the presence of the carbonyl groups around the rims of

the barrel offers interesting possibilities for interacting with charged species. In general, it is found that the stability constants for cucurbituril–guest complexes are consistently larger than the stability constant for the cyclodextrin–guest complex of the same guest molecule, with the difference being particularly marked when the guest is cationic.26, 27 Structures of different-sized cucurbituril molecules are shown in Figure 4. The mode of binding in cucurbiturils and their ability to act as pseudorotaxanes is exemplified by the structure shown in Figure 5, showing the host–guest complex between cucurbit[6]uril and a para-xylenediammonium ion. The real power and attraction of cucurbituril complexes in payload delivery is twofold. First, they can be triggered to release payload in a variety of ways, but in particular, by the

Figure 5 The host–guest complex between cucurbit[6]uril and a para-xylenediammonium ion.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

8

Supramolecular devices

G4

Generation numbers

G3 G2 Core

G1 G0

Branching points

Termini

Dendrimer

Figure 6

Focal point (chemically addressable group)

Dendron

Schematic dendrimer and dendron structures.

formation of competing (stronger) complexes, particularly in the presence of cationic species such as metal ions or ammonium ions. One interesting recent example shows how the presence of metal ions can either trigger release of payload or stabilize the host–guest complex, depending on the host–guest stoichiometry.28 The fact that a range of different-sized cucurbituril molecules are becoming more readily available offers greater scope for tuning the interactions with a particular guest molecule. Second, the outer surface of the cucurbituril molecules can be relatively easily derivatized, offering possibilities both for increasing biocompatability and for inducing cell recognition. Recent work in this area has included publications by the Kim group that demonstrate the ability to use modified cucurbiturils to selectively deliver hydrophobic drugs to the interior of cells by endocytosis29 (in this case the delivery of plaxitel to human ovarian carcinoma cells) and the ability to induce specific gene transfection in human hepatoma-derived cells.30 The term “dendrimers” was originally applied to a relatively new class of monodisperse hyperbranched polymers. The first examples were produced only in the late 1970s and early 1980s by the groups of Vogtle and Tomalia, respectively, but they have now grown into a substantial field in their own right, with applications in medicine, biosciences, and nanotechnology.31 The basic structure of an idealized dendrimer is a monodisperse, regularly branched polymer. Typically, they may be thought of as being made by polymerization of a monomer that has one functional group of type A and two functional groups of type B, which are capable of reacting with groups of type A (a so-called AB2 monomer). If such a monomer is repeatedly reacted in such a way that every B group reacts with an A group, it produces a regularly branched wedgelike structure, which is often referred to as

a dendron. If multiple dendrons (typically between two and four) are linked to a central core, the resulting globular molecule is referred to as a dendrimer. The relevant structures are shown schematically in Figure 6. From the point of view of supramolecular medicine, dendrimers have a number of interesting properties. Most of the available functionalization is on the surface, which must, by definition, be covered in B groups or by species other than the original AB2 monomers, which have been reacted with the B groups to coat the surface. This means that the surface is highly multivalent, having a high density of the multiple copies of the same group and so can interact strongly with other surfaces (the concept of multivalency is discussed more fully in Multivalency, Concepts). However, the architecture of a dendrimer also means that the surface is relatively crowded compared to inner layers, so we can have a closely packed hydrophilic surface and a relatively loosely packed hydrophobic core. This leads to the possibility of using dendrimers as solubilizing agents for hydrophobic small-molecule guests. Since dendrimers are made up of layers of the same (branched) molecule, each successive layer contains twice as many AB2 units as the previous one. The usual practice is to refer to the number of layers in terms of G numbers (short for generation numbers). A dendrimer consisting of just one layer of AB2 units attached to a core is referred to as G-1. A dendrimer with a further layer added is G-2, and so on. Clearly, the number of generations that can be added will potentially be limited by steric effects, but depending on the spacing and flexibility between branching points, it is possible to produce dendrimers up to G-6 and beyond. Larger dendrimers (such as G-5) are typically of an appropriate size to interact directly with cells, and many dendrimers have the ability to permeate cell membranes, providing the possibility of using them for intracell

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

Supramolecular chemistry in medicine payload delivery. In this context, the fact that dendrimers have well-defined structures is important—their biodistribution and cell membrane interaction characteristics can be tuned by varying their size (generation number) and surface functionality/charge. Generally, larger dendrimers have lower biopermeability and high in vivo clearance rates via renal excretion. Polar surface groups such as alcohols and ethers (PEG) increase biocompatibility and are generally nontoxic. Negatively charged surface groups such as carboxylic acids do not interact with most cell surfaces, but positively charged surface groups such as amines interact strongly with most biological membranes. Small dendrimers with amine surface groups (G-3) damage the membrane, resulting in cytotoxicity.32 Putting these properties together, one possible mode of use of dendrimers is as nanoscale containers hosting drugs or other payload in their center—providing solubilization, improved pharmacokinetics (slow release/clearing), membrane permeability, and so on. They also allow the possibility of targeting and/or responsive behavior by surface modification. The external surface of a dendrimer is, by definition, multivalent and can be used as a nanoscaffold, being multiple functionalized by, for example, attaching multiple copies of a drug or contrast agent to the surface. Surface attachment can be covalent or noncovalent (e.g.,by attaching a number of cyclodextrin molecules to the surface of a low-generation dendrimer as a scaffold and then using the cyclodextrin to reversibly bind payload). A further advantage derives from the conformational flexibility of most dendrimer structures. This allows self-optimizing multivalent receptor binding.33 The concept of the dendrimer and the related concept of the dendron (regularly branched wedge) have found such widespread application that research has now evolved to include structures that include these motifs as elements, such as dendronized polymers and other hybrid structures.34 Their potential to produce well-defined supramolecular structures of use in payload delivery has been recognized for a number of years.35 Of particular interest is the emergence of structures made by linking two or more dendrons having different surface functionalities,36 to produce synthetic amphiphiles that can self-assemble to give a wide variety of controllable liposome structures suitable for drug delivery.37 The fact that dendrimers are typically of a suitable size to interact with established biological pathways and can be readily functionalized to tailor their recognition means that there has been a large amount of research aimed at producing dendrimer-based materials with very specific interactions in vivo, which have been the subject of a number of reviews.38, 39 The specific features that need to be considered when designing dendrimer-based systems for

9

use in medicine are now reasonably well understood (effects of generation number, surface functionality, hydrophobicity of the interior, kinetics of payload delivery, and clearance of dendrimer after payload delivery), and a number of dendrimer-based systems are now finding their way into clinical use.40 A substantial number of specific applications in medicine are covered in the extensive review by Astruc et al.31 In addition to their use in simple payload delivery, the fact that dendrimers can interact directly with biological systems means that they are also the subject of interest because of the response they can provoke themselves and so are the subject of research in the field of vaccines and immunostimulatory agents.41

3.3

Micelles, liposomes, and vesicles

Micelles are small supramolecular structures constructed from amphiphilic molecules such as lipids. In aqueous media they consist of a hydrophilic outer surface and hydrophobic core and typically fall in the size range 5–100 nm. As such, they have the potential to transport lipophilic molecules through aqueous media. Vesicles are small, membrane-enclosed sacks produced by the self-assembly of amphiphilic molecules, often in a double layer. Indeed, the cell membrane may be regarded as a (sophisticated) vesicle consisting of a bilayer of phosphalipids. Synthetic vesicles constructed from phosphalipids and closely related materials are often referred to as liposomes. Since they contain a finite volume of solvent (typically water) at their core, they have the potential to transport either hydrophilic payload molecules dissolved in the water they contain or lipophilic molecules dissolved in the hydrophobic bilayer, or, potentially, both. Their size can vary considerably, but it has been found experimentally that diameters around 100 nm are optimal for delivery of chemotherapeutics to tumors. One attractive feature of liposomes, in common with other self-assembled structures, is that tailoring of the constituent molecules allows the chemist a degree of control over the resulting supramolecular structures. Representative structures are shown schematically in Figure 7. A good example of this is given in the work of Percec and coworkers previously referred to in Section 3.2,37 where the authors could achieve good control not only over the average size of the vesicles produced but also over the polydispersity (range of sizes). The size of any drug delivery vehicle can have a major effect on its biodistribution and, in particular, on the rate at which the delivery vehicle is cleared from the active site. In cases where a drug delivery vehicle (supramolecular or otherwise) is used, the pharmacokinetics become more complex since, having reached the active site, the drug delivery vehicle

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

10

Supramolecular devices

Liposome

Micelle

Bilayer sheet

Figure 7 Common self-assembled structures formed from phospholipids.

needs to stay there long enough to release its payload before being cleared. It has been found that the size of the delivery vehicle has a major effect on this rate in tumors (as discussed above), and this has been measured for a range of drug delivery vehicles, including large micelles and liposomes.42 The size of the effect can be significant in that large micelles or liposomes may have a retention time in a tumor measured in weeks while the typical retention time of a small-molecule drug on its own would be measured in minutes. The use of micelles and liposomes as delivery vehicles of chemotherapeutics has recently been reviewed,43 and it is worth noting that a number of such systems are already in clinical use with more in trials and that they show both significant increases in dose potency of the drugs delivered and reductions in some of the common side effects of chemotherapy such as nausea and vomiting.

3.4

Polymers and hydrogels

Hydrogels are three-dimensional networks of fibrils that constrain the mobility of water molecules contained within them, largely by capillary interactions. The fibrils traditionally consist of polymers that are sparsely cross-linked to give a three-dimensional network. The cross-linking may be by chemical means (covalent bonds) or by supramolecular interactions. The amount of polymer present can vary widely, with hydrogels consisting of more than 90% water being possible. The physical and mechanical properties vary widely with the water content and cross-link density, so that, in appearance, hydrogels can range from

tough, rubbery materials to ultraviscous liquids. They have several attractive features that make them of interest in medicine. First, since they are by definition predominantly hydrophilic, they tend to have good biocompatibility. Second, it is possible to incorporate water-soluble payload into them simply by preparing the hydrogel in a solution that contains the desired payload. Subsequent introduction of the hydrogel into aqueous media will not result in the immediate release of payload since the rate of diffusion out of the hydrogel is extremely slow (often effectively zero on the relevant timescale). In view of these properties, hydrogels based on classic covalent polymers have seen use as drug delivery vehicles, where the slow biodegradation of the polymers causes drug release over a long period. This system is not ideal, however, since the rate of drug release is difficult to predict and control because the molecular weight distribution of the breakdown products is generally ill-defined. Nevertheless, they have been successfully used both as drug delivery vehicles and in a wide range of other medical applications,44, 45 including sensors and as scaffolds for cell growth. More interestingly, hydrogels can be formed by the supramolecular self-assembly of relatively small molecules. These can have attractive properties compared with gels made from chemically cross-linked polymers. In some cases they undergo shear thinning, which means that the material exists as a relatively stiff gel under ambient conditions but that the viscosity of the gel reduces significantly under shear. This means that the material can be injected through a hypodermic needle (high shear) but then spontaneously reforms a stable hydrogel that localizes the payload.46 One particular advantage is that self-assembled hydrogels produced from peptides can be made sufficiently biocompatible that they can carry living cells as payload and localize them in a specific area while at the same time protecting them from rapid clearance from the body. This makes them extremely useful in promoting tissue regeneration, and examples have been shown where hydrogels containing suitable combinations of cells and growth factors can greatly enhance the healing of various tissues and even promote healing of lesions in the central nervous system (spinal cord), which generally does not occur spontaneously. Another advantage of supramolecular hydrogels made from relatively low-molecular-weight gelators is that they allow better control of payload release, since degradation of the gel results in the production of known (low-molecularweight) fragments. The introduction of mechanisms to control the degradation of the gel provides for the possibility of triggered and/or controlled release of the payload. One example is provided by the production of gelators in which payload is incorporated into the low-molecular-weight gelator (LMWG) itself using a spacer that is susceptible to cleavage under specific conditions in vivo. The specificity

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

Supramolecular chemistry in medicine of the conditions for cleavage provides the scope for triggered release of the payload. For example, the cleavage may be enzymatic, and examples have been shown using proteolysis to cleave an amide linkage in the gelator. This is of relevance since proteolysis (enzymatic peptide cleavage) is often deregulated in the diseased state (compared with the healthy state).47 Since the structure–property relationship is very specific in drug molecules, it is important that the chemistry used to incorporate the drug molecule does not significantly modify the structure of the drug. One recent example of this approach included the use of “self-immolative” spacers to link the drug to the gelator,48 while a second approach at the “proof of concept” stage uses the modification of relatively common drugs such as acetaminophen (paracetamol) to produce LMWGs that can then be used to produce gels, which, in turn, can carry payload. This approach has the benefit that it can, in principle, be used to deliver two or more drugs in conjunction and that the metabolic pathways of the products of hydrogel degradation are already well known. Other mechanisms may also be used to trigger degradation of hygrogels and therefore payload release, making them “responsive” payload delivery vehicles. Many examples include triggers such as changes in pH or temperature, while a recent example that is particularly elegant from the point of view of supramolecular chemistry uses an inclusion complex between a trans-azobenzene group and β-cyclodextrin as a cross-linker to form a hydrogel.49 Subsequent irradiation with UV light at 365 nm causes isomerization to cis-azobenzene, which does not form a stable inclusion complex with the cyclodextrin and which therefore triggers breakdown of the hydrogel. It has been shown that the release of protein incorporated as payload can be successfully triggered in this way. Inevitably, it is also the case that there is no clear dividing line between the LMWGs discussed above and gels formed from what might be regarded as classic covalently bonded polymers. In particular, the use of hyperbranched and dendritic gelators is an area intermediate between the two extremes that has been the subject of interesting work.50 This is also an example of the difficulty of compartmentalizing topics in this area. It is also the case that a particular gelator may be sensitive to more than one parameter controlling its tendency to form a gel. In this context, a supramolecular hydrogel that is sensitive to a number of discrete stimuli has been produced and its behavior analyzed in terms of “logic gate” operations.51 It has been demonstrated that controlled release from the gel can be triggered by applying the appropriate combinations of stimuli. This approach offers considerable scope for fine-tuning the release response in such gels.

3.5

11

Interplay and interactions

The preceding sections have looked at major classes of structures often involved in medicinal supramolecular chemistry and provided examples highlighting each. This tends to give the impression of a number of disconnected fields and downplays the importance of the effects of supramolecular interactions in the wider context. In this section, we look at a number of examples that seek to highlight these effects. The first of these is the use of supramolecular interactions as a trigger or response. A relatively simple and localized example is the use of the pH-responsive supramolecular interactions of a tethered cyclodextrin moiety to control access to, and release from, hollow silica nanoparticles.52 In this case, the important supramolecular interaction is localized on a payload delivery vehicle and responds specifically to a particular environmental change (pH). While this is interesting and potentially important, it can be seen as exemplifying the “isolated” supramolecular interaction. Often of greater significance in biological systems and medicine is the simultaneous occurrence of multiple supramolecular interactions. These can lead to the selfassembly of large structures such as supramolecular polymers53 and, in particular, to the production of synthetic analogs of naturally occurring materials. This can be particularly useful in regenerative medicine where, for example, self-assembled nanofibres containing appropriate growth factors can be used to promote cartilage regeneration that does not occur naturally.54 However, the self-assembly resulting from supramolecular interactions can be exploited in a much wider variety of applications, some of which offer possibilities to achieve complex structures that could not be synthesized otherwise. One example of this is the behavior of small cyclic peptides composed of amino acids with alternating D and L stereochemistries. These self-assemble by stacking to produce tubelike structures that can insert themselves into cell walls, thereby killing the cell. Interestingly, it has been shown that this behavior can be made preferential for bacterial cells, thus providing antibacterial selectivity. The fact that the tubes are produced by self-assembly makes it relatively easy to produce a wide range of structures, and evaluation of these structures is ongoing. Although the basic idea of using self-assembled stacks of cyclic peptides goes back more than a decade, its application continues to be refined. A recent example demonstrates that attaching specific saccharide side chains promotes specific interaction with bacterial cell walls, thereby retaining potent bactericidal activity while reducing mammalian cell toxicity.55 This demonstrates how supramolecular interactions can be effective at more than one level in medicine. In this case, not

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

12

Supramolecular devices

only is the basic structure supramolecular in nature but also its specificity is determined by the supramolecular interactions between the attached saccharides and the bacterial cell walls. Another interesting example (among many) of the power of self-assembly is provided by recent work in which supramolecular interactions between an amide-containing vinylic monomer and a naturally occurring peptide result in the formation of monomer micelles templated around the peptide. Subsequent polymerization and cross-linking of the monomer followed by removal of the peptide and any unreacted monomer by extensive dialysis results in a polymer nanoparticle that contains a very specific peptide-shaped interaction site. It has been shown such nanoparticles function in living systems to specifically facilitate the removal of the relevant peptide from the blood stream.56 These molecularly imprinted polymer nanoparticles have been dubbed “plastic antibodies,” and along with the many possibilities they offer for medical applications, they also provide very elegant experimental confirmation of the validity of Fischer’s “lock and key” hypothesis. It is worth noting that the self-assembly templating makes it relatively easy to change the target peptide without significantly altering the method. At a still more general level, we need to consider the supramolecular interactions between synthetic bodies and naturally occurring biomolecules. Of particular concern is the interaction with proteins in the bloodstream, since coating of a particle with protein is often the first step in the body’s system for removal of the particle. The ability to control the rate at which this occurs has a critical effect on pharmacokinetics. One simple but important technique in this respect is the attachment of short chains of PEG to the surface of the particle or other structure to significantly increase water solubility, biocompatibility, and in vivo circulation time.57 However, in a setting as complex as a living system, few changes have only a single (desirable) effect and, in the case of PEG-ylation (coating of the particle with PEG) of antitumor drug delivery vehicles, an additional (undesirable) effect is that accumulation in the tumor because of the EPR effect is reduced. Recent work has seen refinement of this approach with high levels of PEG-ylation present initially in a system that progressively sheds PEG. This has the effect of slowing initial clearance of the payload delivery vehicle but encouraging localization of the payload in the tumor.58 It is worth reflecting that, while the complex processes involved in regulating pharmacokinetics in vivo cannot be changed simply, the tools for tuning the effects of those processes (degree of PEG-ylation, mechanism of PEG shedding, etc.) are entirely in the hands of the synthetic supramolecular chemist. Another example of the ability to tune complex interactions to our advantage is provided by recent work that uses binding to a specific protein of a micellar container made

from amphiphilic dendrimers. The binding of the protein triggers disassembly of the micelle, resulting in payload release.59 Since it is possible to select or tune which proteins bind to the micelle, and since overexpression of certain proteins is a characteristic of certain disease states, this provides very specific targeting based entirely on SMC.

3.6

Gene therapy

Once it was realized that many diseases had their origin in “defective” gene sequences, the idea quickly took hold that such diseases could be cured by replacement of the defective sequence by a “normal” one. Development in this area occurred incredibly rapidly and has been a cause of considerable debate on the ethical implications. This is particularly true of germline gene therapy (which involves genetic modification of eggs or sperm), where the effects of gene therapy can potentially eliminate heritable diseases but where the changes in gene sequence are also passed on to future generations.60 For the purposes of this discussion, we restrict ourselves to the discussion of somatic gene therapy, in which the sequence manipulation takes place in the somatic cells of a patient and the changes are not inherited by subsequent generations. The most common approach to gene therapy involves inserting a normal gene into a nonspecific site in the sequence, but other approaches have also been identified, including homologous recombination in which one gene is swapped for another, gene repair through reverse mutation, and controlling gene regulation. A carrier is used to deliver the gene to target cells in the patient before transfection into the cell and release of the DNA. Viruses have been found to be particularly effective carriers, but there have been significant concerns about their safe use,61 particularly when retroviral systems are used. Supramolecular chemistry using nonviral encapsulating agents has shown considerable promise, and a range of amphiphilic molecules has been used as agents in DNA delivery, including small molecules, nanoparticles, and polymers.62–66

4

4.1

SUPRAMOLECULAR CHEMISTRY IN DRUG DISCOVERY Supramolecular assays in high-throughput screening

Traditional immunoassay has used radiotracers to provide tags to study the interaction between an antibody and an antigen (Figure 8 shows the basic principles of such an assay).67 Despite their very high sensitivity (particularly

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc102

Supramolecular chemistry in medicine

Surface

Surface

R

Surface

R R R

R

R

R R

R R

R

R R

R

R

R R R

13

R Analyte is bound, linking some Labeled antibodies to the surface

R Antibody

Radiolabeled

Analyte

antibody

Figure 8

A typical radioimmunoassay.

when using scintillants to signal radioactive decay events), concerns about the use of radioisotopes have lead to a search for less controversial probes. Fluorescent probes have replaced radioisotopes in many conventional bioassays.68 In this case, it is not simply a question of locating a luminescent tag but rather of relying on combined recognition events coupled with energy transfer between two chromophores by fluorescent resonant energy transfer (FRET). However, the sensitivity of this technique is limited by autofluorescence from biological backgrounds. Leif and coworkers reported the first attempts to use europium complexes [Eu(phen)(diketonate)3 ] complexes as tags for antibodies.69 These proved kinetically unstable in the pH regime required for bioassay but did show long-lived metal-centered luminescence. Their results showed that the ideal lanthanide probe complex should combine a high extinction coefficient for the sensitizing chromophore with a relatively long wavelength absorption (ideally >340 nm, permitting the use of glass instead of quartz optics) and kinetic stability on the timescale of the experiment. These criteria are fulfilled by a number of commercial assays, including the DELFIA (dissociation-enhanced lanthanide fluorescent immunoassay) and CYBERFLUOR assays. The DELFIA (Dissociation Enhanced Lanthanide FluoroImmunoAssay)70 is a heterogeneous assay that uses a lanthanide complex based on aminocarboxylate ligands, such as EDTA (ethylenediaminetetraacetic acid), EGTA (ethylene glycol tetraacetic acid), or DTPA (diethylenetriaminepentaacetic acid), linked to the antibody by reaction of appended isothiocyanate groups with nucleophilic residues (particularly amines) on the protein surface. These ligands are useful in two respects. First, they bind tightly to the lanthanide ion at physiological pH, meaning that dissociation on the experimental timescale is not likely to be a problem. Second, the flexible ligand framework allows the europium to be removed and coordinated to when the assay is to be “developed.” Figure 9 shows the assay itself. A multiwell

plate coated with unlabeled antibody is allowed to react with the antigen. A known quantity of the lanthanideconjugated antibody is then added, and the mixture is incubated. When the target antigen is present, it acts to link the lanthanide-conjugated antibody to the surface of the plate, via the antigen, which is itself bound to the surfacebound, unlabelled antibody. Washing the plate removes any uncomplexed antibody, meaning that the remainder can be used to quantify the amount of europium present (and hence the concentration of the target antigen). This is achieved by lowering the pH to 50%) at reasonable module costs (100–150 US$ m−2 ). This type of cells corresponds to the region IIIa is shown in Figure 1. However, such parameters are hardly reached within next decades. Power conversion efficiencies of the best tandem solar cells got stuck at around 40% while their price remains much higher than the target 100–150 US$ m−2 in spite of the using inexpensive light concentrators.5 There is other alternative that is defined by the area IIIb is shown in Figure 1. This area corresponds to the solar cells yielding reasonably high power conversion efficiencies of 8–16% at very low module costs (40–60 US$ m−2 ). Low cost fabrication is envisioned for organic solar cells. Indeed, laboratory prototypes of organic solar cells demonstrated power conversion efficiencies exceeding 8% and approaching 11% in some cases (Table 1). Further improvements of organic solar cells in terms of performance, lifetime, module design, and production technologies might lead to a breakthrough in the renewable energies. At the end, the energy generated by solar light conversion should become cheaper than the energy we currently produce by combustion of fossil fuels. There are also many additional advantages of the organic thin-film solar cells that can be illustrated as follows: •

•

light weight nature of thin-film solar cells makes them ideal suiting for portable electronics applications; integration into cloths (power suits) and military canopies has been already demonstrated; and high sensitivity at low light intensities allows for indoor applications to collect scattered light (e.g., use them as decorative energy-generating wallpaper).

mechanical flexibility allows one to adapt them to any curved surfaces;

Different examples of organic solar cells have entered the phase of commercialization recently.29 In this chapter, we focus on the supramolecular and interface chemistry of organic semiconducting materials used to create the active layers in organic solar cells. In the first part, we give general information about different types of organic solar cells. The typical photoactive materials, device architectures, and operation principles are discussed. A short overview of supramolecular assembling effects in dye sensitized organic solar cells is given in a second part. The third part is dedicated to supramolecular chemistry of photoelectrochemical cells based on covalently and noncovalently linked donor/acceptor molecular architectures. Supramolecular ordering in solar cells based on small molecules, diblock copolymers, and fullerene/polymer blends is discussed later on. Special attention is paid to the interface effects on the anode and cathode side of the photovoltaic device. Finally, we provide an outlook concerning our vision of the future development of organic photovoltaics. The content of this chapter and provided citations are organized for didactical purposes only and do not reflect the chronology of the research in the field and/or have no claim of completeness. The further interested reader is referred to monographs addressing different aspects of organic photovoltaics.29–33

1.2

Schematics of different organic photovoltaic devices

The photoelectrochemical solar cells form the first family of organic photovoltaic devices. Typically, the active layer of such devices consists of nanostructured and dye sensitized electrodes, whereas the other electrode (counter electrode) is separated by an electrolyte or hole conductor. The highest efficiency of around 11% is achieved in “dye sensitized solar cells (DSSCs)” using TiOx nanostructured electrodes.34, 35 Schematic layout of a typical DSSC device is shown in Figure 2.36 The active layer of the DSSCs consists of a mesoporous nanocrystalline metal oxide (typically titanium dioxide or zinc oxide) deposited on fluorine-doped tin oxide (FTO) electrode and covered with some organic or organometallic dye. The dye molecules (D) absorb light while populating

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics Table 1

Some state-of-the-art material combinations for different types of organic solar cells.

Type of the cell

Dye sensitized solar cells (based on TiO2 photoelectrodes sensitized with various dyes)

Description of the materials

Power conversion efficiency (%) (EQE, %)

References

Ruthenium dyes Black dye HOOC COOH N

N

SCN

Ru N NCS SCN

11.1 (80–90)

6

11.4 (90)

7

7.7 (80)

8

COOH

“Ruthenium supersensitizer” COOH SC6H13 S

NaOOC N N Ru

N

SCN NCS N S SC6H13

Organometallic dyes Phthalocyanine dye TT1 + JK2

COOH

t Bu

N

N

JK2

N

N

N

Zn S

N

N N N

t Bu

S

t Bu

TT1

HOOC CN

(continued overleaf )

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

3

4

Supramolecular devices

Table 1

(Continued ).

Type of the cell

Description of the materials

Power conversion efficiency (%) (EQE, %)

References

Porphyrin dye

N

N

N

Zn

COOH

N

11.0 (90)

9

9.8 (85–90)

10

1.5 (54)

10

− (56)

11

N

Organic dye C217 dye

S S

N

CN

S HOOC

O

O

C217

Porphyrin + gold nanoparticles + C60 fullerene

Fullerene/porphyrin photoelectrochemical solar cells

Porphyrin-loaded peptide + C60 fullerene O H O H O H O H O H O H O H O H O H O H O H O H O H O H O H O NH O N N N N N N N N N N N N N N O N O O

NH

O

O

NH

O

O

NH O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

O

NH

O

m

Porphyrin

Fullerene

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

Table 1

(Continued ).

Type of the cell

Carbon nanotube-based photoelectrochemical solar cells

Description of the materials

Power conversion efficiency (%) (EQE, %)

References

Porphyrin self-assembled with single-wall carbon nanotube (SWCNT)

− (13)

12

0.11 (17)

13

−(32)

14

7.4 (68)

15

6.5 (65)

16

Stacked cup carbon nanotubes

Stacked cup carbon nanotubes + porphyrin Low band gap polymer + [70]PCBM O

HEO S

S

Fullerene/polymer bulk heterojunction solar cells

O

H 3C

n

S F S EHOOC

OEH EH=ethylhexyl PTB7

[70]PCBM

P3HT + bis(indeno)fullerene derivative

C6H13 S S C6H13 BisIndeno-C60

n P3HT

(continued overleaf )

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

5

6

Supramolecular devices

Table 1

(Continued ).

Type of the cell

Description of the materials

Power conversion efficiency (%) (EQE, %)

References

Evaporation processible CuPc + C60 fullerene

N

N

N

N Cu N

Small molecule/small molecule bulk heterojunction solar cells

N

N

4.4 (76)

17

5.2 (45)

18

4.0

19

N

C60

CuPc

Si N

Solution processible porphyrin + fullerene derivative

Si

HN

NH N

SIMEF

BP

Evaporation processible low band gap oligomer + C60 fullerene NC

Bu

CN

Bu

Bu

Bu

S

S

NC

CN

S S

S

C60

Solution processible low band gap molecule + [70]PCBM H3C

O

O

O N

S

4.4 (65)

O O N

S O [70]PCBM

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

20, 21

Supramolecular chemistry for organic photovoltaics Table 1

(Continued ).

Type of the cell

Description of the materials

Power conversion efficiency (%) (EQE, %)

References

Merocyanine dye + C60 fullerene CN

NC CH3 CH3

N

4.9 (73)

22

2.9 (65)

23

O HB194 C60

Hybrid bulk heterojunction solar cells based on the inorganic nanocrystals and conjugated polymers

P3HT + CdSe

Bulk heterojunction solar cells based on carbon nanotubes and conjugated polymers

SWCNT + P3HT C6H13 S S C6H13 P3HT

Bulk heterojunction solar cells based on solution processible functionalized graphene and conjugated polymers

0.22

24, 25

0.6–0.8

26, 27

n SWCNT COOX

R

C6H13

R

XOOC

O

S

R +

S C6H13

R

n

XOOC

COOX

P3HT SPFGraphene

C6H13 *

S

S

N S N S

S

Polymer/polymer bulk heterojunction solar cells

C6H13

n

n

*

1.8

P3HT F8TBT

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

28

7

8

Supramolecular devices e− E

e−

Load

Glass substrate dyed nc-TiO2

e−

Jsc

Pt

(D+/D*)

Redox electrolyte

e−

Ecb

e−

∆V hn

Eph = hn e−

I3−

I−

e− ∆VL

(D+/D*) D = N719

FTO

Figure 2

FTO

Schematic layout of a typical DSSC device. (Reproduced from Ref. 36.  American Chemical Society, 2009.)

the excited state (D*). These excitons dissociate rapidly via electron transfer from the excited dye to the titanium dioxide. The formed dye cations D+ are reduced back to the neutral D molecules by the redox electrolyte, for instance, I− anions. In this case, I3 − species are formed as a product. Injection of electrons from a counter Pt/FTO electrode converts I3 − back to I− anions. There is a vast variety of photoelectrochemical solar cells comprising different materials in the active layer. Most of them are based on the donor/acceptor composites assembled on the transparent conducting electrodes such as FTO or indium tin oxide (ITO).37 The highest power conversion efficiencies of approximately 1.5% were obtained using fullerene/porphyrin composites.10 These devices show remarkably high external quantum efficiency (EQE) of around 54% and a broad spectral response covering all visible range and extending to the near infrared region up to approximately 1000 nm. The open-circuit voltage (VOC = 0.4 V) and fill factor (FF < 40%) need further optimization. There is a special type of photoelectrochemical devices based on donor/acceptor dyads and triads self-assembled on the metal (gold) or metal oxide (ITO and FTO) electrodes.38, 39 A multistep photoinduced electron transfer occurring in such devices resembles in some way a natural photosynthetic process. The power conversion efficiencies of such devices are negligibly low because the monolayer of the photoactive material does not absorb much light. However, internal quantum efficiencies (IQEs) of such devices approach 80–90% in many cases.

With the development of organic semiconductors which support electron or hole transport (in analogy of p- and n-type inorganic semiconductors) bilayer type p–n heterojunction devices have been constructed. In the case when organic p- and n-type materials are deposited consecutively in two layers we get lateral heterojunction devices which mimic classical p-n junction junction solar cells based on silicon. The first such cell was designed by Tang and published in 1986 (Figure 3).40 Copper phthalocyanine was utilized as electron donor (p-type) material, whereas perylene derivative was used as electron acceptor counterpart (n-type material). Efficiency of planar heterojunction devices is limited by low exciton diffusion lengths in organic semiconductors that typically stay below 20 nm.41–45 This means that excitons generated more than 10–20 nm away from the interface between the p- and n-type materials do not contribute to the charge generation. As a consequence of this drawback, the power conversion efficiencies of the planar bilayer heterojunction devices typically stay in the range of 1.0–1.5% and approach 2–3% only in rare examples.46, 47 Revolutionary breakthrough was made when bulk heterojunction concept was invented.48–50 According to this concept, p- and n-type materials are mixed together and organized on nanoscale to form three-dimensional (3D) interpenetrating networks capable of efficient charge generation and transport. The bulk heterojunction solar cells have enormously large interface between p- and n-type materials compared to the planar heterojunction devices described above. For the optimal performance, a size of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

N

N

9

O

Ag N

N

N

PTCBI

N Cu N N

N

N

CuP

CuPc

O

N

In2O3 N

Glass substrate

PTCBI

(a)

(b)

Figure 3 (a) Molecular structures of the p- and n-type semiconductor materials and (b) layout of the planar heterojunction device designed by Tang.40

the interconnected grains formed by the phases of pristine materials should stay in the range of the exciton diffusion length (5–20 nm). Such morphology allows more or less all excitons generated in the active layer to reach the interface and contribute to the generation of free charge carriers. This is why, IQE of some bulk heterojunction solar cells approaches 100%.51 An idealized structure of organic bulk heterojunction solar cell is shown in Figure 4. Active layer of this device comprises interpenetrated phases of electron donors (yellow) and electron acceptors (gray spheres) capable of efficient charge transport toward the electrodes. In order to avoid charge recombination at the electrodes, some buffer layers of pristine materials (or some other charge selective materials as discussed later) should be introduced under the electrodes. For instance, positive electrode that extracts holes from the active layer should form a direct contact only with a p-type material. At the same time, just a pure phase of the n-type material should be adjacent to the negative electron-collecting electrode. Unfortunately, such arrangement of p- and n-type materials can be easily constructed only by physical vapor deposition (PVD). Solution processing requires some principally different buffer materials soluble in orthogonal solvents with respect to the active

−

e−

p

e−

p

e− e−

e−

p

p

p

e−

p

+

Figure 4 cell.

Schematic layout of an ideal bulk heterojunction solar

layer components, for example, water-soluble poly(3,4ethylenedioxythiophene) : poly(styrenesulfonate) (PEDOT : PSS)52 or titanium oxide TiOx processed from alcohol precursor solution.53 Intensive research has been devoted to the field of bulk heterojunction solar cells during the last decade. As a result, many different materials were found to be suitable for construction of bulk heterojunction solar cells. Among them are examples of polymer/polymer, polymer/small molecule, polymer/carbon nanotube,54 polymer/inorganic nanoparticle, as well as small molecule/small molecule composites.55 Overview of the most efficient material combinations designed for different types of organic solar cells is provided in Table 1.

1.3

Basic working principles of organic photovoltaic devices

Working principles of organic solar cells are well described in a recent review56 and some monographs.29–33, 57 More or less all types of organic solar cells described above comprise two components in the photoactive layer. One component serves as electron donor, whereas the other works as electron acceptor. Absorption of photons by the active layer components results in the electron transfer from donor to acceptor. This process called photoinduced charge transfer is a fundamental principle of operation of all known organic photovoltaic devices as well as the natural photosynthetic systems. In many cases, donor material is capable of efficient p-type transport and therefore can be called as p-type organic semiconductor. At the same time, electron acceptor material is denoted as n-type semiconductor in many cases. For instance, organic or organometallic dyes in the case of DSSC devices serve as an electron donor material. However, such dye has no role in hole transport and therefore

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

10

Supramolecular devices by organic dye molecules absorbed on the mesoporous titanium dioxide.58 Absorption of photons by donor or acceptor materials leads to the formation of corresponding excited species D* or A*. These excitons have to diffuse in order to reach the interface between the donor and acceptor materials where charge separation might take place (step II in Figure 5). Therefore, it is crucially important to adjust the morphology of these devices in way when more or less all excitons can reach the donor/acceptor interface and contribute to the charge generation.59 This can be achieved by keeping the size of the p- and n-type material domains below the exciton diffusion length which is 5–20 nm for organic materials.41–45 In the case of the DSSCs, the dye molecules are directly bound to the titanium dioxide that is the electron acceptor. Therefore, light absorption can be immediately followed by the charge separation.58 At the donor/acceptor interface (step III in Figure 5), exciton D* is quenched via electron transfer to the lowest unoccupied molecular orbital (LUMO) level of the acceptor molecule (A0 ). On the contrary, exciton A* is quenched via hole transfer to the highest occupied molecular orbital (HOMO) level of the donor molecule (D0 ). Both pathways result in the formation of the same charge separated state D+ · · · A− . Positive and negative charges in this ion pair are bond by Coulomb attraction forces and also denoted as “geminate polaron pair.”60–62 This pair can dissociate in the electric field induced by the potential jump at the heterojunction and/or by the difference in the electrode work functions. At the same time, the energy difference

cannot be treated as p-type organic semiconductor. On the contrary, titanium dioxide works as electron acceptor material and since it has excellent electron transport properties, it is an n-type semiconductor. In the case of photoelectrochemical devices based on the porphyrin/fullerene composites, the former functions as electron donor, whereas the latter is the electron acceptor. Bulk heterojunction organic solar cells are always composed of a blend of p- and ntype organic semiconductors. Conjugated polymers, metal phthalocyanines, conjugated oligomers, and some organic dyes are typical p-type components. Electron acceptor counterparts (n-type components) are usually represented by fullerenes or their derivatives, perylene diimides, naphthalene diimides, and conjugated polymers. A principle of operation of organic photovoltaic devices is shown in Figure 6. The charge generation mechanism is more or less the same for all types of organic solar cells. The first step (step I in Figure 5) is the light absorption in the active layer of organic photovoltaic device. In some cases, both donor and acceptor components harvest photons quite efficiently. For instance, this is true for composites of phenyl-C61 -butyric acid methyl ester [70]PCBM and low band gap polymers that have complementary absorption spectra.15 However, in some cases one component (usually this is electron donor material) is predominantly or even exclusively responsible for light harvesting. For example, titanium dioxide, as a wide band gap material (Eg > 3 eV), does not contribute much to the active layer absorption in dye sensitized solar cells. Therefore, all light is collected I LUMO

II

III

−

V

Hole transfer −

hn

p+

+

+

HOMO

Acceptor (A0)

LUMO

IV

D0

A*

−

hn +

− A*

Electron transfer − e−

+ D+

A−

+

−

+

−

− −

+

HOMO

Donor (D0)

D*

−

+

Geminate polaron pair

+ D*

A0

1

2

3

4

5

Figure 5 Operation mechanism of organic heterojunction solar cells. Step I: absorption of photons and generation of excitons; step II: migration of the excitons toward the donor/acceptor interface; step III: charge separation at the donor/acceptor interface and formation of geminate ion pairs; step IV: dissociation of the ion pairs and moving of the charges in the opposite directions in the internal electric field in the device; step V: collection of the charges at the electrodes in the device, 1—positive electrode (ITO); 2—electron blocking hole transporting layer (PEDOT : PSS, MoO3 , WO3 , V2 O5 ); 3—active layer of the device; 4—hole-blocking electron transporting layer (TiOx and ZnO); and 5—negative electrode (Al, Ag, Ca, Mg, and Ba). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

11

function metal (e.g., calcium, lithium, magnesium, barium) matches well the LUMO of the acceptor component and therefore should collect negative charges (electrons) easily, whereas high work function electrode (e.g., gold, platinum, nickel) matching the HOMO of the donor component should collect positive charge carriers (holes) without large barriers.56, 57 However, the situation is not that simple in real devices. For instance, ITO electrode can extract both positive and negative charges from the active layer of the device. At the same time, top electrodes composed of aluminum or silver can also extract both holes and electrons. Such poor selectivity of the charge collection results in a low photovoltaic performance of the device because of the massive charge recombination at the electrodes. To avoid this loss, some buffer layers should be introduced at the interfaces between the electrodes and the active layer. Electron blocking layers composed of vacuum processed (PVD), vanadium (V), molybdenum (VI), and tungsten (VI) oxides were extensively utilized.65–67 Titanium dioxide, cesium carbonate, zinc oxide, or fullerene derivatives behave as electron transporting and hole-blocking materials.53, 67–71 There is a number of alternative charge transport materials that can be adopted from the experience accumulated in the field of organic light emitting diodes (OLEDs).72 There are recent reviews summarizing information on the available buffer layer materials.73–76 Thus, bulk heterojunction cells

between the donor and acceptor LUMO levels (in the case if electron transfer takes place) or HOMO levels (in the case of hole transfer) released as a heat also helps to some extent with dissociation of geminate polaron pairs.63 The photoinduced free charges should be transported to the respective electrodes (step IV in Figure 5). In bulk heterojunction solar cells holes move in the phase of p-type material, whereas electrons are transported in the n-type counterpart. Therefore, it is essential to attain percolated pathways for charge transport in both phases (Figure 4). In the case of photoelectrochemical cells, the situation is rather different. As soon as the active layer of such cells is deposited on one of the electrodes, a single type of charges can be extracted directly and collected at the electrode. For instance, the titanium dioxide phase collects electrons and transports them efficiently to the negative electrode. However, other type of charges has to be transported toward the counter electrode by the ions of the electrolyte. The I− /I3 − couple is usually employed as a redox component of the electrolyte to take up this job.64 Organic dye methyl viologen capable of transformation from dication to monocation is used to transport electrons through the electrolyte of photoelectrochemical solar cells.38, 39 At the final stage (step V in Figure 6) of the photocurrent generation process charges have to be collected at the electrodes. In bulk heterojunction devices, the lower work

70 60

FF = (Imax × Vmax)/(Isc × Voc) 10 IPCE or EQE (%)

Current density (mA cm−2)

15

5 0 −5

Imax

Isc −10

40 30 20 10 0

−0.25

0.00

0.25 Voltage (V)

0.50 Vmax

400

0.75

Voc

Light power conversion efficiency h h =

(a)

50

Pelectrical Plight =

=

FF·Isc·Voc

Plight

Imax·Vmax Plight × 100%

500

600 700 Wavelength (nm)

800

External quantum efficiency (EQE) or incident photon to collected electron efficiency (IPCE)

× 100% EQEl = IPCEl =

N (collected electrons)λ

× 100%

N (photons)λ (b)

Figure 6 (a) Typical current–voltage characteristics of organic P3HT/PCBM solar cell and equation used for calculation of the maximal light power conversion efficiency η. (b) The external quantum efficiency (EQE) so-called incident photon to collected electron efficiency (incident photon to collected electron efficiency—IPCE) spectrum of a typical P3HT/PCBM device and a simple equation used for calculation of EQE at each wavelength. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

12

Supramolecular devices

can give outstanding FFs of 60–70% and high power conversion efficiencies if electrode and buffer layer materials are chosen correctly.

LUMO

e−

∆EET ~ 0.25 eV

LUMO

VOC (max)

1.4

Characterization of organic photovoltaic devices

The main tool for characterization of organic photovoltaic cells is measuring their current–voltage characteristics (I–V curves) in dark and under illumination. Under standard conditions, the light-on curves of the photovoltaic devices should be measured using standardized light sources (solar simulators) with known irradiation spectrum which has to be close as much as possible to the air mass 1.5 (AM1.5) spectrum. The light intensity should be set to 100 mW cm−2 and the photovoltaic cell should be kept at 25 ◦ C. From the experimental I –V curves (Figure 6a), it is possible to extract all main parameters of the device: short-circuit current density (ISC ) is determined as current at the zero applied voltage; open-circuit voltage (VOC ) is measured at the point where current is equal to zero; FF is determined using the equation shown as inset in Figure 6(a) and; finally, the power conversion efficiency of the device (η). The power conversion efficiency of photovoltaic cell is calculated as electrical power produced by the device divided by the power of the light irradiating the device area. The electrical power is calculated as a maximal product of the current and voltage in the fourth quadrant. The current and voltage at the maximal power point on the I –V curves are defined as Imax and Vmax , respectively. The short-circuit current density in organic photovoltaic cells is limited by a number of photons absorbed in the active layer of the device. Therefore, the active layer materials should exhibit wide absorption spectra to harvest the solar light efficiently and to produce high current densities. Considering the overlap between the absorption spectrum of the device active layer of a certain thickness and the solar AM1.5 emission spectrum, it is possible to estimate the maximal short-circuit current of any photovoltaic device.77, 78 The maximal open-circuit voltage in bulk heterojunction organic solar cells is defined by the energy onset between the HOMO level of the donor (p-type component) and LUMO level of the acceptor (n-type component) minus two energy gains of circa 0.25 eV each (0.5 eV in total) facilitating electron and hole transfers between the components (Figure 7).79 This dependence was well illustrated using sets of fullerene derivatives with different LUMO energy levels and electron donor polymers with different HOMO energy levels.80, 81 The FF of photovoltaic devices depends strongly on the charge transport characteristics of the photoactive blend and

HOMO ∆EHT ~ 0.25 eV Donor

h+

HOMO

Acceptor

Figure 7 Open-circuit voltage in organic bulk heterojunction solar cells.

the interfaces between the active layer and the electrodes. The FFs of 60–70% are considered as state-of-the-art values for organic photovoltaic devices.31 The characterization of photovoltaic devices remains incomplete without measuring their EQE spectra also called IPCE spectra. To obtain such spectra, the devices are irradiated with monochromatic light of known intensity which generates photocurrent that has to be measured with high accuracy. The calculation of an average fraction of incident photons that produced free electrons collected at the device electrodes gives us EQE (IPCE) values at each wavelength (Figure 6b). If we recalculate the fraction of the produced in the device charge carriers per every absorbed (but not incident) photon we can obtain IQE spectra. IQE is always higher than EQE because it does not account for the optical losses (e.g., scattering, reflection). Integration of EQE spectrum of a photovoltaic device over known AM1.5 solar irradiation spectrum provides very precise way for calculation of the device short-circuit current density. This method should always be used for checking the ISC values obtained from the I–V measurements that appear to be less accurate.

2

SUPRAMOLECULAR ASSEMBLING EFFECTS IN DYE SENSITIZED ORGANIC SOLAR CELLS

Dye sensitized organic solar cells rely on the supramolecular interactions between the photon absorbing organic dye and mesoporous titanium dioxide. The most widely used approach is based on the introduction of one or several carboxyl groups into the dye molecular framework. These carboxyl groups form hydrogen bonds with the hydroxyl groups and oxygen atoms exposed on the surface of the titanium dioxide. As a result, TiO2 nanocrystals become covered with a self-assembled monolayer (SAM) of organic dye molecules (Figure 8). Affinity of carboxyl-loaded compounds to the titanium dioxide is so strong that it is enough

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

O O O O

FTO

O

TiO2

O O O O O

Dye

Dye Dye

Dye Dye

Figure 8 A schematic illustration of the photoelectrode with organic dye molecules anchored to TiO2 surface via COOH groups

Conductive glass

Barrier

Layer

Dye molecules Electrolyte Graphite or gold coated glass

(a)

and 0.5 M tert-butylpyridine in methoxypropionitrile. The fabricated device exhibited short-circuit current (ISC ) of 7.87 mA cm−2 , an open-circuit voltage (VOC ) of 0.75 V, and a FF of 0.49 which correspond to the overall power conversion efficiency (η) of 2.9%.82 Surprisingly, the advanced morphology of the titanium oxide nanotube array did not result in a superior performance of the devices in comparison with the conventional DSSCs based on much more disordered nanocrystalline TiO2 . One of the main reasons is that a thickness of the active layer of the nanotubebased device (360 nm) is circa 10–20 times less than in the case of optimized state-of-the-art dye sensitized solar cells (3–7 µm). A 3D dye-sensitized solar cells were fabricated recently using similar approach.83 Commercial optical fibers were covered with zinc oxide nanowire arrays soaked with the ruthenium N719 dye. The ZnO nanowires (NWs) grow in the normal direction to the optical fiber surface and enhance the surface area for the interaction of light with the dye molecules. The device was finalized by attachment of a platinum counter electrode, packaging and filling with the electrolyte. The architecture of 3D DSSCs is schematically shown in Figure 10(a). The highest power conversion efficiencies were achieved for rectangular fiberbased hybrid structures shown in Figure 10(b). It is notable that the illumination of the device from one end of the fiber along the axial direction results in much higher power conversion efficiencies than in the case of light illumination normal to the fiber axis from outside the device. Internal reflection within the fiber creates multiple opportunities for energy conversion at the interfaces in the case of illumination parallel to the fiber axis. The 3D dye sensitized solar cells yielded power conversion efficiency of 3.3%. It was claimed that this value is 120% higher than the highest value reported for ZnO NWs grown on a flat substrate surface and 47% higher than that of ZnO NWs coated with a TiO2 film.83

Photons

Photons

just to immerse the TiO2 electrode into the dye solution for some time to achieve a good coverage. Some examples of efficient organic and organometallic sensitizers for DSSCs are given above in Table 1. One of the interesting examples is the application of aligned TiO2 nanotubes instead of conventional mesoporous titanium dioxide.82 The concept of the device is presented in Figure 9(a). To make an aligned nanotube array, 500-nm thick titanium film was sputtered on the top of a FTO electrode. The obtained titanium metal film was anodized in a 0.5% HF electrolyte at the applied bias voltage of 12 V. Illustrative field emission scanning electron microscopy (FESEM) image of the transparent nanotube arrays is shown in Figure 9(b). The titanium dioxide surface of the nanotubes was activated by the treatment of electrodes with the TiCl4 solution. The device was finalized by the following deposition of the ruthenium N719 dye, assembling of the platinum counter electrode (25-nm film on glass) and filling the cell with the electrolyte containing 0.5 M LiI, 0.05 M I2 , 0.6 M N-methylbenzimidazole, 0.10 M guanidinium thiocyanate,

13

400 nm

(b)

Figure 9 (a) Schematic layout of a DSSC device based on a transparent titanium dioxide nanotube array covered with organic dye molecules and (b) FESEM image of titanium dioxide nanotubes grown by anode oxidation of a 500 nm thick titanium film. (Reproduced from Ref. 82.  American Chemical Society, 2006.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

14

Supramolecular devices a

Cladding

b

λ nm−1

d Dye-coated ZnO NWs

J / mA cm−2

6

4

(a)

ITO

ZnO

Pt film

6

2

4 EEF

1

2 0

0 1 2 3 4 5 6 7 8 Sample number

1 NA

El ec tro de

0 Electrolyte

8

3

0

2

DSSC

4

Electrode

Bundle of optical fibers

PA

2

8

h% / %

Optical fiber

0.0

0.1

0.2

0.3

(b)

0.4

0.5

0.6

0.7

V/V

Figure 10 (a) Schematic layout of the architecture of the 3D DSSCs. (b) I–V curves for rectangular-based fiber device illuminated normal to the fiber axis from outside the device (blue line) and parallel to the fiber axis (red line). (Reproduced from Ref. 83.  Wiley-VCH, 2009.)

Dye sensitized solar cells built without any transparent metal oxide electrodes (ITO or FTO) were presented recently by Kashiwa et al.84 The layout of the device is shown in Figure 11. To fabricate such device, titanium dioxide paste was coated directly on a glass substrate. On the top of TiO2 layer, tetrapode-shaped nanocrystals of zinc oxide were deposited by electrospray technique. A layer of metal titanium was sputtered on the top. Following etching of the zinc oxide template produced pores in the titanium layer.

These pores were large enough to allow facile deposition of the N719 dye on the surface of the underlying titanium dioxide nanoparticles. The device prepared in such a way yielded power conversion efficiency of 7.4%. It was shown that organic dyes can be attached to the titanium dioxide surface indirectly using suitable organic mediators. The most illustrative was the fabrication of DSSCs using complicated porphyrin–ferrocene dyad as a dye.85 This dyad possesses no carboxyl groups capable of direct bonding to the titanium dioxide. Therefore,

1. TiO2 layer (coating)

4. HCl aqueous solution treatment, and rinse Top view

2 A 1

B 3

2. Tetrapod-shaped ZnO (electrospray)

4

5. Dyestaining

Porous Ti

3. Ti sputter (a)

Top view

A

B

(b)

Figure 11 (a) Schematic layout of the device. (b) Zinc oxide templated process for fabrication of porous titanium electrode. (Reproduced from Ref. 84.  American Institute of Physics, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

Light

ET

HS

N N Zn N N

O O C

N

N

O O C

N

N

N

FTO/TiO2

OH

Fe

e−

O O C

O O C

O

N

N

O C HN

l−

C

N

N

O

N

N

N N Zn N N

O O C

15

HN

l 3−

e−

Fe

N

N N Zn N N

O C HN

Donor1-Donor2

1

Pt Fe

e−

(a)

(b)

Figure 12 (a) Molecular structure of 4-(1H -imidazol-1-yl)benzoic acid 1. (b) Operation mechanism of the fabricated supramolecular device. (Reproduced from Ref. 85.  American Chemical Society, 2009.)

titanium dioxide was initially stained with 4-(1H -imidazol1-yl)benzoic acid 1 (Figure 12). This modification resulted in the attachment of imidazolyl units to the titanium dioxide surface. It is known that imidazolyl groups have strong affinity for complexation with metal porphyrins.86 The modified TiO2 electrode with exposed imidazolyl units was stained with the dye simply by its dipping into a solution of the porphyrin–ferrocene dyad. The developed two-step approach for the dye deposition on the titanium dioxide allows for application of a range of readily available metal macrocycles as sensitizers in DSSCs.

Very similar approach was applied previously to fabricate “quantum dot-sensitized solar cells.”87 Treatment of TiO2 with 3-mercaptopropionic acid produced efficient coverage of the material with SH groups (Figure 13). Following selfassembling of the CdSe nanoparticles proceeded smoothly and yielded finally DSSCs with EQE approaching 12%. The examples above illustrate importance of the supramolecular chemistry tools for the development of novel materials and constructing new device architectures in the field of dyesensitized solar cells. This approach readily demonstrates the versatility of an organized supramolecular structures in such applications.

HOOC HOOC HOOC

OOC SH OOC SH OOC SH

e

h

OTE

CdSe TiO2

SH

Chemically modified TiO2 film

h

OOC

hn

OOC SH

OOC

SH

OOC

OOC

S S S

SH

OTE (a)

SH

h OOC

e

Bifunctional linker molecule

TiO2

SH

CdSe (b)

OTE/TiO2/CdSe electrode

Figure 13 (a) Photoactive TiO2 -based electrode covered with CdSe nanoparticles and (b) a route for linking of CdSe nanoparticles to the titanium dioxide. (Reproduced from Ref. 87.  American Chemical Society, 2006.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

16

Supramolecular devices

3

SUPRAMOLECULAR ASSEMBLING EFFECTS IN PHOTOELECTROCHEMICAL CELLS BASED ON COVALENTLY AND NONCOVALENTLY LINKED DONOR/ACCEPTOR MOLECULAR ARCHITECTURES

3.1

with the formation of closely packed bilayer (Figure 16).93 This bilayer works quite well in photoelectrochemical cells due to the presence of five electron donor ferrocene units providing IQE of 12%. Many other examples of donor–acceptor systems self-assembled on metal or metal oxide photoelectrodes interested reader might find in reviews covering this field.39, 94–96 The best photoelectrochemical solar cells based on the covalently linked dyads are characterized by appreciably high IQEs of 30–50%. However, EQEs remain negligibly low (EQE 50% and IQEs >85%.159 Next step toward the understanding of the morphology effects in the MDMO-PPV/P3HT based solar cells came as a result of a combined AFM and Kelvin probe force microscopy study of the toluene and chlorobenzene-cast films.160 By comparing surface work functions at different areas of the films it was proven that the round-shaped

ITO

ITO

(a)

200 nm

200 nm

ISE 5.0 kV 6.0 mm × 100 k SE(U)

(b)

500nm

ISE 5.0kV 13.2mm × 90.0k SE(U)

500nm

e−

e−

Al electrode e−

e−

e−

h+

h+

e−

h+ h+

h+

e−

h+

ITO covered glass (c)

Al electrode h+ h+

h+

h+

h+

ITO covered glass

h+

(d)

Figure 49 SEM cross-section images of MDMO-PPV : PCBM blend films cast on ITO-glass from (a) chlorobenzene and (b) toluene solution. The brighter objects (a) are polymer nanospheres, whereas the darker embedments are PCBM clusters. Schematic of film morphology of (c) chlorobenzene- and (d) toluene-cast MDMO-PPV : PCBM blend active layers. (Reproduced from Ref. 160.  American Chemical Society, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics (a)

(b)

100 nm

100 nm

41

Figure 50 TEM images of P3HT/[60]PCBM blends (a) before and (b) after thermal annealing. (Reproduced from Ref. 167.  WileyVCH, 2007.)

additives, and other tricks with the P3HT/[60]PCBM blends in solution were shown to induce formation of elongated P3HT nanofibers that are supposed to improve percolation in the blends thus facilitating charge extraction to the electrodes.165, 166, 168, 169 In some cases, the formation of nanofibers was revealed using AFM topography images (Figure 51). Electron tomography has revealed a 3D structure of the P3HT/[60]PCBM blends with nanometer resolution.170 It was demonstrated that thermal and solvent vapor annealing results in the formation of genuine 3D nanoscale interpenetrating networks of donor and acceptor materials with high crystalline order (Figure 52). These favorable morphological changes account for a considerable increase of the power conversion efficiency of the devices after thermal or solvent-assisted annealing. Electron tomography allows for precise determination of concentration gradients of both

P3HT and [60]PCBM through the thickness of the photoactive layer. It has proved, in particular, that concentration of P3HT is higher at the ITO/PEDOT : PSS electrode for the annealed blend films (Figure 53). A thorough understanding of the morphology effects in organic bulk heterojunction solar cells based on the P3HT/[60]PCBM composite resulted in gradual improving of their light power conversion efficiency up to the level of 4.0–4.5%.77, 171 Unfortunately, the optimal morphology of the fullerene/ polymer bulk heterojunction solar cells seems to be thermodynamically unstable. Continuous heating of the devices under real outdoor conditions leads to progressing segregation of the fullerene derivative from the polymer, which results in a dramatic degradation of the photovoltaic performance. This thermodynamic instability is a fundamental problem limiting significantly the lifetime of organic solar

55.49°

5.50 nm

200 nm

200 nm –4.50 nm

(a)

44.51°

(b)

Figure 51 (a) AFM height and (b) phase images of P3HT/[60]PCBM composite processed using solvent vapor annealing. (Reproduced from Ref. 168.  Wiley-VCH, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

42

Supramolecular devices TA

SAA

Reconstructed volume

Close to bottom...

...middle...

...close to top

As spin-coated

Figure 52 Results of electron tomography applied to P3HT/[60]PCBM photoactive layers: as spin coated, thermally annealed (TA) at 130 ◦ C for 20 min and solvent assisted annealing (SAA) for 3 h. First three rows contain slices taken out of a reconstructed volume of the corresponding film. All slices are lying in the horizontal (x, y) plane of the film at a different depth (z location): first slice close to the top of the film (i.e., to the electron-collecting electrode), second in the middle of a film, and third close to the bottom of the film (the hole-collecting PEDOT : PSS/ITO electrode). The dimensions of the slices are around 1700 × 1700 nm. Images in the fourth row are snapshots of the corresponding film’s whole reconstructed volume, that is, a stack of all of the slices through the whole thickness of a film, with dimensions of around 1700 × 1700 × 100 nm. (Reproduced from Ref. 170.  American Chemical Society, 2009.)

P3HT volume percentage (%)

48 44 40 36 32 28 24 Close to bottom

Slice position

Close to top

Figure 53 Vertical gradient of P3HT volume concentration through the thickness of the blend film as reconstructed from the electron tomography measurements. (Reproduced from Ref. 170.  American Chemical Society, 2009.)

cells. To overcome this problem, cross-linkable fullerene derivatives have been designed with the aim to fix initially induced morphology of the blend. However, these compounds gave much lower photovoltaic performances.172, 173 Alternative way of stabilization of bulk heterojunction solar cells was presented by Frechet and colleagues.174 Statistical copolymerization of 3-hexylthiophene with 4 mol% of 3,4-dihexylthiophene yielded modified P3HT samples (P3HT-modif) with decreased crystallinity and regioregularity. It was shown that P3HT-modif gives highly ordered blends with [60]PCBM while conventional P3HT does not (Figure 54a and b). This is quite surprising observation since less ordered polymer forms more ordered composite because of some unknown reasons. At the same time, solar cells based on the P3HT-modif showed greatly improved thermal stability. Annealing of the P3HT-modif/[60]PCBM

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics C6H13

C6H13

43

C6H13

S

S

S

S C6H13

n C6H13

n

P3HT

m

P3HT-modif

1 µm

1 µm

200 nm

200 nm

(a)

(b) 4.0 3.5

RR-P3HT Poly(1-co-2)

2.0

5 0 −0.2 0.0

1.5 −2

1.0 0.5

30 min 0.2

0.4

5

0.6

0.8

−5 −10 −15

−0.2 τ (mA cm−2)

2.5

τ (mA cm )

hPC AM 1.5G

3.0

Potential (v)

0 0.0

300 min 0.2

0.4

0.6

0.8

−5 −10 −15

Potential (v)

0.0 0 (c)

50

100 150 200 Annealing time (min)

250

300

Figure 54 TEM images of annealed films of (a) P3HT/[60]PCBM and (b) P3HT-modif/[60]PCBM. An interpenetrating network of polymer and fullerene is observed with P3HT-modif (b), while no obvious features are seen in the polymer region of the control case (a). These insets show a wider view of the films and the micron-scale phase segregation of PCBM (dark region) from the P3HT in the control case. No micron-scale features are observed when P3HT-modif is used. The film edge is shown to confirm focus in the inset (b). Performance of photovoltaic devices fabricated with the control RR-P3HT or poly(1-co-2). Degradation of the solar cell performance as a function of the annealing time at 150 ◦ C; typical J –V behavior (insets) observed after 30 and 300 min of annealing illustrate the difference between the materials (c). (Reproduced from Ref. 174.  American Chemical Society, 2006.)

devices at 150 ◦ C for 300 min resulted just in minor degradation of the efficiency in contrast to the reference P3HT/[60]PCBM cells that degraded substantially (Figure 54c). In our opinion, improved morphology and thermal stability of the P3HT-modif/[60]PCBM blends appear because of some specific supramolecular effects that improve compatibility between the polymer and the fullerene. This aspect requires deeper exploration for understanding. The same group reported designing and application of a special plasticizer improving miscibility of the fullerene and the polymer components.175 This plasticizer is based on a diblock copolymer comprising fullerene and P3HT subunits (Figure 55a). The addition of 17 wt% of this

plasticizer to the active layer improved significantly thermal stability of the devices. No phase separation was observed after continuous (10 h) annealing of the P3HT/[60]PCBM composite at 140 ◦ C in the presence of the plasticizer. At the same time, solar cells with the plasticizer did not degrade dramatically in contrast to the reference devices based on the conventional P3HT/[60]PCBM composite (Figure 56). Even more promising plasticizer based on triblock copolymer PTPA-P3HT-PTPA (Figure 55b) has been designed very recently.176 Incorporation of this material to the P3HT/[60]PCBM blend enhances significantly its photovoltaic performance up to the level of 4.4% and improves thermal stability of the device. It is particularly evident from the optical microscopy images that the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

44

Supramolecular devices

O Ph

O

O

O

3.0%

O

O

6

2.5%

N O

O

hp AM 1.5 G

6

m

O

n

N

Hex S

S

1.5% 1.0% 0.5%

S

O

2.0%

p n

n

Hex

Hex

0.0%

n

m = 25, n = 2 – 10, p = 25

0.1

poly(1)-block-poly(2)

(a)

(b)

1

10

Annealing time (h)

Figure 55 Molecular structure of the plasticizer poly(1)-block -poly(2) (a); Power conversion efficiency of P3HT/[60]PCBM solar cells with (,
) and without (, •) 17 wt.-% of poly(1)-block -poly(2) plasticizer as a function of annealing time before (open marks: , ) and after (closed marks:
, •) deposition of the aluminum electrode (b). (Reproduced from Ref. 176.  American Chemical Society, 2010.) 140 °C 20 min

140 °C 10 h 200 µm

200 µm

200 µm

200 µm

200 µm

0%

200 µm

C6H13

s

Bu

m 3

S

S

n

N

Bu

N

PTPA-P3HT-PTPA 10%

(a)

m

3

5%

s

C6H13

(b)

Figure 56 Molecular structure of PTPA-P3HT-PTPA (a); optical microphotographs of P3HT/[60]PCBM blends with 0, 5 and 10% of plasticizer PTPA-P3HT-PTPA annealed at 140 ◦ C for 20 min. and 10 hrs (b). (Reproduced from ref. 176.  American Chemical Society, 2010.)

plasticizer inhibits efficiently formation of the [60]PCBM crystallites during the thermal annealing of the blends (Figure 56). Other less efficient diblock copolymer plasticizer was reported recently.177 It was possible to design some small molecular “antiplasticizers” that decrease compatibility between P3HT and [60]PCBM. The addition of small amounts of these compounds induces phase separation in fullerene/polymer blends and improves their photovoltaic performance without applying thermal or solvent vapor annealing.178

It has been reported recently that fullerene derivatives with appended thiophene and furane units (Figure 54) show improved miscibility with P3HT.179 Spectroscopy and microscopy studies suggested strong supramolecular interactions between the thiophene-containing fullerene derivatives and the polymer. It was also proposed that the fullerene and the polymer in this case self-assemble together forming thermodynamically stable individual phase. This phase precipitates from the blend in the form of welldefined aggregates dispersed in the composite films.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

O S

O CH3

(a)

O

S

O

I

II

O O

S

CH3

O

III

Nonannealed H3C

45

Annealed −5.00

−5.00 20.0 nm

20.0 nm

10.0 nm

10.0 nm

0.0 nm −2.50

0.0 nm −2.50

S S

0

P3HT

(b)

0

2.50

5.00

0

2.50

0 5.00

(c)

Figure 57 (a) Molecular structures of thiophene and furane-appended fullerene derivatives I–III; (b) schematic illustration of the complex formation between the compound I and P3HT; (c) AFM images of nonannealed and annealed blends of I/P3HT showing power conversion efficiencies of 2.5 and 3.7% in solar cells, respectively. Bright regions on the images are supposed to correspond to I + P3HT aggregates. (Reproduced from Ref. 179.  Wiley-VCH, 2010.)

The appearance of these aggregates does not hamper photovoltaic performance of the fullerene/P3HT blends; short-circuit currents and FFs of the devices become even improved in some cases. This behavior makes a sharp contrast with the MDMO-PPV/[60]PCBM and P3HT/[60]PCBM systems described above where the formation of numerous PCBM aggregates in the films immediately kills their photovoltaic performance. It might be envisioned that solar cells comprising thiophene or furane-appended fullerene derivatives (I–III) exhibit superior thermal stability because of hampered fullerene/ polymer segregation. Unfortunately, this issue was not addressed in this study.179 Unusual results came out from a systematic study of a library of fullerene derivatives in bulk heterojunction solar

cells in combination with P3HT.180 It was found that all of the presented fullerene derivatives shown in Figure 57 have virtually the same frontier energy levels (in particular, LUMO energies affecting VOC of solar cells) regardless the structure of organic addend attached to the cages of C60 and C70 fullerenes. However, even very slight modifications of the molecular structures of fullerene derivatives induce significant changes in their physical properties such as solubility in organic solvents. This is well illustrated by the solubility values measured for all fullerene derivatives in chlorobenzene, which are given in brackets near numbers of the compounds, and are shown in Figure 58. The solubility of the fullerene derivatives affected strongly the morphology of their blends with P3HT (Figure 59). For instance, the films composed of the least

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

46

Supramolecular devices O O

O O

R O

R

R O

S

O

O

O O 25 (4) 1 R = CH3 [60]PCBM (50) 2 R = Et (19)

6 7 8 9 10 11

17 18 19

R = CH3 (10) R = Et (5) R = Pr-n (43) R = Pr-i (22) R = Bu-n (30) R = Bn (106)

R = Et (23**) R = Pr-n (45) R = Bu-n (70)

22 (31) O O

R

O

O

O O S

CH3

O

O

O

CH3 O

O

CH3

CH3

26 (11)

S

O

23 (23) H 20 R = Pr-n (130) 21 R = Bu-n (124)

3 (36) 12 (5) O O

CH3

O C8H17-n

O CH3

27 (9)

R

O

O O

O S

O

24 (25) 13 14 15 16

4 (58)

R = CH3 (12) R = Et (10) R = Pr-n (35) R = Bu-n (30)

5 (80) [70]PCBM

Figure 58 Molecular structures of fullerene derivatives forming a library of acceptor materials investigated in solar cells in combination with P3HT. Solubility values determined in chlorobenzene are given in brackets near numbers of the compounds. (Reproduced from Ref. 180.  Wiley-VCH, 2009.)

soluble fullerene derivative 7 (solubility S = 5 mg ml−1 ) and P3HT showed large aggregates approaching 30–100 µm in size. Increase in the solubility of the fullerene component by a factor of 2 going to the compound 6 (S = 10 mg ml−1 ) resulted in the remarkable decrease in the cluster size in the blend films by a factor of 10. Fullerene-based materials with the solubility of 22 mg ml−1 (compound 9) and 30 mg ml−1 (compound 10) led to further improvement of the blend morphology erasing any signs of the phase segregation at least on the micrometer scale. Strong variation of the film morphology induced by different solubility of the fullerene derivatives (molecular structures are shown in Figure 59) affected significantly photovoltaic performance of the P3HT/fullerene composites. Rather clear correlations of short-circuit current, opencircuit voltage, FF, and light power conversion efficiency with the solubility of the fullerene-based materials were revealed (Figure 60).

The presented examples prove that supramolecular assembling of the fullerene derivative and the polymer in the blends is governed by relative solubility of the materials. In particular case of P3HT, the best solar cell performances were obtained with the use of fullerene derivatives that exhibited solubility values close to the solubility of P3HT itself (70–90 mg ml−1 ). Therefore, it was suggested that balanced active layer morphology in fullerene/polymer composites can be achieved through a combination of electron donor and electron acceptor materials with matching solubilities. As a main consequence of this conclusion, any novel electron donor material might require specific fullerene counterpart with fitting solubility to be combined in order to achieve the highest photovoltaic performance. Further investigation of the library of fullerene derivatives in combination with poly(3-alkylthiophenes) with different side chains revealed unexpected results.181 It was found that the dependence of the solar cell parameters on

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

7 / P3HT S = 5 mg ml−1

6 / P3HT S = 10 mg ml−1

600

600

400

400

200

200

0 0

200

400 µm

600

9 / P3HT S = 22 mg ml−1

0 0

200

400 µm

600

10 / P3HT S = 30 mg ml−1

600

600

400

400

200

200

0

0 0

200

400 µm

600

47

0

200

400 µm

600

Figure 59 Optical microscopy images for the blends of four different fullerene derivatives with P3HT. Obvious improvement of the active layer morphology with increase in the solubility of the fullerene derivatives can be observed. (Reproduced from Ref. 180.  Wiley-VCH, 2009.)

the fullerene component solubility might have an unexpected double-branched shape as it is shown for poly(3pentylthiophene) in Figure 61. The fullerene derivatives with the solubility fitting the range of 20–60 mg ml−1 form two separate groups. One group of fullerene-based compounds clearly outperforms another one in solar cells as it follows from Figure 61. Moreover, very similar dependences were revealed for poly(3-alkylthiophenes) with heptyl, octyl, decyl, and dodecyl side chains as well. The observed double-branched behavior was related to variations in the active layer morphology induced by peculiarities of the molecular structures of fullerene derivatives. For instance, fullerene derivatives 3 and 8 have very similar solubility in chlorobenzene. However, the films of their blends with poly(3-decylthiophene) (P3DT) identically annealed at 140 ◦ C for 5 min have very different morphologies (Figure 62). Fullerene derivative 8 seems to be well compatible with P3DT polymer, which results in homogeneous film structure and relatively high photovoltaic performance (η = 2.0%). On the contrary, compound 3 seems to be badly miscible with the polymer, which results in a large-scale phase separation in the blend (Figure 62) and poor photovoltaic performance (η = 0.3%).

The obtained results suggest strongly that the morphology and photovoltaic performance of the fullerene/polymer composites is governed to a large extent by supramolecular interactions between the components. On one hand, the best performing systems (upper “branch” in Figure 61) comprise fullerene derivatives that are well compatible with the polymer because of some attractive intermolecular interactions. On the other hand, the worse performing systems (lower “branch” in Figure 61) are based on badly compatible fullerene and polymer components that might be a consequence of some missing intermolecular interactions. Molecular structures of some promising latest generation polymer materials for bulk heterojunction solar cells are shown in Figure 63. First of all, these are cyclopentadithiophene-based copolymers with carbon or silicon bridging atoms (PCPDTBT and Si-PCPDTBT), then fluorene-based (PFO-DBT and PSiF-DBT) and carbazolebased copolymers (PCDTBT), and finally, one of the most promising existing polymers abbreviated as PTB7. Power conversion efficiencies of these materials in organic solar cells are listed in Table 2. Supramolecular organization of these new electron donor polymers with electron acceptor fullerene derivatives is

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular devices

Isc (mA cm–2)

10 8 6

600 11

550

20 21

0

60

80

100

120

60

45 40 35 30

2

11* 15

17**

3 21 20

2

13 26 6 27 7 0 2512 20 0

4 60

80

100

120

140

–1

(d)

Fullerene solubility (mg ml )

60

80

100

120

140

5 1 18

2 9 22 11* 10 17** 8* 8 24 14 23 16

19 11

4

1

40

(c)

3 15

11

24 22 23 10 8 6 8* 14 7 27 16 12 25 13 26 0 20

40

4 19

20

Fullerene solubility (mg ml–1)

5

18 1 9

20

(b)

3

55 50

0

140

Fullerene solubility (mg ml–1)

(a)

21

6 7 12 27 25

300 40

11

450

350

20

5 19

500

400

6 27 7 12 25

0

23 15 2 9 16 1 10 8 13 17** 3 18 4 26 22 14 8*24 11*

650

4 2

FF (%)

5

15 1 22 3 10 18 19 8* 11* 9 17** 24 8 4 13 2 16 23 14 26

Voc (mV)

12

h (%)

48

40

60

21

80

Fullerene solubility (mg

100

120

20

140

ml–1

)

Figure 60 The short-circuit current (ISC ), open-circuit voltage (VOC ), fill factor (FF), and light power conversion efficiency (η) as functions of the solubility of the fullerene-based materials. (Reproduced from Ref. 180.  Wiley-VCH, 2009.) We

ll c

3.5

om

pa

tibl

e

h (%)

3.0

2.5

atible comp Badly

2.0

1.5 0

20

40

60

80

100

120

140 −1

Solubility of fullerene derivative (mg ml )

Figure 61 Unusual double-branched dependence of the solar cell power conversion efficiency on the solubility of the fullerene derivatives combined with poly(3-pentylthiophene) as a donor polymer.181

a rather complicated issue. Some time ago, Zhang et al. demonstrated that morphology of the PFO-DBT/[60]PCBM blends depends strongly on the solvent composition used for the film casting.186 For example, the films processed from

pristine chloroform yielded high FFs (63%) in combination with low short-circuit current densities and moderate power conversion efficiencies (Table 2). However, addition of few volume percents of chlorobenzene to the chloroform solution of the PFO-DBT/[60]PCBM blend resulted in significant improvement of the short-circuit current density and overall device performance. Somewhat disappointing was decrease in the FF down to 49% (Table 2). Mixing of chloroform solution of PFO-DBT/[60]PCBM blend with a number of other solvents affected significantly the photovoltaic performance of the system (Table 2).186 It was demonstrated that additives of toluene, chorobenzene, or xylene change significantly morphology of the PFO-DBT/[60]PCBM composite (Figure 64). The PFODBT/[60]PCBM blend films spin coated from chloroform/chlorobenzene mixture (40 : 1 v/v) were much more homogeneous compared to the films cast from pure chloroform. This might be a reason for improved photovoltaic performance of this system. On the contrary, toluene and xylene additives led to coarsening of the film topologies and decrease in the solar cell performance (Table 2 and Figure 63).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics O

49

8/P3DT – well compatible

O C10H21

nm

5.0

Height

S

25.0

3D

4.0

+

20.0

S C10H21

3.0

2.0

P3DT

8

15.0

n

10.0

S = 43 mg ml−1

1.0

5.0

0 pm 0

O S

1.0

2.0

3.0

4.0

5.0

0

3/P3DT – badly compatible O

CH3

nm

10.0

C10H21

O

Height

9.0

250

8.0

S

7.0

3D

200

6.0

+

S

5.0

C10H21

n

3

150

4.0

P3DT

100

3.0 2.0

50

1.0

S = 36 mg ml−1

0 pm

(a)

0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

0

(b)

Figure 62 (a) The blends of fullerene derivatives 3 and 8 with P3DT reveal very different morphologies in spite of the similar solubility of these compounds in chlorobenzene. (b) Molecular compositions of the blends are shown (on the left side) together with the height AFM images and their 3D profiles (on the right side). Table 2

The performance of some third-generation copolymers in organic solar cells.

Materials PCPDTBT/[60]PCBM PCPDTBT/[70]PCBM PCPDTBT/[70]PCBM + ODT PCPDTBT/[70]PCBM + DIO PCPDTBT/[70]PCBM + DBO PCPDTBT/[70]PCBM + DCO PCPDTBT/[70]PCBM + DCNO PCPDTBT/[70]PCBM + DACO Si-PCPDTBT/[70]PCBM PFO-DBT/[60]PCBM (CF) PFO-DBT/[60]PCBM (CF + CB) PFO-DBT/[60]PCBM (CF + XY) PFO-DBT/[60]PCBM (CF + TO) PSiF-DBT/[60]PCBM PCDTBT/[70]PCBM PTB7/[70]PCBM PTB7/[70]PCBM + DIO

ISC (mA cm−2 )

VOC (mV)

FF (%)

9.0 11.0 14.7 15.8 15.3 3.6 8.0 7.8 12.7 14.9 3.2 6.2 2.6 2.6 9.5 10.6 10.2 14.5

650–700 650–700 640 610 640 350 630 640 680 576 1000 1030 1020 1010 900 880 760 740

47 47 49 53 48 33 47 47 55 61 63 49 36 41 51 63 50.5 69

η (%) 2.7 3.2 4.5 5.1 4.7 0.4 2.4 2.4 5.1 5.2 2.1 2.8 1.0 1.1 5.4 6.1 3.9 7.4

References 182 183

184 185 186

187 51 15

ODT, 1,8-octanedithiol; DIO, 1,8-diiodooctane; DBO, 1,8-dibromooctane; DCO, 1,8-dichlorooctane; DCNO, 1,8-dicyanooctane; DACO, 1,8diacetoxyoctane; CF, chloroform; CB, chlorobenzene; XY, xylene; TO, toluene.

The results described by Zhang et al.186 related to the effect of additives of high boiling temperature solvents were further explored by Peet et al. who demonstrated

that photovoltaic performance of the PCPDTBT/[70]PCBM blend can be enhanced dramatically from 2.8 to 5.5% by incorporating a few volume percents of 1,8-octanedithiol

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

50

Supramolecular devices

N

X

S

N

N

S

N

X *

S

*

S

n

S

S

n

X = C: PCPDTBT X = C: PFO-DBT X = Si: PSiF-DBT

X = Si: Si-PCPDTBT

HEO C8H17

C8H17

S N

S

S

N

N n

S S

S

F S

n

PCDTBT

EHOOC

OEH EH = ethylhexyl PTB7

Figure 63 Molecular structures of some third-generation copolymers well-performing as electron donor materials in organic bulk heterojunction solar cells.

(a)

(c)

0

0

2.50 µm

2.50 µm

5.00

5.00

2.50

2.50

0 5.00

(b)

0

2.50 µm

0 5.00

5.00

5.00

2.50

2.50

0 5.00

(d)

0

2.50 µm

0 5.00

Figure 64 Height images obtained by tapping-mode AFM on PFO-DBT/[60]PCBM blend films spin coated from solutions in (a) pure chloroform; (b) chloroform–chlorobenzene 40 : 1 v/v mixture; (c) chloroform–toluene mixture; and (d) chloroform–xylene mixture. The height scale is 15 nm for all images. (Reproduced from Ref. 186.  Wiley-VCH, 2006.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics Voltage (V) 0.0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

−2

Current (mA cm−2)

−4 −6 −8 −10

None HS-C8H16-SH

−12

Cl-C8H16-Cl

−14

I-C8H16-I

Br-C8H16-Br

−16

NC-C8H16-CN CH3O2C-C8H16-CO2CH3

Figure 65 Light-on I–V curves obtained for organic bulk heterojunction solar cells based on the PCPDTBT/[70]PCBM blends processed from chlorobenzene solutions with various additives. (Reproduced from Ref. 183.  American Chemical Society, 2008.)

(ODT) in chlorobenzene solution of the materials subjected to spin coating.188 Further more detailed study showed that ODT improves photovoltaic performance of the PCPDTBT/[70]PCBM composite up to 4.5% while 1,8diiodooctane (DIO) is a superior additive that gives 5.1% (Table 2).183 A number of other 1,8-disubstituted octane derivatives were also evaluated. Promising results were obtained with 1,8-dibromooctane (DBO) which outperformed ODT and yielded 4.7% efficiency. The effect of different processing additives based on 1,8-disubstituted octanes on the morphology and photovoltaic performance of the PCPDTBT/[70]PCBM blends is illustrated by I–V curves shown in Figure 65 and AFM topography images shown in Figure 66. It is seen from Figure 66 that the most homogeneous films are obtained using ODT, DIO, and DBO as processing additives which are actually giving the best photovoltaic performance. Therefore, the effect of additives developed for PCPDTBT/[70]PCBM blends correlates well with the results of the previous study performed for PFODBT/[60]PCBM composites.186 It is known that supramolecular assembling of the polymer and the fullerene derivative in the blend is governed by the properties of the solvent used for film casting.159 This is well illustrated for MDMO-PPV/[60]PCBM blends shown in Figure 48. In the case of the solvent mixtures, major contribution comes from the high boiling temperature component even if its content is rather low. For instance, the boiling temperature of ODT (270 ◦ C) is 140 ◦ C higher than that of chlorobenzene (130 ◦ C). This is a reason why chlorobenzene evaporates much faster than ODT and the

51

freshly deposited films are soacked with pure ODT that has to be removed under continuous evacuation and/or applying elevated temperatures. It was suggested that modification of the active layer morphology by additives is related to their different ability to dissolve the fullerene and the polymer components of the blend.183 For example, ODT dissolves well PCBM (solubility value of ∼19 mg ml−1 was measured for [60]PCBM) while thiophene-based conjugated polymers are completely insoluble in this solvent.168 In other words, the additive reorganizes the fullerene and the polymer domains in the blend, which improves charge generation and transport to the electrodes. Effect of the additives on the fullerene/polymer supramolecular organization is shown in Figure 67. Finally, it should be noted that processing additive (DIO) improved the performance of the PTB7/[70]PCBM blend from 3.9% to impressive value of 7.4%, which stays as one of the highest efficiency values for bulk heterojunction solar cells published in scientific journals up to date.15 This example illustrates that the use of processing additives is a powerful tool for controlling the supramolecular organization of the donor and acceptor materials in bulk heterojunction solar cells.

6.2

Supramolecular assembling of photoactive materials for bulk heterojunction solar cells

Interesting approach to design of light-harvesting electron donor materials for organic solar cells was presented by Cheng et al.189 A conjugated thiophene–fluorene copolymer was modified with chelating terpyridyl units (PTPy) that have strong affinity for complex formation with ruthenium trichloride thus forming a precursor polymer PTPyRu-Cl. Following replacement of chlorine atoms with SCN groups produced final polymer PTPy-Ru-NSC bearing light-harvesting ruthenium complex moieties as side chains attached to the polymer backbone (Scheme 1). A combination of PTPy-Ru-NSC with pristine C60 fullerene resulted in organic solar cells which exhibited power conversion efficiency of 0.12% and EQE of 12%. This polymer absorbs light efficiently in the whole visible range and therefore deserves further investigation in combination with other fullerene-based acceptor materials such as [60]PCBM and [70]PCBM. Similar approach was explored by Liang et al. who applied hydrogen bonding to construct supramolecular donor polymers (Figure 68).190 In this case, PMMA-based polymer was modified with electron donor block bearing pyridyl end group thus forming hydrogen acceptor polymer P-HA. It was assumed that pyridyl unit of P-HA should form efficient hydrogen bonds with the carboxyl group

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

52

Supramolecular devices

(a)

(b)

0.5 µm

0.5 µm

(c)

(d)

0.5 µm

0.5 µm

(e)

0.5 µm

(f)

0.5 µm

Figure 66 AFM topography of the PCPDTBT/[70]PCBM blend films processed with additives: (a) 1,8-octanedithiol, (b) 1,8dicholorooctane, (c) 1,8-dibromooctane, (d) 1,8-diiodooctane, (e) 1,8-dicyanooctane, and (f) 1,8-octanediacetate. (Reproduced from Ref. 183.  American Chemical Society, 2008.)

PCPDTBT

PCBM Additive

Figure 67 Reorganization of the PCBM and PCPDTBT in the blend induced by processing additive. (Reproduced from Ref. 183.  American Chemical Society, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

Bu4N+

Bu4N+

Cl

N N

N

N

N

N

N

SCN NCS SCN Ru N N N

Cl Cl Ru Cl N N N

Cl Cl Ru N

53

N

SCN NCS Ru NCS N N

N

O O

O

O

O

O

(1) NaSCN/DMF (2) Bu4NBr

RuCl3 /DMF

S S

S

n n

n PTPy

Scheme 1

PTPy-Ru-Cl

PTPy-Ru-NCS

Synthesis of low band gap polymer bearing pendant Ru-NCS complex moieties. H3C CH2 C

N

n

O O (CH2)10 O

C6H13

H-acceptor polymers

S

N

Side-chain H-bonded N

Dye

Dye

Dye

Dye

Dye

Dye

Dye

complexation

O

Dye

Dye

H-bonded complexes

Dye

H3C

CH3

S

C6H13 S

O S

H-donor dyes NC N (a)

(b)

P-HA

COOH D-HD

Figure 68 (a) The formation of supramolecular electron donor polymer via hydrogen-bonding complexation of nonconjugated polymer with small molecular organic dye and (b) molecular structures of the best performing polymer (P-HA) and dye (D-HD) combination. (Reproduced from Ref. 190.  Wiley Periodicals, Inc, 2009.)

of a small molecular hydrogen-donor dye (D-HD). The resulting supramolecular polymer yielded power conversion efficiency of 0.5% in bulk heterojunction solar cells with PCBM. This value is quite impressive taking into account relatively simple molecular structures of the applied materials. Molecular structures of the best performing Pt-containing polymers are shown in Figure 69. A composite of the polymer P-Pt-1 with [60]PCBM was evaluated in bulk heterojunction solar cells which yielded VOC = 820 mV, ISC = 15.4 mA cm−2 , FF = 39% and η = 4.9%.191 Outstandingly

high short-circuit current originated from unusually high EQEs of the device approaching 87% at maximum. However, the absorbance of the photoactive films was below 0.2 in the whole visible range which is hardly compatible with the measured EQEs and ISC . Authors of a critical comment on this publication argued that on the basis of optical modeling results for the reported film thickness (70 nm) maximal EQE should not exceed 45%, ISC should stay below 6.8 mA cm−2 , and the power conversion efficiency has to be lower than 2.2%.192 Subsequent independent study delivered power

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

54

Supramolecular devices

N

S

N N

C8H17

S

S

N

S

S

Et3P

PBu3

S

Pt S

Pt

S C8H17

Et3P

Bu3P

n

n P-Pt-2

P-Pt-1

Figure 69

Molecular structures of the best performing Pt-containing polymers.

conversion efficiency of 1.4% for the blends of P-Pt-1 with [60]PCBM.193 This experimental result agrees with the optical modeling data.192 Nevertheless, Pt-containing polymers seem to have high potential in photovoltaics as it follows from a subsequent report on a similar material P-Pt-2 (Figure 69) which demonstrated power conversion efficiencies up to approximately 4.3% in solar cells.194

7

7.1

SUPRAMOLECULAR CHEMISTRY OF INTERFACES IN ORGANIC SOLAR CELLS Hole-collecting electrode/active layer interface

Typically, organic solar cells are built on transparent hole-collecting electrodes made of thin ITO conducting layers deposited on glass. However, ITO work function

S

n

O

PEDOT-PSS

N

S O

Si Cl Cl Cl

OH

N

N

n

O

O

(3.9–4.2 eV) enables collection of both holes and electrons. This might be utilized for construction of inverted and even semitransparent organic solar cells with both electrodes made of appropriately modified ITO.73 At the same time, dual nature of ITO facilitates charge recombination at the electrode. Recombination losses at the electrode/active layer interface can be minimized by covering ITO with some buffer layers that improve selectivity of the charge collection. Such buffer layer materials can have different nature. On one hand, buffer layer might transport holes efficiently and block electrons (defined as electron blocking hole transport layer, EBHTL). On the other hand, buffer layer material might be conductive and have its work function well adjusted to the HOMO energy level of the electron donor component of the blend thus minimizing energy barrier for hole extraction from the active layer. Water-processible polymer material named PEDOT : PSS was developed and used as conductive buffer layer for modification of ITO electrode (Figure 70).195 It was shown in many studies that PEDOT : PSS is a very

Cl Cl

TPD-Si2

Si Cl

S N

S

N Si Cl Cl Cl

PABT-Si2

Cl Cl

Figure 70

Si

TFB

n

Cl

Molecular structures of PEDOT : PSS, TPD-Si2 , PABT-Si2 , and TFB.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics versatile material and its electronic properties including work function and conductivity can be modified by variable degree of doping.196 Electrical conductivity of PEDOT : PSS layer was improved significantly using additives of high boiling temperature solvents (methyl sulfoxide, N-methylpyrrolidinon, glycerol, ethylene glycol, etc.).197 At the same time, commercial grade PEDOT : PSS was shown to form good and rather selective contact to many electron donor conjugated polymers with different HOMO level energies.81 Other p-doped conducting polymers such as polyaniline-camphorsulfonic acid (PANICSA) and polypyrrole-dodecylbenzene sulfonic acid (PpyBDSA) were also successfully applied as buffer layers for ITO modification.198, 199 Poly(3,6-dimethoxy-thieno[3,2-b] thiophene) was also considered as a promising charge transport material for organic electronics.200 Acidity of all above-mentioned polymer-based conductive buffer layer materials is their strongest drawback. Corrosion of ITO electrode and oxidative doping of the active layer materials results in rapid degradation of electronic devices comprising PEDOT : PSS and related materials.201 Alternative group of the buffer materials is based on self-assembling molecules bearing –SiCl3 pendant groups which anchor to the ITO surface.201, 202 Molecular structures of hole transporting electron blocking materials TPDSi2 and PABT–Si2 are shown in Figure 70. It was shown that PABT-Si2 /TFB composite forms a robust and highly cross-linked layer, which affords the same performances in organic bulk heterojunction solar cells as EBHTL based on PEDOT : PSS. However, PABT-Si2 /TFB is a superior material because it is not acidic and therefore it should not damage ITO and active layer materials deposited on top. It is, therefore, expected that unlike the PEDOT : PSS-based devices, the solar cells fabricated using PABT-Si2 /TFB composite as an EBHTL material exhibit good long-term stability.202 There is a large group of metal oxides which can be used as EBHTL materials in organic bulk heterojunction solar cells. Particularly important are divanadium pentoxide (V2 O5 ) and molybdenum trioxide (MoO3 ) that were first applied by Yang and colleagues.203 It was shown that PEDOT : PSS can be successfully replaced by more chemically inert V2 O5 or MoO3 without compromising the device performance. Tungsten trioxide (WO3 ) is also a promising EBHTL material successfully applied in inverted and semitransparent organic solar cells.65 It might be envisioned that using metal oxides instead of PEDOT : PSS will improve long-term stability of organic solar cells.. However, to our best knowledge this has not been experimentally demonstrated yet. Thermally processed metal oxides were used successfully to fabricate efficient inverted and semitransparent organic solar cells.204 Rapid development of this research field is reflected in recent reviews.73, 75, 76

7.2

55

Electron-collecting electrode/active layer interface

Conventional organic bulk heterojunction solar cells do not have hole-blocking electron transport layers (HBETL) at the cathode/active layer interface. Metals with reasonably low-work function (e.g., calcium, aluminium, magnesium, or barium) are typically evaporated directly onto fullerene/polymer blend forming electronic contacts with both donor and acceptor materials under optimal conditions. This situation leads to significant recombination of positive and negative charge carriers at the active layer/cathode interface. Insertion of some buffer interlayer between the cathode and photoactive blend might suppress charge recombination and improve significantly photovoltaic performance of bulk heterojunction organic solar cells. This was first realized when thin lithium fluoride (LiF) interlayer was applied to MDMO-PPV/[60]PCBM devices. The deposition of ˚ of LiF improved FF from an average 55 to 60–61% 3–12 A and open-circuit voltage from 760–770 to 820–840 mV (Figure 71). Short-circuit current also increased which is in accordance with the expectation for more selective diode contact, that is, better ohmic to electrons and better blocking to holes.205 It was suggested that deposition of aluminum on the top of LiF releases metallic lithium which has fairly low work function and forms ohmic contact with PCBM.206, 207 Similar behavior was documented for CsF,208 CaF2 ,209 MgF2 ,210 NaF,211 and KF.212, 213 Cesium carbonate (Cs2 CO3 ) introduced by Yang and colleagues is even more promising buffer layer material.214–217 It can be processed either by spin coating with subsequent thermal annealing or directly by thermal evaporation. Both procedures give very selective electron-extracting contacts when aluminum or silver is deposited on top. Moreover, ITO modified with cesium carbonate also behaves as selective electron-extracting contact. This allowed fabricating efficient inverted devices and semitransparent bulk heterojunction solar cells.73, 218 Further tremendous development of the HBETL materials for bulk heterojunction solar cells has been accomplished recently. Heeger and colleagues’ suggested the application of solution processed TiOx interlayer at the cathode/active layer interface.53 Titanium oxide transports perfectly electrons and block holes; at the same time it serves as an optical spacer, which redistributes the optical electric field strength creating maximal light absorption in the photoactive layer of the device (Figure 72a). Optical spacer effect improves charge carrier generation in the active layer, which increases short-circuit current of the device. At the same time, TiOx behaves as good HBETL material which suppresses charge recombination

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

56

Supramolecular devices 6 Current density (mA cm−2)

250

Current density (mA cm−2)

4

2

0

200 150 100 50 0 1.0

−2

1.2

1.4

1.6

1.8

2.0

Voltage (V)

3 Ang. 6 Ang. 12 Ang. No LiF

−4 −6 0.0

0.2

0.4

0.6

0.64

1.0

840

0.62 0.60

820

0.58

Voc (mV)

Fill factor (%)

0.8

Voltage (V)

(a)

0.56 0.54

800 780

0.52 0.50

760

0.48 LiF-3A LiF-6A LiF-12A No LiF (b)

LiF thickness (A)

LiF-3A LiF-6A LiF-12A No LiF (c)

LiF thickness (A)

Figure 71 Effect of LiF interlayer on the I–V characteristics of MDMO-PPV/[60]PCBM solar cells (a), (b), and (c) are box plots with the statistics of the FF and the VOC from six separate solar cells. (Reproduced from Ref. 205.  American Institute of Physics, 2009.)

and improves the device FF. The resulting light power conversion efficiency of the P3HT/[60]PCBM solar cells with TiOx interlayer reached 5.0% which is an up-to-date record value for this material combination.53 Zinc monoxide (ZnO) is also suitable HBETL material for organic bulk heterojunction solar cells.219 In some way, it resembles TiOx since it also behaves as optical spacer improving light harvesting in the active layer of the devices.220 A remarkable effect of the SAMs deposited on the top of ZnO buffer layer has been reported recently.221 Indeed, pristine ZnO nanoparticles utilized as HBETL yielded rather moderate performances in P3HT/[60]PCBM solar cells with aluminum, silver, and gold cathodes. However, modification of ZnO interlayer with self-assembling carboxylic acid monolayers improved the performance of solar cells dramatically. For example, the performance of solar cells with ZnO/Au cathode was improved from 0.72 to 3.22% after modification of ZnO with 4-methoxybenzoic acid monolayer. The effect of monolayers was ascribed to the formation of interfacial dipoles improving electronic contacts in the zinc oxide/metal junctions (Figure 73).

Polymer-based materials are also suitable for construction of HBETLs in organic bulk heterojunction solar cells.74 It was reported recently that introduction of buffer layer composed of polymers PFPNBr or PFNBr-DBT15 (Figure 74) under cathode in organic solar cells based on PFO-DBT/[60]PCBM composite (Figure 63) improves open-circuit voltage by 350 mV.222 However, the VOC of 950 mV described in this publication using polymer-based HBETL is still lower than the value of 1030 mV reported previously for the same system without any buffer layers applied.186 Therefore, the reported by Luo et al.222 effect of the polymer-based HBETLs can be doubted on the basis of this comparison. Nevertheless, it was recently demonstrated that introduction of thin (5 nm) interlayer of PF-PO(Et)2 (Figure 74) improves the performance of the conventional P3HT/ [60]PCBM blend from 2.0 to 3.4% mainly due to increase in the open-circuit voltage and FF from 450 to 640 mV and from 51 to 59%, respectively.223 Another polymer abbreviated as PFNBr-PFMEE in Figure 74 improved the power conversion efficiency of P3HT/[60]PCBM bulk heterojunction solar cells from 3.0 to 3.8%. The open-circuit

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

ITO

PEDOT Active layer

Glass

Al

ITO

PEDOT Active layer

Al

E

2

(Arb. Unit)

Glass

(B) Device with optical spacer Optical spacer

(A) Conventional device

57

(a)

OR OR

OR

Ti OR OH

OH

Ti

Aluminum

TiOx

P3HT:PCBM

PEDOT:PSS

ITO

Glass

OH

OH O Ti

O Ti Ti

O

O O

O

Ti

3.3eV TiOx 3.7eV ITO 4.8eV

PEDOT:PSS P3HT:PCBM

4.4eV

4.3eV

TiOx

Al

5.0eV 5.2eV 6.1eV

(b)

8.1eV

Figure 72 (a) Schematic representation of the spatial distribution of the squared optical electric field strength |E|2 inside the devices with a structure of (A) ITO/PEDOT/active layer/Al and (B) ITO/PEDOT/active layer/optical spacer/Al. (b) Schematic illustration of the device structure with a brief flow chart of the steps involved in the preparation of the TiOx layer. The energy levels of the single components of photovoltaic cell are also shown. (Reproduced from Ref. 53.  Wiley-VCH, 2006.)

voltage of the cells modified with PFNBr-PFMEE interlayer approached 680 mV, which is quite impressive value for P3HT/[60]PCBM blends.224 Polyethylene glycol (PEG) was also applied as interface modifying material in P3HT/[60]PCBM solar cells.225 However, in this case PEG was added directly into the fullerene/polymer blend solution. Spin coating of the blend with 5% of PEG additive resulted in initial precipitation of P3HT/[60]PCBM composite followed by formation of thin PEG layer on the top (Figure 75). The observed vertical phase separation between P3HT/[60]PCBM blend on one hand and PEG on the other might be explained by different

surface tensions of these materials (low surface tension of PEG brings it to the organic/air interface) as well as by very high solubility of PEG in 1,2-dichlorobezene which exceeds the solubility of both P3HT and PCBM by few orders of magnitude. The observed increase in the solar cell power conversion efficiency from 2.2 to 4.0% was attributed to the formation of interfacial dipoles between aluminum cathode and PEG layer which improved electron extraction from the active layer. We note also that PEG cannot participate as true HBETL material because of its insulating nature.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

58

Supramolecular devices

Metal (AI, Ag, Au) SAM

record value of 70%. The power conversion efficiency was also increased from 3.1 to 3.8%. The attachment of fluorocarbon residues as end-capping groups for P3HT chains also altered significantly volume distribution of the polymer and PCBM in the active layer of the device. It was shown that a design of matching surface tensions of the donor and acceptor counterparts results in well-balanced vertical and lateral blend morphologies and improved photovoltaic performance.226 Fullerene-based HBETLs were also widely applied in design of inverted organic solar cells. PyF bearing carboxylic group (C60 -Py-COOH) was used to form SAM on the top of titanium dioxide layer deposited on ITO to form selective electron-collecting electrode (Figure 77). The use of C60 -Py-COOH as TiO2 surface modifier improved power conversion efficiency of the P3HT/[60]PCBM blend devices from 2.8 to 3.8%. Other compounds such as terthiophenecarboxylic acid, benzoic acid, and dodecanoic acid were evaluated as alternative SAM modifiers which yielded lower performances compared to C60 -Py-COOH.227 A hybrid HBETL material composed of polymerized fullerene derivative (C-PCBSD, Figure 78) and ZnO was introduced by Hsieh et al.71 Inverted P3HT/[60]PCBM organic solar cells with ITO cathode covered with ZnO buffer layer yielded power conversion efficiency of 3.5%. On the contrary, devices comprising hybrid HBETL represented by ZnO/poly-C-PCBSD junction yielded much higher power conversion efficiency of 4.4% mainly due to the increased photocurrent density (ISC = 12.8 mA cm−2 )

Carboxylic acid-based self-assembled molecules

ZnO

X

-OCH3 -CH3 X = -H -SH -CF3 C -CN O HO

P3HT:PCBM PEDOT:PSS ITO

BA-X

Glass

Figure 73 Schematic layout of a bulk heterojunction P3HT/[60]PCBM device with ZnO interlayer modified with different carboxylic acid monolayers. (Reproduced from Ref. 221.  Wiley-VCH, 2008.)

A somewhat similar example of self-assembling buffer layer was reported by Wei et al.70 A special fullerene derivative modified with a long perfluorinated carbon chain (F-PCBM, Figure 75) was mixed with conventional PCBM and P3HT in the active layer of organic bulk heterojunction solar cell. It was demonstrated that the FPCBM spontaneously migrates to the surface of the organic layer during the spin casting owing to the low surface energy of the fluorocarbon chain, and forms a very thin FPCBM buffer layer at the interface between the cathode and fullerene/polymer blend (Figure 76). The F-PCBM behaves as HBETL material and improves dramatically electrical characteristics of the photovoltaic cell. In particular, the FF of the devices with F-PCBM buffer layer approaches O O P O

−

Br N

N(CH3 )3+Br−

N Br−

O P O O

N(CH3 )3+Br− O

O O O

n

PFPNBr

n

n

PFNBr-PFMEE

PF-PO(Et)2

Br− N C8H17

N

Br−

C8H17

C8H17

N

C8H17 S

n

S

N S m

PFNBr-DBT15

Figure 74

Molecular structures of some polymer-based HBETL materials.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

PEG monolayer

P3HT:PCBM + 5% PEG

O H

O

H

O

n OH

H

O n OH

H

O

(a)

O

n OH

O O

O n OHH H n OH

O

Metal Cathode (AI 100nm) O O

n OH n OH

n OH

H

O

n OH

H

H

n OH

H

P3HT:PCBM

P3HT:PCBM

PEDOT:PSS ITO/glass

PEDOT:PSS ITO/glass

n OH

H

Solvent evaporation

O

H

PEDOT:PSS ITO/glass

n OHH n OH H

O

n OH

H

The interaction of AI and PEG lowered the injection barrier height and contact resistance

Dried film:

Wet film:

59

n OH

PEDOT:PSS ITO/glass (b)

(c)

(d)

Figure 75 The formation of vertically segregated P3HT + PCBM/PEG system where PEG behaves as a buffer layer modifying aluminum cathode/active layer interface. (Reproduced from Ref. 225.  Royal Society of Chemistry, 2009.) F

F F F F F F F

AI

F FF F F FF F F FF FF F F F

O

F-PCBM

F FF FF FF

O

P3HT

(a)

PCBM PEDOT:PSS ITO

(b)

Figure 76 (a) Molecular structure of F-PCBM and (b) a schematic layout of organic bulk heterojunction solar cell with self-assembled F-PCBM buffer layer. (Reproduced from Ref. 70.  Wiley-VCH, 2008.)

Ag PEDOT:PSS

O

COOH N

P3HT/ PCBM SAM TiO2 ITO Glass

S

S S

O

COOH

O

O

COOH COOH

Figure 77 A schematic layout of inverted bulk heterojunction organic solar cell with ZnO buffer layer modified with self-assembled monolayers (SAMs) of C60 -Py-COOH, terthiophenecarboxylic acid, benzoic acid, and dodecanoic acid. Inset shows molecular structures of the modifiers. (Reproduced from Ref. 227.  Royal Society of Chemistry, 2008.)

and fill factor (FF = 58%). At the same time, use of polymerized fullerene interlayer improved the efficiency of inverted solar cells based on low band gap PCPDTBT copolymer (molecular structure is shown in Figure 63) from 1.9 to 3.4%. This example illustrates high potential of the fullerene-based hole-blocking electron transport materials in organic photovoltaics. There are many other examples of EBHTL and HBETL materials applied in bulk heterojunction solar cells.

C-PCBSD

Figure 78 Molecular structure of polymerizable fullerene derivative C-PCBSD.

Interested reader might be referred to recent reviews addressing this subject in more details.73–76, 228

8

CONCLUSION AND OUTLOOK

In their publication, Scharber et al. presented theoretical modeling results suggesting that it is possible to design

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular devices

Power conversion efficiency, h (%)

60

6. M. K. Nazeeruddin, P. Pechy, T. Renouard, et al., J. Am. Chem. Soc., 2001, 123, 1613–1624.

11 10 9 8 7 6 5 4 3 2 1 0

7. Y. Cao, Y. Bai, Q. Yu, et al., J. Phys. Chem. C, 2009, 113, 6290–6297. 8. J.-J. Cid, J.-H. Yum, S.-R. Jang, et al., Angew. Chem. Int. Ed., 2007, 46, 8358–8362. 9. T. Bessho, S. M. Zakeeruddin, C.-Y. Yeh, et al., Angew. Chem. Int. Ed., 2010, 49, 6646–6649. 10. T. Hasobe, H. Imahori, P. V. Kamat, et al., J. Am. Chem. Soc., 2005, 127, 1216–1228. 1985

Figure 79

1990

1995

2000 Year

2005

2010

Evolution of organic solar cells.

some electron donor material which might give power conversion efficiency well above 10% in a single junction cells with [60]PCBM.81 At the same time, double junction organic solar cells were predicted to give power conversion efficiencies up to 15% using PCBM as electron acceptor material.229 If the rate of the development of organic solar cells presented in Figure 79 continues in future, the 10–12% organic solar cells are reported within the next couple of years. The scientific challenge and progress in supramolecular (self) assembly still accelerate this progress heavily. The achievement of higher efficiency organic solar cells will go hand in hand with our understanding of the supramolecular interactions within the organic solar cell materials and thin-film interfaces.

11. T. Hasobe, P. V. Kamat, V. Troiani, et al., J. Phys. Chem. B, 2005, 109, 19–23. 12. T. Hasobe, S. Fukuzumi, and P. V. Kamat, J. Phys. Chem. B, 2006, 110, 25477–25484. 13. T. Hasobe, S. Fukuzumi, and P. V. Kamat, Angew. Chem. Int. Ed., 2006, 45, 755–759. 14. T. Hasobe, H. Murata, and P. V. Kamat, J. Phys. Chem. C, 2007, 111, 16626–16634. 15. Y. Liang, Z. Xu, J. Xia, et al., Adv. Mater., 2010, 22, E135–E138. 16. G. Zhao, Y. He, and Y. Li, Adv. Mater., 2010, 39, 4355–4358. 17. F. Yang, K. Sun, and S. R. Forrest, Adv. Mater., 2007, 19, 4166–4171. 18. Y. Matsuo, Y. Sato, T. Niinomi, et al., J. Am. Chem. Soc., 2009, 131, 16048–16050. 19. K. Schulze, C. Uhrich, R. Sch¨uppel, et al., Adv. Mater., 2006, 18, 2872–2875. 20. B. Walker, A. B. Tamayo, X.-D. Dang, et al., Adv. Funct. Mater., 2009, 19, 3063–3069. 21. X.-D. Dang, A. B. Tamayo, J. Seo, et al., Adv. Funct. Mater., 2010, 20, 3314–3321.

ACKNOWLEDGMENTS

22. N. M. Kronenberg, V. Steinmann, H. B¨urckst¨ummer, et al., Adv. Mater., 2010, 22, 4193–4197.

This work was supported by the Russian Foundation for Basic Research (RFBR Grant 10-03-00443a), Russian President Foundation (MK-4916.2011.3) and the Federal Agency for Science and Innovations (Contract No. 02.740.11.0749).

23. B. Sun and N. C. Greenham, Phys. Chem. Chem. Phys., 2006, 8, 3557–3560. 24. E. Kymakis, E. Koudoumas, I. Franghiadakis, and G. A. J. Amaratunga, J. Phys. D: Appl. Phys., 2006, 39, 1058–1063. 25. S. Kazaoui, N. Minami, B. Nalini, et al., J. Appl. Phys., 2005, 98, 084314.

REFERENCES

26. D. Yu, Y. Yang, M. Durstock, et al., ACS Nano, 2010, 4, 5633–5640.

1. J. Zhao, A. Wang, M. A. Green, and F. Ferrazza, Appl. Phys. Lett., 1998, 73, 1991–1993.

27. Z. Liu, D. He, Y. Wang, et al., Sol. Energ. Mater. Sol. Cells, 2010, 94, 1196–1200.

2. S. E. Shaheen, D. S. Ginley, and G. E. Jabbour, MRS Bull., 2005, 30, 10–20.

28. C. R. McNeill, A. Abrusci, J. Zaumseil, et al., Appl. Phys. Lett., 2007, 90, 193506.

3. M. A. Green, K. Emery, Y. Hishikawa, and W. Warta, Prog. Photovolt: Res. Appl., 2010, 18, 144–150.

29. S. Sun and N. S. Sariciftci, eds., Organic Photovoltaics: Mechanism, Materials and Devices, CRC Press, 2005.

4. http://www.spiritus-temporis.com/thin-film-solar-cell/ production-cost-and-market.html (accessed 4 October 2010).

30. P. Wurfel, Physics of Solar Cells, Wiley-VCH Verlag GmbH, Weinheim, 2005.

5. M. A. Green, K. Emery, Y. Hishikawa, and W. Warta, Prog. Photovolt: Res. Appl., 2010, 18, 346–352.

31. C. Brabec, U. Scherf, and V. Dyakonov, eds., Organic Photovoltaics: Materials, Device Physics, and Manufacturing Technologies, Wiley-VCH Verlag GmbH, Weinheim, 2008.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics

61

32. A. Heeger, N. S. Sariciftci, and E. B. Namdas, eds., Semiconducting and Metallic Polymers, Oxford Graduate Texts, Oxford University Press, Oxford, New York, 2010.

57. J. Nelson, The Physics of Solar Cells, Imperial College Press, 2003.

33. C. Brabec, V. Dyakonov, J. Parisi, and N. S. Sariciftci, eds., Organic Photovoltaics: Concepts and Realization (Springer Series in Materials Science), Springer, Berlin, Heidelberg, New York, Hong Kong, London, Milan, Paris, Tokio 2003.

59. J. Peet, A. J. Heeger, and G. C. Bazan, Acc. Chem. Res., 2009, 42, 1700–1708.

34. J. Desilvestro, M. Gratzel, L. Kavan, et al., J. Am. Chem. Soc., 1985, 107, 2988–2990. 35. N. Vlachopoulos, P. Liska, J. Augustynski, and M. Gratzel, J. Am. Chem. Soc., 1988, 110, 1216–1220. 36. S. Yanagida, Y. Yu, and K. Manseki, Acc. Chem. Res., 2009, 42, 1827–1838. 37. T. Hasobe, Phys. Chem. Chem. Phys., 2010, 12, 44–57. 38. H. Imahori and Y. Sakata, Eur. J. Org. Chem., 1999, 2445–2457. 39. H. Yamada, H. Imahori, Y. Nishimura, et al., J. Am. Chem. Soc., 2003, 125, 9129–9139.

58. M. Gratzel, Acc. Chem. Res., 2009, 42, 1788–1798.

60. V. D. Mihailetchi, L. J. A. Koster, J. C. Hummelen, and P. W. M. Blom, Phys. Rev. Lett., 2004, 93, 216601. 61. A. C. Morteani, P. Sreearunothai, L. M. Herz, et al., Phys. Rev. Lett., 2004, 92, 247240. 62. L. J. A. Koster, E. C. P. Smiths, V. D. Mihailetchi, and P. W. M. Blom, Phys. Rev. B, 2005, 72, 085205. 63. J. Nelson, Phys. Rev. B, 2003, 67, 155209. 64. G. Boschloo and A. Hagfeldt, Acc. Chem. Res., 2009, 42, 1819–1826. 65. C. Tao, S. Ruan, G. Xie, et al., Appl. Phys. Lett., 2009, 94, 043311. 66. D. W. Zhao, P. Liu, X. W. Sun, et al., Appl. Phys. Lett., 2009, 95, 15330.

40. C. W. Tang, Appl. Phys. Lett., 1986, 48, 183–185.

67. H.-H. Liao, L.-M. Chen, Z. Xu, et al., Appl. Phys. Lett., 2008, 92, 173303.

41. M. Theander, A. Yartsev, D. Zigmantas, et al., Phys. Rev. B, 2000, 61((),12), 957–12963.

68. S. K. Hau, H.-L. Yip, N. S. Baek, et al., Appl. Phys. Lett., 2008, 92, 253301.

42. D. E. Markov, C. Tanase, P. W. M. Blom, J. Wildeman, Phys. Rev. B, 2005, 72, 045217.

and

69. G. K. Mor, K. Shankar, M. Paulose, et al., Appl. Phys. Lett., 2007, 91, 152111.

43. D. E. Markov, E. Amsterdam, P. W. M. Blom, J. Phys. Chem. A, 2005, 109, 5266–5274.

et al.,

70. Q. Wei, T. Nishizawa, K. Tajima, and K. Hashimoto, Adv. Mater., 2008, 20, 2211–2216.

44. J. J. M. Halls, K. Pichler, R. H. Friend, et al., Appl. Phys. Lett., 1996, 68, 3120–3122.

71. C.-H. Hsieh, Y.-J. Cheng, P.-J. Li, et al., J. Am. Chem. Soc., 2010, 132, 4887–4893.

45. C. Schlebusch, B. Kessler, S. Cramm, and W. Eberhardt, Synth. Met., 1996, 77, 151–154.

72. Y. Shirota and H. Kageyama, Chem. Rev., 2007, 107, 953–1010.

46. D. Wohrle and D. Meissner, Adv. Mater., 1991, 3, 129–138. 47. R. Fitzner, E. Reinold, A. Mishra, et al., Adv. Funct. Mater., 2011, 21, 897–910. 48. N. S. Sariciftci, L. Smilowitz, A. J. Heeger, and F. Wudl, Science, 1992, 258, 1474–1476.

73. L.-M. Chen, Z. Hong, G. Li, and Y. Yang, Adv. Mater., 2009, 21, 1434–1449. 74. F. Huang, H. Wu, and Y. Cao, Chem. Soc. Rev., 2010, 39, 2500–2521. 75. R. Steim, F. R. Kogler, and C. J. Brabec, J. Mater. Chem., 2010, 20, 2499–2512.

49. N. S. Sariciftci and A. J. Heeger, US Patent 5331183.

76. L.-M. Chen, Z. Xu, Z. Hong, and Y. Yang, J. Mater. Chem., 2010, 20, 2575–2598.

50. G. Yu, J. Gao, J. C. Hummelen, et al., Science, 1995, 270, 1789–1791.

77. H. Hoppe, S. Shokhovets, and G. Gobsch, Phys. Stat. Sol., 2007, (RRL) 1, R40–R42.

51. S. H. Park, A. Roy, S. Beaupre, et al., Nat. Photon., 2009, 3, 297–303.

78. P. A. Troshin, H. Hoppe, A. S. Peregudov, et al., ChemSusChem, 2011, 4, 119–124.

52. A. C. Arias, M. Granstrom, D. S. Thomas, et al., Phys. Rev. B, 1999, 60, 1854–1860.

79. C. J. Brabec, A. Cravino, D. Meissner, et al., Adv. Funct. Mater., 2001, 11, 374.

53. J. Y. Kim, S. H. Kim, H. H. Lee, et al., Adv. Mater., 2006, 18, 572–576.

80. F. B. Kooistra, J. Knol, F. Kastenberg, et al., Org. Lett., 2007, 9, 551–554.

54. V. Sgobba and D. M. Guldi, J. Mater. Chem., 2008, 18, 153–157.

81. M. C. Scharber, D. Muhlbacher, M. Koppe, et al., Adv. Mater., 2006, 18, 789–794.

55. H. Hoppe and N. S. Sariciftci, Polymer solar cells, in Advances in Polymer Science, Photoresponsive Polymers II, eds. S. R. Marder and K.-S. Lee, Publ.: Springer, BerlinHeidelberg, 2008, pp. 1–86.

82. G. K. Mor, K. Shankar, M. Paulose, et al., Nano Lett., 2006, 6, 215–218.

56. P. W. M. Blom, V. D. Mihailetchi, L. J. A. Koster, and D. E. Markov, Adv. Mater., 2007, 19, 1551–1566.

84. Y. Kashiwa, Y. Yoshida, and S. Hayase, Appl. Phys. Lett., 2008, 92, 033308.

83. B. Weintraub, Y. Wei, and Z. L. Wang, Angew. Chem. Int. Ed., 2009, 48, 8981–8985.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

62

Supramolecular devices

85. N. K. Subbaiyan, C. A. Wijesinghe, and F. D’Souza, J. Am. Chem. Soc., 2009, 131, 14646–14647.

113. A. Kongkanand, R. M. Dominguez, and P. V. Kamat, Nano Lett., 2007, 7, 676–680.

86. F. D’Souza, G. R. Deviprasad, M. E. Zandler, J. Phys. Chem. A, 2002, 106, 3243–3252.

114. D. Eder, Chem. Rev., 2010, 110, 1348–1385.

et al.,

89. D. Bonifazi, O. Enger, and F. Diederich, Chem. Soc. Rev., 2007, 36, 390–414.

115. C.-C. Chu and D. M. Bassani, Photochem. Photobiol. Sci., 2008, 7, 521–530. ´ Herranz, F. Giacalone, L. S´anchez, and N. Mart´ın, 116. M. A. in Fullerenes Principles and Applications, eds. F. L. De La Puente and J.-F. Nierengarten, RSC Pub., 2007, pp. 152–191.

90. M. Isosomppi, N. V. Tkachenko, A. Efimov, J. Mater. Chem., 2005, 15, 4546–4554.

117. L. Sanchez, N. Martin, and D. M. Guldi, Angew. Chem. Int. Ed., 2005, 44, 5374–5382.

87. I. Robel, V. Subramanian, M. Kuno, and P. V. Kamat, J. Am. Chem. Soc., 2006, 128, 2385–2393. 88. D. M. Guldi, Chem. Soc. Rev., 2002, 31, 22–36.

et al.,

91. H. Imahori, J. Phys. Chem. B, 2004, 108, 6130–6143. 92. Y. J. Cho, T. K. Ahn, H. Song, et al., J. Am. Chem. Soc., 2005, 127, 2380–2381. 93. Y. Matsuo, T. Ichiki, S. G. Radhakrishnan, et al., J. Am. Chem. Soc., 2010, 132, 6342–6348. 94. D. M. Guldi, G. M. A. Rahman, V. Sgobba, and C. Ehli, Chem. Soc. Rev., 2006, 35, 471–487. 95. D. M. Guldi and M. Prato, Chem. Commun., 2004, 2517–2525. 96. G. Accorsi and N. Armaroli, J. Phys. Chem. C, 2010, 114, 1385–1403. 97. V. Sgobba, G. Giancane, S. Conoci, et al., J. Am. Chem. Soc., 2007, 129, 3148–3156. 98. S. Conoci, D. Guldi, S. Nardis, et al., Chem. Eur. J., 2004, 10, 6523–6530. 99. C. Luo, D. M. Guldi, M. Maggini, et al., Angew. Chem. Int. Ed., 2000, 39, 3905–3909. 100. D. M. Guldi, I. Zilbermann, G. Anderson, et al., J. Mater. Chem., 2005, 15, 114–118. 101. D. M. Guldi, I. Zilbermann, A. Lin, et al., Chem. Commun., 2004, 96–97. 102. D. M. Guldi, I. Zilbermann, G. A. Anderson, et al., Chem. Commun., 2004, 726–728. 103. D. M. Guldi, G. M. A. Rahman, F. Zerbetto, and M. Prato, Acc. Chem. Res., 2005, 38, 871–878. 104. D. Tasis, N. Tagmatarchis, A. Bianco, and M. Prato, Chem. Rev., 2006, 106, 1105–1136. 105. D. M. Guldi, G. M. A. Rahman, N. Jux, et al., Angew. Chem. Int. Ed., 2005, 44, 2015–2018. 106. D. M. Guldi, G. M. A. Rahman, J. Ramey, et al., Chem. Commun., 2004, 2034–2035. 107. G. M. A. Rahman, A. Troeger, V. Sgobba, et al., Chem. Eur. J., 2008, 14, 8837–8846.

118. N. D. McClenaghan, Z. Grote, K. Darriet, et al., Org. Lett., 2005, 7, 807–810. 119. C. H. Huang, N. D. McClenaghan, A. Kuhn, et al., Org. Lett., 2005, 7, 3409–3412. 120. E. H. A. Beckers, P. A. van Hal, A. Schenning, et al., J. Mater. Chem., 2002, 12, 2054–2060. 121. C. H. Huang, N. D. McClenaghan, A. Kuhn, et al., Tetrahedron, 2006, 62, 2050–2059. 122. M. Murakami, K. Ohkubo, T. Hasobe, et al., J. Mater. Chem., 2010, 20, 1457–1466. 123. M. Ohtani, P. V. Kamat, and S. Fukuzumi, J. Mater. Chem., 2010, 20, 582–587. 124. R. Bhosale, A. Perez-Velasco, V. Ravikumar, Angew. Chem. Int. Ed., 2009, 48, 6461–6464.

et al.,

125. R. S. K. Kishore, O. Kel, N. Banerji, et al., J. Am. Chem. Soc., 2009, 131, 11106–11116. 126. T. Hasobe, H. Imahori, S. Fukuzumi, and P. V. Kamat, J. Phys. Chem. B, 2003, 107, 12105–12112. 127. T. Hasobe, Y. Kashiwagi, M. A. Absalom, et al., Adv. Mater., 2004, 16, 975–979. 128. T. Hasobe, K. Saito, P. V. Kamat, et al., J. Mater. Chem., 2007, 17, 4160–4170. 129. A. S. D. Sandanayaka, T. Murakami, and J. Phys. Chem. C, 2009, 113, 18369–18378.

T. Hasobe,

130. N. Hadjichristidis and S. Pispas, Adv. Polym. Sci., 2006, 200, 37–55. 131. S. Sioula, N. Hadjichristidis, and E. L. Thomas, Macromolecules, 1998, 31, 5272–5277. 132. S. Sioula, N. Hadjichristidis, and E. L. Thomas, Macromolecules, 1998, 31, 8429–8432. 133. U. Scherf, A. Gutacker, and N. Koenen, Acc. Chem. Res., 2008, 41, 1086–1097. 134. R. A. Segalman, B. McCulloch, S. Kirmayer, and J. J. Urban, Macromolecules, 2009, 42, 9205–9216.

108. V. Sgobba, G. M. A. Rahman, D. M. Guldi, et al., Adv. Mater., 2006, 18, 2264–2269.

135. L. Botiz and S. B. Darling, Mater. Today, 2010, 13, 42–51.

109. G. M. A. Rahman, D. M. Guldi, R. Cagnoli, et al., J. Am. Chem. Soc., 2005, 127, 10051–10057.

136. S. M. Lindner, S. Huttner, A. Chiche, et al., Angew. Chem. Int. Ed., 2006, 45, 3364–3368.

110. V. Sgobba, A. Troeger, R. Cagnoli, et al., J. Mater. Chem., 2009, 19, 4319–4324.

137. M. Sommer, S. M. Lindner, and M. Thelakkat, Adv. Funct. Mater., 2007, 17, 1493–1500.

111. G. Giancane, A. Ruland, V. Sgobba, et al., Adv. Funct. Mater., 2010, 20, 2481–2488.

138. M. Sommer, A. S. Lang, and M. Thelakkat, Angew. Chem. Int. Ed., 2008, 47, 7901–7904.

112. D. M. Guldi, G. M. A. Rahman, V. Sgobba, et al., J. Am. Chem. Soc., 2006, 128, 2315–2323.

139. M. Sommer, S. H¨uttner, U. Steiner, and M. Thelakkat, Appl. Phys. Lett., 2009, 95, 183308.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Supramolecular chemistry for organic photovoltaics 140. S. B. Darling, Energy Environ. Sci., 2009, 2, 1266–1273. 141. D. K. Susarova, P. A. Troshin, D. H¨oglinger, et al., Sol. Energ. Mater. Sol. Cells, 2010, 94, 803–810. 142. P. A. Troshin, R. Koeppe, A. S. Peregudov, et al., Chem. Mater., 2007, 19, 5363–5372. 143. R. Koeppe, P. A. Troshin, A. Fuchsbauer, et al., Fuller. Nanotub. Carb. Nanostruct., 2006, 14, 441–446. 144. R. Koeppe, P. A. Troshin, R. N. Lyubovskaya, and N. S. Sariciftci, Appl. Phys. Lett., 2005, 87, 244102. 145. P. A. Troshin, S. I. Troyanov, G. N. Boiko, et al., Fuller. Nanot. Carb. Nanostruct., 2004, 12, 435–441. 146. A. B. Tamayo, B. Walker, and T.-Q. Nguyen, J. Phys. Chem. C, 2008, 112, 11545–11551. 147. E. A. Kleymyuk, P. A. Troshin, Y. N. Luponosov, et al., Energy Environ. Sci., 2010, 3, 1941–1948. 148. P. A. Troshin, S. A. Ponomarenko, Y. N. Luponosov, et al., Sol. Energy Mater. Sol. Cells, 2010, 94, 2067–2072. 149. A. L. Kanibolotsky, I. F. Perepichka, and P. J. Skabara, Chem. Soc. Rev., 2010, 39, 2695–2728. 150. L. Schmidt-Mende, A. Fechtenkotter, K. Mullen, et al., Science, 2001, 293, 1119–1122. 151. M. Carrasco-Orozco, W. C. Tsoi, M. O’Neill, et al., Adv. Mater., 2006, 18, 1754–1758. 152. M. K. R. Fischer, I. L. Duarte, M. M. Wienk, et al., J. Am. Chem. Soc., 2009, 131, 8669–8676. 153. M. Helgesen, R. Søndergaard, and F. C. Krebs, J. Mater. Chem., 2010, 20, 36–60. 154. F. G. Brunetti, R. Kumar, and F. Wudl, J. Mater. Chem., 2010, 20, 2934–2948.

63

167. W. Ma, C. Yang, and A. J. Heeger, Adv. Mater., 2007, 19, 1387–1390. 168. Y. Yao, J. Hou, Z. Xu, et al., Adv. Funct. Mater., 2008, 18, 1783–1789. 169. S. Berson, R. De Bettignies, S. Bailly, and S. Guillerez, Adv. Funct. Mater., 2007, 17, 1377–1384. 170. S. S. van Bavel, E. Sourty, G. de With, and J. Loos, Nano Lett., 2009, 9, 507–513. 171. G. Dennler, M. C. Scharber, and C. J. Brabec, Adv. Mater., 2009, 21, 1323–1338. 172. M. Drees, H. Hoppe, C. Winder, et al., J. Mater. Chem., 2005, 15, 5158–5163. 173. J.-F. Nierengartena and S. Setayesh, New J. Chem., 2006, 30, 313–316. 174. K. Sivula, C. K. Luscombe, B. C. Thompson, and J. M. J. Frechet, J. Am. Chem. Soc., 2006, 128, 13988–13989. 175. K. Sivula, Z. T. Ball, N. Watanabe, and J. M. J. Fr´echet, Adv. Mater., 2006, 18, 206–210. 176. J.-H. Tsai, Y.-C. Lai, T. Higashihara, molecules, 2010, 43, 6085–6091.

et al.,

Macro-

177. J. U. Lee, J. W. Jung, T. Emrick, et al., Nanotechnology, 2010, 21, 105201. 178. C. Su Kim, L. L. Tinker, B. F. DiSalle, et al., Adv. Mater., 2009, 21, 3110–3115. 179. P. A. Troshin, E. A. Khakina, M. Egginger, et al., ChemSusChem, 2010, 3, 356–366. 180. P. A. Troshin, H. Hoppe, J. Renz, et al., Adv. Funct. Mater., 2009, 19, 779–788. 181. P. A. Troshin, D. K. Susarova, E. A. Khakina, et al., Adv. Funct. Mater., 2011. submitted.

155. J. L. Delgado, P.-A. Bouit, S. Filippone, et al., Chem. Commun., 2010, 46, 4853–4865.

182. D. Muhlbacher, M. Scharber, M. Morana, et al., Adv. Mater., 2006, 18, 2884–2889.

156. K. M. Coakley and M. D. McGehee, Chem. Mater., 2004, 16, 4533–4542.

183. J. K. Lee, W. L. Ma, C. J. Brabec, et al., J. Am. Chem. Soc., 2008, 130, 3619–3623.

157. H. Hoppe and N. S. Sariciftci, J. Mater. Chem., 2006, 16, 45–61.

184. J. Hou, H.-Y. Chen, S. Zhang, et al., J. Am. Chem. Soc., 2008, 130, 16144–16145.

158. H. Hoppe, M. Niggemann, C. Winder, et al., Adv. Funct. Mater., 2004, 14, 1005–1011.

185. M. C. Scharber, M. Koppe, J. Gao, et al., Adv. Mater., 2010, 22, 367–370.

159. S. E. Shaheen, C. J. Brabec, N. S. Sariciftci, et al., Appl. Phys. Lett., 2001, 78, 841–843.

186. F. Zhang, K. G. Jespersen, C. Bj¨orstr¨om, et al., Adv. Funct. Mater., 2006, 16, 667–674.

160. H. Hoppe, T. Glatzel, M. Niggemann, et al., Nano Lett., 2005, 5, 269–274.

187. E. Wang, L. Wang, L. Lan, et al., Appl. Phys. Lett., 2008, 92, 033307.

161. H. Hoppe, T. Glatzel, M. Niggemann, et al., Thin Solid Films, 2006, 511, 587–592.

188. J. Peet, J. Y. Kim, N. E. Coates, et al., Nat. Mater., 2007, 6, 498–500.

162. F. Padinger, R. S. Rittberger, and N. S. Sariciftci, Adv. Funct. Mater., 2003, 13, 85–88.

189. K. W. Cheng, C. S. C. Mak, W. K. Chan, et al., J. Polym. Sci. A: Polym. Chem., 2008, 46, 1305–1317.

163. W. Ma, C. Yang, X. Gong, et al., Adv. Funct. Mater., 2005, 15, 1617–1622.

190. T.-C. Liang, I.-H. Chiang, P.-J. Yang, et al., J. Polym. Sci., Part A: Polym. Chem., 2009, 47, 5998–6013.

164. G. Li, V. Shrotriya, J. Huang, et al., Nat. Mater., 2005, 4, 864–868.

191. W.-Y. Wong, X.-Z. Wang, Z. He, et al., Nat. Mater., 2007, 6, 521–526.

165. G. Li, Y. Yao, H. Yang, et al., Adv. Funct. Mater., 2007, 17, 1636–1644.

192. J. Gilot, M. M. Wienk, and R. A. J. Janssen, Nat. Mater., 2007, 6, 704.

166. A. J. Moul´e and K. Meerholz, Adv. Mater., 2008, 20, 240–245.

193. P.-T. Wu, T. Bull, F. S. Kim, et al., Macromolecules, 2009, 42, 671–681.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

64

Supramolecular devices

194. N. S. Baek, S. K. Hau, H.-L. Yip, et al., Chem. Mater., 2008, 20, 5734–5736.

213. S. Okada, K. Okinaka, H. Iwawaki, et al., Dalton Trans., 2005, 1583–1590.

195. B. L. Groenendaal, F. Jonas, D. Freitag, et al., Adv. Mater., 2000, 12, 481–494.

214. J. Huang, G. Li, E. Wu, et al., Adv. Mater., 2006, 18, 114–117.

196. H. Frohne, S. E. Shaheen, C. J. Brabec, ChemPhysChem, 2002, 3, 795–799.

et al.,

215. J. Huang, T. Watanabe, K. Ueno, and Y. Yang, Adv. Mater., 2007, 19, 739–743.

197. S.-I. Na, G. Wang, S.-S. Kim, et al., J. Mater. Chem., 2009, 19, 9045–9053.

216. J. Huang, W. J. Hou, J. H. Li, et al., Appl. Phys. Lett., 2006, 89, 133509.

198. Y. Yang and A. H. Heeger, Appl. Phy. Lett., 1994, 64, 1245–1247.

217. J. Huang, Z. Xu, and Y. Yang, Adv. Funct. Mater., 2007, 17, 1966–1973.

199. J. Gao, A. J. Heeger, J. Y. Lee, and C. Y. Kim, Synth. Met., 1996, 82, 221–223.

218. J. Huang, G. Li, and Y. Yang, Adv. Mater., 2008, 20, 415–419.

200. M. Turbiez, P. Frere, P. Leriche, et al., Chem. Commun., 2005, 1161–1163.

219. J. Gilot, M. M. Wienk, and R. A. J. Janssen, Appl. Phys. Lett., 2007, 90, 143512.

201. H. Yan, P. Lee, N. R. Armstrong, et al., J. Am. Chem. Soc., 2005, 127, 3172–3183.

220. J. Gilot, I. Barbu, M. M. Wienk, and R. A. J. Janssen, Appl. Phys. Lett., 2007, 91, 113520.

202. A. W. Hains, C. Ramanan, M. D. Irwin, et al., ACS Appl. Mater. Interfaces, 2010, 2, 175–185.

221. H. L. Yip, S. K. Hau, N. S. Baek, et al., Adv. Mater., 2008, 20, 2376–2382.

203. V. Shrotriya, G. Li, Y. Yao, et al., Appl. Phys. Lett., 2006, 88, 073508.

222. J. Luo, H. B. Wu, C. He, et al., Appl. Phys. Lett., 2009, 95, 043301.

204. G. Li, C.-W. Chu, V. Shrotriya, et al., Appl. Phys. Lett., 2006, 88, 253503.

223. Y. Zhao, Z. Y. Xie, C. J. Qin, et al., Sol. Energy Mater. Sol. Cells, 2009, 93, 604–608.

205. C. J. Brabec, S. E. Shaheen, C. Winder, et al., Appl. Phys. Lett., 2002, 80, 1288–1290.

224. S.-I. Na, S.-H. Oh, S.-S. Kim, and D.-Y. Kim, Org. Electron., 2009, 10, 496–500.

206. B. N. Limketkai and M. A. Baldo, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 71, 085207.

225. F.-C. Chen and S.-C. Chien, J. Mater. Chem., 2009, 19, 6865–6869.

207. S. K. M. Jonsson, E. Carlegrim, F. Zhang, et al., Jpn. J. Appl. Phys., 2005, 44, 3695–3701.

226. J. S. Kim, Y. Lee, J. H. Lee, et al., Adv. Mater., 2010, 22, 1355–1360.

208. G. E. Jabbour, B. Kippelen, N. R. Armstrong, and N. Peyghambarian, Appl. Phys. Lett., 1998, 73, 1185–1187.

227. S. K. Hau, H.-L. Yip, O. Acton, et al., J. Mater. Chem., 2008, 18, 5113–5119.

209. J. Lee, Y. Park, S. K. Lee, et al., Appl. Phys. Lett., 2002, 80, 3123–3125.

228. H. Ma, H.-L. Yip, F. Huang, and A. K.-Y. Jen, Adv. Funct. Mater., 2010, 20, 1371–1388.

210. C. H. Lee, Synth. Met., 1997, 91, 125–127.

229. T. Ameri, G. Dennler, C. Lungenschmied, and C. J. Brabec, Energy Environ. Sci., 2009, 2, 347–363.

211. Y. S. Lee, J.-H. Park, Y.-H. Kwak, et al., Mol. Cryst. Liq. Cryst., 2003, 405, 89–95. 212. J. Lee, Y. Park, D. Y. Kim, et al., Appl. Phys. Lett., 2003, 82, 173.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc105

Crystal Engineering Edward R. T. Tiekink University of Malaya, Kuala Lumpur, Malaysia

1 Introduction 2 On the Way Molecules Pack—the Heart of the Matter 3 Applications Part I—Multicomponent Crystals 4 Applications Part II—Coordination Polymers 5 Conclusions and Prospects Acknowledgments References Further Reading

1

1 2 16 24 33 33 33 38

INTRODUCTION

It is the great promise of developing useful applications in fields as diverse as pharmaceuticals (polymorph control and cocrystal formation), photoluminescent materials, metalorganic framework structures (gas storage and catalysis), nonlinear optical (NLO) materials, and so on that draw together chemists, physicists, and materials scientists under the aegis of the relatively new discipline of crystal engineering. So, what is crystal engineering? An early and succinct definition was proffered by Desiraju in the seminal volume Crystal Engineering. The Design of Organic Solids,1 whereby crystal engineering is defined as the understanding of intermolecular interactions in the context of crystal packing and in the utilization of such understanding in the design of new solids with desired physical and chemical properties. While referring primarily to organic systems, this definition applies equally in

the realm of metalorganic crystal engineering, and has stood the test of time. Put simply, whereas the synthetic chemist aims to generate molecules by the formation of covalent bonds (incorporating coordinate or dative bonds for the metalorganic chemist), the crystal engineer seeks to arrange molecules in the crystalline state in a predesigned manner by exploiting intermolecular interactions. Building on the concepts of supramolecular chemistry, Dunitz’s elegant description encompasses much of the above by referring to a crystal as a supermolecule par excellence.2 The challenge confronting the crystal engineer is therefore apparent in that intermolecular interactions must be controlled. The prevalence, directionality, and the strength (i.e., decidedly weaker compared to covalent bonds) of the myriad of intermolecular forces that have been identified make their control arguably akin to herding cats —try and get felines arranged in an orderly manner—the successes of crystal engineering exercises based on hydrogen bonding and/or coordinate bonding notwithstanding. It would seem that the first time the term “crystal engineering” was mentioned in the open literature was in relation to the work of Pepinsky reported over 50 years ago in an abstract titled Crystal Engineering: A New Concept in Crystallography.3 While probably not so closely related to the present accepted definition—Pepinsky was interested in deliberately modifying crystals in order to optimize their anomalous scattering effects for absolute structure determination of organic molecules—Pepinsky’s concept found ready application in Schmidt’s often cited work relating to the solid-state photochemical reactions.4 It is probably the latter work wherein molecules needed to be arranged specifically to allow, for example, photodimerization to occur in the crystalline phase, that laid the foundation to subsequent crystal engineering endeavors, both organic and metalorganic.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

2

Supramolecular materials chemistry

It is arguable that crystal engineering came of age as a recognizable discipline in its own right in the late 1990s with the appearance of journals catering to the burgeoning number of research groups undertaking research in this area. The first journal Crystal Engineering (Elsevier) published its initial papers in 1998 and this was closely followed by the Royal Society of Chemistry’s electronic only journal CrystEngComm (1999) and the American Chemical Society’s Crystal Growth & Design (2001). While Crystal Engineering ceased publication at the end of 2003, the remaining journals enjoy considerable ongoing success with large numbers of submissions and citations, resulting in these journals having high impact factors for chemistry journals, let alone for journals publishing crystallographic studies.5 Despite there being specialist journals on the subject, perhaps, surprisingly, the number of general reviews on crystal engineering per se are still relatively few (see Section Further Reading).6 These are augmented by a number of books and meeting transactions, each collecting state-of-the-art reviews on aspects of crystal engineering (see Section Further Reading). The dearth of specialist reviews on crystal engineering underscores the huge scope of the subject and the difficulties inherent with undertaking such a task. A general and illustrated approach, incorporating both organic and metalorganic crystal engineering themes, is quite likely the best manner by which such a review can be undertaken, and this approach is adopted herein. There are three pillars of crystal engineering. Naturally, these are interrelated and in a sense represent a continuum.6 The first paradigm in crystal engineering is the requirement to understand the nature of intermolecular forces or more correctly, the determination of factors that dictate the way molecules pack in the crystalline phase. While perhaps not as glamorous as the other two paradigms, it is only with a complete understanding of the ways molecules pack that rational crystal design can be achieved on a routine basis. This leads to the second axiom—the application of these principles to control the way molecules pack, that is, crystal design or supramolecular synthesis. The third paradigm, the primary motivation behind crystal engineering, is the optimization of the crystal’s properties to meet the desired application. Herein, the three paradigms are discussed in terms of global crystal packing considerations as well as what might be termed emerging intermolecular forces and other crystal packing principles that are described in the context of both organic and metalorganic systems, and from the perspective of a chemical crystallographer. Then the focus will turn to highlights in contemporary organic crystal engineering followed by an overview of metalorganic crystal engineering, both topics covered in more detail elsewhere in the present volume. While a historical perspective is offered for each of the topics

covered, as is appropriate, emphasis is given to describing more recent developments. Hence, some areas are skimmed over quite briefly or even omitted altogether.

2

2.1

ON THE WAY MOLECULES PACK—THE HEART OF THE MATTER Preamble

The underlying premise of crystal engineering is that it is possible to tailor crystal packing patterns in order to optimize desired physiochemical properties. This requires a detailed knowledge of global crystal packing considerations and the moderation of these by specific intermolecular interactions. Despite the fact that hydrogen bonding is widely exploited in crystal engineering, examples abound whereby these are usurped by nominally weaker intermolecular interactions. This underlines the idea that other or a combination of other intermolecular interactions can be/are important in determining the ultimate crystal structure or structures adopted by a particular molecule. The above is no better illustrated than for the supramolecular association formed by carboxylic acids, one of the most widely studied classes of compounds. Many might assume the prevalence of the eight-membered {· · ·HOC=O}2 homosynthon in their solid-state structures; see Figure 1(a) showing dimer formation in the crystal structure of tetrolic acid.7 However, experimental observations tell another story. In the most recent survey,8 it was shown that in only 31% of the 5690 organic structures where this synthon can potentially occur, it actually did. More often than not, the homosynthon shown in Figure 1(a) is usurped by competing supramolecular synthons. The observation of polymorphic structures of clonixin, 2-(2-methyl-3-chloroanilino)nicotinic acid,9 Figure 2(a) is highly illustrative of some of the principles pertinent to this section. For the neutral form of clonixin,9 the homosynthon is observed in two of the polymorphs with the arrangement stabilized by intramolecular N–H· · ·O hydrogen bonds formed between the amine and carbonyl groups (Figure 2b). In another polymorph, the pyridine-N atom comes into play. Thus, a supramolecular chain mediated by O–H· · ·N(pyridine) interactions is observed wherein the amine forms an intramolecular N–H· · ·O interaction to the carbonyl-O atom (Figure 3a). The situation is further complicated by the observation of a fourth form containing the zwitterionic version of clonixin.9 The supramolecular chain formed in this structure is mediated by N–H· · ·O hydrogen bonds (Figure 3b), and to a first approximation resembles the chain shown in Figure 3(a) but with relocation of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

3

(a)

(a)

(b)

Figure 1 Supramolecular association in tetrolic acid with the carboxylic acid carrying no additional functional groups. (a) Dimer formation via the eight-membered {· · ·HOCO}2 homosynthon and (b) catemer formation in the crystal structure of a second form of tetrolic acid. The O–H· · ·O hydrogen bonds are shown as orange dashed lines. In each of the motifs, the carboxylic acids are syn–syn. It is noted that for the catemer motif, anti–anti and, more rarely, alternating syn–anti carboxylic acids are known. N N H O H

Cl

O

(a)

(b)

Figure 2 Supramolecular association in the neutral forms of clonixin, 2-(2-methyl-3-chloroanilino)-nicotinic acid. (a) Chemical structure of clonixin and (b) dimer formation via the eight-membered {· · ·HOCO}2 homosynthon. The O–H· · ·O and N–H· · ·O hydrogen bonds are shown as orange dashed lines.

(b)

Figure 3 Supramolecular association in the neutral and zwitterionic forms of clonixin, 2-(2-methyl-3-chloroanilino)-nicotinic acid. (a) Supramolecular chain formation via O–H· · ·N(pyridine) hydrogen bonds in the neutral form and (b) supramolecular chain formation via pyridinium-N–H· · ·O hydrogen bonds. The O–H· · ·N and N–H· · ·O hydrogen bonds are shown as orange dashed lines.

acidic hydrogen atoms. While the now classic structural forms of clonixin are exceptional, they serve to emphasize the difficulties in controlling structural motifs, especially in cases of flexible molecules for which appellations such as conformational and synthon polymorphs are applicable.10 Returning to carboxylic acids in general, in the 474 circumstances where there were no competing synthons, the homosynthon was observed 93% of the time with the carboxylic acid catemer making up the balance.8 Figure 1(b) shows catemer formation in the crystal structure of a second polymorphic form of tetrolic acid.7 This second structure highlights what might be termed capricious behavior in these systems, whereby different polymorphs may present very different supramolecular structures (see Polymorphism: Fundamentals and Applications, Supramolecular Materials Chemistry for further discussions of polymorphs); modeling studies designed to determine the influence of solvent upon which form, dimer versus catemer, have been recently reported for tetrolic acid and related systems.11 The importance of exhaustive systematic evaluation of crystal structure is grandly exemplified here, whereby many research groups over many years have tackled the question as to why the homosynthon forms in preference to the catemer even though each features two O–H· · ·O hydrogen bonds. Conclusions proffered by Das and Desiraju indicate that the catemer might be preferred in cases where there is a supporting interaction,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

4

Supramolecular materials chemistry

for example, C–H· · ·O,12 which tips the balance in favor of chain formation. Herein, a discussion of global crystal packing considerations is followed by examples whereby these, rather than conventional intermolecular interactions, can be utilized to control supramolecular aggregation patterns. This is followed by an overview of different types of intermolecular interactions that have come to be recognized as important supramolecular glues.

2.2

Global crystal packing considerations: shape as a factor

This section could be subtitled the contribution of Kitaigorodskii to crystal engineering or, perhaps more controversially, the lessons still to be learned from Kitaigorodskii’s contribution.13 Simply stated, regardless of the nature of their putative intermolecular interactions, it is the close packing of molecules that should be the first consideration when discussing the construction of crystal structures (ice being an exceptional exclusion owing to the open framework sustained by hydrogen bonding). However, molecules tend not to be perfectly spherical and hence there are deviations from the ideal 12 (3 + 6 + 3) neighbors found in cubic and hexagonal close packing arrangements. In a systematic survey of organic molecules in the Cambridge Structural Database (CSD; also see The Cambridge Structural Database System and Its Applications in Supramolecular Chemistry and Materials Design, Supramolecular Materials Chemistry),14 a process also known as data mining, it was demonstrated that for a given reference molecule there are often 14 close neighbors although this number, the molecular coordination number (MCN), ranges from 8 to 22 in the systems surveyed.15 The prevalence of the adoption of close packing arrangements is borne out by statistical studies that show the majority of structures, that is, over 83%, crystallize in one of six close packing space groups (i.e., P 21 /c —including the nonstandard setting of P 21 /n—35.0%, P 1 23.2%, C2/c 8.1%, P 21 21 21 7.8%, P 21 5.4%, and Pbca 3.5%) out of the possible 230 space groups (or 219 if the 11 pairs of enantiomorphic space groups, for example, the affine space group types P 31 and P 32 , are counted once only)16–18 ; up-to-date space group frequency tables are maintained by the CSD.19 The complementarity between molecules is paramount so that during self-assembly, protrusions in one molecule are accommodated by voids in neighbors.20 Supporting this precept is the observation that despite the fact that (organic) molecules may have very odd shapes and be of low symmetry, the crystal packing efficiency hovers around the ideal value (for spheres) of 0.74.20, 21 A term “packing fitness”20 relates to the preceding and might

imply that all factors being equal, molecules would prefer to adopt spherical shapes in order to meet close packing requirements. In a case study prompted by interest in delineating the factors behind the contrasting behavior exhibited by hydrocarbon/fluorocarbon mixtures compared to their aromatic analogs, which are known to form multicomponent crystals, it was concluded that molecular shapes were important.22 For these systems that are maintained in the condensed phase by dispersion forces, pairs of aromatic molecules have disk-like shapes that allow for more efficient crystal packing via π · · ·π interactions, at least more so than for the somewhat cylindrical shapes adopted by pairs of aliphatic counterparts22 ; a recent survey based on the CSD14 correlating molecular shape (spheres, disks, and rods) with packing patterns is highly recommended reading.23 While the foregoing offers an explanation for consequent chemistry, that is, multicomponent crystal formation or not, the next two examples, drawn from phosphinegold(I) thiolate chemistry, go further in that an influence on intra- and intermolecular bonding is apparent. Two distinct coordination motifs are found in structures conforming to the general formula R3 PAu(S2 COR
), Figure 4(a), where, in addition to the linear P–Au–S coordination geometry defined by the monodentate ligands, an intramolecular Au· · ·O or Au· · ·S contact is observed (Figure 4). On the basis of hard acid/soft base considerations, an intramolecular Au· · ·S interaction might be expected in these structures. However, this imperative is usurped in the majority of examples found for this formulation.24 The explanation proffered was that the less favorable intramolecular Au· · ·O contact was formed to allow for the formation of a more spherical/globular molecule which in turn led to more favorable crystal packing (Figure 4b). In contrast, when steric pressures exerted by the R/R
groups were great, the energetically more favorable intramolecular Au· · ·S contact is formed but at the expense of forming a cylindrical molecule with consequent less efficient crystal packing (Figure 4c).24, 25 More remarkable were the observations in a subsequent study of a related phosphinegold(I) thiolate molecule in which the thiolate ligand carries hydrogen-bonding functionality. Two structural motifs are observed in the tetramorphic tricyclohexylphosphinegold(I) 2-mercaptobenzoate system (c-C6 H11 )3 PAu(SC6 H4 CO2 H-2) (Figure 5a).26 A dimeric aggregate mediated by the eight-membered {· · ·HOC=O}2 homosynthon is found in one polymorphic form (Figure 5b). But in another polymorph, a monomeric species is observed, whereby the intermolecular homosynthon is replaced by an intramolecular O–H· · ·S hydrogen bond (Figure 5c).26 Hence, a presumably favorable hydrogen-bonding arrangement leading to a rod-shaped aggregate is disrupted by a less favorable O–H· · ·S

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering c -C6H11

R R

P

Au

S

c -C6H11 S

R

O Au

P

S

OH

c -C6H11 (a)

O (a)

5

R′

(b)

(b)

(c)

Figure 4 Structural variation in phosphinegold(I) xanthates, R3 PAu(S2 COR
). (a) Generic chemical structure of R3 PAu(S2 COR
), indicating the possibility of rotation about the thiolate S–C bond leading to close intramolecular Au· · ·O or Au· · ·S interactions, (b) globular molecule found for Et3 PAu[S2 CO(c-C6 H11 )] for efficient crystal packing but with an unfavorable Au· · ·O interaction, and (c) cylindrical molecule found for (c-C6 H11 )3 PAu[S2 CO(c-C6 H11 )] for less efficient crystal packing but with a more favorable Au· · ·S contact.

interaction but with the adoption of a more spherical/ globular molecular shape. Of particular interest was the observation that both forms crystallized concomittantly27 suggesting, at most, a small difference in crystal stabilization energies. Reverting back to organic molecules, the absence of expected hydrogen bonding in the structure of the organic molecule alloxan, 1,3-diazinane-2,4,5,6-tetrone, has attracted significant attention over the years.28, 29 As can be gleaned from Figure 6, the chemical structure of alloxan presents many opportunities for strong N–H· · ·O hydrogen bonding in its crystal structure.28, 29 However, these are, at best, quite weak in the observed crystal structure with nonconventional C=O· · ·C=O interactions playing a prominent role28 ; dipolar C=O· · ·O=C interactions have long been recognized as supramolecular synthons of comparable energy to hydrogen bonds.30 Shape has also been evoked as being important in determining the global

(c)

Figure 5 Structural variation in polymorphic forms of (c-C6 H11 )3 PAu(SC6 H4 CO2 H-2). (a) Chemical structure of (c-C6 H11 )3 PAu(SC6 H4 CO2 H-2), indicating the possibility of rotation about the C–C carboxylic acid bond leading to intermolecular O–H· · ·O or intramolecular O–H· · ·S interactions, (b) dimeric rod-shaped aggregate found in one polymorphic form of (c-C6 H11 )3 PAu(SC6 H4 CO2 H-2) mediated by an eight-membered {· · ·HOC=O}2 homosynthon, and (c) globular molecule found in another polymorphic form of (c-C6 H11 )3 PAu(SC6 H4 CO2 H-2) featuring an intramolecular O–H· · ·S hydrogen bond. The O–H· · ·O and O–H· · ·S hydrogen bonds are shown as orange dashed lines. O HN

NH O

O O

Figure 6 tetrone.

Chemical structure of alloxan, 1,3-diazinane-2,4,5,6-

crystal packing in this example.29 The crystal structure of alloxan, in space group P 41 21 2, has been correlated with similarly shaped molecules, such as fluorobenzene

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

6

Supramolecular materials chemistry

and benzonitrile, where hydrogen bonding is not possible.29 Molecules of this size and shape are ideal for packing along the crystallographic 41 screw axis and calculations show that the energy of stabilization afforded by the weak bifurcated N–H· · ·O and by the C=O· · ·O=C interactions are comparable.29 Thus, what is lost in terms of stabilization energy provided by strong N–H· · ·O hydrogen bonds is gained by the global close packing.29 Of further interest is the observation that the perdeuterated derivative,31 which has the same room temperature structure as for all hydrogen compounds,28 undergoes a phase transition below 35 K.31 The low-temperature structure crystallizes in P 21 21 21 , a subgroup of P 41 21 2, and features strong N–H· · ·O hydrogen bonds as well as the dipolar C=O· · ·O=C interactions of the same magnitude as in the P 41 21 2 structure. The authors note that the crystal structures of the perdeuterated phases are virtually superimposable and that the calculated lattice energies differ by less than 2%.31 As a corollary to the above, the deliberate design of molecules with awkward shapes that pose difficulties for efficient crystal packing continues to attract interest.20, 32, 33 Motivation for these studies revolves around determining the propensity of such molecular species to crystallize in different crystalline forms, for example, as polymorphs, solvates, and so on, and when within one crystal form, the likelihood to crystallize with multiple molecules in the asymmetric unit (Z
> 1). Before going on to describe some of the different types of intermolecular interactions that have been identified in recent times, two further points need to be made with respect to the foregoing. In terms of shape, the preferences for molecules to crystallize about a center of inversion, crystallographically imposed or not, whereby optimal/ complimentary interactions between molecules occur is well documented.18 In addition, the above notwithstanding, that is, shape considerations are important, the argument cannot be sustained that conventional robust and directional hydrogen-bonding interactions are not structure directing and provide important synthons for the design of crystal structure. This is vindicated quite emphatically in the recently described tetramorphic system of a N, N
-diaryl urea derivative, 1-(3-methylsulfonylphenyl)3-pyridin-2-ylurea (Figure 7a), whereby the eightmembered amide {· · · HNC=O}2 homosynthon persists in all four structures leading to a zero-dimensional aggregate (Figure 7b). However, the supramolecular association between these to generate the three-dimensional crystal structure varies from polymorph to polymorph,34 emphasizing the fact that while hydrogen bonding is often easily recognizable and provides a convenient focus for the interpretation of a crystal structure, hydrogen bonding does not always lead to three-dimensional architectures, meaning other supramolecular glues in operation must be identified

O N

N H

N H

S

(a)

(b)

Figure 7 Chemical structure of (a) 1-(3-methylsulfonylphenyl)3-pyridin-2-ylurea and (b) dimeric aggregate sustained by the eight-membered amide {· · ·HNC=O}2 homosynthon. Hydrogen bonds are shown as orange dashed lines.

and categorized in order to develop a complete understanding of the overarching principles of crystal packing.

2.3

Intermolecular interactions—hydrogen bonding

As mentioned above, one of the pillars of crystal engineering endeavors is the understanding of the way molecules pack, in particular, in systems where hydrogen bonding does not operate in all three dimensions. While it is possible to calculate the relative energies of stabilization associated with different intermolecular interactions for an individual molecule in a given crystal structure (see below), hard and fast values are not forthcoming owing to their dependence on a myriad of factors. Instead, rather large overlapping ranges of energies are usually cited. For neutral organic molecules, the energies of intermolecular interactions35 involving hydrogen atoms can be conveniently classified into three general groups: 1. strong conventional or traditional hydrogen-bonding interactions occurring between hard acids and hard bases with energies in the range 20–65 kJ mol−1 (e.g., O–H· · ·O and N–H· · ·O), 2. intermediate strength contacts that occur between hard acids and soft bases with energies in the range 4–16 kJ mol−1 (e.g., C–H· · ·O and O–H· · ·π), and 3. weak contacts that occur between soft acids and soft bases with energies in the range 1–8 kJ mol−1 (e.g., C–H· · ·π). Paraphrasing some of the discussion above, the task of the crystal engineer would be made significantly easier

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering if molecules assembled in accord with the preceding hierarchy. However, as these energies indicate, it is not difficult to conceive situations whereby a strong hydrogen bond is overwhelmed by a combination of a number of other, nominally weaker, intermolecular interactions. Guru Row’s classification of intermolecular interactions based on charge density studies also provides a useful guide to their relative strength.36 A mainstay of crystal engineering, certainly organic crystal engineering, has been the conventional hydrogen bond.37 Developed in the context of crystal engineering, empirical rules now exist describing hydrogen-bonding propensity. Building on Robertson’s axiom38 that, when able, hydrogen bonds will form, Etter condensed hydrogenbonding patterns into three general rules (repeated here verbatim)39 : 1. all good proton donors and acceptors are used in hydrogen bonding, 2. six-membered ring intramolecular hydrogen bonds form in preference to intermolecular hydrogen bonds, and 3. the best proton donors and acceptors remaining after intramolecular hydrogen-bond formation form intermolecular hydrogen bonds. The prevalence of hydrogen bonds led to their characterization by graph set analysis (see Network and Graph Set Analysis, Supramolecular Materials Chemistry). While conventional hydrogen bonding is well understood, there are variations that need to be considered. At one extreme, charge-assisted hydrogen bonds,40 for example, O–H· · ·N+ , with energies in the range 60–160 kJ mol−1 ,35 enhance the strength of supramolecular synthons formed by neutral analogs.41 Less well understood are weaker interactions involving hydrogen that might be classified as weak hydrogen bonds.42 Among weaker intermolecular interactions involving hydrogen generally regarded as having importance are those probably first recognized in biological systems, for example, C–H· · ·O, C–H· · ·halide, and C–H· · ·π interactions, as these are known to contribute significant stability to protein and other macromolecular structures.43–46 While naturally weaker than conventional hydrogen bonding, these types of interactions are willing participants in various supramolecular synthons and known to be important in stabilizing crystal structure. When reviewing hydrogen-bonding interactions involving halides, those involving the heavier congeners are well accepted, even if weak, but more controversial are interactions involving organic fluorine such as D–H· · ·F.47–52 Such interactions are at the weak extreme of hydrogen bonding with even O–H· · ·F considered rare.47 The weakness of D–H· · ·F is ascribed to the low polarizability of fluorine, a situation accentuated when the donor is weak,

7

such as C–H.52 Clearly, further systematic evaluation of closely related structures is required to shed light on the significance of these types of interactions. A nice example of shape recognition sustained by the cooperation between nonconventional weak interactions is the intermolecular embraces between aromatic systems, as delineated largely by Dance and Scudder.53–55 The original paper in this area53 described sixfold phenyl embraces (6PE) occurring between two X-PPh3 residues, as shown in Figure 8 in the structure of Cl3 GaOPPh3 .56 These supramolecular synthons arise as the shape of the propeller-like triphenylphosphine residue is ideally suited for interpenetration with a second triphenylphosphine to facilitate the formation of six cooperative C–H· · ·π contacts (described as edge-to-face interactions); the energy of stabilization for a 6PE is estimated to be more than 50 kJ mol−1 .55 Subsequently, related embraces featuring less efficient interactions between two X-PPh3 residues (e.g., offset face to face) were categorized and the concept expanded to cover a wide range of molecules containing different aromatic residues as well as extending into two and three dimensions.53, 55 It is probably pertinent at this stage to cite a comment by Dunitz and Gavezzotti57 to the effect that most organic molecules feature hydrogen atoms at the periphery so that C–H· · ·X interactions are inevitable; a similar observation applies to many metalorganic molecules. The question then arises: actually how important are these interactions

Cl Cl

Ga

O

P

Cl (a)

(b)

Figure 8 Chemical structure of (a) the adduct formed between trichlorogallium(III) and triphenylphosphineoxide, Cl3 GaOPPh3 and (b) dimeric aggregate sustained by a sixfold embrace encompassing six C–H· · ·π interactions shown as purple dashed lines. The aggregate in this system is disposed about a crystallographic site of symmetry 3. However, in most examples, such aggregates approximate S3 symmetry.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

8

Supramolecular materials chemistry H3N

Cl Pt

Cl

NH2 O

(a)

OH

(b)

Figure 9 Hydrogen bonding between water and platinum. (a) Chemical structure of trans-[Pt(NH3 )(N-glycine)Cl2 ]·H2 O, (b) a portion of the crystal packing highlighting (i) O–H· · ·Pt hydrogen bonding and (ii) a centrosymmetric 12-membered {· · ·O=COH· · ·OH}2 synthon. The O–H· · ·Pt and O–H· · ·O hydrogen bonding are shown as purple and orange dashed lines, respectively.

or, posed in another manner, are weak intermolecular interactions sometimes overinterpreted, especially with the availability of powerful interpretative programs that provide vast quantities of numerical data that might be utilized uncritically (see below). The word of caution notwithstanding, interactions involving hydrogen still dominate crystal engineering endeavors and novel observations still emerge. For example, the recent report of a hydrogen bond formed between a water molecule and platinum as found in the neutral complex, trans-[Pt(NH3 )(Nglycine)Cl2 ]·H2 O (Figure 9a).58 This is an experimentally authenticated example of a structure featuring an O–H· · ·Pt hydrogen bond and represents a case of inverse hydration (Figure 9b). In this case, the hydrogen atom is directed to a filled d orbital normal to the square plane about the platinum(II) center. In addition to hydrogen bonding, strong or weak, there is what might be termed a class of “emerging” intermolecular interactions and these are discussed below.

2.4

Emerging intermolecular interactions

2.4.1 Halogen bonding An intermolecular interaction attracting significant attention is the so-called halogen bond. This involves the noncovalent interaction between a Lewis base (D) and an electrophilic halogen (X)59–63 and compliments the participation of

halogens in hydrogen-bonding interactions.49 As with other weak intermolecular contacts discussed above, halogen bonds have also been recognized as being important in biological systems.64 It is of interest to note a report by Guthrie in 1863 that described a H3 N· · ·I2 species.65 Hassel (Nobel Laureate, 1969) provided crystallographic evidence for the halogen bond in the 1950s.66 More recently, it has become evident that halogen bonds can be exploited as a structure-directing synthon in crystal engineering. While such a D· · ·X–Y (Y = C, N, another halogen, etc.) interaction may be counterintuitive, the anisotropic polarization of electron density about a halogen is such that there is an electrophilic region at the terminus of the Y–X vector and larger electron-rich (nucleophilic) regions normal to this vector61 ; this phenomenon has been termed polar flattening.67 While this explains the formation of D· · ·X–Y, it also provides a rationale for both the strength and directionality of halogen bonds, which can rival hydrogen bonding in these attributes. Estimates of the energy of attraction for halogen bonds range from 8 to 12 kJ mol−1 for an N· · ·Cl contact to approximately 50 kcal mol−1 for a charge-assisted I− · · ·I2 interaction.61, 68 Such considerations inevitably led to studies of the competition between hydrogen bonding and halogen bonding. In systematic cocrystallization experiments involving the reagents shown in Figure 10(a), it proved possible to form 1 : 1 multicomponent crystals between 1,2,4,5tetrafluoro-3,6-diiodobenzene and 1,2-bis(4-pyridyl)ethane, leading to a supramolecular linear chain mediated by N· · ·I halogen bonds (Figure 10b).69 In a separate experiment, a 1 : 1 multicomponent crystal was prepared having the constituents, 1,4-dihydroxybenzene and 1,2-bis(4pyridyl)ethane (Figure 10a), which were connected via O–H· · ·N hydrogen bonds also into a supramolecular chain (Figure 10c).69 In a competitive cocrystallization experiment containing all three reagents in the same solvent system (acetone), the halogen-bonded multicomponent crystal was isolated with the unreacted 1,4-dihydroxybenzene characterized as remaining in solution. Hence, it can be concluded that halogen bonding is more potent in the formation of a supramolecular chain than hydrogen bonding, at least for this system.69 The electron density distribution around the Y–X bond described above indicates that the electropositive crown61 available for interaction with a lone pair of electrons should be greater for the more polarizable halogens so that putative N· · ·I interactions should be stronger than their N· · ·Cl counterparts. This expectation is borne out experimentally.70, 71 Furthermore, it is also possible to moderate the magnitude of N· · ·X by systematically varying the nature of the halocarbon “acid.”70, 71 With this background, it is not surprising that supramolecular architectures based on halogen bonding appear in the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

9

OH

I F

F

F

F I

OH

(i)

(ii) N

N (a)

(iii)

(b)

(c)

Figure 10 Competition between hydrogen and halogen bonding. (a) Chemical structures of (i) 1,2,4,5-tetrafluoro-3,6-diiodobenzene, (ii) 1,4-dihydroxybenzene, and (iii) 1,2-bis(4-pyridyl)ethane; (b) supramolecular chain in the 1 : 1 multicomponent crystal formed ˚ and (c) supramolecular chain in the 1 : 1 multicomponent crystal between (i) and (iii) mediated by N· · ·I halogen bonds [2.79 A]; formed between (ii) and (iii) mediated by O–H· · ·N hydrogen bonds. Halogen and hydrogen bonds are shown as orange dashed lines.

literature on a more regular basis, in particular featuring charge-assisted halogen bonding. An example of a 2-D architecture bearing a close resemblance to the honeycomb array sustained by hydrogen bonding is shown in Figure 11. Here, charge-assisted I− · · ·I interactions involving iodide and threefold symmetric 1,3,5-trifluoro-2,4,6triiodotrifluorobenzene lead to a (6,3) topology, with the cation residing in the voids.72 Halogens bound to a metal, main group or transition metal, can also function as effective donors or acceptors of electron density by forming D· · ·X–M synthons.73, 74 An illustrative example is found in the crystal structure of trans-bis(4bromopyridine)dichloroplatinum(II) (Figure 12a).75 As shown in Figure 12(b), the presence of Pt–Cl· · ·Br interactions leads to a 2-D architecture. While the previous examples have generally focused on interactions involving nitrogen as the donor, examples exist where other atoms, such as oxygen, function as the donor.73, 74 Recently, the heavier congeners, sulfur and selenium, by virtue of their ability of forming halogen-bonding interactions, have been implicated as being important for iodothyronine deiodinase activity.76 Furthermore, recent crystal engineering studies have demonstrated the cooperation between hydrogen bonding and halogen bonding in a multicomponent crystal containing a thioamide and a halocarbon.77 In this multicomponent crystal, a supramolecular tape comprising 2,3-dihydro-1H-1,3-benzodiazole-2-thione (Figure 13a), is formed via eight-membered {· · · HNC=S}2 homosynthons

(Figure 13b). One iodide from 1,2,3,4-tetrafluoro-5,6diiodobenzene (Figure 13a) is linked to each sulfur via a C=S· · ·I halogen bond (Figure 13b).77

2.4.2 π· · ·π and C–H· · ·π interactions The π· · ·π interaction between aromatic rings is well established in crystal engineering,78 with an energy of stabilization in the range 8–40 kJ mol−1 ,35 and, as with many other supramolecular synthons, charge assistance (e.g., π + · · ·π), contributes significantly to their stability.79 More recently discussed are π· · ·π stacking interactions occurring between chelate rings in metal complexes.80 This arises as planar chelate rings with delocalized π electron density have significant aromatic character, that is, metalloaromaticity. An example of a complex where π(chelate)· · ·π(chelate) stacking interactions are observed, namely, in the structure of bis(3-allylacetylacetonato)copper(II) (Figure 14a),81 leading to the supramolecular chain is shown in Figure 14(b). In their analysis of neutral transition metal complexes with a square planar geometry, Sredojevi et al. found 955 (38.6%) examples where π(chelate)· · ·π(chelate) stacking interactions occur out of a possible 2473 complexes that met their search criteria.80 Interestingly, when these interactions occur, it is more likely there will be multiple interactions, suggesting a measure of cooperativity. Having unsaturated groups (e.g., C≡N and OCOEt) on the α or β carbon atom of the chelate ring tends to reduce the propensity

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

10

Supramolecular materials chemistry Cl

I +

MEt4 I (a)

F

F

I

I

Br

−

N

N

Pt

(a)

Br

Cl

F

(b)

Figure 12 Halogen bonding leading to the formation of a two-dimensional array. (a) Chemical structure of trans-bis(4bromopyridine)dichloroplatinum(II) and (b) supramolecular 2-D ˚ halogen bonds, shown architecture based on M–Cl· · ·Br [3.33 A] as orange dashed lines. I F

I

F

F

(b)

Figure 11 Halogen bonding leading to the formation of a honeycomb array. (a) Chemical structures of the constituents 1,3,5tritrifluoro-2,4,6-triiodobenzene and tetraethylammonium iodide; in the cation the nitrogen atom is partially substituted by phosphorus (34%) and (b) supramolecular 2-D architecture based on a honeycomb mediated by charge-assisted I− · · ·I halogen bonds ˚ shown as blue dashed lines; the iodide anions are [3.48–3.58 A], shown as olive green spheres.

of forming π (chelate)· · ·π (chelate) stacking interactions, and, naturally, the presence of bulky substituents hindered their formation.80 As with what might be termed regular π · · ·π interactions, both parallel and antiparallel conformations are found, although parallel conformations are less common for interactions occurring between sixmembered rings.80 In the complex illustrated in Figure 14, ˚ consistent the π(chelate)· · ·π (chelate) distance is 3.20 A with the observation that the π (chelate)· · ·π (chelate) separations approximate those seen in stacking interactions involving aromatic rings.78 Bearing this out is the observation that mixed π stacking interactions have also been documented.82–84 An example of a mixed π · · ·π(chelate) contact is found in the structure of the square planar complex bis(acetone-1-naphthoylhydrazinato)copper(II) (Figure 15a).82 These assemble molecules into a supramolecular chain where the separation between aromatic ˚ (Figure 15b).82 Consistent with the notion rings is 3.70 A that chelate rings are indeed metalloaromatic, C–H· · ·π

H N S

(a)

F

N H

(b)

Figure 13 Complimentary hydrogen and halogen bonding. (a) Chemical structures of 1,2,3,4-tetrafluoro-5,6-diiodobenzene (left) and 2,3-dihydro-1H -1,3-benzodiazole-2-thione (right) and (b) supramolecular 1-D architecture-based N–H· · ·S hydrogen bonding (orange dashed lines) and C=S· · ·I halogen bonds ˚ blue dashed lines) in the 1 : 1 multicomponent (S· · ·I = 3.31 A; crystal.

(chelate) interactions might be expected in the crystal structures of metal complexes as for the now well-established C–H· · ·π interactions in organic systems.85, 86 The presence of C–H· · ·π(chelate) interactions has indeed been identified and commented upon in the literature.87–90 An early report of C–H· · ·π(CdS2 C) interactions, where the π system is a four-membered chelate ring, was

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

11

O

O

N

N

Cu O

O

Cd S

(a) (a)

S S

CH3OCH2CH2O

S

OCH2CH2OCH3

(b)

Figure 14 An example of π(chelate)· · ·π(chelate) stacking interactions operating in a square planar transition metal complex. (a) Chemical structure of bis(3-allylacetylacetonato) copper(II) and (b) supramolecular 1-D architecture-based π(chelate)· · ·π (chelate) interactions (purple dashed lines).

O N N

Figure 16 An example of C–H· · ·π(chelate) interaction operating in a crystal structure. (a) Chemical structure of (2,2
bipyridyl) bis[O-(2-methoxyethyl)dithiocarbonato] cadmium (II) and (b) supramolecular 2-D architecture based on C–H· · ·π (CdS2 C) interactions (purple dashed lines).

N Cu

(b)

N O

N (a)

P N Rh C

O

O

O (a) (b)

Figure 15 An example of π· · ·π(chelate) stacking interactions operating in a square planar transition metal complex. (a) Chemical structure of bis(acetone-1-naphthoylhydrazinato) copper(II) and (b) supramolecular 1-D architecture-based π· · ·π (chelate) interactions (purple dashed lines).

described in the crystal structures of 2,2
-bipyridine adducts of cadmium xanthates (Figure 16a).87 As an example, Figure 16(b) shows a partial crystal packing diagram where C–H· · ·π (CdS2 C) interactions are highlighted in the crystal structure of Cd[S2 CO(n-Bu)]2 (2,2
-bipyridine).87 Additional examples of C–H· · ·π(MS2 C) interactions occurring in metal (transition and main group) xanthates were recognized during a comprehensive review of their molecular and supramolecular structures.24 In a systematic analysis of C–H· · ·π interactions occurring between metal acetylacetonate complexes and phenyl rings, it was demonstrated that the acetylacetonate chelate

(b)

Figure 17 An example of a C–H· · ·π (chelate) interaction. (a) Chemical structure of (acetylacetonato)carbonyl[phenyl-bis(2pyridyl)phosphine]rhodium(I) and (b) dimeric aggregate based on C–H· · ·π(chelate) interactions (purple dashed lines).

could function as a hydrogen atom donor to a phenyl group, that is, metal–ligand-C–H· · ·π (MLC–H· · ·π) and also as an acceptor of a hydrogen atom from a phenyl group, C–H· · ·π (chelate).89 An example of the latter is illustrated for (acetylacetonato)carbonyl[phenyl-bis(2pyridyl)phosphine]rhodium(I)91 in Figure 17. The ability

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

12

Supramolecular materials chemistry

of the acetylacetonate chelate to function as an acceptor is moderated by the nature of the metal. Thus, if the metal is “soft,” there is greater delocalization of π-electron density which results in an enhanced capacity of the chelate ring to donate a hydrogen; the energy of the MLC–H· · ·π interactions was estimated to lie in the range 2.5–10 kJ mol−1 . While definitive trends in C–H· · ·π(chelate) interactions are yet to discerned, the energies of association are calculated to be in the range 6–11 kJ mol−1 .89 In a further study, the competition between the formation of π (chelate)· · ·π (chelate) stacking and MLC–H· · ·π interactions was evaluated in square planar transition metal complexes.90 While both are observed, the former predominated. When steric hindrance precluded the formation of π(chelate)· · ·π (chelate) contacts, MLC–H· · ·π interactions were apparent.90 The importance of steric interactions in controlling supramolecular aggregation is discussed in more detail below.

2.4.3 Lone pair· · ·π and metal· · ·π interactions Lone pair· · ·π (aryl) interactions are also gaining prominence as a supramolecular synthon. In their 2002 review of tellurium structures, Zukerman-Schpector and Haiduc highlighted the presence of lone pair· · ·π (aryl) interactions as being important in providing stability to crystal structures in a dimension where no other intermolecular interactions were apparent.92 An example of lone pair· · ·π(aryl) interactions operating in the crystal structure of bis(2,4,6-trimethylphenyl)-ditellurium93 is shown in Figure 18. Subsequently, lone pair· · ·π (aryl) interactions have been identified in systematic analyses of tin94 and lead95 compounds. While these interactions are readily recognized as arising from the charge transfer from the metal-bound lone pair of electrons to the lowest unoccupied molecular orbital of the aryl ring, more intriguing was the observation that intermolecular metal· · ·π(aryl) interactions also occur in gold complexes, that is, in a situation where no lone pair of electrons is involved.96 Te

Te

(a)

(b)

Figure 18 An example of a lone pair· · ·π (aryl) interaction. (a) Chemical structure of bis(2,4,6-trimethylphenyl)ditellurium(II) and (b) supramolecular 1-D architecture based on lone pair· · ·π(aryl) interactions (purple dashed lines).

(a)

Cl

Au

C

N

(b)

Figure 19 An example of a Au· · ·π (aryl) interaction. (a) Chemical structure of chlorido(4-ethynylphenylisocyano) gold(I) and (b) supramolecular 1-D architecture based on gold· · ·π(aryl) interactions (purple dashed lines).

An example of a gold· · ·π(aryl) interaction leading to a supramolecular chain is shown for chlorido(4-ethynylphenylisocyano)gold(I)97 in Figure 19. Such interactions involve the donation of electron density from the aromatic ring to the metal center. Further systematic studies considered the influence on lone pair· · ·π(aryl) interactions of increasing the electron density in the aryl ring, leading to heteroaromatic rings such as pyridine and pyrazine.98 In these circumstances, the strength and frequency of lone pair· · ·π(heteroaromatic) interactions should be diminished compared to lone pair· · ·π (aryl) interactions. Conversely, for gold complexes, such interactions should be enhanced. Both trends were evident.98 Of the metals surveyed, lone pair· · ·π(aryl) interactions were most prevalent in tellurium structures, featuring in up to 6% of all structures where such interactions could potentially occur.98 An interesting twist on the above is found in a rare example featuring a lone pair· · ·π interaction where the π-system is a metal chelate observed in the crystal structure of bromido-bis(dimethyldithiocarbamato)-4methoxyphenyl-tellurium(IV) (Figure 20a), characterized as a dichloromethane solvate.99 Although intermolecular Te· · ·S or Te· · ·Br contacts are potentially formed in this structure, as observed for most examples in a survey of over 160 related tellurium 1,1-dithiolate structures,100 these are not found; but instead a lone pairπ (TeS2 C) interactions lead to the formation of a dimeric aggregate (Figure 20b). It is noted that the authors commented on this feature of supramolecular aggregation in their original publication.99 It is likely that the lone pair· · ·π (TeS2 C) interaction is found in this particular structure owing to a coincidence of factors such as the electron-rich nature of TeS2 C chelate due to the presence of the strongly coordinating dithiocarbamate ligand, the small steric influence exerted by the nitrogenbound methyl groups, and the relatively unencumbered

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

13

for example, a halide, oxygen, nitrogen, and sulfur (Lewis base), now well known as a supramolecular synthon.103–105 The moderation of secondary interactions by steric factors forms the basis for the next section.

O

Br Te S N

2.5

S S

S

Steric bulk as a factor in determining supramolecular aggregation

N

(a)

(b)

Figure 20 A rare example of a lone pair· · ·π(TeS2 C) interaction. (a) Chemical structure of bromido-bis (dimethyldithiocarbamato)-4-methoxyphenyl-tellurium(IV) and (b) dimeric aggregate based on lone pair· · ·π(TeS2 C) interactions (purple dashed lines). The molecule crystallizes as a dichloromethane solvent which is not illustrated for reasons of clarity.

coordination environment about the tellurium atom. Lone pair· · ·π interactions are not restricted to metalorganic systems alone but have been recognized in macromolecular structures where, for example, a deoxyribose-oxygen lone pair· · ·π (guanine) interaction was cited as being crucial in stabilizing the left-handed Z-DNA duplex,101 and are also apparent in crystal structures comprising organic molecules.102 In a literature survey of lone pair· · ·π interactions involving a range of elements carrying at least one lone pair of electrons, Mooibroek et al.102 concluded that lone pair· · ·π interactions were more prevalent than anion· · ·π and cation· · ·π interactions, and made the point that while not often commented on in the original publication, as noted for the metal-based systems,92, 94–96 such interactions are not surprising given their prevalence in nature, in particular involving water. In the situation where the lone pair of electrons resides on a metal center, such lone pair· · ·π interactions may be considered as a subset of secondary interactions, that is, interactions of the type M· · ·X that occur between a heavy element center (Lewis acid) and,

Secondary interactions are known to feature in a large number of crystal structures containing, in particular, main group element compounds.103–105 Systematic evaluation of closely related crystal structures has revealed that their presence can be moderated, that is, turned on or off, by careful substitution of metal- or ligand-bound organic substituents.106–108 Put simply, if able, secondary interactions will form and give rise to recognizable supramolecular architectures. However, if the organic substituents are too bulky, such interactions will not form. As an illustration of this principle, a triorganotin system is described which also emphasizes some of the packing considerations mentioned above and provides a nice one-dimensional example of Kitaigorodskii’s Aufbau principle of crystal packing.109, 110 Kitaigordoskii’s principles of close packing are based on the ability of molecules to fill a space with an adequate number of energetically favorable intermolecular interactions. There are four stages in the close packing principle13 : Stage 0 : describes the molecular packing unit. The energy needed for molecule-to-molecule interaction, Stage 1 : determines the packing of the molecules in one dimension, using stage 0, Stage 2 : from one dimension to two dimensions, and Stage 3 : three-dimensional packing combining stage 2 with the symmetry of the third dimension. Triorganotin carboxylates usually adopt monomeric structures with cis-C3 SnO2 tin atom geometries or polymeric chains with trans-C3 SnO2 tin atom geometries.111 In the former, weak intramolecular Sn· · ·O interactions are formed in contrast to the intermolecular secondary Sn· · ·O interactions in the latter. Referring to the chemical structure shown in Figure 21, in solution, regardless of the nature of the tin-bound substituent, the molecules are monomeric as demonstrated by Sn-117 NMR spectroscopy.106 However, in the solid state, a monomeric species is only found when R is large, for example, cyclohexyl (Figure 22). In contrast, a supramolecular polymeric chain is formed when R is small, for example, methyl (Figure 22). During crystallization, the situation may be envisaged whereby individual molecules precipitate and adjust positions so as to optimize favorable Sn· · ·O interactions. When the steric bulk allows,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

14

Supramolecular materials chemistry O R R

Sn

O N

R

OH

N

Figure 21 Generic chemical structure for triorganotin 2-[(E)-2(2-hydroxy-5-methylphenyl)-1-diazenyl]benzoates. The hydroxyl group forms an intramolecular hydrogen bond to the anazo nitrogen atom and does not participate in intermolecular interactions of note. The rotation about the O–C-carboxylate bond allows for the formation of either intra- or intermolecular Sn· · ·O interactions.

these are intermolecular and a supramolecular chain results. In circumstances where the bulk is too great, molecules also arrange themselves into a chain but with no formal Sn· · ·O contacts between them; an intramolecular Sn· · ·O contact is formed instead. This is borne out by the linear ˚ repeat distances between successive tin atoms of 4.89 A ˚ when for the R = methyl derivative compared with 5.17 A R = cyclohexyl.106 The discussion of steric control over supramolecular aggregation is becoming more prominent in organic crystal engineering112–114 with firm foundations of its influence delineated in resolving the dimer/catemer conundrum for carboxylic acids,115, 116 discussed earlier. While not classified as secondary interactions, metallophilic interactions, that is, attractive intermolecular interactions occurring between d10 [e.g., gold(I) and silver(I)] and d8 [e.g., platinum(II)] metal centers should be mentioned here.117 These too can be moderated by steric factors as for secondary interactions. The most notable and investigated of the metallophilic interactions are aurophilic (Au· · ·Au) interactions between gold centers, which occur owing to the overlap of the filled 5d10 orbitals with the vacant 6s and 6p orbitals, and can represent interactions approaching the energy of stabilization provided by conventional hydrogen bonds.118 Indeed, the closeness in energy provided by each inspired Schmidbaur and colleagues’ pioneering crystal engineering studies investigating the cooperation and even competition between these synthons.119 In keeping with steric control over secondary interactions discussed above, steric factors are also important in controlling the formation of aurophilic interactions so that when the bulky groups are present, Au· · ·Au interactions are precluded. The classic example is found in the structures of the linear phosphinegold(I) chlorides where Au· · ·Au interactions are found for Et3 PAuCl but not for (c-C6 H11 )3 PAuCl120 ; it is noted in passing that aurophilic interactions give rise

to interesting photophysical properties, both in solution and in the solid state.120 Furthermore, in connection with studies designed to ascertain the nature of awkward shapes on crystal packing discussed above,20, 32, 33 a recent data mining investigation of X–Au–Y compounds featuring Au· · ·Au interactions indicated that there was a greater propensity to form crystal structures with more than one molecule in the crystallographic asymmetric unit (Z
> 1) compared to gold-containing structures not featuring aurophilic interactions.121 This observation was rationalized in terms of the relative sizes of the X and Y ligands in X–Au–Y. When there was a significant disparity in these, Z
> 1 structures were favored.121

2.6

Structural mimicry

Structural mimicry122 is a concept that draws together many of the principles of crystal packing reviewed above. Basically, when close packing considerations predominate, that is, in the absence of strong and directional supramolecular synthons, volume considerations hold sway as probably first discussed in terms of the chloro/methyl exchange ˚3 rule.123 With molecular volumes of approximately 19 A 3 ˚ for a methyl group, the concept for chloride and 24 A is that these substituents are interchangeable, that is, will give rise to isostructural crystal structures when the only difference between two molecules relates to these substituents. However, in the most comprehensive study of this phenomenon, based on a data mining investigation of the CSD,14 it was demonstrated that in fact in only about 25% of cases where isostructurality based on chloro/methyl exchange could occur, that it did.124 In retrospect and in view of the varied nontraditional intermolecular forces that have now been recognized, this result is not surprising. Related to the above are distinct supramolecular synthons that give rise to similar supramolecular aggregates/crystal packing. An example of structural mimicry has in fact already been discussed herein, that is, for supramolecular association in one polymorphic form of neutral clonixin and in a zwitterionic form.9 In this case, similar supramolecular chains are formed via O–H· · ·N(pyridine) hydrogen bonds in the neutral form and via pyridinium-N–H· · ·O hydrogen bonds in the zwitterionic form giving rise to a similar topology for each chain (see Figure 3).9 Two further examples are described below. In a systematic evaluation of topological similarity in a series of diarylethynylmethanol structures (Figure 23a), several supramolecular synthons were identified; in no circumstances were conventional O–H· · ·O hydrogen bonding observed.125 In the case when Y = methyl, synthons

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

15

R = Methyl

R O

Sn

R R Sn

O

O

OH

N OH

N

N

R

O

R

R

R

R

R

Sn

O

R Sn O

O

OH

N

R OH

N

N

N

O

O

R

R R

Sn

O N

R OH

N

N

R = Cyclohexyl

Figure 22 A linear example of Aufbau crystal packing for triorganotin 2-[(E)-2-(2-hydroxy-5-methylphenyl)-1-diazenyl]benzoates. Molecules precipitate from solution and adopt orientations to optimize intermolecular interactions, in this case secondary Sn· · ·O contacts. When the steric bulk of the tin-bound substituents allows for the close approach of molecules, as in the R = methyl example, a supramolecular chain is formed comprising trans-C3 SnO2 tin atom geometries. When the steric bulk precludes close association, ˚ monomeric packing is found with cis-C3 SnO2 tin atom geometries. The linear repeat distance between successive tin atoms is 4.89 A ˚ for the R = cyclohexyl derivative. for the R = methyl structure compared to 5.17 A Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

16

Supramolecular materials chemistry

2.7

H HO

Y (a)

Y Y = CH3 or Cl

(b)

(c)

Figure 23 Structural mimicry in diarylethynylmethanol derivatives. (a) Generic chemical structures: when Y = methyl, the molecule is bis(4-tolyl)ethynylmethanol, and when Y = chloride, the molecule is bis(4-chlorophenyl)ethynylmethanol, (b) for Y = methyl, dimeric aggregate based on O–H· · ·π interactions, and supramolecular chain based on ethynyl-C–H· · ·π interactions, and (c) for Y = chloride, dimeric aggregate based on ethynyl-C–H· · ·π interactions, and supramolecular chain based on O–H· · ·π interactions. The O–H· · ·π and C–H· · ·π interactions are shown as purple dashed lines.

shown in Figure 23(b) were observed, that is, O–H· · ·π interactions, leading to a dimeric aggregate, and ethynylC–H· · ·π interactions, leading to a supramolecular chain. Dimeric and chain aggregates were also observed in the Y = chloride structure but in this case, these were mediated by ethynyl-C–H· · ·π and O–H· · ·π interactions, respectively (Figure 23c). Variations in supramolecular synthons were rationalized in terms of steric bulk as well as halogen· · ·halogen interactions.125 Global crystal packing considerations are shown to be important in the construction of crystal structures of 2oxa-steroids.126 Two of the synthetic steroids are shown in Figure 24(a) with the difference between the structures being an hydroxyl group. In the crystal packing, supramolecular chains are formed in each case. In the case of the hydroxyl derivative, these are mediated by conventional lactone-O· · ·H–O plus lactone-O· · ·H–C interactions (Figure 24b). When the hydroxyl group is absent, the chain is mediated by lactone-O· · ·H–C interactions alone (Figure 24c). It is salient to note that the illustrated chains are aligned along the crystallographic a-axis in these crystal structures and that these have very similar lengths, that ˚ respectively.126 is, 6.2321(3) and 6.2214(7) A,

Analysis and visualization of supramolecular synthons

From the foregoing, it is clear that a wide variety of supramolecular synthons are now recognized and that seemingly strong and directional interactions such as hydrogen bonding may be overwhelmed by nominally weaker interactions acting in tandem. Identifying all interactions requires detailed analysis of geometric parameters obtained from programs such as Spek’s widely used PLATON routine.127 Further insight into intermolecular interactions is conveniently gained by an exploration of electrostatic potentials mapped on a Hirshfeld surface128, 129 with the use of Crystal Explorer.130 Here, all intermolecular contacts in operation are indicated on a Hirshfeld surface calculated for the molecule within its crystalline environment and the relative importance of each different intermolecular contact quantified. It is interesting to note that in systems in which hydrogen-bonding interactions coexist with other, nominally weaker intermolecular interactions, the contribution to the energy of stabilization associated with the hydrogen bonding computes to values usually less than 40%.10, 131, 132 This bears out the observations made above that global crystal packing considerations are also important in determining the ultimate crystal structure. Finally, for further descriptions of noncovalent interactions operating in crystal structures along with their mathematical description, see Noncovalent Interactions in Crystals, Supramolecular Materials Chemistry.

3 3.1

APPLICATIONS PART I—MULTICOMPONENT CRYSTALS Preamble

In the realm of organic crystal engineering, it is the formation of multicomponent crystals and the understanding of the phenomenon of polymorphism that lie at the forefront of contemporary research. The unifying feature of studies highlighted in this section is the formation of hydrogenbonding interactions between the constituent molecules or ions, with one notable exception. Motivations for this study of multicomponent crystals have varied over time but it is probably fair to say that at present the focus of much attention rests with optimizing pharmaceutical products, incorporating the desire to control polymorphic behavior. Herein, an illustrated and brief overview of aspects of multicomponent crystals is presented. This overview is not meant to be comprehensive as a chapter in this volume is dedicated to this topic (see Cocrystals: Synthesis, Structure, and Applications, Supramolecular Materials Chemistry). Rather it is designed to highlight the scope of crystal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

17

O

O O (a)

(b)

Y

Y = OH or H

(c)

Figure 24 Structural mimicry in 2-oxa-steroids. (a) Generic chemical structures: when Y = hydroxyl, the molecule is 6α-hydroxy2-oxa-androst-4-ene-3,17-dione, and when Y = hydrogen, the molecule is 2-oxa-4-androsten-3,17-dione; (b) two views of the supramolecular aggregation in the crystal structure of 6α-hydroxy-2-oxa-androst-4-ene-3,17-dione via O–H· · ·O and C–H· · ·O interactions; and (c) two views of the supramolecular aggregation in the crystal structure of 2-oxa-4-androsten-3,17-dione via C–H· · ·O interactions only as the O–H· · ·O hydrogen bond in (b) has been replaced by a C–H· · ·O contact. The O–H· · ·O and C–H· · ·O interactions are shown as orange and blue dashed lines, respectively.

engineering studies as applied to organic molecules. Before embarking on this task, a comment concerning nomenclature relating to multicomponent crystals is appropriate. The term multicomponent crystal is an all encompassing term describing crystals containing more than one molecule, for example, a solvate, a salt, and so on. Sometimes these might be referred to as cocrystals. The debate on an overarching definition for the term cocrystal has been summarized recently,133 and there seems some way to go before there is a general accord (see discussion on nomenclature of crystal engineering; Concepts and Nomenclature in Chemical Crystallography, Supramolecular Materials Chemistry). While there is no doubt that the term cocrystal has come to the fore, this term is used herein exclusively in the context of pharmaceutical products, and the generic term used otherwise.

3.2

Additives for crystal growth

One of the earliest desired outcomes of cocrystallization experiments was the isolation of suitable crystals for

single-crystal X-ray crystallographic studies. Here an additive was cocrystallized with a molecule for which the structure determination was required with the aim of forming a multicomponent crystal. Etter’s contribution to this field is succinctly summarized in her 1988 article titled Triphenylphosphine Oxide as a Crystallization Aid.134 From a crystal engineering perspective this contribution highlights the importance of triphenylphosphine to function as a proton acceptor and the utility of its bulky shape to preclude the formation of sheet-like crystals.

3.3

Cocrystallization as a means for separating enantiomers

Separation of enantiomers can often be challenging especially when chromatographic methods fail. Cocrystallization can and indeed is used to achieve chiral resolution. The classic example of the utility of cocrystals in this regard is the industrial-scale enantiomeric separation of 3(ammoniomethyl)-5-methylhexanoate (Figure 25a).135, 136

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

18

Supramolecular materials chemistry

a

0

O

−

+ O NH2

c H

OH OH

(a)

O

(b)

Figure 25 Multicomponent crystals for enantiomeric resolution. (a) Generic chemical structures for 3-(ammoniomethyl)-5methylhexanoate (upper structure) and (S)-mandelic acid and (b) supramolecular layer in the ac plane comprising alternating rows of 3-(ammoniomethyl)-5-methylhexanoate and (S)-mandelic acid.

The S-form is a key intermediate in the synthesis of the anticonvulsant drug Lyrica (pregabalin). The resolving agent of choice is (S)-(+)-mandelic acid (Figure 25a), whereby the result of a 1 : 2 cocrystallization leads to the isolation of 1 : 1 multicomponent crystals formed between the S-form of zwitterionic 3-(ammoniomethyl)-5-methylhexanoate. The structure is sustained by a complex pattern of hydrogen bonds (Figure 25b); presumably, the solubility of the other 1 : 1 cocrystal involving the other hand of the zwitterion is greater.136 Simple recrystallization of the isolated 1 : 1 multicomponent crystal allows for the isolation of the S-form of the synthetic precursor.135

3.4

Cocrystallization for absolute structure determination

Cocrystallization of chiral molecules with molecules carrying heavy atoms (to exploit anomalous dispersion effects) resulting in multicomponent crystals has proven useful to enable the determination of absolute structure via crystallographic studies, obviating the need to form a heavy atom salt or for chemical derivatization. In a case study, Bhatt and Desiraju demonstrated the utility of this approach by, for example, forming multicomponent crystals comprising cholesterol and 4-iodophenol.137

3.5

Multicomponent crystals for NLO applications

Multicomponent crystals involving organic molecules have also been investigated for their utility as NLO materials

owing to their potential applications in photonic devices. In crystalline materials, the second-order effect of NLO properties, that is, second-order harmonic generation (SHG), is dependent on two key factors. SHG depends on the magnitude of hyperpolarizability (β) of the molecule in question and on a noncentrosymmetric arrangement of molecules in the crystal. Having ionic and hydrogen bonding coexisting in a crystal structure is one approach at optimizing NLO properties and this was accomplished by forming multicomponent crystals.138–140 Exploiting the ability of nitrophenols to function simultaneously as hydrogen-bond donors and as strong acids, and the ability of 4-aminopyridines to function as hydrogen-bond acceptors and as strong bases, both having conjugated π -systems, cocrystallization experiments between the neutral molecules met the desired goals of forming salts.138 In this early study, admittedly based on a small sample, multicomponent crystals displayed a greater tendency to form noncentrosymmetric crystals.138

3.6

Salt formation for stabilizing unusual and unstable ions

In continuing and most elegant studies, Mak and coworkers have exploited crystal engineering principles to capture and thereby stabilize hitherto rare or even uncharacterized species within a crystalline manifold.141, 142 For example, monocyclic oxocarbon dianions Cn On 2− for n = 3, 4, 5, and 6 have been characterized in this manner. As an illustrative example, the squarate dianion, C4 O4 2− (Figure 26a), has been trapped in a crystal structure with formulation [Et4 N]2 [C4 O4 2− ]· 4(NH2 )2 C=S·2H2 O.141 The anions are located in a three-dimensional host lattice whereby zigzag

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

3.7 O

O−

O

O

b −

(a)

C

(b)

0

(c)

Figure 26 Multicomponent crystals designed to stabilize rare chemical species. (a) Generic chemical structure of the squarate dianion, C4 O2 2− , (b) hydrogen-bonding environment about the squarate dianion, and (c) view of the incorporation of the ammonium cations in the channels defined by the host lattice. The O–H· · ·O and N–H· · ·S hydrogen bonds are shown as orange dashed lines, and N–H· · ·O are shown as blue dashed lines.

layers comprising hydrogen-bonded (N–H· · ·S) thiourea molecules are linked by N–H· · ·O hydrogen bonds to supramolecular chains comprising squarate dianions and water molecules connected by O–H· · ·O (Figure 26b); the ammonium cations are located in the channels defined by the host lattice (see Figure 26c).141 In allied studies, Ohashi has applied crystal engineering principles to species of interest to the cosmetics industry. Hydroquinone (Figure 27a) is an effective whitening agent with potential application as a cosmetic. Mitigating its use is its susceptibility to oxidation which results in the formation of quinine, a dark substance that renders the original material unusable. The incorporation of hydroquinone in molecular complexes of surfactants such as benzyl(hexadecyl)dimethylammonium chloride (Figure 27a), protects the dihydroxyl compound from oxidation and affords the opportunity for its development as a cosmetic.143 In the crystal structure, the hydroquinone and chloride species are connected into a supramolecular chain via O–H· · ·O and O–H· · ·Cl hydrogen bonds (Figure 27b). The chains slot into voids defined by the ammonium cations, and are held in place by C–H· · ·π interactions, Figure 27(c). In related work Ohashi and colleagues also demonstrated that incorporating different perfumes in molecular complexes of surfactants reduced their volatility.144

19

Luminescent multicomponent crystals

Arene-perfluoroarene multicomponent crystals, mentioned above in the context of shapes being important for crystal packing (see Section 2.2), of 4,4
-diphenylethynyl2,2
-bipyridine and 4,4
-bis(pentafluorophenylethynyl)-2,2
bipyridine (Figure 28a) are included here for the sake of completeness and represent the only examples of multicomponent crystals not featuring hydrogen-bonding interactions described in this chapter.145 In their 1 : 1 multicomponent crystal, characterized as a hemi-benzene solvate, extended π· · ·π interactions are observed owing to substantial overlap between the aromatic rings arranged in a head to tail manner (Figure 28b). Solid-state photoluminescence experiments of the resultant excited electron donor–acceptor complex exhibited a significant redshift compared to the pure components, each of which packed in a zigzag manner.145

3.8

Pharmaceutical cocrystals

3.8.1 Rationale for cocrystal formation Without question the greatest activity in this study of multicomponent crystal formation relates to their relevance to the pharmaceutical industry. In recognition of this important subdiscipline of crystal engineering, materials formed in this category are termed pharmaceutical cocrystals, after the definition proffered by Zaworotko and colleagues,146 which encompasses a crystal containing stoichiometric amounts of an active pharmaceutical ingredient (API) and a pharmaceutically acceptable cocrystal former, each of which is a solid under ambient conditions. Examples of pharmaceutically acceptable cocrystal formers are drawn from molecules fit for human consumption and therefore would be included in the FDA’s list of Generally Regarded As Safe (GRAS) compounds,147 giving 100 or more acceptable cocrystal formers. There are two fundamental motivations for the development of pharmaceutical cocrystals, one altruistic (leading to better drugs) and the other perhaps motivated by more basic commercial concerns (patent extension and protection). It is a fact of life that most pharmaceuticals are available as crystalline formulations under ambient conditions, that is, as pure compounds, salts, or solvates. The reasons for this vary from ease of reproducible synthesis, ease of processing, stability for storage, and so on. From an intellectual property (IP) perspective, a drug protected by IP subsequently found in a new crystalline form might be patented separately as a new material, thereby extending patent rights or undermining the original patent and royalties.148 It is clear that the understanding of the process(es)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

20

Supramolecular materials chemistry

OH

OH

+N

Cl

−

(CH2)15CH3

(a)

(b)

0

a

(c)

Figure 27 Multicomponent crystals designed to stabilize reactive chemical species. (a) Generic chemical structures of each component of benzyl(hexadecyl)dimethylammonium chloride sesquikis(hydroquinone), (b) supramolecular chains comprising hydroquinone and chloride anions, and (c) view of the incorporation of the supramolecular chains shown in (b) within the host assembly comprising the ammonium cations. The O–H· · ·O and O–H· · ·Cl hydrogen bonds are shown as orange and blue dashed lines, respectively.

leading to the formation of crystalline materials is one of the most challenging issues facing contemporary crystal engineering (see also Polymorphism: Fundamentals and Applications, Supramolecular Materials Chemistry for a discussion on polymorphism). As many APIs carry functional groups capable of forming a myriad of intermolecular interactions let alone possess significant conformational flexibility, it is not surprising that these are prone to form polymorphs. In 1965, McCrone provided a definition for polymorphism that has prevailed, namely,149 A solid crystalline phase of a given compound resulting from the possibility of at least two different arrangements of the molecules of that compound in the solid state.

Over and above IP issues, polymorphism is of importance as different crystalline arrangements will naturally result in distinctive chemical and physical properties such as solubility, stability (shelf life), bioavailability, and so on.150 Important examples that underscore the importance of polymorphism and IP issues, are found in the

well-documented cases of Zantac (ranitidine hydrochloride),151, 152 Cefdinir (an oral cephalosporin),153 and Paxil (paroxetine hydrochloride hemi-hydrate).154 The need to control/restrict polymorph formation from an IP perspective is obvious. This being stated, the pharmaceutical industry is faced with an enormous task in consideration of crystal structure prediction (CSP) calculations that reveal a given compound may present a hundred or more polymorphic crystal structures with energies of stabilization within tens of kJ mol−1 (see also Crystal Structure Prediction, Supramolecular Materials Chemistry).155 Cocrystal formation may provide an avenue by which the propensity of APIs to form polymorphs is controlled or at least restricted. Surveys of the CSD14 have indicated that the prevalence of polymorphism in single-component organic molecules (circa 1.4%) is about the same as exhibited by all organic multicomponent crystals sustained by hydrogen bonding (circa 1.9%),156 and that the tendency of each class to form multiple crystal forms is significantly less than exhibited by APIs.33, 157 The working hypothesis is that cocrystal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

21

F

F F

F F

N N

N N

F F (a)

F F

F

(b)

Figure 28 Luminescent multicomponent crystals. (a) Generic chemical structures of 4,4
-diphenylethynyl-2,2
-bipyridine (upper structure) and 4,4
-bis(pentafluorophenylethynyl)-2,2
-bipyridine and (b) stacking in the 1 : 1 multicomponent crystal leading to solidstate luminescence owing to substantial overlap between the aromatic rings arranged in a head-to-tail manner. The structure was characterized as a hemi-benzene solvate (not shown).

formation with APIs will result in drugs with a reduced likelihood to form polymorphs. The formation of a pharmaceutical cocrystal represents a noncovalent derivatization of the original API. This has an impact on IP as by definition the physiochemical properties of the original API have been altered. Different crystals will have different packing arrangements with direct implications for stability toward heat and time, humidity with relevance to storage, different solubility and dissolution rates with relevance to bioavailability, and so on.158, 159 Given that there is a broad range of GRAS compounds, and even other drugs, that can function as cocrystal formers, it is likely that physiochemical properties of pharmaceutical cocrystals can and will be optimized by careful screening. Whatever the motivation for pharmaceutical cocrystal formation, there is no dispute that this technology is making a great impact on the pharmaceutical industry and will continue to do so. Before going on to describe methods for the synthesis of pharmaceutical cocrystals, a brief historical excursion is warranted. It should be appreciated that in the open literature, the concept and exploitation of pharmaceutical cocrystals dates back over 60 years when an antiseptic drug marketed as Flavazole, which is in fact a 1 : 1 cocrystal of sulfathiazole and proflavin (Figure 29), was developed as an improvement of a previously utilized 99 : 1 formulation.160

3.8.2 Synthesis of pharmaceutical cocrystals Synthetic strategies for preparing pharmaceutical cocrystals revolve around supramolecular chemistry in that matching of supramolecular synthons is paramount. In recognition of

NH2

S

S O

N O

H2N

N H

N

NH2

Figure 29 Components of an early pharmaceutical cocrystal used as a drug. Generic chemical structures of sulfathiazole (upper structure) and proflavin.

the fact that a quarter of all drugs feature the carboxylic acid functional group,8 and following on from earlier discussions on carboxylic acids (see Section 2.1), it is apposite to revisit hydrogen-bond synthons involving this group. It has already been mentioned that the carboxylic acid dimer or eightmembered {· · ·HOC=O}2 homosynthon is only observed in a fraction of the structures where it could exist, indicating its propensity to participate in other supramolecular synthons.8 A prominent heterosynthon involving carboxylic acid is found in the seven-membered ring formed with a group containing an aromatic nitrogen atom (Figure 30). This is in fact a very robust synthon occurring in 98% of the structures containing both carboxylic acid and aromatic nitrogen (in the absence of competing hydrogen-bond participants)8 and points to a successful synthetic strategy when pharmaceutical conformers contain these functionalities. Such a retrosynthetic approach and the establishment

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

22

Supramolecular materials chemistry

O H

N

O

H

Figure 30 An example of a heterosynthon, where the hydrogen bonding occurs between two distinct functional groups, in this case, a carboxylic acid and a pyridine-nitrogen atom. The longer and weaker C=O· · ·H interaction complements the O–H· · ·N hydrogen bond and closes a seven-membered {· · ·HO–C=O· · ·HC=N} ring.

of a hierarchy of supramolecular synthons have been pioneered by Zaworotko and colleagues.161, 162 Physical methods of synthesis vary from traditional cocrystallization experiments to green chemistry approaches. While it might be counterintuitive that the often-used purification technique, crystallization, can lead to the isolation of cocrystals rather than pure components, it is the complementarity of disparate functional groups leading to heterosynthons, rather than of the relevant homosynthons of the pure cocrystal formers, that tips the energy balance in favor of the cocrystal. Simple solid-state grinding experiments (mechanochemistry)163 have proven enormously successful in forming cocrystals, a method that can

be enhanced by the addition of a small quantity of solvent (liquid-assisted grinding—LAG) which accelerates the rate of reaction (see also Mechanical Preparation of Crystalline Materials. An Oxymoron?, Supramolecular Materials Chemistry).164 A recent study comparing conventional solvent evaporation with LAG showed all compounds that could be crystallized could also be prepared by LAG.165 It is noted that having powdered samples rather than single crystals is no longer an impediment to three-dimensional structure determination.166 Slurry screening,167 sublimation,168 sonochemical,169 and other strategies156 have also been utilized for the synthesis of cocrystals. The success of these strategies is vindicated by the ever increasing list of drugs that have been derivatized as pharmaceutical cocrystals.170

3.8.3 Utility of pharmaceutical cocrystals The efficacy of the cocrystal approach to produce more reliable APIs will be demonstrated with a case study of the drug carbamazepine (Tegretol ) (Figure 31a),171 a prominent antiepileptic drug that has been in use for over 30 years. At doses over 100 mg day−1 the water-insoluble drug has a high dose requirement. In addition, the drug

O

H2N N

O NH S (a)

O

O

(b)

(c)

Figure 31 The 1 : 1 carbamazepine:saccharin cocrystal system. (a) Carbamazepine (upper structure) and saccharin, (b) supramolecular aggregation in the cocrystal—original (triclinic) form. The central eight-membered {· · ·H–N–C=O}2 homosynthon is flanked by two eight-membered {· · ·H–N–H· · ·O· · ·H–N–S–O} heterosynthons, and (c) supramolecular aggregation in the cocrystal—monoclinic form. The components are linked into two molecule aggregates via an eight-membered {· · ·H–N–C=O}2 heterosynthon and these are linked into a supramolecular chain via carbamazepine-urea-N–H· · ·O-sulphonamide hydrogen bonds. The hydrogen-bonding interactions are shown as orange dashed lines. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering is known to be polymorphic with four anhydrous forms known,172 as well as pseudopolymorphs, that is, various solvates, and a hydrate.173, 174 While a number of pharmaceutical cocrystals containing carbamazepine have also been reported,173 the 1 : 1 cocrystal with saccharin was selected for careful evaluation171 ; a representation of the primary supramolecular synthons operating in the 1 : 1 carbamazepine:saccharin cocrystal is shown in Figure 31(b). The 1 : 1 carbamazepine:saccharin cocrystal was evaluated for polymorphism, using high throughput crystallization trials, physical stability, and bioavailability.171 While no obvious differences were noted in terms of chemical and physical stability between the anhydrous form of carbamazepine and the cocrystal, the cocrystal did not produce any different crystalline forms, exhibited favorable dissolution properties over the anhydrous form of carbamazepine, and was more resistant to hydrate formation. The above is cited as a “proof of concept” in terms of controlling polymorphism and imparting desirable solubility properties, and motivates further studies in pharmaceutical cocrystals. Other case studies have been reviewed very recently.156

3.8.4 Polymorphism and pharmaceutical cocrystals It is appropriate to commence this section with another quote from McCrone,149 . . .every compound has different polymorphic forms, and that, in general, the number of forms known for a given compound is proportional to the time and money spent in research on that compound.

This section aims to highlight the veracity of this statement. While one of the aims of pharmaceutical cocrystal formation is to control polymorphism, this phenomenon is known among cocrystals.156 An outstanding example, and quite plainly vindicating the comments of McCrone cited above, relates to the 1 : 1 carbamazepine:saccharin pharmaceutical cocrystal discussed in Section 3.8.3. As mentioned above, this particular cocrystal underwent extensive polymorph screening with no new form being discovered.171 Subsequently, owing to the interest in this cocrystal system that continues to entice attention, a new form was reported by another group using functionalized cross-linked polymers as heteronuclei.175 The original cocrystal featured homo- and heterosynthons to form a four-molecule aggregate (Figure 31b), whereas homosynthons are absent in the second form in which two molecule aggregates linked by an heterosynthon assemble into a supramolecular chain (Figure 31c).175 A related phenomenon is found in the generation of polymorphs of pure cocrystal formers from failed cocrystallization experiments. The most famous example relates the isolation of crystals of a second form of aspirin from

23

1 : 1 cocrystallization experiments of aspirin and levetiracetam from hot acetonitrile.176 While there has been controversy over the reality of the second form,177, 178 independent evidence for the existence of two forms has appeared recently.179 The case of benzidine (Figure 32) is another prominent example where, in addition to formation of the desired cocrystals, four polymorphs were formed.180 While the prevalence of this phenomenon is not known, that is, isolating polymorphs from cocrystallization experiments, Arora and Zaworotko have suggested cocrystallization as a new methodology to be considered for high throughput and thorough polymorph screening.156 A related phenomenon to polymorphism is the occurrence of multiple molecules in the crystallographic asymmetric unit (Z
> 1). Discussion on this topic continues but such circumstances probably reflect the isolation of a kinetic form rather than the final thermodynamic form. Steed has referred to structures with high Z
> 1 as fossil relics of more stable forms.32 A complimentary point of view refers to such structures as being a snapshot along a crystallization pathway.181 With relevance to the likelihood of cocrystal formation, in a search of the CSD,14 Steed and colleagues demonstrated that pure organic molecules of biological interest that crystallized with Z
> 1 were more likely to form cocrystals than molecules crystallizing with Z = 1.182

3.9

Solid-state reactions in multicomponent crystals

The pioneering efforts4 of Schmidt in conducting solidstate photochemical reactions has already been noted in Section 1 (see also Templated [2 + 2] Photodimerizations in the Solid State, Supramolecular Materials Chemistry). Multicomponent crystals too have been used as templates for arranging substrates in appropriate orientations to allow various reactions in the solid state to be conducted, sometimes enabling synthesis of molecules otherwise not possible in solution chemistry. The prototype reaction is the [2+2] photodimerization of bis(4pyridyl)ethene held in place by resorcinol molecules via O–H· · ·N hydrogen bonds. This reaction is so facile that it can be used as an undergraduate laboratory experiment.183 Ladderanes can be generated by the same methodology as shown in Figure 33.184 Far from being restricted to organic molecules, such solid-state reactions can also be conducted in metalorganic systems with great effect.185, 186 H2N

NH2

Figure 32 Polymorphs of cocrystal formers from cocrystallization experiments. Generic chemical structure of benzidine.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

24

Supramolecular materials chemistry

O

H

N N

CH3O O

H

H

O OCH3

N N

H

O

hn

N

(a)

CH2Cl2

N

(b)

N

N

(c)

Figure 33 Solid state [2+2] photodimerization reaction leading to a ladderane. (a) Schematic outline of the reaction whereby two 1,6-bis(4-pyridyl)hexatriene molecules are orientated by two 5-methoxyresorcinol molecules with the separation between parallel C=C ˚ On irradiation with UV for 120 h, (2+2) photodimerization reaction takes place and after bonds being in the range of 3.78–3.82 A. recrystallization from dimethyl sulfoxide (DMSO), 100% stereospecific conversion is evident. (b) Four component supramolecular aggregate sustained by O–H· · ·N hydrogen bonds (shown as orange dashed lines) and (c) molecular structure of the product, 1,2,11,12tetrakis(4-pyridyl)-(5)-ladderane (solvent molecule of crystallization omitted for reasons of clarity).

In an intriguing example, Vittal and colleagues demonstrated that an entire coordination polymer could be irradiated in situ, that is, within a single crystal, leading to a [2+2] photodimerized product, as shown in Figure 34.187 The foregoing provides a neat segue to the next section devoted to outlining exciting developments in metalorganic crystal engineering.

4

APPLICATIONS PART II—COORDINATION POLYMERS

4.1

Preamble

It is proposed that a new and potentially extensive class of scaffolding-like materials may be afforded by linking together centers with either a tetrahedral or an octahedral array of valences by rod-like connecting units.188

So starts the Abstract of the landmark paper on which rests much of the crystal engineering endeavors based on metalorganic systems. In their typically unassuming manner, Hoskins and Robson outlined a net-based approach

for the rational design of coordination polymers based on Wells’ description of inorganic structures in terms of networks.189 Arguably one of the fastest growing areas in chemistry, research on coordination polymers generates numbers of publications that seem to grow exponentially with time as noted in some recent books and themed journal issues (see Section Further Reading and Coordination Polymers, Supramolecular Materials Chemistry); a generic or conservative approach is adopted in this document whereby the term coordination polymer encompasses 1D, 2-D, and 3-D MOFs (metalorganic frameworks) (see Vittal and colleagues190 and Concepts and Nomenclature in Chemical Crystallography, Supramolecular Materials Chemistry for discussions of terminology). Over and above overcoming synthetic challenges and aesthetic appeal, not to be underestimated, coordination polymers present an enormous variety of potential practical applications. Some of the applications relate to porous materials for gas storage and catalysis, separation science, physiochemical properties, such as magnetic and luminescent materials, and so on. Herein, after some introductory comments, brief summaries of several key applications of coordination polymers are given and references made to recent review articles.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering CF3

CF3 CF3

O O Zn

N

hn N N

Zn

O O O O

(a)

Zn

Zn OO OO

CF3 CF3

CF3

O O

O O

N

OO OO Zn

25

O O N

O O OO Zn

N

Zn OO O O

N N

Zn

O O O O CF3 CF3

(b)

(c)

Figure 34 Solid-state [2+2] photodimerization reaction within a coordination polymer conducted within a single crystal. (a) Schematic outline of the reaction whereby two bridging bis(4-pyridyl)ethene molecules are orientated within a coordination polymer with the ˚ On irradiation of a single crystal with UV for 3 h, a [2+2] photodimerization separation between parallel C=C bonds being 3.75 A. reaction takes place. (b) A view of the coordination polymer highlighting the orientation of the bis(4-pyridyl)ethane molecules within the double chain and (c) a view of the reaction product showing the tetradentate 1,2,3,4-tetrakis(4-pyridyl)cyclobutane ligand generated in situ.

4.2

Opportunities and challenges

Allowing for conformational flexibility, practitioners of crystal engineering with organic molecules need to contend with polymorphism and pseudopolymorphism when designing crystal structures. On the other hand, those dealing with metalorganic systems have a far wider range of variables. These, often interrelated, include chemical and bonding considerations relating to the metal center, such as coordination number and geometry, oxidation state, etc, as well as to the mode of synthesis: solvent system, temperature, pressure, pH, counterions, and so on. Coupled with these variables are the almost inexhaustible range (denticity, flexibility, donor atoms, size, etc.) of organic bridging

ligands that provide the struts between metal nodes required for the generation of high-dimensional architectures. Hence, the possibilities of synthetic outcomes are large even for the most carefully designed synthesis. There are additional challenges in crystal engineering with coordination polymers. One of these relates to the concept of supramolecular isomerism (see also Supramolecular Isomerism, Supramolecular Materials Chemistry).191 For a given chemical reaction, different synthetic outcomes are always possible, as noted above, and it is well known that species with different chemical compositions, that is, incorporating various solvent/guest molecules, might give rise to distinctive supramolecular arrangements. While this versatile chemistry is often

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

26

Supramolecular materials chemistry

N (a)

N

Cu

n

(b)

(c)

(d)

Figure 35 Supramolecular isomerism—influence of solvent. (a) Chemical structure of the basic building block, copper(I) 2ethylimidazolate, (b) supramolecular chain isolated from a polar solvent, (c) triple-stranded helix shown in space filling mode resulting from the combination of three symmetry-related chains illustrated in (b), and (d) zigzag chain obtained from a less-polar solvent.

encountered and has been termed supramolecular isomerism,191 more intriguing are circumstances where different supramolecular architectures are observed for species with identical chemical composition, a phenomenon also called supramolecular isomerism 191 or probably more precisely termed as genuine supramolecular isomerism.192 While, in a sense, supramolecular isomerism resembles polymorphism, the presence of different conformations,

geometries, topologies, and so on demands a distinctive term (see Concepts and Nomenclature in Chemical Crystallography, Supramolecular Materials Chemistry for discussion on nomenclature in crystal engineering). An example of solvent-induced supramolecular isomerism is found in the copper(I) salt of 2-ethylimidazolate (Figure 35a).193 From a polar solvent, the supramolecular chain (Figure 35b) assembles into a triple-stranded

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering helix (Figure 35c). But when formed from a less-polar solvent system, a comparatively simple zigzag supramolecular chain is formed (Figure 35d).193 It is noted that supramolecular isomerism also has precedents in organic crystal engineering.194 An additional challenge in crystal engineering is interpenetration (also see Interpenetration, Supramolecular Materials Chemistry).195–197 While the aim of many crystal engineering studies on coordination polymers is to design open frameworks to allow subsequent chemistry, in keeping with the notion that stable crystal structures must present efficient crystal packing, voids in crystal structures are often filled by other species, such as solvent and/or counterion. The phenomenon of entanglement or interpenetration describes the situation whereby one net in a crystal structure is mutually interpenetrated by another, or others, so that otherwise free volume is occupied. It should be noted that in order to disentangle one interpenetrated net from another requires the breaking of bonds. A simple one-dimensional example of interpenetration occurring in the structure of bis((µ2 -1-bromo-3,5-bis(imidazolylmethyl)benzene)bis(acetato-O)zinc(II)) dehydrate is shown in Figure 36(a).198 Here complex molecules are connected into a linear chain via Br· · ·Br interactions ˚ (Figure 36b). Two symmetry-related chains mutu(2.58 A) ally interpenetrate in a one-dimensional parallel mode

27

(Figure 36c).198 As with supramolecular isomers, there are precedents for interpenetration in purely organic compounds but these are comparatively rare and usually go unrecognized.199 From the foregoing, it is evident that many challenges face the crystal engineer in the design of coordination polymers. At the same time, careful control of synthetic outcome allows the fine tuning of desired properties and presents the crystal engineer with many opportunities.192

4.3

Synthetic strategies for coordination polymer generation

As with multicomponent crystals, the synthesis of coordination polymers requires a retrosynthetic approach with the choice of metal(s) and connecting ligand(s) within the so-called secondary building unit (which can be a cluster, a chain, or even a layer) and spacer molecule(s) being paramount. The metal is chosen for several reasons including the following: for coordination geometry and disposition of coordination sites to secure dimensionality, for charge for the design of neutral or charged species, and for it to be of appropriate character for subsequent application (e.g., magnetic and photophysical). In the same way, ligands to operate as the organic struts need to be selected

Br

N

N

OAc OAc Zn N

N

N

N Zn

N

.2H2O

N

AcO (a)

OAc Br

(b)

(c)

Figure 36 Interpenetration. (a) Chemical structure of the basic building block, bis((µ2 -1-bromo-3,5-bis(imidazolylmethyl)benzene) ˚ and (c) one-dimensional parallel bis(acetato-O)zinc(II)) dihydrate, (b) one-dimensional chain sustained by Br· · ·Br interactions (2.58 A), interpenetration represented in space-filling mode. Water molecules of crystallization are omitted from (b) and (c). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

28

Supramolecular materials chemistry

(a)

N

N

N (b)

N

[M]

[M]

N

[M]

N

N

N

N

N

N

[M]

[M]

N

N

N

N

N

N

[M]

[M]

N N N

(c)

N

Figure 37 Simple connectors for the construction of coordination polymers. (a) Chemical structures of (a) pyrazine (being an example of a linear connector), (b) pyrimidine (angular connector), and (c) 1,3,5-triazine (trigonal connector).

with appropriate donors for the metal in question, suitable geometric disposition, and number of donors to enable the construction of the coordination polymer. In its most basic form, the methodology can be summarized in the following terms. As an illustration of the above in its simplest form, coordination polymers can be envisaged as the coupling of metal complexes with ditopic ligands. Referring to the three chemical structures shown in Figure 37, the relative dispositions of the nitrogen donors in the illustrated species will obviously control the relative positions of coordinated metal centers and the topology of the resultant supramolecular architecture. Tetrahedral and octahedral connectors are obvious extensions to these and can lead to three-dimensional architectures. In terms of metal centers, different architectures may be envisaged depending on the coordination propensity/requirements. As shown in Figure 38 for the ubiquitous 4,4
-bipyridine ligand,200, 201 linear chains, ladders, and two-dimensional arrangements can be designed depending on the coordination requirements and preexisting donor sets about the metal center. Elaboration on this theme includes the use of prefabricated metalorganic polyhedra and polymetallic clusters as the nodes in the construction of coordination polymers.202 Reactions are usually conducted in solution and often under hydrothermal conditions. This being stated, solventfree cogrinding of the reactants (mechanochemistry),203 microwave-assisted,204 and sonochemical205 methods are receiving increasing attention as alternative methods for synthesis. Beyond synthesis, there is increasing interest in fabricating thin films,206, 207 membranes,208 and nanofibers209 comprising coordination polymers. The success of these synthetic strategies is clearly evidenced by the myriad of coordination polymers that have been generated and by the wide range of successful applications that have been discovered. Several examples of these are outlined below with emphasis given to cases where the specific application depends on the three-dimensional character of the coordination polymer, and where the robust

(a)

N

N

N

N

[M]

[M]

(b)

[M]

(c)

N

N

[M]

N

N

N

N

[M]

N

N

[M]

Figure 38 Possible supramolecular architectures based on the 4,4
-bipyridine ligand. (a) Linear chain when [M] is a metal/secondary building unit forming two diagonally opposite bonds, (b) supramolecular ladder, and (c) supramolecular square.

coordination polymer is stabilized by covalent and/or coordinate (dative) bonds rather than other intermolecular interactions such as hydrogen bonding. Applications of coordination polymers have been described in recent comprehensive reviews,210–212 including the industrial context (see also Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry).213 Finally, while the focus below is on metalorganic systems, it is noted that increasingly all organic frameworks are appearing in the literature,214, 215 including the recently reported and intriguing “edible MOFs” containing alkali metal ions as the nodes.216

4.4

Gas storage and purification

Gas cylinders loaded with adsorbents (e.g., activated carbon and zeolites) require lower pressures compared with

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering conventional gas cylinders. Three-dimensional coordination polymers by virtue of their tunable porosity, flexible frameworks (allowing for expansion/compression), and large surface area have found utility as storage media for a range of gases, at least on a laboratory scale (see Gas Storage and Separation in Supramolecular Materials, Supramolecular Materials Chemistry). These include gases important as alternative fuels such as hydrogen and methane.

Zn

29

An early report of the physisorption of hydrogen dates back to 2003 when Yaghi et al .217 described the uptake of hydrogen by Zn4 O(1,4-benzene dicarboxylate)3 ,218 Figure 39(a); the etiology of this and members of the isoreticular series, by design, makes interesting reading.219 In Zn4 O(1,4-benzene dicarboxylate)3 , [ZnO4 ]6+ secondary building blocks are connected by six dicarboxylate ligands (Figure 39b) to form a cubic framework with 91% porosity

6+

Zn O Zn Zn O

O 2−

O

O

(a)

(b)

b

C

0

(c)

Figure 39 Gas storage in coordination polymers. (a) Chemical structures of the [Zn4 O]6+ secondary building block and terephthalate dianion in the structure of Zn4 O(1,4-benzene dicarboxylate)3 , (b) representation of the cubic cavity, and (c) view down the a-axis highlighting the porous framework. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

30

Supramolecular materials chemistry

OH2

R R

O O Cu O O O O

R R

O Cu O OH2

O

4_

O

O

O

O

O O

O

(a)

triangular and six square windows allowing access by gas molecules. One way of describing the three-dimensional structure is based on a honeycomb array with internal orthogonal bridges (Figure 40b). This material exhibited a methane gas adsorption capacity of 230 v/v (290 K and 35 bar) compared with the DOE target of 180 v/v.221 With the ability to incorporate gases and so many options to vary porosity, it is not surprising that coordination polymers also find utility for gas separation and purification.222, 223 Examples of gas separation and purification facilitated by coordination polymers are increasing in number.210–213 A very recent example of gas separation was reported by Li and colleagues whereby CO2 was selectively adsorbed over other gases.224 The coordination polymer Zn2 (4,4
-biphenyldicarboxylate)2 (1,2-bis(4pyridyl)ethane) di-dimethylformamide solvate (Figure 41), features eight-membered [ZnOCO]2 ring secondary building blocks (Figure 41a), each of which is connected to four neighbors to generate a brick-like net. Two nets interpenetrate to form a two-dimensional layer and these are bridged by the ditopic 1,2-bis(4-pyridyl)ethane molecules to form a three-dimensional arrangement that defines channels O

Zn O

O

Zn

R

R O

O

_ O 2

O

O

(b)

Figure 40 Gas storage in coordination polymers. (a) Chemical structures of the dicopper(II) paddlewheel secondary building block and 5,5
-(9,10-anthracenediyl)diisophthalate tetra-anion in the structure of the [µ8 -5,5
-(9,10-anthracenediyl)diisophthalato]diaqua-dicopper(II) dimethylformamide solvate, and (b) representation of the honeycomb array with internal orthogonal bridges. The solvent molecules are not shown and the water-bound hydrogen atoms were not located in the original study.

in its unit cell when guest (solvent) molecules are absent (Figure 39c).218 With hydrogen uptakes as high as 7.1 wt% at 77 K and 40 bar, this remains the most efficient cryogenic storage material.220 Methane is another gas attracting attention in this regard owing to its prevalence (>95%) in natural gas. A notable success is the exceeding of the US Department of Energy’s (DOE) target of methane uptake by the coordination polymer [µ8 -5,5
-(9,10-anthracenediyl)diisophthalato]-diaquadicopper dimethylformamide solvate Figure 40(a).221 This material, features dicopper(II) paddlewheels as the secondary building units (each capped by two water molecules) linked by 5,5
-(9,10-anthracenediyl)diisophthalate ligands. Twelve carboxylate ligands bridge six paddlewheels resulting in the formation of cuboctahedral cages with eight

N (a)

N

0

a

C

(b)

Figure 41 Gas storage in coordination polymers. (a) Chemical structures of the [ZnOCO]2 secondary building block, the 4,4
biphenyldicarboxylate dianion, and ditopic 1,2-bis(4-pyridyl)ethane molecule in Zn2 (4,4
-biphenyldicarboxylate)2 (1,2-bis(4pyridyl)ethane) di-dimethylformamide solvate, a precursor for selective adsorption of CO2 and (b) representation of the connections between doubly interpenetrated layers afforded by the ditopic molecules.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering (Figure 41b). The solvent molecules are readily removed from the original structure, a process that is accompanied by a structural change but with retention of crystallinity.224 Adsorption selectivity measurements revealed that the desolvated material was highly selective for CO2 over other gases such as CH4 , N2 , O2 , and CO. An interesting feature of the above example is the presence of interpenetration, albeit in two dimensions only. This demonstrates that rather than being an impediment to a particular application, interpenetrated structures offer further opportunities for fine tuning of, for example, pore size.225 Other coordination polymers also display selectivity in adsorption profiles as summarized recently.210–212 Extensions to these applications were realized recently where allied experiments on Zn4 O(1,4-benzene dicarboxylate)3 (Figure 39), indicate that this coordination polymer may also function as a chromatographic column for the separation of organic dyes.226

4.5

Catalysis

As with gas storage and separation, the utility of coordination polymers to function as catalysts, as first demonstrated by Fujita et al.,227 is attracting attention as coordination polymers provide an alternative to zeolites, with restricted pore sizes and mesoporous materials210–213, 228, 229 ; opportunities for biomimetic design have also been raised.228 Assuming issues associated with chemical and thermal (including hydrothermal) stability can be resolved, coordination polymers offer the potential to function as specialized catalysts owing to the presence of variable catalytic sites, that is, metal centers and organic components, as well as the possibility of pore size and shape moderation these components offer. Examples of Lewis acid (metal center), Lewis base (organic residue), Brønsted acid, base, and enantioselective catalysis have been documented; it is noted that crystallographic characterization of the coordination polymers is sometimes lacking.228, 229 Carbon–carbon bond formation reactions leading to polymerization have also been investigated.230

4.6

Incorporation of nanoparticles in three-dimensional coordination polymers

Motivated by hydrogen gas storage and catalytic considerations, for example, an emerging development is the incorporation of nanoparticles as guests in three-dimensional coordination polymers. Utilizing the inherent high surface area and porous nature of the host, nanoparticles can be stabilized and yet still be available for subsequent chemistry.231 Again the prototype metalorganic framework, Zn4 O(1,4benzene dicarboxylate)3 (Figure 39), features in pioneering

31

work in that it can function as a template whereby volatile synthetic precursors are introduced into the coordination polymer and then decomposed to yield nanoparticles of metals such as palladium and gold.232 With variations in the dimension and surface attributes of pores possible, there is enormous scope to control the size and shape of the nanoparticles formed therein.

4.7

Luminescence, magnetochemical, and nonlinear optical properties

Coordination complexes are well known to exhibit photophysical properties and, with the appropriate choice of metal, magnetochemical properties.210–212 It is therefore not surprising that supramolecular architectures up to and including three-dimensional coordination polymers attract considerable interest in this regard. While lower-dimensional coordination polymers (and multicomponent crystals, see Section 3.7) can exhibit luminescence properties (encompassing fluorescence, phosphorescence, and scintillation), perhaps uniquely, threedimensional species can potentially exhibit a full range of emissive phenomena, especially when containing d10 elements and lanthanides.233 Emissive processes envisaged in three-dimensional coordination polymers includes metal-based emission, ligand-based emission, metal–ligand charge transfer, ligand–metal charge transfer, emission responses due to adsorbed molecules, exciplex formation, and, in the case of lanthanides, the antennae effect.233 NLO properties were mentioned earlier in Section 3.5 in the context of organic crystal engineering but there is also great scope for generating noncentrosymmetric coordination polymers as second-order NLO materials.234 Perhaps, compared to other crystal engineering applications, it is a little surprising that this facet has not received more attention as, conceptually, the facile combination of nonsymmetric ligands with secondary building units can lead to noncentrosymmetric crystal structures for coordination polymers of varying dimensions. As with luminescent crystals, coordination complexes, in particular, single-molecule magnets,235, 236 have been widely explored for magnetochemical properties (encompassing ferromagnetism, antiferromagnetism, and ferrimagnetism). Opportunities afforded by three-dimensional coordination polymers revolve about tailoring spin quantum number and magnetic anisotropy by organizing the arrangement of paramagnetic metal centers (e.g., by choice of metal(s) and/or oxidation state(s), controlling separation, symmetry, etc.) connected by appropriate bridging ligands, usually short, to promote effective communication between the metal centers; it is also possible to incorporate openshell organic ligands to induce magnetic behavior. A review

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

32

Supramolecular materials chemistry polymer {[Mn3 (4,4
-bipyridine)3 (OH2 )4 ][Cr(CN)6 ]2 .2(4,4
bipyridine)· 4(H2 O)},238 Here neutral two-dimensional architectures are formed by the Mn2+ and Cr(CN)6 3− ions which feature Mn-CN-Cr-CN-Mn magnetic pathways.239 The three-dimensional structure is completed by pillaring afforded by the 4,4
-bipyridine molecules with the noncoordinating 4,4
-bipyridine and water molecules occupying space in the pores; Figure 42(a) and (b) shows two view of the structure with guest molecules removed from the framework. Removal of the guest 4,4
-bipyridine molecules as well as coordinated and guest water molecules did not disrupt the basic framework structure.238 The desolvated material exhibited the ability to adsorb various gases with a high heat of hydrogen adsorption that being noteworthy. Ferromagnetic behavior dependent on the nature of the guest was observed indicating such materials show promise as sensors.238 Finally, coordination polymers are also being examined for their utility as ferroelectric materials but three-dimensional examples are still sparse.240

a

0

c

(a)

4.8

While not nearly as well developed as some of the applications outlined above, there is significant potential for the use of coordination polymers in diagnostic and therapeutic medicine, assuming that issues associated with toxicity can be overcome. Utilizing high storage capacity along with tunable pore size, applications such as drug delivery agents by controlled release of drugs from within, new drugs comprising coordination polymers constructed from benign metals and bioactive struts, and as magnetic resonance imaging contrast agents are envisaged.241–243

b

0

Potential biomedical applications

a

4.9 (b)


Figure 42 Magnetic sensors. Two views of {Mn3 (4,4 bipyridine)3 (OH2 )4 ][Cr(CN)6 ]2 .2(4,4
-bipyridine)· 4(H2 O)} highlighting (a) the pillaring by 4,4
-bipyridine of the two-dimensional arrays comprising Mn2+ and Cr(CN)6 3− ions and (b) the square and distorted triangular channels, which are occupied by guest water and 4,4
-bipyridine molecules, respectively, but which are not shown. Removal of the guest molecules as well as coordinated water molecules results in a similar architecture that adsorbs gas molecules accompanied by ferromagnetic responses.

of magnetism as it pertains to three-dimensional coordination polymers has appeared recently.237 An interesting development is the introduction of magnetism into an otherwise nonmagnetic framework via guest molecules giving rise to the possibility of functional materials.210–212 An example of the latter is found in the bimetallic coordination

Postsynthetic modification

It is appropriate to conclude this section with another quote from the seminal paper of Hoskins and Robson,188 Relatively unimpeded migration of species throughout the lattice may allow chemical functionalization of the rods subsequent to the construction of the framework.

Here, the concept was put forward that once formed, in situ synthetic modification of three-dimensional coordination polymers is possible, a process now realized and termed postsynthetic modification.244 Motivations are obvious in that naturally distinct chemical and physical properties will ensue but also that this methodology allows the introduction of functional groups that may not survive normal methods of synthesis. Often, postsynthetic modification encompasses the reaction of an incorporated substrate with an introduced reagent.244 Introducing new chemical

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

O

O

O H N

O-t-Bu

ACKNOWLEDGMENTS

O NH2

O

33

∆

− CO2 H _

Current financial support for the author’s own work in crystal engineering at the University of Malaya through the High Impact Research (UM.C/625/1/HIR/033/1 & UM.C/ 625/1/HIR/033/2) and University of Malaya Research Grant (RG125/1-AFR) schemes is gratefully acknowledged.

H O

O

O

O

Figure 43 Postsynthetic modification. Thermolysis of Zn4 O (substituted biphenyl-4,4
-dicarboxylate)3 · 5H2 O in dimethylformamide results in the deprotection of the t-butylcarbamate group, via the evolution of CO2 and isobutylene, to yield the amine.

REFERENCES 1. G. R. Desiraju, Crystal Engineering. The Design of Organic Solids, Elsevier, Amsterdam, 1989. 2. J. D. Dunitz, Pure Appl. Chem, 1991, 63, 177.

functionality by deprotection is an emerging aspect of postsynthetic modification. An example of the latter is found in a coordination polymer based on the Yaghi prototype Zn4 O(1,4-benzene dicarboxylate)3 structure.218 Synthesis with a biphenyl-4,4
-dicarboxylic acid featuring substitution with a bulky protected amino group, t-butylcarbamate, at one of the two positions (Figure 43), gave a coordination polymer with formulation Zn4 O(substituted biphenyl4,4
-dicarboxylate)3 ·5H2 O245 with a framework architecture resembling that shown in Figure 39. Facile thermolysis of this material in dimethylformamide, to retain crystallinity, resulted in the evolution of CO2 and isobutylene, and the generation of amine-substituted biphenyl4,4
-dicarboxylate within the framework (Figure 43). As a consequence of the reaction, the void volume increased to ˚ 3 in the resulting amine-substituted approximately 4200 A ˚ 3 in the prespecies compared with approximately 3900 A 245 cursor structure. A recently reported study described the utilization of photochemical methods to deprotect a functional group leaving hydroxyl functionality within the framework.246

5

CONCLUSIONS AND PROSPECTS

The foregoing represents an illustrated overview of the three paradigms of crystal engineering: understanding the nature of intermolecular forces in order to control the way molecules pack (supramolecular synthesis) leading to the all-important applications, in both organic and metalorganic crystal engineering. Desiraju describes these as stages in crystal engineering and suggests that these represent the what, how, and why of the discipline.6 Clearly, significant advances have been made in all three aspects but, as pointed out in many reviews and as indicated above, many exciting opportunities exist for future development of crystal engineering.

3. R. Pepinsky, Phys. Rev., 1955, 100, 952. 4. G. M. J. Schmidt, et al., Solid state photochemistry, in Describing a Symbiotic Relationship between X-Ray Crystallography and Synthetic Organic Photochemistry, Monographs in Modern Chemistry, ed. D. Ginsburg, Verlag Chemie, Weinheim, New York, 1976, vol. 8. 5. For information on citation statistics on journals. http:// thomsonreuters.com/products services/science/free/essays/ impact factor/ Copyright 2011. 6. G. R. Desiraju, Angew. Chem. Int. Ed., 2007, 46, 8342. 7. V. Benghiat and L. Leierowitz, J. Chem. Soc. Perkin II , 1972, 1763. 8. T. R. Shattock, K. K. Arora, P. Vishweshwar, and M. J. Zaworotko, Cryst. Growth Des., 2008, 8, 4533. 9. M. Takasuka, H. Nakai, and M. Shiro, J. Chem. Soc., Perkin Trans. 2 , 1982, 1061. 10. N. J. Babu, S. Cherukuvada, R. Thakuria, and A. Nangia, Cryst. Growth Des., 2010, 10, 1979. 11. J. Chen and B. L. Trout, J. Phys. Chem. B , 2008, 112, 7794. 12. D. Das and G. R. Desiraju, Chem. Asian J., 2006, 1, 231. 13. A. I. Kitaigorodskii, Organic Chemical Crystallography, Consultants Bureau, New York, 1961 (English translation of the Russian original published by the Press of the Academy of Sciences of the USSR, Moscow, 1955). 14. F. H. Allen, Acta Cryst., 2002, B58, 380. 15. E. V. Peresypkina and V. A. Blatov, Acta Crystallogr., 2000, B56, 501. 16. A. D. Mighell, V. L. Himesm, and J. R. Rodgers, Acta Crystallogr., 1983, A39, 737. 17. A. J. C. Wilson, Acta Crystallogr., 1993, A49, 795. 18. C. P. Brock and J. D. Dunitz, Chem. Mater., 1994, 6, 1118. 19. For information on CSD space group statistics. http://www. ccdc.cam.ac.uk/products/csd/statistics/ Copyright 2004–2011. 20. J. D. Dunitz, G. Filippini, and A. Gavezzotti, Tetrahedron, 2000, 56, 6595. 21. J. D. Dunitz, G. Filippini, and A. Gavezzotti, Helv. Chim. Acta, 2000, 83, 2317.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

34

Supramolecular materials chemistry

22. J. D. Dunitz, G. Filippini, and A. Gavezzotti, Helv. Chim. Acta, 2000, 83, 4073.

53. I. Dance and M. Scudder, J. Chem. Soc., Chem. Commun., 1995, 1039.

23. W. D. S. Motherwell, CrystEngComm., 2010, 12, 3554.

54. I. Dance, Mol. Cryst. Liq. Cryst., 2005, 440, 265.

24. E. R. T. Tiekink and I. Haiduc, Prog. Inorg. Chem., 2005, 54, 127.

55. I. Dance and M. Scudder, CrystEngComm, 2009, 11, 2233.

25. G. Siasios and E. R. T. Tiekink, Z. Kristallogr., 1993, 204, 95. 26. D. R. Smyth, B. R. Vincent, and E. R. T. Tiekink, Cryst. Growth Des., 2001, 1, 113. 27. J. Bernstein, R. J. Davey, and J. O. Henck, Angew. Chem. Int. Edn., 1999, 38, 3440. 28. W. Bolton, Acta Crystallogr., 1964, 17, 147. 29. J. D. Dunitz and W. B. Schweizer, CrystEngComm., 2007, 9, 266. 30. F. H. Allen, C. A. Baalham, J. P. M. Lommerse, P. R. Raithby, Acta Crystallogr., 1998, B54, 320.

and

31. R. M. Ibberson, W. G. Marshall, L. E. Budd, et al., CrystEngComm, 2008, 10, 465. 32. J. W. Steed, CrystEngComm, 2003, 5, 169. 33. I. Y. H. Chan, V. T. Nguyen, R. Bishop, et al., Cryst. Growth Des., 2010, 10, 4582. 34. K. Fucke, N. Qureshi, D. S. Yufit, et al., Cryst. Growth Des., 2010, 10, 880. 35. A. Nangia, J. Chem. Sci., 2010, 122, 295. 36. P. Munshi and T. N. Guru Row, CrystEngComm, 2005, 7, 608. 37. G. A. Jeffrey, An Introduction to Hydrogen Bonding, Oxford University Press, Oxford, 1997. 38. J. M. Robertson, Organic Crystals and Molecules, Cornell University Press, Ithaca, 1953. 39. M. C. Etter, Acc. Chem. Res., 1990, 23, 120. 40. P. Vishweshwar, N. J. Babu, A. Nangia, et al., J. Phys. Chem. A, 2004, 108, 9406. 41. M. D. Ward, Struct. Bond., 2009, 132, 1. 42. G. R. Desiraju, Chem. Commun., 2005, 2995. 43. G. A. Jeffrey and W. Saenger, Hydrogen Bonding in Biological Structures, Springer, Berlin, 1991. 44. M. Nishio, M. Hirota, and Y. Umezawa, The CH/πInteraction, Wiley VCH, New York, 1998. 45. G. R. Desiraju and T. Steiner, The Weak Hydrogen Bond in Structural Chemistry and Biology, Oxford University Press, Oxford, 1999.

56. N. Burford, B. W. Royan, R. E. v. H. Spence, J. Chem. Soc., Dalton Trans., 1990, 1521.

et al.,

57. J. D. Dunitz and A. Gavezzotti, Chem. Soc. Rev., 2009, 38, 2622. 58. S. Rizzato, J. Berg`es, S. A. Mason, et al., Angew. Chem. Int. Ed., 2010, 49, 7440. 59. A. Gavezzotti, Molec. Phys., 2008, 106, 1473. 60. W. T. Pennington, T. W. Hanks, and H. D. Arman, Struct. Bond., 2008, 126, 65. 61. P. Metrangolo, G. Resnati, T. Pilati, and S. Biella, Struct. Bond., 2008, 126, 105. 62. P. Metrangolo, F. Meyer, T. Pilati, et al., Angew. Chem. Int. Ed., 2008, 47, 6114. 63. P. Metrangolo and G. Resnati, Science, 2008, 321, 918. 64. Y. Lu, Y. Wang, and W. Zhu, Phys. Chem. Chem. Phys., 2010, 12, 4543. 65. F. Guthrie, J. Chem. Soc., 1863, 16, 239. 66. O. Hassel, Science, 1970, 170, 497. 67. M. Fourmigu´e and P. Betail, Chem. Rev., 2004, 104, 5379. 68. A. Karpfen, Struct. Bond., 2008, 126, 1. 69. E. Corradi, S. V. Meille, M. T. Messina, et al., Angew. Chem. Int. Ed., 2000, 39, 1782. 70. R. Bailey Walsh, C. W. Padgett, P. Metrangolo, et al., Cryst. Growth Des., 2001, 1, 165. 71. G. M. Espallargas, F. Zordon, L. A. Mar´ın, et al., Chem. Eur. J., 2009, 15, 7554. 72. P. Metrangolo, F. Meyer, T. Pilati, et al., Chem. Commun., 2008, 1635. 73. L. Brammer, G. M. Espallargas, and S. Libri, CrystEngComm, 2008, 10, 1712. 74. R. Bertani, P. Sgarbossa, A. Venzo, et al., Coord. Chem. Rev., 2010, 254, 677. 75. F. Zordan and L. Brammer, Cryst. Growth Des., 2006, 6, 1374. 76. C. A. Bayse and E. R. Rafferty, Inorg. Chem., 2010, 49, 5365. 77. H. D. Arman, R. L. Gieseking, T. W. Hanks, W. T. Pennington, Chem. Commun., 2010, 46, 1854.

and

46. T. Steiner, Angew. Chem. Int. Ed., 2002, 41, 48.

78. C. Janiak, CrystEngComm, 2000, 2, 3885.

47. J. D. Dunitz and R. Taylor, Chem. Eur. J., 1997, 3, 89.

79. A. Das, A. Dipankar Jana, S. Kumar Seth, et al., J. Phys. Chem. B , 2010, 114, 4166.

48. K. Reichenb¨acher, H. I. S¨uss, and J. Hulliger, Chem. Soc. Rev., 2005, 34, 22. 49. A. Kovacs and Z. Varga, Coord. Chem. Rev., 2006, 250, 710.

80. D. N. Sredojevi´c, Z. D. Tomi´c, and S. D. Zari´c, Cryst. Growth Des., 2010, 10, 3901.

50. E. D’Ora and J. Novoa, CrystEngComm, 2008, 10, 423.

81. R. K. Murmann, R. Glaser, and C. L. Barnes, J. Coord. Chem., 2005, 58, 279.

51. T. V. Rybalova and I. Y. Bagryanskaya, J. Struct. Chem., 2009, 50, 741.

82. Z. D. Tomi´c, V. M. Leovac, S. V. Pokorni, et al., Eur. J. Inorg. Chem., 2003, 1222.

52. T. S. Thakur, M. T. Kirchner, D. Bl¨aser, et al., CrystEngComm, 2010, 12, 2079.

83. Z. D. Tomi´c, S. B. Novakovi´c, and S. D. Zari´c, Eur. J. Inorg. Chem., 2004, 2215.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering 84. Z. D. Tomi´c, D. N. Sredojevi´c, and S. D. Zari´c, Cryst. Growth Des., 2006, 6, 29. 85. M. Nishio, CrystEngComm, 2004, 6, 130. 86. For a database on C–H· · ·π interactions in both organic and metalorganic systems. http://www.tim.hi-ho.ne.jp/dionisio/ (accessed 2 July 2011). 87. D. Chen, C. S. Lai, and E. R. T. Tiekink, Z. Kristallogr., 2002, 218, 747. 88. Y.-F. Jiang, C.-J. Xi, Y.-Z. Liu, et al., Eur. J. Inorg. Chem., 2005, 1585. 89. M. K. Milˇciˇc, V. B. Medakovi´c, D. N. Sredojevi´c, et al., Inorg. Chem., 2006, 45, 4755. 90. D. Sredojevi´c, G. A. Bogdanovi´c, Z. D. Tomi´c, and S. D. Zari´c, CrystEngComm, 2007, 9, 793. 91. E. Galdecka, Z. Galdecki, K. Wajda-Hermanowicz, and F. P. Pruchnik, Bull. Pol. Acad. Sci. Chem., 1994, 42, 205. 92. J. Zukerman-Schpector and I. Haiduc, CrystEngComm, 2002, 4, 178. 93. A. Edelmann, S. Brooker, N. Bertel, et al., Z. Naturforsch., B: Chem. Sci., 1992, 47, 305. 94. I. Haiduc, E. R. T. Tiekink, and J. Zukerman-Schpector, in Tin Chemistry—Fundamentals and Frontiers, Chapter 3.9, eds. A. G. Davies, M. Gielen, K. H. Pannell, and E. R. T. Tiekink, John Wiley & Sons, Chichester, 2008, p. 392. 95. E. R. T. Tiekink and J. Zukerman-Schpector, Aust. J. Chem., 2010, 63, 535. 96. E. R. T. Tiekink and J. Zukerman-Schpector, CrystEngComm., 2009, 11, 1176. 97. G. Jia, N. C. Payne, J. J. Vittal, and R. J. Puddephatt, Organometallics, 1993, 12, 4771. 98. E. R. T. Tiekink and J. Zukerman-Schpector, CrystEngComm., 2009, 11, 2701. 99. S. Husebye, S. Kudis, and S. V. Lindeman, Acta Crystallogr., 1996, C52, 429. 100. E. R. T. Tiekink and J. Zukerman-Schpector, Coord. Chem. Rev., 2010, 254, 46. 101. M. Egli and R. V. Gessner, Proc. Natl. Acad. Sci. U. S. A., 1995, 92, 180. 102. T. J. Mooibroek, P. Gamez, and J. Reedijk, CrystEngComm, 2008, 10, 1501. 103. N. W. Alcock, Adv. Inorg. Chem. Radiochem., 1972, 15, 1. 104. N. W. Alcock, Bonding and Structure: Structural Principles in Inorganic and Organic Chemistry, Ellis Horwood, New York, 1990. 105. I. Haiduc and F. T. Edelmann, Supramolecular Organometallic Chemistry, Wiley-VCH, Weinheim, 1999. 106. R. Willem, I. Verbruggen, M. Gielen, et al., Organometallics, 1998, 17, 5758. 107. E. R. T. Tiekink, CrystEngComm, 2003, 5, 101. 108. E. R. T. Tiekink, CrystEngComm, 2006, 8, 104. 109. J. Perlstein, Chem. Mater., 1994, 6, 319. 110. C. O’Dowd, J. D. Kennedy, and J. Organomet. Chem., 2002, 657, 20.

M. Thornton-Pett,

35

111. E. R. T. Tiekink, Appl. Organomet. Chem., 1991, 5, 1. 112. C. Bilton, J. A. K. Howard, N. N. L. Madhavi, et al., Acta Crystallogr., 2000, B56, 1071. 113. G. L. Succaw, T. J. R. Weakley, F. Han, and K. M. Doxsee, Cryst. Growth Des., 2005, 5, 2288. 114. S. Varughse and S. M. Draper, Cryst. Growth Des., 2010, 10, 2571. 115. S. S. Kuduva, D. C. Craig, A. Nangia, and G. R. Desiraju, J. Am. Chem. Soc., 1999, 121, 1936. 116. D. Das, R. K. R. Jetti, R. Boese, and G. R. Desiraju, Cryst. Growth Des., 2003, 3, 675. 117. M. J. Katz, K. Sakai, and D. B. Leznoff, Chem. Soc. Rev., 2008, 37, 1884. 118. H. Schmidbaur, Nature (London), 2001, 413, 31. 119. B.-C. Tzeng, A. Schier, and H. Schmidbaur, Inorg. Chem., 1999, 38, 3978. 120. E. R. T. Tiekink and J.-G. Kang, Coord. Chem. Rev., 2009, 253, 1627. 121. K. M. Anderson, A. E. Goeta, and J. W. Steed, Inorg. Chem., 2007, 46, 6444. 122. M. R. Edwards, W. Jones, W. D. S. Motherwell, and G. P. Shields, Mol. Cryst. Liq. Cryst., 2001, 356, 337. 123. G. R. Desiraju and J. A. R. P. Sarma, Proc. Ind. Acad. Sci. (Chem. Sci.), 1986, 96, 599. 124. J. van de Streek and W. D. S. Motherwell, J. Appl. Crystallogr., 2005, 38, 694. 125. C. Bilton, J. A. K. Howard, N. N. L. Madhavi, et al., Acta Crystallogr., 2000, B56, 1071. 126. A. Anthony, M. Jask´olski, and A. Nangia, Acta Crystallogr., 2000, B56, 512. 127. A. L. Spek, Acta Cryst., 2009, D65, 148. 128. F. L. Hirshfeld, Theor. Chim. Acta, 1977, 44, 129. 129. M. A. Spackman and D. Jayatilaka, CrystEngComm, 2009, 11, 19. 130. For information on crystal explorer. http://www. hirshfeldsurface.net/ Copyright  2004–2010. 131. K. A. Lyssenko, P. Y. Barzilovich, Y. V. Nelyubina, et al. Russ. Chem. Bull. Int. Edn., 2009, 58, 31. 132. B. Chattopadhyay, A. K. Mukherjee, N. Narenda, et al., Cryst. Growth Des., 2010, 10, 4476. 133. J. Zukerman-Schpector and E. R. T. Tiekink, Z. Kristallogr., 2008, 223, 233. 134. M. C. Etter and P. W. Baures, J. Am. Chem. Soc., 1988, 1101, 639. 135. M. S. Hoekstra, D. M. Sobieray, M. A. Schwindt, et al., Org. Process Res. Dev., 1997, 1, 26. 136. B. Samas, W. Wang, and D. B. Godrej, Acta Crystallogr., 2007, E63, o3938. 137. P. M. Bhatt and G. R. Desiraju, CrystEngComm, 2008, 10, 1747. 138. K.-S. Huang, D. Britton, M. C. Etter, and S. R. Byrn, J. Mater. Chem., 1997, 7, 713. 139. P. Srinivasan, Y. Vidyalalshmi, and R. Gopalakrishnan, Cryst. Growth Des., 2008, 8, 2329.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

36

Supramolecular materials chemistry

140. H. J. Ravindra, K. Chandrashekaran, W. T. A. Harrison, and S. M. Dharmaprakash, Appl. Phys. B , 2009, 94, 503. 141. C.-K. Lam and T. C. W. Mak, Tetrahedron, 2000, 56, 6657. 142. J. Han, S.-Q. Zang, and T. C. W. Mak, Chem. Eur. J., 2010, 16, 5078. 143. N. Iimura, Y. Fujimura, A. Sekine, et al., Bull. Chem. Soc. Jpn., 2005, 78, 418. 144. N. Iimura, Y. Ohashi, and H. Hirata, Bull. Chem. Soc. Jpn., 2000, 73, 1097. 145. A. Hori, S. Takatani, T. K. Miyamoto, and M. Hasegawa, CrystEngComm, 2009, 11, 567. 146. D. R. Weyna, T. Shattock, V. Reddy, and M. J. Zaworotko, Cryst. Growth Des., 2008, 8, 4533. 147. For the FDA database on GRAS compounds. http:// www.fda.gov/Food/-FoodIngredientsPackaging/Generally RecognizedasSafeGRAS/default.htm (accessed 10 June 2011). 148. F. Lara-Ochoa and G. Espinosa-Perez, Cryst. Growth Des., 2007, 7, 1213. 149. W. C. McCrone, in Physics and Chemistry of the Solid State, eds. D. Fox, M. M. Labes, and A. Weissberger, Wiley Interscience, New York, 1965, vol. 2, pp. 725. 150. J. Bernstein, Polymorphism in Molecular Crystals, Clarendon Press, Oxford, 2002. 151. G. R. Desiraju, J. Ind. Chem. Soc., 2003, 80, 151. 152. J. Bernstein, J. Polymorphism, 2006, 365. 153. W. Cabri, P. Ghetti, G. Pozzi, and M. Alpegiani, Org. Process Res. Dev., 2007, 11, 64. 154. J. Lucas and P. Burgess, Pharm. Law Indust., 2004, 2, 6. 155. S. L. Price, Acc. Chem. Res., 2009, 42, 117. 156. K. K. Arora and M. J. Zaworotko, in Polymorphism in Pharmaceutical Solids, ed. H. G. Brittain, Pharmaceutical co-crystals: A new opportunity in pharmaceutical science for a long-known but little studied class of compounds, Informa Healthcare, New York, 2009, vol. 2, p. 281. 157. S. L. Morissette, S. Soukasene, D. Levinson, et al., Proc. Natl Acad. Sci., 2003, 100, 2180. 158. D. J. Good and N. Rodr´ıguez-Hornedo, Cryst. Growth Des., 2009, 9, 2252. 159. N. Schultheiss and A. Newman, Cryst. Growth Des., 2009, 9, 2950. 160. J. McIntosh, R. H. M. Robinson, F. R. Selbie, et al., Lancet, 1945, 249, 97. ¨ Almarsson and M. J. Zaworotko, Chem. Commun., 161. O. 2004, 1889. 162. J. A. Bis, P. Vishweshwar, D. R. Weyna, Zaworotko, Molec. Pharm., 2007, 4, 401.

and

M. J.

163. T. Friˇscˇ i´c and W. Jones, Cryst. Growth Des., 2009, 9, 1621. 164. N. Shan, F. Toda, and W. Jones, Chem. Commun., 2002, 2372. 165. D. R. Weyna, T. Shattock, P. Vishweshwar, and M. J. Zaworotko, Cryst. Growth Des., 2009, 9, 1106.

166. S. H. Lapidus, P. W. Stephens, K. K. Arora, et al., Cryst. Growth Des., 2010, 10, 4630. 167. G. G. Z. Zhang, R. F. Henry, T. B. Borchardt, and X. Lou, J. Pharm. Sci., 2007, 96, 990. 168. C. C. Mattheus, J. Baas, A. Meetsma, et al., Acta Crystallogr., 2002, E58, o1229. 169. J. R. G. Sander, D.-K. Bucar, R. F. Henry, et al., Angew. Chem. Int. Ed., 2010, 49, 7284. 170. N. Shan and M. J. Zaworotko, in Burger’s Medicinal Chemistry and Drug Discovery, 7th ed., eds. D. J. Abraham and D. P. Rotella, Polymorphic Crystal Forms and Cocrystals in Drug Delivery (Crystal Engineering), John Wiley & Sons, New York, 2010, p. 187. 171. M. B. Hickey, M. L. Peterson, L. A. Scoppettuolo, et al., Eur. J. Pharm. Biopharm., 2007, 67, 112. 172. A. L. Grzesiak, M. Lang, K. Kim, and A. J. Matzger, J. Pharm. Sci., 2003, 92, 2260. 173. S. G. Fleischman, S. S. Kuduva, J. A. McMahon, et al., Cryst. Growth Des., 2003, 3, 909. 174. M. C. Meyer, A. B. Straughn, E. J. Jarvi, et al., Pharm. Res., 1992, 9, 1612. 175. W. W. Porter, S. C. Elie, and A. J. Matzger, Cryst. Growth Des., 2008, 8, 14. 176. P. Vishweshwar, J. A. McMahon, M. Oliveira, et al., J. Am. Chem. Soc., 2005, 127, 16802. 177. A. D. Bond, R. Boese, and G. R. Desiraju, Angew. Chem. Int. Ed., 2007, 46, 615. 178. A. D. Bond, R. Boese, and G. R. Desiraju, Angew. Chem. Int. Ed., 2007, 46, 618. 179. E. J. Chan, T. R. Welberry, A. P. Heerdegen, and D. J. Goossens, Acta Cryst., 2010, B66, 696. 180. M. Rafilovich and J. Bernstein, J. Am. Chem. Soc., 2006, 128, 12185. 181. V. S. Senthil Kumar, A. Addlagatta, A. Nangia, et al., Angew. Chem. Int. Ed., 2002, 41, 3848. 182. K. M. Anderson, M. R. Probert, C. N. Whiteley, et al., Cryst. Growth Des., 2009, 9, 1082. 183. T. Friˇscˇ i´c, T. D. Hamilton, G. S. Papaefstathiou, and L. R. MacGillivray, J. Chem. Edu., 2005, 82, 1679. 184. X. Gao, T. Friˇscˇ i´c, and L. R. MacGillivray, Angew. Chem. Int. Ed., 2004, 43, 232. 185. I. G. Georgiev and L. R. MacGillivray, Chem. Soc. Rev., 2007, 36, 1239. 186. M. Nagarathinam, A. M. P. Peedikakkal, and J. J. Vittal, Chem. Commun., 2008, 5277. 187. N. L. Toh, M. Nagarathinam, and J. J. Vittal, Angew. Chem. Int. Ed., 2005, 44, 2237. 188. B. F. Hoskins and R. Robson, J. Am. Chem. Soc., 1990, 112, 1546. 189. A. F. Wells, Structural Inorganic Chemistry, 4th ed., Clarendon Press, Oxford, 1975. 190. K. Biradha, A. Ramanan, and J. J. Vittal, Cryst. Growth Des., 2009, 9, 2969. 191. B. Moulton and M. J. Zaworotko, Chem. Rev., 2001, 101, 1629.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Crystal engineering

37

192. J.-P. Zhang, X.-C. Huang, and X.-M. Chen, Chem. Soc. Rev., 2009, 38, 2385.

221. S. Q. Ma, D. F. Sun, J. M. Simmons, et al., J. Am. Chem. Soc., 2008, 130, 1012.

193. X.-C. Huang, J.-P. Zhang, Y.-Y. Lin, and X.-M. Chen, Chem. Commun., 2005, 2232.

222. S. Shimomura, S. Bureekaew, and S. Kitagawa, Struct. Bond., 2009, 132, 51.

194. S. Aitipamula and A. Nangia, Chem. Eur. J., 2005, 11, 6727.

223. J.-R. Li, R. J. Kuppler, and H.-C. Zhou, Chem. Soc. Rev., 2009, 38, 1477.

195. S. R. Batten, CrystEngComm, 2001, 3, 67. 196. V. A. Blatov, L. Carlucci, and D. M. Proserpio, CrystEngComm, 2004, 6, 377. 197. For a database maintained by S. R. Batten on metal based interpenetrated species, see http://www.chem.monash.edu. au/staff/sbatten/interpen/index.html. 198. J. Fan, W.-Y. Sun, T. Okamura, et al., Cryst. Growth Des., 2004, 4, 579. 199. I. G. Baburin, V. B. Blatov, L. Carlucci, et al., Cryst Growth Des., 2008, 8, 519.

224. J. Zhang, H. Wu, T. J. Emge, and J. Li., Chem. Commun., 2009, 46, 9152. 225. J. L. C. Rowell and O. M. Yaghi, Angew. Chem. Int. Ed., 2005, 44, 4670. 226. S. Han, Y. Wei, C. Valente, et al., J. Am. Chem. Soc., 2010, 132, 16358. 227. M. Fujita, Y. J. Kwon, S. Washizu, and K. Ogura, J. Am. Chem. Soc., 1994, 116, 1151.

200. L. Brammer, Chem. Soc. Rev., 2004, 33, 476.

228. D. Farrusseng, S. Aguado, and C. Pinel, Angew. Chem. Int. Ed., 2009, 48, 7502.

201. K. Biradha, M. Sarkar, and L. Rajput, Chem. Commun., 206, 4169.

229. J.-Y. Lee, O. K. Farha, J. Roberts, et al., Chem. Soc. Rev., 2009, 38, 1450.

202. J. J. Perry, IV, J. A. Perman, and M. J. Zaworotko, Chem. Soc. Rev., 2009, 38, 1400.

230. T. Uemura, N. Yanai, and S. Kitagawa, Chem. Soc. Rev., 2009, 38, 1228.

203. A. Pichon, A. Lazuem Garay, and S. L. James, CrystEngComm, 2006, 8, 211.

231. M. Meilikhov, K. Yusenko, D. Esken, et al., Eur. J. Inorg. Chem., 2010, 3701.

204. J. Klinowski, F. A. Almeida Paz, P. Silva, and J. Rocha, Dalton Trans., 2011, 40, 321.

232. S. Hermes, M.-K. Schr¨oter, R. Schmid, et al., Angew. Chem. Int. Ed., 2005, 44, 6237.

205. W.-J. Son, J. Kim, J. Kim, and W.-S. Ahn, Chem. Commun., 2008, 6336. 206. D. Zacher, O. Shekhah, C. W¨oll, and R. A. Fischer, Chem. Soc. Rev., 2009, 38, 1418. 207. A. Schoedel, C. Scherb, and T. Bein, Angew. Chem. Int. Ed., 2010, 49, 7225. 208. Y. Hu, X. Dong, J. Nan, et al., Chem. Commun., 2011, 47, 737. 209. R. Ostermann, J. Cravillon, C. Weidmann, et al., Chem. Comun., 2011, 47, 442. 210. G. F´erey, Chem. Soc. Rev., 2008, 37, 191. 211. R. J. Kuppler, D. J. Timmons, Q.-R. Fang, et al., Coord. Chem. Rev., 2009, 253, 3042. 212. C. Janiak and J. K. Vieth, New. J. Chem., 2010, 34, 2366. 213. A. U. Czaja, N. Trukhan, and U. M¨uller, Chem. Soc. Rev., 2009, 38, 1284. 214. H. M. El-Kaderi, J. R. Hunt, J. L. Mendoza-Cort´es, et al., Science, 2007, 316, 268. 215. M. Mastalerz, Angew. Chem. Int. Ed., 2008, 47, 445. 216. R. A. Smaldone, R. S. Forgan, H. Furukawa, et al., Angew. Chem. Int. Ed., 2010, 49, 8630. 217. N. L. Rosi, J. Eckert, M. Eddaoudi, et al., Science, 2003, 300, 1127. 218. H. Li, M. Eddaoudi, M. O’Keeffe, and O. M. Yaghi, Nature (London), 1999, 402, 276.

233. M. D. Allendorf, C. A. Bauer, R. K. Bhakta, R. J. T. Houk, Chem. Soc. Rev., 2009, 38, 1330.

and

234. O. R. Evans and W. B. Lin, Acc. Chem. Res., 2002, 35, 511. 235. R. Sessoli and A. K. Powell, Coord. Chem. Rev., 2009, 253, 2328. 236. T. Glaser, Chem. Commun, 2011, 47, 116. 237. M. Kurmoo, Chem. Soc. Rev., 2009, 38, 1353. 238. A. Hazra, P. Kanoo, and T. Kumar Maji, Chem. Commun., 2011, 47, 538. 239. W. Kaneko, M. Ohba, and S. Kitagawa, J. Am. Chem. Soc., 2007, 129, 13706. 240. W. Zhang, H.-Y. Ye, and R.-G. Xiong, Coord. Chem. Rev., 2009, 253, 2980. 241. R. C. Huxford, J. Della Rocca, and W. B. Lin, Curr. Opin. Chem. Biol., 2010, 14, 262. 242. A. C. McKinlay, R. E. Morris, P. Horcajada, et al., Angew. Chem. Int. Ed., 2010, 49, 6260. 243. J. Della Rocca and W. B. Lin, Eur. J. Inorg. Chem., 2010, 3725. 244. Z. Wang and S. M. Cohen, Chem. Soc. Rev., 2009, 38, 1315.

219. M. O’Keeffe, Chem. Soc. Rev., 2009, 38, 1215.

245. R. K. Deshpande, J. L. Minnaar, and S. G. Tefler, Angew. Chem. Int. Ed., 2010, 49, 4598.

220. S. S. Kaye, A. Dailly, O. M. Yaghi, and J. R. Long, J. Am. Chem. Soc., 2007, 129, 14176.

246. K. K. Tanabe, C. A. Allen, and S. M. Cohen, Angew. Chem. Int. Ed., 2010, 49, 9730.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

38

Supramolecular materials chemistry

FURTHER READING Reviews on general aspects of crystal engineering G. R. Desiraju, Angew. Chem. Int. Ed. Engl., 1995, 34, 2311. C. B. Aakeroy, Acta Crystallogr., 1997, B53, 569. G. R. Desiraju, Chem. Comm., 1997, 1475. B. Moulton and M. J. Zaworotko, Adv. Supramolecular Chem., 2000, 7, 235. D. Braga, Chem. Comm., 2003, 2751. G. R. Desiraju, J. Molec. Struct., 2003, 656, 5. D. Braga, L. Brammer, and N. R. Champness, CrystEngComm, 2005, 7, 1. C. B. Aaker¨oy, N. R. Champness, and C. Janiak, CrystEngComm, 2010, 12, 22. G. R. Desiraju, J. Chem. Sci., 2010, 122, 667. Books and monographs on crystal engineering D. D. MacNicol, F. Toda, and R. Bishop, eds., Solid-state supramolecular chemistry: crystal engineering, in Comprehensive Supramolecular Chemistry, Pergamon Press, Oxford, 1996, vol. 6. G. R. Desiraju, ed., Crystal Design: Structure and Function, Wiley, New York, 2003. E. R. T. Tiekink and J. J. Vittal, eds., Frontiers in Crystal Engineering, John Wiley & Sons, Ltd, Chichester, 2006.

D. Braga, F. Grepioni, and A. G. Orpen, eds., Crystal Engineering: From Molecules and Crystals to Materials: Proceedings of the NATO Advanced Study Institute, Held May 1999, in Erice, Italy (NATO Science Series, Ser. C), Kluwer Academic Publishers, Dordrecht, The Netherlands, 1999. Crystal engineering, in Transactions of the American Crystallographic Association, eds. R. D. Rogers and M. J. Zaworotko, American Crystallographic Association, Buffalo, NY, 1999, vol. 33. Books and themed journal issues on coordination polymers S. R. Batten, S. R. Neville, and D. R. Turner, eds., Coordination Polymers: Design, Analysis and Application, RSC Publishing, Cambridge, 2009. M.-C. Hong and L. C. John, eds., Design and Construction of Coordination Polymers, Wiley & Sons, Inc., Hoboken, NJ, 2009. L. R. MacGillivray, ed., Metal-Organic Frameworks: Design and Application, John Wiley & Sons, Inc., Hoboken, NJ, 2010. A. Chatterjee, ed., Structure Property Correlations for Nanoporous Materials, CRC Press, Boca Raton, FL, 2010. Functional Metal-Organic Frameworks. Gas storage, separation and catalysis, in Topics in Current Chemistry, ed. M. Schr¨oder, Springer Verlag Kg, 2010, vol. 293. J. R. Long and O. M. Yaghi, Metal-Organic Frameworks, Themed Issue of Chemical Society Reviews, Royal Society of Chemistry, 2009, pp. 1201–1508.

D. Braga and F. Grepioni, eds., Making Crystals by Design, Wiley-VCH Verlag GmbH & Co. KgaA, Weinheim, Germany, 2007.

B. K. Teo, ed., Functional Hybrid Nanomaterials—Design, Synthesis, Structure, Properties and Applications, Themed Issue of Coordination Chemistry Reviews, Elsevier Science B.V.l 2009, pp. 2785–3066.

E. R. T. Tiekink, J. J. Vittal, and M. J. Zaworotko, eds. Organic Crystal Engineering—Frontiers in Crystal Engineering, John Wiley & Sons, Ltd, Chichester, 2010.

K. Biradha, ed., Coordination Polymers: Structure and Function. New J. Chem., 2010, 34, 2355–2527.

K. R. Seddon and M. J. Zaworotko, eds., Crystal Engineering: The Design and Application of Functional Solids: Proceedings of the NATO Advanced Study Institute, September 1996, in Digby, Nova Scotia, Canada (NATO ASI Series, Ser. C), Kluwer Academic Publishers, Dordrecht, The Netherlands, 1999.

S. Kitagawa and S. Natarajan, eds., Targeted fabrication of MOFs for hybrid functionality. Eur. J. Inorg. Chem., 2010, 34, 3683–3874.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc107

Concepts and Nomenclature in Chemical Crystallography Leonard J. Barbour1 , Dinabandhu Das1 , Tia Jacobs1 , Gareth O. Lloyd2 , and Vincent J. Smith1 1 2

University of Stellenbosch, Stellenbosch, South Africa University of Cambridge, Cambridge, UK

1 Introduction 2 Scientific Terminology and the English Language 3 Why is Terminology Important? 4 Aim and Scope 5 Terminology 6 Conclusion Acknowledgments References

1 2 3 3 3 32 32 32

“When I use a word,” Humpty Dumpty said, in a rather scornful tone, “it means just what I choose it to mean— neither more nor less.” Through the Looking Glass by Lewis Carroll

1

INTRODUCTION

For several reasons, the past two decades have witnessed an explosion of reports involving crystal structure analysis—an undertaking that was once the domain of a limited number of specialists. User-friendly single-crystal X-ray

diffractometers equipped with CCD detectors, and capable of producing high-quality data very rapidly, have become commonplace in many academic institutions. Parallel developments have ensured that researchers now have easy access to personal computers that substantially surpass, in both speed and graphics capabilities, the mainframe machines that were once essential for crystal structure solution, refinement, and inspection. As a result, many powerful and user-friendly computer packages now exist that facilitate the analysis, presentation, and verification of crystal structures. Taken together, these advances in chemical crystallography have opened up the field of solid-state chemistry to researchers who do not regard themselves as specialist crystallographers. The advantage of this is that researchers can focus most of their attention on the preparation of new materials. An obvious disadvantage is that crystallography is still fraught with many pitfalls that can ensnare the unwary nonspecialist. Fortunately, most crystal structure analyses can be regarded as being relatively routine, even to the point that several software packages now carry out the solution and refinement of such structures automatically. Programs such as CHECKCIF1 can generally alert the crystallographer to potentially embarrassing pitfalls such as incorrect space group assignment, misassigned atoms, and unmodeled electron density. Indeed, the bar for the quality of crystal structure analyses has been raised significantly, with most journals now insisting on submission of a CHECKCIF report as supporting information, together with an explanation of potentially serious softwaregenerated alerts that may be indicative of possible errors.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

2

Supramolecular materials chemistry

Crystallography was also once mainly used to confirm molecular structure, but the focus has shifted decidedly to the investigation of structure–property relationships of materials. Such phenomena concern the supramolecular chemistry of crystalline materials, necessitating detailed analysis of intermolecular interactions, and packing arrangements. In this regard, the many concepts that relate to the structures and properties of crystals involve a number of terms that might be especially bewildering to the nonspecialist. Many of these terms have been borrowed and/or adapted from mineralogy (a much older branch of crystallography and involving a different community of researchers), while others have been coined at will to describe either new concepts, or concepts for which suitable terms did not exist before. As a result, crystallographic nomenclature is far from systematic, and the lexicon of modern crystallography has evolved as a haphazard collection of terms. To compound the problem, different conventions regarding spelling and hyphenation of certain words exist between American English and British English (generally, British journals tend to be more flexible, accepting either American or British spelling).

2

SCIENTIFIC TERMINOLOGY AND THE ENGLISH LANGUAGE

Although English is not the most spoken language on earth, it is de facto the most widely used medium of international communication in science and technology.2 It is not necessary to enumerate the specific reasons for this here, but it is useful to discuss some of the idiosyncrasies of English. The English language is in a continuous state of flux—although there are many rules, there are aspects of the language that are simply subject to conventions, preferences, and fashion. Furthermore, many English words are borrowed and adapted from other (older) languages and, as such, the language is not constructed systematically. In particular, when new “English” scientific terms are required, they are most often concocted as compounds of Greek and Latin affixes. For example, the term polymorph is a compound of two Greek words and translates to “many shape” in common vernacular. There are two important advantages of constructing a new “English” word from non-English roots: (i) we obtain one relatively compact word from several words and (ii) it is unlikely that the word will be confused for something else when it is used in a sentence (provided, of course, that we all agree on what it actually means). Ultimately, this provides a convenient way of avoiding confusion and awkwardness. Consider calling a polymorph a “many shape”—one can immediately envisage the significant

degree of awkwardness in generating the noun (both singular and plural) and adjectival forms equivalent to polymorph, polymorphs, polymorphism, and polymorphic. Although there are apparently no rules that dictate when to resort to either Greek or Latin, it is generally possible to discern the meaning of an unfamiliar word by knowing only a small number of commonly used affixes. The Greek and Latin affixes that have been used as the roots of many of the terms covered in this chapter are listed in Table 1.

Table 1 Greek and Latin affixes commonly used in the nomenclature of chemical crystallography. Affix a ab ad allo amphi anti apo aristo chrom com con co crypto (or krypto) desmo di enantio epi exo glomerate homeo in iso lithos mer meso meta mono morph para periodic poly pseudo quasi sorb tactic, taxis tecto tetra topos tri trope typic zeo

Meaning not, without, lacking away from, down, from, off to, toward different around, at both sides, both against, opposite (1) from, away from, (2) lacking, (3) (chem) derived from or related to best color together together, with with, together hidden, secret adhesion, ligament, bond two opposite above outside, outer part tightly clustered, compacted, collective similar not, lacking, opposite of same stone (can also be used for a crystal ) part intermediate after, over, between one shape or form beside regular intervals many, much false resembling suck pertaining to arrangement or pattern enclosure four place three (from tropos)—turning, change type to boil

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

3

WHY IS TERMINOLOGY IMPORTANT?

Scientists gather information systematically in order to assemble and classify the various phenomena that govern nature. They then communicate their findings to others in a precise and succinct manner. It is self-evident that classification involves terminology and, while a picture may paint a thousand words, a well-coined term can also compress a great deal of information into a small amount of space. In order to be useful, a term should be needed and it should be possible for someone to intuitively gather from its construction what it means. It should also be widely accepted as the unambiguous term of choice for a particular concept. When new terms are coined, it is almost inevitable that their usage will garner mixed receptions—as a case in point, some newly coined terms have sparked much controversy and even heated debate within the chemical crystallography community. In the context of chemical crystallography, organizations such as the International Union of Pure and Applied Chemistry (IUPAC) and the International Union of Crystallography (IUCr) can play a vital role by standardizing definitions. Indeed, many of the terms discussed in this chapter have been defined by the IUPAC and/or the IUCr. However, some of these definitions appear to be outdated, and others have been defined rather vaguely. For example, the IUPAC Compendium on Chemical Terminology (the so-called Gold Book)3 is generally biased toward covalent and coordination chemistry, and this is understandable for historical reasons. However, the field of supramolecular chemistry now encompasses parallel concepts that involve much weaker interactions, but many definitions have not yet been updated despite common usage of well-established terms in a broader context than before. In 1948, the IUCr set up a Commission on Crystallographic Nomenclature in order “to advise on matters of scientific nomenclature as they pertained to the field of crystallography.”4 However, since 1978, nomenclature issues “arising in specific fields of crystallography that come to the attention of, and are recognized as important by, the Commission would be studied by ad hoc committees of experts appointed by the Commission.”4 To date, several commissions have been created under this new system in order to address specific subfields (e.g., aperiodic crystals, inorganic and mineral structures, etc.), but there has been no recent attempt to convene a commission on general nomenclature. It is our view that such a commission would add significant value to the field of chemical crystallography.

4

3

AIM AND SCOPE

The aim of this contribution is to present a survey of the concepts that arise during the crystallographic studies of crystals —with the rather ambitious goal of achieving disambiguation regarding nomenclature. We discuss the terms that we regard as being highly relevant to modern chemical crystallography, the origins of the terms, disagreements over the use of some terms, and even provide examples (several of which have been taken from our own work) where necessary. Although much of the information in this chapter can be found elsewhere, we feel that it serves a useful purpose to compile a relatively comprehensive compendium on terminology into a single source. This chapter is not intended to cover the fundamental nomenclature of crystallography: it is assumed that the reader is already familiar with concepts such as the unit cell, point groups, space groups, site symmetry, special positions, general positions, and symmetry elements. Many of the concepts in chemical crystallography overlap significantly with the field of supramolecular chemistry. We have therefore used our discretion in deciding which supramolecular terms are relevant to the solid state. In general, we have not attempted to review all the work relating to specific terms; instead we have tried to confine our discussions to issues that we feel are relevant to nomenclature. It is inevitable in such discussions that terms would be used that are defined elsewhere in the text—in such cases, we have italicized the terms. We have not included the nomenclature of graph sets since this volume already contains a comprehensive chapter on the field (see Network and Graph Set Analysis, Supramolecular Materials Chemistry).

5 5.1

TERMINOLOGY Adduct

According to the IUPAC Gold Book, an adduct is a “new chemical species AB, each molecular entity of which is formed by direct combination of two separate molecular entities A and B in such a way that there is a change in connectivity, but no loss, of atoms within the moieties A and B.” Originally, the term was most likely intended to describe intramolecular concepts in covalent chemistry such as addition reactions and the combination of Lewis acid–base pairs (excluding metal–ligand coordination, for which the Gold Book suggests the use of coordination entity rather than adduct or complex ). Nowadays, we also interpret connectivity to encompass the weaker supramolecular interactions (e.g., hydrogen bonding, dipole–dipole, π –π, cation–π , etc.), and the use of the term “adduct” can thus be broadened to include intermolecular association

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

4

Supramolecular materials chemistry we discourage its use. Indeed, to avoid confusion, the two separate meanings can be adequately represented with the words mixture and component, respectively.

5.3

Figure 1 Formation of an interstitial solvent-accessible cavity as a result of columnar packing of metallocycles. The cavity is occupied by a binary 1 : 1 adduct consisting of a methanol molecule hydrogen bonded to a water molecule. The adduct can also be referred to as a monoadduct or a heterodimer.5

of entities (e.g., Figure 1) without the need to alter its IUPAC definition. However, it is a matter of personal interpretation whether or not the nondirectional van der Waals forces should also qualify as “connectivity,” and several examples exist where this is indeed taken to be the case. Our recommendation is to consider only directional interactions, otherwise the term “adduct” would become so general as to be meaningless since every multicomponent molecular crystal would then be composed of adducts. Binary, ternary, and quaternary adducts refer to the association of two, three, or four different entities, respectively. A monoadduct is a 1 : 1 association of two different entities and a bisadduct is a 2 : 1 combination. However, there does not appear to be a systematic method of referring to higher-order adducts (e.g., a 1 : 2 : 3 association of three different entities A, B, and C). The term “ternary adduct” does not adequately reflect the relative ratios of A, B, and C and any combination of mono-, bis-, and tris- is awkward at best. Indeed, such an adduct could just be referred to as “a 1 : 2 : 3 adduct of A, B, and C.” Adducts can also be referred to as dimers, trimers, tetramers, and so on (i.e., n-mers). However, an adduct usually refers to an association in which at least two entities are different, whereas n-mers may consist of either the same or different entities.

Allotropism or allotropy (allotropia (Greek): variation) is the phenomenon whereby an elemental substance can exist in more than one form within a particular state—that is, a special case of isomerism. Oxygen displays allotropy in the gas phase (O2 and O3 ), and well-known examples of solid-state allotropy include carbon (diamond, graphite, and fullerenes), phosphorus (white, red, violet, black, and diphosphorus), and sulfur (e.g., S2 , S4 , cyclo-S8 , and many more).6 The term can only be applied to pure elements and not to compounds; the different forms are referred to as being allotropes of one another, and they involve different bonding arrangements. It has been suggested that the term allotropy should be replaced by polymorphism.7 Indeed, strong parallels can be drawn between allotropy and polymorphism —for example, both concepts involve monotropism and enantiotropism (see Section 5.53). However, allotropes are not necessarily polymorphs (diamond and graphite are not polymorphs of each other because they are different compounds), and the IUPAC recommendation is therefore that allotropy should be retained.3 The difference and the similarity between allotropy and polymorphism can be illustrated by considering sulfur. The rhombic and monoclinic crystalline forms both consist of puckered S8 rings, and these two modifications can interconvert by heating and cooling. It is tempting to call this relationship crystal allotropy, but the correct term is polymorphism because both structures involve the same compound (i.e., atomic connectivity). When heated to above 160 ◦ C, the S8 rings open by means of a free radical reaction to form polymeric chains. In contrast to crystal allotropy, the relationship between the polymeric chains and the S8 rings can be termed chemical allotropy. However, since polymorphism is the preferred term for crystal allotropy, chemical allotropy can be shortened to allotropy.

5.4 5.2

Admixture

Admixture is an outdated term and can refer either to the product of mixing two or more substances or to one of the components of the mixture. Owing to these two different interpretations, the term can be ambiguous and

Allotrope

Amorphous

An amorphous solid is a material with sufficient structural rigidity to resist changes in shape and volume, but with insufficient periodicity of its constituent atoms, molecules, or ions to be considered crystalline. Examples of amorphous solids include wax, glass, gels, thin films, and nanostructured materials. A material may undergo a solid–solid

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography phase transition between an ordered crystalline phase and a disordered amorphous phase —a process called metamictization. In such cases, the amorphous phase is considered to be a polymorph of the crystalline phase. Highly crystalline solids tend to yield multiple sharp peaks in X-ray or neutron powder diffractograms, while an amorphous solid typically produces a broad peak that is characteristic of the radial distribution of the molecules—the more random the distribution, the broader is the peak. Highly crystalline and highly amorphous phases are at opposite ends of a continuum in terms of the degree of long-range order, and the exact boundary between crystalline and amorphous is therefore a vague concept. Since the bulk properties of a material generally depend on its internal structure (or phase), it is tempting, and potentially very useful, to quantify the degree of internal order using a suitable term. However, some caution is advised against using degree of crystallinity for this purpose because it is already an accepted term, mainly amongst polymer scientists and geologists, who use it to describe a material that consists of a mixture of amorphous and crystalline phases. The expression specifies the amount of material, as a percentage volume, that is crystalline. Degree of amorphousness can be used to represent the inverse fraction. We do not recommend using degree of crystallinity or degree of amorphousness to qualitatively describe the average level of ordering of a predominantly amorphous material, or to quantify the mosaicity (see Section 5.47) of a single crystal (see Section 5.20).

5.5

Apohost

In the context of host–guest chemistry, an apohost is a guest-free solid phase of a host. For example, p-tertbutylcalix[4]arene is a well-known host compound that forms solid-state supramolecular complexes with a wide variety of guests.8 A guest-free (apohost) phase can be crystallized by sublimation under vacuum (Figure 2),9 but a different apohost phase can be grown by slow cooling of a solution of the host in warm tetradecane.10 This example demonstrates that an apohost can be polymorphic, and the cautious writer is well advised to refer to an apohost phase rather than the apohost phase.

(a)

(b)

Figure 2 (a) Offset face-to-face arrangement of two adjacent molecules of p-tert-butylcalix[4]arene in a crystal grown by sublimation. (b) The molecules pack as bilayers and the yellow surfaces represent isolated unoccupied voids within the crystal structure (Reproduced from Ref. 8.  Royal Society of Chemistry, 2005.)

coordination octahedra of the Co and Re atoms assume an arrangement similar to that of the octahedron in the rutile structure type, which crystallizes in P 42 /mmm. In this particular case, Cmmm is a maximal isomorphic subgroup of P 42 /mmm and the structural resemblance is facilitated by the symmetry relationship between the two space groups. However, a formal symmetry relationship such as this does not appear to be a requirement for aristotype–hettotype relationships.

5.7

Aristotype and hettotype

According to the IUCr on-line dictionary, “an aristotype is a high-symmetry structure type that can be viewed as an idealized version of a lower-symmetry structure. The lowersymmetry structure is called a hettotype.” For example, CoReO4 crystallizes in the space group Cmmm, and the

Awkwardness

The awkwardness of a molecule can be simply expressed as the ratio of its surface area to its volume.11 This parameter is easily calculated from atomic coordinates and van der Waals radii and has been invoked in studies aimed at establishing trends in molecular packing.12

5.8 5.6

5

Bifurcated hydrogen bond

The term bifurcation is used to describe hydrogen bonds between a single donor and two acceptors, or two donors and a single acceptor. When one hydrogen atom hydrogen bonds to two acceptors simultaneously, we refer to it as a bifurcated donor (Figure 3a). When two hydrogen atoms hydrogen bond to a single acceptor, we refer to it as a bifurcated acceptor (Figure 3b).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

6

Supramolecular materials chemistry

(a)

(b)

Figure 3 Hydrogen bonding involving (a) a bifurcated donor and (b) a bifurcated acceptor.

5.9

Bilayer

A bilayer consists of two layers of molecules that are closely packed but loosely associated with each other. Although this is a well-established concept in molecular biology (e.g., lipid bilayer membranes), analogous arrangements of molecules can be identified in some crystal structures. There is no formal definition of a bilayer in the context of chemical crystallography, but it seems reasonable that the following two criteria should be satisfied: (i) the entities of which the two layers are composed should be the same and (ii) successive bilayers should be separated from one another such that individual bilayers can be distinguished. This concept is exemplified by the packing of p-tert-butylcalix[4]arene (see Figure 2b).

5.10

Birefringence and pleochromism

Birefringence (double refraction) occurs when an anisotropic material causes linearly polarized light to experience two different refractive indices, each according to the direction of polarization relative to the optical axis. Light polarized perpendicular to the optical axis is subject to the ordinary index of refraction no , while light polarized parallel to the axis is subject to the extraordinary index of refraction ne . Therefore, unpolarized light rays passing through a birefringent crystal will be split into two linearly polarized components that are displaced relative to each other (Figure 4). This effect is related to pleochromism. When an anisotropic material has different absorption coefficients for light polarized in different directions, the phenomenon is known by the general term pleochroism (i.e., having many colors). The effect is usually strongly dependent on the wavelength of the light, and when a pleochroic crystal is placed between two crossed polarizers, it appears to be colored. The color depends on the thickness of the crystal, its internal symmetry, and its orientation relative to the direction of polarization. Dichroic crystals exhibit two distinct colors and are usually tetragonal, trigonal, or hexagonal. Trichroic crystals are usually

Figure 4 Birefringence in a calcite cleavage rhomb from the de Kock mineral collection, held at the University of Stellenbosch.

triclinic, monoclinic, or orthorhombic, while cubic crystals do not exhibit pleochroism. If a material possesses different absorption coefficients for left- and right-handed circularly polarized light, the phenomenon is known as circular dichroism. Similarly, linear dichroism occurs for orthogonally opposed plane polarized light.

5.11

Centrosymmetric and noncentrosymmetric structures

The concepts and nomenclature related to chiral and achiral crystal structures have been elegantly dealt with by Flack,13 and this section is almost entirely a partial summary of that article. Flack notes that the crystallographer may be confronted by three types of chiral entities: (i) a molecule, (ii) a crystal structure, and (iii) a space group. The potential for misunderstanding can be minimized by adhering to clear terminology and using the term “chiral” correctly. Whether or not an object is chiral can be deduced from its symmetry group; a chiral entity only obeys symmetry operations of the first kind (i.e., rotations and translations) and not operations of the second kind (i.e., rotoinversions, which are combinations of a rotation with a center of symmetry). An achiral entity obeys symmetry operations of both kinds in equal numbers. Crystallographically relevant rotoinversions include 1, 2, 3, 4, and 6 axes. A 1 rotoinversion is the same as an inversion center, while a 2 rotoinversion is equivalent to a mirror plane. In order for an object to be chiral, there must exist, at least in principle, another object where the relationship between the two entities is such that they are mirror images of each other, and not directly superimposable. The two objects are then said to be enantiomorphous, their relationship to each other is enantiomorphic, and each is the enantiomorph of the other. Although this concept is well established for molecules, Flack notes that the expression “chiral space group” has often been incorrectly used for the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

Figure 5

7

Perspective view along [001] of the packing of 2-chloroaniline.

“space group of a chiral structure”; a chiral crystal structure is not the same as a chiral space group. There are 11 space groups that have enantiomorphs (e.g., the relationship between P 61 and P 65 is enantiomorphic) and these 22 space groups can correctly be called chiral. However, there are a total of 65 space groups that contain only symmetry operations of the first kind and that are therefore compatible with chiral crystal structures; these are termed the Sohncke space groups and they include the 22 chiral space groups. For example, note that the Sohncke space group P 21 is its own enantiomorph and it is therefore not a chiral space group, although it describes a chiral crystal structure. Note also that a chiral crystal structure does not have to contain chiral molecules—that is, it is possible for achiral molecules to pack in a chiral fashion (e.g., 2-chloroaniline14 is achiral, but packs in a chiral manner and also crystallizes in the chiral space group P 31 —see Figure 5). All centrosymmetric and some noncentrosymmetric crystals have their origins fixed in space by the symmetry elements of the space group. When this is not the case, at least one of the three crystallographic directions are said to be polar and the coordinates in the polar direction(s) then need to be fixed. Examples of polar space groups are P 1 (x, y, and z must be fixed), Pm (x and z must be fixed), and P 21 , (y must be fixed). Whether a space group is centrosymmetric or noncentrosymmetric can be deduced from its point group. Similarly, the point group of a noncentrosymmetric space group determines whether it is polar, a member of the Sohncke space groups, or both. These relationships are given in Table 2. Note that space groups with 4 and 6 symmetry elements are noncentrosymmetric, do not belong to the Sohncke set of space groups, and they are also nonpolar. Owing to the types of anisotropy that they possess, polar and chiral crystals generally exhibit interesting properties

such as piezo-, pyro-, and ferroelectric effects, optical activity, second harmonic generation, birefringence, and so on.

5.12

Chromism

Chromism refers to the ability of a material to change color as a result of heat (thermochromism), light (photochromism), solvation (solvatochromism), sorption of a vapor (vapochromism), mechanical action (mechanochromism), pressure (piezochromism), magnetic field (magnetochromism), time (chronochromism), aggregation (aggregachromism), and phase transitions or topotaxis (crystallochromism). Such color changes may or may not be reversible. Triboluminescence is the generation of light due to mechanical action, and chromoisomerism refers to polymorphs that are different in color.

5.13

Close packing

With regard to the three-dimensional arrangement of spheres, close packing refers specifically to the two arrangements (cubic and hexagonal close packed) that achieve the maximum packing efficiency of 74% (or a packing coefficient of 0.74). The packing efficiency is the volume fraction, expressed as a percentage, of the crystal that is occupied by atoms. In the context of molecular crystals, close packing simply refers to the natural tendency of molecules to pack efficiently by optimizing their mutual orientations, given their shapes and the possibilities for energetically favorable intermolecular interactions.15 The term is normally regarded as being merely descriptive, and there is no formal requirement in terms of the magnitude of the packing efficiency. Organic molecules generally pack with efficiencies that range between 60 and 65%. For a given molecule, there is, of course, a strong correlation between close packing and density.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

8

Supramolecular materials chemistry Table 2

5.14

The 32 crystallographic point groups and their anisotropy characteristics.

Point group number

Hermann–Mauguin symbol

1

1

2

1

3

2

4

m

5

2/m

6

222

7

mm2

8

mmm

9

4

10

4

11

4/m

12

422

13

4mm

14

42m or 4m2

15

4/mmm

16

3

17

3

18

32 or 321 or 312

19

3m or 3m1 or 31m

20

3m or 3m1 or 31m

21

6

22

6

23

6/m

24

622

25

6mm

26

6m2 or 62m

27

6/mmm

28

23

29

m3

30

432

31

43m

32

m3m

Crystal system Triclinic

Centrosymmetric

Polar

Sohncke group

×

×

×

×

× ×

Monoclinic ×

× ×

Orthorhombic

Cluster

Depending on the context, a cluster can either refer to a chemical entity or to a specific crystal habit that describes an aggregated collection of crystals. The latter context is elaborated in Section 5.30, and the former can be exemplified by a metal cluster, which has been defined as “a group of two or more metal atoms

× ×

×

× ×

Tetragonal × × ×

×

× ×

Trigonal × × ×

×

× ×

Hexagonal × ×

× × ×

Cubic ×

in which there are substantial and direct bonds between the metal atoms” (e.g., Figure 6).16 Although the use of cluster to refer exclusively to compounds containing metal–metal bonds was advocated by Cotton, the term has also been applied to polyhedral compounds based on main group elements (e.g., fullerenes and boranes).17 Any further discussion of the term cluster in the context of a type of coordination entity (e.g., “heteronuclear”

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

9

chagrin. However, in this regard, it is not clear whether a crystal composed of both neutral and charged species should be called a salt or a cocrystal (but we suspect that using salt in such cases would provoke less condemnation). Several groups have correctly noted that there is no universally accepted definition of cocrystal and then proceed to circumspectly state merely what they understand it to be. In an attempt to exclude solvates (and hydrates), Zaworotko and coworkers20 have stated that they “and others have operated under the assumption that a cocrystal is a multiple component crystal in which all components when pure are solid under ambient conditions.” Aaker¨oy and Salmon21 delineated the following three criteria to which they subscribe: Figure 6

Metal cluster of Os5 (CO)16 .16

and “naked ” clusters) is beyond the scope of this review. Owing to their remarkable ability to both donate and accept hydrogen bonds, water molecules often form water aggregates in hydrated crystals —these aggregates are generally termed water clusters.18, 19 Interestingly, and perhaps because of the relevance of water to biological processes, this terminology is usually not applied to other kinds of solvent molecules.

5.15

Cocrystal

The term cocrystal has probably elicited more heated debate within the chemical crystallography community than any other. When two or more chemically distinct species find it more favorable to crystallize together (i.e., as part of the same crystal structure) rather than separately, they form what can generally be described as a multicomponent crystal. Depending on the nature of the components, there are various more precise terms that can be used. For example, if one of the components is a solvent, the crystal can be called a solvate. If the solvent is water, then the crystal is more specifically a hydrate. If necessary, one can be similarly specific about other solvents (e.g., a methanolate when the solvent is methanol). If the components are ionic, then the crystal is a salt. If the components are enantiomers (or quasienantiomers) of one another, then we have a racemate (or quasiracemate). If the components are present in nonstoichiometric proportions, the material is a mixed crystal or a solid solution. In general, it seems that, if none of the above apply, the term cocrystal is most appropriate—but this is a rather fragile description of a cocrystal because it is based on excluding certain types of chemical species. Indeed, most of the discord stems from disagreement about what should be excluded, although disqualifying salts apparently attracts the least

1.

Only compounds constructed from discrete neutral molecular species are considered. Therefore, all solids containing ions, including complex transition metal ions, are excluded. 2. The components should be solids at ambient conditions. Therefore, all solvates are excluded. 3. The crystal must be a structurally homogeneous crystalline material containing two or more neutral building blocks in well-defined stoichiometric amounts. In principle, we agree with the in-house working definitions given by Zaworotko and Aaker¨oy, especially with regard to excluding solvates (we note that these researchers have not suggested that their definitions should be adopted universally but we nevertheless use their specifications as the basis for further discussion). However, if one were to propose a definition that disqualifies solvates, simply requiring cocrystal components to be “solids at ambient conditions” is perhaps not the best approach. First, “ambient conditions” differ in different laboratories, and this is even the case within the same laboratory (i.e., even if the temperature is kept constant, the pressure will vary from day to day). If we were to take “ambient conditions” to mean standard temperature and pressure (STP), these conditions might still seem rather arbitrary in the context of materials science. For example, the 1 : 1 multicomponent crystal consisting of pyrogallol (melting point = 131–134 ◦ C) and pyrimidine (melting point = 20–22 ◦ C) was isolated from ethanol22 as the common solvent for the two components. Depending on the “ambient” conditions, pyrimidine may well be a liquid but it was not the solvent from which the crystals were grown. It is therefore relatively easy to argue that the crystal cannot be a solvate (i.e., ethanol is not included in the structure). However, adhering to a set of rules that excludes components that are liquids below ambient conditions precludes the material from being termed a cocrystal, whereas a crystal consisting of pyrogallol and pyrazine (melting point = 52 ◦ C) would

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

10

Supramolecular materials chemistry

qualify as a cocrystal even though pyrazine is structurally and chemically similar to pyrimidine. Although we are reluctant to prescribe a definition of a cocrystal, we suggest a refinement to point 2 of Aaker¨oy’s working definition, that is, that the “ambient conditions” criterion be dropped in favor of stating that no components should have played the role of solvent during the crystallization process. Of course, this will not placate those who advocate that solvates may be called cocrystals. Furthermore, our refinement would preclude crystals that have been subjected to solvent exchange as a single-crystalto-single-crystal (SCTSC) process (thus, a solvate would become a cocrystal without having significantly altered its composition in principle), as well as a material that has absorbed vapor. Desiraju has criticized the word cocrystal, asking “what is co- to what?,” to which Dunitz has replied: “One of the uses of the prefix co- is to indicate togetherness, and that is its function in co-crystal. We must all agree that the hyphen is essential: co-crystal, but under no circumstances cocrystal.” Hyphenation of the term cocrystal is sensibly dealt with in Bond’s paper where he states, “To hyphenate or not to hyphenate seems to be largely an area of disagreement between British and American English. This issue is of little scientific consequence, since ‘co-crystal ’ (British preference) and ‘cocrystal ’ (American preference) are clearly identifiable as equivalent terms.”23 We have reported a crystal structure that arguably elevates the concept of the cocrystal to a higher level of complexity (Figure 7).24 The structure contains two different columnar supramolecular motifs that have both

Figure 7 Projection of a hybrid structure showing the relative distribution of two different columnar motifs (black and gray) that have both been observed to exist exclusively in separate crystal structures. The columnar motifs are composed of molecules of Dianin’s compound—to reduce clutter, we have only shown the phenolic moieties.24 (Reproduced from Ref. 24.  American Chemical Society, 2009.)

been shown to occur exclusively in different crystals. Therefore, what we observed could be described as the cocrystallization of different extended motifs rather than merely different molecules. In our report, we did not propose a name for this phenomenon, but instead noted that it illustrates a parallel concept to that of cocrystals. While it may be tempting to coin a suitable term (e.g., supracocrystals, ultracocrystals, hypercocrystals, etc.) for such interesting but rare occurrences, it is sometimes sufficient to just point out that they are analogous to known phenomena rather than to muddy the waters with unwarranted terminology, especially when it is not that useful. Finally, racemates, quasiracemates, and solid solutions should not be called cocrystals because these terms are already more informative.

5.16

Complex (in the supramolecular sense)

The IUPAC Gold Book defines a complex as “a molecular entity formed by loose association involving two or more component molecular entities (ionic or uncharged) or the corresponding chemical species. The bonding between the components is normally weaker than in a covalent bond.” Although not recommended by the IUPAC, coordination compounds are widely called complexes and, being thus indelibly entrenched in the lexicon of inorganic chemistry, this usage will undoubtedly persist.3 Owing to the phrase “loose association” in its IUPAC definition, the term complex is a natural and valuable descriptor in the context of supramolecular chemistry because it allows one to refer to a multimolecular collective as a single entity.3 The choice of the components that constitute the collective may even be highly subjective, such as the van der Waals association between calix[6]arene and two C60 or C70 molecules, which can be referred to as a complex for lack of a better term (Figure 8).25 In many cases, the rather general term complex can be replaced by

Figure 8 A supramolecular complex between calix[6]arene and two C60 molecules.25

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography a more specific term. For example, encapsulation of xenon within a functionalized (±)-cryptophane-11126 yields an adduct that can correctly be referred to as a complex, but in this case the more precise term hemicarceplex could also be used.

5.17

Composite, agglomerate, and aggregate

These three terms are related to one another and are often encountered in the context of either synthetic or naturally occurring materials. A composite is a synthetic or naturally occurring material consisting of two or more components. There appears to be little distinction between a composite and an aggregate except that composite seems to be preferred when the material is useful. For example, engineered composites exploit the different properties of the individual components in order to benefit from the ensemble; an active pharmaceutical ingredient not amenable to tableting in its pure form might be combined with a pharmacologically inert excipient to yield a stable and cohesive composite. To aggregate means to gather objects together and, in noun form, an aggregate might refer to a collection of primary particles cemented together to yield larger entities. There appears to be little difference between an aggregate, a conglomerate, and an agglomerate according to the Merriam Webster Dictionary, which states that these three words can be used either as verbs or as nouns and that they are synonyms. In mineralogy, conglomerate and agglomerate both refer to a mass of course-grained (gravel-sized) particles cemented into a finer-grained matrix (the matrix needs not have the same composition as the particles).27 The distinction between the two terms is that a conglomerate contains rounded particles, while an agglomerate (or breccia) contains angular particles. This distinction is of limited use in chemical crystallography, where conglomerate has a different and very specific meaning (see Section 5.18).

but it could also be derived from simply mixing crystals of enantiomers in equal quantities. The most famous example of spontaneous resolution is the crystallization of racemic ammonium sodium tartaric acid, first described by Pasteur in 1848.29 Instead of crystallizing together, the two enantiomers crystallize separately to yield a mixture of enantiomerically pure crystals. Therefore, dissolution of one crystal selected from a conglomerate produces an optically active solution. Despite being a misnomer, conglomerate has long been part of the lexicon of chemical crystallography and, in terms of its accepted definition within the context of structural solid-state chemistry, it serves a useful purpose, is unambiguous, and should be retained. Moreover, the terms conglomerate and racemic conglomerate appear to be synonymous, and conglomerate should therefore not be used to describe any mechanical mixture of crystals. Using a simple but informative scheme (Figure 9), Bishop and Scudder outlined four possible outcomes of allowing a 1 : 1 solution of a racemic mixture to crystallize.30 Scenario iv results in a mechanical mixture of enantiomorphous racemic crystals in which the enantiomers are not related to each other by crystallographic symmetry. Therefore, even though each crystal contains an equal mixture of enantiomers (i.e., dissolution of one crystal does not produce an optically active solution), the molecules must necessarily pack in accordance with one of the 65 Sohncke space groups. Bishop and Scudder called the mechanical mixture of enantiomorphous racemic crystals a false conglomerate, but in a subsequent article,31 they pointed out that the term kryptoracemate (i.e., hidden racemate,

A A*

A (i)

(iii)

5.18

11

A*

A

A*

(ii)

A A* (iv)

Conglomerate and mechanical mixture

In the context of the crystalline solid state, a mechanical mixture (or physical mixture) is a mixture of crystals or crystallites of different phases (i.e., constituting different crystal structures) that are separated by grain boundaries. From its construction, the word conglomerate suggests a bulk material composed of several entities that are tightly bound together. However, in chemical crystallography, the term has been defined as “a mechanical mixture of crystals of pure enantiomers,” even though this does not make linguistic sense.28 A conglomerate may result from spontaneous resolution of a racemate during crystallization,

A

A*

A B*

A* B

Figure 9 Four possible outcomes of crystallizing a racemic solution consisting of a 1 : 1 mixture of A and its enantiomer A∗ . (i) Formation of a homogeneous collection of racemic crystals. (ii) Formation of inversion twins that contain domains of both enantiopure A and enantiopure A∗ . (iii) Formation of a conglomerate. (iv) Formation of a false conglomerate consisting of a mixture of kryptoracemates (i.e., a mixture of crystals of A and B∗ together with crystals of B and A∗ , where B = A and B∗ = A∗ , but there is no crystallographic symmetry relationship between A and B∗ or between B and A∗ ).30 (Reproduced from Ref. 30.  American Chemical Society, 2009.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

12

Supramolecular materials chemistry

referring to the hidden racemic nature of the crystals)32, 33 had already been associated with this phenomenon. Indeed, Bernal34 had previously defined the phenomenon of kryptoracemic crystallization as “the deposition, from a racemic solution, of enantiomorphic crystals containing racemic pairs of molecules as the asymmetric unit. For such a crystallization mode to occur, the number of molecules in the unit cell, Z, must be an even multiple of the number of general positions of the space group and, while some atoms may share a noncrystallographic pseudoinversion center, the degree to which they conform to such an inversion center varies for different sets of atoms.” We do not see any conflict between the terms kryptoracemate and false conglomerate because the former refers to a crystal and the latter to a mechanical mixture of crystals. Therefore, a false conglomerate is a mixture of kryptoracemates. Figure 5 shows the packing of 2-chloroaniline in a space group P 31 , which, for the achiral molecules, is energetically identical to the packing arrangement in the enantiomorphous space group P 32 . If we assume that there are no external factors that favor one of these enantiomorphous space groups over the other (e.g., chiral nucleation sites), the crystallization dish should contain approximately equal quantities of crystals in P 31 and P 32 . It is tempting to call such a mixture of enantiomorphous crystals a conglomerate. However, dissolution of one crystal selected at random from the mixture will not produce an optically active solution. Therefore, the mechanical mixture of P 31 and P 32 crystals should not be called a conglomerate in this case. However, we suggest that the term packing conglomerate be used to distinguish a mixture of enantiomorphous crystals composed of achiral molecules (i.e., where the packing arrangement is chiral) from a (chemical) conglomerate.

5.19

Cooperativity

The term cooperativity has been used in a number of different contexts. In enzyme chemistry, cooperativity refers to the change in affinity for a substrate by a binding site in response to binding at another site. In this context, the term has limited relevance to chemical crystallography. The IUPAC Gold Book defines a cooperative transition as one “that involves a simultaneous, collective displacement or change of state of the atoms and/or electrons in the entire system. Examples include an order–disorder transition of atoms or electrons, as in an alloy, a ferromagnet or superconductor; a Jahn–Teller or ferroic transition; a martensitic transition.”3, 35 The steel alloy Martensite undergoes a gradual and coordinated rearrangement of its crystal structure as a phase interface advances through the solid. Classical examples of martensitic transitions include the transition of carbon-doped iron from a face-centered cubic to

H

H

H

H

H O

H O

H O

H O

c

a

b

d

Figure 10 Cooperativity in a hydrogen-bonded water chain. (Adapted from Ref. 36.  Royal Society of Chemistry, 1957).

a body-centered tetragonal structure, and the transition of tetragonal ZrO2 to monoclinic ZrO2 . The diffusion of vinyl bromide into one of the apohost polymorphs of p-tertbutylcalix[4]arene9 is reminiscent of a martensitic transition—the process occurs as a single-crystal to single-crystal (SCTSC) phenomenon, implying a cooperative mechanism involving concerted movement of molecules as adjacent ˚ relative host bilayers slide laterally by approximately 6 A to one another in a monoclinic to tetragonal transformation. Hydrogen bond cooperativity was first suggested in 1957 by Frank and Wen.36 Using a water chain as an example (Figure 10), they proposed that the formation of the hydrogen bond a· · ·b causes a to become more acidic and b more basic (i.e., relative to unbound water), thus increasing the likelihood of hydrogen bonds a· · ·c and b· · ·d.

5.20

Crystal

This section discusses the term crystal and several other closely related terms or concepts. The word crystal is derived from the Greek krystallos, meaning “clear ice” and it was originally used to describe transparent quartz, which was assumed to be a type of ice formed by extreme cold. It may seem that asking the question “what is a crystal?” is superfluous. A crystal is generally understood to consist of a regular, three-dimensional arrangement (usually also referred to as long-range orientational order) of atoms, molecules, or ions. Owing to this internal periodicity, crystals tend to diffract electrons, X-rays, or neutrons to produce Bragg peaks in the Fourier spectrum. Taken together, these properties appear to distinguish crystals quite clearly from other substances, but the issue is more complicated than that. Despite defining many concepts related to the crystalline state, the IUPAC does not explicitly define a crystal, although various IUCr documents hint at such a definition. For example, the 2006 edition of the International Tables for Crystallography (Volume A, Section 8.1.4) states that crystals “. . . are finite real objects in physical space that may be idealized by infinite three-dimensional periodic ‘crystal structures’ in point space.” However, the relatively recent discovery of aperiodic crystals prompted the formation of an IUCr Commission to promote and encourage “. . . both experimental and theoretical research on aperiodic crystals,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography including quasicrystals, modulated crystals, composite crystals, and polytypes.”37 In its terms of reference, the commission’s report begins with “In the following, by crystal we mean any solid having an essentially discrete diffraction diagram, and by aperiodic crystal we mean any crystal in which three-dimensional lattice periodicity can be considered to be absent. As an extension, the latter term will also include those crystals in which three-dimensional periodicity is too weak to describe significant correlations in the atomic configuration, but which can be properly described by crystallographic methods developed for actual aperiodic crystals.” This statement redefines a crystal in terms of its diffraction pattern, while maintaining a real-space description of aperiodic crystals. From the above, it seems that crystals can be either periodic or aperiodic and that the conventional notion of a crystal pertains to the former variety. That is, the term crystal generally refers to a periodic crystal, unless otherwise stated (indeed, aperiodic crystals constitute a very small proportion of all crystals and their study is a far more specialized field). The interested reader is encouraged to peruse a series of short discussions on the topic “what is a crystal ?” in Zeitschrift f¨ur Kristallografie.38–45 1.

2.

Structure. A crystal structure is not a tangible entity, but rather the arrangement of atoms, molecules, or ions in a crystal. The structure is generally regarded as being conceptually infinite. Lattice. In crystallography, we encounter the term lattice in the context of both real space and reciprocal space. The real-space lattice (usually just referred to as the crystal lattice) is a conceptually infinite array of points that describes the positional and orientational

3. 4.

5.

6.

7.

arrangement of identical atoms, molecules, or ions in three-dimensional space. The term crystal lattice is specific and should not be used to mean crystal structure. A three-dimensional real-space lattice can be uniquely specified using up to six lattice parameters (a, b, c, α, β, and γ ), which together quantify its periodicity in the respective directions. The reciprocal lattice is a useful concept owing to the inverse relationship between diffraction geometry and the realspace lattice of a crystal structure, and its lattice parameters are usually denoted by adding asterisks (i.e., a ∗ , b∗ , c∗ , α ∗ , β ∗ , and γ ∗ ). Superlattice. See Section 5.44. Lattice System. The everyday-word system denotes a collective, and it can be broadly applied across many different disciplines. However, the combination of words including system can impart a specific meaning. A lattice system is a class of lattices that share the same point group (Table 3). Crystal System. Once again, when combined with particular terms, system can take on a very specific meaning. The term crystal system should only be used when referring to entries in the second column of Table 3. Crystal Family. In three dimensions, a crystal family is the same as the crystal or lattice system, except in the case of the rhombohedral/trigonal/hexagonal systems, which are combined into one hexagonal family (see Table 3). Crystal Class. According to Volume A of the International Tables for Crystallography, a “(geometric) crystal class is the set of all crystals having the same point group symmetry. The word ‘class’ therefore denotes

Table 3 Lattice systems, crystal systems, crystal families, and their point groups in three dimensions. Lattice system

Crystal system

Crystal family

Triclinic

Hexagonal

Point groups 1, 1

Monoclinic

2, m, 2/m

Orthorhombic

222, mm2, mmm 4, 4, 4/m, 422, 4mm, 42m, 4/mmm

Tetragonal Rhombohedral

Trigonal Hexagonal

Cubic

13

3, 3, 32, 3m, 3m Hexagonal

6, 6, 6/m, 622, 6mm, 62m, 6/mmm 23, m3, 432, 43m, m3m

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

14

Supramolecular materials chemistry

111

111

111 101 111

110 111

101 100 010 111

010 110

110

101

110

101

101

111

001 111 {110}+{111}+{100}

011 001 {110}+{111}+{100}

001

001

001

111 111

111

111

010 010

100

110

111

011 {110}+{111}

010

100

111

100

111 100

111 111

111

111 111 Cuboctahedron {110}+{111} Elongated {100}+{111}

Figure 11

111 Elongated {100}+{111}

A selection of crystal morphologies.47 (Reproduced from Ref. 47.  Nature Publishing Group, 2004.)

a classificatory pigeonhole and should not be used as synonymous with the point group of a crystal.”46 8. Crystallinity. A solid is crystalline if it possesses three-dimensional order that is measurable on the atomic or molecular scale. 9. Morphology. The morphology of a crystal is the geometric description of its outward appearance in terms of the crystal faces (i.e., size, shape, and angular relationships of the facets—see Figure 11). 10. Habit. See Section 5.30. 11. Crystallite. The distinction between a crystal and a crystallite is subjective, and there does not appear to be a quantitative definition that differentiates one from the other. A crystallite is generally understood to be a microscopic crystal. It usually exists as one of many similar particles that adhere loosely to one another, being separated by grain boundaries. A crystallite is therefore a primary particle of a polycrystalline (or microcrystalline) solid, which is often referred to as a powder. 12. Texture. Texture refers to the degree of crystallographic alignment in a polycrystalline material. A sample has no texture if the crystallographic alignment of the particles is completely random. The amount of texture of a material depends on the percentage of particles that possess a preferred orientation. Texture has a profound influence on the properties of many materials. 13. Single Crystal. The term single crystal is most often used as a compound noun and it usually refers to an individual (monocrystalline or monolithic) crystal

that can be characterized by means of single-crystal diffraction methods. The term is therefore based on practical rather than absolute aspects related to crystal size and quality. The lattice structure of a single crystal is sufficiently continuous such that no grain boundaries are present within the internal bulk of the particle, although it may possess even considerable mosaicity. When the term single crystal is used as a compound adjective (e.g., “a single-crystal transformation”), we recommend that it should be hyphenated in order to retain its sense of representing a single concept. Otherwise, when “single” is the adjectival modifier of the noun “crystal,” the hyphen is not required (e.g., “a single crystal was selected”). Similar arguments can be made for “a axis”, “unit cell”, “solid state,” and so on (e.g., “the a axis of the unit cell”, “the a-axis setting”, “the unit-cell transformation”, “a solid-state effect,” and “the organic solid state”).

5.21

Crystallochromy

Crystallochromy relates to the effect of the solid state on the color of dye materials and is due to the different packing arrangements of molecules.48–50

5.22

Deformation

A plastic deformation is a nonreversible change in shape and an elastic deformation is reversible.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

5.23

Desmotropism

The term desmotropy was first introduced by Jacobson in 188851 and stems from Greek, meaning “linked change” or “bond change” in the context of chemical crystallography. Desmotropy refers to the isolation of the tautomeric forms of a compound in different crystal structures and is therefore only observable in the solid state. It does not refer to the coexistence of the different tautomers in the same crystal structure. Furthermore, there are occurrences in the literature where desmotropy is incorrectly referred to as tautomeric or tautomerizational polymorphism 52, 53 —the definition of polymorphism (see Section 5.53) precludes the making or the breaking of bonds.

5.24

15

Disorder

Conceptually, crystals are infinite, three-dimensional, and periodic entities constructed from atoms, ions, and molecules. Periodicity implies that every object contained in its smallest repeating unit, the unit cell, is systematically and infinitely repeated in three dimensions. However, real crystals are prone to lattice and other defects and parts of molecules or even entire molecules often exist in multiple crystallographically independent (or energetically similar) orientations. Moreover, the crystal structure derived from the diffraction pattern is the spatial average over the whole crystal and any deviation from the ideal three-dimensional regularity of the crystal is presented as disorder. Generally, there are two categories of disorder: substitutional and positional disorder.54 Substitutional disorder occurs when the same site in at least two different unit cells is occupied by different atom types. Another form of substitutional disorder is partial site occupancy. In both cases, the anisotropic displacement parameters may appear to be either unusually large or unusually small. However, in cases where solvent molecules have high mobility, they normally tend to have larger displacement parameters. Positional disorder occurs if a single moiety occupies more than one site in a unit cell and has a stepwise distribution over these sites (also known as dynamic disorder) or if molecules possess multiple, well-defined energetically similar conformations in different unit cells (static disorder). Both examples of positional disorder are dealt with in the same manner during crystal structure refinement. Figure 12 illustrates a case of static disorder.54 Long-range order is the most distinct characteristic of crystalline materials and is simply the regular or periodic repetition (by means of translation) of atoms or molecules in three dimensions. Long-range order determines the structural anisotropy and macroscopic properties of crystalline solids.55

Figure 12 Static positional disorder in the structure of 1benzyl-1 -N ,N-dimethylaminomethylferrocene methiodide.56 The two disordered components are shown in blue and orange.

Short-range order refers to the absence of the repetitive order and translational periodicity. It extends over only a few atoms.55

5.25

Epitaxy

The growth of one crystal onto the surface of another is called epitaxy (or epitaxis)—that is, the surface of one crystal seeds the growth of the other form. When the two substances have the same structure, we refer to it as homoepitaxy and when they are different the phenomenon is called heteroepitaxy.

5.26

Eutectic

A eutectic reaction is an isothermal, reversible reaction between two or more solid phases, which results in a single liquid phase. The pure component thus formed is called the eutectic.3

5.27

Glass transition and glass transition temperature

When a noncrystalline or amorphous phase (e.g., a glass) melts, it undergoes a change from a hard brittle (rigid glassy) state to a more flexible state. This change is known as a glass transition and the temperature interval over which this occurs is known as the glass transition temperature Tg .57

5.28

Grain boundary

In the context of crystallography, a grain boundary is the interface between two crystallites.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

16

5.29

Supramolecular materials chemistry

Grinding

Crystalline materials are often subjected to mechanical grinding (generally by means of a mortar and pestle or a ball mill)—either to effect a chemical or topochemical reaction (i.e., mechanochemistry) or to reduce the size of particles prior to powder diffraction or sorption experiments.58 Although the term grinding appears to be preferred by most authors, it is more or less synonymous with pulverization, crushing, milling, and comminution. However, the past tense verb ground is more awkward than pulverized, crushed, milled, and comminuted because it is homonymous. When grinding is carried out in the presence of a liquid (e.g., solvent), the process is referred to as kneading or liquid-assisted grinding.59–61

5.30

Habit and form

The two most commonly observed herringbone patterns are the gamma packing arrangement (Figure 13d) and the sandwich herringbone motif (Figure 13e).

5.32

Hygroscopy

Hygroscopic substances either absorb or adsorb water from the atmosphere. Deliquescence occurs when a hygroscopic solid absorbs enough water from its surroundings to dissolve, thus forming a saturated solution in equilibrium with the atmosphere (i.e., the vapor pressure of the solution is equal to the partial pressure of water vapor in air). Not all hygroscopic solids deliquesce, but all water-soluble salts deliquesce under sufficiently humid conditions.64 Efflorescence is the opposite of hygroscopy and refers to the loss of water from a hydrated solid when it is exposed to the atmosphere.

The description of the external shape of a crystal (or aggregate) is referred to as its habit. Crystal habit should not be confused with crystal form. The term habit applies to an individual crystal or to an aggregate of crystals and implies the mode of growth. The crystal habit results from kinetic factors that determine the relative rates of growth along various directions of the crystal. Crystal form strictly refers to the internal structure of a crystal and is the basis for naming new structural modifications as Form I or Form II.3, 48 It is noteworthy that morphology and habit should only be used as a last resort for the identification of polymorphic forms. In mineralogy, there are many terms that describe the habit of a crystal or aggregate—for a wide variety of crystal shapes or habits see http://www.khulsey.com/jewelry/crystal habit.html.

In the field of inclusion chemistry, several host compounds are known to adopt identical packing arrangements that accommodate a variety of different guest molecules. Examples include Dianin’s compound (Figure 14),65 p-tertbutylcalix[4]arene,66 calix[4]arene,67 2,7-dimethyl-octa3,5-diyne-2,7-diol,68 and many metal–organic framework s (MOFs). In such cases, the terms isostructural and isotypic do not apply because the guest molecules are different. Although homeotypic (see Section 5.34) may be more appropriate, it is not specific about what is “essentially similar” between the structures. We have suggested the use of isoskeletal as a special case of isostructurality where only the packing of the host molecules is considered.68

5.31

5.34

Herringbone pattern (and angle)

The herringbone pattern is ubiquitous in the world around us—it is encountered in diverse areas such as brickwork (mainly paving, Figure 13a), fabrics, weaved baskets, and molecular packing in crystals. The herringbone motif consists of elongated entities that are stacked in a slanted manner to form linear columns, and adjacent columns are reversed with respect to the stacking direction. This type of packing of molecules is facilitated by common crystallographic symmetry elements such as screw axes and glide planes (for example, see Figure 13b). The herringbone angle is defined as the angle between the best-fit lines through the molecules in adjacent columns (Figure 13c)—that is, twice the tilt angle of the molecules within a stacked column.

5.33

Isoskeletal

Isostructurality

Isostructural compounds crystallize with similar unit cell dimensions, the same space group, and nearly identical coordinates for common atoms. In fact, since the discovery of isostructurality of crystals in 1819 by Mitscherlich, the phenomenon has generated much controversy, which can be attributed directly to his use of the word isomorphous.69 The IUCr does not define isostructurality —instead it still uses the term isomorphous as synonymous with the definition of isostructurality given above. However, K´alm´an and coworkers have pointed out that isomorphous should strictly refer to the external shape similarities of crystals or crystalline substances.70 They advocate that, in the context of structural similarities, isostructural should be used exclusively for organic crystals, with isotypic being reserved for organometallic, metal–organic, and inorganic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

17

(a)

f

(b)

(c)

(d)

(e)

Figure 13 Aspects of the herringbone motif : (a) brick paving, (b) packing of 4,4 -diiodobiphenyl,62 (c) the herringbone angle φ, and the (d) gamma and (e) sandwich motifs.63

crystals. Therefore, the fundamental difference between the terms isomorphous and isostructural is that the former describes crystal morphology (i.e., outward appearance), while the latter is used where two or more crystals share the same three-dimensional packing arrangement. Thus, by implication, it is (largely) possible to superimpose isostructural crystal structures.69–71 Two or more crystals may be considered isostructural even though small differences exist in the molecules, such as the replacement of a hydrogen atom with a methyl group. On the other hand, homeotypic (or homeostructural ) structures allow for larger moieties to be replaced such as (but not limited to) the guest molecules in a host–guest complex (also see isoskeletal ). The degree of isostructurality can be determined by calculating indices such as the unit cell similarity index (), the isostructurality index (Ii (n)), and the volumetric index Iv .72, 73

5.35

Isotropy and anisotropy

In chemical crystallography, these two antonymic terms refer to the directional dependence of a physical property. Isotropic properties or features are directionally independent, while anisotropic properties or features vary according to direction. Properties to which these terms can apply include dimensions, thermal expansion, flexibility, transmission of light (see birefringence and pleochromism), tensile strength, second harmonic generation, and so on.

5.36

Liquid crystal

Materials that exhibit fluid phases with a level of long-range order are referred to as liquid crystal s (LC). Although they are not solids, LCs share some important attributes

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

18

Supramolecular materials chemistry occurs when a substance undergoes a phase transition from the solid to the liquid state.76 The transition from solid to liquid is an endothermic process that is effectively a first-order transition with a discontinuous step in the firstorder derivative of the Gibbs free energy equation.3 The melting point of a substance can also be used as a means of identification and to assess its purity.

5.38

The mesomorphic state refers to anisotropic liquids (LCs) that are intermediate between a true liquid and a pure crystalline phase. More explicitly, it refers to the degree of molecular order that lies in-between the ideal threedimensional, long-range positional and orientational order found in crystals and the random arrangement found in isotropic liquids, gases, and amorphous phases. Anisotropic liquids that exhibit the mesomorphic state are usually elongated, aromatic organic molecules. The mesomorphic state can be divided into the smectic and the nematic states.

(a)

5.39

(b)

Figure 14 Isoskeletal arrangements of the host compound 2,7dimethyl-octa-3,5-diyne-2,7-diol with (a) carbon tetrachloride and (b) benzene as the guests. These two structures are isostructural with respect to only the host.68

with conventional crystals—in the interests of brevity, this section only provides a brief treatment of LCs and does not do justice to the importance and complexity of these fascinating substances. The behavior of LCs is midway between that of a liquid and that of a solid in that the molecules experience molecular motion yet retain the longrange order characteristic of crystals. The molecules of LCs are shaped anisotropically and are usually rodlike or disklike. Their shapes contribute to a large extent to their bulk behavior (of the LCs), while the ordering is dependent on the temperature. Thermotropic LCs generally undergo two phase changes, namely the nematic and smectic phases.74 The nematic phase is characterized by long-range orientational order, while the smectic phase is characterized by well-defined layers of a particular orientation. Furthermore, the nematic phase is threadlike and flows like a viscous liquid, while the smectic phase is soaplike and has an oily nature.75

Metal–organic framework (MOF)

Coordination polymers are a general class of compounds in which metal centers are linked by bridging ligands to form conceptually infinite 1D strands, 2D nets, or 3D frameworks. The terminology of such nonmolecular, extended, and solid-state structures has been discussed in a recent perspective article by Biradha et al.77 Yaghi introduced the term MOF to distinguish robust 3D coordination polymers that result from a rational design strategy that exploits the rigidity derived from well-defined metal centers (or clusters) and modifiable organic linkers.78 He termed this approach reticular synthesis.79 A more comprehensive treatment of MOFs can be found in Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry.

5.40

Metallocycle versus metallacycle

A metallocycle is a cyclic complex containing at least one metal atom (e.g., Figure 15a). A metallacycle is a derivative of a cyclic organic compound in which at least one of the (usually carbon) atoms is replaced by a metal center (e.g., Figure 15b).80

5.41 5.37

Mesomorphic state

Metamict

Melt

A melt (noun) is a liquid or a homogeneous mixture of two or more liquids close to its freezing point. Melting (verb)

Metamiction (or metamictization) is a mineralogical term that describes the gradual and ultimately complete destruction of a crystal structure to yield an amorphous phase (i.e.,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

N

N

N

N

Ag

Ag

N

N

N (a)

PPh3 CO Os CO PPh3 RS

N (b)

Figure 15 (a) A metallocycle of [Ag2 L2 ]2+ ,81 where L = 1,4bis(2-methylimidazol-1-ylmethyl)benzene and (b) an osmabenzene metallacycle.80

the metamict). Since the grinding of crystals sometimes yields amorphous material, it seems that this mineralogical term could be adopted for use in the field of chemical crystallography.

5.42

Metastable

The term metastable is a relative one and the context within which it is defined here relates to polymorphism. When a crystal form is in a metastable (equilibrium) state (not changing with time), it is susceptible to settling into lowerenergy states (different crystal forms). The metastable phase may persist for some time without changing, even though a more stable phase does exist. It is the equivalent of reaching a local minimum on a potential energy surface and persisting there, while the global minimum is reasonably close.82

5.43

Mixed crystal, solid solution, racemate, quasiracemate, pseudoracemate, and inversion twin

A mixed crystal (or a solid solution) is a monomorphic phase composed of two or more different atoms or molecules that are nonstoichiometrically interchanged within the crystal structure. The atoms or molecules present as the major components are known as the solvent phase, while the atoms or molecules present as the minor components are known as the solute phase. The terminology (mixed crystals) probably originates from mineralogy since minerals are rarely pure substances and commonly contain elemental mixtures. Solid solutions can sometimes be obtained by melting and then cooling two or more compounds simultaneously or by recrystallizing from a suitable solvent. The composition of the new phase is largely influenced by temperature, pressure, and the relative mutual solubilities of the components. There are two categories for solid

19

solutions: namely substitutional and interstitial. In the case of substitutional solid solutions, solvent atoms or molecules are directly replaced by solute atoms or molecules in the crystal structure, while interstitial solid solutions occur when solute atoms or molecules are trapped in the interstices between the solvent molecules. In both cases, the properties of the new phase are altered because of the deformation of the crystal lattice, which, in turn, disrupts the physical and electrical characteristics. The Hume–Rothery rules, which describe the conditions favorable for solid solution formation state that solid solutions are favored when the (solvent and solute) components have similar atomic radii or molecular volumes (within 15%), electronegativities, and valency, and when their crystal structures are isostructural. Mixtures of amorphous materials are not solid solutions.83–89 The enantiomers of chiral molecules may be present in mixtures, the most important compositions of which are 1 : 0 and 0 : 1 for the enantiopure compounds, and 1 : 1 for the racemate.13 Indeed, the term racemate should only be used for a 1 : 1 mixture of paired enantiomers, and enantiomeric mixture should be used for any other proportion.13 The enantiomers are usually (but not necessarily—see kryptoracemate) arranged in pairs related by centrosymmetry —in this manner, the packing arrangement achieves higher symmetry and a higher stability than when the enantiomorphs crystallize separately.90 Enantiomeric mixtures with rational proportions other than 1 : 1 have been termed anomalous racemates, but Flack points out that this makes it difficult to distinguish them from pseudoracemates and quasiracemates. As a solution, he suggests the use of the less compact but more precise terminology M:N mixed enantiomeric crystal structure.13 A quasiracemate is “the crystalline product of a 1 : 1 association between quasienantiomers.”3 Quasienantiomers are pairs of compounds, one of which has a molecular structure that is closely related to the enantiomer of the other. Quasiracemates tend to crystallize with pseudoinversion symmetry with packing preferences very similar to their racemate relatives (see Figure 16). Since quasiracemates generally involve heteromeric pairs of enantiopure compounds, quasiracemate formation provides an entry point for controlling the chirality of solid-state processes.91, 92 In 1897, Kipping and Pope93 reported what they termed pseudoracemates and the model that they proposed was consistent with what we now refer to as inversion-twinned crystals. As pointed out by Flack,13 Kipping and Pope were limited by the technology of the time and what they actually observed was a solid solution in which either enantiomer is able to occupy any molecular position in the crystal structure. Flack suggests that a more suitable name for this type of solid solution is a disordered racemic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

20

Supramolecular materials chemistry H3CO

O

CO2H

O

CO2H

(a) Br (b)

(c)

Figure 16 The quasienantiomers (S)-2-(3-methoxyphenoxy)propionic acid (a) (R)-2-(3-bromophenoxy)propionic acid (b) crystallize together to form a quasiracemate (c).92

(crystal) structure, or even the more generic term disordered mixed enantiomeric (crystal ) structure. It therefore seems that the term pseudoracemate may be abandoned for the present. It is pertinent to point out that Flack also strongly discourages the use of the term racemic twinning instead of twinning by inversion.13 Since a conglomerate needs to consist of a mechanical mixture, and a mechanical mixture requires grain boundaries between the particles, a collection of inversion-twinned crystals is not a conglomerate.

rotational symmetry. The diffraction patterns exhibit fivefold and other noncrystallographic point symmetries.96 When dealing with the diffraction pattern of a supercell, no distinction is made between reflections and their satellites, and all reflections are prioritized equally for indexing. A smaller reciprocal unit cell is therefore obtained. Depending on the maximum order of the satellite reflections that are observed, many of the reflections may be absent. The direct space unit cell is larger and is referred to as the supercell, while the resulting structure is called the superstructure.97

5.44

5.45

Modulated structures

Modulated structures are derived from structures whose atoms or molecules are shifted with respect to neighboring atoms or molecules, resulting in the overall threedimensional translational symmetry being extinguished. However, the repositioning (of atoms or molecules) in modulated structures is not arbitrary since distinct rules are adhered to within the modulations.94 Commensurately modulated structures strongly depend on the periodicity of the modulation wave where the periodicity is an integral number of lattice translations of the unit cell.94 Incommensurately modulated structures have a periodicity that is not an integral number of lattice translations of the unit cell and is therefore incommensurate with the periodic basic structure.94 Aperiodic crystals are crystals that possess long-range order but not translational symmetry. It is well known that translational symmetry distinguishes the crystalline state from other states. The discovery of aperiodic crystals has led to a wider range of materials being included within the notion of a crystal. Amorphous materials are not aperiodic crystals.94 Aperiodic composites can be described as two or more crystalline subsystems that have incommensurate periodicities along one or more crystallographic directions, that is, the ratio of the periodicities is not a rational number.95 Quasicrystals possess long-range order, but not threedimensional translational periodicity. The long-range order gives rise to sharp Bragg peaks, while the absence of translational symmetry is reflected in the noncrystallographic

Monomorphism

A substance that exists as a single crystalline phase is monomorphic. Since it is imprudent to assume that a particular compound cannot have polymorphs, the term monomorphism should be used to signify phase purity rather than the hypothetical property of lacking polymorphism.

5.46

Morphotropism

Morphotropism (or morphotropic change) is derived from the Greek “morph” and “tropos” meaning “form change.” Morphotropism is the change that occurs when, by chemical substitution, the limit of change in the crystal structure is exceeded and an alternative atomic arrangement in space gives rise to a new series of isostructural materials. The change may result from a chemical substitution of hydrogen atoms by Br, Cl, or I for example. It may also be induced by a high degree of deformability of the replacement moiety.98–100

5.47

Mosaicity

Conceptually, a crystal structure consists of the defect-free tiling of unit cells in the three crystallographic directions to form an infinite coherent entity. This idealized view is far from reality because real crystals are neither infinite nor defect free. Indeed, a single crystal can be described as a three-dimensional tiling (or mosaic) of microscopic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography coherent particles aligned in almost the same orientation relative to their conceptually identical internal structures. The overall degree of particle misalignment is referred to as the mosaicity (or mosaic spread ) of the crystal, and should not be confused with long-range order. In a typical diffraction experiment, constructive interference only occurs within each coherent particle and the overall diffraction pattern of the crystal is the sum of the individual diffraction patterns of the particles. The mosaicity of a crystal can be estimated from the half-height widths of the diffraction peaks and is typically reported as an angle (the higher the mosaic spread, the more the crystal deviates from the conceptual ideal).

5.48 N -mer A dimer is an entity that contains two subunits (i.e., monomers), which may be connected by covalent bonds or by much weaker interactions (e.g., hydrogen bonds, π –π interactions, etc.); the association may be subjective and can usually be gleaned from the context of the discussion. Although the monomers are often identical, the term dimer only means “two parts” and does not require the parts to be identical. More precision can be provided by distinguishing between homomeric and heteromeric dimers (i.e., homoor heterodimers) where the monomers are either identical or different, respectively. Although what is regarded as a monomer is generally subjective and best left to the author’s discretion, it seems sensible that monomers should be of approximately the same scale. For example, it might not be strictly incorrect to refer to a binary adduct between a small solvent molecule and a much larger organic compound as a dimer, but more appropriate terms may exist in such cases (e.g., host–guest complex, carcerand, etc.). Terms such as trimer, tetramer, pentamer, hexamer, and so on are used for entities consisting of multiple subunits where the number of monomers is known. In such cases, the additional qualifiers homomeric and heteromeric are still useful, but not as precise as in the case of a dimer. The compact term n-mer, where n is replaced by a number, can be used where the alternative may be cumbersome (e.g., “14-mer” instead of “tetradecamer”), and the more general term polymer can be used when the number of monomers is unknown or conceptually infinite.

5.49

Neutron normalization

A typical crystal structure analysis is carried out using Xray diffraction, which ultimately yields a three-dimensional electron density map. Relative atomic coordinates are then inferred from the maxima of the electron density peaks.

21

Since the X-ray scattering power of an atom depends, in part, on its number of electrons, X-ray diffraction is relatively insensitive to hydrogen atoms. Furthermore, each hydrogen atom in a molecule is usually covalently bonded to a much heavier atom, and its region of maximum electron density therefore tends to be polarized slightly toward the parent atom. As a consequence, the coordinates of hydrogen atoms, when obtained from electron density maps, can generally be regarded as being unreliable. In most cases, it is possible to calculate the most likely positions of the hydrogen atoms, provided that the hybridization and bonding environment of the parent atom are known. However, refinement programs usually utilize models that favor lower residual statistics (i.e., a better fit to the electron density map). Neutrons are diffracted by the nuclei of the atoms, and hydrogen has a relatively strong neutron-scattering power. Therefore, the positions of the hydrogen atoms in a crystal structure can be determined far more precisely with neutron diffraction than with X-ray diffraction. Owing to a number of limitations, neutron diffraction is not as widely used as X-ray diffraction. However, the Cambridge Crystallographic Database (Version 5.32, November 2010) contains 1486 structures determined using neutron diffraction, and bond lengths to hydrogen atoms in various environments are therefore well known. When carrying out calculations based on experimentally measured crystal structures (e.g., lattice energy calculations, Hirshfeld surface analyses, etc.), it is essential to provide the relevant computer programs with atomic coordinates that are as accurate as possible. Since organic molecules tend to have many hydrogen atoms at their extremities, it is particularly important not to rely on hydrogen atom positions determined by X-ray diffraction, but rather to adjust the positions of the hydrogen atoms such that they are in agreement with values obtained for the relevant environments from neutron diffraction experiments. This process of adjusting bond lengths is referred to as neutron normalization.

5.50

Nucleation

Primary nucleation refers to the (homogeneous) spontaneous formation of nuclei of the crystallizing phase. The critical cluster or critical nucleus is the minimum size that a continuously growing nucleus has to surpass in order to make the transition to a stable crystalline phase. Secondary nucleation or seeding describes the process whereby nuclei are induced on or near crystals (i.e., the seeds) of the solute that are already present in a supersaturated solution. Moreover, the seeds are not required to have the same crystal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

22

Supramolecular materials chemistry

or molecular structure as the target crystals. Seeds are usually added to supersaturated solutions in order to promote crystallization of the target phase. Alternatively, they can be added to molten liquids prior to cooling, or they can be exposed to a vapor of the material to be crystallized. Seeding can be used to control crystal size and size distribution, while factors such as seed quantity and seed size affect the successful implementation of this protocol.75, 101, 102

5.51

Phase and state

In the context of chemical crystallography, the distinction between the terms phase and state can be subtle. Conventionally, the term state refers to the three states of bulk matter (i.e., solid, liquid, and gas, but additional states of matter may arguably include plasmas, LCs, gels, and supercritical fluids). In chemical crystallography, a phase is a physically homogeneous portion of solid-state matter that is discontinuous from other phases. Substances are able to exist in more than one solid phase, each with its unique arrangement of atoms, molecules, or ions. For example, we may refer to the different polymorphs of a given compound as having different phases, but they exist in the same state. A pure component undergoes a first-order phase transition or discontinuous change when it changes from one phase to another, for example, the gas-to-liquid transition. Essentially, during the phase transition, one phase increases at the expense of the other and this usually follows the direction of heat transfer.3 A second-order phase transition or continuous change is characterized by a change in the heat capacity of a substance without the evolution of heat. While the first derivatives of the Gibbs energies (or chemical potentials) are continuous, the second derivatives with respect to temperature and pressure, that is, heat capacity, thermal expansion, and compressibility are discontinuous, for example, the transition

from ferromagnetic to paramagnetic at the Curie temperatures represents such a transition and is observed as the lowering of the baseline of the thermogram obtained using differential scanning calorimetry.3

5.52

Polycrystalline

Although the term means “many crystals,” it is generally accepted that polycrystalline refers to a crystalline powder in which the individual particles are crystallites.

5.53

Polymorphism

Polymorphism, taken from the Greek words “poly” (many) and “morph” (form or shape), was identified in 1822 by Mitscherlich, who is generally credited with the first observations of the phenomenon.48, 99, 103 Polymorphism occurs when a chemical substance can exist in at least two different crystalline phases (i.e., different crystal structures of the same chemical composition, e.g., Figure 17), with the requirement that those crystalline phases lead to the same liquid or vapor phases. Explicitly, this definition precludes solvates (and hydrates) as being polymorphs of any chemical substance, but polymorphs of solvates (solvatomorphs) are not excluded. If polymorphism had been discovered after the advent of structural crystallography, one wonders if the phenomenon would instead have been called polytaxis (i.e., “many arrangements” rather than “many shapes”). However, polymorph has been in use for long enough for this to now be a moot point. There has been an extended debate104, 105 about what constitutes polymorphism (e.g., see tautomeric and conformational polymorphism below). Similarly, there has also been a heated debate about the suitability of the term pseudopolymorph for solvates. In the latter case, the question

(b) (a)

Figure 17 Packing diagrams that show two polymorphs of hydroquinone. (a) The rhombohedral form 116 and (b) the monoclinic form.117 The projections are along the unique axes in both cases. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography is whether or not solvates may be regarded as “false polymorphs,” as the term implies. Some proponents of pseudopolymorph are in favor of a more inclusive and broader definition, arguing that the definition should not be formulated exclusively in terms of structure and packing arrangements.104, 106 Conversely, it is argued by opponents that the term solvate is already fully descriptive and that an erroneously constructed term serves no useful purpose.105, 107 It has even been suggested that this term be omitted from the supramolecular lexicon—undoubtedly a difficult task after the fact.48, 108 The authors recommend the term polymorphism where the definition holds strictly, that is, hydrates or solvates should not be associated with polymorphism of their hosts. 1.

2.

Pseudopolymorph —pseudo meaning “false” is often and mistakenly/incorrectly associated with the term polymorph when it is used to describe solvated or hydrated forms of compounds. Solvated and solventfree phases are distinct and should therefore not be referred to as polymorphs of one another.105 Conformational polymorphism occurs when conformationally flexible molecules adopt different conformations in the different forms (in borderline cases, this can be assessed by inspecting differences in pertinent torsion angles, with due regard to standard errors). Consider two or more conformers, which can be distinguished with respect to rotations about single bonds, which are in rapid equilibrium under precrystallization conditions. Were this not the case, they would then be distinct isomers and therefore different chemical substances—and, by implication, their structures cannot be polymorphs. It should also be noted that crystal packing interactions can stabilize particular conformers that might be relatively unstable in solution. Frequently, two or more low-energy conformations may appear in different crystal structures.48, 109, 110 Vujovic and Nassimbeni reported what they considered to be a temperature-dependant change in the conformation of methyl paraben and tentatively asked whether or not this constitutes polymorphism.111 On the basis of two structure determinations of the same crystal at 293 and 113 K, they argued that the torsion angles that signify the orientation of the ester group are statistically different in the two structures. Indeed, the simulated powder diffractograms for the two structures are not superimposable. Threlfall and Gelbrich responded112 that the crystal structure of methyl paraben at 118 K does not represent a new polymorph. They began by pointing out that the true definition of a polymorph is that it represents a distinct solid phase and not a distinct crystal structure.

3.

4.

5.

6.

23

They cited the fact that the structure changes slightly with slight changes in temperature or pressure and that these changes may accumulate to produce a substantial change in the structure over a wide temperature or pressure interval. Although we agree with this assessment, the highly pertinent question asked by Vujovic and Nassimbeni highlights the fact that there is a degree of subjectivity in deciding whether or not two structures share the same packing arrangement. Conformational synmorphism occurs when “the different conformers of a molecule are distributed randomly throughout the crystal lattice. Such a situation usually exists when two or more conformers have similar overall molecular shapes. Thus, at any particular molecular site, a number of conformations may be adopted, the relative population being determined by the relative intermolecular and intramolecular energies involved.”48 It is not difficult to appreciate that this concept shares some similarities with solid solutions and disorder. Concomitant polymorphs crystallize simultaneously from the same solvent or melt and in the same crystallizing flask under identical crystal growth conditions. However, the possibility that concomitant polymorphs grow in stages by means of a solvent-mediated phase transformation cannot easily be ruled out.48 Disappearing polymorphs occur when a preferred polymorph (usually the metastable form) can no longer be isolated because of the appearance of a second, more stable form. The original form does not completely disappear but can no longer be isolated free of the seeds of the new, more stable form. However, extreme measures have to be taken to isolate the metastable form again.113 The isolation of specific tautomers in different crystal structures is called desmotropy. However, it is not difficult to appreciate that desmotropy is closely related to polymorphism, and the term tautomeric polymorphism has indeed been invoked in this regard. Since the definition of polymorphism (simply stated) involves different packing arrangements of the same compound(s), it is inevitable that instances will arise that challenge the accepted notion of what constitutes polymorphism. For example, how different should the “different” packing arrangements be and how similar should the “same” molecules be? The former question has already been explored in the text on conformational polymorphism above and the latter has been discussed at length in a report by Bhatt and Desiraju.114 Although the chemical structures of two tautomers of a compound are strictly different, the two forms might interconvert rapidly (in a temperaturedependent manner) in solution or in the melt. This

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

24

Supramolecular materials chemistry

implies, of course, that they are then the same compound—and invokes the requirement that polymorphs should yield the same liquid (or vapor) phase without specifying that dynamic isomerism should be discounted. It seems reasonable to merely observe that tautomeric polymorphism is a special case of polymorphism that involves a special relationship between the molecules present in the crystal structures (i.e., that they can interconvert), and that in such cases we should not expect to find identical molecular structures in the different polymorphic forms. 7. Packing polymorphism is a redundant term. It has presumably been used to indicate different packing of the same compound, but the term polymorphism is exactly that.92, 93 Bernstein has discussed the superfluous use of packing polymorphism and similarly redundant terms such as pharmaceutical polymorphism, synthon polymorphism, and structural polymorphism.108 8. Enantiotropism and monotropism describe the relationship between polymorphs especially where two polymorphs are being explored. The relationship is enantiotropic when each crystal form has a particular temperature range within which it is stable. A transition point exists where the two forms are equistable and where it is possible for the two forms to interconvert. Above the transition point temperature, the form that is stable at higher temperatures exists exclusively, while below the transition temperature the form that is stable at low temperature is stable. It is more usual for the high-temperature form to persist outside or beyond its stability range than for the low-temperature form to do so. When one form remains metastable (i.e. with respect to the other form) at all temperatures, then the relationship is considered to be monotropic. The only transition point exists beyond the melt and is therefore unreachable.115 9. A dimorphic or trimorphic system has two or three known polymorphs, respectively. However, it should be noted that these terms should be used with care since it is not possible to predict when or if another form will be discovered. 10. Form (with regard to naming different polymorphs): in the absence of well-defined rules governing the naming of polymorphs, it is usual to use Arabic and Roman numerals as well as Latin and Greek letters, for example, “α form,” “form 1,” “form I,” “phase 1,” or “phase I”. Different authors use different notations with or without knowledge of prior work.48 11. A substance can exhibit polyamorphism (analogous to polymorphism) if it can exist in more than one amorphous state.

12.

13.

14.

15.

16.

5.54

A polymorphic transition is reversible transition of a solid phase at known temperature and pressure to another solid phase (of the same chemical composition with a different crystal structure).3 Displacive phase transitions occur without the breaking of bonds and involve only small motions of atoms to change the symmetry of a crystal structure.91 Isopolymorphism is used when the various polymorphic forms of a substance are isostructural or homeostructural to corresponding polymorphs of another substance.75 An inversion point is the temperature (or pressure) at which one polymorph transforms into another polymorph at constant pressure (or temperature).3 A lambda transition occurs when the heat capacity of a substance is discontinuous (second-order transition) or has a vertex (higher-order transition) at the transition temperature.3

Polytypism

An element or compound is polytypic if it exists in two or more layerlike crystal structures that differ in their layer-stacking sequences. Polytypes are nearly identical in structure and composition and polytypism is a special case of polymorphism: the two-dimensional translations within the layers are essentially preserved, whereas the lattice spacing normal to the layers varies between polytypes and is indicative of the stacking period. There is increasing evidence that some polytypic structures are characterized either by small deviations from stoichiometry or by small amounts of impurities.118–121

5.55

Powder diffraction

There are several terms related to powder diffraction methods that are occasionally used incorrectly. For example, an experimentally measured powder diffraction pattern consists of a two-dimensional plot of intensity versus the Bragg angle 2θ . This plot may be referred to as a diffractogram, a powder pattern, or a powder trace (or variations of these terms), but it should never be called a spectrum because a spectrum is generally regarded as a signal that varies with frequency. Given a crystal structure in the form of unit cell parameters, the space group, and a set of atomic coordinates, it is possible to simulate a powder diffractogram. However, do we refer to the simulated pattern as having been calculated or simulated ? This is largely a matter of personal preference but we can offer an opinion. From the crystal structure information, it is possible to calculate the relative intensities of all reflections, as well as their Bragg

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography

Relative intensity

oriented, but this is only possible if the sample contains a suitably large number of crystallites. Furthermore, the orientations are dependent on the habit (or morphology) of the crystallites. If the crystallite habit (or morphology) is isotropic, then the probability of attaining a randomly oriented sample is high. Conversely, when the crystallite habit is anisotropic (e.g., plate- or needlelike), then the probability of experiencing preferred orientation is high. However, preferred orientation can also be mistaken for inadequate powder average that results when too few crystallites with a random distribution of orientations are irradiated.55, 123

5.57

10

15

20

25 2θ/°

30

35

40

Figure 18 Calculated (bottom) and simulated (top) powder diffractograms of 5-bromo-2-(hydroxymethyl)pyridine.122

angles for a given radiation wavelength. When these data are plotted as line graphs (Figure 18, bottom), they represent a calculated diffractogram, which does not take peak profiles into account (peak widths are dependent on properties of the material other than atomic coordinates, as well as experimental factors). It is often desirable to fit Gaussian profiles to the individual peaks and to then sum the intensities along the 2θ continuum in order to facilitate a visually more compelling comparison to a measured diffractogram. This process involves a degree of imitation, and may therefore be referred to as a simulation. Of course, a simulation involves several calculations (i.e., peak positions, intensities, and widths), so either trace may be called a calculated pattern. However, the diffractogram shown at the top in Figure 18 could also be referred to as a simulated trace.

Protic, aprotic, and amphiprotic

A protic (or protogenic) substance possesses dissociable protons (e.g., HCl, benzoic acid) and can therefore undergo self-ionization in the form of autoprotolysis. Conversely, aprotic solvents cannot undergo autoprotolysis and examples include pentane, carbon tetrachloride, and dioxane. Amphiprotic (or amphoteric) substances can serve either as Brønsted acids or bases (e.g., water, ammonia). Autoprotolysis involves a proton transfer between two identical molecules: 2 NH3 → NH4 + + NH2 − Note that these properties generally depend on the medium. For example, H2 SO4 is protogenic in water but amphoteric in a superacid. According to the IUPAC Gold Book, the terms protogenic and amphoteric are preferred to protic and amphiprotic.3

5.58

Pseudosymmetry

Pseudosymmetry is noncrystallographic symmetry that occurs when a molecule or pattern of molecules within a crystal structure exhibits symmetry that is not incorporated into the space group.

5.59 5.56

25

Self-assembly and self-organization

Preferred orientation

Preferred orientation is a problem encountered when using powder diffraction methods. It occurs when most of the crystallites in a polycrystalline sample have a nonrandom distribution of orientations, resulting in only a certain number of the Bragg reflections moving into the reflecting position and thus contributing disproportionately to the scattered intensity. Ideally, the crystallites should be randomly

Lehn has outlined a hierarchical order of the terms templating, self-assembly, and self-organization to describe the spontaneous formation of supramolecular assemblies composed of preprogrammed molecular components (i.e., tectons).124 Self-assembly concerns the antientropic aggregation of individual components to yield an ordered multimolecular entity. Intermolecular interactions that are intrinsically preprogrammed (see synthon) into the building

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

26

Supramolecular materials chemistry

(a)

(c)

(b)

Figure 19 Examples of (a) self-inclusion where the same molecule plays two distinct roles (i.e., as both host and guest)68 , (b) interpenetration of two metal–organic frameworks,10 and (c) interdigitation of two p-tert-butylcalix[4]arene molecules.128

blocks drive the self-assembly process by means of molecular recognition (i.e., tectonics), and this phenomenon is the basis of crystallization. Self-organization involves the interaction between the components of self-assembled entities as well as the collective behavior resulting from these interactions (e.g., phase transitions).

5.60

Self-inclusion, interpenetration, and interdigitation

Self-inclusion can occur in host–guest chemistry and refers to situations where the host and guest molecules are the same (Figure 19a), provided that the distinction between their respective roles as host and guest is clear. In all cases of self-inclusion Z  >1125 because the molecules must be crystallographically distinct in order to fulfill two different roles. Interpenetration most often involves two or more (identical or different) coordination polymer networks or hydrogen-bonded frameworks that are entangled with one another (Figure 19b). Interpenetration results from coordination polymeric or hydrogen-bonded networks that would

possess considerable volumes of void space in isolation. However, this void space is then consumed by interpenetration. The only requirement on the entangled networks is that their separation can only be carried out through the breaking of network connections.126 Interdigitation is the interleaving of the digitlike moieties of molecules in a crystalline material (Figure 19c), that is, when part of a molecule penetrates into the cavity or void space of a neighboring molecule.127

5.61

Single-crystal to single-crystal (SCTSC) transformation (lithotropism)

Any change in the structure of a crystal that can be measured by means of single-crystal diffraction methods is called an SCTSC transformation. Such changes in structure can include polymorphic transformations, topochemical reactions and the absorption or desorption of guest molecules. In the last-mentioned case, the structural changes might be minimal but the composition changes.5, 129 It may serve a useful purpose to suggest a

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography single word that encompasses the remarkable ability of some single crystals to undergo structural changes—in this regard, we can suggest the neologism lithotropism (e.g., a lithotropic process).

5.62

Site occupancy factors (SOFs)

During crystal structure analysis, the site occupancy factor (SOF) of each atom is ideally assigned a value imposed by the site symmetry of its position in the unit cell. For example, an atom located on a general position is assigned an SOF of one, while an atom situated on an inversion center has SOF = 0.5. However, it should be noted that the SOFs are lower than their symmetry-imposed values in cases where atoms are disordered. Furthermore, inclusion compounds such as solvates are often nonstoichiometric as they can lose solvent molecules during crystal selection and mounting. In such cases, it may then become necessary to model guest positions with lower than ideal SOFs.54

5.63

Solutions

Most of the crystals that are relevant to chemical crystallography are precipitated from solution. It is therefore important to understand the terminology of solutions. Solvent. A solvent may be a solid, liquid, or gas that dissolves another solid, liquid, or gas. 2. Solute. The dissolved species is called the solute. 3. Precipitate. When the solution becomes supersaturated with the solute, the latter will precipitate, usually as crystals or crystallites. 4. Cosolvent. A cosolvent is an additional solvent added to a solution, usually in small quantities, and with the purpose of enhancing the solubility of the solute. 5. Antisolvent. An antisolvent is an additional solvent added to a solution in order to decrease the solubility of the solute. 6. Supersaturation. When the concentration of the solute exceeds its solubility limits under given conditions (e.g., temperature), the solution is said to be supersaturated.

with many different solvents is known as solvatomorphism (see Section 5.53 for the use of pseudopolymorph in the context of solvates). Decomposition/desolvation/dehydration —On heating, solvates or hydrates release the vapor or gas of the included solvent or water molecules. The loss of the guest as a gas, liquid, or vapor is referred to as decomposition or desolvation in the case of solvates or dehydration in the case of hydrates. Decomposition/desolvation/dehydration is the phase transition from a single phase (solvated or hydrated ) into two or more phases (desolvated or partially desolvated, and the evolved vapor or gas). Furthermore, desolvation and dehydration are usually accompanied by a related thermal event observable by differential scanning calorimetry.3, 57, 130 It is often necessary to refer to a solventless phase of a compound that is known to form solvates, and a number of terms can be used (e.g., solventless, unsolvated, desolvated, nonsolvated, solvent free, and apohost). Solvent free and solventless are more general in that these terms do not hint at how the solventless phase came about, while unsolvated and desolvated (especially the latter) imply that solvent molecules have been removed from a solvate. Apohost is even more general because not all inclusion compounds are solvates.

5.65

1.

5.64

Solvate

Solvates are solid-state host–guest complexes in which solvent molecules are included into the interstitial spaces, cavities, and channels of other molecules as a result of crystallization. When the included solvent is water, the term hydrate applies. The ability of a compound to form solvates

27

Solid-state and solvent-mediated phase transition/transformation

In the case of a dimorphic compound, and under specific conditions of temperature, pressure, and composition, one phase may be stable, while another is metastable. It is noteworthy that even though one of the phases is metastable it is not precluded from persisting for any length of time. However, an inevitable consequence of the existence of a metastable phase is that at some point it will undergo a phase transition and transform into the stable phase. The metastable phase can achieve this via two routes: (i) a solid-state phase transition or (ii) a solvent-mediated phase transition. During the solid-state phase transition, the metastable phase undergoes a rearrangement of its molecules or atoms until the transformation is complete. In the case of the solvent-mediated phase transition, the metastable phase is in contact with a solvent and is able to dissolve, nucleate, and regrow as the stable phase from the same solution.131

5.66

Sorption

The study of porous crystalline materials has recently become an area of intense research activity, but, in the context of permeable solids, should we refer to the uptake of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

28

Supramolecular materials chemistry

one substance by another as absorption and adsorption? In the discussion that follows, we use the generic term sorption (Latin for “suck”) for the process whereby a liquid or gaseous substance (the sorbate) is taken up (i.e., sorbed ) by a solid substrate (the sorbent). We can clearly distinguish between two types of sorption. Surface sorption is a two-dimensional phenomenon, the extent of which depends on the surface area of the sorbent. The surface area can be manipulated by mechanical means such as grinding, which influences the particle size distribution of the material, and thus the number of sorption sites available to the sorbate. On the other hand, internal sorption is a three-dimensional phenomenon, which depends on the volume of the sorbent. In principle, the distribution of sorption sites is intrinsic to the three-dimensional structure of the material and their number therefore depends on the amount (i.e., volume) of sorbent and not particle size. Furthermore, as a consequence of sorption, the sorbate becomes part of the three-dimensional structure of the sorbent, thereby changing the composition of the bulk material—a process that has great significance with regard to our concept of solidstate structure. It therefore seems reasonable that these two quite distinct sorption phenomena should warrant at least slightly different terminology. Unfortunately, this is not the case and, for historical reasons, it may be too late to do anything about it. The antonymic Latin prefixes “ab” (away from) and “ad” (toward) do not shed much light on the differences between absorption and adsorption. Absorb is an everyday English word that can be used in a variety of different contexts (e.g., objects can absorb light and people can absorb information). On the other hand, adsorb has a more specific and scientific meaning. With regard to molecular uptake, the IUPAC definition of absorption is “the process of one material (absorbate) being retained by another (absorbent); this may be the physical solution of a gas, liquid, or solid in a liquid, attachment of molecules of a gas, vapor, liquid, or dissolved substance to a solid surface by physical forces.” The same source defines adsorption as “an increase in the concentration of a dissolved substance at the interface of a condensed and a liquid phase due to the operation of surface forces. Adsorption can also occur at the interface of a condensed and a gaseous phase.” Here “interface” specifically refers to the boundary between two phases. Adsorption is further subdivided into chemisorption, resulting from “chemical bond formation (strong interaction) between the adsorbent and the adsorbate in a monolayer on the surface” and physisorption in which “the forces involved are intermolecular forces (van der Waals forces) of the same kind as those responsible for the imperfection of real gases and the condensation vapors, and which do not involve a significant change in the electronic orbital patterns of the species involved.” Although the term van der Waals

adsorption is synonymous with physisorption, its use is not recommended by the IUPAC. From their IUPAC definitions, it appears that adsorption should be reserved for surface sorption, while absorption should refer to the penetration of the sorbate beyond the sorbent surface. Activated carbon is probably the oldest known substrate for applications involving surface sorption and it is useful to characterize such materials in terms of their total surface areas using well-known techniques such as Brunauer–Emmett–Teller (BET) measurements. According to the IUPAC guidelines, the term adsorption is highly appropriate in this case. On the other hand, zeolites are microporous aluminosilicates that contain large void spaces as a result of their three-dimensional openframework structures, and they can incorporate a wide variety of sorbates into their internal volumes. Since BET measurements can readily be applied to zeolites as well, the extent of their porosity is also conventionally expressed in terms of the total surface area available to sorbates. It is perhaps for this reason that the term adsorption has been used in preference to absorption where zeolites are concerned. Owing to the parallels between studies of zeolites and more recent emphasis on micro- and mesoporous MOFs, the term adsorption seems to be taking precedence for volumetric sorption. In our view, it is unfortunate that no terminological distinction is made between two related, but quite different processes. Even though the IUPAC definitions of absorption and adsorption are relatively unambiguous, references to sorption phenomena will most likely continue to follow entrenched practices. The reverse of both absorption and adsorption is called desorption.

5.67

Supercooling

Supercooling occurs when a liquid or gas is cooled below its freezing point without solidifying, usually due to the lack of nucleation sites for crystallization. For example, water freezes first where it is in contact with its container even though the temperature in the interior region (i.e., far from possible nucleation sites) of the liquid may be well below the freezing point. Supercooling should not be confused with freezing point depression, which is due to the influence of solute molecules.

5.68

Supramolecular Isomerism

The existence of more than one type of network superstructure constructed from the same molecular building blocks is known as supramolecular isomerism (see also Supramolecular Isomerism, Supramolecular Materials Chemistry).132 From the definition, it is apparent that

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography supramolecular isomerism is related to structural isomerism at the molecular level. In other words, the relationship between supramolecular isomerism and molecules is similar to that between molecules and atoms. There are four classifications of supramolecular isomerism namely: structural, conformational, catenane, and optical isomerism. Structural —The components of the network are unchanged but a different superstructure results. In such a situation, the networks are effectively different compounds even though their empirical formula and chemical components are identical. Conformational —Conformational changes in flexible ligands (e.g., bis(4-pyridyl)ethane) generate a different but related network architecture. Conformational polymorphism is a closely related subject. Catenane —The varying degrees in which networks interpenetrate or interweave can result in significant variations in the overall structure and properties. These variations depend upon the molecular building blocks that are utilized. Effectively, interpenetrated and noninterpenetrated structures are different compounds. Optical Networks —these networks are by their very nature chiral and then crystallize in Sohncke space groups. They are analogous to homochiral compounds. This type of supramolecular isomerism lies at the heart of spontaneous resolution of chiral solids.132

5.69

Synthon, moiety, tecton, chiron, and motif

A moiety is a part of a molecule (often called a functional group) and not the molecule itself. When two or more molecules interact with one another in a complementary fashion that can reasonably be expected, given the functional groups, the pattern of the interaction is called a supramolecular synthon. A timeless example of synthons that direct the overall structure is the recognition of guanine by cytosine (Figure 20) and adenine by thymine in DNA. Noting the generality of such phenomena, Desiraju and Nangia have stated that such patterns may be used in a predictive sense.133 According to the Merriam Webster Dictionary, a motif is a single or repeated design. Although not formally entrenched in the nomenclature of chemical crystallography, motif is clearly a useful word for describing a pattern of atoms or molecules in a crystal structure. Once the salient features of a particular arrangement have been described, the pattern can be referred to compactly as a motif in subsequent text. Although considerable flexibility in the use of motif should be tolerated, the word is far more useful for referring to a geometrical feature as a subset of a crystal structure rather than to the entire structure.

29

H R

N

H

O

N

H

N

R

N N N

N

O

H N H

Guanine

Cytosine

Figure 20 The interaction between guanine and cytosine, which occurs in DNA. The supramolecular synthon is highlighted in blue.

It is also more useful to reserve the use of motif for patterns involving homo- or heterogeneous multicomponent entities. Tectons are the fundamental features that drive the process of self-assembly —they are either the peripheral molecular recognition sites or the scaffolds that orient these sites. Much of crystal engineering is based on the utilization of tectons to influence desirable properties of materials. A supramolecular synthon is the result of the combination of tectons —that is, the synthon is the pattern formed by interacting tectons. A supramolecular synthon that favors one enantiomer over another is called a supramolecular chiron. It was defined by Nangia as “the minimal homo- or heterochiral molecular unit or an ensemble capable of generating ordered superstructures by self-assembly through hydrogen bonding or other noncovalent forces, and leading to topologically distinct enantio- or diastereopure architectures.”134, 135

5.70

Thermosalient effect

The thermosalient effect describes the situation whereby crystals jump or hop during a thermally induced polymorphic phase transition. The phenomenon is observable on a macroscopic level and occurs in both inorganic and organic crystals. The mechanism of hopping or jumping has not yet been elucidated.48

5.71

Thermotropism

A literature survey reveals that the term thermotropism is mostly used to describe temperature-dependent ordering of LCs —usually referring to phase changes that occur between the temperature below which the LC crystallizes and the temperature above which it becomes an isotropic liquid. Since the study of crystals is a separate (but somewhat related) field, the term can easily be borrowed to describe thermally induced polymorphic transitions in crystals.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

30

5.72

Supramolecular materials chemistry

Topotactic (topochemical)

Topotactic processes are described more comprehensively in Mechanical Preparation of Crystalline Materials. An Oxymoron? and Templated [2 + 2] Photodimerizations in the Solid State, Supramolecular Materials Chemistry. Since its original definition in 1959,136 the term topotaxy has evolved137–139 to mean a solid-state chemical reaction (not necessarily reversible) that completely transforms a single crystal such that the product resembles that of the initial material both in terms of its three-dimensional structure and the orientation thereof. Topotactic is synonymous with topochemical.140

5.73

Transparent, translucent, and opaque

These three terms are often used to describe (albeit subjectively) the degree of optical clarity of a crystal. A transparent crystal allows a significant proportion of light to pass through such that objects on the other side can still be discerned clearly. A translucent crystal transmits some light but much of the light is diffused by the crystal, thus preventing objects on the other side from being clearly discernable. An opaque (or turbid ) crystal allows little or no light to pass through.

5.74

Twinning

A twinned crystal is a regular aggregate in which different domains of the same species are joined together (Figure 21) according to a specific symmetry operation, that is, the twin law.54, 97 The twin law is a description of the orientation of the different domains relative to each other and the fractional contribution of each component. There are four classifications of twinning, namely merohedral, pseudomerohedral, reticular merohedral, and nonmerohedral twinning.141, 142 The reciprocal lattices of merohedral twins superimpose exactly and the presence of twinning is not immediately obvious from the diffraction pattern. The twin law is a symmetry operator of the Laue group but not of the point group of the crystal structure. Merohedral twins are possible in the low-symmetry tetragonal, trigonal, hexagonal, and cubic systems. Also see twinning by inversion. The reciprocal lattices of pseudomerohedral twins superimpose almost exactly, depending on how well the higher metric symmetry is fulfilled. This occurs when the metric symmetry is higher than the symmetry of the crystal structure. The twin law belongs to a higher crystal system than the structure. In the case of a monoclinic crystal, where β ≈ 90◦ , the combined lattice appears orthorhombic. The

Figure 21 Twinned Museum, London).

calcite

prisms

(Natural

History

lattice possesses twofold symmetry about the a and c directions, which is not present in the point group of the crystal structure. In the case of reticular merohedral twins, the reciprocal lattices do not overlap exactly and only some of the reflections are affected by the twinning. For nonmerohedral twins, only certain zones of data are affected by overlap. The diffraction patterns of nonmerohedrally twinned crystals are difficult to index since the individual reflections may result from different domains of the twin. The twin law is usually a symmetry operation belonging to a higher symmetry supercell. An example of nonmerohedral twinning is an orthorhombic crystal where a ≈ 2b. A metrically tetragonal supercell can be obtained by doubling the length of b so that there is a pseudofourfold axis about c.

5.75

Voids and porosity

Solid-state inclusion chemistry concerns the formation of host–guest compounds, their characterization by means of an extensive arsenal of analytical techniques, and their description. In most cases, guest molecules occupy lattice sites in crystals and it is therefore possible to obtain accurate crystallographic information regarding their locations within the guest-accessible spaces formed by the host framework. These spaces may consist of cavities or channels, and a number of computer programs are now able to map the surfaces that define the boundaries of the guest-accessible volume. Channels can be one-, two-, or three-dimensional (two- and three-dimensional channels

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

Concepts and nomenclature in chemical crystallography interconnect to form networks of channels). Conceptually, channels connect the interior of the crystal to its surroundings—for example, in an idealized (i.e., defect-free) crystal, a one-dimensional channel begins at one extremity of the crystal and terminates at the opposite extremity, where it once again opens up to the surrounding atmosphere. When the guest-accessible regions within the crystal are not conceptually continuous in this manner, they can be referred to as cavities or guest pockets. By definition, a void is an empty space and we therefore recommend only using the term void to describe a cavity from which the guest molecules have been removed. A pore is an opening and the term pore should not be used for channel, cavity, or pocket, but rather the window between adjacent pockets that are interconnected. Porosity is a topical area within the broader discipline of crystal engineering, and it has been discussed at length. A porous material is one that allows the transport of another substance through its bulk in a dynamic fashion. This passage of matter could be either into, out of, or within the material. Permanent porosity exists if the guest can be removed without collapse (i.e., rearrangement to yield close packing) of the host framework in its evacuated state. The abstraction of solvent molecules from a solvate is typically accompanied by such collapse of the host framework to yield a more closely packed arrangement of the host molecules. Although this process clearly involves the passage of one substance through another, referring to the host in such cases as being porous significantly diminishes the utility and impact of the term. Indeed, it can be argued that the host does not have a coherent structure during the process of its collapse and concomitant reorganization, and that the term porous does not refer to a particular solid-state phase in this case. It is also possible to delete atomic coordinates of guest molecules in silico and to then depict the host framework in its hypothetically “porous” state—this has been referred to as virtual porosity and should be discouraged as a de facto demonstration of permeability. Such images are useful to describe the structure of the host framework or to illustrate the space available to the guest, but it should be made clear to the reader that this does not imply porosity unless supported by evidence of guest transport. Calculations of guest-accessible volumes generally consider the host atoms to be immobilized and spherical, with van der Waals radii simply assigned according to the type of element with little consideration given to bonding environments. In some cases, the distinction between cavities (which are discrete) and channels (which are continuous) can be subjective. Guest-accessible space is generally assessed as the volume available to a spherical probe, the radius of which can have a significant effect on the shape and volume of the mapped space. It should also be

31

noted that crystal structure determinations provide static atomic coordinates together with a semiquantitative assessment of atomic displacements. Furthermore, recent reports have shown that some porous frameworks are highly flexible, and in such cases static atomic coordinates also provide poor estimates of the space available to guest molecules during uptake or release. The IUPAC defines a porous solid as “any solid material that contains cavities, channels, or interstices,” but goes on to state that “in a particular context, a more restrictive definition may be appropriate.”3 Nanoporous refers to a material consisting of a regular framework having pores with widths in the range 2 to 1000 nm. These materials can further be subdivided into three categories: Microporous Mesoporous Macroporous

50 nm

5.76 Z, Z
, Z

, and Z r Simply stated, the parameter Z is the number of formula units in the unit cell, Z  is the number of formula units in the asymmetric unit, and Z  is the total number of molecules in the asymmetric unit. Since crystals do not always consist of discrete molecules and the assignment of the formula unit can sometimes be subjective, a considerable amount of confusion may arise when discussing the different variants of Z. Although the formula unit is left to the discretion of the crystallographer, the most sensible guideline that we can offer is to ensure that the formula unit does not contain fractions of either molecules or atoms, unless there is a good reason to include fractional components. The parameters Z  and Z  are useful when discussing issues that influence packing (e.g., awkwardness) and the interested reader is referred to a comprehensive web site on Z  .143 The number of chemically distinct residues (this is relevant to cocrystals and solvates/hydrates) in a structure is given as Z r .144

5.77

Zeolite

The term zeolite means boiling stone and is derived from the Greek words zeo (to boil) and lithos (stone). It was introduced by the Swedish mineralogist Cr¨onsted after observing that a mineral that he was studying released water upon heating. Zeolites have the structural formula Mx/n [(AlO2 − )x (SiO2 )y ]

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc108

32

Supramolecular materials chemistry

Figure 22

AFI-type zeolite framework.

where n is the valence of cation M. The value (x + y) gives the total number of tetrahedra (AlO4 and SiO4 ) per unit cell and y/x is the Si/Al atomic ratio, which is always greater than or equal to 1. The negative charge carried on the aluminum units gives rise to L¨owenstein’s rule, which states that Al–O–Al linkages are forbidden. The overall charge is balanced by cations that reside as guests within the pores, alongside neutral species. These inorganic aluminosilicates have continuous channels rather than discrete cavities in which guest molecules can be accommodated. They are similar to tabulate structures, although the inorganic host lattice is much more robust. Zeolites are microporous (pore diameter 1) or differences between bonds or angles that are symmetrically related in the isolated molecule, but where the molecular symmetry does not coincide with crystallographic symmetry.12 Such investigations demonstrate that the influence of crystal packing on bond lengths is very small and bond angles are typically affected by less than a few degrees. Computational studies, comparing optimized molecular geometries of the isolated molecule and in the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

4

Supramolecular materials chemistry

crystal provide a similar picture.13–15 Therefore, it is usually safe to keep bond lengths fixed at those calculated for the isolated molecule. The same can be said for bond angles, apart from some special cases where a few degrees of change in an angle leads to an important geometrical distortion of a strong intermolecular interaction (such as a hydrogen bond) or where a small change in an angle in the middle of an extended molecule leads to large relative displacements of the extremities of the molecule, changing its overall shape and close packing. Changes in torsion angles, on the other hand, can often be made at a low intramolecular energy cost, such that rotation about single bonds, changes in the puckering of rings and pyramidalization of atoms away from the isolated molecular geometry is often possible and must be considered during the calculations. QM calculations on molecular conformations found in known crystal structures suggest that
Uintra is usually less than about 4 kJ mol−1 ,16 which can sometimes correspond to 20–30◦ rotation about an exocyclic single bond. Biphenyl is a classic example: the isolated molecule is twisted, with an angle of approximately 45◦ between phenyl rings. However, the planar molecular geometry leads to better crystal packing, so that the phenyl rings are only slightly twisted in the low-temperature phase and are coplanar in the room temperature crystal structure.17–20 Clearly, such distortions of the molecular geometry must be considered when generating crystal structures, and during the lattice energy minimization of each trial structure. The final stability ranking of crystal structures must then account for the differences in Uintra as well as Uinter . Methods of accounting for molecular flexibility are discussed in later sections dealing with the crystal structure generation and lattice energy minimization methods.

2.1.2 Multiple conformations A second consideration is the presence of multiple stable conformations of the isolated molecule, where each conformation corresponds to a different local minimum on the intramolecular energy surface. Each of these molecular geometries might lead to stable crystal structures. One strategy in such cases is to obtain energy-minimized molecular structures for each conformation and then to perform the rest of the global lattice energy minimization procedure separately starting with each conformation (Figure 1).21, 22 For small molecules with few flexible degrees of freedom, it should be possible to produce an exhaustive set of stable conformations through systematic conformational energy scans with respect to torsion angles about the relevant bonds. If the resulting number of conformations is manageable, crystal structures can then be generated with each conformation. As above, the final evaluation

of the relative energies of predicted crystal structures must then account for differences in the intramolecular energies of the conformations in addition to the relative intermolecular energies. While the use of all local-energy-minimized molecular conformations as starting points for crystal structure searches might be possible for small molecules, the number of possible conformations can increase rapidly with the number of intramolecular degrees of freedom and rapidly becomes unmanageable for large molecules. The problem is compounded if more than one flexible molecule is present in the asymmetric unit (i.e., multicomponent crystals or Z
> 1), in which case all combinations of conformations would have to be considered and these combinations can run into the hundreds, each producing 104 or more trial crystal structures.23 For this reason, few CSP studies have been reported on molecules with more than four to five intramolecular degrees of freedom and predicting the structures of cocrystals or salts of flexible molecules is a particular challenge, at least in terms of computing resources and data management. The observed conformations in crystal structures of similar molecules can help assess which molecular conformations are likely to lead to the low-energy crystal structures. The number of conformations that need to be considered could also be reduced on the basis of their calculated energies, assuming that those above a certain Uintra will be unable to make up for their high energy through improved intermolecular interaction. The problem is in defining a threshold energy that will not rule out relevant conformations, and this choice will depend on the types of possible inter- and intramolecular interactions. Where a flexible molecule has the opportunity to replace an intramolecular hydrogen bond with an intermolecular hydrogen bond through a change in conformation, the difference in Uintra between two structures with similar total crystal energies will be in the region of the strength of a hydrogen bond, possibly as high as 30–40 kJ mol−1 . The polymorphism of ortho-acetamidobenzamide, which is discussed in more detail below, illustrates a case of large conformational energy differences between crystal structures. In the absence of strong noncovalent intramolecular interactions, the range in Uintra that must be considered will be considerably smaller.

2.2

Generating crystal structure: global searches of the lattice energy surface

Numerous methods have been developed for the purpose of searching the lattice energy surface for all local energy minima, corresponding to the stable arrangements of the molecule into a crystal lattice.24 For any problem involving

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

Crystal structure prediction a global search for local minima, an important consideration is the dimensionality of the energy surface: here, this is the number of variables required to describe the crystal structures. The translational symmetry in the crystal is determined by the dimensions and shape of the unit cell, for which six variables are generally required: the cell lengths a, b, and c, and angles α, β, and γ . In a given unit cell, the crystal structure is then determined by the fractional coordinates of the atoms. Because we are dealing with molecular crystals and the molecular structure has been calculated, only six variables are required to describe how each molecule is positioned within the unit cell: three variables defining the coordinates of the center of mass and three angles defining the molecular orientation with respect to some global reference frame. For Z molecules in the unit cell, this results in 6Z + 6 structural variables. Because of the translational symmetry in the crystal, three combinations of the molecular translational coordinates result in bulk translation within the crystal and do not affect its internal structure or energy. These three coordinates can be fixed, say by fixing the translational position of one molecule, reducing the dimensionality of the problem to 6Z + 3 independent variables. The general strategy is now to generate trial crystal structures by sampling combinations of these variables, looking for those that lead to structures with the lowest energies.

2.2.1 Use of space group statistics It is possible to try to search the (6Z + 3)-dimensional lattice energy landscape for all crystal structures corresponding to local minima. Williams called this the P1 method 25 because all crystal structures are initially generated with no symmetry imposed. Searches must, in principle, be repeated with all possible values of Z, the number of molecules in the unit cell; most observed crystal structures occur with Z ≤ 8.26 The final space group symmetry of each structure can be determined after energy minimization. The high dimensionality of the search space makes this is a computationally very demanding task and there is a high risk of failing to locate important low-energy structures. To simplify the problem, most methods make use of space group symmetry during the search. Having chosen a space group to search within, only the molecules in the asymmetric unit need to be positioned independently and the other molecules in the unit cell are generated by symmetry operators. To avoid ambiguity and allow generalization to structures with more than one molecule in the formula unit, van Eijck and Kroon introduced the notation Z

as the number of crystallographically nonequivalent molecules in a crystal structure and G as the number of independent molecules considered in a crystal structure search.8 (The distinction between G and Z

is

5

that a search with G > 1 where two or more molecules are chemically identical will locate crystal structures with Z

= G and structures with Z

< G, where the symmetry of the lattice energy minimized structure is higher than that of the space group assumed when generating structures.) Since G ≤ Z, the dimensionality of the search space is usually reduced relative to the P 1 method. Furthermore, in all but triclinic space groups, symmetry also constrains some of the six unit cell parameters, such as angles fixed at 90◦ and equivalent unit cell lengths; the number of structural variables decreases to 6G + X, where X is between 1 and 6, depending on the crystal system. In principle, the space group constrained method requires the search to be repeated in all space groups. Z
< 1 structures (where the asymmetric unit is a fraction of a molecule) should be located in searches with G = 1 when molecules find special positions during energy minimization, but separate searches are required for Z
> 1. The approach is helped by the observation that molecules do not populate the 230 space groups equally (80% of homomolecular organic crystal structures are found in the six most popular space groups with Z
≤ 127 ) and crystal structure searches are usually confined to the most popular space groups. The number of space groups to include depends on the acceptability of the risk of missing an observed crystal structure in a rare space group, as well as the computing resources that are available for the search. As discussed in the previous section, soft intramolecular degrees of freedom (usually rotation about acyclic single bonds) must sometimes be considered in the generation and energy minimization of crystal structures. The approach described in Section 2.1.2 for molecules with multiple stable conformations separates the intramolecular variables from the 6Z + 3 that define the crystal packing of the molecule by first considering the isolated molecule, selecting a set of conformations, and then performing separate searches with each. Alternatively, internal degrees of freedom can be treated along with the 6G + X (or 6Z + 3) crystal packing variables, varying the molecular structure simultaneously with the bulk packing variables.

2.2.2 Search methods A conceptually simple approach to generate trial crystal structures is to sample the structural variables at equally spaced intervals within a chosen range of values (say ˚ producing a set of cell lengths in the range 4–40 A), structures on a grid across the lattice energy surface. The grid-based approach illustrates the increasing difficulty of fully sampling structural space as the number of variables required to describe the crystal structure is increased. Take, for example, a grid involving M values for each of D structural variables. The total number of crystal structures

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

6

Supramolecular materials chemistry

that are generated by such a grid is M D ; if space group symmetry is not used, a search for crystal structures with four molecules in the unit cell would require the construction of M (3+6×4) = M 27 structures. As an example of using space group symmetry, a grid search for P 21 /c structures with G = 1 results in M 10 structures: even without using space group probabilities to reduce the search space, searching all 230 space groups in turn is an easier task than the P 1 method with a few values of Z. However, a grid method is still only feasible for fairly coarse sampling, M, and low dimensionality D. An alternative method is to randomly assign values to each variable (within reasonably chosen ranges)8, 28, 29 and, more recently, quasi-random sequences that are designed to provide a more uniform distribution of points than random sampling.30, 31 An advantage of such Monte Carlo (or quasiMonte Carlo) methods is that the search can start with a lower number of trial structures than the systematic grid search and the completeness of the search can be monitored, incrementally increasing the number of random points until a satisfactory sampling of structures is achieved. If the randomly generated structures are lattice energy minimized on the fly, the completeness can be monitored by keeping track of the number of distinct minima found as the number of randomly generated structures is increased, and the number of times each local minimum is reached. It has been observed that the number of times that a particular structure is encountered tends to decrease with increasing lattice energy, indicating that the deepest minima tend to be broader than higher energy minima.8 In general, this benefits the structure-generation procedure, since it is these low-energy structures that are ordinarily of most interest. If any low-energy structures are located only once, this indicates that the spacing between initial trial structures is similar to the breadth of the minima, which has the risk that local minima will be missed. The Monte Carlo approach has proven to be fairly effective at structure generation; tests using from 5000 to 50 000 randomly generated trial structures per space group found that this approach is effective for problems with dimensionality up to about D = 20.8 For rigid molecules, this approach should therefore be applicable to G = 1 or 2 in all crystal systems (G = 2, such as a homomolecular Z
= 2 structure or a 1 : 1 salt, in a triclinic space group has D = 18). The search is amenable to parallel computing and the generation of these quantities of structures on a computing cluster is feasible on the timescale of days. Higher-dimensional problems, such as G > 2 or very flexible molecules, might require more efficient algorithms to provide complete sets of structures in a reasonable time. More sophisticated approaches that have been implemented include genetic algorithms,32–34 Monte Carlo simulated annealing35 and conformation-family Monte Carlo.36 These

algorithms use some fitness function, usually the calculated lattice energy of structures, to guide the search toward regions of the energy surface that produce the most stable structures. Apart from the above methods, where the crystal packing problem is approached almost purely as a mathematical problem in global minimization, some methods have been developed using more chemical understanding. Gavezzotti developed the PROMET approach,37 which starts by building molecular clusters around the symmetry operators from common space groups, selecting the best of these to build into full crystal structures. The process seems closer to the real process of crystal nucleation and growth—only the structures with the most stable “nuclei” survive to be built into fully periodic crystal structures. Holden and Ammon’s MOLPAK method38 enumerates commonly observed packing types—the arrangement of molecules in a coordination sphere around a reference molecule in a crystal structure. Crystal structures are generated in each packing group, within which orientational degrees of freedom of the molecules must be sampled. Overall, a large variety of methods have been developed for the purpose of generating crystal structures. The choice of method might depend on the type of system being studied and the nature of computing resources available for the calculations. Nevertheless, the final aim of all of these is to produce an exhaustive list of the low-energy crystal packing possibilities. Assuming that the search has achieved this goal, the next stage is to assess the structures based on their calculated energies.

2.3

Lattice energy calculations

There are two requirements of the model used to calculate the energies of crystal structures in a CSP study. First, the position of the equilibrium points on the energy surface must be a good representation of their positions on the true energy surface. Where a crystal structure is known, the best way to judge this is to lattice energy minimize the known crystal structure using the same model that would be used during CSP. The resulting energy-minimized version of the known crystal structure is the closest that CSP could predict. Therefore, if the geometrical distortion upon energy minimization is so large that the structure is not recognizable, then CSP would be pointless. If no crystal structure is known, then, prior to performing CSP calculations, the energy model can only be judged on how well crystal structures of similar molecules are reproduced. A second requirement of the model is to provide accurate enough energies to correctly order the stability of the computer-generated crystal structures. Unfortunately, one recurring aspect of CSP studies on molecular organic

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

Crystal structure prediction P 21/c (Z′ = 1), global minimum

P1 (Z ′ = 2), +1.54 kJ mol−1

(b)

(a)

P 212121 (Z′ = 1), +2.47 kJ mol−1

Pbca (Z′ = 1), +2.69 kJ mol−1

(d)

(c)

Increasing lattice energy

Figure 2 Packing diagrams of four of the low-energy-predicted crystal structures of phenobarbital, resulting from CSP searches with G = 1, 2, and 3. Hydrogen bonds are indicated as dashed green lines. Structures are taken from the supplementary information of Ref. 22. (Reproduced from Ref. 22.  Royal Society of Chemistry, 2007.)

Subset from initial structure generation

crystals is that, for a given molecule, many crystal packing possibilities exist within a very small range in lattice energy. As an example, Figure 2 shows a selection of crystal structures of phenobarbital resulting from CSP searches with G = 1, 2, and 3 independent molecules.22 The illustrated crystal structures are a selection of 23 distinct crystal structures that are found within 3 kJ mol−1 of the global minimum (Figure 2a) and a further 49 possible structures located in the next 2 kJ mol−1 interval; that is, a total of 72 structures separated by less than 5 kJ mol−1 . In this case, the search with G = 2 and 3 was recognized to be incomplete; therefore, more structures exist in this small energy range. This is not an unusual example; for small molecules, it is fairly common to find more than 100 crystal packing possibilities within 5 kJ mol−1 of the global minimum.39 This finding is one very interesting aspect of CSP—most molecules seem to have many options for the final product of crystallization that are nearly equienergetic. Even very different arrangements of hydrogen bonds are often scarcely distinguishable in terms of lattice energy (Figure 2). While the multitude of possible structures can be fascinating from the point of view of structural analysis, it is a severe obstacle to the goal of reliable CSP. Is it possible to calculate the relative energies of so many crystal structures to within the required accuracy on a timescale that can usefully be employed in a research program? Many types of energy calculations have been applied in CSP studies, starting from the simplest atom–atom descriptions of interactions. While fairly simple “brute force” (grid and random structure generation) approaches are often successful at addressing the crystal structure generation problem, the assessment of relative energies requires the development of sophisticated methods to achieve the required accuracy for CSP. One approach has been the development of more elaborate, anisotropic (nonspherical) atom–atom models.40 Recently, the application of QM solid-state electronic structure calculations has also been demonstrated to be very successful.41, 42 Because of the increased computational cost required for the methods that have been found to be most successful at correctly ranking predicted crystal structures, a hierarchical approach is often required (Figure 3). Lattice energy minimization of a large number of trial structures (usually 104 –106 , depending on the complexity of the system) is required during the structure generation process and this can only be achieved using fairly simple models to evaluate the energy of each structure. Since this energy model is usually not accurate enough for the final ranking, the lowest-energy subset of generated structures is then taken forward to the next stage—lattice energy minimization with a higher quality, but computationally more demanding model for the energy. The first model must be sufficiently accurate that all relevant structures fall within the energy window used to

7

Fast, lower quality energy model

High quality energy model

Re-minimization

Figure 3 A hierarchical strategy for crystal structure prediction. Each horizontal line indicates the lattice energy of a computergenerated crystal structure. The lowest energy subset of structures generated using a faster, lower quality energy model are reminimized using a higher quality, but slower, model to provide the final ranking of predicted crystal structures.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

8

Supramolecular materials chemistry

define the subset of structures that is passed onto the higherquality model. Furthermore, the initial model should not be so simplified that the lattice energy surface that it describes lacks any of the local minima that exist on the real energy surface, or else the corresponding structures will be missing from the list of generated structures. To make sensible choices for the energy models used throughout the calculations, we now examine the commonly adopted approaches to calculating lattice energies.

2.3.1 Atom–atom models Kitaigorodskii developed many of the principles of close packing in molecular crystals,12, 43 and the initial quantification of these ideas involved the use of additive volume increments for functional groups and atoms in molecules, using average intermolecular atomic radii to determine the closest contacts between molecules. Such concepts can be used to quantify the efficiency of close packing (via packing coefficients) and atom–atom distances are often calculated as a means of looking for the important intermolecular interactions in a crystal. The description of molecules in crystals as a superposition of spherical atoms also leads naturally to the atom–atom approach to calculating lattice energies. The atom–atom method has been crucial to much of the development of our understanding of the interactions in crystals and how these interactions sum up to give total lattice energies, relative stabilities of different crystal structures, and crystal properties. As a first approximation, the total intermolecular interaction energy of a set of molecules can be written as the sum of pairwise molecule–molecule interactions, UMN (M and N labeling the molecules involved). In turn, each molecule–molecule interaction is calculated as a sum of interactions, Uik , between their constituent atoms: Uintermolecular =

N mol
M 1 searches had greater difficulties in ensuring completeness of the search. These problems with two independent molecules were encountered again for the cocrystal (XV) included in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

18

Supramolecular materials chemistry

Table 1 Rates of success in the first four blind tests of crystal structure prediction. Molecule numbers refer to the numbering used in the blind test publications and Figure 6. yr

1999

1999 1999 1999 2001 2001 2001 2004 2004 2004 2004 2007 2007 2007 2007

Category

Target

1

I stable form metastable forma II III VII IVd V VI VIIIe XIe IX X XII XIII XIV XV

2 3 Extra 1 2 3 1 1 2 3 1 2 3 4

Space group of observed structure, Z
= 1 unless indicated

Participating research groups

Number of groups who generated the observed structure somewhere in their search

Successes within the three allowed predictions

Successes as the first predicted structure

—b —

0 4

0 3

—b —b —b 10 (of 14)c 11 (of 14)c 4 (of 10)c 11 4 9 7 10 9 9 5

1 1 1 3 4 0 4 0 1 0 4 4 3 2

0 1 1 1 3 0 3 0 1 0 2 4 3 1

11 P 21 /c Pbca P 21 /c P 21 /c P 21 /c P 21 /c P 21 21 21 P 21 /c C2/c P 21 /c, Z
= 2 P 21 /c, Z
= 1/2 P 21 /c Pbca P 21 /c P 21 /c P 21 /c

8 11 6 16 15 11 14 18 15 15 13 14 12 12

a

Molecule I was polymorphic. 1999 blind test results were not analyzed in enough detail to separate search failures from failures in the ranking of structures. c In the 2001 blind test, some participants shared lists of generated crystal structures, energy minimizing and re-ranking them separately for the final predictions. Therefore, the number of independent lists of structures was smaller than the number of participants. d The results for one of the participants are published separately from the CSP2001 publication (Ref. 72). These participants were the only ones to rank the observed crystal structure as their first prediction. e A partial structure report for molecule VIII was discovered part-way through the 2004 blind test, so this target might not be considered a true blind test. Most participants continued their calculations without using the information contained in this structure report. XI was introduced as a replacement molecule mid-way through the blind test. b The

the 2007 blind test and less than half of the groups that attempted predictions located the observed structure somewhere in their sets of structures. For the cocrystal, this problem was addressed by some participants by using expected hydrogen bond dimers as the search unit for CSP, reducing the degrees of freedom to the unit cell, orientation, and position of the dimer. This can be a useful approach when reliable dimer geometries can be identified, but is not a general solution for all types of multicomponent crystal structures. Until the 2007 blind test, flexible molecules also presented problems for the search methods; success rates for generating the observed crystal structure were less than 50% for molecules VI and X. The problems are a combined effect of the additional degrees of freedom and the poor performance of energy models at assessing the energies of structures containing different molecular geometries. The 2007 blind test saw an improvement in the flexible molecule, with most groups locating the observed structure in their lists. While this is promising, it was also found that the conformational flexibility of molecule XIV was very limited, and only small variations about a single

conformation had to be considered. Therefore, the improved success is not an indication that problems with flexible molecules are solved. These blind test results have not only identified some important limitations of the search methods that are in use for CSP but have also highlighted some methods that have been found to be reliable across the series of tests. The third and fourth blind tests highlighted structure generation methods based on simulated annealing and random or quasi-random sampling as being most consistently successful.41, 83

3.2

Overall prediction success in the blind tests

When it comes to the final choice of structures, the best that could be said after the first three blind tests was that crystal structures were occasionally successfully predicted. Some rigid molecules were predicted successfully by several groups; flexible molecules were a significant problem (only one successful prediction in category 3 during the first three

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

Crystal structure prediction blind tests) and no single method consistently led to correct predictions. The results of the fourth blind test, in 2007,41 were a significant improvement. Thirteen successful predictions were made over the four categories, including two correct predictions for the cocrystal (XV) and three correct predictions for the flexible molecule XIV. It is important to note that all of the successful predictions in this blind test were achieved using more sophisticated methods than the traditional isotropic atom–atom approach. For several groups, this involved an anisotropic treatment of atom–atom interactions, in either or both the electrostatic and repulsion terms. The most notable result was the first use of the DFTD method for ranking structures in a blind test, and this led to correct predictions for all four targets, each as the lowest energy structure. This is the first time that one method has achieved this level of success in a blind test and, while the calculations required approximately two orders of magnitude more computing time than most atom–atom-based methods, the results demonstrate that this is a very promising approach to CSP. The blind tests are a valuable exercise for measuring progress in CSP. However, there are dangers in treating these tests as the sole measure of developments in the field. Despite consistency in some of the target categories, there is inevitable variability in the difficulty of targets between blind tests. It is impossible to select target molecules with a truly consistent level of difficulty because the specific problems involved in predicting the crystal structure of each molecule are usually not clear until the calculations are underway, or until the blind test has finished. Therefore, year-to-year changes in the success rates for the necessarily small set of target structures can reflect specific difficulties with certain molecules as much as developments in the methods. One of the most important roles that the blind tests serve is to highlight the causes of failed predictions and the limitations of individual methods and, in this way, to push forward the development of more reliable methods. The measure of success, that the observed crystal structure must be one of the best three predictions, is a tough criterion and encourages the continued development of existing methods, as well as completely new approaches. However, the level of certainty required for blind test success is not necessary for all applications of CSP in materials research and crystal engineering.

4

CONCLUSIONS

The aims of CSP have motivated the development of a range of powerful methods for generating the possible crystal structures that a molecule can adopt and for assessing these possibilities to find the most likely structures

19

to result from an actual crystallization of that molecule. The goal of reliably successful prediction of crystal structures will require robust methods for both the generation and assessment of crystal structures. The intelligent use of these calculations requires an understanding of the underlying methods and their limitations. The current methods that are being applied to this problem have been reviewed here with the aim of highlighting current capabilities of the methods. These methods are developing rapidly and CSP methods are already at a stage where they can play an important role in guiding experiments. For small molecules, the structure generation problem is almost solved using random or quasi-random searches of the energy landscape and only unusual structures, such as those with high Z
, should be problematic. The extension to very flexible molecules and multicomponent crystals of flexible molecules might reach the limit of these simple methods and require more advanced approaches. The ranking of structures is very sensitive to the energy model that is used and most results point to the need for more sophisticated methods than standard transferable force fields. Here, the most promising results are based on anisotropic atom–atom methods for intermolecular interactions, coupled with quantum mechanical molecular energy calculations for flexible molecules, and fully quantum mechanical solid-state calculations, currently based on DFT. The progress that has been made over the past few years is reflected in the results of the blind tests. Judging from the results of these experiments, developments in the methods have taken us from the occasional successful prediction to more reliable rates of successful prediction. The successes in the most recent blind test justify the global energy minimization approach to CSP. There is a still a long way to go in the quest for generally applicable methods that are not limited to the classes of small molecules that are represented in these blind tests, but the recent rate of development is very encouraging. Indeed, at the time of writing of this chapter, a fifth blind test of CSP is underway, which includes several more complex target systems, including a molecular salt, hydrate and a very flexible molecule with seven to eight rotatable single bonds; the inclusion of such targets demonstrates the developing confidence that such methods are becoming widely applicable.

ACKNOWLEDGMENTS I thank all of my colleagues who have contributed to the development and application of methods used in crystal structure prediction and the Royal Society for current funding of a University Research Fellowship.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

20

Supramolecular materials chemistry

REFERENCES

30. P. G. Karamertzanis and C. C. Pantelides, J. Comput. Chem., 2005, 26, 304.

1. M. U. Schmidt, M. Ermrich, and R. E. Dinnebier, Acta Cryst., 2005, B61, 37.

31. R. G. Della Valle, E. Venuti, A. Brillante, and A. Girlando, J. Chem. Phys., 2003, 118, 807.

2. A. D. Bond and W. Jones, Acta Cryst., 2002, B58, 233.

32. W. D. S. Motherwell, Mol. Cryst. Liq. Cryst., 2001, 356, 559.

3. A. J. Cruz-Cabeza, S. Karki, L. Fabian, et al., Chem. Commun., 2010, 46, 2224. 4. J. K. Harper and D. M. Grant, Cryst. Growth Des., 2006, 6, 2315. 5. E. Salager, G. M. Day, R. S. Stein, et al., J. Am. Chem. Soc., 2010, 132, 2564.

33. V. E. Bazterra, M. B. Ferraro, and J. C. Facelli, J. Chem. Phys., 2002, 116, 5984. 34. S. Kim, A. M. Orendt, M. B. Ferraro, and J. C. Facelli, J. Comput. Chem., 2009, 30, 1973. 35. R. J. Gdanitz, Chem. Phys. Letters, 1992, 190, 391.

6. S. R. Chemburkar, J. Bauer, K. Deming, et al., Org. Process Res. Dev., 2000, 4, 413.

36. J. Pillardy, Y. A. Arnautova, C. Czaplewski, et al., Proc. Nat. Acad. Sci. U. S. A., 2001, 98, 12351.

7. J. Bauer, S. Spanton, R. Henry, et al., Pharmaceut. Res., 2001, 18, 859.

37. A. Gavezzotti, J. Am. Chem. Soc., 1991, 113, 4622.

8. B. P. van Eijck and J. Kroon, Acta Cryst., 2000, B56, 535. 9. F. H. Allen, Acta Cryst., 2002, B58, 380. 10. F. H. Allen and W. D. S. Motherwell, Acta Crystallogr. B, 2002, 58, 407. 11. T. Beyer and S. L. Price, CrystEngComm, 2000, 2, 183. 12. A. I. Kitaigorodskii, Molecular Crystals and Molecules, Academic Press, New York and London, 1973, vol. 29. 13. A. Warshel, E. Huler, D. Rabinovich, and Z. Shakked, J. Molec. Struct., 1974, 23, 175. 14. M. Colapietro, A. Domenicano, G. Portalone, et al., J. Phys. Chem., 1987, 91, 1728.

38. J. R. Holden, Z. Y. Du, and H. L. Ammon, J. Comput. Chem, 1993, 14, 422. 39. G. M. Day, J. Chisholm, N. Shan, et al., Cryst. Growth Des., 2004, 4, 1327. 40. S. L. Price, CrystEngComm, 2004, 6, 344. 41. G. M. Day, T. G. Cooper, A. J. Cruz-Cabeza, et al., Acta Cryst., 2009, B65, 107. 42. M. A. Neumann, F. J. J. Leusen, and J. Kendrick, Angew. Chem. Int. Ed., 2008, 47, 2427. 43. A. I. Kitaigorodskii, Organic Chemical Crystallography, Consultants Bureau, New York, 1961. 44. D. E. Williams, J. Comput. Chem., 2001, 22, 1.

15. N. J. Harris and K. Lammertsma, J. Am. Chem. Soc., 1997, 119, 6583.

45. D. E. Williams, J. Comput. Chem., 2001, 22, 1154.

16. F. H. Allen, S. E. Harris, and R. Taylor, J. Comput.-Aided Mol. Des., 1996, 10, 247.

47. A. J. Stone, and A. J. Misquitta, Int. Rev. Phys. Chem., 2007, 26, 193.

17. G. Casalone, C. Mariani, A. Mugnoli, and M. Simonetta, Mol. Phys., 1968, 15, 339.

48. G. M. Day and S. L. Price, J. Am. Chem. Soc., 2003, 125, 16434.

18. A. Almenningen, O. Bastiansen, L. Fernholt, et al., J. Mol. Struct., 1985, 128, 59.

49. J. B. O. Mitchell and S. L. Price, J. Phys. Chem. A, 2000, 104, 10958.

19. J. L. Baudour and M. Sanquer, Acta Cryst. B, 1983, 39, 75.

50. W. T. M. Mooij, B. P. van Eijck, and J. Kroon, J. Phys. Chem. A, 1999, 103, 9883.

20. A. Dzyabchenko and H. A. Scheraga, Acta Cryst. B, 2004, 60, 228. 21. C. Ouvrard and S. L. Price, Cryst. Growth Des., 2004, 4, 1119.

46. D. E. Williams, J. Mol. Struct., 1999, 486, 321.

51. P. G. Karamertzanis and C. C. Pantelides, Mol. Simul., 2004, 30, 413. 52. A. J. Stone and M. Alderton, Mol. Phys., 1985, 56, 1047.

22. G. M. Day, W. D. S. Motherwell, and W. Jones, Phys. Chem. Chem. Phys., 2007, 9, 1693.

53. S. W. Brodersen, F. J. J. Leusen, and G. Engel, Phys. Chem. Chem. Phys., 2003, 5, 4923.

23. C. H. G¨orbitz, B. Dalhus, and G. M. Day, Phys. Chem. Chem. Phys., 2010, 12, 8466.

54. A. D. Buckingham and P. W. Fowler, Can. J. Chem., 1985, 63, 2018.

24. P. Verwer and F. J. J. Leusen, Rev. Comput. Chem., 1998, 12, 327.

55. G. M. Day, W. D. S. Motherwell, and W. Jones, Cryst. Growth Des., 2005, 5, 1023.

25. D. Gao and D. E. Williamas, Acta Crystallogr. A, 1999, 55, 621.

56. W. T. M. Mooij, B. P. van Eijck, and J. Kroon, J. Am. Chem. Soc., 2000, 122, 3500.

26. C. P. Brock and J. D. Dunitz, Chem. Mater., 1994, 6, 1118.

57. B. P. van Eijck, W. T. M. Mooij, and J. Kroon, J. Phys. Chem. B, 2001, 105, 10573.

27. C. P. Brock and J. D. Dunitz, Acta Crystallogr., 1982, B38, 2218. 28. D. E. Williams, Acta Crystallogr., 1996, A52, 326.

58. B. P. van Eijck, W. T. M. Mooij, and J. Kroon, J. Comput. Chem., 2001, 22, 805.

29. M. U. Schmidt and U. Englert, J. Chem. Soc., Dalton Trans., 1996, 2077.

59. P. G. Karamertzanis and S. L. Price, J. Chem. Theor. Comput., 2006, 2, 1184.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

Crystal structure prediction

21

60. R. W. Lancaster, P. G. Karamertzanis, A. T. Hulme, et al., J. Pharm. Sci., 2007, 96, 3419.

73. A. Dey, N. N. Pati, and G. R. Desiraju, CrystEngComm, 2006, 8, 751.

61. G. M. Day and T. G. Cooper, CrystEngComm, 2010, 12, 2443.

74. A. Dey, M. T. Kirchner, V. R. Vangala, et al., J. Am. Chem. Soc., 2005, 127, 10545.

62. S. Grimme, J. Antony, S. Ehrlich, and H. Krieg, J. Chem. Phys., 2010, 132, 154104.

75. T. S. Thakur and G. R. Desiraju, Cryst. Growth Des., 2008, 8, 4031.

63. M. A. Neumann and M.-A. Perrin, J. Phys. Chem. B, 2005, 109, 15531.

76. G. R. Desiraju, Angew. Chem. Int. Ed. Engl., 1995, 34, 2311.

64. A. Gavezzotti and G. Filippini, J. Am. Chem. Soc., 1995, 117, 12299.

77. P. T. A. Galek, L. Fabian, W. D. S. Motherwell, et al., Acta Crystallogr. B-Struct. Sci., 2007, 63, 768.

65. B. P. van Eijck, J. Comput. Chem., 2001, 22, 816.

78. P. T. A. Galek, F. H. Allen, L. Fabian, and N. Feeder, CrystEngComm, 2009, 11, 2634.

66. A. T. Anghel, G. M. Day, and S. L. Price, CrystEngComm, 2002, 4, 348. 67. L. Errende, M. Etter, R. Williams, and S. Darnauer. J. Chem. Soc., Perkin Trans. 2, 1981, 233. 68. D. Buttar, M. H. Charlton, R. Docherty, and J. Starbuck, J. Chem. Soc., Perkin Trans. 2, 1998, 763. 69. P. G. Karamertzanis, G. M. Day, G. W. A. Welch, et al., J. Chem. Phys., 2008, 128, 244708. 70. G. M. Day and W. D. S. Motherwell, Cryst. Growth Des., 2006, 6, 1985. 71. C. F. Macrae, P. R. Edgington, P. McCabe, et al., Appl. Crystallogr., 2006, 39, 453. 72. J. A. R. P. Sarma and G. R. Desiraju, Cryst. Growth Des., 2002, 2, 93.

79. P. T. A. Galek, L. Fabian, and F. H. Allen, CrystEngComm, 2010, 12, 2091. 80. A. Agresti, Categorical Data Analysis, Wiley, New York 1990. 81. P. T. A. Galek, L. Fabian, and F. H. Allen, Acta Crystallogr., 2010, B66, 237. 82. J. P. M. Lommerse, W. D. S. Motherwell, H. L. Ammon, et al., Acta Cryst., 2000, B56, 697. 83. G. M. Day, W. D. S. Motherwell, H. L. Ammon, et al., Acta Crystallogr. B, 2005, 61, 511. 84. W. D. S. Motherwell, H. L. Ammon, J. D. Dunitz, et al., Acta Cryst., 2002, B58, 647.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc109

Noncovalent Interactions in Crystals Paola Gilli and Gastone Gilli University of Ferrara, Ferrara, Italy

1 1 Introduction 1.1 The birth of structural chemistry 2 A Chemical Classification of Crystals 2.1 Chemical forces in crystals 2.2 Metallic crystals 2.3 Covalent crystals 2.4 Ionic crystals 2.5 Molecular crystals 3 Nonbonded Forces in Molecular Crystals. A Classification 4 Mostly Physical Intermolecular Forces in Crystals 4.1 vdW nonbonded forces 4.2 Electrostatic multipolar forces 4.3 Hydrophobic forces 5 Mostly Chemical Intermolecular Forces in Crystals 5.1 Charge-transfer (CT) or electron donor–acceptor (EDA) interactions 5.2 Hydrogen bond (H-bond) 6 Conclusions References

1 1 2 2 2 3 6 8 9 10 10 12 14 15 15 26 37 37

1.1

INTRODUCTION The birth of structural chemistry

The theory of atomism dates back to Leucippus and Democritus, the ancient Greek philosophers who flourished about 440–420 BC. They believed everything was composed of atoms which were indivisible and indestructible particles that, by colliding in the void, “produce vortices, which generate bodies and ultimately worlds.”1 The atoms were obviously invisible, though the bodies they formed were not. More than two thousands years later, at the end of the eighteenth century, the atomic theory was revived to explain the new facts that chemistry was then discovering and, during the following century, it was possible to establish that the combinations of no more than 100 different atoms (often preorganized in clusters then called substances and now chemical compounds or molecules) can, in fact, account for the structure of matter in any of its aggregation states. Atoms and molecules were still invisible, however, and structural formulae were obtained by purely chemical methods (separation, purification, elemental analysis, molecular weight determination, and degradative or synthetic chemical reactions), which were extremely slow by modern standards. This traditional way of working was suddenly upset by the discovery that atoms and molecules can actually be seen by X-ray diffraction when properly arranged in ordered crystalline lattices.2 It was the beginning of a revolution. After the first structure determinations of diamond and alkaline halogenides by Bragg and Bragg,3, 4 a considerable number of crystal structures of elements and simple covalent and ionic compounds started to accumulate. At the end of the 1930s one of the leading figures of the new field of theoretical chemistry, Linus Pauling, could publish his famous book “The Nature of the Chemical Bond and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

2

Supramolecular materials chemistry

the Structure of Molecules and Crystals: An Introduction to Modern Structural Chemistry,”5 which marked the birth of a new discipline, that of structural chemistry. Its most striking innovation with respect to classical chemistry was that not only the topology but also the metric of molecules was now accessible, and that the molecular geometries so obtained could be accounted for in terms of the recently discovered principles of quantum mechanics. Since then, diffraction of X-rays and, later on, thermal neutrons has developed rapidly allowing the scientific world to obtain, for the first time, a detailed knowledge of the intimate atomic distributions of condensed matter. At present, the number of crystal structures determined has grown in such a way that they must be collected in computerbased structural databases, which contained at the end of 2010, 550 000 organic and organometallic compounds (CSD: Cambridge Structural Database),6, 7 135 000 inorganic salts and minerals (ICSD: Inorganic Crystal Structure Database),8 140 000 metals and intermetallics (CRYSTMET: Crystallographic Metals Data File),9 together with some 70 000 biological macromolecules (PDB: Protein Data Bank).10 These databases constitute a substantial library of crystal and molecular geometries but can also be seen from another point of view: they are an even greater archive of atomic interactions, which keeps track of all interatomic forces connecting atoms into stable molecules and all intermolecular recognition forces connecting molecules into stable molecular aggregates. It is then not surprising that crystal packing analysis is still the method of choice for the study of the chemical bond in its various and intriguing manifestations.

2 2.1

A CHEMICAL CLASSIFICATION OF CRYSTALS Chemical forces in crystals

Structural crystallography is traditionally divided into subfields: metals and alloys, minerals, molecular and macromolecular crystals are taught in different courses, and their findings published in different journals. This is a kind of functional partitioning where the results are classified according to their practical uses. There are, however, more basically founded reasons for this subdivision, which arise from the fact that atoms in crystals may be held together by various sorts of chemical bonds. In this sense, we are used to talk of metallic, covalent, ionic, and van der Waals (vdW ) or molecular crystals. These divisions are discussed briefly following the lucid classification given by M¨uller in his recent book on structural chemistry.11 For the few applications of band theory, the main reference is to the clear exposition given by Hoffmann in his

Table 1 Types of crystals formed by metals, nonmetals, and their combinations. Metals Metals Nonmetals

Metallic crystals Ionic crystals

Nonmetals Ionic crystals 1. Covalent crystals 2. Molecular crystals

attractive booklet on the electronic properties of solids and surfaces.12 We start with the basic consideration that the atoms of the periodic table (the elements) can be divided into three subgroups: noble gases, main-group elements, and transition elements (mostly neglected in this simplified treatment), out of which only those belonging to the first subgroup have the ns 2 , np 6 octet configuration of eight electrons in the outer-shell (but only two for helium), which imparts high stability and very low or null reactivity (octet rule or Kossel–Lewis rule). All other elements have different numbers of outer electrons (valence electrons: from 1 to 7) and then tend to form chemical bonds with other atoms by losing (metallic elements) or gaining electrons (nonmetallic elements) in such a way as to reach the configuration of the noble gas respectively preceding or following in the periodic table. Metallic and nonmetallic elements are divided by the so-called Zintl line (a vertical line in between the third and fourth main group) and are respectively characterized by low and high ionization energies (EI , formerly ionization potentials) and high and low electron affinities (Eea ). These few considerations are sufficient to rationalize the main types of chemical bonds occurring in crystals according to the simplified scheme of Table 1.

2.2

Metallic crystals

Metals are crystalline solids endowed with fairly unique qualities, such as malleability and ductility, heat capacity, lustre, and thermal and electric conductivity. The strength of the metal bonds, as evaluated by the vaporization heat,
Hvap , ranges from 15.5 for Cs to 191 kcal mol−1 for Mo and is found to augment with the number of valence electrons that can be ionized, the charge of the cation, and its charge/size ratio. Any metallic bond theory must explain how so much bonding can be produced by so few valence electrons (from 1 to 3, neglecting d orbitals in transition elements) and account for the quite peculiar metallic properties. In 1900, the scientist Paul Drude13, 14 came up with the first effective theory of the metallic state (called the sea of electrons theory) in which metals are ordered arrays of metal ions surrounded by “a sea” of their

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals AOs

3

MOs

N=1 N=2 N=3 W type (bcc) I m 3 m

Cu type (ccp) Fm 3 m

N=4

Mg type (hcp) P 63 /m m c

(c)

Figure 1 Metals tend to crystallize as very simple structures. For example: (a) Al, Ca, Ni, Cu, Ag, Au, Pt, and Pb adopt cubic close-packing (ccp); (b) Be, Mg, Ti, Zr, Co, Zn, and Cd hexagonal close-packing (hcp); and (c) Cr, Fe, Mo, W, and all alkali metals body-centered cubic (bcc) crystal structures.

lost valence electrons and held together by ion-electron electrostatic forces which increase with the number of the electrons involved. In this model, electrons are free to move around (i.e., they are not associated with any particular ion or couple of ions) and the electron cloud is therefore isotropic. This accounts for the well-known fact that the metal ions normally crystallize in simply packed lattices, which are cubic close-packed (ccp) or hexagonal closepacked (hcp) for stronger metal bonds and body-centered cubic (bcc) for weaker ones, as is the case for alkali metals having a single valence electron (Figure 1). The Drude theory was no more than a successful qualitative model. A truly quantitative treatment became possible only some 30 years later, following the new developments of quantum mechanics and, in particular, of the molecular orbital (MO) theory. The new model was named band theory because electrons are no longer free to move in the lattice but allowed to move only along “bands”, which are highly delocalized MOs formed by the atomic orbitals (AOs) overlapping along the infinite atomic rows of the metal lattice. As an example, Figure 2(a) shows the formation of a one-dimensional s-band by the overlapping of two, three, and, eventually, N infinite s AOs lined up along a unique lattice direction. The band formed will contain N orbitals divided in two half bands, a lower one of bonding MOs and an upper one of antibonding MOs. The solid becomes stabilized because all valence electrons can occupy the bonding band of lower energy (valence band ) while the higher antibonding band will remain empty, the two half bands being divided by a horizontal line known as the Fermi level. Electrons cannot move within the fully occupied valence band but can become electrically conducting when thermally promoted to the antibonding one that, for this reason, is called the conduction band. All AOs of the metal are more or less split into bands as a result of the crystal lattice but only the upper bands can become involved in electric transmission, as shown in Figure 2(b) for the series from Na to Al. Na is conducting

N=∞ Symbols:

singly occupied, Fermi level

doubly,

(a)

Energy

(b)

3p

3p

3s

empty MO

3p

3s

3s Mg

Na

(b)

Al

Conduction Band Energy

(a)

Eg Eg

Valence Band (c)

Insulator

Semicoductor

Metal

Figure 2 (a) Formation of a one-dimensional s-band by overlapping two, three, and eventually infinite s AOs lined up along an unique lattice direction; (b) band structure for metals of the first three groups of the periodic table (Na, Mg, and Al); (c) classification of crystals as insulators, semiconductors, and metals according to the decreasing value of their energy gap, Eg .

because its 3s band is only half filled; in Mg this band is full and conduction is granted by the overlapping of the full 3s with the empty 3p band; finally, Al is conducing because its upper 3p band is only half filled. With nonmetallic elements of group IV (see below), new types of conductivity appear that are classified in Figure 2(c) in terms of the energy difference between the upper limit of the full valence band and the lower limit of the empty conduction band, a quantity known as the energy gap, Eg . High, moderate, and null energy gaps are seen to characterize insulator, semiconductor, and metal behavior, respectively.

2.3

Covalent crystals

Formation of covalent or partially covalent crystals is well represented by the structures of nonmetallic main-group

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

4

Supramolecular materials chemistry

Gray a-As R 3 m

Figure 3 The rhombohedral structure of gray α-arsenic consisting of layers of interconnected six-membered rings in the chair conformation.

elements. To reach the octet configuration, any element X belonging to the Nth main group must participate in 8 − N covalent bonds. As a rule, all bonds are single bonds with 8 − N different neighbors, the formation of multiple bonds being normally forbidden except for the elements of the second period, the main exceptions to the single-bond rule being then limited to aromatic graphite layers and molecular crystals formed by the O=O and N≡N molecules. Hydrogen (group I) and halogens (group VII) need only one bond and can form individual H–H or X–X molecules that are held together by the weak vdW forces typical of molecular crystals (Section 4.1). Chalcogens (group VI: S, Se, Te) need two bonds and can form a variety of ring or chain molecules equally packed in molecular crystals. Phosphorus (group V) needs three bonds and

Diamond

C60

(c)

(a)

preferably crystallizes as white phosphorus, a molecular crystal containing tetrahedral P4 molecules. Hence, true covalent crystals, i.e. crystals with a three-dimensional network of covalent bonds, can be associated only with the As, Sb, Bi (group V) and C, Si, Ge, Sn (group IV) atoms. Arsenic, as well as antimony and bismuth, has five outer electrons and, to reach the octet, the four positions of the sp 3 -hybridized atom become occupied by three pyramidal single bonds and one lone pair. The stable form is the rhombohedral gray α-arsenic which consists of corrugated layers of interconnected six-membered rings in the chair conformation (Figure 3). The layers are stacked in such a way that any As atom protruding from the lower level is at the center of the upper hexagon with ˚ which is which it forms three bonds at a distance (3.12 A) ˚ intermediate between the intralayer As–As bonds (2.52 A) ˚ The interaction and the sum of the vdW radii (3.70 A). can be conceived as a triple ≡As:→As donation of three As lone pairs to a fourth As atom acting as electron acceptor by expanding its hybridization state from sp 3 (tetrahedral) to sp 3 d 2 (octahedral). Atoms of different layers are then linked by a nonbonded charge transfer (CT) interaction of the type n → ν that is discussed in detail in Section 5.1.3. The network of such CT interactions causes the formation of a half-filled electron band, which imparts to arsenic semiconducting properties and a typical metallic lustre. Carbon gives the greatest variety of covalent allotropic forms (Figure 4) all of which are important in supramolecular chemistry. Carbon has an outer 2s 2 , 2p 2 configuration.

(d)

C70

(e)

(b)

Graphite

(f)

Carbon nanotubes

Figure 4 Allotropic forms of carbon: (a) the structure of cubic diamond; (b) the hexagonal structure of graphite; (c) buckminsterfullerene, C60 ; (d) the peanutlike structure of C70 ; two types of carbon nanotubes showing (e) metallic and (f) semiconducting properties. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals

5

s∗ (sp 3)

s∗ (sp 3)

Energy

s∗ (sp 3)

sp 3

sp 3

sp 3

sp 3

Eg

sp 3

s (sp 3) s (sp 3)

Silicon Eg ≅ 1.1 eV

s (sp 3)

C–C bond

Diamond Eg ≅ 6 eV

(a)

s∗ (sp 2)

s∗ (sp 2)

Energy

+ −

C AOs

− +

+ −

− + + − + + + − − + + + + +

s(sp 2) (b)

sp 3 Eg

Benzene

p∗ + + + −

− +

+ +

− −

+ −

p∗ C AOs

p

p s (sp 2) Graphite

Figure 5 (a) The strong C–C bond induces a large σ and σ ∗ separation (left) which is only slightly reduced by the band spread (center) and causes diamond to be an insulator with a large band gap of ≈6 eV. Since the Si–Si bond is much weaker, silicon becomes a semiconductor with a band gap of only 1.1 eV; (b) the π and π ∗ MOs of benzene (left) spread in bands in graphite (right).

When sp 3 hybridized, it forms four tetrahedral covalent bonds promoting the cubic diamond structure (Figure 4a). When sp 2 hybridized, it can form only three trigonal σ bonds resulting in infinite layers of condensed hexagons whose packing dictates the hexagonal structure of graphite (Figure 4b), a phase slightly more stable than diamond. The fourth electron remaining in the p AOs perpendicular to the layer facilitates the formation of a fully delocalized π bond system which is the basis of a half-filled electron band with two-dimensional electric conductivity (Figure 5b). The layers are held together by weak vdW forces with a C–C ˚ which imparts graphite talclike distance as large as 3.35 A, lubricant properties and an inclination to form intercalation compounds with alkali metals, such as LiCn (n = 6, 12, 18) or KCn (n = 4, 24, 36, 48). In these compounds, the metal atoms can easily donate their valence electron to the half-filled valence band of the graphite (what will be called an s → π ∗ CT interaction in Section 5.1.3) causing the material to be a better electric conductor than graphite itself. Electrochemical intercalation of Li+ ions into graphite is at the very root of the technology of lithium ion batteries. The number of carbon allotropes (originally diamond, graphite, and amorphous carbon) has steeply increased after the discovery of fullerenes, cagelike modifications

of general formula Cn (n even) prepared by recrystallization from benzene of the products obtained at the electric arc between graphite electrodes in controlled helium atmosphere.15, 16 The main product is buckminsterfullerene C60 , a soccer-ball-like compound consisting of 12 pentagons and 20 benzenelike hexagons (Figure 4c). The second in yield is C70 (a peanutlike solid with 12 pentagons and 25 hexagons; Figure 4d), while a number of less stable compounds with 12 pentagons and even n ≥ 32 are also known. Similar high-temperature methods have permitted the synthesis of carbon nanotubes, a new allotropic form of carbon consisting of bent graphitelike layers closed in a cylinder that can be as long as 0.1 mm and most frequently closed at their ends by half-fullerene spheres (Figure 4e and 4f). Graphite layers, fullerenes, and nanotubes can be thought of as configurational isomers of the same covalent [C(sp 2 )]n molecule, which crystallizes by forming van der Waalspacked molecular crystals. Silicon and germanium are of great importance as intrinsic semiconductors. Being unable to form double bonds, they can only crystallize in the cubic diamond structure (Figure 4a) with sp 3 -hybridized atoms linked by four identical covalent bonds. Of particular interest is their band structure (Figure 5a). In diamond the C–C bond is so strong ˚ that, in spite of the band spreading, the (C–C = 1.54 A)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

6

Supramolecular materials chemistry

Zn

S

S Zn Zinc blende F 43 m (a)

Figure 6

(b)

Wurtzite P 63 m c

Prototypes of diamondlike structures: (a) the cubic zinc blende (sphalerite) structure; (b) the hexagonal wurtzite structure.

σ and σ ∗ MOs remain widely separated by a large band gap of some 6 eV, which makes diamond an insulator. Bonds become increasingly weaker in silicon and germa˚ and, accordingly, nium (Si–Si = 2.35; Ge–Ge = 2.45 A) their crystals are semiconductors with a decreasing energy gap of 1.1 and 0.7 eV, respectively. The diamondlike structures are adopted by many atomic couples having, altogether, four electrons per atom, such as SiC (IV–IV), GaAs (III–V), CdS (II–VI), or CuCl (I–VII). In analogy with silicon and germanium, some of these compounds are semiconductors with widespread technological applications in electronics. Their crystal structures can be referred to the two allotropic forms of diamond (the normal cubic form and the rare meteoritic hexagonal form) in which the carbon atoms are alternately substituted by couples of heteroatoms to give crystals having the zinc blende (sphalerite) and wurtzite structural types, respectively (Figure 6). Possible elemental combinations that are able to give these types of structures are summarized in Table 2. In diamondlike compounds, the average number of four valence electrons is obtained by binary mixing of two nonmetallic elements. Another method would be to transfer electrons from a metallic to a nonmetallic element in such a way so as to produce anions with the correct number of Table 2 Elemental combinations with an average of four valence electrons per atom that, accordingly, crystallize in the zinc blende (sphalerite) or wurtzite structural types. Group numbers

Zinc blende type

Wurtzite type

IV–IV III–V II–VI I–VII

β-SiC BP, GaAs, InSb BeS, CdS, ZnSe CuCl, CuBr, AgI

SiC AlN, GaN BeO, ZnO, CdS (high temperature) CuCl (high temperature), β-AgI

electrons. This strategy provides the class of polyanionic compounds or Zintl phases.11 A great number of these phases are known but it is sufficient to mention here a few examples that illustrate the mechanism of their formation. Silicon has four valence electrons, makes four bonds, and crystallizes as diamond; in CaSi2 , however, it transforms into the Si− ion with five electrons that can only form, like arsenic, three bonds. Accordingly, the crystal (Figure 7a) turns out to be built up of arseniclike layers intercalated by planes of cations. In a similar way, compounds NaTl and MgB2 can be interpreted as Na+ Tl− and Mg2+ B2 − which, containing Tl− and B− ions with four electrons, can be supposed to crystallize in carbonlike forms. In fact, NaTl includes a diamondlike thallium substructure (Figure 7b) surrounded by cations, and MgB2 an even more interesting graphitelike boron substructure intercalated by planes of cations (Figure 7c).

2.4

Ionic crystals

The electrostatic (or Coulomb) energy associated with two interacting ions i and j , separated by a distance rij and having integer charges zi and zj in units of e = 1.6022 · 10−19 C is
 1 (1) zi zj e2 rij−1 Eel = 4π ε0 where ε0 is the electric permittivity of a vacuum and the ˚ e−2 = constant (1/4π ε 0 ) has value 332.145 kcal mol−1 A −1 ˚ −2 1389.70 kJ mol A e (Section 4.2.2). In crystal lattices there are many more interactions than in a single pair of ions. For example, simple inspection of the NaCl structure (Figure 8b) shows that the Na+ ion has 6 Cl− neighbors at a distance of R, 12 Na+ at a distance

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals

7

Na

Ca

TI Si

(b)

NaTI F d 3 m

B Mg CaSi2 R 3 m (a)

(c)

MgB2 P 6/m m m

Figure 7 Examples of Zintl phases: (a) the crystal structure of CaSi2 showing the arseniclike substructure of Si− ions; (b) the unit cell of NaTl showing the diamondlike substructure of Tl− ions; (c) section of the structure of MgB2 showing the graphitelike substructure of B− ions.

√ √ of R 2, 8 Cl− at a distance of R 3, and so on, so that the electrostatic contribution to the lattice energy, Eel,U , becomes
Eel,U =


NA 4π ε0

 

zi zj e2 rij−1

j

NA zi zj e2 R −1 4π ε0
 6 8 6 24 12 × − √ + √ − √ + √ + ··· 2 3 4 5 1
 NA = (2) zi zj e2 R −1 A 4π ε0 =

where NA is the Avogadro number, zi zj = −1 in this NaCl example, and A, the geometrical Madelung constant, is the mathematical sum of the converging series in parentheses.

Geometrical Madelung constants are positive quantities which are independent of ionic charges and cell dimensions and depend only on the type of structure, as illustrated in Table 3 for some common structural types. Two points need to be considered: (i) since A is always greater than unity, an ion pair is always less stable alone than when embedded in a regular crystal, which is therefore the more stable form in all ionic compounds; and (ii) in a number of crystals the bond is not fully electrostatic but includes a covalent contribution; in such a case (2) can still be used but in the form Eel,U = (NA /4π ε0 )qi qj e2 R −1 A, where qi and qj are fractional (instead of integer) charges which are generally unknown but whose evaluation is perhaps possible by quantum-mechanical methods. The overall lattice energy, UL , of an ionic crystal is calculated in the approximation UL = Eel,U + Edisp + Erep + E0

Table 3 Coordination numbers (CN) and geometrical Madelung constants (A) for some most common ionic structural types. Structure type

CN

A

CsCl NaCl ZnS (wurtzite) ZnS (zinc blende)

8:8 6:6 4:4 4:4

1.76267 1.74756 1.64132 1.63805

Structure type

CN

CaF2 (fluorite) TiO2 (rutile) SiO2 (β-cristobalite) Al2 O3 (corundum)

8:4 6:3 4:2 6:4

A 2.51939 2.408 2.298 4.1719

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

(3)

8

Supramolecular materials chemistry

where Edisp and Erep are the dispersion–attraction and exchange–repulsion terms of the atoms involved (see Section 4.1) and E0 is the zero-point energy at the absolute zero, which is considered to be negligible for most ionic crystals except in the presence of atoms of very small mass, such as in metal hydride or water crystals. The final equation for UL can be written as UL = NA

 
i,j

1 4π ε0



Cs Na Cl CsCl P m 3 m Cl

−6 zi zj e2 rij−1 − ci,j ri,j

(a)

(b) NaCl F m 3 m

 F

+ai,j exp(−bi,j ri,j ) + E0
=

NA 4π ε0



zi zj e2 R −1 A + 



−6 −ci,j ri,j

O

i,j

+ ai,j exp(−bi,j ri,j ) + E0

(4)

Note that other more simplified equations are also available, such as the Born–Land´e and Kapustinskii equations, which reportedly provide reasonably good agreement with the experiments. Experimental values are obtained by means of thermodynamic cycles, in particular the wellknown Born–Haber cycle. A large number of lattice energies have been measured so far, which agree in showing that binding energies in ionic crystals are considerably larger than those in metallic crystals, particularly in consequence of their strong dependence on the product, zi zj , of the ionic charges. As a representative sample, lattice energies amount to 188, 629, 3804, and 3137 kcal mol−1 for NaCl, CaF2 , Al2 O3 , and SiO2 , respectively. Another important factor that can affect the value of UL is the structural type chosen by the crystal, which acts through its specific value of the Madelung constant, A (Table 3). Figure 8 illustrates the structural types adopted by some of the most common ionic compounds. It is worthwhile to remember that the choice of the structural type, coordination number (CN), and coordination polyhedron is essentially determined by the value of the radius ratio (rCAT /rAN ) according to the rules illustrated in Table 4.

Table 4 Radius ratios (rCAT /rAN ), coordination numbers (CN) and coordination polyhedra, and ionic structural types adopted by some common AB and AB2 ionic compounds. rCAT /rAN

CN—polyhedron

>0.732 0.732–0.414 icv–scr ∼ = scr–icv > scr–scr. This concept is well illustrated in Table 11, showing that the formation enthalpies of typical n → σ ∗ adducts of molecular iodine (3–12 kcal mol−1 ) are systematically smaller than those of a representative series of Lewis’ adducts (14–35 kcal mol−1 ) which, in turn, are smaller than those of a true CT chemical reaction (e.g., 71 kcal mol−1 for the BF3 + F−
BF4 − reaction); Stronger CT interactions take place when the HOMO–LUMO energy difference is small or null (Figure 10). Accordingly, electron-donating substituents on the donor and electron-accepting substituents on the acceptor will both stabilize the complex.

Scheme 3 illustrates some of the most common CT adducts ordered according to the nature of the donor. A first group (i–iii; 3.Ia–IIIc) concerns n and n− donors that, being increvalent, can form strong complexes with equally increvalent ν acceptors and moderately strong ones with sacrificial σ ∗ and π ∗ acceptors; a second group (iv; 3.IVa–c) includes a small number of complexes that involve metal interactions with sacrificial σ donors; the last group (v–vi; 3.Va–VIb) reports complexes involving localized or delocalized π bonding pairs that are often of relevant technological interest. They are illustrated in greater detail in the following: (i)

(ii)

n → ν complexes of n donors with acceptors having vacant AOs. The examples shown illustrate the increvalent addition of NH3 to BF3 to form the Lewis’ adduct H3 N→BF3 (3.Ia) and that of Cl3 PO to pentacoordinated pyramidal SbCl5 to give the corresponding octahedral complex (3.Ib); n → σ ∗ complexes of n donors with σ ∗ acceptors. Typical examples are the halogen bonds pyridine → I–Cl (3.IIa) and dioxane → Br–Br (3.IIb), which

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

20

Supramolecular materials chemistry

(i) n

n donor–acceptor complexes H

(3.Ia)

H

B

N H H

+

N H

F F

H

F

Ammonia

I

3.03 Å 2.83 Å I I

From I

−

p∗ and n

n (3.IIIa) O

N

O

H N

O

H

R

X H

O

N+

p

O

N N H

H

O S O Cl

Scheme 3

O

I I I

I

I

Planar sheets in iodine crystals

O

+ C

N

R

H

O

R

p∗(k) R3N

n

R

N

O

R

R

R

R R

N R

N H

C(O)R2

SN2 nucleophilic addition

O

p∗(k) Alloxan self-complex p∗ donor–acceptor complexes D

M (3.IVb)

M

s∗ donor–acceptor complexes Br Å (3.Vb) 1 ÅBr 6 3 01 . 3 .3 2

3.

Br Br

p

S

(3.IVc)

s s∗ Dihydrogen bond (M = Al, B, Ga, transition metal)

18

4

Å

s p∗ Graphite metal intercalation (M = Li, K)

H

C

C s∗ Benzene

(3.VIa)

Cl S

3.162 Å

Cl Tetrathiafulvalene Cl o -chloranil

(3.Vc)

R

C

R

D

H D R

C Br2

p∗ donor–acceptor complexes S

I I

I

I

I

H

CH3 CH3 + n Durene nitrosyl

(vi) p

I

I

R

O 2.720 Å O O

2.271 Å

CH3 CH3

(3.IIc)

I

I

(3.IIIb)

H

n Agostic interactions in organometallics (M = transition metal)

Å

I

I

I

Y Hydrogen bond (X, Y = O, N, C)

O

N H

s

(3.Va)

I I

(3.IVa)

O

O

I

(3.IIIc)

N

s ∗, and s

n+ , p

I

I

O

O

M

(v) p

SbCl5

(3.IIe) R 2.4 0– 3 X H .10 Å R Y

H C

Cl Cl

Å Br

O

O

n

n, s

(iv) s

(3.IIb)

Br2

p∗ Picric acid self-complex

H

O

2.17 Å Cl O Sb Cl Cl

p∗(k) donor–acceptor complexes

O 2.913 Å H O N O O

N

.31

2.90 Å 2.90 Å

− [I I I] − I2 complex to I3 ion (3.IId)

(iii) n

O

F

O 1,4-Dioxane

ICI

−

(3.Ib) Cl CI3PO

2 O .71 Å 2 Br

I Cl 2.51 Å

Pyridine

B

P

s∗ donor–acceptor complexes

2.29 Å (3.IIa) N

Cl

BF3

s∗ and n −

(ii) n

Cl

FF

H H

H

H p

D

s∗ Hydrogen bonds (D = O, N, and C)

H (3.VIb) H

H N

2,3-Diazabicyclooctene tetracyanoethylene

N 2.883 Å N 2.986 Å N N

N

Some of the most common types of CT complexes.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals

(ii )

(iii)

)

(iii

(iv)

(v)



(v )

(vi)

appear to cause a considerable contraction of the N· · ·I and O· · ·Br distances with respect to their ˚ respectively). The example vdWs (3.68 and 3.37 A, (3.IIc) shows that the crystal structure of iodine is based on a net of such n → σ ∗ interactions where iodine is both n donor (perpendicularly to the I–I bond) and σ ∗ acceptor (collinearly with it). As already noticed, H-bonds can also be regarded as n → σ ∗ complexes, as exemplified in (3.IIe); n− → σ ∗ complexes of charged n− donors with σ ∗ acceptors. Because of the negative charge, these donors can produce symmetrical bonds of considerable strength (medium complexes) which are similar to the (−)charge-assisted H-bonds described below. This is illustrated in (3.IId) for the addition of I− to I2 with the formation of the symmetric triiodide anion; n → π ∗ complexes of n donors with π ∗ acceptors endowed with strong electron-withdrawing substituents. This is illustrated by the typical picric acid self-complex (3.IIIa); n → π ∗ (k) complexes of n donors with π ∗ acceptors whose empty π ∗ MOs are located on strong heteronuclear double bonds, such as C=O carbonyls. According to the circumstances, this is equivalent to a nucleophilic addition to the positively-charged ketonic carbon that gives rise to a simple adduct, such as the alloxan self-complex (3.IIIb), up to a full reaction of nucleophilic addition (3.IIIc); σ → ν/σ ∗ and s → π ∗ interactions in a number of different and rather specialized CT complexes involving metal centeres. The examples shown include σ → ν agostic interactions (3.IVa) and σ → σ ∗ dihydrogen bonds (3.IVb) in organometallic compounds as well as s → π ∗ interactions associated with graphite–alkali metal intercalation (3.IVc); π → ν + complexes of neutral π donors with positively charged acceptors having vacant AOs; the example shown (3.Va) illustrates the strong association of tetramethylbenzene (durene) with the nitrosyl NO+ cation; π → σ ∗ complexes of neutral π donors with σ ∗ acceptors. This includes halogen bonds to multiple C–C bonds as in the benzene → Br2 adduct (3.Vb), and all D–H· · · π bonds (3.Vc) formed by normal H-bond donors, D–H, with multiple bonds and aromatic rings (Section 5.2.3 and Scheme 5); π → π ∗ complexes between π donors and π ∗ acceptors, as illustrated by the adducts between TTF and o-chloranil (3.VIa) and 2,3-diazabicyclooctene and TCNE (3.VIb). These complexes include the important classes of organic metals and semiconductors.

21

5.1.4 Some practical applications of CT interactions CT interactions and crystal packing The role played by CT interactions in the packing of molecular crystals is less known than that of H-bonded or vdW interactions. Awaiting a more general treatment, the problem is illustrated here by discussing the single but representative structure of the 4-nitropyridine-Noxide–picric acid cocrystal,51 whose intermolecular contacts shorter than vdW include two O–H· · ·O and seven C–H· · ·O H-bonds (Figure 11a) together with seven O:→ C/N CT contacts of both n → π ∗ and n → π ∗ (k) types (Figure 11b). The crystal packing analysis is simplified

C10

C7 N4

O2

C11

O8

O10 C10

O1

O3

O2

O1

C3

O6

O5

O4

O7

O3 C5

C3

O9

C8

C7

O5

O4

O10 C8

C7

(a) C5

N2 O6 C1 O3 O3 C1 N2 O5

O6

C5 O7

C9

O10

C7 C11

C10 C11

N4 C11

O8

N4

O10

(b)

Figure 11 Intermolecular contacts shorter than vdW in the crystal structure of the 4-nitropyridine-N -oxide–picric acid complex: (a) O–H· · ·O and C–H· · ·O H-bonds; and (b) n → π ∗ and n → π ∗ (k) CT interactions. (Reproduced by permission from Bertolasi et al., 2011;51  American Chemical Society, 2011).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

22

Supramolecular materials chemistry Table 12 Some typical π electron donors and π ∗ electron acceptors whose formulae are depicted in Scheme 4. π ∗ Acceptors

π Donors 4.Ia 4.Ib 4.Ic 4.Id 4.Ie 4.If 4.Ig

Pyrene Hexamethylbenzene Tetramethyl-p-phenylenediamine Tetramethylbenzobisdioxole Tetrathiafulvalene Tetraselenofulvalene Tetramethyltetraselenofulvalene

— HMB TMPD TMDO TTF TSF TMTSF

4.IIa 4.IIb 4.IIc 4.IId 4.IIe 4.IIf

by considering all H-bonds as O–H← :O and C–H← :O σ ∗← n CT interactions. In this way, it becomes evident that the two molecules possess 10 neutral n donors (all located on the NO2 , N–O, and OH oxygens), and that all of them are fully saturated by the available acceptors (two σ ∗ (O–H), seven σ ∗ (C–H), and seven π ∗ or π ∗ (k) acceptors), giving so rise to a total of 16 short contacts, i.e., 1.60 interactions per donor. Hence, the data suggests that the saturation of the maximum number of electron pairs by the available acceptors must be an important factor in determining the crystal packing arrangement. This fact has been recently supported by the crystal packing analysis of picric acid and 14 of its adducts with N-bases and tentatively called electron-pair saturation rule.51 In summary, data point to the idea that bonding and nonbonding electron

Dichlorodicyano-p-benzoquinone Dibromodicyano-p-benzoquinone p-Chloranil Trinitrobenzene Tetracyanoethylene Tetracyanoquinodimethane

DDQ DBQ p-CA TNB TCNE TCNQ

pairs are to be considered as points of residual reactivity on the molecular surface that can then react with the available electron acceptors by incipient (molecular interaction) or full (dative bond) nucleophilic addition to fulfill the saturation rule. π → π ∗ CT interactions in organic molecular semiconductors This class of π → π ∗ adducts is of particular interest for the role they have played in developing a new class of materials of great potential technological interest: the organic molecular semiconductors.52–55 Some of the most common π donors and π ∗ acceptors are listed in Table 12 and depicted in Scheme 4. Electron donors (D) and acceptors (ii) Some typical p∗ acceptors

(i) Some typical p donors N

O

N

O CN Cl

X

Cl

NO2

(4.Ic) TMPD O

O (4.Ia) Pyrene

CN Cl Cl O O NO2 O2N (4.IIa) DDQ (X = Cl) (4.IIc) p -CA (4.IId) TNB (4.IIb) DBQ (X = Br) X

O O (4.Id) TMDO R R

X X

X X

R R

NC

CN

NC

CN

NC

CN

NC

CN

(4.Ie) TTF (X = S, R =H) (4.Ib) HMB

(4.If) TSF (X =Se, R= H)

(4.IIe) TCNE

(4.IIf) TCNQ

(4.Ig) TMTSF (X=Se, R=CH3) (iii) Packing of p– p∗ CT complexes

Scheme 4

(4.IIIa) Alternated stack

(4.IIIb) Segregated stack

Pyrene-TCNE

TTF-TCNQ

Donor

Acceptor

(4.IIIc) Anion-radical salt

(MEM)(TCNQ)2

Cation

Anion

(4.IIId) Cation-radical salt

(TMTSF)2(CIO4)

π → π ∗ CT interactions in organic molecular semiconductors.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals (A) are neutral molecules with widely delocalized systems of multiple bonds which are endowed with electrondonating or attracting substituents. They interact by π → π ∗ donation with formation of charged radicals, D•+ and A•− , respectively. The electric conductivity σ ( −1 cm−1 ) of a material is the product of the number of charge carriers (electrons and holes) for their mobility µ (cm2 V−1 s). According to their conductivity, materials divide into metals (103 ≤ σ ≤ 106 −1 cm−1 ), semiconductors (10−7 ≤ σ ≤ 102 −1 cm−1 ), and insulators (10−18 ≤ σ ≤ 10−7 −1 cm−1 ). As a rule, molecular crystals consist of closed-shell molecules that do not possess intrinsic carriers and are therefore insulators, though good conductivities can be achieved by injecting charge carriers by means of suitable dopants, as is done with many unsaturated compounds, such as metallophthalocyanines and polyacetylene materials.55 Since the discovery of the monoclinic π → π ∗ CT adduct between TTF (tetrathiafulvalene (4.Ie)) and TCNQ (tetracyanoquinodimethane (4.IIf)) at the beginning of the 1970s, no intrinsic organic semiconductors had ever been described. The TTF–TCNQ crystals were found to display typical metallic properties with a roomtemperature conductivity σ RT ∼ = 600 −1 cm−1 , strongly anisotropic along the [010] direction, which monotonically increases with decreasing temperature till a maximum of σ m /σ RT ∼ = 20 at Tm = 58 K. Below Tm the conductivity starts to decrease, the adduct entering the semiconductor region and finally becomes a full insulator near 5 K. The crystal structure56 revealed an interesting aspect: the two molecules do not intermix but are assembled in homomolecular segregated stacks of parallel TTF and TCNQ molecules aligned along the b axis and making with it an angle around 45◦ (4.IIIb; Figure 12). Because of this inclination, molecules do not superimpose but are shifted along their longer axis of more than a C–C distance. Interplanar distances are, however, con˚ for TCNQ stant in both columns with values of 3.17 A ˚ for TTF molecules, that is significantly shorter and 3.47 A ˚ than the corresponding vdW distances of 3.40 and 3.60 A, respectively. The amount of charge transferred from TTF to TCNQ, evaluated from the changes of bond distances and IR stretching frequencies, averages to 0.6 electrons at room temperature (which means that the electron transfer occurs, on average, in 60% of the adducts). This is the origin of the conductivity enhancement, because the electron transfer populates previously empty orbitals and partially empties fully occupied orbitals, transforms nonbonding or antibonding interactions into bonding ones, causes the tightening of donor–acceptor contact distances and, finally, lays the basis for the formation of an electronic band structure that runs along the axis of the segregated stack and

23

c

b

a (a)

a

a

c c

b b (b)

(c)

Figure 12 The structure of the tetrathiafulvalene–tetracyanoquinodimethane (TTF–TCNQ) mixed crystals (CSD refcode: TTFTCQ01).56

is nearly perpendicular to the plane of the aromatic rings involved in the π → π ∗ interaction (monodimensional conductivity). After the discovery of the TTF–TCNQ complex, several hundred potential organic semiconductors were synthesized, crystallized, and studied, for which the reader is directed to the recent literature.53–55 These studies have led to the following general rules: 1.

The crystals can contain either alternated (4.IIIa) or segregated stacks (4.IIIb), but only the latter can form electronic bands that can impart to the stack metallic or semiconducting behavior; 2. Segregated stacks consist of separated sequences of positive D•+ D•+ D•+ D•+ or negative A•− A•− A•− A•− radical ions which are stabilized by the formation of the band structure but destabilized by the repulsion between equal-sign charges. The repulsion can be reduced by decreasing the charge transferred below one (e.g., 0.6 in TTF–TCNQ), by slightly shifting the donors or acceptors with respect to the other (bent stacks as in (4.IIIb) or zigzag stacks as in (4.IIIc) and (4.IIId)), or by cocrystallization of neutral molecules within the stack. From this latter scheme arises a great variety of structures which have the stack repetition

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

24

Supramolecular materials chemistry

unit [(D•+ )·(Dn )] or [(A•− ) · (An )] with n ranging from 0 to 3; 3. Following the Herbstein’s classification,53 there are three main classes of compounds able to give segregated stacks: (a) Radical cation salts with general formula {[(D•+ ) · (Dn )] · [Anion]− }, where [Anion]− is a closed-shell counteranion such as Cl− , Br− , I− , I3 − , ClO4 − , PF6 − , SnCl6 − , and so on. The packing of one of them, (TMTSF)2 (ClO4 ), is sketched in (4.IIId); (b) Radical anion salts with general formula {[(A•− ) · (An )] · [Cation]+ }, where [Cation]+ is a closedshell countercation such as alkaline metal ions and a large number of protonated nitrogen bases. The packing of (MEM)(TCNQ)2 (MEM = N-ethylmethylmorpholinium) is sketched in (4.IIIc); (c) Radical cation–radical anion salts with general formula {[(D•+ ) · (Dn )] · [(A•− ) · (An )]}, where n and n are now small numbers in the range 0–1. TTF–TCNQ (with n = n = 0) is the reference term of this series which has prompted the synthesis and testing of all related adducts. It may be of interest to remark that, in spite of all attempts done, no other compound has proved to have significantly better physical properties than those of TTF–TCNQ. n → σ ∗ CT interactions and the halogen bond (X-bond) As a rule, σ ∗ electron acceptors are found associated with monovalent atoms forming simple covalent bonds. The choice is practically limited to the Y–H and Y–X (X = halogen) cases, giving rise to the Y–H· · ·:D hydrogen bonds (H-bonds) and Y–X· · ·:D halogen bonds (X-bonds), where :D is a suitable electron donor, most often a nonbonding pair located on O, N, S, or on a second X atom and, less frequently, a π bonding pair of a multiple bond. In spite of their striking chemical similarities, H-bonds and X-bonds are normally discussed as distinct topics, even because Hbonds are of great importance in biological processes while X-bonds are essentially a chemical phenomenon whose real role in nature is still subject to speculation. For certain, X-bonds have a significant importance in agricultural and environmental chemistry because halogenocarbons are among the most efficient and widely used pesticides (herbicides, fungicides, insecticides). Though the first mention of an X-bonded complex, H3 N· · ·I2 , is imputed to Guthrie in 1863,57 convincing evidence of a CT complex of this sort, benzene· · ·I2 , was given by Benesi and Hildebrand only 85 years later on the observation of its visible CT band.46, 47 The discovery prompted American scientists to organize a number of

ad hoc meetings on the topic, so inducing Mulliken, one of the world leading theoreticians at the time, to lay down the basis of his quantum-mechanical theory of CT phenomena.44, 45 During the 1950s and in the early 1960s, Hassel and coworkers undertook a systematic program of X-ray structural determination of the crystal adducts between dihalogens and various Lewis bases (now known as Hassel’s compounds)58, 59 and in 1968 Bent published his renowned review on the structural chemistry of donor–acceptor interactions in the solid state including, besides Hassel’s adducts, all other types of CT complexes which were known at the time.60 From then on, X-bonds have been the subject of countless studies by a variety of experimental techniques, among which to be recalled are the thermodynamic measurements of binding constants in nonpolar solvents50, 61 and the structural determination by MW spectroscopy in the supersonic expansion regime.62, 63 From the mid 1990s, X-bonding has had an impressive outbreak following the concomitant explosion of molecular interaction and recognition studies. For a critical appraisal the reader is referred to a number of recent review publications in the fields of supramolecular chemistry and crystal engineering,64–66 liquid crystals,67 biological molecules,68 and biotechnological applications.69 Binding energies of neutral X-bonds span a plausible range of 1–15 kcal mol−1 and are heavily affected by the nature of the halogen. In X–X dihalogens, the electron entering the σ ∗ MO decreases the bond order and lengthens the distance of the X–X bond with an energy cost that is the higher the larger the X–X bond energy is. Since this energy decreases down the group, the accepting ability increases in the order Cl2 < Br2 < I2 (while F2 is reportedly unable to form complexes). Analogously, in interhalogen X–Y molecules the better acceptor is always the less electronegative atom and its accepting power is seen to decrease according to the series: →I–F > →I–Cl > →I–Br > →I–I; →Br–F > → Br–Cl > →Br–Br; and →Cl–F > →Cl–Cl. Halocarbons, C–X, are usually weak acceptors for any X because of the low electronegativity of the carbon, unless the carbon itself is bonded to strongly electronattracting substituents. For instance, the DFT-computed interaction energies of the complexes H3 C–I· · ·NH3 , H2 FC–I· · ·NH3 , HF2 C–I· · ·NH3 , and F3 C–I· · ·NH3 are 2.5, 3.2, 4.3, and 6.4 kcal mol−1 , respectively.70 Not by chance, the fluorination of halocarbons is a well-established technique for X-bond strengthening in crystal engineering, as illustrated in the example of Figure 13.64 Politzer and coworkers71–74 have promoted X-bond studies based on the electrostatic potential V (r), a quantity derived from the electron density ρ(r) that, for a molecule of M atoms, is defined as the sum of its nuclear and electronic contributions, V (r) = VN (r) +

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals

25

Figure 13 The structure of the N,N,N  ,N  -tetramethyl-p-phenylenediamine–1,8-dibromoperfluorooctane mixed crystals. (Reproduced from Metrangolo et al., 2005;64  American Chemical Society, 2005).

Ve (r), where VN (r) = k=1,M Zk /|Rk − r| and Ve (r) = − ∫ dr  ρ(r  )/|r  − r|. Electrostatic potentials have many applications in chemistry,75 particularly in the field of molecular interactions, which can be thought of as due to the mutual attractions among opposite-sign regions of the surface potential, VS (r), that is the electrostatic potential on the outer contour of the molecule (say, on the surface with ρ(r) ≤ 0.001e bohr−3 which nearly includes the 97% of the electron charge). In GS atoms, VS (r) is positive everywhere due to the charge of the nucleus but, when atoms combine in molecules, one or more negative regions may develop in

CF4

CF3Br

association with the most electronegative atoms. It is known since 1992 that surface potentials of alkyl halides display a number of characteristic features71–74 that can be well exemplified by the series of X-bonded complexes F3 C–X with X = F, Cl, Br, and I as depicted in Figure 14. As expected, the fluorine hemispheres of CF4 are completely negative. When a fluorine is substituted by a chlorine, however, a positive potential develops on the outermost part of the surface surrounding the C–Cl axis which, for its aspect, has been called σ hole and considered to be the distinctive mark of the electron-acceptor properties of the σ ∗ MO. Similar results are obtained for CF3 Br

CF3Cl

CF3I

−0.00900 −0.00585 −0.00270 0.00045 0.00360 0.00675 0.00990 0.01305 0.01620 0.01935 0.02250 0.02565 0.02880 0.03195 0.03510 0.03825 0.04140 0.04455 0.04770 0.05085 0.05400

Figure 14 The molecular electrostatic potentials (hartrees) at the 0.001e bohr−3 isodensity surface of CF4 , CF3 Cl, CF3 Br, and CF3 I. (Reproduced from Clark et al., 2007;73  Springer, 2007). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

26

Supramolecular materials chemistry

and CF3 I but with the difference that the σ holes on bromine and iodine become progressively larger and more positive. Since then, much has been done and said on σ holes and their properties and, in particular, it has often been remarked that binding energies calculated for a series of X-bonded complexes with a common electron donor correlate well with the maximum VS (r) of the acceptor at the center of the hole. This does not necessarily imply, however, that the driving force in complex formation is necessarily electrostatic as, on the other hand, shown by a recent energy partitioning performed on the very molecules of Figure 14 complexed with formaldehyde, which has established that the electrostatic term, though important, is systematically about three times smaller than the covalent CT contribution.76

during the period he was working out the modern theory of the chemical bond. This explains why the ideas that they originally developed have not changed so much from that time and are still the rational basis for any further definition. Such a definition, as completed and formalized by Vinogradov and Linnel in their 1971 book84 and slightly adjusted by us in 2009,93 reads as follows:

5.2

This definition has essentially lasted 90 years, a considerably long period for any scientific statement. Thirty years after its first formulation, however, Mulliken44, 45 remarked that the H-bond could also be considered as a R–D·–·H ←: A–R σ ∗← n CT interaction, where the lengthening of the D–H bond, that is the proton transfer, is the direct consequence of the D←: A charge transfer of a typical scr ← ivl CT interaction (Section 5.1.2). Since the two models equally account for the experimental facts, the choice between them mostly remains a matter of practical opportunity and this definitely inclines toward the PT transfer concept because, giving direct access to the acid–base theory, leads more easily to an overall H-bond theory.93, 94

Hydrogen bond (H-bond)

5.2.1 Introduction For all the nineteenth century, hydrogen was firmly monovalent. The first to suggest it could not be so was Albert Werner who, in 1902, suggested that the proton of the ammonium salts was linked to both ammonia and the anion according to the scheme (H3 N· · ·H)X, that he indicated as Nebenvalenz (or secondary valence).77 Only eighteen years later, Latimer and Rodebush78 published a seminal paper marking the birth date of the new interaction, that they called the weak bond but that was later christened hydrogen bond (H-bond ) by Pauling (1931)79 and hydrogen bridge (H-bridge) by Huggins (1936).80 The newly discovered H-bond has become one of the most successful scientific ideas of our age. Originally introduced to explain the deviations from ideality of liquids and solutions, a specialized topic of apparently marginal relevance, in a few years succeeded in answering so many unsettled questions that it is even difficult today to imagine what could be our understanding of nature before knowing the role played by the H-bond: it makes sea water liquid, joins cellulose microfibrils in trees as high as sequoias, shapes DNA into genes, and polypeptide chains into wool, hair, muscles, or enzymes, which makes it the single most effective determinant of the overall structure of the ecosphere. It is not surprising, therefore, that the H-bond has been studied, from its discovery, by practically all experimental and theoretical methods at our disposal, giving rise to countless publications later summarized in a dozen comprehensive books81–93 (for a more complete list of books and reviews, see Section 1.4 of Ref. 93).

5.2.2 H-bond definitions People who discovered and first developed the H-bond were all young researchers at the laboratory of G.N. Lewis,

The H-bond is a three-center-four-electron shared-proton interaction having the general form R–D·–·H · · ·:A–R  , where D is the proton donor (an electronegative atom, such as F, O, N, C, S, Cl, Br, and I) and :A the proton acceptor or lone electron pair carrier (a second electronegative atom or the π-bond of a multiple bond). The H-bond can also be seen as a single proton sharing two lone electron pairs from two adjacent electronegative atoms or groups: R –D − : · · · H + · · · : A–R  .

5.2.3 H-bond nomenclature and physical characterization A rudimentary H-bond nomenclature is given in Scheme 5. H-bonds can be classified according to their strength as weak, moderate, or strong; proton position as symmetric (or centric, or proton-centred) or asymmetric; X–H–Y angular value as bent or linear (5.Ia,b); chemical symmetry as homonuclear or heteronuclear (5.IIa,b); connectivity as two-centered, three-centered (or bifurcated), or chelated (5.IIIa–c); and topology as intermolecular (5.IVa–c) or intramolecular (5.IVd). As a rule, homonuclear H-bonds tend to be stronger than heteronuclear, two-centered than multicentered, and linear than bent ones. It is useful to classify all the infinite H-bonds occurring in nature in more homogeneous groups. According to the most recent classification,92 the H-bonds of main-group elements, which are by far the most numerous, can be profitably divided in four groups (Scheme 5): Group 1: Conventional H-bonds (5.Va) involves the most electronegative atoms that, for this reason, can also form the strongest H-bonds, while the others cannot because of the intrinsic weakness (i.e., too low electronegativity) of the donor, Group 2:

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals H

D

A

D (5.la)

A

H

X

H

N

N

H

O

(5.lVa)

D

X

(5.IIa)

(5.Ib)

O

H

A D

(5.IIIa)

X H (5.IIb)

Y

(5.lVb) H H F

F H

O

F

O O

Group 3

D, A = N, O, S, F, Cl, Br (5.Va)

:A

WD = C, P, Se, Si :O C H :N C H

H

WA = organic F, Cl, Br

H

S, Se, Te, P, As, Sb, organic :C

H

(5.Vb)

CH3

Group 4

D

H

O

Weak HB p-Acceptors D

D

H

CH3

O

Weak HB Acceptors

H

O

(5.lVd)

Group 2

WD

A H (5.IIIc)

O

Weak HB Donors

:A

H D

(5.lVc)

Group 1

H

H

A1 (5.IIIb) A2

Conventional HBs D

Scheme 5

H

27

:WA

(5.Vc)

p-Bond C

C

H

D H

H C

C

D (5.Vd)

Nomenclature and classification of main-group-element H-bonds.

Weak H-bond donors (5.Vb), or of the acceptor, Group 3: Weak H-bond acceptors (5.Vc), and Group 4: Weak Hbond π-acceptors (5.Vd). Note that this classification can be expanded to also include the recently discovered Hbonds involving metal centers, a class of bonds which has been actively studied in the last 20 years95–100 and for which the reader is addressed to more general treatments.92, 93 H-bonds can also be characterized by their physical parameters, the most important being their strengths, quantifiable in terms of both dissociation energies, EHB , and dissociation enthalpies,
HDIS . Energies are normally obtained from ab initio or DFT quantum-mechanical calculations with full geometry optimization, though semiempirical methods based on experimental H-bond geometries are also widely used.101, 102 Large compilations of H-bond enthalpies are available, mostly obtained by van’t Hoff analysis of the equilibrium constants at different temperatures in nonpolar solvents81, 83, 86 or in the gas phase.103 As thermodynamic measurements are rather demanding, Hbond strengths are often estimated from other more accessible experimental quantities, the leading ones being (i) the red shift of the ν(D–H) stretching vibration as measured by IR spectroscopy; (ii) the chemical shift of the H-bonded proton, δ(D–H), as measured by 1 H NMR spectroscopy; (iii) the changes of the D–H· · ·:A group geometry as determined by X-ray and neutron crystallography or retrieved from the inorganic (ICSD) or organic/organometallic (CSD) structural databases.6–8 In this last case, H-bond strengths are mostly appreciated in terms of the D· · ·A contact dis = tance, dD···A , or, more accurately, of the sum dD···A dD – H + dH···A , a quantity which naturally accounts for the effects of D–H–A angle changes.93, 101, 102

The H-bond energy is, in itself, a rather mysterious affair (sometimes called the H-bond puzzle) because it is found to range from less than 1 to more than 45 kcal mol−1 , often without any evident reason. Jeffrey90 has classified all H-bonds in the three groups of weak, moderate, and strong H-bonds according to whether their energies are in the ranges 1–4, 4–15, and 15–45 kcal mol−1 . The physical properties of these three groups are summarized in Table 13 together with the most common types of H-bond donors, acceptors, and H-bonded complexes occurring in each group. The table illustrates the important fact that all H-bond properties are closely connected and that, in particular, energies and geometries are mutually related and encompassed between two extremes: (i) weak, long, dissymmetric, proton-out-centered, and mostly bent D–H· · ·:A bonds of most probable electrostatic nature; and (ii) strong, short, symmetric, protoncentered, and linear D· · ·H· · ·A bonds interpretable as three-center-four-electron interactions of mostly covalent nature. The point of view that the continuous transition from weak and asymmetric H-bonds to strong and proton-centered ones coexists with a gradual transformation from electrostatic to covalent nature of the bond itself is known as the electrostatic-covalent H-bond model (ECHBM ).104, 105

5.2.4 The nature of the H-bond. The dual H-bond model In summary, the entire data points to a model of the H-bond where, while the binding energy increases from less than 1 to about 45 kcal mol−1 , the binding force changes in nature from electrostatic to covalent. In consequence, the first stage

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

28

Supramolecular materials chemistry

Table 13

Physical characterization of the H-bonds and the most common types of H-bond donors, acceptors, and H-bonded complexes.

Strong H-bond nature

Moderate

mostly covalent D: · · ·H+ · · ·:A1/2−

electrostatic-covalent

electrostatic Dδ− –Hδ+ · · ·:Aδ−

D–H < H · · ·A 1.5–2.2 2.5–3.2 130–180 4–15 10–25% 2 O .67

phenols > alcohols and amides > anilines > amines and, in each class, the acidity can be enhanced by halogenation or nitration. Thiols (6.5 ≤ pKa ≤ 11), enols (8.5 ≤ pKa ≤ 12), and oximes

13

36

HB acceptors (:A) pKBH+ The pKa slide rule

9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49

D

A Medium strong

Medium H-bond strengths

H

Medium weak

∆pKa

A Weak

Evaluation of the H-bond strength for the O-H...O bond in the water dimer

Figure 18 An operational version of the pKa slide rule. See text for explanation. (Reproduced from Gilli et al., 2009;112  American Chemical Society, 2009). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals donor (upper scale) is put in vertical correspondence with water as an acceptor (central scale) allowing us to appraise (lower scale) that the
pKa is 17.4 (black harrow) and then that the H-bond formed will be of medium strength.

6

CONCLUSIONS

Some 900 000 crystal structures have been collected during the last century, which constitute an immense archive of atomic interactions that keeps track of all interatomic forces connecting atoms into molecules and molecules into supramolecular aggregates. Hence, crystal packing analysis may represent an invaluable tool for the study of the chemical bond in its various manifestations, provided a comprehensive classification of the forces acting within the crystal has been made previously available. In this chapter, this aim is pursued in two stages. In the first stage, the classical partitioning in metallic, covalent, ionic, and molecular crystals is critically reconsidered to establish a sound basis for the second stage, which is dedicated to a much more detailed analysis of the weaker noncovalent or nonbonded forces that determine the packing of molecular crystals. These are divided in two main groups: (i) mostly physical forces, namely, van der Waals nonbonded, multipolar electrostatic, and hydrophobic forces and (ii) mostly chemical forces, namely, D:→A chargetransfer (CT) or electron donor–acceptor (EDA) interactions, and D–H· · ·:A hydrogen bonds (H-bonds). The treatment of CT interactions includes their QM interpretation, the listing of the most common classes of CT donors, acceptors, and complexes, together with a more detailed analysis of some typical cases, such as halogen bonding and organic metals and semiconductors. H-bond properties are interpreted in terms of a novel comprehensive Hbond theory that takes advantage of the modern concepts of dual H-bond, PA/pKa equalization, and chemical leitmotifs (charge-assisted H-bond, CAHB; resonance-assisted H-bond, RAHB; and polarization-assisted H-bond, PAHB). Finally, the fact that X–H · · · :Y H-bonds can also be interpreted as X–H←:Y σ ∗←n CT interactions opens the interesting perspective of a complete merging of all the mostly chemical forces into a unique CT group where the electron becomes the unified particle (or token) of intermolecular interaction exchange.

REFERENCES 1. B. Russel, History of Western Philosophy, George Allen & Unwin, Ltd., London, 1961, p. 88. 2. M. Laue, Physik. Z., 1913, 14, 421–423.

37

3. W. L. Bragg, Proc. R. Soc. Lond., 1913, A89, 248–277. 4. W. H. Bragg and W. L. Bragg, Proc. R. Soc. Lond., 1913, A89, 277–291. 5. L. Pauling, The Nature of the Chemical Bond and the Structure of Molecules and Crystals: An Introduction to Modern Structural Chemistry, Cornell University Press, Ithaca, NY, 1939, 1940, 1960. 6. F. H. Allen, Acta Crystallogr., 2002, B58, 380–388. 7. F. H. Allen, S. Bellard, M. D. Brice, et al., Acta Crystallogr., 1979, B35, 2331–2339. 8. G. Bergerhoff, R. Hundt, R. Sievers, and I. D Brown, J. Chem. Inf. Comput. Sci., 1983, 23, 66–69. 9. L. D. Calvert and J. R. Rodgers, Computer Phys. Commun., 1984, 33, 93–98. 10. F. C. Bernstein, T. F. Koetzle, G. J. B. Williams, et al., J. Mol. Biol., 1977, 112, 535–542. 11. U. M¨uller, Inorganic Structural Chemistry, John Wiley & Sons, Inc., New York, 2006. 12. R. Hoffmann, Solids and Surfaces. A Chemist’s View of Bonding in Extended Structures, VCH Publishers, Inc., New York, 1988. 13. P. Drude, Ann. Phys., 1900, 306, 566–613. 14. P. Drude, Ann. Phys., 1900, 308, 369–402. 15. H. W. Kroto, J. R. Heath, S. C. O’Brien, et al., Nature, 1985, 318, 162–163. 16. H. W. Kroto, Angew. Chem. Int. Ed., 1992, 31, 111–129. 17. A. Bondi, J. Phys. Chem., 1964, 68, 441–451. 18. R. S. Rowland and R. Taylor, J. Phys. Chem., 1996, 100, 7384–7391. 19. F. London, Z. Phys. Chem., 1930, B11, 222–251. 20. F. London, Z. Phys., 1930, 63, 245–279. 21. R. A. Buckingham, Proc. R. Soc. Lond., 1938, A168, 264–283. 22. J. E. Lennard-Jones, Proc. R. Soc. Lond., 1931, 43, 461–482. 23. G. Filippini and A. Gavezzotti, Acta Crystallogr., 1993, B49, 868–880. 24. A. Gavezzotti and G. Filippini, J. Phys. Chem., 1994, 98, 4831–4837. (UNI–FF). 25. A. J. Pertsin and A. I. Kitaigorodsky, The Atom-Atom Potential Method. Applications to Organic Molecular Solids, Springer Verlag, Berlin, 1987. 26. A. Gavezzotti, Molecular Aggregation: Structure Analysis and Molecular Simulation of Crystals and Liquids, Oxford University Press, Oxford, 2007. 27. R. S. Paton and J. M. Goodman, J. Chem. Inf. Model., 2009, 49, 944–955. 28. N. L. Allinger, Y. H. Yuh, and J. H. Lii, J. Am. Chem. Soc., 1989, 111, 8551–8566. (MM3). 29. S. J. Weiner, P. A. Kollman, D. A. Case, et al., J. Am. Chem. Soc., 1984, 106, 765–784. (AMBER). 30. W. Damm, A. Frontera, J. Tirado-Rives, and W. L. Jorgensen, J. Comput. Chem., 1997, 18, 1955–1970. (OPLS).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

38

Supramolecular materials chemistry

31. T. A. Halgren, J. Comput. Chem., 1996, 17, 490–641. (MMFF94).

58. O. Hassel and C. Rømming, Q. Rev. Chem. Soc., 1962, 16, 1–18.

32. G. Del Re, J. Chem. Soc., 1958, 4031–4040.

59. O. Hassel, Science, 1970, 170, 497–502.

33. J. Gasteiger and M. Marsili, Tetrahedron, 1980, 36, 3219–3228.

60. H. A. Bent, Chem. Rev., 1968, 68, 587–648.

34. Z. Berkovitch-Yellin and L. Leiserowitz, J. Am. Chem. Soc., 1982, 104, 4052–4064. 35. W. G. H. Hol and R. K. Wierenga, The α-helix dipole and the binding of phosphate group of coenzymes and substrates to proteins, in X-ray Crystallography and Drug Design, eds. A. S. Horn and C. J. De Ranter, Clarendon Press, Oxford, 1984, pp. 151–168. 36. R. K. Wierenga, Biochemistry, 1985, 24, 1346–1357. 37. H. S. Frank and M. W. Evans, J. Chem. Phys., 1945, 13, 507–532. 38. C. Chothia, J. Mol. Biol., 1976, 105, 1–14. 39. D. Eisenberg and A. D. McLachlan, Nature (London), 1986, 319, 199–203. 40. R. B. Hermann, Proc. Natl. Acad. Sci. U S A, 1977, 74, 4144–4145. 41. J. A. Reynolds, D. B. Gilbert, and C. Tanford, Proc. Natl. Acad. Sci. U S A, 1974, 71, 2925–2927. 42. T. M. Raschke, J. Tsai, and M. Levitt, Proc. Natl. Acad. Sci. U S A, 2001, 98, 5965–5969. 43. K. N. Houk, A. G. Leach, S. P. Kim, and X. Xiyum, Angew. Chem. Int. Ed., 2003, 42, 4872–4897.

61. C. Ouvrard, J.-Y. Le Questel, M. Berthelot, C. Laurence, Acta Crystallogr., 2003, B59, 512–526.

and

62. A. C. Legon, Angew. Chem. Int. Ed., 1999, 38, 2686–2714. 63. A. C. Legon, 7736–7747.

Phys. Chem. Chem. Phys.,

2010,

12,

64. P. Metrangolo, H. Neukirch, T. Pilati, and G. Resnati, Acc. Chem. Res., 2005, 38, 386–395. 65. P. Metrangolo, F. Meyer, T. Pilati, et al., Angew. Chem. Int. Ed., 2008, 47, 6114–6127. 66. P. Metrangolo and G. Resnati, eds., Halogen Bonding: Fundamentals and Applications, Springer, Berlin, 2008. 67. H. L. Nguyen, P. N. Horton, M. B. Hursthouse, et al., J. Am. Chem. Soc., 2004, 126, 16–17. 68. P. Auffinger, F. A. Hays, E. Westhof, and P. S. Ho, Proc. Natl. Acad. Sci. U S A, 2004, 101, 16789–16794. 69. S. Fetzner and F. Lingens, Microbiol. Rev., 1994, 58, 641–685. 70. G. Valerio, G. Raos, S. V. Meille, et al., J. Phys. Chem., 2010, A104, 1617–1620. 71. T. Brink, J. S. Murray, and P. Politzer, Int. J. Quantum Chem., 1992, 44, 57–64.

44. R. S. Mulliken, J. Phys. Chem., 1952, 56, 801–822.

72. P. Politzer, P. Lane, M. C. Concha, et al., J. Mol. Model , 2007, 13, 305–311.

45. R. S. Mulliken and W. B. Person, Molecular Complexes. A Lecture and Reprint Volume, John Wiley & Sons, Inc., New York, 1969.

73. T. Clark, M. Hennemann, J. S. Murray, and P. Politzer, J. Mol. Model , 2007, 13, 291–298.

46. H. A. Benesi and J. H. Hildebrand, J. Am. Chem. Soc., 1948, 70, 3978–3981. 47. H. A. Benesi and J. H. Hildebrand, J. Am. Chem. Soc., 1949, 71, 2703–2707. 48. R. T. Sanderson, J. Chem. Educ., 1954, 31, 2–7. 49. R. T. Sanderson, Chemical Bonds and Bond Energy, 2nd edn, Academic Press, New York, 1976. 50. M. K. Kroeger and R. S. Drago, J. Am. Chem. Soc., 1981, 103, 3250–3262.

74. P. Politzer, J. S. Murray, and T. Clark, Phys. Chem. Chem. Phys., 2010, 12, 7748–7757. 75. P. Politzer and D. G. Truhlar, Chemical Applications of Atomic and Molecular Electrostatic Potentials, Plenum Press, New York, 1981. 76. M. Palusiak, J. Mol. Struct. Theochem, 2010, 945, 89–92. 77. A. Werner, Justus Liebig Ann. Chem., 1902, 322, 261–296. 78. W. M. Latimer and W. H. Rodebush, J. Am. Chem. Soc., 1920, 42, 1419–1433. 79. L. Pauling, J. Am. Chem. Soc., 1931, 53, 1367–1400.

51. V. Bertolasi, P. Gilli, and G. Gilli, Cryst. Growth Des., 2011, 11, 2724–2735.

80. M. L. Huggins, J. Org. Chem., 1936, 1, 405–456.

52. S. V. Rosokha and J. K. Kochi, Acc. Chem. Res., 2008, 41, 641–653.

81. G. C. Pimentel and A. L. McClellan, The Hydrogen Bond , Freeman, San Francisco, 1960.

53. F. H. Herbstein, Crystalline Molecular Complexes and Compounds: Structure and Principles, Oxford University Press, Oxford, 2005, vols. 1, 2.

82. W. C. Hamilton and J. A. Ibers, Hydrogen Bonding in Solids: Methods of Molecular Structure Determination, Benjamin, New York, 1968.

54. F. H. Herbstein and M. Kapon, Cryst. Rev., 2008, 14, 3–74.

83. G. C. Pimentel and A. L. McClellan, Ann. Rev. Phys. Chem., 1971, 22, 347–385.

55. J. Simon and J.-J. Andr´e, Molecular Semiconductors. Photoelectrical Properties and Solar Cells, Springer Verlag, Berlin, 1984. 56. T. J. Kristenmacher, T. E. Phillips, and D. O. Cowan, Acta Crystallogr., 1974, B30, 763–768. 57. F. Guthrie, J. Chem. Soc., 1863, 16, 239–244.

84. S. N. Vinogradov and R. H. Linnell, Hydrogen Bonding, Van Nostrand-Reinhold, New York, 1971. 85. M. L. Huggins, Angew. Chem. Int. Ed., 1971, 10, 147–152. 86. M. D. Joesten and L. J. Schaad, Hydrogen Bonding, Marcel Dekker, New York, 1974.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

Noncovalent interactions in crystals 87. P. Schuster, G. Zundel, and C. Sandorfy, eds., The Hydrogen Bond: Recent Developments in Theory and Experiments, North-Holland, Amsterdam, 1976, vols. I–III. 88. J. Emsley, Chem. Soc. Rev., 1980, 9, 91–124. 89. G. A. Jeffrey and W. Saenger, Hydrogen Bonding in Biological Structures, Springer Verlag, Berlin, 1991. 90. G. A. Jeffrey, An Introduction to Hydrogen Bonding, Oxford University Press, Oxford, 1997. 91. S. Scheiner, Hydrogen Bonding: A Theoretical Perspective, Oxford University Press, New York, 1997. 92. G. R. Desiraju and Th. Steiner, The Weak Hydrogen Bond in Structural Chemistry and Biology, Oxford University Press, Oxford, 1999. 93. G. Gilli and P. Gilli, The Nature of the Hydrogen Bond: Outline of a Comprehensive Hydrogen Bond Theory, Oxford University Press, Oxford, 2009. 94. P. Gilli and G. Gilli, J. Mol. Struct., 2010, 972, 2–10. 95. L. Brammer, D. Zhao, F. T. Ladipo, and J. BraddockWilking, Acta Crystallogr., 1995, B51, 632–640. 96. L. Brammer, 3145–3157.

J.

Chem.

Soc.

Dalton

Trans.,

2003,

39

107. P. A. Frey, S. A. Whitt, and J. B. Tobin, Science, 1994, 264, 1927–1930. 108. E. P. Hunter and S. G. Lias, Proton Affinity Data, in NIST Chemistry WebBook , NIST Standard Reference Database Number 69, eds. P. J. Linstrom and W. G. Mallard, National Institute of Standard and Technology, Gaithersburg, MD, 2005. 109. A. E. Martell, R. M. Smith, and R. J. Motekaitis, NIST Critically Selected Stability Constants of Metal Complexes, NIST Standard Reference Database Number 46, Version 8.0, National Institute of Standard and Technology, Gaithersburg, MD, 2004. 110. P. Gilli, V. Bertolasi, L. Pretto, et al., J. Am. Chem. Soc., 2004, 126, 3845–3855. 111. P. Gilli, L. Pretto, and G. Gilli, J. Mol. Struct., 2007, 844–845, 328–339. 112. P. Gilli, L. Pretto, V. Bertolasi, and G. Gilli, Acc. Chem. Res., 2009, 42, 33–44. 113. P. Gilli, V. Ferretti, V. Bertolasi, and G. Gilli, in Advances in Molecular Structure Research, eds. M. Hargittai and I. Hargittai, JAI Press, Greenwich, CT, 1996, vol. 2, pp. 67–102.

97. L. M. Epstein and E. S. Shubina, Coord. Chem. Rev., 2002, 231, 165–181.

114. G. Gilli, F. Bellucci, V. Ferretti, and V. Bertolasi, J. Am. Chem. Soc., 1989, 111, 1023–1028.

98. N. V. Belkova, E. S. Shubina, and L. M. Epstein, Acc. Chem. Res., 2005, 38, 624–631.

115. V. Bertolasi, P. Gilli, V. Ferretti, and G. Gilli, J. Am. Chem. Soc., 1991, 113, 4917–4925.

99. G. J. Kubas, Metal Dihydrogen and σ -Bond Complexes, Kluwer Academic/Plenum Publishers, New York, 2001.

116. G. Gilli, V. Bertolasi, V. Ferretti, and P. Gilli, Acta Crystallogr., 1993, B49, 564–576.

100. V. I. Bakhmutov, Dihydrogen Bonds: Principles, Experiments, and Applications, John Wiley and Sons Inc., Hoboken, NJ, 2008.

117. V. Bertolasi, P. Gilli, V. Ferretti, and G. Gilli, Acta Crystallogr., 1995, B51, 1004–1015.

101. E. R. Lippincott and R. Schroeder, J. Chem. Phys., 1955, 23, 1099–1106. 102. R. Schroeder and E. R. Lippincott, J. Phys. Chem., 1957, 61, 921–928. 103. M. Meot-Ner (Mautner) and S. G. Lias, Binding energies between ions and molecules and the thermochemistry of cluster ions, in NIST Chemistry WebBook , NIST Standard Reference Database Number 69, eds. P. J. Linstrom and W. G. Mallard, National Institute of Standard and Technology, Gaithersburg, MD, 2005.

118. V. Bertolasi, P. Gilli, V. Ferretti, and G. Gilli, Acta Crystallogr., 1998, B54, 50–65. 119. V. Bertolasi, P. Gilli, V. Ferretti, and G. Gilli, Chem. Eur. J., 1996, 2, 925–934. 120. P. Gilli, V. Bertolasi, V. Ferretti, and G. Gilli, J. Am. Chem. Soc., 2000, 122, 10405–10417. 121. P. Gilli, V. Bertolasi, L. Pretto, et al., J. Am. Chem. Soc., 2002, 124, 13554–13567. 122. P. Gilli, V. Bertolasi, L. Pretto, et al., J. Am. Chem. Soc., 2005, 127, 4943–4953.

104. P. Gilli, V. Bertolasi, V. Ferretti, and G. Gilli, J. Am. Chem. Soc., 1994, 116, 909–915.

123. P. Gilli, V. Bertolasi, L. Pretto, and G. Gilli, J. Mol. Struct., 2006, 790, 40–49.

105. G. Gilli and P. Gilli, J. Mol. Struct., 2000, 552, 1–15.

124. G. Gilli and V. Bertolasi, Structural chemistry of enols, in The Chemistry of Enols, ed. Z. Rappoport, John Wiley & Sons, Chichester, UK, 1990, Chapter 13, pp. 713–764.

106. W. W. Cleland and M. M. Kreevoy, Science, 1994, 264, 1887–1890.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc110

The Cambridge Structural Database System and Its Applications in Supramolecular Chemistry and Materials Design Elna Pidcock1 , James A. Chisholm1 , Peter A. Wood1 , Peter T.A. Galek1 , L´aszl´o F´abi´an2 , Oliver Korb1 , Aurora J. Cruz-Cabeza1 , John W. Liebeschuetz1, Colin R. Groom1 , and Frank H. Allen1 1 2

Cambridge Crystallographic Data Centre (CCDC), Cambridge, UK University College Cork, Cork, Ireland

1 2 3 4

Introduction The Cambridge Structural Database CCDC Software for Supramolecular Applications The Materials Module of Mercury: Development and Applications 5 Materials Design and Drug Discovery: Two Sides of the Same Coin 6 Cocrystals: Descriptors of Formation and Their Predictive Value 7 Applications of the CSD in Crystal Structure Prediction (CSP) 8 Conclusion References

1

2927 2928 2929 2933 2938 2941 2943 2944 2945

INTRODUCTION

The essential link between Lehn’s definition of supramolecular chemistry: the designed chemistry of the intermolecular bond 1 and crystal structures was made by Dunitz2 : A crystal is, in a sense, the supermolecule par excellence: a lump of matter, of macroscopic dimensions, millions of molecules

long, held together in a periodic arrangement by just the same kind of interactions as are responsible for molecular recognition and complexation at all levels—ion–ion, ion–dipole, dipole–dipole interactions, hydrogen bonding, London forces, and so on. In its turn, this statement defines the crystallographic databases3–7 as libraries of experimental observations of 3D supermolecules and clearly indicates the areas of fundamental research for which these databases are uniquely suited. No experimental technique except crystallography comes close to providing such direct and precise observations of such diverse supramolecular building blocks and architectures: the local and extended motifs formed by noncovalent interactions, their chemical constitutions, and their geometrical characteristics. Within this context, the subdiscipline of crystal engineering8 has seen a renaissance and a broadening of scope9, 10 during the past 15 years, generating a focus for crystallographic supramolecular research that now has its own voice through the successful introduction of specialist journals such as Crystal Growth and Design and CrystEngComm. The crystallographic databases are heavily used and frequently referenced there, and in other journals that embrace papers in crystal engineering, as sources of crystallographic knowledge and supramolecular systematics. In mid-2010, the number of structures available in the crystallographic databases exceeded 700 000, covering the chemical spectrum from monatomic metals to proteins and viruses. The vast majority of these structures arise from

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

2

Supramolecular materials chemistry

formal journal publications, although it is well known, and to the detriment of science, that many thousands of structures determined annually are not published and for a variety of reasons.11 Nevertheless, the crystal structure information base is large, and increases at around 8% per annum: more than 50 000 novel structures will be archived in 2010. Compilation of the crystallographic databases began in the late 1960s and early 1970s and they were among the first compilations of numerical scientific data. If they had not been created early, when just a few thousand structures were known, then they might not have been created at all. This chapter concentrates on the Cambridge Structural Database (CSD7 ) of organic and metal-organic structures, its scientific research applications in the broad area of supramolecular chemistry, and, particularly, on software applications developed at the Cambridge Crystallographic Data Centre (CCDC) that underpin those applications. The CSD is the largest of the crystallographic databases, having recently recorded its 500 000th entry—about 70% of all published crystal structures of any type. The CSD benefits from massive chemical diversity, as crystallography is the method of choice for determining the structures of novel molecules: common molecules proliferate, but unusual molecules are commonplace. This broad structural content makes the CSD crucial in crystal engineering and supramolecular chemistry, and the database has been heavily used for systematic studies of hydrogen bonds and other interactions, including their geometry, motif formation, and the competition between different potential interactions. Supramolecular knowledge derived from the CSD is therefore a key component in the design of solid forms, and has vital practical applications in drug development and formulation. At the molecular level, conformational information from the CSD is also vital in constructing starting models for computational polymorph prediction and for structure solution from powder diffraction data. Allied to this, the CSD is heavily used in structural biology and drug discovery, where supramolecular knowledge underpins studies of protein–ligand interactions and protein–ligand docking. Indeed, drug discovery, crystal engineering, and materials development have many close parallels, so close that two recent guides to intermolecular interactions written for medicinal chemists by Stahl and coworkers12, 13 will also resonate with supramolecular chemists and crystal engineers. Many of these varied applications are highlighted in this chapter, and are facilitated by the specialist software and knowledge bases that are available as part of the distributed CSD System, by Internet applications, or by standalone programs in crystallography and the life sciences.

2 2.1

THE CAMBRIDGE STRUCTURAL DATABASE Information content

The CSD records the crystal structures of organic and metal-organic compounds determined by single-crystal X-ray or neutron diffraction, and around 1800 fully refined powder structure determinations are also included. Compilation of the CSD began in 1965 and the principal objective remains to record the key numerical results of each analysis: the 3D atomic coordinates, unit cell parameters, and space group for each structure—the fundamental data which provide structure visualizations and from which molecular and intermolecular geometry can be calculated. Each CSD entry also contains (i) the formal 2D chemical diagram and a searchable chemical connection table, (ii) full bibliographic reference, including the DOI, (iii) chemical name(s) and formulae, and (iv) other relevant information that may be present in the published report including inter alia: experimental temperature or pressure, melting point, solvent of crystallization, crystal color and habit, polymorphism and phase transitions, and X-ray determination of absolute configuration. Each CSD entry is given a unique identifier, the CSD reference code (refcode), containing six letters and a possible two additional digits, for example, ABCDEF, PQRSTU03, and refcodes are used as structural mnemonics in this chapter. At the time of writing, work is underway to extend this information content, particularly to incorporate additional data items that can be generated from deposited electronic information.

2.2

Database creation and statistical overview

The vast majority of CSD entries arise from formal journal publications, but the current database does contain over 6000 structures that have been deposited directly by crystallographers and are unavailable elsewhere. The CSD covers more than 1300 literature sources and, since the mid-1990s, an electronic crystallographic information file (CIF—the internationally agreed format14 ) is normally deposited with the journal and the CCDC. Data for more than 220 000 earlier non-CIF structures were manually keyed from printed journals or supplementary information and occasional structures are still entered in this way. The CCDC now has formal arrangements with more than 100 major journals, interacting with journal editors, providing CCDC deposition numbers for inclusion in published papers, and acting as the accredited data repository for some journals. Webbased CIF deposition mechanisms were introduced in the late 1990s and are continually being upgraded. Electronic access to the original deposited CIFs is freely available via

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications the “Request-a-Structure” facility on the CCDC Website at www.ccdc.cam.ac.uk. All data entering the CSD are subject to thorough evaluation designed to detect data omissions or numerical and factual errors in the original CIF or published paper. Owing to the CCDC’s journal relationships, many of these issues are dealt with before publication. Other problems are resolved post-publication in collaboration with the authors. Any changes to the original published data are clearly flagged in the CSD entry. Data-processing software is currently undergoing significant improvements to make maximum use of encrypted structural knowledge and scientific expertise in data evaluation and database creation that has accrued over the past 40 years. These improvements are essential in light of continually increasing data volumes and structural complexity. Figure 1 shows the growth of the CSD since 500000

400000

300000

200000

100000

0 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008

Figure 1 Cumulative annual growth of the Cambridge Structural Database 1971–2009 in terms of crystal structure entries. Table 1

CSD summary statistics (1 January 2010). Structures

Total number of structures Number of different compounds Number of literature sources Organic structures Transition metal present Li–Fr or Be–Ra present Main group metal present 3D coordinates present Neutron studies Powder diffraction studies Low/high temperature studies Absolute configuration determined Polymorphic structures Number of atoms with 3D coordinates

501 857 456 701 1336 215 106 266 333 25 739 31 470 462 146 1437 1768 192 352 7672 16 253 36 334 442

1971. The current year-on-year growth rate is just over 8% (≈40 000 new entries expected in 2010), giving a database doubling period of around nine years. Table 1 lists general CSD statistics (January 2010), showing the breakdown of entries in terms of their broad chemistry and type of study. More detailed current CSD statistics are collated from time to time (at least annually) on the CCDC website at www.ccdc.cam.ac.uk.

3

100.0 — — 42.9 53.1 5.1 6.3 92.0a 0.3 0.4 38.3 1.5 3.2 —

a Taken as a percentage of structures for which 3D coordinates are present in the CSD.

CCDC SOFTWARE FOR SUPRAMOLECULAR APPLICATIONS

This section begins by discussing the distributed software and internet applications that are directly involved in searching the CSD and in visualizing and analyzing CSD structures and information, as well as two knowledge bases that provide rapid access to geometrical information derived from the CSD. Together with the CSD itself, these software products and knowledge bases are known as the CSD System and are summarized in Section 3.1. Novel ongoing developments in Mercury that are aimed specifically at materials design are described and exemplified in Section 4 in more detail. Later parts of Section 3 introduce (i) programs that are used extensively in drug discovery for exploring protein–ligand interactions and for protein–ligand docking (Section 3.2) and (ii) software for structure determination from powder diffraction data (Section 3.3), and a number of other software and database services that can be accessed via the CCDC website (Section 3.4). Software products in these sections are also finding important uses in crystal engineering and solidstate formulation, as exemplified in Section 5.

3.1 % CSD

3

CSD system software

3.1.1 ConQuest ConQuest 15 permits searches of all CSD information fields. The most common searches use the 2D chemical connectivity records to locate substructural fragments or specific complete structures. Connectivity information can be combined with 3D geometrical constraints to locate, for example, substructures having specific conformation(s) and, most importantly for a crystallographic database, to locate hydrogen bonds and other intermolecular interactions. Searches of other fields, for example, bibliographic records and other CSD information items, can be performed individually, or in Boolean combination with other searches including connectivity searches. ConQuest displays CSD hit information, including 2D and 3D graphics and numericals, text fields, and DOI links, for browsing by the user.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

4

Supramolecular materials chemistry

Apart from locating relevant crystal structures, ConQuest also outputs user-defined geometrical descriptors for all occurrences of the structure or substructure of interest. These data may be analyzed using the CCDC’s Vista program (Section 3.1.3), or can be read by Microsoft Excel and other statistical packages.

3.1.2 Mercury Mercury 15–17 is the CCDC’s 3D structure visualizer. It has all the expected structural graphics facilities, for example, various atom and bond display styles, atom and bond coloring options, atom labeling, structure rotation and translation, and view directions. Mercury also has many special features to calculate, display, and manipulate crystal structure information, including facilities to (i) view crystal packing across any number of unit cells, display intermolecular interactions, generate and visualize H-bonded motifs, display extended intermolecular interaction patterns, and locate and display structural voids; (ii) calculate and display centroids and least squares and Miller planes; (iii) display crystallographic symmetry elements, atomic displacement ellipsoids, simulated powder patterns, and the predicted crystal morphology; (iv) measure and display interatomic distances, angles, and torsions; (v) display and overlay multiple structures; and (vi) display full CSD entry information. Mercury is fully linked to ConQuest (e.g., to highlight substructures located in a search), as well as to the CCDC’s knowledge bases Mogul18 and IsoStar19 summarized in Section 3.2. Currently, Mercury is being further enhanced for use in crystal engineering and materials design, and these improvements are discussed and exemplified in Section 4.

3.1.3 Vista Vista permits geometrical parameters retrieved from the CSD by ConQuest to be displayed as a spreadsheet, and presented as histograms or scattergrams using either Cartesian or polar axes, together with summary statistics for the variables involved. A number of statistical functions are built into Vista, including linear regression and principal components analysis, and publication quality graphics can also be generated. At the time of writing, a new set of software tools was being developed to supersede Vista and these tools will be released early in 2011.

3.1.4 WebCSD The WebCSD 20 application makes the CSD accessible via the Internet and local Intranet facilities using standard browsers—no local software installations are required. In addition to full text and numeric searching, including

reduced-cell searching, WebCSD allows 2D chemical substructure queries to be defined using an embedded sketcher as in ConQuest—thus allowing rapid retrieval of structures of interest. A 2D structure-based similarity search option is also available. Crystal structure information is accessed either through searches, or by simply browsing the CSD, and is accessible in a single pane (Figure 2). This display can show a 3D visualization, the 2D chemical diagram, and full bibliographic and text information with DOI links. WebCSD provides a choice of two different 3D viewers as embedded Java applets: Jmol21 and OpenAstexViewer.22 These viewers provide a range of display styles and functions, as well as tools to measure distances, angles, and torsions. The default viewer, Jmol, also supports a number of crystallographic options. Users may also download specific CSD entries for use as input to locally installed visualizers such as Mercury.

3.1.5 Knowledge-based libraries derived from the CSD Two basic uses of the CSD are to study intramolecular geometry, particularly the conformational preferences exhibited by molecules and chemical fragments, and intermolecular interactions, including hydrogen-bonding and other interactions. To provide rapid “click of a button” access to standard derived information, the CSD System also contains two major knowledge-based libraries. Mogul Mogul 18 is a knowledge base of intramolecular geometry containing >20 million bond lengths, valence angles, and torsions organized into >1.5 million searchable distributions, each relating to a specific chemical environment. Structures can be drawn in the Mogul interface or imported as 3D models in a variety of common formats. Clicking on two, three, or four bonded atoms in the model then generates a distribution of bond lengths, valence angles, or torsions for the specific chemical environment. Where this yields a small population, the software can extend the search to similar chemical environments for acceptance/rejection by the user. Figure 3 shows the distribution of Car -Car -S-C torsions in arylsulfones. The reference torsion from the input model is shown by the vertical line, and there is full hyperlinking from the distribution back to individual CSD entries. Options also exist to generate parameter distributions for all bonds, angles, or torsions in an input model. This is valuable in checking the geometry of a novel crystal structure during refinement (as in the CRYSTALS package,23 ) or to check the geometry of a protein–ligand docking pose, against accumulated experimental knowledge. The importance of crystal structure conformations in drug design is stressed in a recent review.24

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications

5

Figure 2 WebCSD results pane showing CSD information for caffeine: 3D structure and text information. The chemical diagram can be viewed by clicking on the tab. Mogul search - Torsion angle - C6 C1 S1 C7 30

Value in query: 77.337°

Number of hits

24

18

12

6

0 0

45

90 Torsion angle (°)

135

180

Click to (de)select bars; click and drag to (de)select a range

Figure 3

Distribution of Car -Car -S-C torsion angles in arylsulfones from the Mogul knowledge base.

IsoStar IsoStar 19 is a library of visual and numerical information about nonbonded interactions. IsoStar provides 2D scatterplots showing the distribution of a contact group (e.g., an H-bond donor or other functional group) around a central group, as exemplified by the “native” and contoured

plots of N–H donors around an amide central group shown in Figure 4(a) and (b). IsoStar data are derived (i) from the CSD (>20 000 scatterplots), (ii) from protein–ligand ˚ complexes in the PDB6 having a resolution better than 2 A (>5500 scatterplots), and (iii) ab initio theoretical energy minima for more than 1500 key interactions. The IsoStar

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

6

Supramolecular materials chemistry

ASN568

(a)

(b)

Figure 4 IsoStar scatterplots of an N–H contact group around an amide central group: (a) standard plot (all contacting N–H bonds shown) and (b) contoured plot.

dialogue pages illustrate the 48 contact groups and 300 central groups that are available for selection. IsoStar also provides geometry statistics and sliders to adjust nonbonded distance ranges. All scatterplot points are hyperlinked to the original CSD structure. IsoStar is a browser-based application: the library files may be hosted locally or can be accessed over the Internet from CCDC servers.

3.2

Drug discovery: protein–ligand interactions and protein–ligand docking

3.2.1 Superstar SuperStar 25 is a program for generating maps of interaction sites in protein-binding cavities or around small molecules for a selection of functional group probes. SuperStar maps are entirely knowledge based, using experimental information about supramolecular interactions derived from the CSD and the PDB and stored in IsoStar19 (Section 3.1.5, IsoStar). In Figure 5, Superstar has been used to predict the preferred interaction points (shown in gold) of a C=O group within the binding cavity of tyrosine kinase. These positions are overlayed with the structure of an oxindolebased inhibitor present in the structure of the kinaseinhibitor complex in the PDB.6 The positions of the two inhibitor C=O groups are well predicted, together with other areas of the cavity that might also be attractive to additional C=O groups in a modified inhibitor structure.

3.2.2 GOLD GOLD26 is a protein–ligand docking program that uses a genetic algorithm to dock novel ligands into a protein active site, and uses information from the CSD relating to Hbonds to help direct the docking operation. GOLD permits full ligand flexibility, partial protein flexibility, automatic consideration of cavity-bound water molecules, and allows the user to study an ensemble of protein conformations simultaneously. A wide choice of scoring functions is

ALA564

Figure 5 Interaction “hot spots” (in gold) for a C=O group probe within the binding site of tyrosine kinase, with the experimentally determined position of an oxindole inhibitor superimposed. The correspondence of the two inhibitor C=O groups with two of the SuperStar hot spot predictions is clearly observed.

available, and the software has been extensively validated.27 More general applications of GOLD in materials design are discussed in Section 5.1.

3.3

DASH: structure solution from powder diffraction data

DASH28 uses direct space methods to solve structures from X-ray powder patterns and guides the user through all the necessary steps, including indexing, space group determination, Pawley fitting, simulated annealing, and Rietveld refinement. During simulated annealing, many thousands of trial structures are generated by altering the position, orientation, and conformation of the molecule. For more complex problems, for example, multicomponent systems or molecules with >10 flexible torsions, external knowledge can reduce the search space, and DASH is directly linked with Mogul18 (Section 3.1.5, Mogul) to provide conformational knowledge. Thus, use of Mogul in DASH was vital in determining the structure of the industrially important β-form of pigment yellow 181,29 for which single-crystal analysis has so far been unsuccessful and which contains 10 torsions. Structure solution with torsional bounds gave a success rate of ∼20% (11/50) compared to the 2% (1/50) when all torsion angles were allowed to access the full 360 ◦ range. In a DASH study (E. Pidcock, Manuscript in preparation) of some pharmaceutically relevant molecules with differing numbers of flexible torsions, application of modal torsion angle bounds always increased the number of correct solutions, and increased the efficiency of the simulated annealing procedure by approximately 40%.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications

3.4

Other web-based facilities

4.1

In addition to the “Request-a-Structure” facility and hosting the library files for the IsoStar client-server application, the CCDC also provides several other programs and databases for free download or web access as detailed at www.ccdc.cam.ac.uk/free services/. These include the widely used RPluto program30 and a version of the Mercury visualizer, together with (i) CSDSymmetry,31 a downloadable database containing information about molecular and crystallographic symmetry within the CSD; and (ii) WebCite, a searchable database of literature references to research studies in which the CSD was the principal information source. The “free services” section of the CCDC website also contains extensive information regarding applications of the CSD in chemistry teaching, including details of a 500-entry teaching subset and a number of teaching modules.32–34

4

7

Motif topology searches

4.1.1 Development Graph sets are important and well-established topological descriptors of hydrogen-bonded arrays36, 37 and this notation can now be generated automatically by both RPluto30 and Mercury.17 Now, Mercury can also search for interaction motifs formed between specific functional groups. Two options are provided: (i) motifs are selected from a list taken from the literature that have proved useful in crystal engineering, or (ii) functional groups can be specified and Mercury will identify which motifs form and how common they are. Analysis (ii) could be carried out manually by sketching and then searching (ConQuest) for all possible motifs. However, the number of possibilities can be high and the manual approach can be prone to error. Thus, Mercury automatically generates motif search queries: the user selects or sketches one or more functional groups and close contacts and the program uses a combinatorial algorithm to assemble the groups and contacts into all possible topological combinations or sequences to generate dimer motifs, together with ring and chain motifs of user-specified sizes. Figure 6 shows some possible ring motifs that could form between a carboxylate and a protonated amino group, and which were generated to study the interaction preferences of aliphatic amino acids.38

THE MATERIALS MODULE OF MERCURY: DEVELOPMENT AND APPLICATIONS

The rational design of solid-state forms is a topic of major current interest, particularly in the development of marketable formulations of novel active pharmaceutical ingredients (APIs). Many drug formulations are crystalline and are often delivered as salts or cocrystals with permitted excipients, but usually avoiding solvates as far as possible. The CSD is the ultimate library of solid-state forms, and recent developments in Mercury17 permit scientists from a wide variety of backgrounds to better understand the factors affecting the formation and stability of crystalline solids. Examples of relevant research are described in a recent review35 which discusses, inter alia, salt selection, hydrogen bond competition effects, hydrated versus anhydrous structures, and motif searching. In this section, we describe the current status of ongoing major developments in the Materials module of Mercury together with some example applications of the new functionality in supramolecular materials design.

4.1.2 Applications This functionality means that motifs of different topologies can be located concurrently, with the results grouped by motif geometry, making it trivial to assess, for example, the ratio of dimers to infinite chains formed by transamide groups. A further advantage is the presentation of the frequency of occurrence of each motif geometry, allowing immediate assessment of whether an interaction is common or not, for example, when examining a structure for the likelihood of polymorphism. Since the well-publicized case of ritonavir,39 the presence of uncommon weak interactions in an observed

O −

H + H N H −

O

O

O −

O H+ N H H

H N H O

−O

O

O

H + O H

− +

N H

H

H O H

N H

+

+

H

H

H

+

H

+

H N H

O H

N

N H

−

+

N H

H

H

H

−

O

O

−

O

O −

O O

O

H

+

H N H

−

O

Figure 6 Some possible H-bonded ring motifs involving a protonated amino group and a carboxylate group and generated automatically by the Materials module of Mercury. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

8

Supramolecular materials chemistry Table 2 Frequencies of occurrence (f ) found for donor and acceptor interactions in Leflunomide, using the motif search tool in the Materials module of Mercury.

N O (a)

CH3

NH O

CF3

Donor Amide NH Amide NH

Acceptor

f (%)

Amide C=O Isoxazole N

25.3 21.4

N(obs) 4921 39

N (obs) is the number of observations, and the search is based on v5.31 of the CSD. (b)

(c)

Figure 7 Leflunomide (a) and its two polymorphic forms in the CSD: (b) the stable form (Z  = 2) (VIFQIL40 ) and (c) the metastable form (VIFQIL0140 ).

polymorph is viewed as a warning of the potential for alternative packing arrangements involving stronger interactions (Section 4.4.2). The relative frequencies of occurrence of interactions can also be extremely useful when trying to engineer interactions in crystals. Thus, the motif search can be used to assess the likelihood of polymorphism in the antirheumatic agent leflunomide (Figure 7a), which has a single strong donor, the amido-NH, and two acceptor sites, the amide C=O and the isoxazole-N. With a single donor, the possible outcomes for leflunomide are a solid form showing the amide–amide interaction, a solid form showing the amide–isoxazole interaction, and a form with Z  > 1 exhibiting both interactions. The motif tool shows that both the trans-amide homosynthon and the amide-NH· · ·isoxazole-N interaction occur in the CSD with similar frequencies (Table 2). Therefore, neither interaction is clearly preferred and multiple polymorphs may be likely. In fact, two forms of leflunomide are already known (Figure 7b and c): the more stable form (VIFQIL40 ) is a Z  = 2 structure having both the trans-amide homosynthon and the amide-NH· · ·isoxazole-N interaction, while the metastable polymorph (VIFQIL0140 ) shows only the amide–isoxazole interaction. Other applications include evaluating synthon competition to aid salt selection41 and assessing putative structures from crystal structure prediction (CSP).42

(a)

(b)

4.2

Packing feature searches

4.2.1 Development Although much work has focused on robust hydrogen bonds, other packing features include dipole–dipole, π –π and cation–π interactions and halogen bonds, while shapefitting considerations are also important. Questions that are frequently asked are “how common is this feature?” and “what is its preferred geometry?” So Mercury now provides a packing feature search. This is a very general search where the user defines any structural feature of interest in a displayed structure by selecting the atoms and bonds that define the feature. Although the name suggests the selection of intermolecular features, the selection can be intramolecular, such as a specific arrangement of functional groups within a molecule. The program automatically constructs a search query from the user’s selection without any need for query sketching. Chemical constraints are applied on the basis of the selected atom and bond types and geometric constraints are based on the interatomic distances and angles, which are chosen to best represent or capture the relative position and orientation of the atoms.42 Figure 8(a) and (b) shows an example of the packing feature, with dotted lines showing the selected distances. The CSD is then searched to identify crystal structures that contain a similar spatial arrangement of atoms using the 3DSEARCH algorithm.43 Once hits are found they are automatically overlaid with the original selected packing feature, and a quantitative measure of geometrical similarity (a root mean square deviation (RMSD) value) is reported. Thus, in Figure 8(c), the hit structure (AFACAL44 ) contains the

(c)

Figure 8 (a) Selection of a packing feature in the Materials module of Mercury, (b) conversion of user’s selection into a search query defined in part by a number of interatomic distance constraints, and (c) an example of the packing feature as located in AFACAL.45 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications

(a)

(b)

Figure 9 Packing feature searches (a) the saccharinate salt structure (VAWPIT45 ) used as a search template, and (b) three hits representing examples of potential counterions located by the search.

˚ same packing feature as the query, with RMSD = 0.4 A obtained from overlaying the 21 matching atoms.

4.2.2 Applications Packing feature searches provide unique insights when the spatial arrangement of structural features or functional groups is crucial. For example, in the search for another solid formulation for lamivudine, a nucleoside analog reverse transcriptase inhibitor, the existing saccharinate salt structure (VAWPIT,45 Figure 9a) can be used as a template. By selecting the three acceptor groups of the saccharinate, along with two carbon atoms to provide some hydrophobic filler, other molecules which contain these features can easily be located in the CSD. Three examples in Figure 9(b) show that a range of possibilities for potential counterions are on offer, depending on whether significant or small changes to the saccharinate structure are desired. In experiments performed by Martins et al.,46 the maleate salt was obtained, and the lamivudine framework was not perturbed significantly due to the similarity in size and functional features of the saccharinate and maleate ions.

4.3

Crystal packing similarity

4.3.1 Development Related crystal structures, such as polymorphs, often display a degree of similarity—perhaps a common molecular dimer, a one-dimensional stack, or a two-dimensional layer. Locating and examining such recurring features

9

in polymorphs, multicomponent systems, and structures of related compounds can provide insights into the supramolecular chemistry, packing preferences, and structural relationships between crystals.47, 48 The similarity tool in Mercury will analyze sets of related structures to provide measures of similarity and obtain automatic overlays of common molecular packing regions. This feature can also identify whether two crystal structures are the same, an important (and surprisingly tricky) consideration in CSP,49 and can also identify isostructural solvates. Similarity measures include the number of molecules that can be overlaid and the positional RMSD for the overlaid atoms. The method employed in Mercury uses the COMPACK algorithm.42 Other tools are also available such as XPac50 and CRYCOM.51 To begin the search, a finite cluster of molecules is generated from one of the structures (the “reference” structure). By default, this consists of a central molecule plus a coordination shell containing the nearest 14 molecules. The size of the cluster can be increased to improve the representation of molecular packing. For Z  > 1, the program can generate all symmetry-unique clusters. These clusters are then converted into a search query using the same method as that used for packing feature searches (Section 4.2.1). Chemical constraints are based on atom and bond types, and geometric constraints are based on interatomic distances and angles. A default tolerance of 20% is placed on distance constraints and 20 ◦ on angle constraints but these are adjustable. The search is then performed on the target structure using the 3DSEARCH algorithm,43 which has been developed to allow partial matching. Thus, if all 15 molecules cannot be located, the algorithm persists to find the maximum number of molecules that can be matched. It is also possible to compare structures that contain nonidentical compounds. To achieve this, a maximum common subgraph algorithm is used to identify commonality between molecules. The algorithm then uses only the common part to search for similarity in molecular packing.

4.3.2 Applications This tool is particularly useful for identifying similarities between polymorphs, salts, hydrates, cocrystals, or any family of structures where either one component is the same or there is some chemical similarity between the molecules of interest. Establishing the similarity between crystal structures generated as the result of CSP trials is a typical use: thousands of potential structures are generated by CSP methods and some of these structures, despite having different space groups, are identical. The packing similarity tool is able to cluster structures having a high degree of similarity hence identifying likely members of predicted polymorphic families.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

10

Supramolecular materials chemistry

(a)

(b)

Figure 10 The three packing motifs identified in carbamazepine cocrystals,48 using the packing feature tool of the Materials module of Mercury. Each motif is shown as a cartoon (a), and a structure, exemplifying the motif, is shown (b).

The tool is also valuable in identifying packing features common to different structures such as a dimer interaction or a recurring stack of molecules. A study of the packing modes of 50 carbamazepine cocrystals was carried out,48 and all 50 structures could be classified as having one of three different packing motifs (Figure 10). These motifs were not based on hydrogen-bonding interactions, and therefore a similarity determination based only on geometric relationships between molecules was invaluable. The identification of the packing motifs led to the conclusion that the interactions defined by the packing motifs were competitive with H-bonded interactions, and that the cocrystal structures could be rationalized in terms of a compromise between H-bonding and packing requirements.

4.4

Hydrogen bond propensity

4.4.1 Development The assessment of crystal packing through H-bond analysis is a central theme of crystal engineering,10, 52–54 and the frequent occurrence of suitable donors (D) and acceptors (A) in drug molecules makes H-bonding particularly important in the pharmaceutical industry. Also for pharmaceuticals, polymorphism can be critical,55–58 affecting intellectual property protection, regulatory issues, and formulation control.58 The heightened focus on solid form control in the pharmaceutical industry has prompted the development and integration of effective computational tools that complement experimental polymorph screening. Thus, a methodology has been developed59 to extract appropriate H-bond information from the CSD and use it to predict the likelihoods of formation of each potential Hbond in a structure. Each likelihood is termed a hydrogen bond propensity (π ), and propensity data for expected or unexpected H-bonds has proved to be highly valuable in assessing the relative stability of potential or known crystal forms.60

The hydrogen bond propensity method59 treats each potential D–A interaction as a separate dichotomous probability. A potential H-bond propensity is modeled by a strict probability function, trained using knowledge of H-bonds in related crystal structures. The propensity is captured in parameters within the model function that are based on the respective environments of the D and A groups. Assessment of a target compound uses only its 2D molecular formula, and the outcomes are thus truly predictive. Highly likely/unlikely H-bonding interactions are quickly revealed, pointing to any issues of stability. Furthermore, propensities for the H-bonds observed in known polymorphs can be compared to give an indication of relative stability. The procedure involves four stages: data sampling, model fitting, model validation, and target assessment. So far, only the CSD has been used as the source of training data, but the approach permits the inclusion of additional 3D structures, for example, from local in-house databases. Appropriate structures for use in statistical modeling are obtained using chemical similarity searches relevant to the target as well as comparisons of atom-environment fingerprints61 for the D and A groups. The training sample is critical to the propensity model: structures retrieved from the CSD must be relevant to the chosen target and be sufficiently numerous to minimize statistical uncertainty. Two key pieces of information are derived from the CSD sample: (i) a binary true/false observation for every potential H-bond and (ii) a set of molecular/chemical descriptors providing explanatory data. Four kinds of attributes are used to express influences on H-bonding: functional group categories for both donor and acceptor, a steric density function for the D and A groups, a competition function, and an aromaticity function. All descriptors are computed using chemical connectivity information. Together these descriptors are the parameters that comprise the propensity model function,59 which takes the form of a logit function: a standard probability distribution used to describe binary data. The results are ranked by propensity (π) to enable rapid inspection of the most and least likely D–A permutations. Very low propensity H-bonds are often the most revealing: if present in a structure, they suggest that other more stable forms may well be possible. A structure that uses mainly high propensity H-bonds often indicates the converse.

4.4.2 Applications An early example of H-bond propensity analysis60 was provided by ritonavir (Figure 11a), the anti-HIV agent developed by Abbott Laboratories.39, 62, 63 The two marketed formulations were based on ethanol–water solutions and therefore crystal form control was not undertaken

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications Ureido O

Carbamate Thiazoyl A O N H

O

Thiazoyl B

H N

S N

11

N H

HO

N N

O S

Amide

Hydroxyl (a)

(a) NH2 Triazine

Cl

N 2 N 1

Amine a O

N4

O

NH2 Amine b

HO

Methoxycarbonyl

Phenol (b)

Cl

Figure 11 H-bonding functional groups in (a) ritonavir and (b) lamotrigine and methylparaben. (b)

during development. However, in 1998, the product had to be withdrawn, with huge financial implications, due to the appearance of an unforeseen polymorph (form II) with lower solubility and much reduced bioavailability. Propensity analysis60 showed that the possible H-bonds in ritonavir occupy a wide π -range: those involving the amide, carbamate, and hydroxyl groups all have higher propensities (π > 0.32) as expected, and H-bonds involving the thiazoyl acceptor have the lowest likelihoods (π < 0.23). The knowledge that form I exhibits a hydroxyl–thiazoyl interaction (π = 0.11) and a ureido–ureido interaction (π = 0.22) warns of stability issues, despite the remaining pair of Hbonds in form I having higher propensities. In contrast, the model predicts that all four H-bonds in form II (π > 0.32) are more likely than not to be formed. If the structure of the initial form I had been known during development, then the propensity evidence for potential alternative crystal form(s) could have been recognized, and the subsequent critical difficulties might have been mitigated or avoided.

Table 3

Hydrogen bond propensities can also be used to assess multicomponent crystals such as solvates and cocrystals, where the second component often introduces alternative donor/acceptor functionality, and several lamotrigine cocrystals have recently been studied.64 In pure lamotrigine (Figure 11b), two H-bond motifs occur (Figure 12a: EFEMUX01,65 the 500 000th structure archived to the CSD in 2009): an aminopyridine dimer and an amino-NH2 · · ·Ntriazine motif. H-bond predictions for pure lamotrigine, using a learning set of 721 related CSD structures, are presented in Table 3. All potential H-bonds have high propensity (π > 0.6): all donors and acceptors (except chloro) are strong, and represent permutations of the observed motifs. Pure lamotrigine was then compared

H-bond propensity predictions for parent lamotrigine. Functional groups are labeled according to Figure 11(b).

Donor Amine Amine Amine Amine Amine Amine Amine Amine Amine Amine

Figure 12 (a) Hydrogen bonds in lamotrigine (CSD entry EFEMUX0165 ). (b) Hydrogen bonds in the lamotrigine methylparaben cocrystal,64 which differ completely from those in the structure of parent lamotrigine.

Acceptor NH2 a NH2 b NH2 a NH2 b NH2 a NH2 b NH2 a NH2 b NH2 a NH2 b

Triazine Triazine Triazine Triazine Triazine Triazine Cl8 Cl8 Cl7 Cl7

N1 N1 N2 N2 N4 N4

Propensity (π)

Lower bound

Upper bound

0.928 0.909 0.905 0.881 0.702 0.647 0.419 0.360 0.353 0.299

0.913 0.890 0.884 0.855 0.630 0.569 0.377 0.317 0.311 0.258

0.940 0.926 0.923 0.904 0.764 0.719 0.461 0.405 0.398 0.343

Lower and upper bounds of propensity are calculated at the 95% confidence level based on a χ 2 distribution. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

Bond formed? √ × × √ √ × × × × ×

12

Supramolecular materials chemistry

Table 4

H-bond propensity predictions for the lamotrigine methylparaben cocrystal, presented as in Table 3.

Donor Amine Amine Amine Amine Amine OH Amine OH Amine Amine Amine Amine OH Amine Amine OH OH OH Amine Amine Amine Amine OH OH

Acceptor NH2 a NH2 b NH2 a NH2 b NH2 a NH2 b NH2 a NH2 a NH2 b NH2 b NH2 a NH2 b

NH2 a NH2 b NH2 a NH2 b

Triazine Triazine Triazine Triazine C=O Triazine C=O Triazine OCH3 OH OCH3 OH C=O Triazine Triazine OCH3 OH Triazine Cl15 Cl15 Cl16 Cl16 Cl15 Cl16

N1 N1 N2 N2 N1 N2

N4 N4

N4

Propensity (π)

Lower bound

Upper bound

0.904 0.886 0.883 0.862 0.793 0.787 0.761 0.748 0.693 0.684 0.651 0.641 0.602 0.563 0.515 0.470 0.459 0.335 0.316 0.277 0.274 0.238 0.154 0.129

0.881 0.858 0.853 0.826 0.725 0.734 0.684 0.685 0.553 0.601 0.505 0.551 0.504 0.463 0.414 0.324 0.365 0.245 0.273 0.234 0.232 0.197 0.123 0.101

0.923 0.909 0.908 0.892 0.848 0.832 0.823 0.802 0.804 0.757 0.773 0.723 0.692 0.657 0.615 0.620 0.556 0.440 0.362 0.323 0.321 0.284 0.190 0.163

Bond formed? √ × × √ √ × √ √ × × × × × × × × × × × × × × × ×

The bond formed column has the following: red = observed in pure form, green = observed in cocrystal.

with a potential methylparaben (methyl 4-hydroxybenzoate) cocrystal (Figure 12b) to see if the methyl ester and phenol H-bonding groups of the coformer are likely to disrupt the strong homomolecular interactions. A propensity model, generated (P. T. A. Galek, J. A. Chisholm, P. A. Wood, and E. Pidcock, Manuscript in preparation) using 1282 CSD training structures, achieved 82% correct predictions, a typical value for the method. In Table 4, the interactions in red are those observed in lamotrigine and those in green are potential cocrystal interactions. It can be seen that the NH2 · · ·N-triazine homointeractions remain highly likely, but the competitive hetero-interaction involving lamotrigine-NH2 and the C=O of methylparaben appears likely to displace the homoaminopyridine dimer. The methylparaben phenol donor is also highly competitive. Thus, although propensity margins are small (10–15%), cocrystallization does appear feasible. In practice, a 1 : 1 cocrystal of lamotrigine with methylparaben is obtained,64 and all interaction motifs of the parent are disrupted: the aminopyridine dimer is disrupted by a lamotrigine(NH2 )· · ·methylparaben(C=O) interaction, while the methylparaben phenol donates to the lamotrigine triazine-N (Figure 12b). The strong aromatic N acceptor remains unpartnered by a strong Hdonor, but forms a favorable but weak CH · · ·N bond.

Shape complementarity between the components can also be observed.

5

5.1

MATERIALS DESIGN AND DRUG DISCOVERY: TWO SIDES OF THE SAME COIN Supramolecular design as a docking problem

A compressed crystalline powder or tablet is the most commonly used pharmaceutical dosage form. The design and prediction of solid forms are therefore of great importance in the pharmaceutical industry as the specific form used for a drug defines many of its physical properties such as solubility, stability, shelf life, hygroscopicity, habit, and compressibility.66, 67 Multicomponent forms like cocrystals and salts are of particular interest as these can potentially provide significant physicochemical improvements in relation to pure forms of the API.68, 69 When selecting the form to be used for development, knowledge of a wide range of pure and multicomponent forms will provide the best opportunity to select an acceptable candidate. Current methods for finding new solid forms have limitations: there is a limit to the number of experimental crystallization screens that can

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications be performed, and CSP is computationally expensive and less reliable for multicomponent systems. Other techniques for identifying or predicting multicomponent API crystal structures are therefore highly attractive. A new method for designing or predicting multicomponent forms has recently been proposed70 which involves utilizing the known stable frameworks of an API. The approach is based on the observation that a multicomponent small-molecule crystal structure represents a type of host–guest-binding system related to that observed when an API interacts with, or docks into, a protein-binding site. There are well-established techniques for predicting the conformations and interactions of drug molecules within protein-binding sites, but relatively few studies have attempted to apply these methods to docking into an environment other than a protein. Previous studies have applied molecular docking methods to artificial receptors,71 metalbased frameworks,72 and organic crystal growth surfaces.73 Thus, the efficacy of applying the off-the-shelf molecular docking package, GOLD26 (Section 3.2.2), to the problem of predicting crystalline guest-framework systems has been investigated.70 In order to predict new multicomponent forms using GOLD, it is necessary to test a large number of guestframework pairs and rank them by some scoring function, thus identifying the most likely structures to be observed. The first step is to be able to accurately reproduce the position and orientation of guests in experimentally observed guest-framework crystal structures. To evaluate the success of GOLD in this problem domain, a set of 14 channelbased, guest-framework crystal structures containing small organic guests was retrieved from the CSD. The tests

Table 5 System cbz1 cbz2 cbz3 cbz4 chol1 chol2 dian1 diol1 imin1 mof1 phen1 urea1 urea2 urea3

Pose prediction results.70 ˚ RMSD (A) 0.3 2.1 0.3 0.5 1.6 1.6 0.3 0.5 1.6/2.0 1.2 0.8 1.1 1.2 0.4

For each framework–guest system, the non-H atom RMSD of the best solution with respect to the crystal structure is reported, bold entries indicate near-optimal predictions.

13

used fixed framework structures into which computationally generated 3D conformations of the guest molecules were then docked. The program CORINA74 was used to generate 3D conformers to avoid bias toward the experimentally observed X-ray structures. Further, the scoring function used for this investigation was a standard one (CHEMPLP75 ) trained to reproduce protein–ligand interactions within a binding site; the only modification was to remove very specific C–H · · ·O scoring terms designed for kinase structures. The docking results70 (Table 5) show that half of the systems exhibit near-optimal predictions with very good RMSD values for the overlay of experimental and predicted guest positions. The structures that show near-optimal predictions (Figure 13a) correspond to cases that involve relatively strong intermolecular interactions or distinct shape-fitting requirements. In guest-framework systems involving no strong interactions and only weak shape restrictions (Figure 13b), the docking software has very little information and can only obtain a rough prediction of the guest position and conformation. These tests only docked a single guest molecule into each of the frameworks, so in systems where there are multiple guests (Figure 14a), or symmetry-generated guest–guest interactions, the predictions generally have sub-optimal matches to the experimental structures. To correctly model this type of system, it will be necessary to dock multiple guests concurrently and deal with both host–guest and guest–guest interactions. Also, in cases where the guest molecule is actually disordered in the crystal structure (Figure 14b), the docking result is relatively poor, again because there is little information relating to either interactions or shape restrictions. Nevertheless, the results of this preliminary investigation clearly show that it is possible to successfully apply molecular docking principles to the prediction of framework-guest systems when there are either strong interactions or shape-fitting restrictions in the

(a)

(b)

Figure 13 Structural overlays of the predicted (magenta) and observed (cyan) guest positions in the carbamazepine:furfural (cbz1, a) and urea:1-bromo-6-chlorohexane (urea1, b) channelbased systems.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

14

(a)

Supramolecular materials chemistry

(b)

Figure 14 Structural overlays of the predicted (magenta) and observed (cyan/green) guest positions in (a) an ethanol-containing metal-organic framework and (b) a 4-pyridyl iminophenol:acetone channel-based system.

framework. The potential also exists to improve the docking performance by allowing the docking of multiple guests into the same framework, thus dealing with guest–guest interactions explicitly. The next step in making this method more directly applicable to the prediction of multicomponent systems, including those of APIs, is to re-parameterize a scoring function to reproduce the interactions and conformations in small-molecule crystal structures. This process is currently being undertaken using structures extracted from the CSD with a variety of intermolecular interactions present. The effectiveness of the method is also analyzed in a blind test situation where a given framework is known to bind multiple compounds from a dataset of molecules, but the specific compounds are not known before running the docking calculations.

5.2

Drug discovery using the CSD and materials design software

The design requirements for supramolecular architectures and drug compounds are closely similar, and there are a number of ways in which the CSD can benefit both communities. First, the CSD is a rich source of 3D chemistry, and any CSD-derived fragments arising from new design ideas have the crucial advantage of having achievable geometries. Also, these geometries are usually associated with low strain energies, and this can be quickly checked using the Mogul knowledge base.18 Secondly, the CSD is the ultimate resource for validating molecular geometries and intermolecular interactions that might arise from computational modeling studies. Such studies include molecular docking, in which many ligands are computationally evaluated for their fit to a protein active site, or to a crystal cavity as in Section 5.1. The time allowed per ligand is necessarily short and not all docking poses are good, so that

postprocessing validation to select viable poses is highly valuable. These similar requirements of drug design and materials research suggest that computational informatics tools designed for one group will prove useful to the other. Indeed, some of the Materials module of Mercury developments, described in Section 4, are of significant value in drug design, although they were initiated for use in drug development and formulation. One drug design application that requires 3D fragments of accessible geometry is “re-scaffolding,” where the central portion of a putative drug molecule is replaced by a new chemical scaffold, which maintains binding efficiency by preserving the existing orientations of the binding groups. These new scaffolds may impart improved physicochemical properties or avoid existing patent art. Figure 15 illustrates a re-scaffolding exercise using an inhibitor of the serine protease factor Xa, a protein important in modulating the blood coagulation response and hence an important drug target. The molecule to be re-scaffolding is loaded into the Materials module of Mercury, the atoms to be preserved as link atoms from the retained binding fragments are selected, and the CSD is searched to find other scaffolds that bridge between these link atoms. A possible new scaffold is illustrated in Figure 15(b). A second application is “fragment linking”: fragments are identified, perhaps via high-throughput X-ray crystallography or nuclear magnetic resonance (NMR) experiments, which bind to different pockets in a protein. These fragments and their associated binding modes are then used as a starting point for structure-based drug design: searching for a linker with low strain energy that can be placed between the fragments so that fragment binding is not compromised. The preferential use of CSD-derived libraries for both re-scaffolding and fragment-based design has been highlighted,76 and it is possible to obtain a CSD-derived fragment library for use with the LeadIT platform.77 However, these tasks can also be carried out using the packing feature search tool (Section 4.2) in the Materials module

O N

N O

O

N N

O

(a)

(b)

N

O

Figure 15 Re-scaffolding of a Factor Xa inhibitor: (a) selecting the atoms to be re-scaffolded in the Materials module of Mercury, and (b) a possible new scaffold located by the CSD search.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications of Mercury. This tool can also be used in a third way for “contact motif searching” in which the CSD is searched for a protein–ligand interaction motif, which is known to be beneficial for good activity. This may generate ideas for different fragments that can adopt the same binding motif. Apart from its involvement in design tasks, the packing feature search tool can also be used for the validation of modeling results. In all but the highest level computational studies certain types of intermolecular interactions have to be treated rather crudely. Thus, the geometries of intermolecular hydrogen bonds are well parameterized and are usually modeled with reasonable accuracy, but some common and more diffuse interactions are less well treated, for example, π –π stacking, which can adopt various configurations depending on the interaction partners. This interaction is common in both small molecules and in protein–ligand complexes but, due to its dependence on quadrupole effects, cpu-intensive ab initio methods are required for reliable calculations.78, 79 Alternatively, the Materials module of Mercury can be used to validate stacking interactions by selecting the relevant atoms and locating CSD matches using distance and angle tolerances. The RMSDs, calculated over all matched atoms between the query and each hit, provide goodness-of-fit data, which can be displayed as a histogram and compared with different modeling outputs. Figure 16 shows two binding modes obtained when docking a ligand to serine protease factor Xa. The best-scored GOLD pose (score = 89.9) is not the ligand pose observed in PDB entry 1f0r,80 instead it is the second best GOLD pose (score = 84.6) that matches the crystal structure. The packing feature search can be applied to each of the two stacking

15

interactions of Figure 16(a): pose 1 (A) and pose 2 (B) of the ligand’s terminal aryl group interacting with the indole of an active site tryptophan. The histogram (Figure 16b) shows the RMSD distribution of close matches in the CSD for both stacking interactions, and also for the interaction found in the crystal structure 1f0r. Clearly, the stacking interaction in pose 1 is poorly represented by good CSD matches, whereas pose 2 is well represented and agrees closely with the interaction observed in 1f0r.

6

COCRYSTALS: DESCRIPTORS OF FORMATION AND THEIR PREDICTIVE VALUE

The most successful approach to the design of cocrystals is based on the identification of reliable supramolecular heterosynthons that will form in preference to any possible homosynthon.81 The CSD is an essential tool for finding these synthons and assessing synthon competition.58, 82, 83 Nevertheless, molecules that could form identical supramolecular synthons often differ in their tendency toward cocrystal formation. The low success rate of synthon-based screening experiments (about 20%) indicates that factors other than synthon complementarity should be considered. Known cocrystals contain information about these additional influences, and statistical analysis of cocrystal structures in the CSD has been used to identify them. A database of 974 cocrystal structures was extracted from the CSD, and the molecules that form the cocrystals were

B 80 Crystal Goldscore 89.9 Goldscore 84.6

70 60

Trp216 -34

Frequency

A

50 40 30 20 10 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

(a)

(b)

RMSD

Figure 16 Using the Materials module of Mercury packing feature search to identify the correct docking pose in the binding cavity of serine protease factor Xa. (a) The stacking interaction between the ligand and the protein for docking poses A and B (see text). (b) The histogram displays the low RMSD matches of the two docking poses and compares them with the pose in the protein–ligand crystal structure. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

16

Supramolecular materials chemistry

characterized by a set of molecular descriptors.84 The 130 descriptors included simple atom and bond counts, topological and shape descriptors, and descriptors derived from the charge distribution and surface area of the molecules. The resulting database describes pairs of molecules that form cocrystals with each other by their calculated molecular properties. The values of influential descriptors tend to be either similar or complementary for pairs of cocrystallizing molecules, while the values of descriptors that have no influence on cocrystal formation are statistically independent for such pairs. Consequently, molecular properties that have an effect on cocrystal formation could be identified by calculating correlation coefficients between pairs of descriptors. Correlation coefficients were calculated for all possible descriptor pairs and 2D density plots of correlated descriptor pairs were used to visualize the observed trends and elucidate their statistical significance. The strongest descriptor correlations (r = 0.3–0.4) relate to the shape and polarity of cocrystal formers. Molecular shapes were described using the box model of Pidcock and Motherwell.85 In this model, a rectangular box is drawn around the molecule so that its edges are parallel to the molecule’s principal axes of inertia. L, M, and S denote the long, medium, and short axes of this box respectively, and the ratios S/L and M/L describe the overall shape of the molecule: a small S/L ratio corresponds to flat molecules, a small M/L ratio to rod-shaped molecules. The density plots (Figure 17) clearly show that the values of these two ratios tend to be similar for the molecules in a cocrystal. Most of the data points are close to the diagonal, and the frequency of observations decreases gradually with increasing distance from the diagonal. The polarity descriptors that showed the strongest correlations were molecular dipole moment and the fractional volume of polar atoms relative to the total molecular volume (FPV), which can be approximated by the ratio FNO = (No. of N atoms + No. of O atoms)/No. of

non-H atoms. As with the shape descriptors, these polarity descriptors show a marked preference for similar values in cocrystallizing molecules and show a gradually decreasing frequency of known cocrystals with increasing differences between the molecules.84 Having identified the most influential molecular descriptors, a strategy was developed to use them predictively. The importance of supramolecular synthons is well established, so rather than aiming for complete prediction of cocrystallization results from molecular descriptors, the goal was to increase the success rate of screening experiments by adding a second step to their design after considering supramolecular synthons. While the CSD archives had known about cocrystals, it has no record of failed experiments, and this lack of negative data prevented the use of standard classification methods, such as fitting of scoring functions or training Bayesian classifiers. Instead, a simple cutoff scheme was employed. Cocrystal formation between two molecules was considered likely if the differences between their S/L, M/L, FNO, and dipole values were all less than a specific cutoff. The cutoff values were defined as the ninth decile of the difference between the molecular descriptors of cocrystal formers in our data set (0.275 for S/L, 0.310 for M/L, 0.294 for FNO, and 5.94 Debye for dipole moments). This means that the difference between the cocrystal formers is less than the cutoff value in 90% of cocrystals in the CSD. This scheme was tested using the results of screening experiments on 218 compound pairs (W. Jones, T. Friˇscˇ i´c, and S. Karki, Private communication). As these results were largely unpublished, they could be used as an independent test set. The experiments were designed with synthon considerations in mind, so that the improvement provided by the subsequent use of molecular descriptors could be probed (L. Fabian, Manuscript in preparation). The success rate among the 218 experiments was 22% (Table 6), that is,

1.0

1.0

3 4

0.8

M/L (mol. 2)

S/L (mol. 2)

0.8 0.6 6

8

0.4

1 2 4

0.2

2

1

0.4 0.2

0.0

0.0 0.0

(a)

5 3

0.6

0.2

0.4 0.6 S/L (mol. 1)

0.8

1.0

0.0 (b)

0.2

0.4 0.6 M/L (mol. 1)

0.8

1.0

Figure 17 Contour plots showing the relative frequency of (a) S/L and (b) M/L value pairs in the cocrystal data set. A uniform random distribution would correspond to a constant contour value of 1 over the whole plot area. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications

17

Table 6 Confusion matrix showing the performance of descriptor-based filtering in cocrystal screening experiments. Cocrystal formation

Prediction

Experiment

Likely to form Unlikely to form Experiment total

Cocrystal formed

No cocrystal

40 8 48

67 103 170

48 of the 218 compound pairs formed a cocrystal. About half of the 218 experiments were within the cutoff limits and were labeled as “likely” to give a cocrystal. The success rate among these “likely” experiments is 37% (40/107), a > 50% improvement over the original value. Only 17% (8/48) of cocrystal-forming compound pairs were predicted to be unlikely cocrystal formers. The remaining 83% could be retrieved with less than 50% of the investment if only the “likely” experiments were performed. These results clearly demonstrate the utility of molecular descriptors and CSD-based cutoff values in the design of cocrystal screening experiments. A simple filtering method gave a significant reduction in the number of experiments and an increase in the success rate with only a small decrease in coverage. More complex methods, such as ranking of coformers, instead of a binary likely/unlikely classification, or the use of more elaborate descriptors, may further improve the results. Despite the lack of any such optimization, the original goal to develop a design strategy that complements supramolecular synthons has been successfully achieved.

7

Prediction total

APPLICATIONS OF THE CSD IN CRYSTAL STRUCTURE PREDICTION (CSP)

CSP using only prior knowledge of the chemical connectivity of the molecular component(s) is a complex process, but advances over the last 20 years have been significant, as demonstrated by the periodic CSP blind tests organized by the CCDC.86–89 In the early 1990s, predicting crystal structures for small rigid molecules was a major challenge, but today such structures can often be predicted in a few cpu hours. The challenges, however, continue at a different level, namely: (i) the treatment of molecular flexibility,90 (ii) predicting crystal structures with more than one component in the asymmetric unit (e.g., salts,91 cocrystals,88 solvates,92 and hydrates,93 ) (iii) improving the accuracy of the energy models,94, 95 and (iv) predicting likely polymorphs of even the simplest molecules.96 A typical CSP procedure

107 111 218

can be divided into three main stages (Figure 18): (i) the generation of accurate molecular models, (ii) the generation of all stable crystal structures for each model, and (iii) an accurate ranking of crystal structures, normally based on energy. Crystal structure knowledge encapsulated in the CSD has always been fundamental to the development of CSP methods, and may increase in importance as more challenging systems are attempted. Reasonable models of small rigid molecules can readily be generated with simple modeling software. Ab initio methods are normally used to generate accurate molecular energies, geometries, and other properties such as atomic charges. However, the main problem in model building arises with flexible molecules. Different conformations give rise to different crystal packings and it is not possible to know a priori which conformation(s) will be observed in experimental crystal structures. One could generate all possible stable conformations and sample all possible packing possibilities, but this becomes extremely costly as the number of flexible torsions increases. In some cases, the lowerenergy conformers may not be observed in crystal structures (e.g., amino acids)97 and an initial CSD-guided conformation selection is fundamental for the success of the CSP calculations. CSP studies of molecules with >3 flexible torsions are still rare, and, as more challenging predictions are attempted, the torsional information and Mogul statistics18 (Section 3.1.5, Mogul) become increasingly relevant and important in CSP. The generation of all possible crystal packing arrangements for a given 3D molecular model is a purely mathematical problem. Different CSP programs use different algorithms (e.g., stochastic searches and genetic algorithms) to sample all the structural parameters (e.g., molecular position and orientation in the asymmetric unit, unit cell lengths, and cell angles) and minimize a global function, such as the total energy of the resulting crystal structures. Most programs use simple atom–atom potential functions to evaluate intermolecular interactions. In most cases, generic force fields for the most common atom types in organic crystal structures are derived empirically from crystal structure data in the CSD, but some newer methods derive tailor-made force fields for each molecule from high-level

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

18

Supramolecular materials chemistry

H H

H H H H

H H H

H

N H N

O

H

Molecular sketch

3D molecular model

Generated crystal structures

Observed crystal structure

Space group statistics Torsion angles statistics (MOGUL)

Crystal structure data for interatomic potentials

CSD Observations to help structure ranking

Synthons

Figure 18 CSD information that can contribute to the three main stages in crystal structure prediction: (a) the generation of 3D molecular models, (b) the generation of crystal structure packings, and (c) the final ranking of crystal structures.

ab initio calculations.98 In the search algorithm, space group information from the CSD is normally used to reduce the number of searchable structural parameters. In multicomponent systems, reliable synthon information from the CSD can also be used to generate fixed supermolecules for which possible crystal packing arrangements can be postulated. The supermolecule approach can reduce the computing time considerably, because for each extra independent molecule the potential energy surfaces of possible crystal structures will depend on six extra parameters, making the search stage even more challenging. While CSD observations can help differentiate crystal structures with likely hydrogen bond motifs from those which are less likely, the overall influence of the CSD in the ranking stage is less clear. Often, very similar crystal structures with the same hydrogen-bonding motifs are predicted with minimal energy differences, and the use of very accurate models for energy calculations is often the only way of improving the ranking. However, the H-bond propensity studies discussed in Section 4.4 show that the CSD may have a future role in the prediction of metastable polymorphs.

8

CONCLUSION

Over nearly half a century, the crystallographic databases have moved from being rather peripheral objects of interest to research tools of major significance to an increasingly broad community of scientists. They have played a major role in the development of the new subdisciplines of chemoinformatics, bioinformatics, and crystal engineering, and they are increasingly used as the experimental proving ground for ever more ambitious computational procedures that improve our capabilities in drug discovery, materials design, and in drug development and formulation. In this chapter, we have shown how the CSD has played a major role in these developments, a role which continues to expand as new generations of scientists discover novel applications for the huge accumulation of precise 3D crystal structure information, which continues to be generated worldwide. While this chapter has summarized the current state of the art, it has also highlighted new challenges, for example, in searching the CSD in more advanced ways, in advancing CSP for more flexible molecules and multicomponent systems, and in improving our understanding of intermolecular interactions

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

The CSD system and its applications that are not mediated by hydrogen. We would envisage that a chapter similar to this one, but written a decade hence, would provide at least some of the answers to these challenges.

19

26. G. Jones, P. Willett, and R. C. Glen, J. Mol. Biol., 1995, 245, 43–53. 27. J. W. M. Nissink, C. Murray, M. Hartshorn, et al., Proteins, 2002, 49, 457–471. 28. W. I. F. David, K. Shankland, J. van de Streek, et al., J. Appl. Cryst., 2006, 39, 910–915.

REFERENCES

29. E. Pidcock, J. van de Streek, and M. U. Schmidt. Z. Krist., 2007, 222, 713–717.

1. J. M. Lehn, Angew. Chem. Int. Ed. Engl., 1990, 29, 1304–1319.

30. W. D. S. Motherwell, G. P. Shields, and F. H. Allen, Acta Cryst., 2000, B56, 466–473.

2. J. D. Dunitz, Pure Appl. Chem., 1991, 63, 177–185.

31. J. W. Yao, J. C. Cole, E. Pidcock, et al., Acta Cryst., 2002, B58, 640–646.

3. P. S. White, J. R. Rodgers, and Y. Le Page, Acta Cryst., 2002, B58, 343–348. 4. A. Belsky, M. Hellenbrandt, V. L. Karen, and P. Luksch, Acta Cryst., 2002, B58, 364–369. 5. H. M Berman, W. K. Olson, D. L. Beveridge, et al., Biophys. J., 1992, 63, 751–759. 6. H. M. Berman, J. Westbrook, Z. Feng, et al., Nucleic Acids Res., 2000, 28, 235–242.

32. G. M. Battle, F. H. Allen, and G. M. Ferrence, J. Chem. Educ., 2010, 87, 809–812 33. G. M. Battle, F. H. Allen, and G. M. Ferrence, J. Chem. Educ., 2010, 87, 813–818 34. G. M. Battle, G. M. Ferrence, and F. H. Allen, J. Appl. Cryst., 2010, 43, 1208–1223.

7. F. H. Allen, Acta Cryst., 2002, B58, 380–388.

35. J. Chisholm, E. Pidcock, J. van de Streek, et al., CrystEngComm, 2006, 8, 11–28.

8. G. M. J. Schmidt, Pure Appl. Chem., 1971, 27, 647–678.

36. M. C. Etter. Acc. Chem. Res., 1990, 23, 120–126.

9. G.R. Desiraju, Crystal Engineering—The Design of Organic Solids. Elsevier, Amsterdam, 1989.

37. M. C. Etter, J. C. Macdonald, and J. Bernstein. Acta Cryst., 1990, B46, 256–262.

10. G. R. Desiraju, Angew. Chem. Int. Ed. Engl., 1995, 34, 2311–2327.

38. L. F´abi´an, J. A. Chisholm, P. T. A. Galek, et al., Acta Cryst., 2008, B64, 504–514.

11. F. H. Allen, Cryst. Rev., 2004, 10, 3–15.

39. J. Bauer, S. Spanton, R. Henry, et al., Pharm. Res., 2001, 18, 859–866.

12. C. Bissantz, B. Kuhn, and M. Stahl, J. Med. Chem., 2010, 53, 5061–5084 13. B. Kuhn, P. Mohr, and M. Stahl, J. Med. Chem., 2010, 53, 2601–2611. 14. S. R. Hall, F. H. Allen, and I. D. Brown, Acta Cryst., 1991, A47, 655–685. 15. I. J. Bruno, J. C. Cole, P. R. Edgington, et al., Acta Cryst., 2002, B58, 389–397. 16. C. F. Macrae, P. R. Edgington, P. McCabe, et al., J. Appl. Cryst., 2006, 39, 453–457. 17. C. F. Macrae, I. J. Bruno, J. A. Chisholm, et al., J. Appl. Cryst., 2008, 41, 466–470. 18. I. J. Bruno, J. C. Cole, M. Kessler, et al., J. Chem. Inf. Comput. Sci., 2004, 44, 2133–2144. 19. I. J. Bruno, J. C. Cole, J. P. M. Lommerse, et al., J. Comp. -Aided Mol. Design, 1997, 11, 525–537. 20. I. R. Thomas, I. J. Bruno, J. C. Cole, et al., J. Appl. Cryst., 2010, 43, 362–366. 21. JMol viewer (R.Hanson), http://chemapps.stolaf.edu/jmol/ docs/examples-11/jcse/explore.htm 22. OpenAstexViewer, http://openastexviewer.net/ 23. CRYSTALS program package for crystal structure determination: http://www.xtl.ox.ac.uk 24. A. Brameld, B. Kuhn, D. C. Reuter, and M. Stahl, J. Chem. Inf. Model., 2008, 48, 1–24. 25. M. L. Verdonk, J. C. Cole, and R. Taylor, J. Mol. Biol., 1999, 289, 1093–1108.

40. D. Vega, A. Petragalli, D. Fernandez, and J. A. Ellena, J. Pharm. Sci., 2006, 95, 1075–1082. 41. D. A. Haynes, J. A. Chisholm, W. Jones, and W. D. S. Motherwell. CrystEngComm, 2004, 6, 584–588. 42. J. A. Chisholm and W. D. S. Motherwell, J. Appl. Cryst., 2005, 38, 228–231. 43. J. A. Chisholm and S. Motherwell, J. Appl. Cryst., 2004, 37, 331–334. 44. A. Y. Lebedev, V. V. Izmer, D. N. Kazyul’kin, et al., Org. Lett., 2002, 4, 623–625. 45. R. Banjeree, P. M. Bhatt, N. V. Ravindra, and Desiraju, Cryst. Growth Des., 2005, 5, 2299–2305.

G. R.

46. F. T. Martins, N. Paparidis, A. C. Doriguetto, and J. Ellena, Cryst. Growth Des., 2009, 9, 5283–5292. 47. T. Gelbrich and M. B. Hursthouse, CrystEngComm, 2006, 8, 448–460. 48. S. L. Childs, P. A. Wood, N. Rodriguez-Hornedo, et al., Cryst. Growth Des., 2009, 9, 1869–1888. 49. G. M. Day, J. Chisholm, N. Shan, et al., Cryst. Growth Des., 2004, 4, 1327–1340. 50. T. Gelbrich and M. B. Hursthouse, CrystEngComm, 2005, 7, 324–336. 51. A. V. Dzyabchenko, Acta Cryst., 1994, B50, 414–425. 52. C. B. Aaker¨oy, Acta. Cryst., 1997, B53, 569–586. 53. G. R. Desiraju, Acc. Chem. Res., 2002, 35, 565–573. 54. M. D. Hollingsworth, Science, 2002, 295, 2410–2413.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

20

Supramolecular materials chemistry

55. J. Haleblian and W. McCrone, J. Pharm. Sci., 1969, 58, 911–929.

76. P. Maass, T. Schulz-Gasch, M. Stahl, and M. J. Rarey, J. Chem. Inf. Model., 2007, 47, 390–399.

56. D. J. W. Grant, The theory and origin of polymorphism, in Polymorphism in Pharmaceutical Solids, ed. H. G. Brittain, Marcel Dekker, Inc., New York, 1999, pp. 1–31.

77. http://www.biosolveit.de/press/archive/2010-03-03 CCDC. html, http://www.ccdc.cam.ac.uk/pp/csd recore/2010/, 2010.

57. C. R. Gardner, C. T. Walsh, and O. Almarsson, Nature Rev. Drug Disc., 2004, 3, 926–934.

78. C. A. Hunter and J. K. M. Sanders, J. Am. Chem. Soc., 1990, 112, 5525–5534. 79. S. L. Price and A. J. Stone, J. Chem. Phys., 1997, 86, 2859.

58. J. Bernstein, Polymorphism in Molecular Crystals, Oxford University Press, Oxford, 2002.

80. S. Maignan, J.-P. Guilloteau, S. Pouzieux, et al., J. Med. Chem., 2000, 43, 3226–3232.

59. P. T. A. Galek, L. F´abi´an, F. H. Allen, and N. Feeder, CrystEngComm., 2009, 11, 2634–2639. 60. P. T. A. Galek, L. F´abi´an, F. H. Allen, et al., Acta Cryst., 2007, B63, 768–782.

81. C. B. Aaker¨oy, J. Desper, B. Leonard, and J. F. Urbina, Cryst. Growth Des., 2005, 5, 865–873. ¨ Almarsson and M. J. Zaworotko, Chem. Commun., 2004, 82. O. 1889–1896.

61. A. Bender, H. Y. Mussa, and R. C. Glen, J. Chem. Inf. Comput. Sci., 2004, 44, 1708–1718.

83. M. Rafilovich, J. Bernstein, M. B. Hickey, and M. Taubner, Cryst. Growth Des., 2007, 7, 1777–1782.

62. D. J. Kempf, K. C. Marsh, J. F. Denissen, et al., Proc. Natl. Acad. Sci. U.S.A., 1995, 92, 2484.

84. L. F´abi´an, Cryst. Growth Des., 2009, 9, 1436–1443.

63. S. R. Chemburkar, J. Bauer, K. Deming, et al., Org. Proc. Res. Dev., 2000, 4, 413–417. 64. M. L. Cheney, N. Shan, E. R. Healey, et al., Cryst. Growth. Des., 2010, 10, 394–405. 65. B. Sridhar and K. Ravikumar, Acta Cryst., 2009, C65, o460–o464. 66. L.-F. Huang and W.-Q. Tong, Adv. Drug Deliv. Rev., 2004, 56, 321–334. 67. D. Singhal and W. Curatolo, Adv. Drug Deliv. Rev., 2004, 56, 335–347. 68. M. B. Hickey, M. L. Peterson, L. A. Scoppettuolo, et al., Eur. J. Pharm. Biopharm., 2007, 67, 112–119. 69. M. K. Stanton and A. Bak, Cryst. Growth Des., 2008, 8, 3856–3862. 70. O. Korb and P. A. Wood, Chem. Commun., 2010, 46, 3318–3320. 71. A. Steffen, C. Thiele, S. Tietze, et al., Chem. Eur. J., 2007, 13, 6801–6809. 72. E. J. Cussen, J. B. Claridge, M. J. Rosseinsky and C. J. Kepert, J. Am. Chem. Soc., 2002, 124, 9574–9581. 73. S. Imajo, M. Ishiguro, K. Ishiguro, et al., Bioorg. Med. Chem., 1994, 2, 1021–1027. 74. J. Sadowski, J. Gasteiger, and G. Klebe, J. Chem. Inf. Comput. Sci., 1994, 34, 1000–1008. 75. O. Korb, T. St¨utzle, and T. E. Exner, J. Chem. Inf. Model., 2009, 49, 84–96.

85. E. Pidcock and W. D. S. Motherwell, Cryst. Growth Des., 2004, 4, 611–620. 86. J. P. M. Lommerse, W. D. S. Motherwell, H. L. Ammon, et al., Acta Cryst., 2000, B56, 697–714. 87. W. D. S. Motherwell, H. L. Ammon, J. D. Dunitz, et al., Acta Cryst., 2002, B58, 647–661. 88. G. M. Day, W. D. S. Motherwell, H. L. Ammon, et al., Acta Cryst., 2005, B61, 511–527. 89. G. Day, T. Cooper, A. Cruz-Cabeza, et al., Acta Cryst., 2009, B65, 107–125. 90. T. G. Cooper, K. E. Hejczyk, W. Jones, and G. M. Day, J. Chem. Theor.Comp., 2008, 4, 1795–1805. 91. P. G. Karamertzanis and S. L. Price, J. Phys. Chem. B, 2005, 109, 17134–17150. 92. A. Cabeza, G. Day, W. Motherwell, and W. Jones, J. Am. Chem. Soc., 2006, 128, 14466–14467. 93. A. T. Hulme and S. L. Price, J. Chem. Theor.Comp., 2007, 3, 1597–1608. 94. P. G. Karamertzanis, G. M. Day, G. W. A. Welch, et al., J. Chem. Phys., 2008, 128, 244708–244717. 95. T. G. Cooper, K. E. Hejczyk, W. Jones, and G. M. Day, J. Chem. Theor.Comp., 2008, 4, 1795–1805. 96. S. L. Price, Phys. Chem. Chem. Phys., 2008, 10, 1996–2009. 97. T. G. Cooper, W. Jones, W. D. S. Motherwell, and G. M. Day, CrystEngComm, 2007, 9, 595–602. 98. M. A. Neumann, J. Phys. Chem. B, 2008, 112, 9810–9829.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc112

Cocrystals: Synthesis, Structure, and Applications Christer B. Aaker¨oy and Prashant D. Chopade Kansas State University, Manhattan, KS, USA

1 2 3 4 5

Why Cocrystals? 1 Design and Synthesis of Cocrystals 2 Halogen-Bond-Based Cocrystals 8 Examples of Halogen-Bonded Cocrystals 8 Cocrystallization Using Harmony between Halogen Bonds and Hydrogen Bonds 11 6 Applications of Cocrystals 12 7 Conclusion 15 Notes 15 References 15

1

WHY COCRYSTALS?

What is most likely going to happen when a solution containing two different molecular solutes (that are solids at ambient conditions) is allowed to evaporate to dryness? Unless a chemical reaction involving covalent bond breaking/formation takes place between the two solutes, one would expect the appearance of two separate homogeneous molecular solids. This inherent structural selfishness of molecules1 is relied upon whenever recrystallization is employed as a method of purification. Recrystallizations are essential to covalent synthesis, but in the supramolecular context, the same event provides an opportunity to move in a completely different direction—cocrystallization is an attempt at combining different molecular species within a single crystalline lattice without making or breaking covalent bonds (Scheme 1).

So what is the main reason that the synthesis of cocrystals has garnered such widespread scientific interest? The simple answer is that if we were able to position molecules exactly where we want them to be, and if we can synthesize heteromolecular architectures with specific metrics and stoichiometries, we may find a direct route from molecular structure, via solid-state structure, to tunable physical properties. It has already been shown2–6 that it is possible to develop practical strategies for supramolecular synthesis that are based on, and driven by, readily available information about molecular structure. A compilation of successful cocrystallization strategies would offer an invaluable toolbox for incorporating active ingredients (functional molecules) within a crystalline solid that subsequently can provide improved physical properties such as aqueous solubility, mechanical or thermal stability, or reduced hygroscopicity.

1.1

Nomenclature

Crystal engineering is a scientific area in constant flux, which helps explain why unambiguous definitions have not yet been developed and/or accepted. The term cocrystal is not well defined, and the existing literature contains terms such as molecular complexes, multicomponent solids, cocrystals, molecular adducts, molecular salts, clathrates, and inclusion compounds that frequently describe one and the same family or type of chemical compounds. We will not attempt to add to the ongoing discussion,7, 8 and we limit our overview to structurally homogeneous crystalline materials containing two or more neutral building blocks that are present in definite stoichiometric amounts and that are made from reactants that are solids at ambient conditions [1] (therefore hydrates and other solvates are excluded from this overview). In addition, we do not discuss

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

2

Supramolecular materials chemistry

Cocrystallization 0.62%

Recrystallization 99.38%

Scheme 1 Recrystallization (right) yields homomeric solids. The cocrystallization process (left) leads to a heteromeric product. Percentages are based on the frequency of occurrence of the relevant keywords in SciFinder Scholar ( ACS).

conventional “charge-transfer” complexes,10 instead focus on hydrogen- and halogen-bonded cocrystals.11–28

strategies for supramolecular synthesis often rely on the observed correlations between hydrogen-bond patterns and functional groups.

2

2.1.2 From pattern recognition to practical crystal engineering

2.1

DESIGN AND SYNTHESIS OF COCRYSTALS Background

2.1.1 Developing hydrogen-bond design strategies for the synthesis of cocrystals The hydrogen bond is arguably the intermolecular interaction that offers most opportunities for chemical or geometric fine-tuning. Numerous studies have demonstrated how hydrogen bonds can be employed as a supramolecular synthetic tool as well as for providing precise orientation and organization of molecules within a crystalline solid. Although most molecular solids composed of neutral molecules are single-component architectures, recent efforts have shown how both binary and ternary cocrystals7, 8, 29 can be synthesized with the help of complementary intermolecular interactions located on different molecular fragments.20 Obviously, these hydrogen-bond-based recognition mechanisms are extraordinarily selective (which is well known from biological systems) and, as a result, it is possible to extract important information about the balance between potentially competing intermolecular forces through systematic structural studies.30–33 Some of the pioneering work in this area was carried out by Etter and coworkers34–39 and they deliberately employed cocrystallizations, followed by spectroscopic and diffraction-based methods, for studying packing patterns, hydrogen-bond motifs, and intermolecular forces. In this way, it may be possible to develop a more refined understanding of how different intermolecular interactions can be combined into specific design strategies for the construction of binaryand higher order cocrystals. In principle, guidelines and

The fact that some intermolecular patterns appear more often than others clearly indicates that some motifs are likely to be stronger and more important than others. The difficulty in recognizing and classifying observed structural motifs created by multiple intermolecular interactions can be alleviated through the use of a graph-set notation (see Network and Graph Set Analysis, Supramolecular Materials Chemistry).40 Within this system, each network can be broken down into four principal motifs, chains (C), dimers (D), rings (R), and intramolecular hydrogen bonds (S), and additional notion offers information on the number of donor and acceptor atoms that are present within each motif. Although this classification does not provide direct chemical information about the participating functional groups, it does offer a self-consistent method that facilitates a systematic description of a wide range of structures.41 The simplicity of the nomenclature has arguably helped it become so recognizable and popular. For example, the R2 2 (8) motif, an eight-membered ring with two hydrogenbond donors and two acceptors as found, for example, in the carboxylic head-to-head dimer (Scheme 2), has entered common crystal engineering parlance. Another concept of considerable importance to crystal engineering in general and to cocrystal synthesis O

H O R

R O H

O

Scheme 2 Example of an R2 2 (8) motif, typically found in dimers of carboxylic acids.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystals: synthesis, structure, and applications in particular was introduced by Desiraju42 in 1995. “Supramolecular synthons” describe recognition events that take place between functional groups when discrete molecules assemble into supermolecules and illustrate nicely the conceptual similarities between retrosynthetic organic synthesis and supramolecular assembly.43 The task of identifying supramolecular synthons is critical to our ability to develop reliable synthetic strategies for the assembly of solid-state architectures with specific topologies, stoichiometries, and metrics.

2.2

The importance of electrostatics in hydrogen-bond-driven synthesis

2.2.1 Balancing supramolecular reagents for reliable formation of cocrystals Isonicotinamide has been identified as a reliable supramolecular reagent (SR) that, in combination with carboxylic acids,20, 44 is capable of producing cocrystals in a high supramolecular yield,45 where the primary intermolecular interaction is an O–H· · ·N hydrogen bond between the acid and the N-heterocyclic nitrogen atom (Scheme 3). The physical basis for this supramolecular motif can be ascribed to a desire of the system to maximize electrostatic interactions. The pyridyl nitrogen atom and the –OH moiety constitute the best hydrogen-bond acceptor and donor,36, 37 respectively, and these moieties display a strong preference for each other. The key to understanding and predicting the supramolecular behavior of a family of (N-heterocycle/amide) SRs is the relative hydrogen-bonding (HB) ability46 (based on basicity and molecular electrostatic surface potential) of the N-heterocycle (Scheme 4). O R

O O H

H H N

O R N

N N H

H O

O

H

Scheme 3 Typical tetrameric supermolecule in binary 1 : 1 cocrystals of isonicotinamide and carboxylic acids. ~ −260 kJ mol−1

If the pyridyl site is replaced with a more basic moiety (while retaining the amide as the second binding site), a carboxylic acid will bind to the N-heterocycle rather than to the amide (C. B. Aaker¨oy, J. Desper, M. F. Fasulo, and M. M. Smith, unpublished results). On the other hand, if the basicity of the nitrogen atom is lowered sufficiently, the resulting O–H· · ·N hydrogen bond may at some point become so weak that the carboxylic acid abandons the N-heterocycle in favor of an acid· · ·amide heterodimeric interaction. To establish whether this idea would translate to other molecules, a new ditopic SR, 4-(pyrazol-1-yl)methylbenzamide Scheme 4, 1), was synthesized and allowed to react with a series of carboxylic acids.47 All crystal structures obtained display the same primary motif where the ditopic pyrazole/benzamide SR binds to a carboxylic acid via its amide moiety. This is in stark contrast with the interactions that have been observed between carboxylic acids and SRs composed of an amide moiety and a substantially more basic N-heterocyclic nitrogen atom than the pyrazole moiety (e.g., pyridine or benzimidazole). This study demonstrates that it is possible to organize molecules within supramolecular aggregates by following relatively simple design principles based on solution-based pKa values or calculated molecular electrostatic potential surfaces.

2.2.2 Cyanophenyloximes: reliable tools for hydrogen-bond-directed supramolecular synthesis of cocrystals A systematic structural and spectroscopic examination of the products resulting from cocrystallization reactions between three types of phenyloximes R–C=N–OH (Scheme 5) and a series of N-heterocyclic hydrogen-bond acceptors demonstrates that the acidity of the oxime –OH hydrogen-bond donor is crucial to the efficacy of the supramolecular assembly process.48 The supramolecular yield for each family of oxime is given in Table 1. Note that the term “supramolecular yield” is clearly distinct from the regular usage of the term “yield” (specific to one particular reaction). It is intended to give a sense of how frequently one can expect a desired supramolecular synthon to appear if two specific functional groups

~ −301 kJ mol−1 O N H

H

~ −232 kJ mol−1 O

H N

H pK a = 3.61

O N

N

N

3

H N

N

H

N pK a = 5.36

pK a = 1.65 1

Scheme 4 Molecular electrostatic potentials and pKa values for three ditopic SRs. A more negative potential is indicative of a more basic nitrogen atom. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

4

Supramolecular materials chemistry ~137 kJ mol−1

~135 kJ mol−1 ~174 kJ mol−1 H H O ~ −224 kJ mol−1 O N N NC ~ −212 kJ mol−1

H O N ~ −204 kJ mol−1

(a)

H

(b)

(c)

Scheme 5 The magnitude of the maxima and minima (kJ mol−1 ) on the electrostatic potential surface. The minimum in the electrostatic potential for cyanophenyloxime, c, occurs on the C≡N nitrogen atom. The three families also display distinctly different pKa values: (a) 11.30–11.44, (b) 10.80–11.05, and (c) 8.97–9.01. Table 1

Supramolecular yields for each type of oxime. Cyanophenyloximes

Acetophenoneoximes

16 16 100

24 2 8

Number of total attempts Number of cocrystals formed % Supramolecular yield

(in this case, hydrogen-bond donor and hydrogen-bond acceptor) are present together in a significant number of different reactions. Despite the fact that each oxime family was allowed to react with the same set of ditopic N-heterocyclic compounds, the results are dramatically different. The significantly more acidic cyanophenyloximes were extremely effective (16/16) at forming cocrystals, whereas the less acidic acetophenonoximes/benzaldoximes produced cocrystals only in 3 out of a total of 48 attempts. It is certainly plausible that under certain circumstances more cocrystals between the less acidic oximes and heterocycles employed in this study can be prepared. However, the trends and overall picture that emerge are highly unlikely to change and cyanophenyloximes are comparable to carboxylic acids in terms of success rate. The importance and validity of using experimental pKa values and calculated electrostatic potential surfaces as a basis for predicting the supramolecular yield of an O–H· · ·N interaction for driving the formation of cocrystals (within a functional group class) is further supported.

2.3

Heteromeric interactions are better than homomeric interactions

An examination of reported cocrystals found in the CSD49 shows that most of them have been prepared using strategies that rely on the deliberate combination of chemical entities (or functional groups) located on different molecules capable of forming complementary heteromeric motifs (Scheme 6) rather than homomeric interactions (which leads to recrystallizations instead of cocrystallizations).21, 25, 27, 28, 50

Benzaldoximes 24 1 4

H

H

H N

O

O H

O

N

O H O (b)

(a)

O

NC H N (c)

H N

C

N O

H N (d)

Scheme 6 Four dimeric motifs constructed via heteromeric intermolecular interactions. Heterodimers formed between (a) carboxylic acid:aminopyridine, (b) carboxylic acid:benzamide, (c) phenol:pyridine, and (d) cyanooxime:pyridine.

Since a strong heteromeric interaction is often needed in order to overcome the “selfishness” of molecules, it is not surprising that the most widely used synthons for the directed assembly of binary cocrystals contain a carboxylic acid in combination with a suitable N-containing heterocycle. Such pairs are obviously capable of producing some of the strongest hydrogen bonds that can commonly form between organic molecules. For example, there are currently 17 cocrystals in the CSD [2] with pyrazine, 15 with phenazine, 76 with 4,4 -bipyridine, 98 with pyrimidine, and 36 cocrystals with either azopyridine, quinoline, phenantroline, and a hydroxyl group-based counterpart. In all these cases, the driving force for the assembly of the cocrystal is the heteromeric carboxylic acid· · ·N(heterocycle), O–H· · ·N,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystals: synthesis, structure, and applications hydrogen bond. For other N-heterocyclic compounds that are functionalized with substituents that are also capable of, in theory, forming hydrogen bonds with a carboxylic acid, the results are still very consistent; 11 of 12 carboxylic acid:isonicotinamide cocrystals contain the acid· · ·pyridine hydrogen bond—a good example of a supramolecular reaction with high structural fidelity. Conversely, there are very few reported examples of cocrystals where a primary noncovalent interaction is not present between the two different molecules in the lattice. Rare examples include 4-nitrobenzamide pyrazinecarboxamide (1 : 1)20 and 3,5-dinitrobenzoic acid:4-(N,N-dimethylamino)benzoic acid (1 : 1).51 In the latter structure, each forms a homomeric dimer instead of the much more commonly occurring heteromeric dimer (when two different carboxylic acids are present).52 On the basis of the extensive structural data obtained from the CSD, it is obvious that in order to bring together two different discrete neutral molecular species into the same crystalline lattice, there should be some thermodynamically based reason (originating from a heteromeric molecular-recognition event) for the subsequent solid-state assembly. Although stray structures that cannot readily be explained within such a context will appear from time to time, there is no doubt that the overall structural trends, patterns of behavior, and reproducible motifs can be developed into reliable and versatile supramolecular synthetic strategies.

2.4

Various examples of binary hydrogenbonded cocrystals

Although there are no “discrete” aggregates within a solidstate framework, it can still be helpful to classify an assembly as being 0D, 1D, 2D, or 3D on the basis of

(a)

(b)

(c)

(d)

5

the type of intermolecular interactions that exist between molecules within the lattice. It is often quite clear which interactions are directly responsible for the actual assembly of a heteromeric aggregate and, based on this information, a topological assignment can be made. Some examples of 0D assemblies (Scheme 7) in cocrystals include heteromeric carboxylic acid:carboxylic acid dimers (1 : 1),53 pyridine:carboxylic acid dimers (1 : 1),54 2-aminopyrimidine:carboxylic acid trimers (1 : 2),55 pyridine:dihydroxybenzene trimers (2 : 1),56 bipyridine: carboxylic acid trimers (1 : 2),23 2-aminopyrimidine:carboxylic acid tetramers (2 : 2),55, 57 bipyridine:resorcinol tetramers (2 : 2),24 3,5-dinitrobenzoic acid:nicotinic acid tetramers (2 : 2),58 isonicotinamide:carboxylic acid tetramers (2 : 2),45 2-pyridone:carboxylic acids pentamers (4 : 1),59 melamine:thymine tetramers (1 : 3),60 melamine: barbital hexamers (3 : 3),61 and tripenylphosphine oxide with a variety of hydrogen-bond donors (1 : 1).35, 62, 63 Examples of cocrystals containing chains, ribbons, and other infinite 1D motifs (Scheme 8) include bipyridine: dihydroxybenzene,66 melamine:cyanuric acid,67 bipyridine: (fluorinated)dibromobenzene,68 2-aminopyridine: dicarboxylic acids,69 triaminopyrimidine:barbituric acid,22 2-aminopyrimidine:dicarboxylic acid,34 1,2,3-trihydroxybenzene:hexamethylenetetramine,70 diols:diamines,71 bisbenzimidazole:dicarboxylic acids,72 and 2-amino-5-nitropyrimidine:2-amino-3-nitropyridine.44 Many infinite 2D assemblies have also been constructed including (but not limited to) piperazine:carboxylic acid,76 trithiocyanuric:bipyridine,77 pyridyloxamide:dicarboxylic acid,78 picolylaminocyclohexenone:dicarboxylic acid,79 triazine:uracil,80 tris-(4-pyridyl)triazine:trimesic acid,81 bipyridine:ureylene dicarboxylic acid,26 and isonicotinamide: dicarboxylic acid75 (Scheme 9). Finally, carbamazepine:tetracarboxylic acid-adamantane83 represents one of the few examples of clearly defined

Scheme 7 Three 0D motifs found in cocrystals created by heteromeric hydrogen-bond interactions with primary synthon as (a) acid · · ·acid,64 (b) OH · · ·N,24 (c) amide · · ·acid,20 and (d) trimeric supermolecule with acid · · ·amide and acid · · ·pyridine heterosynthon.65 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

6

Supramolecular materials chemistry 1,3,5-tris(4-carboxyphenyl)adamantane (TCA) with bipyridine (bipy), phenazine, or 4,4 -azopyridine shows formation of one of a kind of 3D Borromean arrayed structure via combination of homomeric and heteromeric hydrogenbonded synthons (Scheme 10).84

(a)

2.5 (b)

From binary to higher order cocrystals

Despite their success, N-heterocyclic amides are not sufficiently versatile to make them ideal SRs. First, they can form self-complementary amide· · ·amide and amide· · ·pyridine hydrogen bonds, which makes it inherently difficult to combine any of them with molecules that lack moieties that can compete successfully with such synthons. Second, if the two binding sites are attached to the same backbone (as is the case with isonicotinamide), it is not possible to tune the charges on individual atoms of the two sites independently, which reduces versatility. Thus, there is a need for second-generation SRs that can be refined such that they offer more opportunities for modular supramolecular synthesis through enhanced structural selectivity and specificity.

(c)

Scheme 8 Infinite 1D motifs created by heteromeric hydrogenbond interactions. Cocrystals of (a) trans-1,2-bis(4pyridyl)ethylene and resorcinol,73 (b) bipyridine and glutaric acid,74 (c) isonicotinamide and oxalic acid.75

and predictable 3D motifs observed in a hydrogen-bondbased binary cocrystal. Recently reported cocrystals of

(a)

(b)

Scheme 9 2D motifs found in cocrystals of (a) active pharmaceutical ingredients (APIs) and succinic acid82 and (b) isonicotinamide and succinic acid.75 2D sheets generated through O–H· · ·N and N–H· · ·O hydrogen bonding.

COOH COOH

HOOC

N bipy

TCA

(a)

N

(b)

Scheme 10 (a) 3D network in TCA:bipy cocrystal. (One layer is omitted for clarity). (b) Space-filling view of a Borromean weave consisting of three TCA·bipy honeycomb nets in 2TCA·bipy cocrystal.84 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystals: synthesis, structure, and applications

7

O(84) N(84) O(81) N(61) O(82)

Stronger base N

O(83)

N(51) N N

O(72)

Weaker base

O(73)

N(53)

N(73) O(71)

(a)

O(74) O(75)

N(75) O(76)

(b)

Scheme 11 (a) An asymmetric ditopic supramolecular reagent with two different binding sites (hydrogen-bond acceptors). (b) The stronger acid binds to the best hydrogen-bond acceptor (left-hand side), and the weaker acid binds to the second-best acceptor (right-hand side).

Against this background, we recently synthesized4 a series of asymmetric bis-heterocycles where two different binding sites85 (hydrogen-bond acceptor sites) are linked by a methylene bridge in order to provide increased solubility in a range of solvents. They also lack strong hydrogenbond donors and, consequently, homomeric intermolecular interactions are unlikely to prevent the desired heteromeric interactions (Scheme 11a). The basicity of each heterocycle can be independently altered through suitable covalent substitution, which provides a practical handle for fine-tuning differences in intermolecular reactivity. The ability of these SRs to form ternary supermolecules with predictable connectivity was put to the test by allowing each SR to react with pairs of different carboxylic acids in a 1 : 1 : 1 ratio (Scheme 11b). The target in each case is a cocrystal containing 1 : 1 : 1 ternary supermolecules where the primary intermolecular interactions can be rationalized according to the best donor/best acceptor protocol. All five structures obtained so far contain supermolecules with the desired connectivity and stoichiometry. Differences in basicity and electrostatic potential translate into supramolecular reactivity and selectivity that subsequently carry over into the solid state.

Scheme 12

2.6

Synthesis of cocrystals and the supramolecular yield

The fact that in the crystal structures of 4-bromo-4 cyanobiphenyl86 and 4-bromobenzonitrile,87 the molecular components are aligned in a head-to-tail manner with relatively short Br· · ·N contacts indicates that there are stabilizing intermolecular interactions between cyano and bromo moieties88 (Scheme 12). However, there is still no example of successful synthesis of binary cocrystals driven by CN· · ·Br interactions. Such interactions can organize molecules within a lattice but have yet to bring about the assembly of heteromeric cocrystals. There is clearly a difference between observing a large number of short contacts in molecular crystal structures composed of only one type of building block and translating such interactions into useful synthetic tools for constructing heteromeric architectures. The success and efficiency for any set of supramolecular reactions can be judged by the frequency of occurrence of desired intermolecular interactions and connectivities in the resulting solid. The probability that a certain motif will appear in a crystal structure is, in many ways, a

A chain of molecules organized in a head-to-tail manner in 4-bromobenzonitrile.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

8

Supramolecular materials chemistry

measure of the yield of a supramolecular reaction. Just as a covalent synthetic chemist searches for ways in which a specific reaction can be promoted or prevented, a supramolecular chemist tries to identify the experimental regime where a synthon prevails despite competition from other noncovalent forces. Much current work in organic crystal engineering is now targeted toward synthesizing cocrystals using supramolecular reactions based on reliable synthons, and, so far, the vast majority of organic molecular cocrystals have been assembled via conventional, strong hydrogen bonds. (Charge-transfer complexes are not included in this discussion.)52, 59 Weaker hydrogen bonds and many other intermolecular interactions such as nitro· · ·iodo, cyano· · ·nitro, and halogen· · ·halogen have not yet been found to be useful tools for construction of cocrystals. However, the halogen bond has emerged as an important alternative that can operate in conjunction with hydrogen bonds in the directed assembly of new cocrystsls (Section 3).

3 3.1

HALOGEN-BOND-BASED COCRYSTALS What is halogen bonding?

HB is undoubtedly the most frequently utilized noncovalent interaction in molecular-recognition processes. However, halogen bonding (XB) is a noncovalent interaction that is in some ways analogous to HB, and it may therefore be used as a practical tool for cocrystal synthesis. In HB, a hydrogen atom is shared between an atom, group, or molecule that “donates” and another that “accepts” it. In XB, it is a halogen atom X that is shared between a donor atom D and an acceptor A. Thus, the two types of interaction can be described as in Scheme 13. XB is a noncovalent intermolecular interaction between halogen atoms acting as halogen-bond donors (X) and electron-rich moieties such as the nitrogen atom in pyridine acting as a halogen-bond acceptor (A) (Scheme 13).89 The strong preference for linear D–X· · ·A interaction parallels that known for D–H· · ·A hydrogen bonds, and X

preferences for the geometries at the acceptor are also consistent with those observed in HB. XB is a highly directional interaction, even more so than HB and short interactions are more directional than long ones.90–92 According to the definition given above, the term halogen bonding covers a vast class of noncovalent interactions, the strength of which can vary in the range 10–200 kJ mol−1 .93, 94

3.2

Halogen-bond donor tectons

Theoretical and experimental studies have confirmed that electron density is anisotropically distributed around halogen atoms in organic halides. In covalently bonded halogen atoms, the effective atomic radius along the extended C–X bond axis is smaller than in the direction perpendicular to this axis, and a region of positive electrostatic potential is present along the covalent bond. This “σ hole” attracts the lone pair of the halogen-bond acceptor (A) closer to the halogen atom (X), and accounts for the linear orientation of the halogen bonds.93, 95–97 This “σ hole” polarization is strongly enhanced in the order Cl < Br < I,96 and varies with the carbon atom hybridization in the increasing order of C(sp3 )-X < C(sp2 )-X < C(sp)X.98, 99 Halogen-bond donor capacity of halogen atoms is strongly enhanced in the presence of electron-withdrawing atoms as in iodoperfluorocarbons, for example, perfluorinated iodoalkanes, F2n+1 Cn -I or pentafluoroiodobenzene, F5 C6 -I.100, 101 This molecular-recognition “activation” is also demonstrated by the fact that a survey of the CSD produces only five cocrystals for 1,4-diiodobenzene, whereas the activated version, 1,4-diiodotetrafluorobenzene, produces 46 entries.

4

EXAMPLES OF HALOGEN-BONDED COCRYSTALS

The following overview of halogen-bonded cocrystals uses a classification of motifs as being 0D, 1D, 2D, or 3D based on the primary halogen bond interactions.

A

D

X

A

D

HB: D–H···A XB: D–X···A X = I, Br, Cl A = N, O, S, Se, Cl, Br, I D = C, N, halogen, and so on.

Scheme 13

General scheme for the formation of halogen bonds.89

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystals: synthesis, structure, and applications

F N O

F

I

F I

F

F

R2

F

F

R1

I 3

1, (TMIO)2·DITFB F F I N

N I S

F

FF

Discrete assemblies

Some examples of 0D assemblies (Scheme 14) in halogenbonded cocrystals include heteromeric isoindoline nitroxide-1,1,3,3-tetramethylisoindolin-2-yloxyl (TMIO): diiodotetrafluorobenzene (DITFB) trimers (2 : 1).102 These trimeric cocrystals were studied to demonstrate the ability of XB to assemble paramagnetic molecules, such as TMIO, in supramolecular arrays. Reported crystal structures show expected N–O· · ·I interaction consistent with n → σ ∗ electron donation from the nitroxide N–O• moiety. Further development in this type of system could lead to potential application in the fields of molecular magnetism, spintronics, and quantum computing. When 2-methylbenzothiazole was cocrystallized with 1,2-diiodotetrafluoroethane (DITFE), to study the binding ability of sulfur versus nitrogen, the resulting structure shows that the cocrystal is assembled via N· · ·I XB interactions, and S· · ·I interactions were not present in the complex. This observed preference of activated iodine for N over S has been attributed to soft–hard theory. Powerful electron-withdrawing ability of the fluorine pulls and holds electron cloud from the iodine atom, making it less soft when it interacts with a base, causing N· · ·I (hard–hard) interactions to prevail over S· · ·I (soft–hard).103 Pyridine itself does not form cocrystals with pentafluoroiodobenzene (PFIB) 3 at ambient temperature. When pyridine was substituted (4–7) with electron-donating groups such as methyl groups, cocrystals were formed (Scheme 15). This cocrystal formation has been attributed to enhancement of the electron donor character of the aryl rings and thus an increased electrostatic interaction with the pentafluorophenyl moiety of 3. Although both the halogen-bond donor and acceptor compounds are liquids at room temperature, desired cocrystals were solids with melting points far above room temperature due to a large contribution to the lattice energy from

F

F

F

R5

N

R1 = R3 = R5 = CH3; R2 = R4 = H R1 = R5 = CH3; R2 = R3 = R4 = H R2 = R4 = CH3; R1 = R3 = R5 = H R3 = CH3; R1 = R2 = R4 = R5 = H

R1 I

F

Scheme 14 Discrete assembly of trimeric cocrystals based on halogen bonds.102, 103

R4

4 5 6 7

S

2, DITFE·2-methylbenzothiazole

4.1

R3

F O N

F

9

R2

N R5

R3 R4

Scheme 15 Cocrystal formation of iodopentafluorobenzene with electron-rich pyridine derivatives.101

XB and aryl–perfluoroaryl interactions between substituted pyridines.101

4.2

From 1D chains to 3D interpenetrated networks

One-dimensional (1D) chains are formed when both the donor and the acceptor molecules are bidentate. The noteworthy strength and high directionality of halogen bonds formed by activated iodo- or bromoperfluorocarbons have been the basis for the assembly of numerous 1D motifs with ditopic XB donors and ditopic XB acceptors (Table 2). Linear 1D polymers are formed when the axes of the donor and acceptor sites are parallel and coaxial.68, 104–110 For example, 4,4 -bipyridine and its ethylene analog 4,4 ethane-1,2-diyldipyridine afford linear chains with DITFB. Similarly, 1,4-dibromotetrafluorobenzene (TFDBB) yields 1D chains held together via XB. Moreover, when 1,4diiodobenzene, a nonactivated XB donor, is cocrystallized with ditopic XB acceptors, such as 4,4 -bipyridine, and its ethylene analog 4,4 -ethane-1,2-diyldipyridine, 1D chains with XB between nitrogen and iodine as primary noncovalent interaction are produced (Table 2). When the binding sites are parallel, but no longer collinear, stepped infinite chains are formed.68, 111–114 For example, 4,4 -bipyridine, and its ethylene analog 4,4 ethane-1,2-diyldipyridine, afforded zigzag chains upon cocrystallization with 1,3- and 1,2-diiodotetrafluorobenzene. When the axes of the binding sites in the components are not parallel, the self-assembled chains yielded a herringbone arrangement.115–117 These studies clearly demonstrate the ability of halogen bonds to govern structural outcome based on angles of binding site axes in the components.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

10

Supramolecular materials chemistry

Table 2 Various halogen-bond donor and acceptor ligands employed to obtain 1D infinite chains. +

Self assembled 1D chain XB donors

F

F

I F

I

N

I

O N

Br

N

N

F

I

F

XB acceptors

N O

F N

Br F

F

N

N

CN

NC

X Y X or Y = NH, S, O

Haloperfluoroalkanes are particularly robust tectons widely studied in the literature. For example, α,ωdiiodoperfluoroalkanes (α,ω-DIPFAs) give rise to infinite 1D chains with a variety of bidentate halogen-bond acceptor molecules, since fluorine atoms inductively boost the electron density acceptor ability of the terminal iodine substituents in α,ω-DIPFAs (Scheme 16). When α,ω-DIPFAs were cocrystallized with a series of terminal dicynoalkanes, the outcome was 1D infinite chains, 8, with N· · ·I XB being the primary interaction governing

self-assembly (Scheme 16). All the cocrystals show highly consistent geometric characteristics when the chain lengths of donor and acceptor modules were varied systematically. The donor and acceptor modules alternate in infinite chains, and the pitch shows a very good linear correlation with the number of the methylene groups and the number of the difluoromethylene groups. The low affinity between hydrocarbons (HCs) and perfluorocarbons (PFCs) induces remarkable segregation, and the infinite chains produce alternating layers of PFCs and HCs with impressively similar overall structures.118 To study XB donor strength of aliphatic nitrogen versus oxygen with activated iodine, 1,4,10,13-tetraoxa7,16-diazacyclooctadecane (Kryptofix 2.2., K.2.2) was cocrystallized with α,ω-DIPFAs. The result was 1D chains with N· · ·I XB synthons (Scheme 16, 9). In all cocrystals of α,ω-DIPFAs with Kryptofix 2.2, the nitrogen atoms were involved in XB; no structural interference was observed from the oxygen atoms, which clearly demonstrates the selectivity of XB in the presence of potentially competing heteroatomic donor sites.119 When various substituted pyrazines were cocrystallized with α,ω-DIPFAs to study halogen-bond acceptor ability of sp2 -N with activated iodine, inevitably 1D chains formed with N· · ·I XB being the primary synthon (Scheme 16, 10). Site selectivity of α,ω-DIPFA’s activated iodine was tested with trans-4,4 -azobipyridine, having two nitrogen atoms with varying basicity. Interestingly, a site-selective supramolecular synthesis occurs, since only the pyridyl nitrogen atoms become involved in the recognition process, giving rise to 1D chains (Scheme 16, 11). This selectivity can be rationalized on the basis of a higher basicity and steric accessibility of the pyridine nitrogens.100 Resnati and coworkers have shown that 2D architectures were obtained when tetradentate XB acceptors at the nodes and bridging linear bidentate XB donors were combined.120 Recently, cocrystals of tridentate halogenbond acceptor, 1,3,5-tris-4-pyridyl(ethenyl)benzene, were obtained with tridentate and bidentate XB donors (triiodotrifluorobenzene and DITFB, respectively). Both cocrystals have the expected N· · ·I as a primary intermolecular interaction governing 2D network (Scheme 17).121 Self-assembly of pyridyl-based tetradentate XB acceptors with bidentate XB donors, α,ω-DIPFAs, gives rise to 3D architectures with N· · ·I being a primary motif (Scheme 18).121 Similar to hydrogen-bond-based networks, the halogenbonded 2D and 3D networks described above often contain large cavities, as is the case for similarly sized networks assembled through HB.123, 124 The voids are filled by solvent molecules or through interpenetration in the overall crystal packing.125

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystals: synthesis, structure, and applications

O

F F

O

NH

I(

HN

O

O

O

)n

O

NH

I

HN

O

F F

11

O

9 O

)

H

HN

H

O

)

H H n

F

F F

I

NC (

)n

(

(

F

F F

N

N

I

I

)nCN H H

CN

)

I

n

H N

F F

F F

I

F F

I

n

(

O

H

(

N

NC

O

NH

NC

H H

)

N

N

N N

N

H

(

N 10

n

H

CN

8 F F N N

N

Scheme 16

I(

N

)n

I

N

N N

N

F F 11

Diagram showing cocrystals given by α,ω-DIPFAs with aliphatic and aromatic dinitrogen hydrocarbons.100, 118, 119

5

N F I

I

COCRYSTALLIZATION USING HARMONY BETWEEN HALOGEN BONDS AND HYDROGEN BONDS

+ F

F I

N

N

Scheme 17 Primary halogen-bond motif causing a 2D selfassembled structure.121

Success in supramolecular synthesis requires us to identify hierarchies of intermolecular interactions and then to develop supramolecular synthetic strategies that exploit synthons that can operate alongside without interfering with each other.42 However, a strategy that relies exclusively on hydrogen bonds could soon run into problems for precise synthesis of more complex cocrystals or extended networks, as it would become unavoidable to encounter crossover reactions between multiple hydrogen-bond-based synthons that compete with each other in one reaction mixture. Accordingly, it may be valuable and, in the long term, more efficient to develop supramolecular strategies that can accommodate two different noncovalent interactions in such a way that they are unlikely to interfere with each other. Therefore, molecular-recognition studies based on molecules with both hydrogen- and halogen-bond donor have started to gain attention recently.126–129

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

12

Supramolecular materials chemistry

N

O N O

O

+

I

(CF2)4

I

O N

Scheme 18

N

Three-dimensional networks formed by the self-assembly of bidentate XB donors and tetradentate XB acceptors.89, 122

~ −260 kJ mol−1

~165 kJ mol−1

N OH

H

N ~ −291 kJ mol−1 N

N

F

F

F 12

(a)

~71 kJ mol−1

F I 13

example, 13; in all three crystal structures, Etter’s rule36 is obeyed. Scheme 19(b) shows a tetrameric cocrystal formation due to alliance between HB and XB. To further examine the competition between HB and XB, isonicotinamide, a bifunctional probe44 was cocrystallized with I2 and TFDIB, respectively, as XB donors. In each case, the XB donors were able to compete successfully with the potential amide N–H(anti)· · ·N(py) hydrogen bond and both structures contained networks built around a core of homomeric amide· · ·amide dimers extended through N· · ·I XB (Scheme 20).127 The above-mentioned studies show that HB and XB can coexist in the same structure and have the capability to compete successfully with weak CH· · ·N/O hydrogen bonds (Scheme 19b). Another series of studies by the Goroff group130, 131 and recently by our group129 also shows how HB and XB can be brought together in effective synthetic strategies.

6

(b)

Scheme 19 (a) Electrostatic charges for the hydrogen-bond donor/acceptor sites in 12 and 13, based on AM1 calculated MEPs. (b) Cocrystal of 12·13 showing tetrameric supermolecule constructed through hydrogen bonds and halogen bonds.126

To develop effective supramolecular synthetic strategies around a hierarchy of synthons that comprise both hydrogen and halogen bonds, we employed a ditopic structural probe molecule, 12, containing two sites (pyridyl and benzimidazole) that can act as either hydrogen-bond or halogen-bond acceptors. The counterpart molecule containing a weak and a strong hydrogen-bond donor, as well as one potential halogen-bond donor, is used (Scheme 19). When the probe, 12, is cocrystallized with HB/XB donor hybrids, for

APPLICATIONS OF COCRYSTALS

Most organic specialty chemicals (e.g., dyes, explosives, and detergents) exist as solids at ambient conditions, and the precise structural details of a particular solid define most of its physical properties such as solubility, thermal/mechanical stability, and particle morphology (which influences downstream processability and formulation). For example, the performance of active pharmaceutical ingredients (APIs) and biocides is governed by bioavailability and absorption, and while there are technologies that can alter the time a drug/toxin remains in the biological target, there are few options for modulating solubility. Particle size reduction may improve kinetic solubility but it can also introduce problems such as agglomeration and instability during formulation. However, by fine-tuning the crystalline environment of a compound without altering its molecular structure, we could potentially “dial-in” desirable physical properties, which would be highly significant to manufacturers/consumers of organic specialty chemicals. Cocrystals

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystals: synthesis, structure, and applications

13

(a)

(b)

Scheme 20

Extended self-assembled network formed between isonicotinamide and (a) I2 , (b) TFDIB.127

represent a hitherto untapped resource for new solid forms of functional materials, and although several areas involving functional materials such as nonlinear optical compounds,132, 133 liquid crystals, (see Liquid Crystals Formed from Specific Supramolecular Interactions, Soft Matter)134–138 energetic materials,139 nutraceuticals,140, 141 and organic semiconductors,142 but by far away most attention has been given to the synthesis and characterization of pharmaceutical cocrystals.143–146

6.1

Pharmaceutical cocrystals

Since most drug molecules contain one or more polar groups that can engage in HB (either as a donor or as an acceptor), it is not surprising that most pharmaceuticals reported to date have been prepared using heteromeric synthons involving different molecules with complementary binding sites. Several well-known commercial compounds have been targeted in cocrystal synthesis such as gabapentine,147 aspirin,23 norfloxacin,148 stanolone,149 caffeine [3], saccharine [4], efavirenz,150 and many others. The synthetic strategy is typically hydrogen-bond driven, and the active ingredient can either act as a powerful hydrogenbond donor (as in saccharin 4-methyl pyridine N-oxide151 ) (Scheme 21) or a hydrogen-bond acceptor (as in benzoic acid carbamazepine152 ) (Scheme 22). Although most publications on pharmaceutical cocrystals to date focus on synthesis and structural characterization,

Scheme 21 The primary hydrogen bond in the saccharin/4methyl pyridine N-oxide cocrystal.

Scheme 22 The primary hydrogen-bond interaction in the 1 : 1 cocrystal benzoic acid carbamazepine.

considerable efforts have been made to improve on one or more physical properties of the solid form of a particular API.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

14

6.2

Supramolecular materials chemistry

Changing physical properties through cocrystallization

Cocrystals of drugs and drug candidates represent new types of materials for pharmaceutical development, and they may enable the incorporation of an API within a solid “casing” that leads to enhanced mechanical or thermal stability, reduced hygroscopicity, or improved bioavailability. Furthermore, cocrystallizations allow such properties to be realized without requiring any changes at the molecular level of the API, which is essential as the active component derives its specific reactivity or function from what is frequently a carefully and painstakingly constructed molecular structure. The pharmaceutical industry has only recently become fully aware of these opportunities (only a handful of relevant patents have been filed to date) and the deliberate synthesis of cocrystals is poised at making a dramatic transition from fundamental research to applied medicinal therapeutics. The physical property that is the most commonly targeted for manipulation is undoubtedly the aqueous solubility,153 and although considerable progress has been made, we are not yet at a stage where the solubility of a cocrystal of an API can be predicted with any degree of certainty. Nevertheless, numerous key components of the necessary experimental and theoretical foundation for understanding aqueous solubility of cocrystals have been provided by Rodr´ıguez-Hornedo and coworkers. In several seminal papers,154, 155 they have brought together invaluable information on solubility products, solubility complexation, and phase-solubility diagrams in order to provide cohesive models for explaining cocrystal solubility. These models indicate, for example,156 that (i) the aqueous solubility of a cocrystal AB is accurately described by the solubility

Scheme 23

product of the cocrystal components and by solution complexation constants, and (ii) the equilibrium constants can be determined from solubility methods. Zaworotko and coworkers have also used animal pharmacokinetic (PK) studies to demonstrate how the PK profile of a poorly soluble antiepileptic drug, lamotrigine, can be altered significantly through the use of cocrystal formation.157 Finally, in order to find a way of fine-tuning melting point and aqueous solubility of a representative, A, of a family of anticancer compounds, Aaker¨oy and coworkers reported the synthesis of five cocrystals with aliphatic, even-chained, dicarboxylic acids.82 The supramolecular synthesis was driven by the well-known COOH· · ·py hydrogen-bond-based synthon, and in each reported case, infinite API· · ·diacid· · ·API· · ·diacid chains were obtained, and these were subsequently arranged into 2D layers driven by API-based self-complementary amide· · ·amide hydrogen bonds (Scheme 23). The structural consistency in the series of five cocrystals implies that differences in physical properties of the cocrystals may be, to a first approximation, directly related to difference in molecular properties of the cocrystallizing agent (the diacid). The subsequent study of thermal properties confirms this approach (Scheme 24) as the highest melting cocrystal contains the dicarboxylic acid with the highest melting point. The relationship between molecular structure and the bulk property shows a strong positive correlation. Second, measurements of intrinsic aqueous solubility of A1–A5 (Scheme 25) show a modest increase by a factor of 2.5 (which is achieved without altering the inherent biological activity of the API), but, more importantly, the trend in solubility can readily be rationalized directly in terms of the aqueous solubilities of the dicarboxylic acids.

2D layers in two isostructural cocrystals of a bis(4-pyridinecarboxamide)-based API and different dicarboxylic acids.82

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Cocrystal melting point (°C)

Cocrystals: synthesis, structure, and applications

R 2 = 0.9844

195

A1

185 175

A2 A3

165 155

A5

A4

145 135 125 125

135

145

155

165

175

185

195

Dicarboxylic acid melting point (°C)

Scheme 24 Melting points of five cocrystals A1–A5 as a function of melting point of the corresponding carboxylic acid (1 = succinic acid, 2 = adipic acid, 3 = sebacic acid, 4 = suberic acid, 5 = dodecanedioic acid). 0.6000

Solubility (g L−1)

work by Boese and coworkers, it is clear that in their elegant work using low-temperature crystallizations they intentionally prepare cocrystals with a very clear and deliberate strategy.9 [2] CSD search carried out on ConQuest Version 1.12. (Updated till February 2010.) [3] There are currently about 40 cocrystals of caffeine in the CSD. [4] There are about 10 cocrystals reported for saccharin in the CSD.

REFERENCES 1. J. D. Dunitz, in Perspectives in Supramolecular Chemistry: The Crystal as a Supramolecular Entity, eds. G. R. Desiraju, John Wiley & Sons, Amsterdam, 1995. 2. C. B. Aaker¨oy, J. Desper, and J. F. Urbina, Cryst. Growth Des., 2005, 5, 1283.

0.5000 0.4000

3. C. B. Aaker¨oy, J. Desper, M. M. Smith, and J. F. Urbina, Dalton Trans., 2005, 2462.

0.3000

4. C. B. Aaker¨oy, J. Desper, and J. F. Urbina, Chem. Commun., 2005, 2820.

0.2000 0.1000

5. C. B. Aaker¨oy, J. Desper, and J. F., Urbina, CrystEngComm, 2005, 7, 193.

0.0000

6. C. B. Aaker¨oy, J. Desper, and B. Levin, CrystEngComm, 2005, 7, 102.

A

Scheme 25

A1

A2

A3

A4

A5

Aqueous solubilities of A and A1–A5.

The more hydrophobic the acid (longer chains), the lower the solubility of the API cocrystal.

7

15

CONCLUSION

The design, synthesis, and applications of cocrystals still face many challenges but these areas comprise a research field that offers a crucial complement to covalent synthesis, and may offer new methods for optimizing bulk properties of any solid molecular material. The fact that the area is experiencing an exponential increase in relevant publications bodes well for the future, and with an ability to construct heteromolecular architectures with desirable metrics, this may soon translate into blueprints for materials design and for constructing viable biological mimics.

NOTES [1] This requirement may be deemed overly restrictive but it does offer an important distinction between solvates and cocrystals. However, in some cases, notably in the

7. G. R. Desiraju, CrystEngComm, 2003, 5, 466. 8. J. D. Dunitz, CrystEngComm, 2003, 5, 506. 9. See, e.g., M. T. Kirchner, R. Boese, A. Gehrke, and D. Blaeser, CrystEngComm., 2004, 6, 360. 10. For information about classic charge-transfer compounds, see J. Rose, Molecular Complexes, Pergamon Press, Oxford, 1967. 11. S. Shan, E. Batchelor, and W. Jones, Tetrahedron Lett., 2002, 43, 8721. 12. P. Vishweshwar, R. Thaimattam, M. Jaskolski, and G. R. Desiraju, Chem. Commun., 2002, 1830. 13. P. Vishweshwar, A. Nangia, and V. M. Lynch, CrystEngComm, 2003, 5, 126. ¨ Almarsson and M. J. Zaworotko, Chem. Commun., 14. O. 2004, 1889. 15. G. R. Desiraju and J. A. R. P. Sarma, Chem. Commun., 1983, 45. 16. F. Pan, W. S. Wong, V. Gramlich, et al., Chem. Commun., 1996, 2. 17. V. R. Pedireddi, W. Jones, A. P. Chorlton, and R. Docherty, Chem. Commun., 1996, 997. 18. S. H. Dale, M. R. J. Elsegood, M. Hemmings, and A. L. Wilkinson, CrystEngComm, 2004, 6, 207. 19. V. R. Pedireddi, J. PrakashaReddy, and K. K. Arora, Tetrahedron Lett., 2003, 44, 4857. 20. C. B. Aaker¨oy, J. Desper, and B. A. Helfrich, CrystEngComm, 2004, 6, 19.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

16

Supramolecular materials chemistry

21. C. B. Aaker¨oy, A. M. Beatty, M. Nieuwenhuyzen, and M. Zou, Tetrahedron, 2000, 56, 6693.

36. M. C. Etter, Acc. Chem. Res., 1990, 23, 120.

22. J.-M. Lehn, M. Mascal, A. DeCian, and J. Chem. Soc., Chem. Commun., 1990, 479.

J. Fischer,

38. M. C. Etter and G. M. Frankenbach, Chem. Mater., 1989, 1, 10.

23. R. D. Bailey Walsh, M. W. Bradner, S. Fleischman, et al., Chem. Commun., 2003, 186.

39. This observation was first made by J. Donohue, J. Phys. Chem., 1952, 56, 502.

24. L. R. MacGillivray, J. L. Reid, and J. Am. Chem. Soc., 2000, 122, 7817.

J. A. Ripmeester,

40. M. C. Etter, J. C. MacDonald, and J. Bernstein, Acta Cryst., Sect. B., 1990, 46, 256.

25. J. A. Zerkowski, J. C. MacDonald, and G. M. Whitesides, Chem. Mater., 1997, 9, 1933.

41. J. Bernstein, R. E. Davis, L. Shimoni, and N.-L. Chang, Angew. Chem., Int. Ed., Engl., 1995, 34, 1555.

26. J. J. Kane, R.-F. Liao, J. W. Lauher, and F. W. Fowler, J. Am. Chem. Soc., 1995, 117, 12003.

42. G. R. Desiraju, Angew. Chem. Int. Ed. Engl., 1995, 34, 2311.

27. P. Vishweshwar, A. Nangia, and V. M. Lynch, J. Org. Chem., 2002, 67, 556.

43. A. Nangia and G. R. Desiraju, Acta Crystallogr., 1998, A54, 934.

28. C. Huang, L. Leiserowitz, and G. M. Schmidt, J. Chem. Soc., Perkin Trans., 1973, 2, 503.

44. C. B. Aaker¨oy, A. M. Beatty, B. A. Helfrich, and M. Nieuwenhuyzen, Cryst. Growth Des., 2003, 3, 159.

29. Many binary cocrystals based on principles of molecular recognition have, however, been reported, e.g., (a) S. Shan, E. Batchelor, and W. Jones, Tetrahedron Lett., 2002, 43, 8721; (b) R. D. Bailey Walsh, M. W. Bradner, S. Fleischman, et al., Chem. Commun., 2003, 186; (c) L. R. MacGillivray, J. L. Reid, and J. A. Ripmeester, J. Am. Chem. Soc., 2000, 122, 7817; (d) P. Vishweshwar, R. Thaimattam, M. Jaskolski, and G. R. Desiraju, Chem. Commun., 2002, 1830; (e) J. A. Zerkowski, J. C. MacDonald, and G. M. Whitesides, Chem. Mater., 1997, 9, 1933; (f) J. J. Kane, R.-F. Liao, J. W. Lauher, and F. W. Fowler, J. Am. Chem. Soc., 1995, 117, 12003; (g) J.-M. Lehn, M. Mascal, A. DeCian, and J. Fischer, J. Chem. Soc., Chem. Commun., 1990, 479; (h) P. Vishweshwar, A. Nangia, and V. M. Lynch, CrystEngComm, ¨ Almarsson and M. J. Zaworotko, 2003, 5, 126; (i) O. Chem. Commun., 2004, 1889; (j) G. R. Desiraju and J. A. R. P. Sarma, Chem. Commun., 1983, 45; (k) C. Huang, L. Leiserowitz, and G. M. Schmidt, Chem. Soc., Perkin Trans., 1973, 2, 503; (l) F. Pan, W. S. Wong, V. Gramlich, et al., Chem. Commun., 1996, 2; (m) V. R. Pedireddi, W. Jones, A. P. Chorlton, and R. Docherty, Chem. Commun., 1996, 997; (n) P. Vishweshwar, A. Nangia, and V. M. Lynch, J. Org. Chem., 2002, 67, 556; (o) C. B. Aaker¨oy, A. M. Beatty, M. Nieuwenhuyzen, and M. Zou, Tetrahedron, 2000, 56, 6693; (p) S. H. Dale; M. R. J. Elsegood, M. Hemmings, and A. L. Wilkinson, CrystEngComm, 2004, 6, 207; (q) V. R. Pedireddi, J. PrakashaReddy, and K. K. Arora, Tetrahedron Lett., 2003, 44, 4857.

45. C. B. Aaker¨oy, A. M. Beatty, and B. A. Helfrich, J. Am. Chem. Soc., 2002, 124, 14425.

37. M. C. Etter, J. Phys. Chem., 1991, 95, 4601.

46. P. Gilli, L. Pretto, V. Bertolasi, and G. Gilli, Acc. Chem. Res., 2009, 42, 3344. 47. C. B. Aaker¨oy, J. Desper, and B. M. T. Scott, Chem. Commun., 2006, 1445. 48. C. B. Aaker¨oy, J. Desper, D. J. Salmon, and M. M. Smith, Cryst. Growth Des., 2006, 6, 1033. 49. F. A. Allen, Acta Crystallogr. Section B, 2002, 58, 380. 50. Some examples of co-crystallizations are, e.g., (a) F. Pan, W. S. Wong, V. Gramlich, et al., Chem. Commun., 1996, 2; (b) J. A. Zerkowski, J. C. MacDonald, and G. M. Whitesides, Chem. Mater., 1997, 9, 9; (c) R.-F. Liao, J. W. Lauher, and F. W. Fowler, Tetrahedron, 1996, 52, 3153; (d) N. Shan, A. D. Bond, and W. Jones, Cryst. Eng., 2002, 5, 9; (e) V. R. Pedireddi, W. Jones, A. P. Chorlton, and R. Docherty, Chem. Commun., 1996, 997; (f) P. Vishweshwar, A. Nangia, and V. M. Lynch, J. Org. Chem., 2002, 67, 556. 51. C. V. K. Sharma, K. Panneerselvam, T. Pilati, and G. R. Desiraju, J. Chem. Soc., Perkin Trans., 1993, 2, 2209. 52. V. R. Pedireddi and J. PrakashaReddy, Tetrahedron Lett., 2002, 43, 4927. 53. J. A. R. P. Sarma and G. R. Desiraju, J. Chem. Soc., Perkin Trans., 1985, 2, 1905. 54. T. Sugiyama, J. Meng, and T. Matsuura, J. Mol. Struct., 2002, 611, 53.

30. F. A. Allen, P. R. Raithby, G. P. Shields, and R. Taylor, Chem. Commun., 1998, 1043.

55. D. E. Lynch, G. Smith, D. Freney, et al., Aust. J. Chem., 1994, 47, 1097.

31. L. Leiserowitz and M. Tuval, Acta Crystallogr., 1978, B34, 1230.

56. L. R. Nassimbeni, H. Su, E. Weber, and K. Skobridis, Cryst. Growth Des., 2004, 4, 85.

32. O. Ermer and J. Neud¨orfl, Helv. Chim. Acta., 2001, 84, 1268.

57. D. E. Lynch, T. Latif, G. Smith, et al., Aust. J. Chem., 1998, 51, 403.

33. C. P. Brock and J. D. Dunitz, Acta Crystallogr., Sect. A., 1991, 47, 854.

58. J. Zhu and J.-M. Zheng, Chin. J. Struct. Chem., 2004, 23, 417.

34. M. C. Etter and D. A. Adsmond, J. Chem. Soc., Chem. Commun., 1990, 589.

59. C. B. Aaker¨oy, A. M. Beatty, and M. Zou, Cryst. Eng., 1998, 1, 225.

35. M. C. Etter and P. W. Baures, J. Am. Chem. Soc., 1988, 110, 639.

60. R. F. M. Lange, F. H. Beijer, R. P. Sijbesma, Angew. Chem., Int. Ed., 1997, 36, 969.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

et al.,

Cocrystals: synthesis, structure, and applications

17

61. J. A. Zerkowski, C. T. Seto, and G. M. Whitesides, J. Am. Chem. Soc., 1992, 114, 5473.

86. P. Kronenbusch, W. B. Gleson, and D. Britton, Cryst. Struct. Commun., 1976, 5, 17.

62. For some more examples of tripenylphosphine oxide with various hydrogen bond donors see, e.g., (a) M. Y. Antipin, A. I. Akhmedov, Y. T. Struchkov, et al., Zh. Strukt. Khim., 1983, 24, 86; (b) D. E. Lynch, G. Smith, K. A. Byriel, and C. H. L. Kennard, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 1993, 49, 718.

87. D. Britton, J. Konnert, and S. Lam, Cryst. Struct. Commun., 1977, 6, 45.

63. M. C. Etter and T. W. Panunto, J. Am. Chem. Soc., 1988, 110, 5896. 64. K. Chadwick, G. Sadiq, R. J. Davey, et al., Cryst. Growth Des., 2009, 9, 1278. 65. C. C. Seaton, A. Parkin, C. C. Wilson, and N. Blagden, Cryst.Growth Des., 2009, 9, 47. 66. C. Glidewell, G. Ferguson, R. M. Gregson, and A. J. Lough, Acta Crystallogr., Sect. C: Cryst. Struct. Commun., 1999, 55, 2133. 67. A. Ranganathan, V. R. Pedireddi, and C. N. R. Rao, J. Am. Chem. Soc., 1999, 121, 1752. 68. A. De Santis, A. Forni, R. Liantonio, et al., Chem. Eur. J., 2003, 9, 3974. 69. F. Garcia-Tellado, S. J. Geib, S. Goswami, and A. D. Hamilton, J. Am. Chem. Soc., 1991, 113, 9265.

88. H. Quast, Y. Gorlach, E.-M. Peters, et al., Chem. Ber., 1986, 119, 1801. 89. P. Metrangolo, F. Meyer, T. Pilati, et al., Angew. Chem. Int. Ed., 2008, 47, 6114. 90. A. C. Legon, Angew. Chem., 1999, 111, 2850. 91. A. C. Legon, Angew. Chem. Int. Ed., 1999, 38, 2686. 92. A. C. Legon, in Halogen Bonding Fundamentals and Applications, eds. P. Metrangolo and G. Resnati, Springer, Berlin, 2008, pp. 17–64. 93. J. P. M. Lommerse, A. J. Stone, R. Taylor, and F. H. Allen, J. Am. Chem. Soc., 1996, 118, 3108. 94. G. A. Landrum, N. Goldberg, and R. Hoffmann, J. Chem. Soc. Dalton Trans., 1997, 3605. 95. T. Clark, M. Hennemann, J. S. Murray, and P. Politzer, J. Mol. Model., 2007, 13, 291. 96. F. F. Awwadi, R. D. Willet, K. A. Peterson, and B. Twamley, Chem. Eur. J., 2006, 12, 8952.

70. M. Tremayne and C. Glidewell, Chem. Commun., 2000, 2425.

97. A. Karpfen, in Halogen Bonding Fundamentals and Applications, eds. P. Metrangolo and G. Resnati, Springer, Berlin, 2008, pp. 1–16.

71. O. Ermer and A. Eling, J. Chem. Soc., Perkin Trans., 2, 1994, 925.

98. J.-W. Zou, Y.-J. Jiang, M. Guo, et al., Chem. Eur. J., 2005, 11, 740.

72. C. B. Aaker¨oy, J. Desper, B. Leonard, and J. F. Urbina, Cryst. Growth Des., 2005, 3, 865.

99. K. Gao and N. S. Goroff, J. Am. Chem. Soc., 2000, 122, 9320.

73. T. Friscic and L. R. MacGillivray, Chem. Commun., 2009, 773.

100. D. Fox, P. Metrangolo, D. Pasini, et al., CrystEngComm, 2008, 10, 1132.

74. V. R. Pedireddi, S. Chatterjee, A. Ranganathan, C. N. R. Rao, Tetrahedron, 1998, 54, 9457.

101. A. Wasilewska, M. Gdaniec, and T. Połonski, CrystEngComm, 2007, 9, 203.

and

75. P. Vishweshwar, A. Nangia, and V. M. Lynch, Cryst. Growth Des., 2003, 3, 783. 76. T.-J. M. Luo and G. T. R Palmore, Cryst. Growth Des., 2002, 2, 337. 77. V. R. Pedireddi, S. Chatterjee, A. Ranganathan, C. N. R. Rao, J. Am. Chem. Soc., 1997, 119, 10867.

and

102. G. R. Hanson, P. Jensen, J. McMurtrie, et al., Chem. Eur. J., 2009, 15, 4156. 103. Q. Chu, Z. Wang, Q. Huang, et al., New J. Chem., 2003, 27, 1522. 104. J. Hulliger and P. J. Langley, Chem. Commun., 1998, 2557.

78. T. L. Nguyen, F. W. Fowler, and J. W. Lauher, J. Am. Chem. Soc., 2001, 123, 11057.

105. J. A. R. P. Sarma, F. H. Allen, V. J. Hoy, et al., Chem. Commun., 1997, 101.

79. J. Xiao, M. Yang, J. W. Lauher, and F. W. Fowler, Angew. Chem., Int. Ed., 2000, 39, 2132.

106. R. B. Walsh, C. W. Padgett, P. Metrangolo, et al., Cryst. Growth Des., 2001, 1, 165.

80. F. H. Beijer, R. P. Sijbesma, J. A. J. M. Vekemans, et al., J. Org. Chem., 1996, 61, 6371.

107. B. Borgen, O. Hassel, and C. Romming, Acta Chem. Scand., 1962, 16, 2469.

81. B.-Q. Ma and P. Coppens, Chem. Commun., 2003, 2290.

108. N. Masciocchi, M. Bergamo, and A. Sironi, Chem. Commun., 1998, 1347.

82. C. B. Aaker¨oy, S. Forbes, and J. Desper, J. Am. Chem. Soc., 2009, 131, 17048. 83. S. G. Fleischman, S. S. Kuduva, J. A. McMahon, et al., Cryst. Growth Des., 2003, 3, 909. 84. Y. Men, J. Sun, Z.-T. Huang, and Q.-Y. Zheng, Angew. Chem. Int. Ed., 2008, 48, 2873. 85. G. Kartum, W. Vogel, and K. Andrussov, Dissociation Constants of Organic Acids in Aqueous Solution, Butterworth & Co. Publishers, London, 1961.

109. J. L. Syssa-Magal´e, K. Boubekeur, P. Palvadeau, et al., CrystEngComm, 2005, 7, 302. 110. D. Cin´ci´c, T. Fris´ci´c, and W. Jones, Chem. Eur. J., 2008, 14, 747. 111. R. Liantonio, S. Luzzati, P. Metrangolo, et al., Tetrahedron, 2002, 58, 4023. 112. R. D. Bailey, L. L. Hook, R. P. Watson, et al., Cryst. Eng., 2000, 3, 155.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

18

Supramolecular materials chemistry

113. R. B. Walsh, C. W. Padgett, P. Metrangolo, et al., Cryst. Growth Des., 2001, 1, 165.

136. H. Loc Nguyen, P. Horton, M. Hursthouse, et al., J. Am. Chem. Soc., 2004, 126, 16.

114. A. B. C. Lucassen, M. Vartanian, G. Leitus, and M. E. van der Boom, Cryst. Growth Des., 2005, 5, 1671.

137. P. Metrangolo, C. Pr¨asang, G. Resnati, et al., Chem. Commun., 2006, 3290.

115. R. Liantonio, P. Metrangolo, T. Pilati, and G. Resnati, Acta Crystallogr. Sect. E, 2002, 58, 575.

138. D. W. Bruce, P. Metrangolo, F. Meyer, et al., New J. Chem., 2008, 32, 477.

116. P. Metrangolo, F. Meyer, T. Pilati, et al., Acta Crystallogr. Sect. E, 2007, 63, 4243.

139. R.-Q. Zhou, D.-X. Liu, D.-L. Cao, et al., Huozhayao Xuebao, 2007, 30, 49. CODEN: HUXUFP ISSN:1007–7812.

117. S. C. Blackstock, J. P. Lorand, and J. K. Kochi, J. Org. Chem., 1987, 52, 1451.

140. M. Zaworotko, H. Clarke, A. Kapildev, et al., PCT Int. Appl., (2008), 155. CODEN: PIXXD2 WO 2008153945 A2 20081218 CAN 150 : 63966 AN 2008 : 1503326 CAPLUS.

118. P. Metrangolo, F. Meyer, T. Pilati, et al., Chem. Commun., 2004, 1492. 119. P. Cardillo, E. Corradi, A. Lunghi, et al., Tetrahedron, 2000, 56, 5535. 120. G. Marras, P. Metrangolo, F. Meyer, et al., New J. Chem., 2006, 30, 1397. 121. M. Vartanian, A. C. B. Lucassen, L. J. W. Shimon, and M. E. van der Boom, Cryst. Growth Des., 2008, 8, 786. 122. P. Metrangolo, F. Meyer, T. Pilati, et al., Chem. Eur. J., 2007, 13, 5765. 123. S. R. Batten and R. Robson, Angew. Chem., 1998, 110, 1558. 124. S. R. Batten and R. Robson, Angew. Chem. Int. Ed., 1998, 37, 1460. 125. R. Liantonio, P. Metrangolo, F. Meyer, et al., Chem. Commun., 2006, 1819. 126. C. B. Aaker¨oy, M. Fasulo, N. Schultheiss, et al., J. Am. Chem. Soc., 2007, 129, 13772. 127. C. B. Aaker¨oy, J. Desper, B. A. Helfrich, et al., Chem. Commun., 2007, 4236. 128. C. B. Aaker¨oy, J., Desper, M. Fasulo, et al., CrystEngComm, 2008, 10, 1816. 129. C. B. Aaker¨oy, N. C. Schultheiss, A. Rajbanshi, et al., Cryst. Growth Des., 2009, 9, 432. 130. N. S. Goroff, S. Curtis, J. A. Webb, et al., Org. Lett., 2005, 7, 1891. 131. C. Wilhelm, S. A. Boyd, S. Chawda, et al., J. Am. Chem. Soc., 2008, 130, 4415. 132. C. Bosshard, F. Pan, M.-S. Wong, et al., Chem. Phys., 1999, 245, 377. 133. K.-S. Huang, D. Britton, M. C. Etter, and S. R. Byrn, J. Mater. Chem., 1997, 7, 713. 134. P. Y. Lee, D. G. Hamilton, E. A. McGehee, and K. A. McMenimen, J. Am. Chem. Soc., 2003, 125, 10586. 135. C. Dai, P. Nguyen, T. Marder, et al., Chem. Commun., 1999, 24, 2493.

141. N. Schultheiss, S. Bethune, and J.-O. Henck, CrystEngComm, 2010, 12, 2436. 142. A. N. Sokolov, T. Friscic, and L. MacGillivray, J. Am. Chem. Soc., 2006, 128, 2806. 143. N. Shan and M. Zaworotko, Drug Discov. Today, 2008, 13, 440. 144. W. Jones, W. D. S. Motherwell, and A. V. Trask, MRS Bull., 2006, 31, 875. 145. N. Shan and M. J. Zaworotko, Drug Disc. Today, 2008, 13, 440. 146. N. Schultheiss and A. Newman, Cryst. Growth Des., 2009, 9, 2950. 147. M. Wenger and J. Bernstein, Cryst. Growth Des., 2008, 8, 1595. 148. S. Basavoju, D. Bostr¨om, and S. P. Velaga, Cryst. Growth Des., 2006, 6, 2699. 149. N. Takata, K. Shiraki, R. Takano, et al., Cryst. Growth Des., 2008, 8, 3032. 150. S. Mahapatra, T. S. Thakur, S. Joseph, et al., Cryst. Growth Des., 2010, 10, 3191. 151. B. K. Saha, R. Banerjee, A. Nangia, and G. R. Desiraju, Acta Crystallogr., Sect. E, 2006, 62, 2283. 152. S. L. Childs, P. A. Wood, N. Rodriguez-Hornedo, et al., Cryst. Growth Des., 2009, 9, 1869. 153. D. J. Good and N. Rodriguez-Hornedo, Cryst. Growth Des., 2009, 9, 2252. 154. N. Rodr´ıguez-Hornedo, S. J. Nehm, K. F. Seefeldt, et al., Mol. Pharm., 2006, 3, 362. 155. A. Jayasankar, D. J. Good, and N. Rodr´ıguez-Hornedo. Mol. Pharm., 2007, 4, 360. 156. S. J. Nehm, B. Rodr´ıguez-Spong, and N. Rodr´ıguezHornedo, Cryst. Growth Des., 2006, 6, 592. 157. M. L. Cheney, N. Shan, E. R. Healey, et al., Cryst. Growth Des., 2010, 10, 394.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc113

Polymorphism: Fundamentals and Applications Srinivasulu Aitipamula1 and Ashwini Nangia2 1 2

Institute of Chemical and Engineering Sciences, Jurong Island, Singapore University of Hyderabad, Hyderabad, India

1 Introduction 2 Polymorphism 3 Kinetic and Thermodynamic Aspects 4 Screening and Characterization 5 Structure–Property Correlation 6 Multicomponent Crystals 7 Growing New Polymorphs 8 Conclusions and Future Thoughts Acknowledgments Related Articles References

1 1.1

1 3 5 9 12 13 14 16 16 16 16

INTRODUCTION How crystal structure affects physical and chemical properties of solids

interactions, and van der Waals interactions. The stability and bulk properties of a solid material depend on the arrangement of the molecules in the crystal lattice and also on the way in which they interact with neighboring molecules. Therefore, a complete understanding of the relationship between the solid-state properties and the crystal structure is of utmost importance in research on solid-state supramolecular materials. One of the most reliable ways to understand the internal structure of matter is by determining the crystal structure and analyzing it in terms of molecular arrangement and intermolecular interactions. In the case of polymorphic structures, which represent different crystal structures of the same compound, crystal structure analysis is a unique opportunity to study structure–property relationships of the same molecular species organized in different supramolecular environments.2 For example, differences in the crystal structures of paracetamol polymorphs have been correlated with physicochemical changes in tableting and mechanical properties.3

1.2 Many organic, metal–organic, and inorganic compounds can exist in different forms in the solid state, which can be crystalline or amorphous. Whereas crystalline solids represent a periodic arrangement of molecules in three dimensions, amorphous materials lack long-range order and are less stable than the crystalline forms.1 Crystalline materials are built up by a regular arrangement of molecules connected by several types of intermolecular interactions, such as hydrogen bonds, ionic or electrostatic interactions, metal–ligand bonds, heteroatom and halogen-based

Different crystalline forms: polymorphs, hydrates, solvates, salts, and co-crystals

The most common crystalline forms are polymorphs, hydrates, and solvates (pseudopolymorphs). Polymorphs are formed when a substance crystallizes in two or more crystal structures. Polymorphism significantly impacts on physicochemical properties of materials, such as stability, density, melting point, solubility, bioavailability, and so on. Hence the characterization of all possible polymorphs, identifying the stable (thermodynamic) polymorph, and design of reliable processes for consistent production are critical in modern day drug development. The incorporation of solvent molecules into the crystal lattice produces new crystalline forms called solvates or

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

2

Supramolecular materials chemistry

pseudopolymorphs.4 If the included solvent is water, they are termed as hydrates. Approximately one-third of active pharmaceutical ingredients (APIs) are capable of forming crystalline hydrates.5 The presence of a solvent or water molecule in the crystal lattice imparts unique properties to these materials. Pharmaceutical compounds are exposed to solvents during various processes and manufacturing, and hence there is a strong possibility that these compounds will form hydrates/solvates. Apart from having different physical and chemical properties, solvates offer another route to control and study polymorphism, for example, desolvation resulting in a new polymorph.6 Salt formation of acidic and basic molecules is a wellknown step in drug development to enhance the physicochemical properties of drugs: for example, water solubility, dissolution rate, bioavailability, chemical stability, compatibility with excipients, filtration, tableting, and so on. It is estimated that at least one-half of the marketed APIs are administered as salts.7 In general, salts are formed by proton transfer from an acid to the conjugate base. HCl salts are most commonly used for basic drugs. The presence of a particular counterion in the crystal lattice imparts unique properties to the drug. Thus the properties of APIs can be manipulated and optimized by selecting appropriate salt formers. Co-crystal structures were reported in the 1990s,8, 9 but they were referred to as molecular complexes or molecular compounds in the older literature.10 Their potential utility in making new medicines, the so-called pharmaceutical cocrystals, has been realized only recently.11 Pharmaceutical co-crystals belong to multicomponent solids that are molecular complexes of two or more compounds which are solids under ambient conditions.11 From a preparative viewpoint, pharmaceutical co-crystals are made from APIs and a pharmaceutically acceptable co-crystal former, preferably from the GRAS (generally regarded as safe) list of the US Food and Drug Administration (USFDA).12 This new class of materials is remarkably impacting pharmaceutical research by broadening the intrinsic values of APIs by modification of their physicochemical properties to create new intellectual property.13 A major advantage of co-crystals compared to salts is that co-crystallization can be implemented even for drugs containing nonionizable functional groups. Different solid forms of a chemical substance, more specifically in relation to pharmaceuticals,14 are schematically shown in Figure 1.

1.3

Polymorphism in materials science and pharmaceutical development

The bulk properties of any material depend on the internal structure of the solid. Therefore, polymorphic

Amorphous

Molecule Solvent molecule Protonated molecule

Polymorphs

Deprotonated acid Nonvolatile molecule

Solvate

Salt

Multicomponent solids

Cocrystal

Figure 1 Various solid-state forms of a chemical substance. (Reproduced from Ref. 14.  Wiley-VCH, 2006.)

structures with different internal structures would likely impact the properties of materials. Brillante et al.15 have reported four polymorphs of an organic semiconductor, dibenzotetrathiafulvalene, and studied single-crystal growth and phase transformations among the different phases. Polymorphism has direct impact on the properties of nonlinear optical materials; for example, a new second-harmonic generation (SHG) active polymorph of p-nitroaniline was crystallized from the molten phase using polymers.16 Polymorphism is ubiquitous in the pharmaceutical industry. The approval of a new drug substance by the USFDA requires that all data on polymorphs of the lead molecule be documented. In general, a given substance can have any number of polymorphs but, under given experimental conditions, only one polymorph will be stable and all other polymorphs will be metastable. Since the presence of a metastable polymorph during processing often leads to phase transformation to the more stable polymorph, it is desirable to choose the thermodynamic polymorph for drug development. Polymorphism in dyes and pigments is a classic phenomenon. The performance of a pigment is generally measured by its tinctorial strength which can be strongly influenced by particle size distribution, morphology, and structure of the pigment.4 Since different polymorphs can have different morphologies and particle size, the properties of polymorphic forms of pigments are different. Some wellknown polymorphic pigments are the trimorphs of copper phthalocyanine and quinacridone. The α-polymorph of quinacridone is less stable, whereas β- and γ -quinacridone are important for industrial applications.17

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications

1.4

Intellectual property and patenting of polymorphs

Patenting all the solid forms of an API is a rigorous exercise in pharmaceutical companies for safeguarding their intellectual property. In layman terms, patenting offers the right to exclude others from exploiting a patented invention, and affords an economic advantage to the inventor. Polymorphs offer new intellectual property opportunities in pharmaceuticals. The efforts to file multiple patents on various solid forms of an API are intended to protect them from generic competitors. A patentable invention must fulfill three requirements: novelty, utility, and nonobviousness. Crystalline polymorphs represent solid materials of a substance with different crystal structures. Therefore, any new form (polymorph) of an API is a new material. Its structure is nonobvious because it is difficult to predict how many polymorphic forms a compound can have, especially in terms of the details of their molecular packing. Most importantly, polymorphs have a utility because of their different physical and chemical properties. For example, the first patent on the histamine H2 receptor, ranitidine hydrochloride (RHCl) was filed by Glaxo SmithKline (GSK) in 1978 (US Patent No. 4 128 658), and two more patents followed, based on the discovery of a novel polymorph (Form 2) of the same drug in 1985 (US Patent No. 4 521 431) and 1987 (US Patent No. 4 672 133). It is claimed that Form 2 has favorable filtration and drying characteristics. These characteristics were attributed to the ability of Form 2 to crystallize with a needle-like habit, as opposed to the platelike crystals of Form 1. The central point in patent litigation between GlaxoSmithKline and Novopharm was about the polymorphs of RHCl. Several other patent litigations involving APIs such as cefadroxil, terazosin hydrochloride, paroxetine hydrochloride, aspartame, and so on involved polymorphism in drug development.

1.5

Frequency of polymorphism in drugs and pharmaceuticals

Polymorphism is possible for any compound, but the conditions under which these polymorphs are stable may be difficult to know a priori. McCrone stated almost half a century ago that “the number of forms known for a given compound is proportional to the time and money spent in research on that compound”18 and this statement is turning out to be truer than ever. Compounds formerly believed to exist in a single form are yielding a second or even a third polymorph. For example, novel polymorphs of aspirin,19 maleic acid,20 and 1,3,5-trinitrobenzene,21 have been found after several decades of their first crystal structure being reported. Polymorphism is not as widespread in terms

3

of overall crystal structures being solved and reported. According to the Cambridge Structural Database (CSD), the numbers are 4–5% for organic compounds, 5–6% for organometallics, and 2% for coordination compounds.22 Polymorphism is more widespread in pharmaceutical solids, with more than 50% of drug molecules estimated to be polymorphic.23 However, it should be mentioned here that the drug molecules are often not archived in the CSD for proprietary reasons, and therefore these statistics vastly underrepresent the true prevalence of polymorphism in pharmaceuticals. The greater occurrence of polymorphism in drugs could also be due to the exhaustive experimental efforts in studying these molecules.

2 2.1

POLYMORPHISM Definition

A widely accepted definition of polymorphism is that proposed by McCrone as “a solid crystalline phase of a given compound resulting from the possibility of at least two different arrangements of the molecules of that compound in the solid state.”18 The definition of polymorphism is still elusive and there are several other definitions. For example, Buerger stated that two polymorphs are different forms of the same chemical compound which have distinctive properties.24 Even though there are slight differences in these definitions, they all emphasize the different packing arrangements of the same molecule in the solid sate, and the different crystal structures leading to differences in physical and chemical properties.

2.2

Early examples

Polymorphism in elements is generally referred to as “allotropism,” the phenomenon of an element existing in two or more physical forms. The importance of allotropism in elements such as carbon, phosphorous, and sulfur can be easily understood by the differences in their physical properties. For example, diamond is the hardest natural mineral and it is also an insulator, whereas graphite is a soft material and an electrical conductor (Figure 2). The identification of polymorphism in inorganic materials dates back to the eighteenth century. Some of the early examples include arsenate/phosphate salts, NaH2 PO4 ·H2 O/NaH2 AsO4 ·H2 O and Na2 HPO4 ·H2 O/Na2 HAsO4 ·H2 O,25 sodium beryllium fluoride (Na2 BeF4 ),26 ammonium paratungstate (NH4 )10 W12 O41 ·H2 O,27 and so on. Organic compounds are susceptible to forming various intermolecular interactions and hence there is an increased chance of polymorphism. Benzamide is one of the earliest

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

4

Supramolecular materials chemistry

O

N+

O− H N

CN

O

HO

H N

Cl

S a0

CH3 ROY

a0 = 0.357 nm

a0

Tolfenamic acid O

Cl

O

O S O

N H O

NH2 OH DCH

Furosemide

Figure 3 Some common compounds that exhibit conformational polymorphism.

Figure 2 Carbon allotropes graphite (left) and diamond (right). (Reproduced with permission from sci.waikato.ac.nz/farm/ images/graphite.jpg, University of Exeter, and Getty Images, Singapore.) (a)

examples of a polymorphic organic compound.28 It forms two polymorphs from the same crystallization batch: that is, block shaped and featherlike needles.

2.3

Classification of polymorphs

Polymorphs may be classified into different types. The advantage with such classification is that the differences between alternative crystal structures are easily identified and this helps us to understand the origins of polymorphism. However, it should be mentioned that these classifications are not rigid, because polymorphs may belong to one or more of them.

2.3.1 Conformational polymorphs Conformational polymorphism refers to the occurrence of different molecular conformations in different polymorphs.29 Flexible molecules with several degrees of freedom and low-energy conformers are likely to adopt different conformations leading to conformational polymorphs. Many APIs are conformationally flexible and possess rotatable C–C, C–N, and C–O bonds, and conformational polymorphism in these compounds can be observed (Figure 3). 5-Methyl-2-[(2-nitrophenyl)amino]-3thiophenecarbonitrile, which is commonly known as ROY for its red, orange, and yellow crystals, holds the current record for the highest number of polymorphs (i.e., seven)

(b)

Figure 4 Overlay of conformations in polymorphs of (a) ROY and (b) DCH.

to be characterized by single-crystal X-ray diffraction (SCXRD).30 Interestingly, different conformations of ROY are present in these polymorphs (Figure 4). Various other classic examples of conformational polymorphs include the common drug furosemide (or Lasix), tolfenamic acid, 4,4diphenyl-2,5-cyclohexadienone (DCH), and so on.

2.3.2 Hydrogen-bond or synthon polymorphism Synthon polymorphs are polymorphic structures that differ in their hydrogen bonding.31 For example, tetrolic acid molecules assemble via the acid–acid dimer synthon in the α form, whereas the same carboxylic acids form a catemer synthon in the β form (Figure 5).32 Well-known polymorphic compounds such as sulfathiazole, temozolomide, furosemide, and isonicotinamide all show differences in their hydrogen-bond synthons.33 The analysis of polymorphic crystal structures as synthon polymorphs facilitates the understanding of differences in hydrogen bonding and intermolecular interactions leading to different crystal structures. The potential of synthon classification for polymorphs goes beyond structural analysis to an understanding and control of crystallization.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications

(a)

5

(b)

Figure 5 Molecular packing diagrams of (a) α and (b) β forms of tetrolic acid which are carboxylic acid dimer and catemer synthon polymorphs. CH3 CH3 O S

H3C O H3C

N

Figure 6

N N H

O

H3C O

CH3 O S

N N

H3C N

H

O CH3

Two tautomers of omeprazole. (a)

2.3.3 Configurational polymorphism When different configurations of the same molecule crystallize in separate crystalline forms, they are termed as configurational polymorphs. As a caveat, since configurational isomers can be considered to be separate chemical entities, these solids may not be strictly classified as polymorphs. Yet another classification of polymorphs that is closely related to configurational polymorphism is tautomeric polymorphism which occurs when different tautomers of a compound exist in different polymorphs. There are cases where configurational isomers or tautomers can interconvert in solution, and the fact that their relative population and stability depends on factors such as temperature, solvent, pH, and so on makes them appear as different crystalline polymorphs. In contrast to other, previously described classes of polymorphs, tautomeric polymorphs are rare. An early report of tautomeric polymorphism by Desiraju is the compound 2-amino-3-hydroxy-6-phenylazopyridine, which exists as hydroxyazo and quinonehydrazone crystals of different colors.34 Polymorphs of the antiulcer drug omeprazole (Figure 6) also represent tautomeric polymorphs that were characterized by SC-XRD.35

Figure 7 Crystal structure of pyrazine-N,N
-dioxide polymorphs. The structure-building C–H · · · O trimer unit is planar in the orthorhombic form (a) but helical in the monoclinic form (b). (Reproduced from Ref. 37.  Royal Society of Chemistry, 2007.)

Z
> 1. Rapid crystallization techniques such as sublimation and melt crystallization are expected to produce polymorphs with high Z
,36 and this also suggests that slow crystallization experiments will result in crystal structures of Z
0.5 or 1. The occurrence of shorter C–H· · ·O (and O–H· · ·O) interactions in the high Z
crystal structures of polymorphs that make a variable Z
pair was analyzed in the CSD. In general, the high Z
structure has shorter C–H· · ·O interactions and O–H· · ·O hydrogen bonds when compared to the corresponding low Z
polymorph.37 For example, the orthorhombic polymorph of pyrazine-N ,N
-dioxide (Z
= 2) has shorter C–H· · ·O ˚ compared to the Z
= 1 interactions (shortest = 2.01 A) ˚ (Figure 7). monoclinic polymorph (shortest = 2.12 A)

3 2.3.4 Polymorphism due to variable Z
The number of symmetry-independent molecules in the crystallographic asymmetric unit cell (Z
) has some relevance in the context of organic polymorphs. High Z
structures have been variously referred to as “snapshot pictures” and “fossil relics” of early events in crystallization. About 12% of crystal structures in the CSD have

(b)

KINETIC AND THERMODYNAMIC ASPECTS

Understanding the origin of polymorphism requires a thorough knowledge of thermodynamic and kinetic factors that affect the crystallization process. In general, it is the free energy of polymorphs that determines their stability relationships. At a given temperature and pressure, there is only one polymorphic form that is stable and all other

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

6

Supramolecular materials chemistry

polymorphs are metastable. The thermodynamically stable phase will have the lowest free energy. Metastable polymorphs are unstable with respect to the thermodynamic polymorph, but are stable below the transition point. The importance of stability issues of polymorphs in drug development are best highlighted by the well-known Ritonavir incident at Abbott Laboratories.38

3.1

The phase rule and phase diagrams

As different polymorphs of a given substance represent different phases under a given set of experimental conditions, the well-known Gibbs’s phase rule (1)39 is applicable to polymorphic systems F =C−P +2

(1)

where F is the number of degrees of freedom of the system, C is the number of components, and P is the number of phases that exist in equilibrium. C = 1 for polymorphs because there is only one component in polymorphic systems. For a dimorphic system, P = 2 when both the polymorphs are in equilibrium, which means that one of the parameters can be varied. This implies that at constant pressure, the temperature at which the two polymorphs coexist can vary, which is defined as the transition point. The phase rule also implies that, since the system cannot have a negative number of degrees of freedom (F < 1), only a maximum of three polymorphs can coexist in equilibrium, and the set of conditions under which this occurs is defined as the triple point. The application of the phase rule to a dimorphic system suggests that, at a given pressure, phase transition from one polymorph to the other may occur by changing the temperature (F = 1). If such a phase transition is reversible, the two polymorphs are said to be enantiotropes and the energy of transition on heating is endothermic. When the phase transition is irreversible, the two polymorphs are termed as monotropes, in which case only one form is stable whatever the temperature, and the transformation of the metastable form to the stable one is exothermic.

3.2

Thermodynamic relationships in polymorphs

Knowledge of the information pertaining to thermodynamic relationships among polymorphs of a given substance is crucial for the successful development of that substance as a drug material. The most stable polymorph will have the lowest free energy. At constant temperature and pressure, the free energy of a solid phase can be represented by the Helmholtz energy (2)

A = E −TS

(2)

where A is the Helmholtz free energy, E is the internal energy, T is the absolute temperature, and S is the entropy. According to the above equation, at absolute zero the term TS is equals to zero and the Helmholtz energy equals the internal energy. This means that the most stable polymorphic form should have the lowest internal energy. On the basis of this equation, the thermodynamic stability between polymorphs is conveniently represented in schematic energy–temperature diagrams (E –T diagrams).40 For practical reasons, such diagrams are drawn on the basis of the Gibbs free energy rather than the Helmholtz free energy, because the data required to produce the diagrams are readily accessible experimentally. Therefore, for a polymorphic system, the Gibbs free energy relationship can be defined as G = H −TS

(3)

where H is the enthalpy and G is the Gibbs free energy. Accordingly, at absolute zero the most stable polymorph will have the lowest Gibbs free energy. Schematic E–T diagrams for enantiotropic and monotropic polymorphs are shown in Figure 8. Figure 8(a) represents the E–T diagram for an enantiotropic system. In general, two polymorphs are related enantiotropically when there is a transition point at which the two polymorphs can undergo reversible solid–solid phase transformation. Furthermore, the transition point should be below the melting point of both polymorphs. The free energy curves of both the polymorphs intersect at the transition point Tp , when Form I transforms to Form II. The E–T diagram for monotropic polymorphs is represented in Figure 8(b). In this case, there is no transition point below the melting points of the two polymorphs, and the free energy curves do not intersect. This means that one form (Form I) is always stable below the melting point of both polymorphs. It is also apparent from the diagram that Form II can undergo a spontaneous exothermic transformation into Form I, and this is thermodynamically feasible at any temperature, because GI < GII at all temperatures. A series of rules have been formulated for understanding the relative thermodynamic stabilities of polymorphs. These rules also help to determine whether a polymorphic system belongs to the monotropic or the enantiotropic category.4 Tammann was the first to develop these rules in the 1920s,41 and they were later extended by Burger and Ramberger who applied these rules to several polymorphic systems.42, 43

3.2.1 Heat of transition rule The heat of transition rule states that, if an endothermic phase transition is observed at a particular temperature, the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

II

∆H f,II ∆Hf, I

HIiq H II HI

Energy

∆Hf , I ∆Hf,II

∆Ht,I/II

Energy

Polymorphism: fundamentals and applications

HIiq HII HI

II I

I

GII GI

GI GII G Iiq −273 (a)

7

Tp,I/II

mpI mpII

Temperature (°C)

GIiq −273 (b)

mpII mpI Temperature (°C)

Figure 8 Energy versus temperature (E–T ) diagram of a dimorphic system: (a) enantiotropic, (b) monotropic. G: Gibbs free energy, H : enthalpy, liq: liquid phase, and Tp : transition point of two polymorphs. (Reproduced from Ref. 40.  Elsevier, 1996.)

thermodynamic transition point lies below that temperature, and the two polymorphs are said to be enantiotropically related. Two polymorphs are related monotropically if an exothermic phase transition is observed at a particular temperature, and there is no thermodynamic transition point below this temperature. The same situation can also occur when the two polymorphs are enantiotropically related and, in addition, have a thermodynamic transition temperature that is higher than the experimentally observed transition temperature. As the endothermic and exothermic phase transitions are easily determined by differential scanning calorimetry (DSC), the heat of transition rule for polymorphs is frequently applied to establish the relationship between polymorphs. Burger and Ramberger42, 43 found that this rule is valid in 99% of cases examined.

3.2.2 Heat of fusion rule According to this rule, if the higher melting polymorph has the lower heat of fusion, then the two polymorphs are enantiotropic; if the higher melting polymorph has the higher heat of fusion, the two polymorphs are related monotropically.40 The application of this rule is based on the assumption that the heat of transition can be approximated by the difference between the heats of fusion of the polymorphs. When the melting points of two polymorphs are not too close (m.p. ∼30 ◦ C), then Burger and Ramberger42, 43 suggest the entropy of fusion rule instead.

3.2.3 Entropy of fusion rule As the entropy of fusion can be measured by DSC, the enthalpy of fusion and the melting point of a crystal can be obtained in one experiment (Sf = Hf /Tf ). The entropy

of fusion rule states that if the modification with the higher melting point has the lower entropy of fusion, the two forms are enantiotropic. The monotropic relationship arises when the lower melting form shows a lower entropy of fusion.43

3.2.4 Density rule The formulation of this rule was based on Kitaigorodskii’s44 close packing principle which states that, at absolute zero for nonhydrogen-bonded system, the most stable polymorph will have the highest density because of stronger intermolecular and van der Waals interactions. This implies that the form with the greatest density is thermodynamically stable at absolute zero. The density rule states that two polymorphs are related monotropically if the higher melting form has the higher density; otherwise, they are enantiotropically related. It is apparent that exceptions to the density rule are possible, because energetically favorable hydrogen bonds may overcome the packing efficiency, and also compensate for the loss of van der Waals energy and thus stabilize the polymorph having a lower density. Hydroquinone, ritonavir, and acetazolamide are some of the well-known examples in which the more stable polymorph has lower density.

3.2.5 Infrared rule In general, the formation of strong hydrogen bonds results in a reduction in entropy and an increase in the frequency of the vibrational modes of the same hydrogen bonds. For polymorphic structures that contain strong hydrogen bonds, the infrared rule states that the polymorph with the higher bond-stretching frequency may be assumed to have the greater entropy and should belong to the most stable

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

8

Supramolecular materials chemistry

form. One of the problems associated with the application of this rule is that, if a molecule forms more than one type of hydrogen bond, then it is difficult to correlate the frequencies observed with a particular polymorphic form. In such cases, other rules described thus far help to understand the relative stabilities of polymorphs. As there are exceptions to one or more of these rules, it is important to caution that the relative stabilities of polymorphs should be estimated by applying more than one of the above rules. A combined application of these rules provides valuable data to construct qualitative E–T diagrams, which in turn give information about the operating regions to obtain a particular polymorph.

3.3

Ostwald’s rule/law of stages

As mentioned in the previous sections, at a given set of experimental conditions only one polymorph is stable and all others can be considered as metastable. This means that metastable polymorphs can convert to the more stable polymorph under the given conditions. Such behavior was highlighted by Ostwald in his famous rule (also known as Ostwald’s law of stages or OLS), which states that “when leaving a metastable state, a given chemical system does not seek out the most stable state, rather the nearest metastable one that can be reached without loss of free energy.”45 This means that, during the crystallization process, a series of polymorphs may be obtained, and the polymorph that crystallizes at the end is the most stable polymorph. OLS is generally valid for all polymorphic systems. However, if the nucleation rates and growth rates of stable and metastable polymorphs are equal, then they tend to crystallize together (a case of concomitant polymorphism).46 In such a scenario, OLS cannot be applied but the metastable polymorphs that appear concomitantly with other polymorphs can convert to the stable polymorph under different experimental conditions.

3.4

the ratio of the concentration of a solution to that of the saturated solution. In general, as the degree of supersaturation increases at a given temperature, the rates of nucleation and crystal growth increase. The implications of this theory are that, by controlling the supersaturation at a given temperature, a particular polymorph can be grown selectively. For example, in a study on phenylbutazone polymorphs, Datta and Grant showed that the polymorph that crystallizes from a methanolic solution of phenylbutazone depends on the initial degree of supersaturation of the solution.47 Whereas the relative nucleation rates of the polymorphs determine the polymorphic outcome at low supersaturation, relative crystal growth of the polymorphs is a determining factor at high superstaturation. The effect of temperature has both thermodynamic and kinetic implications and significantly influences the solubilities of polymorphs.48 In general, crystallization at higher temperatures may produce one polymorph, whereas crystallization at lower temperatures will produce a second polymorph. Polymorphic phase transformations occur with respect to changes in temperature. For example, a less stable polymorph of an enantiotropic polymorphic system can convert to the more stable polymorph upon heating. Therefore, it is of fundamental relevance in the pharmaceutical industry to control the temperature to avoid undesirable phase transformations among polymorphic forms during drug development. Organic molecular crystals are sustained by a wide range of intermolecular interactions, such as ionic, hydrogenbonding, and weak van der Waals interactions, and these interactions are flexible enough to undergo changes upon the application of high pressure. Pressure-induced phase transformations49 of a number of polymorphic systems have provided valuable insights into structure–property relationships. From a thermodynamic perspective, the lowpressure polymorph is generally less dense than the highpressure one, and this means that the application of high pressure to metastable polymorphs can induce phase transformation to a more stable phase.

Effect of operational conditions on polymorphism 3.5

The susceptible nature of metastable polymorphs to undergo phase transformation to the most stable polymorph warrants further in-depth studies concerning the effect of various parameters on the polymorphic outcome of crystallization experiments. For example, it is of great importance to study the effect of parameters such as supersaturation, temperature, pressure, ultrasonication, and so on. Supersaturation is an important parameter which drives crystallization and influences the kinetics of crystal nucleation and growth, and thus determines the outcome of crystallization. The degree of supersaturation can be defined as

Kinetics, crystal nucleation, and crystal growth

Crystallization is a key experimental technique for the purification and separation of organic compounds. Crystallization may be viewed as the aggregation of close to a mole of molecules into energetically favorable packing motifs through intermolecular interactions in an ordered, periodic, and infinite lattice. However, all crystal structures are not necessarily at the global free energy minimum. The result of crystallization generally depends on kinetic and thermodynamic factors. Slow crystallization methods such as natural

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications evaporation from solution, slow cooling of a crystallization batch, slow crystallization from the melt, and slow sublimation refer to thermodynamic conditions. On the other hand, rapid crystallization yields kinetic polymorphs. The crystallization of a particular polymorph is usually determined by both kinetic and thermodynamic factors. Therefore, it is important to identify the conditions under which polymorphs are obtained and stable. For simplicity, the process of crystallization can be divided into two steps: crystal nucleation and crystal growth.50 The process of forming crystals from a supersaturated solution begins with the formation of nuclei by the association of molecules into aggregates. This can be termed primary nucleation. Once the primary nuclei are formed, they continue to grow by the association of additional molecules through a process known as secondary nucleation, which affects the overall crystallization process. Secondary nucleation is influenced by various factors such as agitation, temperature, and concentration gradient. In polymorphic systems, the form that crystallizes first can be considered as due to fast nucleation and hence the one that has overcome the lowest free energy barrier in accordance with Ostwald’s rule of stages. The stable thermodynamic form is generally obtained toward the end of crystallization under slow nucleation conditions. The process of nucleation starts with the formation of clusters of molecules such that nuclei of different polymorphs can form and coexist (simultaneously) in a given crystallization batch leading to a mixture of crystalline forms. However, subsequent crystallization primarily depends on the combination of the relative nucleation rates and the relative crystal growth rates of the polymorphs and, hence, the form that grows faster (faster growth rate) is obtained finally. In addition, if conditions are favorable for a metastable polymorph to grow faster than the stable polymorph, the former will be the final outcome of crystallization. In such cases, if the growth rate of the kinetic polymorph can be inhibited, say by additives or auxiliaries, then the result will be the thermodynamic polymorph.

4

SCREENING AND CHARACTERIZATION

Polymorphism has received widespread general interest in pharmaceutical drug development because of its impact on the physicochemical properties of APIs and also because of their economic significance. It is considered mandatory to perform a thorough and exhaustive solid-form screening of all possible polymorphs of a lead drug molecule at the preformulation stage. The purpose is to discover as many solid forms as possible and to identify the most suitable

9

(stable) form for further development. This section deals with the various methods used for polymorph screening and also the analytical techniques used to characterize new polymorphs.

4.1

Polymorph screening

Crystallization from solution is the most common method for polymorph screening because it generally yields single crystals for X-ray diffraction. Ideally, it is necessary to investigate a wide variety of solvents of varying polarity, boiling point, hydrogen-bond propensity, and also mixtures thereof. As discussed in previous sections, the method of crystallization influences the thermodynamic stability of the resulting polymorphs. For example, slow evaporation of a saturated solution generally yields a thermodynamic form, whereas rapid evaporation favors kinetic forms. Even though slow evaporation is time consuming and single crystals can require significant effort to grow, this method is preferred for obtaining the thermodynamic polymorph. A popular method is antisolvent crystallization in which very high supersaturation is achieved by the addition of a second solvent, that is, an antisolvent, which decreases the solubility of the solutes. Supersaturation levels can be controlled by varying the amount of antisolvent added. For example, nucleation and growth rates of L-histidine polymorphs are influenced by the composition of the solvent (mixture of water and ethanol).51 Various other factors such as antisolvent addition rate and initial concentrations of the solution also control crystallization behavior. Supercritical CO2 can also be used as an antisolvent.52 As exemplified by the polymorphic behavior of glycine,53 the pH and ionic strength of the solvent play a crucial role in the crystallization of small molecules. For example, crystallization of glycine from water–alcohol solutions in the pH range 3.8–8.9 always produced the α polymorph, whereas, when the pH was adjusted to below 3.8 or above 8.9, the γ polymorph nucleated. The preferential crystallization of these polymorphs in the given pH range was independent of the two supersaturations chosen. These studies helped to generalize the fact that the pH and solvent are two key variables determining the polymorphic outcome of a crystallization screen. Even though solution crystallization is always the method of choice, it has limited applicability in cases where the molecules have a strong tendency to form solvates. Solventfree techniques such as melt crystallization and sublimation are suitable for certain compounds. For example, Das and Barbour54 obtained four polymorphs of a hexahost, hexakis(4-cyanophenyloxy)benzene, upon melt crystallization, which otherwise forms a series of solvates with the solvent of crystallization. Sublimation is yet

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

10

Supramolecular materials chemistry

another alternative to solvent-free crystallization, which is the preferred technique to generate guest-free crystals and polymorphs. For example, a new polymorph of the antidepressant drug venlafaxine hydrochloride was obtained by sublimation under reduced pressure.55 A recent survey of crystal structures deposited in the CSD suggested that the likelihood of obtaining structures with Z
> 1 increases to 18% under solvent-free conditions at high temperature56 compared to the overall statistical frequency of 12% for Z
> 1. Therefore, chances are good of obtaining new polymorphs from solvent-free crystallization techniques. A very recent technique that utilizes fully automated robotic systems is high-throughput screening (HTS),57 which is capable of performing thousands of crystallization experiments per week with only a few grams of the API. The compounds obtained from these experiments are analyzed by X-ray powder diffraction and Raman microscopy, and then systematically classified (e.g., as polymorph, cocrystal, salt, solvate, hydrate, etc.).14 One of the key aspects of HTS is that the choice of solvent and solvent mixtures should normally be different for every substance, taking into account solubilities of the compounds used for screening, possible chemical reactions, and also solvent–solute interactions. In general, the data obtained from HTS provide valuable information on the existence of multiple forms of a given substance. Such data can be used as the starting point for a comprehensive polymorphism investigation, co-crystal formation, and scaleup of solid forms. HTS of the anticonvulsant drug carbamazepine resulted in the identification of all known polymorphs and solvates.58 No new form that is more stable at room temperature than Form III was found. However, the results demonstrate the suitability of HTS for discovering polymorphs and identifying the conditions under which they can be produced.

4.2

Characterization of polymorphs

Once there is evidence of polymorphism, it is necessary that the different forms are characterized thoroughly in order to understand the stability relationships and their interconversions. A wide range of analytical techniques is available.59 Recent advances in microscopy have provided instruments that can identify new polymorphs more precisely. Using a PC interfaced to a hot-stage polarizing microscope, thermal events can be visualized and archived and even be used to create a video file. Hot-stage microscopy (HSM) facilitates the identification of phase transformations, monotropic and enantiotropic relationships, crystalline to amorphous phase transitions, the appearance

of kinetic and thermodynamic polymorphs, sublimation, melting, chemical reaction, and morphological changes. A very small amount of the solid is required to record a video of the thermal processes. Fourier transform infrared (FT–IR) and Raman spectroscopy (400–4000 cm−1 ) are frequently used for the characterization of polymorphs. These techniques are used to identify differences in molecular conformation and hydrogen bonding in the solid state. Polymorphic structures containing strong hydrogen bonds, such as O–H· · ·O and N–H· · ·O, are easily differentiated by IR spectroscopy, and the complementary Raman technique can be used to distinguish conformational polymorphs. Near infrared (NIR) spectroscopy (4000–12 000 cm−1 ) is able to distinguish hydration states by means of the diagnostic overtone bands at 5150 and 6900 cm−1 . A major advantage of Raman and NIR is that they can be used to analyze tablets and capsules intact without any sample preparation. Solid-state 13 C NMR spectroscopy (SS-NMR) is a powerful and sensitive technique for polymorph characterization and differentiation.60 Since different polymorphs have different crystal structures, the chemical environment of at least a few of the atoms will differ from one structure to another. This technique also provides valuable crystallographic information, such as the number of crystallographically independent molecules in the crystal structure due to doubled peaks for the same C atom. In addition to 13 C NMR, 15 N CP-MAS NMR (cross-polarized magicangle spinning) is useful for characterizing polymorphic systems, for example, polymorphs of sulfathiazole, those containing N-heterocycles, and neutral co-crystal versus salt formation.61 DSC, thermogravimetric analysis (TGA), and HSM are some of the most commonly used thermal methods for solid-state analysis. Solvent loss from the crystal, melt crystallization, sublimation, and solvent-mediated phase transformations are easily followed by HSM, which gives qualitative information on polymorphic behavior, whereas DSC/TGA are used to quantify polymorphic stability, energies of phase transitions, and monotropic or enantiotropic relationships. Information from DSC such as enthalpy of fusion and melting behavior can be used to understand thermodynamic relationships by applying Burger’s rules. Polymorphic phase transformations and melting behavior of various polymorphs of the antidiabetic drug tolbutamide62 are shown in Figure 9. The thermogram of Form IL shows two peaks: a small endotherm at 40 ◦ C, followed by another at 128 ◦ C. The first peak was ascribed to a kinetically reversible polymorphic transition to Form IH , and the second endotherm is due to melting of Form IH . In the case of Form II, a sharp endothermic melting peak at 117 ◦ C was observed, followed by an exotherm for the recrystallization of Form IH from the melt of Form II. An endotherm at

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications

11

128 °C 117 °C

Endo up

40 °C

84

89

94

99

104

Form IH

Form IL

Form II

106 °C

Form III Form IV 94 °C 30

40

50

60

70

80

90

100

110

120

130

140

Temperature (°C)

DSC of tolbutamide polymorphs I–IV. (Reproduced from Ref. 62.  Wiley-Liss, Inc, 2010.)

128 ◦ C, due to the melting of Form IH , was subsequently observed. Forms III and IV undergo an endothermic transition at 106 ◦ C and an exothermic transition at 88 ◦ C to Form IH , respectively. X-ray diffraction techniques, such as powder X-ray diffraction (PXRD) and SC-XRD, are invaluable methods for the unambiguous quantification of polymorph structure, because each polymorph gives its own characteristic diffraction pattern, which can be considered a fingerprint for that particular solid form.63 If single crystals suitable for SC-XRD are difficult to grow, PXRD is the most reliable method for distinguishing polymorphs. Hydrogen bonding and molecular packing from X-ray diffraction are used to establish structure–property relationships of polymorphs. Figure 10 shows PXRD of the polymorphs of the antibacterial drug nitrofurantoin (NF).64 With the advent of powerful charge coupled device (CCD) based X-ray diffractometers in the 1990s, it is now possible to carry out a complete structural analysis of even very minute and weakly diffracting crystals, notably metastable polymorphs. The data obtained from SC-XRD may be used to simulate the PXRD pattern of the same material. Calculated diffraction patterns are often used to judge the phase purity of the bulk polymorphic material. One of the added advantages of the diffraction technique is that experiments can be carried out at different temperatures, pressures, and humidities to identify phase stability, phase transformations, transition temperature, and hydration states. Though crystal structure analysis provides invaluable information pertaining to the organization of molecules

NF a–form NF b–form Relative intensity

Figure 9

10

20

30 2q (°)

40

50

Figure 10 Comparison of the calculated PXRD patterns of α and β polymorphs of nitrofurantoin.

in the solid state, application of this technique to all classes of compound is often not possible. Growing crystals suitable for SC-XRD of some compounds (e.g., pigments, drugs, etc.) turns out to be very difficult. Crystal structure solution from the PXRD data is one of the alternatives for solving this problem. Rietveld refinement methods65 are frequently used to obtain a reasonable structure solution using the full experimental PXRD pattern as a starting point to index the cell and obtain a structural model. Very often, a 3D structure of the same molecule in a different polymorph is used as a reference for further structure refinement.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

12

4.3

Supramolecular materials chemistry

Quantification of polymorph mixtures

Despite the best of efforts and careful experiments, two or more polymorphs are obtained concomitantly in some cases while rapid phase transformation may occur during isolation for others. Several of the techniques already discussed are useful to quantify the composition of polymorphic mixtures. Harris60 reviewed the use of SS-NMR for the characterization and quantification of polymorphic forms. The advantage of CP-MAS NMR spectroscopy is that there is no need to prepare a standard curve and this simplifies the quantification procedure. PXRD represents another important technique to quantify polymorphic forms.66 A standard calibration plot is prepared using known amounts of pure polymorphs and then an unknown mixture is estimated. This method relies on the fact that the integrated intensity of a diffraction peak is proportional to the amount of each component present, and the relative amounts of each of the polymorphs are determined by the relative intensities of the well-resolved peaks. However, since peak intensities are sensitive to particle size and vary due to orientation effects, the method often poses challenges for accurate quantification of polymorphic mixtures. One of the alternatives to overcome this problem is to use reference diffraction patterns of the known polymorphs as the standard curve for Rietveld refinement.67 In this method, the PXRD patterns of the known crystal structures are refined against the experimental PXRD pattern of the mixture to obtain the relative amounts of polymorphs. A limitation of this method is that single-crystal X-ray structures of the pure components solids are required.

5

STRUCTURE–PROPERTY CORRELATION

Polymorphism is a characteristic of the solid state and the concept is immensely important in the pharmaceutical industry. Properties such as stability, solubility, bioavailability, manufacturability, tableting, formulation, and even toxicity are a function of the polymorph crystal structure.

5.1

Stability

Drug stability is an important consideration in pharmaceutical unit operations and prolonged storage. It is possible that one polymorph is unstable/metastable (a kinetic polymorph) compared to another form of the same drug (the thermodynamic polymorph). For example, differences in the chemical and thermal/photochemical stability of polymorphs are observed for drugs such as carbamazepine,68 furosemide,69

and enalapril maleate.70 In the case of furosemide, Form I is photostable compared to Form II, which undergoes photolytic degradation to 4-chloro-5-sulphamoylanthranilic acid. It is generally true that the thermodynamically stable polymorph is more chemically stable than a metastable polymorph. This is attributed to higher packing efficiency and higher density of the stable polymorph. However, exceptions are not uncommon in polymorphic systems in which the presence of strong hydrogen bonds and better packing may result in a metastable polymorph having higher density.

5.2

Solubility and dissolution rate

Solubility and dissolution rate are two important aspects that have direct impact on the bioavailability of a drug substance. Polymorphs will have different solubilities and dissolution rates. It is generally true that the solubility of a metastable polymorph will be higher than the stable polymorph by a factor of 2–3.71 The influence of solubility on drug development can be best understood by the wellknown case of the antiretroviral drug ritonavir. This drug was first formulated and manufactured using Form I, the only polymorph of ritonavir, in 1996. It was suddenly realized after two years on the market that the metastable form converted to the thermodynamically stable form during manufacturing.38 This new polymorph had reduced bioavailability, resulting in withdrawal of the drug from the market in 1998 and Abbott had to develop a new gel capsule formulation Norvir. Differences in solubility and dissolution rate alter drug bioavailability.72 For example, the higher dissolution rate of chloramphenicol palmitate73 Form B facilitates better oral absorption of the drug while the stable Form A is biologically inactive. In carbamazepine polymorphs,74 the bioavailability order differs from that of dissolution rates. It was reasoned that conversion of the metastable polymorph to the more stable dihydrate was responsible for this behavior.

5.3

Manufacturability

The final product flowability, compressabilty, and compaction depend on the mechanical properties and morphologies of the crystalline form. The fact that these properties are different for different polymorphs suggests that the manufacturability of a drug product depends on the polymorphic state. The effect of polymorphism on tableting of ibuprofen and acetaminophen has been studied.75 It was found that crystal habit affects drug flowability. Polymorphs of paracetamol show different compressability, Form II being easier to compress as a tablet.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications

5.4

Toxicity

sodium monohydrate,79 and amiloride hydrochloride dihydrate80 are polymorphic. As in the case of single-component solids, the physical properties of polymorphic hydrates can be significantly different, as illustrated by amiloride hydrochloride dihydrate. Both of these hydrates have similar melting points, FT–IR spectra, and solubilities. However, one of the polymorphs is physically more stable than the other upon milling or compression. Database analyses have been carried out to estimate the percentage of crystal structures that contain solvent molecules in the crystal lattice. In a database study performed a decade ago, Nangia and Desiraju found that about 15% of organic crystals are solvated.81 The dioxane solvates of paracetamol significantly impact its physicochemical properties.82 The structural interrelationship and reversible phase transformations between the two polymorphs were studied. Interestingly, desolvation of either form generates the stable polymorph of paracetamol. Pharmaceutical co-crystals offer rational strategies to engineer drug properties.83 Co-crystals of the analgesic drug ethenzamide84 with 3,5-dinitrobenzoic acid, saccharin, and ethylmalonic acid are dimorphic, and a co-crystal with another analgesic drug, gentisic acid, is trimorphic (Figure 12). Polymorphs of ethenzamide and gentisic acid co-crystal show differences in dissolution rates. Interestingly, the parent API is not known to be polymorphic and it is difficult to crystallize, while co-crystallization readily gives a family of co-crystals and their polymorphs. Polymorphism is an elusive phenomenon and it is difficult to find conditions under which polymorphs can be obtained. For example, despite a thorough investigation using high-throughput crystal screening, neat grinding, and solvent-drop grinding with several solvents, there was no evidence of polymorphism in the carbamazepine : saccharin co-crystal.85 However, a novel polymorph of the co-crystal was found when functionalized cross-linked polymers were utilized as heteronuclei for crystal growth.86 Among the methods for polymorph screening of cocrystals, solvent-mediated phase transformation is an effective method to search for co-crystal polymorphs, as shown

In addition to exhibiting differences in various physicochemical properties already discussed, polymorphs also show differences in toxicity as exemplified by the anthelminitic drug mebendazole.76 This drug exists in three different polymorphs A, B, and C, which have different solubility and therapeutic effects. Form B has higher toxicity compared to the other two forms because of its very fast dissolution. Form C is the preferred polymorph for pharmaceutical drug development since its solubility and bioavailability help to circumvent the toxicity associated with Form B and the low solubility of the thermodynamic Form A. Therapeutic trials and biological activity tests using Forms A and C in the treatment of hookworm and trichuris infections showed that Form A is the least toxic against Trichinella spiralis. However, the antihelmintic activity of the drug ceases if the composition of the insoluble Form A is >30% in the polymorphic mixture.

6

MULTICOMPONENT CRYSTALS

In contrast to those on single-component crystals, studies concerning polymorphism in multicomponent crystals are scarce. Since pharmaceutical hydrates are frequently encountered during drug development, polymorphism in these solids has also been observed in several cases. For example, two polymorphs of the antibacterial drug nitrofurantoin monohydrate have been reported.64 As shown in Figure 11, water molecules play an important role in the stabilization of hydrated structures, such that there are no direct strong hydrogen bonds between NF molecules and they are bonded only via water molecules. Their hydrogenbond patterns are significantly different: the monoclinic polymorph is planar, whereas the orthorhombic form is sustained by a zigzag tape motif. Several other hydrates also exist in different polymorphic modifications. For example, fluprednisolone monohydrate,77 succinyl sulfathiazole monohydrate,78 nedocromil

(a)

13

(b)

Figure 11 Comparison of the hydrogen bonding patterns in polymorphs of nitrofurantoin monohydrate: (a) monoclinic form; (b) orthorhombic form. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

14

Supramolecular materials chemistry

Form I

Form II

Form I

COOH

O

OH NH

Form III

HO

O

S

O

Gentisic acid

CONH2 O

Form II

Saccharin

Ethenzamide COOH

O HO

O2N

O OH

NO2

3,5-dinitrobenzoic acid

Ethylmalonic acid

Form I

Form II

Figure 12

Various polymorphic co-crystals of ethenzamide.

by Horst et al.87 With the pace of ongoing research to develop pharmaceutical co-crystals, it is expected that more examples of polymorphism in co-crystals will be reported. Similar to the toxicity of the mebendazole Polymorph B,76 toxicity due to co-crystallization was investigated recently.88 Cases of renal failure in cats and infants were attributed to the melamine added to pet foods and baby formulas to increase the amino acid assay based on nitrogen content. Melamine is suspected to form a 1 : 1 co-crystal with cyanuric acid in vivo via strong N–H· · ·O hydrogen bonds, which is the cause of renal toxicity due to its complete insolubility in water. Disruption of amino acid metabolism from increased melamine content could be another reason for renal failure.89

7

Form I and Form II

GROWING NEW POLYMORPHS

As described in the previous sections, the choice of crystallization method is crucial to what form is produced. Therefore it is important to perform crystallization experiments using a variety of methods to obtain a comprehensive knowledge of the polymorphic behavior of a substance. Recently developed crystallization techniques such as laserinduced crystallization, polymer-induced heteronucleation, crystallization from supercritical fluids, capillary growth, and HTS have been shown to be useful in exploring the solid-form diversity. In this section, we describe a few crystallization techniques that are being used to grow new polymorphs.

7.1

Trial-and-error methods

First, the most common and convenient method of obtaining crystalline materials is by crystallization from solution. With a goal to finding as many crystal forms of a material as possible, a wide variety of solvents displaying different physiochemical properties should be tried until there is evidence of a new crystalline form. As the method is by trial and error, a wide variety of experiments need to be conducted by varying the parameters such as solvent polarity, evaporation or cooling rate of the solvent, and initial solution concentration. A typical polymorph screen can have as many as 100 or more solvents, combinations of these solvents, and multiple crystallization conditions to cover the wide range of polymorph space. Because of the presence of a large number of variables, discovering new polymorphs is generally a time-consuming and resourceintensive process. However, as described in Section 4.1, a recent combinatorial approach such as the HTS57 is a promising tool for performing thousands of crystallization experiments within a short time with a minimal amount of sample.

7.2

Role of solvent and additives

The choice of solvent plays an important role in the crystallization of any substance. Properties of the solvent such as hydrogen-bonding capability, polarity, dipole moment, boiling point, dielectric constant, viscosity, density, and so on influence the crystallization of polymorphs. Furthermore,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications interactions between the solvent and solute molecules are also important. Since most of the polymorphic phase transformations occur in solution, it is important to establish the role of solvent on the crystallization of polymorphs. A combination of structural information obtained from singlecrystal structure, crystal morphology, and solute–solvent and solute–solute interactions yields a rational decision tree to select a particular set of crystallization conditions, as illustrated by the dramatic enrichment in the chiral crystallization of racemic mandelic acid.90 The use of tailor-made additives91 to control crystallization of a particular polymorph is well known. In general, the additive is structurally similar to the parent molecule which can favorably interact with the molecules or be selectively adsorbed on its crystal faces, thus inhibiting the growth of the initial nuclei and favoring the crystallization of another form. For example, crystallization of 1,3,5-trinitrobenzene in the presence of a structurally related compound trisindane resulted in the isolation of two novel polymorphs of this classical molecule.21 The effect of L-, D-, and DL-aspartic and glutamic acids on the crystallization of glycine from aqueous solutions has been studied.92 It was observed that these additives inhibit the growth of the metastable α form and promote preferential crystallization of the thermodynamically stable γ form. Among drug substances, a metastable polymorph of the antibacterial drug sulfamerazine was selectively crystallized in the presence of structurally related sulfonamide drugs such as additives, for example, sulfamethazine, sulfadiazine, and N4-acetylsulfamerazine.93 It was noted that a trace amount of N4-acetylsulfamerazine as impurity prevented solutionmediated phase transformation from Form I to Form II even after suspension for two weeks in acetonitrile.

7.3

Attempted co-crystallization experiments

One of the best ways to grow new polymorphs of a given substance is by diversifying crystallization experiments. Several examples wherein attempted co-crystallization of two compounds resulted in the serendipitous crystallization of a new polymorph of one of the components have been reported in the recent literature. New polymorphs of aspirin19 and maleic acid,20 trinitrobenzene,21 and four polymorphs of benzidine94 were unexpectedly obtained during attempted co-crystallization experiments. In contrast to the well-understood role of structurally related tailor-made additives,91 the selective growth of a particular polymorph during attempted co-crystallization experiments is difficult to explain because the components have different molecular structures and complementary functional groups. However, the technique has shown early promise in the discovery

15

of new polymorphic forms. For example, attempted cocrystallization of the calcium channel blocker felodipine with isonicotinamide resulted in a metastable polymorph of the drug.95

7.4

Laser-induced crystallization

Nonphotochemical laser-induced nucleation (NPLIN) is a recent crystallization technique in which intense laser pulses induce supersaturated solutions to nucleate. It was proposed that NPLIN involves the alignment of molecules or groups of molecules in the applied optical field to organize into a crystal-like entity, thereby resulting in the growth of a particular crystal.96 NPLIN has also been demonstrated to control polymorphism in which the polymorph formed depends on the polarization state of the laser beam. For example, application of the NPLIN technique to supersaturated solutions of glycine resulted in the thermodynamically more stable γ polymorph, whereas crystallization of supersaturated solutions of glycine without any laser light produced the metastable α polymorph. The crystal structure of the γ polymorph has helical chains consisting of head-to-tail arrangements of glycine molecules, whereas the α polymorph consists of hydrogen-bonded double layers. It is supposed that NPLIN interactions between the laserinduced electric field and solute molecules align the solute molecules in a polar fashion resulting in the crystallization of the observed polymorph.

7.5

Polymer-induced heteronucleation

The effect of polymer-induced heteronucleation (PIHn) on the preferential crystallization of polymorphic forms is a powerful method to grow novel crystal structures. In this method, compounds are crystallized in the presence of a wide variety of polymers using routine crystallization techniques. The presence of a polymeric template in the solution offers a different crystallization environment which facilitates crystallization of novel polymorphs. This method was demonstrated by identifying the conditions to crystallize a metastable polymorph of acetaminophen.97 The orientation of crystal faces growing on the polymer surface varied with polymer type, which indicates stabilization of specific crystal faces by the polymers through specific interactions. The application of this technique was recently exemplified by the discovery of a novel polymorph of 6-amino-2phenylsulfonylimino-1,2-dihydropyridine,98 which was part of the second Cambridge Crystallographic Data Centre (CCDC) blind test in 2001 (molecule VI). It was found that crystallization of this compound from ethanol using

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

16

Supramolecular materials chemistry

benzyloxy-4-bromobenzene modified polystyrene as heteronuclei resulted in concomitant crystallization of a novel polymorph together with the two known polymorphs of this molecule.

7.6

Crystallization from supercritical fluids

Supercritical fluid (SCF) technology is a useful technique for the preparation of amorphous and metastable forms. SCF can be defined as a substance existing as a single fluid phase above its critical temperature and pressure, allowing it to assume the properties of both a liquid and a gas. Carbon dioxide is the most commonly used SCF. In the crystallization of polymorphic substances, the SCF is used as a solvent or an antisolvent to achieve supersaturation, thereby precipitating submicrometer-sized particles. SCF technology has been applied in the selective preparation of a particular polymorph. For example, by controlling the storage conditions (supercritical domains), pure polymorphs of salmeterol xinafoate were prepared.99 The phase purity of the resulting polymorphs was confirmed by the absence of interconversion in DSC.

7.7

Capillary growth methods

Crystallization from capillaries has evolved as a novel technique to prepare metastable polymorphs. It is believed that the small volumes of solutions produce high levels of supersaturation which promotes the crystallization of metastable polymorphs. The use of small amount of compound and the ability to analyze the crystallized materials in situ by X-ray diffraction techniques are some of the merits of this method over conventional crystallization techniques. The method has been applied to the crystallization of metastable polymorphs of nabumetone and metformin hydrochloride.100

7.8

CGOM workshop

Several new methods of crystal growth and polymorph search and form control were discussed at the recently held International Workshop on Crystal Growth of Organic Materials (CGOM).101 Crystallization in gels and ionic liquids and the mechanism for crystal growth in the presence of impurities/imposter molecules are some new methods for further reading.

8

CONCLUSIONS AND FUTURE THOUGHTS

Polymorphic forms exhibit different physical and chemical properties, which represent valuable solid-state forms

in pharmaceutical and materials science. A deeper understanding of crystallization and polymorphism in the initial stages of drug discovery is necessary because incomplete understanding of solid-state properties can lead to serious and costly setbacks in drug development. A thorough characterization of solid drug forms using multiple analytical techniques as well as the study of phase transitions between polymorphic forms is essential for pharmaceutical development and manufacturing. We have covered the fundamentals of crystallization, thermodynamic parameters, methods of crystallization, selective polymorph growth, and the application of polymorphism in pharmaceutical chemistry. As is true for any article covering a contemporary topic, there may be inadvertent omissions due to limitations of space or oversight. We urge readers to update us for a future article.

ACKNOWLEDGMENTS SA thanks the Science and Engineering Research Council of A*STAR (Agency for Science, Technology and Research), Singapore, for research funding and Prof. Reginald B. H. Tan and Dr. Pui Shan Chow of the Institute of Chemical and Engineering Sciences for encouragement and support. AN thanks the Department of Science and Technology, Council of Scientific and Industrial Research, and University Grants Commission (funded by the Government of India) for continued research grants and infrastructure support.

RELATED ARTICLES Cocrystals: Synthesis, Structure, and Applications Crystal Structure Prediction Noncovalent Interactions in Crystals X-Ray Diffraction: Addressing Structural Complexity in Supramolecular Chemistry

REFERENCES 1. S. R. Byrn, R. R. Pfeiffer, and J. G. Stowell, Solid-State Chemistry of Drugs, SSCI, West Lafayette, 1999. 2. J. Bernstein, Polymorphism in Molecular Crystals, Clarendon: Oxford, UK, 2002. 3. T. Beyer, G. M. Day, and S. L. Price, J. Am. Chem. Soc., 2001, 123, 5086. 4. D. D. MacNicol, F. Toda, and R. Bishop, Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry, Crystal Engineering, Pergamon Press, Oxford, 1996, vol. 6.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Polymorphism: fundamentals and applications 5. P. H. Stahl, The problems of drug interactions with excipients, in Towards Better Safety of Drugs and Pharmaceutical Products, ed. D. D. Braimer, Elsevier, North-Holland Biomedical, Amsterdam, 1980, pp. 265–280. 6. D. E. Braun, T. Gelbrich, V. Kahlenberg, et al., Cryst. Growth Des., 2009, 9, 1054. 7. P. H. Stahl and C. G. Wermuth, Handbook of Pharmaceutical Salt Properties, Selection and use, Wiley-VCH, Weinheim, New York, 2002.

17

33. N. J. Babu, L. S. Reddy, S. Aitipamula, and A. Nangia, Chem.—An Asian J., 2008, 3, 1122. 34. G. R. Desiraju, J. Chem. Soc., Perkin Trans., 2, 1983, 1025. 35. P. M. Bhatt and G. R. Desiraju, Chem. Commun., 2007, 2057. 36. B. Sarma, S. Roy, and A. Nangia, Chem. Commun., 2006, 4918. 37. N. J. Babu and A. Nangia, CrystEngComm, 2007, 9, 980.

8. M. R. Caira, L. R. Nassimbeni, and A. F. Wildervanck, J. Chem. Soc., Perkin Trans 2, 1995, 2213.

38. S. R. Chemburkar, J. Bauer, K. Deming, et al., Org. Process Res. Dev., 2000, 4, 413.

9. M. C. Etter, Acc. Chem. Res., 1990, 23, 120.

39. J. W. Gibbs, Trans. Connecticut Acad. Arts Sci., 1876, 3, 108.

10. F. H. Herbstein, Crystalline Molecular Complexes and Compounds, IUCr Monograph, Oxford University, Oxford, 2005. ¨ Almarsson and M. J. Zaworotko, Chem. Commun., 11. O. 2004, 1889. 12. Everything added to Food in the United States (EUFAS), www.cfsan.fda.gov/∼dms/eafus.html. 13. A. V. Trask, Mol. Pharm., 2007, 4, 301. 14. R. Hilfiker, ed., Polymorphism in the Pharmaceutical Industry, Wiley-VCH, Weinheim, Germany, 2006. 15. A. Brillante, I. Bilotti, R. G. Della Valle, et al., CrystEngComm, 2008, 10, 1899. 16. T. Miyazaki, T. Watanabe, and S. Miyata, Jpn. J. Appl. Phys., 1988, 27, L1724. 17. G. Lincke, Dyes Pigments, 2000, 44, 101. 18. M. C. McCrone, in Polymorphism in Physics and Chemistry of the Organic Solid State, eds. D. Fox, M. M. Labes, and A. Weissberger, Wiley-Interscience, New York, 1965, vol. 2, pp. 725–767. 19. P. Vishweshwar, J. A. McMahon, M. Oliveira, J. Am. Chem. Soc., 2005, 127, 16802.

40. A. Grunenberg, J.-O. Henck, and H. W. Siesler, Int. J. Pharm., 1996, 129, 147. 41. (a) G. Tammann, The States of Aggregation (trans. ed., F. F. Mehl), Constable and Company Ltd., London, 1926, pp. 116–157; Book review by (b) H. S. Taylor, J. Chem. Edu., 1926, 3, 119. 42. A. Burger and R. Ramberger, Mikrochim. Acta, 1979, II, 259. 43. A. Burger and R. Ramberger, Mikrochim. Acta, 1979, II, 273. 44. A. I. Kitaigorodskii, Organic Chemical Crystallography, Consultants Bureau, New York. 1961. 45. W. Ostwald, Z. Phys. Chem., 1897, 22, 289. 46. J. Bernstein, R. J. Davey, and J.-O. Henck, Angew. Chem. Int. Ed., 1999, 38, 3440. 47. S. Datta and D. J. W. Grant, Cryst. Res. Technol., 2005, 40, 233.

et al.,

48. B. Y. Shekunov and P. York, J. Cryst. Growth, 2000, 211, 122.

and

49. F. P. A. Fabbiani, D. R. Allan, W. I. F. David, et al., CrystEngComm, 2004, 6, 504.

20. G. M. Day, A. V. Trask, W. D. S. Motherwell, W. Jones, Chem. Commun., 2006, 54.

21. P. K. Thallapally, R. K. R. Jetti, A. K. Katz, et al., Angew. Chem. Int. Ed., 2004, 43, 1149. 22. Cambridge Crystallographic Data Center, the CSD now contains over 500 000 crystal structure entries, www.ccdc.cam.ac.uk. 23. J. O. Henck, U. J. Griesser, and A. Burger, Pharm. Ind., 1997, 59, 165. 24. M. J. Buerger, Trans. Am. Crystallogr. Assoc., 1971, 7, 1. 25. E. Mitscherlich, Ann. Chim. Phys., 1822, 19, 350. 26. M. C. Marignac, Arch. Sci. Phys. Nat., 1873, 46, 193. 27. M. C. Marignac, Ann. Chim. Phys., 1863, 69, 5. 28. F. W¨ohler and J. Liebig, Annal. Pharm., 1832, 3, 249.

50. R. J. Davey, K. Allen, N. Blagden, et al., CrystEngComm, 2002, 4, 257. 51. G. D. Profio, A. Caridi, R. Caliandro, et al., Cryst. Growth Des., 2010, 10, 449. 52. Y. Tozuka, D. Kawada, T. Oguchi, and K. Yamamoto, Int. J. Pharm., 2003, 263, 45. 53. C. S. Towler, R. J. Davey, R. W. Lancaster, and C. J. Price, J. Am. Chem. Soc., 2004, 126, 13347. 54. D. Das and L. J. Barbour, J. Am. Chem. Soc., 2008, 130, 14032. 55. S. Roy, S. Aitipamula, and A. Nangia, Cryst. Growth Des., 2005, 5, 2268.

30. S. Chen, I. A. Guzei, and L. Yu, J. Am. Chem. Soc., 2005, 127, 9881.

56. B. Sarma, P. Sanphui, and A. Nangia, Cryst. Growth Des., 2010, 10, 2388. ¨ Almarsson, M. L. Peterson, et al., Adv. 57. S. L. Morissette, O. Drug Deliv. Rev., 2004, 56, 275.

31. R. K. R. Jetti, R. Boese, J. A. R. P. Sarma, et al., Angew. Chem. Int. Ed., 2003, 42, 1963.

58. R. Hilfiker, J. Berghausen, F. Blatter, et al., J. Therm. Anal. Cal., 2003, 73, 429.

32. S. Parveen, R. J. Davey, G. Dent, and R. G. Pritchard, Chem. Commun., 2005, 1531.

59. H. G. Brittain, ed., Polymorphism in Pharmaceutical Solids, 2nd ed., Informa Healthcare, New York, 2009.

29. A. Nangia, Acc. Chem. Res., 2008, 41, 595.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

18

Supramolecular materials chemistry

60. R. K. Harris, Analyst, 2006, 131, 351. 61. D. Braga, L. Maini, G. Sanctis, et al., Chem. Eur. J., 2003, 9, 5538. 62. S. Thirunahari, S. Aitipamula, P. S. Chow, and R. B. H. Tan, J. Pharm. Sci., 2010, 99, 2975.

82. R. M. Vrcelj, N. I. B. Clark, J. Pharm. Sci., 2003, 92, 2069.

A. R. Kennedy,

et al.,

83. N. Blagden, M. de Matas, P. T. Gavan, and P. York, Adv. Drug Del. Rev., 2007, 59, 617.

63. H. G. Brittain, Am. Pharm. Rev., 2002, 5, 74.

84. S. Aitipamula, P. S. Chow, and R. B. H. Tan, Cryst. Growth Des., 2010, 10, 2229.

64. E. W. Pienaar, M. R. Caira, and A. P. L¨otter, J. Crystallogr. Spectr. Res., 1993, 23, 739.

85. M. B. Hickey, M. L. Peterson, L. A. Scoppettuolo, et al., Eur. J. Pharm. Biopharm., 2007, 67, 112.

65. K. D. M. Harris, M. Tremayne, and B. M. Kariuki, Angew. Chem. Int. Ed., 2001, 40, 1626.

86. W. W. Porter III, S. C. Elie, and A. J. Matzger, Cryst. Growth Des., 2008, 8, 14.

66. G. A. Stephenson, R. A. Forbes, and S. M. Reutzel-Edens, Adv. Drug Deliv. Revs., 2001, 48, 67.

87. J. H. ter Horst and P. W. Cains, Cryst. Growth Des., 2008, 8, 2537.

67. R. E. Dinnebier and S. J. L. Billinge, eds., Powder Diffraction, Theory and Practice, Royal Society of Chemistry, Cambridge, 2008.

88. B. Puschner, R. H. Poppenga, L. J. Lowenstine, et al., J. Vet. Diagn. Invest., 2007, 19, 616.

68. Y. Matsuda, R. Akazawa, R. Teraoka, and M. Otsuka, J. Pharm. Pharmacol., 1993, 46, 162. 69. M. M. De Villiers, J. G. van der Watt, and A. P. Lotter, Int. J. Pharm., 1992, 88, 275. 70. R. Eyjolfsson, Pharmazie, 2002, 57, 347. 71. M. Pudipeddi and A. T. M. Serajuddin, J. Pharm. Sci., 2005, 94, 929. 72. D. Singhal and W. Curatolo, Adv. Drug Deliv. Rev., 2004, 56, 335. 73. T. Maeda, H. Takenada, Y. Yamahira, and T. Noguchi, Chem. Pharm. Bull., 1980, 28, 431. 74. Y. Kobayashi, S. Ito, S. Itai, and K. Yamamoto, Int. J. Pharm., 2000, 193, 137. 75. N. Rasenack and B. M¨uller, Int. J. Pharm., 2002, 244, 45. 76. M. Brits, W. Liebenberg, and M. M. de Villiers, J. Pharm. Sci., 2010, 99, 1138. 77. J. K. Haleblian, R. T. Koda, and J. A. Biles, J. Pharm. Sci., 1971, 60, 1485. 78. E. Shefter and T. Higuchi, J. Pharm. Sci., 1963, 52, 781. 79. R. K. Khankari, Ph.D. Thesis, University of Minnesota, Minneapolis, MN, 1993. 80. M. J. Jozwiakowski, S. O. Williams, and R. D. Hathaway, Int. J. Pharm., 1993, 91, 195. 81. A. Nangia and G. R. Desiraju, Chem. Commun., 1999, 605.

89. G. Xie, X. Zheng, X. Qi, et al., J. Proteome Res., 2010, 9, 125. 90. R. K. Mughal, R. J. Davey, and N. Blagden, Cryst. Growth Des., 2007, 7, 218. 91. I. Weissbuch, L. Leiserowitz, and M. Lahav, Cryst. Growth Des., 2003, 3, 125. 92. S. K. Poornachary, P. S. Chow, and R. B. H. Tan, Cryst. Growth Des., 2008, 8, 179. 93. Y. Gong, B. Collman, S. M. Mehrens, et al., J. Pharm. Sci., 2008, 97, 2130. 94. M. Rafilovich and J. Bernstein, J. Am. Chem. Soc., 2006, 128, 12185. 95. B. Lou, D. Bostr¨om, and S. P. Velaga, Cryst. Growth Des., 2009, 9, 1254. 96. X. Sun, B. A. Garetz, and A. S. Myerson, Cryst. Growth Des., 2006, 6, 684. 97. C. P. Price, A. L. Grzesiak, and A. J. Matzger, J. Am. Chem. Soc., 2005, 127, 5512. 98. S. Roy and A. J. Matzger, Angew. Chem. Int. Ed., 2009, 48, 8505. 99. S. Beach, D. Latham, C. Sidgwick, et al., Org. Proc. Res. Dev., 1999, 3, 370. 100. S. L. Childs, L. J. Chyall, J. T. Dunlap, et al., Cryst. Growth Des., 2004, 4, 441. 101. 9th International Workshop on Crystal Growth of Organic Materials, 4–7 August 2010, Singapore, www.cgom9.com.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc114

Mechanical Preparation of Crystalline Materials. An Oxymoron? Dario Braga1 , Elena Dichiarante2 , Fabrizia Grepioni1 , Giulio I. Lampronti1 , Lucia Maini1 , Paolo P. Mazzeo1 , and Simone D’Agostino1 1 2

University of Bologna, Bologna, Italy PolyCrystalLine s.r.l., Medicina, Bologna, Italy

1 Introduction 2 The Oxymoron 3 Mechanochemical Preparation of Coordination Networks 4 Mechanochemical Preparation of Cocrystals 5 Conclusions References

1

1 2 4 8 11 13

INTRODUCTION

In this chapter, we deal with the utilization of mechanochemical processes between molecular and ionic crystals to obtain new crystalline materials. Mechanochemical processes can be grossly divided into two categories depending on the bonding interactions involved: (i) breaking and forming of noncovalent bonds to yield supramolecular adducts (viz. cocrystals, salts, molecular complexes, host–guest systems, etc.)1 and (ii) breaking and forming of covalent bonds by the action of mixing reactants.1–9 These processes can take place in the course of a reaction between different molecular materials (intersolid) or within a molecular material (intrasolid).10 A third important class of reactions is that combining inter- and intrasolid processes that exploits

supramolecular (mainly hydrogen bond driven) aggregation to link in space the reactants with the correct separation to allow subsequent covalent reaction, most often activated by UV radiation.11, 12 Mechanochemical reactions have been known for a long time.13–15 They are usually carried out by manual or motorized grinding of two or more crystalline reactants without the involvement of solvents.16–19 Mechanochemical methods are commonly used at industrial level mainly with inorganic solids.20–23 When the mixing or grinding involves molecular crystals and in the presence of a small amount of solvent or directly with the liquid reactant (kneading or liquid-assisted grinding, LAG), the rate of the solid-state reaction is often substantially enhanced.24–26 Sometimes solvent dependence is observed.27 The same may occur in the case of metalorganic frameworks (MOFs),28–35 with the notable difference that, while molecular coordination complexes can exist also in solution, coordination polymers, due to their infinite nature, do not exist in solution and tend to be insoluble. Similarly, molecular crystals, on the basis of hydrogen bonded networks, though generally soluble, exist only in the solid state; for both materials, characterization requires essentially solid-state techniques. In the following, we describe solid-state reactions between molecular crystals and/or salts of organic, organometallic, and metallorganic species to generate new crystal forms (cocrystals, salts, adducts, and solvates). The possibility of using mechanochemical methods to prepare coordination networks and coordination compounds will

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc115

2

Supramolecular materials chemistry

also be dealt with. The potential applications of grinding and kneading methods in a variety of research areas, for example, pharmaceutical compounds, pigments, solid fuel cell electrolytes, and magnetic materials and gas sensors will also be discussed. In view of the vastity of the subject, we have chosen to confine our attention to few subareas, namely those of solidstate interconversions of organic tautomers, preparation and interconversions of solid-state MOFs, and preparation and characterization of cocrystals. Examples taken from our own work as well as from that of others will be used to discuss the most common problems encountered in the study and characterization of solid-state reaction products.

2

THE OXYMORON

The main difficulty of mechanical reactions is the control of the reaction conditions: grinding time, temperature, the pressure exerted by the operator (e.g., in manual grinding), and so on, are the important variables that are not easy to define and monitor during the reaction process. Mechanical stress by fracturing the crystals increases the surface area and facilitates interpenetration and reaction depending on the ability of molecules to diffuse through the crystal surfaces, so that intersolid reactions are often controlled by the rate of diffusion (or by the rate of elimination from within the molecular material of solvent molecules or byproducts). Furthermore, the heat generated in the course of the mechanochemical process can induce local melting of crystals or melting at the interface between the different crystals, so that the reaction might actually be taking place in the liquid phase and not in the solid even though solid products are ultimately recovered. Since mechanical treatment of preformed crystals leads to polycrystalline samples, product characterization often

poses additional problems to the experimentalists. The characterization by single-crystal techniques, for example, is more or less “forbidden” for species obtained mechanochemically (the oxymoron!). It is therefore essential to be able to structurally characterize the product material by other methods. This can be done in one or more of three ways: (i) attempt to circumvent the problem by growing single crystals of the product from appropriate solvents by the use of seeding techniques after the product has been obtained mechanochemically; (ii) structural determination from powder diffraction data alone when diffraction data of good quality are available and the system under investigation is not exceedingly complex; and (iii) characterization by spectroscopic means only, that is, a combination of information derived from solid-state NMR, IR, Raman, and so on. As pointed out in (i), seeding can provide a route to the growth of single crystals of suitable size of the desired material.36–39 It is often sufficient (Figure 1) to use few milligrams of the powdered material resulting from the mechanical treatment to initiate the crystal growth process. This is particularly important in the case of crystal polymorphism. Seeds of isostructural or quasi -isostructural species that crystallize well can also be employed to induce crystallization of unyielding materials (heteromolecular seeding).40, 41 Examples of seeding have been recently reported for the preparation of single crystals of complexes of gabapentin and ZnCl2 and CuCl2 42 ; or complexes with 4-aminosalicylic acid and piracetam, which had been obtained by mechanical reaction.43 Approach (ii), on the other hand, relies on the possibility of determining the crystal structure directly from powder diffraction experiments. Powder data treatment and ab initio structure determination methods are evolving rapidly.

5 10 15 20 25 30 35 40

Figure 1 A solid–solid process and the strategy to obtain single crystals by recrystallization of the solid reaction product in the presence of seeds of the desired crystals. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc115

Mechanical preparation of crystalline materials However, structure solution from powder data remains often a challenge. Diffraction pattern indexing is the first difficulty, since the choice of a unit cell is not a straightforward and unambiguous process. A very critical step is structure factor amplitudes extraction from the powder pattern: this problem arises from the systematic and accidental overlapping of the three-dimensional diffraction data onto the one-dimensional 2-θ axis. Two methods are generally used for this task, the Pawley and Le Bail profile fittings.44 Several algorithms for structure solution from extracted amplitudes have been suggested: simulated annealing,45 evolutionary algorithms,46 Monte Carlo,47 direct methods,48 and charge flipping.49 Any solution method is, however, going to fail if amplitude extraction is not correct. Thus, the data collection setting has to be chosen adequately in order to avoid preferential orientation as much as possible, because this strongly affects peak intensities. Nowadays, pattern indexing, amplitude extraction, and structure solution algorithms can be generally run on a common desktop PC, and have been implemented in several available software packages, for example, EXPO,50 DASH,51 TOPAS,52 and Xpert High Score.53 Finally, a careful Rietveld refinement must always follow a structure solution. Indeed a structure solution has to be checked by optimizing the quality of the fit, although the most important criterion for judging a refinement is the chemical sense of the structural model.54 The more complex the structure (number of molecules in the asymmetric unit, number of degrees of freedom of the molecule itself), the longer the calculation time for the structure solution, and the more difficult the Rietveld refinement. Furthermore, structure determination from a powder pattern may become an impossible task if the powder sample contains unknown impurities. In these cases, it may be necessary to recur to other techniques such as transmission electron microscopy (TEM), which allows very tiny crystals ( T1

3

T1

T2 > T1

0 (a)

Figure 2

Time

kobs

a

T2

(b)

T2

P 0 (T 1 ) P 0 (T 2 )

Px

p (mmHg)

Anti-Arrhenius behavior of the kinetics of a host–guest complex during enclathration.

G(g)

H(s, b0) Host, H

∆H trans

Guest, G

H(s, a)

G(I)

H(s, b)

Figure 3 The formation of an inclusion compound from the solid apohost H(s,α) with a liquid guest G to form the clathrate H·G(s,β) , with a guest to host ratio G:H equal to 1.

Step 2

The guest is converted from the liquid to the vapor phase: G(l) → G(g)

Step 3

Hevap is positive, endothermic (if the guest is solid, Hevap becomes Hsub , the sublimation enthalpy). The gaseous guest enters the empty β 0 structure to form the inclusion compound in its β-phase. H(s,β 0 ) + G(g) → H·G(s,β) The enthalpy of sorption, Hsorp , is negative, exothermic.

The H of enclathration, Hclath , is given by Hclath = Htrans + Hevap + Hsorp We note that the overall process, the driving force, is the change in enthalpy, which overrides the negative entropy contribution, yielding a negative Gibbs free energy change.

Lipkowski, in his analysis of the energetics of formation of Werner clathrates, adds an enthalpic term for the dilation of the β-lattice when it accommodates the guest. This term is dependent on the size of the guest as well as the percentage occupancy.13 This process has been analyzed in detail for the formation of the inclusion complexes of urea with n-paraffins.14

4

THERMAL STABILITY

The decomposition of an inclusion compound is essentially the reverse of its formation discussed above. Thus, when the inclusion compound decomposes under heating: H·Gn(s,β) → H(s,α) + nG(g) the H of the reaction is a combination of the host–guest interactions and the enthalpy change of the concomitant phase change of the host from the β 0 to the α phase. In cases where the guest is volatile, this is readily followed by TGA and differential scanning calorimetry (DSC). This is

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc116

4

Supramolecular materials chemistry

TGA

B

Power

A

exo

Mass %

∆m

endo

∆H

DSC

Ton1

Ton2 T (°C)

Figure 4 TGA and DSC recordings of an inclusion compound while it is subjected to a selected temperature program under a controlled atmosphere.

illustrated in Figure 4, which shows the mass loss of the solid inclusion compound while it is subjected to a selected temperature program under a controlled atmosphere of a carrier gas, usually dry nitrogen or helium. The measurement of mass loss is important in inclusion compounds, because this yields an accurate value of the guest:host ratio n. Since such compounds are often nonstoichiometric, its value is useful in assigning sensible site occupancy factors in the crystal structure analysis of such compounds. An example of this is in the crystal structure of the metal–organic framework formed between 1,3,5-benzenetricarboxylic acid (H3 BTC) and zinc sulfate in a water–DMF (dimethylformamide) solution which yielded two MOFs, [Zn6 (µOH)2 (BTC)(DMF)2.5 (H2 O)2 ]·[Zn(H2 O)3 (DMF)3 ]·3.1H2 O and [Zn2 (HBTC)(BTC)(H2 O)3 ]·DMA·3H2 O.15 The channels in the first crystal structure are occupied by uncoordinated guest and water molecules whose site occupancy factors were modeled after the TGA results. Dehydration–rehydration experiments on the second compound were followed by powder diffraction, and TGA was employed to demonstrate that the compound regains all coordinated and guest water molecules that had been lost on heating. Thermal gravimetry is particularly important in the structure analysis of cyclodextrin inclusion compounds that contain large numbers of waters of crystallization and that are often disordered over multiple sites of the crystal structure. This is illustrated by the structure of the inclusion complex of S-p-tolyl t-butylthiosulfinate with β-cyclodextrin. The stoichiometric formula for this complex was determined as C42 H70 O35 ·C11 H16 OS2 ·12.2H2 O from 1 H NMR data (for the host–guest ratio) and thermogravimetric analysis (for the water content). Crystallographic analysis led

to initial location of 10 oxygen atoms corresponding to water molecules in a difference electron density map. These were assigned full site-occupancies on the basis of their equitable peak heights. An additional five sites for water oxygen atoms were subsequently found and their site occupancy factors were refined while their isotropic thermal parameters were set to the average of those of the ordered oxygen atoms. At the end of the refinement, the sum of the site occupancy factors (s.o.f.s) was 12.46, in accord with the value 12.5, which was estimated by thermogravimetric analysis.16 During a DSC analysis, the endotherm A (Figure 4), associated with the guest release, yields the change in enthalpy H for the guest release reaction, but should be treated with caution, because reliable values rely on good calibration of the DSC and careful experimental procedures. The area under the curve and the onset temperatures depend on several factors, which include the geometry of the apparatus, the heating rate, the flow rate of the purging gas, and the particle size distribution of the sample. The last mentioned factor is the most sensitive one and, ideally, samples should be sieved in order to operate only with a narrow band of crystallite size. This is not possible if the guest is volatile, owing to the prompt desolvation of the inclusion compound. We have studied the effect of particle size on the kinetics of decomposition at a constant heating rate.17 The inclusion compound formed by the host 1,4-di-(5H dibenzo[a,d]cyclohepten-5-ol)-buta-1,3-diyne with acetone was subjected to thermal gravimetry at a heating rate of 10 ◦ C min−1 . Crystallites of known size, varying in volume from 0.02 to 1.6 mm3 , showed that the decay curve was displaced to higher temperatures with increasing particle size. We noted that the maximum rate of decomposition, estimated from dm/dT curves, changed by as much as 14.4 K from the smallest to the largest crystallites. To obtain reliable values of the change in enthalpy (H ) of the desorption reaction, one may measure the vapor pressure of the volatile guest as a function of temperature. This method is slow, in that the vapor pressure must be allowed to reach equilibrium values at each selected temperature, but the appropriate plot of ln p versus 1/T yields good straight lines which give accurate values of H . An example of this is the desolvation of the inclusion compound formed by the host 9,9
-dihydroxy9,9
-bifluorene with ethanol, whose vapor pressure was measured between 25 ◦ C and 80 ◦ C, yielding a H of 17.9 kJ mol−1 .18 We have noted several cases where the decomposition reaction yields only a single endotherm. This is because the outgoing guest liquefies and dissolves the host—a phenomenon that can be confirmed by hot-stage microscopy. This occurred in the desolvation of the inclusion compounds

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc116

Physico-chemical aspects of inclusion compounds

A

ZA

A

ZA

formed by the bulky host 1,1,6,6-tetraphenylhexa-2,4diyne-1,6-diol with lutidine isomers19 and tetrahydrofuran.20

5

5 SELECTIVITY The phenomenon of selectivity of a host for a particular guest is a process driven by molecular recognition. Thus, when a host, H, reacts with a guest, G, in solution:

(a) B

XA

A (b) B

XA

A

A

A

the equilibrium constant, K, can be evaluated from the knowledge of the activities of the host–guest complex and those of the free species. K may be approximated by using concentrations instead of activities: K=

[H·nG] [H] [G]n

with the standard state defined as 1 mol L−1 . The selectivity of a host binding two particular guests G1 and G2 is given by KG1 /KG2 .21 The standard Gibbs free energy, G◦ = (−RT )lnK, is a measure of the molecular recognition occurring in the host–guest reaction. However, when the host–guest compound precipitates from the saturated solution of the mother liquor, the evaluation of G is difficult, as it involves the accurate determination of the solubilities of the host in the guest mixtures. This is time consuming and may be impractical due to the paucity of materials. In practice, it is more convenient to adopt the concept of a selectivity coefficient KA:B , first enunciated by Ward22 and defined as KA:B = (KB:A )−1 = ZA /ZB × XB /XA (XA + XB = 1; ZA + ZB = 1) where for two guests A and B, XA and XB are the mole fractions in the liquid mixture and ZA and ZB are the mole fractions of the guests in the crystal. The experimental technique is to set up a series of competition experiments whereby a fixed quantity of host compound is exposed to mixtures of guests which systematically vary in composition from XA = 0 to XA = 1. The crystals derived from each mixture are analyzed by a suitable analytical technique (TGA, NMR, and gas chromatography) and the mole fractions ZA and ZB of the enclathrated guests are determined. The ensuing selectivity profiles are of four kinds as shown in Figure 5. Figure 5(a) shows zero selectivity, with the points lying on the diagonal line representing KA:B = 1. This occurs

(c) B

ZA

ZA

H + nG
H·nG

XA

A (d) B

XA

A

Figure 5 Typical selectivity curves obtained from competition experiments.

when the topology of the space that accommodates the guest is sufficiently adaptable for either A or B and the resulting structures form a series of solid solutions that are isomorphous. An example is the enclathration of m- and p-xylene by cholic acid, in which the selectivity profile has a low selectivity coefficient, Kp-xylene:m-xylene = 2.5.23 Figure 5(b) shows the case where A is preferentially enclathrated over the whole concentration range with a large selectivity coefficient. This is the situation where the separation of the two guests can be carried out, and in practice it requires a selectivity coefficient KA:B ≥ 10. There are numerous examples of this effect, and a salient example is that of the host 1,1
-bis-(4-hydroxyphenyl)cyclohexane which entraps 3,5xylidine preferentially over 2,6-xylidine.24 Figure 5(c) is the result obtained when the selectivity is concentration dependent and the host favors the guest in greater concentration. This is exemplified by the selectivity profile exhibited by the host 1,1,6,6-tetraphenylhexa-2,4diyne-1,6-diol with the isomeric guests 3-aminobenzonitrile and 4-aminobenzonitrile.25 It is noteworthy that, in this case, the same result is obtained when the competition is carried out in solution or from direct solid–solid reactions. Figure 5(d) results from the host preferentially enclathrating the minor component from the liquid mixture. This is rare, but has been observed in the competition between 3-picoline and 4-picoline with a resorcinarene host, and has been attributed to the solubilities of the compounds formed in the relevant guest mixtures.26 Unusual selectivity profiles, in which the mole fraction in the crystal, Z, varies in discrete steps, have been recorded. Inclusion compounds of formula H·nA·(4 − n)B, in which

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc116

6

Supramolecular materials chemistry

1

(R, S)-AH + (R)-B

H·4B

0.8

H·A·3B

ZB

0.6

↓ −

+

{(R)-A , (R)-BH } + {(S)-A− , (R)-BH+ } p-salt n-salt

H·2A·2B 0.4 H·3A·B

0.2 0

H·4A 0

0.2

0.4

0.6

0.8

1

XB

Figure 6 Unusual selectivity profile of inclusion compounds of formula H·nA·(4 − n)B in which n integrally varies from 0 to 4. (Reproduced from Ref. 22.  American Chemical Society, 2001).

(R)-Host + (R, S)-Guest ↓ (R)-Host·(R)-Guest + (S)-Guest In the above equation, we assume complete recognition of the (R)-Host for the (R)-Guest which precipitates, leaving the (S)-Guest in the mother liquor. This process, however, is seldom completely efficient, and the procedure may require several cycles to obtain the desired result. This is illustrated in a McCabe–Thiele type plot in Figure 7, which shows a system with a modest selectivity coefficient K(R)-G:(S)-G = 5. Starting with a racemic mixture, XR = 0.5, we carry out three consecutive steps of crystallization and retrieval of the guest, yielding a guest of ≈99% enantiomeric purity. R

1 0.9 0.8 0.7 0.6

ZR

n integrally varies from 0 to 4, have been characterized for the system where the host is trans-9,10-dihydroxy-9,10bis(p-tert-butylphenyl)-9,10-dihydroanthracene and the two guests are DMF (A) and DMSO (dimethyl sulfoxide) (B).27 The result is shown in Figure 6. The stoichiometry is therefore controlled by the composition of the guest mixture. All five resultant compounds were fully characterized and their structures were elucidated. The compounds are isostructural with respect to the location of the host molecules and the guests essentially lie at the same sites, exhibiting partial disorder. This result has possible applications in crystal engineering, in that the host:guest stoichiometry can be controlled, with concomitant results in the properties of the resultant compounds, such as their optical and electronic properties as well as their thermal stabilities and kinetic behavior. A simplified form of this stepwise formation of inclusion compounds with mixed guests occurs in the selectivity by the host 9,9
-(ethyne-1,2-diyl)-difluoren-9-ol with the guests ethanol/acetonitrile.28 In this case, the stoichiometry of the mixed guests remains fixed at ZEtOH = 0.56 over a large composition of the mother liquor from XEtOH = 0.1 to XEtOH = 0.9.

The method relies on the p and the n salts having different solubilities and they must not form solid solutions or double salts. The more insoluble salt is allowed to precipitate, is filtered, and the purified acid is recovered by adding mineral acid. Enantiomeric resolution via diastereomeric salt formation is well established and many systems have been studied and cataloged.29 Inclusion chemistry may also be employed for the separation of enantiomers by enclathrating racemic guests with a chiral host:

0.5 0.4

6

ENANTIOMERIC RESOLUTION

A special case of host selectivity is the resolution of a racemic mixture of chiral guests. This has important implications for the pharmaceutical industry, where the production of enantiomerically pure drugs has become important. The most common method of chiral resolution is via the formation and separation of diastereomers. For example, a racemic acid AH may be treated with a chiral base B.

0.3 0.2 0.1 0 0

S

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 XR

1 R

Figure 7 A McCabe–Thiele type plot which shows a system with a selectivity coefficient K(R)-G:(S)-G = 5.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc116

1

1

0.9

0.9

0.8

0.8

0.7

0.7

0.6

0.6

ZR -BUAM

ZR -BUAM

Physico-chemical aspects of inclusion compounds

0.5 0.4

0.5 0.4

0.3

0.3

0.2

0.2

0.1

0.1 0

0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (a)

7

XR -BUAM

XR - BUAM

(b)

1 0.9 0.8

ZR -BUAM

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (c)

XR -BUAM

Figure 8 Selectivity curves for (R,R)-(−)-trans-4,5-bis-(hydroxydiphenylmethyl)-2,2-dimethyl-1,3-dioxolane: (a) molecule 1, (b) molecule 2, and (c) overall selectivity, respectively.

We have investigated the mechanism of enantiomeric enrichment by enclathration by studying TADDOL type hosts (derivatives of α,α,α
,α
-tetraaryl-1,3-dioxolane-4,5dimethanol) which enclathrated 2-butylamine mixtures of varying enantiomeric composition.30 In particular, the host (R,R)-(−)-trans-4,5-bis-(hydroxydiphenylmethyl)2,2-dimethyl-1,3-dioxolane entraps different proportions of both enantiomers of 2-butylamine, depending on the initial composition of the mother liquor. The system is subtle, in that the compounds crystallize in P 1 with Z = 2 and the two independent host molecules behave differently. The selectivity curves of molecules 1 and 2, and the overall selectivity are shown in Figure 8. Analysis of the variation in the torsion angles which control the conformation of the phenyl rings of the host and the concomitant volume of the space available for the enclathrated 2-butylamine guests shows that enhanced enantiomeric selectivity arises from smaller guest volumes. The mechanism of enantiomeric enrichment, by both chiral inclusion and diastereomeric salt formation, is a phenomenon that is not clearly understood. We believe that the analysis of selectivity profiles, together with detailed studies of the structural results, is a rich field which will

yield important information on the process of molecular recognition that drives this phenomenon.

7

CONCLUSION

Various aspects of the physico chemical properties of organic inclusion compounds have been considered. Analysis of their thermal stability and kinetics of formation and decomposition yield a better understanding of their reactivity. Their mechanism of host selectivity is driven by the process of molecular recognition and can be elucidated by analyzing the secondary interactions responsible for their supramolecular structure.

REFERENCES 1. F. H. Herbstein, Crystalline Molecular Complexes and Compounds, University Press, Oxford, 2005, vol. 1, Chapter 1. 2. E. Weber and H.-P. Josel, J. Inclusion Phenom., 1983, 1, 79. 3. M. E. Brown Ed., Handbook of thermal analysis and calorimetry, Principles and Practice, Elsevier, Amsterdam, 1988, vol. 1, Chapter 3.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc116

8

Supramolecular materials chemistry 4. A. Jacobs, N. L. Z. Masuku, L. R. Nassimbeni, and J. H. Taljaard, CrystEngComm, 2008, 10, 322–326.

20. M. R. Caira, L. R. Nassimbeni, F. Toda, and D. Vujovic, J. Chem. Soc., Perkin Trans. 2, 2001, 2119–2124.

5. L. R. Nassimbeni, G. Ramon, and E. Weber, J. Therm. Anal. Cal., 2007, 90, 31–37.

21. J. W. Steed, D. R. Turner, and K. J. Wallace, Core Concepts in Supramolecular Chemistry and Nanochemistry, Wiley, Chichester, 2007.

6. A. Jacobs, L. R. Nassimbeni, H. Su, and B. Taljaard, Org. Biomol. Chem., 2005, 3, 1319–1322. 7. D. V. Pinakov, V. A. Logvinenko, Yu. V. Shubin, and G. N. Chekova, J. Therm. Anal. Cal., 2009, 96, 501–505. 8. V. Logvinenko, V. Drebushchak, D. Pinakov, G. Chekhova, J. Therm. Anal. Cal., 2007, 90, 23–30.

and

9. A. K. Galway and M. E. Brown, Proc. R. Soc. London, Ser. A., 1995, 1, 501–512.

22. A. M. Pivovar, K. T. Holman, and M. D. Ward, Chem. Mater., 2001, 13, 3018. 23. K. Nakano, E. Mochizuki, N. Yasui, et al., J. Org. Chem., 2003, 2436. 24. L. R. Nassimbeni and H. Su, Acta Crystallogr., 2002, B58, 251–259.

10. V. Logvinenko, Thermochim. Acta, 1999, 340/341, 293–299.

25. M. R. Caira, L. R. Nassimbeni, F. Toda, and D. Vujovic, J. Am. Chem. Soc., 2000, 122, 9367–9372.

11. L. J. Barbour, K. Achleitner, and J. R. Greene, Thermochim. Acta, 205, 171–177.

26. M. R. Caira, T. Le Roex, and L. R. Nassimbeni, CrystEngComm, 2006, 8, 275–280.

12. M. R. Caira, T. Le Roex, L. R. Nassimbeni, and E. Weber, Cryst. Growth Des., 2006, 6, 127–131.

27. L. J. Barbour, M. R. Caira, T. Le Roex, and L. R. Nassimbeni, J. Chem. Soc., Perkin Trans. 2, 2002, 1973–1979.

13. P. Starzewski, W. Zielenkiewicz, and J. Lipkowski, J. Inclusion Phenom., 1984, 1, 223–232.

28. T. le Roex, L. R. Nassimbeni, and E. Weber, Chem. Commun., 2007, 1124–1126.

14. W. Schlenk Jr, Liebigs Ann. Chem., 1949, 565, 204–240. ¨ 15. K. Davies, S. A. Bourne, L. Ohrstrom, and C. L. Oliver, J. Chem. Soc., Dalton Trans., 2010, 2869–2874.

and

29. (a) J. Jacques, A. Collet, and S. H. Wilen, Enantiomers, Racemates and Resolutions, Krieger Publishing Company, Malabar, Florida, 1991; (b) S. H. Wilen and E. L. Eliel, Tables of Resolving Agents and Optical Resolutions, University of Notre Dame Press, Notre Dame—London, 1971; (c) D. Kozma, ed., Optical Resolution via Diastereomeric Salt Formation, CRC Press, New York, 2002.

et al.,

30. N. B. B´athori and L. R. Nassimbeni, Cryst. Growth Des., 2010, 10, 1782–1787.

16. N. Stellenboom, R. Hunter, M. R. Caira, et al., Supramol. Chem., 2009, 21(7), 611–617. 17. A. Jacobs, N. L. Z. Masuku, L. R. Nassimbeni, J. H. Taljaard, CrystEngComm, 2008, 10, 322–326. 18. L. J. Barbour, M. R. Caira, L. R. Nassimbeni, Supramol. Chem., 1995, 5, 153–158.

19. M. R. Caira, L. R. Nassimbeni, F. Toda, and D. Vujovic, J. Chem. Soc., Perkin Trans. 2, 1999, 2681–2684.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc116

Clathrate Hydrates Saman Alavi, Konstantin Udachin, Christopher I. Ratcliffe, and John A. Ripmeester National Research Council, Ottawa, Ontario, Canada

1 2 3 4

Introduction The Clathrate Hydrates Guest Species Theoretical Description, Guest–Host Interactions, and Modeling 5 Clathrate Hydrate Preparation and Characterization 6 Future Directions 7 Conclusion References

1

1 2 7 9 12 13 13 13

INTRODUCTION

The publication of this series, interestingly enough, coincides with the 200th anniversary of the first report of the synthesis of a clathrate hydrate, that of chlorine, by Humphrey Davy.1 He reported that a solution of chlorine in water froze at a temperature higher than the ice melting point. Upon decomposition, this unique new material returned unchanged to the starting materials. Clathrate hydrates were likely the first synthetic materials that could be classified as supramolecular. The clathrate nature of hydrates was not confirmed until the early 1950s,2 and by 1960 a theoretical description of clathrates became available.3 This point in time can be taken as the start of clathrate science as based on structural and molecular principles, thus opening this field to a

broader area of researchers. Since then, clathrate science has been reviewed a number of times, both the general aspects4 as well as more specialized accounts on structure,5 dynamics,6 and thermodynamics.7 In 1934, Hammerschmidt8 showed that gas hydrates are implicated in a natural gas pipeline blockage and since then hydrate control has been a major activity in the gas and oil industry with a substantial associated cost factor. The recent (2010) Deep Star Horizon oil well problems in the Gulf of Mexico again brought to the fore that gas hydrates often show up where they are not wanted. Gas hydrate control comes under the discipline of flow assurance in petroleum and chemical engineering. A comprehensive text in this area is the book by Sloan and Koh.9 The most important issues around clathrate hydrates currently involve the natural gas hydrates (NGH), first reported in the late 1960s.10 Because of the natural gas resource potential, research involving NGH has ramped up tremendously as they may be a source of natural gas in times of increasing energy demand with decreasing conventional global energy resources.11 Other areas of importance include geohazards related to NGH,12 as well as the possible impacts of NGH decomposition on global climate change.13 Thus, gas hydrate research now includes the fields of earth and ocean sciences as well as various engineering disciplines and the fundamental sciences. Researchers deal with hydrates on a scale from the molecular (nanometer) to that of gas hydrate reservoirs (meter to kilometer). Since the outstanding physical property of clathrate hydrates is the efficient storage of gases, interest continues in applications where this property will be of use. Typically, a volume of gas hydrate can hold 160 volumes of gas at STP, thus, gas storage, especially energy gases such as methane and hydrogen, continues to be actively pursued.14

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

2

Supramolecular materials chemistry

Table 1

The clathrate hydrates.

Guest species

Classification

Comments

Atoms and neutral molecules (but not amines, except for t-butylamine)

Clathrate hydrates (sI, sII, sH, sT, type VI, HS-1, bromine (type III)) Clathrate hydrates (sI)

van der Waals interactions, weak H-bonding

4, 5, 20–22, 28

Anion in cages, H+ in water lattice HF, H+ in water lattice Cations in cages, OH− in water lattice Strong guest–host H-bonds with numerous structural motifs

31

Strong acids HClO4 , HBF4 , HPF6 HXF6 (X = P, As, Sb) + HF Strong bases N(CH3 )4 + + OH− Amines

HF Clathrates (type VII) Clathrate hydrates (type VII, several others) Semiclathrates

The structure II hydrate of hydrogen, reported for the first time a few years ago, has had an interesting history as a research topic as researchers attempted to produce an efficient hydrogen storage medium at modest temperatures and pressures.15 Since hydrate cages of each type have preferred guest sizes, the separation of gas mixtures according to molecular size is also an active area of research, for instance, for separating CO2 from nitrogen in flue gas, or CO2 from H2 in fuel gas.16 The interest in the possibility of extraterrestrial clathrate hydrates on comets and the outer planets continues, for example, methane hydrate on Titan and CO2 hydrate on Mars.17 Moreover, this has sparked an interest in clathrate hydrate phases at high pressure.18

1.1

Classification of hydrates

From Powell’s rather general definition of clathrates,19 clathrate hydrates include any hydrate structure where guest species reside in cages in a water lattice. Other hydrates, either stoichiometric or not, can easily be distinguished from clathrate hydrates by using this rather broad definition. Jeffrey classified the hydrates on a structural basis, labeling not only the gas hydrates as clathrates but also the structurally related hydrates based on polyalkyl-onium salts.5 On the other hand, he labeled hydrates of simple amines, where there was an obvious hydrogen bond between guest and host, as alkylamine semiclathrates. He identified seven different clathrate structures, both actual and hypothetical (not experimentally observed), labeled as types I–VII. Davidson, on the other hand, took the view that clathrate hydrates should include only those structures where small neutral guest molecules interacted weakly with the host lattice with nondirectional interactions.4 According to his suggestion, other hydrate structures such as the ionic salt hydrates and hydrogen-bonded amine hydrates should be labeled as semiclathrates. Table 1 provides a summary of clathrate hydrates and shows a list of compounds where guest species reside

Refs

5, 29, 33 5, 30 5

in hydrate cages. Likely, the best scheme will be one based on structure as others tend to be more arbitrary. With time, schemes based on host–guest interactions become dated as the ability to detect weak interaction changes with the development of more advanced characterization techniques. In order to keep this article to a reasonable length, we do not deal with the vast number of salt hydrates unless these in some way have direct bearing on the clathrate hydrates under discussion.

2

THE CLATHRATE HYDRATES

2.1

Hydrate cages and geometries

As indicated above, the clathrate hydrates adopt a number of different structures. These are often visualized as consisting of one or more polyhedra, which share faces or edges to fill three-dimensional space. Each polyhedron has four-connected hydrogen-bonded water molecules at each vertex, and hence clathrate hydrates can be considered as ices. Some typical hydrate cages are shown in Figure 1.

51268

4151063

51264

51263

51262

512

435663

425861

Figure 1 Some of the polyhedra encountered in clathrate hydrate structures. Each vertex is a water oxygen and hydrogen atoms lie in between. Although the faces of the polyhedra are represented by planes, this is not necessarily so in actual structures.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates

3

Table 2 Cage types and geometries showing major and minor axes of the included ellipsoidal voids defining the free van der Waals volumes. Cage

˚ Major axis (A)

˚ Minor axis (A)

512 (D) 512 62 (T) 512 (D) 512 64 (H) 512 (D) 43 56 63 (D
) 512 68 (E)

4.84 6.12 4.79 6.29 4.87 4.96 8.44

4.84 5.07 4.38 6.29 4.53 4.26 6.80

Structure type

Structure I Structure II Structure H

As Jeffrey has pointed out,5 the polyhedra comply with Euler’s relationship: faces + vertices = edges + 2. There are several shorthand notations in use for the different polyhedral cages. The simplest one refers to the number of faces present in the polyhedra (U, D, T, P, H, Hp, and E) indicating 11, 12, 14, 15, 16, 17, and 20 faces respectively. More informative is the mnj notation, where n represents the number of faces of type m. For instance, the descriptor for the common dodecahedral cage is 512 , indicating that the cage has 12 pentagonal faces. Some structures may have several kinds of D or T cages, which are then distinguished by primes. The cages that make up the hydrate structures often are taken to be spherical, or nearly so. Such an assumption allows for a simplification in the treatment of guest–host interactions, as the guests also can be given spherical equivalent radii, for example, from viscosity data or virial coefficients. Unfortunately, this approach blurs the fact that guest species may have quite distinct shapes and chemical properties, leading to hydrate compounds with

Maximum deviation from the planes ˚ of the faces (A) 0.068 0.158 0.016 0.054 0.010 0.223 0.223

profound differences as well. Additionally, some clathrate hydrate cages show ellipiticity with considerable differences between the major and minor axes of the cage. Another shortcoming of such an approach is that it leads to the idea that the guests indeed behave as having gaslike properties, for example, rapid, isotropic dynamics. The incorrect nature of this idea is most easily seen when guests are confined in cages that have lower than spherical symmetry. The guest dynamics then tend to be restricted spatially, much as expected for molecules in an anisotropic potential characteristic of the solid state. Thus, instead of emphasizing the near-spherical nature of the cages and listing average cage diameters, we present the lower symmetry cages as ellipsoids with distinct major and minor axes, as shown in Table 2.

2.2

A structural overview

Three structure types, sI, sII, and sH (Figures 2–4) encompass the vast majority of guest molecules known to

Cubic structure I Cubic structure II

Pm3n

Fd3m

a = 12 Å

a = 17 Å

51262

51264

512

512

Figure 2 Cubic structure I clathrate hydrate, showing the two constituent cages, the space group, and typical lattice parameter.

Figure 3 Cubic structure II clathrate hydrate, showing the two constituent cages, the space group, and typical lattice parameter.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

4

Supramolecular materials chemistry Structure H P6/mmm

a = 12 Å c = 10 Å

51268

where there is a single tetrakaidecahedral cage consisting of a truncated octahedron (46 68 ).29 Tetrametylammonium hydroxide pentahydrate also has this structure.30 All other structures require at least two cage types in order to fill three-dimensional space. We note also that, although most structures in Table 3 have 512 cages, the symmetry of this cage changes from one structure type to another. This cage is often considered to be the key building block of clathrate hydrate structures, although there are several structures (sT, types VI, VII) that do not have any D cages.

2.3 5

Clathrate hydrates as layered structures

43 56 63

12

Figure 4 Hexagonal sH clathrate hydrate, showing the three constituent cages, the space group, and typical unit cell parameters.

form clathrates.5b Cubic structures I and II correspond to the classical gas and liquid hydrates already known in the nineteenth century,5 whereas sH,20 sT,21 and HS-122 are relatively recent discoveries. A number of recently reported single crystal structures of hydrates are as follows: sI—CH4 ,23 CO2 ,24 trimethylene oxide,25 ethane 26 ; sII—tetrahydropyran,26 benzene and Xe,26 natural gas,23 propane23 ; sH—methylcyclohexane and methane,26 and pinacolone and H2 S.27 There are various other unique clathrate structures, for instance, those of bromine28 (Figure 5), dimethyl ether (sT),21 and t-butylamine.5 The details of the structures, that is, the cages that are present in each structure and their symmetries, are given in Table 3. Several ionic hydrate structures are composed of a single cage type, namely those of the hexafluoro acids, HXF6 ·HF·5H2 O (X = P, AS, Sb), space group Im3m,

Tetragonal structure

P42/mnm a = 23 Å c = 12 Å 4P·16T·10D·172H2O

51263

51262

512

Figure 5 Tetragonal bromine clathrate hydrate, showing the three constituent cages, the space group, and lattice parameter.

A convenient way to picture some of the hydrate structures is to consider them as arising from stacked layers.32 The fundamental layer is a sheet of face-sharing pentagonal dodecahedra (D cages), Figure 6. In the case of sII hydrate, these fundamental layers, stacked in an ABCABC sequence, alternate with layers composed of D and H cages, see Figure 7. For the sH hydrate, the fundamental layers are stacked in an AA manner and the alternate layers are composed of D
and E cages (Figure 8). Other hydrates that can be described in this way include Jeffrey’s Type IV with AA stacking and a hypothetical hexagonal form of sII with ABAB stacking. One complex structure realized some years ago with choline hydroxide as the guest has elements of both sII and sH and is shown in Figure 9.34 The structure obtained is trigonal R3 with unit-cell parameters ˚ and c = 90.525(2) A. ˚ Choline is located a = 12.533(1) A in both of the large cage types, E and H, present in the lattice. In the E cage, it does not interact with the cage by hydrogen bonding, as there are only van der Waals contacts between the guest and water molecules. The H cavity is too small to accommodate the choline molecule completely in a hydrophobic way, and the guest hydroxyl displaces one of the host-lattice water molecules, forming hydrogen bonds with the framework. The closed-cage version of this structure, where guests are not hydrogen bonded to the cages and the D and D
cages are complete, forms yet another structure that possibly may be found naturally. The space group of this idealized structure is R3m with unit cell param˚ c = 90.525 A, ˚ γ = 120◦ , and compoeters a = 12.533 A,
sition E × 4H × 2D × 11D × 102H2 O. The layer stacking sequence is CABBCAABCCABBCAABC. All of these structures can be considered as polytypes. It is clear that a variety of other hypothetical structures can be formed by modifying the stacking arrangement.

2.4

Stability

The region of stability for a specific clathrate is usually defined by a phase diagram. The most

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates Table 3

Clathrate hydrate structures.

Structure type

Unit cell content space group

Typical unit cell parameters parametersa

Cage

Symmetry of cages

Structure I

2D·6T·46H2 O Pm3n

˚ 11.9 A

512 (D) 512 62 (T)

m3 4 2m

5, 24–26 31

Structure II

16D·8H·136H2 O Fd 3m

˚ 17.0 A

512 (D) 512 64 (H)

3m 4 3m

5, 23 26

Structure H

E·3D·2D
·34H2 O P 6/mmm

˚ a = 12.2 A ˚ c = 10.0 A

3 56 63 (D
)

512 (D)

4 512 68 (E)

mmm 6 m2 6/mmm

20, 26 27

3E·12H·6D
·33D· 306H2 O R3m

˚ a = 12.5 A ˚ c = 90 A α = β = 90o γ = 120◦ ˚ a = 35 A ˚ c = 12.4 A

512 (D) 4 5 6 (D
) 512 64 (H) 512 68 (E)

3, 3m 3m 3m, 3m 3m

34

42 58 61 (U) 512 62 (T) 512 63 (P) 1 4 510 63 (T
)

2,2,1 2,2,1 3,3,1,1 1,1,1,1

21

mmm,m m,m m2m,m

28

Structure II/H polytype (ideal form) Structure T

12U·12T·24T
· 12P·348H2 O P 321

3 6 3

Refs

2DA ·8DB ·8TA ·8TB ·4P·172H2 O P 42 /mmm

˚ a = 23 A ˚ c = 12.1 A

512 (D) 512 62 (T) 512 63 (P)

Hexafluoroacidsc

2T·2HF·10H2 O Im3m

˚ 7.7–7.8 A

68 46 (T)

29

t-butylamined

16Hp·156H2 O I 43d

˚ 18.8 A

43 59 62 73 (Hp)

5

3D·2T·2P·40H2 O P6/mmm

˚ a = 11.2 A ˚ c = 11.5 A

512 (D) 512 62 (T) 512 63 (P)

22

Bromine hydrateb

HS−1e

5

angles are 90◦ unless indicated otherwise. Jeffrey’s type III. c Jeffrey’s type VII. d Jeffrey’s type VI. e Related to Jeffrey’s type IV (Ref. 5). a All b

Layer of dodecahedral cages

Layer of D (512) cavities

Cubic structure II Layer of H (51264) cavities and D (512) cavities

Figure 6 A sheet of face-sharing pentagonal dodecahedra. Different stacking arrangements give rise to a variety of hydrate structures, for example, cubic sII, hexagonal sH, and so on.

Figure 7 polytype.

Cubic sII clathrate hydrate, represented as a layer

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

6

Supramolecular materials chemistry

Layer of D (512) cavities

90.5 Å

Layer of D (512) cavities

Structure H

Layer of E (51268) cavities and D′ (435663) cavities Layer of E (51268) cavities and D′ (435663) cavities

Figure 8 polytype.

Hexagonal sH clathrate hydrate represented as a layer

Layer of H (51264) cavities and D′ (512) cavities

general representation is a three-dimensional, pressure–temperature–composition (P–T –X) diagram, although for convenience it is usual to work with two-dimensional slices (P–T or T –X) through the threedimensional representation. For gaseous guests, the usual presentation is that of a P–T phase equilibrium line where the hydrate phase is in equilibrium with water or ice plus gaseous guest. Phase equilibria for methane hydrate are shown in Figure 10, both as found in terrestrial as well as marine settings.

Figure 9 Ideal version of a clathrate hydrate of choline hydroxide polytype layer representation.

Phase equilibria for gas mixtures, such as natural gas, depend on the gas composition, an area well studied because of the importance of controlling gas hydrate plugs in oil and gas pipelines.9 Hydrates respond to various materials used to lower the freezing point of ice in the same way, that is, the stability region is shifted to lower

0

600

Methanehydrate stability zone

Base of permafrost zone

Methanehydrate phase boundary

800 1000

Base of gashydrate

1200

1600

l

1400

−20

−10

400

10

20

30

Hydrothermal gradient Methanehydrate stability zone Methanehydrate phase boundary

600 800 Sea floor

1000

Water sediment

1200 1400 1600

0

Temperature (°C)

Figure 10

Depth below sea surface (m)

400

200

Geothermal gradient in permafrost

a erm oth t Ge dien gra

Depth below ground surface (m)

200

0

Gashydrate stability zone

Base of gashydrate

−20

−10

0

10

20

30

Temperature (°C)

Phase equilibrium of cubic sI methane hydrate. Both terrestrial and marine versions of the phase diagram are shown.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates

10

0

20

Mole % 30 40 50 60

3

l1 + l 2 + g

Temperature (°C)

−10

I + l1 + g −20

−30

hII + l1 + g hII + l2 + g

I+ hII + g hT + hII + g

h T + l2 + g

I + hT + g −40

0 10

20

30

40

50

60

70

80

90 100

Weight % DME

Figure 11 Temperature–composition diagram of dimethylether hydrates. There are two hydrates, a cubic sII, DME·17H2 O, and sT, DME·8H2 O, both of which melt incongruently.

Table 4

3.1

Neutral species

The guest species that are suitable for clathrate hydrate cages cover a wide spectrum of chemical functionalities: from hydrogen and the noble gas atoms to molecules ˚ Table 5 with a van der Waals diameter up to ∼9 A. shows a classification scheme with representative examples. Although the important guest–host interactions are often taken to be hydrophobic, such as for the alkanes that have a very low solubility in water, guests may also be completely miscible with water, such as some of the ethers and ketones. A useful rule of thumb is that water molecules must interact with each other preferentially as compared to interacting with the potential guest. As described earlier, the main structural families are those of sI, sII, and sH, with a few structures that are unique to a particular guest. Some guests that fit tightly into the large cage of sII require a “helpgas” to fill some of the small cages in order to be stable. All of the larger molecules that form sH also require a “helpgas” to occupy at least some fraction of the two small cages in the structure.

Decomposition temperatures for some clathrates.a

Guest Ar Kr Xe CO2 CH4 C3 H8 CH3 Cl CH3 SH CBrClF2 a

GUEST SPECIES

80 100

l2 l1

7

Structure II II I I I II I I II

T (K) at P = 100 kPa 149 223.4 262.8 218.2 194.5 261.6 280.7 283.2 280.8

From Ref. 4.

temperatures at constant pressure. Classical antifreezes for hydrate control are methanol, glycol, and solutions of common salts. For guests that are liquid at room temperature, the phase behavior may be followed from a T –X diagram (Figure 11), giving information on congruent or incongruent melting, the hydrate composition, and the eutectic temperatures. In addition to temperature and pressure, the stability of the clathrate hydrate phase depends strongly on the size and chemical nature of the guest species. Table 4 shows the decomposition temperature at a pressure of 100 kPa of the clathrate hydrates with a number of typical guest molecules. The decomposition temperature at this pressure spans a temperature range of 130 K.

3.2

Ionic guests

Clathrate hydrates with ionic guest species are examples where hydrophobic interactions do not appear to be important in directing the structure to one of the classical clathrate hydrate motifs. In general, ionic species have received far less attention than the neutral species mentioned in the previous section. Mootz et al.31 reported that a number of strong acids such as perchloric, tetrafluoroboric, and hexafluorophosphoric acids all formed hydrates that had the sI structural motif. For the former two guests, both cages were occupied, for the latter, only the large cages. The proton was then assumed to reside in the water lattice, giving a positively charged host lattice with anions in the cages. Another group of guests, HXF6 , X = P, As, Sb, together with HF form a unique clathrate structure, again with the anions in cages and protons as well as HF in the lattice.29 The cages are the equivalent of sodalite cages (46 68 ), as found in zeolites, and are able to fill three-dimensional space without participation by any other cage type. Initially, these materials were reported as hexahydrates, although Davidson and Garg33 pointed out that it was not possible to have a fully four-connected lattice of water molecules with a sodalite cage. By substituting one HF for every water molecule, apparently this does become a possibility. Tetramethylammonium hydroxide pentahydrate forms much the same structure as the hexafluoro acids as based on sodalite cages, but with cations in the cages and a hydroxide

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

8

Supramolecular materials chemistry Table 5

Neutral guest species.

Guest type Inorganic Monatomics—noble gases Diatomics Triatomics Larger inorganics Organics Hydrocarbons Alkanes, alkenes, alkynes Cycloalkanes, aromatics Fluorinated hydrocarbons

Other halogenated hydrocarbons Mercaptans and sulfides Ethers Ketones Aldehydes Esters Amines Other heterocycles

ion substituting for one water molecule in the framework.5 Mootz and coworkers reported several other clathrates of tetramethylammonium hydroxide, a decahydrate and a heptahydrate, as well as some nonclathrate hydrates.30

3.3

Guest size–structure relationship

3.3.1 Hydrates with a single guest The main criterion that determines the structure of clathrate hydrates is the size of the guests present. As the hydrate lattice is stabilized through weak guest–host interactions by the presence of guest molecules, these must interact reasonably closely with the cage walls. Although often hydrate cages are described or approximated as being spherical, this clearly is a poor approximation for cages with lower symmetry, as seen in a previous section. Most of the hydrates with a single guest fall into the two large structural families: sI and sII. There is a correlation of guest size with structure, as seen in Figure 12. The smallest guests (H2 , Ar, Kr, N2 , and O2 ) form sII hydrate with both large and small cages filled by a single guest and hydration numbers of ∼5.8–6.0 waters/guests. Atoms or molecules such as Xe, H2 S, methane, and CO form sI hydrates with most of the large cages and about 60–80% of the small cages filled, giving hydration numbers close to ∼6 waters/guests. As the guest size increases, the occupancy of the small cage decreases until it is essentially empty

Examples

Ar, Kr, Xe H2 , O2 , N2 , CO, Cl2 , Br2 , BrCl CO2 , H2 S, COS, N2 O, H2 Se, SO2 , ClO2 PH3 , AsH3 , SF6 , SeF6 , SO2 F2 , ClO3 F, NF3 CH4 , C2 Hn (CH3 )6−n , n = 0–6 (CH2 )n , n = 3–8 Methylcycloalkanes, adamantane CHn F4−n , n = 0–3, C2 Hn F6−n , n = 0–5, some partially fluorinated propanes, cycloalkanes CHn Cl4−n , n = 0–3, CFCs, CH3 Br, EtBr RSH (R = CH3 , Et), (CH3 )2 S, (CH2 )n S, n = 2–4 (CH3 )2 O, (CH2 )n O, n = 2–5, 1,3-dioxolane, p, m-dioxane, propylene oxide, t-BuOCH3 Acetone, cyclobutanone, pinacolone Formaldehyde, acetaldehyde Butyrolactone t-Butylamine Isoxazole, isothiazole

for guests such as CH3 Br and the hydration numbers then increase to ∼7–8 waters/guests. For the next size range, there is a transition zone, where guests may form either the sI or the sII hydrate with the guest only occupying the large cage in each structure. Guests for which both structures are known include TMO and cyclopropane, and hydration numbers are either close to 7 for sI or 17 for sII. Larger guests in sizes up to approximately isobutane and cyclobutanone all form the sII hydrate with a hydration number close to 17. It should be noted that the hydration numbers derived from the unit cell data (Table 3) are the ideal maximum values for complete occupancy of all cages. Experimental hydration numbers lie between these and the minimum values required for lattice stability. Such minimum values need to be determined under carefully controlled equilibrium conditions at an invariant point in the phase diagram, although this is seldom done in practice. However, one must expect that cage occupancies and, therefore, the composition of hydrates will change with guest pressure35 or with guest concentration for water-miscible systems.36 The sII hydrates of a single guest type appear to be an exception as most measurements have shown compositions to be very close to 17 waters/guests, a near-stoichiometric filling of the large cages. Several small molecules do not give hydrates, for example, methanol and methyl formate, although either will fit into sI or sII cages. As mentioned before, several other

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates

4

Ar Kr N2,O2 CO, CH4 Xe, H2S

4

IIa 5 2/3 5 3/4 Ia

Largest van der Waals diameter (Å)

4.1 5

CO2 Ethane, ethylene

I or II II

Ib

II 6

7

8

Cyclopropane cyclobutane

6 2/3

Cyclopentane propane

IIb

H

17 Isobutane n-butane (g), neopentane IIc No Hyd. benzene, cyclohexane Adamantane cyclooctane, methyl butanes

Ha No Hyd.

Methyl pentanes (g) methylcyclohexane dimethylcyclohexane

Figure 12 Guest size–structure relationship for sI, sII, and sH clathrate hydrates.

guests, such as bromine, dimethyl ether, and t-butylamine give unique clathrate structures. The principles directing these unique structures remain unknown as each of these guests is also capable of forming classical clathrate structures as far as the guest size is concerned.

9

THEORETICAL DESCRIPTION, GUEST–HOST INTERACTIONS, AND MODELING Introduction

There are two main theoretical/computational approaches used to describe the clathrate hydrate structure and stability for different guest molecules. The first approach is based on the work of van der Waals and Platteeuw3 and uses thermodynamics and statistical mechanical modeling to determine hydrate stability for a variety of guest molecules, and hydrate temperature/pressure conditions. This approach leads to mathematical expressions, which include parameters that are assigned by empirically fitting the predictions of the theory to experimental guest occupancies or hydrate phase stability boundaries. The second approach uses classical molecular simulation techniques to study the structure and stability of clathrate hydrate phases. Quantum mechanical molecular simulation techniques are not considered in this review. Canonical (NVT ) or isothermal–isobaric (NPT) molecular dynamics (MD) simulations are used to determine the structure of hydrate phases with known guest occupancies in the cages. In molecular simulations, knowledge of intra- and intermolecular potential interactions in the system is needed to solve Newton’s equations of motion for the hydrate system. Free-energy MD simulations and grand canonical Monte Carlo (GCMC) simulations can be used to determine the guest occupancies, adsorption isotherms, and phase boundaries under different temperature and pressure conditions. Each approach has advantages and disadvantages, which are summarized below.

3.3.2 Hydrates with two guests The guest size–structure relationships are somewhat complex if more than one guest is present. Figure 12 divides guests into a number of subgroups. When guests from one subgroup are combined with others from the same subgroup to form a hydrate, the same structure results. However, when a guest from subgroup Ia is combined with one from subgroup Ib, either sI or sII may form, depending on P and T conditions. When guests from the subgroup IIa or Ia are combined with guests from subgroups IIb or IIc, sII hydrates form. For these, there is considerable partitioning of large guests in large cages and small guests in small cages. When guests from subgroup Ia or IIa are combined with guests from subgroup Ha, hydrates of sH result. In cases where small guest molecules provide an essential stabilizing function, such as for the larger sII and sH guests, the small guest molecules are referred to as “helpgas” molecules.

4.2

The statistical mechanical “solid solution” theory

In 1958, van der Waals and Platteeuw developed a statistical mechanical theory for predicting the stability region of the clathrate hydrate phase with different guest molecules under different temperature and pressure conditions. The van der Waals–Platteeuw (vdWP) theory is based on the thermodynamic condition of equilibrium between a hydrate phase water/ice (β) phase and i guest species encapsulated in the hydrate at the phase boundary:
Hyd µW (T · p) = µBW (T · p) + µi (T · p) (1) i·guests

The specific form of this general formula depends on the pressure/temperature conditions that determine the aqueous/ice phase in contact with the hydrate and the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

10

Supramolecular materials chemistry

gas/(nonaqueous liquid)/(aqueous solution) state of the guest(s). Depending on the mole ratio of water to guest in this mixture, excess gas-phase or liquid-phase guests may be present at equilibrium. At appropriate low temperatures, clathrate hydrates can be synthesized by compressing liquid or gaseous guests with water or ice. In the vdWP theory, the chemical potentials of the hydrate phase, aqueous solution/water/ice, and guest species are written in the grand canonical ensemble formalism. The pure solid ice, nonaqueous liquid, aqueous solution, and gas phases present in the system are treated with statistical mechanical formalisms appropriate for these phases. In the vdWP theory, the chemical potential of the hydrate phase is written in a form similar to that of the Lennard–Jones–Devonshire cage theory of liquids where the hydrate cage (cavity) structures encapsulate the guest molecules. The original form of the vdWP theory makes a number of simplifying assumptions, namely, 1. 2.

3. 4.

The guest molecules in the cages do not distort the host lattice. The guest molecules are located in cages, each of which holds a single guest. More than one type of cage is possible in the hydrate structure. Interactions between guest molecules in adjacent cages are neglected. Classical statistical mechanics applies to the analysis of the hydrate system.

These assumptions lead to the following condition for phase equilibrium between the hydrate phase and the water/guest phase(s), Hyd−B µW (T

· p) = −RT




 ν m 1n 1 −

m




 θ mj 

(2)

j

where ν m is the number of cavities of type m per water molecule and θ mj is the fractional occupancy of guest species j in cage type m. Assuming a single guest in each cage, the fractional guest occupancy is given as θ mj (T , P , y) =

Cmj (T )fj (T , P , y) n comp 1+ Cmi (T )

(3)

i=1

where Cmj is the Langmuir coefficient and fj (T , P , y) is the fugacity of species j . This is an expression for the Langmuir adsorption isotherm of the hydrate phase that gives the fractional occupancy of the cages as a function of the activity of the guest molecule in nonhydrate phases. The Langmuir constant can be calculated on the basis of the “free volume” of the guest inside the hydrate cage. For a

spherically symmetric cage–guest potential, the Langmuir constant is given as Cmj (T ) =

4π kT





Rm

exp o

−Wmj (r) 2 r dr kT

(4)

The integral is over the guest–cage effective potential Wmj (r) within the spherical confines of the hydrate cage m, determined by the effective radius Rm . This form assumes the guest geometry is spherical or the guest rotates isotropically inside the cage and has an effective spherical potential in the cage. In the case of nonspherical clathrate hydrate cages, this assumption is often not correct. This is reflected in the nonisotropic lineshape of 13 C NMR spectra of nonspherical guests (such as CO2 ) in sI large cages and observations from MD simulations of these clathrate hydrates. The guest–cage potential W (r) is often taken to have the Kihara potential form,

φ(r) =

  ∞   4ε

σ − 2a r − 2a

12

 −

σ − 2a r − 2a

6 

for r ≤ 2a for r > 2a

(5) where ε is the minimum of the potential well, σ is the soft core radius, and a is the hard core radius of the potential. The parameters of the guest–cage potential are often determined empirically by fitting the predicted phase diagram of the hydrate to the experimental phase boundary. Significant discrepancies between the values of the cage water and guest potential parameters from the vdWP theory and transferable potentials of the guests used in molecular dynamics simulations are often seen.37 This can reflect the fact that the assumptions made in deriving (4) are approximate, and fitting of the experimental results by the potential parameters compensates for this. None of the four assumptions listed above are essential to the statistical mechanical theory of hydrate equilibrium and many generalizations have been introduced that allow extensions of the scope of the theory. Multiple guest cage occupancies,38, 39 harmonic and anharmonic oscillations of water molecules about their equilibrium positions in the hydrate,40 nonspherical guest molecules,41 size dependence of the lattice constants on the guest,42 guest–guest interactions,6 and guest–host interactions43 have all been introduced to extend the scope of the theory. The vdWP theory is the basis for software programs, such as CSMGem44 and GasHyDyn for predicting hydrate equilibria. When applied to cases where the approximations of the vdWP theory reflect the nature of the clathrate system, the theory is successful in predicting hydrate stability conditions and phase boundaries.45

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates

4.3

11

Computational approaches

With advances in molecular dynamics and Monte Carlo simulation methodology and ever increasing speed and availability of computers, molecular simulations have become alternatives for studying hydrate phase equilibria. Canonical and isothermal–isobaric MD simulations are, by nature, carried out on closed systems with fixed numbers of guest molecules in the clathrate hydrate cages. In the most general form of molecular simulations, no specific assumptions on the nature of the cage–guest interactions need to be made. These MD simulations are performed to determine molecular level structure and dynamics in the clathrate hydrate phase. To compliment NPT MD simulations, free-energy MD or GCMC simulations are performed to calculate guest occupancies in cages (adsorption isotherms) and hydrate temperature/pressure stability regions.46 Thermodynamic integration in MD simulations is used to determine the free energies associated with different guest occupancies under given temperature/pressure conditions. Grand canonical Monte Carlo simulations use guest insertion, deletion, and displacement moves in the hydrate cages to generate configurations of guests consistent with the grand canonical ensemble partition function under the given temperature/pressure conditions. A limitation of GCMC simulations is that the hydrate water framework is often frozen during the process, which is problematic for cases where the guests interact strongly with the cage water molecules, see below. In the case of clathrate hydrate phases of simple spherical guests such as methane or small nonspherical, but quickly rotating, guests such as ethane and propane, the radial guest–cage assumption commonly used in the vdWP theory can be valid if these guests are in the quasi-spherical hydrate cages, such as the large H cages of the sII hydrate. However, the assumption of the spherical guest–cage potential is not valid for nonspherical cages such as the large oblate T cages of sI and prolate E cages of sH clathrates or the small cages in sII and sH clathrates. Even relatively small molecules such as CO2 do not have spherical potentials in these cages as is evidenced by the solid-state 13 C NMR spectra47 and the MD simulations of these hydrates.48 Further complications to the application of the vdWP theory are guest molecules with conformational flexibility such as n-butane,49 1-propanol,50 tetrahydropyran,26 and cyclooctane.51 X-ray crystal structure measurements and MD simulations49, 50, 52 show that the stable conformation of the n-butane and 1-propanol guest molecules in the sII clathrate hydrate large cages are gauche and not anti, which is the most stable in the gas phase. In these cases, it becomes difficult to assign physical significance to the values of Kihara potential parameters obtained by fitting

Figure 13 A snapshot of a large cage of an MD simulation of the binary CO2 + ethanol sI clathrate hydrate. The ethanol molecule is hydrogen bonded to two cage water molecules.

the predictions of the vdWP theory to the phase boundaries of these guests or adsorption isotherms. Clathrate hydrates, where the guest molecules interact strongly (specifically, through hydrogen bonding) with the cage water molecules, are also problematic for the vdWP theory. In MD simulations, we observe that hydrogenbonding guests behave very differently from their nonhydrogen-bonding analogs. For example, in the pairs THF and cyclopentane,53 TBME and neohexane,54 and 1-propanol and n-butane,52 the guests with ether or alcohol functional groups hydrogen bond with the cage water molecules, introduce Bjerrum L-defects in the water hydrogen bonding lattice, perturb the water hydrogen bond network, and destabilize the hydrate phase. Hydrogen bonding affects the guest dynamics in the cages considerably. In MD simulations, no additional assumptions are needed to account for hydrogen bonding guests, and different guests can be treated with similar levels of approximation. A snapshot of a guest–water hydrogen bond in the binary structure I ethanol + CO2 55 clathrate hydrate is shown in Figure 13. This clathrate hydrate has approximately 10% of the large sI T cages occupied by ethanol and the others with CO2 . An ethanol molecule, which is hydrogen bonded to water molecules in the large cage, is shown. The ethanol forms one proton-accepting and one protondonating hydrogen bond with the cage water molecules (from a large T cage hexagonal and pentagonal faces, respectively). The water molecules hydrogen bonded with ethanol initially formed a pentagonal face, separating a CO2 -containing small sI D cage from a CO2 -containing large sI T cage. The hydrogen-bonding ethanol guest perturbs its cage and other neighboring cages in the hydrate water framework. The perturbation of the cage holding the guest can, in principle, be accounted for in the vdWP theory. However, the perturbation of the neighboring cage structures is more difficult to implement into the vdWP theory. The general nature of the MD and GCMC simulations, which does not require making different assumptions for

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

12

Supramolecular materials chemistry

different types of guest molecules, makes them a necessary tool for predicting the properties of clathrate hydrates with flexible guest molecules, nonspherical cages, and hydrogenbonding guests. The details of the structure of hydrates for different guests warn against the indiscriminate use of a single statistical mechanical formalism for guest species for which the assumptions are not necessarily valid. In particular, one must be careful not to treat hydrogenbonding guests within the same level of approximation in the vdWP theory as structurally similar, but nonhydrogen-bonding guest analogs. For example, effective Kihara potential parameters may be introduced to describe the hydrate phases of these guests, but the qualitatively different nature of the guest–host interactions among these two classes of guest makes this procedure fraught with danger. The strength of the vdWP theory is its ease of use and the speed with which predictions of the clathrate hydrate phase diagram can be generated. If the scope of validity of this theory is recognized, it can be a valuable tool in the hand of researchers in predicting hydrate stability conditions.

5

5.1

CLATHRATE HYDRATE PREPARATION AND CHARACTERIZATION Preparation and sample quality

The challenge in preparing gas hydrates is that in many cases the guest and host phases are not mutually soluble to a sufficient degree so that direct and efficient conversion can take place when the two bulk phases are placed in contact. Thus, various kinds of mixing are often part of the hydrate preparation process, or if ice is used, grinding to expose fresh, unreacted surfaces to guest material. It also is more difficult than might be apparent initially to obtain pure hydrate phases that are not contaminated by an excess of one of the components.56 It is even more difficult to obtain a pure sample of hydrate that is actually homogeneous with a composition characteristic of equilibrium conditions. A key concept to remember is that crystallization, as occurs during clathrate formation, is not an equilibrium process. There is no guarantee that the phase that appears on nucleation (kinetic control) will be the expected phase or that the composition will be homogeneous and characteristic of the thermodynamic conditions. For some measurements, it is not important whether one component or the other (guest or host) is present in excess; however, for others it is imperative that the sample is a pure hydrate with a composition that reflects equilibrium conditions. The most demanding samples are those to be used for thermodynamic

measurements, for example, heat capacity and thermal conductivity, which require high sample purity, knowledge of the composition, and preparation at a quadruple point.57 Very few samples receive this much attention; however, all researchers should be aware of the variables that affect sample quality and homogeneity.

5.2

Characterization of clathrate hydrates

The complete characterization of clathrate hydrates is a difficult task.58 The materials are crystalline; however, they are nonstoichiometric. Sample handling must always be carried out where hydrate decomposition is negligible, for example, at low temperatures or at pressures where the hydrate is stable. A complete description of a hydrate requires knowledge of the structure, composition, and distribution of the guests over different cages and phase equilibrium data. Although the space group and unit cell dimensions can be determined in a straightforward way from powder X-ray diffraction, the stoichiometry can only be obtained by more specialized diffraction techniques.58 Because of the low data-toparameter ratio and the fact that thermal parameters are closely coupled to site occupancies, it is difficult to apply standard techniques such as Rietveld analysis to try and obtain complete structures from powder data unless a number of assumptions are made, for example, centering the guests in the hydrate cavities. Recently, direct methods applied to powder data have proved to give results in reasonable agreement with single crystal data.58 Sometimes, single crystals can be found in recovered natural samples; however, there may be many different guest molecules present.23 With careful analysis of high quality data, three or four guests that coexist in cages may be determined with perhaps 5% accuracy. With some patience, the preparation of single crystals of synthetic hydrates for diffraction is straightforward, as powder samples in a closed environment will slowly anneal and ripen to give larger crystals. These eventually may become suitable for singlecrystal diffraction. Annealing is especially important for samples prepared from ice as it can take weeks or months to eliminate defects incorporated during sample preparation. Spectroscopic methods such as FT-IR,59, 60 Raman,61, 62 and solid-state NMR63–65 have proven to be valuable for the characterization of hydrates. One of the most useful features of these methods is that the presence of a guest molecule in a hydrate cage gives a measurable shift of the characteristic spectral frequencies from those in the bulk phase. For instance, the isotropic chemical shifts for many small molecules with NMR-active nuclei show easily distinguishable resonances for guests in different cages.66

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates The chemical shift–cage size relationship is understood in terms of collision-induced deshielding, and has been particularly well worked out for xenon.67 The anisotropic chemical shift, usually observable for spin 1/2 nuclei also, is sensitive to cage symmetry, thus giving a useful way of identifying both the structure and the relative cage occupancies of hydrates.48, 63–65 In the case of larger, flexible guest molecules matters are somewhat more complicated as molecular conformations can be different from those in the bulk phases.49, 68 Raman and IR techniques are also useful in that some vibrational frequencies of the guest molecule are sensitive to the nature of the cage. Raman is particularly straightforward to apply because of the simple scattering geometry that can be employed60, 61 However, the interpretation of spectral frequencies is not necessarily straightforward, and band intensities are not direct quantitative measurements of the number of scattering species as account must be taken of scattering cross sections.61 The geometry for IR measurements is less convenient as these have to be made either with transmission or reflection techniques.60, 61

13

Other areas of current interest include the microscopic understanding of hydrate processes such as hydrate formation (nucleation and growth),72, 73 hydrate decomposition, and self-preservation.74

7

CONCLUSION

In this chapter, we have summarized some recent advances in the understanding of clathrate hydrate structure and dynamics. It is becoming increasingly clear that details of the geometrical structures of the cages in the different clathrate hydrates and the specific molecular nature of the guest molecules play important roles in determining hydrate stability and dynamics. The traditional picture of clathrate hydrates as having quasi-spherical cages occupied by inert guest molecules that interact with the cage with a quasi-spherical potential does not capture many aspects of clathrate behavior.

REFERENCES 6

FUTURE DIRECTIONS

Despite the fact that clathrate hydrate science can be considered to be a mature field, numerous challenges remain. In part, this is because hydrates exist in nature and hence exhibit the complexity of mineral phases, where the structure and composition reflect not only the conditions of synthesis but also subsequent exposure to changing conditions after synthesis. Another part is that the principal component of these materials is water, likely the most important material on earth, and the clathrates offer a possibility to study the diverse ways in which water interacts with other materials. In this respect, it is interesting to note that new structures continue to be found. Structurally related families of materials such as the silica-based zeosils and clathrasils tend to show a far greater structural diversity.69 In part, this is because of the covalent nature of the silica lattices that give far more robust frameworks than the water-based ones; however, each structure in these classes of materials could be considered as a candidate for clathrate synthesis. Further, following up on Jeffrey’s early work5 on amine hydrates (amine semiclathrates), numerous amines remain to be explored for hydrate phases. In particular, since it has also been observed that some amine clathrates70 as well as stoichiometric alcohol71 hydrates show phase transitions to double clathrate hydrates in the presence of helpgases such as xenon, H2 S, and methane,27 it becomes a worthwhile area for exploration.

1. H. Davy, Philos. Trans. R. Soc. Lond., 1811, 101, 1. 2. (a) W. F. Clausen, J. Chem. Phys., 1951, 19, 259, 662; (b) M. von Stackelberg and H. R. Mueller, J. Chem. Phys., 1951, 19, 1319; (c) W. F. Clausen, J. Chem. Phys., 1951, 19, 1425; (d) L. Pauling and R. E. Marsh, Proc. Natl. Acad. Sci. U. S. A., 1952, 38, 112. 3. J. H. van der Waals and J. C. Platteeuw, Adv. Chem. Phys., 1958, 2, 1–57. 4. D. W. Davidson, in Water. A Comprehensive Treatise, ed. F. Franks, Plenum Press, New York, 1973, vol. 2, p. 115. 5. (a) G. A. Jeffrey, in Inclusion Compounds, eds., J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 1, 135–185; (b) G. A. Jeffrey in Comprehensive Supramolecular Chemistry, eds., D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, 1996, vol. 6, pp. 757–788. 6. (a) D. W. Davidson and J. A. Ripmeester, in Inclusion Compounds, eds., J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 3, pp. 69–123; (b) J. A. Ripmeester and C. I. Ratcliffe, in Inclusion Compounds, eds., J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Oxford University Press, Oxford, 1991, vol. 5, pp. 37–85. 7. Y. A. Dyadin and V. R. Belosludov, in Comprehensive Supramolecular Chemistry, eds., D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, 1996, vol. 6, pp. 789–824. 8. E. G. Hammerschmidt, Ind. Eng. Chem., 1934, 26, 851. 9. E. D. Sloan Jr and C. A. Koh, Clathrate Hydrates of Natural Gases, 3rd edn, CRC Press, 2007. 10. Y. F. Makogon, F. A. Trebin, A. A. Trofimuk, et al., Dokl. Acad. Sci. U.S.S.R. Earth Sci. Sec., 1972, 196, 197–200.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

14

Supramolecular materials chemistry

11. A. V. Milkov, Earth Sci. Rev., 2004, 66, 183–197.

36. D. N. Glew and N. S. Rath, J. Chem. Phys., 1966, 44, 1710.

12. M. Maslin, M. Owen, R. Betts, et al., Philos. Transact. A Math. Phys. Eng. Sci., 2010, 368, 2369.

37. Z. Cao, J. W. Tester, K. A. Sparks, and B. L. Trout, J. Phys. Chem. B, 2001, 105, 10950–10960.

13. F. M. O’Connor, O. Boucher, N. Gedney, et al., Rev. Geophys., 2010, 48, RG4005, doi:10.1029/2010RG000326.

38. J. B. Klauda and S. I. Sandler, Chem. Eng. Sci., 2003, 58, 27–41.

14. (a) J.-S. Gudmundsson, M. Parlaktuna, and A. A. Khokhar, SPE Prod. Fac., 1994, 9, 69–73; (b) Y. Seo, J.-W. Lee, R. Kumar, et al., Chem. Asian J., 2009, 8, 1266–1274.

39. H. Tanaka, T. Nakatsuka, and K. Koga, J. Chem. Phys., 2004, 121, 5488–5493. 40. H. Tanaka, J. Chem. Phys., 1993, 98, 4098–4109.

15. (a) W. L. Mao, H. Mao, A. F. Goncharov, et al., Science, 2002, 297, 2247; (b) L. J. Florusse, C. J. Peters, J. Schoonman, et al., Science, 2004, 306, 469–471; (c) H. Lee, J. Lee, Y. D. Kim, et al., Nature, 2005, 434, 743; (d) T. Sugahara, J. C. Haag, P. S. R. Prasad, et al., J. Am. Chem. Soc., 2009, 131, 14616–14617.

43. V. R. Belosludov, O. S. Subbotin, D. S. Krupskii, et al., J. Phys. Conf. Ser., 2006, 29, 1–7.

16. P. Linga, A. Adeyemo, and P. Englezos, Environ. Sci. Technol., 2008, 42, 315–320.

44. L. Ballard and E. D. Sloan, Jr, Fluid Phase Equil., 2004, 216, 257–270.

17. J. I. Lunine and D. J. Stevenson, Icarus, 1987, 70, 61–77.

45. W. R. Parrish and J. M. Prausnitz, Ind. Eng. Chem. Proc. Des. Dev., 1972, 11, 26–35.

18. J. S. Loveday, R. J. Nelmes, D. D. Klug, et al., Can. J. Phys., 2003, 81, 539–544. 19. H. M. Powell, J. Chem. Soc. (Lond.), 1948, 61. 20. (a) J. A. Ripmeester, J. S Tse, C. I Ratcliffe, and B. M. Powell. Nature, 1987, 325, 135; (b) K. A. Udachin, C. I. Ratcliffe, G. D. Enright, and J. A. Ripmeester, Supramol. Chem., 1997, 8, 173. 21. K. A. Udachin, C. I. Ratcliffe, and J. A. Ripmeester, Angew. Chem. Int. Ed., 2001, 40, 1303–1305. 22. L. Yang, C. A. Tulk, D. D. Klug, et al., Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 6060–6064. 23. K. A. Udachin, H. Lu, G. D. Enright, et al., Angew. Chem. Int. Ed., 2007, 46, 8220–8222. 24. K. A. Udachin, C. I. Ratcliffe, and J. A. Ripmeester, J. Phys. Chem. B, 2001, 105, 4200–4204. 25. K. A. Udachin, C. I. Ratcliffe, and J. A. Ripmeester, J. Phys. Chem. B, 2007, 111, 11366–11372. 26. K. A. Udachin, C. I. Ratcliffe, and J. A. Ripmeester, J. Supramol. Chem., 2003, 2, 405–408. 27. S. Alavi, K. A. Udachin, and J. A. Ripmeester, Chem. Eur. J., 2010, 16, 1017–1025. 28. K. A. Udachin, G. D. Enright, C. I. Ratcliffe, J. A. Ripmeester, J. Am. Chem. Soc., 1997, 11481–11486.

and 119,

29. D. W. Davidson, L. W. Calvert, F. Lee, Ripmeester, Inorg. Chem., 1981, 20, 2014.

J. A.

and

30. D. Mootz and R. Seidel, J. Incl. Phenom. Mol. Recogn. Chem., 1990, 8, 139–157. 31. D. Mootz, E.-J. Oellers, and M. Wiebcke, J. Am. Chem. Soc., 1987, 109, 1200. 32. E. A. Smelik and H. E. King Jr, Z. Kristallogr., 1996, 211, 84–89. 33. D. W. Davidson and S. K. Garg, Can. J. Chem., 1972, 50, 3515–3520. 34. K. A. Udachin and J. A. Ripmeester, Nature, 1999, 397, 420–423. 35. G. H. Cady, J. Phys. Chem., 1983, 87, 4437–4441.

41. H. Tanaka, J. Chem. Phys., 1993, 101, 10833–10842. 42. J. B. Klauda and S. I. Sandler, Ind. Eng. Chem. Res., 2000,39, 3377–3386.

46. Y. Okano and K. Yasuoka, J. Chem. Phys., 2006, 124, 024510. 47. C. I. Ratcliffe and J. A. Ripmeester J. Phys. Chem., 1986, 90, 1259–1263. 48. (a) S. Alavi, P. Dornan, and T. K. Woo, ChemPhysChem, 2008 9, 911–919; (b) H. Mohammadi-Manesh, S. Alavi, T. K. Woo, et al., Phys. Chem. Chem. Phys., 2009, 11, 8821–8828. 49. M. Luzi, J. M. Schicks, R. Naumann, et al., Proc. 6th Int. Conf. Gas Hydrates, (ICGH 2008 ), Vancouver, BC Canada, July 6–10, 2008. 50. K. A. Udachin, S. Alavi, J. A. Ripmeester, J. Chem. Phys., 2011, 134, 121104. 51. K. A. Udachin, C. I. Ratcliffe, G. D. Enright, and J. A. Ripmeester, Angew. Chem. Int. Ed., 2008, 47, 9704. 52. S. Alavi, S. Takeya, R. Ohmura, et al., J. Chem. Phys., 2010. 133, 074505. 53. S. Alavi, R. Susilo, and J. A. Ripmeester, J. Chem. Phys., 2009, 130, 174501. 54. R. Susilo, I. L. Moudrakovski, P. Englezos, and J. A. Ripmeester, ChemPhysChem, 2009, 10, 824. 55. S. Alavi, R. Ohmura, and J. A. Ripmeester, J. Chem. Phys., 2010, 134, 054702. 56. (a) Y. P. Handa, J. Chem. Thermod., 1986, 18, 915–921; (b) Y. P. Handa, D. W. Davidson, and J. A. Ripmeester, J. Phys. Chem., 1986, 90, 6549. 57. J. A. Ripmeester, in Physics and Chemistry of Ice, ed., W. F. Kuhs, RSC Publishing, Cambridge, 2007, 59–72. 58. S. Takeya, K. A. Udachin, I. L. Moudrakovski, et al., J. Am. Chem. Soc., 2010, 132, 524–531. 59. H. H. Richardson, P. J. Wooldridge, and J. P. Devlin, J. Chem. Phys., 1985, 83, 4387. 60. R. Kumar, S. Lang, P. Englezos, and J. A. Ripmeester, J. Phys. Chem. A, 2009, 113, 6308–6313. 61. A. K. Sum, R. C. Burruss, and E. D. Sloan, Jr, J. Phys. Chem. B, 1997, 101, 7371.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Clathrate hydrates

15

62. C. A. Tulk, D. D. Klug, and J. A. Ripmeester, Ann. N. Y. Acad. Sci., 2000, 912, 859.

69. C. H. Baerlocher, L. B. McCusker, and D. H. Olson, Atlas of Zeolite Framework Types, Elsevier, 2007.

63. J. A. Ripmeester and C. I. Ratcliffe, J. Phys. Chem., 1990, 94, 8773.

70. Y. Park, M. Cha, W. Shin, et al., J. Phys. Chem. B, 2008, 112, 8443–8446.

64. J. A. Ripmeester and C. I. Ratcliffe, Energy Fuels, 1998, 12, 197–200.

71. D.-Y. Kim, J.-W. Lee, Y. T. Seo, et al., Angew. Chem. Int. Ed., 2005, 44, 7749–7752.

65. J. A. Ripmeester and C. I. Ratcliffe, J. Struct. Chem., 1999, 40, 654–662.

72. M. R. Walsh, C. A. Koh, E. D. Sloan, et al., Science, 2009, 326, 1095–1098.

66. M. J Collins, C. I. Ratcliffe, and J. A. Ripmeester, J. Phys. Chem., 1990, 94, 157–162.

73. J. A. Ripmeester and S. Alavi, ChemPhysChem, 2010, 11, 979–980.

67. (a) C. J. Jameson and D. Stueber, J. Chem. Phys., 2004, 120, 10200–10214; (b) D. Stueber and C. J. Jameson, J. Chem. Phys., 2004, 120, 1560–1571.

74. S. Takeya and J. A. Ripmeester, Angew. Chem. Int. Ed., 2008, 47, 1276–1279.

68. J.-W. Lee, H. Lu, I. L. Moudrakovski, et al., J. Phys. Chem., 2011, 115, 1650–1657.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc117

Synthetic Clathrate Systems Roger Bishop The University of New South Wales, Sydney, Australia

1 Introduction 2 Modification of Classic Clathrand Hosts 3 Dianin’s Compound and its Cousins 4 Inclined Aromatic Planes 5 Multilegged Hosts 6 Helical Tubuland Diols 7 Highly Porous Organics 8 Hosts with Trigonal Symmetry 9 Halogenated Diheteroaromatic Hosts 10 Convergence of Clathrate Families 11 State of Play and Conclusions Acknowledgments References

1

1 2 4 6 10 11 12 14 17 18 19 21 21

(SO2 )·(H2 O)X , the first of the important clathrate hydrate family of materials (see Clathrate Hydrates, Supramolecular Materials Chemistry). Many further cases of inclusion materials were discovered by happy accident during the next two centuries, for example, additional clathrate hydrates, the Hofmann clathrates, phenol inclusion compounds, Dianin’s compound, urea tubulates, choleic acids, cyclodextrins, and interpenetrated hydroquinone inclusion compounds.2, 3 These substances remained problematic despite being the object of much painstaking scientific study. Many were unstable under ambient conditions and therefore it proved difficult to determine accurate ratios of their two components A and B. Furthermore, the substances did not follow the usual rules of covalent bonding, leading to their representation in the form (A)X ·(B)Y . It was surmised that one component somehow trapped the other, but no experimental methods were available to analyze this phenomenon.

INTRODUCTION 1.2

1.1

Clathrates

In the beginning

Arguably, the first two inclusion materials described scientifically are the boiling stone discovered by Axel Cronstedt in 1756 and the anomalous ice prepared by Joseph Priestley in 1778.1 On heating the mineral stilbite in a flame, Cronstedt observed the release of vapor and named this material zeolite (Greek zein lithos: boil stone). Since then, this substance has proved to be the prototype of an entire group of porous minerals and their synthetic analogs. We also now appreciate that Priestley’s anomalous ice was the compound

Solution of this mystery required both the development of suitable instrumentation and the advent of workers interested in deciphering the dot of ignorance. Among this research, the remarkable series of X-ray crystallographic papers by Powell et al. that commenced in 19434 on The Structure of Molecular Compounds stands out.5–7 The term clathrate was introduced on the basis of several of these pioneering structural determinations. There may thus arise a structural combination of two substances, which remain associated not through strong attraction between them but because strong mutual binding of the molecules of one sort only makes possible the firm enclosure of the other. It is suggested that the general character of this type of combination should be indicated by the description

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

2

Supramolecular materials chemistry

“clathrate” compound—“clathratus,” enclosed or protected by cross bars of a grating.8 Over time, the term clathrate has become strongly intertwined with both Cram’s host–guest concept and our increasing understanding of intermolecular forces within crystals. By 1962, Brown had established the fundamental topologies adopted by inclusion compounds,9 and a comprehensive classification of all host–guest structural types was proposed in 1983.10 Introduction of the supramolecular synthon concept by Desiraju11 and more recent analysis of weak hydrogen bonds and other interactions12 have been of immeasurable value. The primary causes of clathrate formation are geometrical: the poor packing of awkwardly shaped hosts, and/or dominant directional host hydrogen bonding (or other interhost attractions), thereby generating void spaces for guest occupancy. Weber and Josel10 have pointed out that there is a structural continuum (1–4) between clathrates on one hand and complexes on the other. 1. Clathrate: dominant host structure, less important host– guest interactions, guests enclosed by host molecules. 2. Coordinatoclathrate: important host structure, assisted by host–guest coordination (e.g., hydrogen bonding), guests enclosed by host molecules. 3. Clathratocomplex: important host–guest structure (e.g., hydrogen bonding), assisted by host–host interactions, importance of 3D enclosure diminished. 4. Complex: the different components coordinate (e.g., hydrogen bond) with each other, terms host and guest are no longer appropriate, 3D enclosure reduced to minimum. Thus, the host molecules dominate clathrate compounds, and the host–guest (and guest–guest) interactions are a secondary factor. The term clathrate has gradually been applied more broadly over the years, and today the descriptors clathrate, lattice inclusion compound, and solvate tend to be used interchangeably by many.

CH3 O

O

This chapter does not attempt to catalog all the fascinating clathrate and coordinatoclathrate structures reported in the literature, or their many applications.13–16 It focuses instead: (i) on those clathrates that can be analyzed and whose host molecules can be modified to synthesize new clathrands, and (ii) on those systems whose characteristics may be otherwise employed toward the design of new and original clathrands. Review articles are highlighted where possible. The only metal containing clathrates discussed, for reasons of space and avoidance of duplication, are the tetraarylporphyrins. Where possible, the underlying reasons for clathrate formation, and comparison with different groups of compounds, are explored. Such an analysis of the wider clathrate field has not been carried out previously.

2

MODIFICATION OF CLASSIC CLATHRAND HOSTS

2.1

Tri-o-thymotide and trianthranilide hosts

Several early clathrands of quite disparate molecular structure, such as cycloveratrylene (see Cyclotriveratrylene and Cryptophanes, Molecular Recognition), perhydrotriphenylene (PHTP) 1,17 and tri-o-thymotide (TOT) 2, share trigonal symmetry as a common characteristic. TOT is an extremely versatile host that exists in solution as an equilibrium mixture of enantiomeric propeller and enantiomeric helix conformations. Consequently, many TOT clathrates are formed in chiral space groups as conglomerates and 2 can be employed as a resolving agent for racemic guests.18 While PHTP appears to be a unique host structure, a few synthetic TOT analogs do also include guests. The fickle nature of clathrate formation is demonstrated by interchange of the methyl and isopropyl substituents of TOT, or the use of thioester links, both of which result in loss of inclusion properties. Replacement of isopropyl by 2-butyl groups, however, affords an excellent clathrand.19 R4

O C

CH3 O

O 1

Scope of chapter

Pri O

Pri

1.3

O Pri

NR3

NR1 R4 O

C

C NR2

O

R4

CH3 2

3a R1 = R2 = CH2Ph; R3 = R4 = H 3b R1 = R2 = R3 = CH2Ph; R4 = H 3c R1 = CH2Ph; R2 = R3 = R4 = CH3

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

3

A more successful synthetic modification was found instead to be replacement of the ester links of the macrocyclic ring by amide functionalities, thereby producing the trianthranilide family. Three examples of these new clathrate hosts 3a–c are illustrated.20

2.2

Urea and thiourea

The inclusion properties of urea were discovered by Bengen in 1940 and this tubulate host has since become one of the most studied. Thiourea and selenourea form related, but slightly different, clathrate structures.21, 22 Figure 1 illustrates the structure of the (thiourea)3 ·(carbon tetrachloride) compound.23 Many 1,3-diarylurea derivatives also include guest molecules, but these produce hydrogenbonded complexes with acceptor guest species, rather than clathrates.24

2.3

Bile acid derivatives

The bile acids are a group of natural compounds produced in the liver from cholesterol and whose function is to facilitate formation of micelles that assist in the emulsification of dietary fat. They can also form solid inclusion compounds known collectively as choleic acids. The steroidal skeleton has an arched shape with two angular methyl groups and the alkyl side chain subtended from its convex hydrophobic β-face. Two common examples are cholic acid 4 and deoxycholic acid 5, in whose molecular structures the polar hydrogen-bonding groups are subtended from the concave hydrophilic α-face. Since the fused carbocyclic rings are rigid, only relatively straightforward chemistry is required to modify but retain these key supramolecular characteristics in synthetic analogs. Around 17 of these

Figure 2 Part of the structure of crystal form II of (4)·(ethyl acetate) showing a typical host bilayer assembly with enclathrated guest molecules. Color code: H, light blue; O, red; host C, green; guest C, purple or orange (crystallographically independent guest molecules). Host–host hydrogen bonding is indicated by dashed lines.31

steroidal compounds have been shown to act as inclusion hosts. The Miyata group has made the development and interpretation of this chemistry very much their own through their skillful analysis of over 300 inclusion crystal structures.25–31 Concave face to concave face association by means of hydrogen bonding results in bilayer formation in many of these inclusion crystals, but a very wide range of different host–guest structural assemblies is possible. One crystal form of the compound (4)·(ethyl acetate) is illustrated in Figure 2.31 CH3 R3 CH3 CH3

R1

CO2H

R2 4 R1 = R2 = R3 = OH 5 R1 = R3 = OH; R2 = H

2.4

Figure 1 A cross-sectional slice across four tubes of the (thiourea)3 ·(carbon tetrachloride) clathrate structure. Host color code: C, bright green; S, yellow; N, dark blue; and H, light blue. Guest: C, purple and Cl, pale green. Only one orientation of the disordered guest is shown.23

Porphyrin sponges

Porphyrin compounds are often difficult to obtain in quantity, but the nonnatural tetraarylporphyrins 6 are a happy exception.32, 33 These symmetric substances have a large rigid core with rotatable aromatic substituents around its periphery. They therefore constitute a subset of the inclined aromatic planes host family (Section 4). The central ring may be empty or occupied by a wide range of metals (M), and many different aryl substituents (Ar) can be employed. These building blocks generally complement each other in two dimensions but not the third. In consequence, several hundred clathrate compounds, often termed porphyrin sponges, have been obtained through variation of these parameters. The structural characteristics of these compounds have been analyzed34–38 and reviewed.33, 39–41

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

4

Supramolecular materials chemistry

Figure 3 Repeat unit of the crystalline coordinatoclathrate (7)·(acetone)4 . All four acetone guest molecules (colored blue, yellow, orange, and purple) are crystallographically independent, and host–guest O–H· · ·O hydrogen bonds are indicated by dashed lines. The topology of this host molecule should be compared to the general tetraarylporphyrin structure 6.43

Using a different synthetic approach, Aoyama has attached just two resorcinol substituents on an octaethylporphyrin core, thereby creating an open hydrogen-bonded structure that yields coordinatoclathrate10 inclusion compounds.42 HO

OH Ar

Ar N

M

N

of this substance, Dianin’s compound 8, were thoroughly investigated by Baker et al., who reported the formation of some 50 clathrate compounds in 1956.45 Racemic 8 crystallizes in space group R3 with host molecules assembled into hydrogen-bonded (O–H)6 cycles, such that the enantiomeric molecules alternate in both location and directionality.46 Hence each (O–H)6 cycle subtends three (+)-isomers from one face and three (−)-isomers on the other. These subtended molecules intermesh with those of identical building blocks to create linear chains of cages. Adjacent chains pack parallel to each other, assisted by aromatic edge-face (EF) interactions. Each cage is effectively sealed by the (O–H)6 cycles at its top and bottom, and is constricted at its waist by the six intermeshing molecules of 8. Thus, crystals of Dianin’s compound contain parallel hour glass-shaped cages that enclose guests, normally with H : G ratios of 6 : 1 or 3 : 1 depending on the guest size.

3.2

Systematic modification of Dianin’s compound

The robust nature of the Dianin clathrates encouraged systematic modification of the host to attempt the synthesis of further clathrands.47–49 Baker et al. prepared the methylated derivative 9 but found that the inclusion behavior of Dianin’s compound was not duplicated. This experiment is believed to be the first ever attempt to duplicate clathrate

N Ar

Ar 6

R1 OH

HO 7

O

O R4

DIANIN’S COMPOUND AND ITS COUSINS

R3

OH

√ 8 R1 = R2 = R3 = R4 = H x 9 R1 = R2 = R3 = H; R4 = CH3 x 10 R1 = R2 = R4 = H; R3 = CH3 x 11 R1 = R3 = R4 = H; R2 = CH3 √ 12 R2 = R3 = R4 = H; R1 = CH3

OH

√ 13 R1 = R3 = CH3; R2 = H √ 14 R2 = R3 = CH3; R1 = H x 15 R1 = R2 = CH3; R3 = H

R1 O

3

R2

CH3

R3 CH3

The novel compound 7 mimics tetraarylporphyrin topology by employing a quite different planar central core and four bis(aryl) peripheral substituents. Additional hydroxy groups lead to coordinatoclathrate formation, such as the compound (7)·(acetone)4 shown in Figure 3.43 The principal intermolecular attraction in this solid is host guest O–H· · ·O hydrogen bonding, and the guests occupy the spaces present between the peripheral phenyl groups.

R1

CH3

R2

CH3 CH3

R2

S

CH3 CH3

R3 CH3

CH3

OH X

3.1

The structure of Dianin’s compound

Condensation of phenol and mesityl oxide yields a product first described by Dianin in 1914.44 The inclusion properties

x 16 X = NH2 √ 17 X = SH

√ 18 R1 = R2 = R3 = H √ 19 R1 = R2 = H; R3 = CH3 x 20 R1 = R3 = H; R2 = CH3 √ 21 R2 = R3 = H; R1 = CH3

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems properties through synthesis. MacNicol later took up this challenge of making synthetic analogs with remarkable success. (Here, successful clathrand outcomes are marked √ and failures X.) Completing the tetramethyl series of compounds 9–12, the derivative 12 again crystallized in space group R3 and acted as a host. He found that compound 13,50 and Collet and Jacques found that 14,51 of the dimethyl series 13–15 behaved similarly. Replacement of the phenolic hydroxy group with an amino group as 16 failed, but the use of the thiol group in 17 succeeded. Perhaps the most striking breakthrough was in replacing the chroman ring oxygen by a sulfur atom to produce not only the thia-analog 18 of Dianin’s compound, but also the derivatives 19 and 21 belonging to the tetramethyl series of compounds 19–21. The seven new synthetic analogs crystallize in space group R3, are essentially isostructural with Dianin’s compound, and include their guests in a similar manner. In each case, the construction of the host cage comprises (O–H)6 or (S–H)6 cycles at the top and bottom and six intermeshing groups around its waist. Since the latter groups have been changed, these modifications have a considerable effect on the cage sizes and shapes, as illustrated in Figure 4. Whereas 8 has an hour glass-shaped cage, removal of the methyl group at the constricted waist to yield 13 results in an ellipsoidal-shaped cage.52 Substantial changes occur for the methylated thia-analog 21, where the cage is now substantially shorter but wider. It now has a Chinese lantern topology.53 These remarkable studies represent the first examples of the deliberate synthesis of new cage clathrands. Both enantiomers must be present for the formation of the Dianin’s compound crystal structure, and an entirely different noninclusion structure is formed if enantiopure material is used.54, 55 The robust nature of the Dianin

(a)

(b)

5

cage is underscored, however, by its willingness to survive pseudoracemate (quasiracemate) formation. For example, crystallization of equimolar quantities of the Jacques resolved compounds (−)-8 and (+)-14 yielded the mixed clathrate [(−)-8]3 ·[(+)-14]3 (CCl4 ) with only minor loss of symmetry in space group R3.48 Similarly, Dianin’s compound and its thiol analog produced the compound [(+)-8]3 [(−)-17]3 also in R3. This substance employs the novel supramolecular synthon (O–H)3 (S–H)3 .56 The cage structure not only survives in the guest-free compound, but also in its 3 : 1 compound with morpholine in which proton transfer has taken place. Here, the (O–H)6 cycles have been replaced by O–H· · ·O− · · ·H–N+ –H· · ·O–H· · · O− · · ·H–N+ –H· · · hydrogen-bonded cycles (Figure 5).57

(a)

(b)

Figure 5 (a) The original (O–H)6 hydrogen-bonding motif present in Dianin’s compound 8.46 (b) The hydrogen-bonded cycle O–H· · ·O− · · ·H–N+ –H· · ·O–H· · ·O− · · ·H–N+ –H· · · present in the 3 : 1 inclusion compound of 8 and morpholine.57 Ring hydrogen bonds are indicated by the dashed lines. The two phenolic hydroxy groups expelled from the original (O–H)6 ring each hydrogen bond to an O− group. Parts of the host structure are omitted for clarity in both cases.

(c)

Figure 4 Comparisons of host cage topologies. (a) The hour glass-shape of Dianin’s compound 8.46 (b) The ellipsoidal cavity of compound 13.52 (c) The Chinese lantern-shaped cavity of 21.53 In all cases, the molecules at the front and rear of the cage, and H atoms, have been omitted for clarity. Color code: C, green (opposite enantiomers light or dark); O, red; S, yellow. The (O–H)6 cycles are indicated by red and white dashes. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

6

Supramolecular materials chemistry

4

INCLINED AROMATIC PLANES

4.1

Twisted aromatic hosts

CO2H HO2C

A wide range of host molecules has structures in which aromatic groups have a twisted two-bladed propeller topology.58 These frequently involve exact or pseudo C2 molecular symmetry. Tetraphenylene 22 and its dibenzo derivative 23 are archetypical examples.59 Hydrocarbons such as axial compound 24, spirane 25, and allene 26 are further cases.60 As an example, the clathrate structure formed by allene 26 and dioxane is illustrated in Figure 6.61 Alternatively, twist can be incorporated by appropriate placement of hydrogen-bonding roof groups on an aromatic framework, as in the case of 27.60 Scissor structures are produced when the same strategy is applied to an axial

22

23

HO

28

HO HO Pri

29

CHO OH CH3

Pri

OH CH3 OH CHO OH 30

molecular core, for example, to produce the diacid 28 and bis(phenol) 29 molecules.62, 63 The latter are handed atropisomeric bifunctional structures: blade and grip for the scissors, and hydrogen-bonding group and propeller for the host molecule. Consequently, the roof and scissor compounds, and their many analogs, function as potent coordinatoclathrate inclusion hosts. These rich inclusion properties continue in more complex structures, such as the cottonseed oil component gossypol 30 and some of its synthetic derivatives.64

24 HO2C CO2H

.

4.2 25

OH

26

27

Figure 6 Crystal structure of (26)·(dioxane) showing the guest molecules occupying channels between layers of the aromatic allene hosts. Two crystallographically independent dioxane molecules are present (C, purple or orange) and the guest H are omitted for clarity.61

Propeller-shaped hosts

Molecules containing a propeller-shaped array of substituents, such as triarylphosphine and triarylmethane, often

Figure 7 The propeller-shaped host tris(5-acetyl-3-thienyl) methane 31 forms a 3 : 1 clathrate with cyclohexane.68 Color code: H, light blue; host C, green; O, red; and S, yellow; guest C, purple.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

7

S

CH3-CO

CO-CH3 CH

S HO

S 31

HO CO-CH3

32 33

X HO

Ph

Si

HO2C

OH 34a X = CH2-CH2 34b X = CH=CH 34c X = O

associate favorably in the solid state by means of phenyl embrace interactions.65, 66 Nonetheless, such three-bladed groups are bulky and may cause ineffective packing leading to guest inclusion, as in the cases of triphenylmethane and tris(5-acetyl-3-thienyl)methane 31.67, 68 Many alternative aromatic or heteroaromatic groups may be used. The representative example of (31)2 ·(cyclohexane) is shown in Figure 7. Two means of facilitating inclusion behavior are to conjoin two of the aromatic rings to generate singly bridged triarylmethane frameworks, and/or replace the central hydrogen atom with a hydroxy or carboxy substituent. A silicon atom can also replace the methane carbon. Compounds 32–36 are representative examples of these host molecules.69–71 Guest inclusion therefore can be of the clathrate type,8 or it can result from a combination of shape effects and hydrogen bonding (coordinatoclathrate type).10 The extensive range of propeller-shaped host molecules has been shown to yield several hundred cases of inclusion.

4.3

Dumbbell, and wheel and axle hosts

Connection of two bulky propeller-shaped groups by a short linker produces a dumbbell-shaped structure (e.g., 37) and use of a longer linker creates wheel-and-axle geometry (e.g., 38 and 39). The latter group of compounds is constructed from two relatively rigid end groups (wheels) separated by a rigid connection (axle). These types of material were first reported by Toda72, 73 in 1968 and since have been developed into an extremely wide ranging family of host compounds.74–76 Some common end groups used are diaryl, triaryl, adamantyl, or triptycyl moieties, whereas the linkers are often allenyl, alkenyl, alkynyl, azo, or transition metal complex functionalities. However, an exceptionally wide range of end group and linker units may be used,

Ph 36

35

which has led to an enormous number of compounds (not all of which function as hosts). Soldatov has recently reviewed the structures of compounds related to the wheel-and-axle topology.77

OH C

C

HO

HO

C

C OH

38 37 HO C

C

C

C

OH 39

The majority of such compounds intended to act as host molecules carry hydroxy groups at the join of the bulky group and linker.73, 76 This is a consequence of both synthetic practicability and deliberate design. Most of the preparations use organolithium or Grignard methodology to produce the alcohol functionalities. These groups occupy a rather crowded environment due to the bulky end groups and therefore host–host hydrogen bonding is problematic. Instead, the awkwardly shaped molecules prefer to associate with smaller or more compact guest molecules by means of hydrogen bonding (coordinatoclathrates) or weaker interactions (clathrates). Examples of coordinatoclathrates formed by the dumbbell host 3778 and wheel-and-axle host 3979 are illustrated in Figures 8 and 9, respectively. Both enclosure and hydrogen-bonding contributions result in the stability of these particular inclusion compounds.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

8

Supramolecular materials chemistry

S C

C

HO

HO

OH 40

Figure 8 Crystal structure involving a dumbbell-shaped host: the coordinatoclathrate compound (37)·(2,6-lutidine). Color code: O, red; N, dark blue; host C, green; guest C, purple. The H–O· · ·N hydrogen bonds are indicated by dashed lines, and H atoms are omitted for clarity.78

S

OH 41

Although the majority of wheel-and-axle hosts utilize hydroxy groups, this is by no means obligatory and the awkward shape of the molecule alone can be sufficient to confer clathrate-forming properties. This is particularly true where larger wheels are employed. Compound 42 uses tris(o-tolyl)phosphine end groups and a novel gold acetylide axle,82 while 43 contains triptycyl end groups and a central 1,4-disubstituted phenyl axle in order to probe molecular rotor behavior.83 Ar3P

Au

C

C

Au

PAr3

42 Ar = m-tolyl

C

C

C

C

43

The last three examples emphasize just how robust the wheel-and-axle concept is, and reveal how much it can be abused while still retaining clathrate behavior. The complex 44, formed from two molecules each of PdCl2 and a dipyridyl tetrapropoxycalix[4]arene, yields a chloroform clathrate.84 The calixarene groups act here as the wheels, and the coordinated palladium functionalities provide a novel double-axle. The orthogonal view shown in Figure 10 clearly reveals the wheel-and-axle topology of this host. OPr OPr OPr PrO

Figure 9 The principal intermolecular attraction in the (39)·(pyridine)2 coordinatoclathrate structure is host–guest O–H· · ·N hydrogen bonding (dashed lines), and the guests occupy the concave space resulting from the wheel-and-axle moieties. Color code: H, light blue; O, red; N, dark blue; host C, green; guest C, purple.79

O N

The addition of a “differential gear,” usually an aromatic group, at the center of the molecule takes the wheel-andaxle analogy one step further. This modification greatly increases the probability of guest inclusion compared to the original design, by combining wheel-and-axle topology with additional inclined aromatic planes. Two examples of the many reported are 40 from Weber et al.80 and compound 41 belonging to the fused thiophene family introduced by Kobayashi.81

Cl Pd

CH3 NH2

N Cl

Cl Pd

N

Cl

N

N

N

45 OPr OPr OPr PrO 44

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

N(CH3)2

Synthetic clathrate systems

9

do not pack efficiently in the solid state without molecular inclusion. If appropriate functionalities are placed at the extremities of such a molecule, then the molecule becomes a molecular tweezer and a guest may be secured between its tips.88 Structure 47 has a U-shape with the benzofuran tips parallel and ideally separated for intercalation of a 1,3,5-trinitrobenzene guest by means of π· · ·π interactions (Figure 11).89 If construction is continued further to generate a macrocycle containing bridged aromatic groups, then the structure becomes a cyclophane.90, 91 These are generally more flexible compounds, such as 48, and many are excellent clathrands. This does not necessarily mean, however, that guest inclusion is intramolecular. In the case of (48)·(benzene)2 ·(water), a single benzene guest type is enclosed within the macrocycle (Figure 12) while the water and a second benzene type occupy intermolecular sites.92

O

Figure 10 The dipyridyl tetrapropoxycalix[4]arene–PdCl2 complex 44 has two axles, but its crystal structure (rotated 90◦ with respect to the molecular structure diagram) reveals clear wheel-and-axle topology.84

O

O O

47

Further, it is not necessary for the two wheels to be identical: for example, the azo-linked derivative 45 forms clathrates with a variety of aromatic hosts.85 Ph

CH3 Ph

O

N

N

O

H O Ph

CH3

(CH2)4

CH3

OH

O CH3

O

H OH

(CH2)4

O

48

Ph 46

Finally, compound 46 is one of a group of chiral host molecules (TADDOLs) devised by Toda and Tanaka and synthesized from tartaric acid.86, 87 These were designed and employed as chiral agents for resolution experiments. Often hydrogen-bonded complexes, rather than coordinatoclathrates, are formed using TADDOLs. However, their molecular structure is again an excellent illustration of just how ubiquitous the wheel-and-axle motif is in inclusion chemistry.

4.4

Tweezers and paracyclophanes

Molecules designed to have a V-shape, or to form a molecular cleft, frequently have awkwardly shaped structures that

Figure 11 The molecular tweezer compound (47)·(1,3,5trinitrobenzene), showing the guest positioned between the benzofuran tips of the host molecule where they are held by means of π · · ·π attractions. Color code: O, red; N, dark blue; H, light blue; host C, green; and guest C, purple.89

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

10

Supramolecular materials chemistry Ar

Ar

Ar

Ar H O

O

Ar HO H

H Ar

OH

O

L O

Ar L

L Ar

H

L

(b)

Ar

L Ar

Ar (a)

L

Ar

Ar = aromatic substituent L = flexible atomic linker

Figure 12 The crystal structure of (48)·(benzene)2 ·(water). Two crystallographically independent benzene molecules are present: one (C, purple) enclosed within the cyclophane host, and the other (C, orange) between the cyclophane molecules. The water molecules are indicated by black spheres. All hydrogen atoms are omitted for clarity.92

5 5.1

Figure 13 The hexa-host concept.93 Diagrammatic comparison of the (O–H)6 hydrogen-bonded motif present in the R3 Dianin’s compound and β-hydroquinone inclusion structures (a), and a hexaaryl substituted benzene using linkers (L) (b). In both cases, the substituents are subtended alternatively up then down (ababab) around the central ring, and the O–O and L–L distances are comparable.

MULTILEGGED HOSTS Hexa-hosts

The hexa-host molecules are the first synthetic clathrands with no direct structural relationship to a previously known host. They were designed from scratch in the first supramolecular clathrand synthesis.93 MacNicol realized that appropriate hexa-substituted benzenes would be more robust analogs of the (O–H)6 hydrogen-bonded motif found in inclusion compounds such as the β-structure of hydroquinone 4994 and Dianin’s compound 8 (Section 3).46 If the aromatic substituents were attached to the central core by a flexible linker atom or group (L: e.g., S or CH2 –O), then the new molecule would act as a covalent mimic of the earlier hydrogen-bonded motif (Figure 13). HO

OH 49

R R

R

R

R

50 R = S-C6H5 51 R = CH2-O-C6H4-m-CH3 52 R = CH2-S-CH2-C6H5 53 R = CH2-N(CO-CF3)-CH2-C6H5

R

Structures 50–53 are representative of the 50 or so of these highly successful hexa-hosts that have been shown to form clathrate compounds. The crystal structure of (52)·(dioxane), illustrated in Figure 14, is a representative example.95 Hexa-host synthesis and inclusion has been reviewed in detail.96 There have been occasional reports of hexa-host molecules adopting alternative conformations.

Figure 14 Part of the crystal structure of the hexa-host clathrate (52)·(dioxane) projected on the ab plane. Color code: O, red; S, yellow; host C, green; and guest C, purple. The H atoms are omitted, and the central aromatic ring of 52 is colored dark green, for clarity.95

Das and Barbour, for example, observed that hexakis (4-cyanophenyloxy)benzene forms clathrates with aaabbb and aaabab conformations in addition to the usual ababab type. Desolvation yielded the same apohost phase in each case and therefore these less common alternatives indicate host adjustment to best accommodate a particular guest.97

5.2

Spiders and beyond

The use of multilegged molecules for clathrate formation is not limited to a single ring bearing just six appendages. A fully substituted naphthalene carries 8 groups, anthracene 10, and triphenylene and coronene each have 12 substituents. MacNicol has demonstrated that appropriate derivatives of all these ring systems can function as successful hosts and the representative

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

11

Y Ar S

Ar S

Y

Y

Y

SAr

ArS

Y

Y

SAr S S Ar Ar 54 Ar = 3,4-diCH3-C6H3 ArS

Y

Y

Y Y Y

Y

Y

Y

Y

Y Y Y

YY

Y Y

55 Y = S 56 Y = O

Ar X

Cl

Ar X

ArX

Cl Cl

Cl XAr

ArX

XAr

ArX

XAr

ArX

Cl Cl Cl

Cl Cl

Cl

XAr X Ar

Cl Cl

Cl Cl

X Ar

58

57 X = S or O Ar = 3,5-diCH3-C6H3

examples 54–57 are shown here.98 The eight-legged spider hosts based on the naphthalene core have been investigated in the greatest detail.98, 99 Derivative 54 showed particularly impressive versatility by including guests ranging in size from isopropanol, through 1,2bis(trimethylsilyl)ethyne and tetraethyllead (Figure 15), to squalene.100 Decakis(cyclopentylthio)anthracene 55101 and the triphenylene derivative 5698 are typical examples of 10- and 12-substituted clathrand systems, and this inclusion behavior persists using the coronene core of the examples 57.102 This methodology represents a very general concept that extends beyond just considerations of symmetry. Most interestingly, the spiro fluorene derivative 58 with 14 chlorine substituents also proved to be an effective host molecule.103

6 6.1

Host structure

The first helical tubuland diol 59 was reported in 1979 by Bishop and Dance as being a potent host despite its extremely simple molecular structure.104 In this family of hosts (e.g., 59–62) infinite hydrogen-bonded chains · · ·O–H· · ·O–H· · ·O–H· · ·O–H· · · surround threefold screw axes to create rigid rod-like spines (Figure 16a). The diol molecules subtended outwards from these spines form threefold stacks of eclipsed molecules, and the second hydroxy group of each participates in another identical spine. A low-density hydrogen-bonded network structure thereby results, with the diol molecules surrounding parallel tubes filled with guests along the c direction (Figure 16b). This host arrangement is similar to that in unidirectional zeolites such as laumontite but it lacks their thermal stability. A consequence of the helical tubuland host structure is that it requires homochiral diol molecules for its construction in the chiral space group P 31 21 (or its enantiomorph P 32 21). Thus, each racemic diol undergoes self-resolution during crystallization to produce a conglomerate: a 1 : 1 mixture of (+)- and (−)-inclusion crystals.105

6.2

Figure 15 Conformation adopted by the host molecule in the spider inclusion compound (54)3 ·(tetraethyllead). Color code: C, green; S, yellow; H, light blue.100

HELICAL TUBULAND DIOLS

Molecular determinants

It appeared likely that this elegant type of assembly would also occur in the crystals of other diol molecules. Apart from chemical synthesis of these new compounds, this supramolecular synthesis would require retention of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

12

Supramolecular materials chemistry

(a)

(b)

Figure 16 (a) Part of one helical spine in the structure (62)3 ·(chloroform) involving · · ·O–H· · ·O–H· · ·O–H· · ·O–H· · · hydrogen bonding of the diol 62 molecules surrounding a threefold screw axis along c. (b) Projection of the crystal structure in the ab plane and showing four tubes in cross section. The rigid hydrogen-bonded spines appear as triangular shapes and eclipsed stacks of 62 molecules surround each of these. For clarity, only one disorder component of the guest is illustrated. Color code: O, red; H, light blue; Cl, brown; host C, green; and guest C, purple.106

unchanged hydrogen-bonding motif in order to retain the identical crystal space group. Synthetic work was guided by the formulation of a set of six structural rules (molecular determinants) that were believed to govern helical tubuland formation. Alicyclic diols obeying all these rules, and being sterically capable of fitting the required lattice, are expected to crystallize in the required manner. This approach parallels the determination of the pharmacophore of a bioactive molecule and permits complex syntheses to be undertaken with confidence in the eventual outcome. These rules have been refined in the light of synthetic experience and are currently summarized as follows: 1.

2.

3.

4.

5.

The tertiary alcohol groups must have a methyl substituent. This group induces the hydroxy groups to hydrogen bond around a threefold screw axis. Further, it has the correct size and rigidity to support the tube wall structure and prevent its collapse. The alicyclic structure must have a small amount of twist. This may be built into a rigid alicyclic skeleton, or it may be achieved by means of conformational flexibility in the ring system. It aids the generation of helicity during self-assembly. A bridge on the opposite side of the skeleton to the hydroxy groups is optional. It can therefore be modified, or omitted, to produce new diols with different tube dimensions. A molecular bridge must separate the two hydroxy groups. This ensures that both hydroxy groups act independently and form separate helices. It also buttresses the tube walls against internal collapse. Substituent groups around the periphery should be avoided. However, there is an exception if they are remote from the hydroxy groups and also occupy otherwise void space within the tubes.

6.

The diol molecules must be capable of achieving average C 2 rotational symmetry in the solid state. The diol may have actual C2 symmetry or it may achieve this in solution. Some diols, with only a pseudo-C2 axis due to their molecular substitution, can attain average C2 symmetry by means of disorder in the crystal. HO

OH

H3C

CH3

OH

HO D3C

59

60

H3C

CH3

HO

OH

CD3

HO

OH

H3C

CH3

CH3 61

62

This synthetic program did, indeed, prove to be successful.106–108 It has so far produced 12 members of the helical tubuland diol family with different tube cross sections and a range of inclusion properties. Further cases are predicted. Four examples 59–62 of the diverse range of diol molecules that form this host structure are illustrated. The inclusion and structural properties of the helical tubuland diols,108, 109 together with their synthesis and crystal engineering design,110 have been reviewed.

7 7.1

HIGHLY POROUS ORGANICS Directionality and tectons

A highly successful approach for designing new crystals is the use of tectons, (molecular building blocks) that will undergo controlled self-assembly from solution. Small

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

13

Figure 17 View down the b axis of the crystal structure (65)·(butanoic acid)2 showing channels containing parallel columns of butanoic acid guest molecules. Color code: H, light blue; O, red; N, dark blue; host C, green; guest C, purple. Host hydrogen bonding is indicated by dashed lines.114

Figure 18 Part of the crystal structure of (67)·(benzophenone)2 showing the organic zeolite-like host structure and the benzophenone guests occupying parallel channels. Color code: H, light blue; O, red; host C, green; and guest C, purple.121

molecules, such as trimesic acid 63111 and adamantane1,3,5,7-tetracarboxylic acid 64,112 have specifically positioned functional groups that form strongly directional supramolecular synthons with their neighbors.113 A precise and predetermined open crystal structure is thereby obtained. By using larger and much more sophisticated tectons, such as 65, Wuest developed the molecular tectonics strategy into a wide range of important hosts and other materials. The resulting microporous materials frequently contain guest molecules, and this approach is particularly suited to the synthesis of compounds that include polar guests such as carboxylic acids.114–119 One such example, (65)·(butanoic acid)2 , is illustrated in Figure 17.

acceptor atoms.120 In contrast, the bis(resorcinyl)anthracene 67 crystallizes as two-dimensional hydrogen-bonded sheets that stack to produce parallel guest-containing channels in an organic zeolite-type structure (Figure 18). Once again, ester and ketone coordinatoclathrates, stabilized through host–guest hydrogen bonding, are favored. However, clathrate compounds are also obtained.121 Multicomponent systems were also investigated. Crystallization of a mixture of compounds 67 and 68 produced the cocrystalline adduct (67)·(68), but without cavities or guest inclusion. However, a mixture of 67, 69, and anisole yielded the solid (67)·(69)·(anisole). Here, the two anthracene derivatives form 1 : 1 hydrogen-bonded sheets that stack and then trap the guest molecules in cages.122 Pursuing the organic zeolite analogy, compound 67 was found to promote highly stereospecific Diels–Alder reaction between enclathrated acrolein (or acrylates) and 1,3-cyclohexadiene.123 The resorcinol groups of 67 can be replaced by 2,5-dihydroxyphenyl groups, and the anthracene unit of 66 by an acridine substituent, with the retention of clathrand properties in both cases.124 The structures and chemistry of organic zeolite analogs have been reviewed.125, 126

CO2H

HO2C

CO2H

HO2C CO2H

HO2C

63

CO2H 64

R

R R

R

R= N 65

7.2

H

O

Zeolite mimics

HO

HO

HO

OH

HO

OH

66

Aoyama’s resorcinylanthracene derivative 66 forms onedimensional hydrogen-bonded chains that can associate with each other in two alternative arrangements, both of which contain guest-occupied voids. These self-assembly modes are greatly facilitated by the anthracene substructure adopting an orientation orthogonal to the resorcinol unit. Many coordinatoclathrates are formed with esters, ketones, and other guests that contain hydrogen bond

67 O N

N

N

N

68

N

N

N

N

69

O

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

14

7.3

Supramolecular materials chemistry

Fullerene inclusion

The β-structure of hydroquinone 49 is constructed from two identical interpenetrating sublattices.94 Inspection of this crystal superlattice suggested to Ermer that hypothetical removal of one sublattice would create a microporous material with dimensions closely matching those of C60 . This was duly confirmed by successful preparation of (49)3 ·(C60 ) as black crystals. The stabilizing π · · ·π host–guest interactions are particularly favorable in this clathrate structure due to the electron-donating properties of 49 and the electronaccepting nature of the fullerene.127 These charge transfer characteristics also permitted synthesis of the clathrate (49)9 ·(C70 )2 ·(benzene)2 .128 X

O N

X O 70 X = CH 71 X = N

N N

X

O X

72

An alternative approach is to employ trigonal host molecules with concave part structures. Thus, both triptycene 70 and its aza-derivative 71 (Figure 19) form clathrates with C60 .129 The channel-forming properties of 2,4,6-tris(4bromophenoxy)-1,3,5-triazine 72 (X = Br; Section 8.3) permits formation of crystalline one-dimensional arrangements of both C60 and C70 . A significant observation is that many fullerene clathrates also contain stabilizing molecules of the aromatic solvent of crystallization.130

Much elegant work on inclusion and purification of fullerenes has also been carried out utilizing receptor molecules belonging to the calixarene (see Calixarenes in Molecular Recognition, Molecular Recognition) and cyclodextrin (see Cyclodextrins: From Nature to Nanotechnology, Molecular Recognition) families.

8 8.1

HOSTS WITH TRIGONAL SYMMETRY Cyclotriphosphazenes

Allcock discovered that a number of derivatives of the cyclotriphosphazene ring system, such as 73 and 74, function as efficient host molecules.131–133 The homologous cyclotetraphosphazene derivatives do not include. Cyclotriphosphazenes have trigonal symmetry and their spiro substituents result in a paddle-wheel topology. Hence, they have significant three-dimensional character and cannot stack directly over each other. Instead, the molecules form layers that pack in a translated manner to leave guest˚ diameter (Figure 20). The occupied channels of around 5 A paddle-wheel blades have a considerable degree of conformational folding ability around their oxygen-linking atoms, and this can be used to alter their mutual orientation. This flexibility allows the host to adapt easily to the packing requirements of different guests. Synthetic variation of the aromatic groups also modifies the channel dimensions. The tris(o-phenylenedioxy) derivative 74 (Figure 20) can absorb and store large quantities of methane or carbon dioxide in preference to oxygen and nitrogen.134

O

O

O

N

N

P

P

O

N

O

O

P

P N

O

O O

N

P

P

O

O 73

8.2

Figure 19 The host–guest packing arrangement in (71)·(C60 ). The azatriptycene molecules are viewed here edge-on and as staggered columns. Color code: host H, light blue; N, dark blue; C, purple; fullerene C, green; and H, omitted.129

O

N

74

Piedfort assemblies

Trigonal symmetry is a notable characteristic of many clathrands with diverse molecular structures, as seen above and also in Section 2.1. This led MacNicol to conclude that this favorable property could be utilized for the synthesis of entirely new clathrand hosts. This premise was quickly validated through the behavior of the cyclododecane derivatives 75 and 76,135 and it has since proved to be one

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

Figure 20 The topology of the cyclotriphosphazene lattice present in the clathrate compound formed by host 74 and carbon dioxide, and highlighting the channels formed by the host stacking. Color code: C, green; N, blue; O, red; and P, purple. Hydrogen atoms and the disordered guest molecules are omitted.134

of the most reliable synthetic strategies. A particularly interesting outcome was obtained from crystallization of 2,4,6-trisubstituted-1,3,5-triazine derivatives, such as 77, where the central rings of two molecules undergo dimeric face–face association (Figure 21). This novel arrangement was named the Piedfort assembly by analogy with the numismatic term for special double-thickness coins.136

Figure 21 A view of the Piedfort unit present in the crystal structure of (77)·(1,4-dioxane)2 . The carbon atoms of the two trisubstituted-1,3,5-triazine molecules are colored green or orange for clarity. Color code: H, light blue; O, red; N, dark blue.136

The Piedfort unit is encountered in the crystal structures of a number of other 1,3,5-trisubstituted aromatic molecules, such as 78137 and 79: the latter chiral coordinatoclathrand being a particularly interesting example. Here, the guests are sandwiched between adjacent Piedfort units.138

R

O

OCH3

S R S

S

O OCH3

75 R = H 76 R = CH3

O

R

CH3O

78

CH3 CH3

CH3

CH3 OH CH3

O N O

CH3

CH3

N N

O

CH3 77

CH3

HO

CH3 CH3

15

CH3

OH

CH3 CH3

79

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

CH3

16

Supramolecular materials chemistry

8.3

Channel and cage structures

Hosts with trigonal symmetry, bearing three arms around a central core, are prone to producing hexagonal channelcontaining inclusion structures (with or without the involvement of Piedfort units). The host network of the 2,4,6tris(4-halophenoxy)-1,3,5-triazine 72 (X = Cl, Br, or I) examples are stabilized by means of the robust (halogen)3 cyclic supramolecular synthon.139 The 1 : 1 clathrates formed between the chloro- and bromo-hosts and 1,3,5trihalobenzene guests contain parallel guest channels that are strengthened further through use of C–H· · ·π and Chalogen · · ·π interactions.140 The gradual transition from channel to cage structures as the halogen is altered has been analyzed in detail. Thus, 72 (X = Cl) forms channels, the bromo-host can form either structural type, and 72 (X = I) prefers cage inclusion structures.141 The ethynyl group is known to mimic the behavior of halogen in certain crystal structures, and the tris(ethynyl) compound 80 is an excellent example of this phenomenon.142 Here, each (iodine)3 cycle is replaced by an (ethynyl)3 cyclic supramolecular synthon, with minimal alteration to the cage structure (Figure 22).

O N O

N N

Figure 22 Part of the crystal structure of (80)2 ·(hexachlorobenzene) showing the guests situated in the hexagonal cavities. Host molecules above and below the cage sites are omitted for clarity. Color code: Host H, light blue; N, dark blue; C, green; O, red; guest C, purple; and Cl, orange.142

derivative 81 formed channel clathrates with a concomitant switch from hexagonal to a triclinic space group.143 The fluorinated compounds 82 and 83 are also excellent clathrands, and especially notable for their inclusion of the large guests C60 , C70 , permethylferrocene, and a buckyferrocene derivative.144 One remarkable host in this trigonal family is the perchlorotriphenylmethyl radical 84. This exceptionally stable radical forms clathrates with guests such as benzene (Figure 23), halobenzenes, and dioxane.145

O Br

80

R2

R3

Cl

F F

F

R

1

R

F N

F F

O

F

F F

81

R3

O

Br

R1

N N

Cl Cl

Cl Cl

R1

C

Cl

F O

F

F

F F

The guest channel structures of the 2,4,6-tris(aryloxy)1,3,5-triazines are amenable to systematic modification through alteration of the aryloxy group substitution. The perfluorinated version of compound 72 (X = Br) resulted in loss of inclusion properties, but the fully fluorinated

Cl

2

O

R

R4

Br

Cl

N N

N

R4

O

Cl

4

O

F

Cl

Cl

Cl Cl Cl

Cl

R3

R2

82 R1, R3, R4 = H; R2 = F

84

83 R , R , = H; R R = F 1

4

2,

3

Trigonal host symmetry is such a powerful tool in the design of new clathrate compounds that it can be considerably abused with some success. Thus, the threearmed nontrigonal compounds 85146 and 86147 also form clathrate inclusion compounds.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

17

Alicyclic linker

Halogen atom

Halogen atom

Aromatic wings

Figure 23 Part of the crystal structure of the 1 : 1 perchlorotriphenylmethyl radical 84·benzene clathrate projected on the bc plane. Color code: host C, green; Cl, orange; guest C, purple; and H, light blue.145

O

Figure 24 Schematic design of a generic molecule, constructed from three functional subunits, that will act as a synthetic diheteroaromatic host molecule.148

the awkward shape to the molecule. This effect is compounded by its slightly twisted C2 symmetry. The halogen atoms also provide additional hot spots for host–host and host–guest supramolecular interactions.148

Cl OH N

O 85

9.1

9.2

Modular construction

Cl

N

9

S

86

HALOGENATED DIHETEROAROMATIC HOSTS Host design

Halogenated diheteroaromatic molecules of the general type illustrated in Figure 24 pack together awkwardly and consequently form clathrate compounds. The molecule is constructed from three subunits, all of which have defined functions in achieving this end. The central section is a small bicyclic ring that imparts actual or pseudo-C2 symmetry to the host, which therefore can be investigated as either the racemate or as the single enantiomer compound. This linker group also imparts a degree of conformational freedom to the host, thereby allowing it to adjust to the presence of guests of differing sizes, shapes, and functionalities. The two aromatic wings encourage host–host and/or host–guest association by means of aryl EF C–H· · ·π and aryl offset face–face π · · ·π interactions. Finally, incorporation of halogen substituents (Cl, Br, or I), either in the exo-benzylic positions shown in Figure 24 or on the aromatic wings themselves, reduces the ability of the aromatic interactions to propagate throughout the crystal and imparts

Synthesis of these hosts is a modular process, since any subunit can be replaced by another, while leaving the remaining two subunits unchanged. Hence, a family of closely related host molecules is rapidly obtainable. The synthetic chemistry is often simple and the majority of these clathrands can be prepared in only two or three steps. Typical examples 87–92 of hosts belonging to the halogenated diheteroaromatic family are illustrated. Only one enantiomer of these chiral molecules is shown. It is necessary to use racemic mixtures of these compounds for preparing clathrates, because of the prevalence of centrosymmetric packing motifs present in their crystal structures.149 Nineteen potential hosts have been reported, 18 of which successfully gave clathrate products (a 95% success rate). It should be emphasized that this large family of host molecules has broken away from the hydrogen-bonding approach traditionally used to synthesize most clathrands. Analysis of their clathrate crystal structures has uncovered a number of previously unrecognized supramolecular synthons, such as the π-halogen dimer (PHD)150 and the aryl edge–edge C–H· · ·N dimer interactions.151 Hence, the clathrate crystal structure will depend on competition between these, plus a wide variety of other weaker (and sometimes less directional) supramolecular synthons. Typically, these might comprise aryl EF C–H· · ·π, aryl offset face–face π· · ·π, halogen· · ·halogen, halogen· · ·π , C–H· · ·halogen, C–H· · ·N, and C–H· · ·O interactions.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

18

Supramolecular materials chemistry

Br

Br H N

N H

N

N

N

N

Br

Br

87

88 Br

Br N

Br Br

H

Br

N

S Br

N

N

H

Br

Br

Br

89

Br

90

Br

I

I N

N N

N I

Br

I

91

Somewhat of an “uncertainty principle” is thereby achieved. The probability of predicting clathrate formation has been maximized, but concomitant prediction of the precise means of achieving this from the interplay between weak interactions has fallen to around zero.

92

In contrast, none of the hydrocarbon analogs (e.g., 95) of compounds 87–92 include, whereas 96 does.154 Addition of the two further phenyl rings now has switched the latter into the twisted aromatic hosts family (Section 4.1). R

10

R X

CONVERGENCE OF CLATHRATE FAMILIES

X Y

Y

X

This chapter has so far classified synthetic clathrates into families of closely related materials, depending on their related host structures and behavioral properties. The field has now reached sufficient maturity, however, that a number of compounds can be recognized as being located at the cusp of related families. This section describes clathrates where such close connections are apparent. Tanaka and Toda found that the tetrahalo derivatives 93b–e acted as hosts, whereas the hydrocarbon 93a did not.152 Compound 92 above and its corresponding hydrocarbon analog behave in an identical manner. In fact, many tetrahaloaryl compounds, with quite disparate carbocyclic cores, but a halogen atom at each extremity, constitute a previously unrecognized host family. The series of compounds 94 constitute a further recent example, where 94b–d form clathrates (e.g., Figure 25) but 94a does not.153 This latter group of molecules also represent a structural convergence with the dumbbell host family (Section 4.3).

R

X

R

93a R = H 93b R = F 93c R = Cl

93d R = Br 93e R = I

94a X = H, Y = Cl 94b X = Cl, Y = Cl 94c R = Br, Y = Cl 94d R = Br, Y = Br

N N

N N

95

96

The structure of the tetraphenylmethane derivative 97 represents convergence of the propeller-shaped (Section 4.2) and the tetrahedral tecton (Section 7.1) host families.155

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems

19

Figure 25 Part of the crystal structure of clathrate (94b)2 ·(pxylene). Color code: H, light blue; Cl, orange; host C, green; and guest C, purple.153

Figure 26 Part of the crystal structure of the clathrate (98)2 ·(PhC≡CH). The central ring of the hexaphenylbenzene host molecules is colored darker to assist with viewer orientation. Color code: H, light blue; host C, green; and guest C, purple.156

Hexaarylbenzenes combine design elements of the inclined aromatic plane (Section 4) and the hexa-host (Section 5.1) strategies. These clathrands are able to include terminal alkynes by utilizing multiple C–H· · ·π interactions (including the sp C–H· · ·π motif). This outcome is illustrated in Figure 26 for the case of the clathrate formed between hexamethylbenzene 98 and phenylethyne.156

Unlike the above cases, the relationship between structural and supramolecular properties can be far more cryptic and not immediately apparent. One example is the key role played by the (at first glance unimportant) methyl groups of the helical tubuland diol hosts 59–62. Another is the closely related host behavior of hydroquinone 49 and the diol 102 in the presence of small guests. Although these are aromatic and aliphatic compounds, respectively, both of these hosts form clathrates in a similar manner: namely by trapping small guests between two interpenetrated sublattices in their respective crystal structures.159 There is also a close relationship between the inclusion crystal structures of hydroquinone 49, Dianin’s compound 8, and the alicyclic diol 103. This is revealed if the two motifs used by MacNicol in the hexa-host design, namely the aromatic ring and (O–H)6 cycle structures, are regarded as equivalent supramolecular nodal points. The network connectivity (see Network and Graph Set Analysis, Supramolecular Materials Chemistry) and familial relationship of these three lattices is then revealed.160

CN

NC

CN

C

CN

97

98

Draper has recently reported a study that cuts across three of the host structural systems discussed earlier, namely compounds 99a–c (axial aromatic, Section 4.1), 100a–c (wheel and axle, Section 4.3), and 101a–c (trigonal symmetry, Section 8).157, 158 The majority of these molecules exhibit inclusion properties and their crystal structures provide a revealing comparison of these host types.

99

CH3

HO

R=

R

OH 102

CH3

HO

OH

CH3

103

a N

R 100

N N

11

N

b N

R

N N

N

R 101 R

CH3

c N R

N N N

STATE OF PLAY AND CONCLUSIONS

Much clathrate research has involved studies at only ambient temperature and pressure. There is considerable scope for widening these experimental parameters. Indeed, Powell used high pressure to obtain the clathrates of hydroquinone with argon, krypton, and xenon161 way back in 1950, and solvothermal methods are now widely used for metal containing systems. Protective coating of unstable crystals and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

20

Supramolecular materials chemistry

low-temperature X-ray determinations are long-established crystallographic techniques, but the increasing availability of high intensity synchrotron radiation now permits the structures of thin needle, thin plate, and small crystals to be solved much more readily. Quite astonishing structural determinations can be obtained in the hands of highly skilled workers. For example, Yufit and Howard have recently reported the X-ray structures of novel chloroform cocrystals with methanol, cyclopentanone, cyclohexanone, diethylamine, and dimethylformamide at low temperatures.162 Further, the Boese group have determined the structures of many acetylene cocrystals and have reviewed this important work recently.163 Examples are the compounds of acetylene with benzene (1 : 1), formaldehyde (1 : 1), acetonitrile (2 : 1) (Figure 27), and ammonia (1 : 1). These striking successes indicate the potential for future studies on unstable and lowmelting clathrate compounds that may involve only very weak interactions. Uncovering and determining new clathrate structures does not only depend on the synthesis of new host molecules that has been the theme of this chapter. Changing the crystallization guest species, temperature, or pressure may very well yield different crystal forms of a given host. This event is by no means a rare occurrence, and the possibility of alternative crystal forms is a topic that is assuming increasing importance.164, 165 In contrast, synthesis lags well behind. The traditional means of obtaining new clathrand hosts has been by means of serendipity, and Herbstein has suggested in his recent book that perhaps around 90% of binary compounds have been obtained by chance.16 This chapter describes clathrand molecules whose structural features permit their

Figure 27 The (acetylene)2 ·(acetonitrile) clathrate crystal structure.163 Note that one of the organic molecular partners contains three nonhydrogen atoms, and the second only two. Color code: H, light blue; N, dark blue; acetylene C, green, and acetonitrile C, purple.

logical synthesis, and outlines the different methods used by researchers toward this end. Clear progress has been made since Powell’s pioneering crystallographic studies, and a number of systems are now understood well enough for reliable predictions to be made. New compounds of these classes may be synthesized by modification of a previously known host, or even from first principles. Closer examination, however, will reveal that the approaches used to date represent only a few drops in the vast ocean of organic chemistry. Synthesis strictly from first principles is still a remarkably rare occurrence. More frequently, a combination of clues from the literature, past experience, and personal intuition is employed. Consider the potent convulsant anisatin 104166 and the antihypertensive diterpenoid isosteviol 105 (Figure 28),167 both of which have been found to behave as clathrands. We can understand after the event why this took place, but their behavior certainly could not have been anticipated with confidence—let alone accurate prediction be made of the precise supramolecular details of inclusion. O

CH3 OH CH3

O H CH3

CH3

O H

HO O

OH OH

H CH3 CO2H

O

104

105

Figure 28 One of the clathrates formed by the antihypertensive diterpenoid isosteviol 105, namely (105)2 ·(benzene). Color code: H, light blue; O, red; host C, green; and guest C, purple. Interhost C=O· · ·H–O hydrogen bonding is indicated by dashed lines.167

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems Many natural products, synthetic pharmaceuticals, and other organic compounds form clathrate compounds due to their unwillingness to adopt Kitaigorodsky close packing in the solid state.168 In such cases, guest molecules frequently are incorporated in preference to leaving void spaces in the crystal. One cause of this behavior is the presence of strong intermolecular attractive forces, such as hydrogen bonding or metal–ligand coordination, that have strong directional requirements. Structures like the clathrate hydrates (see Clathrate Hydrates, Supramolecular Materials Chemistry) and MOFs (see Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry) immediately spring to mind, but this property also has been exploited for organic systems by using molecular tectonics and related strategies for obtaining open host assemblies. A second factor concerns those host molecules that have an awkward shape for close packing just by themselves. Many of the aromatic families described in this chapter fall into this category. The problem here is just what is meant precisely by awkward in strictly supramolecular terms. Some cases are obvious but many others are highly cryptic. Analysis of host supramolecular properties will become an increasingly important aspect of synthetic clathrate design in the future.

ACKNOWLEDGMENTS I thank Dr. Marcia Scudder for preparation of the color figures that illustrate this chapter, and the Australian Research Council for financial support of my research work in this area.

REFERENCES

21

10. E. Weber and H.-P. Josel, J. Inclusion Phenom., 1983, 1, 79. 11. G. R. Desiraju, Angew. Chem. Int. Ed. Engl., 1995, 34, 2311. 12. G. R. Desiraju and T. Steiner, The Weak Hydrogen Bond in Structural Chemistry and Biology, Oxford Science Publications, Oxford, 1999. 13. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, eds. Inclusion Compounds, Academic Press, London, 1984, Vols. 1–3. 14. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, eds. Inclusion Compounds, Oxford University Press, Oxford, 1991, Vols. 4–5. 15. D. D. MacNicol, F. Toda, and R. Bishop, eds. Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, Pergamon Press, Oxford, 1996, vol. 6. 16. F. H. Herbstein, Crystalline Molecular Complexes and Compounds: Structures and Principles, Oxford University Press: Oxford, 2005, Vols. 1–2. 17. M. Farina, G. Di Silvestro, and P. Sozzani, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 371–398, Chapter 12. 18. R. Arrad-Yellin, B. S. Green, M. Knossow, and G. Tsoucaris, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 3, pp. 263–295, Chapter 9. 19. R. Gerdil, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 239–280, Chapter 8. 20. W. D. Ollis and J. F. Stoddart, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 2, pp. 169–205, Chapter 6.

2. L. Mandelcorn, Chem. Rev., 1959, 59, 827.

21. M. D. Hollingsworth and K. D. M. Harris, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 177–237, Chapter 7.

3. M. Hagan, Clathrate Inclusion Compounds, Reinhold, New York, 1962.

22. S.-O. Lee and K. D. M. Harris, Chem. Phys. Lett., 1999, 307, 327.

4. H. M. Powell, G. Huse, and P. W. Cooke, J. Chem. Soc., 1943, 153.

23. J. F. Fait, A. Fitzgerald, C. N. Caughlan, and F. P. McCandless, Acta Crystallogr., Sect. C, 1991, 47, 332.

5. H. M. Powell, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 1, pp. 1–28, Chapter 1.

24. M. C. Etter, Z. Urba˜nczyk-Lipkowska, M. Zia-Ebrahimi, and T. W. Panuto, J. Am. Chem. Soc., 1990, 112, 8415.

1. V. R. Belosludov, M. Y. Lavrentiev, and Y. A. Dyadin, J. Inclusion Phenom. Mol. Recognit. Chem., 1991, 10, 399.

6. Y. A. Dyadin, I. S. Terekhova, T. V. Rodionova, and D. V. Soldatov, J. Struct. Chem., 1999, 40, 645. 7. J. E. D. Davies, W. Kemula, H. M. Powell, and N. O. Smith, J. Inclusion Phenom., 1983, 1, 3. 8. H. M. Powell, J. Chem. Soc., 1948, 61. 9. J. F. Brown Jr., Sci. Am., 1962, 207(7), 82.

25. M. Miyata and K. Sada, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 147–176, Chapter 6. 26. N. Yoswanthananont, M. Miyata, K. Nakano, and K. Sada, in Separations and Reactions in Organic Supramolecular Chemistry, eds. F. Toda and R. Bishop, Wiley, Chichester, 2004, pp. 87–122, Chapter 4.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

22

Supramolecular materials chemistry

27. K. Nakano, E. Mochizuki, N. Yasui, et al., Eur. J. Org. Chem., 2003, 2428.

51. A. Collet and J. Jacques, J. Chem. Soc., Chem. Commun., 1976, 708.

28. K. Nakano, K. Aburaya, I. Hisaki, et al., Chem. Record, 2009, 9, 124.

52. J. H. Gall, A. D. U. Hardy, J. J. McKendrick, and D. D. MacNicol, J. Chem. Soc., Perkin Trans. 2, 1979, 376.

29. K. Aburaya, K. Nakano, K. Sada, et al., Cryst. Growth Des., 2008, 8, 1013.

53. D. D. MacNicol, A. D. U. Hardy, and J. J. McKendrick, J. Chem. Soc., Perkin Trans. 2, 1979, 1072.

30. M. Miyata, N. Tohnai, and I. Hisaki, Acc. Chem. Res., 2007, 40, 694.

54. M. J. Brienne and J. Jacques, Tetrahedron Lett., 1975, 2349.

31. K. Nakano, M. Katsuta, K. Sada, and M. Miyata, CrystEngComm, 2001, 3, 44.

55. G. O. Lloyd and M. W. Bredenkamp, Acta Crystallogr., Sect. A, 2005, 61, o1512.

32. W. R. Scheidt, M. E. Kastner, and K. Hatano, Inorg. Chem., 1978, 17, 706.

56. T. Jacobs, M. W. Bredenkamp, P. H. Neethling, et al., Chem. Commun., 2010, 46, 8341.

33. W. R. Scheidt and Y. J. Lee, Struct. Bonding (Berlin), 1987, 64, 1.

57. G. O. Lloyd, M. W. Bredenkamp, Chem. Commun., 2005, 4053.

34. M. P. Byrn, C. J. Curtis, S. I. Khan, et al., J. Am. Chem. Soc., 1990, 112, 1865.

58. E. Weber, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 535–592, Chapter 17.

35. M. P. Byrn, C. J. Curtis, I. Goldberg, et al., J. Am. Chem. Soc., 1991, 113, 6549. 36. M. P. Byrn, C. J. Curtis, Y. Hsiou, et al., J. Am. Chem. Soc., 1993, 115, 9480. 37. P. Dastidar, Z. Stein, I. Goldberg, and C. E. Strouse, Supramol. Chem., 1996, 7, 257. 38. S. George, S. Lipstman, S. Muniappan, and I. Goldberg, CrystEngComm, 2006, 8, 417. 39. M. P. Byrn, C. J. Curtis, Y. Hsiou, et al., in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 715–732, Chapter 21. 40. I. Goldberg, Top. Stereochem., 2006, 25, 49. 41. I. Goldberg, CrystEngComm, 2008, 10, 637. 42. K. Kobayashi, M. Koyanagi, K. Endo, et al., Chem.—Eur. J., 1998, 4, 417. 43. A. Jacobs, T. le Roex, L. R. Nassimbeni, and F. Toda, Org. Biomol. Chem., 2006, 4, 2452. 44. (a) A. P. Dianin, Zh. Russ. Fiz. Khim. Obshch., 1914, 46, 1310; (b) A. P. Dianin, Chem. Abstr., 1915, 9, 1903. 45. W. Baker, A. J. Floyd, J. F. W. McOmie, et al., J. Chem. Soc., 1956, 2010.

and

L. J. Barbour,

59. T. C. W. Mak and H. N. C. Wong, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 351–369, Chapter 11. 60. P. Dastidar and I. Goldberg, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 305–350, Chapter 10. 61. E. Weber, W. Seichter, and I. Goldberg, Chem. Ber., 1990, 123, 811. 62. E. Weber and M. Czugler, Top. Curr. Chem., 1988, 149, 45. 63. E. Weber, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Oxford University Press, Oxford, 1991, vol. 4, pp. 188–262, Chapter 5. 64. M. Gdaniec, B. T. Ibragimov, and S. A. Talipov, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 117–145, Chapter 5. 65. I. Dance and M. Scudder, Chem.—Eur. J., 1996, 2, 481.

46. J. L. Flippen, J. Karle, and I. L. Karle, J. Am. Chem. Soc., 1970, 92, 3749.

66. M. Scudder and I. Dance, J. Chem. Soc., Dalton Trans., 2000, 2909.

47. D. D. MacNicol, J. J. McKendrick, and D. R. Wilson, Chem. Soc. Rev., 1978, 7, 65.

67. L. bin Din and O. Meth-Cohen, J. Chem. Soc., Chem. Commun., 1977, 741.

48. D. D. MacNicol, Structure and design of inclusion compounds: the clathrates of hydroquinone, phenol, Dianin’s compound and related systems, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press: London, 1984, vol. 2, pp. 1–45, Chapter 1.

68. L. Pang and F. Brisse, Can. J. Chem., 1994, 72, 2318.

49. P. Finocchiaro and S. Failla, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 593–642, Chapter 18. 50. A. D. U. Hardy, J. J. McKendrick, and D. D. MacNicol, J. Chem. Soc., Chem. Commun., 1976, 355.

69. E. Weber, N. D¨orpinghaus, and I. Cs¨oregh, J. Chem. Soc., Perkin Trans. 2, 1990, 2167. 70. I. Cs¨oregh, E. Weber, L. R. Nassimbeni, et al., J. Chem. Soc., Perkin Trans. 2, 1993, 1775. 71. E. Weber, W. Seichter, K. Skobridis, et al., J. Inclusion Phenom. Macrocycl. Chem., 2006, 55, 131. 72. F. Toda, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 465–516, Chapter 15.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Synthetic clathrate systems 73. F. Toda and K. Agaki, Tetrahedron Lett., 1968, 9, 3695. 74. F. Toda, D. L. Ward, and H. Hart, Tetrahedron Lett., 1981, 22, 3865. 75. H. Hart, L.-T. W. Lin, and D. L. Ward, J. Am. Chem. Soc., 1984, 106, 4043. 76. L. J. Barbour, M. R. Caira, L. R. Nassimbeni, Supramol. Chem., 1995, 5, 153.

et al.,

77. D. V. Soldatov, J. Chem. Crystallogr., 2006, 36, 747. 78. S. A. Bourne, L. R. Nassimbeni, and F. Toda, J. Chem. Soc., Perkin Trans. 2, 1991, 1335. 79. A. Jacobs, N. L. Z. Masuku, L. R. Nassimbeni, J. H. Taljaard, CrystEngComm, 2008, 10, 322.

and

80. E. Weber, K. Skobridis, A. Wierig, et al., J. Chem. Soc., Perkin Trans. 2, 1992, 2123. 81. Y. Mazaki, K. Awaga, and K. Kobayashi, J. Chem. Soc., Chem. Commun., 1992, 1661. 82. M. I. Bruce, K. R. Grundy, M. J. Liddell, Organomet. Chem., 1988, 344, C49.

et al.,

J.

83. C. E. Godinez, G. Zepada, and M. A. Garcia-Garibay, J. Am. Chem. Soc., 2002, 124, 4701. 84. F. Maharaj, R. Bishop, D. C. Craig, et al., Cryst. Growth Des., 2009, 9, 1334. 85. V. Mach´acek, V. Bertolasi, P. Simunek, et al., Cryst. Growth Des., 2010, 10, 85. 86. F. Toda and K. Tanaka, Tetrahedron Lett., 1988, 29, 551. 87. M. R. Caira and K. Tanaka, Top. Heterocycl. Chem., 2009, 18, 75. 88. S. C. Zimmerman, Top. Curr. Chem., 1993, 165, 71. 89. M. Harmata and C. L. Barnes, J. Am. Chem. Soc., 1990, 112, 5655. 90. F. Diederich, Cyclophanes, Royal Society of Chemistry, Cambridge, 1991. 91. Y. Murakami, J. Kikuchi, and Y. Hisaeda, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Oxford University Press, Oxford, 1991, vol. 4, pp. 448–478, Chapter 11. 92. C. Krieger and F. Diederich, Chem. Ber., 1985, 118, 3620.

23

100. G. A. Downing, C. S. Frampton, J. H. Gall, and D. D. MacNicol, Angew. Chem. Int. Ed. Engl., 1996, 35, 1547. 101. C. S. Frampton, W. M. McGregor, D. D. MacNicol, et al., Supramol. Chem., 1994, 3, 223. 102. G. A. Downing, C. S. Frampton, D. D. MacNicol, and P. R. Mallinson, Angew. Chem. Int. Ed. Engl., 1994, 33, 1587. 103. J. H. Gall, D. D. MacNicol, P. R. Mallinson, and P. A. Welsh, Tetrahedron Lett., 1985, 26, 4005. 104. R. Bishop and I. Dance, J. Chem. Soc., Chem. Commun., 1979, 992. 105. A. T. Ung, D. Gizachew, R. Bishop, et al., J. Am. Chem. Soc., 1995, 117, 8745. 106. S. Kim, R. Bishop, D. C. Craig, et al., J. Org. Chem, 2002, 67, 3221. 107. I. G. Dance, R. Bishop, S. C. Hawkins, et al., J. Chem. Soc., Perkin Trans. 2, 1986, 1299. 108. R. Bishop, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 85–115, Chapter 4. 109. F. H. Herbstein, Crystalline Molecular Complexes and Compounds: Structures and Principles, Oxford University Press, Oxford, 2005, vol. 1, pp. 251–268. 110. R. Bishop, Acc. Chem. Res., 2009, 42, 67. 111. F. H. Herbstein, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 61–83, Chapter 3. 112. O. Ermer and L. Lindenberg, Helv. Chim. Acta, 1991, 74, 825. 113. J. P. Glusker, Top. Curr. Chem., 1998, 198, 1. 114. M. Simard, D. Su, and J. D. Wuest, J. Am. Chem. Soc., 1991, 113, 4696. 115. X. Wang, M. Simard, and J. D. Wuest, J. Am. Chem. Soc., 1994, 116, 12119. 116. O. Saied, T. Maris, and J. D. Wuest, J. Am. Chem. Soc., 2003, 125, 14956.

93. D. D. MacNicol and D. R. Wilson, J. Chem. Soc., Chem. Commun., 1976, 494.

117. J. D. Wuest, Chem. Commun., 2005, 5830.

94. D. E. Palin and H. M. Powell, J. Chem. Soc., 1947, 208.

119. T. C. W. Mak and B. R. F. Bracke, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 23–60, Chapter 2.

95. A. D. U. Hardy, D. D. MacNicol, S. Swanson, and D. R. Wilson, J. Chem. Soc., Perkin Trans. 2, 1980, 999. 96. D. D. MacNicol, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 2, pp. 123–168, Chapter 5. 97. D. Das and L. J. Barbour, Cryst. Growth Des., 2009, 9, 1599. 98. D. D. MacNicol and G. A. Downing, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 421–464, Chapter 14. 99. L. J. Farrugia, J. H. Gall, D. D. MacNicol, MacSween, Chem. Commun., 2010, 46, 5241.

and

R.

118. M. J. Zaworotko, Chem. Soc. Rev., 1994, 23, 283.

120. K. Endo, T. Ezuhara, M. Koyanagi, et al., J. Am. Chem. Soc., 1997, 119, 499. 121. K. Endo, T. Sawaki, M. Koyanagi, et al., J. Am. Chem. Soc., 1995, 117, 8341. 122. Y. Aoyama, K. Endo, T. Anzai, et al., J. Am. Chem. Soc., 1996, 118, 5562. 123. K. Endo, T. Koike, T. Sawaki, et al., J. Am. Chem. Soc., 1997, 119, 4117. 124. T. Tanaka, K. Endo, and Y. Aoyama, Chem. Lett., 2000, 29, 1424.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

24

Supramolecular materials chemistry 148. R. Bishop, Crystal engineering of halogenated heteroaromatic clathrate systems, in Frontiers in Crystal Engineering, eds. E. R. T. Tiekink and J. J. Vittal, Wiley, Chichester, 2006, vol. 1, pp. 91–116, Chapter 5.

125. Y. Aoyama, Top. Curr. Chem., 1998, 198, 131. 126. Y. Aoyama, Bull. Chem. Soc. Jpn., 2009, 82, 419. 127. O. Ermer, Helv. Chim. Acta, 1991, 74, 1339. 128. O. Ermer and C. R¨obke, J. Am. Chem. Soc., 1993, 115, 10077. 129. E. M. Veen, P. M. Postma, H. T. Jonkman, et al., Chem. Commun., 1999, 1709. 130. H. I. S¨uss, M. Lutz, and J. Hulliger, CrystEngComm, 2002, 4, 610.

149. R. Bishop, Top. Heterocycl. Chem., 2009, 18, 37. 150. R. Bishop, M. L. Scudder, D. C. Craig, et al., Mol. Cryst. Liq. Cryst., 2005, 440, 173. 151. C. E. Marjo, M. L. Scudder, D. C. Craig, and R. Bishop, J. Chem. Soc., Perkin Trans. 2, 1997, 2099.

131. H. R. Allcock, Acc. Chem. Res., 1978, 11, 81.

152. K. Tanaka, D. Fujimoto, and F. Toda, Tetrahedron Lett., 2000, 41, 6095.

132. H. R. Allcock, in Inclusion Compounds, eds. J. L. Atwood, J. E. D. Davies, and D. D. MacNicol, Academic Press, London, 1984, vol. 1, pp. 351–374, Chapter 8.

153. K. Tanaka, S. Wada, and M. R. Caira, Tetrahedron, 2007, 63, 9213.

133. T. Kobayashi, S. Isoda, and K. Kubono, in Comprehensive Supramolecular Chemistry, Solid-State Supramolecular Chemistry: Crystal Engineering, eds. D. D. MacNicol, F. Toda, and R. Bishop, Pergamon, Oxford, 1996, vol. 6, pp. 399–419, Chapter 13.

155. S. Basavoju, S. Aitipamula, and G. R. Desiraju, CrystEngComm, 2004, 6, 120.

134. P. Sozzani, S. Bracco, A. Comotti, et al., Angew. Chem. Int. Ed., 2005, 44, 1816. 135. D. D. MacNicol and S. Swanson, Tetrahedron Lett., 1977, 2969. 136. A. S. Jessiman, D. D. MacNicol, P. R. Mallinson, and I. Vallance, J. Chem. Soc., Chem. Commun., 1990, 1619. 137. V. S. S. Kumar, F. C. Pigge, and N. P. Rath, CrystEngComm, 2004, 6, 531. 138. M. Czugler, E. Weber, L. P´ark´anyi, et al., Chem.—Eur. J., 2003, 9, 3741.

154. J. Ashmore, R. Bishop, D. C. Craig, and M. L. Scudder, CrystEngComm, 2008, 10, 131.

156. E. Gagnon, A. Rochefort, V. M´etivaud, and J. D. Wuest, Org. Lett., 2010, 12, 380. 157. S. Varughese, G. Cooke, and S. M. Draper, CrystEngComm, 2009, 11, 1505. 158. S. Varughese and S. M. Draper, Cryst. Growth Des., 2010, 10, 2571. 159. R. Bishop, D. C. Craig, I. G. Dance, et al., J. Struct. Chem., 1999, 40, 663. 160. V. T. Nguyen, R. Bishop, I. Y. H. Chan, et al., CrystEngComm, 2008, 10, 1810. 161. H. M. Powell, J. Chem. Soc., 1950, 468.

139. R. K. R. Jetti, P. K. Thallapally, F. Xue, et al., Tetrahedron, 2000, 56, 6707.

162. D. S. Yufit and J. A. K. Howard, CrystEngComm, 2010, 12, 737.

140. R. K. R. Jetti, A. Nangia, F. Xue, and T. C. W. Mak, Chem. Commun., 2001, 919.

163. M. T. Kirchner, D. Bl¨aser, and R. Boese, Chem.—Eur. J., 2010, 16, 2131.

141. B. K. Saha, R. K. R. Jetti, L. S. Reddy, et al., Cryst. Growth Des., 2005, 5, 887.

164. S. R. Vippagunta, H. G. Brittain, and D. J. W. Grant, Adv. Drug Delivery Rev., 2001, 48, 3.

142. B. K. Saha and A. Nangia, Cryst. Growth Des., 2007, 7, 393.

165. D. Braga, M. Curzi, S. L. Giaffreda, et al., in Organic Nanostructures, eds. J. L. Atwood and J. W. Steed, WileyVCH Verlag, Weinheim, 2008, pp. 155–177.

143. K. Reichenb¨acher, H. I. S¨uss, H. Stoeckli-Evans, et al., New J. Chem., 2004, 28, 393. 144. K. Reichenb¨acher, A. Neels, H. Stoeckli-Evans, et al., Cryst. Growth Des., 2007, 7, 1399. 145. J. Veciana, J. Carilla, C. Miravitlles, and E. Molins, J. Chem. Soc., Chem. Commun., 1987, 812. 146. A. Bacchi, M. Carcelli, T. Chiodo, et al., Cryst. Growth Des., 2009, 9, 3749. 147. L. Ballie, L. J. Farrugia, D. D. MacNicol, MacSween, CrystEngComm, 2004, 6, 326.

and

166. M. G. Wong, J. M. Gulbis, M. F. Mackay, et al., Aust. J. Chem., 1988, 41, 1071. 167. O. V. Andreeva, B. F. Garifullin, A. T. Gubaidullin, et al., J. Struct. Chem., 2007, 48, 540. 168. A. I. Kitaigorodsky, Molecular Crystals and Molecules, Academic Press, New York, 1973.

R.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc118

Network and Graph Set Analysis ¨ Lars R. Ohrstr¨ om Chalmers University of Technology, Gothenburg, Sweden

1 What this Chapter can do for you 2 Understanding your Products and Structures 3 Comparing your Products and Structures to Networks in the Literature 4 The Network Analysis Approach to Making Something New by Design 5 The Topology Approach to Network Analysis: a Brief Guide 6 Graph Set Analysis: a Brief Guide 7 Conclusions Notes References

1

1 3 6 8 9 13 15 15 15

WHAT THIS CHAPTER CAN DO FOR YOU

We approach our scientific endeavors with a Kennedyean attitude1 ; we do not ask what science can do for us, we ask what we can do for science. On the contrary, however, the purpose of this chapter is essentially to do something for you, so that ultimately you can improve the presentation and understanding of whatever supramolecular projects you are pursuing. At a first glance, both network analysis2 and graph set analysis3 deal with chemical objects that have already been created and might thus superficially be dismissed as nomenclature exercises of little value to the creative chemist.

However, rejecting both network and graph set analyses as mere workouts in taxonomy is completely wrong; both opportunities of a deeper understanding and of an efficient presentation may be lost in the process. Moreover, these two very different approaches also have a forward looking component, thus enabling us, in principle, to design solid-state molecule-based materials, to do crystal engineering (see Crystal Engineering, Supramolecular Materials Chemistry). Another obstacle, the mathematical formalism that is essential to incorporate these methods into the greater framework of science, should not discourage us either. While this is absolutely fundamental for the development and understanding of the nomenclature and formal definitions of the descriptors and symbols we use to analyze the network topologies, the average chemist only needs to use the basic concepts. The more advanced terminology, such as the gender of a net, comes with the bargain, so to speak, of incorporating the mathematical formulation of periodic graphs into structural chemistry, but this needs not disturb us as we keep this chapter at the basic level.

1.1

What is network and graph set analysis?

In brief, network analysis in the sense employed in this work deals with the network topology of crystalline compounds that can be described as 2D or 3D nets.4 There are no particular criteria in terms of chemical bonding or chemical constituents; DNA, proteins, organic molecules, or coordination compounds can construct the net, but two things are essential: 1. A crystal structure should be obtained with reasonable resolution. 2. The description of the structure as a net is useful.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

2

Supramolecular materials chemistry

Figure 1 (a) network description of a cubic closed-packed metal structure as a 12-connected net may be misleading, although there are a reasonable number of examples of 12-connected nets in the literature. (b) a network description of the diamond structure is appropriate (dia net), and simple inorganic materials like this are often referred to as network solids.

Figure 2 Of these three three-connected (3-c) nets, two are topologically identical, as no bonds have to be broken when transforming one into the other. From left to right, their symbols are hcb, fes, and fes.

While the first condition is self-evident, the second is necessarily fuzzy and subjective. We take two examples: many metals can be described as close-packed structures, but it is also possible to draw connecting lines between nearest neighbors and regard both the hexagonal and cubic close packing as networks (Figure 1). The latter exercise, however, does very little to help us understand the structure and might actually be misleading. On the other hand, diamond can be described as a cubic close packing of half the carbon atoms, with the other half filling in 50% of the tetrahedral holes created by the close packing. This is also misleading, and drawing C–C bonds and regarding this structure as a network is both easier for understanding what the structure looks like and chemically more reasonable.

O O (a)

H H R2,2(8)

It should perhaps be mentioned here for clarity that topology 5 deals with properties that are invariant for an object even if it is subject to stretching, bending, or compression (Figure 2). Thus, two nets are topologically different if they cannot be transformed into each other without breaking any of the network bonds. Graph set analysis, on the other hand, deals specifically with hydrogen-bonded systems, and is used to describe hydrogen-bond links between different molecules. Thus, the graph set symbols that we calculate for such an interaction describes the intermolecular synthon,6 the specific bond, but not the molecules that bind the network together (Figure 3). In principle, one could adopt a similar nomenclature describing synthons formed by coordination bonds but these

N

O

O

O (b)

H H

O O

R2,2(8)

Figure 3 Both the carboxylic acid dimer synthon and the carboxylic acid—amide synthon—have the same graph set symbol R2,2(8), where “R” is for ring, “2,2” stands for the number of hydrogen-bond donor and acceptor atoms, and (8) is the number of atoms in the ring. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis are commonly, and less formally, described in terms of their secondary building units (SBUs).7 In this chapter, we adopt the nomenclature of O’Keeffe and coworkers for network topology (which, for example, uses the symbol dia for the diamond net),8 and for the graph set analysis we rely on the inventors of this concept, Etter and Bernstein.3, 9

What can network analysis do for you?

N

N

N

N Cr

N

N N

N N

N N

N

1.2

N

N

3

N

N

N

Cr N

N

N

N

N

N

Cr N

N

N N

There are four important points from which you could profit from network analysis: 1. 2. 3. 4.

Understand the products you get. Compare these materials to what others have made. Efficiently communicate your results to colleagues. Truly make something new by design.

In this chapter, we first exemplify these arguments (in Sections 2–4) and then go into details of how the actual network analysis is done and what software to use (Section 5). Throughout these examples, you will find network topology symbols as three-letter codes (i.e., dia), and occasional references to specific terms that are defined in more detail in the latter part of the text. In Section 5, we also explain some of the mathematical concepts of this method, especially the meaning of the sometimes rather longish set of figures that usually accompany the topology description such as the point symbol specifying the dia net as 66 . Finally, in Section 6, a brief introduction to graph set analysis is given.

2

2.1

UNDERSTANDING YOUR PRODUCTS AND STRUCTURES Molecule-based magnetic materials

One paradigm for constructing a molecule-based material that has bulk magnetic properties is that there have to be strong ferromagnetic interactions between the unpaired spins in all three directions of space. This implies a 3D network; this is true for the Prussian blue–based compounds10 and most likely for the V(TCNE)-based materials,11 although the latter are notoriously amorphous. In 2008, Joel Miller and coworkers reported the compound Cr(II)[C3 (CN)5 ]2 ·2CH3 CN·H2 O, 1, isostructural with M(II)[C3 (CN)5 ]2 (M = Mn, Fe).12 As such, these compounds are not bulk magnets, but the antiferromagnetic couplings transmitted through the polynitrile ligands are rather strong.

N

Figure 4 Schematic presentation of the connectivity in Cr(II)[C3 (CN)5 ]2 ·2CH3 CN·H2 O, 1.12 Only three of the six pentacyanopropenide radical ligands around Cr are shown explicitly.

The structure, however, was intriguing at first, with six ligands coordinated to every chromium, and each of these ligands in turn bound three metal ions (Figure 4). The resulting three-dimensional network is not easy to grasp, and in such a situation, the authors have three possibilities: 1. Dismiss the whole problem with the phrase “a complicated 3D network.” 2. Make a rather lengthy description of the main features of the network such as rings, planes, and so on to try to generate a mental picture of the structure in the reader’s mind. 3. Analyze the network topology of the structure. Applying the third alternative, it was established that this structure is, in fact, an example of a network from a wellknown “type structure” that many chemists are familiar with from general chemistry courses,13 namely the rutile (the most common form of TiO2 ) structure. From a topology point of view, this is one of the most symmetric, and also most abundant, of the six- and three-connected nets (6c,3c nets) and it has the symbol rtl.14 Thus, a seemingly complicated structure can be reduced to a well-known net related to a basic structure found in general chemistry textbooks. This is of course a great advantage, and it is very important to use the possibility of building on existing science in order to not reinvent the wheel. Thus, lengthy descriptions of the structure of, for example, diamond, are best avoided in articles dealing with materials that form dia- or diamond-nets (such examples do exist!). A comparison between the rtl net in 1 (with the threeconnected vertices placed on the central carbons of the ligand) and the most symmetric form can be seen in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

4

Supramolecular materials chemistry

Figure 5 The rtl or rutile net in Cr(II)[C3 (CN)5 ]2 ·2CH3 CN·H2 O, is pictured to the left with solvent molecules shown as dashed spheres. Six-connected vertices (nodes) are at Cr and three-connected vertices at the central ligand carbon (see Figure 4). To the right, the ideal rtl net is shown for comparison.

Figure 5. As we deal with topology, there are slight differences between the two, and this might be the correct place to make a crystallographic comment and point out that a material containing a certain net (as the rtl) does not have to have the same space group as the most symmetric form of the net. In this case, compound 1 crystallizes in space group Pbca and the “ideal” rtl net has space group P 42 /mnm. The connoisseurs would call this the embedding of the net.15 It is also not necessary that the number of vertices (or nodes)[1] in the ideal net correspond to the number of crystallographic independent atoms or molecules at the center of each vertex in the real structure.

2.2

Hydrogen-bonded networks

For these materials, it is useful to consider both network analysis and graph set analysis. As an example, we take the [M(H2 biimidazole)3 ]n+ building blocks.16 These have six acidic hydrogens available for hydrogen bonding and can be connected by a variety of synthons, some of the most efficient having the graph set symbol R(2,2)9 (Figure 6). A reaction between the [M(H2 biimidazole)3 ]n+ building block and benzene-1,3,5-tricarboxylic acid (benzene-1,3,5tricarboxylic acid) results, if all goes well and no interfering hydrogen bonds occur, in networks where one metal

n+ HN HN

NH

N N

N

N

N

N H

N H 2

HOOC

NH N

M

N

N H

O

N

N H

O

M

X(

R(2,2)9

COOH

)m COOH Benzene-1,3,5-tricarboxylic acid 3

Figure 6 The [M(H2 biimidazole)3 ]n+ cation can be connected by a variety of synthons, notably those having the graph set symbol R(2,2)9, that is, anions of benzene-1,3,5-tricarboxylic acid shown to the right, or sulfate ions. Note that protonation states might vary for the two components but the graph set symbol remains the same. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis complex connects to three benzene-1,3,5-tricarboxylic acids and every benzene-1,3,5-tricarboxylic acid connects to three complex ions, thus a three-connected net. (We disregard the exact position of the protons, we may have COOH . . . Hbiimz or COO− . . . H2 biimz; this is irrelevant to the following discussion.)

2.2.1 Predicting or arguing about the structure from a network synthesis If we want to predict or understand the outcome of a reaction between the building blocks in Figure 6, we need to consider three things: 1. The geometry and chirality of the tris-bischelated metal complex.

83-eta net

103-srs net

103-bto net

Figure 7

5

2. The flatness and geometry of the connecting benzene1,3,5-tricarboxylic acid building block. 3. The geometry of the most common three-connected nets. First, we can conclude that conditions (1) and (2) mean that every vertex with its three next neighbors will lie in a plane with 120◦ angles between propagation vectors of equal length. This sets a basic requirement on which nets are possible, and we may note that the simplest solution is the 2D honeycomb net (see Figure 7). For 3D nets, this condition means that the choice, in principle, is restricted to two nets forming 8-membered rings of vertices, the chiral eta net and the non-chiral etb net, and three nets forming 10-membered rings of vertices, the srs, ths, and bto, of which both srs and bto are chiral (Figure 7). (These nets are also known as (8,3)-a,

83-etb net

103-ths net

63-hcb net (2D layer)

Possible nets combining [M(H2 biimidazole)3 ]n+ building blocks and benzene-1,3,5-tricarboxylic acid.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

6

Supramolecular materials chemistry

(8,3)-b, SrSi2 or (10,3)-a), ThSi2 or (10,3)-b, and (10,3)-c, respectively.4a ) Of these, the nets with 8-membered rings as well as the bto net are rare,17 and we therefore concentrate on the difference between the srs and ths nets. The angles of the out-of-plane rotation of the propagation vectors in the srs net is 71◦ , whereas the two most common embeddings of the ths net has out-of-plane turnings of 90◦ and 60◦ , respectively, and one set of vectors planar, forming a characteristic zigzag chain (see Figure 7).18 Now, in attaching two [M(H2 biimidazole)3 ]n+ building blocks on two carboxylates of the benzene-1,3,5-tricarboxylic acid with an ideal, perfectly planar, R(2,2)9 hydrogen bond, we have two cases, either these complexes have the same (

or ) or opposite chirality (
). Two factors disfavor the srs net for these systems: spontaneous resolution of the complexes (formation of a conglomerate) is required as only the same chirality of the complexes continues to twist the propagation vectors out of plane and this twist angle is a bit removed from the ideal 71◦ . On the other hand, with three connections to the tricarboxylate building block, we have, for example,
 configurations around the flat benzene unit. This gives two distinctive sets of out-of-plane rotations, two of the
 type and one of the

type. While none of these are going to give the flat zigzag chains of the ths net, two different dihedral angles between the propagation vectors is a distinctive feature of this net. We thus conclude that the ths is more likely to form from these building blocks rather than the srs net. In reality, there is significant tolerance in the hydrogen bonds, so there is a possibility for the system to adapt; thus, the 2D hcb net has been observed with both racemic and enantiopure layers.19 Nevertheless, the only 3D net observed for these systems is the ths net, predicted by these rudimentary considerations of geometry.19, 20 Details of the geometries and possibilities of the three-connected 10-gon nets can be found elsewhere.18 A final consideration is the density of the net, and the possibilities of interpenetration (see Interpenetration, Supramolecular Materials Chemistry), that is, two or more interwoven nets in the same structure.21 Both the srs and the ths nets have good geometrical properties for interpenetration as an identical net fits exactly into the voids of the first (these are called “self-dual” nets; for a more stringent definition see ref. 22), whereas this is not true for the other 3D nets in Figure 6.23 The srs net is also less dense than the ths net, and this may also be a factor to consider. In conclusion, we see that it is possible to discuss why a certain structure is formed using arguments based on the geometry, density, and other properties of the ideal nets. In Section 4, we see how this can be turned around to help

us synthesize new materials, but first we discuss another important aspect of network analysis.

3

COMPARING YOUR PRODUCTS AND STRUCTURES TO NETWORKS IN THE LITERATURE

An essential part of science is putting your work in perspective by correct referencing to earlier work and correctly assessing similarities and differences compared to what others have accomplished. Whether your are dealing with coordination polymers in general (see Coordination Polymers, Supramolecular Materials Chemistry), MOF-5 (see Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry) in particular, or hydrogen-bonded network materials, correctly assigning the topology of your net and, perhaps, also those of related compounds in the literature is of great help. Note also that network topology is related to Supramolecular Isomerism, Supramolecular Materials Chemistry.

3.1

Two isoreticular metal–organic frameworks

The term reticular refers to anything with a net or network property; in Somalia and neighboring countries, for example, one can, with a bit of luck, find the reticulated giraffe[2]. The term reticular chemistry has been coined to describe the chemistry of making network compounds.24 So far, this term has been used only sparingly in scientific literature. However, we find the related word isoreticular very useful, as there is no other term describing two compounds that form nets with the same topology but with different chemical constituents. Thus, the famous series of compounds by the Yaghi group, commencing with MOF-5 (see Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry) and a pcu net (pcu for primitive cubic, also known as a α-polonium net), and the subsequent compounds with different aromatic 1,4-dicarboxylates but the same Zn4 O6+ core all form pcu nets and are isoreticular.25 Likewise, the series of compounds based on [M[(Co(en)(ox)2 ]2 ] forming scu nets with different M2+ ions (the scu-net is based on squares and cubes) are all isoreticular, but crystallize in different space groups and with varying unit cells.26 Thus, isoreticular compounds are not simply isostructural. To show how this is useful, we compare the two metal–organic frameworks MOF-CJ327 and UCT-1,28 both prepared with the anion of benzene-1,3,5-tricarboxylic acid, 3 . Formulas and crystallographic parameters are given in Table 1.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis

7

Table 1 Comparison of the two metal-organic frameworks MOF-CJ327 and UCT-1,28 prepared with the anion of benzene-1,3,5-tricarboxylic acid, 3 . MOF-CJ327 4

UCT-128 5

H3 O[Zn3 (OH)(3 )2 (H2 O)2 (DMF)]

Formula Space group ˚ a [A] ˚ b [A] ˚ c [A] α[◦ ] β[◦ ] γ [◦ ]

I 4cm 20.588 20.588 17.832 90 90 90

At first glance, only the c cell parameters look similar in these two structures, and this may well be a coincidence. A closer inspection shows the similarities in the Zn-clusters; UCT-1 seems to have a double version of the Zn3 cluster in CJ3, except for the different numbers of DMF molecules. This effect is as a result of nine crystallographically independent Zn(II) ions in UCT1, and observing the details of the structures reveals that these clusters are indeed very similar, and the H3 O+ ion in CJ3 corresponds to the [Zn(H2 O)3 (DMF)3 ]2+ ion in UCT-1. Further, are the two networks the same too? An atom-by-atom comparison of the two structures is very tedious and prone to errors; however, there are computational tools that can do this for us. Thus, a reliable procedure using the SYSTRE29 and TOPOS30 programs (see Section 5.4) established that the two network topologies are indeed the same, yielding identical vertex symbols (4·4·6·6·6·6·6·6·6·6·62 ·62 ·82 ·82 ·87 , 4·62 ·62 basically counting the different rings in the net, see Section 5.4) and coordination sequences up to the 10th nearest neighbor (average C10 = 1242.3333, see Section 5.4). These structures both contain a six- and three-connected net with the symbol sab, and the vertex assignment and the ideal

O

O

Zn

H O

form of the net (calculated with SYSTRE29 ) are shown in Figure 8. From this example, we conclude that network analysis is a very convenient way of comparing network structures when the complexity of the compounds makes visual comparisons difficult.

3.2

Hydrogen-bonded tubular inclusion compounds

Synthetic clathrate systems are described in another chapter of this work (see Synthetic Clathrate Systems, Supramolecular Materials Chemistry). Here we just give an example showing how these are sometimes well described by 3D nets, and how network analysis can be used to compare different compounds. The diol derivatives of bicyclo[3.3.1]nonanes (Figure 9) are such systems, and the unique molecular shape of this building block has been used to form self-assembled supramolecular structures and inclusion complexes with various guest molecules.31 Some derivatives of this ring skeleton, the so-called tubuland diols, give controllable crystal structures with a variety of inclusion guests.32

O

Zn

O

O

O

O O

O

O

O

O

[Zn(H2 O)3 (DMF)3 ]·[Zn6 (OH)2 (3 )4 (DMF)2.5 (H2 O)2 ]·3.1H2 O Cmc21 29.555 28.856 17.719 90 90 90

O

O

O

Zn O DMF

O

O

O

O

Figure 8 Schematic vertex assignment (a) and the most symmetric form of the sab net (b) found in MOF-CJ3 and UCT-1. Note that vertices are chosen where there are no atoms in the center of the Zn-cluster (six connected, blue) and in the middle of the benzene rings (three connected, red).28 . Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

8

Supramolecular materials chemistry

H O

O

O

HO

HO

OH

OH rac - 7

H

O H

6

H

O

rac - 8

H

O H

H

O

O H

Figure 9 Diol derivatives (7–8) of bicyclo[3.3.1]nonane (6) and vertex assignment in the 3D nets formed by 7 and 8. Note that neither 7 nor 8 forms closed hydrogen-bonded rings; all closed paths contain the blue intramolecular link. Moreover, both 7 and 8, although prepared as racemates (rac), crystallize in enantiomerically pure forms (conglomerates).

Figure 10 The utg net in the conglomerate of 7 and the eta net (or 8,3-a net) in the conglomerate of 8.35a Dashed lines are the bicyclononane frameworks. The inclusion molecules in 8 are not shown.

The solid-state structures of these compounds are, at least at a cursory glance, dominated by OH . . . OH hydrogen bonds. Interestingly, the supramolecular chemistry of such systems can be traced back to the structure determination of 1,3-dihydroxybenzene in 1936,33 and Alexander Wells’ subsequent analysis of this structure, interpreting the topology as a utp-net (or (10,3)-d net with his own nomenclature).34 Both the endo,endo-bicyclo[3.3.1]nonane-2,6-diol 7,35 and exo,exo-2,6-dimethyl-bicyclo[3.3.1]nonane-2,6-diol 8 chloroacetic acid clathrate31b crystallize as conglomerates (in chiral space groups P 41 21 2, and P 31 21, respectively) and thus exhibit spontaneous resolution of the racemate. The nets formed in these structures are based on exactly the same kind of three-connected hydrogen-bond pattern, but network analysis shows them to have two very different topologies, namely the chiral nets (82 .12)-utg and (83 )-eta (Figure 7), see Figure 10. From this exercise, we can thus not only conclude that the nets in these two similar diols are different but also that

8 should be more prone to form inclusion compounds as the eta net is much less dense than the utg net.

4

4.1

THE NETWORK ANALYSIS APPROACH TO MAKING SOMETHING NEW BY DESIGN The supramolecular synthon (SS) and SBU

The ideal net topologies are not only useful when understanding and comparing structures, they can also serve as blueprints in the design of new network compounds. Thus, while the most symmetric nets, as the dia net, were an inspiration for the early work by Robson and his group,36 the catalogs of different nets that can be found on-line in the RCSR (Reticular Chemistry Structural Resource database)8 and in other sources4, 37 can serve as an inspiration to twenty-first-century synthetic chemists.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis

O O Fe O OO

O

9

O Fe OO

+

O Fe O O O

Figure 11 The acs net is constructed by joining regular trigonal prisms (left), and a retrosynthetic analysis suggested the two secondary building blocks (SBUs) to the right as suitable starting points.

Similar to the supramolecular synthon (SS) and SBU approaches, the retrosynthetic analysis of supramolecular and coordination-bond systems are described elsewhere (see Synthetic Clathrate Systems and Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties, Supramolecular Materials Chemistry); we do not dwell on how to connect the vertices but rather go straight to an example.

4.2

Design of a metal–organic framework

The simplest and most common nets can be constructed by joining simple, highly symmetric, geometrical coordination figures. Thus, the dia net is made from tetrahedra and the pcu net is made from octahedra, and examples of these nets are well known. However, what would you get if you joined six-connected vertices with trigonal prismatic geometry instead of octahedral? This question was answered by Delgado-Friedrich et al. in 2003, and the resulting network topology was given the symbol acs (Figure 11).38 The Yaghi group then prepared this net. The starting point was to find a suitable SBU, and the choice was the [(COO)6 (Fe3 O)] core and benzene dicarboxylates as linkers, also shown in Figure 11.39 This approach was successful, both with benzene1,3-dicarboxylate and benzene-1,4-dicarboxylate,39 thus proving the point that rational preparation of network compounds is possible with the help of ideal nets as blueprints. In this context, we should also mention that this design concept might also help in narrowing down the possibilities in case the structure has to be solved from powder diffraction data only and thus be implemental in obtaining structural information.40

A few key concepts have been introduced and given some preliminary explanations. In this section, we give step-bystep instructions on how to obtain the network topology symbols.

5.1

First, it has to be decided where the vertices (or nodes) are in your network. Often this is straightforward for coordination polymers; simply putting the vertices on the metal atoms, or in the center of the cluster, is usually the obvious choice. However, in a number of examples shown here, this is not possible. For example, the diols 7 and 8 both connect to four other diols by four sets of hydrogen bonds. It is possible, and quite correct, to put vertices in the centers of the bicyclonane frame and make a four-connected net. However, such a net will contain links that pass through “empty space” in the structure, and the choice of the oxygen atoms as vertices will generate a network that is more informative and better related to the structure. We suggest the following:37 A path along the net should follow the strongest directional intermolecular interactions (or bonds) and be significantly stronger than all other directional, short-range intermolecular interactions. The general idea, however, is that the resulting network should be useful in interpreting, understating, or comparing your structure. A few basic ideas that should kept in mind are as follows: 1.

2. 3.

5

THE TOPOLOGY APPROACH TO NETWORK ANALYSIS: A BRIEF GUIDE

We have now seen a number of examples of network analysis and different reasons to embark on such studies.

Node, or vertex, assignment

5.2

Links between vertices should follow the molecules and not go through empty space in the structure (solvent filled voids). Unnecessarily complicated multinodal nets should be avoided. They should be simplified if possible. Note that some structures are not well described as nets.

Construction of the net

After choosing the vertices, the net is constructed by joining these vertices by links, and this can be visualized in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

10

Supramolecular materials chemistry

most software that can handle crystallographic information. For comparison, crystallographic parameters for a large number of nets can be found on-line in the RSCR14 ; a more limited collection is found in a 2005 monograph also adhering to the O’Keeffe three-letter code nomenclature,37 and there is the seminal work of Alexander Wells from 1977,4a and O’Keeffe and Hyde from 1996.4b A more mathematical database is also freely available on the Internet.41 Before deleting or hiding all atoms, the vertices should be saved and the structure should be reduced to a net, displaying the simplified vertices-and-links-only network superimposed on the real structure to check that the simplified net does indeed correspond to the real one.

5.3

Network topology by inspection

With a little experience, and by comparing the experimental net with ideal nets from the databases, most of the common high-symmetry nets can be identified without the use of additional computations. However, some caution is required, as some nets contain the same building blocks. For example, in Figure 12, we compare the dia net, the most common of the four-connected nets based on the tetrahedron, and the lon net (the lonsdaleite net) also based on tetrahedrons. These two nets may look quite similar if viewed from certain directions, or if very distorted, because both contain only six-rings.

5.4

Topology designators and network analysis software

When the nets get complicated or when there is the slightest doubt about the assignment, we use the formal identification of the network topology. To this end, we use two kinds of topology designators: ring sizes (see Section 5.4.2) and the number of nearest neighbors (coordination sequence, Section 5.4.1). The most convenient way to find these is to use specialized software, and the following alternatives are available currently (2010): OLEX,42 TOPOS,30, 43 and SYSTRE.29 These programs do not perform exactly the same tasks and the output varies. We refer the reader to the respective manuals and help files. In our experience, SYSTRE gives the most rapid identification of a known network but with no graphics output. OLEX and SYSTRE come with complete graphics interfaces. SYSTRE also calculates the most symmetric, ideal form of an unknown net.

5.4.1 Coordination sequences This is completely straightforward: Starting from any vertex in the network, you can calculate the number of nearest linked neighbors. For the four-connected dia net, this number will obviously be four. Repeating this operation with these four new vertices gives another 12 new neighbors, and the same applies for the lon net. However, for the third generation of linked neighbors, the dia net yields 24 new vertices, whereas the lon net has 25 new vertices. This is thus a good way of differentiating numerically between

Figure 12 Comparison of the dia net, the most common of the four-connected nets based on the tetrahedron, and the rather rare lon net (the lonsdaleite net) also based on tetrahedrons. Table 2

Comparing the four-connected dia, lon, and qtz nets. Coordination sequence

Ring size designators

net

cs1

cs2

cs3

cs4

cs5

cs6

cs7

cs8

cs9

cs10

cum10

Point symbol

dia lon qtz

4 4 4

12 12 12

24 25 30

42 44 52

64 67 80

92 96 116

124 130 156

162 170 204

204 214 258

252 264 318

981 1 027 1 231

66 66 64 82

Vertex symbol 62 ·62 ·62 ·62 ·62 ·62 62 ·62 ·62 ·62 ·62 ·62 6·6·62 ·62 ·87 ·87

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis

11

p=6

p=3

p=4 n=4

n=6

n=3

Figure 13 The three most common 2D nets can be assigned as follows: (6,3), (4,4), and (3,6) nets with the use of the n,p nomenclature (p is the connectivity at each vertex and n designates the shortest circuit found in the net).

these two nets. Normally, the coordination sequence up to the 10th generation is counted, and the accumulated value at this point is given as the C10, TD10, or cum10 value. A comparison between dia and lon can be found in Table 2.

5.4.2 Shortest rings and circuits A simple way to create a nomenclature for networks uses two numbers, the connectivity at each vertex (p), and the number of vertices (n) in the shortest cycle through the net that get you back to the starting point (Figure 13). While this is a good starting point, it is by no means an adequate method to name individual nets. It is better suited to classify nets in categories such as the (10,3)-nets (see Figure 6). Two more advanced designators, automatically calculated by OLEX and TOPOS, are now in common use, the point symbol and the vertex symbol. The point symbol This designator is a set of numbers with superscripts that lists the shortest cycle that joins each pair of links emerging from a vertex. In Table 2, we can see that both the dia and the lon nets contain only hexagons, while the quartz, or qtz net, for two of the six possible combination of links protruding from a qtz vertex, the shortest circuit is an octagon. The sizes of these cycles are written in normal type and the number of times each cycle size occurs is indicated by a superscript[3] giving the point symbol 66 (also “Schl¨afli” or “short symbol” but these terms are not recommended,44 see Section 5.5). This has the additional advantage that we can calculate the connectivity by summing these superscripts, making the p connectivity designator redundant. For easy reference, we give the correspondence between these sums and the connectivity in Table 3.

It has been suggested that, for all nets except the most common one, the point symbol should precede the network topology symbol when the net is first mentioned in a text, such as 64 82 -qtz.4d The vertex symbol This set of numbers lists how many of these shortest rings[4] join each pair of links emerging from a vertex and is called the vertex symbol (also “extended” or “long Schl¨afli symbol” but these terms are not recommended,44 see Section 5.5). Usually this is enough to distinguish two nets, but note in Table 2 that both the dia and the lon nets have the same vertex symbol and that they can only be distinguished by the coordination sequence. (This has certain logic if you know how these nets are derived: the dia net can be constructed from cubic close packing filling every second tetrahedral hole with new vertices and the lon net can be similarly built from the hexagonal close packing.) The six-rings making up the vertex symbol dia net as can be seen in Figure 14. The difference between cycles and rings Finally, we need to point out a subtle difference between cycles and rings. Some vertex symbols, as that for the bnn net (for boron nitride) contain stars “∗ ” or “∞” instead of numbers. The bnn net is a common five-connected net with trigonal bipyramidal vertices as shown in Figure 15. The point symbol is 46 64 , and the vertex symbol is 4·4·4·4·4· 4·6·6·6·∗ . The three different types of shortest cycles in this structure are color coded in Figure 15, showing one type of 4-gon and two types of 6-gons, and it is relatively easy to work out the point symbol by inspection of this figure. However, in defining the rings of the long symbols, a pair of links that are joined by cycles that are short circuited back to the vertex by a third link are not counted and instead

Table 3 The correspondence between the sums of superscripts in the point symbols and the connectivity. Sum of superscripts Connectivity

3 3

6 4

10 5

15 6

21 7

28 8

36 9

45 10

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

12

Supramolecular materials chemistry

C D

C B

D

A

D

B

D

B A

C B

D

B A

A

A

C

C

Figure 14

D

C

C

B A

C B

A

A

D

B A

C B

D

D

A

C D

C B

D

C B

A

D

B A

The six-rings making up the vertex symbol 62 ·62 ·62 ·62 ·62 ·62 of the dia net.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis

13

Figure 15 The bnn net has a short symbol 46 64 and a long symbol 4·4·4·4·4·4·6·6·6·∗ . The three different types of cycles in this structure are color coded, showing one type of 4-gon and two types of hexagons. Note that the black cycle is short circuited by the green link and thus is not a “ring” in the mathematical sense, giving the “∗ ” of the vertex symbol.

noted as “∗ ”or “∞” in the vertex symbol (as these are not “rings” in the mathematical sense). This is illustrated in Figure 15 by the black path short circuited by the green link.

H O

O

O

H

O H

N

O H

H

H

O N O

C(8)

5.5

Recommended usage

A more detailed and stringent explanation of the mathematical background and of the terms and nomenclature can be found in a number of papers by O’Keeffe, Delgado-Friedrich, and coworkers.22, 23, 45 As a final note on networks, we draw the reader’s attention to the recommendations on the naming of these different descriptors emerging from the combined efforts of three different research groups (Table 4).44

6

GRAPH SET ANALYSIS: A BRIEF GUIDE

As alluded to earlier, this notation is convenient when describing complex arrays of hydrogen-bonded entities. It is also a useful tool for identifying the supramoleular synthons that hold a hydrogen-bonded network together. As with network analysis, the assignments can be carried out more or less automatically with the RPLUTO or MERCURY software available free of charge from the CCDC.46 Here we give a very brief description, and we refer the interested reader to the original work by Etter and Bernstein.3, 9

Figure 16

6.1

S(6)

O

D

Some examples of graph set symbols.

Graph set notation

The notation in this system uses a designator and one or two indices. The designators are chains (C), rings (R), intramolecular hydrogen-bonded patterns (S), and other finite patterns (D). This is followed by a subscript (d) and a superscript (a) designating the number of hydrogen-bonding donor atoms and accepting atoms. Note that these numbers are often, as in this chapter, instead written consecutively separated by a comma. In addition, the degree of the pattern (n) is added in parenthesis. This is the total number of atoms in the pattern. The sub- or superscript is omitted if there is only one donor or acceptor. Some “R” examples have been given in Figures 3 and 6 and in Figure 16, the use of the other designators is shown.

6.2

Graph set notation in network analysis

For structures with more complex hydrogen-bond patterns, a hierarchy of graph set symbols is generated, and it is beyond the scope of this chapter to go into details. As an example, we take the bicyclo[3.3.1]nonane diols 7 and 8

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

14

Supramolecular materials chemistry

Table 4 Recommended usage of names for topological descriptors.44 (Reproduced with permission from Ref. 44.  Royal Society of Chemistry, 2010.) Recommended term

Other terms used (not recommended)

Applied to

2D Schl¨afli symbol Vertex symbol (VS)

Vertex type

Delaney symbol or D-symbol

Extended Schl¨afli symbol

Regular tilings Polyhedra and 2-periodic nets All tilings

Comments

Examples from RCSR

Uses shortest rings

sql: {4,4}, hcb: {6,3} rco: 3,43 , sql: 44 , hcb: 63 , cem: 33 ,42

3D Schl¨afli symbol Vertex symbol (VS) Point symbol (PS)

Extended Schl¨afli symbol long Schl¨afli symbol Schl¨afli symbol Circuit symbol Vertex symbol

Face symbol Delaney symbol of D-symbol

Extended Schl¨afli symbol

Regular tilings 3-Periodic nets

Uses shortest rings

3-periodic nets

Uses shortest cycles

pcu: {4,3,4} pts: (4.4.82.82.88.88 )(4.4.87.87.87.87 ) pcu: 4.4.4.4.4.4.4.4.4.4.4.4.*.*.* pts: 42 .84 pcu: 412 .63 rco: [38 .418 ], pts: [42 .82 ]+[84 ], pcu: [46 ]

Polyhedra and cages (tiles) All tilings

already discussed (Section 3.2) and analyze the graph sets with MERCURY. Apart form the symbol C(1,1)8 describing two diols hydrogen bonded to each other and related C symbols, rings of various sizes are also found. Recall that the nets and point symbols were (82 .12)-utg for 7 and (83 )-eta for 8 and compare these to the R6,6(24) and R8,8(40) symbols found for both 7 and 8. The number of atoms in the graph set rings must be equal to the number of hydrogen-bond donors and acceptors (6+6 and 8+8,

respectively) plus the path through the bicyclononane (6 atoms). In this way, we can calculate that the octagons for both 7 and 8 contains (24 − 6 − 6)/6 = 2 links through the bicyclononanes and the 12-gons (40 − 8 − 8)/6 = 4 (disregarding for a moment that MERCURY will display the hydrogen bonds corresponding to any of the graph set symbols it has calculated, which will be quite obvious from the display on the screen). This illustrates the similarities of these compounds well, but is less useful in distinguishing them as the 12-gons also

100 R 2, 2( 7)

90 80 70

R

60

2,

R

2(

2,

)

8)

12

2( R

50

2( 10 )

)

R

0 R ,2(2 2

2,

40 2, 1( 6)

30

R

20

) 14 2( 2,

10 0 R 2,

2,

1,

2(

2( 9)

6) 2( 2, 4) R 2( 2, R 5) 1( 2, ) R 1(4 6)

R

R

Figure 17 The probability of formation of a large number of hydrogen-bond motifs,47 grouped and color coded by the different graph set symbols motifs of these supramolecular synthons. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Network and graph set analysis appear in 8 by taking a longer route through the network. However, had one of these nets instead been a (103 )-srs net, we would immediately have seen the difference on the graph set symbols as there are no R symbols corresponding to 10-gons in either 7 or 8, and as there is a minimum of 10 vertices in the shortest rings in the srs net, the R6,6(24) symbol would not have been possible.

6.3

Reoccurring graph set symbols and SSs

Graph set symbols are also useful for seeing how the same patterns reoccur with different supramolecular bonding motifs. There is, however, no clear correlation telling us that a particular graph set symbol generates an especially stable and robust SS. In Figure 17, we have used the data from the CSD (see The Cambridge Structural Database System and Its Applications in Supramolecular Chemistry and Materials Design, Supramolecular Materials Chemistry) analysis by Allen and coworkers for the probability of formation (frequency relative to the number of possible motifs that could have formed) of a large number of hydrogen-bond motifs,47 and have plotted these in a histogram, grouping and color coding the different motifs according to their graph set symbols.

7

CONCLUSIONS

I hope to have shown how network analysis is an essential tool, not only for the understanding of individual crystal structures but also for synthesis planning. On a more general level, it forms the basis for systematization and generalization of individual observations, the very heart of the scientific endeavor, in this area of chemistry. What the network topology does for the overall structure and the graph set analysis does for individual hydrogen bond synthons and the way this is generalizable into increasing levels of complexity makes it invaluable for the comparison of such extended supramolecular structures.

NOTES [1] Vertex is used throughout this text to comply with existing mathematical notation. [2] Giraffa camelopardalis reticulata The Latin descriptor “reticulata” seems to be fairly common in biology. [3] In the TOPOS text output this is written as {6∧6}. [4] Here we need to note that in mathematics there is a distinction between shortest rings and shortest cycles. A ring is a cycle that cannot be described as the sum of two smaller cycles. For details see reference 44.

15

REFERENCES 1. . . . ask not what your country can do for you, ask what you can do for your country. John F. Kennedy inaugural address 1961, USA government. 2010. http://www.whitehouse.gov/about/presidents/johnfkennedy (accessed 19 June 2011). 2. A. F. Wells, Acta Cryst., 1954, 7, 535–544. 3. M. C. Etter, J. C. Macdonald, and J. Bernstein, Acta Cryst. B, 1990, 46, 256–262. 4. (a) A. F. Wells, Three-dimensional Nets and Polyhedra, John Wiley & Sons, Inc., New York, 1977; (b) M. O’Keeffe and B. G. Hyde, Crystal Structures I: Patterns and Symmetry, Mineralogical Society of America, Washington, 1996; (c) N. W. Ockwig, O. Delgado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, Acc. Chem. Res. 2005, 38, 176–182; ¨ (d) L. Ohrstr¨ om and K. Larsson, Molecule-Based Materials: The Structural Network Approach, Elsevier, Amsterdam, 2005, p. 324; (e) S. R. Batten, D. R. Turner, and M. S. Neville, Coordination Polymers: Design, Analysis and Application, RSC, Cambridge, 2009; (f) O. M. Yaghi, M. O’Keeffe, N. W. Ockwig, et al., Nature 2003, 423, 705–714. 5. Encyclopædia Britannica Online, Encyclopædia Britannica Inc., Chicago, IL, 2005. http://www.britannica.com/ (accessed 19 June 2011). 6. G. R. Desiraju, Angew. Chem. Int. Ed. Engl., 1995, 34, 2311–2327. 7. M. Eddaoudi, D. B. Moler, H. Li, et al., Acc. Chem. Res., 2001, 34, 319–330. 8. (a) M. O’Keeffe, M. A. Peskov, S. Ramsden, and O. M. Yaghi, Acc. Chem. Res. 2008, 41, 1782–1789; (b) M. O’Keeffe, O. M. Yaghi, and S. Ramsden, Reticular Chemistry Structure Resource, Australian National University Supercomputer Facility, 2009, http://rcsr.anu.edu.au/ (accessed 19 June 2011). 9. J. Bernstein, R. E. Davis, L. Shimoni, and N. L. Chang, Angew. Chem. Int. Ed. Engl., 1995, 34, 1555–1573. 10. J. S. Miller, Mrs Bulletin, 2000, 25, 60–64. 11. (a) J. S. Miller and A. J. Epstein, Coord. Chem. Rev. 2000, 206, 651–660; (b) K. I. Pokhodnya, A. J. Epstein, and J. S. Miller, Adv. Mater. 2000, 12, 410. ¨ 12. Q. Y. Zhu, A. M. Arif, L. Ohrstr¨ om, and J. S. Miller, J. Mol. Struct., 2008, 890, 41–47. 13. C. E. Housecroft and E. C. Constable, Chemistry: An Introduction to Organic, Inorganic, and Physical Chemistry, Pearson Education, 2006. 14. M. O’Keeffe, O. M. Yaghi, and S. Ramsden, Reticular Chemistry Structure Resource, Australian National University Supercomputer Facility, 2009. http://rcsr.anu.edu.au/ (accessed 19 June 2011). 15. O. Delgado-Friedrichs, M. D. Foster, M. O’Keeffe, et al., J. Solid State Chem., 2005, 178, 2533–2554. 16. K. Nakasuji, M. Tadokoro, J. Toyoda, et al., Mol. Cryst. Liq. Cryst. Sci. Technol., Sect. A, 1996, 285, 241–248. 17. N. W. Ockwig, O. Delagado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, J.Am.Chem.Soc., 2005, 38, 176–182.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

16

Supramolecular materials chemistry

¨ 18. L. Ohrstr¨ om and K. Larsson, Dalton Trans., 2004 347–353.

34. A. F. Wells, Acta Cryst., 1954, 7, 842–848.

19. (a) L. F. Yang, M. L. Cao, Y. Cui, et al., Cryst. Growth Des. 2010, 10, 1263–1268; (b) M. Tadokoro, S. Fukui, T. Kitajima, et al., Chem.Commun. 2006, 1274.

35. (a) C.-J. Wallentin, E. Orentas, M. T. Johnson, et al., CrystEngComm 2009, 11, 1837–1841; (b) V. T. Nguyen, I. Y. H. Chan, R. Bishop, et al., New J. Chem. 2009, 33, 1736–1741.

20. Y. Cui, M.-L. Cao, L.-F. Yang, et al., CrystEngComm, 2008, 10, 1288–1290. 21. (a) I. A. Baburin, V. A. Blatov, L. Carlucci, et al., Cryst. Growth Des. 2008, 8, 519–539; (b) V. A. Blatov, L. Carlucci, G. Giani, and D. M. Proserpio, CrystEngComm 2004, 6, 377–395; (c) S. R. Batten, CrystEngComm 2001, 3, 67.

36. (a) R. Robson, Dalton Trans. 2000 3735–3744; (b) R. Robson, Dalton Trans. 2008 5113–5131. ¨ 37. L. Ohrstr¨ om and K. Larsson, Molecule-Based Materials: The Structural Network Approach, Elsevier, Amsterdam, 2005. 38. O. Delgado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, Acta Cryst. Sec. A, 2003, 59, 22–27.

22. O. Delgado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, Solid State Sciences, 2003, 5, 73–78.

39. A. C. Sudik, A. P. Cote, and O. M. Yaghi, Inorg. Chem., 2005, 44, 2998–3000.

23. C. Bonneau, O. Delgado-Friedrichs, M. O’Keeffe, O. M. Yaghi, Acta Cryst. A, 2004, 60, 517–520.

and

40. H. M. El-Kaderi, J. R. Hunt, J. L. Mendoza-Cortes, et al., Science, 2007, 316, 268–272.

24. (a) A. Thrierrs and J. C. Mutin, Bulletin De La Societe Francaise Mineralogie Et De Cristallographie 1968, 91, 210–221; (b) O. M. Yaghi, M. O’Keeffe, N. W. Ockwig, et al., Nature (London, United Kingdom) 2003, 423, 705–714.

41. S. J. Ramsden, V. Robins, S. T. Hyde, and S. Hungerford, EPINET: Euclidean Patterns in Non-Euclidean Tilings, The Australian National University, 2005–2009. http://epinet.anu.edu.au/ (accessed 19 June 2011).

25. M. Eddaoudi, J. Kim, N. Rosi, et al., Science, 2002, 295, 469–472. 26. C. Borel, K. Davies, P. Handa, et al., Cryst. Growth Des., 2010, 10, 1971–1978.

42. (a) O. V. Dolomanov, OLEX, http://www.ccp14.ac.uk/ccp/ web-mirrors/lcells/index.htm, 2004 (accessed 19 June 2011); (b) O. V. Dolomanov, A. J. Blake, N. R. Champness, and M. Schroder, J. Appl. Crystallogr. 2003, 36, 1283–1284 (accessed 19 June 2011).

27. J. H. He, Y. T. Zhang, Q. H. Pan, et al., Micropor. Mesopor. Mater., 2006, 90, 145–152. ¨ 28. K. Davies, S. A. Bourne, L. Ohrstr¨ om, and C. L. Oliver, Dalton Trans., 2010 2869–2874.

43. V. A. Blatov, A. P. Shevchenko, and V. N. Serezhkin, J. Appl. Cryst., 2000, 33, 1193.

29. O. Delgado Friedrichs, SYSTRE 1.1.4beta. 2007. http://gavrog.sourceforge.net/ (accessed 19 June 2011).

45. (a) V. A. Blatov, O. Delgado-Friedrichs, M. O’Keeffe, and D. M. Proserpio, Acta Cryst. A. 2007, 63, 418–425; (b) O. Delgado-Friedrichs and M. O’Keeffe, J. Solid State Chem. 2005, 178, 2480–2485; (c) O. Delgado-Friedrichs and M. O’Keeffe, Acta Cryst. A. 2005, 61, 358–362; (d) O. Delgado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, Acta Cryst. A. 2003, 59, 22–27; (e) O. Delgado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, Acta Cryst. A. 2003, 59, 515–525; (f) O. Delgado-Friedrichs, M. O’Keeffe, and O. M. Yaghi, Acta Cryst. A. 2006, 62, 350–355.

30. V. A. Blatov, TOPOS 4.0, Ac.Pavlov St. 1, 443011 Samara, Russia. 2009. http://www.topos.ssu.samara.ru/ (accessed 19 June 2011). 31. (a) A. T. Ung, R. Bishop, D. C. Craig, et al., Chem. Mater. 1994, 6, 1269–1281; (b) A. T. Ung, D. Gizachew, R. Bishop, et al., J. Am. Chem. Soc. 1995, 117, 8745–8756; (c) S. Stoncius, E. Butkus, A. Zilinskas, et al., J. Org. Chem. 2004, 69, 5196–5203; (d) S. Stoncius, E. Orentas, E. Butkus, et al., J. Am. Chem. Soc. 2006, 128, 8272–8285; (e) V. T. Nguyen, R. Bishop, D. C. Craig, and M. L. Scudder, Supramol. Chem. 2001, 13, 103–107. 32. R. Bishop, Acc. Chem. Res., 2009, 42, 67–78.

44. V. A. Blatov, M. O’Keeffe, and D. M. Proserpio, CrystEngComm, 2010, 12, 44–48.

46. RPLUTO, The Cambridge Crystallographic Data Centre, Cambridge, http://www.ccdc.cam.ac.uk/free services/rpluto/ (accessed 19 June 2011). 47. F. H. Allen, W. D. S. Motherwell, P. R. Raithby, et al., New J. Chem., 1999, 23, 25–34.

33. J. M. Robertson, Proc. Roy. Soc (A), 1936, 157, 79.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc120

Coordination Polymers Edwin C. Constable University of Basel, Basel, Switzerland

1 Introduction 2 Principles of Assembly and Assembly Algorithms 3 Illustration of the Assembly of 1D and 2D Coordination Polymers 4 Applications of MOFs 5 Conclusions References

1

1 1 5 11 13 13

INTRODUCTION

The design, synthesis, study, and application of coordination polymers and networks are an area of chemistry that is currently drawing ever-increasing attention. Figure 1 shows the number of publications per year related to this topic and clearly demonstrates that a dramatic and continuing increase in activity has occurred since 1990. A recent book provides an excellent and more detailed overview of this fascinating area of chemistry.1 For the purpose of this chapter, coordination polymers and networks are defined as extended species composed of metal centers linked by bridging organic ligands; compounds with “inorganic” bridges such as cyanide, halide, or phosphate are specifically excluded from this review. The term metal–organic framework (MOF ) is also used sometimes to describe these coordination polymers and networks, although the original usage referred to threedimensional structures with carboxylate bridging ligands.2–4

The bridging ligands are also described by some authors as tectons.5 Coordination polymers have a long history, dating back at least to the discovery of Prussian blue (Berlin blue, Parisian blue, Turnbull’s blue) by Diesbach at the very beginning of the eighteenth century.6 However, the modern period, characterized by the design of ligands which will lead to a predetermined extended structure, can be dated to seminal publications by Richard Robson in the late 1980s and early 1990s.1, 7–9 A recent overview indicates the vision present in this early work by Robson.10 One of the contributing factors to the explosive increase in the study of these systems was the ability to perform routine and rapid single crystal structural determinations to fully characterize solid state materials.

2

PRINCIPLES OF ASSEMBLY AND ASSEMBLY ALGORITHMS

The principle of assembly of a coordination polymer is stunningly simple. A metal-containing fragment (mononuclear or multinuclear) is viewed as a node that is connected by bridging ligands. The node has a number of available coordination sites, c. Each ligand comprises two or more spatially separated n-dentate metal-binding domains, which are simply conventional mono- or multidentate donor sets. The number of available coordination sites c is simply related to the denticity of the metal-binding domains. The possible combinations of various domains in the commonly encountered values of c = 2, 3, 4, or 6 and n = 1, 2, or 3 are summarized in Table 1. One feature to note is that the number of available sites c is not necessarily the same as the total coordination number. Some sites may be blocked by preexisting, adventitious,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

2

Supramolecular materials chemistry 1400

1200

1000

800

600

400

200

1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

0

Figure 1 Number of publications per year containing the terms “coordination polymer,” “MOF,” or “metal–organic framework.” (Data collected from Web of Science, January 2011.) Table 1 Formal matching of metal-binding domains of various denticities with metal-containing nodes possessing c available coordination sites. Entry 2 corresponds to a discrete rather than polymeric system. Entry no.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Available sites c

2 — 3 — — 4 — — — 6 — — — — — —

Number of binding domains of denticity n n=1

n=2

n=3

2 — 3 1 — 4 2 1 0 6 4 3 2 1 — —

— 1 — 1

— — — — 1 — — 1 — — — 1 — 1 — 2

— 1 — 2 1 — 2 1 3 —

or nonlabile ligands. A typical example can be seen in the use of the {Ru(bpy)2 }2+ node (bpy = 2,2
-bipyridine); the kinetically inert d 6 center will bind an additional one

bidentate (Entry 2, c = 2, n = 2, discrete system) or two monodentate metal binding domains (Entry 1, c = 2, 2 × n = 1) to give a total coordination number of 6. There is, of course redundancy in Table 1, as the specific cases with {Ru(bpy)2 }2+ illustrated above correspond to entries 1 and 2 in the table, but also to entries 13 and 15 in terms of the total coordination number. A second point to note is that the table is essentially a topological analysis and the number of possible combinations is increased when the topographical complexity of stereoisomers is taken into consideration. For example, entry 13 is degenerate as it can refer to cis- or trans-stereoisomers of an octahedral center with the added complication of
and  enantiomers of the cis-form. Similarly, the geometrical properties of the metalcontaining nodes play a critical role in dictating the eventual coordination polymer. Thus the use of c = 4 nodes with a tetrahedral or square-planar geometry commonly results in the formation of 2D and 3D networks, respectively.

2.1

Metal-binding domains

A selection of typical metal-binding domains of various denticities is presented in Figure 2. These may be combined as desired to give bridging ligands with desired properties. Combinations of different metal-binding domains may be used to develop strategies for the selective synthesis

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Coordination polymers

3

Monodentate N

N

N

N

N

O

N

N

N

N

− O

− + O N

O−

CN

N

N

N N

− + O P R R

PR2

S−

Bidentate O

S

O−

S −

N

N

N

N

Terdentate

N N

Figure 2

N

Metal-binding domains commonly encountered in coordination polymers. The donor atoms are indicated in bold.

N

N

PATH A

n

N

N PATH B

N

N

N

N PATH C

N

N

N

N

N

N

N

PATH E N

N

N N

N NH

N N

N N

N

HN

N N

Figure 3 domains.

PATH D

N

Strategies for the development of multinucleating ligands for use in coordination polymers based on pyridine metal-binding

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

4

Supramolecular materials chemistry

of heterometallic polymers based on simple principles of coordination chemistry (hard and soft acid base principle, denticity, etc.). Once again, the attempt to classify the metal-binding domains leads to some degeneracy. Thus, the diazines and triazines can be viewed as a product of the “condensation” of two monodentate pyridines to give prototype bridging ligands. The selected metal-binding domains may be linked by flexible or rigid spacers leading to control over the properties and dimensionality of the eventual coordination polymer. This is illustrated in Figure 3 in which pyridine metal-binding domains are combined to give new families of ligands with varying topological and topographical properties. Path A represents the elaboration of linear NN
bridging ligands which differ in the distance between the nitrogen atoms, thus allowing control of the distance between metal atoms in coordination polymers. Path B shows how, by linking the pyridine units to give the series of isomeric oligopyridines, not only the distance between the nitrogen donors but also the vectorial relationship of the nitrogen lone pairs can be controlled (from 180◦ in the 4,4
bipyridine to 120◦ in the 3,4
-bipyridine illustrated in Figure 3). The vectorial dependence may also be tuned in the higher oligomers in the series developed in Path A by using the various substitution patterns of the bridging aryl rings. Path C represents a parallel development to Path A but with increasing flexibility in the bridge. Path D shows the connection of 4-pyridyl metal-binding domains to give new ligands which will bridge three or four nodes,

respectively. This pathway also illustrates the control over the spatial orientation of the four nitrogen donors into square-planar or tetrahedral arrangements. The final path in Figure 3 shows how the use of polyazines can lead to comparable bridging ligands to Path D. Similar pathways exist for the other metal-binding domains, for example, the carboxylate domain leads to the commonly utilized oxalate and higher polycarboxylate ligands. By combining the pyridine and carboxylate metal-binding domains, bridging species such as isonicotinate, are obtained with different metal selectivities at the N and OO
donor ends.

2.2

Metal-containing nodes

In addition to the mononuclear nodes commonly known from coordination chemistry, multinuclear nodes are often encountered in coordination polymers (Figure 4). In general, these are self-assembled under the reaction conditions rather than introduced intact; although this is not necessarily the case with polyoxometallate nodes. The dimetallic node is of some interest as it can enter into the network through coordination of additional donors at the two axial sites, or through the substituents on the carboxylate ligands. In other words, this unit can act as either a two node or a four node.

2.3

Expanded ligands

Much of the drive in the design and synthesis of coordination polymers is centered on the incorporation of

Mononuclear

Multinuclear

O O

O

O

O O

O

O OO O

O O

O

O O

O

O O

O O

Figure 4

Mononuclear and multinuclear nodes commonly encountered in coordination polymers.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Coordination polymers designed functionality. Recently the concept of using metalcontaining spacers between the metal-binding domains was introduced and this was described as the expanded ligands approach.11 This can be seen as an evolution of the approach indicated in Figure 3: the 4
-(4-pyridyl)2,2
:6
,2

-terpyridine (pytpy) complex 1 is a structural analog of 4,4
-bipyridine in which the N· · ·N distance of the ˚ and in metal-binding domain has increased from 7.1 to 18 A which the photophysical and redox properties of the central metal become an integral part of the coordination polymer framework.

N

N

M N

N

N

Figure 5 Illustration of the assembly of 1D and 2D networks from metal centers with two, three, and four vacant coordination domains and ligands with two commensurate metal-binding domains. Note that the example with c = 4 has only one possible outcome: 3D networks are also possible.

1

3

3.1

n

N

N N

5

ILLUSTRATION OF THE ASSEMBLY OF 1D AND 2D COORDINATION POLYMERS Simple examples

Although the formalism of Table 1 is deceptively simple, the reality is rather more complex. Using these simple building blocks, we can now develop topological representations of the structures resulting from the combination of nodes and linkers. Figure 5 illustrates schematically how a node with two available coordination sites interacts with a bridging species with two commensurate metal-binding domains. This could correspond to entry 1 (c = 2, n = 1), entry 9 (c = 4, n = 2) or entry 16 (c = 6, n = 3) in Table 1. In these cases the result is a 1D coordination polymer.

3.1.1 1D coordination polymers An example of entry 1 is seen in the 1D coordination polymer formed between 4,4
-bipyridine (4,4
-bpy) and silver(I) presented in Figure 6. The total coordination

Figure 6 The 1D structure with 4,4
-bipyridines H2 NC6 H4 CO2 H)(NO3 )}n (RefCode: AQAVOD).12

bridging

number of the silver(I) centers is 3, with the remaining coordination site occupied by water which plays no role in the topological description.12 The coordination polymer is formed by self-assembly of the components and serves to illustrate that even if the basic algorithm is fulfilled it is very difficult to determine the number of ancillary (or adventitious) ligands that might be involved. This flexibility in coordination mode is one of the reasons why so many coordination polymers involve silver(I) or other d 10 metal centers.13 In Figure 7, the same assembly principle is illustrated for entry 9 with bidentate carboxylate groups forming terephthalate as the metal-binding domains.14 In this case, the metal node is a copper(II) center in which two of the six available coordination sites are occupied by a di(2-pyridyl)amine ligand. This structure also serves to illustrate that the topologically 1D system exhibits a zigzag topography. Schubert and coworkers have reported extensively on solution phase metallopolymers assembled from the interaction of flexible ligands of general structure 2 with transition metal salts.15–17 These novel materials exhibit useful electronic and photophysical properties, but their

silver(I)

centers

present

in

{[Ag2 (4,4
-bpy)2 (H2 O)2 ](4-

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

6

Supramolecular materials chemistry

A 1D structure with bidentate carboxylate metal-binding domains bridging copper(II) centers (RefCode: AFUQIB).14

Figure 7

formulation has been the subject of some dispute. Currently, it is believed that the metallopolymers are kinetic products of the coordination reaction but that over time equilibration occurs to give a wide range of oligomers and metallomacrocyclic species.18–20 These equilibration reactions are sensitive to solvent, temperature and the nature of any anions present in the reaction mixture and a series of metallomacrocyclic products with cobalt(II), cobalt(III), and iron(II) centers have been structurally characterized.18, 21–23 Analogous 1D metallopolymers with rigid bis(2,2
:6
,2

-terpyridine) ligands do not suffer from the macrocyclization problem but in solution, are relatively polydispersed.24, 25

N

N N N

O

O

N

O

n 2

O

N

N

N 3

The expanded ligand approach works well for 1Dcoordination polymers and by using 1 (M = Ru or Fe) as an expanded 4,4
-bipyridine, examples with {Cu(NO3 )2 (H2 O)} (M = Fe; Figure 8a)26 and {Ag(NO3 )(CH3 CN)} (M = Ru; Figure 8b)27 nodes have been prepared. The exquisite control possible in these reactions allows the preparation of fully ordered systems in which the alternation of metal centers is assured and the electronic and magnetic properties of these new materials are of great potential interest. The direct reaction of pytpy with copper(II) salts also leads to a 1D polymer, but in this case each copper(II) center is coordinated to an NN
N

domain of one ligand and an N domain of a second ligand (Figure 8c).28

Figure 9 presents the commonly identified patterns which may be identified in topologically 1D systems. The expansion of the 1D definition to various more complex chain structures indicates the structural complexity that can be achieved using coordination polymers.

3.1.2 Helical 1D coordination polymers When more flexible bridging ligands are used for the assembly of 1D coordination polymers, a possible consequence is the formation of helical systems arising from the twisting of the ligand–metal framework about the direction of propagation. If achiral ligands are used, the result could be a racemic crystal with equal amounts of P and M (left- and right-handed) helices or partial/complete resolution into homochiral crystals. Much of the interest in this area has centered on the design of ligands such that homochiral packing is favored and enantiopure single crystals are obtained. An example is seen with ligand 3 which reacts with HgBr2 to give homochiral crystals containing the chiral 1D chain is presented in Figure 10.29 The chirality arises from the orientation of the pyridyl groups with respect to the 2,5-diphenylcyclopenta2,4-dienone scaffold. By using chiral ligands, good stereo control is often achieved and homochiral crystals may be obtained.30

3.1.3 2D coordination polymers The basic assembly algorithm is presented in Figure 5. However, the precise coordination geometry at the node plays a critical role in controlling the true nature of the products. The simplest approach to the square grid in Figure 5 is to use four-coordinate metal centers as the nodes or ligands with four metal-binding domains as nodes. A good example is seen in the 2D sheet obtained from the reaction of PbI2 with the tetra(4-pyridyl)porphyrin as shown

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Coordination polymers

7

(a)

(b)

(c)

Figure 8 The 1D chains present in (a) {Fe(pytpy)Cu(NO3 )2 (H2 O)}n 2n+ (GIVDEV),26 (b) {Fe(pytpy)Cu(NO3 )2 (H2 O)}n 2n+ (WICSIL),27 and (c) {Cu(pytpy)Cl2 }n (OGOSIN).28

Figure 9

Figure 10

1D coordination polymers of increasing complexity.

The 1D helical chain obtained from the reaction of 3 with HgBr2 . The crystals consist of homochiral chains (BAXPOG).29

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

8

Supramolecular materials chemistry

(a)

(b)

Figure 11 (a) The square grid from the reaction of tetra(4-pyridyl)porphyrin with PbI2 . The iodide ligands lie above and below the plane of the sheet (BAKKIH)31 and (b) a similar topology with linear connectors (YECFAN).32

Figure 12

An example of a 2D pattern using the expanded ligand 4 (BODGOR) with grid-lines added to show the network.33

in Figure 11(a). Each lead in the center is coordinated to a pyridyl nitrogen from each of four ligands distributed in a square-planar geometry and with two axial iodide ligands.31 The combination of four-coordinate nodes (c = 4) with ligands bearing four metal monodentate binding domains results in an overall 1 : 1 stoichiometry. A second example is seen using a linear 4,4
-bipyridine to connect Cd(NO3 )2 units in the example presented in Figure 11(b).32 In this case, the combination of a four-coordinate node with a ligand bearing two monodentate metal-binding domains results in the stoichiometry of ML2 . Although the topology would appear to predicate the use of square-planar or octahedral metal centers, this is not actually the case if less flexible ligands are used, or if the geometries of the ligands and the metal centers are well matched. An example involving the expanded ligand approach is found using the expanded ligand 4

(M = Ru) which forms a 2D structure on reaction with CuCl2 (Figure 12). This also illustrates that the result is not always a perfectly square grid, and in this case a rhombic grid is obtained (taking the ruthenium centers as the nodes).33

N N

N

N N

N M

N N

N

N

4

Replacing nodes or ligands with a connectivity of four by three connecters, a honeycomb pattern is obtained. This

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Coordination polymers

Figure 13

An example of a honeycomb pattern with stoichiometry M2 L3 in [Cu2 (pyrazine)3 ][SiF6 ] (LETWAI).34

Figure 14

2D coordination polymers of increasing complexity.

is shown in Figure 13 for a three-coordinate copper(I) center with c = 3 and pyrazine. The use of pyrazine with two monodentate metal-binding domains results in an overall stoichiometry for the honeycomb structure of M2 L3 .34 Figure 14 presents some of the 2D structures which are encountered indicating the increasing level of complexity that can be achieved.

9

3.1.4 3D coordination polymers As in previous cases, this chapter only presents the simplest examples in order to illustrate the principles used for the assembly of 3D coordination polymers. For a more detailed discussion of the possible structural nets the reader is recommended to read Ref. 1. Two of the simplest 3D nets are the diamondoid and octahedral structurals and it is

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

10

Supramolecular materials chemistry

(a)

(b)

Figure 15 Examples of (a) diamondoid nets from a tetrahedral copper(I) center and 4,4
-bipyridine (XACHAO)35 and (b) a cubic (PtS) lattice from octahedral copper(II) nodes and 4,4
-bipyridine ligands (GORWUF).36 In the latter case, the lattice is completed by bridging SiF6 anions linking together the 2D array of {Cu(4,4
-bpy)2 } sheets.

constructive to illustrate the coordination polymer approach to these. The simplest approach to the diamondoid structure is to replace each of the tetrahedral carbon atoms in diamond by a tetrahedral metal center. The metal centers are then linked by a (linear) ligand with two two-monodentate metal-binding domains. This approach is exactly shown in Figure 15(a), which presents a copper(I) complex with 4,4
-bipyridine ligands exhibiting the expected diamondoid net (Figure 15b).35 By moving from a tetrahedral copper(I) center to an octahedral copper(II) node, a cuboid system is obtained with the same 4,4
-bipyridine ligand.36 This latter compound shows a high storage capacity for methane in the interstitial voids.

CO2− CO2−

CO2−

CO2−

CO2−

CO2−

X X CO2−

CO2−

X = H, Hal, OR, NH2, etc. CO2− CO2−

CO2−

CO2−

CO2−

CO2−

CO2−

3.1.5 The MOFs Although Kondo37 and Li38 reported the first examples of the porous materials that were later termed as MOFs. The area began its explosive growth with the isolation of materials with surface areas of several thousand square meter per gram.39 There are many different types of MOF and this chapter presents only the simplest parent structures. For good overviews of the structural diversity and the applications of MOFs, the reader is directed to specialist secondary literature.40 The archetypical MOFs are based on dicarboxylate organic linkers of the type is shown in Figure 16. The node is a Zn4 O cluster in which the edges of the Zn4 tetrahedron are bridged by six bidentate carboxylate donors. The resultant Zn4 O(O2 CR)6 cluster acts as a node with a connectivity of six, giving rise to cubic structures. A prototype MOF, known as MOF-5 prepared by Yaghi illustrates both the assembly principles and the structural features. The MOF is prepared by the reaction of zinc(II)

CO2−

Figure 16 Typical dicarboxylate linkers used to vary the voids in MOFs by changing the edge length of the lattice.

salts with terephthalic acid in basic conditions resulting in the self-assembly of the Zn4 O(O2 CR)6 node (Figure 17a). The substituents on the carboxylate ligands of the clusters are oriented in an octahedral manner and the terephthalate ligands define the edges of a cubic network (Figure 17b). The pore volume in MOF-5 approaches 91% and the large cavity is seen in Figure 17(c).41 By increasing the length of the spacer between the two carboxylate groups of the bridging ligand, it should be possible to increase the size of the cavity, although this approach has an inherent problem which is discussed in Section 3.1.5.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Coordination polymers

(a)

11

(b)

(c)

Figure 17 (a) The Zn4 O(O2 CR)6 node present in the archetype MOF (MOF-5, SAHYIK)41 ; (b) the cuboid lattice present in MOF-5; and (c) a space-filling representation showing the large pore volume.

3.1.6 Interpenetration Without doubt, the most common method used for the characterization of coordination polymers is single crystal X-ray crystallography. In the case of 1D and 2D materials, there are no special problems beyond those normally associated with crystallography (unsuitable crystals, disorder, twinning, solvent loss, etc.). However, some specific problems are encountered particularly with 3D materials. One of the targeted applications of 3D MOF coordination polymers is the storage of gas molecules such as dihydrogen in voids within the coordination network. However, the principle that “Nature abhors a vacuum”42 operates in lattices that are designed to very large pores and leads to the phenomenon of interpenetration. Indeed, additional complexity can arise even with 1D and 2D systems. For example, the reasonably flexible ligand 5

reacts with AgClO4 to form the expected 2D (6,3) network (Figure 18).43 The hexagonal void has dimension 4.7 nm in height and 3.5 nm in width, which allows space for the formation of a 3D coordination polymer formed by triply interpenetrated (6,3) layers which further entangle with four other layers, ultimately giving an unprecedented 11-fold interpenetration and 5-fold catenation. Typical examples of 3D interpenetration are presented in Figure 19 and this topic is discussed in detail in Interpenetration, Supramolecular Materials Chemistry.

4

APPLICATIONS OF MOFs

Much of the recent interest in coordination polymers is centered on the MOFs. The large voids present within the lattices have suggested that these materials are prime

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

12

Supramolecular materials chemistry 27Å

47Å

35Å

N

S

N

N N

S S

N

N 5

Figure 18 The basic (6,3) network constructed from silver(I) and ligand 5 and the triply interpenetrating structure (reproduced from Ref. 43 With permission). The fascinating world of interpenetrating structures is the subject of a chapter or a book in its own right. For an easy to digest introduction to the topic, the website and related articles from Stuart Batten are strongly recommended.44–46 solution.

Figure 19 Typical representations of 3D interpenetrated structures involving (a) the (3.5) net, (b) the α-Po net, (c) diamondoid nets, and (d) rutile nets. (Redrawn by Dr. S. Batten from Ref. 44.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Coordination polymers candidates for the storage of small energy-rich molecules in the context of developing new energy strategies. Much attention is on the use of MOFs as storage material for dihydrogen47, 48 and other gases.49 Hydrogen uptake in the 7–10 wt% range is readily achieved at low temperatures and moderate pressures.

15. M. Chiper, R. Hoogenboom, and U. S. Schubert, Macromol. Rapid Commun., 2009, 30, 565. 16. P. R. Andres and U. S. Schubert, Adv. Mater., 2004, 16, 1043. 17. J.-F. Gohy, B. G. G. Lohmeijer, and U. S. Schubert, Chem. Eur. J., 2003, 9, 3472 and refs. therein. 18. E. C. Constable, K. Harris, C. E. Housecroft, M. Neuburger, Dalton Trans., 2011, 40, 1524.

5

CONCLUSIONS

Coordination polymers of various dimensionality are readily prepared using the simple principles of metallosupramolecular chemistry. Although the structural diversity means that it is often a challenge to predict the outcome of a particular reaction, the properties of the new materials reward the synthetic effort. Particular effort is centered on the so-called MOFs which are currently the center of intense investigation for their use as storage materials for gases and as readily prepared functional materials for a wide variety of applications.

REFERENCES 1. S. R. Batten, S. M. Neville, and D. R. Turner, Coordination Polymers, RSC Publishing, Cambridge, 2009. 2. D. J. Tranchemontagne, J. L. Mendoza-Cortes, M. O’Keeffe, and O. M. Yaghi, Chem. Soc. Rev., 2009, 38, 1257.

13

and

19. H. S. Chow, E. C. Constable, R. Frantz, et al., New J. Chem., 2009, 33, 376. 20. H. S. Chow, E. C. Constable, C. E. Housecroft, et al., Polyhedron, 2006, 25, 1831. 21. C. B. Smith, E. C. Constable, C. E. Housecroft, B. Kariuki, Chem.Commun., 2002, 2068.

and

22. E. C. Constable, C. E. Housecroft, and C. B. Smith, Inorg. Chem.Commun., 2003, 6, 1011. 23. E. C. Constable, C. E. Housecroft, M. Neuburger, et al., Dalton Trans., 2005, 2259. 24. D. Hinderberger, O. Schmelz, M. Rehahn, and G. Jeschke, Angew. Chem. Int. Ed., 2004, 43, 4616. 25. B. Guilleaume, J. Blaul, M. Ballauff, et al., Eur. Phys. J. E , 2002, 8, 299. 26. J. E. Beves, E. C. Constable, C. E. Housecroft, et al., CrystEngComm, 2008, 10, 344. 27. J. E. Beves, E. C. Constable, C. E. Housecroft, et al., CrystEngComm, 2007, 9, 456. 28. J. E. Beves, E. C. Constable, S. Decurtins, et al., CrystEngComm, 2009, 11, 2406. 29. U. Siemeling, I. Scheppelmann, B. Neumann, et al., Chem. Commun., 2003, 2236.

3. S. Kitagawa, R. Kitaura, and S.-i. Noro, Angew. Chem. Int. Ed., 2004, 43, 2334.

30. L. Han and M. Hong, Inorg. Chem. Commun., 2005, 8, 406.

4. G. F´erey, Chem. Soc. Rev., 2008, 37, 191.

31. C. V. K. Sharma, G. A. Broker, J. G. Huddleston, et al., J. Am. Chem. Soc., 1999, 121, 1137.

5. M. W. Hosseini, Acc. Chem. Res., 2005, 38, 313. 6. J. E. Berger, Kerrn aller Fridrichs=St¨adtschen Begebenheiten, Manuskript, Berlin, ca.1730 (Berlin, Staatsbibliothek zu Berlin—Preußischer Kulturbesitz, Handschriftenabteilung, Ms. Boruss. quart. 124).

32. M. Fujita, Y. J. Kwon, S. Washizu, and K. Ogura, J. Am. Chem. Soc., 1994, 116, 1151. 33. J. E. Beves, E. C. Constable, S. Decurtins, et al., CrystEngComm, 2008, 10, 986.

7. S. R. Batten and R. Robson, Angew. Chem. Int. Ed. Engl., 1998, 37, 1460.

34. L. R. MacGillivray, S. Subramanian, and M. J. Zaworotko, Chem. Commun., 1994, 1325.

8. B. F. Hoskins and R. Robson, J. Am. Chem. Soc., 1989, 111, 5962.

35. H. Lin and P. A. Maggard, Inorg. Chem., 2009, 48, 8940.

9. R. W. Gable, B. F. Hoskins, and R. Robson, J. Chem. Soc., Chem. Commun., 1990, 1677. 10. R. A. Robson, J. Chem. Soc., Dalton Trans., 2000, 3735.

36. S. Noro, S. Kitagawa, M. Kondo, and K. Seki, Angew. Chem. Int. Ed., 2000, 39, 2082. 37. M. Kondo, T. Yoshitomi, K. Seki, et al., Angew. Chem. Int. Ed. Engl., 1997, 36, 1725.

11. E. C. Constable, Coord. Chem. Rev., 2008, 252, 842.

38. H. Li, M. Eddaoudi, T. L. Groy, and O. M. Yaghi, J. Am. Chem. Soc., 1998, 120, 8571.

12. R. H. Wang, M. C. Hong, J. Luo, et al., Inorg. Chim. Acta, 2004, 357, 103.

39. M. J. Zaworotko, New J. Chem., 2010, 34, 2355.

13. A. N. Khlobystov, A. J. Blake, N. R. Champness, et al., Coord. Chem. Rev., 2001, 222, 155. 14. L. Karanovic, D. Poleti, J. Rogan, et al., Acta Crystallogr. C: Cryst. Struct. Commun., 2002, 58, m275.

40. C. Janiak and J. K. Vieth, New J. Chem., 2010, 34, 2366. 41. H. Li, M. Eddaoudi, M. O’Keeffe, and O. M. Yaghi, Nature (London), 1999, 402, 276. 42. B. de Spinoza, The Ethics, 1677.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

14

Supramolecular materials chemistry

43. Q.-Y. Yang, S.-R. Zheng, R. Yang, et al., CrystEngComm, 2009, 11, 680. 44. S. Batten, http://www.chem.monash.edu.au/staff/sbatten/ interpen/index.html. 45. S. R. Batten, Encyclopedia of Supramolecular Chemistry, eds. J. L. Atwood and J. W. Steed, Marcel Dekker, New York, 2004, pp. 735–741. Interpenetration.

46. S. R. Batten, CrystEngComm, 2001, 3, 67. 47. L. J. Murray, M. Dinca, and J. R. Long, Chem. Soc. Rev., 2009, 38, 1294. 48. S. S. Han, J. L. Mendoza-Cort´es, and Chem. Soc. Rev., 2009, 38, 1460. III.

W. A. Goddard,

49. J.-R. Li, R. J. Kuppler, and H.-C. Zhou, Chem. Soc. Rev., 2009, 38, 1477.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc121

Zeolitelike Metal–Organic Frameworks (ZMOFs): Design, Structure, and Properties Mohamed H. Alkordi and Mohamed Eddaoudi King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia

1 Historical Perspective 2 Metal–Organic Frameworks (MOFs) 3 Metal–Organic Frameworks Utilizing Polytopic, Nitrogen-Donor Ligands 4 Metal–Organic Frameworks Utilizing Polytopic, Carboxylate-Based Ligands 5 Metal–Organic Frameworks Utilizing Polytopic, Heterofunctional Ligands: the Single-Metal-Ion MBB Approach 6 From Zeolites and MOFs to Zeolitelike MOFs (ZMOFs) 7 Zeolitic Metal–Organic Frameworks Based on the Edge-Expansion Approach 8 Zeolitic Metal–Organic Frameworks Based on the Directed Assembly of Supermolecular Building Blocks (SBBs) 9 Zeolitelike Metal–Organic Frameworks as Platforms for Applications 10 Conclusion References

1

1 2 3 4

6 8 10

12 15 18 18

HISTORICAL PERSPECTIVE

It was due to the pioneering studies of Alfred Werner, Nobel laureate of 1913, that chemists first realized the type, geometry, and isomerism in octahedrally coordinated

transition metal ions. In the family of compounds known as Werner clathrates, solid-state host–guest composites with an octahedral metal complex of the formula MA4 X2 , the complex serves as a clathrating agent toward appropriately sized organic guests. In a classical Werner clathrate, M is a divalent metal ion, A is a neutral amine donor ligand (usually a pyridine derivative), and X is an anionic ligand (NCS− , CN− , NCO− , Cl− , Br− , I− , and NO3 − ). This family of compounds is based on N-donor ligands (neutral pyridines) (Figure 1) While Werner complexes are based on discrete coordination complexes, another type of coordination compound with a distinctive extension into 2D layers through utilization of ditopic bidentate ligands was developed and first reported by K. A. Hofmann in 1897, Ni(NH3 )2 Ni(CN)4 2(C6 H6 ).2 Hofmann-type compounds also exhibit inclusion properties of suitably sized guest molecules. Iwamoto and coworkers prepared a number of derivatives of Hofmann clathrates (Figure 2). Those derivatives maintain the parent Hofmann-type compound topology of 2D sheets by substituting octahedrally coordinated metal ions for the octahedral Ni, or altogether substituting both Ni ions by two different metal ions, one with square planar and the other with octahedral coordination sphere. The general chemical formula of such compounds is M(NH3 )2 M
(CN)4 ·2G, where M = Mn, Fe, Co, Ni, Cu, or Zn; M
= Ni, Pd, or Pt; and the guest molecule G can be pyrrole, thiophene, benzene, or aniline.4–7 The emergence of Hofmann compounds clearly demonstrates the ability to extend discrete molecular species, Werner compounds, into 2D layers of coordination polymers simply through utilization of bridging ditopic ligands. The presence of capping monodentate ammonia molecules

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

2

Supramolecular materials chemistry

(a)

(b)

Figure 1 (a) An example of the Werner complex, Ni(NCS)2 (4-methylpyridine)4 and1 (b) crystal packing showing inclusion of benzene guest molecules (highlighted). Ni (green), C (gray), N (blue), S (yellow), and H (white).

(a)

(b)

Figure 2 Crystal structure for one Hofmann-type clathrate.3 (a) Single 2D periodic layer outlining the square planar Ni(CN)4 and the octahedral Fe(CN)4 (NH3 )2 . (b) Benzene guest molecules (highlighted in orange) are enclathrated by the 2D layers. Fe (red), Ni (green), C (gray), and N (blue).

in Hofmann-type compounds prohibited layers from interconnection into infinite, 3D structures. Thus, it is obvious that replacement of the terminal ligand, ammonia, by a bridging ligand will permit the construction of 3D networks. In fact, the early developed Prussian blue,8 octahedral mixed-valance iron metal ions bridged by cyanide with the general formula Fe4 [Fe(CN)6 ]3 ·×H2 O, its first single-crystal X-ray diffraction characterization was conducted by Ludi and coworkers,9 confirmed the potential to construct 3D coordination polymers through control of metal coordination and assisted metal–ligand assembly. Earliest examples of structurally investigated coordination polymers dates back to 1936, where Keggin and Miles reported the structure of Prussian blue and related compounds,10 and Griffth in 1943, who reported the crystal structure of silver oxalate.11

2

METAL–ORGANIC FRAMEWORKS (MOFs)

MOFs are regarded as a subset of coordination polymers sharing topologies with existing minerals, zeolites, or, in many cases, with novel topologies. Coordination polymers, a term first introduced by J. C. Bailar in 1964,12 are commonly encountered in chemical literature. The seminal contributions by Robson et al. in 198913–15 marked the beginning of a rapidly evolving interest in coordination polymer research as part of materials science. The feasibility of constructing crystalline materials containing solvent-filled cavities with relatively larger dimensions, yet with essentially similar topologies of inorganic salts or minerals, was first referred to by Robson et al. in 1989.13–15 The first example reported by Robson et al. is a diamondlike network based on coordination of tetrahedrally arranged

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks

Figure 3 Hoskins and Robson’s first report of a 3D coordination polymer based on diamondlike net. The tetrafluoroborate counterions were highly disordered and not localized in the X-ray crystal structure.13

The crystalline nature of such materials permits unequivocal structural characterization through diffraction techniques. Widely employed are X-ray and neutron diffraction, essential to structure–function relationship studies. In addition, the ability to generate coordinatively unsaturated metal ions has been explored for catalytic transformations and applications pertaining to guest–substrate interactions. The organic moiety itself provides the capability to enhance surface area as well as providing functional sites for guest interactions, a property proven useful in gas or liquid selective sorption and molecular sensing. Accordingly, MOFs offer a great potential to answer some enduring challenges in gas storage, separation, selective guest sensing, enhanced heterogeneous catalysis, magnetism, and nonlinear optics.16–29

3 carbonitriles in 4,4
, 4

, 4


-tetracyanotetraphenyl methane and tetrahedral Cu(I) ions. The resulted diamondlike network has relatively low density because of the expansion of the linear linker, benzonitrile molecules, connecting tetrahedral vertices (Figure 3). MOFs have emerged as a novel class of porous solid-state materials with hybrid chemical composition of organic molecules bearing a wide array of functional groups capable of metal coordination and/or chelation and metal ions/clusters. Several attributes of such materials have captured the interest of scientists in academia and industry because of the wide potential applications in current demanding technologies. MOF attributes are directly correlated to the tunable hybrid chemical composition, crystallinity, and modular nature. In order to construct highly ordered solid-state materials, a plethora of design principles and experimental techniques were devised. Although these kinds of constructs might share some similarity to previously known silicates, sol–gel, zeolites, and mesoporous material, one distinct feature of metal–organic materials (MOMs) is the relatively mild reaction conditions employed in the synthesis, permitting maintained integrity of well-defined building blocks generated in situ from molecular or ionic precursors and thus allowing enhanced control and predictability of the composition and topology of the final construct. Of particular interest regarding structural properties of MOMs are the ability to generate permanently porous solids that possess unprecedented high surface areas and structural rigidity on removal of guest molecules.

3

METAL–ORGANIC FRAMEWORKS UTILIZING POLYTOPIC, NITROGEN-DONOR LIGANDS

The earliest examples of coordination polymers constructed through the node-and-spacer approach utilized metal ions as nodes and neutral N-donor ligands as spacers. Examples of neutral N-donor ligands utilized in the construction of coordination polymers include pyrazine,30a 4,4
-bipyridine,30b pyrimidine,30c triazine,30d and hexamethylenetetramine (HMTA) 30e (Figures 4–6). Examples of permanently porous MOFs based on neutral N-donor ligands are scarce,31a mainly due to the relatively flexible nature of the N–M bonds around the metal center.32 In addition, framework interpenetration is commonly observed in this class of coordination polymers, limiting the maximal attainable pore size in isoreticular compounds constructed through expansion of the linker.31b In fact, MOFs based on anionic N-donor ligands and metal clusters as nodes have been synthesized,

N

N N

N

Pyrazine

N Pyrimidine

1,3,5-Triazine

N

N 4,4′-Bipyridine

Figure 4

N N

N

N N

N

Hexamethylenetetramine

Examples of N-donor polytopic linkers.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

4

Supramolecular materials chemistry

Figure 5 [Cu(pyrimidine)2 ]BF4 framework30c , hydrogen atoms, and tetrafluoroborate counterions omitted for clarity. Cu (green), C (gray), H (white), and N (blue).

Figure 7 Crystal structure of the framework obtained by Long et al. 33b based on the tetrazolate linker and Mn. The network is obtained through association of cubelike Mn4 (µ4 -Cl)(tetrazolate)8 building blocks interconnected through tritopic phenyl rings. Disordered solvent molecules and solvated Mn ions are omitted for clarity. Mn (magenta), Cl (green), C (gray), O (red), H (white), and N (blue).

4

Figure 6 Crystal structure of the framework attained from HMTA as an axial, tritopic, bridging ligand to Cu(II) propionate clusters, aliphatic carbons of propionate ions and hydrogen atoms omitted for clarity.30e Cu (green), C (gray), N (blue), and O (red).

which surmounted challenges faced by coordination polymers based on neutral N-donor ligands. A representative example is the tetrazolate-based MOF reported by Long et al., constructed from a solvothermal reaction of Mn and 5,5
, 5

-benzene-1,3,5-triyltris(1H -tetrazole)33 (Figure 7). The attained framework was experimentally proven to maintain its structural integrity on removal of guest molecules.

METAL–ORGANIC FRAMEWORKS UTILIZING POLYTOPIC, CARBOXYLATE-BASED LIGANDS

Early on, difficulties were encountered in activation of coordination polymers constructed from neutral N-donor organic ligands. Accordingly, the focus has shifted toward anionic carboxylate-based ligands as an attractive alternative to construct robust materials that retain their structural integrity on removal of guest molecules. Moreover, owing to the anionic nature of carboxylate ligands, a wide variety of positively, neutral, or negatively charged frameworks could be accessed depending on the ligand-to-metal ratio in the overall formulation and on the oxidation state of the metal ions employed. In addition, owing to the various coordination modes of carboxylate ions (Figure 8), it was anticipated that a plethora of compounds could be assembled from the same starting materials depending on the reaction conditions that potentially could be controlled to direct the desirable coordination mode of carboxylate ions. An important feature of carboxylate-based linkers is their tendency to generate, in situ, various metalcarboxylate clusters (Figure 9). Utilization of such clusters, highly connected building blocks, permits the construction of solid-state materials not accessible from single-metalion building blocks. The metal-carboxylate cluster can be regarded as a molecular building block (MBB); the points of extension of the MBB define a conceptual geometrical entity, commonly referred to as a secondary building unit (SBU). On the basis of the geometry of the MBBs depicted

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks OH

O

O

O

O

M

M

M

M

M

O

O M

M

O

O

O

OH

M

O

O

HO

O

O

OH

OH

Figure 8 Different coordination modes of carboxylate ions, M denotes metal ion.

N HO

O

O

N

OH O

O OH

(a)

5

OH

Figure 10 Three polytopic carboxylic acid ligands utilized in construction of MOFs with exceptional structural stability on guest removal and desirable gas sorption properties.

(b)

(c)

Figure 9 Metal-carboxylate clusters commonly encountered and utilized as MBBs. Left is the actual cluster and the right is the conceptual secondary building unit (SBU) depiction. (a) Binuclear tetracarboxylate paddle wheel M2 (RCO2 )4 , (b) basic chromium acetate trimer M3 O(RCO2 )6 (H2 O)3 , and (c) basic zinc acetate, M4 O(RCO2 )6 , R represents the organic linker, the points of extensions are highlighted in yellow. Metal ion (green), C (gray), N (blue), and O (red).32

in Figure 9, corresponding SBUs can be visualized. The paddlewheel-like metal-carboxylate MBB, found in copper acetate, can be regarded as a squarelike SBU. The metalcarboxylate trimer MBB found in chromium acetate and analogous species can be regarded as a trigonal prismatic SBU. The metal-carboxylate tetranuclear MBB, found in basic zinc acetate clusters, can be regarded as an octahedral SBU.32, 34–49 The wide array of either commercially available or synthetically accessible polytopic carboxylic acids provides a seemingly endless supply for materials scientists to design and construct a variety of carboxylate-based solid-state materials. Three representative carboxylic acid ligands that

Figure 11 Crystal structure of MOF-5,50 guest solvent molecules omitted for clarity. The yellow sphere represents a ˚ that can fit inside the regular cubic cages sphere of radius 7.1 A of the framework without touching the van der Waals radii of the closest H atoms. Zn (green), C (gray), O (red), and H (white).

have been utilized to construct three different porous MOFs with unique gas sorption properties are shown in Figure 10. Yaghi et al., in 1999,50 reported the synthesis of a prototypal MOF, commonly known as MOF-5, based on the basic zinc acetate MBB (Figure 11). The structure could simply be described in terms of octahedral SBUs interconnected through the linear organic linker (benzene ring) to result in primitive cubic (pcu) topology. Apart from this example representing the successful implementation of the SBU design strategy to approach highly symmetric and crystalline construct, the physical properties of the structure are remarkable because of its maintained structural integrity on desolvation and, accordingly, the ability to utilize the porous nature of the construct toward gas sorption applications. The modularity of this approach was fully exploited through utilization of various analogs of terephthalic acid, all maintaining the relative disposition of the carboxylate functionality but with either an expanded or substituted

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

6

Supramolecular materials chemistry

aromatic ring system. The resulting compounds share the same underlying topology of the prototypal MOF-5 but with far more enhanced sorption properties and/or surface area as a direct result of organic linker functionalization and/or expansion.51–53 The family of reported compounds is termed isoreticular metal–organic frameworks (IRMOFs). The framework constructed from reaction of the tetratopic carboxylic acid linker, 5,5
-(E)-diazene-1,2-diyldibenzene1,3-dicarboxylic acid, and In(NO3 )3 resulted in an MOF with square–octahedral (soc) topology containing the basic chromium acetatelike [In3 O(CO2 )6 (H2 O)3 ] clusters regarded as trigonal prismatic SBUs.54 This framework exhibits interesting hydrogen sorption properties because of the highly localized charge density around the indiumcarboxylate clusters. Two distinct types of channels exist in the soc-MOF, represented in Figure 12. Another noticeable example of an MOF constructed from a polytopic carboxylic acid and the binuclear metalcarboxylate paddlewheel-like cluster MBB, equivalent to a square planar SBU, is the one commonly referred to as HKUST-1 MOF, with the general formula [Cu3 (TMA)2 (H2 O)3 ]n where TMA denotes trimesic acid (1,3,5-benzenetricarboxylic acid) (Figure 13,55 ). The MBB is defined by four carboxylate ions coordinating two Cu(II) ions in a paddlewheel-like structure with four points of extension defining a square-planar SBU. The authors reported the capability of exchanging the axial water molecules coordinated to the Cu(II) paddlewheel, postsynthetically, with pyridine molecules, opening the door for further investigations toward the viability of preferential sorption of small molecules and/or catalysis on coordinatively unsaturated metal centers.

Figure 12 Crystal structure of the soc-MOF. The yellow sphere represents sphere that can fit inside the cages present in the framework without touching the van der Waals radii of the closest atoms of the framework. Solvent molecules and nitrate counterions omitted for clarity. In (green), C (gray), O (red), and N (blue).54

Figure 13 The crystal structure of HKUST-1 MOF showing the interconnectivity of the paddlewheel-like Cu2 (RCO2 )4 to the tritopic, triangular 1,3,5-benzenetricarboxylate linkers. Guest solvent molecules omitted for clarity. Cu (green), C (gray), O (red), and H (white).55

5

METAL–ORGANIC FRAMEWORKS UTILIZING POLYTOPIC, HETEROFUNCTIONAL LIGANDS: THE SINGLE-METAL-ION MBB APPROACH

In the previously described approaches toward construction of MOFs, the attainment of materials with expanded structure is facilitated through expanding the linker while maintaining the same metal-ion cluster as the N-connected nodes. Although this approach has met with success, further reduction to the framework’s density could be attained through replacing the metal ion cluster by Nconnected single-metal ion in conjunction with expanded linker. However, as demonstrated in several examples of frameworks based on single-metal-ion nodes and polytopic, monodentate linkers, the framework’s structural stability is hampered by the flexibility of coordinated ligand molecules around the metal-ion center.31 Therefore, we opted to utilize chelating polytopic linkers to impart structural and chemical stability to the single-metal-ion node, an approach we termed the single-metal-ion MBB approach. In this approach, rigidity and directionality are implemented at the MBB through chelation of the metal ion by a chelating ligand. As chelate metal-ion complexes have much higher stability compared to complexes of monodentate ligands, it was anticipated that MOFs constructed from such MBBs would express superior chemical and physical stability as compared to those constructed from monodentate complexes. Another, equally important, design element in the single-metal-ion MBB approach is the directionality

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks O O HO

OH

O

O

O

O

HO N

NH

1H-Imidazole-4,5dicarboxylic acid

Figure 14

HO

N

OH

OH N

Pyridine-2,5dicarboxylic acid

7

N

N

O

OH

N

Pyrimidine-4,6dicarboxylic acid

Pyrimidine-2carboxylic acid

Examples of heterofunctional bis-bidentate organic linkers.

imparted by the linker in bridging the MBBs. To induce the required directionality, we opted to utilize ditopic bidentate linkers containing both aromatic nitrogen atoms and carboxylic acid groups. It was expected that because of the more pronounced directionality of a metal-nitrogen coordination bond that N atoms of the linker would direct the connectivity of the MBBs while carboxylate at the αposition to the N atom would establish the five-membered ring chelate with the coordinated metal ion. In contrast to the polytopic carboxylate-based linkers (where multinuclear metal-carboxylate MBBs are frequently encountered), our approach utilizes single-metal-ion MBB as a direct result of the chelating nature of the ligands utilized.56–59 Examples of heterofunctional, bis-bidentate ligand molecules that have been successfully utilized by our group, among others, in the construction of functional solid-state materials

according to the single-metal-ion MBB approach are shown in Figure 14. Three illustrative examples demonstrating the viability of this approach toward construction of novel functional materials are represented. The first example demonstrates the ability to utilize the octahedrally coordinated In(III) metal ion to generate the 4-connected InN2 (CO2 )4 MBB. The octahedrally coordinated In(III) permitted the formation, in situ, of 4-connected building units that on coordination to 2,5-pyridinedicarboxylic acid (2,5-PDC) resulted in a 2D MOF having a Kagom´e topology based on 4-connected vertexes (Figure 15). The 2,5-PDC was chosen as a ligand containing concurrently the chelating and bridging functionality, to facilitate the construction of the targeted MOF based on 4-connected inorganic MBBs. In this MBB, the In–N bonds and In-(5-carboxylate) bonds direct the

Figure 15 (From left to right): the crystal structure of Kagom´e MOF, the coordination of In(III) ions by four linkers, coordination sphere around In(III), and the resulted 4-connected single-metal-ion MBB. In (green), C (gray), O (red), N (blue), hydrogen atoms, and solvent molecules omitted for clarity.59

Figure 16 (From left to right): the crystal structure of metal-organic cube (MOC-1), the coordination of Ni(II) ions by three chelating linkers, coordination sphere in fac-NiN3 O3 , and the resulted tri-connected single-metal-ion MBB. Ni (green), C (gray), O (red), N (blue), and H (white).60 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

8

Supramolecular materials chemistry

Figure 17 (From left to right): the crystal structure of metal–organic octahedron, the coordination of In(III) ions by four linker molecules, simplified coordination sphere in cis-InN2 O4 , and the resulted seesawlike single-metal-ion MBB. In (green), C (gray), O (red), N (blue), and H (white).59

topology (positioning each In at 120◦ via the PDC) and the PDC α-carboxylate oxygen atoms complete the In(III) coordination sphere, resulting in a quasi-planar 4-connected building unit.59 The second example of interest is the construction of cubelike metal–organic polyhedra (MOP) utilizing 1H imidazoledicarboxylic acid and Ni(II) metal ions.60 Owing to the favorable fac-NiN3 O3 heterocoordination sphere, chelation of Ni(II) ions by three bis-bidentate linkers resulted in a single-metal-ion MBB that could be visualized as a 3-connected node. On extension of such MBB, generated in situ, to three neighboring MBBs directed by the N-atoms of the linker (acting as a linear spacer), a cubelike MOP was constructed (Figure 16). The octahedronlike MOP was isolated from a reaction of In (NO3 )3 ·xH2 O and pyridine-2,5-dicarboxylic acid.59 The octahedrally coordinated In(III) ions serve as the 4connected, seesawlike, vertices of the octahedron MOP. The discrete M6 L12 octahedron is constructed from cisInN2 O4 MBB, regarded as a trans pyramidal building unit, that results from distortion of the conceptual seesawlike building unit in which monodentate oxygen atoms are trans (Figure 17).

6

FROM ZEOLITES AND MOFs TO ZEOLITELIKE MOFs (ZMOFs)

Zeolites are purely inorganic microporous aluminosilicate or aluminophosphate materials that occur naturally as minerals or have been synthetically prepared. The anionic charges of aluminosilicates are balanced by guest cations, for example, Na+ , K+ , Ca2+ , and Mg2+ . Connectivity of tetrahedrally coordinated atoms (T-O-T) through the ditopic angular (about 145◦ ) oxide is key to afford the noninterpenetrating porous structures of zeolites. Zeolitic structures demonstrate a wide variety of regular cages, channels, and windows as a result of the unique connectivity of tetrahedral nodes through an angular linker.

These very aspects of zeolites have enabled their utilization in various industrial applications, for example, heterogeneous catalysis, ion exchange (cation exchange from aqueous solution, particularly useful in water softening), separation of mixtures in petrochemical industry, and gas storage.61–74 Owing to the ability to exchange guest cations with others on contact in solution, zeolite beds are widely employed as water softeners. Moreover, a number of zeolites demonstrate the ability to confine molecules in their micropores and consequently to induce changes in the structure and/or reactivity of guest molecules, facilitating specific chemical transformations. One prominent example of industrial applications of zeolites is encountered in the petrochemical industry, the catalytic cracking of crude oil. The hydrogen-exchanged zeolites act as powerful solidstate Lewis acids, facilitating acid-catalyzed reactions of confined guest molecules such as isomerization, alkylation, and cracking of hydrocarbons. Confinement of guest molecules in such highly charged chemical environments drastically alters their reactivity and hence provides an amenable pathway to otherwise difficult or costly chemical transformations.75 The regular windows and channels in zeolites with specific dimensions are the basis for their shape-selective properties and accordingly their utilization in purification of gas mixtures or mixtures of branched and linear hydrocarbons. The ability to preferentially adsorb certain molecules and simultaneously excluding others led to the introduction of the term “molecular sieves” for zeolites and their applications in separation techniques based on their size-exclusion properties. Purification of p-xylene by MFI zeolite76 (Figure 18) demonstrates the ability of zeolites to separate mixtures based on the molecular size, and hence shape, as certain types of molecules will diffuse through the regular channels and windows of the zeolite, having been separated from other more steric or branched molecules. The systematic classification of the relatively large number of existing zeolites, about 170 zeolites,77d relies on the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks

(a)

Figure 18

9

(b)

Crystal structure of the MFI zeolite (a), and corresponding topology (b), utilized in purification of p-xylene.76

(a)

(b)

Figure 19 Crystal structure of the ACO zeolite (a) and its corresponding topology (b) showing connectivity of tetrahedral atoms that could be constructed from periodic double 4-ring (D4R) as the zero-dimensional periodic building unit (PerBU), in bold.

crystalline nature, and hence three-dimensional periodicity, of such frameworks to generate specific zero-, one-, or two-dimensional structurally invariant periodic building units (PerBUs). The PerBUs are built from smaller units composed of a limited number of tetrahedrally coordinated atoms by applying simple operation(s) to the smaller unit, for example, translation, rotation. PerBUs include chains, tubes, layers, double 4-rings (D4Rs), double 6-rings (D6Rs), and cages, Figure 19, for example. Another structural characteristic of zeolites is the framework density, ˚ 3, defined as the number of tetrahedral atoms per 1000 A which could be used as a measure to compare between different zeolites in a manner that relies on framework density to the openness of the framework. This measure is especially significant in comparing openness of MOFs with zeolitic topologies to inorganic zeolites. The larger dimensions of zeolitic MOFs will naturally result in fewer tetrahedral nodes per specific volume compared to zeolites, and hence lower framework densities are expected, and indeed observed, for zeolitic MOFs. One particular aspect of synthetic zeolites is the ability to obtain a variety of different topologies through modification of the synthesis conditions. Most prominently, incorporation of structure-directing agents (SDAs) facilitates isolation of particular zeolites and can even result in

novel ones.77 Owing to the anionic nature of aluminosilicates, synthetic pathways that employ cationic SDAs such as alkyl ammonium salts that can direct the assembly of the building units and/or limit the size of cages or windows of the zeolite have met with great success in isolation of numerous synthetic zeolites. However, targeting novel zeolites with larger cages for applications pertinent to encapsulation of relatively large functional molecules faces limitations imposed by the observed upper limit of void volume in zeolites, ∼50% as well as the tendency to obtain zeolites with one-dimensional pore systems, that is, channels, instead of expanded cage dimensions.78 This limitation restricts practical applications of inorganic zeolites as host matrices to relatively small-size functional molecules.79 Therefore, our group, among others, has opted to explore alternative routes to construct novel materials with zeolitelike topologies and characteristics, with special interest are anionic crystalline materials that exhibit forbidden interpenetration, guest-exchange capabilities, and good thermal and chemical stability. Turning to reticular chemistry, it appeared that two particular strategies, namely edge-expansion and vertex decoration,80 could be implemented as design principles to construct such zeolitelike materials.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

10

Supramolecular materials chemistry

7

ZEOLITIC METAL–ORGANIC FRAMEWORKS BASED ON THE EDGE-EXPANSION APPROACH

F

Br

OH

O N

N

N

OH

Employing edge-expansion strategy to design and construct zeolitelike materials with anionic nature and extra-large cages compared to zeolites relies on substituting the bridging, angular oxide ion with a larger molecular linker that can provide similar angularity in bridging coordinated metal ions. The imidazole and pyrimidine rings emerged as potential molecular species in the edge-expansion approach, able to serve as ditopic angular linker when coordinated to two metal ions (Figure 20). Utilizing pyrimidine, or its derivatives, as a linker in constructing MOFs with zeolitic topologies through the edge-expansion approach has met with some success. One of the earliest examples81 of an MOF that demonstrates BCT zeolitic topology is the one constructed from 2amino-5-bromo-pyrimidine and Cu(II) (Figure 21). Other pyrimidine derivatives have been successfully utilized to construct ZMOFs through solvothermal reactions with appropriate metal ions (Figure 2282–86 ).

O

OH

N

N

N

N

N

NH2 HO

N

N OH

N

N

O

OH

Figure 22 Pyrimidine derivatives utilized in construction of MOFs with zeolitic topologies.

HN

N

3Å O Si

M

N

N

6Å

M

Si

M

N

N

M

6Å

Figure 20 The edge-expansion approach to construct MOFs with zeolitic topologies. Distances indicated in the figure are average Si· · ·Si, in zeolites, and M· · ·M distances in dinuclear imidazolate and pyrimidine complexes (CSD V5.30, Jan 2011).

Figure 21 Crystal structure of the MOF constructed through reaction of 2-amino-5-bromo-pyrimidine and Cu(II) with zeolitic BCT topology. Cu (green), Br (brown), N (blue), C (gray), and H (white). Disordered solvent guest molecules omitted for clarity.

Figure 23 Association of a tetrahedrally coordinated Cu(II) ion to four imidazolates to construct SOD zeolitic framework, four- and six-member ring windows to the zeolite sod -cage are shown.87b Cu (green), N (blue), and C (gray).

In a close analogy to pyrimidine, the imidazole ring offers the potential to construct MOFs with zeolitic topologies because of the proper bridging angle provided by the imidazolate ion. Several reports of coordination polymers having zeolitic topologies based on imidazolate ligands have emerged in relatively recent literature87 (Figure 23). The potential of imidazole rings in constructing MOF with zeolitic topologies based on edge-expansion of structurally analogous zeolites was first recognized in a report by Tian et al. in 2002 (Figure 24). 88a Yaghi and coworkers have demonstrated the versatility of the edge-expansion approach through synthesis of a large family of zeolitic imidazolate frameworks (ZIFs), Figure 25, for example. The ZIFs obtained include both previously known zeolitic topologies such as SOD, RHO, ANA, BCT, DFT, GME, GIS, MER, and LTA, as well as novel ones such as POZ and MOZ.89–91

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks

Figure 24 Topology and natural tiling of the Co(II)-imidazolate nog-framework reported by Tian et al.88a

Figure 25 Topology and natural tiling of ZIF-70 with GME zeolitic topology constructed from 2-nitroimidazole and Zn(II) ions.89

To impart rigidity and chemical stability to the targeted ZMOF, our group opted to utilize the imidazole derivative 1H -imidazole-4,5-dicarboxylic acid (H3 ImDC),92 as the organic linker providing the proper bridging angle between coordinated metal ions through its nitrogen atoms, while, simultaneously, acting as a chelating ligand through metal-ion coordination to the ring’s α-carboxylates (Figure 26). Accordingly, construction of anionic materials with zeolitic topologies having extra-large cages, because of the expanded edges, was successful. Solvothermal reactions of In(NO3 )3 · × H2 O and H3 ImDC afforded anionic,

11

ZMOFs that demonstrate the targeted structural and chemical properties. Moreover, analogous to the commonly encountered roles of SDAs in the syntheses of zeolites, our group has demonstrated the ability to direct the resulting topology through incorporation of different SDAs in the initial reaction mixture. This approach marked the hybridization of two types of materials known previously in solid-state chemistry, zeolites, and MOFs, and hence the term ZMOF was introduced to designate materials with hybrid chemical composition and zeolitic topologies (Figure 27). The MBB present in ZMOFs is In(Hn ImDC)4 , n = 0, 1, octahedrally coordinated In3+ ions interconnected through the bis-bidentate Hn ImDC linker. The carboxylate functionalities present on the imidazole ring serve to impart structural rigidity and chemical stability to the constructed materials because of the well-known chelate effect in stabilizing metal-ion complexes. Moreover, owing to the anionic nature of the carboxylate ions, overall anionic ZMOFs could be constructed. The anionic nature of ZMOFs based on the ImDC linker enables utilization of cationic SDAs directing the construction of a variety of zeolitic topologies. This was successfully demonstrated experimentally by isolating three different ZMOFs constructed from essentially the same MBB but in the presence of different SDAs. Namely, the nitrate salts of doubly protonated 1,3,4,6,7,8-hexahydro2H -pyrimido[1,2-a]pyrimidine (HPP), imidazole (Him), and 1,2-diaminocyclohexane (DACy) were utilized as SDAs in synthesis of rho-, sod-, and usf-ZMOFs, respectively.92 In a similar approach, pyrimidine-based ligands were utilized to construct ZMOFs with RHO and SOD-zeolitic topologies.85 Coordination of pyrimidine-2-carboxylate to Cd2+ ions provided the charge-neutral rho-ZMOF, while the reaction of pyrimidine-4,6-dicarboxylic acid and In3+ resulted in the anionic sod-ZMOF (Figure 28). In the pyrimidine-based linkers, the aromatic nitrogen atoms direct the topology of the ZMOF in a similar manner of oxide ion in zeolites, while the α-carboxylates serve to chelate the metal ion and result in the overall charge-neutral ZMOF.

Figure 26 The In(ImDC)4 MBB present in ZMOFs could be visualized as tetrahedral connected node interconnected through bisbidentate angular ImDC linker. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

12

Supramolecular materials chemistry

(a)

(b)

Figure 27 Crystal structures (a) of the sod, rho, and usf-ZMOFs and (b) the corresponding SOD, RHO, and USF zeolitic topologies. In (green), C (gray), O (red), N (blue), and H (white).92

(a)

(b)

Figure 28 Crystal structures of the In-based sod-ZMOF (a) and Cd-based rho-ZMOF (b). Cd or In (green), N (blue), C (gray), O (red), H atoms, and disordered guest molecules are omitted for clarity. Yellow spheres represent the guest-accessible voids inside the ZMOFs.

8

ZEOLITIC METAL–ORGANIC FRAMEWORKS BASED ON THE DIRECTED ASSEMBLY OF SUPERMOLECULAR BUILDING BLOCKS (SBBs)

The ability to control coordination number, and geometry, around metal ions, regarded as nodes, through association to polytopic ligand molecules, regarded as linkers, affords construction of predesigned finite and rigid MOPs. Extension of discrete MOPs into extended frameworks can

be attained from MOPs containing peripheral functionalities that can be employed as coordinating species toward metal ions or as hydrogen-bond donors/acceptors in conjunction with other hydrogen-bond acceptor/donor species. This strategy calls for utilization of MOPs as SBBs in the construction of MOFs, a strategy with great potential to enhance our control over the targeted framework.93, 94 In this approach, the SBB, formed in situ through association of MBBs, is utilized as a building unit with a larger dimension and more complex connectivity.95 In general, when constructing MOFs from SBBs, the points of extension of the SBB define a geometric building unit that is equivalent to augmenting a node in a network.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks Programming the SBB with a hierarchy of appropriate information to promote the synthesis of targeted structures, while simultaneously avoiding other easily attainable nets,96 represents a significant advancement in framework design.97 Of special interest in crystal chemistry are edgetransitive nets since they are simple networks composed of one kind of edge.98 Recently, one particular MOP, the small rhombihexahedron or “nanoball,”44 assembled from [Cu2 (O2 CR)4 ] paddlewheel MBBs, as square SBUs, is utilized as an SBB to construct novel MOFs. The organic linker 1,3benzenedicarboxylic acid (1,3-BDC) provides suitable functionality and angularity (120◦ ) between the two carboxylic groups to construct the nanoball based on the paddlewheel metal-carboxylate MBB. Proper substitution at position 5of the 1,3-BDC provides the handle to link such SBBs through covalent or coordination bonds. It is through proper choice of the functional group introduced to the 1,3-BDC that material scientists were able to control connectivity and therefore topology of the MOF based on the small rhombihexahedron MOP as SBB.94(a–c) More recently, our group utilized a tetrazole functionalized 1,3-BDC to construct a (3,24)-connected MOF with unprecedented rht topology (Figure 2994(d) ). The small rhombihexahedron is constructed from coordination of carboxylate functionalities in the 1,3-BDC, while tetrazolate-coordinated Cu(II) ions form trinuclear trigonal clusters, facilitating connectivity of the MOPs, regarded as SBBs, into the rht-MOF. Similarly, the SBB approach can be utilized to construct ZMOFs. Through the single-metal-ion MBB approach, nondefault98 ZMOFs can be targeted using heterochelation to generate rigid and directional single-metal-ion-based MBBs, while the SBB approach toward construction of such materials remains largely unexplored. A particular subset of zeolite nets exhibits a common SBU composed of eight tetrahedra bridged through oxide ions in a cubelike arrangement, commonly referred to as a D4R. Such an SBU is capable of extending into an infinite network through the coordination of the SBU’s vertices to eight oxide ions.99 The analogy of this SBU to the

13

MOC suggests that MOCs could be used as SBBs to target zeolite nets based on D4Rs. Our approach encompasses using MOCs as 8-connected building blocks, which can be regarded as D4Rs to construct ZMOFs. This approach could alternatively be visualized as augmenting the nodes of related 8-connected edge-transitive nets, that is, substituting an 8-connected node by the vertex figure, a cube in this case. The D4Rs can be connected through linear linkers to construct nets based on zeolites LTA or ACO or through 4-coordinate nodes to result in AST- or ASV-like topologies. The aforementioned zeolites (ACO, AST, ASV, and LTA) are especially interesting to reticular chemistry as their nets correspond to the augmentation of the edgetransitive nets bcu, flu, scu, and reo, respectively, where the D4Rs serve as the cubelike vertex figures.98 Specifically, bcu and reo are both semiregular 8-connected nets, and scu and flu are edge-transitive (4,8)-connected nets (Figure 30). The metal-organic cube (MOC)100 consists of eight vertices occupied by tri-connected nodes bridged through 12 4,5-imidazole dicarboxylate (Hn ImDC, n = 0–1) linkers. Expanding the coordination of such vertices can result in interconnected tetrahedra similar to the D4R units in zeolites. As the MOCs possess peripheral carboxylate oxygen atoms, they have the potential to coordinate additional metal ions and/or participate in hydrogen bonding to construct extended structures, specifically ZMOFs.101 Reaction of (H3 ImDC) and Zn(NO3 )2 ·6H2 O in a N,N
dimethylformamide (DMF)/H2 O mixture in the presence of excess zinc and guanidinium nitrate result in colorless polyhedral crystals containing the expected anionic Zn-based MOCs (Figure 31). The presence of excess zinc and guanidinium ions permits the linkage of the MOCs through the oxygen atoms of the organic linker to form extended zeolite-like framework with AST topology. The as-synthesized compound is formulated as {[Zn8 (ImDC)8 (HImDC)4 ]Zn4 (DMF)8 (H2 O)4 (guanidinium)8 }, ast-ZMOF, where the anionic MOCs are formulated as [Zn8 (ImDC)8 (HImDC)4 ]102 . Guanidinium ions were employed to direct the assembly of such highly functional anionic MOCs through H-bonding interactions,

Figure 29 (From left to right) Combination of 5-(1H -tetrazol-5-yl)-isophthalic acid and Cu(II)-carboxylate paddlewheel (MBBs) generate the rhombihexahedron, SBB. Association of the SBBs through tetrazolate-Cu(II) trinuclear clusters, MBBs, generate the rhtMOF. Cu (green), N (blue), C (gray), O (red), hydrogen atoms and disordered guest molecules are omitted for clarity. Yellow and purple spheres represent the largest, guest-accessible, voids within the rht-MOF. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

14

Supramolecular materials chemistry

SBB

lta-ZMOF

ast-ZMOF

Figure 30 (Middle) Single-metal-ion-based MBBs (tri-connected nodes and linear spacers) facilitate the assembly of a MOC, which is utilized as 8-connected SBB to generate ZMOFs. Zeolitic nets with AST (top left) and LTA (top right) are constructed based on relations with regular (8-connected)-based nets (bottom). The flu net (bottom left) corresponds to ast-ZMOF (middle left) and the reo net (bottom right) to lta-ZMOF (middle right) when the 8-connected nodes are augmented, or replaced by metal-organic cubes (i.e., D4Rs).

(a)

(b)

Figure 31 (a) Single-crystal structure of ast-ZMOF (guanidinium ions omitted for clarity, yellow sphere represents vdw sphere ˚ that can fit into the AST-cage without touching vdw surfaces of the framework). (b) The MOC SBBs are of a diameter ∼15 A simultaneously linked via edge-to-edge connection through coordinated metal ions (green octahedral) and vertex-to-vertex connectivity through charge-assisted H-bonded guanidinium ions. O (red), N (blue), Zn (green), C (gray).

as an SDA. The guanidinium ions provide suitable molecular recognition sites in conjunction with carboxylate ions, acting as the H-bond donor in charge-assisted hydrogen bond interactions (Figure 31). In the crystal structure of the ast-ZMOF, each guanidinium ion is simultaneously hydrogen bonded to three MOCs through the carboxylate ions present on the vertices of the MOCs. Four guanidinium ions act as a tetrahedral node linking the vertices of four MOCs through multiple chargeassisted hydrogen bonds. Such an arrangement could be visualized as MOCs tetrahedrally-connected through their

vertices, analogous to the connectivity of four D4Rs in AST zeolite. In a separate attempt to direct the linkage between the anionic MOCs bearing carboxylate functionalities, oxophilic alkali and alkaline earth metal ions were introduced in the initial reaction mixture to act as linkers of the MOCs. The choice of oxophilic metal ions was based on the rationale of minimal interference with the construction of the MOC because of the weak competition of those ions toward coordination to nitrogen atoms of the ligand, thus permitting their utilization in the initial reaction

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks

(a)

15

(b)

Figure 32 (a) Single-crystal structure of lta-ZMOF (yellow sphere represents vdw sphere). (b) In the lta-ZMOF, 12 MOCs are connected through a series of sodium ions to generate Cd (green), Na (pink), O (red), C (gray), N (blue), H atoms and guest solvent molecules are omitted for clarity.

mixture with Zn2+ , Cd2+ , Mn2+ , Co2+ , or In3+ ions that exhibit favorable octahedral hetero-coordination facMN3 O3 sphere, essential for construction of the targeted MOC. In ast-ZMOF, six MOCs assemble to create the AST-cage in which spheres with diameters about 15 and ˚ respectively, can fit without touching the van der 12 A, Waals radii of the framework (excluding solvent). Reaction of Cd(NO3 )2 ·4H2 O and H3 ImDC in the presence of Na+ ions results in compound lta-ZMOF formulated as {[Cd8 (HImDC)8 (ImDC)4 ] (H2 Pip)2 Na8 (EtOH) (H2 O)41 } (Pip, piperazine; EtOH, ethanol). In the crystal structure of ltaZMOF (Figure 32), each MOC is linked to eight other cubes through linear vertex-to-vertex connections. Half are connected through hydrogen-bonded water molecules and the other four vertices are connected through a series of four sodium atoms (Figure 32). The framework consists of two types of cages, namely α-cage encapsulated by 12 MOCs and elliptical β-cage enclosed by 6 MOCs. The largest sphere that can fit into these cages without touching the van ˚ for the α-cage der Waals surface of the framework is ∼32 A ˚ and ∼8.5 A for the β-cage. Topologically, the framework can be regarded as having the LTA zeolitic topology or an augmented version of reo network, when both the hydrogen-bonded and sodium-bridged vertex–vertex connections are considered. However, the structure can be interpreted as nbo if only connections through sodium ions are considered.

9

ZEOLITELIKE METAL–ORGANIC FRAMEWORKS AS PLATFORMS FOR APPLICATIONS

Despite the relatively large number of MOFs in current literature with accessible voids stable against guest exchange, the idea of being able to fine-tune an existing structure through postsynthetic modification or by incorporation of

functional guest molecules during the synthesis is especially appealing. This is due to the wide potential applications for a porous functional material that is amenable to postsynthetic modifications or ones that demonstrate targeted functionality due to encapsulated, yet accessible, functional molecules, to generate made-to-order materials. In this approach, a MOF can be regarded as a tunable platform suitable for a variety of desired applications merely through cost-, atom-, and time-efficient modifications. Our group has utilized the large pores of the anionic rho-ZMOF to exchange encapsulated dimethylammonium counterions (present in the as-synthesized MOF) with cationic acridine orange fluorophore for sensing applications (Figure 33).92a The framework possesses α-cages similar to those in RHO-zeolite but with far enlarged ˚ along with 8dimension (diameter of about 18.2 A) ˚ diameter) that member ring windows (with about 9 A allow access of acridine orange molecules to the extralarge cavities, forging a ZMOF impregnated with a functional organic molecule that could potentially be used as sensor. After ion-exchange of dimethylammonium with the cationic acridine orange molecules (due to protonation of AO in aqueous ethanol solution), the favorable electrostatic interactions with the framework preclude leaching of acridine guests from the cages of the rho-ZMOF. The extra-large dimensions allow further diffusion of gaseous molecules or other neutral small molecules chosen due to established acridine–guest interactions and thus the impregnated framework acts as a solid support in the sensing process. Examples of such neutral small molecules include methyl xanthenes or DNA nucleoside bases. We have also explored the potential of encapsulating free-base cationic porphyrin inside the α-cages of anionic rho-ZMOF.103 Encapsulation of 5,10,15,20-tetrakis (1-methyl-4-pyridinio)porphyrin, [H2TMPyP]4+ was accomplished through the ship-in-a-bottle process (Figure 34).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

16

Supramolecular materials chemistry

N

N

N

Na+

(a) 1.2

Na+-rho + Na -rho + AO + + Na -rho + AO + Na

Absorbance (a.u.)

1.0 0.8 0.6 0.4 0.2 0.0 300 (b)

400 500 600 700 Wavelength (nm)

800

Figure 33 (a) Optical images of Na+ -exchanged rho-ZMOF crystals before (left) and after AO exchange (right). (b) Solid-state UV–vis spectra of corresponding rho-ZMOF solids indicating reversible exchange between Na+ ions and AO cationic guest ions.92a

(a)

(b)

Figure 34 Crystal structure of rho-ZMOF (a), hydrogen atoms omitted for clarity, and schematic presentation of [H2 TMPyP]4+ porphyrin ring enclosed in rho-ZMOF α-cage (b, drawn to scale).

In this process, the anionic framework of rho-ZMOF would self-assemble from its molecular and ionic precursors encapsulating the cationic porphyrin present in the initial reaction mixture. This approach was sought due to the myriad applications accessible through anchoring and isolating catalytically active metalloporphyrins into a solid matrix, prohibiting their self-dimerization and oxidative degradation resulting in enhanced catalytic properties and catalyst recyclability. Moreover, an isolated metalloporphyrin encapsulated in a well-defined chemical environment might provide a synthetic model for active sites in various enzymes utilizing metalloporphyrins, of which cytochromes are the most prominent examples.

Moreover, stability of rho-ZMOF in different methanolic solutions of metal nitrate salts enabled us to metallate the encapsulated free-base porphyrin, with a variety of metal ions, forging a versatile precatalyst (the H2 RTMPyP, free-base porphyrin in rho-ZMOF) that, postsynthetically, can be utilized to prepare a variety of encapsulated metalloporphyrins (M-RTMPyP, M = Cu, Co, Zn, Ni, Mn) for applications in different metalloporphyrins. Reactions of In(NO3 )3 ·xH2 O and 4,5-imidazole dicarboxylic acid (H3 ImDC) in a mixture of N,N
dimethylformamide (DMF) and acetonitrile (CH3 CN) in presence of 5,10,15,20-tetrakis(1-methyl-4-pyridinio) porphyrin tetra(p-toluene sulfonate) ([H2 TMPyP] [p-tosyl]4 )

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc123

Zeolitelike metal–organic frameworks

(a)

H2R(4-N-CH3Py)P Mn-R(4-N-CH3Py)P Co-R(4-N-CH3Py)P Cu-R(4-N-CH3Py)P Zn-R(4-N-CH3Py)P

0.5

0.4 Absorbance

have yielded dark red cubic-like crystals, H2 RTMPyP, suggesting the presence of the porphyrin. The powder Xray diffraction pattern of the as-synthesized compound matches that of rho-ZMOF, confirming the construction of the intended crystalline framework. The presence of [H2 TMPyP]4+ inside rho-ZMOF was confirmed by solidstate UV–vis studies of the fully washed crystalline solid, H2 RTMPyP. Interestingly, the porphyrin was not metallated by In3+ present in the assembly conditions. Incubation of H2 RTMPyP in an aqueous solution of Na+ ions showed no release of the porphyrin, as indicated by UV–vis studies of the solution. In contrast, smaller cationic molecules, such as acridines, can be reversibly immobilized (through electrostatic interactions) inside rho-ZMOF cavities by ion exchange due to their smaller dimensions, about G(C2 , R2 )). The three components in (2) and (3), that is, energy of crystal (Gf,C(s) ), energy of building block (Gf,Bn(solv) ), and concentration of building block ([Bn ]), can be used to reverse the crystallization preference of different crystals. The energy difference between two kinds of crystals can be reversed under two physical conditions. However, the chemical potentials (Gf,Bn(solv) and [Bn ]) of the reactants are medium dependent. The energy difference between two sets of reactants (different in type (n) and/or stoichiometry (ν n )) can be reversed in two mediums. Although C1 and C2 have different compositions (i.e., n and/or ν n ), varyingthe medium may reverse their ener getic preferences ( Vn GBn(solv) and/or vn RT ln[Bn ]). This analysis is useful in determining the common structural diversity (different chemical compositions), including

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

Supramolecular isomerism guest-induced isomerism in coordination polymer syntheses. Apparently, large differences in product compositions and/or reaction mediums would give rise to larger G difference between different products. On the other hand, it is expected for the crystalline products (C1 = C2 ) that all variants for a pair of genuine  supramolecular  isomers in (1)–(3) are identical ( Vn GBn(solv) =  Vn RT ln[Bn ] = 0) in a given medium. Variation of reaction medium should have the same energetic effect on different isomers. Therefore, thermodynamic factors of the equilibrium states do not account for mediumdependent genuine supramolecular isomerism. The fact that variation of crystallization mediums can produce genuine supramolecular isomers reflects the kinetic nature of supramolecular isomerism.5

2.2

Chemical aspects

The importance of kinetic factors in isomer crystallization highlights the role played by the crystal–medium interface, particularly during the early stages of crystal growth when crystals are small and surface interactions dominate relative to lattice energy. Chemical variation in the crystallization medium can be achieved by modification of solvent, counterions, and pH, among other factors.

2.3

Physical aspects

Apart from chemical factors that influence the reaction medium, physical variables may also affect crystallization of supramolecular isomers. The most commonly varied physical factors are temperature and pressure, which influence not only the energy of the precursors but also the final isomeric products (Figure 2). Although a given supramolecular isomer may be the thermodynamically favored product under its crystallization conditions, the diffraction and other physical properties of isomer crystals are usually measured at the same temperature and pressure. Thus, physically controlled isomerism of this type remains a kinetic effect. Temperature has long been used as a variable which influences the crystal form obtained. In the case of isomerism, the kinetic isomer at ambient temperature may become thermodynamically favored at a different temperature. One such example19 is provided by the coordination polymers with composition [Cu2 (tcppda)(H2 O)2 ]n (tcppda = N, N, N , N  -tetrakis(4-carboxyphenyl)-1,4-phenylenediamine), 1. The tcppda ligand can adopt two conformations, either square-planar C2h or tetrahedral D2 . On association with the square-planar Cu2 paddlewheels, these give rise to two supramolecular isomers in the form of

−

O O

115 °C −

3

O O− N

N

1 (C 2h )

O O

O− O

lvt

= = =

O −

O O−

120 °C −

N

N

1 (D 2 )

O

O O

−

O

pts

Figure 3 Temperature-dependent isomerism of [Cu2 (tcppda) (H2 O)2 ]n , 1. The tetrahedral conformation of D2 -tccpda and square-planar C2h -tccpda combine with the square-planar Cu2 paddlewheel secondary building unit (SBU) to form lvt or pts networks.

porous coordination polymers with lvt (1A) or pts (1B) network topologies (Figure 3). The relative distribution of the tcppda conformations should be temperature dependent, but they cannot be separated because of the low-energy barrier. However, on coordination, a full separation occurs, with a temperature difference of only 5 ◦ C being required. In most laboratories, pressure-induced isomerism is accompanied by a variation in temperature under solvothermal conditions. High-temperature (and hence high-pressure) solvothermal reactions typically favor high-density isomers. However, the recent increase in the use of new diffraction technology to study high-pressure crystal forms is likely to raise the number of reports of supramolecular isomerism dependent only on variations of pressure.20

3

POLYMORPHISM AND SUPRAMOLECULAR ISOMERISM

Polymorphism is an important topic, which is comprehensively dealt within another chapter in this volume (see Polymorphism: Fundamentals and Applications, Supramolecular Materials Chemistry). Supramolecular isomerism is related to polymorphism in that, in both cases, the chemical constituents of two or more crystalline substances are identical. Indeed, polymorphs may generally be regarded as being supramolecular isomers of one another, although the reverse is not always true.4 Polymorphism of molecular crystals, in particular, remains an important phenomenon, especially in pharmaceuticals. However, although reports of polymorphs have increased significantly

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

4

Supramolecular materials chemistry

5

in recent years (e.g., the Cambridge Structural Database (CSD) contains more than 16 000 reports of polymorphic structures),21 their predictability remains low.22 Desiraju found that polymorphic modifications do not appear uniformly in all categories of molecules.23 Nangia and Desiraju also reported that polymorphism appears to be highly dependent on solvent,24 further complicating the fundamental understanding of the subject. Where guest or solvent molecules are present, the term polymorphism is not applicable. (In the older literature, one encounters the term pseudopolymorphism, which is used to describe solvates especially those of pharmaceutically active compounds. This is misleading and the term should be avoided as such solvates are entirely different compounds.)

4

Moulton and Zaworotko4 suggested that supramolecular isomers could be considered in four categories on the basis of analogies drawn from molecular isomerism.

5.1

Structural supramolecular isomers

The components of the network remain the same, but a different network is generated. In such cases, chemical components and empirical formula are identical. The compound 1,3,5-tris(4-cyanobenzoyl)benzene, 2, provides an example of an organic compound which displays both polymorphism and structural isomerism in its crystal forms.28 Recrystallization of 2 from acetone/water gave two concomitant polymorphs (forms A and B), illustrated in Figure 4. Each polymorph exists in a distinct network structure, held together by C–H· · ·O hydrogen bonds. In form A, these generate a honeycomb network while in form B, a ladder network is produced. As both networks contain the identical molecular building blocks, they are supramolecular isomers of each other. Fromm et al. reported an example of supramolecular isomers of a metal–organic compound which form a ring and a helix network when silver perchlorate is combined with oxydiethane-2,1-diyl dipyridine-4-carboxylate (oddc).

NETWORK ANALYSIS

While some instances of supramolecular isomerism are immediately and visually apparent, it is often difficult to describe the structures briefly enough to be useful, while still fully enough to be complete. A number of topological approaches have been described over the years as aids in the description of extended structures. Both graph set25 and network analyses26 are useful for this purpose. As many supramolecular isomers can be described as 2D or 3D nets, they are amenable to an analysis of their network topology, as described fully in another chapter in this volume (see Network and Graph Set Analysis, Supramolecular Materials Chemistry).27 O

CLASSIFICATION OF SUPRAMOLECULAR ISOMERS

O

Ar

Ar 2 Ar = 4-NCC6H4

(a)

O

Ar

(b)

(c)

Figure 4

(a) 1,3,5-Tris(4-cyanobenzoyl)benzene, 2 forms: (b) honeycomb network, 2A, and (c) ladder network, 2B.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

Supramolecular isomerism N

5

N O

O

O O

O + AgX

3A

(a)

3B

(b)

Figure 5

Isomers of [Ag(oddc)]n (ClO4 )n , 3: (a) molecular ring (3A) where n = 2 and (b) helix with n = ∞, 3B.

Figure 6

1D structure of fused metallacycle rings of [Ag(oddc)2 ] (PF6 ). The chains interpenetrate to form a 2D polycatenated network.

Crystals of both compounds29 are obtained in an H-tube, having an aqueous solution of a silver salt on one side and THF solution of the ligand on the other side, connected via a mixture of both solvents. Crystallization of the ring-shaped compound, [Ag(oddc)]2 (ClO4 )2 3A, took place on the silver salt side of the H-tube, while the formation of a helical network structure having the same empirical composition, 3B, took place on the ligand side (Figure 5). The authors proposed that a concentration effect is responsible for the formation of these supramolecular isomers. Both isomers exist in the same system, but after total diffusion only 3A remains, suggesting that this is the thermodynamically stable product and that 3B can be considered a kinetic product. The same authors later showed30 that the ring is the favored isomer when the perchlorate ion is replaced by nitrate or PF6 − . In the latter case, using AgPF6 as starting material produced either the [Ag(oddc)]2 (PF6 )2 THF solvate metallacycle or a 2D polycatenated structure with composition [Ag(oddc)2 ] (PF6 ), which, although not an isomer of 3A, retains some of its features (Figure 6).

5.2

Conformational supramolecular isomers

When flexible molecules (or flexible ligands in coordination networks) are employed, different network architectures may be generated from the same building blocks. An analogy can be drawn to conformational polymorphism such as that seen in acetone tosylhydrazone, 4.31 A monoclinic and a triclinic form are obtained from anhydrous ethanol, sometimes together. The conformations differ by about 15◦ about the S–C exocyclic bond. H3C

O

N

H3C

NH S

CH3

O 4

O

N

O N NH S 5

CH3

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

6

Supramolecular materials chemistry HO

OH

HO

OH

HO

CH3

CH3

CH3

CH3 OH

HO

OH 6

Guest N

N

H H O H O O H O

N

OO HHO O HH

N N

N

OO H OO H HH

H OH O H OH O

N N

OH O H O OH H N N

O O O HH O H H N N

N N

Guest Guest

N

N H H O HO O HO

Guest H O OH O HH O

N

N N

N N

N

OH O O H O H H N N

OOH O HO HH N N

N N

H H O H O O H O

H OHO H O H O

Guest 7A

7B

HO H O

O H

H

O H H O H

O

N N

O OH

H H O H HO O H O

O

O

O H O H O H O H O H O H

HO H O

H

N N

N N

O OH H N N

Guest N N H H O O HO H O

N N

N N

H O H HO O H O

O H O H O H O H O H O H

O HO OH H O H

H

H O

O

N N H OH O O H O H

7C

Figure 7 C-Methylcalix[4]resorcinarene, 6, forms supramolecular isomers in co-crystals with 4,4 -bipyridine, 7 = 6·2(bipy): hydrogen-bonded wave-like 7A, carcerand-like capsule 7B or 2D brick frameworks 7C. (Reproduced from Ref. 33.  Royal Society of Chemistry, 2001.)

Reports of conformational supramolecular isomers of organic materials include that of the compound 5-methyl2-[(2-nitrophenyl)amino]-3-thiophenecarbonitrile, 5, which has at least six polymorphs.32 The main difference in the six phases is the variation in the torsion angle between the thiophene and the o-nitroaniline moieties. Another example involves the two components C-methylcalix[4]resorcinarene (6) and 4,4 -bipyridine (bipy), which can form three supramolecular isomers (7) in the presence of different

guest molecules.33 The host network in each case is identical in composition, 6·2(bipy), but the conformational flexibility of 6 allows the formation of hydrogen-bonded wave-like 7A, carcerand-like capsule 7B, or 2D brick frameworks 7C, as illustrated in Figure 7. Conformational flexibility in ligands can lead to conformational supramolecular isomers in metal–organic network systems. For example, hydrothermal crystallization using an aminotriazole ligand afforded two 3D supramolecular

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

Supramolecular isomerism

N

coordinate to metal ions in three distinct ways (Figure 8). Compound 8 has abpt in its transoid -configuration linking distorted octahedral copper centers into 1D metal–organic chains, which are then interlaced by [Cu2 (χ-Mo8 O26 )] ladder-like chains to generate a 3D network. In 9, the square-pyramidal Cu(II) centers are firstly linked by cisoid abpt ligands to give 2D metal–organic coordination layers, which are connected by [δ-Mo8 O26 ]4− clusters into a 3D network. Compounds 8 and 9 can be described as 3,4-connected nets with (43 .6.86 )2 (42 .6)2 (42 .84 ) and (4.82 )2 (4.82 .103 )2 topologies. Figure 9 illustrates the two network topologies, and highlights the usefulness of network analysis to identify differences between isomers with identical empirical formulae.

NH2 N

N

N N

transoid -abpt

N

NH2

N

N

NH2 N

N

N

N N

N N

cisoid -abpt-I

7

cisoid -abpt-II

Figure 8 Three potential coordinating conformations for the abpt ligand. (Reproduced from Ref. 34.  Elsevier, 2010.)

5.3 isomers, [Cu2 (abpt)2 (H2 O)2 (Mo8 O26 )]n (8) and [Cu2 (abpt)2 (H2 O)2 (Mo8 O26 )]n · 4nH2 O (9), where abpt = 4-amino3,5-bis(3-pyridyl)-1,2,4-triazole.34 The abpt ligand can

Catananes or interpenetrated networks

Different degrees of interpenetration can result in widely different structures being formed with substantial differences in physical properties, even though the networks N5 N1

N2C

N6D N3

N4 O7 N2 O11 O5 O9 Mo3 O10 O14 Mo2 O2 O3

Mo4

(a)

(b)

O13 O6 O12A

8

Cu1

O9B N6 O4

O1(w)

Mo1 O12 O8 O13A

(c)

9

Figure 9 (a) Coordination environment in 8, showing transoid conformation of abpt. 9 has identical connectivity, but abpt adopts a cisoid conformation. The 3,4-connected topological nets of (b) [Cu2 (abpt)2 (H2 O)2 (Mo8 O26 )]n , 8, and (c) [Cu2 (abpt)2 (H2 O)2 (Mo8 O26 )]n · 4nH2 O, 9. (Reproduced from Ref. 34.  Elsevier, 2010.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

8

Supramolecular materials chemistry chlorobenzene, o-dichlorobenzene, benzene, nitrobenzene, toluene, anisole), appears to require the complementarity of the (4,4) or (6,3) arene in order for the (4,4) coordination network to form.36 The complementary coordination and aromatic networks interpenetrate in a manner described as parallel/diagonal inclined interpenetration as illustrated in Figure 10 for {[Ni(bipy)2 (NO3 )2 ]·chlorobenzene}n , 10. Organic networks can also be interpenetrated, as is illustrated by the networks formed by 1,3,5benzenetricarboxylic acid (BTC), 11, which typically forms a (6,3)-network (“honeycomb”), as shown in Figure 11. In pure BTC,37 threefold self-interpenetration of the hexagonal networks occurs with puckering of each honeycomb network, 11A (Figure 12a). When crystallized in the presence of small solvent molecules, BTC forms an open honeycomb structure with adjacent layers aligned to form small cavities capable of holding molecules such as halogens,38 linear alkanes,39 and p-nitroaniline4 (11B, Figure 12b).

10: {[Ni(bipy)2(NO3)2].(C6H5Cl)}n

5.4 Figure 10 Square-grid coordination and aromatic networks in {[Ni(bipy)2 (NO3 )2 ]·chlorobenzene}n , 10, are complementary and result in an interpenetrated structure.

themselves may be identical. A review of interpenetrated networks was published by Batten and Robson,35 and the topic is treated in greater detail in another chapter in this volume (see Interpenetration, Supramolecular Materials Chemistry). The use of templates can direct the generation of either interpenetrated structures or open frameworks from the same building blocks. For example, the family of squaregrid coordination polymers, {[M(bipy)2 (NO3 )2 ]·arene}n (M = Ni, Co; bipy = 4,4-bipyridine; arene =

N N

N

CN N N C

Ni

CN

N

NC

Networks produced by chiral components and crystallizing in chiral space groups are important for properties such as spontaneous resolution. An analogy can be drawn to homochiral compounds. Formation of chiral helices does not necessarily mean that the resulting crystal will be chiral. As an example, cis-[Ni(f -rac-hmta)][Ni(CN)4 ]n (hmta = 5,5,7,12,12,14hexamethyl-1,4,8,11-tetraazacyclotetradecane) forms two supramolecular isomers, 12A (in which n = 2) and a 1D helical chain, 12B. In 12B, the 1D helical chains pack in right- and left-handed chirality due to the opposite twist arrangements of adjacent [Ni(CN)4 ]2− anions (Figure 13).40 The unit cell contains chains of both chiralities in the centrosymmetric space group P 21 /c.

N Ni

Chiral networks

CN N

Ni

N C

Ni

CN

N N

C

N

C

C

N

C

N

Ni

C N

Ni N

CN

NC

Ni

CN

Ni N

N N

12A

N

CN

N

N N

12B

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

Supramolecular isomerism

∆ 11

Figure 13

9

Λ

Helical chains in 12B, having opposite chirality.

Figure 11 Honeycomb network in BTC, 11. (One BTC molecule has been highlighted for clarity).

6

CRYSTAL ENGINEERING CONTROL AND EXPLOITATION OF SUPRAMOLECULAR ISOMERISM

Crystal engineering aims to control structural characteristics of crystalline materials, with the aim of having greater control over their physical properties. Although hydrogenbonded supramolecular synthons have been studied in depth for more than a decade and although coordination bonds are stronger and more directional than other supramolecular interactions, structural uncertainty is still a characteristic of self-assembled systems. That said, the more we understand the materials we obtain from different forms of synthesis, the closer we are to the goal of control.

(a)

In principle, the structure of a material determines its physical properties. Although it remains difficult to understand the structure–property relationship in a new system, crystal engineering of supramolecular systems offers the possibility of allowing for rational design of new and useful materials. In this context, comparing the properties of isomeric crystals can aid in the analysis of the structural features required for a particular property. Luminescence is one such property which has yielded to the study of supramolecular isomers. The absorption and emission spectra of the supramolecular isomers of [Ni(4-bpd)2 (NCS)2 ], 13, where 4-bpd = 1,4-bis(4-pyridyl)2,3-diaza-1,3-butadiene, showed distinct differences which could be attributed to the differing conformations of the bpd ligand in the 44 square-grid framework versus the 65 · 8 framework structures.41 Both isomers also

(b)

Figure 12 Honeycomb networks in BTC: (a) puckered net in pure BTC, 11A, and (b) planar net in BTC crystals with small guest molecules such as halogens or alkanes, 11B (guest molecules omitted). Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

10

Supramolecular materials chemistry

Dehydration

Rehydration (a)

Five-fold interpenetration

N Ag

Six-fold interpenetration

N G

Ag

N (b)

+ +

N

N Ag

G

G

N

Ag

Ag N

N

G

N

G

N

G = N of atz or O of guest

(c)

Figure 14 (a) SCSC isomerization of [Ag6 Cl(atz)4 ]n (15) and possible mechanisms for (b) self- or guest-assisted bond cleavage and reformation, and (c) catenation rearrangement. (Reproduced from Ref. 5.  Royal Society of Chemistry, 2009.) N

N N N

N

N N NCS

N

N

N

N Ni N

Cu

Cu

N NCS

N N

N

N N

13

N

N N

N

14

15

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

Supramolecular isomerism had reversible solvent uptake properties. In another case, luminescence of three isomers of [Cu(dmtz)], 14 (dmtz = 3,5-dimethyl-1,2,4-triazolate), was explained by consideration of their crystal symmetry, packing efficiency, and cuprophilicity.42 The porous metal–organic framework {[Ag6 Cl(atz)4 ] OH·6H2 O}n , 15, (Hatz = 3-amino-1,2,4-triazole) undergoes a single crystal to single crystal (SCSC) transformation (Figure 14) from a fivefold interpenetrated to a sixfold interpenetrated framework on dehydration. The transformation requires coordination bond cleavage and formation, which is postulated to occur in the presence of a coordinating guest molecule that acts as a mediator.43

11

13. D. B. Cordes, A. S. Bailey, P. L. Caradoc-Davies, et al., Inorg. Chem., 2005, 44, 2544. 14. D.-L. Long, A. J. Blake, N. R. Champness, Chem.–Eur. J., 2002, 8, 2026.

et al.,

15. R. Peng, S.-R. Deng, M. Li, et al., CrystEngComm, 2008, 10, 590. 16. J.-P. Zhang, Y.-Y. Lin, X.-C. Huang, and X.-M. Chen, Chem. Commun., 2005, 1258. 17. P. M Forster, A. R. Burbank, C. Livage, et al., Chem. Commun., 2004, 368. 18. G. R. Desiraju, Angew. Chem. Int. Ed., 2007, 46, 8342. 19. D. Sun, Y. Ke, T. M. Mattox, et al., Chem Commun., 2005, 5447. 20. A. Katrusiak, Acta Crystallogr., 2008, A64, 135.

7

21. CSD statistics report, 1 January 2010. http://www.ccdc. cam.ac.uk/.

CONCLUSION

Supramolecular isomerism provides an useful framework for the analysis and understanding of network structures. At present, prediction of and control over the crystal form obtained remains challenging. Deeper understanding of the thermodynamic and kinetic factors affecting crystallization is required, but advances in crystal engineering provide the hope of increasing control over the results of a given crystallization experiment.

REFERENCES 1. N. J. Turro, N. H. Han, X. G. Lei, et al., J. Am. Chem. Soc., 1995, 117, 4881. 2. D. Henschel, O. Hiemisch, A. Blaschette, and P. G. Jones, Z. Naturforsch B, 1996, 51, 1313. 3. T. L. Hennigar, D. C. MacQuarrie, P. Losier, et al., Angew. Chem., Int. Ed. Engl., 1997, 36, 972. 4. B. Moulton and M. J. Zaworotko, Chem. Rev., 2001, 101, 1629. 5. J.-P. Zhang, X.-C. Huang, and X.-M. Chen, Chem. Soc. Rev., 2009, 38, 2385. 6. E. Tynan, P. Jensen, N. R. Kelly, et al., Dalton Trans., 2004, 3440.

22. R. Saikat and A. J. Matzger, Angew. Chem. Int. Ed., 2009, 48, 8505. 23. J. A. R. P. Sarma and G. R. Desiraju, Polymorphism and pseudopolymorphism in organic crystals: a Cambridge Structural Database study, in Crystal Engineering: The Design and Application of Functional Solids, Eds. M. J. Zaworotko and K. R. Seddon, Kluwer, Dordrecht, 1999, pp. 325–356. 24. A. Nangia and G. R. Desiraju, Chem. Commun., 1999, 605. 25. M. C. Etter, J. C. Macdonald, Crystallogr., 1990, B46, 256.

and

J. Bernstein,

Acta

26. A. F. Wells, Acta Crystallogr., 1954, 7, 535. ¨ om and K. Larsson, Molecule-Based Materials: The 27. L. Ohrstr¨ Structural Network Approach, Elsevier, Amsterdam, 2005. 28. V. S. S. Kumar, F. C. Pigge, and N. P. Rath, New J. Chem., 2003, 27, 1554. 29. K. M. Fromm, J. L. S. Doimeadios, and A. Y. Robin, Chem. Commun., 2005, 4548. 30. J. L. Sagu´e and K. M. Fromm, Cryst. Growth Des., 2006, 6, 1566. 31. C. R. Ojala, W. H. Ojala, S. Y. Pennamon, and W. B. Gleason, Acta Crystallogr., 1998, C54, 57. 32. L. Yu, G. A. Stephenson, C. A. Mitchell, et al., J. Am. Chem. Soc., 2000, 122, 585. 33. L. R. MacGillivray, J. L. Reid, and J. A. Ripmeester, Chem. Commun., 2001, 1034.

7. Z.-G. Li, G.-H. Wang, H.-Q. Jia, et al., CrystEngComm, 2007, 9, 882.

34. Q.-G. Zhai, R. Ding, S.-N. Li, et al., Inorg. Chim. Acta, 2010, 363, 653.

8. Z.-M. Hao and X.-M. Zhang, Cryst. Growth Des., 2007, 7, 64.

35. S. R. Batten and R. Robson, Angew. Chem. Int. Ed. Engl., 1998, 37, 1460.

9. M. Oh, G. B. Carpenter, and D. A. Sweigart, Angew. Chem., Int. Ed., 2002, 41, 3650.

36. K. Biradha, A. Mondal, B. Moulton, and M. J. Zaworotko, J. Chem. Soc., Dalton Trans., 2000, 3837.

10. R. Horikoshi, T. Mochida, M. Kurihara, and M. Mikuriya, Cryst. Growth Des., 2005, 5, 243.

37. D. J. Duchamp and R. E. Marsh, Acta Crystallogr., 1969, B25, 5.

11. D. Braga, M. Curzi, F. Grepioni, and M. Polito, Chem. Commun., 2005, 2915.

38. F. H. Herbstein, M. Kapon, and G. M. Reisner, Acta Crystallogr., 1985, B41, 348.

12. H. Abourahma, B. Moulton, V. Kravtsov, and Zaworotko, J. Am. Chem. Soc., 2002, 124, 9990.

39. F. H. Herbstein, M. Kapon, and G. M. Reisner, J. Incl. Phenom., 1987, 5, 211.

M. J.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

12

Supramolecular materials chemistry

40. L. Jiang, T.-B. Lu, and X.-L. Feng, Inorg. Chem., 2005, 44, 7056.

42. J.-P. Zhang, Y.-Y. Lin, X.-C. Huang, and X.-M. Chen, Dalton Trans., 2005, 3681.

41. C.-C. Wang, W.-Z. Lin, Commun., 2008, 1299.

43. J.-P. Zhang, Y.-Y. Lin, W.-X. Zhang, and X.-M. Chen, J. Am. Chem. Soc., 2005, 127, 14162.

W.-T. Huang,

et al.,

Chem.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc124

Interpenetration Stuart R. Batten Monash University, Clayton, Melbourne, Australia

1 Introduction 2 Interpenetration of 1D Networks 3 Interpenetration of 2D Networks 4 Interpenetration of 3D Networks 5 Heterointerpenetration 6 Self-Penetration 7 Conclusion References

1

1 3 4 7 10 10 12 13

INTRODUCTION

The network approach to understanding solid-state structures has been an important factor in the development of crystal engineering.1–8 In many cases, however, a structure contains not a single net, but rather multiple nets that interpenetrate, particularly if a single net would result in an open framework structure. Indeed, a driving force for the formation of interpenetrating structures is the desire to maximize the packing efficiency of any solid. Consider, for example, the three related structures of general formula Cu(tcm)L, tcm = tricyanomethanide, C(CN)3 − , L = hexamethylenetetramine, 4,4
-bipyridine, and 1,2-bis(4-pyridyl)ethene.9 All contain simple (4,4) sheets of cross-linked Cu(tcm) chains; yet, each finds a different way to fill up the “holes” created by the inefficient packing provided by the rigid coordination networks (Figure 1). In the case of hexamethylenetetramine, adjoining sheets interdigitate (Figure 1a); that is, pendant C(CN)

arms of tcm anions in one sheet project into the holes of the adjoining sheet and vice versa. In the case of 4,4
bipyridine, pairs of sheets interpenetrate (Figure 1b). Each sheet weaves through the spaces of its partner such that they become intimately entangled, unable to be separated topologically without breaking of bonds. Finally, the sheets in the 1,2-bis(4-pyridyl)ethene structure align to give square channels; these channels are then occupied by intercalated molecules of 1,2-bis(4-pyridyl)ethene and solvent (CH3 CN) (Figure 1c). So, each finds a different way to minimize the unoccupied spaces—interdigitation, interpenetration, or intercalation. Of most interest here, of course, is interpenetration. So what is interpenetration? Interpenetration occurs when two or more polymeric networks are entangled such that they cannot be separated, in a topological sense, without the breaking of bonds. It is perhaps best illustrated by looking at what is not interpenetration. The sheets in Cu(tcm)(hexamethylenetetramine), described above, are not interpenetrating even though they are entangled, as they can, in an imaginary, topological sense, be pulled apart into their separate individual nets without the breaking of bonds, much like the components of a rotaxane that could be separated but not those of a catenane. Similarly, the structure of HgI2 L, L = 2,6-bis(4-pyridinylmethyl)-benzo[1,2-c:4,5c
]dipyrrole-1,3,5,7(2H ,6H )-tetrone, contains 1D chains that are interwoven like threads in a cloth, but might be pulled apart without breakage just like the threads in a real piece of cloth (Figure 2a).10 Finally, the structure of [Mn(dca)2 (H2 O)2 ]·H2 O, dca = dicyanamide, N(CN)2 − , contains 1D chains threaded through the windows of 2D (4,4) sheets (Figure 2b).11 Again, the structure is not interpenetrating as the 1D chains penetrate the windows of the sheets, but are not in turn penetrated themselves by the sheets. Thus, the threads could be pulled out of the channels.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

2

Supramolecular materials chemistry

(a)

(b)

(c)

Figure 1 Three closely related structures, all having 2D (4,4) nets, which show three different methods of maximizing packing efficiency: (a) interdigitation, (b) interpenetration, and (c) intercalation.

(a)

(b)

Figure 2 Two examples of entangled networks that are not interpenetrating: (a) interwoven 1D threads and (b) 1D chains passing through channels created by stacked 2D layers.

Although the occurrence of interpenetration can be a nuisance when seeking to construct highly porous materials, there are situations where the presence of multiple networks is of value. The compound Cu(dcnqi)2 and its derivatives contain seven interpenetrating diamond networks.12–15 These networks interpenetrate such that infinite stacks of dcnqi ligands are created, which are responsible for the metal-like electrical conductivity observed for these compounds. Interpenetration has also led to the creation of structures with completely sealed-off, solvent-filled

cavities16 and has been proposed as a way of structurally stabilizing otherwise fragile porous networks.17 Interpenetrating systems can also produce unusual materials in which the networks may move relative to one another upon the insertion or removal of guest molecules (e.g., solvents or gases). This can lead to changes in magnetic properties18 or new classes of porous materials that show hysteresis in their gas sorption properties.19 Furthermore, various strategies have been proposed for the minimization or elimination of interpenetrating

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Interpenetration

Figure 3

Two topologically different ways for pairs of 2D (4,4) sheets to interpenetrate.

nets. These include incorporation of counterions1 or guest molecules,20 addition of steric bulk to the network components,21–23 or synthesis from high-dilution reactions.24 It is clear that if one has to properly understand and design crystal structures in terms of networks, then an understanding is required not only of the topology of the individual networks but also of the topology of interpenetration. That is, networks can interpenetrate in different ways, and this needs to be described along with the individual network topology. For example, Figure 3 shows two ways in which pairs of simple (4,4) nets can interpenetrate that are superficially similar but nonetheless topologically different. In one motif, the windows of one net have two links of the other net passing above the window’s own links and two below, whereas in the other motif, it is three above and one below (or vice versa). Thus, a brief and necessarily selective overview of the different types of interpenetration follows. For more comprehensive reviews on interpenetration, the reader is directed elsewhere.4, 25–32 An understanding of basic network topology is assumed; again, for discussions of network topology, the reader should consult other sources,4, 7, 8 ¨ including the chapter by Ohrstr¨ om elsewhere in this volume (see The Cambridge Structural Database System and Its Applications in Supramolecular Chemistry and Materials Design, Supramolecular Materials Chemistry).

2

3

INTERPENETRATION OF 1D NETWORKS

There are few examples of 1D networks interpenetrating, perhaps due to the requirement of ring motifs of suitable size to be present so that the catenane-like interactions needed can be formed. Nonetheless, 1D networks can interpenetrate in different ways, leading to entanglements

that can be 1D, 2D, or 3D in nature. Thus, a nomenclature has been developed,4, 25, 26 which is useful for describing such motifs. The nomenclature incorporates three aspects and applies to both 1D and 2D networks. The first aspect is that networks can interpenetrate such that their mean lines of propagation (or mean planes in the case of 2D nets) can either be all parallel or be inclined at two or more angles to each other. The second aspect is that the interpenetration can lead to overall entanglements that are either of the same dimension as the individual nets or of higher dimensions. The latter case is sometimes referred to as polycatenation (as opposed to interpenetration)30 ; however, here we refer to both cases as interpenetration as they both clearly fall under the definition given above. Thus, given these two aspects and the fact that nets of different topologies or dimensions can interpenetrate (see below), interpenetration of 1D and 2D nets can be categorized according to the following nomenclature: mD(/nD) → pD parallel/inclined interpenetration where mD (and nD if the interpenetrating nets are different) is the dimension of the individual interpenetrating networks, and pD is the dimension of the overall entanglement. This nomenclature is best understood by considering each of the possibilities for 1D nets, which are shown in Figure 4. Figure 4(a) shows two 1D nets that are parallel and interpenetrate such that an overall 1D motif is formed. Thus, this is 1D → 1D parallel interpenetration and can be found, for example, in the hydrogen-bonded networks contained in the cocrystal structure of 4,4
sulfonyldiphenol/4,4
-trimethylenedipyridine (1 : 1).33 In Figure 4(b), the individual networks are also parallel, but their mean directions of propagation are now offset rather than coincident, and this leads to an entanglement that is overall 2D in nature (1D → 2D parallel

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

4

Supramolecular materials chemistry

(a)

(b) (c)

(d)

(e)

Figure 4 Different interpenetration modes for 1D nets: (a) 1D → 1D parallel interpenetration; (b) 1D → 2D parallel interpenetration; (c) 1D → 3D parallel interpenetration; (d) 1D → 2D inclined interpenetration; and (e) 1D → 3D inclined interpenetration.

interpenetration). This interpenetration mode can be seen in the structures of [Cu2 L3 (CH3 CN)2 ]X2 ·solv, L = 1, 4bis(4-pyridyl)butadiene and X = PF6 − , BF4 − .34, 35 Finally, 3D entanglements can also be achieved through the interpenetration of 1D networks (Figure 4c), and although architecturally this is harder to achieve at least one example of this has been observed.36 Inclined interpenetration of 1D nets cannot result in 1D entanglement. Figure 4(d) shows 1D → 2D inclined interpenetration as observed for coordination polymers of 1(1-imidazolyl)-4-(imidazol-1-ylmethyl)benzene.37, 38 Similarly, 1D → 3D inclined interpenetration (Figure 4e) has been observed for coordination polymers of another linear dipyridyl ligand, namely, 1,4-bis(4-pyridylmethyl) benzene.39, 40

3

INTERPENETRATION OF 2D NETWORKS

The interpenetration of 2D networks can also occur in either a parallel or an inclined manner, leading to 2D or 3D entanglements. One of the most common interpenetration topologies is 2D → 2D parallel interpenetration. For example, the structure of Ag(tcm) contains hexagonal (6,3) sheets that are corrugated, which allows sheets to form individual 2D layers, which each contain two interpenetrating networks with mean planes that are coplanar (Figure 5).41–43 One polymorph of Ag(TEB)(CF3 SO3 ), TEB = 1,3,5-tris(4ethynylbenzonitrile)benzene, contains layers of six interpenetrating (6,3) sheets.44

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Interpenetration

5

(a)

Figure 6 Unusual interpenetration of “(4,4)” sheets in which the net topology is better described as 6-connected. (b)

Figure 5 2D → 2D parallel interpenetration, as displayed by Ag(tcm), viewed from (a) perpendicular to the layer and (b) sideon to the layer.

The twofold parallel interpenetration of sheets in the structure of [Mn(p-XBP4)3 ](ClO4 )2 , p-XBP4 = N,N
-pphenylenedimethylenebis(pyridin-4-one), presents an interesting case.45 Each sheet would, in isolation, be described topologically as a (4,4) sheet; however, this presents problems when describing the interpenetration. One type of topological link in the structure is actually a double bridge of two ligands connecting a single pair of metal ions, and it is through this loop created that the rods of the partner net pass, leading to the interpenetration (Figure 6). Thus, the network topology is better described as a 6-connected network containing two-membered rings (which are normally reduced to a single topological link). The formation of 3D entanglements from parallel 2D sheets is less common and usually requires individual sheets that are either highly corrugated or have a degree of “thickness.” The former case is well illustrated by the structure of [Cu(bpee)1.5 (PPh3 )] PF6 ·1.5CH2 Cl2 , bpee = 1,4bis(4-pyridyl)ethene.46 It has highly undulating (6,3) sheets, each of which interpenetrates with a sheet “above” it at the “top” of its undulation and with a sheet “below” it at the “bottom” of its undulation (Figure 7). The structure of Cu4 (dca)4 (4,4
-bipy)3 (CH3 CN)2 , 4,4
-bipy = 4,4
bipyridine, is an example of the latter case.47 Each 2D net is effectively two (6,3) sheets cross-linked by 4,4
-bipyridine pillars. Each net is penetrated by four others with mean

Figure 7 2D → 3D parallel interpenetration of highly corrugated sheets.

planes that are parallel but not coplanar—two that pass through the “top” face and two that pass through the “bottom” face. As each net is identical, this creates an overall 3D assembly (Figure 7). A topologically interesting mode of interpenetration that has been reported for 2D nets is that of “Borromean” interpenetration.29 The name derives from the Borromean ring motif, in which three rings are interlinked such that removal of any one ring results in the other two being separated from each other; that is, any two of the rings are not catenated, and it is only the presence of the third which ensures that the three rings are inseparable (Figure 8a). A molecular catenane with this topology has been reported,48 and a number of interpenetrating 2D nets also display this characteristic. The compound [Cu(tmeda)2 {Ag(CN)2 }3 ]ClO4 , tmeda = N,N,N
,N
-tetramethylethylenediamine, can be described in terms of (6,3) sheets showing threefold 2D → 2D parallel interpenetration.49 Removal of any one of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

6

Supramolecular materials chemistry

(a)

(b)

Figure 8

(a) Borromean rings and (b) Borromean interpenetration of 2D sheets.

Figure 9

Three topologically different ways for (4,4) sheets to interpenetrate in an inclined manner.

the nets leaves the other two no longer interpenetrating (Figure 8b). Although the 2D nets in [Ag2 L3 ]X2 , L = N,N
-bis(salicylidene)-1,4-diaminobutane, X = NO3 , ClO4 , have a 3D entanglement, again no two nets are themselves interpenetrating, and only addition of a third locks them together.50 The other common way for 2D nets to interpenetrate is in an inclined manner. Since this must always produce a 3D entanglement, this mode can simply be referred to

as 2D inclined interpenetration. Again, within this broad class, there are different topologies of interpenetration. For example, Figure 9 shows three topologically different ways that (4,4) sheets can interpenetrate in an inclined manner. In the first motif, the windows of one series of nets are penetrated by rods from the other nets (and vice versa); in the second case, the windows contain nodes from the other nets; and the third example is a mixture of both.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Interpenetration Usually, this mode of interpenetration involves two sets of parallel sheets inclined to each other. For example, CdL2 (NO3 )2 , L = 1,6-bis(4-pyridyl)hexane, contains two sets of (4,4) sheets interpenetrating in an inclined manner; each window of each sheet is penetrated by three other nets from the opposing stack.51 The structure of [Ni(azpy)2 (NO3 )2 ]2 [Ni2 (azpy)3 (NO3 )4 ]·4CH2 Cl2 , azpy = trans-4,4
-azobis(pyridine), contains a stack of (6,3) nets interpenetrating a stack of (4,4) nets (2D/2D inclined interpenetration).52 Each window in the (6,3) sheets has two (4,4) nets passing through it, whereas those of the (4,4) sheets have only one (6,3) net passing through. The remarkable structure of Co2 (azpy)3 (NO3 )4 ·(CH3 )2 CO·3H2 O, however, contains four different stacks of (6,3) sheets all inclined to each other.53 A more complicated example of 2D interpenetration is observed in the structure of [AgL2 ]SbF6 , L = 3-cyanophenyl 4-cyanobenzoate.54 This compound contains layers of doubly interpenetrating (4,4) sheets (i.e., parallel interpenetration), which then interpenetrate each other in an inclined manner, giving an overall entanglement that could be described as 2D → 2D parallel → 3D inclined interpenetration.

4

7

(a)

INTERPENETRATION OF 3D NETWORKS

For 3D nets, the parallel/inclined interpenetration notation discussed above obviously does not apply, but there is nonetheless a rich variety of possible interpenetration topologies they can exhibit. An excellent illustration of this is in the interpenetration of (10,3)-a networks. This network is inherently chiral due to the presence of fourfold helices, which are all of the same hand within an individual net. This leads to the possibility of chiral or racemic interpenetration; that is, the interpenetration can be such that all the nets are of the same hand, or there are equal numbers of either hand net (in theory, it is also possible to have another interpenetration topology where unequal numbers of each hand net interpenetrate, although this is yet to be observed). The two possibilities for two interpenetrating networks are shown in Figure 10. The structure of Ni3 (btc)2 (pyridine)6 (eg)6 ·∼3eg·∼4H2 O, btc = benzene1,3,5-tricarboxylate and eg = ethylene glycol, is an example of the former—the four interpenetrating networks are all of the same handedness.55 The compound Ag2 L3 (SbF6 )2 , L = 2,3-dimethylpyrazine, is an example of the latter—it contains two interpenetrating nets, one of each hand.56 The structure of Zn(tpt)2/3 (SiF6 )(H2 O)2 (CH3 OH), tpt = 2,4,6tri(4-pyridyl)-1,3,5-triazine, contains eight interpenetrating

(b)

Figure 10 Interpenetration of two (10,3)-a nets that (a) are of opposite handedness and (b) are of the same handedness. The chirality of each net can be seen by examining the fourfold helices going into the page.

networks, four of each hand.57 Even more remarkably, a hydrogen-bonded structure has been reported to contain nine nets of each hand, giving a total of 18 interpenetrating 3D networks.58 Interpenetration has also been observed for other achiral types of (10,3) nets—the structures of Zn3 (tpt)2 X6 , X = Cl, I, contain two interpenetrating (10,3)-b nets,25, 59 while [Cu(4,4
-bipy)1.5 ]NO3 ·1.25H2 O contains six such nets interpenetrating.60 CoL1.5 (NO3 )2 ·H2 O, L = 1,4-bis(3pyridyl)-2,3-diazabuta-1,3-diene, contains four (10,3)-d nets.61 Similar to (10,3)-a, the (8,3)-a net is chiral, and [Cu2 (tae)(4,4
-bipy)2 ](NO3 )2 , tae = dibasic tetraacetylethane, contains four such nets of the same hand interpenetrating.62 By far and away, the most commonly observed form of interpenetration is the formation of multiple diamond networks. The classic examples are the structures of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

8

Supramolecular materials chemistry

Figure 11

Interpenetration of 10 diamond networks.

M(CN)2 , M = Zn, Cd, which are perhaps the earliest known examples of interpenetration in general.1, 63, 64 They contain two interpenetrating networks of tetrahedral metal ions linked by cyanide bridges. The structure of [Ag(ddn)2 ]NO3 , ddn = 1,12-dodecanedinitrile, by contrast, contains 10 interpenetrating networks (Figure 11).65 In nearly all examples of interpenetrating diamond nets, the interpenetration topology is the same, despite the varying numbers of nets. Occasionally, however, unusual cases of “abnormal” interpenetration occur even for this simple, highly symmetric topology. One such case occurs in the structure of β-Cu(dca)(bpee).47 It is best illustrated by considering one net (all nets are equivalent), and the way the other nets (there are five interpenetrating nets in total) interact with it. A defining feature of the diamond net is the adamantanelike cavity it contains, which is composed of four six-membered windows. If one arbitrarily chooses one adamantane cavity of one net in β-Cu(dca)(bpee) and then chooses one of the six-membered windows of that cavity, unique adamantane cavities of the other four nets can

(a)

be defined by the rods that pass through the nominated window of the first net. For two of the interpenetrating nets, the relationship to the first net is the same as seen in the “normal” mode of interpenetration; that is, the cavity of the first net contains a single node from the other, and each window of the adamantane cavity of the first net is penetrated by a rod from the included node of the second net (Figure 12a). For the third and fourth nets, however, the relationship is different. For example, for the interpenetrating net shown in Figure 12(b), one window is not penetrated at all by a rod from the equivalent node, while another window has two rods passing through it (emanating from two different nodes). As all nets are equivalent, this motif is repeated throughout the structure, with the nets swapping roles between “normal” and “abnormal” interpenetration of their adamantane cavities. This is just one illustration of “abnormal” interpenetration of diamond networks; a small number of other examples, with different interpenetration topologies to β-Cu(dca)(bpee), have also been reported.65, 66 Interpenetration of other 4-connected topologies has also been

(b)

Figure 12 Pairs of adamantane cavities from separate networks in the structure of β-Cu(dca)(bpee) showing (a) “normal” interpenetration and (b) “abnormal” interpenetration. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Interpenetration observed. Six interpenetrating, chiral quartz nets have been reported in the structures of M[Au(CN)2 ]3 , M = Zn, Co; all interpenetrating nets are of the same hand.67, 68 The coordination polymers [Ag(sebn)2 ]XF6 , sebn = sebaconitrile, X = P, As, contain four SrSi2 nets.69 These nets all have tetrahedral nodes; for square planar nodes, two CdSO4 nets are found in [Cu(bpee)(H2 O)(SO4 )],70 and three NbO nets interpenetrate in FeL2 [Ag(CN)2 ]2 ·2/3H2 O, L = 3-cyanopyridine.71 The most common 3D network topology after diamond is that of the 6-connected α-Po net. As for diamond, there is a “normal” mode of interpenetration that this net displays in almost every case. Indeed, the normal mode is similar to that of diamond in many respects; for two nets, it involves a node of one net lying inside the distinctive cavity of the second net (the adamantane cavity for diamond and the cubic cavity for α-Po) and projecting a rod through each of the cavity windows (four rods through the four adamantane windows for the 4-connected diamond net and six rods through the six faces of the cube for the 6-connected α-Po net). The special relationship for interpenetration in these two nets has been noted previously; they are two of only four nets known to show this “self-dual” relationship.72

(a)

9

For the α-Po net, this leads to each cubic cavity being catenated with eight others of each interpenetrating net, as illustrated by α-M(dca)2 (pyrazine), M = Mn, Fe, Co, Ni, Cu, Zn, in Figure 13(a).73, 74 For the structure of [Mn(bpea)(H2 O)4 ](ClO4 )2 (bpea)4 , bpea = 1,4-bis(4pyridyl)ethane, however, an “abnormal” interpenetration topology is observed, and each cube is catenated with 10 cubic cavities of the other interpenetrating net (and vice versa) (Figure 13b).75 Thus far, we have considered only uninodal nets; however, interpenetration of multinodal nets does, of course, occur. The PtS net has two topologically different nodes, even though both are 4-connecting (geometrically one is tetrahedral and one is square planar), and two such nets interpenetrate in the structure of Cu(tcp)CuBF4 ·17C6 H5 NO2 , Cu(tcp) = 5,10,15,20-tetrakis(4-cyanophenyl)-21H,23Hporphine copper(II).76 The interpenetration in this structure is unusual in that the nets interpenetrate in an asymmetric manner; that is, one net does not lie in the middle of the cavities of the other, but rather lies closely adjacent to the other net, leaving larger cavities than would be otherwise present (Figure 14). Internetwork interactions are likely important here. Two 3,4-connected boracite nets

(b)

Figure 13 (a) “Normal” interpenetration of α-Po networks in the structures of α-M(dca)2 (pyrazine)—each cubic cavity is catenated with eight others. (b) “Abnormal” interpenetration of α-Po networks in the structure of [Mn(bpea)(H2 O)4 ](ClO4 )2 (bpea)4 —the cube in the central net is catenated with 10 from the surrounding net.

Figure 14

Asymmetric interpenetration of two porphyrin containing 3D PtS networks.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

10

Supramolecular materials chemistry

interpenetrate in Cu3 (tpt)4 (ClO4 )3 ·solv,77 whereas the two 3,4-connected nets in Cu3 L2 (H2 O)3 , L = 4,4
,4

-s-triazine2,4,6-triyltribenzoate or s-heptazine tribenzoate, have the twisted boracite topology.78, 79 Two 3,5-connected nets are found in the structures of Ag(tcm)L, L = pyrazine, dabco, 4,4
-bipyridine,42, 43 and two interpenetrating 3,6-connected pyr nets are found in Zn4 O(TCA)2 ·3DMF·G, TCA = 4,4
,4

-tricarboxytriphenylamine, G = 3H2 O, EtOH.80, 81

5

HETEROINTERPENETRATION

Another limitation of the discussion above (with one exception) has been the consideration of interpenetration by nets that are chemically, topologically, and even dimensionally different. While it is true that in most examples, the interpenetrating nets are all exactly the same (and often related crystallographically), there are a number of fascinating examples where this does not apply. The compound K2 [PdSe10 ] contains two interpenetrating diamond nets bridged by Sex 2− chains.82 In one net x = 4, while in the other x = 6. The two different ligands adopt conformations that result in corresponding Pd· · ·Pd separations in each net being exactly the same, as they have to be for interpenetration to occur (Figure 15). Interpenetration of nets of different topologies is rare. An example of interpenetration of 2D (6,3) and (4,4) sheets was given earlier.52 The inorganic material Bi3 GaSb2 O11 has been described in terms of three 3D nets, one with NbO topology and two with diamond topology (Figure 16).83 Interpenetration of nets of different dimensionalities is also very unusual. The structure of [Cu5 (bpp)8 (SO4 )4 (EtOH) (H2 O)5 ](SO4 )·EtOH·25.5H2 O, bpp = 1,3-bis(4-pyridyl) propane, contains both 1D chains and 2D (4,4) sheets.84 The chains interpenetrate with the sheets in an inclined manner (Figure 17a); that is, 1D/2D inclined interpenetration, which necessarily gives a 3D entanglement. 1D/3D interpenetration has been observed in the structure of [Co(bix)2 (H2 O)2 ](SO4 )·7H2 O, bix = 1,4-bis(imidazol-1ylmethyl)benzene (Figure 17b),85 and [Co(mppe)2 (NCS)2 ]· 2[Co(mppe)2 (NCS)2 ]·5CH3 OH, mppe = 1-methyl-1
-(4pyridyl)-2-(4-pyrimidyl)ethylene, displays 2D/3D interpenetration.86

Figure 15 Interpenetration of diamondoid [Pd(Se4 )2 ]2− and [Pd(Se6 )2 ]2− nets in K2 [PdSe10 ].

Figure 16

6

Interpenetration of two diamond and one NbO net.

SELF-PENETRATION

Interpenetrating networks have an obvious relationship to molecular catenanes; indeed, they can be thought of simply as catenated polymers. Similarly, entangled networks with rotaxane-like relationships have also been reported.87, 88

Another fascinating discrete supramolecular architecture that has attracted considerable attention is the molecular knot. The polymeric version of a knot is a self-penetrating network 26, 89, 90 ; indeed, this is sometimes referred to as polyknotting.30

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Interpenetration

(a)

Figure 17

11

(b)

(a) 1D/2D inclined interpenetration and (b) 1D/3D interpenetration.

(a)

(b)

(c)

Figure 18 (a) The single rutile network formed by α-M(dca)2 . (b) The two interpenetrating rutile nets formed by M(tcm)2 . (c) The single, self-penetrating network formed by M(dca)(tcm), with a penetration of a rod through a six-membered shortest circuit highlighted.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

12

Supramolecular materials chemistry

(a)

The octahedral metals are bridged by 3-connecting dca anions; the dca anions have a central nitrogen atom, which is connected to one metal by a direct M–N bond and to two others by longer N–CN–M links. Only a single network is formed. The tcm anion can also act as a 3-connecting ligand; except in this case, the central carbon atom is connected to each of the three metal atoms by three long C–CN–M links. Again, rutile networks are formed for the binary M(tcm)2 compounds, but this time, the larger ligand allows for the formation of two interpenetrating networks (Figure 18b).94, 95 Notably, the rods of one net pass through the six-membered smallest rings of the other. A self-penetrating network is produced when both ligands are used together to make M(dca)(tcm) (Figure 18c).96 This combination of smaller and larger ligands results in a network that is halfway between the interpenetrating and noninterpenetrating parents. The topology is different from, but closely related to, rutile (both nets have the same Schl¨afli symbol), and in the self-penetrating net, there are again rods passing through six-membered rings. In this case, however, the rods and rings are part of the same network. There are a number of uninodal nets that are also selfpenetrating, notably the (8,4) net97 (Figure 19a) and the (12,3) net90 (Figure 19b). A compound containing two interpenetrating self-penetrating (12,3) nets has even been reported.98

7

(b)

Figure 19 The uninodal self-penetrating (a) (12,3) and (b) (8,4) nets. Catenane-like motifs are highlighted in each.

Self-penetrating networks are single nets, which nonetheless contain rotaxane-like motifs, similar to those seen between interpenetrating networks. Since such motifs can be defined in any 3D net if suitably large rods and rings are defined, a necessary restriction is that rods must penetrate at least some of the smallest topological rings in the structure. Note that all the “smallest rings” in a net are not the same size—a smallest ring is the shortest circuit containing a chosen set of links from a node; different choices of links may result in different sized smallest circuits. The relationship between interpenetration and selfpenetration is best illustrated by considering a series of closely related structures. In the compounds, α-M(dca)2 , 3,6-connected rutile networks are formed (Figure 18a).91–93

CONCLUSION

The use of networks to describe and design crystal structures inevitably leads to the possibility of interpenetration of multiple networks. In such cases it is essential that the topology of interpenetration, as well as the individual network topology, is analysed and understood in detail. For 1D and 2D networks, interpenetration can occur either in a parallel or inclined fashion, leading to overall entanglements of the same or higher dimensionality. Such systems can be described in terms of a “mD(/nD) → pD parallel/inclined interpenetration” notation. Even highly symmetrical 3D nets such as diamond or α-Po can show, on rare occasions, different modes of interpenetration. While in most cases interpenetration occurs between identical networks, interpenetration of a heterogeneous nature is well known. This can involve interpenetration of chiral networks of opposite handedness, interpenetration of topologically identical but chemically different networks, or interpenetration of topologically (and even dimensionally) different networks. Finally, the related phenomenon of self-penetration occurs when the smallest circuits of a net are penetrated

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Interpenetration by rods from the same network, which leads to many fascinating entanglement topologies.

13

25. S. R. Batten and R. Robson, Angew. Chem. Int. Ed. Engl., 1998, 37, 1460. 26. S. R. Batten, CrystEngComm, 2001, 3, 67.

REFERENCES 1. B. F. Hoskins and R. Robson, J. Am. Chem. Soc., 1990, 112, 1546. 2. R. Robson, Dalton Trans., 2000, 3735. 3. R. Robson, Dalton Trans., 2008, 5113.

27. V. A. Blatov, L. Carlucci, G. Ciani, and D. M. Proserpio, CrystEngComm, 2004, 6, 377. 28. I. A. Baburin, V. A. Blatov, L. Carlucci, et al., J. Solid State Chem., 2005, 178, 2471. 29. L. Carlucci, G. Ciani, and D. M. Proserpio, CrystEngComm, 2003, 5, 269. 30. L. Carlucci, G. Ciani, and D. M. Proserpio, Coord. Chem. Rev., 2003, 246, 247.

4. S. R. Batten, S. M. Neville, and D. R. Turner, Coordination Polymers: Design, Analysis and Application, Royal Society of Chemistry, Cambridge, 2009.

31. I. A. Baburin, V. A. Blatov, L. Carlucci, et al., Cryst. Growth Des., 2008, 8, 519.

5. A. F. Wells, Three-dimensional Nets and Polyhedra, WileyInterscience, New York, 1977.

32. I. A. Baburin, V. A. Blatov, L. Carlucci, et al., CrystEngComm, 2008, 10, 1822.

6. A. F. Wells, Further Studies of Three-dimensional Nets, ACA Monograph No. 8 , American Crystallographic Association, Knoxville, TN, 1979.

33. G. Ferguson, C. Glidewell, R. M. Gregson, and E. S. Lavender, Acta Crystallogr., Sect. B , 1999, 55, 573.

7. M. O’Keeffe, M. Eddaoudi, H. Li, et al., J. Solid State Chem., 2000, 152, 3. ¨ 8. L. Ohrstr¨ om and K. Larsson, Molecule-Based Materials, The Structural Network Approach, Elsevier, Amsterdam, The Netherlands, 2005, p. 171.

34. A. J. Blake, N. R. Champness, A. Khlobystov, et al., Chem. Commun., 1997, 2027. 35. M. Maekawa, H. Konaka, Y. Suenaga, et al., J. Chem. Soc., Dalton Trans., 2000, 4160. 36. Z. Shi, X. Gu, J. Peng, et al., Eur. J. Inorg. Chem., 2006, 385.

9. S. R. Batten, B. F. Hoskins, and R. Robson, Chem.—Eur. J., 2000, 6, 156.

37. H.-F. Zhu, W. Zhao, T. Okamura, et al., New J. Chem., 2004, 28, 1010.

10. Y.-H. Li, C.-Y. Su, A. M. Goforth, et al., Chem. Commun., 2003, 1630.

38. H. F. Zhu, J. Fan, T. Okamura, et al., Cryst. Growth Des., 2005, 5, 289.

11. S. R. Batten, P. Jensen, C. J. Kepert, et al., J. Chem. Soc., Dalton Trans., 1999, 2987.

39. M. Fujita, Y. J. Kwon, O. Sasaki, et al., J. Am. Chem. Soc., 1995, 117, 7287.

12. O. Ermer, Adv. Mater., 1991, 3, 608.

40. M. Fujita, O. Sasaki, K.-Y. Watanabe, et al., New J. Chem., 1998, 22, 189.

13. K. Sinzger, S. Hunig, M. Jopp, et al., J. Am. Chem. Soc., 1993, 115, 7696.

41. J. Konnert and D. Britton, Inorg. Chem., 1966, 5, 1193.

14. A. Aumuller, P. Erk, G. Klebe, et al., Angew. Chem. Int. Ed. Engl., 1986, 25, 740.

42. S. R. Batten, B. F. Hoskins, and R. Robson, New J. Chem., 1998, 22, 173.

15. R. Kato, H. Kobayashi, and A. Kobayashi, J. Am. Chem. Soc., 1989, 111, 5224.

43. B. F. Abrahams, S. R. Batten, B. F. Hoskins, and R. Robson, Inorg. Chem., 2003, 42, 2654.

16. S. R. Batten, B. F. Hoskins, and R. Robson, J. Am. Chem. Soc., 1995, 117, 5385.

44. D. Venkataraman, S. Lee, J. S. Moore, et al., Chem. Mater., 1996, 8, 2030.

17. B. Chen, M. Eddaoudi, S. T. Hyde, et al., Science, 2001, 291, 1021.

45. D. M. L. Goodgame, S. Menzer, A. M. Smith, and D. J. Williams, Angew. Chem. Int. Ed. Engl., 1995, 34, 574.

18. G. J. Halder, C. J. Kepert, B. Moubaraki, et al., Science, 2002, 298, 1762.

46. J. M. Knaust, S. Lopez, and S. W. Keller, Inorg. Chim. Acta, 2001, 324, 81.

19. R. Kitaura, K. Seki, G. Akiyama, and S. Kitagawa, Angew. Chem. Int. Ed., 2003, 42, 428.

47. S. R. Batten, A. R. Harris, P. Jensen, et al., J. Chem. Soc., Dalton Trans., 2000, 3829.

20. T. Kitazawa, S. Nishikiori, R. Kuroda, and T. Iwamoto, Chem. Lett., 1988, 1729.

48. K. S. Chichak, S. J. Cantrill, A. R. Pease, et al., Science, 2004, 304, 1308.

21. O. Ermer and L. Lindenberg, Helv. Chim. Acta, 1991, 74, 825.

49. D. B. Leznoff, B.-Y. Xue, R. J. Batchelor, et al., Inorg. Chem., 2001, 40, 6026.

22. O. K. Farha, C. D. Malliakas, M. G. Kanatzidis, and J. T. Hupp, J. Am. Chem. Soc., 2010, 132, 950.

50. M. L. Tong, X.-M. Chen, B.-H. Ye, and L.-N. Ji, Angew. Chem. Int. Ed., 1999, 38, 2237.

23. R. K. Deshpande, J. L. Minnaar, and S. G. Telfer, Angew. Chem. Int. Ed., 2010, 49, 4598.

51. M. J. Plater, M. R. St, J. Foreman, et al., J. Chem. Soc., Dalton Trans., 2000, 3065.

24. M. Eddaoudi, J. Kim, N. Rosi, et al., Science, 2002, 295, 469.

52. L. Carlucci, G. Ciani, and D. M. Proserpio, New J. Chem., 1998, 22, 1319.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

14

Supramolecular materials chemistry

53. M. Kondo, M. Shimamura, S. Noro, et al., Chem. Mater., 2000, 12, 1288.

77. B. F. Abrahams, S. R. Batten, H. Hamit, et al., Angew. Chem. Int. Ed. Engl., 1996, 35, 1690.

54. N. Moliner, C. Munoz, S. Letard, et al., Inorg. Chem., 2000, 39, 5390.

78. D. Sun, S. Ma, Y. Ke, et al., J. Am. Chem. Soc., 2006, 128, 3896.

55. C. J. Kepert and M. J. Rosseinsky, Chem. Commun., 1998, 31.

79. S. Ma, D. Sun, M. Ambrogio, et al., J. Am. Chem. Soc., 2007, 129, 1858.

56. L. Carlucci, G. Ciani, D. M. Proserpio, and A. Sironi, Chem. Commun., 1996, 1393.

80. H. K. Chae, J. Kim, O. D. Friedrichs, et al., Angew. Chem. Int. Ed., 2003, 42, 3907.

57. B. F. Abrahams, S. R. Batten, H. Hamit, et al., Chem. Commun., 1996, 1313.

81. E. Y. Lee, S. Y. Jang, and M. P. Suh, J. Am. Chem. Soc., 2005, 127, 6374.

58. T. R. Shattock, P. Vishweshwar, Z. Wang, and Zaworotko, Cryst. Growth Des., 2005, 5, 2046.

M. J.

82. K. W. Kim and M. G. Kanatzidis, J. Am. Chem. Soc., 1992, 114, 4878.

59. K. Biradha and M. Fujita, Angew. Chem. Int. Ed., 2002, 41, 3392.

83. A. W. Sleight and R. J. Bouchard, Inorg. Chem., 1973, 12, 2314.

60. O. M. Yaghi and H. Li, J. Am. Chem. Soc., 1995, 117, 10401.

84. L. Carlucci, G. Ciani, M. Moret, et al., Angew. Chem. Int. Ed., 2000, 39, 1506.

61. Y.-B. Dong, M. D. Smith, and H.-C. zur Loye, Inorg. Chem., 2000, 39, 4927. 62. B. S. Luisi, V. Ch. Kravtsov, and B. D. Moulton, Cryst. Growth Des., 2006, 6, 2207.

85. L. Carlucci, G. Ciani, and D. M. Proserpio, Chem. Commun., 2004, 380.

63. H. S. Zhdanov, C. R. Acad. Sci. URSS , 1941, 31, 352.

86. D. M. Shin, I. S. Lee, Y. K. Chung, and M. S. Lah, Chem. Commun., 2003, 1036.

64. E. Shugam and H. S. Zhdanov, Acta Physiochim. URSS , 1945, 20, 247.

87. B. F. Hoskins, R. Robson, and D. A. Slizys, J. Am. Chem. Soc., 1997, 119, 2952.

65. L. Carlucci, G. Ciani, D. M. Proserpio, and S. Rizzato, Chem.—Eur. J., 2002, 8, 1520.

88. C. S. A. Fraser, M. C. Jennings, Chem. Commun., 2001, 1310.

66. Y.-H. Liu, H.-C. Wu, H.-M. Lin, et al., Chem. Commun., 2003, 60.

89. S. R. Batten and R. Robson, Catenane and rotaxane motifs in interpenetrating and self-penetrating coordination polymers, in Molecular Catenanes, Rotaxanes and Knots, a Journey Through the World of Molecular Topology, eds. J.-P. Sauvage and C. Dietrich-Buchecker, Wiley-VCH Verlag GmbH, Weinheim, 1999, pp. 77–106.

67. B. F. Hoskins, R. Robson, and N. V. Y. Scarlett, Angew. Chem. Int. Ed., 1995, 34, 1203. 68. S. C. Abrahams, L. E. Zyontz, and J. L. Bernstein, J. Chem. Phys., 1982, 76, 5458. 69. L. Carlucci, G. Ciani, P. Macchi, et al., Chem.—Eur. J., 1999, 5, 237. 70. D. Hagrman, R. P. Hammond, R. Haushalter, and J. Zubieta, Chem. Mater., 1998, 10, 2091.

and

R. J. Puddephatt,

90. B. F. Abrahams, S. R. Batten, M. J. Grannas, et al., Angew. Chem. Int. Ed., 1999, 38, 1475. 91. S. R. Batten, P. Jensen, B. Moubaraki, et al., Chem. Commun., 1998, 439. 92. M. Kurmoo and C. J. Kepert, New J. Chem., 1998, 22, 1515.

71. A. Galet, V. Niel, M. C. Munoz, and J. A. Real, J. Am. Chem. Soc., 2003, 125, 14224.

93. J. L. Manson, C. R. Kmety, Q.-Z. Huang, et al., Chem. Mater., 1998, 10, 2552.

72. O. Delgado-Friedrichs, M. D. Foster, M. O’Keeffe, et al., J. Solid State Chem., 2005, 178, 2533.

94. S. R. Batten, B. F. Hoskins, and R. Robson, J. Chem. Soc., Chem. Commun., 1991, 445.

73. P. Jensen, S. R. Batten, G. D. Fallon, et al., J. Solid State Chem., 1999, 145, 387.

95. S. R. Batten, B. F. Hoskins, B. Moubaraki, et al., J. Chem. Soc., Dalton Trans., 1999, 2977.

74. P. Jensen, S. R. Batten, B. Moubaraki, and K. S. Murray, J. Solid State Chem., 2001, 159, 352.

96. P. Jensen, D. J. Price, S. R. Batten, et al., Chem.—Eur. J., 2000, 6, 3186.

75. C. S. Hong, S.-K. Son, Y. S. Lee, et al., Inorg. Chem., 1999, 38, 5602.

97. M.-L. Tong, X.-M. Chen, and S. R. Batten, J. Am. Chem. Soc., 2003, 125, 16170.

76. B. F. Abrahams, B. F. Hoskins, D. M. Michail, and R. Robson, Nature, 1994, 369, 727.

98. T. Schareina, C. Schick, B. F. Abrahams, and R. Kempe, Z. Anorg. Allg. Chem., 2001, 627, 1711.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc125

Templated [2 + 2] Photodimerizations in the Solid State Elizabeth Elacqua, Rebecca C. Laird, and Leonard R. MacGillivray University of Iowa, Iowa City, IA, USA

1 Engineering Reactivity in the Solid State 2 Approaches to Control the [2+2] Photodimerization 3 Emergence of Supramolecular Chemistry 4 Templated [2+2] Photodimerizations 5 Other Cycloadditions 6 Conclusion References

1

1 1 2 2 11 11 12

ENGINEERING REACTIVITY IN THE SOLID STATE

Crystal engineering is an area of supramolecular chemistry that relies on the use of intermolecular forces to achieve solid-state frameworks with desired physical and/or chemical properties.1 The field is undergoing rapid development owing to an awareness that such an approach to synthesis can afford functional solids with unique properties (e.g., reactivity, optical properties, magnetism). In this context, chemists have worked to design olefins that will crystallize, with certainty, in geometries suitable for intermolecular [2+2] photodimerization reactions. The reaction results in the formation of two carbon–carbon single (C–C) bonds in a rigid, yet flexible, environment that can afford

products, otherwise, unattainable in solution.2 That the reaction occurs in a solvent-free environment also means that the approach can provide an entry to the emerging area of green chemistry.3 It is with these ideas in mind that we describe here supramolecular approaches to control [2+2] photodimerization reactions in the solid state based on the use of templates. We first describe the use of small organic molecules that act as templates and then move to metal–organic complexes. We end with a treatment of other cycloadditions (e.g., [4+4] cycloaddition, Diels–Alder) that may also be controlled using the template approach.

2

APPROACHES TO CONTROL THE [2+2] PHOTODIMERIZATION

Pioneering work of Schmidt involving cinnamic acids resulted in geometry criteria for a [2+2] photodimerization to occur in the solid state.4 In particular, it was determined that the cycloaddition reaction is topochemically controlled in solids with minimal molecular movement. The reaction is generally dictated by parallel alignment and overlap ˚ 4 of olefinic bonds with a separation distance of < 4.2 A. Schmidt also revealed that the geometry and positioning of olefins in the solid state are highly sensitive to subtle changes to molecular structure. More specifically, it was demonstrated how the organization of molecules in the solid state can be “unpredictably” influenced by functional groups or substituents.4 Thus, homologous series’ of cinnamic acids, in contrast to the liquid phase, did not exhibit homologous reactivities in solids. The lack of reaction homology has meant that it has remained difficult

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc126

2

Supramolecular materials chemistry

for chemists to synthesize molecules in solids with similar degrees of synthetic “freedoms” that are realized in solution.

4

4.1

3

EMERGENCE OF SUPRAMOLECULAR CHEMISTRY

Since the work of Schmidt, several advances have been made to circumvent problems associated with crystal packing in attempts to reliably facilitate [2+2] photodimerizations in solids. In particular, intermolecular forces from the field of supramolecular chemistry have been exploited to steer the packing of olefins into geometries for reaction. Covalently attached functional groups that employ directing effects based on π–π stacking, charge-transfer complexation, and hydrogen bonding were initially studied to control packing.1 Despite significant successes, however, somewhat limited control of reactivity, in the context of a general organic synthesis, was achieved since the functional groups lacked an ability to effectively compete with demands of dense packing. Whereas initial work to control [2+2] photodimerizations in the solid state involved the use of substituents, chemists have recently turned to auxiliaries (Scheme 1). Organic molecules and metal–organic complexes in the form of templates5 have been used to assemble and organize olefins via hydrogen bonding and metal coordination, respectively, for reaction. The templates preorganize olefins within supramolecular assemblies of molecules with structures largely independent of crystal packing. As a result, the templates have provided a means to construct relatively complex molecules in solids (e.g., ladderanes and paracyclophanes) with high levels of stereocontrol and in up to quantitative yields.6

Scheme 1

Hydrogen bonds

The first small-molecule templates to control the [2+2] photodimerization in the solid state involved ditopic hydrogen-bond donors. The templates assembled olefins lined with pyridyl groups.5, 6 The solid-state reactivity was, consequently, achieved within a cocrystal or a “structurally homogeneous crystalline material that contains two or more neutral building blocks present in definite stoichiometric amounts.”7 More specifically, MacGillivray and coworkers8 demonstrated that photostable trans-1,2-bis(4-pyridyl)ethylene (4,4
-bpe), on cocrystallization with 1,3-dihydroxybenzene or resorcinol (res), forms a discrete four-component molecular assembly sustained by four O–H· · ·N hydrogen bonds. The C=C bonds of 4,4
-bpe were aligned within the 0D ˚ (Figure 1). On assembly parallel and separated by 3.65 A UV irradiation, the olefins reacted to afford rctt-tetrakis-(4pyridyl)cyclobutane (rctt-4,4
-tpcb) stereospecifically and in quantitative yield. A template based on a crown ether was subsequently employed to organize C=C bonds within a 0D complex in a solid for a [2+2] photoreaction. Specifically, GarciaGaribay, Stoddart, and coworkers demonstrated that reaction of a bisparaphenylene-34-crown-10 (bpp-34-crown-10) with a bis(dialkylammonium)-substituted stilbene (ammstilb) produced the four-component complex 2(bpp-34crown-10)·2(amm-stilb) held together by eight N+ –H· · ·O hydrogen bonds (Figure 2). The cavity of the crown ether was filled with the ends of the two reactants, with the ˚ 9 olefinic bonds being parallel and separated by ∼4.20 A. UV irradiation resulted in a dimerization of amm-stilb to give a single diastereomer in ∼80% yield.

hn Crystal

Self-assembly

Template

TEMPLATED [2+2] PHOTODIMERIZATIONS

Olefin

Product

Schematic of strategy to control reactivity in the solid state using a template.

hn Solid state

Figure 1

Synthesis of 4,4
-tpcb using res as an organic template.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc126

Templated [2 + 2] photodimerizations in the solid state

a 1D hydrogen-bonded tape wherein the C=C bonds of 4,4
˚ (Figure 3b).11 UV bpe were parallel and separated by 3.8 A
irradiation produced 4,4 -bpe in 90% yield. In addition to symmetrical hydrogen-bond donors, MacGillivray and coworkers introduced unsymmetrical Rebek’s imide (Reb-im) as a template (Scheme 2).12 Cocrystallization of Reb-im with 4,4
-bpe produced a discrete four-component assembly with olefins parallel and ˚ The assembly was sustained by separated by 3.78 A. a combination of O–H· · ·N and N–H· · ·N forces. The olefins reacted in a rare single-crystal-to-single-crystal (SCSC) transformation to give 4,4
-tpcb stereospecifically in 100% yield. Monopyridines have also been assembled to undergo [2+2] photodimerization in the solid state (Scheme 3). Cocrystallization of 4-chlorostilbazole (4-Cl stilbz) with either res or 4-ethyl-res afforded discrete three-component assemblies sustained by two O–H· · ·N hydrogen bonds.13 The olefin adopted head-to-head geometries and participated in Cl· · ·Cl forces. The C=C bonds reacted to give the head-to-head cyclobutane rctt-1,2-bis(4-pyridyl)3,4-bis(p-chlorophenyl)cyclobutane (4-Cl dpcb) in quantitative yield. Recently, Ramamurthy and coworkers have employed thiourea as a template to assemble substituted stilbazoles (e.g., 4-CN, 4-F, 4-Br stilbz) to undergo stereospecific photodimerizations in the solid state (Figure 4).14 Cocrystallization of 4-CN stilbz afforded infinite 1D tapes

Figure 2 Capped-stick model of the photoactive assembly 2(bpp-34-crown-10)·2(amm-stilb).

MacGillivray and coworkers also described a dicarboxylic acid that acted as a template (Figure 3).10 Cocrystallization of 1,8-naphthalenedicarboxylic acid (1,8-nda) with either 4,4
-bpe or 2,2
-bpe produced four-component assemblies sustained by four O–H· · ·N hydrogen bonds (Figure 3a). The olefins were parallel and separated by ˚ respectively, which rena distance of 3.73 and 3.91 A,

dered 4,4 -bpe and 2,2 -bpe photoactive. UV irradiation of each solid produced the corresponding rctt-cyclobutanes in quantitative yield. In a related work, Jones demonstrated the ability of a triacid to act as a template. Specifically, cocrystallization of tricarballylic acid (tca) with 4,4
-bpe afforded

(a)

Figure 3

Scheme 2

(b)

Cocrystals of photoactive (a) 2(nda)·2(2,2
-bpe) and (b) (tca)·2(4,4
-bpe). O NH O O OH

N

N

N

N

HO O O HN O

O NH O O OH

hn Solid

N

N

N

N

HO O O HN O

Synthesis of 4,4
-tpcb using Reb-im as a heteroditopic template. OH

N

Cl

OH

N

Cl

OH

N

Cl

hn Solid state

R

OH

N

Cl

R

R = H, CH2CH3

Scheme 3

3

Synthesis of 4-Cl dpcb using res or 4-Et res as a template.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc126

4

Supramolecular materials chemistry

N N

Scheme 4

4.3

Figure 4

1D tapes based on crystalline (thiourea)·(4-F stilbz).

sustained by N–H· · ·N hydrogen bonds. The olefins were preorganized parallel and in a head-to-head geometry. UV irradiation afforded rctt-1,2-bis(4-pyridyl)-3,4-bis(pcyanophenyl)cyclobutane in quantitative yield. Reactive assemblies that afforded head-to-tail dimers were also achieved.

4.2

Reversing the code

The supramolecular “code” involving a template has been reversed such that the templates and olefins act as hydrogenbond-acceptors and -donors, respectively. Code reversal was first demonstrated by MacGillivray and coworkers by using 2,3-bis(4-methylenethiopyridyl)naphthalene (2,3nap) as a template and fumaric acid (fum) as the olefin (Figure 5).15 A four-component assembly sustained by O–H· · ·N hydrogen bonds formed from cocrystallization of 2,3-nap and fum. The olefins were aligned and approxi˚ The solid mately parallel with a C=C separation of 3.84 A. reacted in a SCSC transformation to give ctba stereospecifically in up to 70% yield. A similar approach was later described by Wolf to direct the reactivity of fum wherein 1,8-bis(4-pyridyl)naphthalene (dpn) was a template16 to generate cyclobutanetetracarboxylic acid (cbta) in quantitative yield (Scheme 4).

H O

O

O H

N

O H

N

O O

Code reversal involving 2(dpn)·2(fum).

Templated construction of complex architectures

The ability to employ templates to direct reactivity independent of packing has enabled multiple C=C bonds to be assembled and react to form molecules with complex architectures. In particular, 5-methoxyresorcinol (5-OCH3 -res) was employed by the MacGillivray group to organize both trans-1,4-(4-pyridyl)-1,3-butadiene (4-py-but) and trans1,6-(4-pyridyl)-1,3,5-hexatriene (4-py-hex) into 0D fourcomponent assemblies wherein the C=C bonds lie parallel ˚ 17 UV irradiation of each solid and separated by 6000 m2 g−1 .22 Owing to the combination of extreme porosity and surface areas, tunable pore sizes and modifiable surfaces, MOFs stand out from other porous materials for gas storage and separation applications. In the first part of this chapter, we focus on the application of MOF materials to clean energy, hydrogen and methane storage, and carbon dioxide capture. There are several relevant reviews summarizing the fast growing research efforts in this field. Those by Yaghi,23 Long,24 Kitagawa, Zhou,25 and Hupp26 are highly recommended.

(a)

(b)

1.2

Porous MOFs for clean energy applications

1.2.1 Hydrogen storage in porous MOFs Hydrogen has been described as “the fuel of the future.” As an ideal energy carrier, hydrogen has many merits compared to fossil oil, including high-energy density, abundance, and benignancy in terms of emissions.27 There is great incentive for researchers to develop technologies that might efficiently exploit the potential of hydrogen energy for power generation, particularly for mobile uses. However, this effort is greatly hindered by the lack of a safe, efficient, and economical on-board hydrogen system.28 An ultimate target is the 7.5 wt% storage of H2 , or 70 g of hydrogen for each liter, set out by the US Department of Energy (DOE) for a hydrogen storage system (including tanks and temperature and pressure control equipment) by 2015, although these targets are set somewhat higher for the future. It is required that the systems need to be operated at near-ambient temperatures (−40 to 85 ◦ C) and high pressures (3 nm

Figure 6 A schematic of the birth and spread model of crystal growth (after Mullin2 ).

Amplification

factors that affect agglomeration rates are as follows: the supersaturation, the hydrodynamics, the surface charge, and the strength of adhesion between the particles.

Single crystal

Orientend attachment

2.4

Nonclassical crystallization

The overview provided above covers the classical perspective on crystallization and particularly nucleation. A rapidly growing body of work shows that this view is not complete. The alternative model for nucleation is often termed the two-step nucleation theory.5, 6 In brief, this model suggests that nucleation is a multistep process where the first step involves phase separation via the formation of liquid or amorphous nanoparticles. This is then followed by crystallization within this particle. The activation energy for each of these steps is relatively small, and it is expected that the overall process would be faster when compared with a single-step (classical) process with the same overall activation energy. Experimental and theoretical evidence supports this mechanism, and it has been suggested that the nucleation process is likely to proceed in this way in most, if not all, cases.5 A fascinating consequence of this nucleation mechanism arises when additives are used to stabilize the primary nanoparticles. Aligned assemblies of nanoparticles are typically expected to fuse to form a single crystal; however, an appropriate additive can prevent this, leading to the production of what have been termed mesocrystals (Figure 7).7 The nanoparticles in these systems are sufficiently well aligned to make detection by X-ray diffraction difficult, and it is possible that mesocrystal formation has not been recognized as such in many cases. While most of the reported examples of mesocrystal formation have involved inorganic

Crystal with complex shape

Mesocrystals

Iso-oriented crystal

Fusion (a)

Mesoscale assembly

Fusion (b)

(c)

Figure 7 Schematic representation of classical and nonclassical crystallization. Pathway (a): the classical crystallization pathway where nucleation clusters grow to a primary nanoparticle, which is amplified to a single crystal. Pathway (b): the primary nanoparticles form an iso-oriented crystal, which can fuse to form a single crystal. Pathway (c): the primary nanoparticles can be stabilized by an additive and can form a mesocrystal. Pathway (d): amorphous particles or liquid droplets are formed, which can transform to complicated morphologies. (Reproduced from Ref. 7.  John Wiley & Sons, Ltd, 2008.)

materials, examples of molecular mesocrystals are known and include DL-alanine and sodium ibuprofen.8 The mechanism by which the nanoparticles achieve oriented attachment is being actively investigated, and has been proposed to be caused by factors including physical fields (electric, magnetic, dipole), and steric control via an external matrix.8 The potential for designed supramolecular interactions in this area appears to be significant, and is largely unexplored to date.

2.5

Actions of impurities

For the vast majority of systems, species other than the crystallizing solute will be present as trace (or bulk) impurities, or as deliberately introduced crystal growth modifiers. In almost all cases, these other species will not be mere spectators but will influence the crystallization

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc129

Crystal growth and molecular crystal growth modification behavior in some way. Crystal growth from solution obviously occurs in the presence of the solvent, and this has been shown to be critically important, as discussed below. The mode of action of these additional or impurity species can be many and varied. For example, the following modes of action have recently been proposed,9 with an emphasis on mesocrystal formation: 1. 2. 3. 4. 5. 6. 7. 8. 9.

Adsorption of solution-phase solute species. Influence on soluble cluster formation. Nucleation inhibition of a precipitated nanoparticle phase. Adsorption on, and stabilization of, nucleated particles. Influence on the structure of the nucleated phase (amorphous polymorph). Influence on the nanocrystal shape through facespecific interactions. Influence on the oriented attachment and alignment of nanoparticles. Stabilization of mesocrystals inhibiting fusion to a single crystal. Mechanical reinforcement of the crystal through incorporation of the additive and/or modification of the hybrid structure.

To these could be added the absorption of an additive to a crystal face of a single crystal resulting in a change of growth rate, and incorporation of an additive molecule into the bulk crystal structure. It has been proposed that a single additive can be involved in a number of these modes of action simultaneously. Clearly, additives and impurities can have a complex interplay with crystal nucleation and growth, and there remains much to be understood in this area, particularly with respect to nonclassical crystal growth.

3

MOLECULAR CRYSTALLIZATION

The generic discussion of crystal growth above applies equally to inorganic systems and molecular crystals, although the majority of the literature on nonclassical crystallization to date focuses on inorganic systems, and in particular, on calcium carbonate.7 In terms of solution clustering, there has been a significant amount of work studying molecular systems in organic solvents, with a particular emphasis on the relationship between solution-phase clusters and the polymorph crystallized (see Polymorphism: Fundamentals and Applications, Supramolecular Materials Chemistry). Here, we will focus on recent work studying molecular systems in water. Recently, it has been shown for inorganic species that the formation of hydrates (see Clathrate Hydrates,

5

Supramolecular Materials Chemistry) is essentially thermodynamically favorable but that as more water molecules are “inserted” in the structure the additional Gibbs free energy contribution diminishes, thus limiting the number of hydrates observed.10 While such data do not exist for molecular crystals, many molecules crystallize as hydrates, thus suggesting a favorable interaction with water. In addition, evidence has been accumulated to suggest that the molecular interactions with water in the solvated state are closely related to the formation of the eventual crystalline state.11–13 This is critical to our understanding of the “clustering” that occurs in solution prior to the formation of the crystal. Thus, for molecular crystals, in particular, if one has a good idea of the solution clustering of the molecule one also has a good idea of the water–solute interactions that occur in the solid state. Recent experiments involving synchrotron radiation have been able to probe the solution and show that solution clustering occurs in water.14–17 The transition from solution state to crystalline state appears to be via the removal of water from the solution cluster and this process can be rate limiting.13 Additionally, the amorphous to crystalline transformation has been found to be an activated process for some molecular crystals, suggesting that this too can be rate limiting.18 In molecular crystallization, there is also the necessity to understand how hydrates form anhydrous products and vice versa. This is particularly important in the pharmaceutical industry where the hydration of products may significantly alter their efficacy. Also, there is some limited work in this field showing that both solution-mediated and solidstate transitions can occur.19 To conclude this section, we highlight that similar arguments can be made in the presence of other solvents and to the formation of solvates rather than hydrates. The molecular solid-state structure of solvates can be considered to belong to the “host–guest intermolecular complexes” group with the solvent as the possible guest20 (see Clathrate Hydrates, Cocrystals: Synthesis, Structure, and Applications, and Synthetic Clathrate Systems, Supramolecular Materials Chemistry).

4

TAILOR-MADE ADDITIVES FOR CRYSTAL GROWTH CONTROL

Having discussed the general fundamental principles of crystallization and the ways in which impurities may impact on this, this section will cover seminal works in the area of designing additives to interact with crystal faces using fundamental supramolecular chemistry. The morphology control of molecular crystals was a relatively undeveloped area until the group of Weissbuch introduced the “tailor-made” additive concept in 1980.21

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc129

6

Supramolecular materials chemistry

B

+b

B B

(010)

B

Position 1 A

A

A

A

Position 3 Additive inhibits face B

B

B

B

B

Figure 8 A schematic showing the change in crystal morphology induced by an additive that is selective for face B, thereby reducing the growth rate of face B relative to that of face A, leading to face B dominating the morphology.22

Essentially, by making use of the known molecular crystal structure, additives could be rationally chosen which would impact on certain faces and modify the morphology. In simple terms, an additive can alter the morphology of a crystal when it adsorbs to a specific crystal face (or faces) and changes the relative growth rates of the faces (Figure 8). This concept can be used to resolve racemic mixtures, determine absolute configuration, template nucleation of particular faces, or control polymorphism.22–26 This strategy was shown to be highly successful in many situations, some of which are briefly covered here, using the widely studied glycine system as a case study. An illustrative example is the assignment of the absolute configuration of a chiral molecule by observing the changes induced in the morphology of a prochiral molecular crystal.27 An early example is that reported by Weissbuch et al. studying the crystallization of prochiral glycine in the presence of chiral alanine.28 The α-polymorph of glycine crystallizes from water in a symmetrical morphology that expresses both the (010) and (010) faces. The packing of the molecules perpendicular to the b direction is shown in Figure 9. Replacing a hydrogen atom on the carbon atom with a methyl group generates D- or L-alanine depending upon which proton is replaced. The incorporation of Dalanine at the (010) surface is shown in Figure 9. If one now considers the intermolecular interactions associated with this substitution, it seems clear that the D-alanine fits neatly into the crystal surface in the position shown, with the methyl group directed out of the crystal surface. Other symmetrically related positions in the crystal are likely to be much less favorable. For example, a D-alanine molecule in position 1 or 2 at the (010) face will have the methyl group directed into the bulk of the crystal, leading to highly unfavorable interactions. Similarly, in position 3 or 4, the methyl group would be directed along the crystal surface, rather than project from it, and this would again be unfavorable. Hence, these simple supramolecular considerations suggest that the glycine crystals grown in the presence of

Position 2 Position 4

(a)

(010) +b (010) Position 1

Position 3

Position 2 Position 4

(b)

(010)

Figure 9 (a) The packing arrangement of α-glycine viewed perpendicular to the b direction. (b) Incorporation of a D-alanine molecule (circled) at the (010) crystal face. Position 1 is related to position 2 by an 180◦ rotation about b and a translation. Positions 1 and 3 (and 2 and 4) are related by an inversion. (Reproduced with permission from Ref. 27.  American Chemical Society, 2008.) D-alanine should be inhibited on the (010) face and not the (010) face. Indeed that is what was observed, as is shown schematically in Figure 10.28 The opposite effect is observed for L-alanine as would be expected. Adding a racemic mixture of a chiral additive leads to a symmetric morphology, as shown in Figure 11(b), in this case using the bulkier phenylalanine additive, again in the crystallization of α-glycine.29 Interestingly, increasing the concentration of this additive was found to fully inhibit the formation of α-glycine allowing the γ -glycine form to crystallize (Figure 11), illustrating that additives can induce a range of effects depending on the conditions used.29 Another example of the insights available through the use of tailor-made additives can be found in the study of the polarity of β-glycine crystals. This polymorph of glycine crystallizes in the chiral space group P 21 . Attempts to assign the polarity of the crystals using anomalous X-ray scattering techniques were unsuccessful.29 Instead, crystals

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc129

Crystal growth and molecular crystal growth modification

7

α-Glycine (010)

(b)

(a)

(c)

(a)

Figure 10 Schematic of the morphologies of α-glycine grown (a) from pure water, (b) in the presence of D-alanine, and (c) in the presence of L-alanine. (Reproduced with permission from Ref. 27.  American Chemical Society, 2008.)

600 µm

600 µm (a)

(b)

600 µm

(c)

90 µm

(b)

Figure 12 (a) Yellow (−)-β-glycine crystals, and (b) colorless (+)-β-glycine crystals, grown in the presence of S-N-(2,4dinitrophenyl)lysine. (Reproduced with permission from Ref. 29.  American Chemical Society, 2005.)

monolayers) can also be used to template crystallization of particular faces in inorganic systems, such as NaCl for example.23 It should be mentioned that computational molecular or Monte Carlo modeling is increasingly being used to better understand how additives interact with crystal surfaces to bring about specific outcomes.30 While it is not appropriate to discuss this in detail here, some measure of the power of computational methods in this area is illustrated by the relatively recent report of a combination of molecular and Monte Carlo simulations successfully predicting the macroscopic morphology of urea crystals, if urea–solvent interactions are properly considered (Figure 13).31 The ability to apply computational methods across such a large

(d)

Figure 11 Glycine crystals grown from aqueous solutions in the absence and in the presence of various additives: (a) α-glycine, no additive; (b) α-glycine in the presence of 1% RS -phenylalanine; (c) γ -glycine with 8% RS -methionine; and (d) γ -glycine with 3.5% RS -phenylalanine. (Reproduced with permission from Ref. 29.  American Chemical Society, 2005.)

grown in the presence of a chiral and colored additive were found to selectively include the additive according to the polarity of the individual crystal (Figure 12), thus achieving discrimination that could not be made using Xray techniques.29 Tailor-made additives can also be used to template a particular face by nucleation at the air–water interface modified with a carefully selected monolayer. An example, once again, is the crystallization of α-glycine. The hypothesis for this work was that the R-α-amino acids should induce the (010) face in the glycine crystals while the S-isomer should induce the (010) face, and indeed this was found to be the case.23 Nucleation under monolayers (or Langmuir

[001] [001] 1µm

[111] 1 µm

[110]

[110]

c a b

(a)

(b)

Figure 13 Computed and experimental morphologies of urea in (a) water and (b) methanol. (Reproduced with permission from Ref. 31.  Nature Publishing Group, 2005.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc129

8

Supramolecular materials chemistry

size range is notable and has particular importance in the study of nucleation and crystal growth. As this urea example shows, it is evident that the solvent itself can be used as a tailor-made additive for the control of morphology, the formation or selection of hydrates or solvated molecular crystals,22 or the polymorph obtained.32 Obviously, one main difference between the use of tailormade additives to control morphology and the use of the solvent to control polymorph selection (or anything else) is that the solvent is present in large excess (and this is quite different to the situation that is normally encountered for tailor-made additives). A review of the use of solvents as additives for polymorph control can be found in Blagden and Davey.33 This method relies on an understanding of the various solvation and desolvation pathways in the various solvents and once again, simulations have proven useful in understanding the solvent effects. It has been shown that solvent effects are not only limited to hydrogen bonding but also can involve electrostatic and van der Waals interactions.34 Let us now look at an area that is less investigated, that of crystallization promotion. The majority of this work has involved the promotion of nucleation by the use of close-packed surfactants at the air–water interface.23, 24, 35 The polar heads of these surfactants are chosen to mimic structural motifs of the fast-growing faces of the molecular crystal of interest. It is still unclear why this strategy works, since in solution this would lead to inhibition and classical nucleation theory also gives no reasonable answers.35 An example of this promotion is the crystallization of ice at long chain alcohol monolayers.25 Long chain alcohols were chosen due to the similarity of the oxygen positions in the ab layer of ice and the oxygen positions of the alcohols when packed in the herringbone arrangement. GIXRD (grazing incidence X-ray diffraction) was used to deduce an epitaxial mode of ice nucleation onto the polar section of the monolayer. In terms of solution-phase crystal growth promotion, molecular additives have been shown to promote the growth of inorganic crystals, in part by lowering the energy barrier to dehydration of the cationic species.36 Very recently, a similar concept has been proposed to explain the surprising acceleration of growth of γ -glycine in the presence of tailor-made additives that are known to inhibit the growth of α-glycine.37

5

CONCLUSIONS

to many areas of supramolecular chemistry, and the study of phenomena such as clustering in supersaturated solutions, adsorption of additives on growing crystal surfaces, and so on, remains a significant technical challenge. As just one example of how active and challenging this area is, a recent computational study of the impact of methanol on the selective crystallization of α- and β-glycine suggests that neither of the mechanisms (impact on solution aggregation, versus inhibition of the growth of specific crystal faces) proposed in the literature are actually correct.38 Clearly, however, supramolecular interactions are a key factor in crystal growth and the control of crystallization, and the future of this area will benefit greatly from increased collaboration between supramolecular chemists and crystal growth specialists.

REFERENCES 1. J. D. Dunitz, Pure Appl. Chem., 1991, 63, 177. 2. J. W. Mullin, Crystallization, Butterworth-Heinemann Publishers Ltd, London, 1993. 3. N. Rodriguez-Hornedo and D. Murphy, J. Pharm. Sci., 1999, 88, 651. 4. R. C. Burton, N. Cabrera, and F. C. Frank, Philos. Trans., 1951, A243, 299. 5. D. Erdimemir, A. Y. Lee, and A. S. Myerson, Acc. Chem. Res., 2009, 42, 621. 6. P. G. Vekilov, Cryst. Growth Des., 2010, 10, 5007. 7. H. C¨olfen and M. Antonietti, Mesocrystals and Non Classical Crystallization, John Wiley & Sons, Chichester, U.K., 2008. 8. R. Q. Song and H. Colfen, Adv. Mater., 2010, 22, 1301. 9. D. Gebauer, H. Colfen, A. Verch, and M. Antonietti, Adv. Mater., 2009, 21, 435. 10. L. Glasser and F. Jones, Inorg. Chem., 2009, 48, 1661. 11. R. C. Burton, E. S. Ferrari, R. J. Davey, and D. T. Bowron, J. Phys. Chem. B , 2009, 113, 5967. 12. R. J. Davey, N. Blagden, S. Righini, et al., Cryst. Growth Des., 2001, 1, 59. 13. R. C. Burton, E. S. Ferrari, R. J. Davey, et al., Cryst. Growth Des., 2008, 8, 1559. 14. S. E. McLain, A. K. Soper, A. E. Terry, and A. Watts, J. Phys. Chem. B , 2007, 111, 4568. 15. D. T. Bowron and J. L. Finney, J. Phys. Chem. B , 2007, 111, 9838. 16. S. Hamad, C. E. Hughes, C. R. A. Catlow, and K. D. A. Harris, J. Phys. Chem. B , 2008, 112, 7280. 17. L. Almasy and G. Jancso, J. Molec. Liq., 2004, 113, 61.

When one considers the number of mechanisms by which additives and impurities can have an impact on crystal growth, it is clear that this is a very complex area. The study of surfaces requires a different suite of techniques compared

18. S. Janbon, R. J. Davey, and G. Dent, J. Phys. Chem. C , 2008, 112, 15771. 19. A. L. Gillon, R. J. Davey, R. Storey, et al., J. Phys. Chem. B , 2005, 109, 5341.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc129

Crystal growth and molecular crystal growth modification 20. F. H. Herbstein, Cryst. Growth Des., 2004, 4, 1419. 21. L. Addadi, J. van Mil, E. Gati, and M. Lahav, Orig. Life, 1980, 11, 107. 22. I. Weissbuch, L. Leiserowitz, and M. Lahav, Chirality, 2008, 20, 736. 23. R. Popovitz-Biro, I. Weissbuch, D. Jacquemain, et al., in Advances in Industrial Crystallization, eds. J. Garside, R. J. Davey, and A. G. Jones, Butterworth Heinemann, Oxford, 1991, p. 3. 24. H. Rapaport, I. Kuzmenko, M. Berfeld, et al., J. Phys. Chem. B , 2000, 104, 1399. 25. I. Weissbuch, M. Lahav, and L. Leiserowitz, Cryst. Growth Des., 2003, 3, 125.

9

29. Y. V. Torbeev, E. Shavit, I. Weissbuch, et al., Cryst. Growth Des., 2005, 5, 2190. 30. R. J. Davey, N. Blagden, G. D. Potts, and R. Docherty, J. Am. Chem. Soc., 1997, 119, 1767. 31. S. Piana, M. Reyhani, J. D. Julian, and D. Gale, Nature, 2005, 438, 70. 32. A. Llin`as and J. M. Goodman, Drug Discov. Today, 2008, 13, 198. 33. N. Blagden and R. J. Davey, Cryst. Growth Des., 2003, 3, 873. 34. C. Stoica, P. Verwer, H. Meekes, et al., Cryst. Growth Des., 2004, 4, 765. 35. K. Chadwick, R. J. Davey, and R. Moughal, Org. Process Res. Dev., 2009, 13, 1284.

26. I. Weissbuch, L. Leiserowitz, and M. Lahav, in Crystallization Technology Handbook , ed. A. Mersmann, Marcel Dekker, Inc., New York, 1995, p. 401.

36. S. Piana, F. Jones, and J. D. Gale, Cryst. Eng. Commun., 2007, 9, 1187.

27. M. A. Lovette, A. R. Browning, D. W. Griffin, et al., Ind. Eng. Chem. Res., 2008, 47, 9812.

37. R. Dowling, R. J. Davey, R. A. Curtis, et al., Chem. Commun., 2010, 46, 5924.

28. I. Weissbuch, L. Addadi, Z. Berkovitchyellin, et al., J. Am. Chem. Soc., 1983, 105, 6615.

38. J. Chen and B. L. Trout, J. Phys. Chem. B , 2010, 114, 13764.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc129

Soft Matter Science—a Historical Overview with a Supramolecular Perspective David K. Smith University of York, York, UK

1 Introduction 2 From Colloids to Supramolecular Chemistry 3 Liquid Crystals as Supramolecular Soft Matter Systems 4 From Polymers to Supramolecular Materials 5 Conclusions References

1

1 2 5 7 12 13

INTRODUCTION

Soft matter science is one of the fastest growing areas of modern interdisciplinary science, which builds on diverse experimental and theoretical foundations, and has experienced intense current development and interest.1 Soft matter systems can be easily deformed by thermal stresses and fluctuations,—and, as such, soft materials usually have significant degrees of disorder and are highly responsive. Importantly, soft matter systems have different behaviors across a range of length scales. Molecular-scale components often assemble into nanoscale/microscale architectures, which can sometimes self-organize on the mesoscopic level. All these features can have an impact on the ultimate macroscopic behavior of the material—tuning the observed soft matter properties. As such, soft matter assemblies are quite different from “hard matter” systems, in which the same repetitive crystalline organization patterns the material

all the way from the molecular scale up to the macroscopic level. Soft matter is, of course, familiar to us from everyday life—for example, gels, foams, rubbers, liquids, and plastics are all types of soft matter. Soft matter occurs naturally, and is widespread in biology, while synthetic approaches are also able to generate a variety of soft matter systems with highly tunable properties—as this article illustrates. As a consequence, soft matter systems have a wide range of technological applications—for example, in packaging, adhesives, personal care products, food additives, cosmetics, detergents, lubricants, and so on (Figure 1). Liquid crystals are a typical example of a responsive soft material, which can change optical properties in response to an electric field. This behavior enables liquid crystals to be used as a key component in display screen technology. As a discipline, soft matter science is highly interdisciplinary.2 It is important to understand the chemical/ molecular composition of such systems in order to determine how assembly and self-organization arises. Physics then plays an important role in analyzing, modeling, and understanding the properties of the materials, while much of biology comprises many soft matter systems—from biological fluids such as blood and milk, to individual biological substructures such as cells and connective tissue. Being able to control and understand the complexity of soft matter systems is therefore a significant intellectual challenge, but enables major technological breakthroughs. Furthermore, much of our current electronic technology is based on hard matter systems (circuit boards, transistors, etc.), and, over the next 100 years, there is no doubt that interfacing hard and soft matter systems more effectively with one another will become increasingly important at the physics/chemistry/biology interface.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

2

Soft matter

Figure 1 Some traditional applications of soft matter science—lithium grease as an engine lubricant, LCD screens, gels in personal care, polymeric adhesives, laundry detergents, and cosmetics.

Clearly, understanding and controlling the interactions between molecular-scale building blocks is fundamental to gaining an insight into soft matter science; as such, interactions lie at the heart of these materials.2 Therefore, in the last 20 years, many supramolecular chemists have become involved in designing systems that contain multiple noncovalent interactions, which exhibit soft matter behavior. However, because the field of soft matter science, as described above, is ultimately based on a materials-level description of systems rather than a single scientific concept/approach, this field has emerged from a wide range of different scientific disciplines. The goal of this introductory article is therefore to trace the origins of soft matter science, with specific regard to chemistry and supramolecular/nanoscience. The remainder of this encyclopedia volume then provides detailed examples of the outstanding recent work that combines supramolecular methodology with soft matter science principles in order to generate highly functional and tunable new material systems. The specific advantages that can be built into soft matter systems by taking a supramolecular approach are the capacity to endow them with multiple functionalities, and to tailor the interactions between different molecular and nanoscale components more precisely. In this way, soft matter systems with enhanced high-tech applications ranging from enhanced paints and coatings to smart drug delivery vehicles can be generated.

2

FROM COLLOIDS TO SUPRAMOLECULAR CHEMISTRY

It is widely accepted that the genesis of what is now known as supramolecular chemistry was in the late 1960s, with the pioneering work of Pedersen, Lehn, and Cram, who developed the concepts of using noncovalent interactions to design host molecules (receptors) capable of binding specific guests (substrates) (Definition and Emergence of Supramolecular Chemistry, Concepts).3 However, even

then, it is clear that the field of host–guest chemistry drew on much older principles, such as Emil Fischer’s Lock and Key hypothesis, which was first outlined as early as 1892.4 Indeed, as Jean-Marie Lehn defined supramolecular chemistry as “chemistry beyond the molecule” in the 1970s and 1980s, and the paradigm of self-assembly became ever more important in the field (Self-Assembly and SelfOrganization, Concepts), it became clear that, under this broad definition, the field of supramolecular science was much wider than had perhaps been imagined at its inception.5 This realization culminated in the awarding of the Nobel Prize to the early pioneers of supramolecular chemistry in 1987.6 It is fair to say that using the Lehn definition of supramolecular chemistry as “chemistry beyond the molecule”, the field straddles an enormous range of modern scientific endeavor, and can be considered as an area of chemical science at least as influential as the traditional three areas of organic, inorganic, and physical chemistry. Indeed, it is worth noting that a very high percentage (circa 25%) of papers published in leading general chemical science journals, such as Angewandte Chemie, J. Am. Chem. Soc., and Chem. Commun. are based on supramolecular principles.7 Furthermore, a fundamental understanding of noncovalent interactions and thermodynamically controlled self-assembly provides the ability to approach and solve scientific problems ranging from biochemical reaction mechanisms to the construction and manipulation of functional nanoscale electronic devices. However, amongst the chemical foundations on which the pioneers of supramolecular chemistry were building, although sometimes without acknowledgment, were the basic principles of colloid science. Colloid chemistry had been long established by the time of the initial development of supramolecular chemistry, and had also focused on the behavior of ensembles of molecules—typically mediated through noncovalent interactions.8 Indeed, the concept of self-assembly is fundamental in the study and manipulation of colloidal materials.9 By definition, colloids consist of a dispersed phase (or discontinuous phase) distributed

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

Soft matter science—a historical overview 3 Table 1 Nature of colloids along with (in italics) some typical examples.

Discontinuous Phase

Continuous phase Solid

Liquid

Gas

Solid

Solid dispersion Stained glass

Suspension/gel Toothpaste

Solid aerosol Smoke

Liquid

Solid emulsion Nanocapsule

Emulsion Milk

Liquid aerosol Fog

Gas

Solid foam Molecular sieve

Foam Beer froth

—

uniformly in a finely divided state in a dispersion medium (or continuous phase). If the dispersed phase has at least one dimension in the size range 1–1000 nm, the system is described as colloidal. Table 1 indicates how using different states of matter as discontinuous and continuous phases gives rise to different types of colloids. Colloidal systems have been known for millennia, and have been scientifically studied for well over 100 years. Key Nobel Prizes for Chemistry and Physics that recognized the establishment of the field as a serious arena of scientific study were awarded in 1925/1926 to Richard Zsigmondy,10 Theodor Svedberg,11 and Jean Baptiste Perrin12 for their work on understanding the behavior of disperse systems. In any colloid, perhaps the most interesting point of the structure is the surface, or interface, between the discontinuous and the continuous phase. Clearly, the two different phases are not necessarily mutually compatible. Take, for example, the case of an oil–water (liquid–liquid) mixture. The oil is hydrophobic and the water hydrophilic, and, as such, the oil has a tendency to aggregate, minimizing its surface area and hence its contact with the water. This process of surface area minimization is driven by the entropic benefits of releasing the water molecules that are “frozen” at the apolar surface, combined with the enthalpic benefits of allowing the surface layer of water molecules to maximize their hydrogen-bonding potential as a part of the bulk water phase. Indeed, it is worth noting that much of the thermodynamic understanding of the hydrophobic effect came about as a consequence of the pioneering work of colloid chemists as early as the 1930s13 —work that went on to have major influences on both biological and supramolecular chemists. Clearly, therefore, in order to stabilize an oil-in-water emulsion with small particle sizes (1–1000 nm), as required for a true colloidal system, it is necessary to use some sort of additive that is capable of stabilizing the interface—a so-called surface-active agent, or surfactant. Surfactants designed to stabilize oil-in-water emulsions should contain a hydrophobic segment, capable of interacting with the oil, and a hydrophilic part, capable of interacting with

the aqueous phase. The covalent connection of these two different units within the one molecular structure allows the effective compatibilization of the two discrete phases within the bulk mixture. Of course, detergents of this type have been known in practical terms since ancient Egyptian times, when cassia oil, water, and alkali were mixed together to form simple soaps. In more recent times, colloid chemists have realized that, within aqueous solution, surfactants are capable of forming a variety of structures with different morphologies— spheres, rods, bilayers, and so on—as they attempt to “hide” their apolar functionality from the aqueous phase— and, as such, colloid scientists were investigating “programmed self-assembly” of molecular-scale building blocks in water before supramolecular chemists developed the concept. In a highly influential work, Israelachvili proposed a series of predictive rules that helped rationalize why amphiphilic molecules assemble into nanoscale architectures with different shapes. (Figure 2)14 In essence, the relative sizes of the hydrophobic and hydrophilic blocks determine the packing parameter, P , which controls the assembled architecture according to relatively simple rules of geometry. A sphere has the smallest hydrophobic/ hydrophilic (interior/exterior) ratio, and as such is favored by amphiphiles with smaller hydrophobic groups and larger hydrophilic head groups (allowing maximum space on the surface for the head groups to occupy). Cylindrical micelles have relatively more interior space, while bilayers have even more, and, therefore, as the hydrophobic tail unit increases in size relative to the hydrophilic head group, the shape of the assembled nanostructure changes from spheres to cylinders to bilayers to inverted spheres. This was one of the first examples in which the self-assembly of nanostructures from molecular-scale building blocks could be rationalized in a simple and predictive way. As noncovalent interactions are very important in controlling surfactant aggregation, and given the wide range of real-world applications of colloids, supramolecular chemists have recently increasingly extended their

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

4

Soft matter Spherical micelles

P < 1/3 a lc

n

Cylindrical micelles 1/3 ≤ P < 1/2

Flexible bilayers, vesicles 1/2 ≤ P < 1

Planar bilayers P=1

Inverted micelles (hexagonal (HII) phase)

P>1

Figure 2 Different morphologies of surfactant assemblies as predicted by Israelachvili’s packing parameter, P . (Reproduced from Ref. 15, Hindawi Publishing Corporation, 2011.)

investigations into areas that were traditionally the preserve of colloid scientists—blurring the interface between these two areas (Self-Assembly of Facial Amphiphiles in Water, Soft Matter). In particular, colloidal assemblies are considered to be simple, yet highly tunable, fundamental nanoscale building blocks that can subsequently allow further self-assembly processes to take place. For example, the incorporation of ligand groups into the head group of an amphiphile can generate nanoscale micellar aggregates with multiple ligand groups located at the surface. Such systems are ideal for achieving multivalent interactions (Multivalency, Concepts) with selected targets. For example, they may be useful for creating targeted nanoscale drug delivery vehicles.16 This approach therefore allows

hierarchical control of complex self-assembling nanostructures. In this way, colloid chemistry is a vital tool in the synthetic kitbag of any modern supramolecular scientist, with colloidal self-assembled systems playing important fundamental roles in the “bottom–up” assembly of nanostructured materials. It is not only liquid–liquid colloids that are of interest to the supramolecular scientist. All types of colloidal organizations have now been incorporated into the repertoire of supramolecular chemistry. For example, solid–liquid and solid–gas colloids have been of great interest in the development of nanostructured materials. For example, metal nanoparticles normally grow uncontrollably into bulk metal in order to minimize their surface area—but if a suitable

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

Soft matter science—a historical overview 5

Figure 3 Lycurgus cup, containing gold nanoparticles that change its optical appearance, and a model of taxol-stabilized gold nanoparticles—hard metal nanoparticles with a soft ligand shell (Reproduced from Ref. 19.  American Chemical Society, 2007.)

capping agent is present, which can interact with both the metal surface and the surrounding fluid, then particle growth can be capped, and well-defined nanoparticles obtained.17 Metal nanoparticles, although present in the Lycurgus cup of the ancient Romans (Figure 3), and known from the pioneering work of Faraday,18 have seen a serious surge of interest in the last 20 years, since the ready availability of electron microscopy in most laboratories, and the intense activity in the field of nanoscience and technology. The metallic parts of these systems tend to constitute what is generally defined as “hard” matter systems, as they have a much greater degree of crystallinity than soft matter systems. However, the surface layer is often quite fluidlike in some of its behavior—for example, multiple flexible ligands can be attached with the intention of interfacing them with biological systems (Figure 3).19 Metal nanoparticles are clearly related to other examples of flat solid–liquid and solid–gas interfaces. Although not strictly colloidal as they are not finely divided, such extended interfaces are closely related to colloidal systems. It was the pioneering work of Langmuir which first investigated monomolecular layers,20 and building on his work using controlled interactions to achieve monolayer assembly on solid surfaces can lead to assembled ordered structures at extended interfaces. Such surface-modified materials (Self-Assembly of Surfactants at Solid Surfaces, Physisorption for Self-Assembly of Supramolecular Systems: A Scanning Tunneling Microscopy Perspective, and Chemisorbed Self-Assembled Monolayers, Soft Matter) can subsequently be further patterned on the nanoscale using top–down methods of fabrication, and have widespread applications ranging from the development of biosensors to nanoelectronics. There are many other examples in which a solid-derived colloidal approach is being developed and used—such as nanocapsule formation (Self-Assembly of Polymers into Soft Nanoparticles and Nanocapsules, Soft Matter).21 These colloidal systems can be considered as being

partly crystalline (“hard” matter), but still have significant disorder/responsiveness. This understanding of semicrystalline systems leads us into consideration of another crucial class of material, which, like colloids, evolved separately to, and in parallel with, supramolecular science but is increasingly being employed in the field, and is vital to the understanding of soft matter—liquid crystals.

3

LIQUID CRYSTALS AS SUPRAMOLECULAR SOFT MATTER SYSTEMS

Intriguingly, like colloid science, and Fischer’s Lock and Key hypothesis, the genesis of liquid crystals was also in the late nineteenth century. Indeed, it is fair to say that this period in history was distinguished by an unparalleled and explosive growth of scientific endeavor across many fields. The first example of liquid crystalline behavior was reported in 1888/1889 by Friedrich Reinitzer22 and Otto Lehmann,23 who were investigating a series of cholesterol derivatives. They noted that, just above their melting points, these compounds exhibited intriguing color effects. Perhaps surprisingly, there was relatively little interest in the phenomenon and, although during the early part of the twentieth century, Daniel Vorl¨ander synthesized a wide range of different liquid crystalline materials,24 it was not until the late 1960s that interest in this family of soft materials really surged forward. Once Kelke25 and Gray26 achieved the synthesis of chemically stable liquid crystalline substances with relatively low melting temperatures, such that they could exhibit room temperature liquid crystallinity, the commercial application of liquid crystals became viable, and over the following 40 years, liquid crystals have become commonplace in a wide range of electronic devices. These materials have revolutionized display screen technology and enabled the development of portable electronic devices, which have underpinned the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

6

Soft matter

way in which information technology has transformed the world in the late twentieth and early twenty-first centuries.27 A full description of liquid crystalline (LC) materials can be found elsewhere in this volume (Self-Organization and Self-Assembly in Liquid-Crystalline Materials and Liquid Crystals Formed from Specific Supramolecular Interactions, Soft Matter)—but it is instructive to consider the basics here. In chemical terms, LC molecules (mesogens) are usually anisotropic, for example, having rodlike shapes. In the crystal form, these rods would be perfectly aligned and perfectly packed, while in the liquid state, completely free and random motion of the rods would lead to isotropic behavior. However, intermediate between the crystal and liquid state, molecules such as this can sometimes exhibit a liquid crystal form, in which the molecules have a degree of alignment, but also some freedom to move.28 Typically the LC phase is accessed by heating the solid form or cooling the liquid form, and is intermediate between the two. There are various types of molecular organizations within LC phases, with nematic (all molecules oriented in the same direction), smectic (all molecules oriented in the same direction and organized in layers), and cholesteric (twisted layers of aligned mesogens) being typical (Figure 4). Often, one molecule can exhibit more than one type of LC phase depending on the temperature. Clearly, the forces responsible for LC organization are primarily anisotropic van der Waals and steric forces, which in bulk materials are additive and become very significant over a relatively long range. As such, liquid crystal ordering is related to the process of crystal packing. However, as a consequence of the thermal energy inherent within the system, these forces are somewhat frustrated by entropy, which therefore allows a degree of mobility of the molecular building blocks and some liquidlike behavior. Liquid crystals of this type, which respond to temperature in their bulk phase by going through an LC phase transition, are described as thermotropic materials. It is worth noting that,

(a)

Figure 4

Smectic A

(b)

Smectic C

(c)

in addition to rodlike molecules, disklike molecules can also organize themselves into LC phases—in this case, forming loose stacks that can then further align with one another.29 If the LC molecular building blocks are chiral, then it is possible to obtain twisted aligned phases, in which the molecules are arranged in a helical manner on the nanoscale.30 The molecular-scale chirality is translated into the chiral organization of the LC phase as a whole. If a twisted nematic has a helical pitch that is on the order of the wavelength of light, then interesting optical interference effects can be seen. Importantly, the twisting of some liquid crystal phases can effectively be switched by the use of an electric field, and this has led to some of their most important applications. If the LC phase is placed between polarizers, then the electric field can essentially appear to switch an optical interference pattern (Figure 5) on and off—the basis of a simple pixel, which ultimately allows the development of LC display screen technology. In large part, liquid crystal chemistry, a fundamental area of soft matter science, has evolved somewhat separately from supramolecular chemistry. However, it is clear that the forces between molecules—both short range and long range—are important in mediating the formation of these materials. Obviously, in terms of the thermotropic LC molecules described above, shape-derived steric effects and dispersion forces will dominate the LC phase formation. However, the anisotropic dispersion forces that allow the formation of an LC phase can also be supplemented by other noncovalent interactions. For example, mesogens with terminal dipoles can align in an antiparallel manner, with the true mesogenic species being a dimer.29 In addition to the noncovalent forces responsible for mesogen–mesogen interactions, it is also possible to generate LC materials in which the mesogenic species itself is underpinned by noncovalent interactions. For example, complementary molecules that form hydrogen bonds with one another can then form an extended rodlike molecule,

Nematic

(d)

Cholesteric

Organization of anisotropic rodlike molecules in different liquid crystalline phases.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

Soft matter science—a historical overview 7

Figure 5 filters.

Responsive optical properties of liquid crystalline materials. Typical patterns observed by optical microscopy under polarizing

which shows significantly enhanced LC properties (e.g., lower energy phase transitions).31 Other interactions, such as donor–acceptor interactions32 and halogen bonding33 can also be used to organize mesogenic species at the subLC-assembly stage. In addition, metals have been incorporated into liquid crystalline phases, usually by complexing them to ligands, which then form a complex with shape anisotropy.34 These complexes can then organize into liquid crystalline phases, which preserve some aspects of the behavior of the metal within them, such as redox responsiveness and optical properties. Liquid crystalline behavior can also be exhibited by some molecules in the presence of a solvent phase. Such systems are described as lyotropic liquid crystals.35 This kind of behavior is directly related to the colloidal systems described in Section 2. For example, rodlike (cylindrical) micelles or membrane bilayers, formed by surfactants, constitute anisotropic architectures in a solvent phase. These assemblies can further align with one another and, as a consequence, exhibit LC-type behavior. Unlike thermotropic liquid crystals, which can only be addressed by changing the temperature, with lyotropic liquid crystals, the phase behavior can also be controlled by changing the concentration. Indeed, the concentration of a surfactant can profoundly influence the types of assemblies which it forms, and hence the phase behavior. Clearly, in lyotropic liquid crystal phases, there is a subtle interplay of intermolecular forces at work. Long-range, crystal-packing-type interactions are important (as for thermotropic liquid crystals), along with solvent–LC and solvent–solvent interactions and also LC–LC interactions. Such systems play vital roles in biological organisms, and this is one of the most exciting current interdisciplinary interfaces of liquid crystal and colloid/surfactant science.36 Although liquid crystalline materials clearly employ supramolecular principles in their molecular assembly and organization processes, the field has largely remained distinct from supramolecular chemistry. However, increasingly, chemists from many different backgrounds are

harnessing the power of LC organization to generate impressive and highly functional nanomaterials, and it is hoped that the articles in this volume will encourage supramolecular chemists to pay more attention to LC phenomena, and liquid crystalline chemists to see new possibilities for their work within a supramolecular context.

4

FROM POLYMERS TO SUPRAMOLECULAR MATERIALS

Perhaps the archetypal new materials of the twentieth century are based on polymer technology. Indeed, it is fair to say that, for the past 100 years or so, we have been living in the polymer age.37 Since the pioneering work in the 1800s, in which celluloid was developed as a synthetic replacement for ivory, and rayon as a silk substitute, it has been realized that polymers can generate functional materials with a wide range of different physical properties and potential applications.38 In 1920, Staudinger proposed that polymers had extended-chain molecular structures,39 an idea much ridiculed by many organic chemists who believed polymers were actually colloidal materials similar to those described in Section 2, consisting of small molecules held together by multiple weak (noncovalent) interactions. Indeed, a famous letter was written by Wieland, who doubted Staudinger’s hypothesis of polymers as macromolecules40 : “Dear Colleague, abandon your idea of large molecules, organic molecules with molecular weights exceeding 5000 do not exist. Purify your products such as rubber, they will crystallize and turn out to be low molecular weight compounds.” From the 1920s, however, it increasingly became clear that Staudinger’s hypothesis was correct, and polymeric materials became embedded in every aspect of modern life. Eventually, in 1953, Staudinger was rewarded the Nobel Prize for the genesis of polymer chemistry.41 Polymer chemistry was therefore well established before supramolecular chemistry came into being, and it had

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

8

Soft matter

been known since the very early developments that noncovalent interactions between polymer chains could have a significant impact on the material behavior of the resulting polymer.39 Strong interchain interactions raise the glass transition temperature (the temperature at which the polymer changes from being glassy to rubbery) and the melting point, as greater thermal energy has to be provided to the system in order to destroy the chain–chain interactions and allow the polymer backbone to become mobile. Noncovalent (supramolecular) interactions can also encourage polymer chains to align in a co-axial manner, which enhances their ability to act as fibers (and also their tendency to form mesoscale organized structures and liquid crystalline phases of the kind described in Section 3). Poly(amides) exemplify these principles (Figure 6). For example nylon[6,6], first developed by Carothers in 1935,42 is widely used in fibers for the manufacture of clothing, and has interchain hydrogen bond interactions. Kevlar, developed in 1965,43 replaced the flexible aliphatic linkers in the main polymer chain of nylon with rigid aromatic rings, limiting chain mobility even further and providing π –π stacking interactions between adjacent polymer chains, which can support the hydrogen bond interactions, allowing the polymer to withstand very high temperatures and physical impacts. As such, Kevlar is used in the manufacture of fireproof and bullet-proof clothing. The noncovalent interactions between the polymer chains therefore play a vital role in allowing these applications and controlling polymer behavior—as such, supramolecular organization

O

H N

N H

N H

O

O

H N O

H

N

N

O

O

N

H

H

O

O

H N

O

H N

N H

O

O N H

O

(self-assembly) of polymer chains was in active use long before the genesis of supramolecular chemistry itself. However, in spite of the broad applications of polymer chemistry, and the way in which it was radically changing the modern world, organic and supramolecular chemists generally neglected the field. The reasons for this are expressed, at least in part, in a note from Staudinger’s memoirs44 : “Those colleagues who were aware of my early publications in the field of low molecular weight chemistry asked me why I had decided to quit these beautiful fields of research and why I devoted myself to such disgusting and illdefined compounds such as rubber and synthetic polymers which at that time in view of their properties were referred to as grease chemistry (“Schmierenchemie”).” That is to say, somehow, polymer chemistry had gained the reputation of dealing with impure compounds, and therefore being, in some way, a lesser science, which, although of great industrial utility and practical application, was less extensively investigated in academic chemistry departments. Indeed, in many cases, separate departments dealing with polymer science would develop, along with companies specializing in the technology, rather than the activity being deeply embedded within all chemistry departments/companies. It is interesting to note that, without the intervention of polymer chemists, the development of “supramolecular” science may itself have been considerably delayed—Pedersen himself was a synthetic polymer chemist, trying to generate novel polymeric architectures, when he accidentally found that his systems cyclized in the presence of alkali metal cations (Crown and Lariat Ethers, Molecular Recognition).45

O O

H

H

N

N

O

O

N

H

H H

Nylon 6,6

Kevlar

POLICE

Figure 6

N

Nylon and Kevlar structures showing interchain hydrogen-bonding interactions.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc131

N

Soft matter science—a historical overview 9 Interestingly, however, the interface between polymer chemistry and supramolecular science was one of the earliest areas to give rise to materials with practical applications. In these materials, a three-dimensional polymer matrix is synthesized around a template with which the individual monomeric building blocks can interact. After removal of the template, the porous materials generated in this way can exhibit selectivity for the template (or other desired) molecule. Such molecularly imprinted polymers (MIPs) have applications in the development of sensors and separation science techniques. Interestingly, the first example of MIPs reported in the literature was in 1949 (using silica rather than organic polymers),46 predating the development of supramolecular chemistry. However, recent developments have significantly enhanced the sophistication and dramatically improved the performance of these materials (Molecularly Imprinted Polymers, Soft Matter).47 In recent years, living polymerization has radically changed the nature of polymeric products, and led to a resurgence of interest in using polymers as controllable/ programmable building blocks.48 Living polymerization, first demonstrated in 1956 for the anionic polymerization of styrene,49 is based on chain polymerization in the absence of termination steps. As such, the chain terminus retains its reactivity and is said to be “living.” The initiation step in living polymerization occurs with rapid kinetics, and, as such, and the polymers are therefore all initiated quickly, and grow smoothly until the remaining monomer is exhausted. The average chain length can therefore be predicted by the [monomer]/[initiator] ratio, and, because of the kinetics of the reaction pathways, the polymers that form are very well defined with low degrees of polydispersity (106 M−1 ). Key quantitative early work in the field of supramolecular hydrogen-bonded polymers was carried out by Meijer and Sijbesma et al.,5 Zimmermann, Park, and Nakashima,6 and Lehn, Fouquey, and Levelut,7 with each group developing exquisitely organized supramolecular motifs that contain multiple complementary hydrogen-bonding interactions. (Figure 1a–c, respectively). Unlike conventional covalent polymers, the DP of a supramolecular polymer may be altered in situ by application of a suitable stimulus, such as increasing temperature, that changes the Ka of the components (1). Removal of the stimulus restores the original properties of the material. It is this adaptable feature of supramolecular systems that makes them ideally suited for the production of healable materials. The following sections detail how the understanding gained from these fundamental, solution-state studies has been harnessed to produce adaptable materials that show healing characteristics.

HYDROGEN-BONDED, HEALABLE SUPRAMOLECULAR POLYMERS 2.2

2.1

(1)

Healable polymers with hydrogen-bonding motifs of high binding constant

Background

Because of its relative ease of synthesis from inexpensive starting materials, the ureido pyrimidinone (UPy) supramolecular motif has been widely exploited in recent years (Figure 1a). Appending UPy residues at the terminal positions of telechelic polymers such as end-functional polysiloxanes8 and polyethylene-co-butylene (pE-co-B),9 transforms previously free flowing, liquid starting materials into tough, elastomeric solids (Scheme 1).

The most highly developed family of supramolecular polymers harness hydrogen-bonding interactions to assemble end-functional oligomeric units into polymeric structures.3 Over the past 15 years, a great deal of work has focused on studying the properties of supramolecular hydrogen-bonded polymers in dilute solution (typically 10% w/v). Under these conditions, the degree of polymerization (DP) for the system, and thus the physical properties of the solution

R N O

N

H N

H N O

H

H N H N

N

O

O NH

N H

O H N

O

N H N

O

R

O

O N H O

H N

N

N

H N

H N N

N

O

H

N H

R

O N

R

N R

(a)

(b)

R

(c)

Figure 1 Supramolecular hydrogen-bonding motifs designed by the groups of (a) Meier and Sijbesma, (b) Zimmerman, and (c) Lehn, respectively.5–7 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc134

Self-healing and mendable supramolecular polymers

N O

N

H N H

O

O N H

pE-co -B

N H

H

N

N H

N

O

Hydrogen-bonded supramolecular polymer

Scheme 1 Formation of a supramolecular polymer containing “UPy” hydrogen-bonding residues.8 © Suprapolix bv

Heat

Dissolve

Cool

Dry Suprapolix

3

The well-established, if approximate, relationship between DP in solution and Ka (1) has driven the design of numerous multiple hydrogen-bonding motifs with very high binding constants. This approach to supramolecular polymer design has led to the development of lowto medium-MW materials that possess solution viscosities similar to those of high-MW covalent polymers. In the solid state, however, the connection between the DP and association constant and therefore physical properties is less well understood, and other factors, such as microphase separation and crystallinity may play equally important roles in determining the physical characteristics of the material. The following section details the emergence of supramolecular polymers that exploit hydrogen-bonding residues with weaker and less well-defined binding modes to produce thermo-responsive materials that show potential as healable materials.

2.3

Healable materials based on weak hydrogen-bonding motifs

Figure 2 Example of the phase change behavior between the melt, solid, and solution states of Suprapolix materials. (Reproduced with permission from Ref. 10.  Elsevier, 2004.)

2.3.1 Thermoplastic elastomers with in-chain hydrogen-bonding residues

Materials harnessing the reversible nature of UPy hydrogen-bonding residues are currently (2010) produced on an industrial scale by the company Suprapolix BV .10 These materials are described as being readily soluble in a range of solvents, which aids processing, but they form tough, elastomeric films on removal of the solvent (Figure 2). Suprapolix have visually demonstrated crack-healing in the surface of the supramolecular material by heating the sample to 140 ◦ C. At this temperature, the material is able to flow and fill the fracture-void without further external intervention (Figure 3), regenerating the flat surface of the pristine material.

In a series of papers beginning 1990, Stadler and coworkers studied the effect of introducing multiple, weak hydrogen-bonding residues into a nonpolar polymer such as polybutadiene (pB).11–16 They designed a polymerizable, benzoic acid-modified, urazole hydrogen-bonding motif (Figure 4). The X-ray structure of the urazole monomer 1 revealed a continuous, two-dimensional hydrogen-bonding array created by dimerization of the carboxylic acid groups and by analogous interactions between the urazole moieties. The resulting hydrogen-bonded layers were supported in the third dimension by dipole–dipole interactions between the carbonyl residues of the urazoles (Figure 4).

Suprapolix

115 micron

Figure 3 Thermally induced crack-healing of a Suprapolix “UPy”-based supramolecular polymer (left hand micrograph taken at 25 ◦ C, right hand image at 140 ◦ C). (Reproduced with permission from Ref. 10.  Elsevier, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc134

4

Soft matter O

O H

O

N N

H O N

N O H

O

Hydrogen-bonded carboxylic acid dimer

N N

N

HN H

N

O O

O H

CO2H

O

O

O 1

N N

O N O H

N NH

N HO2C

O

O Hydrogen-bonded urazole dimer

Figure 4

Dipole–dipole interactions between carbonyl groups in different layers

Supramolecular interactions identified by X-ray analysis of the crystal structure of a urazole monomer.14

Copolymerization of the hydrogen-bonding monomer 1 (Figure 4) with butadiene was achieved by conventional anionic polymerization. Copolymers with low levels of 1 ( 0.5) favors the formation of flatter interfaces, and samples can be altered from spheres to lamellae. This can be explained by the following rationale: For a symmetrical diblock copolymer (f = 0.5), the most stable conformation is formed when a planar interface is created between the polymer domains. However, for compositionally asymmetric diblock copolymers, a planar interface is no longer stable as the larger polymer component would have to stretch further than the shorter component, generating an entropic penalty. To overcome this penalty, a curved interface is adopted where the larger component resides on the convex face of the curvature. However, curvature comes at both an enthalpic and entropic cost; as a curved surface creates a larger interfacial area between two incompatible domains (enthalpy) and the shorter chains must stretch to fill the center of their domains (entropy). When the entropy “save” associated with curvature overcomes these two penalties, curved interfaces will be observed. Changing the PDI of one of the blocks shifts the balances of these competing forces and determines which forces become the most dominating (producing more or less curvature in the system). Some studies have shown that increasing PDI can lead to coexisting stable morphologies, such as lamellae, cylinders, and spheres all in the same polymer sample.22 Caution is needed when extrapolating conclusions from such reports, however, as the systems under scrutiny were actually blends of a number of monodisperse BCPs and not single polymer samples with continuous (unimodal) molecular weight distribution. An elevation in PDI is known to increase the domain spacing within a nanostructure; the largest chains partition themselves within the center of a given domain, whereas the shorter chains reside closer to the interface, as illustrated in Figure 8. This has been shown both experimentally and theoretically in a wide number of reports across the literature.23–27 Finally, although increasing the PDI of a BCP will increase the phase space in which ordered nanostructures are observed, it will be more difficult to target specific volume fractions (and thus morphologies) when using

d1

7

d2

Monodisperse

Polydisperse

Figure 8 Cartoon schematic to illustrate the effect of polydispersity on domain spacing.

uncontrolled synthetic techniques that are typical for producing polydisperse materials.

3

SOLUTION MICROPHASE SEPARATION

BCPs can also undergo microphase separation in solution.28–30 Here, phase separation occurs when the different blocks have different affinities for the solvent, or solvents, that are used. At a basic level, the phase separation is driven by the differing selectivity of one or more blocks for the solvent. More generally, when a solute molecule is dissolved in a solution, there is an entropy gain due to mixing. There is also an increase in the interaction free energy as the solute molecules generally prefer to be surrounded by other solute molecules as opposed to solvent molecules. Competition between the entropy gain and free energy increase determines whether the solute is dissolved or phase separation occurs. In water, for example, it is more energetically favorable for the hydrophobic blocks to pack together than it is for water to order around the individual hydrophobic chains. As a result, the entropy gain on adding a hydrophobic solute is normally not sufficient to overcome the large increase in free energy and phase separation is most common. However, for an amphiphilic molecule (or here, a BCP) that has one segment (or block) that is hydrophilic, a minimum in free energy can be achieved such that bulk phase separation does not occur. Rather, microscopic phase separation into well-defined aggregates that are stabilized by the hydrophilic segment occurs. Phase separation can, of course, occur in any selective solvent, not just water. However, in a polar solvent, micelles tend to form at lower temperatures, implying that, for these cases,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

8

Soft matter

(a)

(b)

(c)

Vesicles

100 nm (d)

(e)

(f)

100 nm 0.2 µm

Figure 9 Top: cartoon representations of a spherical micelle, a wormlike micelle, and a vesicle. The red blocks represent the solvophilic blocks, and the blue blocks represent the solvophobic blocks. Bottom: example TEM images showing different micelle morphologies adopted by block copolymers in solution. (a) Spherical micelles formed from poly(ethylene oxide)-b-polycaprolactone (PEO-b-PCL) copolymers.31 (b) Wormlike micelles, vesicles, and “octupi” formed by mixing PEO-b-polybutadiene (PEO-b-PB) block copolymers. (Reproduced from Ref. 32.  American Chemical Society, 2004.) (c) Vesicles formed from PEO-b-PCL copolymers. (Reproduced from Ref. 33.  Royal Society of Chemistry, 2011.) (d) Multicompartment micelles formed from a triblock copolymer. (Reproduced from Ref. 34.  American Chemical Society, 2010.) (e) Stomatocytes formed using PEO-b-polystyrene (PEO-b-PS) copolymers. (Reproduced from Ref. 35.  American Chemical Society, 2010.) (f) Toroidal micelles coexisting with cylindrical micelles and spherical micelles formed from poly(acrylic acid)-b-poly(methacrylic acid)-b-PS (PAA-b-PMA-b-PS) triblock copolymers. (Reproduced from Ref. 36.  Royal Society of Chemistry, 2009.)

micellization is driven by enthalpic forces. On micellization, a range of different morphologies have been reported, including spherical micelles, wormlike micelles, vesicles, helical structures, toroids, multicompartment micelles, and discs. Schematic representations of the most common morphologies (spherical micelles, wormlike micelles, and vesicles) as well as example TEM images of various morphologies are shown in Figure 9. Before considering BCPs in detail, it is worth reflecting on the phase behavior of surfactants (see Self-Assembly of Facial Amphiphiles in Water and Soft Matter Science—a Historical Overview with a Supramolecular Perspective, Soft Matter). Classically, a simple geometric

approach is often used to relate the morphology adopted by amphiphilic molecules in solution, with surfactants being prototypical. This was initially developed by Israelachvili and relates the morphology adopted to the “shape” of the molecule in solution.37 For an amphiphile in water, for example, the major forces at play are the hydrophobic attraction between the hydrophobic groups and the electrostatic or steric repulsion between hydrophilic head groups. Hydrophobic attraction will be determined by the length (l) and the volume (v) of the hydrophobic groups. The magnitude of the repulsive forces between head groups relates to the interfacial area of the molecule (ao ). On the basis of this, a nondimensional packing parameter (P ) can be

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Assembly of block copolymers ao

v

P = 1/3

1/3 < P < 1/2

1/2 < P < 1

P>1

Figure 10 The packing parameter, P , estimates the timeaveraged molecular shape of a surfactant.

defined, such that P =

v ao l

When P < 1/3, individual molecules are conically shaped, as shown in Figure 10. This results in the formation of spherical micelles in solution. As the volume of the hydrophobic tail is increased, P increases. For 1/3 < P < 1/2, nonspherical (cylindrical) micelles are formed. As P increases further, bilayer structures are formed. At P > 1, inverted structures are formed. This packing parameter can be used to rationalize why sodium dodecyl sulfate (SDS) forms spherical micelles in solution, while lysolecithin forms wormlike micelles. While there are similarities between the assembly of surfactants and that of amphiphilic BCPs, there are also significant differences. The differences essentially arise due to the presence of chain entanglements and the reduced mobility of the polymer chains in the assembled structures; both of these are due to the higher molecular weight of amphiphilic BCPs as compared to small molecule surfactants. For small molecule surfactants, micellization occurs above a critical concentration [the critical micelle concentration (cmc)]. This is typical of a closed association process whereby below the cmc only molecularly dissolved unimers are present in solution. Above the cmc, unimers are in equilibrium with micelles. This equilibrium allows for transfer of surfactant molecules between micelles. For BCP systems, while the assembled structures formed can be similar to those formed by surfactants, the cmc is usually much lower. In general, the cmc decreases as the absolute molecular weight of the copolymer increases and as the balance of solvent-selective to nonsolvent-selective block decreases. This has been demonstrated, for example, for a large number of poly(oxyalkylene) amphiphilic BCPs.38 Also, the dynamics of exchange of copolymers between aggregates is much slower than for classical surfactant aggregates. Despite the prevalence of cmc measurements, there is significant evidence in many examples that the transfer of chains between micellar structures is extremely slow and

9

does not occur over the timescale of most experiments. This is especially true where the BCP molecular weights are high, so the overall equilibrium drops dramatically and there is a high energy barrier to redissolution. There are many examples of so-called “frozen” structures from BCPs where the hydrophobic block is a glassy polymer such as poly(styrene). In cases where exchange is very slow, the aggregation numbers of the structures that are formed are determined by those at the first point of phase separation; hence, it is possible to trap out of equilibrium nonergodic structures (i.e., the system has failed to achieve global equilibration) in addition to coexisting morphologies. This is because the only reorganization that can take place is that within an aggregate, rather than equilibration between aggregates. As a result, it has been shown that micelle morphology can be controlled based on how two polymers are mixed.32 If they are mixed before assembly, different structures are formed than if the polymers are mixed on partial assembly.33 Additionally, because of the slow exchange kinetics, metastable structures have significantly longer lifetimes than in the cases of low-molecular-weight structures. In these cases, unsurprisingly, the size and shape of the aggregates formed are determined by the process in which the polymers are dispersed and are kinetically controlled. For amphiphilic BCPs, rather than considering molecules with a hydrophobic tail and a hydrophilic head group, the most studied class is that of the coil–coil BCPs. Differences in solubility of the two blocks in a selective solvent result in repulsive interactions, inducing phase separation. This can happen in organic solvents or water as noted above. The solvent-selective block orientates into the solvent, and the nonselective block forms solvent-poor (or excluded) domains. Phase separation therefore leads to micellization, reminiscent in concept to surfactant assembly. For amphiphilic diblock copolymers, depending on the relative lengths of the solvent-selective and nonselective blocks, the aggregates formed can be referred to as starlike, where the solvent-selective block is the longest, or crew-cut micelles, where the nonselective block is the much longer block. The morphology of the aggregates formed from dispersion in a selective solvent is controlled by the balance between a number of factors including the degree of stretching of the core-forming blocks, the interfacial tension between the micelle core and the solvent outside the core, in addition to the repulsive interactions among corona-forming blocks. It therefore follows that the adopted morphology can be controlled by varying these parameters, which can be achieved by varying the BCP architecture, the composition (in terms of block ratios, block lengths, or block chemistry), the copolymer concentration, the solvent composition, and the presence of additives such as salts, surfactants or other polymers, and temperature.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Soft matter

As described above, the packing parameter is widely used to explain the morphology adopted by amphiphilic molecules. For amphiphilic BCP systems, however, the head groups are generally much bulkier than those for surfactants. Also, the hydrophobic tails generally have a more complex chemical nature than the alkyl chains often used for surfactants. Hence, the above description is perhaps somewhat restrictive, although it is still commonly cited. Instead, a simple concept that has been used for BCP systems is that of relative volume or weight fractions for the different blocks.39 This correlates with concepts used in the bulk (see Section 2). Again, as for the packing parameter, this concept attempts to capture the time-averaged shape of the polymer and explain the morphology adopted. In general, the morphology adopted can be linked to the relative volume fractions of the solvent-selective to nonselective blocks. As the relative volume fraction of the solventselective block increases, curvature is forced into the assembled structure. Spheres, wormlike, vesicular, toroidal, helical, and disclike micelles have all been reported as aggregates formed on the micellization of BCPs. Such structures have been formed from diblock, triblock, multiblock, and graft copolymers, among others. Toroids are relatively rare. Wormlike micelles are uncapped, this requires the formation of more interfacial area. The copolymers in the end-capped region only represent a small fraction of the whole, however. At higher molecular weights, this end-capping region becomes less stable. As a result, toroidal or other closed systems become more favorable.40 On the basis of a series of poly(ethylene oxide)-bpoly(butadiene) (PEO-b-PB) BCPs, the morphology of the assembled BCPs in water was correlated with the ratio of the hydrophilic mass to total polymer mass, whydrophilic (Figure 11).40 Vesicle structures were found for polymers where whydrophilic < 35%. Cylindrical wormlike micelles were found 35% < whydrophilic < 50%, with spherical micelles being found with whydrophilic > 45%. These morphological windows were found to hold for PEO-b-PB copolymers over a range of different molecular weights. From this, it has been stated that, as a general rule, copolymers with a hydrophilic to hydrophobic weight ratio (whydrophilic ) of >1 : 1 usually form micelles, copolymers with whydrophilic < 1 : 2 usually favor vesicle formation, and those with whydrophilic < 1 : 3 may form vesicles, other complex structures, or macroscopic precipitates. Similar links between the volume or weight fraction of PEO have been described for BCPs with other hydrophobic groups.31, 41 Micelles are formed for those polymers with relatively high fEO values giving way to wormlike micelles at intermediate fEO , often coexisting with spherical micelles and, at low fEO , vesicles are formed. However, it must be stressed that these ratios do not provide definitive design rules. The positions of the boundaries between

the different morphologies do not necessarily hold across different polymer systems. For example, for poly(ethylene oxide)-b-poly(caprolactone) (PEO-b-PCL) copolymer systems,31 the boundaries were found to be at lower fEO than the value expected when compared to literature data for PEO-b-PB copolymers.42 For PEO-b-PCL copolymers, which are widely investigated due to the biocompatibility of both blocks, recent studies introduced the use of the reduced tethered density, defined as σ = 1/SPEO πR2g , which links SPEO , the area of the block interface available per PEO chain (determined by the PCL core) to the cross-section of the solvated PEO chain, πR2g to enhance the prediction of the adopted morphology in such systems.43 σ is a measure of the crowding of the PEO chain at the PEO–PCL interface. Recently, for polymers such as PEO-b-PCL, it has been suggested that the two oxygens per hydrophobic caprolactone monomer unit contribute a degree of hydrophilicity and this should be taken into account when calculating the overall hydrophilic volume or weight fractions.44 Here, it was also found that wormlike micelles were only observed within a narrow subphase of the vesicle region, rather than intermediate between spherical micelles and vesicles as was observed elsewhere31 for PEO-b-PCL (A)

(C)

(B)

200

N

160

B+CY

CY

120

NPB

10

B

C

C+S

S

80 B+C 40

0 0.2

0.3

0.4

0.5

0.6

0.7

WPEO

Figure 11 Morphology diagram for PB-b-PEO in water (concentration of 1 wt%) as a function of molecular size and composition. NPB is the degree of polymerization of the PB block and wPEO is the weight fraction of the PEO blocks. Four basic morphologies were identified: bilayers (B), Y junctions (Y), network, N cylinders (C), and spheres (S). (Adapted with permission from Ref. 40.  American Association for the Advancement of Science, 2003.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Assembly of block copolymers polymers. This may be due to the different methods used to assemble the polymers. PCL is a semicrystalline polymer, and it is not clear how the different methods of assembly affect the crystallinity in the final assembled structures. Linking aggregate morphology to the BCP composition is further complicated by the fact that the process of self-assembly is a critical parameter, with the same BCP assembling into different morphologies depending on the assembly process. For example, Sachl et al. have reported that PEO113 -b-PCL56 forms vesicles45 while other reports suggest that this polymer forms micelles.31 A number of different methods are often used to prepare selfassembled structures in solution from amphiphilic BCPs. These include dissolving the polymer in a solvent which is a good solvent for both blocks and then changing the solution conditions such that one block becomes poorly solvated. This can be achieved, for example, by adding an antisolvent for one block or varying the solution temperature. A common method here is dialysis where the solvent composition is slowly altered from a common solvent to a selective solvent. Alternatively, the polymer can be dispersed in a solvent that is selective for only one block. Here, the soluble block must provide sufficient solubility to allow the dispersion and reorganization of the BCP. This latter method can be problematic. Depending on the solubility of the BCP in the solvent, it is possible that the resulting selfassembled aggregates depend on the phase-separated bulk morphology. Also, if the solvent does not swell the nonsoluble block, then it is unlikely that an equilibrium structure will be achieved on a reasonable experimental timescale. For all methods of BCP self-assembly, it is important to consider whether an equilibrium structure is reached. If the hydrophobic block has a high-glass transition temperature, kinetically “frozen” structures are formed. An example of high-glass transition temperature hydrophobic polymer is poly(styrene); there are many “frozen” micelle structures in the literature prepared from PEO-b-PS and PAA-b-PS BCP systems.35, 46, 47 It has recently been highlighted that so-called frozen systems are more common for BCPs with such glassy, hydrophobic blocks.48 Frozen micellar structures have also been reported for BCP systems where the hydrophobic block is a low Tg polymer such as poly(n-butyl acrylate), whose Tg ∼ −50 ◦ C. The energy barrier that must be overcome to allow transfer of polymers between aggregates is determined by the interfacial tension between the solvophobic block and the solvent. In water, the interfacial tension is often very high, leading to many frozen systems. The interfacial tension can be tuned by solvent composition, for example, by adding organic solvents to the water. An alternative approach to encourage equilibration between structures is to reduce the hydrophobicity of the hydrophobic

11

block. This has been demonstrated, for example, by copolymerizing n-butyl acrylate with tert-butyl acrylate, followed by deprotection of the tert-butyl groups to afford a mixed poly(n-butylacrylate-co-acrylic acid) hydrophobic block.49 Where one of the blocks is crystalline or semicrystalline, it can be observed that the morphology adopted in solution does not necessarily correlate with that expected from the weight ratio of the two blocks. For example, the Manners group has reported numerous examples of BCPs with a semicrystalline poly(ferrocenylsilane) (PFS) block.50 For PI-b-PFS BCPs dispersed in hexane (selective for the PI block), cylindrical structures were adopted even where the block ratios might suggest that spherical micelles would be formed. For example, PI320 -b-PFS53 , with a block ratio of 6 : 1, formed cylindrical structures. This deviation from the expected morphology was explained on the basis of the crystallization of the PFS block in addition to the intercoronal repulsion between the coronal PI blocks. The crystallization of the PFS block was demonstrated by WAXS experiments. This result is hardly surprising as the driving force of crystallization is well known to affect the microphase separation of rod–coil BCPs both in solution (as discussed elsewhere51 ) and in the bulk (see Section 2.2). Similarly, for poly(ferrocenyldimethylsilane)b-poly(dimethylsiloxane) (PFDMS-b-PDMS), when micelles were formed below the melting temperature of bulk PFDMS, cylindrical morphologies were always observed, as shown in Figure 12.52 However, when micelles were formed above this melting temperature, spherical micellar aggregates were formed. Additionally, if the PFDMS block was replaced with a related, but amorphous, block (poly(ferrocenylmethylphenylsilane)), spherical micelles were observed.

3.1

Concentration effects

The majority of the discussion in this section has concerned morphologies formed in dilute solutions. Typically, such aggregation is examined at copolymer concentrations of 1 wt% or less. However, concentration is a key parameter in controlling the aggregates formed by BCPs in solution. At higher concentrations, further assembly of the micellar aggregates occurs. Spherical micelles can assemble into cubic-packed structures and wormlike cylindrical micelles into hexagonal-packed structures.38 Lamellar structures are formed at higher concentrations of vesicles. Phase transitions on dilution are complex, with concentrated lamellae often forming a sponge phase, before forming vesicles. Battaglia and Ryan have shown that the dilution of a bulk lamellar gel of a poly(ethylene oxide)-b-poly(butylene oxide) (PEO-b-PBO) BCP results in a series of phase transitions from the lamellar gel to an interconnected sponge

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

12

Soft matter

750 nm

125 nm

(a)

(b)

Figure 12 (a) Transmission electron micrograph of cylindrical micelles of PFDMS50 -b-PDMS300 prepared by the slow addition of hexane to a THF solution and subsequently aerosol sprayed onto a TEM grid. (b) Transmission electron micrograph of spherical micelles of PFDMS50 -b-PDMS300 prepared above the Tm of the PFDMS block (aerosol sprayed from an n-hexane solution). (Adapted from Ref. 52.  American Chemical Society, 2000.)

phase to hexagonally packed vesicles to finally yield dispersed vesicles (Figure 13).53 Elsewhere, for PS-b-PAA BCPs where assembly occurs as water is added to a solution of the polymer in an organic solvent, Eisenberg has demonstrated that the morphology adopted is concentration dependent. In terms of the packing parameter or weight fraction considerations, increasing the concentration is not a parameter and can be considered constant (as there are no changes in either the solvent or the copolymer composition). Hence, the change in morphology must be a result of the increase in aggregation number per micelle as the concentration is increased. This increase in aggregation number requires that the PS chains must stretch more, resulting in an entropic penalty and a change in the morphology. So, for a range of PS-b-PAA BCPs, at a constant water content, an increase in polymer

concentration results in a change in the morphology that is equivalent to adding more water into the system.54

3.2

Effect of additives

Depending on the system, the addition of various additives can affect the morphology. This is especially the case where the hydrophilic block is a polyion and the presence of salts screens the charges.55 It has also been shown that salts can affect the assembly of PEO-based copolymers.38 Depending on the nature of the salt, “salting-out” effects can be observed, which result in the PEO chain becoming less soluble thus leading to an effective decrease in the weight fraction of the hydrophilic block.41 These effects are related to the Hofmeister.56

Gel

Liquid Lyotropic Sponge phase

Lamellae Swelling

Isotropic

Hexagonal-packed vesicles

Single-dispersed vesicles

Unbinding Solid

100%

10%

1% Wcopolymer (%)

0.1%

0.01%

Figure 13 The phase diagram of PEO115 -b-PB103 at various concentrations in water. (Adapted from Ref. 53.  Nature Publishing Group, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Assembly of block copolymers

3.3

Polydispersity

3.4

One major difference between BCPs and small molecule systems is the concept of polydispersity. However, there are not many studies that consider polydispersity as a factor. Eisenberg demonstrated that polydispersity in the PAA block of a PS-b-PAA BCP affected vesicle size.57 This was explained in terms of the BCPs segregating within a vesicle such that the longer PAA blocks preferentially sat on the outside of the vesicles, with the shorter PAA chains segregating to the inside. As a result, increasing the PAA polydispersity resulted in a decrease in vesicle diameter. However, here, the polydispersity was artificially increased by mixing polymers with low polydispersity. As a result, a bimodal system was used as opposed to a continuous molecular weight distribution. Elsewhere the assembly of PEO-b-poly(N,N-diethylaminoethyl methacrylate) (PEO-b-PDEAMA) BCPs with relatively high polydispersities (1.36 < PDI < 1.96) were examined.41 These polymers had monomodal weight distributions and, as they were prepared using PEO-based initiators, the majority of the polydispersity resided within the controllably hydrophobic PDEAMA block. Qualitatively, the same trend was observed as for other systems with micelles being formed at higher fEO , wormlike micelles being formed at intermediate fEO , and vesicles being formed at low fEO . In short, polydispersity appeared to little or no effect on the observed morphologies in solution. Similar micellar structures were formed when compared to other polymers prepared with low polydispersity by other authors.58 In all of these, it has to be noted that where the polydispersities are considered, it may be that calculating the hydrophilic volume or weight fractions can only represent an average of the system. A logical extension on increased polydispersity is to use mixed polymer systems. As noted above, nonergodic structures can arise depending on the mixing method (Figure 14).32, 33

100 nm (a)

(b)

Figure 14 Self-assembly of (a) postmixed or (b) premixed PEO-b-PB samples can lead to nonergodicity. (Adapted from Ref. 32.  American Chemical Society, 2004.)

13

Stimuli-responsive block copolymers

One important subclass of BCPs is stimuli-responsive BCPs (see Stimuli-Responsive and Motile Supramolecular Soft Materials, Soft Matter). On application of an external trigger, these blocks change their solvophilicity and can assemble in solution.59 Examples include pHtriggerable systems, such as poly(N,N-dialkylaminoalkyl methacrylates), which are hydrophilic at low pH where the amino groups are protonated and hydrophobic at high pH where the free amine is present, and thermo-responsive systems, such as poly(N -isopropylacrylamide), which is hydrophilic at low temperature and becomes hydrophobic at ∼32 ◦ C. Using these concepts, BCPs have been synthesized that are hydrophilic under one set of conditions and then amphiphilic under another. Micellization then occurs as the polymers become amphiphilic. Further complexity arises from double-stimuli-responsive BCPs. As one example from many, pH- and temperature-responsive BCPs have been prepared from poly(2-(dimethylamino)ethyl methacrylate)-b-poly(glutamic acid) (Figure 15).60 Both blocks are pH sensitive, the dimethylaminoethylmethacrylate block is positively charged and hydrophilic at low pH, whereas the glutamic acid block is neutral and hydrophobic. Under basic conditions, this is reversed. Hence, at low pH, structures are formed, stabilized by a hydrophilic dimethylaminoethylmethacrylate block. At high pH, structures are formed which are stabilized by a hydrophilic negatively charged poly(glutamic acid) block. This is an example of a so-called schizophrenic BCP.61 Additionally, the dimethylaminoethylmethacrylate block exhibits a lower critical solution temperature (LCST), being hydrophilic at room temperature, but hydrophobic above 40 ◦ C. Hence, in these cases, the assembled structure depends on the pH and temperature of the system.

3.5

Mechanism of vesicle formation

Polymer vesicles have received significant attention in recent years, especially from the perspective of encapsulation. They have been shown to have much stronger walls as compared to lipid vesicles. The wall thicknesses can also be controlled to some degree via the molecular weight of the hydrophobic block used. Vesicles prepared from phospholipids or other small molecule surfactants comprise walls where the molecules form either a well-defined ordered bilayer or an interdigitated structure. The situation is more complicated for BCPs, where demixing of the hydrophobic blocks of the bilayer is entropically unfavorable and it has been questioned how close the analogy can be between phospholipid and polymer systems.62 The degree of stretching of the hydrophobic blocks can be related to the vesicle

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

14

Soft matter

Thermo responsive micelles

DPPGA

High temprature

186

DPPGA

77

DPPGA

Thermo responsive vesicles

186 37

Soluble complexes

186 77

Insoluble complexes

Room temprature

77

37

Free chains

2

3

Electrostatic vesicles

4

5

Free chains

6

7

8

9

10

11

pH

Figure 15 Schematic representation of the different morphologies obtained from poly(2-(dimethylamino)ethyl methacrylate)-bpoly(glutamic acid) block copolymers at different pH and temperature. (Reproduced from Ref. 60.  American Chemical Society, 2010.)

wall thickness, t. This will vary with the degree of polymerization of the hydrophobic block, N, such that t ∼ N b , where b is related to the segregation strength. For BCPs in the strong segregation regime, where the hydrophobic and the hydrophilic blocks are well separated (high χ value) and the chains strongly stretched, b = 2/3. In the weak segregation regime (low χ ), equivalent to the hydrophobic block adopting a random coil, b = 1/2. Examples of vesicles exhibiting both exponents exist, depending on the block chemistry. For PEO-b-PB BCPs, an exponent of b = 1/2 was found.63 This was attributed to partial collapse of the PEO chains toward the interface, shielding the hydrophobic membrane. On the other hand, for PEO-b-PBO, b = 2/3 was found.62 Nevertheless, the absolute thickness of the walls was found to be lower for the PEO-b-PBO BCPs as compared to the PEO-b-PB examples for the same value of N, as expected for the more flexible PBO block. For a range of PEO-b-PCL BCPs, the membrane thickness scaled such that b = 0.62, close to 2/3, the theoretical value for the strong segregation regime.31 From the perspective of encapsulation, the mechanism by which vesicles form is of critical importance. The most accepted mechanisms of vesicle formation in BCP systems are either from bilayer fragments64 or by a sequence of intermediates, where spherical micelles are formed, then wormlike micelles, which then form flattened disclike micelles, which close up and form vesicles.31, 65 Recently, another mechanism has been postulated where small spherical micelles are initially formed because the segregation of

the two blocks lags behind copolymer aggregation. These grow to large, energetically unfavorable spherical micelles. They therefore restructure to give vesicles.66, 67 The mechanism by which vesicles are formed depends on the process by which self-assembly is carried out. The principal methods of vesicle formation, as for other morphologies, involve either phase inversion (by dissolving the polymer in a suitable organic solvent, then exchanging the solvent for water by dialysis or by solvent evaporation from a solvent/water mixture) or a solvent-free route by hydration of the BCP from a surface using agitation, sonication, or an electric field to encourage hydration. The mechanism by which vesicles are formed is very different between these methods and impacts directly on the final utility of these structures for encapsulation. Additionally, it is worth noting that the absolute size of the vesicles formed often depends on the process of their formation. The solvent-free route to vesicle formation requires the diffusion of water into a bulk film of the BCP as well as the diffusion of the BCP into the aqueous phase. For PEO-b-PBO BCPs, the swelling has been reported to initially follow a subdiffusional growth.68 After a certain transition time, the swelling follows a normal diffusional growth. As the water diffuses into the film, a series of phase changes occurs from rods to lamellae to a sponge phase and finally to isolated vesicles, all of which occurs within the film itself. Three different molecular weight polymers were examined. In each case, the subdiffusional growth phase existed. Conventional diffusional growth then

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Assembly of block copolymers occurred once the final equilibrium structure had formed. Here, the rate of diffusion was related to the molecular weight of the polymer, with the higher molecular weight polymers exhibiting a drop of three orders of magnitude in diffusion rate for a single order of magnitude increase in molecular weight. At the surface of the polymer film, finger instabilities known as myelins are formed. These then destabilize to form vesicles in solution. To increase the rate of vesicle formation, a constant concentration gradient is required at the copolymer/water interface. This can be maintained by an energy input such as an AC electric field in the electroformation process.68 Recently, Howse et al. demonstrated the spontaneous formation of vesicles via hydration and dewetting of a thin BCP film from a patterned surface.69 The pattern determines the maximum vesicle size. Vesicles form via hydration of the BCP leading to microphase separation into lamellae, followed by budding from the surface (Figure 16). A low-molecular weight, relatively hydrophilic BCP was used (PEO16 -b-PBO22 , Mw = 2300 Da). Solvent composition-driven assembly represents an interesting process. This highlights the difficulties in assuming a simple packing parameter or volume fraction approach to morphology prediction. In this case, the BCP is initially dissolved in a solvent that is nonselective. A selective (a)

(b)

15

nonsolvent is then added, acting as a precipitant for one of the blocks. Because the organic solvent can partition between the water and the hydrophobic core or the wall of the aggregates, the chain dynamics of BCPs are expected to depend strongly on the solvent composition. An example is the addition of water to a solution of PEO-b-PCL in tetrahydrofuran (THF). The process was followed by turbidity and cryo-TEM, as shown in Figure 17.31 The assembled structure formed in solution depends on the solvent composition. Initially, the copolymer is present as unimers in THF. At a critical water concentration, polymer assembly begins. Initially, aggregation into spherical micelles occurs. As more water is added, the spherical micelles aggregate further, resulting in the formation of wormlike micelles. Further addition of water results in a further morphology switch, resulting in the formation of vesicles. Beyond this point, further addition of water does not lead to another switch in the morphology. These morphological transitions can be explained on the basis of changes in core stretching, interfacial tension, and repulsion between coronal blocks. As water is added, and the overall solvent quality decreases for the PCL block, the interfacial energy between PEO and PCL increases. To reduce the interfacial area, there is an increase in aggregation number of an individual micelle, with a corresponding decrease in the overall number of

Hydration

(c)

L

(d)

(e)

(f)

dmax

10 µm

Figure 16 Upper: schematic representation of the controlled formation of vesicles. (a) Resulting drop profile following dewetting. (b) Hydration resulting in microphase separation—hexagonal rod phase (blue: hydrophilic, red: hydrophobic). (c) Further hydration at the surface resulting in surface lamellae and further internal phase separation. (d) Expansion of exterior bilayer. (e) Detachment. (f) Surface minimization leading to closure and vesicle formation. Lower: a single vertical slice showing a series of vesicles “budding” from the surface. (Adapted from Ref. 69.  Nature Publishing Group, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

16

Soft matter

120

Transmittance

100

(B)

(C)

80 60 (F)

40

(A)

20

(E)

0

(D)

0

20

40

60

% H2O

(a) (A)

(B)

(C)

(D)

(E)

(F)

(b)

Figure 17 (a) Turbidity plot for PEO23 -b-PCL28 at 25 ◦ C obtained by adding water dropwise to a solution of polymer in THF. (b) Points (A)–(F) show points at which solutions were analyzed by cryo-TEM. (Adapted from Ref. 31.  Royal Society of Chemistry, 2009.)

micelles. However, this results in an increase in the stretching of the chains in the core and repulsion between coronal chains, both of which are thermodynamically unfavorable. At some point, the driving force to reduce the interfacial tension is exceeded by the thermodynamic penalties and a morphological transition occurs from spherical micelles to wormlike micelles. This process is repeated until a worm-to-vesicle transition occurs. These transitions can be thought of in terms of a gradual change in the packing parameter or weight fraction of hydrophilic block, with key values of these resulting in morphological shifts. Similar morphological changes have been reported on addition of water to a solution of PS-b-PAA BCPs.65 It is clear that this process will be dependent not only on the BCP but also on the solvent initially used to dissolve

the polymer. Noticeably, depending on the solvent, the coil dimensions of both blocks can vary. Also, depending on the miscibility of the solvent, selective partitioning of the solvent may occur. Finally, the rate of change in the coil dimensions on addition of the precipitating solvent will depend on the compatibility between the solvent and the polymer, that is, χ polymer–solvent . As an example, for PS-b-PAA BCPs, the choice of the initial solvent was shown to be critical. Adding water to a dimethylformamide (DMF) solution of PS500 -b-PAA58 resulted in the formation of spherical micelles.47 However, when either THF or 1,4-dioxane was used, vesicles were formed. This was explained in terms of the solubility parameters of the solvents; those of THF and dioxane are closer to that of polystyrene compared to that of DMF. Hence, it is expected

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Assembly of block copolymers

OEGA

O

O

EHA

O

O

n~8.5

CH3

(b)

FA

O

O (CF2)7

O

(a)

17

CF3

(c)

Figure 18 Cryo-TEM micrographs of micellar aggregates in 0.5 wt% aqueous solutions of amphiphilic triblock copolymers (a) core–shell micelles formed from (OEGA)70 -(EHA)140 -(FA)13 . Scale bar = 50 nm. (b) “Patched” multicompartment micelles formed from (EHA)120 -(OEGA)109 -(FA)23 . Scale bar = 50 nm. (c) Monomers used for the synthesis of triphilic ABC block copolymers. (Adapted from Ref. 34.  American Chemical Society, 2010.)

that the PS in PS500 -b-PAA58 will be better solvated in THF or dioxane compared to that of DMF. Control over the size of vesicles prepared from PS-bPAA BCPs has been demonstrated by Choucair et al.46 Here, for a given polymer, the vesicle size could be varied by the addition of HCl or NaCl (thereby affecting the charge on the PAA block and hence the repulsive interactions between coronal blocks; this allows the aggregation number and therefore vesicle size to be controlled). The length of the coronal PAA block also affected the vesicle size, with shorter PAA chains resulting in larger vesicles. This was explained in terms of the molecular weight distribution becoming wider at higher molecular weights for a fixed polydispersity. Consequently, at higher molecular weights, the difference in length between the longest and shortest chains increases. As noted above, segregation of PAA blocks was observed elsewhere such that the shorter PAA blocks were found on vesicle interiors and the longer chains at the exterior, aiding curvature.57 Finally, these vesicles were prepared via the addition of water to a solution of the PS-b-PAA BCP in an organic solvent. The nature of the solvent was also found to affect the vesicle diameters for a fixed copolymer composition. This was explained in terms of the degree of swelling of the hydrophobic block by different solvents and solvent mixtures as well as the degree of ionization of the PAA block in different solvents. Hence, the solubility parameters and dielectric constants of the solvents are critical. The growth in size of polymer vesicles formed from PS-b-PAA BCPs was attributed to fusion and fission mechanisms by Eisenberg.70 Fusion occurs by contact and adhesion between the two vesicles. Coalescence, followed by destabilization of the central wall, leads to the formation of a single large vesicle. The reverse fission process was also demonstrated.

3.6

Multicompartment structures

Recently, a number of groups have described the assembly of ABC BCPs consisting of one hydrophilic block and two incompatible hydrophobic blocks. It has been shown that phase separation can lead to interesting multicompartmentalized structures, where the hydrophobic core consists of segregated domains of the incompatible blocks. For example, BCPs incorporating both lipophilic and fluorophilic hydrophobic blocks have been prepared.34, 71 Phase separation within the hydrophobic core of the micelles formed in dilute aqueous solution was observed. The phase separation and structures formed depended on the sequence of the blocks. A hydrophilic–lipophilic–fluorophilic sequence resulted in micelles with a core–shell morphology, with the fluorophilic block forming the inner core. On the other hand, when a lipophilic–hydrophilic–fluorophilic sequence was used, “patched” multicompartment micelles were formed (Figure 18).

4

APPLICATIONS OF BLOCK COPOLYMERS

One of the well-known commercialized BCP families is those based on poly(ethylene oxide) and poly(propylene oxide). These polymers have been commercialized by BASF under the trade name Pluronic. These BCPs are widely used as surfactants in a range of products including cosmetics and pharmaceuticals as well as in the petroleum and paper industries. BCPs have also been commercialized as thermoplastic elastomers.72, 73 These include poly(styrene)-b-poly(isoprene)-b-poly(styrene) or poly(styrene)-b-poly(butadiene)-b-poly(styrene) BCPs, sold

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

18

Soft matter

under the trade name Kraton.74 These polymers are widely used in a range of applications including adhesives and coatings, as well as blending agents to produce materials with novel properties. In addition to linear polymers, star BCPs have also been commercialized.75 Recently, Arkema has introduced a BCP family to the market under the trade name Nanostrength.76 These poly(methyl methacrylate)-b-poly(butyl acrylate)-b-poly (methyl methacrylate) polymers rely on the incompatibility between the blocks to generate useful assembled structures for fillers and rheological control, again in thermoplastic and composite applications. There is currently significant interest in the use of BCP systems for drug delivery and encapsulation. The Pluronics have been widely examined here. Other potential applications exist in nanopatterning72 and photonics. While much of this has yet to be commercialized, we anticipate that, with the strides currently being taken in both the synthesis and the assembly, such exploitation is not far away.

in this area is enormous. Although synthetically the complexity of protein structures is currently unavailable, rational design of structures in the bulk and solution phase is often possible.

REFERENCES 1. P. J. Flory, Principles of Polymer Chemistry, Cornell University Press, Ithaca, NY, 1953. 2. P.-G. de Gennes, Scaling Concepts in Polymer Physics, Cornell University Press, Ithaca, NY, 1979. 3. F. S. Bates, Science, 1991, 251, 898. 4. L. Leibler, Macromolecules, 1980, 13, 1602. 5. G. H. Fredrickson and E. Helfand, J. Chem. Phys., 1987, 87, 697. 6. S. B. Darling, Prog. Polym. Sci., 2007, 32, 1152. 7. H. Fried and K. Binder, J. Chem. Phys., 1991, 94, 8349. 8. Z. G. Wang and S. A. Safran, J. Chem. Phys., 1991, 94, 679. 9. A. J. Meuler, M. A. Hillmyer, and F. S. Bates, Macromolecules, 2009, 42, 7221.

5

CONCLUSIONS

In this chapter, we have introduced the area of the assembly of BCPs both in the bulk and in dilute solution. This area is highly active and complex. In the bulk, phase separation is relatively well understood and related to the segregation strength of the two chemically distinct polymer components. Phase diagrams have been generated to allow the accurate prediction of the nanostructures adopted on phase separation. Complications arise if one of the blocks is a rod as opposed to a coil, but the assembly is still relatively well understood. In solution, BCPs can adopt a range of self-assembled structures depending on many parameters including the polymer composition, polymer concentration, solvent choice, temperature, and the presence of additives among others. The structure adopted tends to follow trends akin to those for small molecule surfactants. However, this is not always the case and often it is necessary to scope out from scratch the structure adopted for a new family of polymers over a range of block lengths. On top of this, the process by which self-assembly is carried out is a critical parameter and means that it is often difficult to predict the structure that will be adopted even if a phase diagram has been generated from another route. Nonetheless, a range of structures can be prepared and there is significant interest in the use of structures such as micelles, wormlike micelles, and vesicles for drug delivery or as nanoreactors, for example. The ability to synthesize BCPs with a range of block lengths, chemistries, and architectures means that the scope

10. S. W. Yeh, K. H. Wei, Y. S. Sun, et al., Macromolecules, 2005, 38, 6559. 11. M. W. Matsen and F. S. Bates, Macromolecules, 1996, 29, 1091. 12. M. W. Matsen and F. S. Bates, J. Pol. Sci. B-Pol. Phys., 1997, 35, 945. 13. N. Sary, R. Mezzenga, C. Brochon, et al., Macromolecules, 2007, 40, 3277. 14. D. R. M. Williams and G. H. Fredrickson, Macromolecules, 1992, 25, 3561. 15. A. de Cuendias, R. C. Hiorns, E. Cloutet, et al., Polymer Int., 2010, 59, 1452. 16. M. Reenders and G. ten Brinke, Macromolecules, 2002, 35, 3266. 17. L. B. Li, Y. Serero, M. H. J. Koch, and W. H. de Jeu, Macromolecules, 2003, 36, 529. 18. N. A. Lynd, A. J. Meuler, and M. A. Hillmyer, Prog. Pol. Sci., 2008, 33, 875. 19. C. Burger, W. Ruland, and molecules, 1990, 23, 3339.

A. N. Semenov,

Macro-

20. M. W. Matsen, Phys. Rev. Lett., 2007, 99, 148304. 21. N. A. Lynd and M. A. Hillmyer, Macromolecules, 2007, 40, 8050. 22. A. Noro, M. Iinuma, J. Suzuki, et al., Macromolecules, 2004, 37, 3804. 23. Y. Matsushita, A. Noro, M. Iinuma, et al., Macromolecules, 2003, 36, 8074. 24. N. A. Lynd and M. A. Hillmyer, Macromolecules, 2005, 38, 8803. 25. S. W. Sides and G. H. Fredrickson, J. Chem. Phys., 2004, 121, 4974.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Assembly of block copolymers

19

26. M. W. Matsen, Eur. Phys. J. E., 2006, 21, 199.

53. G. Battaglia and A. J. Ryan, Nature Mater., 2005, 4, 869.

27. D. M. Cooke and A. C. Shi, Macromolecules, 2006, 39, 6661.

54. H. W. Shen and A. Eisenberg, Macromolecules, 2000, 33, 2561.

28. J. Rodriguez-Hernandez, F. Checot, Y. Gnanou, S. Lecommandoux, Prog. Polym. Sci., 2005, 30, 691.

55. A. Choucair and A. Eisenberg, Eur. Phys. J. E., 2003, 10, 37.

and

29. J. F. Lutz, Polym. Int., 2006, 55, 979. 30. J. Z. Du and R. K. O’Reilly, Soft Matter, 2009, 5, 3544.

56. P. Alexandridis and J. F. Holzwarth, Langmuir, 1997, 13, 6074.

31. D. J. Adams, C. Kitchen, S. Adams, et al., Soft Matter, 2009, 5, 3086.

57. O. Terreau, L. B. Luo, and A. Eisenberg, Langmuir, 2003, 19, 5601.

32. S. Jain and F. S. Bates, Macromolecules, 2004, 37, 1511.

58. A. S. Lee, V. Butun, M. Vamvakaki, et al., Macromolecules, 2002, 35, 8540.

33. P. Schuetz, M. J. Greenall, J. Bent, et al., Soft Matter, 2011, 7, 749. 34. K. Skrabania, H. von Berlepsch, C. Bottcher, A. Laschewsky, Macromolecules, 2010, 43, 271.

and

35. K. T. Kim, J. H. Zhu, S. A. Meeuwissen, et al., J. Am. Chem. Soc., 2010, 132, 12522. 36. H. G. Cui, Z. Y. Chen, K. L. Wooley, and D. J. Pochan, Soft Matter, 2009, 5, 1269. 37. J. Israelachvili, Intermolecular & Surface Forces, Elsevier, Amsterdam, 1992.

59. A. E. Smith, X. W. Xu, and C. L. McCormick, Prog. Polym. Sci., 2010, 35, 45. 60. W. Agut, A. Brulet, C. Schatz, et al., Langmuir, 2010, 26, 10546. 61. V. Butun, S. Liu, J. V. M. Weaver, et al., React. Funct. Pol., 2006, 66, 157. 62. G. Battaglia and A. J. Ryan, J. Am. Chem. Soc., 2005, 127, 8757. 63. H. Bermudez, A. K. Brannan, D. A. Hammer, et al., Macromolecules, 2002, 35, 8203.

38. C. Booth and D. Attwood, Macromol. Rapid Commun., 2000, 21, 501.

64. M. Antonietti and S. Forster, Adv. Mater., 2003, 15, 1323.

39. D. E. Discher and A. Eisenberg, Science, 2002, 297, 967.

65. P. L. Soo and A. Eisenberg, J. Pol. Sci. B Pol. Phys., 2004, 42, 923.

40. S. Jain and F. S. Bates, Science, 2003, 300, 460. 41. D. J. Adams, M. F. Butler, and A. C. Weaver, Langmuir, 2006, 22, 4534. 42. Y. Y. Won, A. K. Brannan, H. T. Davis, and F. S. Bates, J. Phys. Chem. B , 2002, 106, 3354. 43. Z. X. Du, J. T. Xu, and Z. Q. Fan, Macromolecules, 2007, 40, 7633. 44. K. Rajagopal, A. Mahmud, D. A. Christian, et al., Macromolecules, 2010, 43, 9736.

66. X. H. He and F. Schmid, Macromolecules, 2006, 39, 2654. 67. D. J. Adams, S. Adams, D. Atkins, et al., J. Contr. Rel., 2008, 128, 165. 68. G. Battaglia and A. J. Ryan, J. Phys. Chem. B , 2006, 110, 10272. 69. J. R. Howse, R. A. L. Jones, G. Battaglia, et al., Nature Mater., 2009, 8, 507. 70. L. B. Luo and A. Eisenberg, Langmuir, 2001, 17, 6804.

45. R. Sachl, M. Uchman, P. Matejicek, et al., Langmuir, 2007, 23, 3395.

71. S. Kubowicz, J. F. Baussard, J. F. Lutz, et al., Angew. Chem. Int. Ed., 2005, 44, 5262.

46. A. Choucair, C. Lavigueur, and A. Eisenberg, Langmuir, 2004, 20, 3894.

72. S. Hadjichristidis, S. Pispas, and G. A. Floudas, Block Copolymers. Synthetic Strategies, Physical Properties and Applications, J. Wiley and Sons, NJ, 2003.

47. Y. S. Yu and A. Eisenberg, J. Am. Chem. Soc., 1997, 119, 8383. 48. T. Nicolai, O. Colombani, and C. Chassenieux, Soft Matter, 2010, 6, 3111. 49. E. Lejeune, M. Drechsler, J. Jestin, et al., Macromolecules, 2010, 43, 2667. 50. D. A. Rider and I. Manners, Polymer Rev., 2007, 47, 165. 51. Y. B. Lim, K. S. Moon, and M. Lee, J. Mater. Chem., 2008, 18, 2909. 52. J. A. Massey, K. Temple, L. Cao, et al., J. Am. Chem. Soc., 2000, 122, 11577.

73. A. V. Ruzette and L. Leibler, Nature Mater., 2005, 4, 19. 74. www.kraton.com (accessed 9 May 2011). 75. www.styrolution.com/products/styrolux.html May 2011).

(accessed

9

76. P. Gerard, L. Couvreur, S. Magnet, et al., Controlled architecture polymers at Arkema: synthesis, morphology and properties of all-acrylic block copolymers, in Controlled/Living Radical Polymerization: Progress in RAFT, DT, NMP, & OMRP, ed. K. Matyjaszewski, American Chemical Society, Oxford University Press, USA, 2009.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc135

Molecularly Imprinted Polymers B¨orje Sellergren1 and Andrew J. Hall2 1 2

INFU, Technical University of Dortmund, Dortmund, Germany Medway School of Pharmacy, Universities of Greenwich and Kent at Medway, Chatham, UK

1 2 3 4 5

Introduction Fundamental Concepts and Approaches Techniques for Imprinting Small Molecules Techniques for Imprinting Large Molecules Morphology Design and Special Imprinting Techniques 6 Applications of Molecularly Imprinted Polymers 7 Conclusions References

1

1 1 4 13 16 21 25 26

INTRODUCTION

Robust molecular recognition elements with antibody-like ability to bind and discriminate between molecules can today be synthesized using various forms of molecular imprinting (Figure 1).1–5 The most common form of imprinting entails the synthesis of reticulated polymers in the presence of templates (T) which, widely defined, may range from being individual ions, small molecules, or biological macromolecules to microorganisms or crystal particles. Functional monomers act as anchors, interacting with the template (1) and holding it in place during the subsequent polymerization (2). Removal of the template from the formed polymer liberates binding sites (3) complementary in shape and binding groups to the template structure. Hence, these molecularly imprinted polymers (MIPs) are functional in

the sense that they exhibit a memory for the template and can selectively bind it or related structures with high affinity, not unlike the way antibodies bind their antigens. In spite of this advanced function, and in contrast to the biological recognition elements, MIPs are remarkably stable against mechanical stresses, high temperatures and pressures, and intense radiation. They are also resistant to treatment with acid, base or metal ions, and stable in a wide range of solvents. The storage endurance of the polymers is also very high. Furthermore, the polymers can be used repeatedly without loss of their “memory effect.” This, in addition to the relative ease of producing MIPs, has led to a boom in the research and industrial interest in this fascinating class of materials, particularly with the aim of finding alternatives or substitutes for the labile biologically derived recognition elements.

2

FUNDAMENTAL CONCEPTS AND APPROACHES

The main appeal of molecular imprinting lies in its simplicity in terms of the required ingredients, equipment, and unit operations. The majority of the reported examples are based on MIPs formed by free radical polymerization in the presence of the template by operations requiring only simple equipment. Hence, MIPs can be produced in essentially any moderately equipped laboratory. The resulting polymers are then used in various recognition-based applications, many of which have been reported in the literature. These range from separation technology where they achieve selective separations at the analytical or preparative scale, through their use as analytical tools in assays or sensors, to biomedicine, where their use to control the delivery of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

2

Soft matter

R +

T

T

T

1

Figure 1

2

−T

3

Principle of molecular imprinting. Binding site design Template molecule − Target imprinting − Fragment imprinting − Analog imprinting Functional monomers − Computational methods − Combinatorial methods − Host–guest chemistry − Covalent imprinting

Figure 2

+T

Scaffold design Organic polymers −

Polymethacrylics, polyacrylics, and polystyrens

Inorganic polymers −

For example, SiOz, TiOz, etc.

Polymerization techniques − − −

Crushed monoliths Beaded MIPs Hierarchical MIPs Composites Thin films

Noncontrolled Controlled

Fibers and tubes

Type of initiator

Perfusive monoliths

Crosslinking − − − −

Morphology design

During polymerization

Membranes Nanostructured MIPs

After polymerization Degree of cross-linking Covalent or noncovalent

Steps involved in MIP design.

drugs, or even as drugs per se, is being explored (see Section 6.4). In spite of this progress, no single imprinting protocol is generic and laboratories serious about developing MIPs must therefore adopt a multifaceted approach if the goal is to address a broad range of target molecules by imprinting. The design of a MIP receptor can be divided into three steps, distinguished by the type of chemistry involved and the length scales of the designed features (Figure 2). The construction of a binding site for a low-molecular-weight target aims to place functional groups at 1-nm length scale, in complementary positions to the target. This design starts with an examination of the structural and functional features of the target molecule for which an MIP is needed, while also considering the context in which the MIP should operate (solvent, temperature, target concentration, static or dynamic mode, etc.) and whether the binding event should trigger an associated function. The template can be either covalently or noncovalently linked to the functional monomer and the same distinction can be made with respect to the interactions occurring between the target molecule and the resulting binding site. This results in various possibilities to design the binding sites which have all been explored, with some examples shown in Figure 3. A limited number of functional groups can be targeted by the covalent imprinting approaches (a) and (b) (Section 3.3)3 ; but the most widely used approach in

imprinting is (c), involving functional monomers that are chosen to associate noncovalently with the template.7, 10 The hybrid approach (d) takes advantage of both noncovalent and covalent imprinting, whereas recognition is based solely on noncovalent interactions.6 Metal ion imprinting shown in (e) is also an important branch of the field; several research groups are involved in developing methods for the sequestration or analysis of metal ions.11 Metal ions can also be used to coordinate to the template, whereby removal of the template leaves behind vacant coordination sites available for template rebinding.12 In the most commonly used noncovalent imprinting approach, the template is mixed with one or several functional monomers that are subsequently polymerized in the presence of an excess of cross-linking monomer. Thereafter, it can be extracted from the polymer and recycled. Generally, the resulting materials can be used directly to perform separations with high affinity and selectivity, for instance, as chromatographic stationary phases. An adequate choice and design of functional monomers and templates are of crucial importance here for achieving success in the imprinting. The functional monomer–template complexes then need to be incorporated in a polymer scaffold by copolymerization with comonomers and cross-linking monomers (e.g., Figure 4).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

M T

O

R1

O

R2

3

B

T

(a) M

M T

O

T

(b) M

R1

O

O

R1 R2

O

H

O H

R2

M (c)

H

T

T

O

M

H N

O H

N N

N N

M

O

T

T (d)

M

O H

N

O

O H

N

R O

M

(e)

M

CH3

CH3 N O O Cu O O O O Cu O O N

Figure 3 Examples of binding-site design strategies using covalent or noncovalent interactions between the template (T) and the functional monomers (M) during synthesis (S) and use (U). (a) S = covalent, U = covalent, for example: boronate ester formation.3 (b) S = covalent, U = noncovalent, for example: carbonate as sacrificial spacer to place two OH groups at complementary positions.6 (c) S = noncovalent, U = noncovalent, for example: Hydrogen bonding between methacrylic acid and adenine.17 (d) S = covalent/noncovalent, U = noncovalent, for example: Use of sacrificial spacer and hydrogen bond interaction for imprinting peptides.8 (e) S = metal ion complex, U = metal ion complex. For example: copper ion recognition.9 O O O O

H N

EDMA O

DVB

H N O

MDA O

O

O

O O

HO O

O

O

O O

O

TRIM

O

PETRA

Figure 4 Cross-linking monomers commonly used in molecular imprinting.

The nature of the scaffold influences the site integrity and the way in which the templated sites are presented in the network, that is, to what extent they are accessible to solvent and guests. Most commonly, MIPs are produced by free radical polymerization in the presence of a high level of cross-linking monomer and a solvent called a porogen.13 Free radical polymerization is a kinetically controlled process which, in the presence of cross-linkers, leads to fixation of the polymer chains in an irregular glass-like arrangement. In addition, the solvent added for solubilizing monomers and template serves the purpose of creating the pores in the polymer in a process called phase separation. The pores are required for transport, that is, for the template to be removed from the polymer and for a guest to occupy the imprinted sites. The size of the pores and the surface area of the pore walls are influenced by the quality

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

4

Soft matter Site A. In macropores C

Site B. In micropores A

Site C. Embedded

G F

Site D. Site complementary to dimer or multimer

B Site E. Induced binding site. E Site F. Nonselective site. D Site G. Residual template

Functional group

Figure 5

Schematic drawing displaying different types of binding sites in an amorphous imprinted polymer.

of the solvent (good or bad solvent for the growing polymer chains) and the level of cross-linking, among other factors, but the end result is a highly heterogeneous amorphous material (Figure 5). Partly in order to address the morphologically related heterogeneity, but also in order to develop MIPs for advanced applications, different materials formats are being developed.14–16 These can be in the form of beads, fibers, tubes, films, membranes, or nanoparticles. An effective development of MIPs hence requires the mastering of multiple disciplines in chemistry and materials science.

3 3.1

TECHNIQUES FOR IMPRINTING SMALL MOLECULES Noncovalent imprinting

Using the noncovalent approach (c) of Figure 3 (see Figure 6), the imprinting of lipophilic, low-molecularweight templates containing basic or acidic functional groups is relatively straightforward.2 For example, a simple commodity monomer such as methacrylic acid (MAA) can be used to create good binding sites for a large variety of template structures containing hydrogen bond- or proton-accepting functional groups [see Figure 6 for the imprinting of 9-ethyladenine (9EA)].17 MAA forms complementary hydrogen bonds or hydrogen-bonded ion pairs with the template, with individual binding constants ranging from single figures for weak hydrogen bonds to several hundreds for cyclic hydrogen bonds or hydrogen-bonded ion pairs formed in weakly polar, aprotic solvents, such as chloroform.

3.1.1 The functional monomer In spite of the versatility of MAA in creating good imprints, it is complemented by a range of other commodity functional monomers (Figure 7) targeting other functional groups (e.g., amine-containing monomers for acidic groups). The optimal type of functional monomer can often be predicted based on theoretical arguments, considering intermolecular interaction theory supported by modeling (Section 3.1.3)18, 19 or physical characterization (e.g., by NMR, IR, UV-vis)10, 20 of the complexes present in the prepolymerization mixture. On the other hand, trial and error still plays an important role and, when combined with high-throughput techniques for synthesizing and evaluating MIPs21 (Section 3.1.4), has proven surprisingly powerful in achieving MIPs displaying adequate performance. The MIPs prepared by noncovalent imprinting typically display highly heterogeneous site distributions, low binding capacities and, often, poor target binding in polar media, such as water. Imprinting relies here on the successful stabilization of relatively weak interactions between the template and the functional monomers. This typically requires the use of solvents of low polarity and the addition of an excess of functional monomer (typically four equivalents, but sometimes higher) in order to ensure that the template molecule is complexed to a maximal degree.2 This, in turn, means that a large proportion of the functional monomer is not involved in complexation of the template and is instead distributed randomly throughout the polymer matrix during the polymerization. This is a major cause for the high levels of nonspecific binding and binding site heterogeneity observed in these materials. For instance, using 9EA as template and MAA as functional

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

H

N

5

H O N

N

N

N

O 9EA/MAA/EDMA: 1/4/20 (mol/mol)

MAA

9EA

H H

N

N

O

H

O

O H

H

N

H O

O

O

H

O

O O

N

N

AIBN, UV CHCl3

O O

O

O

OH

O

H O

EDMA

Figure 6 Noncovalent imprinting of 9-ethyladenine (9EA) leading to highly cross-linked monoliths from which particles are obtained by repetitive crushing and sieving cycles. N

O

N 4-VP

N

O 2-VP

DEAEMA CF3

OH

O OH

O

O

MAA

TFM

OH O OH ITA

O

also include dimerization (or higher-order complexes) of the functional monomers or the interactions between different functional monomers. The latter is a particular problem when, for instance, acidic and basic functional monomers are used jointly to target basic and acidic sites on the template. Nevertheless, synergistic effects have been reported for terpolymer MIPs also in the latter case (e.g., using mixtures of MAA and VPY for imprinting amino acid derivatives).22

O O NH2

OH

N

3.1.2 The template

O MAAM

HEMA

NVP

Figure 7 Examples of commodity functional monomers used in noncovalent molecular imprinting.

monomer, MIPs were obtained which displayed a heterogeneous binding site population for the template in chloroform with one class of high-energy binding sites (K = 77 000 M−1 , Q = 20 µmol g−1 ) and one class of lowerenergy sites (K = 2400 M−1 , Q = 86 µmol g−1 ).17 Given these limitations and the incompatibility of these monomers with a number of templates, it would seem reasonable that improved imprinting could be achieved with an expanded monomer repertoire at hand. This would also match Nature’s solution to molecular recognition problems, relying on hosts incorporating multiple different functional groups converging to complementary positions on the guest. A way forward could thus be a “cocktail” strategy where 20 “essential” functional monomers could be used to create more discriminative imprints with higher target affinity (Figure 8). The limitations of this approach are linked to the simplicity of noncovalent imprinting and its reliance on the self-assembly principle. Multiple equilibria involving the monomers and template exist in the solution prior to polymerization. These do not all involve the template, but

The engineering of MIPs capable of not only recognizing their templates but also another target compound or a group of structurally related compounds is a key activity in MIP development. There are several reasons for choosing a target analog as template. The targets themselves may be toxic and/or highly priced, their direct use as templates may be associated with potential interference problems or reactivity problems, or their use may cause excessive template bleeding. In target analog imprinting, the template should exhibit the following properties: • • •

it should be readily available in large quantities at a low cost, it should exhibit solubility under the imprinting conditions, and it should result in sites showing good cross-reactivity with the target analyte or analytes.

The design of the template is therefore seldom a trivial task but, on the contrary, may involve numerous iterations and the synthesis of several candidates with suitable properties. One of the most widely used MIPs in analytical chemistry targets the general class of 1,2-amino alcohols (1).23

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

6

Soft matter

N NH NH2

NH2

−

O

H N

NH2

− HN

−

HN

HO

+

OH O

Figure 8

High-fidelity MIP formed using a cocktail of amino acid analog monomers. OH H N R1

R3 R2 1

Thanks to a careful template selection (a dummy template such as 2 of minor analytical relevance), the sorbents can cross-react with the majority of β-receptor agonists and antagonists, such as the high-priority growth promoter, clenbuterol (3). OH

H N

Br

OH Cl

It may also pay off to design template libraries directed toward the functionalities of a given target. Complex natural products containing a limited number of functional groups may, for instance, bind to MIPs made using considerably simpler templates exhibiting part or all of these functional groups placed at geometrically complementary positions to the target. This principle was used by Nemoto et al. in targeting the fish poison domoic acid (8) with MIPs prepared using simple commodity di- or triacids as templates.25 It was found that the triacid pentane-1,3,5-tricarboxylic acid (9) led to an MIP displaying sufficient retention of the target to allow efficient clean up of the target from fish.

H N O

H2 N

H N

H2N Cl

Br

OH O

HO

2

O

3 OH

MIPs targeting single high-priority analytes may also find broad usage and the role of the template here is to create a site capable of accommodating only one compound; thus, it can exhibit a closer structural resemblance to the target. For instance, the nitrosylated nicotine metabolite 4(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) (4), a biomarker for nicotine smoke exposure, could be selectively retained by an MIP prepared using the isosteric enamine (5) as template.24 O

OH

N

OH

N N

N N

4

5

8

O

O

HO

OH O

OH

9

Traditional imprinting fails when the targets are complex, poorly soluble, or simply unavailable in sufficient quantities for imprinting. Most biomolecules fall into this category, especially the macromolecular ones, that is, proteins, nucleic acids, and polysaccharides. In these cases, the polymer may need to be synthesized in an aqueous medium to solublize the target and/or to stabilize it in a conformation close to its native low-energy conformation. However, in water, the strong hydration forces and the polar environment commonly prevent stable electrostatic interactions between the target and the functional monomers that are commonly used in molecular imprinting. Moreover, such targets are difficult, if at all possible, to come across in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers 1. Design of a virtual library of functional monomers 2. Design of molecular model of the template to be screened 3. Screening using a LeapfrogTM algorithm Selection of best monomers ideally suited for polymer formation, possessing high affinity for the template

4. Refining by molecular dynamics and mechanics Mimicking specific polymerization or binding conditions

Figure 9 Procedure for computational screening of functional monomers for noncovalent imprinting.

sufficient quantities, precluding the use of the target as template. In order to overcome this problem, fragments complementary to substructures of the target may be used as templates.26–28 Although the resulting sites are by consequence less specific for the target, this strategy uses simple available templates compatible with the imprinting conditions. This may result in sites complementary to the substructure that still bind the larger biomolecules with significant affinities. This concept was first demonstrated for peptides26 and simple vitamins.27 Following an epitope imprinting approach, short peptides were used as templates for binding a larger target containing the epitope sequence as its terminal part.26 This concept has subsequently been combined with surface imprinting techniques to produce MIPs exhibiting cross-reactivities with peptides29 or impressively with proteins28 based on just a short peptide complementary terminal sequence (see Section 4.3).

3.1.3 Computational techniques Computational techniques including molecular modeling and molecular dynamics simulations have grown in importance as a tool for both fundamental studies of MIP systems and for predicting the perfomance of a given MIP starting from the monomer and template components.18, 19 This is an obvious consequence of the steadily increasing computer power and the development of improved software and draws inspiration, to some extent, from in silico drug design. These theoretical methods are based on quantum mechanical treatments allowing for the modeling of intermolecular interactions using static structures in low-energy conformations. Piletsky and colleagues first demonstrated the predictive power of simulations in the design of MIPs for a given target following a simple four-step operation (Figure 9).18 Lacking a prior knowledge of which functional monomer to choose, a virtual library of functional monomers was constructed and screened for all possible interactions with a minimized molecular model of the template. The monomers

7

giving the highest binding score with the template were anticipated to give the polymers with higher affinity and specificity, which, in some cases, could be experimentally verified. In contrast to simulations comprising only the functional monomers and template of a given system, a complete molecular dynamics simulation replicates the components and concentrations employed in the corresponding polymer synthesis and, hence, offers results correlating better with physical characterization data.19 Hence, Nicholls and colleagues performed a comprehensive molecular dynamics simulation taking into account the influence of explicit solvent, multiple molecules of template, functional and crosslinking monomers, and initiator on the nature, number, and dynamics of the noncovalent interactions underlying template complexation. A mixture incorporating the template bupivacaine, MAA, ethylene dimethacrylate (EDMA), and 2,2
-azobis(isobutyronitrile) (AIBN) in chloroform was simulated (Figure 10), and the results correlated with complexation studies performed using 1 H-NMR spectroscopic and radioligand binding studies. The results could be used to partially account for the binding site heterogeneity in MIPs and revealed a previously neglected role of the crosslinking monomer EDMA in engaging in binding site interactions with the template.

3.1.4 Combinatorial techniques and screening Previous work has shown that major performance enhancement of MIPs can be achieved by using combinatorial techniques allowing a careful and systematic screening of different monomer combinations and subsequent optimization of the polymer compositions.21 These techniques rely on the ability to synthesize and test a large number of polymers in parallel, which necessarily implies the use of small batch sizes. The synthesis of such mini-MIPs can be performed in the bottom of small vials or in the wells of multiwell plates, followed by testing in situ of binding and molecular recognition via equilibrium batch rebinding experiments. The general procedure is essentially a scaled-down version of the traditional monolith procedure and can be partly automated. In particular, the use of modern plate technology (plate readers, pipetting robots) can be exploited for the preparation and screening of 96-well MIPs (Figure 11). Recently, the power of this tool in improving MIP performance in water was demonstrated.30 When allowing a library of structurally diverse compounds to contact an MIP, it is often found that the MIP retains a few compounds that may appear structurally unrelated to the template. This nonobvious cross-reactivity has led to the industrial Explorasep  concept which, in brief, comprises screening of 96-well plate MIP libraries grouped according to the interacting functional group of the polymer

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

8

Soft matter

Bupivacaine ×7

MAA × 75

EGDMA × 368

AIBN ×8

CHCI3 × 741

Figure 10 A bupivacaine MIP prepolymerization solution studied by the molecular dynamics (MD) method consist of: (A) representation by molar ratios of components used in a typical protocol (left), (B) building the mixture using a periodic boundary condition and subsequent equilibration at constant volume and pressure (middle), (C) running and collecting data during a 5-ns production phase (right), and, finally, (D) statistical analyses on, for example, the atomic distribution of the various prepolymerization components around each of the multiple template molecules. (Reproduced from Ref. 19.  American Chemical Society, 2009.) UV

Thermostat

Polymerization

Dispensing

Plate reader

Figure 11

Vacuum unit

Release and rebinding tests

Transfer to filterplates

HPLC

High-throughput synthesis and screening technique for combinatorial MIP optimization.

(i.e., acid, base, neutral, etc.).31 For example, it was found that an MAA-based MIP imprinted with nicotinamide (10) also showed selectivity for theophylline (11). O

O H N

N

H2 N N

10

O

N

acceptors in theophylline match as highlighted in red, the hydrogen-bond acceptors in nicotinamide. In other words, nicotinamide should provide a good template for theophylline. One key advantage of the Explorasep approach is that template analogs found by screening will typically be different from the target molecules and, thus, unlikely to interfere with the analysis.

N

11

The strongest interactions between the functional monomer and the template are here expected to involve the pyridine nitrogen and the amide group. If theophylline is superimposed on nicotinamide, good hydrogen-bond

3.2

Host–guest chemistry

As mentioned above, the majority of MIPs have been prepared using a small number of commercially available functional monomers. While successful, it has become clear that better MIPs can be prepared using designed

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

O O N H

N

O NH

O

O

NH

1

HN O

N

N H

9

HN

N NH

O

HN

O

O

12

Figure 12

2,6-Bis-acrylamidopyridine (12) and the proposed polymeric binding site with barbital.

functional monomers, allowing for “stoichiometric” noncovalent imprinting.4 Inspiration for the design of such monomers has come largely from Nature and the fields of host–guest and supramolecular chemistries. In this section, we focus on a few of the more prominent examples. One of the first reports of a designed functional monomer in molecular imprinting came from Takeuchi and colleagues32 who used the bis-amidopyridine monomer (12) in the imprinting of barbital (Figure 12). The monomer presents a donor–acceptor–donor (DAD) array of hydrogen-bond sites, which is complementary to the ADA (acceptor donor acceptor) moiety within the template. The obtained polymeric binding site was postulated to resemble the structure of small molecule receptors prepared by Hamilton et al.33 for the same purpose (for details of hydrogen bonding receptors see Hydrogen-Bonding Receptors for Molecular Guests, Molecular Recognition).

H

N N Cu2+ NH2 H H2O

O

O

N

O

P

N

MIPs prepared with this monomer showed relatively high imprinting factors and a degree of selectivity for barbital over differently substituted barbiturates when tested in the chromatographic mode. Further, analytes where some of the hydrogen-bonding sites had been removed were much less retained on these polymers. A little later, Wulff and Sch¨onfeld introduced a benzamidine-based monomer (13) capable of strong interactions with neutral oxyacids via ion pairing and cyclic hydrogen bonding.34

N NH 13

N

H

H HO H2O

NH2

N Cu2+ NH2 NH2

O

(a)

(b)

N N N N Cu 2+ H H H H2O NH2 O O O O N N P OPO O O

N H

N H

(c)

N

N H

N HO H 2O

N Cu2+ H NH2

N H

(d)

Figure 13 Schematic representation of the preparation and the catalytic centers of imprinted polymer catalysts. (Reproduced from Ref. 35.  American Chemical Society, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

10

Soft matter

NO2

R O

O

H N O

O

N H

H N R

TBA

O N H

N H

N H

16

19

Ka ~ 620 M

−1

Ka ~ 8000 M

−1

O N H

N H

O CF3 N H

14

17

Ka ~ 30 M−1

N H

O

Ka ~ 1300 M−1

N H

20

N H

O N H

N H 18

15

CF3

Ka ~ 8800 M−1

CF3 O

N H

N H

Ka ~ 6500 M−1

Ka ~ 120 M−1

Figure 14 1,3-Bis-aryl and 1,3-aryl-alkyl mono-urea monomers providing for strong binding to oxyanions. Binding constants for TBA (tetrabutyl ammonium) benzoate determined by 1 H-NMR titrations37 in DMSO-d6 have been indicated.

The high affinity of this monomer in chloroform and acetonitrile (Ka > 106 M−1 and circa 104 M−1 , respectively) implies >99% complexation with the template species in these media, making imprinting with this monomer truly stoichiometric. A potential limitation of the monomer is that the use of more polar solvents, such as dimethyl sulfoxide (DMSO), leads to a reduction in affinity (Ka < 10 M−1 ) due to unfavorable conformational effects. However, this monomer has been used to prepare a series of catalytically active MIPs, possessing high activity and efficiency. The latest in this series also contains a metal ion coordination site, together with the benzamidine moiety and mimics the action of carboxypeptidase A (Figure 13).35 Oxyanions have also been targeted through the use of urea-based functional monomers, inspired by the work of Wilcox et al.36 It has been shown that 1,3-diaryl ureas are capable of binding with high affinity to oxyanions in polar media, such as DMSO.37 This affinity is enhanced through placement of electron-withdrawing groups on the aromatic rings and for monomer 20 in Figure 14, a Ka of circa 104 M−1 with tetrabutylammonium benzoate was observed in deuterated DMSO. Such monomers have proven successful in the preparation of MIPs targeting a number of oxyanionic species, for example, penicillin-based antibiotics38 and phosphotyrosine-containing peptides (Figure 15)39 (for

CF3 F3C

F 3C

O

N N

H

O

N

H

H −

O

+

CF3

O P O

N −

H

O

H3N COO−

Figure 15 Schematic drawing of a urea containing imprinted site complementary to the phosphotyrosine side chain of a phosphorylated peptide. (Reproduced from Ref. 39.  WileyVCH, 2008.)

details of anion binding ureas see Amide and Urea-Based Receptors, Molecular Recognition). Turning to host–guest chemistry, an interesting approach to imprinting in aqueous media was reported by Komiyama and colleagues.40 Here, functionalized β-cyclodextrins

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

11

O NH

H N O

O

NH NH O

= 6-AAm-b-CyD

= Template

NH O

Figure 16 Imprinting of peptides using a polymerizable cyclodextrin as functional monomer and methylenebisacrylamide as crosslinking monomer. The polymer was grafted to the surface of porous silica gel. (Reproduced from Ref. 41.  American Chemical Society, 2007.)

were used as functional monomers for the imprinting of steroids and dipeptides, taking advantage of hydrophobic effects. In the latter example, the latent enantioselectivity exhibited by the host molecule was shown to be enhanced by the imprinting process. In an interesting extension, CyD monomers were used for the high-fidelity imprinting of the peptide hormone angiotensin (Figure 16) (for details of cyclodextrins see Cyclodextrins: From Nature to Nanotechnology, Molecular Recognition).41

O O O

O

22

C N R H

3.3

Covalent imprinting approaches 23

As stated previously, the covalent approach a (Figure 3) to imprinting was introduced by Wulff et al. in the early 1970s.3, 42 Here, the template is bound to polymerizable moieties through labile covalent bonds (as in structures 21–23). After synthesis of the polymer, these bonds are cleaved to release the template and create a binding site containing functionality which can rebind the template structure through re-formation of the covalent bonds used during the imprinting step.

B O O O

O B O

O

21

This technique has the advantage that, in principle, each polymerizable template molecule will give rise to one imprinted binding site, with optimal geometrical arrangement of the monomer functionality. This should lead to materials with high binding capacity and higher affinity binding sites. Despite these attractive features, a number of drawbacks exist. Chief among these is the limited number of functional groups that readily form reversible covalent bonds, the most common examples being boronate esters (21), ketals (22), imines (23), and disulfides (Figure 18). It is apparent that the number of templates containing suitable functional groups is also limited. Further, the making and breaking of the covalent bonds should occur rapidly and under relatively mild conditions. Thus, despite its potential advantages, the covalent approach has experienced limited uptake. The semi-covalent43 and sacrificial spacer6 approaches of imprinting bear similarities to the covalent imprinting approach, at least in the imprinting step, where the template is covalently bound to the functional monomer. In the latter approach, labile covalent bonds, such as carbonates

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

12

Soft matter

O O

O

O

H

O H

24

Figure 17

25

Schematic view of the imprinting of cholesterol using a semicovalent, sacrificial spacer approach.

O S S

O

O

S

S

BPA-D

Apo-SH

SH

O

O HO S

Polymerization

O O

S S

O S

Coupling with DTBA

OH

O

S S

S

HS

S

O

Holo-COOH HO

OH

O

S

S

O H O

O H

Reduction

SH

Apo-SH

S S

HS

OH O OH

Incubation with BPA

HO O HO

S

S

140 Bound amount (µmol g−1)

Apo-SH

Holo-COOH

120 100 80 60 40 20 0

Apo-SH(1)

HoloCOOH(1)

Apo-SH(2)

HoloCOOH(2)

Apo-SH(3)

HoloCOOH(3)

Figure 18 Preparation of imprinted binding sites for the recognition of bis-phenol A via covalent imprinting followed by postpolymerization modifications and demonstration of a fully reversible cofactor and template binding. (Modified from Ref. 44.  American Chemical Society, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers as in the imprinting of cholesterol 24, are used (Figure 17). After cleavage of the template, the binding cavity remains decorated with functionality that may interact noncovalently with the template/analyte as in 25. More recently, in a further advancement of such methods, Takeuchi and colleagues44 have explored the use of covalent imprinting, followed by postpolymerization modification of the functional groups placed in the imprinted cavity, to create a range of imprinted materials with affinity for bisphenol A and related compounds (Figure 18). As enzymes depend on cofactors for their catalytic activity, only in the holo-form did the MIP bind the target molecule to a high degree. This contrasted with the apo-form of the MIPs, lacking the cofactor, showing no significant uptake of the target. Interestingly, the disulfide-mediated docking of the cofactor was fully reversible and the stepwise binding of cofactor and target could be repeated several times (Figure 18).

4

TECHNIQUES FOR IMPRINTING LARGE MOLECULES

R

Gel imprinting

Numerous reports have been published in recent years describing protein imprinting in lightly cross-linked gels.45, 46 Recent activities in the field were triggered mainly by a series of reports by Hjert´en and coworkers,47 showing that recognition of integral proteins can be accomplished in imprints in soft polyacrylamide gels. Polyacrylamide is by far the most commonly used matrix offering a proteinlike backbone providing hydrogen bond complementary surfaces.46 Other biocompatible matrices used in protein imprinting are based on polyethyleneglycoldiacrylates or dimethacrylates. Common to all systems is the need to employ a low cross-linking level in order to provide a mesh size of the network large enough for the protein to penetrate. The memory effect of these gels is thus easily erased, preventing repeated use. The guiding principle in the choice of monomer system for macromolecular templates is quite different to that in small molecule imprinting, where functional group complementarity is the important design criteria. For proteins, the preservation of the template in the native folded form is important, which most often restricts the monomer systems to those polymerizable in water. Moreover, the protein surface exposes numerous sites capable of interacting weakly with functional monomers. The resulting imprinted site will thus offer multiple weak interactions which may collectively result in a high affinity for the target protein.48 Nonetheless, several literature reports confirm the viability of this strategy. Figure 19 shows an example of this powerful method for

Myo (h)

Cyt C

Myo (w)

Cyt C

Myo (w)

Myo (h)

R

0 (a)

4.1

13

2

4 nm

(b)

Figure 19 (a) Ion-exchange chromatography of the eluate from a column packed with nonimprinted gel granules and (b) gel antibodies selective for myoglobin from horse. The bed in column (b) does not recognize myoglobin from whale, although the structural differences are very small. (Reproduced from Ref. 47.  Elsevier, 1997.)

generating MIPs capable of discriminating between minimum structural differences in the imprinted protein. The MIP was prepared from acrylamide and methylene bisacrylamide, hence identical monomers as those used to prepare slab gels for PAGE. A departure from this weak imprinting concept is affinity imprinting introduced by Vaidya et al.49 This makes use of an affinity ligand for a given protein, for example, an enzyme inhibitor for a given enzyme, which, if coupled to a polymerizable moiety, can be added as a functional monomer during imprinting. This leads to incorporation of a strong anchoring point for the target protein in association with the imprinted sites. Thus a complex of trypsin and its modified inhibitor N-acryloyl-para-aminobenzamidine was polymerized to form a polyacrylamide gel showing strong affinity for the target enzyme (Figure 20). In an interesting extension of the affinity imprinting approach, Cutivet et al. demonstrated more recently that the resulting polymer gels, when produced in nanogel format by precipitation polymerization, could act as effective inhibitors for the enzyme-catalyzed reaction, suggesting a novel strategy to improve on currently used inhibitor drugs.50

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

14

Soft matter

NH2

NH O

+

O

+

+

NH

NH

NH

NH2

NH

Aps/ Temed

NH

O O

O

Figure 20 Principle of affinity imprinting of proteins shown for the imprinting of trypsin using a polymerizable p-aminobenzamidine affinity ligand. (Reproduced from Ref. 49.  Wiley Periodicals, Inc, 2001.) O H N

H N

NH O

O

N,N ′-methylenebis(acrylamide) (Bis)

N-isopropylacrylamide (NIPAm) O

O

O NH

NH

NH2

N-tert butyl acrylamide (TBAm)

Acrylamide (AAm)

(a)

+ G I GA V L K+V LT

(b)

O

O

NH

NH

O O− Acrylic acid (AAc)

NH3+ N -(3-aminopropyl)methacrylamide (APM)

TG L PA LIS W I K+ R+K+ R+QQ NH2

NH

O NH

O

O NH

O

O NH

O O

Aps Temed

O

HN

NH O

O O

O

O

O

O

O

O

O

HN

HN

O

O

Dialysis NH

O

O NH

O NH

O O

NH O O

O

O

O

HN

(c)

O

Figure 21 (a) Monomers used for nanoparticle synthesis. (b) Amino acid sequence of mellitin. (c) Schematic representation of the imprinting process. Hydrophobic, positive/negative charged, and hydrophilic residues are printed in brown, blue/red, and green. (Reproduced from Ref. 51.  American Chemical Society, 2008.)

The use of precipitation polymerization to produce nanogels was also the technique favored by Shea and colleagues in the imprinting of the bee venom toxin mellitin. Using a combinatorial approach to imprint N-isopropylacrylamide-based nanoparticles, certain terpolymer

compositions containing preferentially AAc and TBAm were found to display effective imprinting of the 26amino acid hydrophobic template peptide (Figure 21).51 From quartz crystal microbalance (QCM)-based testing of the interactions between the imprinted nanoparticles and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers immobilized template peptide, an apparent dissociation constant of Kd = 7.5–25 pM was measured. It is interesting that this is in the same range as that displayed by the corresponding natural antibody (Kd = 17 pM) and, with a size of the nanoparticles being in the range 30–40 nm, they also resemble the size of immunoglobulins, all in all justifying the term plastic antibodies.

4.2

15

and colleagues28 developed a film imprinting procedure where epitopes corresponding to the C-terminal sequence of a target protein were immobilized on a glass surface. Exposing the functionalized surface to a polymerization solution containing acrylamide and ethylenebisacrylamide followed by polymerization resulted in an imprinted film that could be mechanically removed from the template for subsequent recognition studies (see Section 4.3). Ju et al.53 designed a surface molecularly imprinted nanowire by chemical polymerization of dopamine in neutral aqueous solution with anodic alumina oxide (AAO) membrane as a nanomold to imprint bovine and human hemoglobin (Figure 22). Nevertheless, in the above 2D approaches to protein imprinting, only the exterior surface of a substrate or a sacrificial solid support has been imprinted. This results in a small number of imprinted sites and, hence, a limited adsorption capacity in these materials. In order to overcome this limitation, a 3D hierarchical protein imprinting technique offers a promising way forward (Figure 23). This starts from wide-pore silica modified with a submonolayer of adsorbed protein (IgG or HSA), filling with monomers and initiator of the protein–silica template, polymerization, and, subsequently, removal of the protein and silica porogen via fluoride etching and washing. The last step leaves behind an inverse replica of the silica template featuring highly accessible protein complementary binding sites localized preferentially at the pore walls of the polymeric beads.54

Surface imprinting

In spite of numerous reports describing protein-imprinted hydrogels, advances toward generic and robust imprinting techniques have been slow for the reasons mentioned above. Various forms of surface imprinting techniques have been used, with promising results, to address this problem.28, 45, 46, 52, 53 Here, the template protein is confined near an interface separating a monomer-rich and a monomerpoor phase. After polymerization, the template is removed more easily owing to the absence of matrix-related diffusional barriers and the resulting surface-exposed sites can also be more readily occupied by the template. For instance, Ratner and colleagues52 adsorbed template proteins on an atomically flat mica surface followed by spin coating of the modified surface with a solution of a disaccharide that noncovalently complexed with amino acid residues of the protein. The disaccharide was subsequently coated with hexafluoropropylene to create a polymer with permanent surface recognition sites for the protein. Shea

O

O Si

OH HO

CCH4 H3CO Si

OH HO

O

O CH CH3

CCH4

CH CH3 O

O

CH CH3 O

O

O Si O

O Si

OH HO

O

O

CH CH3 O

O

CH CH3

O

O

OH HO

O

Si

O

Si

O O O O

O

Si

O O

O

Si

O

CH CH NH CH CH NH CH CH NH CH CH NH

Si

NH3 Template protein

NH3 OH CH

−

Polymerization removing AAO

+

200 nm

Figure 22 Preparation and specific recognition of surface protein imprinted polydopamine nanowires. (Reproduced from Ref. 53.  Royal Society of Chemistry, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

16

Soft matter

= IgG or HSA IgG

O NH2 O

O N H

N H

HSA NH4HF2

Plasma

30 µm

SiO2

Figure 23

PIgG or PHSA

PN

PHSA

Eluted fractions

Template removal

Target capture

Peptide template

4.3

PIgG

Hierarchical protein imprinting. (Reproduced from Ref. 54.  Wiley-VCH, 2011.)

Polymerization

Figure 24

PN

20 µm

Principle of the hierarchical epitope imprinting technique targeting the terminal sequence of a peptide or protein.

Epitope and fragment imprinting

A viable approach to target inaccessible or labile larger molecules is based on the use of stable templates corresponding to substructures or epitopes of the target.26–28 Although the resulting sites are by consequence less specific for the target, this strategy uses simple available templates compatible with the imprinting conditions. This may result in sites complementary to the substructure that still bind the larger biomolecules with significant affinities. This concept was first demonstrated for peptides26 and simple vitamins.27 Following an epitope imprinting approach, short peptides were used as templates for binding a larger target containing the epitope sequence as its terminal part.26 This concept has subsequently been combined with surface imprinting techniques to produce MIPs exhibiting cross-reactivities with peptides29 or more impressively with proteins28 based on just a short peptide complementary terminal sequence (Figure 24). Shea and colleagues28 imprinted the C-terminus domains of Cytochrome c, alcohol dehydrogenase, and BSA (bovine

serum albumin), using nonapeptide epitopes attached to modified microscope glass slides followed by polymerization of an acrylamide monomer solution covering the slide and subsequent separation of the resulting gel film from the slides. SDS-PAGE analysis of the protein uptake by the corresponding imprinted films demonstrated the success of the imprinting in that each film preferentially bound the protein displaying the complementary epitope.

5

5.1

MORPHOLOGY DESIGN AND SPECIAL IMPRINTING TECHNIQUES Scaffold design

As outlined in Section 2 the optimization of the imprinting scaffold is of crucial importance for improving the performance of MIPs with respect to binding site homogeneity, capacity, mass transfer, and accessibility. Moreover, the development of MIPs in different formats is key to unlocking previously unexplored applications.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

O O

OO

O

Template

OO

O O

OO

O

O O

OO

17

O O

O

O

O

O

O

O

O

O O

O

O O O

O

O

O

O

OO OO

O

O

O

O

O

O O OO

O O

O

O

O

O O

O O

O

O

O O

O

O O OO

O

O

O

O

O

OO

O O

O O

O O

O

O

O

O O

O

O O

O

O O

O O

O

O

O

O

O

O

O

O OO

O

O

O O

O O

O

O

N

NH

O

O

O

O

O O

O

O OO

O

O

O

O

O

N

NH

O

OO O

O

HN

N

O

OO

O

O

O

O

O

O

O

O

O

O O O

OO

O

O

O O O

OO O

O

HN

N

O

O O

O

O

O

O

O

O O

O

O O

O O

O O

O

O

O

O

O O

O

O

O O O O

O

O O

OO

O

O

O

O O

O O

O

O

O

O

O

O O

O

O

O

O

O

O

O

O

O

Grubbs catalyst

O

O

O

O

O

O

O

O

O

O

O

O

O

O O

O

O

O

O

O O

O O

O

O

O

OO

O

O O O

OO

O O

O

Figure 25

Monomolecular imprinting inside a dendrimer.

As will be discussed below, new imprinted material architectures have been described that exhibit superior mass transfer properties and saturation capacities. However, the heterogeneous distribution of binding sites appears to be a more difficult problem to overcome.55 Unavoidably, the kinetically controlled formation of the polymer network leads to a statistical distribution of binding site microenvironments. Several approaches to reduce this cause of site heterogeneity in MIPs have been proposed and demonstrated. By reducing the size of the MIP to form particles in the nanometer range, MIPs with on average one site per particle can be produced.56 In such systems, heterogeneity is mainly due to the nonequivalence of sites between the particles (cf polyclonal antibodies) and, as polyclonal antibodies, the particles can be fractionated by affinityenrichment procedures.57 This is not a limitation in singlemolecule imprinting, as first proposed by Zimmermann in the imprinting in the core of dendrimers (Figure 25).58 Here, the defined stepwise buildup of the dendritic scaffold assures that the binding sites are identical. Nevertheless, this approach requires strong and defined interactions between the binding site functional groups and the template (for details of dendrimers see Supramolecular Dendrimer Chemistry, Soft Matter). Another approach to address the heterogeneity issue is to shrink only one dimension of the MIP forming thin films or two dimensions forming fibers.59 Alternatively, factors related to the polymerization per se may be considered. One way is to use polymer matrices formed

by thermodynamically controlled polymerization reactions. This approach was suggested by Steinke and colleagues who used ring-opening metathesis polymerization (ROMP) to form networks under thermodynamic control in the presence of a template.60 The use of controlled radical polymerization (e.g., ATRP, atom transfer radical polymerization with iniferters, and RAFT, reversible addition fragmentation chain transfer) has been investigated by several groups as a means to produce more homogeneous polymer networks, with several encouraging results.61 The enhanced capacity observed in polymers prepared in the presence of iniferters was attributed by Vaughan et al. to the shorter kinetic chain length and a more narrow distribution of kinetic chains, which could lead to more uniform and a higher population of sites of the right size.

5.2

Particles

As mentioned earlier, MIPs are most commonly prepared as “bulk” monoliths which are then manually crushed to give smaller particles (Figure 26). These particles are then fractionated by size depending on the intended application. Commonly, particles between 25 and 36 µm are used in chromatographic applications, while larger sizes are used in other applications, for example, a size range of 36–63 µm is common for solid-phase extraction (SPE) applications. Such particle sizes are, however, too large for high efficiency in chromatographic separations, where particles of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

18

Soft matter Template Porogen Initiator

Cross-linker Functional monomer

N2

Heat or UV Polymer constituents are added

The template is extracted

Figure 26

The mixture is purged with nitrogen

...and sedimented

Polymerization is initiated

The polymer is removed from the tube...

The particles are sieved... ...and mechanically ground

The crushed monolith approach to MIP particles.

3–10 µm are ideal. Furthermore, the crushing process leads to irregularly shaped “rock-like” particles, which are again unsuitable for chromatographic applications. Thus, it would be advantageous to be able to prepare MIPs as regular, spherical particles with narrow size distributions.14 In traditional polymer chemistry, this is achieved either through suspension or emulsion polymerization, which allows polymer particles in different size ranges to be produced.13 Using water-stable covalent linkages to hold the template, such techniques may be used.62 However, the presence of large amounts of water as the continuous phase in such method tends to render these approaches unsuitable to the formation of noncovalent MIPs, as the water generally interferes with the interactions holding the template–functional monomer complexes together. This is not the case with the benzamidine-based functional monomers of Wulff and colleagues which have been used to prepare MIPs via aqueous suspension polymerization.63 There are also some examples of suspension polymerizations in media other than water, for example, the use of perfluoralkanes as the continuous phase.14 Some of the more important methods used to prepare noncovalent MIPs directly in particulate form are described below. Preformed silica particles have been used as supports to which imprinted layers have been grafted. In earlier approaches, polymerizable units were attached to the silica support followed by “grafting to” polymerization using template, functional monomer, cross-linker and soluble initiator.64 It is now more common to attach initiator or iniferter65 units to the silica support, allowing for a “grafting from” approach. The template molecule may also

be immobilized on the silica surface and polymerization allowed to occur within the porous structure of the beads.66 Etching of the silica support postpolymerization gives organic polymer beads containing surface-confined binding sites and a pore system that is a mirror to that of the original silica (Figure 27).29 Two-stage swelling polymerizations have also been used to create spherical MIP particles. Here, preformed seed polymer particles are swollen in a stepwise manner with porogen, template, and monomers in a stage akin to suspension polymerization, leading to uniformly sized MIP particles.67 Mini-emulsion polymerization is another emerging two-phase bead polymerization technique68 and, more recently, Shea and colleagues have used an inverse microemulsion polymerization technique to prepare nanoparticles with high affinity for a hydrophilic nonapeptide template.69 The relatively straightforward method of precipitation polymerization has become an increasingly popular method for producing spherical MIP particles. While earlier reports led to particles in the submicron size range,70 optimization of conditions allows for larger, micron-sized beads to be prepared (Figure 28).71 Haupt and colleagues have recently reported the use of this method for the preparation of microgels able to act as enzyme inhibitors,50 while Shea and colleagues originally used this method for creating mellitinspecific nanoparticles.51 A drawback of this method is the high dilution required to avoid agglomeration of the growing particles.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

(a)

19

(b)

Figure 27 Scanning electron micrographs of the (a) precursor silica template and of (b) P(FMOC-Phe-Si) after dissolution of the silica template. The micrographs were obtained using a LEO 1530 “Gemini” scanning electron microscope (LEO Elektronenmikroskopie GmbH, Oberkochen, Germany) at a magnification of 1000×. Scale bar represents 20 µm.

5.00 µm

P1

Φ 10.0 µm

P2

Φ

Figure 28 Scanning electron micrographs of theophyllineimprinted (P1, bar = 5 mm) and nonimprinted (P2, bar = 10 mm) microspheres. (Reproduced from Ref. 71.  Wiley-VCH, 2003.)

5.3

Films

As mentioned in Section 5.2, the conditions needed to generate sites of high affinity are often incompatible with the conditions needed to obtain polymer beads. Therefore, tedious trial and error is often needed to arrive at an acceptable compromise. These problems can be overcome by

grafting techniques where the MIPs are grown on preformed support materials of known morphology.72 The “grafting from” approach (Figure 29) relying on immobilized radical initiators leads to chain growth mainly confined to the support surface. In this way, the thickness of the grafted polymer can be effectively controlled and a high density of grafted polymer chains can be achieved compared with that obtained in the technically more straightforward “grafting to” procedure. Thin-film MIPs feature significantly improved mass transfer properties when compared to previously described monolith MIPs in liquid chromatography. MIP-based chemical sensors is another area where MIPs in layer or thin-film format are becoming increasingly important.5, 73 The main problems with this application are the slow responses and poor sensitivities for the target analyte. Mass transfer may be overcome by applying thin-film grafting techniques and sensitivity by coupling the molecular recognition elements to a sufficiently sensitive signal transducer. Ultrathin films of metal oxides were used by Kunitake and Lee to produce such thin films exhibiting fast response and very high selectivity.74 On the basis of thin films of the common matrix component, TiO2 , imprinting of a variety of Lewis basic chiral templates capable of coordination with Ti have been demonstrated. The films can be coated on metal or metal oxide substrates for use as sensing elements for QCM- or ISFET-based readings. Other reports confirm that sol–gel films exhibit very promising features in terms of response times, selectivity, and sensitivity, compared to organic polymer-based coatings.73 Although the molecular-level origin of imprinting remains largely unknown, the use of conducting polymers and electrosynthesized polymers as scaffolds for imprinting, notably protein imprinting, is rapidly progressing.75–78

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

20

Soft matter “Grafting from’’

“Grafting to’’

Figure 29

5.4

R

Grafting approaches to prepare thin organic polymer films on modified support materials.

Monoliths

As opposed to the standard preparation of MIPs as “bulk” monoliths, which must be ground and sized prior to use, there have been some reports on the in situ preparation of porous, monolithic MIPs.15 At first glance, this would appear to be an attractive proposition, as it should simplify the MIP preparation and allow for better performance in chromatographic applications. For instance, if the synthesis conditions could be chosen to create macropores in the monoliths, allowing convective mobile-phase transport, they could be directly applied in the many chromatographic modes already demonstrated for MIPs. However, the melding of the required flow-through and recognition properties is far from trivial, leading to much trial and error before a successful system is discovered. The first reports concerned the preparation of monolithic polymer rods for use in HPLC applications, using L-phenylalanine anilide as the template and leading to enantioselective flow-through monoliths.79 These relied on the use of mixtures of good and poor solvents to generate the flow-through pores. Owing to the unsuitable porogen, they exhibited much poorer enantioselectivities compared to the conventional crushed monoliths. However, for more strongly associating monomer–template systems, a higher selectivity was observed. Thus, in situ dispersion polymerization was used to create monolithic stationary phases selective for the antimicrobial agent, pentamidine.80 While the above reports deal with HPLC applications, there have been a greater number of reports on the preparation of monolithic MIPs within fused silica capillaries, especially for use in capillary electrochromatography (CEC).81 Such preparations can be very time efficient. For example, a monolithic capillary imprinted against S-propranolol was operational within 3 h of beginning the preparation and allowed for baseline separation of propranolol enantiomers, with short analysis times. More recently, a slightly modified approach allowed the preparation of flow-through monoliths for liquid chromatographic enantiomer separations. Thus, Huang et al. observed that small amounts of toluene in

dodecanol produced superporous packings exhibiting high enantioselectivities.82 The complexity in finding suitable conditions for achieving flow-through properties simultaneous to enantioselectivity appears clearly from the sensitive dependence of the separation factor and back pressure on the content of toluene in the porogenic mixture. A way around this problem is again to resort to grafting chemistry starting from a mass-producible monolithic structure. Thus, Irgum and colleagues developed a flow-through monolithic capillary based on polytrimethyloltrimethacrylate (p-TRIM) which, by virtue of the incomplete doublebond conversion, features numerous unreacted pendant double bonds.39, 83 The latter can act as anchor points for the grafting of a subsequent MIP film as was demonstrated for numerous independent templates. After the grafting step, the capillary could be directly connected to an LC instrument and used as an affinity stationary phase in micro-LC.

5.5

Membranes

Membranes tailored at the nanoscopic level are expected to find important applications in the area of environmental or industrial separations and purifications. The imprinted membranes developed thus far can be classified according to the transport mechanism imparted by the imprinting step (Figure 30).84 Thus, imprinting may lead simply to a programmed adsorption of a given compound inhibiting its transport through the membrane, but examples of membranes where imprinting leads to active transport or so-called gate effects also exist. The methods used for producing the imprinted membranes were, for a long time, dominated by a one-step imprinting protocol, combining the commonly used phase inversion techniques for producing anisotropic microporous membranes directly with imprinting. This led to affinity membranes with demonstrated imprinting effects for targets of sizes ranging from small molecules to proteins. Thus, Kobayashi et al. used copolymers of acrylonitrile with acrylic acid for imprinting by phase inversion, yielding anisotropic porous membranes.85

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

A

B A (a)

A

B (b)

A

A

(c)

(d)

21

B

Figure 30 Separation mechanisms for molecularly imprinted membranes: (a) transport of A driven by a concentration gradient is facilitated, whereas the nonspecific transport of another substance B is hindered, (b) transport of A is retarded by binding to MIP sites on the surface of transmembrane pores, whereas another substance B is transported by diffusion or convection (membrane adsorber), (c) the membrane permeability is increased, for example, because of an increase of membrane swelling, (d) the membrane permeability is deceased, for example, because of a decrease of membrane swelling as a consequence of A binding to MIP sites. (Reproduced from Ref. 84.  Elsevier, 2004.)

The binding sites in these materials rely on noncovalent entanglement of polymer chains which exist as long as the membrane is exposed to a nonsolvent. Thus, the memory effects can be erased by solvent exposure and the method hence suffers from a limited robustness. Instead of this one-pot approach, composite membranes are growing in importance. Here the MIP formation is decoupled from the phase inversion process or technique to form a membrane support of choice. This support can then be used for grafting MIPs either in the film format or by attaching preformed MIP particles.84

(a) Application: urine or plasma sample containing analyte OH Cl

H N

(b) Aqueous wash (c) Organic wash

(d) Elution

H2N Cl

Clean and concentrated analyte O − Cl

O

O

OH H H N +

H2N OH

Cl

Brombuterol

6

6.1

APPLICATIONS OF MOLECULARLY IMPRINTED POLYMERS Solid phase extraction and trace analysis

The detection and quantification of high-priority analytical targets typically rely on labor-intensive methods that often fail to satisfy the required method criteria in terms of speed, robustness, and sensitivity. The bottleneck and the source of the poor performance of these methods commonly are found in the sample pretreatment steps, which are often lengthy and laborious. This illustrates the need for developing improved sample extraction techniques capable of cleaning up and enriching samples to easily measurable levels. One of the most powerful and reproducible ways of doing this is by molecularly imprinted solid phase extraction (MISPE) (Figure 31).86 Numerous examples now demonstrate the superior performance of MIPs over more generic phases in delivering clean extracts and reproducible recoveries of, mainly, lowmolecular-weight analytes. Several examples of the successful application of MIPs in solid phase extraction have been described in the literature and the field is presently experiencing a rapid growth with nearly 200 papers published during 2010. The main reason for the success of

Figure 31 Principle of molecularly imprinted solid phase extraction (MISPE).

MISPE is that it offers an affordable and robust mode of programmable affinity SPE to capture single analytes or groups of analytes from complex samples. Thus, clean extracts can be obtained with minimal sample pretreatment, translating into significant time savings, lower detection limits and rugged methods. For example, this was shown in a recent report describing the analysis of the tobacco-specific biomarker NNAL (4) in urine samples.24 This lung carcinogen is an indicator for tobacco exposure in smokers as well as in nonsmokers exposed to second-hand smoke. The low concentrations found, particularly in the latter group, require highly sensitive and selective analytical methods. Thus, an MIP targeting NNAL was developed (vide supra) and implemented in an off-line SPE method for urine analysis. The resulting method was sufficiently sensitive (LOD (limit of detection) = 1.7 pg mL−1 ) precise, accurate, and fast to be implemented in large-scale analysis in support of epidemiological investigations. The time savings of the method was considerable and in the order of several hours.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

22

6.2

Soft matter

Process scale purifications

which mimic the action of the natural female hormone 17β-estradiol and, hence, may affect the reproductive capacity of exposed organisms. MIPs capable of removing all chemicals with estrogen-like activity were proposed by Mattiasson and colleagues as a general tool for this remediation.87 Hence, using the MIP as a substitute for classical adsorbents (e.g., granulated activated carbon), trace amounts of 17-β-estradiol could be effectively removed from wastewater effluents.

The demand for robust and economic alternatives to perform affinity-based separations at a large scale is growing in several branches of chemistry. For instance, environmental and health protection requires techniques for selectively purifying water from persistent organic pollutants causing adverse health effects. The same criteria apply to the food industry with respect to their raw materials, whereas food processing in general may benefit from techniques allowing selective removal of spoilage agents or specific flavors. In addition to such sequestration applications, robust capture agents are also needed. This is a great concern in the biopharmaceutical industry, where purification is the production bottleneck, causing delays in the launching of drugs that could save millions of lives. It is in such applications where stationary phases based on robust and affordable MIPs show distinct advantages. Three examples will be discussed below, referring to the use of MIPs in three widely different contexts. An important governmental challenge is to provide effective water purification technologies capable of assuring the effective removal of trace amounts of persistent pollutants that presently, due to insufficient remediation, escape the water treatment plants.87 As recent reports have suggested, these challenges could benefit from robust MIPs which, due to their high affinity, would be capable of trapping very low concentrations of such pollutants. A particular problem concerns so-called endocrine-disrupting chemicals,

O

OH 17

H

1 2

H

H HO 3

4 17-b-estradiol

One of the long-standing problems in the beverage industry is how to prevent chemical processes in the drinks compromising their taste, quality, and shelf life. For example, riboflavin (vitamin B2) is responsible for driving photooxidation reactions that affect the flavor of beverages such as beer, wine, and milk. Hence, they are packaged in light-shielded containers.88 One obvious alternative is to remove riboflavin but this would require an extremely selective receptor that would leave the remaining constituents of the drink untouched. A water-compatible MIP for riboflavin was capable of at least partly solving this task (Figure 32). Hence treating beer with the MIP resulted in removal of nearly 90% of the vitamin.

O O

O NH

O

N O

NH

O

NH

O

+

N

HN

N

−

O

N

O

OH O

NH

N

HN

N

N

O

OAc

OH

N

N OAc

OH

OAc AcO

OAc AcO

AcO AcO OH OH O O

O O

OH

OH

Figure 32 Hydrophilization of a riboflavin MIP by a hydrolytic post-treatment. (Reproduced from Ref. 88.  Royal Society of Chemistry, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

Co

N

O O −

O Co O O O N

−

O O

O O −

N OH O O HO N

−

O O Fe

Figure 33 A cobalt-imprinted polymer showing total rejection of iron. (Reproduced from Ref. 21.  American Chemical Society, 2009.)

Structural materials, such as carbon steel in power plants’ water cooling systems, form deposits of metal oxides when they interact with coolants. In nuclear power plants, these oxides trap radioactive ions, leading to buildups of radioactivity that require costly cleanups of reactor surfaces. Cobalt, present in some alloys used in the reactors’ water systems, is a major contributor toward this problem because of its long half-life (Figure 33). In order to address this problem, an MIP was developed which displayed a near absolute selectivity for cobalt and completely disregarded iron-based ions.21 The potential for using such trapping agents in decontamination processes in reactors was demonstrated.

6.3

Chemical sensors and assays

As MIPs possess an inherent specificity, there would appear to be many opportunities for their integration into biomimetic sensor devices.5 There are a large number of reports on the incorporation of MIP layers on QCM devices. The MIP layers used in these devices have been diverse, from polyurethanes, through (meth)acrylate and styrenic polymers, to inorganic sol–gel materials. This work has been pioneered mainly by the group of Dickert, who have demonstrated devices that are of use in real-world applications, for example, in monitoring the degradation of engine oil and the content of polyaromatic hydrocarbons in water.89 Other mass-sensitive devices incorporating MIPs have also been reported, for example, bulk- and surface acoustic wave-based devices. Fluorescence-based sensing systems offer the possibility of high levels of sensitivity and MIPs containing reporter molecules have been the subject of considerable research.5 There are two ways in which this may be achieved within MIPs. The simpler of the two is simply imprinting a molecule with inherent fluorescence, as demonstrated by Matsui et al. in the imprinting of cinchonidine and cinchonine.90 More challenging is the placement of a

23

fluorescent reporter group within or near the MIP binding cavity. An example of this type of methodology was reported by Rathbone and colleagues, the MIP exhibiting selective recognition of cyclic GMP.91 Other sensor platforms have also been explored, such as surface plasmon resonance, field-effect transistors, other optical/spectrophotometric methods, for example, Raman spectroscopy and chemiluminescence, and electrochemical techniques. Very recently, Cai et al. have demonstrated that arrays of carbon-nanotube tips with an imprinted nonconducting polymer coating can recognize proteins below the picograms per liter level, using electrochemical impedance spectroscopy.77 Devices for the specific recognition of human ferritin and human papillomavirus-derived E7 protein were described (Figure 34). MIPs hold promise as potential replacements for antibodies in pseudoimmunoassays, offering potential benefits, such as enhanced stability, the ability to function in both aqueous and organic solvents, and the avoidance of the need to use laboratory animals. There are some drawbacks, however, such as the need for the template for MIP production and the polyclonality generally exhibited by MIPs. Radioligand displacement assays dominate this area, such as in the first MIP-based immunoassay reported by the group of Mosbach.92 Here, MIPs prepared against theophylline and diazepam, respectively, were shown to display comparable cross-reactivities to antibodies raised against the two drugs, with respect to a range of structural analogs. Other types of displacement assay have also been reported, for example, those based on optical detection methods after the release of template analogs tagged with chromo- or fluorophores. In one of the more recent reports, Urraca et al. described an automated online fluorescent competitive assay for penicillin-type β-lactam antibiotics.93

6.4

Therapeutic applications

MIPs have attracted attention as potential vehicles for drug delivery, especially for the controlled release of therapeutic agents. The vast majority of such reports have concerned the imprinting of hydrogels.94 Such materials are ideal for use as contact lenses and much effort has focused on the area of ocular drug delivery, for example, for the delivery of timolol in the treatment of glaucoma.95 MIPs have also been shown to exhibit enantioselective-controlled drug release in the delivery of the enantiomers of the beta-blocker propranolol, depending on which enantiomer was used in the synthesis of the MIP.96 Recently, a pH-responsive drug delivery system for enantioselective-controlled delivery of the proton pump inhibitor omeprazole has been reported.97

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

24

Soft matter

Grow nanotube array

Low High Protein entrapment

100 nm Embed the array Low High Template removal 2 µm Polish the array

(b)

50 nm

Low High Protein rebinding

Carbon sheets

Deposit PPn

PPn 70 nm

(a)

Glass substrate

Ti

Supporting polymer

PPn coating

Nanotube

Protein template

hFtn

(c)

Figure 34 Fabrication of protein nanosensor based on imprinting. (a) Schematic of nanosensor fabrication and template protein detection. The supporting polymer is spin coated on a glass substrate containing nanotube arrays. Template proteins trapped in the polyphenol (PPn) coating are removed to reveal the surface imprints. Inset: hypothetical sensor impedance responses at critical stages of fabrication and detection. (b) Scanning electron microscopy image of a polished nanotube array after PPn coating. Inset: cross section of nanotube tip after polishing. (c) TEM images of PPn-coated nanotube tip without (top) and with (bottom) hFtn. (Reproduced from Ref. 77.  Nature Publishing Group, 2010.)

While interest in MIPs as drug delivery materials existed for over a decade, the idea of using MIPs as therapeutic agents in their own right is a more recent development. Haupt and colleagues reported on the preparation of molecularly imprinted microgels capable of inhibiting the enzyme trypsin.50 This was achieved by using a polymerizable version of a known inhibitor as a functional monomer (Figure 20). The monomer bound to the enzyme active site and the imprinted polymer was formed around the enzyme–monomer complex. The MIP microgels were selective for trypsin and were found to competitively inhibit trypsin with Ki = 79 nM, an increase of potency of about three orders of magnitude over the low-molecular-weight inhibitor, benzamidine, on which the functional monomer was based. A second example of MIPs as therapeutics has been reported by Shea and colleagues.51, 98 They have developed MIP nanoparticles capable of binding to a peptide toxin using precipation polymerization (Figure 21). The chosen toxin was mellitin, the principal active component of bee

venom. The nanoparticles exhibit antibody-like affinity for the mellitin ex vivo and have recently been shown to recognize, neutralize, and clear (via the liver) the toxin from mice in in vivo experiments.99

6.5

Synthesis and catalysis

Molecular imprinting offers the intriguing possibility of offering the synthetic chemist the opportunity to perform selective reactions in highly defined stereochemical environments. As MIPs can be designed to perform almost any chemical task and are inherently stable, their use in industrial processes would be especially interesting. From a more fundamental perspective, there is also the drive to create artificial materials that can truly function as enzyme mimics. This section will focus on a few of the more prominent examples of the use of MIPs in synthesis and catalysis.4 The first reports of MIPs employed in chemical synthesis came from Damen and Neckers62 and were followed by

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers the work of Shea et al.100 and Wulff et al.101 In these cases, MIPs can be viewed as “microreactors,” the latter possessing the ability to induce enantiomeric excess through the orienting of reacting groups within the MIP binding cavities. An intriguing possibility is to use MIPs not only as microreactors, but concurrently as drug screening aids, as demonstrated by Yu et al. in a so-called “anti-idiotypic” approach.102 A well-characterized inhibitor of the proteinase kallikrein was used as template to prepare MIPs that were subsequently employed as reactors to create novel inhibitors of the enzyme. Catalysis using imprinted polymers can be broken down into two major classes. The first concerns the imprinting of low-molecular-weight catalysts that are known to show catalytic activity in solution. Such catalysts are rendered polymerizable and then complexed with a suitable template, for example, the reaction substrate or a transition-state analog (TSA). In some ways, these materials can be said to mimic metalloenzymes. Early examples include the work of Lemaire, using polyurethane-supported rhodium catalysts to promote hydride transfer for the formation of alcohols.103 Later examples have come from the groups of Severin, on the catalysis of hydrogenation reactions,104 and Gagn´e, on the platinum catalyzed ene reaction.105 Cammidge et al. have reported the use of MIP-based catalysts for the Suzuki coupling reaction.106 The second type of MIP-based catalysts can be classed as bioinspired as, at least in the early stages, they were based on the catalytic sites to be found in enzymes. For example, chymotrypsin has served as the model of choice, with these systems incorporating some or all of the key features of the active site, that is, the Ser-His-Asp catalytic triad, transition state stabilization, and a stereoselective binding pocket, into the synthetic polymer. The first example, from Mosbach’s group, used imidazole residues as the catalytically active groups in the hydrolysis of 4-nitrophenyl esters of amino acids.107 Indeed, much of the early work was performed using activated substrates of this type. Sellergren and Shea used a hybrid covalent/noncovalent imprinting approach to incorporate most of the catalytically important features of chymotrypsin into the polymer matrix.108 The catalyst showed modest rate enhancements and a degree of enantioselectivity, the hydrolysis of the enantiomer corresponding to the TSA template being more rapid. These catalysts were also effective in hydrolyzing the nonactivated ethyl esters, although at lower rates than for the p-nitrophenyl esters.109 Far more active catalysts have been reported by the group of Wulff, using the benzamidine-based monomer mentioned in Section 3.2 in conjunction with a phosphonate TSA template.34 The complex between these two species

25

was envisaged to create an “oxyanion hole” for transitionstate stabilization, with the basic amidine residues bringing extra stabilization. The hydrolysis of a nonactivated phenyl ester was found to be accelerated more than 100-fold in the presence of the MIP catalyst, which also showed Michaelis–Menten kinetics. However, the turnover was low and the products and template were shown to completely inhibit the catalysis. To avoid the issue of product inhibition, Wulff et al. investigated the hydrolysis of carbonates and carbamates, which release products with no significant affinity for the amidine active site. Rate enhancements for the hydrolysis of diphenyl carbonate and diphenyl carbamate were found to be 588 and 1435, respectively, compared to the uncatalyzed reaction.63 The most recent reports from this group demonstrate that catalysis on a par with or greater than catalytic antibodies is possible with MIPs. These materials, which are described as carboxypeptidase A mimics, contain the amidine moiety together with a metal ion center chosen from Zn(II) or Cu(II).110, 111 The Zn(II)-containing catalyst showed rate enhancements greater than 3000-fold for the hydrolysis of an aromatic carbonate, while switching to Cu(II) allowed for rate enhancements greater than 105 for the same hydrolysis reaction. The incorporation of two amidine groups within the active site (Figure 13), together with a Cu(II) center, has been reported to lead to materials showing rate enhancements as high as 410 000.35 While imprinted materials that can compete with catalytic antibodies have been demonstrated, there remains much to be done before MIPs will be able to supplant the supremacy of enzymes, in terms of reaction diversity, rate enhancement, stereochemical selectivity, and turnover.

7

CONCLUSIONS

Molecular imprinting is a highly interdisciplinary area of research due both to the broad knowledge required to make the materials and the broad scope with regard to applications. MIPs are presently used for routine analysis by government laboratories and food producers and also by doping laboratories, in pharmaceutical analysis and clinical analysis laboratories worldwide in analytical methods with unsurpassed performance in terms of sensitivity, selectivity, and robustness. The robustness of MIPs, together with their low cost, is attractive not only for analytical scale separations but also for preparative or process scale separations. This can concern removal of toxic or unwanted byproducts from process streams in the food industry, wastewater treatment or biopharmaceutical separations. However, while current imprinting techniques commonly succeed in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

26

Soft matter

generating MIPs showing strong affinity for low-molecularweight guests in organic solvents, they often suffer from a heterogeneous “polyclonal” distribution of binding sites, poor performance in water or other protic solvents, low capacity, slow mass transfer, or the binding may simply be too weak for a given application. A particular problem refers to targets that are either very polar and water soluble or nonpolar and poorly functionalized. Host–guest chemistry and covalent imprinting approaches can be productively used to address these problems. Computational techniques and combinatorial imprinting are other important tools for improving performance. Another issue concerns the influence of the polymerization technique and used format on the above properties. The use of thermodynamically controlled polymerization techniques and controlled radical polymerization produces more homogenous networks, which leads to reduced heterogeneity. The development of MIPs as nanometer thin films grafted on preformed support materials, as nanoparticles, core shell particles, thin fibers, or tubes are likely to produce MIPs exhibiting faster on/off rates, and immobilization of the template has been shown to result in enhanced binding site homogeneity and accessibility. These latter formats are also keys for unlocking applications in the biomedical, sensor, and nanotechnology fields, whereas other formats such as membranes will be important in process scale separations using MIPs. It is not only small molecule imprinting that results in effective recognition elements. Recent reports have shown that MIPs toward peptides and proteins can challenge antibodies in performance. Generic methods to produce such receptors are likely to lead to breakthroughs in affinitybased technologies, for example, separations, extractions, and sensing, which currently rely to a dominating extent on costly and labile biologically derived affinity reagents. In light of the above, we are convinced that MIPs will play an ever more important role in chemistry presenting solutions to currently known or unknown problems.

8. J. U. Klein, M. J. Whitcombe, F. Mulholland, and E. N. Vulfson, Angew. Chem. Int. Ed., 1999, 38, 2057. 9. W. Kuchen and J. Schram, Angew. Chem. Int. Ed. Engl., 1988, 27, 1695. 10. B. Sellergren, M. Lepistoe, and K. Mosbach, J. Am. Chem. Soc., 1988, 110, 5853. 11. T. P. Rao, R. Kala, and S. Daniel, Anal. Chim. Acta, 2006, 578, 105. 12. P. K. Dahl and F. H. Arnold, J. Am. Chem. Soc., 1991, 113, 7417. 13. D. C. Sherrington, Chem. Commun., 1998, 2275. 14. A. G. Mayes, Chapter 12: Polymerisation techniques for the formation of imprinted beads, in Molecularly Imprinted Polymers: Man Made Mimics of Antibodies and their Applications in Analytical Chemistry, ed. B. Sellergren, Elsevier Science B. V., Amsterdam, 2001, vol. 23, p. 305. 15. B. Sellergren, Chapter 13: Imprinted monoliths, in Monolithic Materials: Preparation, Properties and Applications, ed. F. Svec, Elsevier Science, Amsterdam, 2003. 16. K. Flavin and M. Resmini, Adv. Nanomater., 2010, 2, 651. 17. K. J. Shea, D. A. Spivak, and B. Sellergren, J. Am. Chem. Soc., 1993, 115, 3368. 18. K. Karim, F. Breton, R. Rouillon, et al., Adv. Drug Deliv. Rev., 2005, 57, 1795. 19. B. C. G. Karlsson, J. O’Mahony, J. G. Karlsson, et al., J. Am. Chem. Soc., 2009, 131, 13297. 20. M. J. Whitcombe, L. Martin, and E. N. Vulfson, Chromatographia, 1998, 47, 457. 21. A. Bhaskarapillai, N. V. Sevilimedu, and B. Sellergren, Ind. Eng. Chem. Res., 2009, 48, 3730. 22. O. Ramstr¨om, L. I. Andersson, and K. Mosbach, J. Org. Chem., 1993, 58, 7562. 23. C. Widstrand, S. Kronauer, H. Bjoerk, et al., Am. Lab. (Shelton, CT, U. S.), 2008, 40, 6. 24. Y. Xia, J. E. McGuffey, S. Bhattacharyya, et al., Anal. Chem., 2005, 77, 7639. 25. K. Nemoto, T. Kubo, M. Nomachi, et al., J. Am. Chem. Soc., 2007, 129, 13626. 26. A. Rachkov and N. Minoura, Biochim. Biophys. Acta, 2001, 1544, 255.

REFERENCES

27. M. Quaglia, K. Chenon, A. J. Hall, et al., J. Am. Chem. Soc., 2001, 123, 2146.

1. C. Alexander, H. S. Andersson, L. I. Andersson, et al., J. Mol. Recogn., 2006, 19, 106.

28. H. Nishino, C.-S. Huang, and K. J. Shea, Angew. Chem. Int. Ed., 2006, 45, 2392.

2. B. Sellergren, eds., Molecularly Imprinted Polymers: Man Made Mimics of Antibodies and their Applications in Analytical Chemistry, Elsevier Science B. V., Amsterdam, 2001, vol. 23, p. 113.

29. M. M. Titirici, A. J. Hall, and B. Sellergren, Chem. Mater., 2003, 15, 822.

3. G. Wulff, Angew. Chem. Int. Ed. Engl., 1995, 34, 1812. 4. G. Wulff, Chem. Rev., 2002, 102, 1. 5. K. Haupt and K. Mosbach, Chem. Rev., 2000, 100, 2495. 6. M. J. Whitcombe, M. E. Rodriguez, P. Villar, E. N. Vulfson, J. Am. Chem. Soc., 1995, 117, 7105. 7. K. Mosbach, Trends Biochem. Sci., 1994, 19, 9.

and

30. B. Dirion, Z. Cobb, E. Schillinger, et al., J. Am. Chem. Soc., 2003, 125, 15101. 31. C. Widstrand, E. Yilmaz, B. Boyd, et al., Am. Lab. (Shelton, CT, U. S.), 2006, 38, 12. 32. K. Tanabe, T. Takeuchi, J. Matsui, et al., J. Chem. Soc. Chem. Commun., 1995, 22, 2303. 33. S. K. Chang, D. Van Engen, E. Fan, and A. D. Hamilton, J. Am. Chem. Soc., 1991, 113, 7640.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Molecularly imprinted polymers

27

34. G. Wulff, T. Gross, and R. Sch¨onfeld, Angew. Chem. Int. Ed. Engl., 1997, 36, 1962.

62. J. Damen and D. C. Neckers, J. Am. Chem. Soc., 1980, 102, 3265.

35. J.-Q. Liu and G. Wulff, J. Am. Chem. Soc., 2008, 130, 8044.

63. A. G. Strikovsky, D. Kasper, M. Gruen, et al., J. Am. Chem. Soc., 2000, 122, 6295.

36. C. S. Wilcox, E. Kim, D. Romano, et al., Tetrahedron, 1995, 51, 621. 37. A. J. Hall, P. Manesiotis, M. Emgenbroich, et al., J. Org. Chem., 2005, 70, 1732. 38. J. L. Urraca, A. J. Hall, M. C. Moreno-Bondi, B. Sellergren, Angew. Chem. Int. Ed., 2006, 45, 1.

and

39. M. Emgenbroich, C. Borrelli, S. Shinde, et al., Chem. Eur. J., 2008, 14, 9516. 40. T. Hishiya, M. Shibata, M. Kakazu, molecules, 1999, 32, 2265.

et al.,

Macro-

41. S.-h. Song, K. Shirasaka, M. Katayama, et al., Macromolecules, 2007, 40, 3530. 42. G. Wulff and A. Sarhan, Angew. Chem. Int. Ed. Engl., 1972, 11, 341. 43. B. Sellergren and L. Andersson, J. Org. Chem., 1990, 55, 3381. 44. K. Takeda, A. Kuwahara, K. Ohmori, and T. Takeuchi, J. Am. Chem. Soc., 2009, 131, 8833–8838.

64. O. Norrl¨ow, M. Glad, and K. Mosbach, J. Chromatogr. A, 1984, 299, 29. 65. B. Sellergren, B. Ruckert, and A. J. Hall, Adv. Mater., 2002, 14, 1204. 66. E. Yilmaz, K. Haupt, and K. Mosbach, Angew. Chem. Int. Ed., 2000, 39, 2115. 67. K. Hosoya, K. Yoshizako, Y. Shirasu, et al., J. Chromatogr. A, 1996, 728, 139. 68. D. Vaihinger, K. Landfester, I. Krauter, et al., Macromol. Chem. Phys., 2002, 203, 1965. 69. Z. Zeng, Y. Hoshino, A. Rodriguez, et al., ACS Nano., 2010, 4, 199. 70. L. Ye, P. A. G. Cormack, and K. Mosbach, Anal. Commun., 1999, 36, 35. 71. J. Wang, P. A. G. Cormack, D. C. Sherrington, and E. Khoshdel, Angew. Chem. Int. Ed., 2003, 42, 5336.

45. D. E. Hansen, Biomaterials, 2007, 28, 4178.

72. C. Sulitzky, B. R¨uckert, A. J. Hall, et al., Macromolecules, 2002, 35, 79.

46. W. Turner Nicholas, W. Jeans Christopher, R. Brain Keith, et al., Biotechnol. Prog., 2006, 22, 1474.

73. A. Mujahid, P. A. Lieberzeit, and F. L. Dickert, Materials, 2010, 3, 2196.

47. S. Hjerten, J. L. Liao, K. Nakazato, matographia, 1997, 44, 227.

74. T. Kunitake and S.-W. Lee, Analyt. Chim. Acta, 2004, 504, 1.

et al.,

Chro-

48. S. Srebnik, Chem. Mater., 2004, 16, 883. 49. A. A. Vaidya, B. S. Lele, M. G. Kulkarni, and R. A. Mashelkar, J. Appl. Polym. Sci., 2001, 81, 1075. 50. A. Cutivet, C. Schembri, J. Kovensky, and K. Haupt, J. Am. Chem. Soc., 2009, 131, 14699. 51. Y. Hoshino, T. Kodama, Y. Okahata, and K. J. Shea, J. Am. Chem. Soc., 2008, 130, 15242. 52. H. Shi, W.-B. Tsai, M. D. Garrison, et al., Nature, 1999, 398, 593. 53. R. Ouyang, J. Lei, and H. Ju, Chem. Commun., 2008, 5761. 54. A. Nematollazadeh, W. Sun, C. S. A. Aureliano, et al., Angew. Chem. Int. Ed., 2011, 50, 495.

75. H. Shiigi, H. Yakabe, M. Kishimoto, et al., Microchim. Acta, 2003, 143, 155. 76. A. Menaker, V. Syritski, J. Reut, et al., Adv. Mater., 2009, 21, 2271. 77. D. Cai, L. Ren, H. Zhao, et al., Nat. Nanotechnol., 2010, 5, 597. 78. W.-H. Zhou, C.-H. Lu, X.-C. Guo, et al., J. Mater. Chem., 2010, 20, 880. 79. J. Matsui, T. Kato, T. Takeuchi, et al., Anal. Chem., 1993, 65, 2223. 80. B. Sellergren, Anal. Chem., 1994, 66, 1578.

55. K. D. Shimizu, Chapter 16: Binding Isotherms, in Molecularly Imprinted Materials: Science and Technology, eds. M. Yan and O. Ramstroem, Marcel Dekker, New York, NY, 2005, p. 419–434.

81. L. Schweitz and S. Nilsson, Chapter 16: Capillary electrochromatography based on molecular imprinting, in Molecularly Imprinted Polymers: Man Made Mimics of Antibodies and their Applications in Analytical Chemistry, (ed. B. Sellergren) Elsevier Science B. V., Amsterdam, 2001, vol. 23, p. 377.

56. G. Wulff, B.-O. Chong, and U. Kolb, Angew. Chem. Int. Ed., 2006, 45, 2955.

82. X. Huang, H. Zou, X. Chen, et al., J. Chromatogr., A, 2003, 984, 273.

57. Y. Hoshino, W. W. Haberaecker, III, T. Kodama, et al., J. Am. Chem. Soc., 2010, 132, 13648.

83. J. Courtois, G. Fischer, B. Sellergren, and K. Irgum, J. Chromatogr. A, 2006, 1109, 92.

58. S. C. Zimmerman, I. Zharov, M. S. Wendland, et al., J. Am. Chem. Soc., 2003, 125, 13504.

84. M. Ulbricht, J. Chromatogr. B, 2004, 804, 113.

59. F. Vandevelde, A.-S. Belmont, J. Pantigny, and K. Haupt, Adv. Mater., 2007, 19, 3717. 60. A. Patel, S. Fouace, and J. H. G. Steinke, Chem. Commun., 2003, 88. 61. A. D. Vaughan, S. P. Sizemore, and M. E. Byrne, Polymer, 2007, 48, 74.

85. T. Kobayashi, H. Y. Wang, and N. Fujii, Chem. Lett., 1995, 927. 86. B. Sellergren and A.-M. Esteban, Chapter 23: The use of molecularly imprinted polymers for sampling and sample preparation, in Handbook of Sample Preparation, eds. J. Pawliszyn, and H. L. Lord) John Wiley & Sons, Hoboken, New Jersey, 2010, vol. 1, p. 445.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

28

Soft matter

87. M. Le Noir, A.-S. Lepeuple, B. Guieysse, B. Mattiasson, Water Res., 2007, 41, 2825.

and

99. Y. Hoshino, H. Koide, T. Urakami, et al., J. Am. Chem. Soc., 2010, 132, 6644.

88. P. Manesiotis, C. Borrelli, C. Aureliano, et al., J. Mater. Chem., 2009, 19, 6185.

100. K. J. Shea, E. A. Thompson, S. D. Pandey, and P. S. Beauchamp, J. Am. Chem. Soc., 1980, 102, 3149.

89. F. L. Dickert, P. Lieberzeit, and M. Tortschanoff, Sens. Actuators, B, 2000, B65, 186.

101. G. Wulff and J. Vietmeier, Makromol. Chem., 1989, 190, 1727.

90. J. Matsui, H. Kubo, and T. Takeuchi, Anal. Chem., 2000, 72, 3286.

102. Y. Yu, L. Ye, K. Haupt, and K. Mosbach, Angew. Chem. Int. Ed., 2002, 41, 4459.

91. N. T. K. Thanh, D. L. Rathbone, D. C. Billington, and N. A. Hartell, Anal. Lett., 2002, 35, 2499.

103. P. Gamez, B. Dunjic, C. Pinel, and M. Lemaire, Tetrahedron Lett., 1995, 36, 8779.

92. G. Vlatakis, L. I. Andersson, R. Mueller, and K. Mosbach, Nature, 1993, 361, 645.

104. K. Polborn and K. Severin, Chem. Eur. J., 2000, 6, 4604.

93. J. L. Urraca Javier, C. Moreno-Bondi Maria, G. Orellana, et al., Anal. Chem., 2007, 79, 4915.

105. J. H. Koh, A. O. Larsen, P. S. White, and M. R. Gagne, Organometallics, 2002, 21, 7. 106. A. N. Cammidge, N. J. Baines, and R. K. Bellingham, Chem. Commun., 2001, 2588.

94. M. E. Byrne, K. Park, and N. A. Peppas, Adv. Drug. Deliv. Rev., 2002, 54, 149.

107. A. Leonhardt and K. Mosbach, React. Polym., 1987, 6, 285.

95. C. Alvarez-Lorenzo, F. Yanez, R. Barreiro-Iglesias, and A. Concheiro, J. Control. Release, 2006, 113, 236.

108. B. Sellergren and K. J. Shea, Tetrahedron Asymmet., 1994, 5, 1403.

96. R. Suedee, T. Srichana, and G. P. Martin, J. Control. Release, 2000, 66, 135.

109. B. Sellergren, R. Karmalkar, and K. J. Shea, J. Org. Chem., 2000, 65, 4009.

97. R. Suedee, C. Jantarat, W. Lindner, et al., J. Control. Release, 2010, 142, 122.

110. J.-Q. Liu and G. Wulff, J. Am. Chem. Soc., 2004, 126, 7452.

98. Y. Hoshino, T. Urakami, T. Kodama, et al., Small, 2009, 5, 1562.

111. J.-Q. Liu and G. Wulff, Angew. Chemie. Int. Ed., 2004, 43, 1287–1290.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc137

Supramolecular Dendrimer Chemistry Monica Soler1 and George R. Newkome2 1

Universidad de Chile, Santiago, Chile The University of Akron, Akron, OH, USA

2

1 Introduction 2 Dendritic Structures 3 Conclusion References

1

1 3 11 11

INTRODUCTION

Dendrimers are a class of well-defined and highly branched three-dimensional macromolecules, containing regularly repeating branch units attached to a core, whose size, shape, and topology can be precisely controlled by tuned synthetic procedures. The name dendrimer comes from the Greek word for tree (δενδρoν = dendron), but these macromolecules have also been known as cascade molecules1 or arborols.2 In the past 25 years,3–6 a variety of dendrimers with different chemical and structural characteristics have been designed, synthesized, and manipulated by chemists to obtain different three-dimensional structures in which functional groups are introduced into the core, at branches at each generation (Gn), and at the branch termini in order to fulfill a prescribed purpose. Owing to the number of terminal groups as well as multifunctional and potential large internal void regions at high generations, numerously applications of dendrimers have been reported, such as catalysis,7–10 biomedical applications,11, 12 drug delivery,11, 13, 14 molecular recognition,15, 16 molecular imaging, and light- and energy-harvesting materials,17 which have

supported their increased interest. Dendrons, which are the constitutional subunits of dendrimers, are each composed of a focal point and dendritic branching (Figure 1), which can be attached to a core (a metal ion, small molecules, or ligand) to form a dendrimer, as well as attached to nanoparticles, surfaces, nanotubes, and linear polymers as side groups.18 This chapter presents a selective glimpse of this dynamic family of spherical macromolecules for newcomers to the topic in order to help them better appreciate the field that has been extensively reviewed elsewhere. This chapter is divided into several parts to emphasize the structural diversity and their potential applications. First, a study of the internal structure of the dendritic architecture, emphasizing the different types, followed by a study of their interactions with other molecules or atoms, such as in the case of host–guest chemistry, molecular recognition, or encapsulation inside the dendrimer. Finally, there is a small section that will address the intermolecular interactions of dendrimers and dendrons to either themselves or other nano-objects. In this past quarter century, tens of thousands of papers have been published producing a wide variety of different dendritic architectures with varied structural components capable of novel supramolecular interactions. Therefore, only an overview describing their structure with representative examples and practical purposes will be discussed, when appropriate.

1.1

Historical perspectives

The first low molecular weight multiarmed molecules, poly(propylenimine) (PPI), were initially described in 1978 by Buhleier, Wehner, and V¨ogtle1 as “cascade Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

2

Soft matter

Building block Branch

Core

Branching center

Terminal groups

Focal point Branching center

Void

(a)

(b)

Figure 1 Schematic representation of a (a) dendrimer and (b) dendron, with the corresponding part names. Generations represented in different colors.

molecules” obtained through a repetitive-stepwise principle or “cascade-like” pathway annexing successive arms generating the desired branched target with an increasing cavity size; these structures were designed as low molecular weight polyamines. In 1979, Denkewalter, Kolc, and Lukasavage reported the divergent synthesis of a highmolecular weight polyamide using the protected aminoacid (L-lysine)19 ; however, they failed to fully characterize these novel and useful branched biopolymers. In 1985, Newkome et al.2 and Tomalia et al.20 independently reported different series of 1 → 3 and 1 → 2 branched macromolecules or branched fractal constructs, respectively, possessing three or more generations with characterized superstructures prepared via a divergent process, that is, prepared from the inside-out. Newkome’s research group synthesized threedimensional spherical structures focused on sp3 -carbon branching and an outer surface covered with polar functional groups, which they termed arborols and noted their intrinsic “unimolecular micelles” character.21, 22 Parallel work from Tomalia’s group described the use of a threedirectional core (ammonia) to generate high branching macromolecules with three or more tiers and reactive end groups, the poly(amidoamine) (PAMAMs) dendrimers, also termed “starburst dendrimers.”20 Their syntheses followed a strategy involving what was then called “time sequenced propagation” techniques commonly known today as the divergent synthetic approach. In 1990, Hawker and Fr´echet described the convergent (outside-in) approach, where dendritic polyethereal macromolecules, based on 3,5-dihydroxybenzyl alcohol as the building block, were synthesized.23 That same year, Neenan reported the convergent synthetic approach of dendrimers based on 1,3,5-trisubstituted benzenes.24 Since then, research on dendrimers has grown to such an extent

that dendrimers are considered an important class of macromolecular architectures.

1.2

Synthesis and structure

Dendrimers possess three distinguishing components: (i) an inner core, (ii) interior layers (generations—composed of repeating branching units and branches radially attached to the central core), and (iii) termini or end groups, attached to the outermost generation. A schematic representation of a dendrimer providing the terms for the different parts is presented in Figure 1. In each generation, the number of branches ideally multiplies; ultimately, the total number depends on the branching valency of both the monomeric building blocks and the core. Eventually, the addition of more generations leads to a dense packing limit, that is the stage of the dendritic growth where the total surface area of the dendrimers is less than the total area required for the stoichiometric amount of building blocks.25 When this limit is reached, additional building blocks can lead to incomplete surface connectivity, which introduces branching defects or structural imperfections. Therefore, the appropriate choice of the initial core, building blocks, terminal groups, and chemical methodology dictates their eventual shape, size, porosity or internal void regions, number of termini, hydrophobic/hydrophilic nature, solubility, functionality both inside and out, and structural stability, having some unique physical and chemical properties due to their molecular superstructure. The intrinsic viscosity of dendrimer solutions is maximized at a specific generation and its shape can change with increasing generation, where with lower generations they adopt a more open planar elliptical shape, but it changes to a more compact spherical shape at higher generations.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Supramolecular dendrimer chemistry

Route A1

3

Route B

Route A2

Figure 2

Schematic representation of the divergent (Routes A1 and A2) and convergent (Route B) approach.

Dendrimers are built in layers or generations, where a central core or zeroth generation (G0) possesses a specific number of active sites, where the branches, building blocks, or dendrons are attached. Figure 2 shows the two main synthetic routes to construct a dendrimer—the divergent approaches (A1 and A2) and convergent approach (B). Divergent approach (A1) builds the dendrimer outwardly, beginning the synthesis at the core and creating the branching point at each new surface moiety using traditional linear monomers, deprotection, and then through a repetitive addition of (dis)similar monomers. The second type of divergent approach (A2) is the attachment of (1 → 3 or 1 → 2) branched monomers, envisioned as a minidendron, to an appropriate core, deprotection, and then repeating the procedure; the advantage of this sequence is the more rapid growth and compact structural motif. The convergent approach (B) (outside-in) attaches preformed dendrons of specific generation to a core, in which the synthesis of the dendron begins at the outside of the dendrimer. Most divergent syntheses require excess monomer and lengthy chromatographic separation for higher generations, while convergent syntheses are limited to create lower generation dendrimers due to the accessibility of the focal functionality to attach to an appropriate polyfunctional core. Owing to the time, effort, and difficulty of such approaches, design improvements or new synthetic approaches for tailored dendritic syntheses has become a major challenge. Two such synthetic improvements are methods introducing either “Lego” chemistry26 or “click” chemistry.27 The “Lego” methodology uses three building blocks, one core (either A3 or B3 ) and two branched monomers (CA2 and DB2 ) introducing the monomers in consecutive reaction steps; whereas, “click” chemistry28, 29 involves a Cu(I)catalyzed, alkyne azide [3 + 2] cycloaddition reaction to form a 1,4-disubstituted 1,2,3-triazole connectivity, as the structural linker of the core and dendron. Dendrimers and dendrons are monodisperse, usually highly symmetric, spherical compounds, and characterized by their structural composition; however, the field can be

divided into either low- or high-molecular weight species of which the first include dendrimers and dendrons, and the latter dendronized or hyperbranched polymers. This chapter will be limited to the first group, where this family is generally monodisperse and possesses specific structural morphology.

2 2.1

DENDRITIC STRUCTURES Inside the dendrimer

Dendrimers, based on their bonding framework, can be classified into two main groups, covalent and noncovalent; the latter is where the core or branches contain other types of bonding interactions apart from covalent bonding, such as H-bonding, van der Waals, or metal–ligand bonding. Once these preformed dendritic structures interact with external entities, they become host of the incoming guests forming the new host–guest adduct. Binding between host and guest molecules occurs as a result of certain forces such as π-stacking, metal–ligand interactions, Hbonding, hydrophobic interactions, ionic interactions, or other such electrostatic attractions. Important applications of host–guest chemistry in dendrimers are drug, gene, and/or vaccine carriers; nitric oxide release vesicles; and for imaging purposes, due to their monodisperse, structurecontrolled macromolecular structure permitting molecular encapsulation.

2.1.1 Host dendrimer Covalent dendrimers Many covalent dendrimers have been reported, and the early ones that have already been introduced above are summarized in Figure 3, being (i) the PPI dendritic structure (also called polypropylenamine, POPAM), based on AB2 branching units owing to the trivalency of the amine nitrogen atoms30, 31 ; (ii) the Newkome-type 1 → 3

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

NH2

N

NH2

N

NH2

NH2

NH2

(b)

HO

NH HN

O

O

O NH

O

O

HN O

O

O O

O

O

NH HN

OH OH

HO OHOH

C5H11

HO HO

HO

HO

HO

HO

O

N H NH

NH O

OH OHOH

OH OH OH

O

O

O HO H OH

OH

OH OH

(c)

OH OH

OH

H2N O

O

O

H2 N

NH

NH

N

N

O

O

NH

HN

N

HN

HN

O

O N H

NH2

N

O

NH

H2N

O

H N

(d)

NH2

O

O

O

O

O

H3 C

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

Figure 3 (a) V¨ogtle-type PPI dendrimer, (b) Newkome-type “arborol” dendrimer, 27-arborol, (c) Tomalia-type PAMAM dendrimer, and (d) Fr´echet-type dendrimer, [G-2]3 -[C].

(a)

NH2 NH2

N

N

N

N

NH2

NH2

4 Soft matter

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Supramolecular dendrimer chemistry C-branched dendritic structures,6 based on either ether and amide linkages32 or simply amide connectivity33, 34 ; (iii) Tomalia-type PAMAM dendritic structure, differing from the POPAM by the presence of amide bonds, but keeping the amine AB2 as the branching center; and (iv) Fr´echettype dendritic structures, based on ether and phenyl groups as AB2 branching center. Description of the different types of covalent dendrimers reported will follow a classification based on the element content, such as only C, C–N, N–O, C–N–O, or with heteroatoms such as P, Si, and so on. Examples of porphyrin-, thiophene-, carbohydrate-, and fullerene-based dendrimers are described separately, since they can be bound to different types of frameworks. Several examples of C-based dendrimers have been found in the literature,3 with the exception of those composed of unsaturated aliphatic hydrocarbon branches, which are rare. Several dendritic structures containing saturated and/or aromatic subunits have been found and will be summarized. Polyphenyl dendrimers are structures based only on the repetition of benzene groups arranged symmetrically around a 1,3,5-substituted benzene core.35 Phenylacetylene dendrimers, which have a branching structure based on arene and triply bonded building blocks, were initially synthesized by Moore,36 following a convergent approach. Another carbon-based dendrimer, the supertriptycene dendrimer (C104 H62 ), is based on condensed arene components, which can be synthesized by repetitive Diels–Alder reactions.37 These dendritic structures are interesting due to their thermostability and ability for guest inclusion. Stilbenoid dendrimers are composed of E-conjugated double bonds substituted with a phenyl group on both carbon atoms of the double bond, with the branching unit being the benzene branching at 1,3,5-positions; they also undergo aggregation and depending on the generation number, several of these pure structures easily form liquid–crystalline phases.38 Examples of dendritic structures with N or O, apart from the previously described PPI-type or Fr´echettype dendritic structures, are the polyester-type dendrimers, based on repetition of ester groups with AB2 C branching multiplicity (C refers to a substituent that will no further branch). Examples of fractal structures containing both N and O atoms, apart from the above mentioned 1 → 3 C-branched and 1 → 2 N-branched structures, are the polylysine-based dendrimers, which are similar to PAMAM with amide bonds, but they have an 1 → 2 C-branching centers. They attract interest due to their lower toxicity, as potential therapeutic agents for use in boron neutron capture therapy and magnetic resonance imaging (MRI). Several dendrimers containing heteroatoms have been well studied, especially Si- and P-based. The Si-based dendrimers were the first to contain second-row heteroatoms and can be classified into polysilanes (Si–Si), carbosilanes (Si–C), carbosiloxanes (Si–O–C), siloxanes (Si–O–Si),

5

and carbosilazane (Si–N) dendrimers.39 The most important type of Si-based dendrimers is that of the carbosilane family due to their excellent chemical, thermal stability, well-studied Si–C chemistry, and high flexibility, some of which have been considered as liquid–crystalline materials and catalysts supports.36 Many reports have been also published on P-based dendrimers, which are based on phosphorus atoms located at the core, branching centers, and/or periphery.40–42 It is an advantage that the phosphorus atoms in different generations can be distinguished on the basis of their different chemical shifts and signal intensities in the 31 P NMR spectrum. These P-based dendrimers, due to their solubility in water and low toxicity, are attractive for a variety of medical applications. Other potential applications are as “modifiers” for material surfaces in the development of DNA chips and as new gelators. With the exception of Si- and P-based dendrimers, very few other heteroelements, other than bismuth and germanium, have been used to build dendrimers. Many covalent dendrimers are functionalized with several organic groups due to their potential utilitarian applications, such as with carbohydrate, porphyrin, and fullerene (Figure 4a and b).43, 44 An example of a hexaarylbenzeneanchored polyester porphyrin dendrimer (coordinated to Zn ion, explained later) is presented in Figure 4(b), where it possesses a covalent dendrimer bond to the porphyrin units.43 In other cases, the porphyrin units are present in diverse positions within the infrastructure. Carbohydratebased dendrimers (glycodendrimers)45 have been reported with many different structures depending on the position of the carbohydrate groups, which can act as the core unit, serve as branching centers, or function as terminal groups, in which the latter instills overall enhanced water solubility. Glycodendrimers are also suitable for supramolecular bonding and transport of active ingredients.46 Porphyrin-based dendrimers were synthesized initially, mainly with the porphyrin, as the core, due to the morphological resemblance of these macrostructures to natural hemoproteins.47, 48 Since then, numerous reports have been published with multiple porphyrin rings acting either as dendritic surface coatings or within the branches. The combination of the peculiar photophysical and redox features as well as dendritic structure has allowed porphyrin-based dendrimers to be employed in a broad range of uses, such as entity for reversible O2 binding, light-harvesting (LH), artificial photosynthesis, catalysis, biomedical applications, and optoelectronics. Thiophene-based dendrimers are composed of thiophene subunits that are regularly repeated throughout the different generations, generally following an AB2 branching multiplicity centered on the thiophene moiety. This family is a candidate for photon harvesting and electronhole transporting materials in novel organic light emitting diodes and solar energy conversion devices.49 Finally, many

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

O R

O

O R O R

O O O O

OO

O

O O O O

O O

RO

O

O

O

RO

O O

O

N H

O O

O

O

H N

O O

O

O O

O R

O

O

O R

O O

O

HN

N

O

O

N

N

HN

O

O

O O

O

R O

O

N H

H N

O R

OR

OR

OR

OR

OR

G2 (R = C3H17)

O R

O

O R

OO

O O

O O O O

O

O

R O

O

R O R O

O O O O

R O

O O

O

O

O O

O O

O

O

O O

O R

R O

O

O

O O

O R

O

NH

N

N

NH

N

O

O

O

O

O O

R O

R

(b)

R

R

R

R

R

R

O O

R

R

O O OO

O

O O

O

O

O

O

O

R

O

O

O

27

R

O O

O

O O O

O

O

R

O

O OO O

O

O

O

R

O

R

O

O

O O O O

O OO O

O

O

O

O

O OO O

O

O

O O

O O O O O

R

O O

O

O

O

O

OOO O

O O OO

O

O O

OO O O

R

R

O O

R

R

R

R

R

R

R

(c)

Dendron

Dcore ~ 18 Å

Figure 4 (a) Organic dendrimers: Fullerene-based dendrimer. (Reproduced with permission from Ref. 44.  Royal Society of Chemistry, 2009.) (b) Zn Porphyrin-based dendrimer, (Reproduced with permission from Ref. 43.  Wiley-VCH, 2006.) (c) Self-assembly of 12 conical dendrons forming a hollow spherical dendrimer. (Reproduced with permission from Ref. 50.  American Chemical Society, 2008).

(a)

RO

RO

RO

RO

R O

R

R O

R O

6 Soft matter

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Supramolecular dendrimer chemistry

7

examples of fullerene dendrimers have been reported, some of which possess porphyrin-C60 dyads constructed on their surface. A recent example based on the functionalization of PPI dendrimers with a C60 derivative showed enhanced absorption by increasing generation number.43 (a)

Noncovalent dendrimers Dendrimers that contain supramolecular interactions are mainly composed of metallodendrimers as well as with H-bonding and/or van der Waals interactions. Zimmerman et al. reported the synthesis of a family of dendrimers formed by the H-bond-mediated, self-assembly of six Fr´echet-type polyether dendrons.51 Also, Percec et al. using hydrogen bond and van der Waal interactions selfassembled functionalized polyether dendrons into cylindrical columnar or spherical assemblies that mimic certain types of viruses (Figure 4c).18, 50 In the early 1990s, Balzani and V¨ogtle52, 53 and Newkome et al.54–56 initiated the incorporation of metal ions into dendritic architectures but in totally different processes. Balzani’s group demonstrated the formation of bimetallic dendritic structures such as with Ru(II) and Os(II) centers using 2,3-bis(2-pyridyl)pyrazine as bridging ligands and bipyridines as terminal ligands; whereas, Newkome’s group reported the connection of dendritic architectures with a terpyridine (tpy) ligand at the focal point and possessing 1 → 3 C-branching multiplicity utilizing connectivity. Newkome et al.55 also reported other dendritic structures such as a dodecaruthenium macromolecule, employing in each terminal branch a tpy ligand coordinated to an Ru tpy ligand terminal group. The interest in metallodendrimers is due to the fact that it allows access to highly ordered materials with attractive magnetic, electric, photo-optical, and catalytic properties depending on the metal(s) utilized. A combination of characteristics of metallodendrimers utilizing different transition metals, for example, with redox-active metals on the surface or within dendritic structures, can undergo multiple electrontransfer processes, which can be used as molecular batteries, sensors, and catalysts.57 Other applications of metallodendrimers are in the area of luminescence materials, molecular switches, and biomedicine.54 The number and structural location(s) of the metal complexes in the dendritic structures are essential factors to control their functions in future applications. Metallodendrimers can be classified according to the metal ion’s position within the dendritic structure, such as (i) the structure’s core; (ii) connectors between branching centers; (iii) branching centers; (iv) terminal groups; or (v) metal ions attach using a supramolecular interaction to a preformed covalent dendrimer; the latter will be considered as a host–guest interaction (Figure 5).

(b)

(c)

Metal ion

(d)

(e)

Figure 5 Different positions that metal ions can play in metallodendrimers, such as the (a) core, (b) connectors, (c) branching centers, (d) termini, or (e) structural auxiliaries.

To synthesize a dendrimer with a metal ion at the core (Figure 5a), two synthetic approaches can be followed. The first synthetic approach is to generate a metal complex whose ligand framework is covalently substituted by dendritic branches. Early examples of the first approach were reported by Aida,58 Diederich et al.,59 and Cardona and Kaifer60 presenting either metallo-porphyrin or metallo-ferrocene dendrimers. The second synthetic approach involves the coordination of a metal ion (core) with preformed dendrons, following a convergent approach. Percec et al.50 reported that the functionalization of the focal point of the conical dendrons with diethylene glycol allowed the encapsulation of LiOTf and RbOTf in the center of the hollow sphere (Figure 6a). Kawa and Fr´echet.61 reported the self-assembly of lanthanide cations (Er3+ , Tb3+ , and Eu3+ ) surrounded by three convergent polyether dendrons. Balzani et al.62 also reported the selfassembly of photoactive G1 and G2 dendrimers built around a Ru(II) ion core and containing 12 and 24 bipyridine dendrons, respectively, with the peripheral naphthyl moieties. The first example of metallodendrimers containing metal ions as part of the connectors (Figure 5b) was synthesized by Newkome et al.,55 which presented a dendrimer containing 12 pseudooctahedral Ru(II) centers via connectivity. An extension of this family was the synthesis of a tetrahedral metallodendrimer, wherein each arm, two ruthenium ions were linearly connected by a connectivity.56 A well-known example of metal ions, as branching centers, (Figure 5c) is the family of dendrimers with metal complexes serving both as core and as branching centers, constructed solely from polypyridine ligands and transition metal ions by Balzani et al.64 Such metallodendrimers could be synthesized either convergently or divergently as well as using different transition metal ions (ruthenium/osmium).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

8

Soft matter Guest encapsulated

Dendron 4 to 1 ratio

Guest (a)

High pH Low pH

(b)

Figure 6 Host–guest chemistry: (a) Representation of the self-assembly of a hollow dendrimer host, metal ions guest. (Reproduced with permission from Ref. 50.  American Chemical Society, 2008.) (b) pH-Dependent encapsulation of pyrene in PPI-core : PAMAMshell dendrimers. (Reproduced with permission from Ref. 63.  American Chemical Society, 2009.)

Metallocenes are an important component of metallodendrimers containing metal ions as terminal groups (Figure 5d) due to their potential application in catalysis, electron-transfer processes, and molecular recognition.65 Many examples with ferrocene and cobaltocene have been reported, some using PPI as the infrastructure or as in the case of ferrocene using silicon-based dendrimers. Many other examples with metal ions as terminal groups with diverse functionality have been reported using a wide variety of metals such as Cr, Ir, Ru, Os, Zn, Cu, Pt, Pd, and so on.56 The supramolecular assembly of dendrimers via Hbonding has been accomplished by Astruc, Boisselier, and Ornelas in which commercial dendritic poly(propylenimine) cores are rapidly and reversibly H-bonded to triallylor tris-amidoferrocenyl phenol dendrons57 ; these novel supramolecular assemblies were used for the electrochemical recognition of H2 PO4 − and adenosine-triphosphate anions.66

2.1.2 Host–guest chemistry Since the beginning of dendrimer chemistry in 1985, dendrimers were considered to be examples of an “unimolecular micelle”2 denoting the possible application where the branched host was capable of guest inclusion due to their molecular infrastructure. Dendrimers are attractive host molecules for molecular recognition, since they adapt a globular conformation at higher generations creating internal cavities of different size and shape for binding guest

molecules such as simple organic molecules, nanoparticles, as well as metal ions. There is considerable interest in the use of dendrimers as unimolecular micellar carriers of water-insoluble drugs or for targeted delivery of drugs using the peripheral groups for tissue or cellular specificity.67 Several reviews on these different topics in biomedicine have been published summarizing recent advances in the field, such as nanocarriers for versatile vectors for gene delivery, NO release, Gd encapsulation for molecular imaging, and others.68, 69 Organic–organic Organic guests inside the dendritic hosts have been reported based on micelle-type studies. Many organic molecules can be captured within the internal cavities of the hostdendrimers, when the dendrimers are designed with a hydrophobic interior, water-soluble, and using a simple precipitation approach, functioning as aqueous solubilization of inherently water-insoluble species.21 The first organometallic dendrimer possessing metal ions as structural auxiliaries (Figure 5e) or forming a host–guest adduct was reported by Newkome et al.,70 where the organic dendrimer dodecaalkyne was treated with dicobalt octacarbonyl moieties to afford the corresponding cobalt-containing metallodendrimer termed “cobaltomicellane.” Dendrimers consisting of PPI, as cores, and PAMAM dendrons, as a shell, were investigated for the encapsulation of pyrene as a function of pH (Figure 6b). It was found that as the pH and generation size increased, additional pyrene was encapsulated in the hydrophobic PPI cavity (Figure 6b).63 Dendrimers can also be constructed in the presence of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Supramolecular dendrimer chemistry guest molecules such as 3-carboxypropyl, tetracyanoquinodimethane (TCNQ), among others; thus, when PPI with 64 amine end groups was terminally capped, a dendritic box was generated. The degree of structural perfection of the PPI determines the number of guests entombed.71 Dendrons can also be considered as guest molecules in molecular recognition phenomena, where the host species, such as cyclodextrins (CD), can interact with only specific groups, such as dansyl group or dendronized viologens, placed in the focal position.72, 73 The introduction of specifically located H-bonding sites within the dendritic infrastructure has been demonstrated to afford convenient docking sites for complimentary guests.74 Organic–inorganic Organic–inorganic host–guest chemistry include many possibilities: the host can be covalent or noncovalent dendrimers by themselves or containing groups such as crown ethers, porphyrins, CDs, or individual molecules, for instance, the previous three examples and curcurbit[n]urils (CB). Depending on which is the host, the guest can be an organic molecule in the case of metallodendrimers or metal ions or nanoparticles in the case of organic dendrimers. Thus, when the above mentioned molecules act as hosts, dendrons with a functionalized focal point or a dendrimer’s terminal groups could function as the guest. Many examples have been reported in the literature based on host–guest interactions.71 If dendrimers possess a hydrophobic interior, organic guests can be easily encapsulated; dendrimer-like “inverted unimolecular micelles” with hydrophilic interior can encapsulate ionic molecules. Sun et al.75 reported synthesizing an “inverted unimolecular micelle” by sonication of aqueous lithium chloride in dodecane leading to the incorporation of water and metal ions inside the dendrimer. Examples of dendrimers behaving as guests have been reported where the dendrimer contains multiple copies of a guest residue on their surface, such as multiferrocene- and multicobaltocenium-dendrimers (PPI skeleton), forming a host–guest adduct with CD hosts.65 Dendrimers have also been used to stabilize and control the growth of nanoparticles by forming interdendrimer complexes.75 Dendrimers are also attractive hosts for the inclusion of catalytically active metal nanoparticles or intradendrimer composites. Dendrimers are well suited to host metal nanoparticles because they have uniform composition and structure. Also, the encapsulated nanoparticles do not agglomerate inside the host, in that they are confined primarily by steric effects; therefore, a substantial fraction of their surface is still available for catalysis. The dendritic branches can also function as selective gates to control small molecules for catalysis, since the terminal groups can be designed to either control solubility or the linking to surfaces and polymers.76

2.2

9

Outside the dendrimer

A dendrimer’s surface can be composed of active or passive terminal functionality that may perform several roles such as (i) gates that control the entrance and egress of molecular guests to and from the internal void regions, (ii) control of solubility, (iii) manage linking to different surfaces or polymers, or (iv) participate in molecular recognition and specific intermolecular interactions.50 Owing to the type of functional surface moieties, preformed dendrimers can aggregate or interact with one another or associate with other entities outside the dendritic structure creating highly regular molecular nanostructures. Some of these supramolecular assemblies will be introduced such as the concept of megamers and core–shell tecto(dendrimers), talking about the self-organization of liquid crystalline (LC) dendrimers, and considering an example of an interaction of preformed dendrimers with other nano-objects such as nanotubes. The surface of dendrimers basically controls its solubility characteristics and is the best area for the attachment of diverse functionality in either a homogeneous or heterogeneous manner. There can be either a preplanned substitution pattern leading to a specific product77, 78 or a combinatorial approach leading to a distribution of related compounds. A heterogeneous combinatorial approach has been demonstrated via a surface coating using a mixture of related dendrons79, 80 possessing an isocyanate focal group; this leads to a rapid construction of nanoscale superstructures that are intermediate between a dendrimer and a classic polymer: in essence a “polycelle.”81, 82 A related approach reported by Baker et al.12 has utilized PAMAMs in which random functionalization lead to multifunctional medical nanodevices. The use of connectivity has also been utilized to generate combinatorially generated metallodendrimers83 ; glycodendrimers45 have also been generated to investigate tunability of multivalent interactions.84

2.2.1 Megamers Some attention has arisen to consider dendrimers as building blocks for the construction of nanoarchitectures possessing higher complexity and dimensions beyond the dendrimer.85 In an early paper, Tomalia et al.20 referred to their “starburst polymers” as a class of macromolecules obtained from the chemical bridging of starburst dendrimers. These macromolecules have been termed megamers, which have been defined as architectures derived from the combination of two or more dendritic macromolecules or poly(dendrimers). Many examples of randomly assembled as well as structure-controlled megamers have been reported, such as oligomeric covalent assemblies of dendrimers (i.e., dimers, trimers) referred

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

10

Soft matter

as low molecular weight megamers, dendritic oligomers such as [dendrimer]n , and dendritic gels (n > 10).86 Even an example of megameric dendrimer clusters capable of reversible cross-linking and dissociating to give back the dendrimers has been recently reported.87 A mathematical study was performed to understand the random packing of spheres, controlled megamers were synthesized following a traditional core–shell construction. These new core–shell nanostructures were called core–shell tecto(dendrimers). Two synthetic approaches have been reported to create such supramolecular structures: (i) the direct covalent reaction of a nucleophilic dendrimer core with an excess of an electrophilic dendrimer shell reagent and (ii) the self-assembly by electrostatic neutralization of cationic dendrimer core excess of anionic shell reagent, followed by the covalent bond formation method.88 The creation of megamers and core–shell tecto(dendrimers) goes beyond individual dendrimers, introducing new levels of molecular complexity to nanochemistry.

2.2.2 Liquid crystalline dendrimers Depending on the molecular structure of the dendrimer and interactions between them, dendrimers can self-organize to create liquid crystals. Dendrimers can also self-organize at interfaces and have been shown to be good building blocks for mono- and multilayers.89 Interest in the field of LC dendrimers has increased due to the multiple possibilities offered by combining the mesomorphic properties of single mesogenic subunits based on supramolecular properties

and architectures of dendrimers yielding highly functional materials.89 These LC dendrimers can form due to the intrinsic properties of their polybranched constitution, such as connectivity, multivalency of the focal core, multiplicity of the branches, and control of the geometrical rate of growth or generations, which can play a crucial role in the formation of the mesophases, but what actually forms the LC mesophase is the inherent desire of the entities to organize themselves into the lowest thermodynamic state. Several LC dendrimers have been reported having nematic, lamellar, columnar, cubic phases as well as other less conventional mesophases. One example of an LC dendritic structure is the so-called side-chain LC dendrimers, where dendrimers have mesogenic units attached to the termini of the flexible infrastructures, such as Si-containing LC, PAMAM LC, and PPI LC, as well as polyester and polyether LC dendrimers. Another LC type is that of mainchain LC dendrimers, where the branching centers are no longer single atom but instead anisotropic molecular moieties, and the anisotropic groups are present at each generation. A third type, that is the shape-persistent LC dendrimers, is formed due to their dendritic architecture, which is completely rigid and conjugated making them interesting due to their expanded and electron-rich core.

2.2.3 Dendrimers with other nano-objects Several examples have been reported talking about the interaction of dendrimers with other nano-objects, such as DNA, proteins, and others.67 Newkome et al.90 published

OCH6H13 N N N Ru N N N

H13H6CO

N N N Ru N N N

1.75 nm

N N Ru N N N N

N N N Ru N N N

H13H6CO

N N N Ru N N N

12+ 12PF6−

OCH6H13 N N Ru N N N N

OCH6H13

−12NaPF6

0

OCH6H13

NaO NaO O

O ONa O O O O

O NaO NaO

O O O ONa

NaO O ONa O ONa O O O O O O ONa O ONa NaO O

−n (NaPF6)12 12− 12Na+

1.92 nm

3.85 nm

Figure 7 Dendrimers associated with rigid nanosize macrocyclic ring forming a nanofiber. (Reproduced with permission from Ref. 90.  Wiley-VCH, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Supramolecular dendrimer chemistry the assembly of nanoscale composite fibers (Figure 7) through electrostatic interaction between the structurally rigid polycations of hexameric macrocycles and spherical polyanionic dendrimers forming a hexagonal rod of precise alternating composition. Martin et al.91 have also reported the self-assembly and multiencapsulation of C60 by TTF-based dendrimers to form well-ordered arrays of donor–acceptor materials, which could be used for the preparation of optoelectronic devices. The association of dendrimers and nanotubes of either carbon nanotubes (CNTs) or noncarbon nanotubes have also been studied. Caminade and Majoral92 have recently published an overview describing the formation of nano-objects, such as the functionalization of CNT with dendrimers or dendrons or the synthesis of SWCNT (single-walled carbon nanotube) using dendrimers, as catalysts. Functionalization of CNT with dendrimers or dendrons can be achieved by either covalent or noncovalent interactions, and they generally use PAMAM dendrimers or dendrons, leading to applications in various fields such as formation of nanoparticles, elaboration of (bio)sensors, and in biology.

3

CONCLUSION

Dendrimers and dendrons are appealing types of nanoscale, highly branched, macromolecules, which, because of their structure and properties, have attracted the interest of many researches worldwide. The preceding text has tried to summarize the different selective supramolecular aspects about their properties, structure, potential diversity, and applications to nonspecialized scientists. An introduction of these dendritic structures has combined a short description of the structure and synthesis with some historical perspectives, followed by a classification of dendritic structures, as covalent and noncovalent entities. Emphases have been given to their host–guest capacity to encapsulate small molecules, ions, or nanoparticles, as well as to interact with themselves or other nano-objects. The continued investigation in many fields of these unique architectures has produced a wide variety of branched fractal constructs, which undoubtedly will continue to spark the imagination of future synthetic architects.

11

4. G. R. Newkome, C. N. Moorefield, and F. V¨ogtle, Dendrimers and Dendrons: Concepts, Syntheses, Applications, Wiley-VCH, Weinheim, Germany, 2001. 5. G. R. Newkome and C. D. Shreiner, Polymer, 2008, 49, 1. 6. G. R. Newkome and C. D. Shreiner, Chem. Rev., 2010, 110, 6338. 7. D. Astruc and F. Chardac, Chem. Rev., 2001, 101, 2991. 8. D. Astruc, C. Ornelas, and J. R. Aranzaes, J. Inorg. Organometal. Polym. Mater., 2008, 18, 1. 9. D. Astruc, C. Ornelas, and J. Ruiz, Acc. Chem. Res., 2008, 41, 841. 10. D. M´ery and D. Astruc, Coord. Chem. Rev., 2006, 250, 1965. 11. W.-D. Jang, K. M. K. Selim, C.-H. Lee, and I.-K. Kang, Prog. Polym. Sci., 2009, 34, 1. 12. I. J. Majoros and J. R. Baker Jr, eds., Dendrimer-Based Nanomedicine, Pan Stanford Publishing, Singapore, 2008. 13. S. H. Medina and M. E. H. El-Sayed, Chem. Rev., 2009, 109, 3141. 14. J. B. Wolinsky and M. W. Grinstaff, Adv. Drug Delivery Rev., 2008, 60, 1037. 15. V. V. Narayanan and G. R. Newkome, Top. Curr. Chem., 1998, 197, 19. 16. W. Ong, M. G´omez-Kaifer, and A. E. Kaifer, Chem. Commun., 2004, 1677. 17. G. D. D’Ambruoso and D. V. McGrath, Adv. Polym. Sci., 2008, 214, 87. 18. B. M. Rosen, C. J. Wilson, D. A. Wilson, et al., Chem. Rev., 2009, 109, 6275. 19. R. G. Denkewalter, J. F. Kolc, and W. J. Lukasavage, US Patent 4,360,646, 1979. 20. D. A. Tomalia, H. Baker, J. Dewald, et al., Polym. J., 1985, 17, 117. 21. C. N. Moorefield and G. R. Newkome, Compt. Rendus Chim., 2003, 6, 715. 22. G. R. Newkome, C. N. Moorefield, G. R. Baker, et al., Angew. Chem. Int. Ed. Engl., 1991, 30, 1176. 23. C. Hawker and J. M. J. Fr´echet, J. Chem. Soc., Chem. Commun., 1990, 1010. 24. T. M. Miller and T. X. Neenan, Chem. Mater., 1990, 2, 346. 25. P. G. de Gennes and H. Hervet, J. Phys. Lett., 1983, 44, L351. 26. V. Maraval, J. Pyzowski, A.-M. Caminade, Majoral, J. Org. Chem., 2003, 68, 6043.

and

J.-P.

27. G. Franc and A. K. Kakkar, Chem. Soc. Rev., 2010, 39, 1536.

REFERENCES 1. E. Buhleier, W. Wehner, and F. V¨ogtle, Synthesis, 1978, 155. 2. G. R. Newkome, Z. Yao, G. R. Baker, and V. K. Gupta, J. Org. Chem., 1985, 50, 2003. 3. F. V¨ogtle, G. Richardt, and N. Werner, Dendrimer Chemistry: Concepts, Synthesis, Properties, Applications, Wiley, Weinheim, Germany, 2009.

28. W. H. Binder, Macromol. Rapid Commun., 2008, 29, 951. 29. R. K. Iha, K. L. Wooley, A. M. Nystr¨om, et al., Chem. Rev., 2009, 109, 5620. 30. M. H. P. van Genderen, E. M. M. de Brabander-van den Berg, and E. W. Meijer, in Advances in Dendritic Macromolecules, eds. G. R. Newkome, JAI Press, Inc., Stamford, CN, 1999, pp. 61–105, Chapter 2. 31. M. H. P. van Genderen, M. H. A. P. Mak, E. M. M. de Brabander-van den Berg, and E. W. Meijer, in Dendrimers

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

12

Soft matter

and other Dendritic Polymers, eds. J. M. J. Fr´echet and D. A. Tomalia, John Wiley & Sons, Ltd., West Sussex, UK, 2001, pp. 605–616, Chapter 26. 32. G. R. Newkome, X. Lin, and J. K. Young, Synlett, 1992, 53. 33. G. R. Newkome, R. K. Behera, C. N. Moorefield, and G. R. Baker, J. Org. Chem., 1991, 56, 7162. 34. G. R. Newkome and C. D. Weis, Org. Prep. Proced. Int., 1996, 28, 485. 35. Y. H. Kim and O. W. Webster, J. Am. Chem. Soc., 1990, 112, 4592. 36. J. S. Moore, Acc. Chem. Res., 1997, 30, 402. 37. K. Shahlai and H. Hart, J. Am. Chem. Soc., 1990, 112, 3687. 38. M. Lehmann, C. Kohn, H. Meier, et al., J. Mater. Chem., 2006, 16, 441. 39. J. Hu and D. Y. Son, Macromolecules, 1998, 31, 8644. 40. J.-P. Majoral and A.-M. Caminade, Chem. Rev., 1999, 99, 845. 41. J.-P. Majoral and A.-M. Caminade, Top. Curr. Chem., 2003, 223, 111. 42. A.-M. Caminade, Y. Wei, and J.-P. Majoral, Compt. Rendus Chim., 2009, 12, 105. 43. S. Cho, W.-S. Li, M.-C. Yoon, et al., Chem.—Eur. J., 2006, 12, 7576. 44. U. Hahn, J.-F. Nierengarten, F. V¨ogtle, et al., New J. Chem., 2009, 33, 337. 45. Y. M. Chabre and R. Roy, Curr. Top. Med. Chem., 2008, 8, 1237. 46. Y. M. Chabre and R. Roy, Advances in Carbohydrate Chemistry and Biochemistry, Elsevier, Oxford, UK, 2010, pp. 165–393, Chapter 6. 47. W. Maes and W. Dehaen, Eur. J. Org. Chem., 2009, 4719. 48. W. S. Li and T. Aida, Chem. Rev., 2009, 109, 6047. 49. G. Ramakrishna, A. Bhaskar, P. Bauerle, and T. Goodson III, J. Phys. Chem. A, 2008, 112, 2018. 50. V. Percec, M. Peterca, A. Dulcey, et al., J. Am. Chem. Soc., 2008, 130, 13079. 51. S. C. Zimmerman, F. Zeng, D. E. C. Reichert, and S. V. Kolotuchin, Science, 1996, 271, 1095. 52. V. Balzani and F. V¨ogtle, Compt. Rendus Chim., 2003, 6, 867. 53. V. Balzani, P. Ceroni, A. Juris, et al., Coord. Chem. Rev., 2001, 219–221, 545. 54. S.-H. Hwang, C. D. Shreiner, C. N. Moorefield, and G. R. Newkome, New J. Chem., 2007, 31, 1192. 55. G. R. Newkome, F. Cardullo, E. C. Constable, et al., J. Chem. Soc., Chem. Commun., 1993, 925. 56. G. R. Newkome, E. He, and C. N. Moorefield, Chem. Rev., 1999, 99, 1689. 57. D. Astruc, E. Boisselier, and C. Ornelas, Chem. Rev., 2010, 110, 1857. 58. Y. Tomoyose, D.-L. Jiang, R.-H. Jin, et al., Macromolecules, 1996, 29, 5236. 59. P. J. Dandliker, F. Diederich, J.-P. Gisselbrecht, et al., Angew. Chem. Int. Ed. Engl., 1995, 34, 2725.

60. C. M. Cardona and A. E. Kaifer, J. Am. Chem. Soc., 1998, 120, 4023. 61. M. Kawa and J. M. J. Fr´echet, Chem. Mater., 1998, 10, 286. 62. M. Plevoets, F. V¨ogtle, L. De Cola, and V. Balzani, New J. Chem., 1999, 63. 63. D. Kannaiyan and T. Imae, Langmuir, 2009, 25, 5282. 64. G. Denti, S. Campagna, S. Serroni, et al., J. Am. Chem. Soc., 1992, 114, 2944. 65. A. E. Kaifer, Eur. J. Org. Chem., 2007, 5015. 66. M.-C. Daniel, F. Ba, J. R. Aranzaes, and D. Astruc, Inorg. Chem., 2004, 43, 8649. 67. B. Mishra, B. B. Patel, and S. Tiwari, Nanomed. Nanotech. Biol. Med., 2010, 6, 9. 68. A. D’Emanuele and D. Attwood, Adv. Drug Delivery Rev., 2005, 57, 2147. 69. A. T. Florence, Adv. Drug Delivery Rev., 2005, 57, 2101. 70. G. R. Newkome, C. N. Moorefield, J. M. Keith, et al., Angew. Chem. Int. Ed. Engl., 1994, 33, 666. 71. J. F. G. A. Jansen, E. M. M. de Brabander-van den Berg, and E. W. Meijer, Science, 1994, 266, 1226. 72. H. W. Gibson, N. Yamaguchi, L. Hamiliton, and J. W. Jones, J. Am. Chem. Soc., 2002, 124, 4653. 73. W. Wang and A. E. Kaifer, Adv. Polym. Sci., 2009, 222, 205. 74. G. R. Newkome, J. Heterocycl. Chem., 1996, 33, 1445. 75. R. M. Crooks, M. Zhao, L. Sun, et al., Acc. Chem. Res., 2001, 34, 181. 76. X. Peng, Q. Pan, and G. L. Rempel, Chem. Soc. Rev, 2008, 37, 1619. 77. G. R. Newkome, H. J. Kim, C. N. Moorefield, et al., Macromolecules, 2003, 36, 4345. 78. A. V. Ambade, Y. Chen, and S. Thayumanavan, New J. Chem., 2007, 31, 1052. 79. G. R. Newkome, C. D. Weis, and B. J. Childs, Des. Monomers Polym., 1998, 1, 3. 80. K. C. F. Leung, F. Aric´o, S. J. Cantrill, and J. F. Stoddart, Macromolecules, 2007, 40, 3951. 81. G. R. Newkome, C. D. Weis, C. N. Moorefield, et al., Angew. Chem. Int. Ed. Engl., 1998, 37, 307. 82. G. R. Newkome, B. J. Childs, M. J. Rourk, et al., Biotechnol. Bioeng., 1999, 61, 243. 83. G. R. Newkome, K. S. Yoo, and C. N. Moorefield, Tetrahedron, 2003, 59, 3955. 84. M. L. Wolfenden and M. Cloninger, J. Am. Chem. Soc., 2005, 127, 12168. 85. S. Svenson and D. A. Tomalia, Adv. Drug Delivery Rev., 2005, 57, 2106. 86. D. A. Tomalia and D. R. Swanson, in Dendrimers and other Dendritic Polymers, eds. J. M. J. Fr´echet and D. A. Tomalia), John Wiley & Sons, Ltd.: West Sussex, UK, 2001, pp. 617–629, Chapter 27. 87. B. M. Kiran and N. Jayaraman, Macromolecules, 2009, 42, 7353. 88. D. A. Tomalia, S. Uppuluri, D. R. Swanson, and J. Li, Pure Appl. Chem., 2000, 72, 2342.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Supramolecular dendrimer chemistry

13

89. B. Donnio, S. Buathong, I. Bury, and D. Guillon, Chem. Soc. Rev., 2007, 36, 1495.

91. G. Fern´andez, L. S´anchez, E. M. P´erez, and N. Mart´ın, J. Am. Chem. Soc., 2008, 130, 10674.

90. P. Wang, C. N. Moorefield, K. U. Jeong, et al., Adv. Mater., 2008, 20, 1381.

92. A.-M. Caminade and J.-P. Majoral, Chem. Soc. Rev., 2010, 39, 2034.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc138

Hyperbranched Polymers in Supramolecular Chemistry Barnaby W. Greenland and Wayne Hayes University of Reading, Reading, UK

1 Introduction: Dendrimers, Hyperbranched Polymers, and Supramolecular Chemistry 2 Supramolecular Hyperbranched Polymers 3 Self-Assembled Dendrimers, Dendritic, and Hierarchical Structures 4 Assembly of Polydisperse Structures Containing Discrete Dendritic Components 5 Hyperbranched Molecules in Hierarchical Assemblies 6 Conclusions and Future Perspectives Acknowledgment References

1

1 2 9 16 18 22 22 22

INTRODUCTION: DENDRIMERS, HYPERBRANCHED POLYMERS, AND SUPRAMOLECULAR CHEMISTRY

Molecules that possess hyperbranched and dendritic structures have been among the most well-studied polymeric architectures over the past 30 years.1 Dendrimers typically comprise of either branched monomers arranged in layers or generations that emanate concentrically from a multifunctional core. This architecture results in high molecular weight, monodispersed polymers that possess a three-dimensional, globular topology, which has been likened to that of large biomolecules such as proteins. To

achieve such intricate dendritic structures often requires lengthy and multistep synthesis using either a large amount of reagents to push reactions toward completion or extensive purification by chromatographic techniques to remove by-products. Interest in dendritic molecules has been maintained as a consequence of the novel and useful macroscopic properties they exhibit when compared to linear polymers of similar chemical composition.2 These properties include high solubility, thermal stability, and low viscosity. In addition, dendrimers offer a range of useful chemical features not found in linear, random-coil polymers, such as internal voids that may be used to encapsulate guest species, and a large number of surface functional groups. The surface groups can cluster together in high effective concentration on the exterior of the dendrimer, isolating the central core region in its own microenvironment, shielded from bulk solvent effects. Efforts to mimic the structural features of dendritic architectures in a more synthetically efficient manner have resulted in the production of hyperbranched polymers.3 Hyperbranched polymers are typically produced by the one-pot polymerization of ABX monomers (where X > 1). Although the statistical nature of the polymerization generally results in polydisperse products and branching imperfections within the structure, the methodology can deliver a three-dimensional topology analogous to that of dendrimers. As a consequence of the globular structure, hyperbranched polymers deliver many of the beneficial features of the dendritic structures in a more cost-efficient manner than that possible using a typical multistep dendrimer synthesis. The emergence of the field of supramolecular chemistry has approximately mirrored that of branched polymer

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

2

Soft matter

synthesis.4 Over the course of the past 40 years, synthetic chemists have harnessed noncovalent interactions to rapidly assemble designed supramolecular systems that range in scale from discrete, mechanically interlocked molecules5 to macroscopic fibers and gel networks.6 In contrast to covalent bonds, supramolecular interactions are kinetically labile. This feature allows a multicomponent system to explore a range of possible assemblies before attaining a thermodynamic minimum. Therefore, in an exactly analogous fashion to nature, synthetic supramolecular systems can spontaneously generate immensely complex structures with macroscale, three-dimensional topologies from simple, nanoscale monomers. This chapter highlights how the fields of dendrimer and hyperbranched polymer chemistries have converged with supramolecular chemistry to produce structurally and topologically diverse products that have wide ranging applications as sensors, gene and drug vectors, and nanoscale machines. This monograph seeks to provide the reader with an understanding of the genesis of the field of supramolecular hyperbranched molecules and employs selected examples to demonstrate the fundamental steppingstones in design rational that have led to highly complex structures that can be routinely synthesized today. The chapter is restricted to predominantly organic structures and does not cover metal–ligand interactions within branching points or cores of the dendritic species. Readers who are interested in hyperbranched organometallic structures, or who are inspired to read more comprehensive overview of the area, are directed to one of the following excellent review articles.1, 7

2

SUPRAMOLECULAR HYPERBRANCHED POLYMERS

Supramolecular hyperbranched polymers have been developed that utilize molecular recognition to facilitate selective binding of a guest molecule or molecules by a (dendritic) host or self-assembly that involves numerous weak noncovalent interactions to generate large functional assemblies. Examples of both these classes of supramolecular hyperbranched polymers are examined in the following sections.

2.1

Supramolecular hyperbranched polymers: molecular recognition

Molecular recognition of an analyte by a branched macromolecule can occur either at the surface of the polymer (so-called exoreceptors) or within the interior of the macromolecule (i.e., endoreceptors) (exemplified for a generic

16

Exo receptor

4

Endo receptor

Encapsulation

Scheme 1 Schematic structures of hyperbranched, branched polymers behaving as exo- and endoreceptors, and encapsulation through nonspecific interactions.

dendrimer in Scheme 1). In a related manner, encapsulation of guest molecules to produce ternary structures may be driven by nonspecific interactions such as solvophobic effects (Scheme 1). Significant examples of each of these methods of generating supramolecular hyperbranched systems are presented in the following sections.

2.1.1 Exorecognition The fractal-like, branched architecture of dendritic molecules results in a large number of closely packed surface groups.8 The proximity of surface groups allows enhancement of each recognition event cooperatively thereby increasing the sensitivity of the system and making dendritic molecules an ideal platform for nanoscale molecular recognition devices. Early examples of surface functionalized dendritic sensors include the series of poly(amido amine) (PAMAM)9 -based glycodendrimers introduced by Okada and coworkers10 (1, Figure 1) and subsequently investigated by several groups.11 Motivation for the synthesis of glycodendrimers arose from the desire to target complex natural carbohydrates that are prevalent in biological processes. During in vitro tests of the binding affinity of glycodendrimer (1) for lectin

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

O

HO

HO

O

HO

HO

HO

OH O

O OH HO

OH HO

HO

HO

OH

HO O OHHO

HO

O

O

OH NH

N

N

N

N

HO

N

OH

HO OH

NH

OH OH HO O

NH OH HO O

O

HO OH

HO HO

O

OH

OH

OH

O

OH

O

OH O

OH OH

OH

OH

OH

O

OH

O

OH O

OH

O HO

OH O OH

OH

OH

OH HO O

OH

NH

HN

OH

OH OH

O HO

NH

OH O

OH

NH

N

OH

O O HO

N

N

OH

OH OH HO

H N N

O

O

HO

N

HO

OH

O

N O

OH

OH HO

O HO

OH

OH O

OH

O OH

O HO

HO NH

N

O

O

NH

OH HO

OH

OH

HO

N

OH OH HO HO OH O O H N O

HO

OH HO

O OH

HO

HO

HO

OH O

OHHO

HN

N

O

HO

O

HN

HN

OH O

O

HO

OH

NH

OH

O

HO

O

HO OH O HO NH

HO

HO

O

HO HO

O

OH OH

HO

HO

HO

HO

HO HO

HO

HO

3

OH HO

OH

HO

OH

OH

OH OH 1 O

N

Figure 1

N

N

N H

N

Structure of the PAMAM-based glycodendrimer synthesized by Okada and coworkers.10

(which has an affinity for the α-D-glucosyl surface groups on 1), a turbid solution was produced, presumably as a consequence of the formation of an insoluble supramolecular network. The supramolecular network could only be disrupted, returning the turbid mixture to a homogeneous solution, by the addition of at least a 1200 M excess of mono-D-glucose. This interesting competition experiment clearly demonstrates the amplification of binding strength achieved by positioning the multiple receptor residues on the surface of dendrimers. Although it was not the primary aim of the research, Okada’s binding assay also demonstrates that the system can be induced to switch between solution and precipitated states, without the

formation of new covalent bonds. These glycodendrimers are thus chemoresponsive, highlighting the reversible nature of supramolecular systems. The versatile inclusion phenomena exhibited by the family of cyclodextrin (CD) macrocycles12 have been harnessed to produce supramolecular dendrimers and hyperbranched polymers. An early demonstration of exorecognition that resulted in an electroresponsive supramolecular dendrimer was published by Kaifer and coworkers in 1997 (Scheme 2).13 The group produced a series of polypropylene imine (PPI)14 dendrimers (from the first to the fourth generation) that were decorated with cobaltacene residues at the periphery. In their multiply charged state (2), the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

4

Soft matter

O O Co

O NH

Co

Co

HN

HN N

4e

N

N

N

3

2 HN Co

O NH

Co

HN NH

O

Co

Co

NH

O

Co

O

O

Scheme 2 Electrochemical switching between a charged, uncomplexed PPI dendrimer and a neutral, supramolecular complex in a PPI cobaltacene dendrimer.13

dendrimers exist as single entities in aqueous solution (0.1 M NaCl) with β-CD. Upon reduction to the neutral species (3), the dendrimers form stable host–guest complexes, with each cobaltacene moiety bound within the cavity of the macrocyclic β-CD. Analysis of the half-wave electropotentials of the dendrimers revealed that the switching between the charged dendrimer species and inclusion complex was fully reversible in the presence of excess β-CD, which is needed to maintain solubility. Zhu and coworkers synthesized a hyperbranched polymer that was capable of forming pseudorotaxanes with α-CD macrocycles (Scheme 2).15 A two-step synthesis was employed whereby a 3-ethyl-3-oxetanemethanol (4) was polymerized under cationic conditions to produce a hyperbranched polyether with multiple hydroxy groups. These

hydroxy residues were used to initiate the polymerization of ethylene oxide (5) to generate oligoethylene glycol arms (black, Scheme 3). Addition of α-CD to the resulting polymer resulted in threading of the macrocycles onto the linear glycol units and the desired hyperbranched (poly)pseudorotaxane (6). During the addition of α-CD, the glycol arms become increasingly constrained and extended. During this process, complex (6) slowly crystallizes from the reaction media (water) with a lamella morphology, which is quite distinct from that observed for either the free polymer or α-CD macrocycles individually. In 2001, Meijer and coworkers reported the synthesis of a family of urea modified PPI-based dendrimers up to the fifth generation.16 These dendrimers (7, Figure 2) proved to be excellent hosts for glycine urea functionalized H O

H O

O

O

O 5 BF3Et2O

O HO 4

O Hyperbranched core

O O

O OH

O

OH

O O

O

O

O OH

O

HO

O

O

O H

O

O

O

O O O H

H

O 6

O H

Scheme 3 Synthesis of hyperbranched poly(3-ethyl-3-oxetanemethanol) (blue) with PEG arms (black) and schematic of the inclusion complex that the polymer forms with α-CD. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry R HN

NH

HN

N

S R N H

R NH

S

S

N

N H

5

H N

H N O

H N

H N

N S

NH

R

HSH N N

O

R

H N

O

H N

R

S N

8

S HN HN R

7

HN HN R

S

(a)

R=

(b)

Figure 2 (a) Structure of the thiourea functionalized PPI dendrimer. (b) Structure of the proposed “pincer” type interaction between glycine-based urea analytes and the surface functional groups of the dendrimer.16

guest molecules (8). 1 H NMR spectroscopic analysis of the host–guest complex revealed that complexation was achieved through a combination of ionic interactions between the deprotonated glycine residue and the outermost branching amine in the dendrimer, reenforcing thiourea/urea bidentate hydrogen bonding interactions. Increasing the generation number of the thiourea functionalized PPI dendrimer enhanced the strength of the host–guest interaction. This enhancement was attributed to the increase in density of thiourea groups around the periphery at higher generations favoring the formation of the proposed “pincer” type supramolecular interactions (8).

O

O− O Cu(II) O −O

N O O N N

N

N

NH HN

N

O−

O O

Cu(II)

N

O

Cu(II)

O−

HN 9 O

O O

N

NH N

O

O O O− −

O O−

2.2

Supramolecular recognition within the interior of dendrimers and hyperbranched polymers

O

Figure 3 Proposed coordination spheres for the three identified copper environments observed during analysis of PAMAM copper complexes.19

2.2.1 Nonspecific interactions The use of dendrimers as micelles was envisaged by Newkome in 1985 and was the driving force behind the group’s original arborol synthesis17 ; however, it was not until 1989 that the first evidence for this phenomenon was reported by Tomalia et al.18 Dramatic decreases in the T1 relaxation times for 13 C nuclei of acetylsalicylic acid were observed when measured in solution with PAMAM dendrimers above the fourth generation, when compared to measurements recorded in solutions that contained lower generation dendrimers. This result indicated that the acetylsalicylic acid probes were at least transiently encapsulated within dendrimers. Turro and coworkers have also demonstrated encapsulation of metal ions within dendritic structures when solutions of Cu(II) and acid-terminated

PAMAM dendrimers were studied—formation of three distinct copper complexes was determined (9, Figure 3).19 Most intriguingly, one of these complexes was identified as resulting from Cu(II) coordinated to four nitrogen atoms. This could only be achieved by the encapsulation of copper within the dendrimer. In 1990, Kim and Webster synthesized a hyperbranched polyphenylene with carboxylic acid surface groups (10, Figure 4). The (poly)lithium salt of 10 was water soluble (>1 g ml−1 ). 1 H NMR spectroscopic analysis of 10 with p-toluidine (11) revealed significant upfield shifts (0.84 ppm) for the aromatic guest molecule (11). This was interpreted to be as a consequence of inclusion of 11

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

6

Soft matter

OH O HO

OH

OH O

O

OH

O O

OH

O

O

OH O

HO

NH2 OH

O O HO

OH

HO

OH O

O

O OH

O O

O

11

OH

10

HO

OH O

Figure 4

OH

O

Generic structure of the polyphenylene hyperbranched polymer (10) synthesized by Kim and Webster.

within the cavities of the hyperbranched polymer driven by hydrophobic effects. Thus, the hyperbranched polymer appeared to behave as a unimolecular micelle. In 1993, Fr´echet and coworkers demonstrated that dendrimers with hydrophilic surface groups were able to render hydrophobic pyrene molecules soluble in aqueous conditions by encapsulation.20 Related encapsulation phenomena have been observed for the inclusion of chromophores within hyperbranched polymers.21 A seminal report from the Meijer laboratory outlined the use of modified PPI dendrimers as a “dendritic box” (12, Figure 5).22 In this study, p-nitrophenolic acid (13) and the dye, methyl violet (14), were coencapsulated into a G4 amine terminated PPI dendrimer. The dendrimer was subsequently reacted with Boc-protected L-phenylalanine (12, Figure 4). Exhaustive dialysis revealed that leakage of the encapsulated molecules (13 and 14) had not occurred, demonstrating that the surface of the dendrimer was sufficiently densely packed to be rendered impermeable to either p-nitrophenolic acid or methyl violet. Remarkably, selective removal of the bulky Boc protecting groups of the L-phenylalanine surface unit with piperidine (15, Scheme 4) allowed only p-nitrophenol to be released, with the larger dye molecule remaining inside the dendrimers until the surface L-phenylalanine layer was cleaved by harsher hydrolysis conditions (16, Scheme 4).

2.2.2 Supramolecular interactions in the interior of hyperbranched polymers—site-specific complexation In a series of papers commencing in 1995, Diederich and coworkers introduced two new classes of dendritic molecules that featured either a cyclophane (dendrophanes)23 (17, Figure 6) or cleft-type (18, dendroclefts 24 ;

see Figure 8) receptors at their core. Cyclophanes are wellknown supramolecular receptors typically comprising several aromatic residues constrained within a macrocycle.25 By varying the chemical structure of the cyclophane, it is possible to produce molecules that are capable of highly specific host complexation within the cavity of the macrocycle. The first series of dendrophanes (up to the third generation) produced by the Diederich group possessed a central core with a strong affinity for aromatic species such as 6-(p-toluidino)naphthalene-2-sulfonate (TNS, 19, Figure 6).23 Detailed 1 H NMR and fluorescence spectroscopic analysis of the complex formed between dendrophane 17 and 19 revealed that binding occurred exclusively within the cyclophane core of the dendrimer with negligible nonspecific interactions observed between the guest and the outer generations of the dendrimer. The binding constants for receptors of this type were determined to lie in the range of 103 –104 M−1 and were independent of the generation of the dendrimers used. The kinetics of binding was also similar for each generation, revealing that the relatively open architecture of the dendrimer imposed by the large core molecule did not hinder access of the host to the interior receptor for any of the generations studied. Fan and coworkers synthesized a hyperbranched polymer with multiple β-CD residues distributed throughout the macromolecule (20, Figure 7).26 The hyperbranched polymer 20 was found to encapsulate both phenolphthalein (21) and methyl orange (22) concurrently. Ultraviolet/visible (UV/vis) analysis of the complex demonstrated that the guest molecules were encapsulated within the cavities of the β-CD residues and the voids in the hyperbranched polymer. The binding constants for the polymer complex with the dyes were greater than those observed for monomeric βCD derivatives with the same guests (21 and 22). This result exemplifies the cooperative effect of positioning multiple

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

7

O O O O O O O O O O O O O O O O O O O NH NH NH O O NH NH NH NH NH NH R NH O O OO R R R R R R R R NH NH O O O O O O O O R R O O NH O O O HN HN HN HN NH R O HN HN O HN R NH HN OO HN NH O O NH O O R HN HN R O HN O R NH NHO O HN R O HN O O O R NH N NH O N NH R O N N O H OO O N HN N Cl R HN NH O NH R O N N N HN Cl Cl O O HO R O N NH R OHN N NH O N HO N O N N CO H Cl N R H 2 O O O N R HN N N I I N O NO N R H O O HN H N HO O H R N N OO HO O N N N N N R H NH I I H OH RO O N NO N O N N N 14 N N H O N R N O N H O R NH N N O O N O H N R H O N N HN O O NH O H R N O O N N N O R H O O HN H N N N O HR N N N O N N N O N O N OH R HO O O N R NH 2 H N N O N N N OH N N OH O R H O N H R NH O H O 13 N N O N O N O N R N NH O N R H O N N O N O HN OH N NHO R HN N N O R H O O O NH R N O HN N N O HN NH R N O NH H N OO O N O R HN N N O HN R N HN O NH O O O R O NH OHN R O NH HN OR NH NH O O R O HN NH NH O O NH O HN O R NH NH O HN NH NH NH HN HN O O HN R O O O O O R O R O O O O O HN R R HN R R O O R R R HN R O O R HN HN HN HN HN O O HN O O HN HN HN O O O O O O O O O O O O O O O O O O O O

O

12

Figure 5 Structure of the dendritic box containing two encapsulated methyl violet residues synthesized by Meijer and coworkers (R = Ph).22

H N

NH

R

NH O

N

O

HN

N H

N

O

NH2

NH

R

O

NH2

O NH

R

R

NH2 H2 O

N NH2

O O

HN O 12

Scheme 4

R = Ph 15

16

Selective modification of the surface groups on Meijer’s dendritic box.22

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

8

Soft matter

OH HO O O

O

O OH

O

O

O

OH

OH

OH

O O

O O

O

O O

O

19

O O O 17

O

O

O O OH

OH

OH

OH

N H

O

O

HO O

Figure 6

SO3H

O O

O O

O

O O

OH

Structure of a first-generation dendrophane and the host molecule 6-(p-toluidino)naphthalene-2-sulfonate (19, TNS).23 HN X HO

OH

HN Y N Y H N X HN

H N

X

O

HN H N X

H Y N

O

Y N

Y H N X

HN

H N

H N

21 X

N Y N

N N

SO3Na

20 22

O X=

N H

O

Y=

Si

O

Si

OH

Figure 7 Structure of a hyperbranched polymer with multiple β-CD residues and the structures of guest molecules phenolphthalein (21) and methyl orange (22).

receptors within one macromolecule. Encapsulation of two distinct guests within a single polymer has exciting implications for the delivery of several pharmaceuticals from a single formulation. Dendrocleft dendrimers contain a rigid cleft-like receptor at the core of the molecule. The conformationally constrained, optically pure core based on spirobifluorene fixes the spatial relationship between the appended dicarboxyamino derivatives, which are both capable of forming hydrogen bonds to a range of guest molecules (18, Figure 8).24 Indeed, the G1 dendrocleft 18 was able to form complexes with the sugar derivatives such as 23, 24, and 25 with binding constants of 160, 225, and 390 M−1 , respectively. The variation in binding constants demonstrated that

the chiral environment imposed by the rigid cleft receptor was able to differentiate between the glucoside derivatives. Dendrimers that feature porphyrin cores have been studied as potential synthetic chlorophyll mimics among other applications.27 One of the most innovative examples of this class of dendrimer in the context of supramolecular chemistry was produced by Zimmerman and coworkers in the late 1990s.28 Dendrimers that possessed alkenyl surface groups on Fr´echet-type benzyl ether29 dendrons were synthesized and then each dendron was attached to a porphyrin core by means of ester linkages (26, Scheme 5). Exposing dendrimer 26 to the first-generation Grubbs’ catalyst under high dilution conditions resulted in ring-closing metathesis

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

9

O 3 O O HN

N

N H

O

H N

O

O

3

O

O

O 3

O

O O

O

O

N H

3

O

3

H N

N

HN

3

O

O

18

O HO O

OH OH OH

C8H17O OH

Figure 8

OH

HO HO

HO HO HO OC H 8 17

OC8H17 HO

23

24

25

Octyl α-L-glucoside

Octyl α-D-glucoside

Octyl β-D-glucoside

Structure of a dendrocleft and three alkylated sugar derivatives.24

of the peripheral alkenyl groups, effectively cross-linking the outer shell of the dendrimer (27). This core–shell nanoparticle (CSNP) was then subjected to basic conditions resulting in the cleavage of the four ester linkages to the central porphyrin residue (28), effectively coring the dendrimer, in a process similar to that used to synthesize conventional molecularly imprinted polymers.30 The cavity contained four carboxylic acid receptors ideally positioned to accept porphyrin-type guest molecules. Binding assays were then conducted between the “cored” molecularly imprinted dendrimer 28 and a series of isomeric porphyrins (29–31). The binding constant was found to decrease from 3.3 × 103 to 1.6 × 103 to 0.9 × 103 M−1 (29–31, respectively) demonstrating that the receptor cavity was selective with respect to the positioning and/or basicity of the heteroatom in the pyridine ring. In contrast to a typical linear polymer, which assumes a random-coil conformation in solution, hyperbranched and dendritic molecules maintain their globular topology. This architecture has been used to facilitate exorecognition by the multiple, densely packed surface groups, and encapsulation of small molecules within voids or at receptors inside these highly branched macromolecules. In each of these examples, the branched host molecule was synthesized through a traditional, covalent bonding forming approaches. The following section describes how

hyperbranched structures can be efficiently synthesized using noncovalent, supramolecular techniques.

3

SELF-ASSEMBLED DENDRIMERS, DENDRITIC, AND HIERARCHICAL STRUCTURES

Despite significant advances in covalently bonded dendrimer design and synthesis, it remains a lengthy and labor-intensive procedure. The reversibility and specificity of noncovalent, supramolecular chemistry is ideally suited to solving this problem. It may be envisaged that by simply mixing a series of monomers that each possess substratespecific receptors, the target supramolecular dendrimer may be accessed through a one-pot procedure (Scheme 6), akin to that of a conventional, covalent hyperbranched polymer synthesis. Furthermore, these discrete branched supramolecular entities could be further manipulated to form an extended, hierarchical superstructure, thus producing a material with macroscopic order from monomeric components (Scheme 6). In addition to the synthetic advantages that the supramolecular approach offers to branched polymer synthesis, the reversible nature of supramolecular bonds

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

10

Soft matter

O

O

O

OO

O O

OO

O

O

O O O

O O

O

O

O

O

O

O

O

O

O O

O

O

O O O

OO

O

O

O

O

O O

O O O

O

O

O

O O

O

NH N

O

O

O

O N HN

O

O O

O

O

O

O

O

O O

O O

OO

O

O O

O

O

O

O

O O

27

O

OO

O O

OO

O O

O O

O O

O

O

O

O

O O

O O

O O

O

O

O

N HN

O

O O O

O

NH N

O

catalyst

O O

O

O

O

O

O O

Grubbs'

O O

O

O O

O

O

O

O

O O

O O

O

O

O O

O

O O

O O

O

O

O

O O

O

O

KOH 26

O

XY Z

O

NH N HO O

X Y Z

XY Z O

NH N

OH X YZ

ZY X HO

N HN Z XY O OH

X

O

Z Y

O

O O

O

29 X = N, Y & Z = CH 30 Y = N, X & Z = CH 31 Z = N, X & Y = CH

O

O

O

O HO

OH

O

O O

O OH O

28 ⊂ 29, 30, or 31

Scheme 5

O HO O

O

O

Z Y X

N HN

O

28

Synthesis of molecularly imprinted dendrimers by Zimmerman and coworkers.28

confers the structures with unique stimuli-responsive properties. At a fundamental level, disruption of the supramolecular bonds by, for example, the application of heat, change in pH, or solvent polarity causes disassembly of the dendritic and/or hierarchical structures. The disruption of the supramolecular bonds may be observed as a change in solution viscosity or, more dramatically, the change from a solution to a gel phase. From an application-driven perspective, the disassembly of the supramolecular structures into low molecular weight components could be harnessed

to release a drug in vivo followed by rapid excretion of the benign component monomers.

3.1

Supramolecular self-assembly of dendrimers and hyperbranched polymers

In 1996, Zimmerman and coworkers produced a seminal study in this field that first involved the synthesis of a series of Fr´echet-type benzyl ether dendrons (up to the fourth generation) that possessed a novel hydrogen

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

4

2 1

Monomers

Supramolecular dendrimer

Hierarchical superstructure

Scheme 6 Schematic synthesis of a self-assembled hierarchical superstructure.

bonding motif at their focal position (32 and 33, Scheme 7, only G1 and G2 shown).32 The supramolecular motif was based on a rigid, fused aromatic system that contained four carboxylic acid residues. The orientation of the acid groups permitted cooperative self-assembly of the dendrons into larger superstructures. The precise architecture of the resulting aggregate was found to be dependent on the generation of the dendron appended to the hydrogen bonding unit. For higher generation dendrons (G2–G4), the equilibrium solution-state structure was found to be

11

the hexameric supramolecular dendrimer that contained a rosette of six hydrogen bonding residues at the core. In contrast, it was found that the system equilibrated to linear supramolecular structures when the least bulky, G1 dendron was attached to the tetracarboxylic acid hydrogen bonding motif. This behavior was rationalized by the need for an additional steric buttress effect provided by the larger dendrons favoring the formation of the spherical supramolecular dendrimer. A recognized drawback of this system is the nonspecific nature of the hydrogen bonding motif. The degenerate arrangement of the four carboxylic acid groups permits the dendrons to adopt either rosette-like or linear structures. This lack of specificity in supramolecular self-assembly is overcome in nature by utilizing more complex recognition moieties that contain a precise arrangement of hydrogen bond donor (D) and hydrogen bond acceptor (A) groups (e.g. DNA base pairs33 ). Thus, a receptor molecule that contains two donor and one acceptor functionalities—DDA—will form a complex with a second receptor that contains the complementary arrangement of AAD functionalities. Furthermore, the two components adopt a single orientation with respect to each other to generate the supramolecular structure with the highest degree of hydrogen bonding interactions (DDA–AAD). R

O

OH

R OH

R

R

R

R

R

O O N

OH OH

R

O

R

R 32, R = G1 33, R = G2

R Hexameric supramolecular dendrimer

Schematic supramolecular dendron

Linear supramolecular polymer

t Bu

t Bu

t Bu

O

t Bu

O t Bu t Bu

O

O

O t Bu

t Bu

G1

O

R = t Bu t Bu

t Bu

O O

O

O G2

t Bu

Scheme 7 Structures of G1 and G2 dendrons with a tetracarboxylic acid hydrogen bonding motif, and a schematic of their self-assembly into hexameric supramolecular dendrimers and linear supramolecular polymers.32 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

12

Soft matter

The use of precisely oriented arrangements of hydrogen bond donor and acceptor motifs to assemble supramolecular dendritic structures was first described by Reinholdt, van Veggel, and coworkers in 1997 during the synthesis of self-assembled metallodendrimers.34 Later studies conducted by Zimmerman and coworkers revealed how assemblies could be produced using initially DDA–AAD35 and then DDAA–AADD31 interactions (Figure 9). These latter hydrogen bonding motifs exhibited extremely high association constants of over 108 M−1 in dry CHCl3 resulting in dendritic structures (34) that were robust enough to be analyzed by gel permeation chromatography (GPC). One of the most widely used supramolecular interactions in the field of mechanically interlocked molecules, so-called mechanostereochemistry,5 is that of the strong hydrogen bonds between ammonium ions and crown ethers, which were first reported by Stoddart and coworkers in the mid1990s.36 This versatile interaction has found use in the construction of a myriad of interlocked systems and bistable molecules. In 2004, Gibson and coworkers made use of ammonium ion/crown ether interactions to template the formation of a hyperbranched polymer.37 The key AB2 monomer, 35 (Figure 10), consisted of a bis(m-phenylene)-32-crown10 receptor derivatized with two paraquat residues. Variable concentration of 1 H NMR titrations 35 revealed that the chemical shift of the aromatic protons within the

crown ether moiety decreased with increasing concentration between 0.333 and 100 mM. Thus, the degree of complexation, and therefore the degree of polymerization (DP) was concentration dependent (see schematic, Figure 10). The binding constant for the system was estimated to be 380 M−1 , giving a DP of 40 (MW = 5.94 Kg mol−1 ) at 100 mmol concentration of 35. This analysis was supported by the nonlinear increase in solution viscosities observed for samples of 35 at increasing concentration.

3.2

Multicomponent supramolecular assembly of dendrimers

The supramolecular dendrimers and hyperbranched system considered in the previous section are assembled from identical dendrons and monomers that contain complementary noncovalent recognition motifs. In 1997, Zimmerman and coworkers produced a three-component dendrimer (36) whereby two dendrons (37) were brought together around a small dicationic core (38) by complementary electrostatic and hydrogen bonding interactions (Scheme 8).38 The ditopic core template is also a known pharmaceutical, pentamidine, which is active against Pneumocystis carinii pneumonia —a serious lung infection. Controlled disassembly of the dendrimer in vivo would result in increased R

O N

t Bu

R= R

t Bu

O

N

N

H N

H N O

O

t Bu

O

O

O

H N

t Bu

N

O N

t Bu

O

O

H t Bu t Bu

t Bu

N

O H

N

N

N

H N

N

O N

H

H

N

N

O

N O

N N H

H O

R O

N

H N

N

O N

H

N

O

N H

O

N

N O H

O N

N

H N

N H

O

N

H

H N

H

O

N

H

N

H O

N

N

N

N

H N

H N O

O

R 34

H

H

N

H

N N

N O R

N

O

O G2

N H

N H

N

O

O

N H

H

H

N

H

H

N

N O

N H

O

O

N H

N H

N

N

O

R

N O

Figure 9 The structure of the hexameric supramolecular dendrimer supported by multiple DDAA hydrogen bonding interactions as described by Zimmerman et al.31 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

O

O

N

O

O

O

O

O

2PF6

O

35 = High concentration

N

O

O

13

2PF6

35

N

N

(a)

(b)

Figure 10 (a) Structure of the AB2 monomer synthesized by Gibson and Huang.37 (b) Schematic of the concentration-dependent formation of the supramolecular polymer consisting of monomer 35.

t Bu t Bu t Bu t Bu

t But Bu

t Bu t Bu

t Bu

t Bu

O

O

t Bu

O

t Bu t Bu t Bu t Bu

O

O

O

O

O

O

O

2

t Bu

O

B(Ph(CF3)2)4

O O

H HN

O

HN H

O

O

H NH

O

2X

NH H

38

N N N CH3O

OCH3 37 t Bu

t Bu t Bu

t Bu

t Bu

O

t Bu

OCH3

O O O

O

O

O

N O

N N

H HN HN H

2X O

O 36

H NH NH H

O

N N N

O

t Bu

O

O

t Bu

O

O

t Bu

O O

OCH3

CH3O

t Bu

O

O t Bu

t Bu

O O

O

t Bu

t Bu

t Bu t t Bu Bu

t Bu

Scheme 8

t Bu

O

O

t Bu t Bu

t Bu

O

t Bu t Bu t Bu

t Bu

O

CH3O

O O

t Bu

t Bu

O

t Bu

t Bu

t Bu

O

O

t Bu

Templated supramolecular synthesis of a three-component dendrimer [X = B(Ph(CF3 )2 )4 ].38

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

14

Soft matter Boc NH

HN Boc Boc NH

Boc HN

O

Boc

O NH

Boc

HN

HN

NH NH HN

HN

Proflavin template

O

O O

Boc NH

O

O

O

O NH

OH

H2 N

Boc NH HN

N H

O

39

NH2

HO

HN Boc

HN NH HN Boc

Cl

O

Boc NH

HN Boc NH

HN

O

O NH Boc

HN Boc

Boc NH

Figure 11

NH

O

HN Boc

Encapsulation of proflavin hydrochloride by L-lysine-based dendrons.41

bioavailability of the active compound, resulting in an interesting example of a nanoscale drug delivery device. Assembling branched structures from multiple distinct components offers a new degree of adaptability to supramolecular structures. Thus, the templated synthesis of dendritic structures has become a fertile field of exploration over recent years. Smith and coworkers produced a series dendrons39 with carboxylic acid groups at the focal point based on the known40 L-lysine-derived dendritic architecture (e.g., 39, Figure 11). These have been found to solubilize the hydrophilic dye, proflavin hydrochloride, in nonaqueous solvents such as CH2 Cl2 . Careful control experiments revealed that the dye could not be induced to dissolve in CH2 Cl2 by the addition of either smallor long-chain carboxylic acids (acetic acid or steric acid, respectively). In addition, the spectroscopic properties of the dye were modified in the presence of the lysine dendrons, demonstrating that the proflavin hydrochloride was encapsulated within the microenvironment of the templated supramolecular dendrimer. In further studies, Smith and coworkers demonstrated that the dye molecules could be induced to dissolve in an organic solvent by the dendrons, which could transport the dye into a separate, aqueous phase.41 The rate of dye transportation was found to increase as a function of generation, showing the degree of isolation of the template from the bulk media is an important factor in the dissolving process. The first example of the use of ammonium ion/crown ether interactions to template the formation of dendrimer assembly was disclosed by Gibson and coworkers in 1998 (40, Figure 12).42 The target pseudorotaxane dendrimer was formed by the addition of dibenzo-[24]-crown-8

(DB-24-C8) functionalized Fr´echet-type benzyl ether dendrons (first to third generation) to a core molecule that featured three benzyl ammonium cations. It was observed that the triply charged, poorly soluble core slowly dissolved in chloroform upon the addition of the dendrons. Analysis of the 1 H NMR spectrum of the solution revealed almost complete consumption of starting materials over a 72-h period, resulting in a high yield of the desired self-assembled dendrimer. Remarkably, evidence of the supramolecular assembly was found in the matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) mass-spectrum, which displayed a signal consistent with the targeted four-component assembly (40), demonstrating the strength of the hydrogen bonds that underpin this dendrimer. Smith and coworkers have also harnessed related ammonium ion/crown ether interactions to template the assembly of two Boc-protected L-lysine dendrons (Scheme 9).43 It was observed that the addition of potassium ions, a competitive guest of the crown ether hosts, triggers disassembly of the supramolecular dendrimer (41) to its component parts (42 and 43). This example of stimuli-responsive dendritic architectures that release small molecules on demand clearly demonstrates advantages inherent in noncovalent dendrimer synthesis over conventional, covalent approaches. This section has used select examples from the literature to demonstrate the synthesis of precisely controlled dendritic structures through the spontaneous assembly of multiple components around a multivalent core. This approach has universal application and there are now examples of dendrons that are induced to assemble around a range of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

15

Bn

O O

Bn O

O

O

Bn

O

O

O

O O

O Bn

O

O

O

Bn

O O

O NH 2 O

O Bn

O

O Bn

O Bn

O O

3PF6

O

O

O H2 N

O O

O O

O

O

O H N O 2

O O

O

O

O

O

O 40

O

O Bn

O

O

Bn

Figure 12

O

O

Bn

Bn

Structure of the four-component pseudorotaxane G2 dendrimer synthesized by Gibson and coworkers.42

Boc HN

Boc NH Boc HN

O

O

NH O

O

O

H3 N

O NH HN

O

O O

HN O

O

O

Boc NH

O

O

O

NH HN

O

2Cl 41

NH Boc

HN Boc

O

O

NH3

HN Boc

NH Boc

2KPF6

Boc NH 2

Boc HN

O

O

NH O O

K

O

O

O NH HN HN Boc

O

O

NH3 H3 N

+ 2PF6 + 2Cl

43

NH Boc 42

Scheme 9 Release of the benzyl diammonium central core template from an L-lysine-based supramolecular dendrimer by the addition of potassium ions, a competitive guest for benzo-[18]-crown-6.43 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

16

Soft matter

template molecules via a plethora of supramolecular interactions. The resulting supramolecular globular structures can act as hosts for a range of small molecules, aiding solubility by encapsulation of a guest at the core, within the unique microenvironment provided by the assembled dendrons. By careful choice of the core template and the supramolecular interactions used to assemble the system, discrete dendrimers can be disassembled by the addition of competitive hosts for the dendritic receptor residues.

4

ASSEMBLY OF POLYDISPERSE STRUCTURES CONTAINING DISCRETE DENDRITIC COMPONENTS

The synthesis of defined supramolecular dendrimers is an aesthetically appealing challenge for chemists; however, the production of less well-defined supramolecular systems from dendritic components offers equally exciting possibilities. These structures can be produced by assembling precisely synthesized dendrons around a polyvalent receptor. Thus, the polydispersity of the final supramolecular structure is as a consequence of variations in the number of receptor sites available for dendron binding.

4.1

Polydisperse dendritic structures

Examples of entirely organic multivalent receptors have been synthesized recently by the Stoddart group44 with the aim of producing supramolecular dendronized polymers. Dendronized polymers45 contain a central backbone with dendritic wedges attached. These macromolecular assemblies can be synthesized readily via several methodologies—by the polymerization of dendrons with suitable functional groups at the focal point, by graftingon dendrons to reactive functionalities on a preconstructed polymer, or by convergently synthesizing dendrons from

the polymer backbone. The steric constraints imposed on the polymer by the large dendritic side groups dramatically alters the conformation of the polymer in solution and the solid state, forcing the polymer from a random-coil conformation into adopting a more linear, tube-type structure. Persistent structures of this nature can afford materials with interesting properties such as liquid-crystalline behavior. The stimuli-responsive supramolecular dendronized polymers produced by Stoddart and coworkers used a socalled hook-and-eye approach, harnessing the interaction between ammonium ions and crown ether receptor units (Scheme 10).44 The assembly was constructed by the addition of Fr´echet-type benzyl ether dendrons with benzyl ammonium at the focal point (first to third generation, only G2 shown, 44, Figure 13) to either polystyrene (45, Mn ≈ 85 kDa)- or polyacetylene (46, Mn ≈ 13 kDa)-based polymers containing DB-24-C8 pendent eye groups. The supramolecular dendronized polymers were found to reach an equilibrium structure in 116 000 Da). More recent studies initiated by the Diederich group have examined the possibility of using low molecular weight dendrons to achieve gene transfection.49 Low generation (i.e., ≤G2) dendrons were synthesized and studied for gene transfection efficiencies (47, Figure 14). The dendrons were comprised of hydrophilic and hydrophobic hemispheres separated by a rigid aromatic core. All of these dendrimers exhibited low toxicity and optimum transfection at extremely low molecular weights (below 2700 Da).

NH3 H3N

NH3 O

NH NH3

O

H N

N H

O N H H N

HN

H N

O O

H3N

NH3

NH3

NH3

O

O

47 N H

Figure 14

17

NH3

9CF3CO2

O

Amphiphilic low molecular weight dendron for gene delivery.49

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

18

Soft matter upper face and four long alkyl chains on the lower face of the macrocycle. Structure 48 was found to dissolve nonpolar porphyrin and fullerene derivatives in aqueous environments, presumably by encapsulation within a micelle decorated by the numerous polar amine residues. Cryo-TEM analysis of 48 revealed that they are discrete aggregates with extremely low dispersity. The micelles were evident ˚ made up as spherical cages with a diameter of about 75 A of seven amphiphilic dendrons.

The amphiphilic nature of the dendrons was thought to be critically important regarding the mode of action. It was speculated that higher order aggregates of dendrons were present in solution. Thus, the system was behaving as though comprised of macromolecules that were far higher in molecular weight. The assembly of dendrons into higher ordered structures provides a new level of function to these molecules. Select examples of this process will be provided in the following sections.

5.2

5

5.1

HYPERBRANCHED MOLECULES IN HIERARCHICAL ASSEMBLIES

Solid-state hierarchical assemblies from dendritic and hyperbranched structures

In 1995, Percec and coworkers published an extensive study concerning the design and synthesis of hyperbranched and dendritic macromolecules that exhibited typical liquidcrystalline (LC) behavior as a consequence of a highly ordered structure in the solid state.54 In subsequent investigations, Percec et al. found that subtle changes in substitution patterns within the branching units of LC dendrons and the addition of sterically more demanding surface groups served to dramatically alter the packing structures of the dendrons.55 For example, the flat-tapered monodendron 49 and the conical dendron 50 exhibit hexagonally packed conical and cubic-packed sphere-type topologies, respectively (Scheme 11).

Solution-state hierarchical assemblies

Important examples of extended, solution-state hierarchical assemblies of branched molecules that are beyond the scope of this monograph include those that harnessed metal–ligand interactions,50 or use of phase boundary interfaces to promote ordering of amphiphilic, surfactanttype dendrons.51, 52 Hirsch and B¨ottcher synthesized a notable example of amphiphilic dendritic molecules 48 (Figure 15)53 selfassembling to form micelles. The dendron contained a central calixarine with two amine-terminated dendrons on the NH2 NH2 NH2

H2N HN H2N

O N H

H N

NH O

H 2N

O

H2N NH2

NH2

N H

H2N NH2 O

NH2

NH2 H2N O O

NH

HN

H N

O

O O

t HN Bu

NH2

NH2

O HN t Bu

H2N

NH2

OO OO

48

Figure 15

Amphiphilic calixarine-based dendron that assembles into well-defined micelles in water.53

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

19

a R

49

a R

50

Scheme 11 Formation of hexagonally packed conical and cubic-packed sphere-type structures from flat-tapered and conical dendrons, respectively. (Adapted from Ref. 55.  American Chemical Society, 1997.)

The Percec group has produced several libraries of structurally related LC dendrons including those based upon phenylpropyl,56 biphenylpropyl,57 and benzyl ether dendrons,58 whereby each member of the library differs by generation and substitution pattern at each branch point.59 Analysis of the solid-state packing of each of the dendrons has allowed the group to develop a nano-periodic table to facilitate the prediction of solid-state structure from molecular connectivity of dendrons of this type. HO HO

HO O

O

O HO O OH O O O O O O

OH

HO O

HO

HO

O

HO

O HO

O

HO

O

O OH

O O O

O

O O

O

B

O O

OH

O O

OH (b)

OH

O HO

O

OH

O

O

OH

O

O OH

O O

O

O

HO

OH

O

O

HO

OH

O

O

O

O

HO

(a)

O

O

O

O

O

OH

O

O

OH O

O

O

A OH

O O

O

O

O

O O

O

O

O O

O

O O

O

O

O O

OH OH

OH

O

O O

The polydisperse nature of hyperbranched materials would seem to make them ill-suited to the spontaneous self-assembly of precise structures. However, in 2004, Yan and coworkers demonstrated the assembly of hyperbranched materials into macroscopic objects.60 The structurally simple polymers were based on the known hyperbranched poly(3-ethyl-3-oxetanemethanol) core (blue in 51, Figure 16), which was modified with oligoethyleneglycol arms. Stirring 51 in acetone, which is a

51

Figure 16 (a) Structure of the hyperbranched polymer synthesized by Yan and coworkers. (b) Optical images of the self-assembled tubes (scale bar 300 µm). (Adapted with permission from Ref. 60.  American Association for the Advancement of Science, 2004.)60 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

20

Soft matter

HO O HO HO HO

O N H HN

OH

NH H N

OH OH OH

O 52

O

molecules.61 The solvent within the gel is constrained by capillary action within an entangled fibrillar-type network, with each fibril comprising an extended supramolecular array of monomers. The first example of a dendritic gelator was published by Newkome and coworkers in 1992 (Figure 17).62 This dumbbell-shaped dendrimer comprised two low generation dendrons with alcohol surface groups separated by a hydrophobic tetradecyl chain (52). These dendrimers were found to form thermoreversible hydrogels. Electron micrographs of the gel revealed the expected fibrilar structure, with each fiber extending over many microns in length. A range of branched gelators have since been extensively investigated. In 2000, Aida and coworkers produced the first organo-gelator based on dipeptide functionalized Fr´echettype benzyl ether dendrons.63 Smith and coworkers subsequently reported a supramolecular dendrimer (53), that itself organized into an extended supramolecular fibrous gel in organic solvents (Scheme 12).64 The gel was formed by mixing 1,12-dodecane diamine with two equivalents of G2 lysine dendrons that had carboxylic acid focal groups. Gels could be formed in a broad range of solvents (see SEM images in Scheme 12), but the stability of the gel with respect to the temperature was found to be dependent on the polar solubility parameter of the solvent (δ a ). For the series of solvents: mesitylene (δ a = 0.3), bromobenzene (δ a = 3.4), tetrahydrofuran (THF) (δ a = 4.8), the maximum temperature at which the gel remained stable decreased in the order 105, 40, and 26 ◦ C, respectively. In addition,

OH

OH OH OH

Figure 17 Dumbbell-shaped gelator produced by Newkome and coworkers.62

good solvent for the arms, but a poor solvent for the core, produced multiwalled macroscopic tubes and ribbons up to 7.5-cm long and 1.5 mm in diameter (Figure 16). Prior to self-assembly, the hyperbranched polymer had a molecular weight of 37 kDa and PDI of 2.7 whereas the analysis of the fraction of the polymer that made up the tubes revealed that this fraction had a molecular weight of 33 kDa and PDI of 1.5. Thus the system is self-sorting, selecting the components of the disperse hyperbranched polymer sample with the correct composition to produce the macroscopic tubes.

5.3

Hyperbranched molecules as gelators

Supramolecular gelators are able to immobilize large quantities of solvent at very low concentrations of the gelating

Boc HN

Boc NH

HN

NH HN Boc

O

O

O

O

O

O NH O

NH3

H3N

Boc

KUN SEI

(a)

O HN HN Boc

53

NH HN Boc

NH Boc

Supramolecular dendrimer

3.0 kV ×55,000 100 nm WD 8 mm

KUN SEI 3.0 kV × 55,000 100 nm WD 8 mm

(b)

KUN SEI

NH Boc

3.0 kV ×55,000 100 nm WD 8 mm

(c)

Scheme 12 Supramolecular dendrimer that self-assembles to form gels in organic solvents. SEM images of the gel formed in various solvents: (a) mesitylene, (b) bromobenzene, (c) THF (scale bar = 100 nm). (Reproduced from Ref. 64.  American Chemical Society, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

21

NH2 N X N

HN N

N

N

X N H N

N

X

X N

O

N

N

O N

54 +

N N

X

NH2

N NH2

NH

N

N

N

HN

H N

X=

55

X N

H N

H N

O

X

N

NH2 N

O

N

N

56

Scheme 13

N X

X N

N

H N

NH

NH

N X

Synthesis of the hyperbranched gelator 56 from N,N  -methyl bis-acrylamide (54) to 1-(2-aminoethyl)piperazine (55).

Gelator 56 formed stable gels at room temperature in a range of polar solvents including dimethylacetamide (DMAc), dimethylformamide (DMF), dimethyl sulfoxide (DMSO), and pyridine. The gels were all thermoreversible, with the temperature at which they break down dependent on the solvent and the feed ratio of the monomers (54 and 55) used to produce the gel. It was proposed that the gel was supported by hydrogen bonding interactions between the carbonyl groups and amine and amide hydrogen atoms either within the arms of the hyperbranched polymer or at the chain ends (Figure 18).

solutions of the supramolecular dendrimers in solvents in which gels did not form (e.g., methanol, CH2 Cl2 ) still exhibited ellipticity in the circular dichroism spectra. This demonstrated that hierarchical superstructures were present in these solutions even though fibrils of high enough aspect ratio to form gels were not formed. The first example of a hyperbranched gelator was reported by Yan, Huang, and coworkers in 2007.65 The polymer was synthesized by the Michael addition of N,N  methyl bis-acrylamide (54) to 1-(2-aminoethyl)piperazine (55), which generated the hyperbranched polymer 56 (Scheme 13).

HN O

NH O N

H N O

N

H N

H N

N O

N

H N H

H N

N

H N

N O

H N

H

O H N

H N

N

N O

H N

N

O

N

N

N N

NH

O N O

O

H

H

O

O

O

H N H

N

N

N

O

H N

H N

N

H N O

H N H

O

O

Figure 18 Proposed intra- and intermolecular hydrogen bonding interactions that support the organogel synthesized by Huang and coworkers.65 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

22

Soft matter

6

CONCLUSIONS AND FUTURE PERSPECTIVES

Notable examples from the literature to demonstrate the convergence of the fields of supramolecular chemistry with hyperbranched polymeric materials are highlighted in this chapter. The globular architecture of branched molecules combined with the high number and density of peripheral functional groups that these systems possess has allowed chemists to design unique ternary structures whereby small molecules can be selectively recognized either at the exterior or within the interior of the polymeric architecture. The encapsulated molecules can be expelled in response to external stimuli, such as temperature, pH, and selective chemomodification of the hyperbranched macromolecule. In addition, supramolecular self-assembly processes offer the possibility to use a mix-and-match approach to produce new, monodisperse systems from carefully designed monomers that take advantage of highly specific intermolecular interactions. Supramolecular branched polymers can thus be programmed to further assemble into hierarchical superstructures. Macroscopically ordered materials are consequently generated from (nanoscale) monomeric building blocks, without the formation of a single new covalent bond. In all of the cases described in this short review, these structures are stimuli-responsive, often undergoing rapid alteration in structure for the cost of little addition energy. Harnessing these intriguing properties to produce industrially important and commercially viable materials now represents the next major target for this exciting field in the forthcoming years.

5. (a) J. F. Stoddart and H. M. Colquhoun, Tetrahedron, 2008, 64, 8231–8263; (b) J. F. Stoddart, Chem. Soc. Rev., 2009, 28, 1802–1820; (c) L. Fang, M. A. Olson, D. Benitez, et al., Chem. Soc. Rev., 2010, 39, 17–29. 6. N. M. Sangeetha and U. Maitra, Chem. Soc. Rev., 2005, 34, 821–836. 7. Dendrimers in supramolecular chemistry: F. Zheng and S. C. Zimmerman, Chem. Rev., 1997, 97, 1681–1712. 8. D. A. Tomalia, A. M. Naylor, and W. A. Goddard III, Angew. Chem. Int. Ed., 1990, 29, 138–175. 9. D. A. Tomalia, H. Baker, J. Dewald, et al., Polym. J., 1985, 17, 117–132. 10. (a) K. Aoi, K. Itoh, and M. Okada, Macromolecules, 1995, 28, 5391–5393; (b) K. Aoi, K. Tsutsumiuchi, A. Yamamoto, and A. Okada, Macromol. Rapid Commun., 1998, 19, 5–9. 11. For examples see: (a) P. R. Ashton, S. E. Boyd, C. L. Brown, et al., Chem. Euro. J., 1996, 2, 1115–1128; (b) M.-G. Beak, K. Rittenhouse-Olson, and R. Roy, Chem. Commun., 2001, 257–258; (c) W. Hayes, H. M. I. Osborn, S. D. Osborne, et al., Tetrahedron, 2003, 59, 7983–7996. 12. For a review of CD based supramolecular polymers see: A. Harada, Y. Takashima, and H. Yamaguchi, Chem. Soc. Rev., 2009, 38, 875–882. 13. R. Castro, I. Cuadrado, B. Alonso, et al., J. Am. Chem. Soc., 1997, 119, 5760–5761. 14. E. M. M. De Brabander-Van den Berg and E. W. Meijer, Angew. Chem. Int. Ed., 1993, 32, 1308–1311. 15. X. Zhu, L. Chem, D. Yan, et al., Langmuir, 2004, 20, 484–490. 16. U. Boas, A. J. Karlsson, B. F. M. de Waal, and E. W. Meijer, J. Org. Chem., 2001, 66, 2136–2145. 17. G. R. Newkome, Z. Q. Yao, G. R. Baker, and V. K. Gupta, J. Org. Chem., 1985, 50, 2003–2004. 18. A. M. Naylor, W. A. Goddard, G. E. Keifer, and D. A. Tomalia, J. Am. Chem. Soc., 1989, 111, 2339–2341.

ACKNOWLEDGMENT

19. M. F. Ottaviani, N. J. Bossmann, N. J. Turro, and D. A. Tomalia, J. Am. Chem. Soc., 1994, 116, 661–671.

The authors thank EPSRC (EP/G026203/1) for postdoctoral funding for BWG.

20. C. J. Hawker, K. L. Wooley, and J. M. J. Fr´echet, J. Chem. Soc. Perkin Trans. 1, 1993, 1287–1297. 21. For a recent example see: C. Liu, C. Gao, and D. Yan, Macromolecules, 2006, 39, 8102–8111.

REFERENCES 1. For review articles on dendrimers see: (a) O. A. Matthews, A. N. Shipway, and J. F. Stoddart, Prog. Polym. Sci., 1998, 23, 1–56; (b) D. A. Tomalia, Adv. Mater., 1994, 6, 529–539; (c) K. Inoue, Prog. Polym. Sci., 2000, 25, 453–571; (d) F. Vogtle, S. Gestermann, R. Hesse, et al., Prog. Polym. Sci., 2000, 7, 987–1041.

22. (a) J. Jensen, E. M. M. De Brabander-Van den Berg, and E. W. Meijer, Science, 1994, 266, 1226–1229; (b) J. Jensen, E. W. Meijer, and E. M. M. De Brabander-Van den Berg, J. Am. Chem. Soc., 1995, 117, 4417–4418.

2. J. M. J. Fr´echet, Science, 1994, 263, 1710–1715.

23. (a) S. Mattei, P. Seiler, and F. Diederich, Helv. Chim. Acta, 1995, 78, 1904–1912; (b) S. Mattei, P. Wallimann, B. Kenda, et al., Helv. Chim. Acta, 1997, 80, 2391–2417; (c) T. Habicher, F. Deiderich, and V. Gramlich, Helv. Chim. Acta, 1999, 82, 1066–1095; (d) F. Diederich and B. Felber, Proc. Natl. Acad. Sci. U.S.A., 2002, 99, 4778–4781.

3. For a review of hyperbranched polymer synthesis see: C. R. Yates and W. Hayes, Eur. Polym. J., 2004, 40, 1257–1281.

24. (a) D. K. Smith and F. Diederich, Chem. Commun., 1998, 2501–2502; (b) D. K. Smith, A. Zingg, and F. Diederich, Helv. Chim. Acta, 1999, 82, 1225–1241.

4. J.-M. Lehn, Supramolecular Chemistry: Concepts and Perspectives, VCH, Weinheim, 1995.

25. F. Diederich, Cyclophanes, Royal Society of Chemistry, Cambridge, UK, 1991.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Hyperbranched polymers in supramolecular chemistry

23

26. W. Tian, X. Fan, J. Kong, et al., Macromolecules, 2009, 42, 640–651.

44. K. C.-F. Leung, P. M. Mendes, S. N. Magonov, et al., J. Am. Chem. Soc., 2006, 128, 10707–10715.

27. For early examples of porphyrin dendrimers see: (a) R. Sadamoto, N. Tomioka, and T. Aida, J. Am. Chem. Soc., 1996, 118, 3978–3979; (b) K. W. Pollak, J. W. Leon, J. M. J. Fr´echet, et al., Chem. Mater., 1998, 10, 30–38.

45. For reviews containing dendronised polymers see: (a) H. Frauenrath, Prog. Polym. Sci., 2005, 30, 325–384; (b) A. Carlmark, C. J. Hawker, A. Hult, and M. Malkoch, Chem. Soc. Rev., 2009, 38, 352–362.

28. (a) M. S. Wendland and S. C. Zimmerman, J. Am. Chem. Soc., 1999, 121, 1389–1390; (b) L. G. Schultz, Y. Zhao, and S. C. Zimmerman, Angew. Chem. Int. Ed., 2001, 40, 1962–1966; (c) S. C. Zimmerman, I. Zharov, M. S. Wendland, et al., J. Am. Chem. Soc., 2003, 125, 13504–13518.

46. For a review of non-viral gene vectors see: S. Han, R. I. Mahato, Y. K. Sung, and S. W. Kim, Mol. Ther., 2000, 2, 302–317.

29. (a) C. J. Hawker and J. M. J. Fr´echet, Chem. Commun., 1990, 1010–1013; (b) C. J. Hawker and J. M. J. Fr´echet, J. Am. Chem. Soc., 1990, 112, 7638–7647. 30. Reviews of molecularly imprinted polymers: (a) J. Steinke, D. C. Sherrington, and I. R. Dunkin, Adv. Polym. Sci., 1995, 123, 81–125; (b) C. Alexander, L. Davidson, and W. Hayes, Tetrahedron, 2003, 59, 2025–2057. 31. P. S. Corbin, L. J. Lawless, Z. T. Li, et al., Proc. Nat. Acad. Sci., 2002, 99, 5099–5104. 32. (a) S. C. Zimmerman, F. Zeng, D. E. C. Reichert, and S. V. Kolotuchin, Science, 1996, 271, 1095–1099; (b) F. Zeng, S. C. Zimmerman, S. V. Kolotuchin, et al., Tetrahedron, 2002, 58, 825–843; (c) P. Thiyagarajan, F. Zeng, C. Y. Ku, and S. C. Zimmerman, J. Mater. Chem., 1997, 7, 1221–1226. 33. (a) J. D. Watson and F. H. C. Crick, Nature, 1953, 171, 964–967; (b) J. D. Watson, The Double Helix, Weidenfeld and Nicolson, London, 1968, p. 195.

47. J. Haensler and F. C. Szoka, Bioconj. Chem., 1993, 4, 372–379. 48. J. F. Kukowska-Latallo, A. U. Bielinska, J. Johnson, et al., Proc. Nat. Acad. Sci., 1996, 93, 227–236. 49. D. Joester, M. Losson, R. Pugin, et al., Angew. Chem. Int. Ed., 2003, 42, 1486–1490. 50. See, for example: (a) M. Enomoto, A. Kishimura, and T. Aida, J. Am. Chem. Soc., 2001, 123, 5608–5609; (b) Y. Kim, M. F. Mayer, and S. C. Zimmerman, Angew. Chem. Int. Ed., 2003, 42, 1121–1126. 51. I. Gitsov and J. M. J. Fr´echet, Macromolecules, 1993, 26, 5608–5609. 52. (a) J. C. M. van Hest, D. A. P. Delnoye, M. W. P. L. Baars, et al., Science, 1995, 268, 1592–1595; (b) J. C. M. van Hest, M. W. P. L. Baars, C. Elissen-Rom´an, et al., Chem. Eur. J., 1995, 268, 1616–1626. 53. M. Kellermann, W. Bauer, A. Hirsch, et al., Angew. Chem. Int. Ed., 2004, 43, 2959–2962. 54. V. Percec, P. Chu, G. Ungar, and J. Zhou, J. Am. Chem. Soc., 1995, 117, 11441–11454.

34. W. T. S. Huck, R. Hulst, P. Timmerman, et al., Angew. Chem. Int. Ed., 1997, 36, 1006–1008.

55. V. S. J. Balagurusamy, G. Ungar, V. Percec, and J. Johansson, J. Am. Chem. Soc., 1997, 119, 1539–1555.

35. Y. G. Ma, S. V. Kolotochin, and S. C. Zimmerman, J. Am. Chem. Soc., 2002, 46, 13757–13759.

56. V. Percec, M. Peterca, M. J. Sienkowska, et al., J. Am. Chem. Soc., 2006, 128, 3324–3334.

36. P. R. Ashton, P. J. Campbell, E. J. T. Chrystal, Angew. Chem. Int. Ed, 1995, 34, 1865–1869.

et al.,

57. B. M. Rosen, D. A. Wilson, C. J. Wilson, et al., J. Am. Chem. Soc., 2009, 131, 17501–17521.

37. F. Huang and H. W. Gibson, J. Am. Chem. Soc., 2004, 126, 14738–14739.

58. V. Percec, W.-D. Cho, P. E. Mosier, et al., J. Am. Chem. Soc., 1998, 120, 11061–11070.

38. Y. Wang, F. W. Zeng, and S. C. Zimmerman, Tetrahedron Lett., 1997, 38, 5459–5462.

59. For a comprehensive review see: B. M. Rosen, C. J. Wilson, D. A. Wilson, et al., Chem. Rev., 2009, 109, 6275–6540.

39. D. K. Smith, Chem. Commun., 1999, 1685.

60. D. Yan, Y. Zhou, and J. Hou, Science, 2004, 303, 65–67.

40. Lead references concerning branched L-lysine derivatives see: (a) R. G. Denkewalter, J. Kolc, and W. J. Lakasavage, US 4289872; (b) J. P. Tam, Proc. Natl. Acad. Sci. U.S.A., 1988, 85, 5409–5413; (c) T. M. Chapman, G. L. Hillyer, E. J. Mahan, and K. A. Shaffer, J. Am. Chem. Soc., 1994, 116, 11195–11196.

61. For a review of supramolecular gelators see: N. M. Sangeetha and U. Maitra, Chem. Soc. Rev., 2005, 34, 821–836.

41. G. M. Dykes, L. J. Brierley, D. K. Smith, et al., Chem. Euro. J., 2001, 7, 4730–4739. 42. (a) N. Yamaguchi, L. M. Hamilton, and H. W. Gibson, Angew. Chem. Int. Ed., 1998, 37, 3275–3279; (b) H. W. Gibson, H. Yamaguchi, L. Hamilton, and J. W. Jones, J. Am. Chem. Soc., 2001, 124, 4653–4665.

62. G. R. Newkome, C. N. Moorefield, G. R. Baker, et al., Angew. Chem. Int. Ed., 1992, 31, 917–919. 63. W. D. Jang, D. L. Jiang, and T. Aida, J. Am. Chem. Soc., 2000, 122, 3233. 64. A. R. Hirst and 10851–10857.

D. K. Smith,

Langmuir,

2004,

20,

65. Y. Zhang, W. Huang, Y. Zhou, and D. Yan, Chem. Commun., 2007, 2587–2589.

43. (a) G. M. Dykes, D. K. Smith, and G. J. Seeley, Angew. Chem. Int. Ed., 2002, 41, 3254–3257; (b) G. M. Dykes and D. K. Smith, Tetrahedron, 2003, 59, 3999–4009. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc139

Supramolecular Chemistry in Polymer Networks Pol Besenius1 and Peter A. G. Cormack2 1 2

Eindhoven University of Technology, Eindhoven, The Netherlands University of Strathclyde, Glasgow, UK

1 Introduction 2 Noncovalent Networks in Bulk Materials 3 Noncovalent Networks in Gels 4 Conclusions and Outlook Acknowledgments References

1

1 1 12 14 15 15

INTRODUCTION

Polymer networks comprising linear macromolecules crosslinked covalently to produce infinitely high-molar-mass, insoluble polymer matrices, have found numerous interesting and valuable applications, ranging from protective organic coatings, high-performance composites, biomaterials and drug delivery systems, to sorbent materials, separation media, and supports for catalysts, to name but a few. The cross-linking units or branching points that give such polymer networks their characteristic material properties need not be restricted to structures based on covalent bonds; cross-links and branch points can also be derived from structures based on noncovalent bonds. Generally speaking, the physical form of polymer networks ranges from highly swollen soft gels, via rubbers to hard materials. The physical form adopted depends primarily upon the crosslinking density (number of cross-links per unit volume), the nature of the monomeric subunits, and the choice of polymer backbone.1

A new class of materials has emerged, which incorporates reversible noncovalent interactions into the polymer backbone or side chains (Figure 1).2–5 Pendent molecular recognition units that can act as noncovalent cross-linkers can lead to novel polymer networks, which exhibit unusual and interesting properties due to the dynamic character and tunable strength of these supramolecular interactions.6 Similar molecular design motivations have led to the recent development of reversible covalent cross-linking, as a potential route to self-healing materials, for example. Adaptable networks based on dynamic covalent chemistry are not covered in this review; we refer the reader to the recent literature on this very intriguing class of materials.7–10 The concept of self-healing materials based on reversible noncovalent approaches11–13 is discussed in detail elsewhere in this volume (Self-Healing and Mendable Supramolecular Polymers, Soft Matter). In the first part of this chapter, we aim to introduce and discuss complementary functional groups that have been developed to allow for reversible network formation in bulk materials. The second part of the chapter deals mainly with polymer matrices in the gel state. In both parts of the chapter, we highlight how material properties at the mesoand macroscopic scales are governed by noncovalent forces on the molecular level, and how supramolecular interactions can offer opportunities in the development of stimuliresponsive materials. Lead examples and applications are highlighted throughout.

2

NONCOVALENT NETWORKS IN BULK MATERIALS

Supramolecular polymer chemistry, as introduced by JeanMarie Lehn, is a relatively new field which has emerged Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

2

Soft matter Side-chain

(a)

Main-chain (b) Combined

(c) Network

(d)

Figure 1 Side-chain and main-chain supramolecular polymers of varying architecture: linear polymers and cross-linked polymer networks. (Reproduced from Ref. 16.  Wiley-VCH, 2001.)

from the marriage of polymer science with noncovalent chemistry.2 Supramolecular polymers can been defined as polymeric arrays of monomeric units that are brought together by reversible and highly directional noncovalent interactions, resulting in polymeric properties both in dilute and concentrated solution and in the bulk.3 Parameters that dictate crucial material properties of covalent polymers, such as degree of polymerization (DP) and configuration, also apply to supramolecular polymers. It should be noted that when referring to monomeric units one is not necessarily limited to low-molar-mass building blocks; larger polymerizable units or macromonomers can also be used. However, in the strictest sense of the above definition, polymeric behavior according to well-established theories in polymer physics arises from supramolecular polymerization. Considerable advances have been achieved in integrating hydrogen bonding, Coulombic interactions, and metal coordination bonds into the design of macromolecular scaffolds, particularly with respect to the tunability of their material properties and morphologies. The appealing tunability feature arises from the fact that the strength of the noncovalent interaction typically varies as a function of the temperature, solvent, or pH, for example. External stimuli can therefore be used to trigger specific changes in

the properties of supramolecular materials. This concept combines attractive features from conventional covalent polymers with properties arising from the reversible and dynamic nature of noncovalent bonds. Normally, one distinguishes between noncovalent main-chain and noncovalent side-chain supramolecular polymers; however, it is possible to form all known architectures, including linear homopolymers, alternating and block copolymers, branched or graft structures, and cross-linked networks or matrices.4 Despite the increased understanding of noncovalent polymerization mechanisms and the undeniable advantages of supramolecular polymers,5 key polymer material properties such as tensile strength [the maximum stress (e.g., external force) a material can withstand while being stretched, before breaking] and perfect elasticity (physical property of a material that returns to its original shape after the stress) require irreversible covalent bonds to link the repeating units in the main-chain. We therefore restrict our review primarily to network supramolecular materials, and refer the reader to Coordination Polymers, Supramolecular Materials Chemistry, Multicomponent Self-Assembled Polymers Based on π-Conjugated Systems, and Hyperbranched Polymers in Supramolecular Chemistry, Soft Matter for a broader coverage of supramolecular polymer architectures. The commercial importance of cross-linked polymers has already been stressed. Noncovalent cross-links introduce new properties, such as reversibility and enhanced control over the network architecture, and lead to tunable and responsive material properties, for example, based on the thermal sensitivity of hydrogen-bonding motifs or based on redox reactions for metal–ligand coordination complexes. In the first instance, we highlight some of the early strategies that were developed to introduce noncovalent, reversible, and tunable cross-links into main-chain covalent polymers, before moving on to recent developments and specific examples in materials science. Thereafter, we use the example of a very well-known, quadruple hydrogenbonding motif as a supramolecular synthon to discuss applications of supramolecular polymeric materials in areas ranging from thermoplastic elastomers, to coatings, tissue engineering, bioinspired polymers, and their mechanical properties. We have restricted our coverage to a few lead examples; references to secondary literature are provided to guide the interested reader toward more elaborate reviews and textbooks.

2.1

Supramolecular cross-linking

Noncovalent network formation has been described extensively using either main-chain or side-chain supramolecular polymers (Figure 1). In the latter case, the general design comprises the incorporation of molecular recognition units

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Supramolecular chemistry in polymer networks

O

H

N O

N

N

N

N

O N

H

O

Figure 2 Formation of a supramolecular network by hydrogen bonding between phenylurazole units pendent as side chains on polybutadiene. (Reproduced from Ref. 3.  American Chemical Society, 2001.)

)

into either a covalent polymer main-chain or as pendent side groups. An early, and widely used, commercial example of sidechain functionalities inducing interactions between polymer chains are ionomers, hydrocarbon macromolecules bearing, for example, carboxylic acid groups [e.g., poly(ethyleneco-methacrylic acid)], which are partially or fully neutralized with metal or quaternary ammonium ions.14 These ionomers are thermoplastic ionic polymers boasting unique physical properties such as enhanced impact strength, toughness, and thermal reversibility. They were developed and commercialized by DuPont, and have recently attracted attention due to their self-healing properties.8 Stadler and coworkers have made important contributions to the field of supramolecular polymer chemistry through their studies of polybutadienes derivatized with hydrogenbonding phenylurazole derivatives (Figure 2).15 Lightscattering experiments, optical measurements, and thermal analysis were used to probe the formation of thermoplastic elastomers and elastomeric blends: at lower temperatures, hydrogen bonding contributes to network formation and elastic behavior, whereas at higher temperatures these

(

O C O H

interactions vanish and the bulk material flow properties are similar to a low-molecular-weight polymer.3 In a number of examples, self-complementary hydrogenbonded motifs (two- or four-point hydrogen bonding) with a high tendency for self-association or dimerization have been used in the development of thermo-reversible polymer networks. Generally speaking, the network and material properties can be tuned by the cross-linking density, the thermodynamic and kinetic stability of the self-associating units, or by their distribution over the covalent mainchain backbone, for example. Section 2.2 highlights some examples based on a very well-known, self-complementary, quadruple hydrogen-bonding motif. In contrast, in a two-component approach, complementary linkers are used that undergo little or no selfassociation. Network formation in the case of complementary binding motifs is achieved by adding an external cross-linker that enables reversible interactions with the recognition units attached to the polymer chain, or alternatively by mixing blends with complementary binding motifs. In the latter case, the two components need not necessarily be polymers; they can also be multivalent lowmolar-mass compounds, where one of the monomeric units should have functionality greater than 2. Both approaches have been reported for the synthesis of liquid crystalline (LC) polymers. For example, benzoic acid–pyridine mesogenic complexes, first described by Kato and Fr´echet, are well studied and lead to thermally stable smectic and nematic phases.16 Reversible-phase transitions in LC networks were obtained by using bifunctional pyridines as hydrogen-bond acceptors, with either benzoic acid pendent polyacrylates17 or low-molar-mass trifunctional benzoic acid-based units as hydrogen-bond donors (Figure 3).16 In )

N

N

H O C O

O CH2 O (CH2)6 O C CH n (

HC C O (CH2)6 O H 2C O n

3

O HOOC

O

O

HOOC

O

O

O

O

COOH

O

COOH

O O

O

O

O

O

COOH

O

COOH

Figure 3 Reversible phase transitions of supramolecular LC networks based on bifunctional pyridines as hydrogen-bond acceptors, with benzoic acid pendent polyacrylates (left); low–molar-mass trifunctional benzoic acid units (right) can also be used as hydrogen-bond donors. (Reproduced from Ref. 16.  Wiley-VCH, 2001.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

4

Soft matter

the absence of supramolecular interactions, the monomeric units did not exhibit liquid crystallinity, which when considered alongside the dynamic properties of the networks, highlights the potential of self-assembled systems to access supramolecular functional materials. Rotello and coworkers have incorporated triple hydrogen-bonding moieties into side-chain polymers, via the postpolymerization functionalization of polystyrenes (PS). The versatility of these “plug and play polymers”18 has been disclosed in a large number of publications, and ranges from ordered gold colloid aggregates to the reversible formation of microspheres and vesicles. In the latter example, PS functionalized with diamidopyridine (DAP) recognition units (marked in blue, Figure 4) selfassemble cooperatively to form, in a thermally reversible

manner, micrometer-sized, spherical cross-linked aggregates.19 Upon the addition of a thymine functionalized polymer (marked in red, Figure 4), these microspheres are converted into vesicle-like aggregates. Differential interference contrast (DIC) microscopy and laser confocal scanning microscopy (LCSM) were used to show that the morphology change was reversible; following the addition of DAP-functionalized polymer, the vesicles converted back to microspheres. Zimmerman and coworkers made use of complementary quadruple hydrogen-bonding complexes to form networks and blend two different side-chain functionalized polymers, poly(butyl methacrylate) (PBMA, marked in light blue) and PS (marked in red, Figure 5).20 The polymer blends were characterized by differential scanning calorimetry (DSC),

O O

O

N N N H H H H N N N O

O O

0.25

n

0.5

0.25

0.5

n

0.5

40.00 µm

O Cl

O

N O N R O

O O O Et N N N Et H H

O O

O N N N H H H O N O N 40.00 µm

Figure 4 Schematic illustrations and LCSM images of reversible morphology control from microspheres to vesicles using specific noncovalent interactions. (Reproduced from Ref. 19.  American Chemical Society, 2004.) * *

m

n

O O

O

4 N H

O

N

N H

N

R

*

*

m

O

O

n

O

O

N

O O

N

O

O

N H N

O O

O

O

H

N H

N Bu

O

Figure 5 Quadruply hydrogen-bonded heterocomplexes for the formation of supramolecular polymer blends. (Reproduced from Ref. 20.  American Chemical Society, 2005.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Supramolecular chemistry in polymer networks size-exclusion chromatography (SEC), and viscometry to confirm the formation of reversible networks. Interestingly, transparent films were obtained in these blends and phase separation was avoided. Even at low loadings of functional groups, the formation of homogeneous blends was confirmed on the basis of two polymer backbones, which without hydrogen-bonding complexes would otherwise be immiscible. This polymer compatibilization strategy has potential for applications in materials science. Inspired by Jean-Marie Lehn’s pioneering work, a diverse range of binding motifs has been employed in supramolecular network formation. The reader is referred to a recent review for a very comprehensive list of references.6 Figure 6 provides a short overview of some possible complementary supramolecular binding motifs: Weck and coworkers21 have cross-linked polynorbornenes bearing pendent cyanuric acid (CA) with bis-triazine- or bis-isophthalamide (ditopic Hamilton wedge)-based crosslinkers (Figure 6a); the groups of Schubert (Figure 6b),22 Craig (Figure 6c),23, 24 and Weck (Figure 6d)25 have also reported metal coordination complexes as cross-linkers. In the latter case, Weck and coworkers have shown that simultaneous side-chain functionalization and crosslinking can be achieved through an orthogonal strategy by combining, for example, the hydrogen-bonding complex diaminopyridine-thymine with palladium-based pincer complexes.25 Using viscosity measurements, the thermaland chemo-reversible cross-linking of these networks was established. When relying on supramolecular polymerizable entities, an important property is not only the equilibrium or association constant (K) of the dynamic bond, describing its thermodynamic stability, but also the kinetic parameters, and the formation (k1 ) and dissociation rates (k−1 ), whereby K = k1 /k−1 . In studies of supramolecular networks, based on different bis-Pd(II) and Pt(II) organometallic crosslinkers and poly(4-vinylpyridine) that form organogels in DMSO (Figure 6), Craig and coworkers23, 24 have shown, via detailed rheological studies, that, if the association constant of the complex remains effectively unchanged, a decrease in the exchange rate can be correlated with an increase in the bulk mechanical properties. Under nonequilibrium conditions, such as those applied by a mechanical stress, molecular dynamics are particularly important and suggest that “slower” dissociation rates mean “stronger” materials. These results have highlighted that the dynamics, rather than the equilibrium structure, determine the viscoelastic properties of the materials under investigation. For an update on how the molecular dynamics can influence material properties, we refer the reader to the recent literature.26 So far, we have restricted our discussion of supramolecular network materials to chemically cross-linked systems.

5

However, it is possible that main-chain supramolecular polymers behave similarly to multiblock copolymers: nanoand microphase segregation between incompatible “soft” core units [e.g., poly(ethylene glycol) or polytetrahydrofuran] and “hard” supramolecular binding motifs (e.g., aromatic units, ionic groups, or metal coordination complexes) can lead to physical cross-linking. Block copolymer phase separation in the bulk is well known and originates from the immiscibility of one block in the other (Assembly of Block Copolymers, Soft Matter). On the basis of wide angle Xray scattering (WAXS) experiments and dynamic mechanical thermal analysis (DMTA), Rowan and coworkers have provided evidence in support of phase-segregated morphologies and elastomeric behavior in films based on nucleobase hydrogen-bonded supramolecular polymers,27 and Zinc-based metallosupramolecular polymers (Figure 7).28 In these specific examples, the network architecture is formed by the hard phase, i.e., the terminal supramolecular motif that serves as nodes or cross-links of the network.29 Palmans and Meijer have reported wholly organic supramolecular materials: the phase segregation of benzene1,3,5-tricarboxamide (BTA)-based helical nanorods within an amorphous polymer such as poly(ethylene butylene) (PEB) resulted in elastomeric behavior (Figure 8).30 The BTA motif has two binding sites (top and bottom “face” of the discotic), which provide the basis for both supramolecular chain extension and simultaneous cross-linking. Furthermore, hydrogen-bonding interactions and order in the nanofibers were studied via infrared (IR) spectroscopy and circular dichroism (CD). The materials are liquid crystalline at room temperature, which leads to their elastomeric properties. In many reported examples, the extent to which enhanced mechanical properties of supramolecular materials are related to phase segregation or can be ascribed to the strength of lateral noncovalent interactions between hard segments, and the quantitative correlation of the two latter phenomena, remains unclear. Hayes and coworkers have very recently investigated the influence of the strength of the hydrogen-bonding motif on the elastomeric properties of telechelic supramolecular polymers: using temperaturedependent small-angle X-ray scattering (SAXS) and rheological analysis, a clear influence of the association constant on the microphase separation and mechanical properties was observed.32 Additionally, Palmans, Meijer, and coworkers have highlighted the influence of the polarity of the telechelic covalent backbone: apolar telechelics led to the formation of stable, phase-segregated nanorods (Figure 8), but by increasing the polarity of the backbone a decrease in nanorod stability and even loss of nanorod formation ability was observed.31

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

6

Soft matter

O

O C8H17

H

9 O

1 12.5 O

O O (CH2)11 H N N N H

O H N

O H N

N N

H

N

C8H7 O

O

N

O

H N

O C8H7

9 O

n

H

H

O

N

N O

O

N

N

H O

O

H

Poly-CA

N N H O

N

N (CH2)11 O O O

O 1

S

O

H

N

(CH2)10 O

N

m

N N H

N

N OH (CH2)11 O

O

N Pr

O

12.5

1 12.5

9

Poly-CA-wedge

O O

N

H OH

HO

Pr C8H17 O

Poly-CA-triazine

Cl

N

H

O (CH2)11 N O

O

9 (a)

N

Pr H

1 12.5

O N H O

H

O

N

N H O

O (CH2)11 N O

O

N

N

O (CH2)10

H

H N

(CH2)10 O

H

N

O

1 12.5

9 O

O C8H17 Pr

n (CH2)6

2 OTf

N

2L Y X

M

X Y

Y X

M

X Y

N 2

N

Ru

N N

N N

M = Pd, Pt X=N Y = alkyl

N O Cl

S

(CH2)6 O

n n

(b)

m

(c)

Ph

O

(CH2)11 O

O O C8H17

m

Ph

Pd N

BF4

S

S

O

n

BF4

(CH2)2

N

Pd

S

S

Ph

Ph

O O (CH2)11 O

O

n

C8H17 O

m

(d)

Figure 6 Various supramolecular cross-linking strategies based on (a) the three-point CA–triazine couple and the six-point CA–isophthalamide (Hamilton wedge) couple; (b) ruthenium–terpyridine coordination complexes; (c) and (d) the platinum- or palladium-based pincer complexes.6 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Supramolecular chemistry in polymer networks

N

N N

N O

O

7

ZnII

N

N

m N N

N N

n

Phase segregation

Figure 7 Structure of Zn(II)-based metallosupramolecular polymer and the proposed multiblock-like phase segregation present in the solid state. The images highlight the elastic nature of the metallosupramolecular polymer. (Reproduced from Ref. 29.  American Chemical Society, 2009.)

O N

O R N H

H N O

H

H

O

H

H O

N

N

H O

N

O

H

N X N N

O

N O

O

R

H

R

R

O

H O N

N H H

R=

X= n

m

Figure 8 Schematic representation of benzene-1,3,5-tricarboxamide nanorods embedded in a poly(ethylene-co-butylene) polymer matrix.30, 31

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

8

Soft matter

O N H

H N N H

N H

N H

N H

N H

N H

N H

H N N H

N H

N H

N H

H N N H

N H

H N

N H

N H

N H

H N O

O

O

H N O

O

H N

H N O

O

O

N H

N H

N H

H N

H N

H N O

H N

H N

H N

O

N H

N H

O

O

H N

H N O

O

O

O

O H N

H N

H N N H

N H H N

H N O

O

O H N

H N N H

N H H N

H N

O N H

O

H N O

O

H N N H

H N O

O O

O

m

N

N

H

H

x

n

Figure 9 Poly(ether-urea) block copolymers phase separate after one-dimensional stacking and aggregation into nanofibres, as visualized by atomic force microscopy (AFM) phase images. (Reproduced from Ref. 34.  American Chemical Society, 2006.)

Another supramolecular motif that has received considerable attention is the urea unit. Inspired by the impressive mechanical properties of hydrogen-bonded polyamides and polyurethanes, bis-urea-based supramolecular polymers have been developed by Sijbesma and Meijer.33, 34 Strong hydrogen bonding gives rise to supramolecular cross-links, and further bundling of the one-dimensional aggregates leads to nanofiber crystallization (Figure 9).35 This process reinforces the materials significantly: owing to the low-glass transition temperature, low-temperature flexibility is characteristic, while elongation of the materials demonstrates their highly elastic behavior with a strain at break ranging from 1000 to 2100%.34 However, crystallization reduces the mobility of the hydrogen-bonding units at room temperature because of the high melting point of the nanofibers. For self-healing applications at room temperature, for example, such crystallization should be avoided in supramolecular network materials.12 Similarly, for ureidopyrimidinone (UPy) supramolecular materials,36 which are introduced in more detail in the next section, the influence of phase segregation and lateral interactions for enhanced mechanical properties has been highlighted.37, 38 By introducing urethane linkers—weak lateral interaction units—and urea groups, with an even stronger tendency for aggregation into UPy-based supramolecular polymers, one-dimensional aggregation of end groups into long fibers is reinforced in each case (Figure 10). Despite the relatively low-molar mass of the amorphous covalent polymer backbone (typically 107 M−1

N

O

O

n Bu

CDCl3

H

N H

N H

O

N

N

N

y

x

N

O

H N

H N

H

O

O

O

N H

O

N H

N

O

n Bu H

O H3C

N

H N

O

H N

n Bu (a)

N H

N H O

y

O

O

CH3

N

i Pr

x

O

H N H

O

O y

7

O

x

10

N N

O (b)

Figure 12 Quadruple hydrogen bonding of 2-ureido-4[1H]-pyrimidinones and their incorporation into (a) covalent main-chain polyolefins (b) and poly(alkyl methacrylates).40, 41

R

CH3 N H O

O

H N

N

H N

N H

N

N H

O

O H N CH3

R O O

O O O n

Polyol

O

O

O H m

O Covalent cross-link Supramolecular cross-link

Figure 13 Schematic representation of networks with covalent and noncovalent cross-links (left), based on the star polyester and polyol depicted on the right. (Reproduced from Ref. 43.  American Chemical Society, 2009.)

potential of supramolecular polymer chemistry for the design of new thermoplastic elastomers, where the aim is to maximize the enhancement of a particular material property while minimizing the incorporation of functional units, and thereby the costs involved. Recently, Wietor and Sijbesma have reported networks based on poly(ε-caprolactone) (PCL) and poly(L-lactic acid) (PLA), which were cross-linked with different ratios of covalent to noncovalent units (Figure 13).43 These

materials combine properties unique to thermosets, such as resistance to solvent swelling, with features typical of thermoplastic elastomers. Partial replacement of covalent crosslinks with reversible cross-links (UPy groups) resulted in superior stress relaxation behavior. The strategy, referred to as preemptive healing, opens the possibility of novel applications in the coatings industry because the ability of highly cross-linked thermoset materials to relax mechanical tensions that originate from film shrinkage during the curing

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Supramolecular chemistry in polymer networks

HN

NH2 HN

NH

O

N

H N NH

H N O

O N H

11

H N

COOH O H N

O N H

O

O

NH OH

O

N H

N

H N

O

H N

N H

N

OH

UPy-GRGDS

O

UPy-PHSRN

N H N

O N H

O

NH2 NH

O N H OH

H N O

O OH O NH2

NH O

N

N H

O N H

O

H N

O O

O n

O 2

PCLdiUPy

Figure 14 The modular approach to construct bioactive materials consists of mixing different UPy-functionalized biomolecules (UPyGRGDS in red and UPy-PHSRN in green) with an oligocaprolactone functionalized on both ends with UPy units (PCLdiUPy). (Adapted from Ref. 44.  Nature Publishing Group, 2005.)

stage, or deformation of the substrate, is rather limited in general. The dynamic nature of supramolecular materials also makes them an appealing target for the development of scaffolds in biomedical applications. Dankers and Meijer used UPy-functionalized telechelic PCL as a modular host matrix for the incorporation of bioactive guest molecules.44 By simply mixing these polymers with UPy-derivatized peptides such as UPy-GRGDS and UPy-PHSRN, bioactive scaffolds were obtained using a variety of processing techniques (Figure 14). In vitro results indicated strong and specific cell binding of fibroblasts to the materials containing both UPy peptides. In addition, the formation of giant cells at the interface between bioactive material and tissue was triggered in vivo. Inspired by the skeletal muscle protein titin, Guan and coworkers have reported a biomimetic concept, which exploits a reversibly unfolding modular cross-linker (Figure 15). Titin’s combination of high-strength

(modulus), high-fracture toughness (energy-dissipating capacity) and elasticity is remarkable, and has been related to its multidomain structure, which can lead to sequential unfolding under an external force. In the realm of synthetic polymers, the group of Guan has introduced UPy groups that are linked covalently via double closed loops (Figure 15). One can imagine that by stretching such a multidomain polymer, the force on the chain will increase quickly with extension. However, the unfolding process of a specific domain that occurs once a sufficient amount of force has been applied results in the breaking of noncovalent interactions. This thereby reduces the force on the polymer chain and prevents breaking of covalent bonds. If it is possible to repeat this process until all the domains have unfolded, the material will be both strong and tough, but also elastic: because of the dynamic nature of the hydrogen-bonding interactions, a removal of the force results in a refolding of the domains. Indeed, atomic force microscopy (AFM), single-molecular

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

12

Soft matter

Extension Retraction

(CH2)19 O HN 5( ) OH N NH O

O O O

N HN NH N NH O

O HN H N O ( )5 O

O O

(CH2)19

Figure 15 Concept of biomimetic design of a modular cross-linker, based on double-looped UPy motifs, for enhancing network mechanical properties. (Reproduced from Ref. 45.  American Chemical Society, 2007.)

force-extension experiments have revealed sequential unfolding in the linear synthetic polymers. In network materials,45 these biomimetic cross-linking modules have led to enhanced elastomeric mechanical properties studied via static stress–strain and dynamic mechanical analysis (DMA): compared to poly(ethylene glycol) (PEG) covalently cross-linked materials, higher crosslinking ratios led to an increase in the modulus and tensile strength as anticipated, but crucially without sacrificing the extensibility. In conclusion, the molecular engineering of cross-linkers is clearly a very powerful concept to modulate material properties on a macroscopic level.

3

NONCOVALENT NETWORKS IN GELS

Cross-linked polymer matrices can be prepared using a variety of synthetic methods, wherein their network structure can be controlled via the cross-linking density and the physicochemical properties controlled via the choice of polymer backbone. A lightly cross-linked 3D network derived from a hydrophilic polymer can, in principle, retain large volumes of water, which causes the network to swell and form a hydrogel. This class of gels has received substantial interest in academia and industry because of widespread applications in the development of biomaterials, carriers of therapeutics in controlled drug delivery, tissue engineering and regenerative medicine, or contact lenses, for example.46 Amongst the most widely used synthetic polymers for the preparation of hydrogels are PEG, poly(hydroxyethyl methacrylate) (PHEMA), and poly(vinyl alcohol) (PVA). Synthetic polymers offer the possibility of surface engineering via the conjugation of

functional groups in order to modify cell adhesion or degradation properties, as well as their swelling or mechanical properties. Polymers of natural origin, including proteins such as collagen and fibrin, and polysaccharides such as alginate, agarose, hyaluronic acid, and chitosan, have also been used. Very recently, Aida and coworkers have disclosed the preparation of hybrid hydrogels, in which anionic clay nanosheets were noncovalently cross-linked with cationic dendrimers, or “molecular glues,” to form materials with high mechanical strength and self-healing properties.47 Embedding stimuli-responsive properties, such as reversibility, into the chemical design of biomaterials, while tailoring mechanical and chemical features, is a key consideration for many biomedical applications of hydrogel materials. In this respect, the self-assembly of low-molarmass “gelating” units holds great promise. These gels are thereby supramolecular in the strictest sense, since they are formed through the self-assembly of small molecular building blocks to form entangled fibrillar networks via a combination of noncovalent interactions. We refer the interested reader to more detailed contributions on molecular gels (Self-Assembling Fibrillar Networks—Supramolecular Gels, Soft Matter) and molecular biomaterials (Designing Peptide-Based Supramolecular Biomaterials, Soft Matter). In order to introduce the potential of supramolecular polymer science in the development of hydrogel materials, and the applications thereof, we focus our discussion on the use of supramolecular engineering in manipulating macroscopic material properties. Using inclusion complexes based on noncovalent cross-links, we introduce, as examples, macroscopic hydrogels on the one hand and particulate systems on the other. Specific applications are highlighted in each case.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Supramolecular chemistry in polymer networks

3.1

13

Inclusion complexes in the formation of macroscopic gels

Interlocked molecules, in particular rotaxanes and pseudorotaxanes, have been shown to be a versatile motif in the formation of supramolecular polymers and materials (Self-Assembly of Macromolecular Threaded Systems, Self-Processes).48 Cyclodextrin (CD)-based inclusion complexes (Cyclodextrins: From Nature to Nanotechnology, Molecular Recognition) are amongst the most commonly used because of their water solubility, biocompatibility, ease of functionalization, and low cost. The cucurbiturils (CBs, see Cucurbituril Receptors and Drug Delivery, Molecular Recognition), which are interesting because of their very high binding affinities, as shown very recently for the preparation of supramolecular network polymers, are increasingly becoming popular.49 Crown ethers have also been used for similar purposes50 ; for an extensive list of published examples, we refer the interested reader to recent review articles.48, 51 Supramolecular polymerization of functionalized αand β-CD (macrocylic hexameric and heptameric Dglucopyranoses connected by α-1,4-linkages) has been reported by Harada, one of the pioneers in the field, to form [2]rotaxane homopolymers or alternating copolymers. Cinnamoyl (CI) and adamantyl (ADA) guest molecules were conjugated covalently onto the rim of the CDs; selective host–guest complexation is then the basis for oligomerization and polymerization (Figure 16). Some of these selfassembled polymers can gelate at higher concentrations to form supramolecular hydrogels.48, 51–53 A large body of work has been devoted to the formation of swollen networks in water by the co-aggregation of covalent linear polymers conjugated with guest and CD molecules, not only to form macroscopic hydrogels but also nano- and microparticles.54 Many of the hydrogels, because of the reversible nature of the embedded cross-links, are responsive to external stimuli, for example, changes in pH, the application of shear forces, changes in temperature or the redox potential, and the irradiation with light. Hydrogel materials that show thixotropic behavior are very interesting for biomedical applications and are potentially useful candidates for the direct injection of delivery matrices. Under stress (shaking or agitation, for example), a thixotropic system thins, becomes less viscous and flows, but after the period of stress the viscosity increases and thickening is restored over time. For the encapsulation and delivery of delicate or thermally non-robust goods (cells or proteins), this appears to be a particularly useful concept: they can be injected through a thin needle under pressure into a tissue because of their thixotropic character.

α-CD

β-CD

H2O

+

CIN-β-CD

ADA-α-CD

Alternating α-, β-CD Supramolecular polymer

Figure 16 Preparation of α- and β-CD-alternating supramolecular polymers. (Reproduced from Ref. 52.  American Chemical Society, 2004.)

Efforts toward this concept have been reported by Li and coworkers. Moderate to high-molar-mass PEG (8000–100 000 Da) was used to form polypseudorotaxanes with α-CD in water, leading to PEG chains threading into α-CD from the two ends to form inclusion complexes. Subsequently, these self-assembled into crystalline and water insoluble domains, while the middle segments of PEO remained uncomplexed and hydrophilic. This process resulted in the formation of hydrogels (Figure 17), which were characterized by WAXS, rheology investigations, and in vitro release kinetics.55 Supramolecular hydrogels with thixotropic, reversible properties were indeed obtained. Initially, the material properties could be tuned by varying the molar mass of the PEG; however, the group has made further refinements and improvements, for example, by using amphiphilic block copolymer inclusion complexation between PEG-co-PCL and α-CD. The sustained release of these supramolecular hydrogels was increased from five days to over one month, even for lower-molar-mass PEO blocks, and could be fine-tuned using different copolymer contents and at different ratios of α-CD, which widens their scope and potential applications.56 Owing to their thixotropic properties, these systems might be applicable as injectable, in situ gelling devices, a valuable delivery strategy that circumvents the need for surgical implantation of the material.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

14

Soft matter

Figure 17 Hydrogel formed between α-CD and high-molar-mass PEG, based on the partially formed inclusion complexes. These domains self-assemble to yield physical cross-links. (Reproduced from Ref. 56.  Springer, 2009.)

3.2

Inclusion complexes in the formation of particulate gels

Davis and coworkers have made an impressive contribution to the area of cross-linked CD-based hydrogels and nanoparticles over the last decade. For drug delivery purposes, a linear PEG-based polymer containing β-CDs in the backbone was conjugated with the anticancer drug camptothecin (CPT) via a biodegradable glycine-based linker. In water, the resulting polymers self-assemble in a multivalent fashion into nanoparticles (∼30 nm) driven by CPT/β-CD inclusion complexes. The nanoparticles (referred to as IT101) were therapeutically effective, with low side effects in rodent tumor models; these are currently undergoing phase II clinical trials.57 Key features of the nanoparticles are their long circulation half-lives (because of their size renal clearance is avoided) and localization in tumors. They enter the tumor cells and slowly release the CPT, which leads to their disassembly into individual polymer chains that are sufficiently small to be cleared via the kidneys. Davis and coworkers also developed cationic polymers that contained β-CD units in their backbones, similar to the drug conjugates (Figure 18). Self-assembly with negatively charged plasmid DNA, or small interfering (si)RNA, led to the formation of polyplexes driven by electrostatic interactions. The β-CD functionalized nanoparticles could be decorated with a range of functional groups in order to improve the particles’ stability or to introduce targeting groups for specific cells. For example, ADA-modified PEG, with or without targeting ligands such as transferrin (Tf) for cancer cell targeting, galactose for hepatocyte targeting, or antibody fragments, were coupled noncovalently to polyplexes via inclusion complexes.58

The nucleic acid delivery efficacy of decorated cationic polymer-pDNA (or siRNA) polyplexes has also been demonstrated in several in vivo tumor models. After intravenous administration, these targeted polyplexes were taken up specifically by tumor cells and showed tumor inhibition via knockdown of mRNA or increased expression of tumor suppressor genes. Currently, one siRNA-based polyplex formulation for the treatment of solid tumors in humans has reached phase I clinical trials.59 In addition to hydrogels, the use of CD-based carrier vehicles for targeted drug and gene delivery shows great potential. Tailoring the stability and thereby degradability of the nanoparticles via noncovalent cross-linking is just one of the crucial properties, besides biocompatibility, biodistribution, and delivery efficacy, which will determine the success of future clinical applications of a supramolecular therapeutic.

4

CONCLUSIONS AND OUTLOOK

Pioneering work in supramolecular chemistry has allowed scientists to design and apply noncovalent interactions in numerous fields, exemplified by the high number and broad scope of the various chapters that are part of this work. Clearly, manipulating reversible interactions at the molecular scale to rationally design and control the properties of functional materials at the meso- and macroscopic level holds great potential. Advances in supramolecular materials are disclosed essentially continuously, thus the prospect of commercialization and applications in everyday life technologies has never been so high. We have used this opportunity to highlight some of the opportunities and challenges that a supramolecular engineering approach poses for the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Supramolecular chemistry in polymer networks

15

Tf Tf

Tf

Tf NH2 H3CO

NH2

HS

I

I

H2N

S

S

Plasmid DNA

OCH3 NH2+

Cysteamine

+

ADA-PEG ADA-PEG-Tf Tf

DMS

NH2

S

S

NH2+

H N NH2

+

N H

x

Figure 18 Synthetic route for the preparation of linear polymers with β-CD in their backbone and schematic representation of an ADA-PEG-Tf and ADA-PEG functionalized polyplex. (Reproduced from Ref. 58.  Nature Publishing Group, 2004.)

development of functional soft matter, more specifically network materials in the bulk or gel phase. In both cases, specific and well-defined interactions between molecules contribute to material properties on two levels: structure and dynamics. It is the dynamic nature of the noncovalent interactions that distinguishes supramolecular materials from their covalent counterparts. Making use of the dynamic properties for the development of noncovalent architectures is particularly exciting for applications such as self-healing materials in coating technology or adaptable, modular biomaterials in regenerative medicine and controlled drug delivery, for example. More often than not the dynamic nature of the supramolecular interactions is taken for granted, and mechanistic studies remain rarer compared to structural investigations. It is safe to say that the thermodynamics of the intermolecular interaction are usually a primary design consideration. However, complementary to the recent developments in reversible covalent chemistry, actively manipulating the dynamics of specific noncovalent interactions will enable the supramolecular engineering toolbox to expand enormously,10 moving toward applications of kinetic trapping60, 61 and operating under out-ofequilibrium conditions.62, 63 Natural building blocks have been a continuous inspiration to chemists for the design of self-assembling units into higher-order aggregates. Similarly, concepts such as energy-dissipative processes and nonlinear mechanical properties keep on challenging the imagination of synthetic, physical, and biological scientists to produce biomimetic structures with new functions for applications in material sciences and biomedical technologies.

ACKNOWLEDGMENTS We thank Wilco P. J. Appel and Marko M. L. Nieuwenhuizen (both from Eindhoven University of Technology) for critical reading of the manuscript.

REFERENCES 1. K. Duˇsek and M. Duˇskov´a-Smrˇckov´a, Prog. Polym. Sci., 2000, 25, 1215–1260. 2. J.-M. Lehn, in Supramolecular Polymers, ed. A. Ciferri, CRC Press, Taylor & Francis, Boca Raton, FL, 2005, pp. 3–28. 3. L. Brunsveld, B. J. B. Folmer, E. W. Meijer, R. P. Sijbesma, Chem. Rev., 2001, 101, 4071–4097.

and

4. G. B. W. L. Ligthart, O. A. Scherman, R. P. Sijbesma, and E. W. Meijer, in Macromolecular Engineering: Precise Synthesis, Materials Properties, Applications, eds. K. Matyjaszewski, Y. Gnanou, and L. Leibler, Wiley-VCH Verlag GmbH, Weinheim, 2007, pp. 351–399. 5. T. F. A. de Greef, M. M. J. Smulders, M. Wolffs, et al., Chem. Rev., 2009, 109, 5687–5754. 6. K. P. Nair and M. Weck, in Molecular Recognition and Polymers. Control of Polymer Structure and Self-Assembly, eds. V. Rotello and S. Thayumanavan, John Wiley & Sons, Inc., Hoboken, NJ, 2008, pp. 103–136. 7. S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, et al., Angew. Chem. Int. Ed., 2002, 41, 898–952. 8. S. D. Bergman and F. Wudl, J. Mater. Chem., 2008, 18, 41–62. 9. C. J. Kloxin, T. F. Scott, B. J. Adizma, and C. N. Bowman, Macromolecules, 2010, 43, 2643–2653.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

16

Soft matter

10. R. J. Wojtecki, M. A. Meador, and S. J. Rowan, Nat. Mater., 2010, 10, 14–27.

36. R. P. Sijbesma, F. H. Beijer, L. Brunsveld, et al., Science, 1997, 278, 1601–1604.

11. K. A. Williams, A. J. Boydston, and C. W. Bielawski, J. R. Soc. Interface, 2007, 4, 359–362.

37. B. J. B. Folmer, R. P. Sijbesma, R. M. Versteegen, et al., Adv. Mater., 2000, 12, 874–878.

12. P. Cordier, F. Tournilhac, C. Souli´e-Ziakovic, L. Leibler, Nature, 2008, 451, 977–980.

and

38. H. Kautz, D. J. M. v. Beek, R. P. Sijbesma, and E. W. Meijer, Macromolecules, 2006, 39, 4265–4267.

13. S. Burattini, B. W. Greenland, D. H. Merino, et al., J. Am. Chem. Soc., 2010, 132, 12051–12058.

39. A. W. Bosman, R. P. Sijbesma, and E. W. Meijer, Mater. Today, 2004, 7, 34–39.

14. A. Eisenberg and J. S. Kim, Introduction to Ionomers, John Wiley & Sons, Inc, New York, 1998.

40. L. R. Rieth, R. F. Eaton, and G. W. Coates, Angew. Chem. Int. Ed., 2001, 113, 2211–2214.

15. L. L. de Lucca Freitas and R. Stadler, Macromolecules, 1987, 20, 2478–2485.

41. C. L. Elkins, T. Park, M. G. McKee, and T. E. Long, J. Polym. Sci. Part A: Polym. Chem., 2005, 43, 4618–4631.

16. T. Kato, N. Mizoshita, and K. Kanie, Macromol. Rapid Commun., 2001, 22, 797–814.

42. K. E. Feldman, M. J. Kade, E. W. Meijer, et al., Macromolecules, 2009, 42, 9072–9081.

17. T. Kato, H. Kihara, U. Kumar, et al., Angew. Chem., Int. Ed. Engl., 1994, 33, 1644–1645.

43. J.-L. Wietor, A. Dimopoulos, L. E. Govaert, et al., Macromolecules, 2009, 42, 6640–6646.

18. F. Ilhan, M. Gray, and V. M. Rotello, Macromolecules, 2001, 34, 2597–2601.

44. P. Y. W. Dankers, M. C. Harmsen, L. A. Brouwer, et al., Nat. Mater., 2005, 4, 568–574.

19. O. Uzun, A. Sanyal, H. Nakade, et al., J. Am. Chem. Soc., 2004, 126, 14773–14777.

45. A. M. Kushner, V. Gabuchian, E. G. Johnson, and Z. Guan, J. Am. Chem. Soc., 2007, 129, 14110–14111.

20. T. Park, S. C. Zimmerman, and S. Nakashima, J. Am. Chem. Soc., 2005, 127, 6520–6521.

46. N. A. Peppas, J. Z. Hilt, A. Khademhosseini, and R. Langer, Adv. Mater., 2006, 18, 1345–1360.

21. K. P. Nair, V. Breedveld, and M. Weck, Macromolecules, 2008, 41, 3429–3438.

47. Q. Wang, J. L. Mynar, M. Yoshida, et al., Nature, 2010, 463, 339–343.

22. M. A. R. Meier and U. S. Schubert, J. Polym. Sci. Part A: Polym. Chem., 2003, 41, 2964–2973.

48. A. Harada, A. Hashidzume, H. Yamaguchi, Y. Takashima, Chem. Rev., 2009, 109, 5974–6023.

23. W. C. Yount, D. M. Loveless, and S. L. Craig, J. Am. Chem. Soc., 2005, 127, 14488–14496.

49. E. A. Appel, U. Rauwald, S. Jones, et al., J. Am. Chem. Soc., 2010, 132, 14251–14260.

24. W. C. Yount, D. M. Loveless, and S. L. Craig, Angew. Chem. Int. Ed., 2005, 44, 2746–2748.

50. T. Oku, Y. Furusho, and T. Takata, Angew. Chem. Int. Ed., 2004, 43, 966–969.

25. J. M. Pollino, K. P. Nair, L. P. Stubbs, et al., Tetrahedron, 2004, 60, 7205–7215. 26. D. M. Loveless, F. R. Kersey, and S. L. Craig, in Molecular Recognition and Polymers. Control of Polymer Structure and Self-Assembly, eds., V. Rotello and S. Thayumanavan, John Wiley & Sons, Inc, Hoboken, NJ, 2008, 37–62.

and

51. A. Harada, Y. Takashima, and H. Yamaguchi, Chem. Soc. Rev., 2009, 38, 875–882. 52. M. Miyauchi and A. Harada, J. Am. Chem. Soc., 2004, 126, 11418–11419. 53. A. Harada, R. Kobayashi, Y. Takashima, et al., Nat. Chem., 2011, 3, 34–37.

27. S. Sivakova, D. A. Bohnsack, M. E. Mackay, et al., J. Am. Chem. Soc., 2005, 127, 18202–18211.

54. F. van de Manakker, T. Vermonden, C. F. van Nostrum, and W. E. Hennink, Biomacromolecules, 2009, 10, 3157–3175.

28. J. B. Beck, J. M. Ineman, and molecules, 2005, 38, 5060–5068.

55. J. Li, X. P. Ni, and K. W. Leong, J. Biomed. Mater. Res. A, 2003, 65A, 196–202.

S. J. Rowan,

Macro-

29. J. D. Fox and S. J. Rowan, Macromolecules, 2009, 42, 6823–6835.

56. J. Li, Adv. Polym. Sci., 2009, 222, 79–113.

30. J. Roosma, T. Mes, P. Lecl`ere, et al., J. Am. Chem. Soc., 2008, 130, 1120–1121.

58. M. E. Davis and M. E. Brewster, Nat. Rev. Drug Discovery, 2004, 3, 1023–1035.

31. T. Mes, M. M. J. Smulders, A. R. A. Palmans, and E. W. Meijer, Macromolecules, 2010, 43, 1981–1991.

59. M. E. Davis, J. E. Zuckerman, C. H. J. Choi, et al., Nature, 2010, 464, 1067–1071.

32. P. J. Woodward, D. H. Merino, B. W. Greenland, et al., Macromolecules, 2010, 43, 2512–2517.

60. J. M. A. Carnall, C. A. Waudby, A. M. Belenguer, et al., Science, 2010, 327, 1502–1506.

33. R. A. Koevoets, R. M. Versteegen, H. Kooijman, et al., J. Am. Chem. Soc., 2005, 127, 2999–3003.

61. J. W. Sadownik and R. V. Ulijn, Chem. Commun., 2010, 46, 3481–3483.

34. R. M. Versteegen, R. Kleppinger, R. P. Sijbesma, E. W. Meijer, Macromolecules, 2006, 39, 772–783.

and

62. G. M. Whitesides and B. Grzybowski, Science, 2002, 295, 2418–2421.

35. L. Bouteiller, O. Colombani, F. Lortie, and P. Terech, J. Am. Chem. Soc., 2005, 127, 8893–8898.

63. B. A. Grzybowski, C. E. Wilmer, J. Kim, et al., Soft Matter, 2009, 5, 1110–1128.

57. M. E. Davis, Adv. Drug Deliv. Rev., 2009, 61, 1189–1192.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc140

Stimuli-Responsive and Motile Supramolecular Soft Materials Paul D. Topham1 and Anthony J. Ryan2 1 2

Aston University, Birmingham, UK University of Sheffield, Sheffield, UK

1 Introduction 2 Chemical Actuation 3 Magnetic Actuation 4 Light Actuation 5 Conclusions References

1

1 2 9 12 17 17

INTRODUCTION

Stimuli-responsive macromolecules are central to all living organisms. The quest for materials with the ability to respond to a change in the environment has thus been the basis of research toward artificial biological systems. In particular, artificial muscles have been the focus of many groups for some decades. Artificial muscles have been identified as candidates in robotics, microfluidics, and, perhaps most pertinently, as biomedical devices and implants. Muscles can be described as actuator devices in that they convert chemical energy into mechanical work, often in a nonlinear fashion. A muscle-like actuator is based on a pronounced, yet abrupt, response with a high contractile force. Actuators in the macroscopic engineering world are based on hard, piston-like materials and are used to convert various energy sources into mechanical work with extremely high efficiency. However, at micro- and nanoscopic length scales, the domain in

which living systems operate, such stiff devices are no longer applicable due to high viscosities and rapid heat dissipation.1 Natural organisms, however, have relied on soft materials to transduce chemical and electrical signals into mechanical work, which have been developed over millions of years of biological evolution. If we are to produce synthetic equivalents, we must also turn to soft, wet systems. The actuators discussed in this chapter are those triggered by chemical reactions (namely pH oscillations and electrochemistry), magnetic fields, and photoirradiation. The use of supramolecular construction is a relatively new concept in the field of soft actuation. Historically, chemically crosslinked polymers make up the vast majority of systems studied in this area, which has set the basic ground rules from which we can build. However, the inception of physically bound systems has changed the design parameters of actuator devices, making available an extremely versatile tool box of materials to produce power output from a variety of energy sources. Conventionally, when we think of supramolecular activity, we immediately think of small molecule self-assembly to form macromolecular networks. In this chapter, we concentrate on polymeric supramolecular activity, that is, the self-assembly of polymers and oligomers, which create physically crosslinked networks of pseudoinfinite molecular mass. A thermodynamically driven self-assembly process forms larger constructs from the individual constituent building blocks, just like small molecule supramolecular construction. Physical crosslinking can be achieved by hydrogen-bonding (see Designing Peptide-Based Supramolecular Biomaterials, Soft Matter), ionic clusters, hydrophobic interactions, and microphase separation. Individual bonds formed are weaker than covalent bonds, but collectively can create

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

2

Soft matter

extremely robust systems. The advantages of supramolecular construction of this type include reversibility, selectivity, well-ordered microstructures, and processing simplicity. However, throughout this chapter, chemically crosslinked systems are discussed to provide the background to the different driving forces available and are the focus of certain sections where there are still limited examples of supramolecular actuator devices.

2

CHEMICAL ACTUATION

Chemically driven actuation encompasses a wide variety of different triggers, such as temperature, pH, ionic strength, and electric field. Consequently, the types of systems that can be created are incredibly diverse. Here, we focus solely on polymer constructs that are powered by changes in pH and electric field. A common feature during the mechanism of actuation in all chemically driven actuators is the dependence on diffusion of ions in and out of the polymer network. It is the transport of ions and solvent molecules that dictates the macroscopic dimensions of a polymer gel. The volume of water absorbed within a polymeric network is a balance between the thermodynamic forces of mixing and the elastic, entropically driven restoring force of the polymer. An external trigger can be used to control the extent of mixing by affecting the Flory–Huggins parameter (a measure of how well two components mix, often denoted by χ) between the polymer and the solvent. In other words, the chemical environment can be changed to create a good or bad solvent for the polymer. For linear polymer chains, such a trigger can induce the coil–globule transition (or vice versa), where the polymer passes from a tightly packed hydrophobic coil with a small radius of gyration (in a bad solvent) to an extended solvated globule (in a good solvent). During the transition, free chains simply dissolve in the good solvent and re-precipitate in the bad solvent. To create actuation, however, a volume change of a macroscopically stable structure is needed. This is obtained via crosslinking, which is a process of tying the responsive chains together, so that while they are macroscopically insoluble in the media in which they are immersed even though the chain segments are well solvated. The different methods of crosslinking are discussed throughout this chapter. The pioneering work of Yoshida2–4 in the field of chemically driven polymer gels is beyond the scope of this chapter, but it should be noted that his research has inspired much of the work discussed herein.

2.1

pH-responsive systems

Alongside temperature, pH is one of the most commonly investigated triggers for chemically driven actuators. Part

of the reason for this is that the pH change required to induce the coil–globule transition in polymers can be extremely narrow (generally, less than half of a pH unit). Consequently, dramatic volume changes can be observed with just a subtle variation in pH. Such variations in pH are ubiquitous in biology. pH-responsive macromolecules are an attractive class of ionizable polymers known as polyelectrolytes. There are two types of polyelectrolytes, namely polyacids and polybases. A polyacid is protonated at low pH and is charge neutral. The polymer adopts a hydrophobic, collapsed conformation. However, if the pH is increased above the pKa of the acidic repeat units, the material becomes deprotonated and ionized. Negative charges located along the polymer chains force counter ions and associated water molecules to be driven in from the external solution by osmotic pressure. The polymer adopts a hydrophilic, swollen conformation. This volume transition can be switched “on” and “off ” by cycling the pH above and below the pKa of the polyacid. Polybases exhibit the same phenomenon whereby they are protonated at low pH and deprotonated at high pH. In contrast to polyacids, however, the ionized, swollen form of the polybase is observed at low pH, due to protonation of the basic groups. This is illustrated in Figure 1 with triblock copolymers comprising poly(methacrylic acid) and poly[2-(diethylamino)ethyl methacrylate] as examples of a polyacid and polybase, respectively. The pH-insensitive endblocks used in each case comprises poly(methyl methacrylate). The advantage of using complementary polyelectrolyte pairs of polyacids and polybases is that devices fabricated from these two building blocks can be bipolar. Under the same pH conditions, one polymer will be swollen, while the other is collapsed, as shown in Figure 1. In principle, judicious selection of the respective pKa values of the responsive chemical group should enable a polyacid and a polybase to be used in tandem, equivalent to the antagonistic pairing common in mammalian muscle structures. This is discussed in more detail in Section 2.1.3.

2.1.1 Synthesis To date, the simplest and most cost-effective method of producing polymer gels is to use free radical polymerization on the desired monomer in the presence of a difunctional monomer (crosslinking agent). This approach yields chemically crosslinked polymer networks, which are not synthetically demanding and are relatively reproducible. One of the greatest disadvantages of using a chemically crosslinked system, however, lies in its inherent processability. Most simply, articles are moulded directly by polymerization of a liquid that results in a solid product. The solid, however, has a heterogeneous distribution of crosslinks resulting from the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials

Polyacid

Acid

m

O

O

OH

Polybase

m

O

Neutral / base

O

O OH

O

O O

O

O O

O

O

O O−

O−

m

m

n/2

O

m

n/2

O

O

O

O O

O

O

O

m

n/2

O

O

O

N+

N+ H

Figure 1

m

m

n/2

O

O

3

N

H

N

Schematic representation of the pH-induced volume change in polyacid and polybase triblock copolymer systems.

latter stages of the reaction. In some cases, more complex approaches such as lithography and laser cutting are used to fabricate parts from large blocks, but these too suffer from the same microscale heterogeneity in crosslink density. Physically crosslinked polymer networks, on the other hand, can be processed into various shapes and sizes from the molten or solution state. Physical gels are unique because they can be dissolved and reprocessed allowing for multiple use. Another advantage of physical gels, particularly those derived from self-assembled structures (via microphase separation), is that there is uniformity in the crosslink density compared to chemical gels. This is clearly shown in Figure 2. An increase in uniformity reduces the amount of network stresses when the polymer cycles through repeated volume changes, potentially increasing both the mechanical strength and reusability of the material. Moreover, devices made via microphase separation can be designed to function in one, two, and three dimensions by using spherical, cylindrical, and lamellae block copolymer5 morphologies, respectively. This is particularly useful for creating artificial muscles, where unidirectional actuation is required. For work describing the various morphologies obtainable (via microphase separation) and the factors affecting them, the reader is referred to Assembly of Block Copolymers, Soft Matter.5–7 The discussion outlined in this section is applicable to all of the systems featured in this chapter and is the central focus of the synthetic strategy when designing smart actuators. In order to obtain a physical gel via microphase separation, block copolymers are required. For chemically

(a)

(b)

Figure 2 Schematic representations of chemically (a) and physically (b) crosslinked polymer networks. Due to the thermodynamic driving process of supramolecular constructs, a physically bound material generally exhibits a more ordered microstructure.

triggered gels, one of the blocks must comprise the responsive polymer, whereas the other block(s) should be made of chemically inactive polymer that aggregate to form the crosslinks. This produces inert domains to prevent dissolution and provide structural integrity for the responsive polymer chains. Symmetrical triblock copolymers provide the simplest route to achieving this. Controlled polymerization techniques, such as anionic,8 group transfer polymerization (GTP),9 and reversible addition-fragmentation chain transfer (RAFT)10 polymerization, are employed to produce the desired architecture. Polymers with a broad molecular weight distribution, produced by conventional free radical polymerization, for example, tend not to produce well-defined microstructures.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Soft matter

Finally, to create well-equilibrated, physically crosslinked systems using microphase separation, the polymer system must be annealed. Annealing provides mobility to the individual polymer chains so that they can adopt the most thermodynamically stable conformation, that is, the one with the lowest free energy. The annealing process can be carried out by elevated temperatures or solvent saturation. In both methods, the annealing source must be removed slowly. Rapid removal can lead to vitrification, whereby the polymer is effectively locked in the random configuration in which it was during the annealing process. Although temperature annealing is the most commonly adopted approach due to efficacy, expense, and inherent level of control, solvent annealing can be more favorable, particularly for thermally unstable polymers or systems with high transition temperatures.

2.1.2 Characterization Characterization of pH-responsive actuators can be broken down into two main categories: (i) microstructure analysis and (ii) quantification of the pH-response itself. There are a wide range of techniques that can be used to ascertain the microstructure of a polymer system. For example, microscopy, such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), and atomic force microscopy (AFM), is employed to take a snapshot of a given area of the sample. Alternatively, scattering techniques are used to provide structural information that is averaged throughout the entire sample, or at least the millimeter3 scattering volume. Microphase separated structures are typically on the nanometer length scale and can be characterized by small angle X-ray scattering (SAXS). SAXS can provide detailed information about the size, shape, and distribution of self-assembled aggregates provided that they are monodisperse and well-defined in structure space. When the crystal grains of a polymer system are not all perfectly aligned, only an average length scale between domains can be obtained. The simplest and most common technique to evaluate pH-response is gravimetry. A polymer gel is weighed before and after immersion into a solution for a given time. From this, the equilibrium water content, or EWC, is calculated, as shown in (1). EWC = (me –m0 )/m0

(1)

where me is the mass of the gel at equilibrium and m0 is the original, dry mass of the sample. The EWC is plotted against the pH of the solution in which it is immersed to provide information about the extent of swelling and the pH at which it occurs.

Although simple to apply, gravimetric analysis can give rise to large experimental errors. For example, excess surface water may not be removed on blotting (which is always done prior to weighing) or excess water could accumulate in the interstices between grain boundaries. To overcome this, the SAXS structural peak from microphase separated polymers can be used. For direct comparison with the EWC measured by gravimetry, the length scale obtained can be subsequently converted to a volume. Alternatively, the mass data can be converted to a length scale. Since the majority of the system is water, a density of 1.0 g cm−3 is assumed and therefore the mass can be approximated as the volume. The cube root is then taken for each value and this is substituted in (1) as before. A SAXS approach significantly reduces the errors associated with gravimetry. Although SAXS is a desirable technique and gives an average over a relatively large sample by a noninvasive technique, it is rather expensive to implement in the laboratory and for dynamic experiments requires the valuable resource of a synchrotron, and therefore not accessible to all research groups. Swann et al.11 demonstrated that laser light scattering could be used as an inexpensive technique, which provides more accuracy than gravimetry in obtaining the EWC of a hydrogel. A pH-responsive triblock was solvent cast onto a diffraction grating and the solvent was slowly removed to adopt an equilibrium structure. Consequently, one surface of the triblock hydrogel contained a diffraction pattern with a period of 2 µm, suitable for scattering laser light. Swelling the hydrogel caused an increase in the period of the diffraction pattern, which could be measured to create an accurate value for the EWC at the microscale. The results obtained from gravimetry, SAXS, and laser light scattering for a polybase hydrogel swelling experiment are shown in Figure 3. It is clear to see that all techniques 1.5

(L eq − L ref) /L ref

4

1.0

0.5

0.0 3

4

5

6

7

8

9

10 11

pH

Figure 3 Comparison of the three characterization techniques: SAXS (hollow circle), laser scattering (red square), and gravimetric analysis (green triangle) for a triblock copolymer hydrogel across a pH range. Leq and Lref denote the equilibrium and original lengths, respectively. (Adapted with permission from Ref. 11.  Royal Society of Chemistry, 2010.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials indicate the pH at which the volume transition occurs and the magnitude of swelling across the pH range; however, the error bars are considerably smaller for the scattering methods than the gravimetric approach.

2.1.3 Applications The full potential of pH-responsive actuators is yet to be exploited fully. It is envisaged that these devices will be used as flow-sorter valves, nanoscale actuators, micropumps, micromanipulators, sensors, artificial muscles, and self-propelling tails. Large steps have been taken toward the latter.12 A bipolymeric strip was fabricated where a polyacid plate was solvent-welded to a polybase plate to create an antagonistic muscle pair. The curvature of the strip depends on the pH of the surrounding environment. At low pH, contraction of the polyacid (and simultaneous expansion of the polybase) drives the strip to bend toward the polyacid. On the other hand, at neutral pH, the strip bends toward the contracted polybase, as shown in Figure 4. When exposed to a pH-oscillating environment, the tail waggles back and forth just like a flagellum. Future work in this area revolves around the attachment of such selfpropelling tails to a drug-loaded vesicle (or capsule) to create artificial spermatozoon-like devices. Devices based on the mass transport of solvent will be intrinsically slow on any scale other than the nanoscale because the actuation involves the diffusion of ions and solvent. The response time of chemically actuated systems scales with the square of the size of the object (because Polyacid triblock

Polybase triblock

Bipolymer strip

pH 4

pH 7

Figure 4 An artificial flagellum from bipolar polyelectrolyte gels working as an antagonistic muscle pair. (Adapted with permission from Ref. 12.  American Chemical Society, 2007.)

5

of diffusion) and a smaller object has a higher surface area for the ingress of reactants and solvent. Moreover, in cyclic processes, the accretion of salt can lead to deleterious effects. Attempts to get around these issues by the fabrication of nanofibres by electrospinning have only been marginally successful.13

2.2

Electrochemical actuation

Electrochemical actuation in polymers is the closest we have come to muscle mimicry; electrical impulses create a chemical reaction that stimulates conformational changes in a polymer to induce solvent and ion interchange, thus leading to macroscopic volume change. In this section, we take a detailed look at two of the major types of electrochemical polymer systems; conductive polymers (i.e., those with conjugated π-systems capable of charge delocalization) and pH-responsive polymers. The mechanisms for operation are very different in each case. In conjugated polymer systems, an electrical current is used to directly change the electronic structure of the polymers chains, which in turn causes an influx or expulsion of ions and solvent molecules. When an electric current is passed through a conductive polymer, electrons can be either ejected from (oxidation) or donated to (reduction) the polymer matrix, depending on the flow of the electricity and the polymer/electrolyte system used.14 The induced charges are stored along the polymeric chains and balancing counterions are forced to infiltrate the matrix to stabilize the system. Accordingly, solvent molecules are driven into the polymer network by the difference in osmotic pressure inside and outside of the gel. The influx of ions and solvent molecules causes an increase in volume. Figure 5 shows a schematic representation of the volume change induced when a polymer, such as polypyrrole, is subject to an electric charge. In this example, the polymer switches from neutral to cationic by the ejection of electrons to produce polarons (positively charged particles). The current and voltage typically required are particularly low (in the region of 30 mA and 1 V, respectively) and the response time can be fast (subsecond). For pH-responsive electrochemical systems, on the other hand, the electrical field indirectly influences the polymer conformation to induce a reversible volume transition. The electricity is used to electrolyze water, generating hydrogen ions at the anode and hydroxyl ions at the cathode. This produces a pH gradient across the solution, which can be used to influence the conformation of pHresponsive polymers. Strategic placement of the polymer in the solution can cause expansion, contraction, or even bending (when placed directly between the electrodes) of a polymer actuator. Switching the direction of the current creates the opposite effect. This second method of actuation

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

6

Soft matter

n

+

Oxidation

+ n e−

Reduction

Anions

Solvent molecules

Charged polymer (polarons)

Figure 5 Schematic representation of the reversible volume change in conjugated polymers, such as polypyrrole, induced by electrical current and the flow of ions and solvent molecules. (Modified from Ref. 14.  MDPI Publishing, 2010.)

is typically slower than the direct application of an electric field to conductive polymers.

Glass PTFE

Galvanostat WE(PPy film)

To generate molecular machines or artificial muscles, an anisotropic volume change is required. This can be done by fixing the conductive polymer to a passive film. Otero’s group pioneered such an approach for polypyrrole-based actuators by the fabrication of bilayer (and multilayer) devices, where a film of the responsive polymer is adhered to an inert flexible film. Electrochemically induced contraction of the polymer causes the actuator to bend because of the strain difference created across the bilayer; the polypyrrole contracts, while the flexible polymer film retains a constant volume. When the electric field is reversed, the polypyrrole expands and the device bends toward the inert film.15 As with the vast majority of conductive polymer films used in electrochemical actuation, electropolymerization was used to create the polypyrrole layer in this example. Typically, in electropolymerization, a low current is passed through a mixture of the monomer and an electrolyte salt, such as sodium para-toluene sulfonic acid or lithium perchlorate (LiClO4 ), dissolved in deionized water in a specifically designed vessel. The desired polymer film forms at the working electrode. Furthermore, anisotropic layers can be fabricated by growing the polymer in contact with a smooth surface, such as polytetrafluoroethylene (PTFE).16 The polymer face in contact with the electrolyte solution is considerably rougher than the opposing face that is forced up against the smooth cell wall. This creates the required anisotropy for a bending actuator without the need to resort to multilayer devices. The experimental setup of such an electrochemical polymerization is shown in Figure 6. Many electrochemically driven actuators are fabricated via supramolecular assembly. This is in contrast to the other sections in this chapter, where most of the actuator devices

PPy film

Electrolyte surface

Spacer

38 mm

2.2.1 Synthesis

CE(Ni plate)

16 mm

5 mm

0.3 mm

PTFE Glass

0.3 mm (spacer thickness) Thin slab vessel (a) (b)

Figure 6 (a) Schematic diagram of the electropolymerization to produce anisotropic films of polypyrrole and (b) the assembly of the vessel. (Reproduced from Ref. 16.  Wiley Periodicals, Inc, 2004.)

are composed of chemically crosslinked polymer networks. For example, copolymers have been allowed to microphase separate via solvent annealing (slow solvent evaporation)17, 18 or physically bound by freeze–thaw cycling (Section 3.1).19 Such systems include poly(vinyl alcohol)–poly(sodium acrylate) composites and polythiophenebased diblock copolymers. According to the H¨uckel model, a longer polymer chain has a lower ionization potential than a shorter one.20 Therefore, if there is a broad molecular weight distribution in the sample, there will be varying degrees of ionization and the average value will be somewhat difficult to control. However, if a controlled polymerization technique is employed to produce polymers with low polydispersity, devices can be tailor-made for specific applications. Designer polymers with predetermined electrical response can be fabricated by adjusting the target degree of polymerization during the synthetic process. This area remains relatively unchartered

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials

1

Current (mA)

territory but is believed to be the focus of much attention in the future. To increase the response rates of diffusion limited actuators, the aspect ratio can be increased to allow more surface area to be accessed. A common approach to achieve this is to produce responsive fibers of the polymer, often exploited for polyacrylonitrile (PAN)-based devices. To further improve their actuator behavior, the conductivity of the fibers can be increased by seeding with metallic particles, such as platinum, or by combination with other conductive fibers, such as graphite.21

Sweep rate: 15 mV s−1 PTSNa TMAPTS TBAPTS in 1.0 mol l aq−1

0

−1

2.2.2 Characterization

−1

One of the most important characteristics of a conductive polymer system is its response to electrical input. To ascertain this, cyclic voltammetry (CV) is used to produce a voltammogram. The potential of the working electrode (composed of the material under scrutiny) is ramped linearly with time, while the induced current between the working and reference electrodes is measured. Potential ramping, also referred to as the sweep rate, is kept constant throughout the cycle. A typical voltammogram for a polypyrrole electrode is shown in Figure 7. The peak and trough indicate reduction and oxidation of the electrode, respectively. A voltammogram is useful in quantifying the position and amplitude of the material response to electrochemical stimulus. The extent of ionization can be controlled by changing the analyte solution (as the volume transition is dictated by the transport of ions), which is quantified by CV. This can be clearly seen in Figure 7. Varying the analyte solution can have a measured effect on the performance of the device. An alternative characterization technique to CV is chronopotentiometry. In chronopotentiometry, a constant current is applied to the material and the potential is measured as a function of time. This can provide information

7

0 Potential (V vs Ag/Ag+)

Figure 7 Cyclic voltammograms of a polypyrrole actuator in three different analyte solutions: p-toluenesulfonic acid sodium salt (PTSNa), tetramethylammonium p-toluenesulfonate (TMAPTS), and tetrabutylammonium p-toluene sulfonate (TBAPTS). (Reproduced from Ref. 16.  Wiley Periodicals, Inc, 2004.)

about how the material responds while varying specific external parameters, such as temperature, analyte solution (and concentration), current, and applied load.14 A characterization tool often used for electrochemical actuation is optical imaging. This allows the displacement of a device to be mapped on changing various parameters. Figure 8 shows an anisotropic polypyrrole device during cyclic electrochemical stimulation. Optical imaging, whether in the form of a movie or as snapshots, is powerful in demonstrating the actuation characteristics to the reader. Moreover, the movement of the device is a function of the work done and the power output of the actuator. Onoda et al.16 attached a series of weights to an electrochemical actuator and monitored the maximum displacement using

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

Figure 8 Optical images used to record the actuator behavior of a polypyrrole device during cyclic voltammetric scanning. The electrode on the right-hand side (seen in images a to e) is the counter electrode. (Reproduced from Ref. 22.  Wiley-VCH, 2007.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

8

Soft matter

optical imaging, which in turn reveals the force produced during the contraction (the power stroke of an actuator).

2.2.3 Application The ultimate vision for electrochemically powered actuators is the artificial muscle. As previously mentioned, the mechanism of muscle operation follows the same fundamental steps: (i) electrical impulse inducing a chemical reaction, (ii) conformational changes in macromolecules caused by the chemical reaction, (iii) solvent and ion exchange between the macromolecular network and the surroundings leading to (iv) an anisotropic volume change. The target for synthetic muscles is based on the filaments in natural muscles that exhibit large one-dimensional deformations (approximately 30% strain) with a force of 0.3 MPa at a near-instantaneous rate (about 0.1 s).23 To date, this target has not been met in its entirety, but large steps have been made toward to it. For example, PAN-based fibers have been shown to have similar elasticity and triggered deformation to human tissue; however, the response times are still at least one order of magnitude slower.21 Polypyrrole

actuators, on the other hand, have been shown to exhibit near-instantaneous response rates, but the elasticity and deformation remain somewhat lacking.14 Besides artificial muscles, electrochemical actuators have been used to create swimming tails that sway at least 90◦ in both directions from their central resting position depending on the direction of the electrical field. The oscillating device could be used to propel micromachines through various analyte solutions or to push switches on command. Otero’s group created a triple layer system comprising conducting polymer–tape–conducting polymer. One of the conducting polymer layers acts as the anode and contracts, whereas the other active layer acts as the cathode and expands. The stress gradient about the central flexible tape causes the device to bend. On application of a constant current of 5 mA, the tail was seen to push obstacles up to approximately 10 g in weight. The optical images and the corresponding chronopotentiograms are given in Figure 9. The chronopotentiograms show that the weight of the obstacle placed in the tail’s trajectory directly influences the observed voltage profile. Moreover, these voltage profiles will change on variations in temperature, analyte species

1

4

2

5

3

6

(a) 3

j:14400 mg i:9600 mg h:8400 mg g:7200 mg f : 6000 mg e:4800 mg d:3600 mg c:2400 mg b:1200 mg a:no obstacle:

Potential (V)

2.5 2 1.5 1 0.5 0 − 0.5 (b)

0

10

20

30 40 Time (s)

50

60

70

Figure 9 Images of an electrochemically induced actuator tail with an obstacle placed within its trajectory (a) and the corresponding potentiograms when the weight of the obstacle was varied (b). The tail comes in contact with the obstacle after approximately 10 s in each case. (Reproduced from Ref. 24.  Wiley-VCH, 2003.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials (both type and concentration), and input current. As such, the actuator can be used as elegant sensors to subtle changes in their environment.

3

MAGNETIC ACTUATION

An inherent problem with the chemically responsive gels discussed so far in this chapter is that the actuator mechanism relies on the diffusion of species in and out of the polymer network. Consequently, actuator response times are kinetically restricted, with the rate of diffusion scaling with the square of the dimensions of the polymer construct. Magnetic elastomers and gels (the latter also known as ferrogels) on the other hand, display a near-instantaneous response as they are triggered by a penetrating magnetic field. Moreover, actuation is triggered remotely (externally), increasing their potential for in vivo applications. A ferrogel comprises an elastic polymer scaffold swollen by a colloidal dispersion of magnetic particles. The most commonly used magnetic particles in ferrogel actuators are magnetite (Fe3 O4 ) nanoparticles. Nanoparticles (with typical diameter in these devices of approximately 10 nm) have distinct advantages over microparticles. The nanoparticles are held in place by strong, yet noncovalent, adhesive forces between themselves and the polymer chains. Such forces prevent the magnetic particles from rotating or flowing from their relative position in the matrix. Owing to the presence of a vast number of surface atoms, a common feature in nanomaterials, their magnetic moment can fluctuate in the absence of a magnetic field so that the material has no net magnetization. This effect is known as superparamagnetic relaxation and is unique to nanosized magnetic particles. When a magnetic field is applied, the nanoparticles become magnetized and align with the field. The nanoparticles are attracted to one another, but are held apart by the polymer matrix. If the magnetic field applied is nonuniform, then the ferrofluid will be attracted to the areas of higher magnetic field strength and the surrounding polymer matrix has the elasticity to move with the fluid. In a magnetic field gradient, the ferrogel will swing toward the high end of the field, thus performing actuation. In contrast to the other devices discussed in this chapter, ferrogels do not change their volume during actuation; instead their shape is distorted depending on the geometry of the gel and the nature of the applied magnetic field. When the magnetic field is switched off, the nanoparticles quickly lose their dipole moments and the material again acquires no overall net magnetization. For a more in-depth discussion on the mechanism behind this process, the reader is directed elsewhere.25–27 When subject to a magnetic field, superparamagnetic particles can also generate heat due to a process known

9

as N´eel relaxation.27, 28 N´eel relaxation involves internal rotation of the magnetic moment within individual particles to reorganize the magnetic vector. During the process, energy is dissipated as heat. This phenomenon can be used to trigger the coil–globule transition in thermoresponsive polymers, allowing for a more direct route of performing temperature-sensitive actuation. Superparamagnetic nanoparticles require a weaker magnetic field than ferromagnetic particles to generate the same heat.29 Moreover, nanodomains have less effect than microdomains on the mechanical properties of the ferrogel,30 further illustrating the necessity for morphology control in such devices.

3.1

Synthesis

To create a ferrogel, one must swell a crosslinked polymer network such as poly(vinyl alcohol) or poly(hydroxyethyl methacrylate) with a ferrofluid. This can be done before, during, or after the crosslinking step in chemically bound gels, but is undertaken before self-assembly in physically bound networks. The trick is to ensure a fine distribution of magnetic particles throughout the polymer matrix, avoiding aggregation where possible. This is generally achieved by sonication of appropriate proportions of ferrofluid and polymer. There are a number of strategies for synthesizing the magnetic nanoparticles. For example, magnetite nanoparticles can be synthesized by reacting FeCl2 with FeCl3 in aqueous media with the addition of perchloric acid (HClO4 ) to induce peptization (stabilization of the colloidal dispersion). This robust synthetic route can be carried out in the presence of molecular polymer chains or a polymer scaffold,31 where in the latter case, the nanoparticles are formed within the hydrophilic voids of the crosslinked network. Commonly, the flexible polymer construct of a magnetic elastomer or ferrogel is a chemically crosslinked network, but there are also a number of examples where supramolecular physical gels (using H-bridges or microcrystals) have been exploited. In the case of ferrogels, physical networks have been reported to be softer and more elastic than chemically crosslinked gels of the same constituents.32 This can be attributed to the freedom offered by physical interactions over chemically rigid junction points. A particularly simple and elegant method of forming supramolecular magnetic constructs is the “freeze–thaw” method.33 Simple linear polymer chains are dissolved in water and the temperature is subsequently decreased. As ice crystals begin to form, the polymer chains get excluded from the ice crystals and the polymer becomes concentrated in the amorphous regions of the sample (as shown in Figure 10). Hydrogen bonds between polymer chains form within these concentrated aggregates, limiting thermal motion around average fixed positions. When thawed,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

10

Soft matter

Void Ice

Void

Ice Polymer chains in solution

Figure 10

Freezing

The freeze–thaw process leading to supramolecular constructs.

the aggregates remain intact due to the strong physical crosslinking points created by the hydrogen bonding. Subsequent freeze–thaw cycles force more polymer chains to participate in the supramolecular activity, creating more robust gels on each repeated cycle. In ferrogels, there is a fine balance between magnetization and elasticity; the more particles present, the higher the magnetization. However, as the content of the nanoparticles increases, the elasticity of the gel decreases, that is, the material becomes more brittle. To overcome this, researchers have investigated the use of a magnetic monomer. Fuhrer et al.34 have functionalized cobalt/carbon (Co/C) nanoparticles with a vinyl group and copolymerized them with hydroxyethyl methacrylate (HEMA) and ethylene glycol dimethacrylate (EGDMA). A very strong and flexible magneto-responsive hydrogel was afforded. It was found that higher magnetic loading could be achieved by this strategy. Additionally, due to covalent incorporation, nanoparticles with higher saturation magnetization, such as cobalt or iron, can be used. When dispersed, such nanosized particles have low stability due to their high surface atom content and are therefore easily oxidized by air or solvents. As a consequence, colloidal work (as discussed above) must involve less sensitive particles, such as magnetite. The approach of covalent incorporation of the magnetic component allows for high loading of more effective particles. When coupled with the advantage of supramolecular chemistry, this combination offers a promising future option in magnetically stimulated actuation that is yet to be explored.

3.2

Thawing

Characterization

The characterization of magnetic elastomers/ferrogels broadly falls into three main parts: (i) analysis of the magnetic nanoparticles/ferrofluid, (ii) ascertaining the mechanical and morphological properties of the polymer construct, and (iii) assessing the actuator response. Two of the most effective techniques for characterizing the ferrofluid composition are TEM and X-ray diffraction (XRD). Both techniques provide a size distribution of the nanoparticles, with TEM being more of a qualitative tool compared to XRD. TEM gives a two-dimensional projection of a

three-dimensional structure and for thin enough samples nanoscale features can be observed directly. Image analysis techniques can then be used to extract quantitative information such as particle size, distribution, and separation. For XRD, on the other hand, inversion of the scattering pattern by Fourier transformation results in an average size distribution for the sample. This method is much more accurate than microscopy, although perhaps not quite as aesthetically pleasing. Figure 11 shows a TEM image and XRD patterns for a maghemite (γ -Fe2 O3 ) ferrofluid. XRD data show peaks relating to interparticle distances, which can be extrapolated to give the characteristics of the ferrofluid, such as the particle diameter. The scattering vector (the x-axis in Figure 11) is a function of the angle of the diffracted X-rays. The polymer construct is also often imaged using electron microscopy (both SEM and TEM) to reveal the gel morphology at the nanoscale. More specifically for supramolecular networks, SEM can be used to show the effect of the freeze–thaw cycling process on the structure of the gel (Figure 12). The nanoscale morphology of the polymer dictates the mechanical properties of the gel. Hatakeyema et al.33 have shown that the scaffold of porous PVA hydrogels becomes thicker on increasing the number of repeated freeze–thaw cycles, which increases the structural integrity of the gels. The elasticity of the gel can be quantified by stress–strain measurements. A plot of stress versus strain provides detailed information about the mechanical properties of the material, such as its yield and break points. Elasticity of a material is described by the Young’s modulus (denoted by E) and is determined from the value of the slope at low strain values, where the stress–strain profile is in a linear regime (Figure 13). The lower the value of the Young’s modulus, the higher the elasticity; smaller stresses are required to stretch the material further. Figure 13 shows the dependence of magnetic particle loading on the stress–strain profile of a chemically crosslinked ferrogel. As the concentration of filler particle increases, the modulus increases, that is, the material becomes more brittle (as discussed in Section 3.1). The magnetic properties of the final elastomer/ferrogel, that is, the actuator response, can be demonstrated both

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials

11

10 nm

1000

311

Intensity (a.u.)

800

220

400

600

440 511

400

200

0 1

2

3

4

5

6

Wavevector (Å–1)

Figure 11 TEM image of maghemite nanoparticles deposited from a dispersion in water and the corresponding XRD pattern (red trace). The blue, sharper XRD pattern is obtained from the same nanoparticles dispersed in hexane, which produced a more ordered array of maghemite. (Reproduced with permission from Ref. 35.  American Chemical Society, 2003.)

10 µm

10 µm (a)

(b)

Figure 12 Scanning electron micrographs of PVA hydrogels after two (a) and five (b) freeze–thaw cycles. (Reproduced with permission from Ref. 33.  Elsevier, 2005.)

quantitatively (magnetization curves) and qualitatively (using image capture/optical microscopy). Magnetization curves show the magnetization of the gel plotted against the applied magnetic field strength. A typical magnetization profile for superparamagnetic species can be fitted to a Langevin function,25 where there is a sharp increase in magnetization on application of small fields, followed

by a reduction to a plateau (Figure 14). This shows that actuators driven by superparamagnetic behavior will exhibit pronounced shape deformation on the application of extremely small magnetic fields. Most examples of magnetic actuators in the literature express the magnetic response using image capture or movies provided as supplementary data. These methods

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

12

Soft matter

3.3

10 wt% 20 wt% 30 wt%

4.5

sn [kP a]

3.0

1.5

0.0 0.0

0.1

0.2 D

0.3

0.4

Figure 13 Neo-Hookean representation to show the mechanical properties of a magnetic elastomer at different magnetic particle concentrations, where σ n is the applied stress and D is the resulting strain. (Reproduced with permission from Ref. 27.  Springer, 2007.)

12

Application

Currently, the magnetic actuators discussed in the literature comprise of chemically crosslinked polymers and not supramolecular constructs. Physically crosslinked magnetically triggered polymers are generally used in the field of drug delivery or hyperthermia. In these cases, the magnetic response is manifested as an increase in temperature. The generated heat can then be used to denature unwanted cells (hyperthermia) or to trigger the coil–globule transition in thermo-responsive polymers to induce drug release. As actuators, magnetic elastomers appear to be a promising option for the future, as they exhibit a smooth, instantaneous shape change much like a mammalian muscle. Moreover, a remote magnetic field can penetrate deep within the body without significant absorption by body tissue, producing a harmless external trigger for in vivo applications. Such flexible machines encompass potential uses in medical implants, heart pump components, and micropumps for drug delivery.

[Gs]

10

4

8 6 4 2 0 −2 −4 −6 −8 −10 [Oe]

−12 −10000

−5000

0

5000

10000

Figure 14 Typical magnetization curve for a superparamagnetic material. The magnetization (y-axis) is expressed in Gauss and the field intensity (x-axis) in Oersted. (Reproduced with permission from Ref. 27.  Springer, 2007.)

of analysis, although not always quantitative, provide the most appealing demonstration of actuation. Visualization of a material’s function is often more striking than data that requires interpretation. Figure 15 shows a cylindrical ferrogel changing shape on application of a nonuniform magnetic field. The actuator instantly bends toward higher field strengths within the field, with the elasticity of the polymer network accommodating the deformation. Such successive bending and stretching cycles can be repeated many times without deterioration of the ferrogel, making them ideal candidates for soft actuators.

LIGHT ACTUATION

Light, much like a magnetic field, can be used remotely to produce a near-instantaneous mechanical response. This is known as photomechanical actuation. Light is an energy source, which can be controlled with great precision; it can be polarized, highly localized in space and time, and the wavelength can be selected with great accuracy. The earliest report of transducing photonic energy into mechanical work was back in 1967, where Lovrien used azobenzene as the active chromophore.37 Since then, a number of different photochromic groups have been investigated, with examples including acetylacetone,38 fulgides,39 stilbenes,40 reversible crosslinking moieties,41 amongst others.42 Undoubtedly, the most studied chromophore of all is azobenzene43, 44 and is the focus of the work described herein. Azobenzene is a molecule based around an isomerizable azo-linkage (–N=N–), as shown in Figure 16. The rod-like trans form is the most stable conformation. However, the molecule can be converted to the metastable cis isomer using ultraviolet light with a wavelength in the region of 365 nm. This process is known as trans–cis photoisomerization, denoting the changing of isomers from trans to cis. During the transformation, UV irradiation drives a large volume change as the molecule shifts from ˚ being rigid and straight (with a molecular length of 9 A) ˚ to being bent, with a length of 5.5 A. However, this is a reversible process; visible light irradiation (λ = 465 nm) can be used to trigger the cis–trans isomerization. Additionally, as the cis configuration is only metastable, the isomerization in this direction can also occur via thermal

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials

13

Figure 15 Actuator behavior of a cylindrical ferrogel. A nonuniform magnetic field has been applied perpendicular to the axis of the cylinder using an electromagnet. (Reproduced with permission from Ref. 36.  Elsevier, 1998.) At the molecular level

R

5.5

Å

R

9.0 Å

N

(a)

R′

hn

N

hn′, kBT

N

N

R′ Within a liquid-crystalline matrix

hn hn′, kBT

(b)

Nematic

Isotropic

Figure 16 The photo-triggered isomerization of azobenzene at the molecular level (a) and within a liquid crystalline array (b). The trans-form is a rigid rod-like structure (left), whereas the cis conformation (right) is bent. (Adapted with permission from Ref. 44.  Royal Society of Chemistry, 2007 and Ref. 45.  American Physical Society, 2002.)

relaxation; heating the sample will return the molecule to the trans configuration. Reversible switching between isomeric forms will create oscillations in volume that can be channeled effectively for mechanical output. The amplitude and sensitivity of the volume transition can be increased by incorporating the azobenzene moiety into a liquid crystalline material. When the azobenzene

component is in its trans state, the material will have a degree of unidirectional order, known as a nematic phase. On switching to the cis state, the material loses its order and the system becomes isotropic. In short, a liquid crystal polymer network that has enough azobenzene character will exhibit a reversible change in volume driven by the nematic–isotropic phase transition, triggered by light irradiation. This enhanced volume change can then be used to power photomechanical devices, particularly artificial muscles where a uniaxial contraction is needed. Moreover, liquid crystal networks containing azobenzene not only have wavelength sensitivity but also display polarization sensitivity, in that the directional contraction is dependent on the polarization direction.46 In addition to UV and visible light, laser light (around 488 nm) can also be used to trigger conformational changes in azobenzene-containing molecules.47 Furthermore, if a laser light source is used that covers a larger wavelength band (such as 457–514 nm), both the trans–cis and cis–trans isomerizations can be simultaneously activated throughout the sample. This leads to a phenomenon known as trans–cis–trans reorientation,48 where the actuator will bend with a swaying motion; away from and toward the light source. A number of studies probing the underlying physics behind the light-induced volume transition of supramolecular azobenzene-containing polymers have been carried out in an attempt to understand the molecular orientation of macromolecules far below their glass transition temperatures. These theories are beyond the scope of this chapter and for an example of such work, the reader is directed elsewhere.49

4.1

Synthesis

There are a number of ways to incorporate an azobenzene moiety into a polymer network. The chromophore

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

14

Soft matter

O

O

N

n

O

N

n

O N N

O O O

O

nO

N N

Photoactive monomers O

N N

O n

O

Photoactive crosslinkers

Photoactive dopant

O

O

N

m

H 2N

O n

O N

O

N

NO2

m

N

Figure 17 Examples of photoresponsive monomers, crosslinkers, and dopants commonly used as the active component in azobenzene-based actuators.

can be located within the monomer (pendant azobenzene units), within the crosslinks or added as a dopant (dispersed molecules/particles noncovalently bound to the polymer construct). Examples of each of these are given in Figure 17. In chemically crosslinked networks, it is extremely common to have azobenzene in both the monomer repeat unit and the crosslinker. Azobenzene monomers are typically polymerized by free radical polymerization, either photolytically or thermally initiated under appropriate conditions. For liquid crystal networks, the polymerization is often performed between polymer-coated, glass slides, commonly using a polyimide as the coating. The coatings are rubbed unidirectionally, which induces the polymer chains to grow in an aligned fashion.50 Another method of producing liquid crystal networks is to adopt a two-stage crosslinking strategy, whereby the second crosslinking step is performed under induced uniaxial extension to chemically trap the required alignment.51 To create a physically bound photoresponsive actuator, Kim et al.52 cast a polymer blend film and employed solvent annealing to allow the self-assembly of a photoactive polymer within a nonresponsive polymer matrix. The group used cationic polymerization to synthesize the azobenzenecontaining polymer before mixing with a commercially available polycarbonate solution, as shown in Figure 18. It is likely that the self-assembly process was driven by π –π stacking (such as that discussed in Multicomponent SelfAssembled Polymers Based on π -Conjugated Systems, Soft Matter) between the azobenzene units of the photoresponsive polymer and the phenyl rings of the polycarbonate (in addition to the azobenzene molecules overlapping with other azobenzene molecules), although no circular dichroism data were shown to confirm this. Upon UV irradiation,

the extent of overlapping is significantly reduced as the azobenzene units are bent away from one another. During this transformation, the modulus of the film decreases because the polymer chains acquire more mobility, and the film expands under tension. When the light source is switched off, the cis–trans isomerization is triggered, the modulus of the material increases and the film contracts to perform work against the tensile force. Other methods of creating physically bound systems include spin-coating block copolymers containing azobenzene repeat units in one of the blocks. The spin-coated layers are then annealed at elevated temperatures to provide sufficient chain mobility and drive subsequent microphase separation. This synthetic strategy allows devices to be fabricated, which exhibit truly one-dimensional expansion and contraction cycles. Moreover, following use, the films can be dissolved up and recast, if necessary, to produce brand new devices from the very same material.

4.2

Characterization

The characterization methods applied to light-powered actuators encompasses a large array of techniques, most of which are common to viscoelastic materials, such as stress–strain curves, differential scanning calorimetry (DSC), and so on. For annealed films, the thickness with respect to the input information (light source) is measured by ellipsometry or neutron reflectivity, both of which are sensitive analytical tools for film thickness. The major analytical technique that differs from those already discussed in this chapter is UV–vis spectroscopy as it can be used to monitor the extent of photoisomerization. The different isomers give rise to different spectral absorption bands, which can be followed during the photo-illumination process. The most common ways to map the behavior of photoresponsive actuators generally involve optical imaging. Tracking software can then be used to monitor the expansion/contraction, tip displacement, and shape. These data are then plotted as a function of exposure time, intensity, or polarization (for liquid crystalline materials) of the light source. Such plots provide detailed information about the control over the output signal on varying the input parameters, essential for widespread application of these devices. Ikeda’s group evaluated the driving force of a photoresponsive belt by applying an external tensile strain and measuring the internal stress on UV irradiation of varying intensities.53 The sample length was kept constant throughout so that the required change in load was directly related to the mechanical force induced by photoisomerization. Their results, shown in Figure 19, clearly demonstrate how the intensity of the UV-light plays a vital role in dictating the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials

15

O

n O

O CH2

CH2

4

O

O

O

O

O

4

Polycarbonate

O

Cationic Polymerization

N

Casting from CH2Cl2

N N

N UV-active film

(a) 5 mN 5 mN UV-on

Irradiation

UV-off (b)

5 mN

5 mN

UV-on

N

N N N

N

N

N

N

N N N

N

UV-off

N N

N N

Strongly aggregated (High modulus)

(c)

Weakly aggregated (Low modulus)

Figure 18 The synthetic strategy employed to create a physically bound photoresponsive actuator (a), the UV-induced macroscopic behavior (b), and the proposed mechanism on UV-light irradiation (c). (Adapted with permission from Ref. 52.  Elsevier, 2005.)

External force

200

UV

120 mW CM−2 60 mW CM−2

100

20 mW CM−2 50 0

(a)

240 mW CM−2

150

F / mN

Sample film

UV-on

(b)

0

1

2 t /min

3

4

5

Figure 19 (a) Experimental setup employed by Ikeda’s group to evaluate the driving force of a photoresponsive belt and (b) the corresponding results when different light intensities were employed. (Reproduced from Ref. 53.  Wiley-VCH, 2008.)

mechanical response of the device. Unsurprisingly, higher light intensities produce larger forces at faster rates. However, these results also show that the device output can be tuned to suit a specific application by judicious choice of the power of the light source. Further control of the actuator response can be obtained by modifying the physical dimensions of the device. Serak et al.46 elegantly demonstrated this with a photodriven cantilever. They monitored the tip displacement using optical imaging and systematically changed the length and thickness of the device in addition to observing the effect of the intensity of the light source. The cantilever was irradiated as shown in Figure 20(a). Only one face of the cantilever was irradiated at any given time. Owing to the high molar absorption coefficient of azobenzene at 360 nm,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

16

Soft matter Start position

Azo-LCN

Light beam

j

Oscillation angle (a)

Holder

Normalized intencity

300

Frequency (Hz)

250 200 150

1.0

the trans-form. As only surface contractile force is observed in the presence of the light source, the cantilever bends around this surface and exposes the other, expanded face to the light source. Consequently, trans–cis isomerization occurs to the surface of the newly exposed face, while the now-hidden side relaxes back into the stable trans conformation. The cantilever subsequently contracts on the exposed surface to bend the material back in the opposite direction. This oscillation can be tuned to be incredibly fast, as shown in the inset of Figure 20(b), where the frequency in this example reached a staggering 270 Hz. Here, the frequency was controlled by both adjusting the length and thickness of the sample, with shorter, thicker actuators exhibiting the faster oscillations. On a final note, it has been shown that increasing the power of the light source increases the amplitude of the oscillations but not the frequency.

0.5

4.3

Application

0.0 0

5

10

15

20

25

30

Azobenzene-based photodriven actuators have shown great potential as candidates for synthetic muscles, microfluidic devices, and remotely accessed optical elements. As

Time (ms)

100 50 0 1

2

3 4 Length (mm)

(b)

5

6

Vis

Belt Alignment direction

UV

40

Axle Frequency (Hz)

30

(a)

Axle

Pulleys

White spot as a marker

20

10

0s

16 s

8s

24 s

0 0 (c)

10

20

30 40 50 Thickness, t (µ)

60

70

Figure 20 (a) Experimental setup of a photodriven oscillator and the corresponding frequencies obtained on changing the (b) length and (c) thickness of the polymer tail. (Adapted with permission from Ref. 46.  Royal Society of Chemistry, 2010.)

almost all of the photons are absorbed within the first micron of material. This leads to the formation of a bilayertype structure where only the surface is transformed into the contracted cis-form and the bulk of the cantilever remains in

(b)

Figure 21 A photodriven plastic motor system developed by Ikeda’s group. (a) Schematic diagram of the setup and (b) photographs taken periodically during 24 s of light irradiation. (Reproduced with permission from Ref. 53.  Wiley-VCH, 2008.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Stimuli-responsive and motile supramolecular soft materials with magnetic actuators discussed earlier in this chapter, the robust examples across the literature mainly feature chemically crosslinked polymer networks. The few physically bound, supramolecular actuators developed have been shown to be the most suitable as optical sensors; with a change in light intensity or wavelength giving rise to a change in the thickness of a polymer layer. On the other hand, chemically crosslinked examples currently have a much broader array of applications including photodriven plastic motors53 and UV-switches used to control the intensity of a laser beam.54 The very first example of a plastic motor consisted of a light-sensitive belt, which was strategically irradiated with light sources of different wavelengths to drive a pulley system as shown in Figure 21. Both trans–cis and cis–trans isomerizations were induced at different sections of the belt to create an expansion/contraction wave in the material, causing rotation on the pulleys.

REFERENCES 1. R. A. L. Jones, Soft Machines: Nanotechnology and Life, Oxford University Press, Oxford, 2004. 2. R. Yoshida, T. Takahashi, T. Yamaguchi, and H. Ichijo, J. Am. Chem. Soc., 1996, 118, 5134. 3. Y. Hara, S. Maeda, S. Hashimoto, and R. Yoshida, Int. J. Mol. Sci., 2010, 11, 704. 4. R. Yoshida, Sensors, 2010, 10, 1810. 5. A. K. Khandpur, S. Foerster, F. S. Bates, et al., Macromolecules, 1995, 28, 8796. 6. A. Nykanen, M. Nuopponen, A. Laukkanen, et al., Macromolecules, 2007, 40, 5827. 7. M. W. Matsen and F. S. Bates, Macromolecules, 1996, 29, 1091. 8. A. Ryan, C. Crook, J. Howse, et al., J. Macromol. Sci. B Phys., 2005, 44, 1103. 9. P. D. Topham, J. R. Howse, O. O. Mykhaylyk, Macromolecules, 2006, 39, 5573.

5

CONCLUSIONS

In this chapter, we have introduced the burgeoning field of supramolecular smart actuators. The principle actuation mechanisms and energy sources (chemical, electrical, optical, and magnetic) have been explored as external triggers to drive isothermal volume changes in responsive polymers. The area is diverse and there is much potential for devices fabricated from artificial muscles to light-driven switches. However, much work remains to be done to realize current ideologies and transform the dreams of researchers into technological reality. The most recent robotics contest found the most powerful synthetic muscles overcome by a schoolgirl in an arm-wrestling match. Clearly, there is a great need to improve the mechanical properties of supramolecular polymer constructs, so that they can compete with chemically crosslinked materials and deliver the promise of greater uniformity in mechanical response. In designing materials, there is a balance required between responsiveness (accentuated by softness) and generating force (accentuated by stiffness). The advantages of supramolecular chemistry, such as processability and control of microstructure, can be used to construct composite materials at the nanoscale such that responsiveness and force generation can be optimized. For example, a muscle, used to operate a lever, need only to respond in one dimension and a three-dimensional response wastes two-thirds of the motion. On the other hand, a gel used to make a microfluidic valve may indeed need to operate in three dimensions. The very nature of supramacromolecular assembly shows the way to construct bespoke materials for specific application and the construction of a toolbox of suitable components that can be used in a modular manner should be our goal.

17

et al.,

10. K. B. Guice, S. R. Marrou, S. R. Gondi, et al., Macromolecules, 2008, 41, 4390. 11. J. M. G. Swann, W. Bras, J. R. Howse, et al., Soft Matter, 2010, 6, 743. 12. P. D. Topham, J. R. Howse, C. J. Crook, et al., Macromolecules, 2007, 40, 4393. 13. L. Wang, P. D. Topham, O. O. Mykhaylyk, et al., Adv. Mater., 2007, 19, 3544. 14. L. V. Conzuelo, J. Arias-Pardilla, J. V. Cauich-Rodriguez, et al., Sensors, 2010, 10, 2638. 15. T. F. Otero and J. M. Sansinena, Adv. Mater., 1998, 10, 491. 16. M. Onoda, Y. Kato, H. Shonaka, and K. Tada, Electr. Eng. Jpn., 2004, 149, 7. 17. Y. Wu, A. M. Ballantyne, P. Wagner, et al., Electrochim. Acta, 2007, 53, 1830. 18. H. Okuzaki, H. Suzuki, and T. Ito, Synth. Met., 2009, 159, 2233. 19. T. Shiga, Y. Hirose, A. Okada, and T. Kurauchi, J. Mater. Sci., 1994, 29, 5715. 20. T. F. Otero, J. Mater. Chem., 2009, 19, 681. 21. H. B. Schreyer, N. Gebhart, K. J. Kim, and M. Shahinpoor, Biomacromolecules, 2000, 1, 642. 22. X. M. He, C. Li, F. G. Chen, and G. Q. Shi, Adv. Funct. Mater., 2007, 17, 2911. 23. R. M. Alexander, Nature, 1992, 357, 360. 24. T. F. Otero and M. T. Cortes, Adv. Mater., 2003, 15, 279. 25. D. Szabo, I. Czako-Nagy, M. Zrinyi, and A. Vertes, J. Colloid Interface Sci., 2000, 221, 166. 26. M. Zrinyi, L. Barsi, and A. Buki, J. Chem. Phys., 1996, 104, 8750. 27. G. Filipcsei, I. Csetneki, A. Szilagyi, and M. Zrinyi, Oligomers Polym. Compos. Mol. Imprint., 2007, 206, 137. 28. R. Hergt, S. Dutz, R. Muller, and M. Zeisberger, J. Phys. Condens. Matter, 2006, 18, S2919.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

18

Soft matter

29. J. K. Yang, J. H. Yu, J. Kim, and Y. H. Choa, Mater. Sci. Eng. A Struct. Mater., 2007, 449, 477.

42. S. Yagai and A. Kitamura, Chem. Soc. Rev., 2008, 37, 1520.

30. V. M. De Paoli, S. H. D. Lacerda, L. Spinu, et al., Langmuir, 2006, 22, 5894.

44. C. J. Barrett, J. I. Mamiya, K. G. Yager, and T. Ikeda, Soft Matter, 2007, 3, 1249.

31. T. Caykara, D. Yoruk, and S. Demirci, J. Appl. Polym. Sci., 2009, 112, 800.

45. P. M. Hogan, A. R. Tajbakhsh, and E. M. Terentjev, Phys. Rev. E Stat. Nonlin. Soft Matter Phys., 2002, 65, 041720.

32. T. Y. Liu, S. H. Hu, K. H. Liu, et al., J. Control. Release, 2008, 126, 228.

46. S. Serak, N. Tabiryan, R. Vergara, et al., Soft Matter, 2010, 6, 779.

33. T. Hatakeyema, J. Uno, C. Yamada, et al., Thermochim. Acta, 2005, 431, 144.

47. Y. B. Li, Y. N. He, X. L. Tong, and X. G. Wang, Langmuir, 2006, 22, 2288.

34. R. Fuhrer, E. K. Athanassiou, N. A. Luechinger, W. J. Stark, Small , 2009, 5, 383.

and

48. U. Hrozhyk, S. Serak, N. Tabiryan, et al., Opt. Express, 2009, 17, 716.

35. L. E. Euliss, S. G. Grancharov, S. O’Brien, et al., Nano Lett., 2003, 3, 1489.

49. N. Mechau, M. Saphiannikova, and D. Neher, Macromolecules, 2005, 38, 3894.

36. M. Zrinyi, D. Szabo, and H. G. Kilian, Polym. Gels Netw., 1998, 6, 441.

50. M. H. Li, P. Keller, B. Li, et al., Adv. Mater., 2003, 15, 569.

43. D. Tabak and H. Morawetz, Macromolecules, 1970, 3, 403.

37. R. Lovrien, Proc. Natl. Acad. Sci. U.S.A., 1967, 57, 236.

51. J. Cviklinski, A. R. Tajbakhsh, and E. M. Terentjev, Eur. Phys. J. E , 2002, 9, 427.

38. T. Watanabe, M. Akiyama, K. Totani, et al., Adv. Funct. Mater., 2002, 12, 611.

52. H. K. Kim, X. S. Wang, Y. Fujita, et al., Polymer, 2005, 46, 5879.

39. T. Kardinahl and H. Franke, Appl. Phys. A Mater. Sci. Process., 1995, 61, 23.

53. M. Yamada, M. Kondo, J. I. Mamiya, et al., Angew. Chem. Int. Ed., 2008, 47, 4986.

40. F. E. Mikes, P. Strop, and J. Kalal, Chem. Ind., 1973, 24, 1164.

54. C. S. Li, C. W. Lo, D. F. Zhu, et al., Macromol. Rapid Commun., 2009, 30, 1928.

41. A. Lendlein, H. Y. Jiang, O. Junger, and R. Langer, Nature, 2005, 434, 879.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc141

Self-Assembling Fibrillar Networks—Supramolecular Gels David K. Smith University of York, York, UK

1 Introduction to Low-Molecular-Weight Gelators (LMWGs) 2 Characterization of Gels 3 Self-Assembly Rules in Gels 4 Molecular Structures of Gelators 5 Unique Features of Gel-Phase Materials and Their High-Tech Applications 6 Conclusions References

1

1 3 5 7 10 19 20

INTRODUCTION TO LOW-MOLECULAR-WEIGHT GELATORS (LMWGs)

Gels are easily recognized soft materials with a wide range of different applications—for example, in cosmetics, pharmaceutical preparations, as greases/lubricants, and in the food industry. Gel-phase materials are colloidal systems in which two different phases coexist; a liquidlike phase contains a sample spanning solidlike network, the presence of which prevents the bulk flow of the material.1 Gels have been well-known materials for many years, but recently, there has been an increasing focus on understanding the behavior of these materials.2 This knowledge can then be used to design carefully tailored gel-phase materials

with high-tech applications, in areas as diverse as tissue engineering and nanoscale electronics.3 The scientific origins of gel-phase materials can be found in observations in the latter part of the nineteenth and early twentieth centuries that certain natural products were capable of making liquid samples more viscous and ultimately converting them into gels. For example, 12-hydroxystearic acid (1), isolated from castor oil, particularly when converted into its lithium salt was shown to immobilize organic fluids, such as petrol, paraffin, and kerosene (Figure 1).5 By the 1950s, organogels (gels formed in organic solvents) based on metal salts of 12-hydroxystearic acid were widely used in grease formulation for the engine lubrication industry and now constitute >50% of the grease market.6 It was also realized more than 100 years ago that protected versions of sorbitol could also form gel networks in organic media.7 Unprotected sorbitol acts as a thickener in water, helping to form hydrogels (gels in water).8 Such materials typically have applications in cosmetic, pharmaceutical, and food products. Remarkably, over 50 years ago, electron microscopy was being used to probe the nanostructure of organogels, and it became understood that the molecular building blocks assembled into easily visualized fibrillar “solidlike” networks, which spanned the “liquidlike” phase (Figure 1).4, 9 As such, the study of gels constitutes a very early example of the “bottom-up” self-assembly approach to nanotechnology. With the rise of polymer chemistry in the 1920s and the increasing reliance of global development on petroleum products, the investigation of gels based on polymers rapidly advanced.10 Polymers are obvious candidates for gelation, as the polymer chains themselves constitute an extended system, and if there is appropriate crosslinking between the polymers (either covalent or noncovalent),

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

2

Soft matter

O OH OH

1

a

1µ

D-form

b

1µ L-form

Figure 1 Chemical structure of 12-hydroxystearic acid and original TEM images from 1965 of lithium 12-hydroxystearate. The D-form exhibits a right-handed twist, while the L-form forms left-handed twisted fibrils. (Adapted with permission from Ref. 4.  American Chemical Society, 1965.)

(a)

Prenucleus

Nucleating aggregate

h tape

Helical growth

h fibril

hribbon

etape

efibril

efiber

eribbon Rod-like monomer

etrans

b2 (b)

Monomer

a b1 Tape

Ribbon

Fibril

Fiber

Figure 2 Hierarchical assembly of gel-phase materials as illustrated in (a) Ref. 12 (Reproduced with permission from Ref. 12.  American Association for the Advancement of Science, 2006.) and (b) Ref. 13. (Reproduced with permission from Ref. 13, A. Aggeli, I. A. Nyrkova, M. Bell, R. Harding, L. Carrick, T. C. B. McLeish, A. N. Semenov and N. Boden, Proc. Natl. Acad. Sci. USA 2001, 98, 11857–11862. Copyright (2001) National Academy of Sciences, U.S.A.)

then a sample-spanning “solidlike” network, which is capable of preventing the bulk flow of the material, is readily established. As such, polymers have been widely exploited in the development of soft materials with gel-like rheological (materials) properties.11 In recent years, however, attention has begun to refocus on small molecules, such as hydroxystearic acid and sorbitol derivatives, which form gels.2 The insights provided by early pioneers of supramolecular chemistry made clear that precisely controlled noncovalent interactions between the molecular-scale building blocks can direct the selfassembly of these molecules into extended nanoscale structures and that these structures can then constitute the sample-spanning “solidlike” network that provides the soft material with its cohesion. In most gels, the molecular

building blocks assemble into “one-dimensional” fibrils. In many cases, the fibrils then aggregate to form fibers, and these fibers often interact with one another to form bundles (Figure 2).13 The fibers or bundles then tangle and interact with one another, hence forming an extended nanoscale “solidlike” network. As such, the self-assembly of gel-phase materials from small molecules is a multistep process—often referred to as hierarchical. Small molecules that form gels are referred to as lowmolecular-weight gelators (LMWGs) and are now the focus of much attention because, as described above, this approach is one of the key strategies by which nanostructures can be assembled from the “bottom-up.”14 Typically, only small amounts of the gelator are required to immobilize surprisingly large amounts of solvent—usually circa

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels 1% wt/vol and, on some occasions, as little as 0.1% wt/vol of gelator. This means that considerable nanoscale organization is generated by very small amounts of gelator and, furthermore, ensures that these materials are highly porous. The multistep self-assembly process means that the formation of soft gels is a reversible process, and, as such, these materials are highly responsive. Furthermore, because the structures of molecular-scale building blocks can be tailored through simple synthetic methods, it becomes possible to easily incorporate additional functionality to provide the resulting gel with additional multifunctional behavior (for more details on the applications of LMWGs, see Section 5). One of the most intriguing aspects of gel-phase materials is that they assemble across multiple length scales, and, as such, modifications at the molecular level can be expressed on the nanoscale (in fibril assembly), on the microscale (in network formation), and ultimately on the macroscopic length scale (through rheological/materials behavior). As such, a wide range of techniques have to be employed to study the aspects of gel behavior and performance across all length scales, in order to gain detailed insight.

2 2.1

CHARACTERIZATION OF GELS Macroscopic characterization

On the macroscopic length scale, there are a range of techniques for probing aspects of gel formation. The simplest relies on visual inspection of the gel, its dependence on factors, such as concentration, and its response to external stimuli, such as temperature—sometimes referred to as “table-top” rheology.15 Simple inversion of a gel sample, and inspection of whether or not the sample flows, allows a basic determination of whether a sample exists as a gel, a sol, or an intermediate viscous liquid (inverted vial 120

Sol

Tgel (°C)

100 80 60

Gel

40 20 Gel

0 0

10 20 30 [Gelator] (mM)

40

Figure 3 Typical phase diagram plot of variation of Tgel with concentration and an image of a gel as examined using the vial inversion technique.

3

method, Figure 3). It is important that such measurements should always be carried out in identical tubes if data are to be compared, as the stress exerted by the gel depends on the mass of the sample and the surface area of the gel, as well as interactions between the gel and the walls of the vessel. Most commonly, the temperature at which a given sample is converted from a gel to a sol is determined—the Tgel value of the material. This temperature is concentration dependent—that is, as the concentration of gelator increases, the “solidlike network” becomes more extensive and, as such, the transition temperature increases (Figure 3). An alternative visual approach to monitor the physical state of a gel is to observe the temperature at which a stainless steel ball of defined mass and diameter placed on top of the gel falls through the sample (falling ball method). In addition to these simplistic visual approaches, it is possible to carry out full rheological investigation of the materials.16 In typical experiments, the magnitudes and ratios of the elastic/storage (G
) and viscous/loss (G

) moduli are determined by using oscillatory shear. The storage modulus (G
) indicates the ability of a deformed material to regain its shape, while, the loss modulus (G

) represents the ability of the material to flow under stress. For a gel, the elastic storage should be independent of oscillatory frequency, and G
should exceed G

by about one order of magnitude. Measurements of viscosity also provide a useful method for comparing gel-phase materials in a quantitative way and generating structure–activity relationships. It is increasingly clear, from the relationship between strain and stress tensors, that the choice of theoretical rheological model for gels is not straightforward.16 Molecular gels can be classified as cellular solids, fractal/colloidal systems, or soft glassy materials, dependent on their behavior. Therefore, there is a need for continued application of rheological methods to molecular gels in order to provide a better insight into the way in which modifying gelator structures on the molecular scale control the rheological properties of the resulting soft material. Differential scanning calorimetry can be used to provide an insight into the thermal behavior of a gel.15 On heating, it is possible to observe an endotherm associated with the gel–sol transition, and on cooling, an exotherm can be observed as the sol reforms the gel. However, it should be noted that the conversion of a gel to a sol is not a first-order phase transition, as it is a multistep hierarchical process, and for this reason, the observed endotherms/exotherms can be very broad. However, if such peaks are observed, then integration can provide data about the enthalpy of the transition (
Hgel–sol on heating and
Hsol–gel on cooling). This method can also provide detailed information on the reversibility and repeatability of the gel–sol–gel transitions.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

4

2.2

Soft matter

Microscale and nanoscale characterization

It is possible to gain significant insights into the organization of gels using electron microscopy methods. Often, scanning electron microscopy (SEM) and transmission electron microscopy (TEM) are performed on dried samples of the gel, that is, the xerogel (Figure 4a). Electron microscopy can provide a direct insight into the nanoscale dimensions of the fibrillar objects present within most gels. However, it should be noted that considerable collapse and reorganization of the gel network may occur during the drying process, as the effective concentration of the gelator increases. Therefore, although SEM is useful for comparing families of structurally related gelators, it must be remembered that drying artifacts can be significant. Furthermore, in the absence of heavy atoms in the gelator structure, samples studied by using TEM are often stained for enhanced visualization. The use of cryo electron microscopy (cryo EM) methods can provide a better insight into the solvated structure of the gel network, as the solvent is removed by sublimation after the sample has been frozen, meaning that there is less thermal energy available for collapse/reorganization of the solidlike gel network.17 When cryo EM methods are carried out, the gel fibers typically appear to be noncollapsed, giving a better insight into network morphology (Figure 4b). X-ray and neutron methods can also be used to provide an insight into the aspects of gel assembly such as fibril dimensions and structure.18 Typically, fibrils and fibers are very small and are disorganized in terms of their relative orientations within a gel (isotropic); therefore, it is necessary to study the scattering of X-rays/neutrons at low angles. These methods can either be applied to dried xerogel samples, which have a greater degree of order, or, using a

200 nm

(a)

Mag = 100.00 K X EHT = 3.00 kV

Signal A = InLens WD = 3.0 mm

synchrotron source, can be used on the solvated gel itself (provided the solvent itself does not absorb). Fitting the data to an appropriate computer model can yield fibril diameters and can sometimes provide information about helical repeat lengths or the bundling of fibrils to yield larger fibers. If the LMWG is chiral, then circular dichroism (CD) spectroscopy is a useful technique for probing nanoscale chiral organization.19 This method also begins to give insight into how the self-assembly of fibrils takes place from the molecular level (i.e., from the bottom-up). In general, as small molecules assemble into chiral nanostructures, there is a significant increase in the intensity of any bands in the spectrum. The assembly process can most easily be monitored by carrying out variable temperature (VT) CD spectroscopy—as the temperature increases and the gel nanostructures disassemble, the CD bands associated with them typically decrease in intensity (Figure 5).21 Variable concentration CD spectroscopy can also be very useful for monitoring chiral self-assembly processes and has been used to provide an insight into the nucleation pathway that takes place in gel nanofiber self-assembly.12 This approach can distinguish between stepwise fiber growth (isodesmic growth), in which each gelator building block is added to the assembly with the same affinity, and a cooperative growth process, in which after the initial formation of an assembled “nucleus,” further fiber growth becomes thermodynamically enhanced. Cooperative assembly of this type is also observed in the aggregation of amyloid fibrils in disease pathologies such as Alzheimer’s disease. In addition to CD spectroscopy performed in the UV–vis region of the spectrum, it is also possible to carry out vibrational CD in the infrared (IR) region, which can provide information about the chiral organization of key functional groups.22

Date: 20 Aug 2009

X 50,000

2.00 kV SEI

100 nm JEOL 9/25/2009 SEM WD 3 mm 12:18:10 PM

(b)

Figure 4 Typical images of a fibrillar gel network, as imaged using: (a) SEM (scale bar 200 nm) and (b) cryo-SEM (scale bar 100 nm) techniques. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels

5

80

q (mdeg)

40 0 190 −40

210

230

250

290 70 °C

−80

50 °C RT

−120 −160

270

l (nm)

Figure 5 Variable temperature CD experiment demonstrating disassembly of chiral nanostructure on application of heat. (Reproduced from Ref. 33.  American Chemical Society, 2005.)

In some cases, nanoscale chirality can be observed by using electron microscopy methods, for example, in the form of self-assembled helical fibrils/fibers. It has been reported that using opposite enantiomers as gelators can give rise to fibers with opposite helicities (Figure 1).23 In this way, chirality on the molecular scale is directly translated into chirality on the nanoscale.

2.3

Probing assembly on the molecular scale

Standard spectroscopic methods can be very useful for probing and understanding the self-assembly of gelators at the molecular scale. NMR spectroscopy can provide a unique and fundamental insight into the molecular recognition pathways that underpin self-assembly.24 The “solidlike” gel network is effectively invisible by NMR, as the NMR resonances of “immobilized” gelator molecules are significantly broadened. However, anything which is mobile within the “liquidlike” phase exhibits sharp peaks. Indeed, it should be noted that the solvent, and any small molecules dissolved within it, are effectively highly mobile on the molecular scale and can diffuse rapidly through the gel, even though the material does not exhibit flow on the bulk, macroscopic scale—this provides a significant insight into the highly dynamic nature of a gel on the molecular scale. Furthermore, it is sometimes possible to observe gelator molecules in equilibrium with the solidlike fibers, as long as the on–off kinetics are appropriate. By using a probe molecule, which does not aggregate with the gel fibers, it is possible to use NMR peak integration to quantify how many of the gelator molecules are mobile within the liquidlike phase, and by inference, how many are actually immobilized within the solidlike fibers.25 The effects of concentration and temperature can also readily be monitored by using NMR. Shifts in the NMRs indicate molecule–molecule interactions within the solution phase

and can, as a consequence, provide a direct insight into the types of noncovalent interactions that ultimately underpin the assembly of the solidlike gel fiber network; for example, hydrogen-bond and π –π stacking interactions can be easily identified. It is also possible to determine whether the fiber formation is an isodesmic or cooperative process using these experiments and hence derive effective equilibrium constants for gelator self-assembly.25 IR spectroscopy is frequently used to provide evidence of hydrogen-bond, π –π stacking, and van der Waals interactions. Shifts in IR bands on changing concentration or temperature, or differences between solid, gelphase, and solution-phase IR spectra, can enable assignment of the types of noncovalent interactions that underpin the nanoscale assembly process. Optical spectroscopy can also be useful, depending on the molecular structure of the gelator; for example, UV–vis spectroscopy is a good way of probing gels assembled via donor–acceptor interactions, whereas fluorescence spectroscopy can provide information about the organization and solvent shielding of luminescent chromophores within solidlike gel fibers.

3 3.1

SELF-ASSEMBLY RULES IN GELS What makes an effective gelator?

A wide range of different types of compounds have been shown to form gel-type materials (Section 4). Although it has often been stated that gelators are discovered serendipitously, it is becoming increasingly possible to predict whether or not a compound will form effective gels.26, 27 In general, gelators must be capable of forming selfcomplementary noncovalent interactions, and these should have a tendency for one-dimensional organization, as this can encourage fibril assembly rather than three-dimensional crystal formation.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

6

Soft matter

O R′

O

N

N

H

H

R

N

N

H

H

O R′

R′

O

N

N

H

H

R

N

N

H

H

R′

Figure 6 Assembly of bis-ureas through the formation of selfcomplementary hydrogen-bonding molecular recognition pathways.

To illustrate the principles of molecular fibril assembly, one of the first classes of synthetic LMWGs were the bisureas, related to polyurea gels developed by the Japanese automotive industry in the mid-twentieth century.6 It is now understood that these molecules assemble into onedimensional fibrils as a consequence of complementary urea–urea hydrogen bonds (Figure 6). As long as the basic bis-urea recognition units are in place, it is possible to change the structure of the R and R
groups and still obtain gelation. The surface groups (R
) control the solvent compatibility of the gel fibers (Section 3.2). The spacer group (R) controls the relative orientation of the two urea groups. For example, aliphatic chains with odd or even numbers of carbon atoms change the relative orientation of the hydrogen-bonding units and hence the propensity of the molecular-scale building blocks to selfassemble—“odd–even” effects of this type are commonly observed in gel formation.28 In many other cases, the nature of the noncovalent interactions responsible for self-assembly is less clear—however, in general terms, such interactions must exist, and interactions of the gelators in three dimensions must be somehow frustrated by the molecular structures. In this way, fibrillar assembly becomes preferred. However, it should be noted that in addition to simple assembly of a molecular-scale fibril, there must also be some fibril bundling and fiber–fiber interactions, responsible for the formation of a sample-spanning network. These interactions are often relatively poorly understood, although it should be noted that organogelators are very often functionalized with aliphatic chains, which opens the possibility that van der Waals interactions between the “hairy” surfaces of the fibers, may encourage network formation.

3.2

Effect of solvent/solubility

It is often tempting to focus most attention on the structure of the gelator when considering nanoscale self-assembly. However, gels are >99% solvent, and the liquidlike phase plays an important role in mediating self-assembly.

For example, in organic solvents, gelators will typically interact through hydrogen bonds, π –π stacking interactions, and van der Waals forces, which are all significant in such media, in order to assemble into organogel fibers. In competitive, polar solvents however, such as for hydrogels formed in water, hydrophobicity will typically play a dominant role in self-assembly, supplemented by electrostatic interactions, with the other noncovalent interactions playing a lesser/supporting role in directing self-assembly. A number of attempts have been made to correlate gelation parameters with empirical solvent parameters, and it has been noted that significant insight can be gained when the nature of the solvent is considered on a molecular level, for example, by considering Kamlet–Taft parameters, which express the ability of the solvent to donate and accept hydrogen bonds, as well as describing its polarizability.29 Furthermore, it has been demonstrated that gelator solubility plays a key role in mediating self-assembly.25 Gels are somewhat intermediate between homogeneous solutions and phase-separated crystals suspended in a solvent. Indeed, gelation can be considered as a dimensionally limited form of crystallization, and for this reason, an understanding of crystal engineering has been used to provide insight into which compounds might be expected to form gels based on the way in which they pack and stack in the crystal phase.30 Clearly, if a compound exhibits solubility, which is too high in a given solvent, it cannot act as an effective gelator, as assembly into nanoscale “solidlike” fibers will not be favored. On the other hand, if a compound is too insoluble, it prefers to crystallize, and, particularly, if crystallization kinetics are fast, it cannot form a solvent-compatible extended and swollen network. Gelator solubility is, therefore, often intermediate, and this explains why most gels are typically require either a heat-cool cycle or sonication to initiate the formation of the material; that is, it is necessary to solubilize the gelator sufficiently before nanoscale assembly can take place. The relationship between crystallization and gelation also explains why a significant number of gels are metastable and evolve over time from highly disperse networks of fibers into more crystalline forms, less capable of gelation.31 It has also been reported that the selfassembly of gel-phase materials depends explicitly on the solvent structure, with an organized solvent shell around a self-assembled molecular fibril playing an explicit role in rigidifying the aggregates and guiding them toward further hierarchical assembly into bundles and sample-spanning gels.12

3.3

Effect of chirality

Many gelators are constructed from chiral building blocks such as sugars and amino acids (Section 4), and, as such,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels there has been considerable interest in the way in which chiral information inherent within the molecular skeleton can inform the self-assembly process (e.g., Figure 1).32 There has also been considerable interest in the behavior of mixtures of enantiomers in terms of gelation. There are three possible outcomes: 1.

2.

3.

Most commonly, using a racemic mixture of gelators leads to a significantly decreased ability to form gels. This is presumably because one enantiomer can just as easily interact with its antipode as with itself, forming a mixture of different diastereomeric complexes which is not optimized for self-assembly into fibrillar objects, or preferentially crystallizes. In rare cases, enantiomeric mixing enhances gelation—in such cases, a diastereomeric complex is proposed to preferentially form between the enantiomer and its antipode, with this complex being better able to subsequently assemble into nanoscale fibrils than the individual enantiomeric building blocks.33 Finally, in a few cases, mixing enantiomers has little effect on gelation. In such cases, self-sorting into nanoscale enantiomeric objects is proposed to occur—with the two enantiomers acting independently of one another.20, 21 In this case, each enantiomer must preferentially interact with itself, rather than its antipode—hence driving the formation of discrete enantiomeric nanoscale objects.

In a key example,34 it was reported that a photoresponsive unit incorporated into LMWGs could be addressed via irradiation in order to lock the chirality of the molecularscale building blocks. Interestingly, this led to an inversion of the chirality of the overall supramolecular assembly. As such, the assembly process could be addressed using both temperature and photoirradiation, giving rise to a highly addressable and switchable nanoscale chiral assembly.

have identical molecular recognition pathways, it can mean that two different types of molecule can be coassembled into the same nanofibers. This method has allowed, for example, a small amount of chiral additive to bias the nanoscale assembly of an achiral gelator in favor of a given enantiomeric form, through a sergeants and soldiers mechanism (see Multicomponent Self-Assembled Polymers Based on π-Conjugated Systems, Soft Matter).35 If gelators preferentially interact with one another (2), this can enhance gelation and lead to multicomponent gels (Section 4.7) and hence generate new nanoscale objects. If gelators self-sort into their own architectures (3), different nanostructures can coexist in the same sample.36 Using mixtures of components with different self-assembly preferences is therefore a key strategy in nanoconstruction that allows complex multicomponent systems to be readily assembled by simple mixing. This does not only apply to mixture of different gelators but also enables the orthogonal assembly of hybrid nanomaterials using mixtures of building blocks programmed with different criteria for selfassembly; for example, surfactant vesicles and gel fibers can be coassembled orthogonal to one another.37 In an elegant work, it was demonstrated that the composition of a dynamic mixed combinatorial library of potential gelators could be biased if one of them formed a more stable gel than the others.38 In this case, self-assembly provides the thermodynamic driving force biasing the library of molecules in favor of one particular product. This demonstrates how gelation can select and amplify specific components from complex mixtures. It is interesting to speculate that such processes may have been important in the development of ordered systems from complex mixtures during evolution on the early earth—cytoplasmic cell interiors are gel-like materials, and it has been suggested that gels played an important role in the development of cellular life.39

4 3.4

Assembly from mixtures of components

The rules for assembly from mixtures of components are similar to those described above in terms of chirality. In general, different gelators can either 1. 2. 3.

interact with one another equally; interact with each other preferentially (social selfsorting); or interact with themselves preferentially (narcissistic self-sorting).

MOLECULAR STRUCTURES OF GELATORS

A wide variety of molecular architectures have been reported to act as LMWGs in organic and aqueous media. The following sections provide a brief overview of the types of functional group most commonly employed (Figure 7). As explained in Section 3, the functional groups that underpin gelation must have the potential to establish noncovalent molecular recognition pathways and generally have partial solubility in the solvent of choice.

4.1 If gelators interact with one another equally (1), this may mean that they get their molecular recognition pathways confused, suppressing gelation. However, if both gelators

7

Peptides and ureas

Peptides and ureas have been widely employed as gelators.40 These functional groups contain both hydrogen-bond

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

8

Soft matter

O R

R′ N H

HO

O

R

R

N

N

H

H

O

HO HO HO

RO

OR

Sugars

Peptides and ureas

Steroids and bile acids

O H3C (CH2)34 CH3

H2N Alkanes

N

N

O NH

N

HN

O

O

O

HN

O

O

O

NH

HN

O

HN

R Nucleobases

O HN

O

O

NH

O O

NH

O

NH

O N H

12 N H

HN

O

O

O NH

HN

O

O O

O

Dendrimers and oligomers

Two-component gel (schematic)

Figure 7

Key functional groups and structural motifs often associated with gelation.

donors and acceptors and are hence appropriate for the selfassembly of gels in organic media through the formation of self-complementary hydrogen-bond interactions. Peptides have been particularly attractive, as the wide range of amino acids makes it possible to generate a variety of assembled soft materials and also to incorporate function in a relatively simple manner. The functionalization of polar peptides with hydrophobic units allows the hydrophobic effect to drive self-assembly in water, and supported by electrostatic and hydrogen bonding interactions, such materials form effective hydrogels. The peptidic nature of these materials offers the possibility of biocompatibility and opens up a range of potential biological applications (Section 5).

within organic media and consequently assemble to form organogels—the subtle synthetic modifications that are possible within the family of sugars have made them ideal for exploring structure–activity and chirality effects on selfassembly.41 In order to form effective organogels, a number of the OH groups will usually be protected to provide the molecule with an appropriate balanced solubility profile. Less highly protected sugars, with their enhanced polarity, multiple hydrogen-bonding interactions, and hydrophobic surfaces, have also been employed as hydrogels. Often, hydrogelation is enhanced by functionalization of the sugar with a significant additional hydrophobic unit capable of further assisting self-assembly.

4.2

4.3

Sugars

Sugars contain multiple –OH groups that, as a result, can form self-complementary hydrogen-bond interactions

Steroids and bile acids

Steroids and bile acids have large, relatively flat, solvophobic surfaces, which are appropriate for stacking and,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels as such, have been employed in gelation systems.42 Once again, modifying the basic molecular framework with additional functional groups can enhance gelation—for example, by introducing additional π –π stacking units or potential hydrogen-bonding interactions to the overall assembly.

4.4

Nucleobases

Nucleobases are, at least in part, responsible for the organization of DNA into a one-dimensional helical selfassembly, and, as such, their ability to form π –π and hydrogen-bond interactions has made them of interest to the designers of molecular gels.43 Modification of nucleobases with aliphatic chains can enhance their solubility in organic media and provide them with the potential to operate as organogelators. Conversely, unmodified guanosine has a preference to assemble into a tetramer (G quartet) around an alkali metal cation. These complexes are then capable of stacking in water through their hydrophobic faces in order to yield effective hydrogel-type materials.38

4.5

Alkanes

Surprisingly, long-chain alkanes are capable of acting as gelators in nonpolar organic solvents, even though the only noncovalent interactions that they can exhibit are relatively weak van der Waals forces.44 Indeed, the aggregation/gelation of such molecules can lead to blockages in crude oil pipelines. There is a tendency for such alkanes to aggregate into two-dimensional platelets, which can form an interlocked network, consequently leading to gel-phase behavior. As such, these materials are a rare example in which gelation is not underpinned by a fibrillar morphology; however, it should be noted that the self-assembly is, nonetheless, dimensionally constrained (in this case to two dimensions). Alkanes are also widely employed as functional groups in other gelation systems. Appending alkanes onto a variety of molecular frameworks provides an opportunity to introduce van der Waals interactions (both within and between fibrils), as well as tuning the overall solubility of the gelator and, consequently, the solvent compatibility of the selfassembled architecture.

4.6

Oligomeric, dendritic, and gels

As described earlier, the formation of gels is dependent on the formation of multiple self-complementary

9

noncovalent interactions between molecular-scale building blocks. Oligomers and dendrimers, which contain repeated monomeric units organized either in linear chains or in a branched architecture, both possess multiple functional groups. As such, these classes of molecules are somewhat intermediate between LMWGs and traditional polymerbased systems45 and are well suited to the formation of gels. For example, there has been considerable development of peptide-derived dendrimers that assemble in a controlled manner into gels.46 Establishing multiple interactions between building blocks can lead to the formation of particularly strong materials as a consequence of the multivalent nature of the binding (see Multivalency, Concepts). Once again, depending on the functional groups chosen within the molecular architecture, it is possible to generate either organogels or hydrogels in this way. The tunability of the molecular structures (e.g., chain length, dendritic generation, etc.) means that studying these classes of gelator can provide detailed insight into structure–activity relationships.

4.7

Multicomponent gels

In a number of cases, gelators have been reported in which an initial complexation event between two or more nongelator molecules assembles a complex that is subsequently capable of further hierarchical assembly into a fibrillar gel. Such systems are referred to as multicomponent gels.47 One of the advantages of multicomponent gels is that they are highly responsive, and gelation can be effectively switched on at a desired moment by simple mixing of the two components. The first example of a two-component gel made use of the well-known interaction between barbituric acid and pyrimidine and clearly illustrates the principles of two-component gel formation (Figure 8). The molecular building blocks (2 and 3) are functionalized in such a way as to sterically prevent the three-dimensional aggregation and thus encourage the formation of a linear supramolecular tapelike polymer with alternating building blocks held together by hydrogen bonds.48 This system formed gels only when both components were present, with each component individually being soluble—demonstrating how two components can act in a synergistic manner. In a significant number of cases, metals have been used to trigger gelation processes via the formation of a complex that can then further support self-assembly and ultimately gelation.49 Such metal-containing gels are particularly interesting because the metal center provides a functional handle by which the gel can be spectroscopically probed and addressed via methods such as electrochemistry.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

10

Soft matter

H

O N

C12H25 C12H25

O N O

H

H N H

H

N N

H H

O C12H25 C12H25

H

N O N O

N H

+ H

C16H33

N

2

N

C12H25 C12H25

N H

H

O N O

H 3

N H H

H

O

N N

C16H33

N

H

H

H H

N H N

N

N H H

H

O

C16H33

N

H

N

C12H25 C12H25

O N O

2,3

H etc.

Figure 8 Two-component gel based on complementary hydrogen-bond interactions between derivatized barbituric acid (2) and pyrimidine (3) building blocks.

5

UNIQUE FEATURES OF GEL-PHASE MATERIALS AND THEIR HIGH-TECH APPLICATIONS

As described elsewhere in this volume, self-assembly is a versatile synthetic strategy to form many different types of nanomaterials. It is therefore worth considering what is special about gel-phase materials and what potential applications may stem from these unique features.50

5.1

Gels as soft materials

Perhaps, the most obvious aspect of gels is that they are soft materials and, as such, have simple bulk applications as a consequence of their rheological properties. The longest established applications of LMWG organogels are in the automotive industry and used as greases/lubricants.6 Gels have also seen historic applications in the field of war, where Napalm, a sticky incendiary gel, first developed during the second world war, was originally an LMWG based on a mixture of naphthenic and palmitic acids51 —although, more recently, polymer gels replaced the original self-assembled formulation. Recently, there has been increasing focus on the application of organogels, which makes use of their soft solid

characteristics. For example, it has been reported that an effective organogelator can be applied to the remediation of crude oil spillages.52 The ideal gelator should be based on simple, environmentally compatible building blocks and should assemble into a soft semisolid material within the organic phase without need for heating and/or sonication (Figure 9). It must be capable of operating in a thin film of oil spread across the surface of water containing dissolved salts (to mimic sea water). The resulting gel could then be scooped from the surface of the sea (e.g., with fishing nets) to remove the oil spillage.53 Gels have also been used for the restoration and conservation of works of art.54 Use of an appropriate organogel allows an organic solvent to come into contact with the artwork, hence removing soiling/dirt from the surface. However, the gel network prevents leaching from the paint underneath and maintains the physical integrity of the material. The gel itself can then be relatively easily removed from the surface of the paintwork, along with any undesired residues.

5.2

Highly solvated porous materials

One of the most intriguing features about gels is that although in many ways they appear to be solidlike materials, and indeed the applications described in Section 5.1

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels

H

H O

O

2.7 nm

O

O

11

O

O

O H

H O

O H

H O

O

O

O

O

O

H

H

H O

O H

O

O

O

O

O

O H

O H 3.01 nm

1

2

3

Figure 9 Sugar-derived gelator functionalized with alkyl chains self-assembles into a gel in organic solvents, even in the presence of an aqueous phase. This gel can be used to instantaneously immobilize a layer of diesel floating on water and, as such, has potential applications for remediation of crude oil spills. (Adapted from Ref. 53.  Wiley-VCH, 2010.)

rely to some extent on their solidlike rheological behavior, the nanostructured gelator network is highly solvated and porous, and the solvent phase has an effective high mobility on the molecular level. Gel networks are therefore potentially useful materials as reaction matrices and as supports for recyclable catalysts.55 In exciting work, it has been demonstrated that selfassembly can have a direct effect on the stereoselectivity of an organocatalytic reaction process.56 L-Proline-L-valine derivatives were used as organocatalysts for the conjugate addition of cyclohexanone to trans-β-nitrostyrene, and it was demonstrated that the version of the catalyst that self-assembled gave the opposite enantiomer to the nonself-assembling form. This difference was attributed to conformational changes of the dipeptide associated with selfassembly. Furthermore, the activity of the self-assembling catalyst was less dependent on conditions such as concentration and temperature, and the catalyst could be more

easily recovered and reused as a consequence of its existence as a gel. It has also been reported that the self-assembly of LMWG building blocks can give rise to changes in the basicity of catalytically proficient units such as L-proline57 and has been demonstrated that the self-assembled gel can exhibit higher activity as a consequence. It was suggested that the combined self-assembly and catalytic roles played by the amino acid building blocks may have significance from the point of view of studies of the origins of life. Hydrogels have been used to generate semi-wet peptide/protein arrays.58 By using an LMWG, hydrogel spots were created in an array format, and it was demonstrated that the semi-wet gel interior was a suitable environment/reaction medium for enzymes, whereas the hydrophobic fiber interiors could be used to monitor the reaction. In a proof-of-principle experiment, lysyl-endopeptide (LEP) was used as the enzyme catalyst. A pentapeptide

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

12

Soft matter

substrate for LEP was designed bearing Lys and a fluorescent DANsen hydrophobic moiety at the C-terminal. When LEP cleaved the peptide bond adjacent to the DANsen, the environmentally sensitive DANsen molecule shifted from the aqueous environment of the hydrogel cavity to the hydrophobic fibrillar network inducing a fluorescence change. Different enzymes/substrates were screened, but only the active couple showed the appropriate response. The concept of using hydrogels as semi-wet reaction media, has since been extended to detect a range of biological targets, including insulin.59

5.3

In early work, it was demonstrated that the incorporation of crown ethers into gelators gave such materials the potential to bind cationic guests.61 On binding cations, the self-assembly of the gelator building blocks was modified—changing the thermal behavior of the gel–sol transition. Different cations gave rise to different effects depending on the stoichiometry of the complex formed with the crown ether. Anion-responsive gels have also been reported.62 It is well known that ureas are good units for anion recognition (see Amide and Urea-Based Receptors, Molecular Recognition) with the hydrogen-bonding N–H groups being capable of binding to small electronrich, basic anionic guests. Furthermore, as described earlier (Figure 6), ureas are ideal units for the construction of gel networks, owing to the self-complementary nature of this hydrogen-bonding molecular recognition motif. As such, when gels incorporating ureas are exposed to anions, the anion has the potential to compete with the molecular recognition pathways that underpin gelation, and anions can therefore lead to gel break down. Such gels have been demonstrated to act as effective sensors of anions like fluoride in organic media. Hydrogels also have the potential to be highly responsive and given that they are formed in aqueous solution and can respond to biologically relevant molecules

Responsive nature—gels as sensors

One of the advantages of soft materials constructed via selfassembly methods is that the self-assembly process is, under the right conditions, reversible and gives such materials the potential to be highly responsive. When coupled with the observation made above that gels are highly porous and solvated materials, with rapid diffusion of small ions and molecules through the gelator network being possible, this gives them significant potential to act as sensors, responsive to the presence of target analytes by undergoing an easily monitored gel–sol (or sol–gel) transition.60

Hydrogelator X

Hydrophilic group S

Bla

S HOOC

N

COOH

Hydrophilic group

N

O HOOC H2O

Hydrogelator Self-assembly

Hydrophobic Hydrophilic

Hydrogel

(a)

(b)

(c)

(d)

500 nm

500 nm

Figure 10 Design of molecule in which β-lactamase (Bla) activity leads to cleavage and removal of a hydrophilic unit, switching on the formation of a hydrogel. The change in viscosity from (a) to (b) represents the response of the system to lysates of penicillin-resistant bacteria that contain β-lactamase; (c) and (d) are TEM images of samples A and B, respectively. (Adapted from Ref. 64.  American Chemical Society, 2007.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels and substrates. For example, it has been demonstrated by a number of groups that carefully chosen peptidederived gelators can undergo enzyme-triggered breakdown/formation.63 The enzyme acts to make or cleave a key bond within the gelator molecular structure, with the result that gelation is either switched on or off. As such, hydrogels can act as sensors of specific biological targets, reporting on their activity. For example, by designing an LMWG precursor that contained a β-lactam unit, it was possible to detect the activity of β-lactamase enzymes.64 If β-lactamase enzymes were present, the βlactam group was cleaved and gelation was switched on (Figure 10). It was demonstrated that this approach could be used to determine which bacterial cells contained βlactamase (by analyzing the cell lysate), and, as such, it was possible to determine which bacteria would be resistant to antibiotics in the β-lactam family, such as penicillin. Furthermore, applying this gel sensor approach avoided

(a)

(b)

ASAP

(d)

13

some of the false positives found when using other assay methods.

5.4

Self-healing potential

The dynamic and responsive nature of self-assembled gels means that they have considerable potential to repair their structures should damage occur. This was elegantly demonstrated using a hydrogel based on a combination of nanoscale clays, anionic surfactants, and dendritic molecules with multiple cationic surface groups that were capable of binding to the surfactant-stabilized clay.65 This gave rise to extended networks that showed high strength, shape persistence, and a remarkable capacity to repair. Notably, the gel could be cut into cubes, and if freshly cut surfaces were placed together, then they became joined with high strength. As such, the cubes could be used as building blocks for self-supporting bridgelike structures (Figure 11),

(c)

G3-binder

(e)

(f)

(g)

Figure 11 Gels derived from the interaction of clay nanosheets (a and d), anionic surfactant sodium polyacrylate (ASAP; b and e), and a polycationic dendritic cross-linker (G3-binder; c and f). This gel is capable of self-healing when freshly cut, as indicated by the free-standing bridge built from alternating blocks of gel with and without a dye additive. (Adapted from Ref. 65.  Nature Publishing Group, 2010.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

14

Soft matter

as clearly illustrated by fusing the blocks of gel that had been made either with or without a marker dye. It is to be assumed that the dynamic nature of the supramolecular gel, combined with the multivalent strong interactions between nanoscale building blocks is the feature that gives rise to this remarkable behavior.

5.5

Capturing self-assembled materials in more robust form

Self-assembly is an ideal methodology for programming controlled structures; however, in the case of gels, the resulting nanostructured materials are, in many cases, relatively weak and easily broken down. Although this responsive nature can be an advantage, as outlined in the sections earlier, there has also been considerable interest in developing methods to “capture” the self-assembled information in more permanent/robust form. Using this approach, molecular self-assembly preferences can be used to program the structuring of “hard” materials. Many attempts to capture gels have focused on the inorganic mineralization of the nanoscale gel scaffold.66 It has been demonstrated that gel fibers can act as effective templates for the formation of inorganic materials such as SiO2 , metal oxides such as ZnO, or non-oxide materials such as CdS.67 This approach has led to fibrillar, and in some cases helical, inorganic materials, in which the nanostructures of the gel fibers were essentially translated into the inorganic material. In key work, it was demonstrated that the “handedness” of the chiral organic nanostructures was directly transferred into the helicity of the silica structures formed using this nanotemplation approach.68 Structuring hard inorganic materials using a self-assembled scaffold mimics aspects of biomineralization and is a method for producing smart materials with enhanced physical properties and applications. As such, this approach has been employed to template the formation of biomaterials such as apatite (i.e., bone).69 The gelator used to template biomineralization was carefully tailored to include specific groups to enhance bone growth—for example, phosphorylated serine groups that act as Ca2+ receptors and Arg-Gly-Asp (RGD) peptides, which act as ligands to encourage cell adhesion (Figure 12). Importantly, all these active units are displayed on the periphery of the self-assembled gel fiber and are available to the surrounding medium to interact with the crystallizing bioinorganic matrix. The crystallographic axes of the resulting apatite crystals were demonstrated to be aligned with the long axes of the gel fibers—analogous to the alignment observed between collagen fibers and hydroxy-apatite crystals in bone. In addition to templating the formation of crystalline materials, the void spaces present within gel-phase materials are of interest for controlling the crystallization of

inorganic materials. For example, calcite crystals can be effectively grown within hydrogels,70 and in a recent report, Steed and coworkers demonstrated that low-molecularweight bis-urea gels could be used to control the crystal growth of pharmaceutical substances.71 It was demonstrated that controlled crystallization occurred without cocrystal formation. The gels used were anion responsive (Section 5.3), and it was shown that the addition of an appropriate anion was an appropriate method by which the gel could be broken down and the pharmaceutical crystals easily reclaimed. Gel-phase materials have also been captured within polymeric matrices. The simplest way of achieving this is to polymerize a fluid monomer phase around a selfassembled gel scaffold—which forms a material in which the nanofibers of the gel network are effectively embedded.72 It has been demonstrated that the presence of a “nanoskeleton” within the bulk polymer phase can significantly modify the material properties of the polymer—for example, enhancing materials strength.73 Such materials have potential applications in polymer-based films, coatings, and paints. In related work, it has been reported that the addition of LMWGs can significantly improve the hardness and elastic properties of bitumen, used in the construction industry, owing to the ability of the gelator to establish a nanoscale network within the material.74 In some cases, it has been demonstrated that after polymerization, the self-assembled gel scaffold can be washed out of these materials, leaving nanoscale cavities within the polymeric matrix75 —this process can be considered as nanoscale molecular imprinting (see Molecularly Imprinted Polymers, Soft Matter). Such materials may ultimately be applied as porous membranes or in separation science/catalysis. Gelator nanostructures can also be “captured” by polymerization of the gel network itself. One way of achieving this has been to introduce photopolymerizable bisalkynes into the molecular structure of the gelator.76 The self-assembly process organizes the bis-alkynes in space and enables the photopolymerization process to take place efficiently along the gel fibers. It has been demonstrated that the structure of the gel network can be directly translated into the robust polymerized material. Metathesis polymerization of alkenes has been used in a similar way, with Grubbs’ catalyst being diffused into the highly porous, solvated gel material in order to initiate polymerization.77 These examples demonstrate how, by careful choice of the functional groups appended to the gelator at the molecular level, the overall gel network can be endowed with useful and new forms of functional behavior—in this case the potential to achieve postassembly polymerization. As such, supramolecular gels are highly promising and simple materials for the nanofabrication of smart materials.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels

15

4 O

2

HO P OH H N O

O N H SH

SH H N

O

O

N H SH

SH H N O

O

H N

N H

O

O

O N H

H N

O

O

(a)

O

NH

3 1

H N

N H

H 2N

O OH O OH

NH 5

(b)

(c)

Figure 12 Gelator used to template mineralization of bone (a and b) self-assembling into cylindrical fibers (c). Unit 1 encourages self-assembly; unit 2 contains thiols capable of polymerizing and stabilizing the self-assembled structure; unit 3 is a glycine spacer; unit 4 contains a phosphate group capable of binding Ca2+ and directing biomineralization; and unit 5 is an RGD peptide that enables cell adhesion. (Adapted with permission from Ref. 69.  American Association for the Advancement of Science, 2001.)

5.6

Hybrid materials incorporating gels

Once again, the porous nature of gel-type materials also makes it relatively easy to generate gels that also contain other nanostructured materials. Such materials can usually be made by simple mixing, followed by heating/cooling or sonication. This allows hybrid materials containing, for example, mixtures of gel fibers and carbon nanotubes or metal nanoparticles to be easily formed.

The incorporation of carbon nanotubes into gel-type materials is of great interest and importance for the development of materials with interesting materials and optoelectronic behavior.78 It has been reported that mixing carbon nanotubes with oligo(p-phenylenevinylene)s (OPVs) can trigger enhanced gelation through network formation.79 It has been suggested that noncovalent interactions between the oligomers and the carbon nanotubes both help to disperse the nanotubes and also underpin the accelerated

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

16

Soft matter

OC12H25

HO H25C12O

H25C12O

OC12H25

H25C12O

OC12H25 OH

OPV1 OC12H25

HO H25C12O

OC12H25

OC12H25 SS

H25C12O

H25C12O

O O

OPV2

100 nm (b)

(a)

3 µm

60 nm (c)

Figure 13 Organogelator OPV1, on mixing with OPV2 (1%), enables the organization of metal nanoparticles to take place along the surface of the tapelike assembly, as illustrated by TEM. (Adapted from Ref. 83.  Wiley-VCH, 2007.)

network forming process. This gives rise to materials with enhanced mechanical properties. Such composites have been demonstrated to exhibit enhanced conductivities, and it has been shown that irradiation with a near IR laser can induce a gel–sol transition’—due to the excitation of the carbon nanotubes within the hybrid material.80 Incorporating surface-functionalized metal nanoparticles into gels has also been demonstrated to have a direct impact on the mechanical behavior of the gel.81 This effect could be correlated with the molecular structure of the nanoparticle surface ligands and a consideration of the way in which they could directly interact with the gel nanofibers through noncovalent interactions. Furthermore, it has been demonstrated that if the gel fibers are decorated with appropriate ligating groups, such as thiols, metal nanoparticles can be specifically arranged along the surfaces of the fibers through S–Au interactions.82 In this way, gels provide an easy mechanism by which the anisotropic arrangement of metal nanoparticles can be achieved. Such materials may have interesting applications in optoelectronics or, if the metal nanoparticles are based on Pd, interesting catalytic properties. Figure 13 demonstrates an example in which gold nanoparticles are bound to tapelike structures of self-assembled OPV organogelators.83 In this case, OPV1 assembled into one-dimensional fibrils, and even when OPV2 was added (1%) fibrillar assembly still occurred. OPV2 contains disulfide units capable of binding to metal nanoparticles, and,

as such, gold nanoparticles could be bound to the tapelike assemblies. In this case, the proximity of the metal particles to the π-conjugated tapes facilitated the electronic communication and the resulting materials begin to resemble one-dimensional hybrid organic/inorganic “wirelike” materials.

5.7

One-dimensional nanoscale materials with optoelectronic applications

The discussion in the previous section indicated that gel fibers can be an effective way of organizing potentially conductive materials in a one-dimensional manner. Indeed, the fact that molecular-scale building blocks are organized into fibers is one of the most attractive features of gels to nanoscale engineers, as it means that the self-assembly processes that underpin gels have the potential to be employed in the development of nanoscale electronics. In particular, a gel fiber with a conjugated, conducting core and an apolar periphery can be considered as a nanoscale object analogous to a macroscopic copper wire, sheathed in an insulating polymer. As such, there has been considerable interest in the development of conjugated gelators that self-assemble into fibrils capable of exhibiting conductivity properties.84 Recently, it has been demonstrated that tetrathiafulvalene-derived (TTF) gelators can self-assemble into a

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels fibrillar network, which on doping with iodine vapor exhibits conductivity.85 The xerogel was prepared on a glass surface and oxidized with iodine vapor—microscopy demonstrated that this process did not cause disruption of the gel morphology. The resultant material had measurable room temperature conductivity. The temperature dependence (200–300 K) of the resistance also indicated semiconductor-like properties. Annealing the xerogel allowed the formation of robust TTF-nanowires on different substrates. Intriguingly, current sensing AFM measurements were used to probe the conductivity of individual nanofibers, and it was reported that thicker fibers are more conducting than the thinner ones, a likely consequence of their better ordering and/or more effective interfiber contacts.86 Alignment of conductive nanofibers is an area of considerable interest. Shinkai and coworkers also developed a TTF-based gelation system composed of one-dimensional fibers, which were unusually highly aligned in one direction.87 It was shown that when doped with iodine, the mixed valence state of the TTF stack had a characteristic absorption band in the NIR region—a necessary prerequisite for conductivity. Kato and coworkers have demonstrated that forming gels within a liquid-crystalline fluid phase can lead to the alignment of gel fibers induced by the liquidcrystalline fluid.88 By combining electroactive TTF-derived F 3C

F3C

CF3 NC CF3

CF3

S F3C

CN-TFMBE

gelators with a liquid-crystalline fluid phase, oriented gel fibers with interesting electrical conductivities could be generated.89 In addition to electronic behavior, the optical properties of gel-phase materials have come under particular scrutiny. For example, it is possible to generate highly fluorescent nanostructured materials in which the electronic demands of substituents on the gelator framework tune the fluorescence wavelength of the soft material (Figure 14).90 Such systems can therefore be color tuned across the visible spectrum, with differently colored nanowires being visualized by using fluorescence microscopy techniques. The ability of self-assembling systems to harvest light has also been reported. For example, by functionalizing perylene with two cholesterol units, the molecule gains the ability to self-assemble into an organogel.91 Tuning the other substituents on the perylene enables the possibility of energy transfer from one type of building block to the other. It was reported that energy transfer was much more efficient within gels based on mixtures of molecules than it was in homogeneous solutions based on equivalent mixtures and that excitation of one molecular building block could lead to efficient emission from the other via intrafiber energy-transfer events. In this way, assembly of the molecular-scale building blocks into one-dimensional F 3C

F 3C

CF3 THIO-G

(a)

CN

CF3 F3C

NC

CF3

THIO-Y

(b)

H3CO

OCH3

CF3

NC CF3

DM-R

(c)

(d)

100 nm

100 nm

100 nm

100 nm

10 µm

10 µm

10 µm

10 µm

(e)

(f) I

F 3C

CN S

F 3C

NC

A

20 µm

(g) I

A

20 µm

17

(h) I

A

20 µm

I

A

20 µm

Figure 14 SEM (a–d) and fluorescence microscopy (e–h) images of gelators indicating how modifying the molecular-scale building blocks is able to tune the optical behavior of the self-assembled nanowires. (Adapted from Ref. 90.  American Chemical Society, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

18

Soft matter matrix. Furthermore, there is a possibility that interactions between gel fibers and cell surfaces may give rise to directed/controlled cell growth and organization. In eye-catching work, Zhang and coworkers reported a peptidic gel that could be used to assist optic nerve regeneration in a hamster animal model system (Figure 15).95 Injection of a warm 1% peptide solution into the damaged optic nerve, followed by in situ self-assembly and gelation, led to functional return of vision. It was proposed that cell–scaffold interactions promoted the healing process. Hydrogels have also been designed by Stupp and coworkers in which the gelator incorporated a pentapeptide (Ile-Lys-Val-Ala-Val) known to encourage the growth and direction of neural progenitor cells and demonstrated that after introduction into the spinal cords of paralyzed mice, these animals regained partial use of their rear legs.96 These examples demonstrate how self-assembling biocompatible soft materials can be used for tissue engineering. In particular, by introducing bioactive peptides, which are displayed on the surfaces of the self-assembled nanofibers in the form of a multivalent array (see Multivalency, Concepts), highaffinity interactions with biological tissue can be stimulated and encouraged.

nanostructures encourages the energy-transfer process and therefore mediates light harvesting within this class of material. It has been demonstrated that by self-assembling molecular building blocks containing perylene bis-imide-acceptor groups within a donor polymer matrix, it is possible to generate an interpenetrating donor–acceptor interface with inherent morphological stability.92 This type of morphology is important in the development of organic solar cells, and it was demonstrated that these materials could be used to generate functioning devices.

5.8

Biocompatibility/biorelevance

As discussed previously, many gels are assembled using bio-derived and/or biocompatible building blocks. Hydrogels assembled from peptide building blocks have a good chance of biocompatibility,93 and there has therefore been considerable interest in using this family of materials in a wide range of biological and medical applications.94 Soft materials such as gels are capable of supporting and directing the growth of cells within and on their

Ionic fluid: CSF 1.3 nm 5 nm

(B)

(A)

(a)

(C)

(b)

(c)

6 kV X30,000 0.5µm

09 32 SE I

(d)

100 90 Percentage (%)

80 70 60

Treated Control Linear (treated) Linear (control)

50 40 30 20 10 0

(e)

1

2

3

4

5

6

7

8

9

10 11

Figure 15 Peptide hydrogelator assembles into a fibrillar network (models (A and B) and SEM (C)). When applied to the damaged optic nerve of hamsters with blinding in their right eye, vision is regenerated and the hamster responds to stimulus (a–d). Data (e) indicated that treated hamsters regained circa 80% of vision, whereas untreated animals regained only 10%. (Adapted with permission from Ref. 95, R. G. Ellis-Behnke, Y.-X. Liang, S.-W. You, D. K. C. Tay, S. Zhang, K.-F. So and G. E. Schneider, Proc. Natl. Acad. Sci. 2006, 103, 5054–5059. Copyright (2006) National Academy of Sciences, U.S.A.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels In addition to applications in enhancing nerve cell growth, gelators have been suggested as appropriate therapeutic agents for the development of injectable joint lubricants for the treatment of osteoarthritis.97 The gelators were shown to mimic some aspects of the behavior of hyaluronic acid, the main component of healthy synovial fluid. This application of self-assembling gelators not only relies on their biocompatibility but also uses the rheological properties of the gel to assist the biomechanics of the human body. In an interesting supramolecular approach to treating uranium contaminated wounds, antiinflammatory amino acids were combined with a bis-phosphonate receptor for UO2 2+ .99 The overall conjugate self-assembled to form a gel, and the material was demonstrated to (i) remove uranium from the wound, (ii) decrease inflammation, and (iii) promote wound healing. In animal studies, mice treated with the self-assembled material had significantly better outcome in terms of body weight and life expectancy.

5.9

Drug delivery

Gels are widely explored in pharmaceutical science as potential drug formulation and delivery vehicles.100 There are some specific advantages to use LMWGs rather than polymer gels for this purpose—specifically their enhanced biocompatibility and biodegradability, as a consequence of their being assembled from small biocompatible building blocks via reversible noncovalent interactions. Hydrogels are the most widely explored systems for applications in drug delivery, and, in particular, there has been considerable interest in incorporating therapeutic agents into the structures of molecular hydrogelators.100 For example, analogs of gonadotropin-releasing hormone, including the approved drug ganirelix (used for the treatment of ovulation control) and degarelix (potential prostate cancer drug), can self-assemble into fibrillar materials.98 The 3-(naphthalen-2-yl)-D-alanyl unit in these systems promotes self-assembly and hydrogelation, with the gels exhibiting sustained slow release of the active drug for up to 35 days (Figure 16). As such, the hydrogel can act as a stable depot for subcutaneous drug administration. These gels showed much better, longer-term release profiles than related analogs that were incapable of self-assembly and gelation. In this way, the formation of a gel plays an active role in controlling drug release kinetics. It has also been demonstrated that simple drug molecules, such as ibuprofen, can be incorporated into a gelator structure—in this case, by modifying the drug molecule with a Gly-Gly dipeptide.101 The hydrogel, after incubation with carboxypeptidase Y, was shown to release ibuprofen

19

+ Action Action

Figure 16 Controlled release of active drug molecules from self-assembled amyloid structures. (Reproduced from Ref. 100, Public Library of Science, 2008.)

as an active ingredient. The knowledge gained from this kind of study can be of great value in the development of drug delivery systems with more expensive or complicated chemotherapeutic agents. The concept of modifying a drug to give it the potential to self-assemble is akin to the prodrug strategy, and pharmaceutical systems developed in this manner require similar approaches in terms of development and regulation. It has been noted99 that key aspects which should be considered in detail in the further development of therapeutic hydrogels include fuller investigations of bioactivity in vivo, toxicity studies, bioresponsiveness, and potential administration routes. Organogels are less widely used than hydrogels for drug delivery, as there are only limited solvents that are pharmaceutically acceptable. However, there is a growing interest in the development of LMWG organogels for drug delivery, and, in particular, a focus on understanding how the structure of the gelator can control the kinetics of drug delivery.102 Most commonly, the LMWG is mixed with the active drug ingredient within the pharmaceutical solvent of interest; however, it is also possible to incorporate the active ingredient into the gelator structure, with the drug being released via bond cleavage in a prodrug-type strategy. It is proposed that such materials may have applications for drug delivery via oral, transdermal, and parenteral (injected) routes of administration.

6

CONCLUSIONS

In summary, gels are an intriguing class of self-assembled material, in which the impact of synthetic changes made on the molecular level is clearly evident on the nanoscale and directly controls the macroscopic materials behavior visible to the naked eye. As such, this family of soft materials provides an ideal playground to explore fundamental principles of supramolecular chemistry. Furthermore, these materials have a wide range of potential applications, ranging from biological science to the development of advanced

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

20

Soft matter

nanomaterials with applications in device technology. The simplicity of nanofabrication using a self-assembly strategy should ensure that over the coming years gel-phase materials are increasingly widely explored. In particular, it seems likely that materials incorporating ever-increasing degrees of functional behavior are developed, in which the function is programmed in via synthetic chemistry at the molecular level, and then expressed on the nano- and macroscopic length scales as a consequence of self-assembly.

18. M. Anne, X-Ray Diffraction of Poorly Organized Systems and Molecular Gels, Chapter 11 in Molecular Gels, Materials with Self-Assembled Fibrillar Networks, eds. R. G. Weiss and P. Terech, Springer, Dordrecht, Netherlands, 2006.

REFERENCES

20. A. R. Hirst, V. Castelletto, I. W. Hamley, and D. K. Smith, Chem. Eur. J., 2007, 13, 2180–2188.

Self-Assembled Fibrillar Networks, eds. R. G. Weiss and P. Terech, Springer, Dordrecht, Netherlands, 2006.

19. G. Gottarelli, G. P. Spada, and E. Castiglioni, Circular Dichroism for Studying Gel-Like Phases, Chapter 13 in Molecular Gels, Materials with Self-Assembled Fibrillar Networks, eds., R. G. Weiss and P. Terech, Springer, Dordrecht, Netherlands, 2006.

1. P. Terech and R. G. Weiss, Chem. Rev., 1997, 97, 3133–3159.

21. V. Kral, S. Pataridis, V. Setnicka, et al., Tetrahedron, 2005, 61, 5499–5506.

2. R. G. Weiss and P. Terech, eds. Molecular Gels—Materials with Self-Assembled Fibrillar Networks, Springer, Dordrecht, 2006.

22. N. Ramanathan, A. L. Currie, and J. Ross Colvin, Nature, 1961, 190, 779–781.

3. A. R. Hirst, B. Escuder, J. F. Miravet, and D. K. Smith, Angew. Chem. Int. Ed., 2008, 47, 8002–8018. 4. T. Tachibana and H. Kambara, J. Am. Chem. Soc., 1965, 87, 3015–3016. 5. R. Zsigmondy and W. Batchmann, Z. Chem. Ind. Koll., 1912, 11, 145–157. 6. C. J. Donahue, J. Chem. Educ., 2006, 83, 862–869.

23. B. Escuder, M. Llusar, and J. F. Miravet, J. Org. Chem., 2006, 71, 7747–7752. 24. A. R. Hirst, I. A. Coates, T. Boucheteau, et al., J. Am. Chem. Soc., 2008, 130, 9113–9121. 25. J. H. van Esch and B. L. Feringa, Angew. Chem. Int. Ed., 2000, 39, 2263–2266. 26. J. H. van Esch, Langmuir, 2009, 25, 8392–8394.

7. M. J. Meunier, Ann. Chim. Phys., 1891, 22, 412.

27. S. De Feyter, M. Larsson, N. Schuurmans, et al., Chem. Eur. J., 2003, 9, 1198–1206.

8. L. A. Estroff and A. D. Hamilton, Chem. Rev., 2004, 104, 1201–1217.

28. C. Lagadec, W. Edwards, and D. K. Smith, Soft Matter, 2011, 7, 110–117.

9. D. H. Birdsall and B. B. Farrington, J. Phys. Colloid Chem., 1948, 52, 1415–1423. 10. H. B. Bohidar, P. Dubin, and Y. Osada, Polymer Gels: Fundamentals and Applications, ACS Symposium Series, vol. 833, American Chemical Society, Washington DC, 2003.

29. P. Dastidar, Chem. Soc. Rev., 2008, 37, 2699–2715. 30. J. R. Moffat and D. K. Smith, Chem. Commun., 2008 2248–2250. 31. D. K. Smith, Chem. Soc. Rev., 2009, 38, 684–694. 32. J. Makarevi´c, M. Joki´c, Z. Raza, et al., Chem. Eur. J., 2003, 9, 5567–5580.

11. G. John, B. V. Shankar, S. R. Jadhav, and P. K. Vemula, Langmuir, 2010, 26, 17843–17851.

33. B. W. Messmore, P. A. Sukerkar, and S. I. Stupp, J. Am. Chem. Soc., 2005, 127, 7992–7993.

12. P. Jonkheijm, P. van der Schoot, A. P. H. J. Schenning, and E. W. Meijer, Science, 2006, 313, 80–83.

34. J. J. D. de Jong, L. N. Lucas, R. M. Kellogg, et al., Science, 2004, 304, 278–281.

13. A. Aggeli, I. A. Nyrkova, M. Bell, et al., Proc. Natl. Acad. Sci. U.S.A., 2001, 98, 11857–11862.

35. A. Ajayaghosh, R. Varghese, S. J. George, and C. Vijayakumar, Angew. Chem. Int. Ed., 2006, 45, 1141–1144.

14. M. George and R. G. Weiss, Acc. Chem. Res., 2006, 39, 489–497. 15. S. R. Raghavan and B. H. Cipriano, Gel Formation: Phase Diagrams using Tabletop Rheology and Calorimetry, Chapter 8, in Molecular Gels, Materials with SelfAssembled Fibrillar Networks, eds. R. G. Weiss and P. Terech, Springer, Dordrecht, Netherlands, 2006. 16. P. Sollich, Soft Glassy Rheology, Chapter 5 in Molecular Gels, Materials with Self-Assembled Fibrillar Networks, eds. R. G. Weiss and P. Terech, Springer, Dordrecht, Netherlands, 2006. 17. D. Danino and Y. Talmon, Direct-Imaging and FreezeFracture Cryo-Transmission Electron Microscopy of Molecular Gels, Chapter 9 in Molecular Gels, Materials with

36. J. R. Moffat and D. K. Smith, Chem. Commun., 2009 316–318. 37. A. Brizard, M. Stuart, K. van Bommel, et al., Angew. Chem. Int. Ed., 2008, 47, 2063–2066. 38. N. Sreenivasachary and J. M. Lehn, Proc. Natl. Acad. Sci. U.S.A., 2005, 102, 5938–5943. 39. J. T. Trevors and G. H. Pollack, Prog. Biophys. Mol. Biol., 2005, 89, 1–8. ˇ c, Top. Curr. Chem., 2005, 40. F. Fages, F. V¨ogtle, and M. Zini´ 256, 77–131. 41. S. Kiyonaka, S. Shinkai, and I. Hamachi, Chem. Eur. J., 2003, 9, 976–983.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-assembling fibrillar networks—supramolecular gels

21

ˇ c, F. V¨ogtle, and F. Fages, Top. Curr. Chem., 2005, 42. M. Zini´ 256, 39–76.

70. H. Y. Li, H. L. Xin, D. A. Muller, and L. A. Estroff, Science, 2009, 326, 1244–1247.

43. K. Araki and I. Yoshikawa, Top. Curr. Chem., 2005, 256, 133–165.

71. J. A. Foster, M.-O. M. Piepenbrock, G. O. Lloyd, et al., Nat. Chem., 2010, 2, 1037–1043.

44. D. J. Abdallah, S. A. Sirchio, and R. G. Weiss, Langmuir, 2000, 16, 7558–7561.

72. R. J. H. Hafkamp, B. P. A. Kokke, I. M. Danke, et al., Chem. Commun., 1997, 545–546.

45. M. Suzuki and K. Hanabusa, Chem. Soc. Rev., 2010, 39, 455–463.

73. E. R. Zubarev, M. U. Pralle, E. D. Sone, and S. I. Stupp, Adv. Mater., 2002, 14, 198–203.

46. D. K. Smith, Adv. Mater., 2006, 18, 2773–2778.

74. B. Isare, L. Petit, E. Bugnet, et al., Langmuir, 2009, 25, 8400–8403.

47. A. R. Hirst and D. K. Smith, Chem. Eur. J., 2005, 11, 5496–5508. 48. K. Hanabusa, T. Miki, Y. Taguchi, et al., J. Chem. Soc., Chem. Commun., 1993, 1382–1384.

75. F.-X. Simon, N. S. Khelfallah, M. Schmutz, et al., J. Am. Chem. Soc., 2007, 129, 3788–3789.

49. F. Fages, Angew. Chem. Int. Ed., 2006, 45, 1680–1682.

76. M. George and R. G. Weiss, Chem. Mater., 2003, 15, 2879–2888.

50. N. M. Sangeetha and U. Maitra, Chem. Soc. Rev., 2005, 34, 821–836.

77. J. R. Moffat, I. A. Coates, F. J. Leng, and D. K. Smith, Langmuir, 2009, 29, 8786–8793.

51. L. F. Fieser, G. C. Harris, E. B. Hershberg, et al., Ind. Eng. Chem., 1946, 38, 768–773.

78. T. Fukushima, A. Kosaka, Y. Ishimura, et al., Science, 2003, 300, 2072–2074.

52. S. Bhattacharya and Y. Krishnan-Ghosh, Chem. Commun., 2001, 185–186.

79. S. Srinivasan, A. Ajayaghosh, 5746–5749.

53. S. R. Jadhav, P. K. Vemula, R. Kumar, et al., Angew. Chem. Int. Ed., 2010, 49, 7695–7698. 54. E. Carretti, M. Bonini, L. Dei, et al., Acc. Chem. Res., 2010, 43, 751–760. 55. B. Escuder, F. Llansola-Rodriguez, and J. F. Miravet, New J. Chem., 2010, 34, 1044–1054. 56. F. Rodriguez-Llansola, J. F. Miravet, Chem. Eur. J., 2010, 16, 8480–8486.

and

B. Escuder,

57. F. Rodriguez-Llansola, B. Escuder, and J. F. Miravet, J. Am. Chem. Soc., 2009, 131, 11478–11484. 58. S. Kiyonaka, K. Sada, I. Yoshimura, et al., Nat. Mater., 2004, 3, 58–64. 59. S. Bhuniya and B. H. Kim, Chem. Commun., 2006 1842–1844. 60. M. O. M. Piepenbrock, G. O. Lloyd, N. Clarke, J. W. Steed, Chem. Rev., 2010, 110, 1960–2004.

and

61. K. Murata, M. Aoki, T. Nishi, et al., J. Chem. Soc., Chem. Commun., 1991, 1715–1718. 62. G. O. Lloyd and J. W. Steed, Nat. Chem., 2009, 1, 437–442. 63. R. J. Williams, R. J. Mart, and R. V. Ulijn, Biopolymers, 2010, 94, 107–117. 64. Z. Yang, P.-L. Ho, G. Liang, et al., J. Am. Chem. Soc., 2007, 129, 266–267. 65. Q. Wang, J. L. Mynar, M. Yoshida, et al., Nature, 2010, 463, 339–343. 66. K. J. C. van Bommel, A. Friggeri, and S. Shinkai, Angew. Chem. Int. Ed., 2003, 42, 980–999. 67. M. Llusar and C. Sanchez, Chem. Mater., 2008, 20, 782–820.

S. S. Babu, V. K. Praveen, and Angew. Chem. Int. Ed., 2008, 47,

80. S. K. Samanta, A. Pal, S. Bhattacharya, and C. N. R. Rao, J. Mater. Chem., 2010, 20, 6881–6890. 81. S. Bhattacharya, A. Srivastava, and A. Pal, Angew. Chem. Int. Ed., 2006, 45, 2934–2937. 82. M. Kimura, S. Kobayashi, T. Kuroda, et al., Adv. Mater., 2004, 16, 335–338. 83. J. van Herrikhuyzen, S. J. George, M. R. J. Vos, et al., Angew. Chem. Int. Ed., 2007, 46, 1825–1828. 84. D. B. Amabilino and J. Puigmarti-Luis, Soft Matter, 2010, 6, 1605–1612. 85. J. Puigmarti-Luis, V. Laukhin, A. Perez del Pino, et al., Angew. Chem. Int. Ed., 2007, 46, 238–241. 86. I. Danila, F. Riobe, J. Puigmarti-Luis, et al., J. Mater. Chem., 2009, 19, 4495–4504. 87. T. Kitahara, M. Shirakawa, S.-i. Kawano, et al., J. Am. Chem. Soc., 2005, 127, 14980–14981. 88. T. Kato, N. Mizoshita, M. Moriyama, and T. Kitamura, Top. Curr. Chem., 2005, 256, 219–236. 89. T. Kitamura, S. Nakaso, N. Mizoshita, et al., J. Am. Chem. Soc., 2005, 127, 14769–14775. 90. B.-K. An, S. H. Gihm, J. W. Chung, et al., J. Am. Chem. Soc., 2009, 131, 3950–3957. 91. K. Sugiyasu, N. Fujita, and S. Shinkai, Angew. Chem. Int. Ed., 2004, 43, 1229–1233. 92. A. Wicklein, S. Ghosh, M. Sommer, et al., ACS Nano, 2009, 3, 1107–1114. 93. M. Zelzer and R. V. Ulijn, Chem. Soc. Rev., 2010, 39, 3351–3357. 94. B. Xu, Langmuir, 2009, 25, 8375–8377.

68. J. H. Jung, Y. Ono, K. Hanabusa, and S. Shinkai, J. Am. Chem. Soc., 2000, 122, 5008–5009.

95. R. G. Ellis-Behnke, Y.-X. Liang, S.-W. You, et al., Proc. Natl. Acad. Sci. U.S.A., 2006, 103, 5054–5059.

69. J. D. Hartgerink, E. Beniash, and S. I. Stupp, Science, 2001, 294, 1684–1688.

96. G. A. Silva, C. Czeisler, K. L. Niece, et al., Science, 2004, 303, 1352–1355.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

22

Soft matter

97. C. J. Bell, L. M. Carrick, J. Katta, et al., J. Biomed. Mater. Res., Part A, 2006, 78, 236–246.

100. S. K. Maji, D. Schubert, C. Rivier, et al., PLoS Biol., 2008, 6, e17, 0240–0252.

98. Z. Yang, G. Liang, M. Ma, et al., Chem. Commun., 2007 843–845.

101. S. Bhuniya, S. M. Park, and B. H. Kim, Tetrahedron Lett., 2006, 47, 7153–7156.

99. F. Zhao, M. L. Ma, and B. Xu, Chem. Soc. Rev., 2009, 38, 883–891.

102. A. Vintiloiu and J.-C. Leroux, J. Controlled Release, 2008, 125, 179–192.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc142

Self-Assembly of Facial Amphiphiles in Water Job Boekhoven, Patrick van Rijn, and Jan H. van Esch Delft University of Technology, Delft, The Netherlands

1 2 3 4 5

Head/Tail Amphiphiles versus Facial Amphiphiles Structural Features of Facial Amphiphiles Self-Assembly of Facial Amphiphiles in Water Facial Amphiphiles as Solubilizing Agents Facial Amphiphiles as Transmembrane Pore-Forming Agents 6 Morphological Sensing and Control by Facial Amphiphilic Structures 7 Conclusions References

1

1 3 3 7 8 13 15 15

HEAD/TAIL AMPHIPHILES VERSUS FACIAL AMPHIPHILES

Amphiphiles, from the Greek amphis for “both” and philia for “friendship,” are chemical compounds containing both a hydrophilic and a hydrophobic segment. Owing to their affinity for both hydrophilic and hydrophobic domains this class of molecules often interacts with interfaces between hydrophobic and hydrophilic phases and they are also often referred to as surfactant (from surface active agent). In the absence of free hydrophobic/hydrophilic interfaces, these amphiphiles can interact with each other because of hydrophobic interactions between the hydrophobic segments and assemble in water into a wide variety of architectures. Because of these properties, amphiphiles are widely found in everyday life fulfilling many different functions.

There are many different amphiphiles with a large variety of chemical structures and sizes, ranging from simple surfactants consisting of one or more hydrophobic alkyl chain connected to a single hydrophilic head group, gemini and bolaform surfactants containing two head groups, up to macromolecular block copolymers containing a hydrophilic and a hydrophobic polymer segment. The most common class of amphiphiles consists of head/tail amphiphiles, where the hydrophilic and the hydrophobic group are distributed along the length axes of the molecule. These amphiphiles adopt a cylindrical or a cone-like structure in water (Figure 1a). To prevent exposure of the hydrophobic moiety to the aqueous phase, these amphiphiles aggregate by bringing their hydrophobic moieties together. Because the hydrophobic moieties are exposed on a cylindrical surface, selfassembly extends in directions perpendicular to the normal of the surface (Figure 1b). In other words, the hydrophobic interactions of normal head/tail amphiphiles are divergent and their self-assembly results in larger aggregates. The morphology of such aggregates can be predicted by the structure of the amphiphile via the structureshaped concept.1 For each head/tail amphiphile, a socalled packing parameter (p) can be calculated as p = VH / lc a0 in which VH is the volume of the hydrophobic segment, lc is the length of the molecule, and a0 is the cross-sectional area of the head group. This packing parameter is directly related to the architecture formed. If p < 1/3 spherical micelles are expected, for 1/3 < p < 1/2 cylindrical micelles are expected, for 1/2 < p < 1 lamellar structures such as bilayers and vesicles are expected, and for p > 1 inverted micelles are expected (Figure 2). Beside the head/tail amphiphiles, there are other classes of amphiphiles with a distinct architecture and different properties. One of these classes of amphiphiles is formed

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

2

Soft matter

O

O

+ N

P O

O

O

OH OH

OH OH

O O

H O O

(c)

(a)

(b)

(d)

Figure 1 (a) Structure of dioctadecylphosphatidylcholine (DOPC), a classical head/tail amphiphile. (b) Hydrophobic interactions of such amphiphiles are divergent, and therefore their self-assembly can extend in any direction away from the cylindrical surface (black lines). (c) Structure of cholic acid, a typical facial amphiphile, in which the hydrophilic and hydrophobic groups are separated along the long axis of the amphiphile. (d) Self-assembly of facial amphiphiles is self-complementary and convergent. Dark gray areas indicate hydrophobic domains, whereas light gray areas indicate hydrophilic domains.

Packing parameter

Structures formed

Shape a0

1 Wedge

Inverted micelle

Figure 2 The structure-shaped concept of head/tail amphiphiles. The packing parameter of the amphiphile determines the structure of the aggregate formed on self-assembly. (Reproduced from Ref. 1.  Royal Society of Chemistry, 1976.)

by the so-called facial amphiphiles. Facial amphiphiles are distinguished from head/tail amphiphiles by the distribution of their hydrophobic and hydrophilic groups. In the classic head/tail amphiphiles, the hydrophobic and hydrophilic

segments are separated along the short molecular axis, but in facial amphiphiles they are separated along the long molecular axis (Figure 1c).2 Because of their separation along the longitudinal axis, the hydrophobic and hydrophilic moieties are almost fully exposed on either face of the amphiphile, hence the name facial amphiphiles. For these type of amphiphiles, it can be argued that the packing parameter p ∼ 1, which according to the structure-shaped concept would lead to the formation of bilayers. However, research in the past decades has shown that facial amphiphiles usually form different kind of assemblies. Because with facial amphiphiles, the hydrophobic moieties are almost fully exposed at one face of the molecule, interaction with one other hydrophobic domain of a similar amphiphile will almost completely cover both hydrophobic domains from the solvent. In other words, self-assembly of these facial amphiphiles is self-complementary and the exposure of their hydrophobic interactions is convergent (Figure 1d). Therefore, the self-assembly behavior of facial amphiphiles is markedly different from the better-known head/tail amphiphiles, but they also show distinct behavior at hydrophobic/hydrophilic interfaces, such as air/water or oil/water. Most interestingly, these different properties are associated with a number of interesting and important biological functions and applications which are unique to facial amphiphiles. In this article, we gave an overview of the self-assembly and interfacial behavior of various natural as well as synthetic facial amphiphiles and discuss their most important biological functions and applications.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water

2

STRUCTURAL FEATURES OF FACIAL AMPHIPHILES

Just like head/tail amphiphiles, facial amphiphiles are found in a variety of chemical structures and sizes, ranging from small molecules such as bile acids to biopolymers such as amphiphilic alpha helices. A common class of small molecular facial amphiphiles are the bile acids. Bile acids and their salts are found mainly in the bile of mammals and are made in the liver by oxidation of cholesterol.3 Cholesterol is a relatively flat fused hydrocarbon ring system and therefore has two hydrophobic faces. Enzymatic oxidation in the liver leads to the attachment of three hydroxyl groups as well as a carboxylic acid to one face of the fused ring system rendering it hydrophilic, whereas the other unchanged face remains hydrophobic (Figure 1c). Many studies on bile acid derivates showed that these compounds are very effective in stabilizing emulsions but can also serve as antimicrobial agents.4 Both topics are discussed in following sections. Facial amphiphilic peptides are another class of facial amphiphiles, which play an important role in many biological processes involving lipid bilayer membranes. Because of the large surface area of the amphiphilic domains, they are prone to interact with the hydrophilic/hydrophobic interface of lipid bilayers, which is necessary to assist in membrane fusion or transmembrane pore formation. In the case of pore-forming antibiotics, the peptides are often relatively small (between 25 and 100 amino acids) and the entire peptide becomes facially amphiphilic on folding into the secondary structure. In the case of membrane fusion or curvature-inducing proteins only the peptide fragment, which interacts with the bilayer membrane, is facially amphiphilic. The largest group of facial amphiphilic peptides consists of the alpha-helical peptides. Facial amphiphilic alpha helices, often referred to as amphipathic alpha helices, are not amphiphilic in their random coil conformation and their amphiphilicity is not directly obvious from their sequence. However, folding of the peptide into its preferred secondary structure, leads to the formation of an alpha helix, of which the hydrophilic amino acids occupy one face and the hydrophobic amino acids are located at the other face.5 Alpha-helical peptides have a periodicity of 3.6 amino acid residues per turn, and because of this, for two turns, roughly every third and seventh amino acids are on the same face of the alpha helix. In order to make a helix amphiphilic, the sequence of amino acids should alternate between hydrophobic and hydrophilic every three to four residues, which becomes more clear in a helical wheel representation (Figure 3). An example of such a facial amphiphilic alpha helix is magainin 2, a 23 amino acid antibiotic peptide.6 Studies have shown that magainin

1)

G

I

3)

G

K

F

L

7)

H

S

A

10) 13)

K G

K K

F A

F

17)

G

E

I

M

(a) 21)

N

S

N

A

S

E G

F

F

K K G H

V

3

N F A I G

K

G

K

V

L

(b)

Figure 3 (a) Amino acid sequence of magainin 2 with hydrophobic amino acids highlighted. (b) Helical wheel presentation of magainin 2, an amphiphilic peptide. (a) K I G A K A K I G A K A K I G A K A-NH2 1

3

+ (b)

2

5

7

9

+ +

4

6

8

11 13 15 17

+ +

+

10 12 14 16 18

Figure 4 (a) Amino acid sequence of antibiotic peptide described by Prasad Kari and coworkers with hydrophobic amino acids highlighted in bold and (b) its beta-strand diagram. Dark shaded circles are hydrophobic amino acids, whereas white circles are hydrophilic. (Reproduced with permission from Ref. 9.  American Society for Biochemistry and Molecular Biology, 2001.)

2 has a random coil conformation in solution.7 However, incorporating magainin 2 into a lipid bilayer induces folding into a facial amphiphilic alpha helix (Figure 3). Another class of facial amphiphilic protein domains are the amphiphilic beta sheets. Like alpha helices, amphiphilic beta sheets can be a substructure of larger proteins, as in beta-barrel proteins, or these beta sheets can be the entire peptide, as is the case with the smaller antibiotic facial amphiphilic beta sheets. If the sequence of beta-sheet peptides alternates between hydrophobic and hydrophilic residues,8 one face of the beta sheet is hydrophilic and the other is hydrophobic. An example of such a facial amphiphilic beta-sheet peptide strand is the synthetic antibiotic described by Prasad Kari and coworkers.9 This 18-sequence peptide has alternating hydrophobic and hydrophilic amino acids and it has been shown that it folds into a facial amphiphilic beta sheet in the presence of a bilayer (Figure 4).

3

SELF-ASSEMBLY OF FACIAL AMPHIPHILES IN WATER

In normal head/tail amphiphiles, the hydrophobic interactions are spatially divergent, but because of their specific molecular architecture the hydrophobic interactions of facial amphiphiles are much more convergent. They

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

4

Soft matter complementary and convergent. This is confirmed by the model of bile acid micelles proposed by Small where two or four molecules assemble because of a hydrophobic collapse between the cholesterol domains.21 The carboxylic acid side groups are on opposite sides of the micelle protecting these sides from water (Figure 5c). Several studies have been conducted on derivatives of cholic acid, which were easily obtained by functionalization of the hydroxyl groups. For instance, to further increase the hydrophilicity, these hydroxyl groups have been replaced by quaternary amines resulting in larger spherical micelles with a higher aggregation number in aqueous solution.25 Other examples involve functionalization with sugars,2 carboxylates,26, 27 amino acids,28 and sulfonates.29 If, however, only the carboxylic acid moiety is functionalized with a long-alkyl chain, the cholic ester stacks to form highly ordered sheets30 or fibers, depending on the solvent.31, 32 Another example of facial amphiphiles that form aggregates with a low aggregation number is the amphiphilic clips shown in Figure 6(a).33, 34 Depending on the counterions, these facial amphiphiles assemble into dimers or larger-sized aggregates in a wide range of solvents including water.35, 36 Self-assembly of synthetic facial amphiphile is, however, not restricted to assemblies with low aggregation numbers. For instance, very large and well-defined supramolecular assemblies are formed by the rigid-flexible macrocycle constructed from a rigid aromatic oligomer as the hydrophobic face and a flexible oligo (ethylene glycol) as

also have relatively large hydrophilic and hydrophobic surfaces, which are roughly equal in size. As a consequence, the properties of facial amphiphiles are clearly different from those of normal head–tail amphiphiles. For instance, assemblies of facial amphiphiles have relatively low aggregation numbers and have excellent solubilizing and emulsifying properties. The latter properties are discussed in more detail in the next section. Facial amphiphiles give rise to not only a wide variety of self-assembled architectures, including spherical-like10, 11 and worm-like micelles12 and vesicles,13, 14 but also less common morphologies unknown for head/tail amphiphiles, such as molecular clips and two molecule micelles. One key property of facial amphiphiles is the low aggregation number found with many micellar assemblies of facial amphiphiles. For instance, the aggregation numbers of bile acids 3, 6, and 7 are approximately 6, 3, and 5, respectively (Figure 5a and b).15, 16 This is in strong contrast with the much larger aggregation numbers for micelles formed by head-to-tail surfactants, e.g. ∼80 for CTAB17 or ∼50 for SDS18 (CTAB, cetyl trimethylammonium bromide, SDS, sodium dodecyl sulphate), making bile acids very efficient in the formation of micelles. It should be noted that the critical micelle concentrations of bile acids are in the same range as for common head/tail surfactants (0.9 mM for CTAB19 and 8.2 mM for SDS20 ). This marked difference in aggregation number is a direct result of the facial amphiphilicity. Since the hydrophobic domain is mainly located on one face, self-assembly is

R4 R3

O

O C A

OH

(a)

R1

OH

1 NaC 2 NaCDC 3 NaDC 4 NaGC 5 NaGDC 6 NaTC 7 NaTDC (b)

D

Hydrophobic interactions

OH

OH

B R2 H

R1

R2

R3

R4

Cmc (mM) Agg #

OH OH OH OH OH OH OH

OH OH H OH H OH H

OH H OH OH OH OH OH

ONa ONa ONa NHCH2CO2Na NHCH2CO2Na NHCH2CH2SO3Na NHCH2CH2SO3Na

13.8 7.0 5.25 12.02 4.26 4.33 2.98

5.82 (c) 2.59 5.42

Figure 5 (a and b) Chemical structure of the sodium salt of several bile acids: 1 cholic acid, 2 chenodeoxycholate, 3 deoxycholic acid, 4 glycocholic acid, 5 taurocholic acid, and 6 tauro deoxycholic acid. (c) Structures of micelles from cholic acid derivatives, proposed by Small and coworkers. Two or four molecules assemble because of hydrophobic interactions between the cholesterol groups. The hydroxyl groups (black dots on cholesterol) and the carboxylic acids side group shield the hydrophobic domain from water. (Refs. 22–24 for cmcs and Ref. 16 for aggregation numbers.) (Reproduced with permission from Ref. 21.  Indian Academy of Sciences, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water O

OCH3

OCH3 N

N R OCH3

O

O

R N

O

O

n

n O

N

O

O

5

O

O

O

O

O

O O

OCH3

CH3-1 R = p -C6H4COOCH3 Na-1 R = p -C6H4COO−Na+ NH4-1 R = p -C6H4COO−NH4+

P = 4.7 nm (a)

(b)

W = 4.1 nm

Figure 6 (a) Facial amphiphilic molecular clips forming dimers. (Reproduced with permission from Ref. 33.  Springer, 2001.) (b) Rigid-flexible cyclic facial amphiphile assembling in a bilayer strap which in turn folds into a tubular structure. (Reproduced with permission from Ref. 38.  American Chemical Society, 2006.)

the hydrophilic face (Figure 6b). In the absence of solvent, these macrocycles tend to aggregate into highly ordered bilayers with the rigid hydrophobic aromatic domains in the core due to π –π stacking.37 In aqueous solution, these molecules assemble into an infinite bilayer strap, which in turn folds into a helix, giving a helical tubular structure with a hydrophilic interior and exterior.38 Not only small molecular facial amphiphiles have been described but also polymeric facial amphiphiles. Amphiphilicity can be induced in polymers by using hydrophilic and hydrophobic blocks. Self-assembly of such head–tail amphiphilic block copolymers has extensively been described and is controlled by phase separation of the hydrophilic and hydrophobic blocks and their volume fractions.39 On the contrary, facial amphiphilic polymers can be obtained from alternating copolymers of hydrophilic and hydrophobic monomers, or by polymerization of facial amphiphilic monomers. An interesting example of such facial amphiphilic polymers is given by McCullough et al.40 These facial amphiphilic polythiophenes consist of alternating hydrophilic ethylene glycol/hydrophobic alkyl chain substituted thiophenes (Figure 7a). This polymer self-assembles into well-ordered monolayers at an air–water interface because of hydrophobic interactions between the alkyl chains as well as π –π interactions between the thiophene moieties. This high degree of organization can be related to the large surface area of the hydrophobic and hydrophilic face as well as the rigidity of the polythiophene backbone. A system analogous to the amphiphilic polythiophenes is based on oligophenylene ethynylene, which is turned into a facial amphiphilic polymer by the attachment of ammonium groups on one face and aliphatic tails on the

other (Figure 7b). These facial amphiphiles self-assemble in aqueous solution to form ordered layers and were found to be efficient stabilizers of oil–water emulsions.41, 42 The stabilization of emulsions by facial amphiphiles is discussed in more detail in the next section. Another class of natural facial amphiphilic polymer able to self-assemble into supramolecular architectures is the socalled leucine zippers.43 Owing to their large hydrophobic and hydrophilic surfaces, these amphiphiles can form architectures with low aggregation numbers. These facial amphiphilic alpha helices are characterized by seven amino acid repeating unit (a–b–c–d–e–f–g), where a and d are hydrophobic and on folding an amphiphilic alpha helix is created. Two of such helices can aggregate through hydrophobic interactions, as well as hydrogen bonding to form a coiled-coil. Several mutants have been investigated and show that mutations in both the hydrophobic and hydrophilic domains result in the formation of dimers, trimers, or tetramers.44 Dendrimers are another type of macromolecule, which are frequently used as building blocks of supramolecular architectures because of their well-defined structures. Numerous examples of amphiphilic dendrons and dendrimers exist, giving rise to various architectures on selfassembly. In some cases, dendrimers behave as facial amphiphiles although this is not obvious from the molecular structure (Figure 8a). The aryl ether dendrimers shown in Figure 8 contain both hydrophobic and hydrophilic substituents at each generation, and as a consequence they adopt a facial amphiphilic conformation in aqueous solution (Figure 8b), which self-assembles into micellar aggregates of 10–40 nm in diameter.45

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

6

Soft matter

CH3

CH3

O

O

O

O

O

S

CH3

O

O

O

H3C

CH3

O

O

O

S

O

S

S

S

S

S

S

CH3

(a)

O

O

O

O

O

O

Air Water

CH3N

CH3N

CH3N

CH3N

Side-on view 11 Å

(b)

CH3N

CH3N

Edge-on view 4Å

Figure 7 (a) Facial amphiphilic polythiophenes described by McCullough and coworkers and (b) the rigid oligophenylene ethynylenes described by Arnt and coworkers are known to form ordered structures at a water–air interface because of hydrophobic interactions and π –π stacking. (Reproduced with permission from Ref. 42.  American Chemical Society, 2003.)

O

O OH

HO O

O O

OH O

O

O

O

O

O

O

O

O

HO

O OH

and COO−

O

−

COO

O

O O

O

O

O

(a)

O

OH

COO −

COO

COO−

COO−

O

COO− COO−

OH

O

Hydrophilic residue

−

COO −

O

Hydrophobic residue

O

COO−

−

COO

COO− COO−

COO− COO− COO−

OH

(b)

Figure 8 (a) Molecular structure of a dendrimers described by Thayumanavan et al. (b) Facial amphiphilicity of the molecule becomes clear in aqueous solution. (Reproduced with permission from Ref. 45.  American Chemical Society, 2006.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water

O

O

O

O

O

O

7

O

O O

O O

O

O

O O

O

O

O

O

O

O

O

O O

O

O

O

O

O O

O

O

O

O

O

O

O

O

7 nm

RO

OR

50 nm

R = H, CH3

(a)

(b)

Figure 9 (a) Molecular structure of the series of facial amphiphilic dendrons. (b) These dendrons aggregate into elongated micelles in water. (Reproduced from Ref. 46.  Royal Society of Chemistry, 2007.)

Another type of facial amphiphilic dendrons based on an aromatic segment and an oligo (ethylene oxide segment) was described by Lee and coworkers (Figure 9a).46 Selfassembly of these amphiphiles gives rise to elongated micelles with an average width of 7 nm (Figure 9b). As the fully extended molecular length is about 4 nm, the diameter of 7 nm is consistent with a bilayer packing of the aromatic segments. From these examples, it is clear that self-assembly of facial amphiphiles results in the formation of a variety of common and less common aggregate morphologies. Most notable is the low aggregation number of micellar assemblies and the high degree of order in interfacial assemblies. These properties are related to the large and comparable size of the hydrophobic and hydrophilic surface areas of facial amphiphiles compared to those of classic head/tail amphiphiles.

4

FACIAL AMPHIPHILES AS SOLUBILIZING AGENTS

Another property of facial amphiphiles, which is related to the large hydrophobic and hydrophilic areas, is their effectiveness as solubilizing and emulsifying agents. In fact, one of the main functions of facial amphiphiles in molecular biology is the solubilization of hydrophobic molecules in micellar assemblies. For instance, the already mentioned bile acids are biologically the most important dissolution agents and assist in the transport and digestion of triglycerides47 and the dissolution of hydrophobic vitamins such as vitamin E.48 Bile acids are commercially used in detergents,

as food additive and in laboratories they are used to dissolve bilayers for protein purification.49 Because of the high efficiency of bile acids to form micelles and stabilize emulsions, researchers started to investigate nonsteroidal facial amphiphiles. An example of such a nonnatural facial amphiphilic compound is described by Sorrells and Menger,50 consisting of a rigid hydrophobic segment functionalized on one face with hydrophilic sulfates (Figure 10). These amphiphiles are effective at stabilizing an oil–water interface. As a consequence, the size distribution in an oil–water emulsion is unaffected for six months in the presence of 0.6 mM of the facial amphiphilic compound. Again the large hydrophobic and hydrophilic surfaces play a key role in this stabilizing behavior. Not only do facial amphiphiles act at the oil/water interface, natural facial amphiphiles also interact strongly with lipid bilayers such as cell membranes. Depending on the nature of the facial amphiphile, its interaction with biomembranes can lead to membrane bending, to pore formation, or even complete dissolution of the membrane. The dissolution of membranes by facial amphiphiles leads to cell death, and therefore the secretion of bile acids to the intestine of vertebrates is tightly regulated. The “carpet” model describes the mechanism of membrane dissolution by facial amphiphiles. Pore formation and membrane bending by facial amphiphiles are described in the next sections. The carpet model was first described for the mode of action of the alpha-helical facial amphiphilic peptides dermaseptin S51 and later for cecropins52 as well as LL-37.53 These compounds display antibiotic behavior

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

8

Soft matter

O +Na−O

O S

O

O +Na−O

O S

O 50 µm

(a)

(b)

Figure 10 (a) A nonsteroid facial amphiphilic compound described by Sorrells et al. (b) Oil droplets in water stabilized by the nonsteroid facial amphiphile. (Reproduced with permission from Ref. 50.  American Chemical Society, 2006.)

“Carpet” mechanism

face can then rotate and reorient toward the hydrophobic core of the membrane. If the concentrations of alpha helices on both sides of the membrane are high enough the membrane can be solubilized in a detergent-like manner.54 As a consequence the bilayer dissolves, causing leakage of the cellular contents leading to cell death (Figure 11). This process of dissolving the bilayer membrane to induce membrane leakage requires high (local) concentrations of facial amphiphilic peptide. This is demonstrated by calcein release from vesicles as a measure of activity of dermaseptin.51 Significant leakage was observed at peptide:lipid ratios of approximately 200 : 1 for soybean vesicles.

5

Figure 11 (a) Carpet mechanism of membrane dissolution; facial amphiphilicity is induced by the membrane bilayer. (b) If the concentration of amphiphiles reaches a threshold concentration they can incorporate into the bilayer. (c) By forming a carpet around a patch of bilayer these amphiphiles can dissolve the bilayer. (Reproduced with permission from Ref. 55.  Elsevier, 1999.)

and are short peptide sequences isolated from frog skin, moths, and humans, respectively. In the carpet model, the peptides interact with the hydrophilic/hydrophobic interface of the membrane, both via electrostatic attraction and hydrophobic interactions. The hydrophilic face of the antibiotic alpha helices is mainly cationic and therefore they have a preference for the negatively charged bacterial phospholipids. At high concentration on the membrane, the alpha helix monomers can form a “carpet” on the bilayer, these high concentrations can either be reached by covering the entire membrane surface or alternatively by local association of the monomers. The hydrophobic

FACIAL AMPHIPHILES AS TRANSMEMBRANE PORE-FORMING AGENTS

In the previous section, we showed that, because of their large surface areas, facial amphiphiles are active at hydrophobic/hydrophilic interfaces. This is not only the case for oil/water interfaces but also for the hydrophobic/hydrophilic interface of phospholipid bilayers. Because of this property, facial amphiphilic molecules have been found to carry out a variety of biological functions related to their interaction with cell membranes. One of the most prominent functions of facial amphiphiles is the formation of pores in bilayer membranes. In the case of peptide antibiotics and toxins, depending if the peptide is specific for bacterial or mammalian cells, respectively, the formation of pores leads to lysis of the cells and ultimately cell death. Many other facial amphiphiles form pores to regulate transmembrane transport, such as the ion channels for selectively transport ions, or release osmotic pressure. Pore formation by facial amphiphiles can occur via several mechanisms, mainly depending on the structure of the amphiphile. Extensive research has been done on the mechanism of action for antibiotic alpha-helical facial

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water

(a)

(b)

(c)

Figure 12 Mechanism of proposed pore formation by amphiphilic alpha helices. (a) Single helices combine with the bilayer. (b) If the concentration monomer is sufficient, the monomers insert in the hydrophobic core and align via the “barrel-stave” model or (c) via the “toroidal” model. (Reproduced from Ref. 57.  John Wiley & Sons, Ltd, 2000.)

amphiphiles such as the magainins56 and others.57 Mostly these alpha helices are relatively short (less than 50 amino acid residues) and show activity against various pathogens, including gram-positive and gram-negative bacteria, fungi, and protozoa. Pore formation by alpha helices has been proposed to proceed via one of two general mechanisms (Figure 12).55 The first model, referred to as the “barrel-stave” model, describes the formation of membrane pores by bundles of alpha helices.58 Like the previously discussed carpet model, single helices bind at the surface of the bilayer with the hydrophobic domains inserting in the hydrophobic core of the membrane, and the hydrophilic segments remaining in the aqueous phase (Figure 12a). The perpendicular insertion of a single alpha helix into the bilayer, that is, with the helical axis parallel with the bilayer normal such that the peptide spans the membrane, is highly unfavorable since this would lead to exposure of the hydrophilic peptide domains to the hydrophobic bilayer interior. However, if the concentration of alpha-helical monomers on the bilayer surface is sufficiently high, the monomers can insert in the bilayer and align their hydrophilic faces to overcome the unfavorable interaction with the hydrophobic membrane core (Figure 12b). An example of pore-forming alpha helix acting via the “barrel-stave” models is the toxin pardaxin. This 33-residue peptide is found in the Red Sea Moses sole (a flatfish) and used to repel Sharks.59 It is found that this peptide binds to bilayer surfaces and forms transmembrane pores consisting of 6 ± 3 peptides. Remarkable is the efficiency

9

of pore formation, studies have shown that calcein leakage takes place at lipid:peptide ratios as low as 2000 : 1.60 A second model, referred to as the “toroidal” model, is fairly similar to the “barrel-stave” model for the first step. The “toroidal” model differs from the barrel-stave model in that the peptides are associated with the lipid head groups and stay bound to these head groups through the entire process, even when they are perpendicularly inserted in the lipid bilayer (Figure 12c). It has been suggested that the previously discussed magainin 2 forms transmembrane pores via the “toroidal” model61 and only four to seven peptides can form a transmembrane pores at lipid:peptide ratios of 100 : 1.62, 63 In both models facial amphiphilicity is the key factor. If the peptides are also long enough, they are able to span the entire membrane. Because of their facial amphiphilicity, self-assembly into pores is self-complementary and as a result pores can form with low aggregation numbers, typically around five monomers. As a result, the diameter of the pores formed by alpha-helical peptides are relatively small, however, sufficiently large to induce leakage of small molecules and ions, and becoming lethal for the cell. The distribution of charge in pore-forming alpha-helical peptides is shown to be important for its biological function and specificity. Facial amphiphilic peptides with a relative large net positive charge distributed along the axis of the peptide backbone show antimicrobial behavior [e.g., magainin (+5)7 , cecropin (+7),64 and dermaseptin S3 (+5)65 ]. In contrast, peptides with a low net positive or negative charge show high lytic behavior toward mammalian cells [e.g., pardaxin (+1)59 and δ-hemolysin (0)].66 It has been proposed that these different modes of action are governed by electrostatic interactions between the peptide and the phospholipid membranes.55 The phospholipid membrane of mammalian cells is zwitterionic and the cell membrane interfacial potential is slightly negative.67 Bacterial membranes, on the other hand, contain acidic phospholipids resulting in much more negative interfacial potential.68 Therefore, cationic facial amphiphiles prefer to interact with bacterial cells, whereas slightly positive or negative charged or uncharged facial amphiphiles prefer to interact with mammalian cells. In addition to amphiphilic alpha helices, peptides with a beta-sheet secondary structure are also known, which are capable of forming transmembrane pores on insertion in lipid bilayers. Although the interaction of beta-sheet type peptides with lipid bilayers operates via many different mechanisms, facial amphiphilicity is one of the key factors for the formation of pores. One specific case where beta sheets are involved in pore formation is in the case of beta barrel-forming proteins.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

10

Soft matter

1 i

2

3

ii

iii

4

(a)

(b) Water-soluble Membrane-bound Prepore Membrane-inserted monomer monomer complex oligomer

Figure 13 (a) The structure of PFO as determined by X-ray crystallography. The colors refer to different domains of the protein. (Reproduced with permission from Ref. 72.  Elsevier, 1999.) (b) Mechanism of pore formation in bilayer (gray). (Reproduced with permission from Ref. 73.  American Chemical Society, 2001.)

Beta barrels are transmembrane proteins that contain betasheet peptide fragments incorporated in the lipid bilayer. The beta-sheet peptide fragment is facial amphiphilic and decreases the interfacial energy between the hydrophobic core of the bilayer and the aqueous phase to allow the formation of a pore with an aqueous interior. There are many different families of beta barrel-forming proteins with pore diameter ranging from a few nanometers to larger than 10 nm, where the pore can be formed by betasheet peptide fragments located on a single protein, or by oligomerization of beta-sheet peptide fragments located on different proteins.69 Such pore formation can have several functions, for example, pore formation can lead to leakage of ions leading to cell death by beta-barrel toxins, but other pores can also regulate the uptake and release of ions and small molecules in a controlled way. The mechanism of incorporation and pore formation differs drastically per family and an extensive discussion would be beyond the scope of this article. A nice example of pore formation by beta-barrel proteins is the Perfringolysin O (PFO) pore-forming protein. PFO is part of the family of cholesterol-dependent cytolysins (CDCs), a class of pore-forming toxins which form pores as large as 30 nm in diameter by oligomerization of 40–50 monomers.70 These CDCs are secreted by bacteria and act as toxins to cholesterol containing membranes. X-ray crystallographic studies have demonstrated that this protein consists of four domains (Figure 13).71 On pore formation, domain 4 shallowly inserts into the bilayer membrane after which two small alpha helices in domain 3 rearrange to form two antiparallel facial amphiphilic beta hairpins. These two beta hairpins also incorporate in the membrane and together with beta hairpins from other monomers, form the wall of the beta barrel (Figure 13).72

Besides the channel-forming peptides and proteins, there are also small molecule channel-forming agents. A wellstudied facial amphiphile is the antibiotic cholesterol derivative squalamine, which was first isolated from the dogfish shark (Figure 14a).74 Squalamine is a broad spectrum antibiotic with a minimal inhibitory concentration (MIC) of 1–2 µg ml−1 against E. coli. It has been proposed that the hydrophilic polyamine chain covers the hydrophobic face of the steroid when inserting in a membrane resulting in an amphiphilic compound (Figure 14b).75 This hypothesis was confirmed by studies with squalamine analogs, in which the position of the polyamine chain was exchanged with the position of the sulfonate group, but nevertheless retained the antimicrobial properties (MIC of 16 µg ml−1 against E. coli ). Several models have been suggested for pore formation by squalamine. Regen and coworkers described a model in which squalamine monomers embedded in one leaflet of the bilayer are in equilibrium with bilayer-spanning dimers.76 These dimers create pores which can transport ions and are therefore lethal to the cell (Figure 14b). The previous discussed cholic acid can also serve as a squalamine scaffold. If, for instance, cholic acid is functionalized with guanidine moieties on the hydroxyl positions, higher antimicrobial activities as compared to squalamine were found (MIC = 0.31 µg ml−1 against E. coli ).77, 78 Inspired by these natural antibiotics, researchers have also investigated facial amphiphiles as pore-forming agents and antibiotic activity. In 2000, Gellman and coworkers investigated nonnatural amphiphilic helices made from βamino acids for antibiotic activity.79, 80 The advantage of β-amino acids over natural amino acids is their increased stability to proteases, because they do not occur in nature.81 Facial amphiphilic helices were designed by taking an

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water

11

OSO3 R

3

C A R1

D

B R2

H

(b) Monolayer leaflet

Ion channel

OSO3

R3 R2 R1 R1 =

(a)

H N

HN

Squalamine

NH2 R2

R

OH

H

3

Figure 14 (a) Squalamine, a natural compound and although facial amphiphilicity is not directly obvious it is proposed to be induced by bilayers (b). (Reproduced with permission from Ref. 76.  American Chemical Society, 1998.)

O

NH

NH

NH N H2 +

O

O

NH

+

O

NH2

NH

NH N H2

O

O

3

NH

O

O

H

i +3 H

i +1

i

i +2

i +4 H

Figure 15 Sequence of β-17 with a hydrophilic cationic residue on position i and i+2. On self-assembly, a facial amphiphilic helix is formed (hydrophilic light gray and hydrophobic dark gray). (Reproduced with permission from Ref. 79.  Nature Publishing Group, 2000.)

hydrophilic residue for every first and third residues, because helices composed of β-amino acids have five residues per two turns (Figure 15). The antibiotic activity of this facial amphiphile (β-17) was found to be comparable to synthetic antibiotic magainin derivatives against different

bacteria. For instance, against E. coli a MIC of 6.3 µg ml−1 was found. Another successful example of an antimicrobial βpeptide is described by DeGrado and coworkers.82 This synthetic peptide consists of a repeating sequence of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

12

Soft matter

NH2

H N

S

NH2

NH2

H

H N

S

H

N

N

N

H N

N

S

H NH

H O

O

O

O

n

n

(a)

(b)

NH2

NH2

H

H N

S

H N

H N

S

H N

NH2

H N

S

H N

H

O n

O

(c)

Figure 16 (a) A rigid polyamide with high antimicrobial activity. (b) Rigidity is induced by hydrogen bonding. (c) Additional rigidity is acquired if the benzene ring is replaced with a hydrogen bond accepting pyrimidine ring. A rigid facial amphiphilic polyurea with high antimicrobial activity against E. coli .

hydrophobic β-leucine, hydrophobic β-valine, and hydrophilic β-lysine. On folding, with three residues per turn, a facially amphiphilic helix is formed. Cell penetrating activity was measured against E. coli and human erythrocytes as bacterial and mammalian cell models, respectively. The 12residue helix showed the highest selectivity for the bacterial cell with MIC values of 1.1 µg ml−1 . These examples illustrate the possibility of mimicking biological antimicrobial alpha helices. DeGrado and coworkers also described an extensive investigation of the effect of the rigidity of the backbone of the polymer on its facial amphiphilic character. These studies are based on aryl amide oligomers containing hydrophilic amines and hydrophobic t-butyl-groups (Figure 16a). Owing to intramolecular hydrogen bonding between the thioether and the amides, the polymer backbone becomes rigid and the hydrophilic side groups remain on one face, resulting in a facial amphiphile.83 Antimicrobial studies showed that oligomers with n = 8 possessed high antimicrobial activity (MIC of 7.5–15 µg ml−1 against E. coli ).84 In an attempt to make the oligomers more rigid, and thereby increase their facial amphiphilic character, a polyamide was synthesized similar to the previous described compound. In this case, however, the benzene ring was replaced with a pyrimidine, bearing two additional

NH3CI

NH3CI NH3CI

Br Br 1 or 2

Figure 17 A rigid oligophenylene ethylene showing antibiotic behavior with a MIC of 0.1 µg ml−1 against E. coli .

nitrogen atoms and capable of forming hydrogen bonds (Figure 16b).85 For n = 1, the triaryl compound showed the highest antimicrobial activity, with a more than 10-fold lower MIC (MIC = 0.8 µg ml−1 against E. coli ) than the less rigid triaryl thioether analog described above. Moreover, this monomer showed high selectively toward the E. coli cells with respect to human red blood cells. The high antibiotic activities of these relatively simple facially amphiphilic oligomers prompted researchers to further explore facial amphiphiles with rigidified oligomer backbones as possible antibiotics. For instance, a similar system based on urea-linked oligomers was investigated by Tew et al. (Figure 16c).86 These compounds show

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water comparable antimicrobial activities (MIC = 0.7 µg ml−1 against E. coli ). Another attempt to increase the rigidity of these compounds by Tew and coworkers used oligophenylene ethylene (Figure 17). These oligomers form ordered layers at air–water interfaces and also displayed antimicrobial behavior.87 Again, the triaryl compound showed the highest antibiotic behavior against E. coli (MIC = 0.1 µg ml−1 )88 and several other antibiotic-resistant bacteria such as methicillin resistant Staphylococcus aureus (MRSA bacteria) and Enterococcus faecalis. This study shows that rigidity of the polymer backbone, analogous to the rigidity induced by hydrogen bonding in alpha helices, is of high importance for antibiotic activity. Next to that, in most examples, the triaryl shows the highest antibiotic activity. This observation suggests that the match between the length of these triaryl molecules and the bilayer thickness allow them to span the bilayer, whereas the longer oligomers might fail to do so because they are too long. It is tempting to speculate that these relatively short and rigid molecules form transmembrane pores similar to other rigid facial amphiphiles, but more insights into the origin of the antibiotic activity are required. Not all facial amphiphilic amphiphiles need support from a bilayer to induce their folding into a channel. In the case of a polycholate, the apolar environment provided by solvents such as hexane and CCl4 induces its folding to hollow helices due to intramolecular interactions between the slightly concave hydrophilic faces (Figure 18a). Most likely, the solvophobic effect also contributes to the driving force for their folding. These helices have a hydrophilic interior with a pore size of roughly 1 nm and a hydrophobic exterior in apolar solvents (Figure 18b).89, 90

O H2N

HO HO

NH

HO HO

O

CH3

n

O

(a) Folding 1.5 nm

Unfolding

(b)

Side view

1 nm Top view

Figure 18 (a) Molecular structure of oligocholate which, on intramolecular self-assembly, forms helices with a hydrophilic interior (b). (Reproduced with permission from Ref. 89.  American Chemical Society, 2005.)

6

13

MORPHOLOGICAL SENSING AND CONTROL BY FACIAL AMPHIPHILIC STRUCTURES

In addition to the well-known antibiotic and pore-forming properties, facial amphiphilic proteins are also involved in several nonlethal functions in connection to bilayer membranes, especially in sensing and controlling membrane curvature. Membrane curvature plays an important role in essential processes such as cell division, exocytosis and endocytosis, vesicle trafficking, and motility. Unlike in model systems such as vesicles and liposomes in which the bilayer membrane adopts a minimal curvature, the internal membranes of cells bear regions with high (local) curvatures, such as tubules, pores, bends, flagella, and so on. Moreover, natural bilayer membranes are continuously changing shape and curvature, because of the high turnover of the processes they are involved in,91 and because many of these processes are related to membrane curvature, the ability to sense and regulate membrane curvature is of primary importance for living cells. Several different proteins for the sensing and regulation of membrane curvature are known, and not all of them involve facial amphiphilic peptides. For instance, membrane curvature can be induced by the pushing or pulling force by the cytoskeletal network or by the insertion of entire proteins which have intrinsic curvature.92 A well-studied class of facial amphiphiles used in curvature sensing are facial amphiphilic alpha-helical peptides, which are part of a variety of proteins involved in, for example, protein coat assembly and disassembly.93 Membrane curvature sensing by facial amphiphilic alpha helices can be explained by the following model. In case of a membrane curvature sensing facial amphiphilic alphahelical domain, the alpha helix generally bears a small net positive charge, in order to interact with the negatively charged phospholipid bilayer. (Highly) curved membranes are known to come with defects in the phospholipid bilayer resulting in packing stress and curvature strain (Figure 19c).1 The hydrophobic face of the facial amphiphilic alpha helix is proposed to incorporate shallowly within these defects thereby alleviating the curvature strain (Figure 19b).93, 94 This hypothesis is further supported by the observation that the ability to sense curvature is affected by the composition of the bilayer membrane. The presence of lyso-lipids, known to incorporate into bilayer defects at convex interfaces, drastically diminishes the curvature sensing ability of membrane curvature sensing proteins.95 A second membrane curvature sensing motif is the Bin/Amphiphysin/Rvs (BAR) domain, which is found in, for instance, amphiphysin and involved in endocytosis, and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

14

Soft matter + AH

+

+

+

+ (a)

(c)

+

AH

+

Low curvature

−

+

+

+

(b)

High curvature

Figure 19 (a) The BAR domain a crescent-shaped protein domain with a cationic concave surface. (b) Facial amphiphilic alpha helices can incorporate in lipid defects alleviating this membrane stress, which is the mechanism for curvature sensing of these facial amphiphiles. (c) Membrane curvature results in defects in the bilayers, resulting in membrane stress. (Reproduced with permission from Ref. 99.  Elsevier, 2010.)

is suggested to be part of a much wider range of proteins including nexins, centaurins, and oligophrenins.96 These domains are able to sense membrane curvature and preferentially bind to curved membranes, and at higher protein concentration these domains can also deform the membrane and induce curvature,97 essential, for example, vesicle budding, and tubulation. The exact sensing mechanism of membrane curvature sensing of these BAR domains is under discussion. The BAR domain is a crescent or bananashaped quaternary protein structure, having positive charges on the concave surface (Figure 19a). Until recently, membrane sensing has been thought to occur by inserting the cationic charges on the membrane surface and the binding constant depends on the similarity between the membrane curvature and the intrinsic curvature of the protein. However, recent studies have shown that mutations changing either the charge of the concave surface or the curvature of the crescent do not change the membrane sensing properties.98 Moreover, the preferred membrane curvature sensed by these mutated proteins does not change either.

BAR domains are also often found in combination with N-terminal facial amphiphilic helices (making them NBARs) and several studies have shown that mutations on these facial amphiphilic helices change the membrane sensing ability drastically. These observations led to the suggestion that membrane sensing by these N-BAR domains actually takes place via a mechanism, which involves facial amphiphilic alpha helices. In fact, it has been suggested that all membrane sensing BAR domains would contain such adjacent facial amphiphilic alpha helices.99 After sensing the membrane curvature with the facial amphiphilic alpha helix, the BAR domain can induce curvature by the overall banana-shaped structure (Figure 20).100 An example of a membrane curvature inducing protein without a BAR domain is epsin, which is known to assists in membrane deformation during clathrin-mediated endocytosis.101 Epsin is a well-characterized member of accessory proteins and its primary function is to add bending stress to membranes which induces curvature, necessary for vesicle formation. A specific domain of this protein, the ENTH domain, can bind specifically to phosphatidylinositol-4,5-bisphosphate [PtdIns(4,5)P2 ], a membrane phospholipid. On binding, unstructured residues at the N-terminus of the ENTH domain fold to a facial amphiphilic alpha helix. This facial amphiphilic alpha helix inserts into the inner leaflet of the bilayer and acts as a wedge pushing aside the surrounding phospholipids. To overcome the induced local bilayer stress, the total membrane reorients and bends at the insertion site of the protein (Figure 21).102, 103 In both membrane curvature sensing and inducing by facial amphiphilic helices, the mechanism has to do with incorporation of the helix into the bilayer. In the case of membrane curvature sensing incorporation is driven by alleviating the membrane stress, present in curved membranes. In the case of membrane curvature inducing proteins a facial amphiphilic alpha helix is forced into the bilayer, for example, binding of a protein domain to the phospholipids, which results in bilayer curving.

300 ns r = 54 nm

(a)

(b)

Figure 20 All atom simulations of inducing membrane curvature by N-BAR domain. (a) The N-BAR domain can bind a flat membrane with facial amphiphilic alpha helices after which it can bend the membrane with a radius of 54 nm (b). (Reproduced with permission from Ref. 100.  Elsevier, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water

Outer

Inner

PtdIns(4,5)P2 a0 helix ENTH

(a)

Outer

Inner

a0 helix

PtdIns(4,5)P2 ENTH

15

including emulsification and solubilization of hydrophobic compounds, as antibiotics and pore-forming agents in bilayer membranes, and as membrane curvature sensing and regulating agents. In supramolecular chemistry, facial amphiphiles have been synthesized ranging from low molecular weight to polymers and dendrimers for the purpose of novel building blocks for supramolecular architectures and to mimic the natural functions of facial amphiphiles. This approach has led to novel building blocks for common morphologies such as vesicles and micelles, and also new architectures, such as coils and complex networks. Many of these synthetic facial amphiphiles have similar properties as their natural counterparts with regard to emulsification and solubilization, and their interaction with bilayer membranes. Therefore facial amphiphiles are gaining more and more interest as new antibiotics and toxins, and transport channel agents.

(b) Outer Inner

PtdIns(4,5)P2 a0 helix ENTH

(c)

Figure 21 Epsin-mediated membrane curvature inducing. (a) Membrane curvature induced by incorporation of a facial amphiphilic alpha helix. (Reproduced with permission from Ref. 101.  Elsevier, 2007.)

7

CONCLUSIONS

In this article, an overview is given concerning facial amphiphiles in nature, bio-, and supramolecular chemistry. In facial amphiphiles, the hydrophilic and hydrophobic parts are separated by the longitudinal axis of the molecule, in contrast to head–tail amphiphiles in which the hydrophilic and hydrophobic groups each occupy one end of an elongated molecule. This different distribution of hydrophilic and hydrophobic groups with respect to the overall molecular shape leads to a number of properties and functions which are unique for facial amphiphiles. In nature facial amphiphiles occur mainly as low molecular weight compounds such as the bile acids, and as a variety of peptides, alone or as part of larger proteins. Interestingly, these peptides gain their facial amphiphilicity from folding into alpha-helical or beta-sheet secondary structures. In nature these compounds serve a variety of functions,

REFERENCES 1. J. N. Israelachvili, D. J. Mitchell, and B. W. Ninham, J. Chem. Soc., Faraday Trans. 2, 1976, 72, 1525. 2. Y. Cheng, D. M. Ho, C. R. Gottlieb, et al., J. Am. Chem. Soc, 1992, 114, 7319. 3. D. W. Russel and K. D. R. Setchell, Biochemistry, 1992, 31, 4737. 4. E. Virtanen and E. Kolehmainen, Eur. J. Org. Chem., 2004, 3385. 5. A. S. Ladokhin and S. H. White, J. Mol. Biol., 1999, 285, 1363. 6. A. Tossi, L. Sandri, and A. Giangaspero, Pept. Sci., 2000, 55, 4. 7. M. Jackson, H. H. Mantsch, and J. H. Spencer, Biochemistry, 1992, 31, 7289. 8. K. S. Rotondi and L. M. Gierasch, Biopolymers, 2006, 84, 13. 9. J. Blazyk, R. Wiegand, J. Klein, et al., J. Biol. Chem., 2001, 276, 27899. 10. A. Coello, F. Meijide, E. Rodr´ıguez Nu˜nez, and J. V´azquez Tato, J. Phys. Chem., 1993, 97, 10186. 11. P. van Rijn, T. J. Savenije, M. C. Stuart, and J. H. van Esch, Chem. Commun., 2009, 2163. 12. P. van Rijn, A. M. Brizard, M. C. Stuart, and R. Eelkema, Chem. Eur. J., 2010, 45, 13417. 13. E. K. Chung, E. Lee, Y. B. Lim, and M. Lee, Chem. Eur. J., 2010, 16, 5305. 14. V. Percec, D. A. Wilson, P. Leowanawat, et al., Science, 2010, 328, 1009. 15. G. Li and L. B. McGown, J. Phys. Chem., 1994, 98, 13711. 16. A. Coelle, F. Meijide, E. R. Nunez, and J. V. Tato, J. Pharm. Sci., 1996, 85, 9.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

16

Soft matter

17. P. Hansson, B. Jonsson, C. Strom, and O. Soderman, J. Phys. Chem. B, 2000, 104, 3496.

46. D.-J. Hong, E. Lee, and M. Lee, Chem. Commun., 2007, 1802.

18. C. Th´evenot, B. Grassl, G. Bastiat, and W. Binana, Colloids Surf. A, 2005, 252, 105.

47. E. Ros, Atherosclerosis, 2000, 151, 357.

19. C. Carnero Ruiz and J. Aguiar, Mol. Phys., 1999, 97, 1095.

48. H. E. Gallo-Torres, Lipids, 1970, 5, 379.

20. A. B. Mandal, B. U. Nair, and D. Ramaswamy, Langmuir, 1988, 4, 736.

49. J. M. Neugebauer, Detergents: an overview, in Guide to Protein Purification, ed. M. P. Deutscher, (Methods in Enzymology, vol. 182), Academic Press, San Diego, 1990.

21. S. Mukhopadhyay and U. Maitra, Curr. Sci., 2004, 87, 1666.

50. F. M. Menger and J. L. Sorrells, J. Am. Chem. Soc., 2006, 128, 4960.

22. R. Shaw, W. H. Elliott, and B. G. Barisas, Mikrochim. Acta, 1991, 3, 137.

51. Y. Pouny, D. Rapaport, A. Mor, et al., Biochemistry, 1992, 31, 12416.

23. B. R. Simonovi´c and M. Momirovi´c, Microchim. Acta, 1997, 127, 101.

52. E. Gazit, A. Boman, H. G. Boman, and Y. Shai, Biochemistry, 1995, 34, 11479.

24. K. Suzuki, T. Hasegawa, Y. Takamura, et al., Langmuir, 1996, 12, 5536.

53. Z. Oren, J. C. Lerman, G. H. Gudmundsson, Biochem. J., 1999, 341(Pt 3), 501.

25. H. M. Willemen, L. C. P. M. de Smet, A. Koudijs, et al., Angew. Chem. Int. Ed, 2002, 41, 4275.

54. Z. Oren and Y. Shai, Biopolymers, 1998, 47, 451.

26. U. Taotafa, D. B. McMullin, S. C. Lee, et al., Org. Lett., 2000, 2, 4117.

et al.,

55. Y. Shai, Biochim. Biophys. Acta., 1999, 1462, 55. 56. M. Zasloff, Proc. Natl. Acad. Sci. USA, 1987, 84, 5449.

27. P. B. Savage, Eur. J. Org. Chem., 2002, 759.

57. A. Tossi, L. Sandri, and A. Giangaspero, J. Pept. Sci., 2000, 55, 4.

28. H. M. Willemen, T. Vermonden, A. Koudijs, et al., Coll. Surf. A: Phys. Chem. Eng. Aspects, 2003, 218, 59.

58. B. Christensen, J. Fink, R. B. Merrifiel, and D. Mauzerall, Proc. Natl. Acad. Sci. USA, 1988, 85, 5072.

29. Z. Zhong, J. Yan, and Y. Zhao, Langmuir, 2005, 21, 6235.

59. Y. Shai, J. Fox, C. Caratsch, et al., FEBS Lett., 1988, 242, 161.

30. Y. Hishikawa, R. Watanabe, K. Sada, and M. Miyata, Chirality, 1998, 10, 600. 31. H. M. Willemen, T. Vermonden, A. T. M. Marcelis, and E. J. R. S¨udholter, Eur. J. Org. Chem., 2001, 2329. 32. H. M. Willemen, T. Vermonden, A. T. M. Marcelis, and E. J. R. S¨udholter, Langmuir, 2002, 18, 7102.

60. D. Rapaport, R. Peled, S. Nir, and Y. Shai, Biophys. J., 1996, 70, 2502. 61. S. J. Ludtke, K. He, W. Heller, et al., Biochemistry, 1996, 35, 13723. 62. M. R. Wenk and J. Seelig, Biochemistry, 1998, 37, 3909.

33. J. A. A. W. Elemans, R. R. J. Slangen, A. E. Rowan, and R. J. M. Nolte, J. Incl. Phenom. Macro., 2001, 41, 65.

63. K. Matsuzaki, O. Murase, N. Fujii, and K. Miyajima, Biochemistry, 1995, 34, 6521.

34. J. A. A. W. Elemans, R. R. J. Slangen, A. E. Rowan, and R. J. M. Nolte, J. Org. Chem., 2003, 68, 9040.

64. L. Silvestro, K. Gupta, J. N. Weiser, and P. H. Axelsen, Biochemistry, 1999, 38, 3850.

35. A. Wu, P. Mukhopadhyay, A. Chakraborty, et al., J. Am. Chem. Soc., 2004, 126, 10035.

65. A. Mor, K. Hani, and P. Nicolas, J. Biol. Chem., 1994, 269, 31635.

36. L. Isaacs, D. Witt, and J. Lagona, Org. Lett., 2001, 3, 3221.

66. J. Verdon, N. Girardin, C. Lacombe, et al., Peptides, 2009, 30, 817.

37. W.-Y. Yang, J.-H. Ahn, Y.-S. Yoo, et al., Nature Mater., 2005, 4, 399. 38. W.-Y. Yang, E. Lee, and M. Lee, J. Am. Chem. Soc., 2006, 128, 3484. 39. D. E. Discher and A. Eisenberg, Science, 2002, 297, 967. 40. (a) N. Reitzel, D. R. Greve, K. Kjaer, et al., J. Am. Chem. Soc., 2000, 122, 5788; (b) D. R. Greve, N. Reitzel, T. Hassenkam, et al., Synth. Met., 1999, 102, 1502.

67. A. J. Verkleij, R. F. A. Zwaal, B. Roelofsen, Biochim. Biophys. Acta, 1973, 323, 178.

et al.,

68. T. D. Brock, Biology of Microorganisms, 2nd edn, Prentice Hall Inc., Englewood Cliffs, NJ, 1974. 69. A. P. Heuck, R. K. Tweten, and A. E. Johnson, Biochemistry, 2001, 40, 9065.

42. L. Arnt and G. N. Tew, Langmuir, 2003, 19, 2404.

70. (a) A. Olofsson, H. Hebert, and M. Thelestam, FEBS Lett., 1993, 319, 125; (b) P. J. Morgan, S. C. Hyman, A. J. Rowe, et al., FEBS Lett., 1995, 371, 77; (c) A. P. Heuck, R. K. Tweten, and A. E. Johnson, Biochemistry, 2001, 40, 9065.

43. E. K. O’Shea, J. D. Klemm, P. S. Kim, and T. Alber, Science, 1991, 254, 539.

71. J. Rossjohn, S. C. Feil, W. J. McKinstry, et al., Cell , 1997, 89, 685.

44. L. Gonzalez, D. N. Woolfson, and T. Alber, Nat. Struct. Mol. Biol., 1996, 3, 1011.

72. O. Shatursky, A. P. Heuck, L. A. Shepard, et al., Cell , 1999, 99, 293.

45. A. Klaikherd, B. S. Sandanaraj, D. R. Vutukuri, and S. Thayumanavan, J. Am. Chem. Soc., 2006, 128, 9231.

73. A. P. Heuck, R. K. Tweten, and A. E. Johnson, Biochemistry, 2001, 40, 9065.

41. R. B. Breitenkamp, L. Arnt, and G. N. Tew, Poly. Adv. Technol., 2005, 16, 189.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-assembly of facial amphiphiles in water

17

74. K. S. Moore, S. Wehrli, H. Roder, et al., Proc. Natl. Acad. Sci. USA, 1993, 90, 1354.

89. Y. Zhao and Z. Zhong, J. Am. Chem. Soc., 2005, 127, 17894.

75. G. Deng, T. Dewa, and S. L. Regen, J. Am. Chem. Soc., 1996, 118, 8975.

90. Y. Zhao, Z. Zhong, and E.-H. Ryu, J. Am. Soc., 2006, 129, 218.

76. M. Merritt, M. Lanier, G. Deng, and S. L. Regen, J. Am. Chem. Soc., 1998, 120, 8494.

91. R. Parthasarathy and J. T. Groves, Soft Matter, 2007, 3, 24. 92. H. T. McMahon and J. L. Gallop, Nature, 2005, 438, 590.

77. C. Li, M. R. Lewis, and P. B. Savage, Antimicrob. Agents. Chemother., 1999, 43, 1433–1439.

93. N. S. Hatzakis, V. K. Bhatia, J. Larsen, et al., Nat. Chem. Biol., 2009, 5, 835.

78. P. B. Savage, C. Li, U. Taotafa, et al., FEMS Microbiol. Lett., 2002, 217, 1.

94. W. S. Davidson, A. Jonas, D. F. Clayton, J. M. George, J. Biol. Chem., 1998, 273, 9443.

and

79. E. A. Porter, X. Wang, H. S. Lee, et al., Nature, 2000, 404, 565.

95. S. M. Davies, R. M. Epand, R. Kraayenhof, R. B. Cornell, Biochemistry, 2001, 40, 10522.

and

80. E. A. Porter, B. Weisblum, and S. H. Gellman, J. Am. Chem. Soc., 2002, 124, 7324.

96. B. J. Peter, H. M. Kent, I. G. Mills, et al., Science, 2004, 303, 495.

81. J. Frackenpohl, P. I. Arvidsson, J. V. Schreiber, D. Seebach, Chembiochem, 2001, 2, 445.

and

97. J. Zimmerberg and M. M. Kozlov, Nat. Rev. Mol. Cell. Biol., 2006, 7, 9.

82. Y. Hamuro, J. P. Schneider, and W. F. DeGrado, J. Am. Chem. Soc., 1999, 121, 12200.

98. V. K. Bhatia, K. L. Madsen, P. Y. Bolinger, et al., EMBO J., 2009, 28, 3303.

83. D. Liu, S. Choi, B. Chen, et al., Angew. Chem. Int. Ed., 2004, 43, 1158.

99. K. L. Madsen, V. K. Bhatia, U. Gether, and D. Stamou, FEBS Lett., 2010, 584, 1848.

84. G. N. Tew, D. Liu, B. Chen, et al., Proc. Natl. Acad. Sci. USA, 2002, 99, 5110.

100. Y. Yin, A. Arkhipov, and K. Schulten, Structure, 2009, 17, 882.

85. H. Tang, T. J. Doerksen, T. V. Jones, et al., Chem. Biol., 2006, 13, 427.

101. C. A. Horvath, D. Vanden Broeck, G. A. Boulet, et al., Int. J. Biochem. Cell. Biol., 2007, 39, 1765.

86. H. Tang, R. J. Doerksen, G. N. Tew, Chem. Commun., 2005, 1537.

102. M. G. Ford, I. G. Mills, B. J. Peter, et al., Nature, 2002, 419, 361.

87. L. Arnt, K. N¨usslein, and G. N. Tew, J. Pol. Sci. Part A: Pol. Chem., 2004, 42, 3860.

103. H. T. McMahon, M. M. Kozlov, and S. Martens, Cell , 2010, 140, 601.

88. G. N. Tew, D. Clements, H. Tang, et al., Biochim. Biophys. Acta, 2006, 1758, 1387.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc144

Self-Assembly of Surfactants at Solid Surfaces Michele Ferrari CNR-Institute for Energetics and Interphases, Genova, Italy

1 Introduction 2 Theoretical Modeling and Simulations 3 Experimental Techniques and Properties 4 Applications 5 Conclusions References

1 1.1

1 4 5 12 21 22

INTRODUCTION Surfactants, adsorption, and self-assembly

Surface-active agents are probably the most popular chemicals in our daily life. These molecules can be described as composed of two parts of different behavior with respect to the solvent in which they are dissolved. While one of them is generally a hydrocarbon chain with alkyl or even more complex hydrophobic character, the other has a more defined behavior containing one or more hydrophilic group, ionic or nonionic, which can be specifically tailored in order to show suitable properties. This so-called amphiphilic character is the key feature for their such wide exploitation. Better known as surfactants, they are commonly found in soaps, detergents, and housekeeping cleaning agents. Their addition to washing water results in a significant lowering of its surface tension, improving the wetting and enhancing its spreading properties.1 It would be difficult to give an exhaustive view of the presence of surfactants in industry, since from oil extraction

to cosmetics and from mining to food, production processes are often related to surface or interfacial phenomena of disperse systems such as foams or emulsions. This large flexibility in their use is due to their amphiphilic molecular structure, which can mainly range from a straight, linear alkyl chain to more complex double or cholesterol derivative hydrophobic moieties with one or more conventional or innovative hydrophilic groups being introduced to tune their properties according to a specific required performance (Figure 1). If we classify surfactants on the basis of the ionic properties (electric charge) of the hydrophilic head in water, they can be grouped into four categories: anionic, nonionic, cationic, and amphoteric or zwitterionic. As organic compounds with an amphiphilic structure, they show solubility in both oils or organic solvents and water; and, in fact, with respect to water, because these molecules have one hydrophilic and one hydrophobic part, they are pushed toward the areas of contact, concentrating at the interfaces between air (or oil) and water through the adsorption process. Further insight on the self-assembly and self-organization including surfactants in liquids can be found in Introduction to Surfactant Self-Assembly, Concepts, Soft Matter Science—a Historical Overview with a Supramolecular Perspective and Self-Assembly of Facial Amphiphiles in Water, Soft Matter, respectively. We can generally give a description of adsorption as the tendency of the molecule of a surfactant to collect at an interface. Liquids or gases can be taken up at a solid substance involving different kinds of molecular attractions. The adsorption properties of surfactants are related to how the molecules reach and are generally found at the interface between a water and an oil phase and an air and a water phase. This molecular property results in the macroscopic properties of wetting, detergency, foaming, and emulsion formation. The molecules of a surface-active

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

2

Soft matter

HO

O O S O Na

O Na

H HO

Chemical or trade name Sodium dodecylsulfate (SDS)

OH

H

Sodium cholate HO

N

O

CH3 H

O

CH3

CH3 CH3

O

CH2

O

CH3 CH3

O

CH3 H

Na

O Na

H

O HO

N -Lauroylsarcosine sodium salt

H Sodium deoxycholate (DOC)

CH3 CH3 CH3 Br N

O

S O

O O

Cetyltrimethylammoniumbromide (CTAB)

O Na

O O

CH3 N

Bis(2-ethylhexyl) sulfosuccinate sodium salt

O

CH3 Lauryldimethylamine-oxide (LDAO)

Figure 1 Examples of surfactant molecular structure evidencing the wide range of both hydrophilic and hydrophobic moieties. (Reproduced from Ref. 2, Detergents and Soaps.com.)

agent tend to adsorb onto the surface of oil droplets and, while the hydrophilic heads stick out into the water phase, the hydrophobic tails stick into the oil phase.2 Surfactant adsorption at solid surfaces is, in practice, exploited to facilitate detergency, control wetting and penetration of solutions, stabilize foams and emulsions, and collect minerals in flotation operations. From this list alone, it is clear that this field is of tremendous importance. Consequently, significant effort has been directed toward acquiring a better understanding of the adsorptive nature of different surfactant molecules at solid surfaces.3 According to forces or bonding effects involved, we briefly describe the mechanisms of surfactant adsorption that have been introduced to explain how surface-active molecules may adsorb onto the solid substrates from aqueous solutions generally involving single molecules or oligomeric aggregates rather than micelles4, 5 : • Ion exchange occurs when similarly charged surfactant ions replace counterions adsorbed onto the substrate from the solution. • Ion pairing is involved when surfactant ions adsorb from solutions onto oppositely charged sites free of counterions.

• •

•

Hydrophobic bonding can be described as an attraction between a hydrophobic group of adsorbed molecules and a molecule present in the solution. In the presence of electron-rich aromatic nuclei in the surfactant, hydrophobic tail attraction between electronrich aromatic nuclei of the adsorbate and positive sites on the adsorbent results in adsorption by polarization of π electrons. Adsorption by dispersion forces such as the London–van der Waals force between the adsorbate and the adsorbent increases with the increasing molecular weight of the adsorbate.

In addition to the adsorption process, in which the molecules reach the interface depending on their structure and relationship with the solvents, amphiphilic molecules show the tendency to organize and coordinate themselves into ordered structures in water or solvent including the formation of aggregates such as micelles, liquid crystals (LCs), or bilayers. Such self-assembly phenomena can be described when the hydrophobic tails of surfactant molecules form a cluster to produce small aggregates, such as micelles, or large layered structures such as bilayers that are similar to a cell wall.6

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces 3

(b)

Figure 2 Surfactant micelles in water environment (a) and bilayer (b) (Reproduced from Ref. 2, Detergents and Soaps.com.)

During the micellization process in water, typically tens to a few hundreds of surfactant molecules organize in a micelle, the hydrophobic tails getting far from the water forming a core, which can encapsulate an oil droplet, while the hydrophilic heads (ionic or non-ionic) remain oriented towards the water, producing an outer shell that is in contact with the aqueous environment (Figure 2). When the surfactants assemble in oil, their aggregate is referred to as a reverse micelle (RM), where on the contrary the heads remain in the core and the tails maintain contact with the oil. The concentration at which surfactants start forming micelles is known as the critical micelle concentration or cmc. Above the cmc, spherical or cylindrical structures are present, while a further increase in concentration can result in the self-organization of micelles into periodic hexagonal, cubic, or lamellar mesophases. Depending on their molecular structure, surfactants can also aggregate to develop and produce extended structures in water such as bilayers. Giving a more general definition of self-assembly, we describe it as a spontaneous organization of materials through noncovalent interactions without an external action involved: asymmetric molecules usually self-assemble, being preprogrammed to organize into well-defined supramolecular assemblies through hydrogen bonding, van der Waals forces, electrostatic forces, π –π interactions, and so on.

I

II A

B

C

III

IV i

ii

V (a)

i A

ii

B

C V

Γ

(a)

As a key characteristic feature, self-assembly influences the role played by surfactant molecules at the surface; their surface activity and the understanding of the structure that surfactants adopt at surfaces is vital for determining their role in practical applications. Importantly, it has been found that surfactants often self-assemble at hydrophilic surfaces to form quasi-twodimensional analogs of the aggregate structures observed in a bulk solution, that is, spherical or cylindrical surface micelles or bilayer-type structures, whereas at hydrophobic surfaces they tend to form monolayers or hemimicellar aggregates. The thermodynamics of surface-active agents representing systems between the ordered and disordered states of matter is therefore of great importance theoretically as well as practically: recent works have in fact generated a relatively detailed and consistent picture of the adsorption mechanisms for different surfactant–substrate combinations along with the way in which structural layer properties can be correlated to general characteristics of the corresponding adsorption isotherms (Figure 3). The progress in this area is, to a significant degree, linked to the exploitation of specialized experimental techniques such as neutron reflection, ellipsometry, fluorescence

IV III I II (b)

III I II

IV III I II

Concentration

Figure 3 Adsorption of nonionic surfactants showing the orientation of the molecules at the interface (a) at different stages and in relationship with the isotherms. (Reprinted from Adv. Coll. Int. Sci. 110, S. Paria and K. C. Khilar. A review on experimental studies of surfactant adsorption at the hydrophilic solid–water interface, 75  2004, with permission from Elsevier.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

4

Soft matter

spectroscopy, and also atomic force microscopy (AFM) for the characterization of adsorbed layer properties at solid surfaces. The presence of different interfacial force characteristics dependent on the structure and interfacial density strongly affects the technical performance, and the increased knowledge about adsorbed layer morphologies can hopefully also facilitate important new methods for the fabrication of nanostructured functional surfaces: in fact, synthetic methods already allow the tailoring of specific molecules in order to fulfill new requirements and aims at a wider range of industrial applications such as skincare, antibacterial properties, vesicle formation, high porosity materials, and so on.

2

THEORETICAL MODELING AND SIMULATIONS

From the theoretical point of view, some attempts have been made to describe the behavior of ionic and nonionic surfactants organized at different solid surfaces. In these works, the self-assembly of ionic and nonionic surfactants on both hydrophilic and hydrophobic surfaces has been investigated by computing the equilibrium free energy for the formation of a given structure and predicting the critical surfactant concentration for the formation of surface aggregates (cac) and the surface morphology: the cac is always predicted to be lower than the bulk phase critical micelle concentration. The key feature of these works is to use, beyond the full sphere and cylinder models proposed in the literature, a series of composite structures composed of monolayers oriented with head groups in contact with the hydrophilic surface and covered with hemispheres, hemicylinders, finite disks, or another monolayer (making the full structure a bilayer), showing that, for several illustrative cases, composite hemicylinders will form at the cac.7 Since the surfactant concentration required for the formation of full spherical and full cylindrical surface aggregates is largely comparable to the bulk cmc, it is likely that the surface structures observed in AFM experiments conducted near the bulk cmc are not full spheres or cylinders, but corresponding composite structures made up of hemispheres or hemicylinders. On the other hand, a theory of surfactant self-assembly on isotropic hydrophobic surfaces is presented by extending the well-established treatment of self-assembly in solution with the aim of considering the replacement of the solid surface–water contact by the solid surface-aggregate core contact.8 A term in addition to the bulk ones, characterized by a single parameter, the displacement tension γ , has

been introduced in the free-energy model for the formation of surface aggregates. Calculations of the cac, aggregate shape, and size have been performed for anionic, zwitterionic, and nonionic surfactants, and, for all types of surfactants under investigation, the cac was always more than 1 order of magnitude lower than the bulk cmc, owing to the favorable energy contribution of the surface to the free energy of aggregation. For ionic surfactants, surface aggregates were always found to be smaller than those in the bulk phase, independently of the value of γ , while those from zwitterionic and nonionic surfactants were either smaller or larger, influenced by the balance between head group repulsion and aggregate core–solid surface attraction. Depending upon the surfactant and the solid surface hemispheres, hemicylinders, finite disks, and continuous monolayers can be predicted from the chemical potential of the surfactant on increasing the total surfactant concentration. Thus various shapes for the same surfactant on a specific solid surface can form by self-assembly, depending upon the total surfactant concentration (Figure 4). This behavior has been confirmed for weakly adsorbing nonionic surfactants on hydrophilic surfaces; in fact, despite the simple nature of the course-grained lattice model, the simulation can predict certain features that are also found experimentally.9 Even without a quantitative comparison, cac was found to be smaller than cmc, and, above the cac, a relatively large increase was observed in

X1 ≈ CAC

X1 > CAC ∼

X1 > CAC

X1 >> CAC

Figure 4 Progression of aggregate microstructures for single ionic surfactants beyond the cac. (Reprinted from Coll. Surf. A, 167, R. A. Johnson and R. Nagarajan, Modeling self-assembly of surfactants at solid : liquid interfaces. I. Hydrophobic surfaces, 31  2000, with permission from Elsevier.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces 5 15

2

10

1

5

H

q ads

3

0

0.01

0.1

1

0

φb

Figure 5 Adsorbed amount theta (circles) and layer thickness (squares) as a function of the bulk volume fraction. (Reprinted with permission from C. M. Wijmans and P. Linse, J. Phys. Chem. 1996, 100, 125831  1996 American Chemical Society.)

the adsorbed amount. A smaller increase was predicted in the adsorbed layer thickness with increasing solution concentration and other features that were not accessible such as the shape and orientation of the adsorbed aggregates (Figure 5).

Computer simulations can also be a key tool in predicting kinetics phenomena of surfactant adsorption on hydrophilic solid surfaces: depending on the head–surface interaction, two different kinetic modes can describe the adsorption process on solid surfaces.10 For the strongly attractive head–surface interaction, four distinct regimes of surfactant adsorption exist; in particular, adsorption in the self-assembly controlled regime displays time dependence with an exponent unrelated to bulk concentrations and diffusion coefficient, whereas for weaker adsorption surfaces a stepwise mode exists. Besides the head–surface interaction, the effects of surfactant diffusivity bulk concentration, the length of the diffusion zone, and the surfactant architecture on the adsorption kinetics are also considered (Figure 6). For further studies on computational techniques, see Computational Techniques (DFT, MM, TD-DFT, PCM), Techniques.

3

EXPERIMENTAL TECHNIQUES AND PROPERTIES

From the experimental point of view, most of the studies are aimed at providing evidence of the role of surfactants

(a)

(aT)

(b)

(bT)

Figure 6 Aggregates at surfaces corresponding to perforated bilayers (only tail groups are shown) (a and aT) for H4T4 and full bilayer (b and bT) for H2T4. (Reprinted from J. Coll. Int. Sci., 313, X. Zhang, B. Chen, Z. Wang Computer simulation of adsorption kinetics of surfactants on solid surfaces 414  2007, with permission from Elsevier.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

6

Soft matter

on solid surfaces, especially at the nanoscale, which is still a challenging field, where verification of the validity and the reproducibility of more macroscopic conclusions is required. The interfacial behavior of nonionic surfactants at solid surfaces is, for example, of considerable interest due to their common and important use in processes such as colloidal stability and detergency, since they are more attractive in comparison with cationic molecules owing to their preferential adsorption to hydrophobic substrates, combined with a lower sensitivity to the presence of salt compared to anionic samples.4 From these studies, it can be highlighted that adsorption at hydrophilic silica surfaces represents a surfaceaggregation process similar to micelle formation in the bulk solution; depending on various parameters such as the surfactant composition and relative sizes of the head groups and the tails of the surfactant, surface aggregation may lead to discrete surface micelles or to fragmented or complete bilayers. Steitz et al. demonstrate some advantages in comparison with AFM studies. Combining the length scales of smallangle neutron scattering (SANS) and the surface sensitivity of grazing incidence diffraction (GID), grazing incidence small angle neutron scattering (GISANS) makes nanostructured surface and thin film studies possible without requiring any specific sample preparation other than thin film deposition. In fact, the film surface, interior, and the substrate–film interface are all accessible in a range from a few to hundreds of nanometers. GISANS beamlines can be found at many synchrotron light source facilities for the characterization of self-assembly and self-organization phenomena at the nanoscale in thin films, while future developments might foresee biological applications, such as biopolymers or viruses attached to surfaces or lipid layers. GISANS has been used to study surfactant films at a hydrophilic silicon wafer in the regime below saturation coverage, that is, in conditions not available to AFM which usually only works well with surface saturation.11 The mean surface concentration of the chosen nonionic surfactant could be varied over a wide range simply by varying the sample temperature at a constant bulk concentration somewhat below the cmc, indicating the existence of transient surfactant aggregates without a preferred structure at half-coverage of the surface. Different techniques have also been adopted to study the self-assembly of two nonionic surfactants, such as pentaethylene glycol monododecyl ether (C12E5) and ndodecyl-β-maltoside (β-C12G2), in the presence of a synthesized silica sol of uniform particle size.12 A wider range of studies including adsorption measurements, dynamic light scattering, and SANS was used to evidence the structure of the surfactant aggregates on silica

in a H2 O/D2 O mixture: SANS measurements were carried out in a range of concentrations from zero surfactant up to maximum surface coverage; the results show that the spherical core-shell model reproduces the SANS data, indicating that the model of a uniform adsorbed layer holds to a lower value of the surface area. C12E5 was found to strongly adsorb onto the silica beads, with a plateau value of the adsorption isotherm above the cmc: the model of micelle-decorated silica beads can be used to understand the scattering profiles with C12E5 adsorbed as spherical micellar aggregates that are influenced by the high surface curvature of the silica, preventing the packing of the hydrophobic chains in a bilayer configuration. On the other hand, the maltoside surfactant β-C12G2 showed very weak adsorption on the silica beads, forming oblate ellipsoidal micelles in the silica dispersion, similar to that observed without silica beads (Figure 7). In situ ellipsometry has been used to investigate the adsorption of zwitterionic dodecyl-N,N-dimethylammonium alkanoates with polymethylene intercharge arms of different lengths on silica, with common aspects being observed in the adsorption features of zwitterionic surfactants with nonionic polyoxyethylenated surfactants.13 Zwitterionic or amphoteric surfactants are an important class of molecules owing to their particular properties such as tolerance to hard water, strong electrolytes, oxidizing and reducing agents, low toxicity, and compatibility with other amphiphiles.

Rsph

d Rsi

Rsi

(a)

(b)

L

R R (c)

L (d)

R (e)

Figure 7 Cartoons of surfactant self-assembly structures on silica beads: (a) spherical core-shell model and (b) micelledecorated silica model with spherical surface micelles. Different forms of surface micelles: (c) spheres, (d) patch, and (e) wormlike. (From D. Lugo, J. Oberdisse, M. Karg, R. Schweinsc and G. H. Findenegg, Soft Matter, 2009, 5, 2928—Reproduced by permission of The Royal Society of Chemistry http://dx.doi.org/10.1039/B903024G.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces 7

1

Dewetting film velocity VR (cm s−1)

10

2

Number of successive recedence trials 3 4 5 6 7 8 9 10 11 12 13 14 15

VR (m)

1

Pyrex, 5×10−4 M DDAA Film thickness: 1.00 mm

VR(1)

0.1 0

10

20

30

40

50

Elapsed time for desorption after, VR(m), (h)

Figure 8 Nonequilibrium surfactant phenomena in dewetting. (Reprinted from J. Coll. Int. Sci., 256, D. W. Fuerstenau, Equilibrium and Nonequilibrium Phenomena Associated with the Adsorption of Ionic Surfactants at Solid–Water Interfaces, 79  2002 with permission from Elsevier.)

Owing to the weak interaction between single molecules and silica, a self-assembly process had a significant influence on surfactant adsorption, induced by the presence of the solid surface. This indicated an aggregation mode similar to the corresponding solution phase behavior, while increasing the number of methylene units between the charged groups was not found to have a significant influence on the adsorption mode, which was also independent of pH.

Contact angle (degrees)

100

80

S A

S A

The evolution of interfacial phenomena involving ionic surfactants at solid–water interfaces is particularly important in controlling the wetting behavior, suspension stability, and suspension rheology, and in this work the adsorption and association of physically adsorbed ionic surfactants on hydrophilic oxides in aqueous media and their effect on certain interfacial phenomena in these systems have been assessed by the adsorption isotherm evidencing up to four regions corresponding to the aggregation mode of adsorbed ions at the solid–water interface.14 In the first stage, surfactant adsorption initially produces hemimicelles at the interface, inducing coagulation in colloidal particles becoming hydrophobic and reaching a maximum at zero ζ potential. The zeta potential is then reversed, increasing surfactant concentration, with consequent redispersion of the particles again being hydrophilic. In these conditions, reverse hemimicelles or bilayers already form from the surfactant ions adsorbed in reverse orientation, at surfactant concentrations lower than the cmc. Owing to their kinetic energy, coarser particles remain in dispersion overcoming the energy barriers acting between them and keeping their hydrophobicity up to the cmc (Figure 8). Advancing and receding contact angle measurements, reflotation, and capillary rise in systems involving dewetting after surfactant adsorption at the solid–water interface reflect the behavior of the surfactant at the gas–solid interface, such as that shown in Figure 9. Going further into the mechanism and kinetics of monolayer self-assembly at the mineral–water interface, in situ vibrational sum-frequency spectroscopy (VSFS) has been used to investigate the adsorption, desorption, and equilibrium monomer exchange processes of sodium dodecanoate at the fluorite(CaF2 )–water interface.15

S A

Liquid Equilibrium Liquid receding advancing

60 Advancing

40 Alumina ionic strength 2 × 10−3 N

Equilibrium

20

pH 7.2 24 ± 1 °C

Receding

0

10−6

10−5

10−4

10−3

Concentration of dodecyl sulfonate (mol l−1)

Figure 9 Effect of sodium dodecyl sulfonate on the equilibrium and liquid advancing and receding contact angle on alumina. (Reprinted from J. Coll. Int. Sci., 256, D. W. Fuerstenau, Equilibrium and Nonequilibrium Phenomena Associated with the Adsorption of Ionic Surfactants at Solid–Water Interfaces, 79  2002 with permission from Elsevier.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Soft matter c1453 (normalized)

8

1.1 1.0 0.9 1.0

(b)

Bulk

0.5

kH

CaF2

f2873

Despite the role of electrostatic interactions, these results also note the importance of cooperative effects of both the water structure and surfactant hemimicelle formation at the interface (Figure 10). Desorption kinetics of partial and completed monolayers were also investigated together with the influence of the presence of electrolytes (uni- and divalent anions) in affecting the nonequilibrium kinetics of self-assembled monolayers. A further insight toward the understanding of surfactant monolayers at SL interfaces can be found in the work by Miranda et al., where the authors used SFG vibrational spectroscopy to study the effect of the liquid environment on the chain conformation (Figure 11).16 The chain conformation of monolayers with lower surface density was very sensitive to different liquid environments in contrast to monolayers with fully packed alkyl chains: a wide range of conformations were found for the monolayers showing more loosely packed features, as a result of a balance between solvophobic effects and the

(a)

0.0 (c) 1.0

f2873

kD 0.5

0.0

t0 −2000

0

2000

4000

6000

120

Time (s)

Figure 10 Monomer exchange kinetics of sodium dodecanoate at the fluorite–water interface (a) normalized nonlinear optical response of the carboxylate headgroup; (b) exchange kinetics (kH ) of a deuterated monolayer; and (c) of a hydrogenated monolayer (kD ). (Reprinted with permission from S. Schr¨odle and G. L. Richmond, J. Am. Chem. Soc., 2008, 130, 5072  2008 American Chemical Society.)

110 100 90 Air

Compared to the second harmonic generation and infrared and Raman spectroscopy, such a nonlinear, laserbased spectroscopy technique has been often applied to access structural information about molecules at gas–solid, gas–liquid, and liquid–solid interfaces such as composition, orientations, and distributions. A typical sumfrequency generation (SFG) setup is composed of two laser beams mixing at a surface and generating an output beam, the frequency of which is equal to the sum of the two input frequencies. SFG can be performed in situ on aqueous surfaces or in a gas environment without significant damage to the sample surface as, among its advantages, it is monolayer surface sensitive. Focusing on the temporal correlation of head group adsorption and alignment of the surfactant’s alkyl chain, the monolayer formation was monitored by exploiting the nonlinear optical response of the adsorbate: initially highly structured interfacial water molecules were revealed by the spectra at the clean fluorite water and, as the surfactant was added, such a structure was disrupted on starting the adsorption process and the consequent formation of a wellordered monolayer up to a negatively charged monolayer far beyond the point of electrostatic equilibrium.

SFG signal (a.u.)

80 70 C10

60 50

C12

40 30

C14

20 10

C16

0 2800

2850

2900

Frequency

2950

(cm−1

3000

)

Figure 11 SFG spectra of DOAC monolayer at the quartzdeuterated alkane interface. Each curve represents a specific tail length. (Reprinted from Thin Solid Films 161, P. B. Miranda, V. Pflumio, H. Saijo, Y. R. Shena, Surfactant monolayers at solid–liquid interfaces: conformation and interaction, 327,  1998, with permission from Elsevier.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces 9 chain–chain interaction due to polar- or hydrogen-bonding liquids inducing gauche defects in the surfactant chains. As a consequence of stronger chain-chain interactions between long chain alkanes and surfactants, the hydrophobic chains are straightened, while in the case of long chain alcohols the hydrophobic effect plays a major role compared to the chain-chain interaction, which are minor when the alcohol is diluted in a nonpolar liquid. Even if literature reports data in contrast to the weak environmental effects on “compact,” self-assembled monolayers, the influence of the liquid environment on the structure of self-assembled monolayers is relevant in the wetting properties of these monolayers. Wetting depends on how the liquid is in contact with it, because the surrounding liquid molecules influence the chain conformation.17, 18 Despite some limitations as seen above, by using AFM “soft-contact” imaging, it is possible to view the structure of adsorbed surfactant layers on solid surfaces in equilibrium with their solution state. This topic is discussed in more detail in Atomic Force Microscopy (AFM), Techniques. AFM is particularly sensitive to the lateral structure in the film, and this has yielded some surprising results, showing that adsorbed surfactant layers contain aggregates that are strikingly similar to elements of complex fluid phases including spherical or globular and cylindrical micelles, branched or mesh structures, as well as (occasionally) conventional bilayers.19 The influence of the surfactant head group on the adsorption of cationic surfactants with primary, secondary, tertiary, and quaternary ammonium bromide head groups attached to a single dodecyl hydrocarbon chain onto mica has been investigated. This shows differences in the aggregation state of the adsorbed layer and also that the size of these two-dimensional aggregates is consistent with the extent of the two-dimensional aggregation, which is the predominant driving force for adsorption.5, 20 AFM imaging techniques have been exploited to identify the size range of aggregate structures (i.e., hemimicellar structures) and to represent the physical state of the interface. The mica–water interface was chosen because the well-known, surface–oxygen atom crystal geometry could act as a calibration-like reference for the imaging of the hemimicellar structures that were visible in places where they were not covered with the surfactant, and, in addition, this kind of solid interface was well characterized from the point of view of the change in the zeta potential of the interface induced by adsorbed surfactants. The interactions between surfactant solutions and solid surfaces play a key role in technologically important processes such as colloidal stabilization, ore flotation, and soil removal; however, the interfacial aggregation of surfactant molecules is not yet well understood.

Direct images of surfactant aggregates at solid surfaces in aqueous solutions were obtained by AFM evidencing structures for quaternary ammonium surfactants (above the cmc) consistent with half-cylinders on crystalline hydrophobic substrates, full cylinders on mica, and spheres on amorphous silica, which can potentially be used to pattern surfaces at nanometer-length scales.21–23 These structures are interfacial aggregates that are surprisingly different from earlier models, appearing as a result of a compromise between the natural free curvature, as defined by intermolecular interactions, and the constraints imposed by specific surfactant–surface interactions. AFM studies can also follow morphology transitions in nonionic, surfactant-adsorbed layers near their cloud points, since temperature plays a significant role in surfactant aggregation states (Figure 12).24 By imaging the adsorbed layer structure by AFM as a function of temperature up to their cloud points, several polyoxyethylene alkyl ether (CnEm) nonionic surfactants have been investigated on silica and graphite surfaces, showing the thermal evolution of the adsorbed layer morphology. On silica, surfactants first formed globules at low temperatures, then rods and then a mesh at higher temperature. Such a structure was maintained in the two-phase region by heating the solutions at temperatures higher than the respective cloud points. On graphite, however, they formed straight and parallel hemicylinders at the examined temperatures with the only exception of C12E3, which formed a laterally unstructured bilayer. Research on industrial areas such as organic lightemitting devices, photovoltaics, and thin film transistors involves investigations about LCs due to their high carrier mobilities, anisotropic transport, and polarized emission resulting from their self-assembling properties and supermolecular structures. Despite the fact that conventional thermotropic liquids are usually composed of rodlike and disk-shaped bodies, banana-shaped LCs represent a new subfield of LCs where studies of film structure and properties of 2D assemblies of banana-shaped LC molecules on the highly oriented pyrolytic graphite (HOPG) surface can be carried out by scanning tunnel microscopy (STM) at the submolecular level.25 For further reading on supramolecular strategies in LC design, see Self-Organization and Self-Assembly in Liquid-Crystalline Materials and Liquid Crystals Formed from Specific Supramolecular Interactions, Soft Matter. STM is a suitable instrument for imaging surfaces at the atomic level, with resolution typically on the order of 0.1 nm laterally and 0.01 nm in depth. In fact, individual atoms within materials can be usually imaged and manipulated. A standard STM can not only operate in ultrahigh vacuum but also in air, water, and various other liquid

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

10

Soft matter

10 −

5

50 nm

0

5

0 nm

10

Figure 13 P-14-PIMB molecules on the HOPG surface. Highresolution STM images showing mirror chiral (banana) structures. (Reprinted with permission from J. R. Gong and L. J. Wan, J. Phys. Chem. B, 2005, 109, 18733  2005 American Chemical Society.)

(a)

the HOPG surface, with the molecule-substrate interactions playing a minor role (Figure 13).

3.1

Novel surfactants and Self assembly (SA)

3.1.1 Gemini surfactants

50 nm (b)

Figure 12 AFM deflection images showing the change in the morphology. With the temperature of C12E23 (0.2 mM) on graphite at (a) 22◦ C, a laterally unstructured monolayer was observed, (b) and at 28◦ C, hemicylinders were observed. (Reprinted with permission from A. Blom, G. G. Warr, E. J. Wanless, Langmuir, 2005, 21, 11850  2005 American Chemical Society.)

or gas environments, and in a temperature range from near zero Kelvin to a few hundred degrees Celsius. For a wider knowledge of this topic, see Scanning Tunneling Microscopy (STM), Techniques. The orientation between the top and lower layers in bilayers has been imaged by STM, evidencing a more complicated packing organization of banana-shaped LCs due to the flexible structure of the molecules and the intermolecular interaction that influences the packed structures of 1,3-phenylene bis[4-(4-n-alkylphenyliminomethyl) benzoates] (P-n-PIMB) molecules into chiral assemblies on

The concept and design of specifically tailored molecules is necessary to allow new industrial and research requirements. Introduced about 25 years ago, Gemini surfactants have since then been promising molecules, being a group of surfactant molecules possessing more than one hydrophobic tail and a hydrophilic head group (Figure 14).26 The Gemini structure can give surface and bulk properties that are different from those of conventional singlechain monomeric surfactants with the same number of carbon atoms per polar “head” group. While monomeric Tail

Tail Ion

Spacer

Ion

(a)

Interface (b)

Figure 14 Representation of Gemini surfactant molecules and their configuration at the interface. (Reprinted from S. K. Hait and S. P. Moulik Current Science, 2002, 82, 1101, with permission from Current Science.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces surfactants generally form spherical micelles, Geminis typically form threadlike aggregates. Active at lower concentrations, they show excellent foaming and wetting properties. Aqueous solutions of Gemini surfactants can solubilize neutral hydrophobic molecules more easily compared to solutions containing their monomeric surfactant counterparts. Also, a variation in the hydrophobic/hydrophilic nature and the position or flexibility of the linker can lead to dramatic effects on their physicochemical properties, due to changes in the organization of micellar structures. Better surface-active properties of the Gemini structure than those of corresponding conventional surfactants of equal chain lengths are among its advantages (surface tension reduced by 10 or more mN m−1 ). In addition, since solubilization is an important phenomenon required in tertiary oil recovery and detergency and Geminis are good solubilizers and homogenizers due to their very low cmc values, worldwide corporations such as detergent and oil-based companies, which need newer, cost-effective products, have contributed substantially to the synthesis and characterization of useful Geminis. As in conventional surfactants, the foaming properties of Gemini surfactants depend upon the chain length of the hydrophobic groups, and also on the length and nature of the spacer between them. Interestingly, Kim et al.27 have reported that C12 cationic Geminis form large amounts of foam in comparison to Dodecyl trimethyl ammonium bromide (DTAB); in some cases, even better than sodium dodecyl sulphate (SDS). Using theoretical modeling by Monte Carlo simulations on ionic and nonionic Geminis with both hydrophobic and hydrophilic spacers,28 the role of hydrophobic spacers was investigated and it was found that short spacers give nonspherical micelles, while long ones lead to rodlike micelles. Surfactants with hydrophilic spacers form spherical micelles, and the morphologies of the ionic and nonionic Geminis are identical regardless of the character of the spacer while the bending stiffness of the hydrophobic spacer was also found to increase the cmc, which instead decreases with the hydrophilic spacer.

3.1.2 Bolaamphiphiles (bolas) Research on the bolas has attracted considerable attention in the last decades since they were found in the membrane of Archaebacteria, which is able to sustain harsh environments including high salt concentrations, extreme temperatures, or very acidic conditions (Figure 15).29 The advantages of the bipolar structure of the bolas, amphiphiles with two polar heads connected by one or two long hydrophobic spacers, and the difficulty in isolating them from natural membranes, have pushed researchers toward chemical synthesis, which can mimic natural

Solution

(a) Solid

Solution

(b) Solid

11

Solution

Solid

Solution

Solid

Figure 15 Possible mechanism for adsorption (a) of bolaform cationic amphiphiles in mixed systems and (b) of single flexible bolaform cationic amphiphiles systems on glass. (Reprinted from J. Coll. Int. Sci. 337, Yun Yan, Ting Lu, Jianbin Huang Recent advances in the mixed systems of bolaamphiphiles and oppositely charged conventional surfactants, 1,  2009 with permission from Elsevier.)

molecules.30 Their surface properties show very poor surface activity and their ability to decrease the surface tension of water is in fact weaker than that of conventional surfactants due to the reverse U-shaped conformation of the bolas at the air/water interface, and it cannot be lowered as much as that in conventional surfactant solutions since the ability for CH2 groups to decrease the surface energy is lower than that of CH3 . Mixtures with conventional surfactants, however, are of special interest in the research of bolaamphiphile/oppositely charged surfactant mixed systems: in this case, the average saturated surface excess is increased, and some CH3 groups from the conventional surfactant play a significant role at the air/water interface in effectively lowering the surface tension of the mixture. Physical properties of such mixtures can be manipulated by inserting special functional groups in bolas leading to optical-, electronic-, or magnetic-responsive materials: as an example, by introducing a redox active group, such as a ferrocenyl head or disulfide bond, the mixed systems can be converted into redox switchable, self-assembled structures.31 Host-guest chemistry can be applied. For example, bola molecules with crown ether heads can host ionic surfactants such as SDS: such a catanionic bola/surfactant system is expected to be responsive to additional salt or specific metal ions and, in fact, the sodium ions released from the SDS can enter the cavity of the crown ether, charging the bola, so as to form ion pairs with the anionic SDS molecules.32 Exploiting the bipolar nature of the bolaamphiphiles, multilayers of oppositely charged bola/surfactants can be produced using the layer-by-layer method, that is, by dipping a mica or silica wafer into the solution of a bola and an oppositely charged polyelectrolyte multilayer.33 Perspectives in advanced functional materials can be foreseen; in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

12

Soft matter

fact, using photo-sensitive molecules, optical properties of films can be suitably reversed.34

3.1.3 Fluorosurfactants These amphiphiles have some unique properties that can be advantageously used in a range of applications as compared to hydrocarbon surfactants; in fact, they halve the surface tension of water and, because of the stability of the carbon–fluorine bond, are more stable and can survive in harsh conditions, although environmental issues can arise because of their very low biodegradation. Owing to the lipophobic nature of fluorocarbons, the interfacial behavior shows that fluorosurfactants tend to concentrate at the liquid–air interface with very low lipophilicity, because the electronegativity of fluorine reduces the polarizability of the fluorinated moiety of the surfactants. Cationic fluorosurfactant adsorption on negatively charged hydrophilic surfaces was investigated in a study by Rojas et al., especially with respect to the adsorbed layer structure, long-range interactions, and adhesion forces. The initial adsorption to the oppositely charged surfaces occurs due to the electrostatic attraction between the charged head (a)

(b)

groups and the surface, and further adsorption, driven by hydrophobic interactions, occurs readily as the surfactant concentration is increased.35 Surface force and ellipsometric experiments indicate that the surfactants self-assemble in the form of bilayer aggregates whose thickness was found to be consistent with the molecular structure, and, furthermore, ellipsometric measurements indicated that no complete bilayers were formed but rather bilayer aggregates were present on the surface even at concentrations well above the cmc. Surface force data for low fluorosurfactant concentrations demonstrated that upon compression the bilayer aggregates assembled on the isolated surfaces are transformed and as a result monolayer structures build up between the surfaces in contact (Figure 16). The force required to achieve bilayer–bilayer contact increases with the bulk concentration of the surfactant due to an increase in the repulsive double-layer force. Also, the force to drive out surfactant molecules to achieve monolayer–monolayer contact increases with surfactant concentration. Above the cmc, some additional aggregates are present on top of the bilayer aggregates coating the surface, and the adhesion found between the monolayer aggregates is one order of magnitude larger than that between the bilayer aggregates although it is still one order of magnitude smaller than the corresponding value for Langmuir–Blodgett monolayer films of similar fluorosurfactants.

4 4.1

(c)

(d)

Figure 16 Assembly of fluorosurfactant adsorbed on two approaching surfaces as a function of the separation at (a) 2, (b) 10, (c) 20, (d) 50 or higher µg l−1 . (Reprinted with permis˚ Emmer sion from O. J. Rojas, L. Macakova, E. Blomberg, A. and P. M. Claesson, Langmuir, 2002, 18, 8085  2002 American Chemical Society.)

APPLICATIONS SA in bio- and nanotechnology

Emerging technologies based on nanoscale materials and devices are already a very dynamic field in research and industry, where efficient methods of organizing materials (molecules, clusters, polymers, or building blocks) have to be introduced with the aim of keeping the nanostructures stable and effective in their engineered and final shape. In the following, some examples are shown illustrating how several applications are expected to impact in numerous fields.36 Self-assembling biomaterials have also been discussed in Peptide Self-Assembly, Self-Processes and Biologically Derived Supramolecular Materials, Nanotechnology. An example of the need to develop durable, biomembrane-mimetic coatings for inorganic and polymeric surfaces resistant to nonspecific protein adsorption (“protein resistant”) is the design of biosensor surfaces at which a ligand-binding event must be detected in the presence of numerous other nontarget proteins. An effective strategy for the self-assembly and stabilization of phospholipid bilayers at solid surfaces for the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces

(a)

(b)

13

(c)

5Å

5Å

50 Å

0

0

0

−5 Å

−5 Å

−50 Å

Figure 17 AFM tapping images of a polymerized layer (a) in air and (b) in water. Below each image is a line scan extracted from the respective image. (c) A region of the film damaged by repeated high force scanning. (Reprinted with permission from E. E. Ross, B. Bondurant, T. Spratt, J. C. Conboy, D. F. O’Brien and S. Scott Saavedra, Langmuir. 2001, 17, 2305.  2002 American Chemical Society.)

formation of a self-assembled, air-stable lipid bilayer membrane on solid supports can be found in the work by Ross et al., where the lipid bilayer was stabilized to surfactants, organic solvents, and transfer across the water/air interface by cross-linking polymerization of moieties in the acyl chains (Figure 17).37 Adsorption studies were carried out with bovine serum albumin (BSA) underlining the potential use of this architecture as a protein-resistant coating for molecular devices. In addition, the insertion of amphiphiles with reactive head groups has been investigated, to which recognition molecules can be conjugated to form a polymerized bilayer. More biologically oriented novel materials can be obtained by the self-assembly and stabilization of a phospholipid bilayer, self-organized by the fusion of fluid vesicles, composed of bissorbylphosphatidylcholine, on an oxide surface.38 A cross-linked structure has been obtained by in situ polymerization, producing a stable substrate to surfactant solutions, organic solvents, and to transfer across the air/water interface, yet retaining the resistance to the nonspecific protein adsorption characteristic of a fluid phosphatidylcholine bilayer. Protein production processes involve adhesion phenomena of polypeptides or proteic material at interfaces ranging from the adhesion onto vessels during the preparation of functional coatings to protein–membrane interactions in single cells. The adhesion strength and location of polypeptides such as poly-D-tyrosine at interfaces can be studied by AFM by using a single-molecule-based method (Figure 18) to determine the unexpectedly high influence of the support (solid, liquid, or gas) on the adhesion properties in an aqueous environment; in the presence of ethanol, the adhesion is reduced by half.39

Tip

Figure 18 AFM molecular functionalized tip. (From T. Pirzer and T. Huge, Adsorption Mechanism of Polypeptides and Their Location at Hydrophobic Interfaces ChemPhysChem, 2009, 10, 2795.  Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.)

Previous theoretical predictions and experimental data based on the presence of salt can support the mechanism of adsorption and location of the polypeptide molecules in which a compensation between free energies from hydrophobic hydration and van der Waals interactions is only possible if the polypeptide is located in both the depletion layer and the hydrophobic hydration layer.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

14

Soft matter

V PMF (kBT )

Bulk water

2

0 −2

Hydrophobic hydration water Depletion layer Surface

−

(a)

(b)

(c)

Br Cl− Na+

I−

Figure 19 Schematic representation of the possible positions of poly-D-tyrosine at hydrophobic interfaces. The polymer could reside (a) in the depletion layer, (b) the hydrophobic hydration layer, (c) or in both. Right: Position of ions with respect to these layers (Potential mean force (PMF)). (From T. Pirzer and T. Huge, Adsorption Mechanism of Polypeptides and Their Location at Hydrophobic Interfaces Chem. Phys. Chem, 2009, 10, 2795.  Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.)

From these results, the hydrophobic interaction between proteins and lipids in cell membranes needs a further insight in its understanding, while in a more general overview these approaches could have applications in a wide variety of processes where the adhesion of polypeptides has to be controlled (Figure 19). Again, microscope techniques such as AFM and TEM and neutron scattering (SANS) have been successfully applied to investigate dynamic processes of molecular selfassembly from two surfactant-like peptides A6K and A9K: in this investigation, even if the main feature of formation of A6K nanofibers and A9K nanorods was directly confirmed by SANS, evident from the diameters and lengths consistent with AFM and TEM, the SANS measurements showed some limitations in detecting the dynamic changes of nanofibrillar lengths over the full-length scale, especially for possible changes in the proportion of peptides in fibrillar fragments. An increase in the length of the hydrophobic peptide tail associated with the entropic gain, cac decrease, and increased electrostatic interaction can explain the differences observed in the structure and dynamic self-assembly process (Figure 20).40 Microscopy and micro/nanoimaging techniques (TEM) are described in more detail in Scanning Electron Microscopy and Scanning Near-Field Optical Microscopy (SNOM), Techniques. Recently, self-assembly has been increasingly finding applications in fuel cells, solid-state battery electrolytes and plasticizers, chemical sensors, photovoltaics, bone implants, membranes, chromatography supports, and chemical delivery systems.41, 42 Synthetic methods43 allow the preparation of ordered porous materials with channel and cavity dimensions at multiple length scales, combining inorganic and organic units and building up and controlling the organization of composite materials from the subnano- to the millimetre scale by a combination of templating and microfabrication techniques, creating hierarchical functional structures that are far removed from the original single components.

A 6K

(a)

A 9K

(b)

Figure 20 Schematic illustration of the dynamic self-assembly process and nanostructures for (a) A6K and (b) A9K. (From J. Wang, S. Han, G. Meng, H. Xu, D. Xia, X. Zhao, R. Schweinsc and J. R. Lu, Soft Matter, 2009, 5, 3870 http://dx.doi.org/10.1039/b901653h the Reproduced by permission of The Royal Society of Chemistry.)

A discussion on template strategies in self-assembly can be found in Template Strategies in Self-Assembly, SelfProcesses. An aerosol route can also be an easy way to capture polymers, particles, drugs, and even bioactive species within a highly structured nanocontainer.44 By variation of the self-assembled mesostructure, the accessibility and diffusivity of the guest species could be kept closely under control.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces

15

micelle concentration, and increasing self-assembly on solid surfaces with increasing temperature near ambient temperature is observed, in contrast to that expected, considering that molecular motion increases with an increase in the temperature like in simpler liquids. By using a lattice gas theory to interpret inhomogeneity, hydrogen bonding, and micelle formation, these phenomena can be found to be dependent on hydrogen bonding between 50 nm

(a)

50 nm

(b)

100

1.1

1.0

60 40

ϑ (°C)

T

80

0.9 20 (d)

Figure 21 TEM images of calcined mesoporous silica nanoparticles obtained by evaporation-induced self-assett; (a) hexagonal, (b) cubic, and (c) vesicular mesophases. (d) Uncalcined silica nanoparticles. (Reprinted from Curr. Opin. Coll. Int. Sci., 4, I. Soten, G. A. Ozin, New directions in self-assembly: materials synthesis over “all” length scales 32,  1999 with permission from Elsevier.)

By introducing a mixture of guest species, the differential curvatures of each succeeding layer of the vesicular mesophase would selectively position different species at different radial positions. This approach could lead to radially graded hierarchical materials that might facilitate vectorial energy transfer as in photosynthesis centers. As such, nanocomposite self-assembly can be considered to be an efficient structure-generating approach for the discovery of nanocomposite materials (Figure 21). By variation of composite mesostructure, material combinations, interfacial chemistry, and characteristic (confining) dimensions, we should expect to discover new material properties. This knowledge is often acquired from self-assembly and self-organization studies (see Template Strategies in Self-Assembly, Self-Processes).

0 0.001

SA in the presence of physical stimuli

Modification of physical parameters can significantly affect the self-assembly of surfactant molecules in terms of their structure and function in solution and, consequently, their interaction with the solid surface. First, anomalous and yet unexplained temperature dependence is found, for example, in aqueous solutions,45 where a decrease in surfactant solubility, an increase in the critical

XS

(a) 100 80 60 40 20 0 0.0001

0.1

0.01

0.001

XS

(b)

100

1.1

80 1.0

60 40

0.9 20 10−8 (c)

4.2

1

0.1

0.01

ϑ (°C)

50 nm

ϑ (°C)

(c)

T

50 nm

10−6

10−4

10−2

0 1

XS

Figure 22 (a) Bulk-phase diagram of aqueous surfactant solutions in the temperature/surfactant mole fraction: predictions of the theory for a solution with W=S hydrogen bonds: solid line, phase coexistence line; dashed line, CMC line; dash dotted line, temperature and concentration range of adsorption isotherms. (b) Experimental bulk phase diagram of an aqueous solution of C8E4 at different cmc; (c) Similar to (a) but for a solution without W=S hydrogen bonds. (Reprinted with permission from H. Bock and K. E. Gubbins, Phys. Rev. Letters, 92, 135701-1,  (2004) by the American Physical Society.)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

16

Soft matter

Nonirradiated

UV-irradiated for 2h

100 nm

100 nm

(a)

Vis-irradiated for 2h

(b)

100 nm

(c)

Figure 23 Freeze replica TEM micrographs for aqueous AZTMA-SDBS mixture under different lighting conditions. (From J. Eastoe and A. Vesperinas Soft Matter, 2005, 1, 338. http://dx.doi.org/110.1039/b510877m. Reproduced by permission of The Royal Society of Chemistry.)

water and the surfactant head groups and are caused by the sudden increase in rotational entropy as a consequence of breaking water–surfactant (WS) H bonds as the temperature is increased. Theory and experiment show very good qualitative agreement when such H bonding is included, and an identical system without WS hydrogen bonds shows a “normal” temperature behavior (Figure 22). Water–surfactant H bonds, in fact, break more rapidly at temperatures near the lower critical temperature and the system becomes normal at higher temperatures. Many practical applications such as coating processes, detergency, waterproofing of fabrics, and lithography in the microelectronics industry are closely related to surface tension phenomena. Thus, clean and effective methods can be used in many new technologies to “tune” surface tension according to a specific process. In order to control and modify surface properties, photosensitive surfactants can be a successful tool in the manipulation of liquids in advanced “nanotechnological” applications. Conformational changes such as cis–transisomerizations, dimerizations, photoscission, polymerizations, or polarity changes can be induced in these surfactants under different lighting conditions (Figure 23).46 The presence of a chromophore group in the hydrophilic or hydrophobic moieties in the surfactant molecular structure makes it sensitive to different physical responses, in particular, for the control of physicochemical parameters of colloidal systems such as surface activity, aggregation structure, viscosity, microemulsion separation, and solubilization. By using photosurfactants, a number of potential applications can be targeted. Most of these applications can benefit from an external stimulus, such as UV light, without other

changes in the composition or thermodynamic conditions of the system. An innovative application can be found in the case of microemulsions, where the use of UV light helps in water separation without the need of extra components or modification of physicochemical conditions. Modulating light intensity as a function of time can influence and then control the rate of the (micro)emulsion destabilization, opening the avenue to potential applications in novel, controlled-release systems. Further studies on these systems should stimulate the design of new and more effective photosurfactants in order to achieve a greater active control over the physicochemical properties of interfaces. For further reading on photoswitching materials, see Photoswitching Materials, Supramolecular Devices.

4.3

SA in modifying solid-surface properties

Surfactant molecules adsorb from a solution on hydrophobic solid–liquid (SL) and liquid–vapor (LV) interfaces, modifying the interfacial tensions (or excess surface free energies) of the SL and LV interfaces and contact angle. The wetting behavior of the surfactant solution can be affected by the carryover of the surfactant ahead of the contact line in case the surfactant molecules would adsorb on the solid–vapor (SV) interface along the contact line, where the bulk solution does not reach the solid surface. The SV surface tension would be changed due to this self-assembled surfactant ahead of the contact line. This phenomenon can be described as the “autophilic effect,” in contrast to the autophobic effect that results in a larger contact angle. These results also invalidate the assumption of there being no surfactant ahead of the contact line, a basis commonly used to estimate adsorption on SL

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces

Lr

U

Figure 24 Surfactant movement to and from the contact line region. (Reprinted with permission from K. S. Varanasi and S. Garoff, Langmuir, 2005, 21, 9932  2005 American Chemical Society.)

interfaces from contact angle data. The self-assembly of certain ionic and nonionic surfactants ahead of contact lines on hydrophilic solids, such as cetyltrimethylammonium bromide (CTAB) solutions on silica surfaces or sodium dodecyl sulfate solutions on alumina surfaces, has been observed and is considered to be responsible for their unusual wetting behavior. Re-self-assembly of surfactant molecules must occur at moving contact lines of soluble surfactant solutions.47, 48 Molecules are transported into and out of the contact line region coming from the three interfaces in contact and the fluid confined between the SL and LV interfaces (Figure 24). A molecular rearrangement is expected around the contact line having an effect on the advancing contact lines creating an unsteady motion and a complex structure in this region, while for the receding angle the contact line should move more steadily. For a wide variety of systems, quasi-static distortions of the contact line have been localized at the micrometer scale as they retreat because of the inability of the surfactant to completely re-self-assemble at localized positions along the contact line. Nonionic and ionic surfactants also in the presence of an electrolyte (CTAB on SiO2) have been investigated on high- and low-energy surfaces. Contact line distortions show micrometer-scale defects that are visible by AFM in the deposited surfactant layer consistent with their size and spatial frequency because the surfactant cannot re-self-assemble moving from the contact line region to the solid–vapor surface.

17

Since the surfactant molecules re-self-assemble into their relaxed structure over most of the contact line, a steadily moving contact line is generated due to a uniform SV surface tension. During this stage, the more hydrophilic structure pins the contact line locally, causing a quasistatical distortion as the substrate is continuously pulled out of the solution. Self-assembled monolayers of surfactant molecules constitute model systems that permit incorporation of diverse chemical and physical properties and ease of preparation. Technological areas involving electronic and optical devices, sensors and transducers, protective and lubricating layers, and patternable materials require ultrathin organic molecular films in which the relationships between structure, forces, and electrical and mechanical properties are continuously under investigation according to their application.49 In frictional studies, humidity strongly influences forces that can differ by orders of magnitude, and these differences can be explained at a molecular level with variations in the surfactant’s chemical composition, degree of unsaturation, chain length, and functional group. Techniques such as X-ray diffraction, scanning calorimetry, contact angle, and simulation data can, in addition, describe other physicochemical parameters, producing differences such as structure, phase transition temperature, compressibility, and surface hydrophobicity of the monolayer. Lateral force AFM is a powerful tool that is utilized to investigate the frictional properties of monolayers formed by different double-chain quaternary ammonium surfactants self-assembled onto mica.50 A relationship between frictional properties at the nano and microscale is possible by comparison between lateral force measurements using a standard AFM tip and measurements modified by replacing the tip with a sphere. From these observations, a relation between frictional properties and molecular properties of thin films has been found; therefore, this is an important step toward their use in many applications including lubrication and tribology (Figure 25). The fluids used in cutting and forming metallic surfaces play a critical role in working operations. In fact, extreme tribological phenomena occur during physical–chemical–mechanical processes. Process performance and finished surface quality are strongly influenced by the chemical composition and mechanical properties of the work material, the tool, and the cutting fluid. Cutting fluids provide lubricating action between the work piece and tool, removing the heat generated during the metalworking process: from a hydrodynamic point of view, during the lubrication, a film of fluid or lubricant always separates the moving surfaces, while under a boundary

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

18

Soft matter

Molecule

Tm

Structure

2C14 2C16

CH3-(CH2)n−1

2C18

CH3-(CH2)n−1

2C20

CH3

CH3-(CH2)11

N+

C12C18 CH3-(CH2)17 2C18 Unsaturated

C9H18 C9H17

Symmetric

C9H18 C9H17 C9H18

N+

CH3

29 °C

CH3 CH3

C9H17

N+

C18H37

CH3

36 °C

46 °C

CH3

2C16, −OH Terminated

HO-CH2-(CH2)15

−2OH

HO-CH2-(CH2)15 CH3-(CH2)15

N+

CH3 CH3

N+

HO-CH2-(CH2)15

FC

48 °C†

CH3

Asymmetric

−OH

34 °C 62 °C* 70 °C

n = 14,16,18,20,22

2C22

4 °C

CH3

N+

Peak

CH3

69 °C

71 °C

CH3

O C8F17(CH2)2OC(CH2)2

94 °C

C8F17(CH2)2OCCHNHCCH2N+(CH3)3 O

O

Figure 25 Correlation between structure, melting temperature, and frictional force for the surfactant monolayer. (Reprinted with permission from Y. Liu and D. Fennell Evans, Langmuir 1996, 12, 1235  1996 American Chemical Society.)

conditions’ perspective the surfaces rub against each other and the wear results in nascent metal surface. Micellar copper is an attractive additive, owing to its performance and production costs, to metalworking fluids in machining and drilling tests carried out to study

(H2O)n

Normal micelle (M) in water and polar solvents

Soft-core reverse (RM) micelle in hydrocarbon formulation

the tribological behavior of the micellar copper oxide fluid.51 During tribochemical processes, soft-core RMs formed in oil-based fluids and hard-core RMs in water-based fluids transfer additives, forming a copper tribofilm during the

(CaCO3)n

Hard-core reverse (RM) micelle in hydrocarbon formulation

(CuO)n

Hard-core reverse (RM) micelle in hydrocarbon formulation

Figure 26 Schematic representation of a normal micelle (M) in water, soft-core reverse micelle, and hard-core reverse micelle (RM) in hydrocarbon formulation. (Reprinted from Tribology International, 38, Z. Pawlak, B. E. Klamecki, T. Rauckyte, G. P. Shpenkov, A. Kopkowski, The tribochemical and micellar aspects of cutting fluids, 1  (2005) with permission from Elsevier.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces friction process, and disintegration of the RMs takes place in a tribochemical reaction creating copper tribofilms at the contact of the tool and the piece, reducing cutting forces and the wear of cutting tools (Figure 26). From the mechanical point of view, using such test fluids reduces the minimum cutting force by about half in comparison with using cutting fluids without copper additives. This also increases the life of the tools up to four times.

4.4

hierarchically structured materials can thus be obtained by applying different self-organization and recognition principles, and the directed assembly forms a basis for tunable nanoporous materials, smart membranes, preparation of nanoobjects, and anisotropic properties, such as proton conductivity. For more details on supramolecular chemistry of membranes, see Supramolecular Chemistry of Membranes, Supramolecular Aspects of Chemical Biology. Examples of functional materials based on the selfassembly of polymeric supramolecules52 include the hexagonal self-organization of conjugated conducting polymers using a comb-shaped architecture and the polarized luminance in solid-state films of rodlike polymers obtained by removing the hydrogen-bonded side chains from the aligned thermotropic smectic phase53 (Figure 27).

Functional self-assembly

Self-assembly of polymeric supramolecules is a powerful tool for producing functional materials combining different properties and may respond to external conditions:

R

R N

O

N

19

H N H

O

O N H

N H

H

O

N N

N H

O

R′

O N H

N

N

N H

O

O N H

N

N

N

R′

H

x

x

Figure 27 A retrosynthetic analysis of a material with microphase separation, stemming from a block copolymer prepared from supramolecular polymeric components. (Reprinted from O. A. Scherman, G. B. W. L. Ligthart, H. Ohkawa, R. P. Sijbesma, and E. W. Meijer, PNAS, 2006, 103, 11850  (2006) National Academy of Sciences, U.S.A.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

20

Soft matter

Alternating different layers of amino-functionalized silica nanoparticles (295 nm diameter) and epoxy-functionalized smaller silica nanoparticles (20 nm diameter), hybrid organic/inorganic surfaces have been constructed using the concept of covalent layer-by-layer assembly (covalent LbL) for the elaboration of films from polymers or dendrimers.54 A hierarchical integration of nanoscale textures by macromolecular building, is obtained in this way by grafting a newly designed highly fluorinated aldehyde and the last layer of amino-functionalized silica particles is then hydrophobized, creating a monomolecular layer via the formation of an imine function (Figure 28). A detailed discussion on nanoparticle studies is given in Self-Assembled Nanoparticles, Nanotechnology. The hydrophobicity of several highly fluorinated surfaces was found to increase with the number of layers, and their water-repellent abilities were directly correlated to the surface topologies (number of layers of silica nanoparticles and their organization on the glass support). Stable, highly water-repellent surfaces (superhydrophobic) SH with a static contact angle with water of 150◦ and very low hysteresis were obtained with the alternation of nine layers constructing covalent LbL “edifices” with functionalized silica nanoparticles of different

sizes, making a significant step toward the elaboration of responsive, sensing, and therapeutic surfaces with improved film stability. The wetting properties of surfactant solutions and engineered and nonaqueous liquids on SH substrates have been reviewed in this work. It has been shown that the combination of highly water-repellent surfaces with suitable liquids provides a wide range of methods and techniques for liquid handling and manipulation of small volumes. The effect on wetting of surfactants, as a single system or mixture, has been described at water–air and also at liquid–liquid interfaces, where the limited available data still require further development. Exploiting the role played by mutual solubility in immiscible phases, the adsorption of amphiphilic molecules on an SH surface can be regarded as an effective tool in switching mechanism between different wetting states (Figure 29).55 Biotechnology can also benefit from SA-based materials to reduce platelet adhesion by an antifouling interface obtained by means of the synthesis of dendritic saccharide surfactant polymers (Figure 30).56 After the initial step, where a maltose dendron was obtained as the starting material, the dendritic surfactant polymers were then synthesized via a two-step method.

(a)

(b)

(c)

(d)

Figure 28 SEM images with different layers of particles and related water contact angles; (a) one, (b) two, and (c) and (d inset) five layers. (Reprinted with permission from S. Amigoni, E. Taffin de Givenchy, M. Dufay and F. Guittard, Langmuir, 2009, 25, 11073  2009 American Chemical Society.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces

21

160

Contact angle (°)

150

140

130

120

110

100 0

200

400

600

800

1000

Figure 29 Dynamic contact angles on mixed organic–inorganic superhydrophobic substrates of different CTAB solutions as a function of the concentration and presence of salts. Advancing (filled symbols) and receding (empty symbols) contact angles for CTAB aqueous solutions: c = 2.0 × 10−5 M (circles), and c = 2.0 × 10−5 M + NaCl 20 mM (triangles), c = 1.6 × 10−2 M (squares), and c = 1.6 × 10−2 M + 20 mM NaCl (diamonds). (Reprinted from M. Ferrari, Aqueous And Non-Aqueous Liquids On Superhydrophobic Surfaces: Recent Developments in Contact Angle, Wettability and Adhesion, vol. 6 ed. K. Mittall, Brill VSP, The Netherlands, 2009, pp. 269–282.)

10 µm

10 µm (a)

(b)

Figure 30 Antifouling surfaces: platelet adhesion on (a) OTS-coated glass coverslip and (b) dendritic surfactant polymer-coated OTSglass coverslip. (Reprinted with permission from J. Zhu and R. E. Marchant, Biomacromolecules, 2006, 7, 1036.  (2006) American Chemical Society.)

Interfacial properties were investigated: the water surface tension was reduced by the surface-active amphiphilic glycopolymers at the air/water interface and water contact angle measurements on octadecyltrichlorosilane (OTS)coated coverslips also showed the adsorption of the amphiphilic glycopolymers at the solid/water interface. AFM observations on HOPG in fluid-tapping mode were carried out for a nanoscale understanding of surfaceinduced self-assembly of the dendritic surfactant polymer. The adsorption of surfactant polymers onto hydrophobic substrates are found to be influenced by the hexyl side chains, while a protective antifouling layer is formed by the maltose dendron side chains inhibiting platelet adhesion.

This knowledge is often acquired from tissue engineering studies (see Supramolecular Systems for Tissue Engineering, Supramolecular Aspects of Chemical Biology).

5

CONCLUSIONS

In this work, the author has reviewed the more recent literature, providing an overview on both theoretical and experimental approaches, in particular, focusing on more advanced techniques and applications. Self-assembly of surfactants at the solid surface has become a unique tool to create specifically tailored interfaces or to modify solid surfaces and their related properties.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

22

Soft matter

The combination of new strategies in synthesis and characterization of the supramolecular species obtained allow the access and design of a very large number of systems with applications ranging from micro and nanoelectronics to biotechnologies, creating monolayers responding to physical stimuli or simply improving the performance of more conventional workshop tools.

REFERENCES 1. M. J. Rosen, Surfactants and Interfacial Phenomena, 2nd edn, J. Wiley and Sons, New York, 1989. 2. http://www.detergentsandsoaps.com/surface-active-agents. html. 3. J. J. Lyklema, Fundamentals of Interface and Colloid Science: Solid-Liquid Interfaces, Academic Press, 1995, vol. 3.

23. B. G. Sharma, S. Basu, and M. M. Sharma, Langmuir, 1996, 12, 6506. 24. A. Blom, G. G. Warr, E. J. Wanless, Langmuir, 2005, 21, 11850. 25. J. R. Gong and L. J. Wan, J. Phys. Chem. B , 2005, 109, 18733. 26. S. K. Hait and S. P. Moulik, Curr. Sci., 2002, 82, 1101. 27. T. S. Kim, T. Kida, Y. Nakatsuji, et al., J. Am. Oil Chem. Soc., 1996, 73, 907. 28. P. K. Maiti and D. J. Chowdhury, Chem. Phys., 1998, 109, 5126. 29. Y. Yan, T. Lu, and J. Huang, J. Coll. Int. Science, 2009, 337, 1. 30. A. Meister and A. Blume, Curr. Opin. Coll. Int. Sci., 2007, 12, 138. 31. L. I. Jong and N. L. Abbott, Langmuir, 2000, 16, 5553. 32. R. Muzzalupo, G. Gente, C. La Mesa, et al., Langmuir, 2006, 22, 6001.

4. S. Paria and K. C. Khilar, Adv. Coll. Int. Sci., 2004, 110, 75.

33. A. Toutianoush, F. Saremi, and B. Tieke, Mater. Sci. Eng., 1999, C8–C9, 343.

5. S. Nishimura, P. J. Scales, S. Biggs, and T. W. Healy, Langmuir, 2000, 16, 690.

34. F. Saremi, G. Lange, and B. Tieke, Adv. Mater., 1996, 8, 923.

6. D. K. Schwartz, Annu. Rev. Phys. Chem., 2001, 52, 107.

35. O. J. Rojas, L. Macakova, E. Blomberg, et al., Langmuir, 2002, 18, 8085.

7. R. A. Johnson and R. Nagarajan, Coll. Surf. A, 2000, 167, 21. 8. R. A. Johnson and R. Nagarajan, Coll. Surf. A, 2000, 167, 31. 9. C. M. Wijmans and P. Linse J. Phys. Chem., 1996, 100, 125831. 10. X. Zhang, B. Chen, and Z. Wang J. Coll. Int. Sci., 2007, 313, 414. 11. R. Steitz, S. Schemmel, H. Shi, and G. H Findenegg, J. Phys. Condens. Matter, 2005, 17, 5665. 12. D. Lugo, J. Oberdisse, M. Karg, et al., Soft Matter, 2009, 5, 2928. 13. I. Harwigsson, F. Tiberg, and Y. Chevalier, J. Coll. Int. Sci., 1996, 183, 380. 14. D. W. Fuerstenau J. Coll. Int. Sci., 2002, 256, 79. 15. S. Schr¨odle and G. L. Richmond, J. Am. Chem. Soc., 2008, 130, 5072. 16. P. B. Miranda, V. Pflumio, H. Saijo, and Y. R. Shena, Thin Solid Films, 1998, 327–329, 161.

36. C. J. Brinker, Y. Lu, A. Sellinger, and H. Fan, Adv. Mater., 1999, 11, 579. 37. E. E. Ross, B. Bondurant, T. Spratt, et al., Langmuir, 2001, 17, 2305. 38. A. zur Muelhen, E. zur Muelhen, H. Niehus, W. Menhert, Pharmac. Res., 1996 13, 9.

and

39. T. Pirzer and T. Huge, Chem. Phys. Chem., 2009, 10, 2795. 40. J. Wang, S. Han, G. Meng, et al., Soft Matter, 2009, 5, 3870. 41. G. A. Ozin, Adv. Mater., 1992, 4, 612. 42. S. Mann, G. A. Ozin, Nature, 1996, 382, 313. 43. I. Soten, G. A. Ozin, Curr. Opin. Coll. Int. Sci., 1999, 4, 32. 44. Y. Lu, H. Fan, A. Stump, et al., Nature, 1999, 398, 223. 45. H. Bock and K. E. Gubbins, Phys. Rev. Lett., 2004, 92, 135701. 46. J. Eastoe and A. Vesperinas, Soft Matter, 2005, 1, 338.

17. N. Kacker, S. K. Kumar, and D. L. Allara, Langmuir, 1997, 13, 6366.

47. N. Kumar, K. Varanasi, R. D. Tilton, and S. Garoff, Langmuir, 2003, 19, 5366.

18. P. E. Laibinis, C. D. Bain, R. G. Nuzzo, and G. M. Whitesides, J. Phys. Chem., 1995, 99, 7663.

48. K. S. Varanasi and S. Garoff, Langmuir, 2005, 21, 9932. 49. A. S. Kuzharova, S. B. Bulgarevichb, et al., J. Frict. Wear, 2007, 28, 218.

V. E. Burlakovaa,

19. L. M. Grant, F. Tiberg, and W. A. Ducker J. Phys. Chem. B , 1998, 102, 4288.

50. Y. Liu and D. Fennell Evans, Langmuir, 1996, 12, 1235.

20. S. Nishimura, P. J. Scales, S. R. Biggs, and T. W. Healy, Coll. Surf. A, 1995, 103, 289.

51. Z. Pawlak, B. E. Klamecki, T. Rauckyte, et al., Tribol. Int., 2005, 38, 1.

21. H. N. Patrick, G. G. Warr, S. Manne, and I. A. Aksay, Langmuir, 1999, 15, 1685.

52. O. Ikkala, and G. ten Brinke, Science, 2002, 295, 2407.

22. W. A. Ducker and E. J. Wanless, Langmuir, 1999, 15, 160.

53. O. A. Scherman, G. B. W. L. Ligthart, H. Ohkawa, et al., PNAS , 2006, 103, 11850.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Self-assembly of surfactants at solid surfaces 54. S. Amigoni, E. Taffin de Givenchy, F. Guittard, Langmuir, 2009, 25, 11073.

M. Dufay,

23

and

Wettability and Adhesion, ed. K. Mittall, Brill VSP, The Netherlands, 2009, vol. 6, pp. 269–282.

55. M. Ferrari, Aqueous and Non-Aqueous Liquids on Superhydrophobic Surfaces: Recent Developments in Contact Angle,

56. J. Zhu and R. E. Marchant, Biomacromolecules, 2006, 7, 1036.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc145

Physisorption for Self-Assembly of Supramolecular Systems: A Scanning Tunneling Microscopy Perspective Kunal S. Mali, Jinne Adisoejoso, Inge De Cat, Tanya Balandina, Elke Ghijsens, Zongxia Guo, Min Li, Mahesh Sankara Pillai, Willem Vanderlinden, Hong Xu, and Steven De Feyter Katholieke Universiteit Leuven, Leuven, Belgium

1 Introduction 2 Importance of the Substrate in Physisorption Phenomena 3 Concentration and Temperature Dependence of Self-Assembly 4 Template Directed Self-Assembly 5 Self-Assembly at Electrified Interfaces 6 Chirality on Surfaces 7 Reactivity on Surfaces 8 Molecular Electronics at the Liquid–Solid Interface 9 Conclusion Related Articles Acknowledgments References

1

1 2 3 5 7 11 15 20 23 24 24 24

INTRODUCTION

Physisorbed monolayers at liquid–solid interfaces play an important role in a number of industrially relevant processes

(e.g., lubrication) and are of intense academic interest, not at least because of the possibilities in nanostructuring and functionalizing surfaces. The structure and dynamics of these monolayers can be followed with a variety of tools such as neutron diffraction and incoherent neutron scattering.1 The invention of scanning probe microscopes though has literally opened our eyes and revealed in real space, the often complex structure and dynamics of molecular patterns. Scanning probe microscopy methods and especially scanning tunneling microscopy (STM, see Scanning Tunneling Microscopy (STM), Techniques) have proven to be excellent tools to probe these molecular layers in detail.2 In this contribution, we will not discuss the immensely popular class of chemisorbed self-assembled monolayers (SAMs).3 (see Chemisorbed Self-Assembled Monolayers, Soft Matter) The molecular systems discussed here are only weakly physisorbed at the liquid–solid interface.4 In addition to important enthalpic contributions, the self-assembly of rather rigid solutes at the liquid–solid interface replacing small physisorbed solvent molecules is entropically favored (see Self-Assembly and Self-Organization, Concepts). The substrate is typically highly oriented pyrolytic graphite (HOPG) or Au(111). These atomically flat surfaces are ideal to unravel the structure and properties of molecular layers by STM, which requires a conductive substrate. Often, only monolayers and no clusters or multilayers are revealed. Successful STM imaging with submolecular resolution calls for molecules to be laterally immobilized for the time it takes for the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

2

Soft matter

Molecules2

Substrate Supramolecular pattern

Solvent

Scheme 1 The outcome of the molecular self-assembly process at a liquid–solid interface is the result of a complex interplay between molecule–molecule, molecule–solvent, molecule–substrate, and solvent–substrate interactions.

STM tip to scan the area (from several seconds to even minutes). Under physisorption conditions, at room temperature, most low-molecular weight molecules are too mobile to be visualized at the liquid–solid interface, except if they are trapped in or being a part of a two-dimensional (crystalline) matrix. A vast majority of these studies deal with the structural aspects of these monolayers, which one could call twodimensional (2D) crystal engineering5 (see Crystal Engineering, Supramolecular Materials Chemistry and TwoDimensional Supramolecular Chemistry, Nanotechnology). Although such studies are absolutely necessary as they bring insight in the complexity of the interplay between molecules, substrate, and solvent (Scheme 1), the next important level is functionality.6 Key challenges are the identification and design of possible functionalities as well as the construction of these functional surfaces, based on self-assembly of weakly adsorbed molecules. That is what this review is about: structure and function of physisorbed monolayers at the liquid–solid interface, focusing on supramolecular host–guest networks and dynamics, chirality, molecular electronics and reactivity, monitored or even induced by STM. The formation and visualization of 2D crystals typically occur in two different environments: either under ultra-high vacuum (UHV) condition or in solution, thus, at liquid– solid interface (see Scanning Tunneling Microscopy (STM), Techniques). In UHV, molecules are deposited by molecular beam epitaxy and the mobility of the molecules is restricted to the surface plane, which allows the molecules to rotate and translate and to find each other in order to form thermodynamically most stable structure at a given temperature. UHV–STM can be operated at temperatures ranging from near 0 K to few 100 ◦ C. The molecule–substrate interactions can be tuned by varying temperature and also weakly adsorbed molecules can be selectively desorbed from the surface. Thus, by tuning the annealing temperature, the network structure can be manipulated.7, 8 The liquid–solid interface also supports the growth of 2D crystals. The liquid phase acts as a reservoir of dissolved species

which can diffuse toward the substrate, adsorb, diffuse laterally, and desorb. These dynamic processes favor the repair of defects. Under equilibrium conditions, relatively large domains of well-ordered patterns are formed. Large domains grow at the expense of small domains via a process called “Ostwald ripening.” Furthermore, the solvent plays a significant role in the network formation. The choice of the solvent affects the mobility of molecules, especially, the adsorption–desorption dynamics via the solvation energy and possibly also via solvent viscosity. In contrast to UHV–STM studies, STM measurements at the liquid–solid interface are typically carried out at room temperature. The effect of temperature on the self-assembly of physisorbed molecules is discussed at a later stage in this chapter.

2

IMPORTANCE OF THE SUBSTRATE IN PHYSISORPTION PHENOMENA

HOPG has been the substrate of choice for studying physisorbed monolayers with STM because it is inert, easy to clean, and very stable under ambient conditions. Apart from these advantages, graphite also has a specific epitaxial interaction with alkyl chains due to the correspondence between the zigzag alternation of methylene groups and the [100] direction of graphite. This interaction has a stabilization energy of about 64 meV per methylene group.9 Besides this favorable epitaxial adsorption, alkyl chains are also stabilized by interchain interactions (so-called interdigitation), ˚ This wherein they stack (laterally) at a distance of ∼4.1 A. 10 favorable interaction inspired Charra et al. to design a molecular clip where two alkylated 1,3,5-tristilbene derivatives are able to interdigitate due to the optimal intramolecular distance between adjacent alkyl chains. They observed that the type of packing obtained depends on the number and location of the interdigitated alkyl chains. It was possible to form supramolecular dimers of two stilbene derivatives, polymers of multiple derivatives, and hexagonal networks by simply using 2, 4, or 6 peripheral alkyl chains. By combining epitaxial molecule–substrate adsorption and intermolecular packing interaction, a variety of surface specific architectures could be constructed.10 Besides graphite, other substrates such as MoS2 , MoSe2 , and Au(111) have also proven to be useful for studying physisorbed supramolecular networks. The latter differs from graphite both in periodicity as well as in specific interaction strength with adsorbates. This was demonstrated by Feringa et al. where they compared the physisorption of a 5,10,15,20-meso-tetraalkyl-porphyrin (H2 P; Figure 1a) both on graphite and Au(111).11 The H2 P molecules formed extended rows on graphite with the porphyrin part lying flat on the surface. Two out of four alkyl chains were adsorbed on the surface, following the [100] direction of the graphite

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems

(a)

( )11

(b)

(c)

(d)

(e)

(f)

(g)

3

( )11 N H N

N H N ( )11

( )11

Figure 1 (a) Chemical structure of 5,10,15,20-meso-tetraalkyl-porphyrin (H2P). (b) STM image of the molecular network of H2P on graphite. (c) and (d) STM images of the two different packings of H2P on Au(111). (e–g) Molecular models corresponding to the arrangement of molecules in (b–d), respectively. (Reproduced with permission from Ref. 11.  American Chemical Society, 2006.)

lattice (Figure 1b, 1e). On Au(111) surface, however, two different packings were observed: “a”-packing where all alkyl chains are adsorbed along the [110] direction of Au(111) (Figure 1c, 1f) and “b”-packing where alkyl chains form a zigzag packing (Figure 1d, 1g) and are adsorbed both along the [110] and [112] directions. Remarkably, only one alkyl chain is adsorbed per molecule and it was concluded that the favorable interaction of the porphyrin and the Au(111) substrate dominates the packing. With the help of X-ray photoelectron spectroscopy and also by close inspection of the contrast in the STM images, it was found that the porphyrin has two nonequivalent N-sites, which have different interactions with the surface. It was proposed that the iminic nitrogens coordinate with the Au(111) surface, while the pyrrolic nitrogens were not involved, thereby tilting the porphyrin molecules out of plane.11

3

CONCENTRATION AND TEMPERATURE DEPENDENCE OF SELF-ASSEMBLY

While investigating surface assemblies at the liquid–solid interface, in addition to molecule–molecule and molecule– substrate interactions, one cannot neglect the role of solute–solvent as well as solvent–substrate interactions. For most STM investigations, the solvent is chosen based on the following criteria: The solvent should (i) dissolve the molecules of interest; (ii) have a low vapor pressure in order

to prevent too rapid evaporation; (iii) be chemically inert; (iv) have a low polarity; and (v) have a low affinity for self-adsorption. Typical solvents include 1-phenyloctane, 1-octanoic acid, 1-octanol, tetradecane, and so on. However, recent studies indicated that the solvent plays a more important role in addition to only dissolving the desired molecules.12 The most dramatic role the solvent can have is that it participates in the molecular self-assembly via coadsorption into a multicomponent system. When studying physisorption phenomena at the liquid–solid interface, the number of molecules adsorbed at the interface as well as those present in the supernatant solution are extremely crucial and can be controlled by manipulating the solute concentration. Especially, when two or more configurations are possible which differ in packing density or adsorption energy, careful control over the concentration can select a specific structure. One of the earliest examples of such concentration control at the liquid–solid interface was reported by Lei et al.13 in the case of an alkoxylated dehydrobenzo[12]annulene (DBA). The original intention was to design a nanoporous hexagonal system which was tunable in size by simply increasing the alkoxy chain length from –OC10 to –OC20 . However, on increasing the alkyl chain length, a more densely packed linear structure was obtained which gradually replaced the porous packing. In fact, the nonadsorbed space inside the void increased on increasing the alkyl chain length. The surface coverage then decreased and the molecules responded by packing more densely together. By decreasing the concentration and thus limiting the amount of molecules which are present, it was possible to form

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

4

Soft matter

(b)

RO

(a)

OR

(c) RO RO

OR OR

R = Cn H2n +1, n = 12,14,16,18,20

3 nm

10 nm

5 nm

10 nm

(d)

High concentration

Low concentration

Figure 2 (a) Chemical structure of the DBA derivative. STM images of a DBA derivative at (b) high concentration forming a dense linear packing, and (c) low concentration forming a porous honeycomb network. (d) A schematic showing both types of packings. (Reproduced with permission from Ref. 45.  Wiley-VCH, 2008.)

a hexagonal porous packing even for the longest alkoxy chains.13 (Figure 2) A similar observation was made by Samori et al.14 when they studied the bicomponent self-assembly of melamine and a bis-functionalized uracyl derivative. A porous hexagonal packing was observed only at submonolayer concentration of both components, which is comprised of hydrogen-bonded motifs of melamine and the uracyl derivative. However, at high concentration, only melamine, which has higher interaction energy per unit area than the uracyl derivative, was found to adsorb on the surface since the system tends to minimize its energy per unit area. When working with submonolayer concentration, the dominating contribution in driving the pattern formation at the surface is the intermolecular interaction. On analyzing the hydrogen bond energy of adsorbed molecules, it was found that the intermolecular hydrogen bond energy between two components is higher than that of the monocomponent phases, explaining the presence of the bicomponent hexagonal network.14 Wan et al. reported that such concentration dependence can even lead to three different packing motifs.15 By using a 5-(benzyloxy)-isophthalic acid derivative (BIC) in 1octanol, a densely packed lamellar structure was observed at higher concentration. On decreasing the concentration, the solvent molecules contribute more actively in the process of self-assembly thus transforming the lamellar structure into a loosely packed quadrangular packing. By diluting

even further, a much more loosely packed porous hexagonal structure was observed. The number of molecules coadsorbed in the hexagonal pattern was found to be more than the one in the case of quadrangular structure, confirming the fact that with decreasing solute concentration, the system attains equilibrium via coadsorption of solvent molecules. Thus, these examples illustrate that the solvent can contribute to the self-assembly process either as a neutral component (acting as a dispersion matrix) or an active component (as a counterpart) in determining the final architecture on surfaces. While the system described in the previous paragraph can be considered as a pseudobicomponent system, it becomes a real challenge when two different building blocks are involved. Lackinger et al.16 studied the self-assembly of mixtures of two types of carboxylic acids, namely, benzenetribenzoic acid (BTB) and trimesic acid (TMA). By changing both the mixing ratio and the overall concentration, as many as six different hydrogen-bonded motifs were obtained. Despite both components being always present in the supernatant solution, molecular arrangements ranging from pure TMA networks over mixed hexagonal structures of both TMA and BTB to finally pure BTB networks were observed. The exact amount of a component in the solution is also expressed on the surface since these dissimilar motifs differ in the number of TMA or BTB molecules adsorbed per unit area. A phase diagram could be constructed. Diluting the mixture while keeping the TMA/BTB ratio constant,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems 55 °C

(a)

10 nm

55 °C

10 nm

55 °C

5

10 nm

COOH

COOH

HOOC

(d)

(b) (c)

25 °C

(f) (g)

(e)

10 nm

25 °C

10 nm

25 °C

10 nm

Figure 3 (a) Chemical structure of benzenetribenzoic acid (BTB). (b–g) A sequence of STM images taken at high and low temperatures (temperature is given at the lower left corner) showing a reversible transformation between the porous and the closed network. (Reproduced with permission from Ref. 17.  American Chemical Society, 2010.)

resulted in phase transitions predicted by the phase diagram, proving that the system is under thermodynamic control. Besides concentration, temperature is an essential parameter governing the self-assembly process as it affects both the thermodynamics and kinetics of the system. Despite this fact, amongst all the important parameters that govern the self-assembly at the liquid–solid interface, the temperature effect on the self-assembly of physisorbed molecules is the least studied. Moreover, in many variable temperature studies, samples are conditioned at elevated temperature, while the STM measurements are still carried out at room temperature. In a recent in situ investigation, Lackinger et al.17 showed that the morphology of BTB monolayers can be reversibly switched by variation of substrate temperature. Repeated heating and cooling cycles resulted in a reversible phase transition between a hexagonal porous pattern and a linear closely packed row network. At room temperature, the hexagonal porous pattern is thermodynamically the most stable network due to solvent coadsorption in the empty voids. The solvent molecules, however, desorb from the surface on increasing the temperature as they are less tightly bound than the BTB molecules. The porous network now lacks the stabilizing contribution of solvent coadsorption and therefore it collapses into the linear row structure. By repeated temperature driven phase transitions, it was therefore possible to close and open the supramolecular cavities as depicted in Figure 3.

4

TEMPLATE DIRECTED SELF-ASSEMBLY

Apart from the creation of exotic self-assembled networks on surfaces, special attention has also been paid to their application and, in particular, to their ability to host specific

guest molecules. Two-dimensional nanoporous networks are specifically designed to host guest molecules in their voids, giving rise to two- or even multicomponent architectures (see Template Strategies in Self-Assembly, SelfProcesses). Selective trapping of two types of guest molecules in a spatially resolved way was illustrated by Wang et al.18 By using tetra-acidic azobenzene (NN4A), a 2D porous kagom´e network could be obtained through hydrogen bonding interactions which had two types of cavities of different size and symmetry. When added as separately, coronene molecules specifically coadsorbed in the A-type voids, while C60 did not have any adsorption site selectivity. However, when all three components, namely, NN4A, coronene, and C60 are simultaneously applied to the surface, it was observed that coronene was entrapped in the Atype voids and C60 in the B-type voids, creating a spatially resolved ternary architecture of entrapped coronene and C60 layers (Figure 4). A powerful combination of physisorption and chemisorption was realized by Buck et al.,19 where host–guest chemistry was brought to a next level. A bicomponent hexagonal nanoporous network, stabilized by hydrogen bonds between melamine and perylene-3,4,9,10-tetracarboxylic di-imide (PTCDI) was formed on a Au(111) surface. This host network was utilized as a template for the specific adsorption of thiol SAMs in the voids. The supramolecular network was estimated to be stable enough to allow the thiol adsorption and three different thiol derivatives were successfully trapped inside the voids: adamantane thiol, dodecane thiol, and ω-(4 -methylbiphenyl-4-yl)propane thiol. After formation of this SAM-network, the hybrid structure was processed further and electrochemical deposition of Cu was performed in the underpotential deposition (UDP) region, which causes an insertion of a Cu monolayer between the Au substrate and the thiol molecules making the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

6

Soft matter

(a)

(b) O

HO C

O C OH N N

HO C

O

C OH O

10 nm

(c)

(d)

Figure 4 (a) Chemical structure of tetra-acidic azobenzene (NN4A), coronene, and C60 . (b) STM image of the coronene/C60 /NN4A ternary architecture. (c) STM image of the ternary architecture at different bias and current, where the upper part reveals the coronene molecules and the lower part reveals the C60 molecules. (d) Molecular model of coronene and C60 molecules coadsorbed in the NN4A network. (Reproduced with permission from Ref. 18.  Wiley-VCH, 2010.)

(b) Thiol

(a)

H N H H

O

O

N N

H

N

H N

N H

N N H

O

O

H

(c)

2+

(d)

2+ 2+

2+ 2+

2+

2+ 2+

(e)

2+

−

Figure 5 (a) Chemical structures and hydrogen bonding motif of melamine and PTCDI. (b) Scheme of filling the cells of the PTCDI–melamine network by thiols. (c) STM image of the hybrid structure on Au(111) with the network filled with adamantane thiol. (d) Illustration of electrochemical Cu deposition in the pores of the network at the thiol/Au interface. (e) STM image of the network after Cu deposition. (Reproduced with permission from Ref. 19.  Nature Publishing Group, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems O CnH2n +1

(a)

7

OH

HO

O

(b) ISA CnH2n +1

H2n +1Cn H2n +1Cn

CnH2n +1

COR

bisDBA-Cn

CnH2n +1

TRI

(c) b a a 3 nm

Figure 6 (a) STM image of a kagom´e network containing three different guest molecules. (b) Chemical structure of the components. (c) Molecular model of the network of A. S. De Feyter et al. (2009). (Reproduced with permission from Ref. 21.  Wiley-VCH, 2009.)

thiol-substrate bond even stronger. Evaluation of the height profile of an UDP and an unaltered region strongly suggested that Cu was only inserted between the thiol groups and the substrate and not between the melamine–PTCDI network and the substrate. Thus, these SAM modified pores serve as active sites for Cu deposition (Figure 5). This method presents a direct approach of spatially confining the adsorption of SAMs on a substrate. While coadsorption of guest molecules of appropriate size and shape in such nanopores is rather trivial, there are conditions where “guest” molecules template the formation of the porous molecular framework. In other words, the porous host nanomesh is only formed in the presence of the “guests” which stabilize the pores. For instance, Beton et al.20 anticipated that a tetracarboxylic acid derivative would self-assemble into a kagom´e network on a graphite surface. However on adsorption, a closely packed parallel packing was observed in which all four carboxylic acid groups form hydrogen-bonded dimers with adjacent molecules. The anticipated kagom´e network was uniquely obtained by the addition of coronene, which acts as a template, since its size and shape are identical with the circular pores of the kagom´e network. A kagom´e network is characterized by two types of voids that differ in size and shape. De Feyter et al.21 were able to “fill up” the pores, or rather could induce the formation of the kagom´e network, by adding different “guest” molecules simultaneously to the surface, leading to the formation of a four-component 2D crystalline network. A rhombic-shaped fused DBA derivative with decyl chains was chosen as a host network. However, self-assembly at the 1-octanoic acid/graphite interface resulted in a densely packed nonporous network. On addition of a mixture of coronene and isophthalic acid (ISA),

a kagom´e network was formed. The hexagonal void in the kagom´e network was filled by a heterocluster consisting of one coronene molecule surrounded by a cyclic hexamer of hydrogen-bonded ISA molecules. The heterocluster of coronene and ISA stabilizes the hexagonal cavity due to its shape and size complementarity. Adding a third triangular guest in the form of triphenylene filled the triangular voids of the kagom´e network giving rise to a four-component host–guest network as shown in Figure 6 (see Crystal Engineering, Supramolecular Materials Chemistry and Two-Dimensional Supramolecular Chemistry, Nanotechnology).

5

SELF-ASSEMBLY AT ELECTRIFIED INTERFACES

A majority of examples presented above comprise of STM investigations of molecules physisorbed at the interface of a low polarity organic solvent and HOPG. However, this does not mean that STM measurements in highly polar liquids are not possible. A special type of microscope called electrochemical scanning tunneling microscope (EC-STM) offers a unique possibility of visualizing the self-assembled networks of physisorbed molecules in highly polar liquids such as aqueous solutions.22, 23 EC-STM employs two additional electrodes to gain independent control over the electrochemical potentials of the STM tip and the substrate. It is the technique of choice when it comes to investigating the molecular ordering at electrified interfaces since it adds a new dimension to influence molecule–substrate interactions by controlling the potential of the substrate, which acts as a working electrode. Simple alkanes such as hexadecane have been known to adsorb parallel to the metal–electrolyte interface, forming

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

8

Soft matter

(a)

(b)

NNN = √3 =

NN

HOOC

HOOC

∆L 2.9 mn O

(c) (d)

(e)

8.6 nm

(g)

(f)

8.6 nm

Ordered

8.6 nm

Individual lamellae

Disordered

Level of order

CH3 V = 300 mV

V = 500 mV

V = 700 mV

Potential Anion adsorption & deprotonation of ISA

Figure 7 (a) Structure of the alkylated isophthalic acid derivative ISA16. (b) High-resolution STM image of ISA16 physisorbed at the Au(111)/0.1 M HClO4 interface. A few molecules are schematically indicated on top of the STM image. (c) Tentative model of the molecular arrangement of a lamella of ISA16 on Au(111). Note that the isophthalic acid groups are arranged along the NNN direction. (d, e, and f) Potential-dependent self-assembly of ISA16. The substrate potential was varied from (d) −250 mV to (f) −650 mV via (e) −450 mV. (g) A schematic of the potential-dependent hierarchical self-assembly. (Reproduced with permission from Ref. 25.  American Chemical Society, 2007.)

well-ordered 2D crystals on Au(111) surface. Moreover, it was also observed that the molecular layers could be reversibly switched between ordered and disordered structures by driving the electrode potential away from the potential of zero charge.24 Functionalized alkane derivatives such as an amphiphile ISA16, which is an ISA derivative, self-assemble in the form of stable lamellar structures at the Au(111)–electrolyte interface, at potentials close to the potentials of zero charge.25 Figure 7 shows STM images of ISA16 adsorbed at the Au(111)–aqueous HClO4 interface. The lamellae consist of bright double rows which can be assigned to the aromatic moieties, whereas the striped features with darker contrast are attributed to the alkyl chains. Thus, ISA16 forms hydrogen-bonded dimers with interdigitated alkyl chains lying parallel to each other. The alkyl chains do not follow the gold lattice, instead the selfassembly is directed by the commensurability of the ISA16 aromatic residues with the gold lattice underneath, in combination with the hydrogen bonding interactions between

carboxylic groups of the molecule (Figure 7c). A notable feature of this stable self-assembled structure is that it exists and survives despite being both the constituent hydrophilic and hydrophobic rows exposed to the aqueous medium. Previous reports illustrated that such adlayers are unstable and tend to transform into hemispherical micellar forms.26 The surprising stability of these adlayers was attributed to the strong in-plane hydrogen bonding interactions between molecules. Another remarkable feature of these adlayers is potential driven structural dynamics. The formation of well-ordered lamellae was observed only within a limited potential range. Increase in the potential (i.e., potential of the substrate), lead to loss of order converting the well-organized domains into individual lamellae (Figure 7d and e). A further increase in the potential leads to the disappearance of the isolated lamellae and only disordered aggregates of molecules (Figure 7f ) could be observed. More importantly, the potential driven order–disorder transition was found to be totally reversible

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems

9

Phthalocyanine-Based Systems, Molecular Recognition). Typically, the formation of mixed adlayers was observed, in which the molecular rows of the distinct components are arranged in an alternate fashion on Au(111) or Au(100) surfaces. The composition as well as surface mobility of such adlayers could be tuned by modulation of the electrode potential.28–31 A recent noteworthy example amongst multicomponent systems at electrified interfaces is the formation of a fullerene supramolecular network on a 2D bicomponent monolayer comprising of porphyrin and phthalocyanine molecules (Figure 8a).31 A “chessboard”-like supramolecular surface pattern which was stable to potential manipulation (Figure 8b, 8c) was obtained for the bicomponent system of ZnPC and ZnOEP at the Au(111)–0.1 M HClO4 interface. This pattern was utilized as a template to host C60 molecules (Figure 8d, 8e). Furthermore, similar studies carried out on Au(100) surface demonstrated that the supramolecular assembly of C60 molecules was strongly influenced by the

allowing the formation of well-ordered nanostructures from disordered domains in a controlled manner, suggesting the hierarchical nature (Figure 7g) of the selfassembly of ISA16 on Au(111) (see Stimuli-Responsive and Motile Supramolecular Soft Materials, Soft Matter). The potential dependence of these uncharged molecules was explained based on the deprotonation of the carboxylic groups of ISA16 at increased potentials.23 A similar protonation/deprotonation mechanism was invoked in an earlier study for explaining the potential-induced structural transformation observed in the adlayers of TMA on the Au(111) surface.27 Besides the monocomponent systems described above, a number of multicomponent systems have also been explored at the metal–electrolyte interface. Yoshimoto et al.28–31 have extensively investigated a number of bicomponent systems, mainly consisting of porphyrinand phthalocyanine-based molecules (see Porphyrins and Expanded Porphyrins as Receptors and Supramolecular (a) N N Zn N

N

N N

N N

Zn

N

N

N

(c)

N

ZnOEP

2.0 nm

ZnPc

(b) 2.0 nm

(e)

3 nm

(d)

3 nm

ZnPc side

ZnOEP side

Figure 8 (a) Molecular structures of ZnOEP and ZnPc. (b) STM image of the bicomponent system of ZnOEP and ZnPc physisorbed on Au(111) surface. A “chessboard”-like molecular arrangement is evident in which molecules of ZnOEP and ZnPc are arranged in the form of alternate rows. (c) Proposed model of the bimolecular chessboard. (d) STM image of the C60 array adsorbed in the bimolecular chessboard. (e) Proposed model for the top and side views of C60 array in the bimolecular chessboard. (Reproduced with permission from Ref. 31.  American Chemical Society, 2008.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

10

Soft matter

O

O

O

O

O

O

O

O O

O N N

N N

Co

N

N

N N

O

O O O

(a)

O

O

O O

O

8 nm

O

(b)

9.0

6.0

3.0

0

(c) 0

3.0

6.0

(d)

9.0 /nm

Figure 9 (a) Molecular structure of the cobalt phthalocyanine molecules containing four crown ethers. (b) STM image of a monolayer of this molecule at the Au(111)/0.05 mM aqueous HClO4 interface. (c) High-resolution STM image of the same monolayer in the presence of Ca2+ ions. Two of the crown ethers of each phthalocyanine host a Ca2+ ion. Bright spots corresponding to unoccupied crown ether ligands are indicated by white circles. (d) Proposed structural model of the host–guest complex; the green-colored crown ethers bind the Ca2+ ions; note that filling the two other crown ethers would probably lead to electrostatic repulsion between the positively charged ions. (Reproduced with permission from Ref. 28.  American Chemical Society, 2004.)

bimolecular packing of ZnPc and ZnOEP depending on the crystallographic orientation of Au.31 The electrified interface between a metal and an electrolyte has also been utilized to study host–guest chemistry of physisorbed molecules. Kunitake et al.32 were the first to report on the host–guest binding properties of simple dibenzo-18-crown-6 molecules at the Au(111)–0.05 M H2 SO4 interface. They could successfully image the inclusion complexes of the crown ether molecules with potassium ions at submolecular resolution. In a similar study, a peculiar substrate dependence was observed on the guest encapsulation of Ca2+ ions in the cavities of tetra-crown ether-functionalized phthalocyanines (Figure 9a) immobilized at the Au(111)–0.5 M H2 SO4 interface.33 Extended domains (Figure 9b) of this molecule could be imaged, and high-resolution STM images revealed that each square feature corresponding to the large aromatic cores of the molecules was surrounded by four bright spots, which correspond to the crown ether rings. After drop-wise addition of an aqueous solution containing 10 mM Ca2+ ions to the

subphase, only two of these bright spots remained visible along the diagonal of each molecule (Figure 9c). It was proposed that the other two spots, which had turned dark in the STM image, correspond to two crown ether rings in which Ca2+ ions were complexed. The fact that only two and not four Ca2+ ions were complexed by each phthalocyanine host is probably due to electrostatic repulsion: on filling the remaining crown ether moieties, the bound Ca2+ ions would get too close to the crown ethers of neighboring hosts that already contained ionic guests (Figure 9d). A remarkable substrate dependence was also observed in this case: instead of a Au(111) surface, when a Au(100) surface was used for the host–guest experiments, no complexation of Ca2+ ions was observed. This lack of capability to bind guests was attributed to a difference in crystallographic orientation of the Au-lattice underneath the monolayers of the phthalocyanine hosts.33 In addition to its use as a tool to investigate selfassembly at electrified interfaces, EC-STM has recently been successfully employed to study dynamics of electron

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems transfer reactions at the metal/electrolyte interface. He and Borguet34 studied interfacial electron transfer dynamics in the monolayers formed by simple tetrapyridyl porphyrin molecules at the Au(111)–0.1 M H2 SO4 interface. A potential pulse perturbation was employed to control the electrochemical oxidation of porphyrin molecules and the molecular level dynamics of this interfacial oxidation–reduction reaction was followed by using EC-STM. They observed a significant level of local heterogeneity at the nanometer scale. Low oxidation overpotential resulted in a random distribution of oxidized molecules which changed in time suggesting the diffusion of charges between the adsorbed molecules. At higher overpotential, only large patches of oxidized molecules could be observed. This molecular level approach is more beneficial than the conventional theoretical and experimental treatment of interfacial electron transfer which mainly relies on ensemble-averaged optical spectroscopic and electrochemical measurements.34

6

CHIRALITY ON SURFACES

As discussed previously, different interactions between molecules, solvent, and substrate play an important role in two-dimensional crystal engineering. Another key factor which can strongly influence the outcome of the selfassembly process is chirality.35 This structural aspect also assumes immense importance within a broad range of research domains such as catalysis and materials science. Separation of enantiomers is still a main concern in the pharmaceutical industry and the study of chirality on surfaces could be of great potential in this field. STM has been considered as an ideal technique to investigate the expression of chirality at a variety of interfaces with submolecular resolution.36–39 We focus on ambient conditions since it resembles the environment in which many important chemical and biological processes take place. A molecule can be called chiral when it lacks an internal plane of symmetry and has a nonsuperimposable mirror image. There exist three different possibilities when a racemic mixture of molecules crystallizes. It may lead to the formation of (i) a conglomerate, in which the molecules form condensates comprised of only one enantiomer; (ii) a racemic compound, in which both enantiomers are ordered in the same condensate; or (iii) a solid solution, in which the condensate contains both enantiomers in a disordered arrangement.35 In three dimensions, more than 90% of the compounds crystallize in centrosymmetric space groups. Thus, only a minor part forms a conglomerate, separating the enantiomers.40 When molecules are confined in two dimensions, chirality can be achieved more easily due to the loss of symmetry

11

elements. In fact, chirality is expressed in almost 80% of the molecules described in the two-dimensional structural database.40 Adsorption of an enantiopure molecule will give rise to the formation of a homochiral monolayer due to stereospecific interactions between the molecules and the substrate. The other enantiomer will form the mirror image structure on the surface. The self-assembly of molecules from a racemic mixture also leads to different possibilities of formation of: a conglomerate, a racemic compound, or a solid solution, analogous to crystallization in 3D. Finally, achiral molecules can also self-assemble into chiral domains. Since the molecules are achiral, an equal amount of both 2D enantiomers are observed. Chiral domains can be found locally, but globally the monolayer stays achiral. In SAMs, chirality can be expressed in different ways: packing of the molecules or the orientation of the monolayer with respect to the underlying substrate. In a joint experimental and theoretical study, Lazzaroni et al.41 studied the self-assembly of a chiral tetrakis (meso-amidophenyl)-substituted porphyrin containing long hydrophobic tails at the periphery of the conjugated π electron system (Figure 10a). The aim of this study was to understand the influence of stereogenic centers on the self-assembly in two dimensions. STM images of this compound at the 1-heptanol/HOPG interface revealed the formation of an enantiomorphous monolayer (Figure 10b). The porphyrin rows were always rotated clockwise (CW) with respect to the normal of the main symmetry axes (Figure 10c). This indicates that the stereogenic centers strongly influence the molecule–substrate interactions. For an achiral molecule both positive and negative domains would have been observed, while for this chiral porphyrin exclusively positive domains were formed. Forcefield molecular dynamics simulations were then carried out to get insight into how the chiral information is transferred from the molecular to the monolayer level. Calculations showed that the methyl group, attached to the chiral center is oriented toward the HOPG surface (Figure 10d). To maximize the monolayer density, the porphyrin cores are shifted with respect to each other, which results in an angle between the porphyrin rows and the graphite reference axis, defining the monolayer chirality. Thus, the monolayer chirality is determined by the chirality and location of the stereogenic center. Having established that stereogenic centers clearly have an influence on the 2D self-assembly, Amabilino et al.42 investigated if also the number of chiral centers have an effect on the expression and amplification of chirality. Six chiral porphyrin derivatives with the number of chiral centers ranging from 0 to 4 R centers were studied at the 1-heptanol/HOPG interface (Figure 11a). For the achiral compound, both positive and negative angles were observed for the porphyrin rows against the graphite reference axes

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

12

Soft matter

(a)

H

H

R

N

N

R

(b)

O

O O

O

N

O

R N H

a

O O

(c)

γ

R = n C18H37

N

O

b

1

HN

NH

N R

H

5 nm

(d)

φ

2 nm

Figure 10 (a) Molecular structure of a chiral (meso-amidophenyl)-substituted porphyrin 1. The red lines in the inset indicate the direction of major symmetry axes of graphite. The blue line running perpendicular to one of the main symmetry axes, is a reference axis selected to evaluate the orientation of the monolayer with respect to the substrate. (b) The unit cell is indicated in purple. The inset shows an STM image of graphite (not to scale). (c) The blue line is the selected reference axis. The green solid line is the propagation axis of the porphyrin rows.  is the angle between the reference axis and the propagation axis. (d) Side view of the optimized structure of 1 adsorbed on graphite. This image illustrates the orientation of the amide groups, with the C=O bonds pointing away from the surface (oxygen atoms in red), and the orientation of the methyl group toward the surface. (Reproduced with permission from Ref. 41.  American Chemical Society, 2008.)

and mirror domains were formed (Figure 11b). The achiral chain does not favor an achiral arrangement, but spontaneous resolution of the 2D enantiomers was observed. All chiral compounds formed chiral porphyrin rows with a positive angle against the reference axis of graphite ((R,R,R)-2 shown in Figure 11c). The only discernible difference in the packing of the compounds is the value of this angle. It decreases with a decreasing number of chiral centers in the molecule. This was also confirmed with molecular modeling. The degree of structural chirality with respect to the surface increases almost linearly with the number of stereogenic centers. As mentioned already, achiral molecules can also be employed to obtain chiral surfaces. Wang et al.43 studied the self-assembly of an achiral aldehyde-substituted oligo (p-phenylene vinylene) (OPV3-CHO) at the air/graphite interface (Figure 12a) focusing on the structure and

structural transition of chiral domains. This achiral OPV3CHO can form different chiral domains with multiple chiral polymorphs: a windmill structure, a linear structure, and a dense windmill structure (Figure 12b–d). However, achiral domains with a linear structure were also observed. In the achiral domains, a racemic compound was found, while for the chiral structures, conglomerates were formed. In the first chiral structure, windmill-like tetramers (Figure 12b) are closely packed due to interdigitation of the alkyl chains. The rotation of the windmills can be both CW and counterclockwise (CCW) and the different domains are homochiral. The second chiral structure observed, was a linear-type structure. The OPV-backbones are positioned side-by-side and form linear rows as displayed in Figure 12(c). The chirality was expressed in the orientation of the molecules, seen as deformed windmills with both handedness. For the third chiral motif, which was termed as dense windmill

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems

(a) C18H37

H

H

R

R

N

N O

O

C18H37

O

O

(R,R,R,R )-1 (R,R,R )-2 (R,R )-3 (R,R )-4 (R )-5 6

N HN

NH N

O

O

N

N H

5,10,15,20-R = CH3 5,10,15-R = CH3 20-R = H 5,15-R = CH3 10,20 = H 5,10-R = CH3 15,20 = H 5-R = CH3 10,15,20-R = H 5,10,15,20-R = H

O

O C18H37

13

R

R

C18H37

H

(b)

(c)

b2 Y2

j2

a2 b

b1 Y1

Y

a

a1

j

3 nm

j1

3 nm

Figure 11 (a) Molecular structures of the porphyrin derivatives. STM images of the porphyrin derivatives 6 (b) and (R,R,R)-2 (c) physisorbed at the 1-heptanol/HOPG interface. The insets show STM images of HOPG (not to scale) corresponding to sites underneath the monolayer. The solid white lines in the insets indicate the direction of the main symmetry axis of HOPG. The dashed red lines in all insets and main images are the selected HOPG reference axes . Unit cells are indicated in black. The solid black line is the propagation axis of the porphyrin rows. ϕ is the angle between the reference axis and the propagation axis. Each inset relates to the area underneath the monolayer. (Reproduced with permission from Ref. 42.  American Chemical Society, 2010.)

(a)

O

OC5H11 O

H C5H11O

H

(c)

(b)

(d)

a A R

5 nm

b R

5 nm

R

5 nm

Figure 12 (a) Molecular structure of OPV3-CHO. High-resolution STM images with CW rotation (b) Self-assembly of chiral windmill structures. (c) Self-assembly of chiral linear structure. (d) Self-assembly of dense windmill structure. (Reproduced with permission from Ref. 43.  National Academy of Sciences, 2010.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

14

Soft matter

(a)

(b)

S R

S R

S 30 nm

5 nm

(d)

a

b

(c)

R

S

R 20 nm

5 nm

Figure 13 (a) and (b) The adlayer structure after annealing. (a) Large-scale STM image of the heated adlayer. (b) Highresolution STM image. (c and d) Self-assembly of OPV3-CHO/ C18 H37 Br on HOPG. (c) Large-scale STM image. (b) Highresolution STM image. (Reproduced with permission from Ref. 43.  National Academy of Sciences, 2010.)

structure (Figure 12d), only half of the alkyl chains were adsorbed on the surface. The polymorphism of the OPV3CHO adlayer was attributed to the two aldehyde substitutions and the appropriate length ratio of the backbone and side chains. To converge the different polymorphs on the surface, thermal annealing was employed. After increasing the substrate temperature to 80 ◦ C, all polymorphs were transformed into the chiral linear structure, which is thermodynamically most stable polymorph (Figure 13a and b). Apart from thermal annealing, addition of alkyl bromides was also attempted to achieve structural unification of the self-assembled structure. The alkyl bromides coadsorbed with the achiral OPV-molecules and formed a new uniform chiral helix structure (Figure 13c and d). The driving force for this structural transition was found to be the halogen bonding between the halogen atom in Cn H2n+1 X (X = Cl, Br, I) and the oxygen of OPV3-CHO. Although the examples presented in the previous paragraph illustrate that achiral molecules do express chirality on confinement into a 2D plane, globally the surface still remained achiral. Nevertheless, it is possible to induce homochirality in such systems in a number of ways. Following is a brief account on two different approaches. Berg and Patrick demonstrated that mirror symmetry can

be broken in a layer of achiral molecules by controlling their arrangement on a single-crystal substrate with a liquid-crystal solvent and a magnetic field.44 When achiral 4-cyano-4 -octylbiphenyl (8CB; Figure 14a) was adsorbed on graphite, its conformational mobility was hindered and the mirror symmetry is broken, thereby allowing the formation of enantiomorphic domains (Figure 14b). To induce homochirality, a magnetic field of 1.2 T was applied parallel to the substrate. The surface was then heated and subsequently cooled. As a result, the orientational order in the bulk liquid-crystal supernate became imprinted on the monolayer, producing a molecular film with macroscopic uniform in-plane alignment. Since the orientational energy of the magnetic field was relatively small compared to the intermolecular and molecule–substrate interactions, the arrangement of the molecules within a unit cell and the registry of the molecules against the surface were unaltered. Applying a magnetic field favored one possible configuration and by simply changing the orientation of the field, a different configuration was favored (Figure 14c–e). This method allows continuous control over absolute handedness and enantiomeric excess without incorporation of any external chiral agent. Katsonis et al.45 used a different approach and demonstrated that homochirality emerges at the interface between a chiral liquid and the graphite surface. The self-assembly of an achiral OPV with a diamino triazine moiety (AOPV4T; Figure 15a) was studied at the liquid–solid interface. When an achiral solvent was used, A-OPV4T formed chiral rosettes consisting of six molecules which were held together by hydrogen bonding. Homochiral domains with either CW or CCW were formed. To induce chirality, enantiomerically pure solvents (R)- and (S)-1-phenyl-1-octanol were chosen (Figure 15b). On adsorption of the achiral molecules in the chiral solvent, both CW and CCW rosettes were observed. However, statistical analysis of large-scale STM images (Figure 15c) indicated that there was a clear bias toward CCW rosettes when (R)-1-phenyl-1-octanol was used (Figure 15c and d), whereas CW rosettes were favored in the case of (S)-1-phenyl-1-octanol (Figure 15e). Although there was no evidence for coadsorption of solvent molecules, hydrogen bonding interactions between the solvent and the adsorbed achiral molecules seem to be of key importance for the induction of the observed homochirality at the surface. The actual mechanism of chiral selection that favors the formation of rosettes with a particular handedness is not known, but it most likely involves, apart from solvent–molecule interactions, chiral desolvation processes and steric restrictions within the monolayer. This study primarily illustrated that a chiral solvent can produce enantiomerically enriched and even homochiral organic surfaces.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems (c) (a)

15

20

CN Number

15 10 5 0 (d)

8 Righthanded

Number

6

(b)

Lefthanded

4 2 0

Number

(e)

8 6 4 2 0 −80

−60

−40

−20

0 20 f /°

40

60

80

Figure 14 (a) Molecular structure of 8CB. (b) STM image of a right-handed domain. (c–e) Histograms of molecular head group orientations (ϕ ) determined by STM for samples prepared with three representative field alignments: (c) field at −19◦ from ; (d) field at 0◦ (parallel to ); and (e) field at +19◦ from . Blue bars indicate right-handed molecules, red bars indicate left-handed molecules. Part (d) shows two degenerate molecular configurations. Other field orientations break the degeneracy which leads to an excess of one chiral population over the other as in parts (c) and (e). (Reproduced with permission from Ref. 44.  Wiley-VCH, 2005.)

7

REACTIVITY ON SURFACES

STM can also be utilized as a tool to induce and even follow chemical reactions between molecules adsorbed on surfaces (see Covalent Capture of Self-Assembled Soft Materials, Soft Matter). Numerous conventional techniques such as nuclear magnetic resonance (NMR) spectroscopy, mass-spectrometry, and absorption/fluorescence spectroscopy are routinely employed to study and understand the mechanisms of chemical reactions. However, these techniques provide only ensemble-averaged output and single-molecule information thus remains unknown. In contrast to the aforementioned techniques, STM has a distinct advantage of resolving the details of surface reactions at the single-molecule level. A variety of reactions have been successfully probed with STM both under UHV and ambient conditions. As the former is somewhat extreme environment, where the molecules have to be sublimed prior to visualization and therefore far removed from

traditional laboratory conditions, the latter, especially the liquid–solid interface, is very familiar to many synthetic chemists. However, the number of studies performed under ambient conditions still remains small. Successful probing of a reaction on the surface using STM requires immobilization of the reactant molecules, which limits the choice of reactions that can be followed. Surface polymerization reactions (see Self-Assembled Polymer Supermolecules as Templates for Nanomaterials, Soft Matter) were among the first category of reactions to be studied with STM. The topochemical photopolymerization of diacetylene units into a long single polydiacetylene requires a specific orientation of the monomers within the assembly on the substrate. It is generally known that diacetylene compounds can undergo an addition reaction on the surface on excitation with external stimuli like UV irradiation or a voltage pulse by the STM tip, thereby forming 1D nanowires (see Self-Assembly of Supramolecular Wires, Self-Processes). Maximal reactivity is expected

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

16

Soft matter (a)

(b)

C12H25O O

OH

O

C12H25O

NH2 C12H25O

*

N N

O O

N NH2

(c)

(d)

(e)

a g b

Figure 15 (a) Molecular structure of A-OPV4T. (b) Molecular structure of 1-phenyl-1-octanol. (c, d) STM images of an A-OPV4T monolayer at the (R)-1-phenyl-1-octanol–HOPG interface. (c) In addition to several domains of CCW rosettes, one CW domain is observed as marked. Scale bar is 10 nm. (d) High-resolution image of the CCW rosette. Individual OPV units are indicated by green lines emphasizing the nonradial orientation. The rotation direction is highlighted by the white arrow. Scale bar is 3 nm. Sequence of dashed and solid white marker lines from left to right: dashed-up and solid-down. (e) Molecular resolution STM image of an A-OPV4T monolayer at the (S)-1-phenyl-1-octanol–HOPG interface. The CW rotation direction is highlighted by the white arrow. Scale bar is 3 nm. Sequence of dashed and solid white marker lines from left to right: dashed-down and solid-up. (Reproduced with permission from Ref. 13.  Wiley-VCH, 2008.)

when the stacking distance between two adjacent diacetylene moieties and the angle between the units and the ˚ and ∼50◦ , respectively.46 Okawa stacking axis is ∼5 A and Aono described the STM tip pulse-induced formation of conjugated diacetylene polymer chains from 10,12pentacosadiynoic acid on the surface of graphite.47 The monolayer obtained on dropcasting the solution on the surface, initially consisted of straight lamellae in which the molecules are arranged in such a manner that the COOH end groups form dimers and thus stabilize the monolayer. By careful placement and stimulation of the STM tip at a predetermined position, a conjugated polymer wire could be created without any additional energy besides thermal energy. The length of the nanowires was found to be controlled by domain boundaries or even artificial defects in the monolayer. The polymerized part of the monolayer has a distinct bright contrast due to the conformational change of the polydiacetylene backbone, which appears to be 0.15 nm higher than the unpolymerized moieties. The alkyl chains of the monomers as well as those of polymerized molecules were found to follow the main symmetry axes of graphite. De Schryver et al.48 showed that by using 5-(10,12-tricosadiinyloxy) isophthalic acid (Figure 16a, ISA-DIA), a similar polymer could be obtained. After UV-illumination or STM tip pulse excitation, identical bright feature was observed,

which was attributed to a polymerized polydiacteylene. A dramatic change in the orientation of the alkyl chains was observed on polymerization, which is in contrast to the observations made by Okawa and Aono47 . Nevertheless, the orientation of the alkyl chains of unpolymerized molecules remained unaffected (Figure 16b–d). By choosing 2,5(10,12-heneicosadiinyloxy) terephthalic acid (Figure 16a TTA-DIA), which contains two diacetylene units per molecule, they were even able to form a 2D nanostructure which chemically connects two adjacent rows of 1D polydiacetylenes, paving the way toward 2D polymers (Figure 16e). Following these observations, Beebe et al. were able to spatially confine the polymerization by using molecular corrals on the graphite surface.49 Polymers were formed both inside and outside the molecular corral but a polymer extending over the edges of the molecular corrals was never observed. This once again strengthens the fact that the process of polymerization propagates via reaction between adjacent diacetylene units because at step edges the required topochemistry was disrupted. By controlling the size and position of the molecular corrals, the length of the polymers could be controlled in a spatially resolved manner. An inorganic nanowire was synthesized by M¨ohwald et al.50 by using arachidic acid and copper(II) sulfate. In the solid state, a copper dimer is known to coordi-

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems

(a)

(b)

17

(c)

HOOC O C9H18 HOOC

C10H21

ISA-DIA COOH

C8H17

C9H18 O

O C9H18

C8H17

HOOC

TTA-DIA 4 nm

4 nm

(d)

(e)

aP aM

4 nm

Figure 16 (a) Chemical structures of ISA-DIA and TTA-DIA, (b, c) STM images of the unpolymerized network of ISA-DIA and TTA-DIA. (d, e) Polymerized network of ISA-DIA and TTA-DIA where the polymerized part appears as a bright stripe. (Reproduced with permission from Ref. 48.  American Chemical Society, 2003.)

nate with four carboxylate groups where two carboxylate chains coordinated to the same copper(II) are not parallel to each other but form an extended chain structure. On spin coating a mixture of arachidic acid and copper(II) on a graphite surface, disordered nanostripes of arachidic acid were revealed by atomic force microscopy (AFM; see Atomic Force Microscopy (AFM), Techniques), showing numerous kinks and turns (Figure 17b) while in the case of the pure arachidic acid, homogeneous unidirectional striped domains were found (Figure 17a). However, on addition of H2 S, the unidirectional orientation of the stripes was restored and two types of stripes were observed which differ in height. These stripes were attributed to both arachidic acid bilayers and copper sulfide nanorods. It was proposed that, H2 S releases the copper from the coordinated network by forming CuS, and thereby restores the pure arachidic acid domains. Individual CuS molecules then form CuS nanorods (Figure 17c–e).

While the reactions described above occur on the surface, many reactions occur at the liquid/solid interface and the reaction product assembles on the surface by competitive adsorption. Such a series of reactions involving the addition of a reactant to a molecular monolayer on a substrate was reported by Samori et al.51 An octadecyl substituted guanine derivative was dropcasted on graphite and a ribbonlike structure was observed. This architecture is stabilized by both the interdigitation of the octadecyl chains through van der Waals interactions and hydrogen bonds between the NH, N, and carboxyl groups. Addition of potassium picrate induced a structural transformation into a G4 -based network, where a potassium ion is trapped inside the void. In this case, different hydrogen bonding sites are interacting than in the case of the ribbon-like packing. This indicates that the octadecyl chains are not adsorbed on the surface and are pointing toward the solution phase, making new hydrogen bonding sites accessible. The [2.2.2] cryptand

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

18

Soft matter

(a)

(b)

(c)

D S

150 nm

Arachidic CuS acid

Arachidate Cu2+

100 nm

(e) 1.50

(d)

250 nm

H2S

0

0.8 nm

−1.50

0.4 nm

0

50

100 (nm)

150

Figure 17 (a) AFM image of a pure arachidic acid monolayer film on graphite. (S stands for a striped domain and D stands for a disordered domain). (b) AFM image of copper(II)arachidate film. (c) Copper arachidate film after the reaction with H2 S. (d) Scheme of the CuS nanorod formation from a copper arachidate template monolayer. Dashed lines indicate coordination bonds between copper ions and carboxyl groups. (e) Height profile corresponding to the line in C. H. Mohwald et al. (2004). (Reproduced with permission from Ref. 50.  American Chemical Society, 2004.)

offers an efficient complexation of potassium into a cryptate [K+ ⊂ 2.2.2] and by adding this cryptand to the G4 network, the guanine reassembled into the original ribbon structure. To continue the assembly cycle, trifluoromethanesulfonic acid, which releases the potassium ions from the cryptate, was added and the G4 assembly was regenerated. Finally, adding more [2.2.2] cryptand on the surface again resulted in the formation of ribbons, proving an in situ reversible assembly and reassembly between two highly ordered structures on the surface (Figure 18). In addition to following the reactants and the products on surfaces, a catalyst and the changes in its topography during a reaction can also be monitored by means of STM.52 The high spatial and temporal resolution of the STM can be used to obtain quantitative information on elementary processes in surface catalyzed reactions. In fact, the surface science approach to catalysis, which was pioneered by 2007 Nobel Laureate in chemistry Gerhard Ertl, has helped to revolutionize the understanding of heterogeneous catalysis down to the atomic level. Thus, the atomic scale insight gained from STM studies of fundamental surface processes relevant to heterogeneous catalysis can be used for the rational design of new catalysts. A more exhaustive review of such studies, often carried out under UHV conditions, can be found in Refs 53 and 54.

Elemans et al. demonstrated the use of STM as a tool to investigate catalytically relevant model systems.52 Mn(II) porphyrins are often used as catalysts for the chemical transformation of alkenes into epoxides in which an oxygen is added across a carbon–carbon double bond to form a three-membered ring. Mn(III) porphyrins adsorbed on a Au(111) surface react with O2 as the apparent height of the porphyrins increased by a factor 3. Despite the fact that these porphyrins were chemically inert in solution, it was presumed that the Au(111) surface played an essential role in the activation of the porphyrins. Indeed, the surface promotes a reduction from Mn(III) to Mn(II), thereby activating the complex for the reaction with O2 . Moreover, close inspection of the monolayer led to the conclusion that the reaction with O2 mostly occurs in pairs of adjacent porphyrins, which is explained by the homolytic dissociation of O2 , distributing both its oxygen atoms over two adjacent porphyrins. Once bound to oxygen, these Mn(IV) porphyrins are suitable catalysts for the epoxidation of alkenes. On adding cis-stilbene to the monolayer, the number of oxidized Mn(IV) porphyrins dramatically decreased, probably due to the transfer of their oxygen to the cis-stilbene forming cis-stilbene epoxide. This was confirmed by gas-chromatographic analysis of the liquid layer (Figure 19).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems

0s (a)

C14H37 N

H N

N

N

N

C18H37 N N H

N H

O

H

N

N

N

O

H N

H N

N

N

N H

N N C18H37

O

H

O

H

H

19

N C18H37

H N H

G ribbon

2 nm

(b)

360 s C18H37 N

H

N

N

N H O N H H N N O

H N

H N

H O

H N

H

N

N C10H37

C18H37 N

N

O

N

2 nm

H N

N

N C H 18 37

N

H

N H

G quartet [2.2.2]Cryptand

60 s (c) C18H37 N

H N

N

N

N

C10H37 N N N H

H O

H

O

N

N

N

H

O

H

H

O

N

N

H

H

N

N H

N

N H

N N C18H37

2 nm

H N

N C18H37

O

O

O O

O O

N

[K+ ⊂ 2.2.2]

G ribbon HTf

(d)

60 s

C10H37 N

N

H

H N

N H O N H H N N O

H N H

N

O

O

N H O O

O O

N

H O N

C18H37 N

N

N

N C18H37

2 nm

H N

O

N H N

H

N

[H+ ⊂ 2.2.2]

N C H 18 37

N

H

+

G quartet

60 s

[2.2.2]Cryptand

(e)

C18H37 N

N

N

N O

2 nm

H N

C18H37 N N N H

H O

H N

N

N N C18H37 G ribbon

H N H

H N N

O

H

H N

O

H N

N N C18H37

H N H

N

O

O

O O

O O

N

[K+ ⊂ 2.2.2]

Figure 18 A sequence of STM images and chemical structures showing the structural evolution of a monolayer of the guanine derivative over a 9 min time scale (time range displays in the upper right part of the images correspond to the time that was needed to reach the equilibrium after addition of reacting agents). (a), (c), and (e) show ribbon-like structure, whereas (b) and (d) exhibit G4 -based architectures. (Reproduced with permission from Ref. 51.  Wiley-VCH, 2010.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

20

Soft matter Cl Mn(III) Mn(II)

O

Mn(II)

Au(111) 2 N N

M

O2

+ O

N

N

M = MnCl

2

O

O

Mn(IV)

Mn(IV)

Au(111) (a)

(b)

5 nm (c)

Figure 19 (a) Molecular structure of the manganese porphyrin catalyst. (b) Proposed catalytic cycle. (c) Series of STM images of a monolayer of the catalyst self-assembled at the interface of Au(111) and n-tetradecane; left: system under argon; middle: 4 h after flushing the bell jar with O2 , causing 10% of the catalysts to be oxidized (visible as red spots); right: 3 h after the addition of cis-stilbene, showing a decrease in the number of oxidized catalysts. (Reproduced with permission from Ref. 52.  Nature Publishing Group, 2007.)

Besides the reactions leading to the formation of a new product, reversible conformational changes can also be induced and monitored on the surface. Azobenzenes, for example, are known to be photoresponsive and the reversible transformation from a trans isomer to a cis isomer on UV light irradiation (see Photoswitching Materials, Supramolecular Devices and Photochemically Driven Molecular Devices and Machines, Nanotechnology) has been extensively studied. Zhu et al.55 used a 4-hydroxy-3 trifluoromethyl-azobenzene derivative which spontaneously forms lamellae of the trans isomer on the surface on deposition from a 1-undecanol solution, stabilized by a hydrogen bond between the –CF3 and the alcohol group. After illuminating the monolayer with UV light, a complete photoisomerization from trans to cis occurred. No more trans domains were observed, which is explained by the increased stabilization of the cis isomers due to the formation of more hydrogen bonds in comparison to the trans isomer. In a somewhat similar investigation, Wang et al.56 incorporated azobenzene groups into a macrocyclic compound, which could change its shape on isomerization. This 4NN -macrocycle can be incorporated into a predesigned hexagonal nanoporous network formed by 1,3,5-tris(10carboxydecyloxy)-benzene (TCDB) and were observed as

rectangular entities inside the voids. On increasing the concentration of the 4NN -macrocycle, the TCDB molecules transform from a hexagonal packing into a rectangular packing and thus trap the 4NN -macrocycles. In both cases, all the azobenzene groups of the macrocycle are in trans conformation although in the latter case, the macrocycle is more stretched. After UV-illumination, it was found that either one or two nonadjacent azobenzene groups were able to isomerize into the cis configuration, thereby changing the shape of the macrocycle and providing an additional five configurational isomers. The variety in shape also induces structural changes of the TCDB network, where a mixture of the hexagonal packing, the rectangular packing, and a disordered packing is observed. The entrapment of photosensitive molecules into nanoporous networks provides an easy method for the observation of the isomerization.

8

MOLECULAR ELECTRONICS AT THE LIQUID–SOLID INTERFACE

In addition to its utility to visualize and manipulate matter down to the atomic scale, STM also offers a unique possibility of probing the electronic properties of the

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems adsorbate molecules. As a consequence, it can be used as a spectroscopic tool [Scanning Tunneling Spectroscopy (STS)]57 to investigate the current–voltage (I−V ) characteristics of individual physisorbed molecules or their supramolecular networks, with submolecular spatial resolution. The key to the success of STS is the ability to position the STM tip freely with respect to the molecule or various parts within the molecule. In STS-mode, the STM tip is positioned on a sample coordinate of choice, the feedback loop, which controls the distance between the tip and the sample according to a given tunneling current set point, is deactivated, and I−V curve is recorded. Thus, a combination of STM and STS at the liquid–solid interface provides in situ access to both structural information as well as electronic properties of single molecules or molecular complexes confined in a well-defined junction between two electrodes. Molecules can be immobilized on solid surfaces via chemisorption; however, it often alters the electronic states of the molecules as well as the substrate. In this context, physisorption of molecules at the liquid–solid interface has many advantages, namely: (i) It allows immobilization of molecules without interfering with the electronic states of the molecules as well as substrate. (ii) Weak molecule–substrate interactions allow substantial molecular dynamics at the interface which is a prerequisite for the construction of flawless large area molecular networks.

(a)

(iii) It allows investigation of very large molecules which cannot be sublimed nondestructively. In the meantime, electronics based on electron transport through single molecules has received considerable impulse in view of the tremendous miniaturization trend in the semiconductor industry as well as the anticipated limits of the conventional silicon-based integrated circuits.58 In fact, the idea that a few molecules or even a single molecule could be sandwiched between two electrodes and perform the basic functions of digital electronics was first put forward by Aviram and Ratner59 in the mid 1970s. They suggested that a current could be modified with a single molecule composed of a donor-bridge-acceptor structure fixed between two metal electrodes. The advent and subsequent development of STM and STS has allowed researchers to realize the electronic functions of single molecules in real time (see Molecular Devices: Molecular Machinery, Single-Molecule Electronics, Supramolecular Devices, and Nanoelectronics, Nanotechnology). In their pioneering approach, Rabe et al.60 developed a prototype single molecule chemical-field-effect transistor (CFET) based on self-assembly of a molecular dyad (Figure 20a) which consists of an electron rich hexa-peri -hexabenzocoronene (HBC) core covalently linked to six electron deficient anthraquinone (AQ) subunits. Figure 20(b) displays an STM image of monolayer

R

I/nA 0.6

0.8 I/nA 0.6

(d)

(b)

R

0.4 0.2 0.0

R

−1.0

0.4

(b)

1.0 U/V

−0.2 −0.4

0.2

R

0.0

1.0

U/V

OFF

−1.0

R

−0.2

GATE

5 nm −0.4

R O R=(CH2)10

21

ON

O O

(e)

(c)

Source O

U

OCH3 DMA

Gate

AQ CT complex

OCH3

HBC Drain

5 nm

DMA

Figure 20 (a) Structure of the molecular dyad HBC–AQ and DMA. (b) STM image of HBC–AQ physisorbed on the surface of HOPG. (c) STM image of a 1 : 10 mixture of HBC–AQ and DMA at the liquid–solid interface. (d) I–V curves measured through the HBC cores before (open triangles) and after (solid circles) the formation of AQ–DMA charge-transfer complexes. Inset: shifted and normalized data (see text). (e) A schematic of the resultant prototype CFET. (Reproduced with permission from Ref. 60.  American Physical Society, 2004.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

22

Soft matter unit cell with the HBC cores surrounded by six additional bright spots arranged in a zigzag row corresponding to the AQ–DMA charge-transfer complexes. Figure 20(d) displays the I−V curves through the HBC cores before and after complexation with DMA. The I−V curve measured through the HBC moieties after the formation of chargetransfer complexes was found to be much more symmetrical than the one in absence of DMA. This significant change has been attributed to the relative change in the Fermi level of the substrate and the molecular orbitals of the adsorbate as a result of the formation of a dipole at the interface which shifts the I−V curve by about 120 mV. Thus, because of the covalent connection between the HBC cores and the chargetransfer complexes, the substrate work function is shifted by 120 mV. This number was found to be consistent with the calculated dipole moments of the AQ–DMA chargetransfer complexes and their 2D density at the interface. The shift in the I−V curves indicates that such charge-transfer

of (HBC–AQ) physisorbed at the liquid–solid interface. The HBC cores can be clearly visualized as bright spots within the rectangular unit cell with four less bright features in between, which are attributed to the AQ moieties. STS experiments carried out by precisely positioning the STM tip over the electron rich and electron deficient segments of the molecule revealed characteristic behavior of these molecular subunits. I−V curves recorded over HBC cores were asymmetric and exhibited rectifying behavior with larger currents at negative sample bias, whereas AQ subunits displayed rectifying behavior with larger tunneling probability at positive sample bias. On addition of an excess of electron donor molecule, 9,10dimethoxyanthracene (DMA), to the liquid–solid interface, the molecular packing changed which was attributed to the formation of charge-transfer complexes between the AQ moieties of (HBC–AQ) and the DMA molecules. The STM image (Figure 20c) of the monolayer revealed a much larger

(a)

(b)

O

N

(d)

O

b O

a

O

2 nm R

4 µm

R

(c)

(e) 10−5

R

R

Vds = ±10 V

10−6

20 µm

−Ids/A

R

10−6

R=

p-channel

2 µm

−10

−5

n-channel 0

5

10−7

10

V g /V

Figure 21 (a) Structure of the molecular dyad HBC–PMI. (b) STM image of HBC–PMI physisorbed at the 1-phenyloctane/HOPG interface. (c) AFM topography images of nanofibre networks of HBC–PMI obtained by spin coating. (d) AFM topography images of the nanoribbons obtained after solvent vapor annealing treatment of HBC–PMI. (e) Transfer characteristics for a transistor based on mesoscopic HBC–PMI nanoribbons. Inset: optical microscope image of nanoribbons grown on top of an array of gold electrodes. (Reproduced with permission from Ref. 64.  Wiley-VCH, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems complexes can act as integrated nanosized gates in a prototype single molecule CFET (Figure 20e).60 At this juncture, it must be noted that realizing electronic functions through single molecule does not make it a molecular electronic device (see Nanotechnology: The “TopDown” and “Bottom-Up” Approaches, Nanotechnology). There must be a way in which this nanoscale component can interact with the macroscopic world. As a consequence, it is important to consider how a molecular electronic component can be wired up into a deviced architecture. To address this issue, efforts were focused on developing molecular-scale wires and circuitry.61–64 In addition to the use of conventional materials such as metals or carbon nanotubes to construct nanoscale wires, a bottom-up approach in which organic molecules self-assemble into extended one-dimensional supramolecular nanostructures is also considered as a versatile alternative.61, 62 In particular, construction of supramolecular assemblies based on π-conjugated systems offers an excellent tool to construct nanoscale architectures in solution and in solid state.61–63 In a recent study, Samori et al.64 investigated molecular dyads based on polycyclic electron donor (D) and acceptor (A) units which proved to be suitable building blocks for forming highly ordered, solution-processable, nanosegregated D–A domains for potential use in (opto)electronic applications. A molecular dyad, based on alkylated HBC and perylene monoimide (PMI) was shown to have a high tendency to self-assemble at multiple length scales. The selfassembly lead to the formation of various supramolecular architectures, namely, macroscopic filaments, networks of nanosized fibers, and 2D monolayers (Figure 21). Solvent– vapor annealing of this dyad leads to the formation of highly uniform mesoscopic ribbons bearing atomically flat terraces. Electrical measurements carried out on these ribbons revealed that they exhibit ambipolar transport with well-balanced p and n-type mobilities. Owing to the higher level of order at the supramolecular level, the devices based on such ribbons showed mobilities which were an order of magnitude higher than the ones with thin films.64 Besides the realization and development of moleculebased devices, the combination of STM (high spatial resolution) and STS (single molecule electronics) can also be used for understanding fundamental issues related to the charge carrier transport through single molecules.65 Moreover, the contrast in STM images is extremely sensitive to the electronic signature of the surface as well as the adsorbate molecules and thus is a convolution of both electronic and topographic effects. This fact has been exploited in understanding the electron transport through donor–acceptor type molecules physisorbed on the surface of HOPG with their molecular axes parallel to the surface. The bias-dependent contrast in the images as well as the STS measurements

23

revealed the nature of different electronic states involved in the tunneling process.66, 67

9

CONCLUSION

With the advent of scanning probe microscopy, especially STM, the research on supramolecular self-assembly on atomically flat conductive surfaces has witnessed a spectacular increase in activity in the last decade. By exploiting the concepts from supramolecular chemistry, it is now possible to construct more and more complex 2D architectures. This holds also for the liquid–solid interface and under electrified conditions. Nevertheless we must admit, serendipity still plays an important role. Supramolecular (chiral) self-assembly highlights the importance of the subtle interplay between intermolecular interactions on one hand and molecule–substrate interactions on the other hand, and thus uncovered some of the basic mechanisms of (chiral) molecular recognition. Key points of interest remain the investigation of how spontaneous resolution can be achieved, controlled, or templated. Amplification of chirality effects observed at the liquid–solid interface is also waiting to be explored. In general, important new insights are to be expected on the overall role of solvent in directing solvent-assisted supramolecular self-assembly processes on surfaces. A recent development of the last decennium is the formation of so-called “nanoporous” networks. Primarily driven by academic curiosity, the research activities have revealed a number of previously unexplored aspects of monolayer formation such as concentration and surface coverage control, self-recognition and self-selection along with the effect of solvents on self-assembly, and porous network formation. In general, the research activities targeting the formation of nanoporous structures provide a wealth of insight in the concepts which drive monolayer formation and stabilization. The main aspect though, which makes the field of 2D nanoporous materials “hot,” is the broad range of potential applications. An exciting development, finally, is the real-space and real-time exploration of physical and chemical phenomena such as molecular dynamics, electronics, and reactivity. Using STM, information of dynamic processes is obtained at the level of single molecules and in some cases new aspects of reaction mechanisms are revealed which remain unexplored with conventional ensemble techniques. A steadily growing number of scientists now appreciate the opportunities brought forward by the research on supramolecular self-assembly at surfaces. This revolutionizing trend promises a bright future to this research field where scientists from many disciplines meet.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

24

Soft matter

RELATED ARTICLES

13. S. B. Lei, K. Tahara, F. C. De Schryver, et al., Angew. Chem. Int. Ed., 2008, 47, 2964.

Supramolecular Electrochemistry Hydrogen-Bonding Receptors for Molecular Guests Photoinduced Electron Transfer Processes in Biological and Artificial Supramolecules Molecular Devices: Energy Transfer Molecular Logic Gates Noncovalent Interactions in Crystals Soft Matter Science—a Historical Overview with a Supramolecular Perspective Multicomponent Self-Assembled Polymers Based on π Conjugated Systems Self-Assembly of Surfactants at Solid Surfaces Reactivity in Nanoscale Vessels Atomic Force Microscopy Measurements of Supramolecular Interactions Nanolithography

14. C. A. Palma, M. Bonini, A. Llanes-Pallas, et al., Chem. Commun., 2008, 42, 5289. 15. X. Zhang, T. Chen, Q. Chen, et al., Chem. Eur. J., 2009, 15, 9669. 16. L. Kampschulte, T. L. Werblowsky, R. S. K. Kishore, et al., J. Am. Chem. Soc., 2008, 130, 8502. 17. R. Gutzler, T. Sirtl, J. F. Dienstmaier, et al., J. Am. Chem. Soc., 2010, 132, 5084. 18. Y. T. Shen, M. Li, Y. Y. Guo, et al., Chem.- Asian J., 2010, 5, 787. 19. R. Madueno, M. T. Raisanen, C. Silien, and M. Buck, Nature, 2008, 454, 618. 20. M. Blunt, X. Lin, M. D. Gimenez-Lopez, et al., Chem. Commun., 2008, 20, 2304. 21. J. Adisoejoso, K. Tahara, S. Okuhata, et al., Angew. Chem. Int. Ed., 2009, 48, 7353. 22. K. Itaya, Prog. Surf. Sci., 1998, 58, 121–247. 23. D. M. Kolb, Surf. Sci., 2002, 500, 722–740.

ACKNOWLEDGMENTS

24. Y. He, T. Ye, and E. Borguet, J. Phys. Chem. B, 2002, 106, 11264.

The authors thank the Interuniversity Attraction Pole program of the Belgian Federal Science Policy Office (IAP 6/27), K.U.Leuven (GOA, Geconcerteerde Onderzoeksacties), I.W.T, the Fund of Scientific Research—Flanders (FWO), and K.U.Leuven for financial support.

25. A. S. Klymchenko, S. Furukawa, K. M¨ullen, et al., Nano Lett., 2007, 7, 791.

REFERENCES 1. S. M. Clarke, Curr. Opin. Colloid Interface Sci., 2001, 6, 118. 2. G. Binnig, H. Rohrer, C. Gerber, and E. Weibel, Phys. Rev. Lett., 1982, 49, 57. 3. A. Ulman, Chem. Rev., 1996, 96, 1533. 4. J. P. Rabe and S. Buchholz, Science, 1991, 253, 424. 5. S. Furukawa and S. De Feyter, Top. Curr. Chem., 2009, 287, 87–135. 6. J. V. Barth, G. Costantini, and K. Kern, Nature, 2005, 437, 671. 7. C. Julian Chen, Introduction to Scanning Tunneling Microscopy (Oxford Series in Optical and Imaging Sciences), Oxford University Press, 1993.

26. S. M. Xu, G. Szymanski, and J. Lipkowski, J. Am. Chem. Soc., 2004, 126, 12276. 27. G.-J. Su, H.-M. Zhang, L.-J. Wan, et al., J. Phys. Chem. B, 2004, 108, 1931. 28. S. Yoshimoto, N. Higa, and K. Itaya, J. Am. Chem. Soc., 2004, 126, 8540. 29. S. Yoshimoto, N. Yokoo, T. Fukuda, et al., Chem. Commun. 2006, 500. 30. K. Suto, S. Yoshimoto, and K. Itaya, Langmuir, 2006, 22, 10766. 31. S. Yoshimoto, Y. Honda, O. Ito, and K. Itaya, J. Am. Chem. Soc., 2008, 130, 1085. 32. A. Ohira, M. Sakata, C. Hirayama, and M. Kunitake, Org. Biomol. Chem., 2003, 1, 251. 33. S. Yoshimoto, K. Suto, A. Tada, et al., J. Am. Chem. Soc., 2004, 126, 8020. 34. Y. He and E. Borguet, Angew. Chem. Int. Ed., 2007, 46, 6098. 35. L. Perez-Garcia and D. B. Amabilino, Chem. Soc. Rev., 2002, 31, 342.

8. S. Chiang, Chem. Rev., 1997, 97, 1083.

36. K.-H. Ernst, Top. Curr. Chem., 2006, 265, 209.

9. A. J. Groszek, Proc. R. Soc. Lond. A. Mat., 1970, 314, 473.

37. N. Katsonis, E. Lacaze, and B. L. Feringa, J. Mater. Chem., 2008, 18, 2065.

10. D. Bleger, D. Kreher, F. Mathevet, et al., Angew. Chem. Int. Ed., 2007, 46, 7404.

38. R. Raval, Chem. Soc. Rev., 2009, 38, 707.

11. N. Katsonis, J. Vicario, T. Kudernac, et al., J. Am. Chem. Soc., 2006, 128, 15537.

39. J. J. A. W. Elemans, I. De Cat, H. Xu, and S. De Feyter, Chem. Soc. Rev., 2009, 38, 722.

12. Y. Yang and C. Wang, Curr. Opin. Colloid. Int. Sci., 2009, 14, 135–147.

40. K. E. Plass, A. L. Grzesiak, and A. J. Matzger, Acc. Chem. Res., 2007, 40, 287.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Physisorption for self-assembly of supramolecular systems

25

41. M. Linares, P. Iavicoli, K. Psychogyiopoulou, et al., Langmuir, 2008, 24, 9566.

55. C. L. Feng, Y. Zhang, J. Jin, et al., Surf. Sci., 2002, 513, 111.

42. P. Iavicoli, H. Xu, L. N. Feldbord, et al., J. Am. Chem. Soc., 2010, 132, 9350.

56. Y. T. Shen, L. Guan, X. Y. Zhu, et al., J. Am. Chem. Soc., 2009, 131, 6174.

43. Q. Chen, T. Chen, D. Wang, et al., Proc. Natl. Acad. Sci. U.S.A., 2010, 107, 2769.

57. K. W. Hipps, Scanning tunneling spectroscopy, in Handbook of Applied Solid State Spectroscopy, Ed., D. R. Vij, Springer, Berlin, Heidelberg, New York, 2006.

44. A. M. Berg and D. L. Patrick, Angew. Chem. Int. Ed., 2005, 44, 1821.

58. M. Lundstrom, Science, 2003, 299, 210.

45. N. Katsonis, H. Xu, R. M. Haak, et al., Angew. Chem. Int. Ed., 2008, 47, 4997.

59. A. Aviram and M. A. Ratner, Chem. Phys. Lett., 1974, 29, 277–283.

46. V. Enkelmann, Adv. Polym. Sci., 1984, 63, 91. 47. Y. Okawa and M. Aono, J. Chem. Phys., 2001, 115, 2317.

60. F. J¨ackel, M. D. Watson, K. M¨ullen, and J. P. Rabe, Phys. Rev. Lett., 2004, 92, 188303.

48. A. Miura, S. De Feyter, M. M. S. Abdel-Mottaleb, et al., Langmuir, 2003, 19, 6474.

61. A. P. H. J. Schenning and E. W. Meijer Chem. Commun., 2005, 3245.

49. S. P. Sullivan, A. Schmeders, S. K. Mbugua, and T. P. Beebe, Langmuir, 2005, 21, 1322.

62. R. van Hameren, P. Sch¨on, A. M. van Buul, et al., Science, 2006, 314, 1433.

50. G. Z. Mao, W. F. Dong, D. G. Kurth, and H. M¨ohwald, Nano Lett., 2004, 4, 249.

63. J. Puigmarti’-Luis, A. Minoia, H. Uji-i, et al., J. Am. Chem. Soc., 2006, 128, 12602.

51. A. Ciesielski, S. Lena, S. Masiero, et al., Angew. Chem. Int. Ed., 2010, 49, 1963.

64. J. M. Mativetsky, M. Kastler, R. C. Savage, et al., Adv. Funct. Mater., 2009, 19, 2486.

52. B. Hulsken, R. Van Hameren, J. W. Gerritsen, et al., Nat. Nanotech., 2007, 2, 285.

65. C. Joachim and M. A. Ratner, Proc. Natl. Acad. Sci. U.S.A., 2005, 102, 8801.

53. R. T. Vang, J. V. Lauritsen, E. Legsgaard, and F. Besenbacher, Chem. Soc. Rev., 2008, 37, 2191.

66. A. Miura, Z. Chen, H. Uji-i, et al., J. Am. Chem. Soc., 2003, 125, 14968.

54. P. R. Davies and M. W. Roberts, Atom Resolved Surface Reactions: Nanocatalysis, RSC Publishing, Cambridge, UK, 2007.

67. H. Uji-i, A. Miura, A. Schenning, et al., ChemPhysChem, 2005, 6, 2389.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc146

Chemisorbed Self-Assembled Monolayers Omar Azzaroni and Roberto C. Salvarezza INIFTA-Universidad Nacional de La Plata-CONICET, La Plata, Argentina

1 2 3 4 5

Introduction Basic Concepts on SAMs Chemisorbed SAMs THIOL SAMs ON Au(111) Building Molecular Structures by Bottom-Up Approach Using SAMs 6 Applications of Self-Assembled Monolayers—a Brief Summary 7 Conclusions and Outlook References

1

1 1 2 3 6 14 15 15

INTRODUCTION

Since the second half of the twentieth century, the preparation and characterization of self-assembled molecular films on solid surfaces have attracted very great and widespread interest, both as a fundamental intellectual and technological challenge to chemists, physicists, and materials scientists. The very possibility of designing and creating surfaces from scratch marks a profound departure from traditional surface science. One major attraction of surface-confined molecular assemblies is its potential to combine and manipulate topological, chemical, and functional features that are essential for a wide variety of technological applications such as microanalysis, biotechnology, nanofabrication, or corrosion protection, just to name a few examples.

As we move further into the new century, self-assembled thin films seem indeed to offer almost unlimited opportunities for fundamental and applied surface science. At present, self-assembled monolayers (SAMs) constituted of chemisorbed species represent fundamental building blocks for creating complex structures by the so-called “bottom-up approach.” The self-assembly of molecules into structurally organized thin films exploiting the flexibility of organic and supramolecular chemistry has led to the generationsynthetic surfaces with well-defined chemical and physical properties. Chemical synthesis offers the appeal of an unparalleled level of control over the selection of functional features while hydrophobic and van der Waals interactions lead to the spontaneous association of the predesigned building blocks into stable, well-defined surface structures. In this work, we first sketch the fundamental aspects of chemisorbed SAMs as a tool for building complex molecular systems. Using thiol SAMs as model systems, we first briefly review the self-assembly, surface structure, and stability under different experimental conditions. We also point out the characteristics of SAMs that make them suitable especially for building active micro- and nanostructured molecular systems on surfaces, and stress their limitations resulting from defects, contaminants, and disorders. Finally, we present examples of interfacial architectures drawn from supramolecular and covalent systems to illustrate the potential of SAMs as robust platforms for functional 3D structures on solid substrates.

2

BASIC CONCEPTS ON SAMs

Self-assembly is the construction of systems without guidance from external sources other than those provided by the environment.1 SAMs are examples of intermolecular self-assembly that takes place at different interfaces.2 Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc147

2

Soft matter Terminal functional group

x

x

x

x

x

x

x

Hydrocarbon chain

Si Sulfur head

S

S

S

S

S

S

S

O O

O

Si O O

(a)

P

Si Si O O O O

(b)

P

O O O O O O

Si/SiO2

Metal

Substrate

O

P O O O

Metal oxide (c)

Figure 1 Chemisorbed SAMs. (a) Thiols, (b) silanes, and (c) phosphonates. Panel (a) describes the typical structure of chemisorbed alkanethiolate self-assembled monolayers on metal surfaces.

In particular, solid surfaces are often used as scaffolds for the construction of these supramolecular systems at gas–solid and gas–liquid interfaces. In fact, these twodimensional structures, usually with a thickness between 1 and 3 nm, are spontaneously formed on a variety of solid surfaces such as metals, semiconductors, and oxides by different molecules or mixtures of different molecules from both gas and liquid phases.3 The thickness can be precisely tuned through the molecular dimension and its arrangement on the surface. In a typical SAM, the molecules (building blocks) are bonded to the solid substrate by a reactive head that provides a strong molecule–substrate link to the system (Figure 1).3 Metals (Me) (such as gold, silver, and copper) easily react with thiol, alkylsulfide, or alkyldisulfide molecules forming a strong covalent S–Me (thiolate) bonds (typically ≈ 2 eV), in comparison the C–C bond is ≈ 3.5 eV (Figure 1a).3, 4 On the other hand, hydroxyl groups present on Me, Si, and glasses can react with alkylchlorosilanes, alkylalkoxysilanes, and alkylaminosilanes yielding covalent Si–O–Me or Si–O–Si (siloxane) bonds (4.6 eV)2, 5, 6 (Figure 1b). Also oxidized surfaces react with alkylphosphonate molecules7–9 and alkylphosphates10, 11 forming P–O–Me bonds (Figure 1c). On the other hand, dispersive forces between molecules stabilize the supramolecular assembly introducing long- or short-range ordering depending on the molecule and the substrate.2–4 The terminal group of the molecule provides chemical functionality to the SAM (Figure 1). They can be used to tailor the physical chemistry of the solid surface changing completely their properties, such as complexation or molecular recognition centers for sensing devices, microarrays, catalysis, and biocatalysis, and also to provide reactive chemical groups for linking other molecules in templating-directed synthesis for building preorganized three-dimensional molecular architectures.3

3

CHEMISORBED SAMs

Silanes, alkylphosphonates, and alkylphosphates form dense SAMs on oxidized surfaces through chemical reactions that involve the surface OH groups of the substrate. Alkylsiloxane SAMs can be formed on the surface of SiO2 /Si,5, 12–15 Al2 O3 /Al,16, 17 quartz,18, 19 and mica.20, 21 The two-dimensional system seems to exhibit chemical,22 mechanical,23 and thermal stabilities.24 The self-assembly process takes place by exposing the oxidized surface to a solution containing the silane molecules. However, the SAM quality strongly depends on the hydrocarbon chain length, solvent, and temperature. In many cases, siloxane films involve extensive cross-linking,25 and they are usually thick, not surface conforming, and have disordered alkyl chains.26 Alkylphosphonates and alkylphosphates self-assemble on common engineering metals such as steel, stainless steel, aluminum, copper, and brass,27 among many other oxidized surfaces such as Al,28 Ti,8 and Zr.29 Phosphonate SAM formation involves the adsorption of the phosphonic acid followed by heating. Layer formation of selfassembling molecules of alkylphosphonic acid on mica from ethanol involves nucleation, growth, and coalescence of densely packed islands of phosphonates.30 The structure of islands depends on the length of alkyl chains. Selfassembly of phosphonates has also been observed from aqueous solution using octylphosphonic acid. The height of these islands was practically equal to the length of molecule indicating the formation of a single molecular layer. Infrared reflection–absorption spectroscopy indicates that phosphonate SAMs on Zr(IV) have alkyl chains in a liquid-like environment.31 A higher degree of order and packing density within the monolayers was found for alkyl phosphates with alkyl chain lengths exceeding 15 carbon atoms self-assembled on Ti surfaces.32 The shift of the symmetric and antisymmetric C–H stretching modes in

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc147

Chemisorbed self-assembled monolayers the IR spectra to lower wave numbers observed is consistent with higher two-dimensional crystallinity. Experimental data also show that the molecules in the SAMs have an average alkyl chain tilt angle of 30◦ to the surface normal and intermolecular spacing of 0.5 nm similar to thiol SAMs on Au surfaces. SAMs based on organophosphonates are more durable than siloxane films.32, 33 Phosphonate SAMs are surface conforming and more ordered than siloxane layers. In fact, it was found that on native titanium oxide loading for the phosphonate was four times greater than that for the siloxanes, and also the hydrolytic stability of the siloxanes was poor.33 The metal bisphosphonate SAMs are shown to be receptive to complexation by organic acids and acid-containing polymers such as fluoropolymers and ethylene-co-methacrylic acid, opening interesting technological applications.27 Selenols (Se) on metals have been explored as an alternative to thiol SAMs. These monolayers are similar to thiol SAMs (Figure 1a) but with the replacement of S atom by Se. A comparative study of benzenethiol and benzeneselenol on Au(111) has shown that the selenol SAMs are more ordered than the thiol SAMs, but they are less strongly bound to the gold substrate than the thiol analog.34 In contrast, it has been recently reported that the selenium-based SAMs are more stable than their sulfur analogs.35 More experimental work on these systems is needed for a better understanding of selenol SAMs. However, our knowledge of silane, phosphonate, and selenol SAMs at the molecular level is considerably smaller than that we have for thiolate SAMs on metals and semiconductors, in particular for thiols on Ag and Au surfaces. These SAMs exhibit well-ordered (crystalline) structures that can be characterized by surface science techniques at molecular or even at submolecular level.3, 4 Therefore, in the next section, we concentrate on the thiol–Au(111) and Ag(111) SAMs, although information concerning thiol self-assembly on other surfaces such as metal and semiconductor surfaces are also included.

4 4.1

THIOL SAMs ON Au(111) Sam Preparation and Structure

Preparation of thiol SAMs can be achieved from gas and liquid phase by using thiols (aliphatic, aromatic), alkyldisulfides, or dialkylsulfides,2–4 on clean single crystal, rough or nanocurved surfaces of metals,3, 4 and semiconductors.36, 37 The SAMs can be studied by using a large variety of surface characterization technique.39 The gas phase self-assembly is performed in UHV (ultrahigh vacuum) chamber by introducing small amounts of

3

the reactive molecules that adsorb onto the clean substrate surfaces. This method is applied to short molecules with high vapor pressures, and it results in SAMs of good quality that can be “in situ” characterized by all the usual surface science techniques.38 STM (scanning tunneling microscopy) and LEED (low energy electron diffraction) results show that self-assembly initially involves, in the case of Au(111), the formation of lying down phases. By increasing the thiol pressure, the system undergoes a phase transition from the lying down to dense phases of standing up molecules.2–4 The stable phases detected √ √ at equilibrium are the 3 × 3 R 30◦ and its c(4 × 2) √ √ supperlatice for Au(111) (Figure 2),2–4, 40 7 × 7 R 19.1◦ for Ag(111).2, 3, 41 On the other hand, for Cu(111) two structural phases are observed, a “honeycomb” phase and a pseudo-(100) reconstructed surface phase.42 The nearest neighbor distances in these cases is between 0.5 nm in Au(111)2–4 and 0.43–0.47 nm in Ag(111),2, 3 close to the distances observed in solid alkanes by X-ray diffraction. Coverage consistent with these structures has also been observed for thiols on Ni(111).43 Dense phases of standing up alkanethiols on GaAs37 and InP44 have also been reported. The tilt angle of the molecules in the standing up configuration is ≈10–19◦ for Ag(111) and Cu(111),2, 45, 46 14–18◦ for Pd,46 and 30–40◦ for Au(111).2, 47, 48 In the case of self-assembly from liquid phase, a clean Au, Ag, or Cu substrate is immersed in thiol containing solutions of ethanol, hexane, toluene, benzene, or neat thiols depending on the substrate–molecule system, for times ranging from minutes to days depending on the thiol molecule.2, 3 Well-organized SAMs with crystalline order usually requires times ranging from 12 to 24 h as chain organization is a slow process. The role of the solvent is crucial to obtain high-quality SAMs. It is well known that the molecular environment around a supramolecular system is also of prime importance to its operation and stability. In fact, some solvents are able to form hydrogen bonds, and exhibit electrostatic and chargetransfer properties. Therefore, they can interact strongly with the supramolecular assembly. For instance, for reactive metals, such as Cu, hexane or toluene is preferred to ethanol to avoid oxide formation, although it has been recently claimed that better results are obtained by using self-assembly in alkaline solutions where the oxide is completely reduced by the thiol.49 In the case of dense standing up phases of dithiol SAMs on Au, the selfassembly from hexane solutions is preferred as it has been observed that ethanol induces oxidation of the terminal SH group leading to disulfide formation.50, 51 In solution deposition (ethanol, hexane), the lying down phases reported on Au(111) substrates are not observed and the system evolves directly to the same standing √ √ up structures ( 3 × 3 R 30◦ and c(4 × 2) supperlatice)

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc147

4

Soft matter

Rectangular c (4× 2)

3 × 3 R 30° (a)

Zig-zag c (4×2)

(b)

Figure 2 STM images showing well-ordered domains of alkanethiols on Au(111): (a) hexanethiol (20 × 20 nm2 ) and mercapto √ (b)√ undecanoic acid (18 × 18 nm2 )-covered Au(111) surfaces. The rows and the dense regions correspond to c(4 × 2) and 3 × 3 R 30◦ domains, as shown in the high-resolution (4 × 4 nm2 ) STM images. (Reproduced from Ref. 40.  American Chemical Society, 2001.)

observed in gas phase.4 For Ag surfaces and also for Cu substrates, the thin native oxides (silver oxide or copper oxides) are spontaneously reduced by the thiol molecules to metallic Ag or Cu while they are oxidized to sulfonates. Afterward the fresh metallic surfaces react with additional thiol molecules resulting in ordered SAMs.52, 53 In the case of Pd, the reactive Pd surface initially breaks the S–C bond of the thiol molecules producing a diluted sulfide layer that practically “passivates” the Pd surface allowing further chemisorption of thiol molecules.54, 55 Thus, in this case the surface is composed by a mixed sulfide–thiol adlayer. Some amount of S has also been observed for thiol self-assembly on Cu surfaces.56 In situ STM in electrolyte solutions has also been shown that thiols form (2 × 2) and √ √ 3 × 3 R 30◦ on Pt(111) surfaces.57 The formation of alkanethiols SAMs on Pt(111) have also been observed by XPS and complementary techniques.58 Dense SAMs of long alkanethiols can be formed on polycrystalline Ni and Ni(111) surfaces in aqueous solution under potential control in order to eliminate the NiO layer from the surface.59 Irrespective of the metal, thiol covered metal exhibits in XPS (X-ray photoelectron spectroscopy) data the characteristics of a “thiolate bond” (162 eV).4 This analysis also shows, in some cases, small amounts of adsorbed S (161 eV), with the exception of the above-mentioned Pd surfaces that have about 0.4 monolayer of S as sulfide, and

free thiols (163 eV). The latter can be removed by carefully rinsing with the solvent. The self-assembly process of these molecules is accompanied by a reconstruction of the metallic surfaces. In fact, experimental and theoretical evidences seem to indicate that the Au(111),4 Ag(111),41, 60 Cu,61 Ni(111),43 and Pd(111)53, 54 surfaces reconstruct resulting in the case of Au(111) in the formation of a metal adatom–thiolates complexes on the surface.4 From the above discussion, we can conclude that SAMs of thiols can be formed on different metal and semiconductor surfaces opening a wide range of possible applications in technology. In the next section, we discuss the presence of defects, contaminants, and the stability of these systems under different environmental conditions that could seriously limit their applications.

4.2

Defects, contaminants, and stability

The stability of the thiol SAMs is extremely important for their applications. Thermal stability is restricted up to ≈ 100◦ C.62 Increasing the temperature above this value results in thiol desorption mainly as disulfides and finally as thiol molecules. Recent results from thermal-programmed desorption have shown that disulfides are desorbed at around 400 K.63 These disulfides arise from the thiol

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc147

Chemisorbed self-assembled monolayers molecules in standing up configuration present in dense, well-defined SAMs. On the other hand, desorption from the lying down molecules of diluted thiol phases has been observed around 500 K. Finally, desorption of an appreciable amount of gold-containing molecules is observed around 700 K, thus supporting the presence of thiolate–gold complexes. The stability range of thiols is smaller than that exhibited by self-assembled silane monolayers that remain stable up to 500◦ C.64 Chemical stability is also an important factor. In ambient conditions and also in liquid media such as ethanol, the presence of light and O2 leads to thiolate oxidation to disulfides and also to sulfonates.65 Both species are easily desorbed from the surface so that continuous exposure results in SAMs degradation. It has been proposed that O3 molecules react with the S heads forming oxidized S products. This is supported by the fact that photolithography performed with UV radiation is usually employed to patterning thiol SAMs covered metals.66 Oxidation takes place at a fast rate for short hydrocarbon chains.67 Recent results of methanethiol SAMs on Au formed form ethanolic solutions of dimethyldisulfide indicate a rapid decomposition to form adsorbed S.68 Rapid degradation yielding S species has been also reported for 4-mercaptopyridine on Au(111) surfaces.69 Degradation of the SAM terminal groups has also been reported. The exposed–SH groups of hexanedithiol on Au(111) can form disulfide species with adjacent chemisorbed species and/or nonchemisorbed dithiol molecules during self-assembly in ethanolic solutions.50 This process is avoided by performing the selfassembly process in hexane, and in the absence of O2 and light. Also light and O2 cause the oxidation of amineterminated SAMs.70 Using time-of-flight secondary ion mass spectrometry (ToF-SIMS), the oxidation of the amino groups to nitroso groups was evident when exposed under ambient conditions. Electrochemical stability is also important in device fabrication and SAM metallization.71 In many devices such as amperometric sensors and biosensors, thiol SAMs are important structural or functional units.72 Thiol exhibits reductive and oxidative desorption at threshold potentials (which depend on the chain length) below or above which they are reductively desorbed or oxidized, respectively. Therefore, the potential reached by the system must be within the stability range of the thiol SAMs. For reductive desorption the stability range of the SAM increases to ≈ 0.04 V/CH2 in aqueous solutions and 0.025 V/CH2 in 95% methanol + 5% water, revealing the important role of hydrophobic interactions in SAM stability.73 In general, no changes have been reported for reductive desorption of thiol SAMs in the pH range 5–13, but the system becomes more instable at lower pH values.71 At a constant pH and for a given thiol, the electrochemical stability

5

range increases as Au71 < Ag71 ≈ Ni74 < Pd75 ≈ Cu.71 Therefore, for electrochemical applications, one can select the metal considering the stability range. Thus, for SAMs metallization from electrodeposition Cu seems to be an excellent candidate. SAMs exhibit structural and conformational defects. Structural defects consist of molecular vacancies, missing rows, and domain boundaries resulting from the nucleation and growth process while conformational defects involve chain disorder.2, 4 Chain disorder for aliphatic short thiols is revealed by the asymmetric stretching of the methylene groups detected at wave numbers >2920 cm−1 .47, 48 On the other hand, well-ordered chains, typical of long thiols (C > 12), exhibit the asymmetric stretching of the methylene groups at wave numbers 1. However, caution has to be taken using this model because the above predicted limits set on the values of S are relatively insensitive to the exact values of V and a but are strongly dependent upon the choice of l.54, 55

7.2

Ionic complexes and self-assembly

In a similar way to how hydrogen-bonded systems form thermotropic and lyotropic mesophases first by selfassembly then by self-organization, so too do ionic systems. However, the interactions for ionic materials are

often stronger than those for hydrogen bonding. Such liquid crystals can be truly ionic or they can be CT systems. Self-assembling ionic materials have drawn interest because of the possibility of creating anisotropic conductors. Ionic material (a) in Figure 34 56 has three aliphatic chains attached to an aromatic ionic system. As with the hydrogen-bonded systems described previously, selfassembly occurs through the ionic groups and the aliphatic chains form a curved periphery to give a columnar phase made up of wedge-like components. Again the columns are probably disordered and not composed of supramolecular discs. However, the structure that is formed resembles insulated arrays of ionic wires. Down the cores of the columns the ionic species are present, whereas the aliphatic insulators coat the outside of the columns. A similar effect is seen for material (b) in Figure 34.57 This simple system comprises a diallylamine quaternary salt and an attached aliphatic chain. In aqueous media, it self-organizes into columns with the ionic head groups toward the outside of the columns, thereby forming a normal hexagonal phase. This self-organized system can be further cyclopolymerized through the diallylamine moiety to give an anisotropic polymeric self-organized conducting material, and if a cross-linking bis-diallylamine unit is mixed with a host diallylamine then a conducting polymer network can be created. Liquid-crystalline CT interactions have also attracted increasing interest in the field of soft materials related to synthetic metals, organic semiconductors, molecular electronics, and nonlinear optics. In systems composed + OC12H25 e

C12H25O RO

NC

CN

NC

CN

e−

Cl−

RO

+

N +

RO

N

BF4−

N

R

S

H

S

CH3

R = C8H17 and C12H25

R = C11H23

(a)

C12H25O

(b)

RO

OC12H25

(c)

TCNQ

OR NO2

RO

O2N

OR RO

NO2 O TNF

OR HAT (d)

Figure 34

21

Ionic and charge-transfer liquid-crystalline materials.56–60

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

22

Soft matter

of good electron donors and good electron acceptors, the formation of CT complexes is characterized either by CT transitions in the UV/vis absorption spectra or at least by a broadening of the absorption. However, relatively little is known about the significance of electrostatic versus CT interactions for self-organization in condensed phases. Especially for systems without spectroscopic evidence for the presence of strong CT interactions, an approach based on electrostatic forces involving the out-of-plane π-electron density seems to be sufficient to explain geometrical requirements for interactions between the aromatic moieties of molecules. As with other systems discussed previously, it is possible for the donor and/or the acceptor not to be liquid crystal, but for the self-assembled complex to be so. Intercalations between donor and acceptor can serve to stabilize mesomorphic behavior. Consider system (c) in Figure 34,58 the acceptor 7,7,8,8-tetracyano1,4-quinodimethane (TCNQ) is not liquid crystalline, but the donor tetra(4-dodecyloxyphenyl)dithiapyranylidene is mesomorphic; however, the identity of the phase is not clear. The two materials associate face-to-face through π –π interactions and possibly through electron transfer. The structure of the resulting phase is a cross between lamellar and columnar, giving a so-called lamello-columnar phase which is made up of one-to-one intercalated donor and acceptor molecules. Often materials that form CT complexes form stronger complexes when they have large conjugated aromatic ring systems, which means they are disclike. For example, hexa-2,3,6,7,10,11-alkoxytriphenylenes complex readily with 2,4,7-trinitro-9-fluorenone (TNF) in a 6 : 4 intercalated arrangement to give a hexagonal ordered columnar phase with a clearly defined melting point and phase transitions.59, 60 Interestingly, the HAT (hexa-alkoxytriphenylenes) homologues that do not exhibit mesomorphic behavior still form columnar phases on the addition of TNF.

7.3

Chromonic liquid crystals

Chromonic liquid crystals are a separately defined family of lyotropic phases (i.e., water- or solvent-based systems), which includes materials such as drugs, dyes, nucleic acids, antibiotics, carcinogens, and anticancer agents.8, 9 In contrast to conventional amphiphiles, such as soaps, detergents, and biological lipids, chromonic materials do not possess significant surfactant properties, that is, they do not usually possess much in the way of aliphatic content. The molecular structures of chromonic materials are often disc-like or board-like; they are essentially composed of aromatic ring systems, with hydrophilic, ionic, or hydrogenbonding solvating substituents. The molecules can be thus

regarded as being insoluble in one dimension, with the basic structural unit being a molecular stack of some form, rather than a micelle. The tendency for the molecules of chromonic phases to aggregate into columns is present even in dilute solution where column formation occurs before a phase is formed. Although there may be a threshold concentration before aggregation begins to occur, there is no optimum column length and hence no critical concentration, with a further distinction being the absence of a Krafft temperature. Since the process of mesophase formation does not depend on the presence of flexible alkyl chains, there is no threshold temperature below which mesophases cannot form because the vital flexibility of the molecules has been frozen out. There are two principal chromonic mesophases, which have become known as the N and M phases, there are other phases, such as the lamellar phase, but these are not so common. The N phase has a nematic array of columnar stacks of the board-like molecules, with the only regular repeat distance being the stacking repeat of 0.34 nm along each column.61 The stacking into supramolecular rods in turn means that the structure of the nematic phase is not dependent on singular molecular species self-organizing. Instead, the N phase is another example of a superphase. Similarly, the M phase possesses columns of the board-like molecules but this time arranged on a hexagonal lattice, making it comparative in structure to a columnar hexagonal phase. The two phases are shown together in Figure 35. There are a number of examples of drugs and dyes which exhibit chromonic phases, see Figure 36. One of the most notable is disodium cromoglycate (DSCG), which goes under the trade name of INTAL and is an antiasthma drug,9, 61–63 additionally it was shown to be an inhibitor of mast cell degranulation and histamine release induced by phospholipase A. The material has been shown to selfassemble into columns and to form supernematic, N, and columnar, M, phases, but if this organization has any effects on the treatment of asthma is as yet unclear. Nevertheless, similar patterns of mesophase formation have been found in other antiasthmatic drugs such as the 7,7 -analog of INTAL. DSCG notably has two ionic substituents which interact strongly with water in the self-organizing process. Its structure as such is not so dissimilar from many dyes. Figure 36 shows a selection of dyes that are also chromogenic, some are rod-like and others disc-like, however, what is common is their ionic substituents, which can be terminal or lateral, singular or multiple. In water, and potentially other hydrophilic solvents, strong solvent solute interactions are expected. Shielding of the ionic substituents by solvent possibly makes way for stacking of the aromatic parts of the ˚ which is very similar to molecules at a distance of 3.4 A, face-to-face distances found in columnar liquid crystals.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

Liquid-crystalline materials

(a)

N Phase

(b)

23

M Phase

Figure 35 The structures of the N (a) and M (b) chromonic phases. The N phase resembles the structure of the nematic phase, whereas the M phase is similar to a hexagonal columnar phase. In each case, the molecules are shown as boards.

O

O

O

O

OH Na+ −OOC

COO− Na+ COO− Na+

Disodium cromoglycate DSCG -INTAL Na+ −OOC N Na+ O−3S

N

Cu

N

N

N N

N

CH3

Methyl orange

N

N

CH3

N

N Na+ −OOC

Na+ O−3S

N

N N

N

Sirius supra brown RLL

Figure 36

8

COO−Na+

Copper-tetracarboxyphthalocyanine

CH3

N N

O

Na+ O−3S

Structures of some drug and dye materials that exhibit chromonic phases.

THE KINETICS OF SELF-ASSEMBLY AND SELF-ORGANIZATION

As noted in the previous few sections, interactions such as hydrogen bonding can be dynamic, whereas stronger interactions, such as ionic, can lead to more stable and long lasting complexation. When the interactions tend toward being dynamic, in addition to thermodynamics, kinetics start to play a role in the formation of liquid crystal phases. For simple molecular materials, the phase sequences are usually reversible as a function of

temperature, except for the recrystallization temperature which is dependent on the kinetics of the nucleation process. Similarly, the process of self-assembly for large molecular systems can also be caught up with the thermodynamic processes. For example, the hexadecyl- and the octadecyl-homologues of the 4 -alkoxy-3 -nitrobiphenyl-4carboxylic acids both exhibit lamellar phases associated with molecular materials. However, for the hexadecyl homolog, the two phases, smectic A and smectic C, are separated by a cubic supramolecular phase. As a consequence, the phase transitions between the three phases are

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

24

Soft matter

9

O2N O CnH2n +1O

O

n = 16 or 18 H

n = 16 Cryst 126.8 SmC 171 Cubic 197.2 SmA 201.9 °C Iso Liq 126.8 Cryst

171.0

198.0

SmC

Iso Liq

SmA

Cubic +0.4 to −30

+3 to −10

170.0 +/− 2

199.8

191.0

+/− 2

193.3 +/− 2

Columnar

Figure 37 The phase types, transition temperatures, and hysteresis for 4 -hexadecyloxy-3 -nitrobiphenyl-4-carboxylic acid.

not necessarily reproducible or reversible. Under certain circumstances, the cubic phase can give way to the formation of a columnar phase. Figure 37 shows the various phase types and pathways for the phase sequences and the associated transition temperatures and hysteresis.19 It was proposed that the acids dimerize and the repulsions between the nitro-groups induced the self-assembly of the dimers into discs, see Figure 38(a). The supramolecular discs could stack together to form a columnar structure; however, repulsions of the nitro-groups would be expected to occur up and down the column, as shown in Figure 38(b). If this happens then the ends of the growing columns would be subject becoming conical, and under these conditions it was proposed that nucleation of another column, effectively through twining might occur, thereby giving a jointed rodlike structure, see Figure 38(c), which is the basis for the formation of a cubic phase. Depending on the heating and cooling rates and the nature of the surfaces of the sample holder, various mesophase sequences can be obtained. Thus, the kinetics of the self-assembling processes are affecting the selforganization processes into condensed phases, because of this the phase transitions are no longer reversible and are subject to experimental history. Such effects are seen with a variety of complex molecular systems that are mesomorphic.

(a)

(b)

TOPOLOGY AND NANOSCALE SEGREGATION IN ROD-LIKE MOLECULAR MATERIALS

The following sections are concerned with the interactions that stabilize mesophase formation. The preceding descriptions demonstrate that liquid crystal formation is primarily dependent on molecular or molecular ensemble shape coupled often with anisotropic interactions or nanophase segregation. A number of the types of interaction have been discussed, and so as the variations of shape and interactions are many, the following description will focus on the weak interactions (van der Waals, polarizability, and polarity) of rod-like systems because these are most often encountered in the design of liquid crystal systems. For low-molar-mass materials with rod-like molecular shapes, the prototypical molecular design of a liquidcrystalline material involves the incorporation of a central aromatic, heterocyclic, or alicyclic core unit, to which are attached terminal aliphatic chains,64–66 thereby engendering dichotomous structures with rigid or semi-rigid sections surrounded or segregated by flexible fatty chains. When molecules with this type of architecture self-organize they generally do so with their rigid, aromatic parts tending to pack together and their flexible/dynamic aliphatic chains orienting together, as depicted in Figure 39 for sweets. “The central parts connected with the sweets associate together, leaving the twisted wrapper ends to correlate separately.” Thereby the overall system becomes so-called locally nanophase segregated. Even at smaller distances, further correlations become possible as shown by the layers of sweets in Figure 39(b). In the central section, the light and dark colors become forced to correlate in clusters by the shape of the sweet itself. The two ends of the wrappers are colored differently to show that the sweets can be either way up in the layers, which by commensurability/incommensurability with respect to the layers can affect domain size and so on. As the possible variations of packing molecules, sweets in the above description, which is relatively large, the main

(c)

Figure 38 (a) Dimerization and self-assembly of 4 -hexadecyloxy-3 -nitrobiphenyl-4-carboxylic acid, (b) the formation of columns by the self-assembled discs, and (c) the macroscopic organization of the columnar units into a cubic phase. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

Liquid-crystalline materials

(a)

25

(b)

Figure 39 A schematic representation of the shapes of low-molar-mass materials that can self-organize to form a (a) nematic phase and a (b) lamellar phase.

target of material design, particularly for the display industry, has been, by default, the variation in the structure of the central core region of the molecules in the belief that the core is more important in influencing mesophase incidence, mesophase temperature range, isotropization point, melting point, mesophase sequence, dielectric and optical anisotropy, elastic coefficients, and the reorientational viscosity associated with the mesophase.67 However, if we consider the common “materials design” comprising a central core unit with two attached peripheral groups, usually aliphatic chains, then it is easy to understand how the differences in interactions between the aliphatic flexible chains and the aromatic rigid cores can lead to so-called nanophase segregation where the like-parts of the molecules pack together and repel the dissimilar parts. As a consequence, the development of property–structure correlations for material design has involved the synthesis of many families of materials where the central core has been kept constant and the peripheral flexible units have been varied (usually in length), or the flexible units have been arranged to be constant and the central core has been varied.64 Figure 39 also demonstrates another aspect of mesophase formation and that is the topological shapes of the molecules. The sweets in Figure 39(a) are fatter in the middle than at the ends, and so they are more likely to pack in a disordered way because they cannot be easily arranged into layers in order to fill space effectively. Conversely, the sweets in the Figure 39(b) show that as the cross-sectional areas of the end and the middles of the sweets are similar

it is easier to organize the sweets into layers. The structure (a) is effectively that of a nematic, whereas (b) is smectic or layered. Thus, two types of self-organization have been created solely by shape, and without consideration of any attractive or repulsive effects. The attractive and repulsive interactions include polar, dipolar and quadrupolar, polarizability and induced dipoles, van-der-Waals interactions, and the effects of nanosegregation. The interactions in typical liquid crystal systems would be expected to be anisotropic, but not necessarily directional as the molecules often choose to minimize the interactions with as many molecules pointing in one direction as the opposite direction. What molecular shape does is to determine which of the interactions are the most important in mesophase formation, and conversely strong molecular associations can determine shape and packing arrangements. In all cases, void volumes are to be avoided if possible, if not molecular frustrations can occur in order to utilize space effectively. Let us take first the relatively simple example of 4pentyloxy-4 -cyanobiphenyl (5OCB). It has a strong dipole parallel to the long axis of the molecule, associated with the donor alkoxy group and the acceptor nitrile. Therefore, the individual molecules will be polarized and in the bulk phase they will be polarizable. The material will be prone to nanosegregation as the aromatic units will associate more strongly through π –π interactions than the s–s interactions of the chains. Parallel correlations of the molecules will be preferred, but to minimize the dipolar interactions, the molecules will be required to pack in a head-to-tail

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

26

Soft matter

O

CN C

O

N

d

d

N

d

C

O d

(a)

(b)

Figure 40 (a) The minimized structure of 4-pentyloxy-4 -cyanobiphenyl and its dimerization in the gas phase at absolute zero. (b) The polarization of 4-pentyloxy-4 -cyanobiphenyl and the quadrupolar couplings that stabilize a staggered antiparallel arrangement.

arrangement. In order to maintain the π –π interactions, the molecules will be required to partially overlap, which has the added benefit of creating quadrupolar couplings which further stabilize this arrangement as shown in Figure 40. The dominant associations are therefore with the aromatic parts of the molecules, and now the molecular pairs have a shape somewhat similar to the sweets in Figure 39(a). Of course, the interactions will all be subject to minimization of distance associated with Lennard–Jones potentials. Nevertheless, this analysis suggests that the material would be expected to exhibit a nematic phase, which it does, with pairing length of approximately of 1.4 × the molecular length, which has been confirmed experimentally by X-ray diffraction. If this model is extended to a simple dialkyl system such as the alkyl 4 -n-alkoxybiphenyl-4-carboxylates, the second aliphatic chain fills the space effectively thereby reducing the void volume, and requiring no dimerization, as shown in Figure 41. Even in this case there is an advantage for antiparallel arrangements of the molecules, but as the molecules will be locally close packed in a low energy hexagonal arrangement there will still be parallel and antiparallel associations. Given that the molecules in liquid crystal phases are in various degrees of dynamic motion, what interactions are allowable/possible? This issue is bound up with the molecular topology, and potentially its dynamical shape. Determination of molecular shape is difficult because the molecules are often oscillatory or in rotational motion, and d

moreover measurements from X-ray diffraction shows that in this motion they share each others space. Thus, modeling of molecular topology is not an exact science. In its simplest and crudest form, the calculation of the shape of a molecule as it rotates about its long axis can be made by projecting the molecular structure of the mesogen onto a flat surface.68 Typically, this involves drawing out the molecular structure in its all trans conformation (other conformations can be used if required) by use of known bond lengths and bond angles. This provides the simple “s” bond framework of the molecular skeleton. The molecule can then be given its body by the addition of van-derWaals radii associated with the electrons of each atomic center. The position of the long axis is determined from the minimum mass inertia axis through the atomic centers. Reflection of the structure across the minimum inertia axis gives the cross-sectional shape of the molecules as it rotates about its long axis. An example of a molecular cross-sectional shape is shown in Figure 42 for N,N  terephthalylidene-bis-4-n-butylaniline (TBBA). TBBA has been selected because it has a symmetrical molecular structure and it exhibits nematic, and smectic A and smectic C phases, where the smectic A phase has its molecules orthogonal to the planes of the layers, whereas in the smectic C phase they are tilted. Once having determined the shape of the cross-sectional area of the molecule as it rotates about a known long axis, the shape can be used to investigate the packing requirements of the mesogen under consideration. For

O

O

d O

O O

d

O d

d O O d

(a)

O

d

O O

O

Parallel orientation

(b)

d

Antiparallel orientation

Figure 41 Polarizabilities of the core sections of the alkyl 4 -n-alkoxybiphenyl-4-carboxylates. (a) The parallel arrangement, and (b) the antiparallel arrangement showing the formation of a quadrupole. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

Liquid-crystalline materials

27

15 N

N

Overall length ~28 Å

Å

10

5

0

5

10

15

20

25

30

Å

Figure 42 inertia.

The cross-sectional area of N,N  -terephthalylidene-bis-4-n-butylaniline (TBBA) as it rotates about its minimum axis of

Interactions first favor tilting but then formation of layers becomes destabilized Polarizable Schiff’s bases

Soft aliphatic chains Layer planes

Aromatic core

Slide surface of adjacent molecules over one-another Favors smectics

Favors nematics

Figure 43 Packing of the shapes of the cross-sectional areas of TBBA showing progressive tilting as a function of passing crosssections over one another. The low tilt version gives the best packing arrangement and the highest degree of interaction between the polar and polarizable parts of the molecules.

example, if cross-sections of TBBA are taken together they can be arranged to pack together in parallel arrangements, see Figure 43 (up or down orientations do not matter as TBBA is symmetrical). The most polarizable sections of TBBA are associated with the aromatic core, and in particular, the Schiff’s base linkages and their related dipoles (shown as light gray circles in the figure). By taking two such cross-sections and passing one over the other in a parallel manner, minimum energy positions can be located where strong dipolar interactions and higher polarizability occur. It can be seen from the figure that a slightly tilted arrangement of the molecules is most stable. As the tilt

increases, the minimum energy positions correspond to ever increasing tilts with respect to the layers. Zero tilt has a greater separation between the polar regions than the small tilt variation, and so it is concluded that the arrangement with a small tilt is the most stable because the molecules pack most closely together with good overlap of the polarizable regions, which is also observed by experiment. Thus, the shape allows certain interactions to occur, but maintains a reasonable degree of separation so that the interactions remain weak. Unlike small molecular materials, as described earlier, the topology of large super- and supramolecular systems

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

28

Soft matter

O

O

O O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O

O O O O

O

O

O O

O

O

O

O O O

O

O

O

O

O

O

O

O

O

O O O

O O

O

O

O

O O

O

O

O

O

O

O

O

O

O

O

O O

O

O

O

O

O

O

O

O O

O O

O

O O O

O

O

O

O

O

O

O

O

O O

O O

O

O

O

O

O O

O O O O

O

O O

O O

g 47 N* 103 Iso Liq

Figure 44

A chiral fullerodendrimer which exhibits a chiral nematic phase.

appears to be the predominant factor in determining the selforganizing process. As a consequence, the structures of such large systems can be somewhat simplified. For example, a supermolecule possessing a spherical scaffold with chiral mesogenic units laterally linked to the scaffold is shown in Figure 44.23, 69 The lengths of the rod-like mesogens are of a similar size to the diameter of the scaffold which is composed of a [C60 ] fullerene with short aliphatic linking units. The mesogenic units are chiral by virtue of having

Figure 45

asymmetric terminal aliphatic chains derived from (S)-2butanol. The linking units are bifurcated, thereby allowing 12 mesogenic units to surround the spherical core of the scaffold. Modeling demonstrates that the mesogens cannot pack together around the fullero-scaffold without twisting. As they are chiral, the twist will be in one preferred direction; hence a molecular boojum is potentially formed with defects at the poles. The material itself, because the mesogens are laterally attached, exhibits a chiral nematic phase.

Schematic representations of the chiral nano-object.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

Liquid-crystalline materials Figure 45 shows a schematic representation of the fullerodendrimer with the mesogenic units spiraling around the spherical scaffold. Such spherical nanostructures that possess a helical surface topological relief can pack together through their defects so that the twist might propagate. Although this is a relatively small scale system, if the scaffolds are increased in size, this may be via the use of a nanoparticle at the center, such as gold, the surface decoration by laterally appended chiral mesogens can also create spheres with topological reliefs. Thus, extremely large-scale lattices may be possible with quasi-bonding between the particles. Thus, if shape is the determinant for the selforganization of supermolecular systems, then obviously surface topography is an important issue to consider for rods, discs, and spheres, and combinations thereof, and large systems should be considered as nano-objects.

10

CONCLUSION

In summary, we have used this conceptual article to describe how the availability of intermolecular interactions influence mesophase formation and stability at the nanoscale length regime. We describe how the whole of the molecular architecture and its interactions have to be considered and as a consequence the longitudinal interactions are equally as important as the lateral ones that are usually manipulated in the process of the molecular design of materials of commercial interest. Thus taking a “holistic” approach, we conclude that it is the strength of the allowable dipolar and quadrupolar interactions, which are dependent on the dynamic and static steric shapes of the molecules, that determines mesophase stability. Upscaling of molecular structure to the meso-length scale appears to result in the smearing out of the intermolecular interactions, and as a consequence molecular shape becomes the important determinant of mesophase formation. Such supersized systems should thus be considered as nano-objects.

REFERENCES 1. P. G. de Gennes, The Physics of Liquid Crystals, Oxford University Press, London, 1974. 2. E. B. Priestley, P. Wojtowicz, and P. Sheng, Introduction to Liquid Crystals, Plenum, New York and London, 1976. 3. J. W. Goodby, in Handbook of Liquid Crystals Vol 2A: Low Molecular Weight Liquid Crystals I, eds. D. Demus, J. W. Goodby, G. W. Gray, et al., Wiley-VCH, Weinheim, 1998, Ch 1, pp. 3–21. 4. J. W. Goodby, in Handbook of Liquid Crystals Vol 2A: Low Molecular Weight Liquid Crystals I, eds. D. Demus, J. W. Goodby, G. W. Gray, Wiley-VCH, Weinheim, 1998, Ch V, pp. 413–440.

29

5. J. Malthˆete, A.-M. Levelut, and H.-T. Nguyen, J. Phys. (Paris) Lett., 1985, 46, L–875. 6. Phasmids and Polycatenar Mesogens, H.-T. Nguyen, C. Destrade, and J. Malthˆete, in Handbook of Liquid Crystals Vol 2B: Low Molecular Weight Liquid Crystals I, eds. D. Demus, J. W. Goodby, G. W. Gray, Wiley-VCH, Weinheim, 1998, Ch 1, pp. 865–885. 7. G. Pelzl, S. Diele, and W. Weissflog, Adv. Mater., 1999, 11, 707. 8. T. K. Attwood, J. E. Lydon, and F. Jones, Liq. Cryst., 1986, 1, 499. 9. J. E. Lydon, in Handbook of Liquid Crystals Vol 2B: Low Molecular Weight Liquid Crystals I, eds. D. Demus, J. W. Goodby, G. W. Gray, Wiley-VCH, Weinheim, 1998, Ch 1, pp. 981–1007. 10. H. Sackmann and D. Demus, Mol. Cryst. Liq. Cryst., 1966, 2, 81. 11. H. Sackmann, in Liquid Crystals of One- and TwoDimensional Order, eds. W. Helfrich and G. Heppke, Springer-Verlag, New York, 1980, p. 19. 12. J. W. Goodby and G. W. Gray, in Handbook of Liquid Crystals Vol 1: Fundamentals, eds. D. Demus, J. W. Goodby, G. W. Gray, et al., Wiley-VCH, Weinheim, 1998, pp. 17–23. 13. S. Chandrasekhar, Liquid Crystals, Cambridge University Press, Cambridge, 1977. 14. R. M. Richardson, A. J. Leadbetter, and J. C. Frost, Mol. Phys., 1982, 45, 1163. 15. P. J. Collings and M. Hird, Introduction to Liquid Crystals; Chemistry and Physics, Taylor and Francis, London, 1997. 16. M. J. Freiser, Phys. Rev. Lett., 1970, 24, 1041. 17. L. A. Madsen, T. J. Dingemans, M. Nakata, and E. T. Samulski, Phys. Rev. Lett., 2004, 92, 145505. 18. B. R. Acharya, A. Primak, and S. Kumar, Phys. Rev. Lett., 2004, 92, 145506. 19. G. W. Gray and J. W. Goodby, Smectic Liquid Crystals— Textures and Structures, Leonard Hill, Glasgow and London, 1984. 20. A. J. Leadbetter, in Thermotropic Liquid Crystals, Critical Reports on Applied Chemistry, ed. G. W. Gray, John Wiley and Sons, Chichester, Vol 22, 1987, pp. 1–27. 21. A. de Vries, A. Ekachai, and N. Spielberg, Mol. Cryst. Liq. Cryst., 1979, 49, 143. 22. I. M. Saez and J. W. Goodby, J. Mater. Chem., 2005, 15, 26. 23. J. W. Goodby, I. M. Saez, S. J. Cowling, et al., Angew. Chem. Int. Ed., 2008, 47, 2754. 24. J. W. Goodby, I. M. Saez, S. J. Cowling, et al., Liq. Cryst., 2009, 36, 567. 25. I. M. Saez and J. W. Goodby, in Liquid Crystalline Functional Assemblies and their Supramolecular Structures, Structure and Bonding Series, eds. D. M. P. Mingos and T. Kato, Springer-Verlag, Berlin Heidelberg, 2008, Vol. 128, pp. 1–62. 26. T. Kato, N. Mizoshita, and K. Kishimoto, Angew. Chem. Int. Ed., 2006, 45, 38. 27. B. M. Rosen, C. J. Wilson, D. A. Wilson, et al., Chem. Rev., 2009, 109, 6275.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

30

Soft matter

28. G. R. Newkome, C. D. Weiss, C. N. Moorfield, I. Weiss, Macromolecules, 1997, 30, 2300.

and

49. C. Tschierske, J. Mater. Chem., 2001, 11, 2647.

29. C. T. Imrie, Liq. Cryst., 2006, 33, 1449.

50. J. N. Israelachvili, D. J. Mitchell, and B. W. Ninham, J. Chem. Soc. Faraday Trans. 2, 1976, 72, 1525.

30. G. S. Attard, R. W. Date, C. T. Imrie, et al., Liq. Cryst., 1994, 16, 529.

51. J. N. Israelachvili, S. Marcelja, and R. G. Horn, Q. Rev. Biophys., 1980, 13, 121.

31. C. V. Yelamaggad, S. Anitha Nagamani, U. S. Hiremath, et al., Liq. Cryst., 2001, 28, 1581.

52. J. N. Israelachvili, Intermolecular and Surface Forces, Academic Press, New York, 1985, pp. 229–271.

32. R. Els¨aßer, G. H. Mehl, J. W. Goodby, and M. Veith, Angew. Chem. Int. Ed., 2001, 40, 2688.

53. C. Tanford, in The Hydrophobic Effect: Formation of Micelles and Biological Membranes, John Wiley & Sons, New York, 1980, pp. 43–59.

33. R. Els¨aßer, G. H. Mehl, J. W. Goodby, and M. Veith, Phosphorus Sulfur and Silicon and the Related Elements, 2001, 168, 341–344. 34. R. Els¨aßer, G. H. Mehl, J. W. Goodby, et al., Mol. Cryst. Liq. Cryst., 2003, 402, 237. 35. G. H. Mehl, R. Els¨aßer, J. W. Goodby, and M. Veith, Mol. Cryst. Liq. Cryst., 2001, 364, 219. 36. J. W. Goodby, M. Bates, I. M. Saez, et al., Proc. Mater. Res. Soc., U.S.A., 2009, 1157, 1134–BB09–1134–BB04. 37. S. A. Ponomarenko, E. A. Rebrov, A. Y. Bobrovsky, et al., Liq. Cryst., 1996, 21, 1. 38. K. Lorenz, D. H¨olter, B. St¨uhn, et al., Adv. Mater., 1996, 8, 414. 39. J. Lenoble, S. Campidelli, N. Maringa, et al., J. Am. Chem. Soc., 2007, 129, 9941. 40. (a) G. W. Gray and B. Jones, J. Chem. Soc., 1953, 4179–4180; (b) G. M. Bennett and B. Jones, J. Chem. Soc., 1939, 420–425; (c) G. W. Gray and B. Jones, J. Chem. Soc., 1954, 683. 41. J. D. Bunning, J. W. Goodby, G. W. Gray, and J. E. Lydon, in Liquid Crystals of One- and Two-Dimensional Order, eds. W. Helfrich and G. Heppke, Springer-Verlag, New York, 1980, pp. 397–402. 42. J. E. Bunning, J. E. Lydon, C. Eaborn, et al., J. Chem. Soc. Faraday Trans. 1, 1982, 78, 713. 43. H. Zocher and V. Birnstein, Z. Phys. Chem., 1929, A142, 113. 44. H. Zocher, in Liquid Crystals and Plastic Crystals, eds. G. W. Gray and P. A. Winsor, Ellis Horwood, Chichester, 1974, Vol. 1, pp. 64–66. 45. K. Praefcke and D. Blunk, Liq. Cryst, 1993, 14, 1181. 46. K. Praefcke, P. Marquardt, B. Kohne, et al., Mol. Cryst. Liq. Cryst., 1991, 203, 149. 47. J. W. Goodby, V. G¨ortz, S. J. Cowling, et al., Chem. Soc. Rev., 2007, 36, 1971.

54. S. Carnie, J. N. Israelachvili, and B. A. Pailthorpe, Biochim. Biophys. Acta, Biomembr., 1979, 554, 340. 55. R. Nagarajan, Langmuir, 2002, 18, 31. 56. M. Yoshio, T. Mukai, H. Ohno, and T. Kato, J. Am. Chem. Soc., 2004, 126, 994. 57. N. Bongartz and J. W. Goodby, Chem. Commun., 2010, 46, 6452. 58. P. Davidson, A.-M. Levelut, H. Strzelecki, and V. Gionis, J. Phys. (Paris), 1983, 44, L–823. 59. H. Bengs, M. Ebert, O. Karthhaus, et al., Adv. Mater., 1990, 2, 141. 60. M. Edert, G. Frick, C. Baehr, et al., Liq. Cryst., 1992, 11, 293. 61. J. E. Lydon, Mol. Cryst. Liq. Cryst. Lett., 1980, 64, 19. 62. J. S. G. Cox, G. D. Woodard, J. Pharm. Sci., 1971, 60, 1458.

and

W. C. McCrone,

63. N. H. Hartshorne and G. D. Woodard, Mol. Cryst. Liq. Cryst., 1973, 23, 343. 64. K. J. Toyne, in Thermotropic Liquid Crystals, Critical Reports on Applied Chemistry, ed. G. W. Gray, John Wiley and Sons, Chichester, Vol. 22, pp. 28–63. 65. D. Demus, H. Demus, and H. Zaschke, Fl¨ussige Kristalle in Tabellen, VEB DeutscherVerlag f¨ur Grundstoffindustrie, Leipzig, 1974. 66. D. Demus and H. Zaschke, Fl¨ussige Kristalle in Tabellen, VEB Deutscher Verlag f¨ur Grundstoffindustrie, Leipzig, 1984. Vol. II. 67. I. Sage, in Thermotropic Liquid Crystals, Critical Reports on Applied Chemistry, ed. G. W. Gray, John Wiley and Sons, Chichester, Vol. 22, pp. 64–119. 68. J. W. Goodby, Mol. Cryst. Liq. Cryst., 1981, 75, 179. 69. S. Campidelli, T. Brandm¨uller, A. Hirsch, et al., Chem. Commun., 2006, 4282.

48. M. Su´arez, J.-M. Lehn, S. C. Zimmerman, et al., J. Am. Chem. Soc., 1998, 120, 9526.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc148

Liquid Crystals Formed from Specific Supramolecular Interactions Duncan W. Bruce University of York, York, UK

1 Introduction 2 Liquid Crystal Generalities 3 Liquid Crystals Formed through Noncovalent Interactions 4 Perspective and Conclusions Acknowledgment Notes References

1

1 2 4 19 20 20 21

INTRODUCTION

The purpose of this chapter is to address the influence of specific, supramolecular interactions between different species on the formation of liquid crystal phases. More precisely, cases will be discussed where the presence of these interactions leads either to the induction of liquid crystallinity where this was not observed in the components separately, or to the modification of liquid crystal properties beyond that normally found in mixture systems. These terms of reference are stated carefully, for there is a sense in which the term supramolecular liquid crystal could otherwise be very misleading for it could imply that in some way that there are liquid crystals that are not supramolecular. Thus, the liquid crystal state exists between the solid state, in which molecules are ordered both positionally and orientationally in three dimensions, and the liquid state in

which they are free to move and to rotate in all directions. Liquid crystallinity is found in molecules that have an appreciable structural anisotropy, and this anisotropy results in additional, anisotropic dispersion forces between molecules that are able to stabilize the liquid crystal phases. Thus, the simple existence of liquid crystal phases is a supramolecular phenomenon, for the ordering of the molecules is driven by noncovalent interactions. Liquid crystals may be classified into two broad groups depending on the way in which the liquid crystal state is achieved. The materials with which most people would identify as liquid crystals by virtue of their use in the ubiquitous liquid crystal display, are known as thermotropic liquid crystals because transitions between the solid, liquid, and liquid crystal states are driven primarily using temperature as the variable. It is these materials that will be the focus of this discussion. In addition, there exist families of materials in which the transitions in and out of the liquid crystal state are driven by a solvent. These are termed lyotropic liquid crystals. In fact, lyotropic liquid crystals are much more common than their thermotropic equivalents and are manufactured on a much bigger scale, examples being soaps, detergents, and emulsifiers. The major class is that of surfactants, whose liquid crystal phases are organized arrays of micelles found typically at concentrations of surfactant in water of >20 wt%. Other classes of lyotropic liquid crystal exist1, 2 and it is instructive to note that DNA forms liquid crystal phases,3 as do many of the lipid molecules that form cell membranes.4 Further, the very high tensile strength of aramid fibers such as Kevlar or even spider silk arises from their processing from the lyotropic liquid crystal state. Although also supramolecular in nature, lyotropic materials are not discussed here but aspects of their behavior

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

2

Soft matter

Unique structural axis

n

Figure 1

(a)

Disc-like molecule

Rod-like molecule

Nematic phase

(c) Colh ( p 6mm)

The two principal shape motifs in liquid crystals.

n

n

n

(b) Stacking of molecules into columns.

(d) Colr (c 2mm)

Figure 3 Schematic showing (a) nematic phase of disklike mesogens; (b) stacking of disklike molecules into columns; (c) top view of the columnar hexagonal (Colh ) phase; (d) top view of one type of columnar rectangular (Colr ) phase where the ellipses represent disks that are tilted within the column. N

SmA

SmC

Figure 2 Representative structure of the arrangement of calamitic mesogens in the nematic (N), smectic A (SmA), and smectic C (SmC) mesophases, with the director (n) shown in each case.

are covered in Chapter Introduction to Surfactant SelfAssembly, Concepts, Soft Matter Science—a Historical Overview with a Supramolecular Perspective and SelfAssembly of Facial Amphiphiles in Water, Soft Matter.

2

LIQUID CRYSTAL GENERALITIES

The two most common molecular motifs that lead to liquid crystal phase behavior are the rod and the disk. Clearly rodlike molecules have one unique axis that is longer than the other two, while disklike molecules have one unique short axis and two longer axes (Figure 1). Rodlike molecules organize into nematic or smectic phases, while disklike systems form nematic or columnar phases. Figure 2 shows schematic diagrams of molecules arranged in a nematic, smectic A (SmA), and smectic C (SmC) liquid crystal phase (= mesophase). There are very many smectic phases of which SmA and SmC are only two.5 The supramolecular order of the nematic phase is purely of an orientational nature with the long molecular axes arranged about a common axis, the director, n, with the order of the phase being described by the function S=

1
3 cos2 θ − 1 2

where θ is the angle between the long axis of individual molecules and the director, summed over the whole sample. For perfect alignment, S = 1, whereas in the normal (isotropic—Iso) liquid state, S = 0. For a nematic liquid crystal phase, typical values for S are 0.3 < S < 0.7. This formula may also be used with smectic liquid crystal phases although in lower-symmetry situations (e.g., SmC), more than one order parameter may be needed to describe the order fully. Disklike liquid crystals may also form a nematic phase (Figure 3), but the most common behavior is to form columns which then organize themselves into some symmetric arrangement, two examples of which are shown in Figure 3.

2.1

Some specific and more subtle aspects of liquid crystal organization dependent on polarity

At the level of detail indicated, all of the behavior described above can be accounted for by anisotropic dispersion forces between molecules. However, polarity also plays a role and some aspects of this are now described. The nematic and smectic mesophases are normally dipolar inasmuch as molecules with terminal dipole moments will organize with the dipoles antiparallel. In some liquid crystals, of which the cyanobiphenyls (Figure 4) are the archetypal example [1], such antiparallel interactions are crucial, because in these materials, it is the antiparallel

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions F

C N

3

F

N C

Figure 4 Cyanobiphenyl liquid crystals showing the characteristic antiparallel correlations present in the mesophase. Disruption of these interactions suppresses the liquid crystal phase.

dimers that are, in effect, the mesogen species. Therefore, when their formation is suppressed (as can be the case on addition of certain other materials, including other liquid crystals), their mesomorphism is also suppressed. This is because the cyanobiphenyls themselves are insufficiently anisotropic to form liquid-crystalline mesophases, whereas the antiparallel dimers are longer and so the anisotropy is increased. This phenomenon can also lead to an effect known as re-entrant behavior.6 Thus, there is an accepted thermodynamic ordering of liquid crystal phases, which is shown in 1 for the phases considered so far: Iso − N − SmA − SmC

(1)

Such knowledge assists in identifying liquid crystal phases, for to find, for example, an SmA phase below an SmC phase in temperature would at best be unexpected. However, in some cyanobiphenyl liquid crystals or their mixtures, the phase sequence shown in 2 is observed: Iso − N − SmA − Nre

(2)

where Iso is the normal (isotropic) liquid phase and Nre is a re-entrant nematic phase—that is, one out of thermodynamic sequence. In these materials, the phase sequence is accounted for by a change in the nature and/or degree of antiparallel correlations. This leads, in effect, to the mesogenic unit changing to a different supramolecular species with a different anisotropy and showing a different mesophase, the effect being driven by temperature. Thus, the supramolecular nature of some systems is dynamic. This dynamic nature is further evidenced by the phenomenon of cybotacticity, in which localized clusters that may be, for example, SmA or SmC in nature, are found in nematic phases. Such behavior is now the subject of intense speculation as proposals

Figure 5 moment.

to explain the nature and behavior of the biaxial nematic phase develop (vide infra). There are, of course, also many examples of materials in which the major dipole is perpendicular to the long axis of the molecule (Figure 5) and there is evidence in these materials for ferroelectric correlations over short length scales.7, 8 Finally here, it is important to note that the supramolecular nature of liquid crystal mesophases, in conjunction with polarity, can also lead to the induction of chirality in nonchiral materials. Current interest dates back to 2006 when Niori et al.10 reported the observation of ferroelectric switching in some achiral, bent-core liquid crystals. The molecules are shown in Figure 6 and are unusual inasmuch as convention suggests that liquid-crystalline molecules should be highly anisotropic. Matsunaga et al.11 had prepared these materials in the 1990s and had noted that they did indeed form a liquid crystal phase. However, what Niori et al. showed was that the symmetry of the liquid crystal phases must be broken in order to observe a ferroelectric response and further that chiral domains could be observed.10 The origin of the ferroelectricity and chirality was subsequently explained by Link et al.12 and has its origins in the fact the average molecular symmetry is C2v and these C2v objects arrange in layers in which they are tilted to the side. As shown in Figure 7, this leads to four possible arrangements. Thus, alternate layers can show the same tilt (synclinic) or an alternating tilt (anticlinic). Further, the apex of the bent molecules can all be arranged in the same direction perpendicular to the layers (ferroelectric) or in alternating manner (antiferroelectric). Thus, the anticlinic, antiferroelectric and the synclinic, ferroelectric organizations are both chiral, for in these cases, the two possible tilt directions lead to enantiomeric arrangements. In this case, chirality can be understood in terms of tilted arrangements of low-symmetry objects within layers. However, there are a limited number of examples where

O

O O

N CnH2n+1(O)

Figure 6

Difluoroterphenyls9 —mesogens with a lateral dipole

O N (O)CnH2n+1

Bent-core liquid crystals.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

4

Soft matter

SmCsPA

achiral molecules. The first such observation was reported by Weissflog14 in 2002 in compound 1, and subsequent examples were from Yokoyama et al. (2),15 G¨ortz and Goodby (3),16 and Pr¨asang et al. (4)17 (Figure 8). Theories based on clustering (related to cybotacticity)18, 19 as well as supramolecular, helical ordering deriving from particular combinations of elastic constants have been advanced,20 but with relatively few molecules showing such behavior, a definitive explanation has not yet been expounded. The above examples provide a brief overview of some aspects of the inherently supramolecular nature of all liquid crystals and so what follows should be seen against that background.

SmCaPF

SmCaPA

SmCsPF

Figure 7 The four possible arrangements of bent-core mesogens in a tilted, lamellar phase after Reddy and Tschierscke.13 The nomenclature [2] is: subscripted “s” is synclinic, while that subscripted “a” is anticlinic—referring to the relative tilt from layer to layer. P implies that the phase is polar and subscript “A” is antiferroelectric and “F” is ferroelectric. As indicated, SmCa PA and SmCs PF are chiral, while clearly SmCs PF and SmCa PF are ferroelectric.

3

Against the background above, it is appropriate to reemphasize the focus of this chapter as suggested in the heading, that is, liquid crystals formed through noncovalent interactions, and these are divided into four categories:

chirality is shown in the absence of layers, in which case there is no meaningful notion of tilt. This refers to the observation of chiral behavior in the nematic phase of CI

O

1

O

O

LIQUID CRYSTALS FORMED THROUGH NONCOVALENT INTERACTIONS

O

O

O

O

O

C12H25O

OC12H25

O N

O

2 O

CI

CI

N

O

O CI

O

OC8H17

CI

C8H17O

3

N N

O

O

O R

O

O

R′

R = R′ or R ≠ R′ R, R′ = CnH2n +1 or CnH2n+1O F F

4 N

F I

I F

CnH2n+1O

Figure 8

N

OCnH2n+1

Bent-core liquid crystals forming chiral nematic phases.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions

Figure 9

• • • •

Arrangement of C6 H6 and C6 F6 in co-crystals.

liquid crystals formed through quadrupolar interactions; liquid crystals formed through charge-transfer interactions; hydrogen-bonded liquid crystals; halogen-bonded liquid crystals.

In this chapter, it is not intended to provide a comprehensive treatment of the liquid crystals associated with each of the above classes, rather the aim is to choose examples that are felt to illustrate the area well and to give the reader a sense of the range of possibilities.

3.1

Quadrupolar interactions

The idea of stabilization of intermolecular interactions using quadrupolar effects is illustrated nicely by the observation in 1960 by Patrick and Prosser21 that benzene and hexafluorobenzene, both of which melt very close to 5 ◦ C, form a co-crystal that melts at 23.7 ◦ C. The crystal structure (here quoted for C6 D6 ·C6 F6 ) shows how the two molecules stack (Figure 9). That this might be of use in liquid crystals was demonstrated many years later so that, for example, Marder, Viney, and coworkers22, 23 prepared a number of rigidrod materials with arene rings linked by acetylene spacers. When brought together in two-component systems where

Figure 10 mesogens.

5

7

6

F

F

F

F

F

F

F F

F

F

F

F

F

interactions

F

F

F

F

F F

F

Arene–perfluoroarene

in

rigid-rod

perfluorinated and hydrogenous rings could be arranged in a complementary fashion, X-ray diffraction showed the preferred interaction between arene and perfluoroarene (Figure 10). Furthermore, it was found to be the case that liquid crystal properties were stabilized. For example (Figure 11), compound 5 melts at 181.7 ◦ C and shows a monotropic nematic phase at 164.3 ◦ C [3] while compound 6 does not have a liquid crystal phase, simply melting at 226.2 ◦ C. However, a co-crystal of the two melts at 178 ◦ C to form a smectic B phase, giving way to a biphasic nematic/isotropic phase at 196.2 ◦ C with nematic regions persisting until 231.5 ◦ C. Similarly, compound 7 simply melts at 182.5 ◦ C, while 8 melts at 196.5 to give an extremely shortrange nematic phase, which immediately gives way to the isotropic state. However, the co-crystal melts at 189.3 ◦ C to a nematic phase, which persists to 199 ◦ C where it forms the isotropic phase cleanly. The fact that this last transition is clean (i.e., there is no evidence of biphasic behavior) shows that it is the co-crystal that is melting. The other classic example of the use of quadrupolar interactions was from the Leeds group,24–28 who coined the term complementary polytopic interaction to describe the phenomenon and to distinguish it subtly from other examples

F

Figure 11

5

F

F F

F

F

F

F

8

Arene–perfluoroarene pairs.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

6

Soft matter C9H19

C9H19 C9H19 C9H19

OC6H13 OC6H13

N N N

N

C6H13O

N

C6H13O

N

C9H19

OC6H13 OC6H13

C9H19

C9H19

9

10

C9H19 C9H19

C9H19 C9H19

11

C9H19

Figure 12 Discotic mesogens that, in certain mixture combinations, showed modified mesomorphism on account of intermolecular quadrupolar interactions.

3.2

CnH2n +1 O

12

O

N N

O

O O CnH2n +1

13

N N N O

Figure 13 actions.

N

Mesogens used in the study of charge-transfer inter-

Charge-transfer interactions

A different mechanism is that of charge transfer, the origin of which in liquid crystals can be traced back to the work of the Halle group in the late 1960s, who recognized that mixtures of calamitic liquid crystals containing both donor and acceptor functions (e.g., 12 and 13, Figure 13) behaved as though a single component.29 Thus, for example, compound 12 (n = 3) showed the following phase behavior: Cr • 99.4 • SmC • 142.9 • SmA • 203.0 • Iso whereas compound 13 behaved as follows:

of quadrupolar effects. For example, consider the discotic mesogens shown in Figure 12. The hexaazatriphenylene, 9, shows two crystalline modifications and melts directly to the isotropic phase with no sign of liquid crystallinity, while 10 is a well-known discogen that melts to a columnar hexagonal (Colh ) mesophase at 70 ◦ C and clears at 100 ◦ C. However, if the two are mixed in a 1 : 1 ratio, then the mixture melts at circa 130 ◦ C and shows a Colh phase that persists to 240 ◦ C—an enormous stabilization. Similarly, when 10 is mixed with 11 (which simply melts to the isotropic liquid at 59 ◦ C, the 1 : 1 mixture now melts at 66 ◦ C into a Colh phase that is stable up to 155 ◦ C when the isotropic phase forms. These are remarkable stabilizations and the absence of any signatures of charge transfer suggests that the electronic states of the components are only disturbed minimally. Clearly then, this is a very powerful interaction and one that has led to remarkable phase stabilization. However, while the above materials and others related to them have been studied computationally in order to understand the effect, it is not apparent to what extent the structural and electronic features that are necessary can be predicted a priori to lead mesogen design.

Cr • 143.0 • SmA • 167.2 • N • 169.8 • Iso However, the binary phase diagram (Figure 14) shows that while there is biphasic behavior at the clearing point over almost all compositions, when the ratio is precisely 1 : 1 then the mixture behaves in a monophasic manner, providing direct evidence of an associated state. More than that, the clearing point of the 1 : 1 mixture is some 5 ◦ C higher than the clearing point of the higher-clearing component, 12. This kind of evidence and behavior will be discussed in greater detail in Section 3.3. However, the majority of the work in this area has been carried out with discotic systems and it is this area that will be discussed here as indicative of what is possible. Important considerations here relate first to mesophase structure, for if charge-transfer organic metals such as TTF–TCNQ (tetrathiafulvalene–tetracyanoquinodimethane, respectively) are considered, perhaps surprisingly at first sight, in the solid phase, these materials arrange themselves in stacks (Figure 15) of partially oxidized TTF and separate stacks of partially reduced TCNQ (there is partial charge transfer

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions

7

C7H15O

Iso

OC7H15 C7H15O

14 C7H15O

200

OC7H15 C7H15O

N

T / °C

SmA 15

150

OC7H15

C7H15O OC7H15

C7H15O

C7H15O

OC7H15

C7H15O

100

O O

SmC

OC7H15

O O

(CH2)14

O

O

C7H15O

Figure 16

0

50 mol% 12

100

Figure 14 Binary phase diagram for compositions of compounds 12 and 13, showing clear evidence of a 1 : 1 association. (Redrawn from Ref. 21.  Nature Publishing Group, 1960.)

OC7H15

Monomeric and dimeric triphenylene derivatives.

O2N

O

NO2 NO2

NO2 O2N

O2N NO2 TNF

TAPA

N O * O O H

Figure 17

Electron-accepting materials—TNF and (–)-TAPA.

with 2,4,7-trinitro-9-fluorenone, TNF (Figure 17), at a loading of 2 : 1 disk : TNF. In the case of the monomeric triphenylene, the clearing point increased 125 K from 89 to 214 ◦ C, while for the dimer, it increased by a more modest 53 K to 200 ◦ C. 2 H NMR spectra were recorded in the mesophase of charge-transfer adducts prepared using perdeuterio-TNF and showed conclusively that the TNF molecules existed within the stacks and that they stayed there to the highest reaches of the mesophase.32 Another important point is that, despite initial calculations33 that show the importance of charge transfer in liquid Figure 15 Crystal structure31 of perdeuterio TTF–TCNQ showing the individual stacks of the two components.

between these components affecting circa 59% of the components).30 However, all the evidence is that in the liquid-crystalline state of mixtures of 12 and 13, the components are mixed intimately as no significant effect is observed on mesophase structure. This was studied in some detail using 2 H NMR spectroscopy. The monomeric (14) and dimeric (15) triphenylenes (Figure 16) were studied in isolation and in combination

OCnH2n+1

16

Figure 18

Structure of Praefcke’s radial pentaynes.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

8

Soft matter

Table 1 Modification of liquid crystal properties of palladium complexes with TNF. 18 (X=Cl)

19

(ND )a Colh

Colo Colh

— Colb

200

a

The parentheses indicate that the phase is monotropic (i.e., thermodynamically metastable). b Symmetry of the columnar phase not identified.

Colh

150

100

Cr

50 5

6

M

N

X

10

11

12

13

OCnH2n +1 OCnH2n +1 X

OCnH2n +1

M

M

N

X

N OCnH2n +1

CmH2m +1(O)

17 M = Pd, Pt X = Cl, Br, I, SCN

14

15

16

(−)-TAPA (Figure 17) led to the induction of the chiral N∗ col phase. Praefcke and coworkers also investigated the behavior of various di- and tetra-nuclear group 10 metal mesogens with TNF and TAPA, the main examples of which are shown in Figure 20. While these complexes were generally mesomorphic in nature, addition of TNF or TAPA did lead to a modification in behavior, examples of which are shown in Table 1.38–40

OCnH2n +1 OCnH2n +1

M

M

CnH2n +1O

N

X

N

OCnH2n +1

X

OCnH2n +1

CnH2n +1O

N

N

O

O

CnH2n +1O M O

18

OCnH2n +1 OCnH2n +1

CnH2n +1O CnH2n +1O

CnH2n +1O

Figure 20

9

Figure 19 Phase diagram for the radial pentaynes (blue curve) and the 1 : 1 radial pentayne : TNF mixtures (black curves).

N

X

8

n

CnH2n +1O M

7

CnH2n +1O CnH2n +1O

CnH2n +1O CnH2n +1O CnH2n +1O

Ncol

Cr

crystal systems, there are aspects of the literature that have suggested that any charge transfer is not the driving force in such materials and that all can be explained by quadrupolar interactions.34 However, the method used in this work had limitations and subsequently was shown to present an incomplete picture.35 Thus, while quadrupolar interactions are likely to make a contribution, it is extremely unlikely that charge-transfer effects can be ruled out as the significant contributor to mesophase induction/modification. The majority of the work has been carried out with disklike systems and their electron-rich nature lends itself to transfer of charge.36 For example, Praefcke et al. studied 1 : 1 adducts of pentaynes (16, Figure 18) with TNF as a function of the alkoxy chain length.37 While the pentaynes themselves were not mesomorphic, addition of TNF led to the induction of a Colh phase for 5 ≤ n ≤ 10 and an Ncol phase[4] for 8 ≤ n ≤ 15 (Figure 19). Use of (O)CmH2m+1

Iso

Iso

T/°C

Pure mesogen Mesogen + TNF

17 (X=Cl)

250

OCnH2n +1 OCnH2n +1

M O

19

Di- and tetra-nuclear mesogens of palladium(II) and platinum(II).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions OC10H21 C7H15 N

O N H

OC10H21

N

20

O

N

CnH2n +1 O

O

H N

N H

Cm H2m +1

O

O

O O

O Pd

9

O

N N O CpH2p +1

Cq H2q +1

23

O

OC10H21 C10H21O

Figure 23 Liquid crystals formed from complementary hydrogen bonding.

OC10H21

Figure 21 Palladium mesogen in which the biaxial SmA phase was induced by addition of TNF.

O H O O H O

OCnH2n +1 OCnH2n +1

C6H13

H O

21

CnH2n +1O CnH2n +1O

O H O

Figure 22 crystals.

3.3

O

By far the most common noncovalent interaction responsible for generating new liquid-crystalline species is the hydrogen bond, and this area has been well reviewed.42, 43 In fact, hydrogen bonding in liquid crystals is a very old concept and it has been known for some time that, for example (Figure 22), the mesogenic behavior of 4alkoxybenzoic acids (21) depends totally on the fact that they exist as hydrogen-bonded dimers. Other simple, elegant examples are the undeca-2,4-dienoic acids (22).44 More exotic systems have also been described, for example, Lehn and coworkers described the complementary system, 23, which was found to exhibit a columnar mesophase45 while neither of the individual components was mesomorphic (Figure 23). Another example of multiple hydrogen bonding by the same group is 6,7-bis(alkyloxy)-2,3-dihydrophthalazine1,4-dione (24).46 The molecule self-assembled into a trimeric, disklike structure via complementary hydrogen

N

N

O H O

24 OCnH2n +1 OCnH2n +1

Simple, early example of hydrogen-bonded liquid

Hydrogen Bonding

H

N H

O

22

Finally here, it is noted that addition of TNF to the palladium mesogen (20) in Figure 21 led to the formation of the biaxial SmA phase (SmAb )—a discovery of note as it was the first demonstration of this phase which had, until then, been no more than a prediction.41

H H

OCn H2n +1 O H O

N N N

C6H13 Cn H2n +1O

O

Figure 24 diones.

Disklike, trimeric assembly of dihydrophthalazine-

bonding (Figure 24) and the disks so formed stacked to give Colh and columnar rectangular (Colr ) mesophases. Conceptually similar is work from Janietz and coworkers47 who prepared a 2,4,6-triarylamino-1,3,5-triazine core (25) which was itself mesomorphic showing Colh phases. When hydrogen bonded to alkoxybenzoic acids, these triazines gave rise to complexes (26) showing Colh mesophases between about 35 and 80 ◦ C for 4-alkoxybenzoic acids (Figure 25) while Colr phases (noncentered, P 2m) were seen between 60 and 98 ◦ C for complexes with 3,4-dialkoxybenzoic acids (not shown). A very nice example comes from the work of Beginn and Lattermann48 who reported the observation of columnar phases in amide 27 (Figure 26). If the hydrogenbonded core is regarded as the structural equivalent of another benzene ring, then it is found that three-ring mesogens bearing six peripheral chains do not form mesophases. However, it is clearly the case that the amide groups allow for intermolecular hydrogen bonding to form stacks of molecules that give rise to the columnar phase. Diamides such as 28 were described originally by Matsunaga and Terada49 and then later reinvestigated by Malthˆete and coworkers.50–52 In many ways, at first sight, it is surprising that these systems are liquid crystalline at all, but the “secret” is in intermolecular hydrogen bonding,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

10

Soft matter OCnH2n+1

CnH2n+1O CmH2m +1O

OCnH2n +1 OCnH2n +1 H CnH2n +1O N H

O

N

N

CnH2n +1O

O

N N

N

CnH2n +1O OCnH2n +1 OCnH2n +1

25

N H

CnH2n +1O

H

H

H

OCmH2m

+1

N H

N

O

N N

O N

H

H O

O

OCnH2n +1 OCnH2n +1

26 OCmH2m +1

Figure 25

Triarylaminotriazines (25) and their hydrogen-bonded complexes (26) with alkoxybenzoic acids.

H N H O

O

O

O

O O H N H

O

O 27

Figure 26

Mesomorphic trialkoxybenzamides.

O CH3

R H

N

CH3

N 28

(a)

CH3 O

N H

O CH3 H

CH3 HN

R

O CH3

N H

O CH3

N H

CH3 O

CH3 H N

O CH3 N

H

CH3 O

CH3 H N

CH3 O

CH3 H N

CH3 O

(b)

Figure 27 The molecules involved in the formation of hydrogen-bonded nematic and columnar phases (a) and a representation of the arrangement of the molecules in the mesophases (b).

which causes the molecules to form columns (Figure 27). These columns can than organize to form nematic and columnar phases, dubbed “spaghetti nematic phases.” There is a degree of conceptual similarity between these materials and those of Beginn and Lattermann above. However, the hydrogen-bonded mesogens that are of most interest in the context of this chapter are those elaborated initially by Kato and Fr´echet in the late 1980s.53–62 In this approach, a pyridine, which may or may not have

liquid crystal properties, was hydrogen bonded with a 4-substituted benzoic acid to form a new species with its own, distinct mesomorphism. For example, Figure 28 shows a complex between a stilbazole ester (29) with a nematic phase and an alkoxybenzoic acid; as the notional monomer, the acid would not be liquid crystalline. The 1 : 1 complex (30) then shows both a smectic phase (seen in neither component) and a nematic phase, the latter clearing at a temperature higher than either

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions

11

O H O

O

O

O O

N

29

Cr 147 N 160 Iso (as the dimer)

Cr 165 N 213 Iso

O O O N

H O O

30 O Cr 136 S 160 N 238 Iso

Figure 28

Hydrogen-bonded complex (30) formation between a mesomorphic stilbazole (29) and an alkoxybenzoic acid.

C8H17

O N

OC8H17

n = 6: Cr 38 N 48 Iso n = 12: Cr 35 N 54 Iso

Cr 92 SmC* 101 SmA* 132 N* 135 Iso

Figure 29 Chiral hydrogen-bonded mesogen, which shows ferroelectric switching in its SmC∗ phase. [5].

component, convincingly demonstrating the existence of the complex.53 To all intents and purposes, these hydrogen-bonded mesogens behave as single-component materials, and this is perhaps well demonstrated by the observation that the use of chiral components can generate SmC∗ phases which are able to be switched ferroelectrically. For example, complex 31 (Figure 29) has a spontaneous polarization of 16 nC cm−2 , which is greater than most SmC∗ materials containing the 2-methylbutyl function.63 A particularly elegant example of the pyridine/acid approach is shown in Figure 30, which represents a very simple situation, namely, that of a complex (32) between an alkoxybenzoic acid and an alkyl pyridine and which shows a nematic phase close to room temperature. These materials have added significance and are discussed again later. In addition to benzoic acids, phenols are also able to hydrogen bond to stilbazoles to give mesomorphic materials

OCnH2n +1

O

O

Figure 32

H O

32

H O

31

N

Figure 30

Simple, low-melting hydrogen-bonded complexes.

CnH2n +1O

N H O

33

CnH2n +1O

C N N H O

34

Figure 31 Hydrogen-bonded cyanophenol.

C N

complexes

of

4-

and

3-

from nonmesomorphic components. For example, cyanoand nitro-phenols were shown to bind effectively and the way in which the mesomorphism depended on the substituent and its position were elucidated. Complexes of 4-cyanophenol (33, Figure 31) showed nematic phases,64 but as the single-crystal X-ray structure shows (Figure 32), the nature of the hydrogen bond with the phenol leads to a bent complex and so, despite the

Molecular structure of complex 33 (n = 8).

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

12

Soft matter

CnH2n+1O N HO N C

C N O H N OCnH2n +1

Figure 33

Postulated antiparallel correlations in hydrogen-bonded complexes of 4-cyanophenol.

CnH2n +1O N H O CN or NO2 100.0 3-NO2 90.0 4-NO2

T /°C

80.0 3-CN

4-CN

70.0

60.0

50.0

0

2

6

4

8

10

n

Figure 34 Plot of clearing point vs chain length for stilbazole complexes of 3- and 4-isomers of cyano- and nitro-phenol. Replotted from data in Ref. 66.

reasonably extended nature of the complex, the clearing point where the stilbazole chain length is C6 H13 is only 64 ◦ C. That such small, nonlinear mesogens were mesomorphic at all was attributed to the presence of antiparallel correlation characteristic of cyano-terminated liquid crystals (Figure 33). Some anisotropy can be recovered simply

by using 3-cyanophenol (34) in preference and here the clearing point recovered to 75 ◦ C.65 If nitrophenols are used instead, even more stable mesophases are formed showing that in these systems, nitro is more effective as a terminal group in promoting mesomorphism.66 Once more, 3-substitution is more effective than 4-substitution. This is shown graphically in Figure 34. A rather nice example of the use of hydrogen bonding in liquid crystals comes from the work of Coco and Espinet.67 4-Isocyanobenzoic acid was prepared and complexed either to gold(I) in a 1 : 1 complex (35) or to transdiiodo-palladium(II) or -platinum(II) (36) to give a 2 : 1 complex (Figure 35). The free, terminal acid function was then allowed to hydrogen bond to decyloxystilbazole resulting in liquid-crystalline materials. Thus the gold complex showed the formation of an SmA phase, while the palladium complex showed a nematic phase. Slightly unexpectedly, the mesomorphism of the platinum complex differed showing an unidentified smectic phase below a nematic one. Then, using a triazine similar to that of Janietz (Figure 25), the group reported homotrinuclear complexes of some metal carbonyls using the isocyanobenzoic acid as the linker (Figure 36).68 All of the complexes (37) showed Colh phases up to around 80 ◦ C. Finally in this overview of hydrogen-bonded liquid crystals, it is noted that there is a large family of carbohydrate mesogens and that where these contain free hydroxyl functions, hydrogen bonding will be a determining factor in their

O Cl Au C N O H

N

35 C10H21O

N

OC10H21

O

H O N C M C N

O H

O 36

Figure 35

N

M = Pd, Pt

OC10H21

Hydrogen-bonded mesogens formed from isonitrile complexes of noble metals.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions CnH2n +1O

OCnH2n +1

Lp M O O CnH2n +1O CnH2n +1O CnH2n +1O

H N H O

OCnH2n +1 H

N H

N

MLp

O

N N

13

MLp = Fe(CO)4, Cr(CO)5,

O N

Mo(CO)5, W(CO)5

H OCnH2n +1

H O

OCnH2n +1 OCnH2n +1 37 MLp

Figure 36

Metal carbonyl mesogens formed from a disklike, hydrogen-bonded scaffold.

behavior. These materials may be studied further through the review literature.69, 70

C8H17O N H O NO2

3.3.1 Stoichiometry One of the attractions of hydrogen-bonded liquid crystals is the ease with which they can normally be prepared. Thus, for many systems it is enough to stir the two components in a suitable cosolvent for a short while (maybe hours) and then remove the solvent. Assuming that the components were pure and weighed out accurately, the target material is then obtained in quantitative yield and purity! Occasionally, such a treatment is insufficient and so it may be necessary to heat the mixture of components into the melt to ensure complex formation. It is then easy to demonstrate that the complex is pure, because according to the condensed phase rule, a single component has a single degree of freedom and so a precise 1 : 1 mixture will show a sharp melting point indicative of a single component. This is demonstrated rather elegantly in the full binary phase diagram constructed for compositions of 4-octyloxystilbazole and 3-nitrophenol (Figure 37).66 Thus, at all compositions other than 1 : 1, biphasic behavior is seen. This is understood as follows, for at phenol : stilbazole ratios 7 kJ mol−1 ). Chirality in bent-core liquid crystals is well known (vide supra), but in bent-core materials with nematic phases it is extremely rare and is still without adequate explanation. Other trimeric complexes were formed between stilbazoles and flexibly linked diiodotetrafluorobenzenes (Figure 52).98 For n = 6, 8, 10, and 12 and p = 2, 4, 6, and 8, examination of the thermal behavior showed that most showed a monotropic SmA phase with melting normally being observed between 88 and 99 ◦ C and clearing between 82 and 89 ◦ C. A curious feature of these complexes is the apparent insensitivity of the melting and clearing points to both n and m. The halogen bonds in these complexes were investigated by XPS, where data showed that the binding energy of the N 1s level increased by around 0.9–1.1 eV on complexation. When given the possibility of hydrogen and/or halogen bonding, the preference shown may have been readily predicted. Thus, in combination with 4-iodotetrafluorophenol, a 2 : 1 complex formed in which a stilbazole was hydrogen bonded to the phenol and halogen bonded to the iodine (Figure 53). However, when crystallized in a 1 : 1 ratio, the stilbazole complexed preferentially to the phenol with the

electron-poor iodine halogen bonding to the oxygen of the alkoxy chain (Figure 54).99 Finally, polymeric examples have also been reported,100 whose components are shown in Figure 55. Polymers were thus prepared between the flexible dimer of iodotetrafluorobenzene (49) and the different bifunctional pyridines (50). The degree of polymerization of these polymers was unreported and, given that they were (nor unsurprisingly) not crystallized, it is likely that in addition to being polydisperse, they also contained appreciable quantities of monomer components. For polymers formed between the diiodo compound and the three simple difunctional pyridines, liquid-crystalline behavior was not observed, but the polymers formed using the dimeric stilbazoles did show nematic phases.

4

PERSPECTIVE AND CONCLUSIONS

As pointed out at the beginning of this chapter, the whole phenomenon of liquid crystallinity is a supramolecular effect and one that is driven by the gentlest of all forces, dispersion forces, leading Collings to term the liquid crystal state nature’s delicate state of matter.101 Very deliberately then, the introduction points out some of

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

20

Soft matter F

F O

I

F

F

F

F

p O

49

I F

F N

N

O

N

N

m

50a

O

N N

50b

Figure 55

O

N N

50c

50d

Components of halogen-bonded polymers.

this general aspect of liquid crystals as clearly it underpins any broader discussion. The remainder then considers how specific intermolecular interactions can cause either the formation of new species (e.g., hydrogen or halogen bonding) or can promote specific associations (e.g., quadrupolar systems or lyotropics) that lead to liquid crystallinity. Some of the structures included here truly celebrate the diverse imagination of those who have worked in the field and it is certainly true that many people have arrived at liquid crystals from very different backgrounds. It is also clear from reading the literature that there are others who appear reluctant to use the term liquid crystal or do not realize that this is sometimes what they have. This is not meant to be a clever statement from someone who works with liquid crystals, rather it is an encouragement to think about the possibilities that liquid crystals can offer, for the long-range order afforded by mesophases has a great deal to commend it when looking for bulk properties. Liquid crystals are famous (and hugely profitable) because of flatpanel displays, but they are so much more than that and it is hoped that in reading this chapter and that of Goodby, Saez, and Cowling (see Self-Organization and Self-Assembly in Liquid-Crystalline Materials, Soft Matter), even more people will be inspired to contribute to this fascinating field.

ACKNOWLEDGMENT I am very grateful to Linda McAllister for her careful and critical reading of earlier drafts of this manuscript.

NOTES [1] Cyanobiphenyl liquid crystals were reported in the early 1970s by Gray and coworkers and were responsible for the commercialization of liquid crystal displays. [2] Using the term ‘SmC’ here is somewhat controversial. It is used as the SmC phase is a tilted, lamellar phase and in one sense a similar arrangement applies here. However, it is argued that the naming of the smectic phases, which was largely accomplished during the 1970s, was achieved on the basis of miscibility and that this is not a test that has been applied to materials showing these phases. Collectively, these four variants are also known as the B2 phase (B ⇒ bent). Ref. 13 provides an excellent overview of bent-core liquid crystals. [3] Monotropic means that the phase is seen only on cooling and, as such, is thermodynamically metastable. [4] The Ncol phase is a nematic phase were the structural units are columns of mesogens—in this case columns containing both the multiynes and TNF. See Refs 38– 40 for more details. [5] Note that while there is no formal chiral variant of the SmA phase, the nomenclature SmA∗ is used to recognize that the symmetry of the phase must be lower as it is composed of chiral units. [6] The view taken in this work is that the angle measured at the pyridine nitrogen using both iodine and the ipsocarbon of the pyridine ring gives a better expression of the halogen bond angle compared to the angle measured at the iodine and involving the pyridine nitrogen and the carbon bound to the iodine.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Liquid crystals from specific supramolecular interactions

REFERENCES 1. J. E. Lydon, J. Mater. Chem., 2010, 20, 10071–10099. 2. T. K. Attwood, J. E. Lydon, C. Hall, and G. J. T. Tiddy, Liq. Cryst., 1990, 7, 657. 3. F. Livolant, S. Mangenot, A. Leforestier, et al., Phil. Trans. R. Soc. A, 2006, 364, 2615–2633. 4. I. W. Hamley, Soft Matter, 2010, 6, 1863–1871. 5. Handbook of Liquid Crystals, eds. D. Demus, J. Goodby, G. W. Gray, et al., Wiley-VCH, Weinheim, 1998. 6. P. E. Cladis, Re-entrant phase transitions in liquid crystals in Handbook of Liquid Crystals, eds. D. Demus, J. Goodby, G. W. Gray, et al., Wiley-VCH, Weinheim, vol. 1, Chapter VII, 6.4, 1998, pp. 379–390. 7. (a) D. A. Dunmur and K. Toriyama, Mol. Cryst., Liq. Cryst., 1991, 198, 201–213; (b) K. Toriyama, D. A. Dunmur, and S. E. Hunt. Liq. Cryst., 1989, 5, 1001. 8. D. A. Dunmur and P. Palffy-Muhoray, Mol. Phys., 1992, 76, 1015–1023.

21

27. N. Boden, R. J. Bushby, Z. Lu, and O. R. Lozman, J. Mater. Chem., 2001, 11, 1612–1617. 28. N. Boden, R. J. Bushby, Z. Lu, and O. R. Lozman, Liq. Cryst., 2001, 5, 657–661. 29. G. Pelzl, D. Demus, and H. Sackmann, Z. Phys. Chem, 1968, 238, 2232. 30. L. Valade and P. Cassoux, Molecular inorganic superconductors, in Inorganic Materials, eds. D. W. Bruce and D. O’Hare, 2nd edn, Wiley, Chichester, 1996, Chapter 1, pp. 1–64. 31. A. Filhol, G. Bravic, J. Gaultier, et al., Acta Cryst. B, 1981, 37, 1225. 32. W. Kranig, C. Boeffel, H. W. Spiess, et al., Liq. Cryst., 1990, 8, 375. 33. W. H. de Jeu, L. Longa, and D. Demus, J. Chem. Phys., 1986, 84, 293. 34. M. A. Bates and G. R. Luckhurst, Liq. Cryst., 1998, 24, 229. 35. M. A. Bates, Liq. Cryst., 2003, 30, 181.

9. G. W. Gray, M. Hird, D. Lacey, and K. J. Toyne, J. Chem. Soc., Perkin Trans. II, 1989, 2041.

36. K. Praefcke and J. D. Holbrey, J. Incl. Phenom. Mol. Recogn. Chem., 1996, 24, 19.

10. T. Niori, T. Sekine, J. Watanabe, et al., J. Mater. Chem., 1996, 6, 1231.

37. K. Praefcke, D. Singer, and A. Eckert, Liq. Cryst., 1994, 16, 53.

11. (a) H. Matsuzaki and Y. Matsunaga, Liq. Cryst., 1993, 14, 105; (b) Y. Matsunaga and S. Miyamoto, Mol. Cryst., Liq. Cryst., 1993, 237, 311.

38. K. Praefcke, D. Singer, and B. G¨undogan, Mol. Cryst., Liq. Cryst., 1992, 223, 181.

12. D. R. Link, G. Natale, R. Shao, et al., Science, 1997, 278, 1924. 13. R. A. Reddy and C. Tschierske, J. Mater. Chem., 2006, 16, 907. 14. G. Pelzl, A. Eremin, S. Diele, et al., J. Mater. Chem., 2002, 12, 2591. 15. T. Niori, J. Yamamoto, and H. Yokoyama, Mol. Cryst., Liq. Cryst., 2004, 409, 475.

39. K. Praefcke, B. Bilgin, N. Usol’tseva, et al., J. Mater. Chem., 1995, 5, 2257. 40. K. Praefcke, S. Diele, J. Pickardt, et al., Liq. Cryst., 1995, 18, 857. 41. T. Hegmann, J. Kain, S. Diele, et al., Angew. Chem. Int. Ed., 2001, 40, 887. 42. C. Paleos and D. Tsiourvas, Angew. Chem. Int. Ed. Engl., 1995, 34, 1696.

17. C. Pr¨asang, A. C. Whitwood, and D. W. Bruce, Chem. Commun., 2008, 2137.

43. T. Kato, Hydrogen-bonded systems, in Handbook of Liquid Crystals, eds. D. Demus, J. Goodby, G. W. Gray, et al., Wiley-VCH, Weinheim, 1998, vol. 2B, Chapter XVII, pp. 969–980.

18. V. G¨ortz and J. W. Goodby, Chem. Commun., 2005, 3262.

44. K. Markau and W. Maier, Chem. Ber., 1962, 95, 889.

19. A. G. Vanakaras and D. J. Photinos, J. Chem. Phys., 2008, 128, 154512.

45. M. J. Brienne, J. Galard, J. M. Lehn, and I. Stibor, J. Chem. Soc., Chem. Commun., 1989, 1868.

20. I. Dozov, Europhys. Lett., 2001, 56, 247.

46. M. Suarez, J. M. Lehn, S. C. Zimmerman, et al., J. Am. Chem. Soc., 1998, 120, 9526.

16. V. G¨ortz and J. W. Goodby, Chem. Commun., 2005, 3262.

21. C. R. Patrick and G. S. Prosser, Nature, 1960, 187, 1021–1021. 22. C. Dai, P. Nguyen, T. B. Marder, et al., Chem. Commun., 1999, 2493. 23. S. W. Watt, C. Dai, A. J. Scott, et al., Angew. Chem. Int. Ed., 2004, 43, 3061–3063. 24. E. O. Arikainen, N. Boden, R. J. Bushby, et al., Angew. Chem. Int. Ed., 2000, 39, 2333–2336.

47. D. Goldmann, D. Janietz, C. Schmidtand, J. H. Wendorff, J. Mater. Chem., 2004, 14, 1521.

and

48. U. Beginn and G. Lattermann, Mol. Cryst., Liq. Cryst., 1994, 241, 215. 49. Y. Matsunaga and M. Terada, Mol. Cryst., Liq. Cryst., 1986, 141, 321.

25. O. R. Lozman, R. J. Bushby, and J. G. Vinter, J. Chem. Soc., Perkin Trans 2, 2001, 1446–1452.

50. (a) D. Pucci, M. Veber, and J. Malthˆete, Liq. Cryst., 1996, 21, 153; (b) C. Garcia and J. Malthˆete, Liq. Cryst., 2002, 29, 1133.

26. N. Boden, R. J. Bushby, G. Cooke, et al., J. Am. Chem. Soc., 2001, 123, 7915–7916.

51. J. Malthˆete, A. M. Levelut, and L. Liebert, Adv. Mater., 1992, 4, 37.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

22

Soft matter

52. P. A. Albouy, D. Guillon, B. Heinrich, et al., J. de Phys. II, 1995, 5, 1617.

non-linear organometallic materials, PhD Thesis, University of Sheffield, UK, 1995.

53. T. Kato and J. M. J. Fr´echet, J. Am. Chem. Soc., 1989, 111, 8533.

77. F. Guthrie, J. Chem. Soc., 1863, 16, 239–244.

54. T. Kato and J. M. J. Fr´echet, Macromolecules, 1989, 22, 3818.

79. O. Hassel and C. Rømming, Quart. Rev. Chem. Soc., 1962, 16, 1–18.

55. U. Kumar, T. Kato, and J. M. J. Fr´echet, J. Am. Chem. Soc., 1992, 114, 6630.

80. O. Hassel, Science, 1970, 170, 497–502.

56. T. Kato, H. Kihara, T. Uryu, et al., Macromolecules, 1992, 25, 6836. 57. T. Kato, H. Kihara, U. Kumar, et al., Angew. Chem. Int. Ed. Eng., 1994, 33, 1644. 58. T. Kato, P. G. Wilson, A. Fuijishima, and J. M. J. Fr´echet, Chem. Lett., 1990, 2003.

78. I. Remsen and J. F. Norris, Am. Chem. J., 1896, 18, 90–95.

81. J. M. Dumas, C. G´eron, H. Peurichard, and M. Gornel, Bull. Soc. Chim. France, 1976, 720. 82. J.-M. Dumas, H. Peurichard, and M. Gomel, J. Chem. Res. (S), 1978, 54–55. 83. A. C. Legon, Angew. Chem. Int. Ed., 1999, 38, 2686–2714. 84. A. C. Legon, Struct. Bond., 2008, 126, 17–64.

59. T. Kato, J. M. J. Fr´echet, P. G. Wilson, et al., Chem. Mater., 1993, 5, 1094.

85. P. Metrangolo, H. Neukirch, T. Pilati, and G. Resnati, Acc. Chem. Res., 2005, 38, 386.

60. T. Kato, A. Fuijishima, and J. M. J. Fr´echet, Chem. Lett., 1990, 912.

86. T. C. Clark, M. Hennemann, J. S. Murray, and P. Politzer, J. Mol. Model., 2007, 13, 291.

61. M. Fukumassa, T. Kato, T. Uryu, and J. M. J. Fr´echet, Chem. Lett., 1993, 65.

87. P. Politzer, J. S. Murray, and P. Lane, Int. J. Quantum Chem., 2007, 107, 3046.

62. T. Kato, H. Kihara, T. Uryu, et al., Ferroelectrics, 1994, 148, 1303.

88. P. Politzer, J. S. Murray, and M. C. Concha, J. Mol. Model., 2007, 13, 634.

63. H. Kihara, T. Kato, S. Ujiie, et al., Liq. Cryst., 1996, 21, 25.

89. H. L. Nguyen, P. N. Horton, M. B. Hursthouse, et al., J. Am. Chem. Soc., 2004, 126, 16.

64. D. J. Price and D. W. Bruce, Adv. Mater. Opt. Electron., 1994, 4, 273.

90. A. Wasilewska, M. Gdaniec, and T. Polloski, CrystEngComm, 2007, 9, 203.

65. K. Willis, D. J. Price, H. Adams, et al., J. Mater. Chem., 1995, 5, 2195–2200.

91. D. W. Bruce, P. Metrangolo, F. Meyer, et al., Chem. Eur. J., 2010, 16, 9511–9524.

66. D. J. Price, K. Willis, T. Richardson, et al., J. Mater. Chem., 1997, 7, 883.

92. C. Pr¨asang, A. C. Whitwood, and D. W. Bruce, Cryst. Growth Des., 2009, 9, 5319–26.

67. S. Coco, E. Espinet, P. Espinet, and I. Palape, Dalton Trans., 2007, 3267.

93. L. C. Roper, C. Pr¨asang, V. N. Kozhevnikov, et al., Cryst. Growth Des., 2010, 10, 3710–3720.

68. S. Coco, C. Cordovilla, C. Dom´ınguez, et al., Chem. Mater., 2009, 21, 3282.

94. L. C. Roper, C. Pr¨asang, A. C. Whitwood, and D. W. Bruce, CrystEngComm., 2010, 12, 3382–3384.

69. J. W. Goodby, V. G¨ortz, S. J. Cowling, et al., Chem. Soc. Rev., 2007, 36, 1971–2032.

95. P. Metrangolo, C. Pr¨asang, G. Resnati, et al., Chem. Commun., 2006, 3290.

70. V. Vill and R. Hashim, Curr. Opin. Colloid. Interfac. Sci., 2002, 7, 395–409.

96. D. W. Bruce, P. Metrangolo, F. Meyer, et al., New. J. Chem., 2008, 32, 477–482.

71. M. Fukumasa, T. Kato, T. Uryi, and J. M. J. Fr´echet, Chem. Lett., 1993, 65.

97. C. Pr¨asang, A. C. Whitwood, and D. W. Bruce, Chem. Commun., 2008, 21372139.

72. K. Willis, J. E. Luckhurst, D. J. Price, et al., Liq. Cryst., 1996, 21, 585–587.

98. J. Xu, X. Liu, J. Kok-Peng Ng, et al., J. Mater. Chem., 2006, 16, 3540.

73. D. J. Price, H. Adams, and D. W. Bruce, Mol. Cryst., Liq. Cryst., 1996, 289, 127.

99. C. Pr¨asang, H. L. Nguyen, P. N. Horton, et al., Chem. Commun., 2008, 6164–6166.

74. D. Demus, H. Demus, and H. Zaschke, Fl¨ussige Kristallen in Tabellen, VEB, Leipzig, 1974, vol. 1.

100. J. Xu, X. Liu, T. Lin, et al., Macromolecules, 2005, 38, 3554.

75. D. J. Price, T. Richardson, and D. W. Bruce, J. Chem. Soc., Chem. Commun., 1995, 1911.

101. P. J. Collings, Liquid Crystals: Nature’s Delicate State of Matter, 2nd edn, Princeton University Press, Princeton, 2001.

76. D. J. Price, Hydrogen-bond induced mesomorphism in heteromeric systems and studies towards optically

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc149

Covalent Capture of Self-Assembled Soft Materials Bing Gong SUNY Buffalo, Buffalo, NY, USA

1 Introduction 2 Covalent Capture of Liquid Crystal Phases 3 Covalent Capture of Gels 4 Conclusions References

1 1.1

1 2 5 7 7

INTRODUCTION Soft materials

Soft materials (or soft matter) encompass a wide variety of substances such as liquid crystals (LCs), colloids, gels, and composite materials.1 In addition, biological materials including connective tissue, bone, cell membrane, and scaffolds, as well as carbohydrate-based materials are also classified as soft matter. The “softness” of these materials arises from their many different physical states that are easily deformed by external mechanical or thermal stresses. With macroscopic behavior being determined by the weak but long-range interactions of their nanometer to micronscale structures, soft materials of various shapes are being designed to explore properties that are often remarkably different as compared to those of unstructured materials. Soft materials have found a wide range of applications in areas ranging from microfluidic device technology to nanofabrication, and to advanced biomedical applications such as controlled drug delivery, tissue engineering, and other smart materials.2, 3 Besides, soft materials also serve

as structural and packaging materials, foams and adhesives, detergents and cosmetics, paints, food additives, lubricants and fuel additives, and rubber in tires.

1.2

Self-assembly in soft materials

The majority of soft materials are based on small organic molecules or polymers of synthetic as well as biogenic origins. In spite of the seemingly large difference between small molecules and macromolecules, the underlying principles leading to soft materials remain the same. First, the spontaneous or directed self-assembly of small or macromolecules leads to nanometer or micron-scale structures. The interaction or further assembly of the nano- and mesoscopic structures then affords the macroscopic soft materials along with their various unique properties. Such a multilevel self-assembly results in structural hierarchies of increasing diversity and complexity. Given the presence of the extremely large variety of soft materials, covering all aspects of soft materials in one chapter is an impractical task. The following discussions are therefore based on examples from two specific systems, LCs, and gels, both of which involve hierarchical structural levels resulting from multilevel self-assembly. With properties of both a liquid and a solid crystal, LCs are formed from molecules capable of adopting (or assembling into) various arrangements in which the molecules are in short- or long-range positional and orientational orders.4 These structural orders correspond to the so-called mesophases. Depending on the type of LC molecules, nematic, smectic, discotic, and a few other phases can be exhibited. For example, disk-shaped (discotic) LCs provide a system illustrating the self-assembling structural hierarchy of soft materials.5 Discotic LC molecules may assemble into a layer-like nematic phase. They may also stack on

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

2

Soft matter

top of one another, forming a discotic columnar phase. The columns further assemble (or pack) into two-dimensional rectangular or hexagonal lattices. A columnar phase is also observed with lyotropic liquid crystals (LLCs) by the assembly of amphiphilic molecules (e.g., surfactants) into cylindrical aggregates of indefinite length.6 Similar to the assembly of columns consisting of discotic LC molecules, the cylinders formed from amphiphiles arrange into a hexagonal lattice called the hexagonal phase. Gels represent another class of soft materials that have attracted intense interest in recent years. They are mostly liquid but behave like solids because of the presence of a three-dimensional cross-linked network that spans the volume of the liquid medium and entraps the liquid molecules. The cross-linked network may result from the aggregation or assembly of small molecules (molecular gels)7, 8 or the entanglement of polymer chains (polymer gels).9, 10 The entrapment of water molecules by a cross-linked network formed from the entanglement of hydrophilic polymer chains leads to a hydrogel.11 Gels, especially hydrogels, occur widely in nature, and exist in the body as mucus, the vitreous humor of the eye, cartilage, tendons and blood clots, and as soft tissue component. Gels can be formed from many different substances. In fact, the wide occurrence and unique properties of gels have prompted researchers to develop synthetically derived gels as soft materials. For example, smart (intelligent) materials based on hydrogels can serve as sensors, actuators, switches, drug delivery systems, separation systems, and bioreactors, many of which allow their functions to be controlled in response to environmental stimuli such as changes in pH, temperature, or ionic strength. Hydrogels have had a major impact on tissue engineering by serving as the matrix for the attachment and spreading of cultured cells.12 Besides, gelators of organic fluids are also attracting increasing interest because of their numerous potential applications as soft materials.13, 14 To gelate solvents, long fibers derived from small molecules or polymers have to become entangled to trap liquid via surface tension. Small molecules undergo selfassembly first, forming extended 1D networks that further aggregate into higher order structures such as various fibers, ribbons, or sheets. Further entanglement of fibers (or other aggregates) without precipitation results in the entrapment and thus, the gelation, of solvent molecules. Obviously, the structural hierarchy accompanying the gelation of liquid bears close resemblance to that of protein structures, involving the self-assembly or aggregation of structures of each level. Linear polymer chains are covalently linked 1D networks. To serve as gelators, polymer chains have to be noncovalently (physically) or covalently (chemically)

cross-linked. Noncovalent cross-linking of polymer chains is possible with block copolymers consisting of solventcompatible segments that prevent undesired precipitation and other segments that aggregate and thus serve as physical cross-links. For hydrogels based on polymers, the soluble segments should be hydrophilic while the cross-linking segments should be able to engage in attractive interaction, such as hydrophobic or ionic interaction, in water. The noncovalent cross-linking of polymer chains, along with the interaction of the hydrophilic segments of such polymer chains with water molecules, give rise to a complicated self-assembling system consisting of multiple levels of structures.

1.3

Reinforcement of soft materials based on covalent capture

In spite of intense interest in developing soft materials such as gels, these materials share a weakness. The softness of these materials means that many of them are weak and fragile, which hampers many potential applications. The fragility associated with soft materials is due to their self-assembling nature, which relies on relatively weak noncovalent interactions that can be easily interrupted by external forces.15 For example, molecular gels or polymer physical gels are very sensitive to thermal changes and also undergo ready deformation under mechanical stress. To get around this limitation, there has been considerable interest in enhancing the thermal and mechanical strength of soft materials. A common practice is to covalently capture or fix the structurally organized nanoscale assemblies into a permanent, robust form.

2

2.1

COVALENT CAPTURE OF LIQUID CRYSTAL PHASES Covalent capture of lyotropic liquid crystals

Capturing the concentration- and temperature-dependent periodic nanostructures exhibited by LLCs has been regarded as a straightforward means toward the generation of organic nanostructures.16 The covalent retention of the different morphologies generated from the self-assembly of surfactants, from micelles, hexagonal, and lamellar LLC phases, to other phases such as the bicontinuous cubic phase, may provide nanostructured polymeric materials with unique features that cannot be obtained based on traditional methods. Specifically, reactive LLCs are formed from surfactant molecules bearing reactive (cross-linkable) functional groups. The advantage of reactive LLCs is that

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

Covalent capture of self-assembled soft materials

Columnar packing

3

Covalent crosslinking

Reactive amphphile

Figure 1

The assembly of LLC amphiphile into columns that are further cross-linked into a film.

R O

R1 O P+ 10

COO−Na+

O R

1

COO−(CH2)2-N

O 6

O R R = −(CH2)11OOCCH=CH2

Figure 2

Br − R2

P+ Br −

BF4− CH3

O

10

2

N

R1 3: R1 = R2 = −(CH2)11OOCCH=CH2 4: R1 = −(CH2)11OOCCH=CH2; R2 = −(CH2)11OH3

Reactive LLC amphiphiles that assemble into columnar phases.

the same surfactant that assembles into the LLC also serves as the monomer (Figure 1). For example, Gin et al. carried out seminal studies on the preparation of nanoporous membranes by cross-linking LLC assemblies.17 It was found that a supported NF (nanofiltration) membrane with cylindrical pores of 1.2 nm diameter, which was made by cross-linking the inverted hexagonal (HII ) phase formed from the self-assembly of a LLC amphiphile 1, showed excellent molecular size selectivity but low water fluxes because of nonuniform alignment of the uniaxial pore domains. Another NF membrane with pores having an effective size of approximately 0.75 nm was prepared by cross-linking a type I bicontinuous cubic (QI ) LLC network formed by another LLC monomer 2.18 The cross-linked QI phase membrane showed superior water permeability because, unlike the membrane based on the HII phase structure, its 3D interconnected nanopores did not need to be aligned (Figure 2). Columnar hexagonal phases of imidazolium and tris(alkoxy)phenyl moieties 3 and 4 containing polymerizable acrylate groups were synthesized by Kato et al. The columnar assembly was retained after photo-crosslinking.19 The obtained transparent polymer film was found to mediate anisotropic ion transport. The observed ion conduction is dependent on the orientation of columns, which indicates that the properties of these polymers are determined by the nanostructure of the resultant polymer, which in turn depends on both the structure of the surfactant monomer and the conditions adopted for the cross-linking reaction (Figure 3). To achieve better retention of the LLC order for a monomethacrylate quaternary ammonium surfactant

O Br − + 13N

O

O 5

O O

6

O

Figure 3 Reactive LLC amphiphile 5 assembles into a columnar phase that is reinforced by the presence of cross-linker 6 on photoirradiation.

monomer 5, which was obtained by treating dimethylaminoethyl methacrylate with tetradecyl bromide, Guymon et al. used a cross-linker, 1,6-hexanedioldimethacrylate 6, during photo-cross-linking. Cross-linker 6 could segregate to the polar regions of the surfactant assembly, likely near the quaternary ammonium group, that is, the methacrylate moiety, of reactive surfactant 5.20 It was found that the presence of a cross-linker led to the retention of the original LLC order after the covalent cross-linking of the hexagonal LLC phases that otherwise could not be retained by other means. That the produced polymer had a robust LLC order was evidenced by its LLC order after being dried and swelled in water.

2.2

Cross-linking of thermotropic liquid crystals

Similar to the retention of LLC phases, thermotropic LC assemblies have also been covalently captured, leading to robust polymer films that retain the original nanostructures. M¨oller et al. cross-linked columnar assemblies formed by amphiphile 7 bearing crown ether units inside the pores of track-etched membranes and obtained membranes containing oriented channels.21 Salt-diffusion experiments demonstrated that the resultant nanoporous membrane displayed

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

4

Soft matter

O O

O R O O

O

O R = −H2C

R O

O (CH2)11-OOC(CH3)C=CH2

O 7

R O

Figure 4

Amphiphile 7 assembles into columns that are aligned with the pores of track-etched membranes.

strongly enhanced ion transport rates as compared to those without oriented channels (Figure 4). Broer and coworkers prepared thermotropic LC monomers with acrylate-containing moieties and achieved preservation of LC order by using in situ photopolymerization.22, 23 It was observed that the resulting liquid-crystalline polymer exhibited anisotropic optical properties that could be varied by structurally tuning spacer length and substituents of monomers. Polymerization conditions were found to have an important effect in this system.23

2.3

In spite of their interesting properties, the columnar assemblies formed by PAHs, especially HBCs, could find few practical applications because of their instability. To overcome this limitation, M¨ullen et al. explored the covalent fixation of the supramolecular columnar assemblies of 10, an HBC bearing six side chains with terminal acrylate units.35 The columnar assembly of this reactive HBC is preserved by covalent cross-linking either in the crystalline state using photon energy or in the mesophase using thermal energy. It was found that the HBC monomers, being aligned in the LC columnar phase, could undergo photopolymerization without the addition of a photo-initiator, which led to the retention of the organization of the liquidcrystalline material in either the crystalline state or the mesophase. An approach that led to carbon nanofibers, nanotubes, and nanoribbons of high aspect ratios based on templatedirected LC assembly and covalent capture was described by Hurt et al.36 By controlling the interactions between discotic molecules and template surfaces, LC assemblies were directed to produce four new high aspect ratio carbon nanomaterials, that is, with varying graphene layer arrangements. Specifically, the combinations of two different LC precursors, thermotropic AR (aromatic resin) mesophase and lyotropic indanthronedisulfonate, and two different nanochannel templates, alumina and pyrolytic carbon, allowed both the shape of the nanocarbon and the graphene layer arrangement to be systematically engineered.

Covalent capture of discotic liquid crystals

Discotic liquid-crystalline materials based on polycyclic aromatic hydrocarbons (PAHs) have a high propensity to form supramolecular columnar assemblies via π –π stacking interactions.24, 25 With high charge-carrier mobilities along the columnar stacks, these materials are promising as electronic components26 with applications such as the development of field-effect transistors and photo-voltaic devices.27–32 A specific class of PAHs that have attracted intense attention are hexa-peri -hexabenzocoronenes (HBCs), a flat aromatic molecule built up from 13 fused six membered rings (Figure 5), because the large aromatic core of these molecules permits some of the highest intrinsic charge-carrier mobility values among mesogens.33, 34

R1 R6

R R2

R

R R=

O O

R3

R5

8

R4 9

R

R R 10

Figure 5 The general structures of hexa-peri -hexabenzocoronene (HBC) 8 and its functionalized derivatives 9. Reactive monomer 10 that assembles into a cross-linkable LC columnar phase. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

Covalent capture of self-assembled soft materials

5

O 11: R = O

O NH

HN

R

Figure 6

3 3.1

N H

R

O

O O

12: R =

Gelators 11 and 12 bearing reactive groups at both termini and in the middle.

COVALENT CAPTURE OF GELS Covalent cross-linking of small molecular gels

The gelation of various solvents by low molecular weight organic compounds represents one of the most spectacular examples of molecular self-assembly.37–40 Small molecular gels differ from macromolecular gels41, 42 in that the gelator assembles into fibrous networks that further entangle and entrap solvent molecules via aggregation mediated by noncovalent forces such as hydrogen bonding or hydrophobic interactions. It is well known that the resultant networks are capable of immobilizing up to 105 liquid molecules per gelator molecule, enhancing the viscosity of organic solvents by up to 1010 fold, in many cases leading to gels with the ability to respond to a variety of stimuli.43, 44 Covalent cross-linking helps to enhance the thermostability of organogels and hydrogels, which has traditionally been performed based on reactions such as free radical or photo-initiated polymerization and cross-linking. For example, polymerizable organogelator 11, based on (1R, 2R)-trans-1,2-bis(ureido)cyclohexane, was designed to examine the effect of covalent cross-linking on enhancing the stability of the corresponding organogel.45 Compound 11 was able to form optically transparent gels with a variety of organic solvents, including cyclohexane, butyl acetate, benzene, tetralin, and 1,2-dichloroethane, at very low concentrations. However, it was observed that gels of 11 remained stable for only 1–10 days, after which precipitation or crystallization occurred, and were easily disrupted by mechanical agitation. On photo-cross-linking, the melting temperatures of the resultant gels, indicative of thermal stability, increased dramatically, up to temperatures well above the boiling points of the solvents, or at very low concentrations of 11 (Figure 6).

In a conceptually similar piece of work,46 photo-crosslinking of gels formed by 12 that contains two diacetylene units led to new robust gels that retained their gel–fiber superstructure. Compound 12 was able to gelate both nonpolar and polar solvents such as benzene, hexane, methanol, 1-butanol, and acetonitrile, leading to colorless gels that turned blue on photo-irradiation. Compound 13, containing two cholesteryl ester units at the ends and a urethane group between a diyne and each cholesteryl group, was found to form a polymerizable organic gel with nonpolar solvents such as cyclohexane or mixtures of hexane and dichloromethane.47 Similar to the above example, the colorless gels formed with the mixtures of hexane and dichloromethane or with cyclohexane were converted into dark blue ones on irradiation with ultraviolet (UV) light. TEM (transmission electron microscopy) showed that the fibril structure in the original gel remained unchanged after polymerization. However, photo-polymerization increased the stability of the gel state. For example, the Tgel for cyclohexane gel [(1) = 5.3 mmol dm−3 ] was 45 ◦ C before polymerization. The polymerized gel maintained the shape after heating it to the boiling point of cyclohexane (80.7 ◦ C) (Figure 7). It needs to be pointed out that the success of the above covalent capture reactions relies on the careful alignment of the bisalkyne groups resulting from the hydrogen-bonding interaction between the amide and carbamate moieties. These systems demonstrate the ability of supramolecular methods to place reactive groups into precisely defined positions. In recent years, many groups have employed the “click reaction” in the supramolecular environment of organogels and hydrogels. One such example is in the stabilization of low molecular weight organogelators 14 and 15 based on the undecynylamide of trans-1, 2-diaminocyclohexane H N

O O R

O O

Figure 7

N H

O O

R

O O 13: R = Cholesterol

Gelator 13 with two chlosteryl ester units and a cross-linkable central diyne moiety.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

6

Soft matter

O

cross-linking via olefin metathesis with Grubbs’ secondgeneration catalyst.49 This approach led to the covalent capture of the fibrillar self-assembled network and yielded a material that was robust, thermally stable, and highly swellable in solvents such as dimethyl sulfoxide (DMSO), toluene, and styrene that were compatible with the gelator. The obtained results suggest that the gel-type behavior and solvent compatibility of compound 18 were retained within the captured material (Figure 9). Another example of covalent capture based on reversible covalent bonding involves the fibers and gels formed by peptide amphiphiles (PAs) derived from 19.50 The role played by different structural units in these designed molecules was investigated by modifying the alkyl tail length and peptide amino acid composition. These PAs were designed to have hydrophobic tails of various lengths and a peptide block consisting of a segment of four consecutive cysteine residues and another segment rich in ionic amino acid side chains. Nanofibers and their corresponding networks varying in morphology, surface chemistry, and potential bioactivity were obtained from the self-assembly of these PAs. The nanofiber networks resulted in the formation of aqueous gels through pH changes. Reversibly cross-linking (polymerizing) the PA fibers via disulfide bond formation served to enhance the stability of the fibers and gels. This system incorporates two switchable events, that is, pH changes and disulfide bond formation, controlling the formation of supramolecular structure and polymerization, which resulted in a remarkably versatile material (Figure 10).

O NH HN

R

R

14: R = −(CH2)8

N3 16

N3

17

15: R = −(CH2)10-N3

Figure 8 The structures of gelators 14 and 15 and the corresponding cross-linkers 16 and 17.

O 7

O NH

HN

7

18 O 7

N H

H N

O

H N 10

O

O

N H

7

Figure 9 Gels of 18 were cross-linked via olefin metathesis in the gel phase.

(Figure 8).48 Unable to gelate acetonitrle at room temperature, compound 14 or 15, in the presence of diazide 16 or dialkyne 17 and CuI, undergo copper-catalyzed azide–alkyne (3+2) cycloaddition (CuAAC) reaction in acetonitrile, leading to the formation of gels with enhanced thermal stability. The advantage of this strategy lies in maintaining the overall structure and the thermoreversibility of the organogels. Reversible covalent bonding was also adopted in the covalent capture of gels and other supramolecular structures. Gels formed by compound 18, which were functionalized with peripheral alkene groups, were stabilized by

H N

C15H31 O

Hydrophobic tail

O N H SH

SH H N O

O N H SH

SH H N O

3.2

Covalent cross-linking of macromolecular gels

In addition to small molecular gels, the covalent capture or cross-linking of macromolecular gels has also

O N H

H N O

O HO P OH O O O H N N N H H O

H N O

COOH 19 COOH

NH Cysteine-rich (cross-linking) segment

H2 N

NH

pH-responsive (ionic) segment

Figure 10 A series of peptide amphiphiles derived from 19 were prepared by varying the three segments as shown. Fibers and gels formed by these peptide amphiphiles were cross-linked based on disulfide bond formation. Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

Covalent capture of self-assembled soft materials OH

OH OH O

HO

7

HO O

NH

O

OH HO

O

O

HO O

NH

O NH

H3COC

O

O

O NH

H3COC

R 20: R = O

O

O

N3

21

Figure 11 Mixing water-soluble polysaccharide derivatives 20 and 21 in a 1 : 1 ratio in the presence of Cu(I) led to rapid gelation of water because of the cross-linking of the polymers resulted from 1,3-dipolar cycloaddition (click) reaction.

been actively attempted and resulted in robust materials. For example, poly(γ -benzyl L-glutamate) (PBLG) in the cholesteric liquid-crystalline (CLC) state was cross-linked by using diamines of different chain types as the crosslinkers in chloroform, methylene dichloride, and dioxane.51 The irreversible cross-linking of PBLG in the CLC phase led to a polymer gel that retained the CLC order (CLC gel), with reversible CLC-isotropic transition at the helix-coil transition region on the change of solvent or temperature. While retaining the native collagen triple helical structure and maintaining high cell viability, the mechanical properties of type I collagen gels were enhanced based on photo-mediated cross-linking in the presence of rat aortic smooth muscle cells (RASMC).52 Modified with acrylate groups, collagen was cast into gels in the presence of a photo-initiator along with RASMC. Cross-linking using visible light irradiation led to gels in which the collagen triple helical content remained unchanged. After the cross-linking reaction, significant improvement of mechanical properties was evidenced by the increase of denaturation temperature beyond the physiologic range, along with a 12-fold increase in shear modulus. Besides, satisfactory cell viability (in the range of 70%) was observed in the photo-cross-linked gels. Similar to that observed for natural type I collagen, the cells were able to contract the cross-linked gel. Click chemistry was employed in the in situ rapid chemical gelation of aqueous solutions of hyaluronan.53 Thus, the side chains of hyaluronan were endowed with either azide or alkyne terminal functionality, leading to water-soluble polysaccharide derivatives 20 and 21 (Figure 11). When 20 and 21 were mixed together in aqueous solution, a 1,3dipolar cycloaddition reaction in the presence of catalytic amounts of Cu(I) resulted in fast gelation at room temperature. Performing the gelation reaction in the presence of drug molecules in aqueous media led to polysaccharide networks that acted as drug reservoirs. Besides, yeast cells entrapped within the polysaccharide networks were found to be homogeneously distributed and smoothly adhered, and remained viable.

4

CONCLUSIONS

The covalent capture of self-assembling soft materials, as exemplified in this chapter by the cross-linking of LCs and gels, is attracting increasing attention as an approach toward robust nanostructures and nanomaterials. In most of the systems examined, the self-assembling architectures are retained while the mechanical strength of the crosslinked materials is enhanced significantly. Although early efforts employed well-established polymerization methods that involve irreversible covalent bond formation, examples reported recently have taken advantage of reversible covalent bonding such as disulfide bond formation and olefin metathesis. The inclusion of reversible, that is, thermodynamically controlled covalent forces in capturing selfassembling structures is of particular value in achieving the desired outcome with minimum disruption of the original nanoarchitectures. In fact, reversible covalent capturing has allowed the creation of not just extended one-, two-, or three-dimensional networks exemplified here, but also has afforded many novel molecular nanostructures, such as those reported by Ghadiri54 and Zimmerman,55 that otherwise would be beyond the reach of current synthetic methods. Compared to currently known systems based on covalent capture, a more fundamental approach that is being explored involves the development of new association units that combine the high specificity of multiple noncovalent forces and the strength of covalent interactions by integrating the spontaneity and error-correcting ability of selfassembling structures with the stability typical of covalent bonds.56–58 The availability of such association units could result in the next major progress in synthetic chemistry by providing a powerful, bottom-up self-assembly approach for fabricating robust nanoscale devices.

REFERENCES 1. I. W. Hamley, Introduction to Soft Matter: Polymers, colloids, amphiphiles and liquid crystals, 2nd edn, John Wiley & Sons, Ltd, Chichester, 2000.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

8

Soft matter 2. S. R. Quake 1536–1540.

and

A. Scherer,

Science,

2000,

290,

29. S. X. Xiao, M. Myers, Q. Miao, et al., Angew. Chem. Int. Ed., 2005, 44, 7390–7394.

3. I. W. Hamley, Angew. Chem. Int. Ed., 2003, 42, 1692–1712.

30. W. Pisula, A. Menon, M. Stepputat, et al., Adv. Mater., 2005, 17, 684–689.

4. P. J. Collings and M. Hird, Introduction to Liquid Crystals, Taylor & Francis, Bristol, PA, 1997. 5. S. Chandrasekhar, B. K. Sadashiva, and K. A. Suresh, Pramana, 1977, 9, 471–480. 6. R. G. Laughlin, The Aqueous Phase Behaviour of Surfactants, Academic Press, London, 1996. 7. J. van Esch, F. Schoonbeek, M. De Loos, et al., Low molecular weight gelators for organic solvents, in Supramolecular Science: Where It Is and Where It Is Going, eds R. Ungaro and E. Dalcanale, Kluwer Academic Publishers, 1999, pp. 233–259. 8. L. A. Estroff and A. D. Hamilton, Chem. Rev., 2004, 104, 1201–1218. 9. J. D. Ferry, Viscoelastic Properties of Polymers, John Wiley & Sons, Inc., New York, 1980. 10. Y. Osada and A. R. Khokhlov, Polymer Gels and Networks, Marcel Dekker, New York, 2001. 11. A. Singh, P. K. Sharma, V. K. Garg, and G. Garg, Int. J. Pharm. Sci. Rev., 2010, 4, 97–105. 12. H. Uludag, P. De Vos, and P. A. Tresco, Adv. Drug Delivery Rev., 2000, 42, 29–64. 13. M. George and R. G. Weiss, Acc. Chem. Res., 2006, 39, 489. 14. P. Terech and R. G. Weiss, Chem. Rev., 1997, 97, 3133. 15. A. R. Hirst, I. A. Coates, T. Boucheteau, et al., J. Am. Chem. Soc., 2008, 130, 9113–9121. 16. D. L. Gin, W. Gu, B. A. Pindzola, and W.-J. Zhou, Acc. Chem. Res., 2001, 34, 973–980. 17. M. Zhou, T. J. Kidd, R. D. Noble, and D. L. Gin, Adv. Mater., 2005, 17, 1850–1853. 18. M. J. Zhou, P. R. Nemade, X. Y. Lu, et al., J. Am. Chem. Soc., 2007, 129, 9574–9575. 19. M. Yoshio, T. Kagata, K. Hoshimo, et al., J. Am. Chem. Soc., 2006, 128, 5570–5577.

31. M. Oukachmih, P. Destruel, I. Seguy, et al., Sol. Energy Mater. Sol. Cells, 2005, 85, 535–543. 32. L. Schmidt-Mende, M. Watson, K. M¨ullen, and R. H. Friend, Mol. Cryst. Liq. Cryst., 2003, 396, 73–90. 33. W. Pisula, M. Kastler, D. Wasserfallen, et al., Chem. Mater., 2006, 18, 3634–3640. 34. A. M. van de Craats, J. M. Warman, K. M¨ullen, et al., Adv. Mater., 1998, 10, 36–38. 35. M. Kastler, W. Pisula, R. J. Davies, et al., Small, 2007, 3, 1438–1444. 36. C. Chan, G. Crawford, Y. M. Gao, et al., Carbon, 2005, 43, 2431–2440. 37. D. J. Abdallah and R. G. Weiss, Adv. Mater., 2000, 12, 1237–1247. 38. P. Terech and 3133–3159.

R. G. Weiss,

Chem. Rev.,

1997,

97,

39. J. H. van Esch and B. L. Feringa, Angew. Chem., Int. Ed., 2000, 39, 2263–2266. 40. O. Gronwald and S. Shinkai, Chem. Eur. J., 2001, 7, 4328–4334. 41. D. Derossi, K. Kajiwara, Y. Osada, and A. Yamauchi, eds, Polymer Gels: Fundamentals and Biomedical Applications, Plenum Press, New York, 1991. 42. G.-M. Guenet, Thermorersible Gelation of Polymers and Biopolymers, Academic Press, London, 1992. 43. F. Ilmain, T. Tanaka, and E. Kokufuta, Nature, 1991, 349, 400–401. 44. Y. Osada and A. R. Khokhlov, Polymer Gels and Networks, Marcel Dekker, New York, 2002. 45. M. de Loos, J. van Esch, I. Stokroos, et al., J. Am. Chem. Soc., 1997, 119, 12675–12676.

20. L. Sievens-Figueroa and C. A. Guymon, Chem. Mater., 2009, 21, 1060–1068.

46. K. Inoue, Y. Ono, Y. Kanekiyo, et al., Chem. Lett. 1999, 28, 429–430.

21. U. Beginn, G. Zipp, A. Mourran, et al., Adv. Mater., 2000, 12, 513–516.

47. N. Tamaoki, S. Shimada, Y. Okada, et al., Langmuir, 2000, 16, 7545–7547.

22. D. J. Broer, H. Finkelmann, and K. Kondo, Makromol. Chem., 1988, 189, 185–194.

48. D. D. Diaz, K. Rajagopal, E. Strable, et al., J. Am. Chem. Soc., 2006, 128, 6056–6057.

23. D. J. Broer and G. N. Mol, Makromol. Chem., 1991, 192, 59–74.

49. J. R. Moffat, I. A. Coates, F. J. Leng, and D. K. Smith,. Langmuir 2009, 25, 8786–8793.

24. S. Ito, M. Wehmeier, J. D. Brand, et al., Chem. Eur. J., 2000, 6, 4327–4342.

50. J. D. Hartgerink, E. Beniash, S. I. Stupp,. Proc. Natl. Acad. Sci. U. S. A., 2002, 99, 5133–5138.

25. B. Alameddine, O. F. Aebischer, W. Amrein, et al., Chem. Mater., 2005, 17, 4798–4807.

51. R. Kishi, M. Sisido, and S. Tazuke, Macromolecules, 1990, 23, 3779–3784.

26. H. Iino, Y. Takayashiki, J. Hanna, et al., Appl. Phys. Lett., 2005, 87, 192105–192107.

52. W. T. Brinkman, K. Nagapudi, B. S. Thomas, and E. L. Chaikof, Biomacromolecules, 2003, 4, 890–895.

27. J. P. Schmidtke, R. H. Friend, M. Kastler, and K. M¨ullen, J. Chem. Phys., 2006, 124, 174704.

53. V. Crescenzi, L. Cornelio, C. D. Meo, et al., Biomacromolecules, 2007, 8, 1844–1850.

28. I. O. Shklyarevskiy, P. Jonkheijm, N. Stutzmann, et al., J. Am. Chem. Soc., 2005, 127, 16233–16237.

54. T. D. Clark, K. Kobayashi, and M. R. Ghadiri, Chem. Eur. J., 1999, 5, 782–792.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

Covalent capture of self-assembled soft materials

9

55. Y. Kim, M. F. Mayer, and S. C. Zimmerman, Angew. Chem. Int. Ed., 2003, 42, 1121–1126.

57. M. F. Li, K. Yamato, J. S. Ferguson, et al., J. Am. Chem. Soc., 2008, 130, 491–500.

56. H. Zeng, R. Miller, R. A. Flowers, and B. Gong, J. Am. Chem. Soc., 2000, 122, 2635–2644.

58. B. Gong, Polym. Int., 2007, 56, 436–443.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc150

Designing Peptide-Based Supramolecular Biomaterials Vineetha Jayawarna and Rein V. Ulijn University of Strathclyde, Glasgow, UK

1 Introduction 2 The Toolbox: Amino Acids and Peptides 3 Design Paradigms for Peptide Self-Assembly 4 β-Sheets/β-Hairpins 5 α-Helices/Coiled-Coils 6 Aliphatic Peptide Amphiphiles 7 Aromatic Peptide Amphiphiles 8 Conclusions References

1

1 3 3 3 7 8 11 14 14

INTRODUCTION

One rapidly emerging area where supramolecular chemistry can have a significant impact is in biomaterials science, and specifically in the design of cell-contacting biomaterials. Indeed, in their natural environments within tissues, cells are supported by a natural supramolecular and dynamic gel-phase network, known as the extracellular matrix (ECM). The key features of the ECM, which are known to influence cell behavior at any particular time, include chemical signals, mechanical properties (i.e., gel stiffness), and topographical signals related to fiber sizes and alignment. These signals can, in principle, be incorporated into a material using noncovalent approaches. Indeed, peptides and their derivatives have been shown to spon-

taneously (or in response to an externally applied trigger) assemble into highly organized fibrous structures with welldefined mechanical, chemical, and structural properties that show similarities with natural ECM (but are much less complex). In this review, we provide an overview of current supramolecular chemistry strategies in the design of ECM mimics based on peptides and their derivatives. Although universal design rules for self-assembling molecules are not yet available,1 and much work relies on serendipitous discovery, molecule/structure relationships for a number of peptide-based systems are now emerging. The natural ECM is a hydrogel material, because of formation of a sample spanning network of fibers, which entraps water, giving rise to gel-like properties. Key to the formation of such gelforming fibers is unidirectional molecular self-recognition, that is, a preferential binding of copies of the same molecule in one direction. Peptides can be designed to interact in defined ways, by exploiting noncovalent interactions, including electrostatic, hydrophobic, π -stacking, hydrogen bonding, as well as steric contributions. By careful choice of building blocks, peptide structures can be encouraged to form fibrous structures based on the well-known secondary or quaternary structures (α-helix, β-sheet, and collagen triple helix) found in natural proteins or completely new synthetic sequences such as aromatic and aliphatic peptide amphiphiles (PAs). Naturally, there is a clear analogy in design strategies used for PAs and structurally simpler selfassembling block copolymers, which are discussed elsewhere in this chapter (Section 6). The diversity that may be achieved is huge, nearly endless peptide sequences are possible even if only the gene encoded amino acids are considered (20n , where n is the number of amino acids in the peptide sequence) (Figure 1). Provided that the appropriate

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

2

Soft matter

Hydrophobic

Hydrophilic/ polar

Aliphatic

Neutral

O H2N

O OH

O

H2N

O

H2N

O

H2N

OH

O

H2N

OH

OH

OH

H2N

O OH

H2N

OH

O

O H2N

H2N

OH OH

S

O

Methionine (M)

Isoleucine (I)

Leucine (L)

Alanine (A)

Aromatic H2N

H2N

H2N OH

OH

Serine (S)

Threonine (T)

Asparagine (N)

H2N

OH NH

OH NH2

Tryptophan (W)

Other O

N H

H2N

H N OH

H2N OH

Lysine (K)

Proline (P)

Peptide building blocks

O

Aspartic acid (D)

Fibrous network

OH

OH OH O

Cysteine (C)

Histidine (H)

H2N

OH SH

Glycine (G)

NH

Arginine (R)

H2N

OH N

NH2

O

Acidic

O

O

O H2N

H2N

OH

Tyrosine (Y)

O

Glutamine (Q)

O

O

OH NH

Phenylalanine (F)

H2N

Basic

O

O

O

Valine (V)

OH

OH NH2

O

OH

Glutamic acid (E)

Cell culture

R

N H

Hydrogel O

n 1.0 µm

Figure 1 Chemical structures of the 20 amino acids found in peptides, grouped according to the characteristics of their side chain. Common names and their one letter abbreviations are used to represent the amino acids. Bottom: an atomic force microscopy image showing typical morphology of a self-assembling peptide network, macroscopic appearance of a gel for cell culture, and appearance of fluorescent-stained gels in a 3D construct.

chemistry is presented at the fiber surface to favor entrapment of water, gel-phase materials emerge. Indeed, peptidebased self-assembling systems are a special category of low-molecular-weight gelation systems, as discussed elsewhere in this chapter (Section 7). Beyond such production of gels, the next level of complexity is to introduce biochemical instructions to enhance and support specific cell behavior. Cellular behavior is known to be responsive to a triad of material cues, which are (i) chemical,2 (ii) mechanical,3 and (iii) topographical.4 Gels from supramolecular fibers based on peptides can, in principle, be designed with control of each of these cues. In terms of chemical information, hydrogels can incorporate bioactive peptides, such as the well-known cell binding RGD sequence, as presented by the natural ECM.5 Material stiffness has been proven to be a crucial parameter, especially for stem cell culture.3 It has been demonstrated that stem cells can be influenced to differentiate to

cells that express markers typical of neuronal (on chemically cross-linked soft hydrogels of 0.1–1 kPa), muscle (medium, 10 kPa), or bone (stiff, 100 kPa). Supramolecular peptide-based hydrogels of varying stiffness covering this entire range of moduli have been reported,6 and it is only a matter of time until these materials will be explored for controlled stem cell differentiation. Indeed, tuning the assembly kinetics through controlling temperature, pH, and ionic strength, or by enzymatically controlled7 formation of building blocks allows for the synthesis of gels with different mechanical properties from identical chemical precursors. Topographical control of nanoscale features is also possible using self-assembling materials, with fibers of varying width and levels of clustering and alignment achievable.7 This chapter gives an overview of the design rules of peptide self-assembly systems for biological applications concentrating on four main classes of self-assembly design

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials

3

rules: β-sheets, α-helix, aliphatic, and aromatic PAs. For each category, the control of chemical and mechanical properties is discussed, followed by examples of designed self-assembly peptide systems in 2D and 3D cell culture (Figure 1).

methods may have advantages for the production of longer peptides or for large-scale production of medium length peptides. There are many excellent text books that cover peptide synthesis approaches and they are, therefore, not covered here.

2

3

THE TOOLBOX: AMINO ACIDS AND PEPTIDES

Nature’s supramolecular structures are largely composed of the same 20 gene encoded amino acids (Figure 1). They are commonly classified on the basis of the chemical nature and polarity of the R group/side chains.8, 9 Thus, the structures shown in Figure 1 are grouped as follows: (i) hydrophobic or nonpolar amino acids that are either aliphatic (Methionine M, Isoleucine I, Leucine L, Alanine A, Valine V) or aromatic (Phenylalanine F, Tyrosine Y, Trytophan W); (ii) hydrophilic or polar amino acids that are neutral (Serine S, Threonine T, Asparagine N, Glutamine Q), basic (Lysine K, Arginine R, Histidine H), or acidic (Aspartic acid D, Glutamic acid E); and (iii) “others,” the amino acids that impact directly upon the structural folding of the peptide chain (Proline P, Glycine G, Cysteine C). While aliphatic residues provide a general hydrophobic environment, aromatic residues promote π –π stacking via overlapped π -orbitals. As we see later on, π –π interactions can play an important role in supramolecular interactions between peptide building blocks. The amino acids that fall into the category “others” offer structural or chemical modifications. Proline, for example, has a cyclic locked conformation, which gives it an exceptional conformational rigidity compared to other amino acids. A hydrogen atom R group in glycine, on the other hand, introduces flexibility through removing the steric hindrance normally found in other amino acid groups. Both glycine and proline are present in the α chains of the collagen subunits and are largely responsible for the triple-helical supramolecular structure of collagen. The side chains of cysteine residues offer a unique chemical reactivity (formation of disulfide bond under oxidizing conditions) and are often used in structures where chemical modification and interpeptide cross-linking are demanded. In the following discussions, we use the single letter codes to represent amino acids. Synthesis of peptide building blocks generally follows one of the three strategies: solid-phase peptide synthesis (SPPS) on polymeric beads,10 solution-phase synthesis, and recombinant production using transgenic organisms.11 Solution-phase synthesis is suited to shorter oligopeptides and PAs, solid-phase methods are preferred for medium length (arbitrarily 8–20 amino acids), while recombinant

DESIGN PARADIGMS FOR PEPTIDE SELF-ASSEMBLY

A number of strategies have been developed to achieve unidirectional self-assembly based on peptides and their derivatives. Most commonly, these approaches are based on using the sequences that are found in naturally occurring systems. For example, β-sheets and β-hairpins (Section 4), α-helices/coiled-coils (Section 5), collagenmimetic peptides,12, 13 and elastin-like peptides14 are under active investigation, exploiting the known sequence designs as discussed below. Chemically designed (nonnatural) peptide self-assemblies involve linking peptides to nonpeptidic components by producing amphiphiles, such as polymer,15 aliphatic (Section 6), and aromatic PAs (Section 7).

4 4.1

β-SHEETS/β-HAIRPINS Design rules: peptide sequence

The β-sheet structure was first identified by Pauling and Corey in the early 1950s.16 β-sheets consist of multiple peptide chains with extended backbone arrangements that enable hydrogen bonding between the backbone amides and carbonyls (Figure 2A). Each chain is referred to as a strand and the strands that link via hydrogen bonds are referred to as a sheet. Depending on the orientation of the individual strands within the β-sheet, a parallel or an antiparallel structure can be obtained. In a parallel β-sheet structure, all the C-termini are located at one end of the structure, whereas in the (more stable) antiparallel configuration, the N- and C-termini are alternating (Figure 2B). The first examples of cell supporting peptide networks were described by Zhang et al. in 1993. They recognized that peptides with distinct, repetitive molecular recognition patterns of alternating cationic-hydrophobic–anionichydrophobic can self-associate through a combination of hydrogen bonding, electrostatic interactions, and the hydrophobic effect. The hierarchical assembly of these systems ultimately results in the formation of macroscopic membranes or gel-phase materials composed of extended one-dimensional fibers. A majority of reports

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

4

Soft matter +

−

R

H

O

+

R

H

O

R

H

O

R

H

N N N O

R

H

O

R

H

R

H

H

R

O

O

R

H

H

R

O

O

Hydrophilic

N

N R

H

N

N

+ − + − + −

Hydrophobic

O

O O

(A)

N

N

N H

H

R

N

N

N

N

O

R

O

(B)

H

N R

O

H

R

Antiparallel b-sheets

100 µm

100 nm (C)

Functional motif

(D)

(E)

Figure 2 Peptide-based gels based on β-sheet structures. (A) Schematic representation of β-sheet forming peptide with alternating charged/hydrophobic residues. (B) β-sheet containing two antiparallel strands. Hydrogen bonds are indicated with dotted red lines. Hydrophilic face in red and hydrophobic face in blue. (C) Representation of self-assembling peptide nanofibers formation with bioactive peptides extended out from the nanofiber. (D) Nanoscale morphology of a self-assembling peptide nanofiber analyzed by using atomic force microscopy. (Reproduced from Ref. 17.  Public Library of Science, 2010.) (E) Fluorescence microscopy image of periodontal ligament fibroblasts on bioactive RADA gel. F-actin is shown in red and nuclei in green. (Reproduced from Ref. 17.  Public Library of Science, 2010.)

on β-sheet-rich peptide systems concentrated on two sequences: EAK 16 (16 amino acids arranged into four repeating units of negatively charged E residues and positively charged K, separated by hydrophobic A, (Table 1, entry 1) and (RADA)4 (16 amino acids arranged into four repeating units of positively charged R and negatively charged D, separated by hydrophobic A).18–22 These sequences fold into sheets through hydrophobic and ionic bonds to produce a hydrophobic and hydrophilic face, respectively. When they interact with water, the hydrophilic face is positioned on the outside of the nanofibers, while the hydrophobic face is buried inside a double sheet (Figure 2B).23 In 1997, the groups of Aggeli and Boden in Leeds separately developed a range of peptides based on three similar design rules: (i) cross-attractive forces (hydrophobic, electrostatic, and hydrogen bonding) between side chains; (ii) lateral recognition between adjacent βstrands to constrain self-assembly to one dimension; and (iii) strong adhesion of solvent to the surface of tables to control solubility.24 On the basis of these rules, and inspired by the well-known formation of insoluble aggregates of poly-glutamine peptides, they developed a range of peptides with charged, hydrophobic glutamine (Q) residues. Under certain conditions, these peptides formed unidirectional fibers and gel-phase materials.

4.2

Controlling self-assembly

The properties of the hydrogel materials that result from β-Sheet self-assembly can be controlled using several approaches. These include varying the peptide concentration, or the building block sequence and length. In addition, assembly kinetics could be controlled by application of external stimuli. Changing peptide concentration demonstrated that the self-assembly process is hierarchical in nature in a concentration-dependent manner, giving rise to tapes, ribbons, fibrils, and fibers, which depend on the number of β-sheet tapes that packed together to form the final assembly25 (Table 1, entry 2). Controlling the assembly could also be realized through systematic variation of the peptide sequence.26 Sequence and compositional dependence were investigated by varying three key aspects: (i) nature of hydrophobic residues; (ii) charged side chains; and (iii) number of these repeats in the sequence. When alternative hydrophobic residues (FKFE), (IKIE), and (VKVE) were introduced, the sequences with more hydrophobic residues formed gels at lower concentrations. Counterintuitively, gelation happened in long peptide sequences (16 amino acid peptides) at higher critical concentration compared to short sequences (8 amino acid peptides) (Table 1, entry 3). Competition between favorable

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials Table 1 S. No.

5

Designer self-assembling systems based on β-sheet and β-hairpins structures. Systema

References

Stimulus

Stiffness G
(Pa)

1

EAK 16: [(AE)2 (AK)2 ]2 EAK 12: AEAK(AE)2 (AK)2 EAK 8: (AE)2 (AK)2

18

Salt (Li+ , Na+ , K+ , Cs+ )

NA

2

P11 -I: Ac-(Q)2 R(Q)5 E(Q)2 -NH2 P11 -II: Ac-(Q)2 RFQWQFE(Q)2 -NH2

25

pH

NA

3

KFE12: (FKFE)3 KIE12: (IKIE)3 KVE12: (VKVE)3 KFE8: (FKFE)2 KFE16: (FKFE)4

26

Ionic strength (1 mM NaCl), pH

NA

4

EAK 16-I: (AEAK)4 EAK 16-II: [(AE)2 (AK)2 ]2 EAK 16-IV: (AE)4 (AK)4

19

Ionic strength (1 mM NaCl)

NA

5

FEFK (FEFK)2 (FE)2 (FK)2

30

Enzyme (thermolysin), temperature (∼90 ◦ C to rt)

25 × 103

6

RADA16-I: Ac-(RADA)4 -NH2

20

pH

1541 ± 72

7

RADA16-I: Ac-(RADA)4 -NH2 RADA16 + PRG: Ac (RADA)4 GPRGDSGYRGDS-CONH2 RADA16 + PDS: Ac(RADA)4 GGPDSGR-CONH2

17

pH

1252 ± 62 −1.6 ± 3.5 214 ± 47

8

RADA + MMP: (RADA)3 PVGLIG(RADA)3 (RADA)2 PVGLIG(RADA)2 (RADA)4 PVGLIG(RADA)4

36

Enzymatic

NA

9

MAX1: (VK)4 VD PPT(KV)4 -NH2

37

pH, ionic strength (150 mM NaCl), heat

10 × 103

10

MAX8: (VK)4 VD PPTKVEV(KV)2 -NH2

39

Ionic strength (160 mM NaCl)

450

11

TSS1: (VK)4 VD PPT(KV)3 K D PP(KV)4 -NH2

42

Temperature

8500

NA, not available. a Backbone sequence in hydrophilic (italic) and hydrophobic (bold) order.

hydrophobic effects and unfavorable entropic effects is likely to explain these observations. Despite the influence of peptide length on gelation properties, it is now known that much shorter peptides of as little as two amino acids can also form β-sheet structures when further stabilized by aromatic interactions, as discussed in Section 7. In a different example, Hong et al.19 used variations of the EAK 16 peptide to demonstrate how the formation of nanostructures can be affected by the sequence (with identical overall amino acid composition) and the pH of the solution. When two modified versions, EAK 16IV and EAK 16-II, achieved through varying molecular hydrophobicity (Table 1, entry 4), were tested under neutral pH conditions, the former formed globular assemblies in comparison to fibrillar assemblies from the latter. With the change in pH, although the former displayed a transition from globular to fibrillar structures, the fibrillar structure of the latter remained unchanged.

As an alternative to rational elucidation of design rules, several researchers have started exploring dynamic combinatorial library (DCL) approaches to peptide discovery.27 DCLs are molecular component libraries that continuously exchange their components toward a thermodynamic equilibrium state.28, 29 In this regard, DCLs hold considerable promise as a tool for self-assembled materials discovery. The catalytic activity of enzymes can play a role in the formation of β-sheet-rich fibrillar hydrogels by dynamic exchange of shorter, nonassembling peptide sequences. In a recent example,30, 31 it was demonstrated how a nonspecific protease could be used to reversibly form self-assembling oligomers. Specifically, the tetrapeptide FEFK was exposed to thermolysin (Table 1, entry 5), which first partially hydrolyzed the tetrapeptide into dipeptides which, in turn, formed into longer peptide sequences. Of the hexa-, octa-, and decapeptides formed, octapeptide gels dominated the mixture, suggesting that they represent

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

6

Soft matter

the most stable assembled structures. Notably, octapeptidebased gels with elasticity moduli up to 25 kPa, which is considerably stiffer compared to octapeptide gels produced via conventional self-assembly methods, could be obtained using this method. Responsiveness to mechanical stress was investigated using RADA16-I20 (Table 1, entry 6). When subjected to sonication, the macroscopic, gel-phase structure was disrupted, while the supramolecular structure remained intact. On release of the stress, reassembly occurs due to the interaction of exposed hydrophobic ends with one another to allow the short fibers to fuse together to form longer fibers again. This self-healing and adaptive behavior is a useful property for cell culture studies, where gels may reform on cells movement. Closely related to β-sheets are β-hairpins that form through two β-strand structures linked by a turn sequence. This turn sequence ensures stable self-assembly, because it forces peptides stack in register with one another and thus maximizes interstrand interactions and onedimensional assembly. An extensively studied 20-residue peptide, named MAX1 adopts β-hairpin secondary structure via a pH-induced intramolecular folding event.32–34 MAX1 consists of alternating positively charged K and hydrophobic V residues arranged along either side of a D PP sequence (P induces the sequence to turn) followed by G or T. The cyclic structure of P and the conformationally flexible G/T residues together allow a complete reversal of direction of the α-carbon backbone. This turn-inducing pair holds to the two residues of the peptide strand via hydrogen bonding to form a tight hairpin.

4.3

Biological functionalization and cell culture

Cell adhesion ligands, which are generally short peptides, provide the biological cues necessary for cellular response. For example, RGD and PDSGR motifs (derived from the ECM proteins, fibronectin and laminin, respectively) were appended at the C-terminus (Figure 2C) (Table 1, entry 7).17 When compared to the unmodified RADA fibrous scaffold (Figure 2D), the peptide scaffolds functionalized with two-unit RGD binding sequence PRGDSGYRGDS and laminin cell adhesion motif PDSGR promoted significantly higher levels of proliferation and migration of human periodontal ligament fibroblast cells (Figure 2E). The cells’ ability to produce ECM protein type I and type III collagen, which are typical matrix proteins extracted by fibroblast, was also demonstrated. The incorporation of bioactive peptides improved the potential of these scaffolds to support cell culture in the absence of additional growth factors. Changes in cell behavior, associated with wound healing, cell differentiation, and so on, are dictated by dynamic

events, which often involve selective matrix degradation by cell-secreted enzymes. Specifically, matrix metalloproteinases (MMPs) play key roles in cell migration, and have been previously incorporated into chemical hydrogels.35 Zhang and Langer et al. demonstrated the incorporation of MMP-2-labile hexapeptide PVGLIG into an RADAbased scaffold36 (Table 1, entry 8). A number of structural/degradable motifs were tested (by varying the position of the MMP-2 substrate and the length of the RADA), and it was found that the sequence that includes centrally positioned MMP-2 linked to three RADA units on each side provides the desired biodegradability. β-Hairpin peptide hydrogel MAX1 has been studied for 2D and 3D cell culture. The initial attempts of injectable delivery of NIH3T3 cells37 and human red blood cells38 (Table 1, entry 9) onto MAX1 demonstrated that these gels support 2D cell culture. However, owing to slow gelation kinetics, nonhomogeneous cell distributions were observed when the cells were cultured in 3D. A modified version of MAX1, MAX8, was designed, which showed better hydrogelation kinetics through the replacement of K with E.39 The introduction of E decreased the net positive charge on the hydrophilic face of MAX1 resulting in rapid hydrogelation after addition of cell culture media. At low concentrations (0.5 wt%), MAX8 supported homogeneous distribution of cells within the hydrogel network (Table 1, entry 10). Using J774 mouse peritoneal macrophages, HainesButterick et al.40 studied the proinflammatory behavior and response patterns of cell culture on MAX1 and MAX8 hydrogels. The gels were found to be noncytotoxic, displaying healthy morphologies and showing reduced TNFα (tumor necrosis factor-α) secretion, showing a minimal inflammatory response. Using multistranded precursors,41 temperature-triggered injectable self-assembly systems (TSS1) were developed42 (Table 1, entry 11). In the presence of DMEM (Dulbecco’s modified Eagle’s medium) cell culture media, the peptide solution could undergo thermally induced intramolecular folding at around neutral pH and produced mechanically rigid, β-sheet-rich fibrillar network. Although self-assembly was hampered at temperatures below 20 ◦ C, at a higher temperature the desolvation of hydrophobic V side chains could overcome the electrostatic repulsion between the protonated K side chains. When tested with C3H10T1/2 mesenchymal stem cells, the TSS1 hydrogel surface was found to be supporting cell adhesion and cellular migration for up to the 24-h experimental period. The injectable potential achieved through shear recovery properties of the TSS1 hydrogel is particularly promising. Overall, it is clear that β-sheet self-assembly is a powerful design paradigm to develop hydrogels with desirable properties for 2D and 3D cell culture. Indeed, RADA peptides have been successfully commercialized as

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials Puramatrix , and are now used routinely as standard laboratory tools for cell culture. Researchers have demonstrated that gelation behavior, mechanical properties (ranging from 102 –104 Pa, i.e., covering a significant range of values found in biological tissues), and biochemical functionalities can be rationally incorporated into these systems. Overall, these developments point to a future where in vivo tissue engineering applications based on these peptides may become a clinical reality.

5 5.1

α-HELICES/COILED-COILS Design rules for coiled-coil systems: peptide sequence

The α-helix peptide structure consists of a spiral chain of amino acids interacting via hydrogen bonding between the backbone NH and CO groups along the peptide chain itself, and therefore is a local secondary structure. The structure was first discovered by Pauling, Corey, and Crick in the

Helix A

Ionic interactions

g

c

early 1950s.43, 44 The so-called coiled-coil structures are formed when multiple α-helices twist around each other. Each of the helical structures in a coiled-coil consists of hydrophobic residues appearing at repeated intervals of three or four residues apart with intervening polar residues corresponding to the structural repeat of 3.6 residues per αhelical turn. As is shown in Figure 3(A), for every two turns of a helix, there are seven amino acid residues that organize into a repeating pattern, labeled a through g (abcdefg). This heptad motif repeat, when viewed end-on, takes the shape of a helical wheel, and when at least two of these wheels come together they form interhelical structure, dictated by the pair-wise interactions between four key positions, a, d, e, and g. While the hydrophobic residues at positions a and d (commonly these are L and I) interact to form the hydrophobic core of the coiled-coil, the charged residues at e and g, are located on either side of the core and participate in electrostatic interhelical contacts (Figure 3A and B). The remainder of the amino acids is exposed to improve the solubility and helicity. Examples of dynamic selfassembly achieved through rational incorporation of amino acids within helical structures have since been reported.45, 46

Helix B

e′

b′

a′

d

7

f

f′ d′

a b

Hydrophobic interactions

e

c′

g′

2 µm

Ionic interactions

(A)

(C)

3.6 residues

f e

c a

b

d

a

g

b

b a Helix A

a′ b′

c′

d′ e′

g′ f′

a′ b′ Day 10

Helix B (B)

100 µm

(D)

Figure 3 Design principles for the α-helix/coiled-coil assembly. (A) Cartoon showing the interactions between two α-helices of a coiled-coil. The helical wheel of the coiled-coil heptad repeat shows the positions a –g in a cross-section. The red (helix A) and blue (helix B) represent the right- and left-handed helixes, respectively. Arrows indicate interactions: hydrophobic (between a and d) and ionic (between e and g) (B) Side view of each α-helix with seven residues (a –g) in two turns. Each turn of the helix contains 3.6 amino acid residues. (C) Two-stranded coiled-coil model and electron microscopy image showing unidirectional fibrous network. Model shows pairs of D (red) and single R (blue) residues tracking around the surface of the coiled-coil. (D) (a) Cryogenic scanning electron microscopy image of coiled-coil hydrogel. (b) Phase-contrast microscopy image of differentiating rat adrenal pheochromocytoma (PC12) cells in hSAFAAA−W hydrogel at day 10. (Adapted from Ref. 47.  Nature Publishing Group, 2009 and from Ref. 48.  Wiley-VCH, 2006.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

8

Soft matter

Solvent-exposed amino acids b, c, and f can dictate fiber thickness and flexibility allowing for the formation of gelphase materials47 (see below).

5.2

Gels based on α-helices and their exploitation in cell culture

A number of examples of two component α-helical systems (i.e., involving two charge complimentary sequences) have been introduced by Woolfson group.48, 49 Two heterodimeric parallel coiled-coils, self-assembling fiber systems SAF-p1 and SAF-p2 were formed from a basic four-heptad framework (IXX LKXK)2 (IXX LEXE)2 (where X positions are variable) provided positively charged Ntermini and negatively charged C-termini to promote staggered assembly with “sticky ends.” The “sticky ends” encourage the formation of coiled-coil interfaces resulting in unidirectional self-assembly and fiber formation. Modifications in relation to fiber morphology achieved through the incorporation of arginine residues to allow salt bridges to form with aspartic acid residues at the interface between coiled-coils (Figure 3C) resulted in fibrous assembly with high degree of order.48 Most α-helix-based systems formed fibers with high aspect ratios; however, only recently the first gel-forming samples were described. The interactions of fibers with solvent, as relevant for hydrogelation, are dictated by the peripheral regions (b, c, and f positions).46 Four α-helical coiled-coil peptides based on a primary sequence of I (a), L (d), and E (e and g positions) with charged residues K, Q, S, and Y alternatively at b, c, and f positions formed hydrogels at low pH and relatively high peptide concentrations. Modifications at the peripheral positions of the coiled-coil restrained lateral aggregation of fibers resulting in more uniform fibrous structures. The solvent-exposed residues of a sticky end SAF were modified with A and Q residues at b, c, and f positions, and hSAFs maintained the original SAF design at a, d, e, and g positions. These modifications were effective in promoting hydrogelation via hydrophobic interactions between fibrils and through stronger hydrogen bonds. Both sequences tested hSAFAAA and hSAFQQQ formed hydrogels of interconnected network of thin fibrils (Figure 3D(a)).47 In terms of stability, while the latter melted on warming, the former strengthened on warming. A modified version of this, hSAFAAA−w , (modification was achieved through replacing A residues at f position with the more hydrophobic W) was found to form a similar but more stable hydrogel near neutral pH at room temperature. This hydrogel supported the growth and differentiation of rat adrenal pheochromocytoma (PC12) cells (Figure 3D(b)), and the differentiation patterns were found to be comparable with those

observed on a Matrigel substrate, which is commonly used for cell culture experiments. As this system promoted gel formation only on mixing the two peptides, hSAFs have the additional benefits of controlled hydrogelation at constant, physiological conditions. Functionalization by incorporating cell binding ligands onto the coiled–coiled scaffolds have been proposed and may be achieved by using noncovalent approaches using peptide tags.50

6 6.1

ALIPHATIC PEPTIDE AMPHIPHILES Design rules for aliphatic peptide amphiphiles: peptide sequence

PAs are composed of polar peptide head groups modified with aliphatic tails. Early examples developed by Stupp et al. typically consisted of four regions (Figure 4A). Region 1 consists of hydrophobic block (an alkyl chain), region 2 consists of short peptide sequence capable of forming intermolecular hydrogen bonding, the charged amino acids in region 3 enhance the solubility of the PAs system in water, and region 4 consists of bioactive epitopes ultimately giving rise to structures that display high density of peptide ligands at their surface (Figure 4B,C). A functional ECM mimicking system was developed to control mineralization, by mimicking some key features of collagen fibers (Table 2, entry 1).51 In this case, four consecutive C amino acid residues were incorporated to enable polymerization via disulfide bonds, thereby locking the structure into place post-assembly. The second part consists of three G residues that act as a linker to provide the necessary hydrophilic end group flexibility. The single phosphorylated S residue promotes mineralization. The final addition to the peptide sequence is the cell adhesion ligand, Arg-Gly-Asp (RGD), specifically incorporated to provide the biological properties to the PA system. A number of bioactive PAs have since been developed based on a similar modular design approach, some of which are discussed below.

6.2

Controlling self-assembly

The scope for controlling the self-assembly characteristics of PAs was investigated by varying amino acid sequence and alkyl tails.52 The variations imposed on the chemical structure resulted in nanofibrous matrices of varying morphology, surface chemistry, and potential bioactivity. Investigations into the effect of hydrophobic tail length on PA self-assembly involved testing five different tail length molecules, containing 0, 6, 10, 16, and 22 carbon atoms while maintaining the same peptide sequence

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials O

H N

R2

O

H N

n OH

nN

n

O

R1

H

O

R3

(a) V

Region 1: hydrophobic tail

Region 2: structural peptide segment

(A)

Region 3: charged solubilizing group

9

Region 4: bioactive ligand

K I

A V

(b)

300 nm

Z axis

a (C)

b (B)

c

(a)

(b)

20 µm

(c)

5 µm

(D)

Figure 4 Systems based on self-assembled PA. (A) Chemical structure showing different regions of the PA molecule drawn using ChemSketch. (B) Cross-sections of PA nanofiber formed from PA molecule. (C) (a) Model showing individual PA molecules assembled into a bundle of fibers interwoven to produce IKVAV PA. (b) Scanning electron micrograph of IKVAV PA nanofiber network formed by adding cell culture media to the PA aqueous solutions. (D) (a) Bone-marrow mononuclear cells grown on RGDS PA. (b) Neural progenitor cells encapsulated in an IKVAV PA. (c) TEM image of a cell entrapped in PA nanofibers (Adapted from Ref. 53.  American Chemical Society, 2006, from Ref. 54.  Wiley Periodicals, Inc, 2010 and from Refs. 55 and 56.  Elsevier, 2005 and 2010.)

(C)4 (G)3 S(PO4 )RGD (Table 2, entry 2). On changing the pH of aqueous PA solutions, only the latter three formed self-supporting gels, demonstrating the importance of the hydrophobic effect in these systems. More recent variations in the PA design suggests the possibility of controlling the shape of assembled nanostructures through variation of the central peptide segment (β-sheet region 2) of the PA molecules. First, a series of N-methylated PA molecules were tested to study how the amino acids nearer to the core of the PA (as opposed to those at the periphery) dictate the architecture of the final nanostructure.53 Molecular variations were achieved through changing the position of the N-methyl glycine derivative and alanine mutant of the tested series (Table 2, entry 3). The importance of the hydrogen bonding between the inner amino acids was confirmed after observing spherical micelles on disturbing the hydrogenbonding ability of the first four amino acids after the alkyl tail. Hydrogen bond formation between the outer amino acid residues, on the other hand, had little or no impact on the stability of the final assembly. Therefore,

the residues further from the aliphatic chain are more exposed and suitable for attachment of epitopes rather than making a structural contribution to stable self-assembly. More recently, noncylindrical architectures were obtained through introducing tetrapeptide sequences with alternating hydrophobic (V) and negatively charged hydrophilic residues (E) in the central peptide region 2.57 This structural modification resulted in disturbed hydrogen bonds that supported the interfacial curvature giving rise to lateral growth to generate flat tape-like structures (Table 3, entry 4). Control of mechanical properties of the amphiphilic PAs over more than one order of magnitude is possible through varying the number and position of V and A residues in the β-sheet region of the PA (Table 2, entry 5).58 Owing to the possibility of incorporating different peptide epitopes into the PA structure without disturbing its cylindrical geometry, PAs are well suited to biomaterials applications. Systems that incorporated bioactive, cell binding peptides, most notably RGDS51, 59, 60 and IKVAV,61, 62 for use in cell culture and regenerative

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

10

Soft matter

Table 2 S. No.

Designer self-assembling systems based on peptide amphiphiles molecules. Sequencea

System R1

R2

R3

References

Stimulus

R4

1

CH3 (CH2 )14 CO–(C4 G3 )S(PO4 )RGD

C16

(C4 G3 )

S(PO4 )

RGD

51

pH

2

H(C4 G3 )S(PO4 )RGD CH3 (CH2 )4 CO–(C4 G3 )S(PO4 )RGD CH3 (CH2 )8 CO–(C4 G3 )S(PO4 )RGD CH3 (CH2 )14 CO–(C4 G3 )S(PO4 )RGD CH3 (CH2 )20 CO–S(PO4 )RGD

H C6 C10 C16 C22

(C4 G3 ) (C4 G3 ) (C4 G3 ) (C4 G3 ) (C4 G3 )

S(PO4 ) S(PO4 ) S(PO4 ) S(PO4 ) S(PO4 )

RGD RGD RGD RGD RGD

52

pH

3

CH3 (CH2 )14 CO–G7 ERGDS CH3 (CH2 )14 CO–G6 AERGDS CH3 (CH2 )14 CO–VEVE

C16 C16 C16

E E

RGDS RGDS

53 57

pH, multivalent cations pH

5

CH3 (CH2 )14 CO–V3 A3 E3 CH3 (CH2 )14 CO–V2 A2 E3 CH3 (CH2 )14 CO–V4 A4 E3 CH3 (CH2 )14 CO–V2 A4 E3 CH3 (CH2 )14 CO–V4 A2 E3 CH3 (CH2 )14 CO–A3 V3 E3

C16 C16 C16 C16 C16 C16

(G7 ) (G6 A) VEVE (V3 A3 ) (V2 A2 ) (V4 A4 ) (V2 A4 ) (V4 A2 ) (A3 V3 )

E3 E3 E3 E3 E3 E3

58

pH, osmolarity

6

CH3 (CH2 )14 CO–C4 G3 S(P)RGD–COOH CH3 (CH2 )14 CO–A4 G3 S(P)RGD–COOH CH3 (CH2 )14 CO–A4 G3 S(P)KGE–COOH CH3 (CH2 )14 CO–C4 G3 SRGD–COOH CH3 (CH2 )14 CO–A3 G2 EQS–COOH CH3 (CH2 )14 CO–A4 G3 ERGDS–COOH CH3 (CH2 )14 CO–C4 G3 EIKVAV–COOH CH3 (CH2 )14 CO–C4 G3 KIKVAV–NH2

C16 C16 C16 C16 C16 C16 C16 C16

(C4 G3 ) (A4 G3 ) (A4 G3 ) (C4 G3 ) (A3 G2 ) (A4 G3 ) (C4 G3 ) (C4 G3 )

S(PO4 ) S(PO4 ) S(PO4 ) S E E E K

RGD RGD RGD RGD QS RGDS IKVAV IKVAV

55

Polyvalent and monovalent ions

7

CH3 (CH2 )14 CO–A4 G3 LRKKLGKA

8

CH3 (CH2 )14 CO–V3 A3 K3 RGDS CH3 (CH2 )14 CO–V3 A3 K3 CH3 (CH2 )14 CO–V3 A3 E3

C16 C16 C16 C16

(A4 G3 ) V3 A3 V3 A3 V3 A3

K3 K3 K3

LRKKLGKA RGDS

63 56

pH pH

4

a R1,

R2, R3, and R4 represent the four regions in a typical PA model as explained in Figure 4(a).

medicine applications have been described, with some examples discussed in the next section.

6.3

Biological functionalization of PAs and applications in cell culture

Beniash et al., demonstrated the formation of a 3D nanofiber network of a library of PAs consisting of RGD, IKVAV, and KGE that undergo self-assembly at relatively low concentrations (∼1 wt%) (Table 2, entry 6).55 The encapsulation of cells in the nanofibrous network was triggered by polyvalent metal ions within a cell culture media under physiological conditions. The viability of cells entrapped within the PA matrix was confirmed, which revealed a degree of internalization of the PA nanofibers by the entrapped cells (Table 2, entry 7).63 Rajangam et al., provided an example that incorporated a heparin-binding peptide motif into a PA. Heparin is known to have the ability to bind angiogenic growth factors of relevance to blood vessel formation. The ability to promote the formation of

blood vessels was confirmed, which was thought to be made possible through heparin molecules binding to the surface of the PA fibers via electrostatic interactions. RGDS-incorporated PAs were tested in vivo as a therapeutic cell delivery system. In this work, RGDS containing PAs were coassembled with negatively and positively charged non-bioactive PAs (termed diluents) (Table 2, entry 8).56 When cultured in vitro with bone-marrow mononuclear progenitor cells, the binary PA mixed to the ratio of 10 wt% RGDS containing molecules to 90 wt% negatively charged diluents molecules promoted optimal cell recognition and adhesion (Figure 4D(a)). Natural ECM is a highly hierarchical structure with features that range in size from the molecular- to the micron-scale. In an effort to capture this entire size range, self-assembly was combined with top-down fabrication to design topographical patterns with microscale and nanoscale features. Specifically, RGD functionalized PAs were flow-aligned and micropatterned using moulds, and UV polymerization of these materials gave rise to highly controlled stable gels with microscale topographical

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials

can contribute to the energy required for self-assembly and is thought to be in part responsible for the formation of amyloid deposits.67 The first gelating aromatic peptide amphiphilic system was introduced by Vegners and coworkers in the mid1990s (Table 3, entry 1).68 Over a decade later, cell culture applications of these peptide derivatives are now gaining momentum because they are simple, cost-effective, and highly versatile.69–71 Examples reported to date involve chemical coupling of a variety of aromatic groups, including from phenyl, naphthyl (Nap), pyrene, 9-fluorenyl methoxycarbonyl (Fmoc) (Figure 5A) to short peptides. It turns out that both the nature of the aromatic ligand and the peptide sequence and length dictated self-assembly kinetics and nanoscale morphologies.71–74 A small library of Fmoc/Nap dipeptides made up of combinations of the amino acids S, G, A, L, K, V, and F encompassing a number of hydrophobicities has been tested for gelation ability. In these samples, the gelation potential and the pH value at which gelation took place varied per amino acid sequence used. Among the ones tested, Fmoc-GF and Nap-GV, -GL peptides did not form gels under any of the conditions, while the remainder formed gels consisting of over 99.5% water (Table 3, entries 2 and 3).71, 73

features.64 These materials could be used to provide directional control of skeletal stem cells. Stem cells underwent morphological changes, alignment, and differentiation when cultured on nanofibrous substrates that had been patterned into micron-scale topographies.

7

AROMATIC PEPTIDE AMPHIPHILES

7.1

Design rules for aromatic peptide amphiphiles: peptide sequence and controlling self-assembly

Aromatic PAs, composed of aromatic group coupled to a peptide, enable self-assembly from much smaller peptides compared to other systems that normally required at least eight amino acids (and often many more) in each peptide chain. This is possible through π -stacking of the aromatic rings in combination with hydrogen bonding from peptide portion. In supramolecular chemistry and biochemistry, the central role of π -stacking interactions in the formation of structure is of course well known.65, 66 These are weakly attractive interactions between planar aromatic rings. In addition to giving order and directionality, this π-stacking Table 3

11

Designer self-assembling systems based on aromatic peptide amphiphiles molecules.

S. No.

System∗

References

Stimulus

Stiffness G
(Pa)

1

Fmoc-LD, AD, ID, LA, LE, LK No gel formed for Fmoc-LA, LE, LK

68

Temperature

NA

2

Fmoc-GG,AA,FG,GF,FF,FF/K a , FF/GG a No gel formed for Fmoc-GF

71

pH ∼7

NA

3

Nap-GG, GA, GS, GV, GL No gel formed for Nap-GV, GL

73

pH ∼2

400 (Nap-GG, GS), 5 × 103 (Nap-GA)

4

Fmoc-L3

75

Enzyme (subtilisin)

NA

5

Pyrene(P), Fluorene(F), Naphthalene(N), GAGAS, GVPVP, VPGVG, VTEEI, VYGGG, YGFGG. VPGVG did not form gel under any protecting group tested

79

pH 2–8, temperature—55–90 ◦ C

∼8–17 945b

6

Fmoc-GA, GG, AA, AG, AV, VG, LA, F A, LG, FG, FV, LF, FF

74, 80

pH 8–9 (GdL)

2900–28 × 104

7

Fmoc-Y(p)-OH

81

Enzyme (alkaline phosphatase) (0.1–10 units µl−1 )

9 × 103 –34 × 103c

8

Fmoc-FF, FF/K a , FF/D a , FF/S a

83

pH ∼7

502–21 × 103d

9

Fmoc-FF /RGD

76

pH ∼7

800–10 × 103e

10

Fmoc-FRGD, RGDF, FG, FF, Fmoc-2-naphthyl

84

Dissolution from DMSO

NA

∗ Aromatic

groups in bold and peptide group in italic. 50 : 50 (mol mol−1 ) ratios were used. b Depending on the peptide sequence as well as aromatic group. c Depending on the enzyme concentration. d Depending on the peptide mixtures. e Depending on the FF/RGD ratio. a

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

12

Soft matter Aromatic groups

Phen(yl)

Napth(yl)

Aromatic group Fluoren(yl) (A)

Peptide group

Pyrene(yl)

p - Interlocked b - sheets

(B)

O O

N H

O

H N

OH

O

Fmoc-FF

(D) H2N

NH O O

(C)

N H

H N O

100 nm

NH OH O N H

O OH O

Fmoc-RGD

3 nm

3 nm (E)

50 micron

Figure 5 Design principles for aromatic peptide amphiphile systems. (A) Commonly used aromatic groups in aromatic peptide amphiphiles. (B) A model of the Fmoc-peptide self-assembly arranged in a π-stacked antiparallel β-sheet pattern. (C) The chemical structures of the hydrogel building blocks: Fmoc-FF and Fmoc-RGD. The proposed supramolecular model demonstrates the formation of the 3-nm fibrils. (D) TEM image of RGD contain hydrogels. (E) Fluorescence microscope image of human adult dermal fibroblasts grown on RGD hydrogel. Green fluorescence shows actin fibers and blue fluorescence shows the nucleus. (Adapted from Ref. 75.  Royal Society of Chemistry, 2010 and from Ref. 76.  Elsevier, 2009.)

Molecular architecture of Fmoc-FF, which self-assembles at neutral pH, was studied in detail77 using a range of spectroscopic techniques. The supramolecular structure was composed of antiparallel β-sheet arranged peptides, which positioned the Fmoc fluorenyl rings on alternating sides of the sheet, allowing π -stacking between adjacent β-sheets (Figure 5B). This π-interlocked β-sheet arrangement or π−β structure gives rise to an extended π -stacked aggregate along the long axis of the fiber. For Fmoc-FF, peptides hydrogen bonded to one another in an organization where the phenyl rings placed above and below each pair of fluorenyl groups giving rise to a twist in each sheet structure. Four sheets then twist together forming a cylindrical structure with 28 monomers per turn with each turn ˚ and a hole in the center of having a width of ∼30 A ˚ 7 A diameter. In this Fmoc-FF, large pKa shifts (of up to 6 pH units) are observed on self-assembly, which provides strong evidence that the polar group is indeed in a

highly hydrophobic environment.78 More recently, enzymetriggered self-assembly of Fmoc-tri-leucine (Fmoc-L3 ) was demonstrated to also follow this π –β self-assembly motif, resulting in tubular structures which display significant charge transport through the extended π-stacked aggregates.75 This nanotubular network formed through selfassembly may in future prove to be a suitable platform to interface biological components with electronics (Table 3, entry 4). Aromatic interactions as favorable forces for hydrogel formation was further elaborated using a library of pentapeptides79 (Table 3, entry 5). On introduction of aromatic moieties Pyr, Fmoc, and Nap to C- and N-termini, a range of pentapeptides that failed to self-assemble on their own, formed supramolecular hydrogels. By comparing the molecular interactions to the hydrogel with its pregelation solution, it was revealed that the aromatic–aromatic interaction from the pyrene, fluorine, and naphthalene rings

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials played important roles. Of the ones that are tested, peptide sequence consisting of VTEEI with pyrene gave rise to very high elasticity values, as high as 17.9 kPa. Recently, Adams and coworkers introduced a new method to produce more homogeneous hydrogels produced by pH change.74, 80 Conventional pH-triggered gelation usually involves gradual addition of acid while stirring or vortexing. In this case, local gelation occurs instantly, and as a result achieving a uniform pH throughout the gel sample is challenging. In Adams’ system, gelation was triggered by hydrolysis of glucono-δ-lactone (GdL), which results in a gradual lowering of pH. As GdL hydrolysis is slower than the rate of dissolution of the peptide derivatives, a uniform pH change is obtained throughout the sample resulting in a homogeneous gelation (Table 3, entry 6). As discussed in a previous section, gelation may also be controlled by the enzymatic conversion of nonassembling precursors to self-assembly building blocks. This approach has been used extensively to control self-assembly for aromatic PAs,7, 81 and so is not discussed here in detail. In the context of cell culture, it is worth mentioning that the gelation kinetics can be controlled simply by changing the concentration of the enzyme. Differences in the gelation time, supramolecular arrangement, and, importantly, mechanical properties have been observed on chemically identical gels (Table 3, entry 7).

7.2

Biological functionalization and cell culture

The first demonstration of cell culture on gels from aromatic PAs was undertaken in our lab71 and simultaneously in Gazit’s group.72 Of a library of aromatic short PAs tested for hydrogelation, Fmoc-FF (either on its own or in combination with Fmoc-GG or Fmoc-K) was found to form stable hydrogels at physiological conditions. Live/dead staining and metabolic assays confirmed the hydrogels ability to support cell viability and retention of phenotype of bovine chondrocytes in both 2D and 3D culture. Liebmann et al., demonstrated the formation of gels by dilution from sulfoxide (DMSO, dimethyl sulfoxide)82 to circumvent potential biocompatibility issues with 1,1,1,3,3,3-hexafluoro-2-propanol used by Gazit.72 It was demonstrated that cell density could be controlled and assessment of cell behavior under controlled conditions was possible. In order to investigate compatibility with a range of cell types, simple chemical functionalities were introduced into Fmoc-FF gels to provide gels with tunable chemical and mechanical properties for in vitro cell culture. A series of hydrogel compositions consisting of combinations of FmocFF and n-protected Fmoc amino acids, K, D, and S were

13

studied (50 : 50 M ratio), thereby introducing the functional groups NH2 , COOH, and OH. Oscillatory rheology results show that all four hydrogels have mechanical profiles of soft viscoelastic materials with elastic moduli dependent on the chemical composition, ranging from 502 Pa (Fmoc-FF/D) to 21.2 kPa (Fmoc-FF). Different cell types were used: human dermal fibroblasts, bovine chondrocytes, and mouse Swiss Albino Embryo fibroblast cells (3T3). Although all gels supported the viability of bovine chondrocytes, Fmoc-FF/S hydrogel was the only gel type that supported viability for all three cell types tested. Cell response was found to relate to the chemical and mechanical properties of the resulted hydrogel supporting the hypothesis that introduction of chemical functionality into Fmoc-peptide scaffolds may provide gels with tunable chemical and mechanical properties for in vitro cell culture (Table 3, entry 8).83 Aromatic PAs with the bioactive RGD ligand could be produced by simply doping an Fmoc-FF gel with varying concentrations of Fmoc-RGD building blocks. These bioactive materials were then used in 3D cell culture to promote the cell adhesion behavior of a fibrous hydrogel76 (Figure 5C) (Table 3, entry 9). The resulted highly hydrated nanofibrous (Figure 5D) hydrogel network assumed a structure where the bioactive ligand, RGD, positioned in tunable densities on the fiber surface. The human adult dermal fibroblast cells responded to the hydrogel positively as demonstrated by homogeneity of cell distribution, high cell viability rates, 3D cell adhesion, and the patterns of cell migration commonly associated with anchorage-dependent cells (Figure 5E). The extent of cell spreading was influenced by the Fmoc-RGD concentration within the hydrogel. The results supporting the 3D cell culture applications from this peptide derivative (rather than 2D culture) open up opportunities for cell therapy and model systems to study fundamental cell biology. Following on from previous efforts to develop selfassembly from short peptide derivatives, Orbach et al.84 extended their library of Fmoc-dipeptides by inclusion of a range of natural and synthetic amino acids with an aromatic nature, some of which including the bioactive RGD sequence (Table 3, entry 10). Nanostructured gels were observed for most of the materials tested. Of these structures, 3D cell culture experiments were carried out for three samples, Fmoc-FRGD, Fmoc-RGDF, and Fmoc-2-naphthyl. The results suggest that while the RGD incorporated structures showed a decrease in cell numbers over time, Fmoc-2-naphthyl showed a moderate increase with some stability in the cell counts up to seven days in culture. Future research to fully appreciate the cell culture possibilities in these systems was proposed.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

14

Soft matter

8

CONCLUSIONS

Cell fate is dictated by a triad of cues, which are topographical, mechanical, and chemical in nature. Indeed, natural cell scaffolds vary in these characteristics across tissue types. This variation is a result of differences in supramolecular chemistry: biomolecules assemble into fibers, which entangle, resulting in gel-phase materials with characteristic stiffness. Peptide-based nanomaterials produced by molecular self-assembly are well suited to mimic these systems as they can provide a structural and functional interface with biology as peptides are nature’s expression language. A number of approaches have been developed over the past two decades, and it is now possible to design materials that control and dictate the response of cells. It is becoming increasingly clear that supramolecular organization and functionality is dictated not only by chemical nature of the building blocks but also by the self-assembly conditions. For example, materials with varying network properties have been produced from identical building blocks by simply varying the assembly route and exploiting the arrested dynamics of gel-phase materials (i.e., kinetic trapping). This latter approach is especially attractive as it allows for the production of gels with the same chemical composition but different mechanical properties, allowing disentanglement of chemical and mechanical stimuli on cell behavior and differentiation. For biomedical applications, reproducibility and manufacturing costs have to be considered. Minimalist approaches involving very simple peptides (e.g., as exemplified by aromatic PAs) can be produced at low cost and will have lower regulatory barriers. The next frontier in this area probably relates to the spatial and dynamic control of mechanical, chemical, and topographical properties within a single material, which will allow more complex cell behaviors to be supported, toward the complexity of functional organs. Overall, it is clear that peptide-based materials will continue to play an important role as cell instructive scaffolds in nanomedicine.

7. A. R. Hirst, S. Roy, M. Arora, et al., Nat. Chem., 2010, 2,, 1089–1094. 8. D. L. Nelson and M. M. Cox, Principles of Biochemistry, 4th edn, 2005. chapter 3–4. 9. H. R. Horton, L. A. Moran, R. S. Ochs, et al., Principles of Biochemistry, 3rd edn, 2002. chapter 2–3. 10. R. B. Merrifield, J. Am. Chem. Soc., 1963, 85, 2149–2154. 11. S. Kyle, A. Aggeli, E. Ingham, and M. J. McPherson, Trends Biotechnol., 2009, 27, 423–433. 12. J. A. Fallas, L. E. R. O’Leary, and J. D. Hartgerink, Chem. Soc. Rev., 2010, 39, 3510–3527. 13. H. J. Lee, J. Lee, T. Chansakul, et al., Biomaterials, 2006, 27, 5268–5276. 14. J. F. Almine, D. V. Bax, S. M. Mithieux, et al., Chem. Soc. Rev., 2010, 39, 3371–3379. 15. H. G. B¨orner and H. Schlaad, Soft Matter, 2007, 3, 394–408. 16. L. Pauling and R. B. Corey, Proc. Natl. Acad. Sci. U.S.A., 1951, 37, 251–256. 17. Y. Kumada and S. Zhang, PLoS ONE, 2010, 5, e10305, 1–7. 18. S. Zhang, T. Holmes, C. Lockshin, and A. Rich, Proc. Natl. Acad. Sci. U.S.A., 1993, 90, 3334–3338. 19. Y. Hong, M. D. Pritzker, R. L. Legge, and P. Chen, Colloids Surf., 2005, 46, 152–161. 20. H. Yokoi, T. Kinoshita, and S. Zhang, PNAS, 2005, 102, 8414–8419. 21. F. Gelain, D. Bottai, A. Vescovi, and S. Zhang, PLoS ONE, 2006, 1, e119 1–11. 22. R. P. W. Davies, A. Aggeli, A. J. Beevers, et al., Supramol. Chem., 2006, 18, 435–443. 23. S. Zhang, Biotechnol. Adv., 2002, 20, 321–339. 24. A. Aggeli, M. Bell, N. Boden, et al., Nature, 1997, 386, 259–262. 25. A. Aggeli, I. A. Nyrkova, M. Bell, et al., PNAS, 2001, 98, 11857–11862. 26. M. R. Caplan, E. M. Schwartzfarb, S. Zhang, et al., Biomaterials, 2002, 23, 219–227. 27. J. W. Sadownik and R. V. Ulijn, Curr. Opin. Biotechnol., 2010, 21, 401–411. 28. R. F. Ludlow and S. Otto, Chem. Soc. Rev., 2008, 37, 101–108.

REFERENCES 1. J. H. Van Esch, Langmuir, 2009, 25, 8392–8394. 2. J. E. Phillips, T. A. Petrie, F. P. Creighton, and A. J. Garc´ıa, Acta Biomater., 2010, 6, 12–20. 3. A. J. Engler, S. Sen, H. L. Sweeney, and D. E. Discher, Cell, 2006, 126, 677–689.

29. J.-M. Lehn, Chem. Soc. Rev., 2007, 36, 151–160. 30. J. Guilbaud, E. Vey, S. Boothroyd, et al., Langmuir, 2010, 26, 11297–11303. 31. A. Saiani, A. Mohammed, H. Frielinghaus, et al., Soft Matter, 2009, 5, 193–202. 32. J. P. Schneider, D. J. Pochan, B. O. Ozbas, et al., J. Am. Chem. Soc., 2002, 124, 15030–15037.

4. M. J. Dalby, Nanomedicine, 2009, 4, 247–248.

33. J. K. Kertsingera, L. A. Hainesa, B. Ozbasb, et al., Biomaterials, 2005, 26, 5177–5186.

5. F. Gelain, A. Horii, and S. Zhang, Macromol. Biosci., 2007, 7, 544–551.

34. L. A. Haines, K. Rajagopal, B. Ozbas, et al., J. Am. Chem. Soc., 2005, 127, 17025–17029.

6. C. Yan and D. J. Pochan, Chem. Soc. Rev., 2010, 39, 3528–3540.

35. J. Patterson and J. A. Hubbell, Biomaterials, 2010, 31, 7836–7845.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

Designing peptide-based supramolecular biomaterials

15

36. Y. Chau, Y. Luo, A. C. Y. Cheung, et al., Biomaterials, 2008, 29, 1713–1719.

60. T. D. Sargeant, M. O. Guler, S. M. Oppenheimer, et al., Biomaterials, 2008, 29, 161–171.

37. J. K. Kretsinger, L. A. Haines, B. Ozbas, et al., Biomaterials, 2005, 26, 5177–5186.

61. G. A. Silva, C. Czeisler, K. L. Niece, et al., Science, 2004, 303, 1352–1355.

38. D. A. Salick, J. K. Kretsinger, D. J. Pochan, J. P. Schneider, J. Am. Chem. Soc., 2007, 14793–14799.

62. V. M. Tysseling-Mattiace, V. Sahni, K. L. Niece, et al., J. Neurosci., 2008, 28, 3814–3823.

and 129,

39. L. Haines-Butterick, K. Rajagopal, M. Branco, et al., PNAS, 2007, 104, 7791–7796. 40. L. Haines-Butterick, D. A. Salick, D. J. Pochan, J. P. Schneider, Biomaterials, 2008, 29, 4164–4169.

and

41. H. Rapaport, G. Moller, C. M. Knobler, et al., J. Am. Chem. Soc., 2002, 124, 9342–9343. 42. R. V. Rughani, D. A. Salick, M. S. Lamm, et al., Biomacromolecules, 2009, 10, 1295–1304. 43. L. Pauling and R. B. Corey, Nature, 1953, 171, 59–61. 44. F. H. C. Crick, Acta. Crystallogr., 1953, 6, 685–689.

63. K. Rajangam, H. A. Behanna, M. J. Hui, et al., Nano Lett., 2006, 6, 2086–2090. 64. A. Mata, L. Hsu, R. Capito, et al., Soft Matter, 2009, 5, 1228–1236. 65. C. G. Claessens and J. F. Stoddart, J. Phys. Org. Chem., 1997, 10, 254–272. 66. G. B. McGaughey, M. Gagn´e, and A. K. Rapp´e, J. Biol. Chem., 1998, 273, 15458–15463. 67. E. Gazit, FASEB, 2002, 16, 77–83. 68. R. Vegners, I. Shestakova, I. Kalvinsh, et al., J. Pept. Sci., 1995, 1, 371–378.

45. M. G. Ryadnov and D. N. Woolfson, Nat. Mater., 2003, 2, 329–332.

69. Z. Yang and B. Xu, Chem. Commun., 2004, 21, 2424–2425.

46. H. Dong and J. D. Hartgerink, Biomacromolecules, 2006, 7, 691–695.

71. V. Jayawarna, M. Ali, T. A. Jowitt, et al., Adv. Mater., 2006, 18, 611–614.

47. E. F. Banwell, E. S. Abelardo, D. J. Adams, et al., Nat. Mater., 2009, 8, 596–600.

72. A. Mahler, M. Reches, S. Cohen, and E. Gazit, Adv. Mater., 2006, 18, 1365–1370.

48. A. M. Smith, E. F. Banwell, W. R. Edwards, et al., Adv. Funct. Mater., 2006, 16, 1022–1030.

73. Z. Yang, G. Liang, M. Ma, et al., J. Mater. Chem., 2007, 17, 850–854.

49. M. J. Pandya, E. Cerasoli, A. Joseph, et al., J. Am. Chem. Soc., 2004, 126, 17016–17024.

74. D. J. Adams, L. M. Mullen, M. Berta, et al., Soft Matter, 2010, 6, 1971–1980.

50. D. N. Woolfson and Z. N. Mahmoud, Chem. Soc. Rev., 2010, 39, 3464–3479.

75. H. Xu, A. K. Das, M. Horie, et al., Nanoscale, 2010, 2, 960–966.

51. J. D. Hartgerink, E. Beniash, and S. I. Stupp, Science, 2001, 294, 1684–1688.

76. M. Zhou, A. M. Smith, A. K. Das, et al., Biomaterials, 2009, 30, 2523–2530.

52. J. D. Hartgerink, E. Beniash, and S. I. Stupp, PNAS, 2002, 99, 5133–5138.

77. A. M. Smith, R. J. Williams, C. Tang, et al., Adv. Mater., 2008, 20, 37–41.

53. S. E. Paramonov, H. W. Jun, and J. D. Hartgerink, J. Am. Chem. Soc., 2006, 128, 7291–7298.

78. C. Tang, A. M. Smith, R. F. Collins, et al., Langmuir, 2009, 25, 9447–9453.

54. H. Cui, M. J. Webber, and S. I. Stupp, Pept. Sci., 2010, 94, 1–18.

79. M. Ma, Y. Kuang, Y. Gao, et al., J. Am. Chem. Soc., 2010, 132, 2719–2728.

55. E. Beniash, J. D. Hartgerink, H. Storrie, et al., Acta Biomater., 2005, 1, 387–397.

80. D. J. Adams, M. F. Butler, W. J. Frith, et al., Soft Matter, 2009, 5, 1856–1862.

56. M. J. Webber, J. Tongers, Biomater., 2010, 6, 3–11.

Acta

81. K. Thornton, A. M. Smith, C. L. R. Merry, and R. V. Ulijn, Biochem. Soc. Trans., 2009, 37, 660–664.

57. H. Cui, T. Muraoka, A. G. Cheetham, and S. I. Stupp, Nano Lett., 2009, 9, 945–951.

82. T. Liebmann, S. Rydholm, V. Akpe, and H. Brismar, BMC Biotechnol., 2007, 7, 88–99.

58. E. T. Pashuck, H. Cui, and S. I. Stupp, J. Am. Chem. Soc., 2010, 132, 6041–6046.

83. V. Jayawarna, S. M. Richardson, A. R. Hirst, et al., Acta Biomater., 2009, 5, 934–943.

59. H. Storrie, M. O. Guler, S. N. Abu-Amara, et al., Biomaterials, 2007, 28, 4608–4618.

84. R. Orbach, L. Adler-Abramovich, S. Zigerson, Biomacromolecules, 2009, 10, 2646–2651.

M. Renault,

et al.,

70. M. Reches and E. Gazit, Israel J. Chem., 2005, 45, 363–371.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc151

et al.,

Self-Assembly of Polymers into Soft Nanoparticles and Nanocapsules Jianbin Tang, Dingcheng Zhu, and Youqing Shen Zhejiang University, Hangzhou, China

1 Introduction 2 Self-Assembly of Polymers in Solution 3 Theoretical Prediction of the Basic Structural Parameters 4 Experimental Aspects 5 Stimulus-Responsive Nanoassemblies 6 Application of Nanoassemblies for Drug Delivery 7 Conclusions References

1

1 1 4 9 13 18 19 20

INTRODUCTION

Polymer architectures containing immiscible segments or chains, such as linear block copolymers, graft copolymers with pendant chains different from the backbone, and treelike dendritic polymers, can self-assemble via phase separation between immiscible blocks into well-defined structures and unique morphologies with characteristic sizes ranging from a few to hundreds of nanometers1 (Figure 1).2 Theoretical models, such as the self-consistent field theory (SCFT) or the mean-field theory (MFT), and computer simulations,3 such as dissipative particle dynamics simulation4 and coarse-grain molecular dynamics,5 have been developed to describe the behaviors of block copolymers, in which the phase behavior is governed by the Flory–Huggins segment–segment interaction parameter, the degree of polymerization, and the composition.

Dissolution or solvation of the bulk materials produces various nanostructures in solution. Among them, spherical and worm-like micelles and vesicles (polymersomes) have recently been attracting great attention due to their many potential applications, particularly in drug and gene delivery.6 For instance, the critical micelle concentration (CMC) of block copolymers is usually much lower than that of lowmolecular weight surfactants; therefore, polymer micelles are advantageous in drug delivery since they hardly dissociate in the blood stream under highly dilute conditions. The walls of the polymeric vesicles, around 10-nm thick, are much thicker than those of typical liposomes (several nanometers), and thus can hold their contents much more tightly. The self-assembly mechanisms of copolymers into nanostructures and their applications have been extensively reviewed elsewhere.1, 7 The basis of the various aspects of this research area is the correlation of the polymer structures to the type and the structural parameters of the formed nanostructures, which are the keys for researchers to design needed nanostructures for applications such as drug delivery. Along this line, this chapter focuses on the principles of formation, theoretical structural-parameters predications, and the basic experimental aspects in preparation and characterization, as well as recent progress and applications of polymer micelles and vesicles.

2

SELF-ASSEMBLY OF POLYMERS IN SOLUTION

Self-assembly of block copolymers in solution is driven by their block selectivity to a particular solvent. The solvophobic blocks aggregate together to form the solvophobic Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc152

2

Soft matter

MIC

BCC

HEX

CYL la3d

LAM

HPL

MLAM

Figure 1 Common morphologies of microphase-separated block copolymers: body-centered cubic packed spheres (BCC), hexagonally ordered cylinders (HEX), gyroid (Ia3d), hexagonally perforated layers (HPLs), modulated lamellae (MLAM), lamellae (LAM), cylindrical micelles (CYL), and spherical micelles (MIC). (Reproduced from Ref. 2.  Wiley-VCH, 1998.)

domain, while the solvophilic blocks are solvated by the solvent and extend into the solvent to form the shell. For an amphiphilic diblock copolymer in water, for example, a micelle consists of a hydrophobic core formed by the hydrophobic blocks and a corona formed by the hydrophilic blocks extending into the water. The specific size and morphology of such self-assembled structures are driven largely by thermodynamic forces—minimizing their free energy,8 but there is also evidence that suggests some degree of kinetic control during self-assembly as well.9 Spherical micelles

The micellar structures form in the solution only when the block copolymer is above a certain concentration. The specific concentration required for micellar structure formation is referred to as the CMC or, less commonly, the critical aggregation concentration (CAC). At polymer concentrations below the CMC, only free, randomly distributed single chains are present in the solvent. Above the CMC, both micelles and single chains coexist at equilibrium. The CMC is a simple thermodynamic measure of aggregate stability, as described in (1).10 The CMC varies depending on the chemical structure of the polymer as well the molecular weights of each block of the copolymer. Higher molecular weight generally, and specifically higher molecular weight of the hydrophobic block, corresponds to a lower value of CMC: CCMC ∼ exp(−nε h /kB T )

(1)

where kB T is the thermal energy, εh is the monomer’s effective interaction energy with the bulk solution (related to χ in polymer physics), and n is the number of ethylene units in the hydrophobic block. Block copolymers self-assemble in a particular solution into various reported morphologies11 primarily dictated by the inherent molecular curvature, which can be described by a dimensionless packing parameter, p 12 (Figure 2), as used in small molecular amphiphiles and defined in (2): p=

v/ lc a0

Cylindrical micelles

‘‘Polymersomes’’

P ≤ 13

Medium curvature 1 ≤ P ≤ 1 3 2

Low curvature 1 2 ≤P ≤ 1

>45%

45 –55%

25 – 40%

(2)

Ic a0 V

High curvature

f hydrophilic

Figure 2 Self-assembled structures from block copolymers in a block-selective solvent estimated from the packing parameter or the mass fraction of the hydrophilic block in the copolymer (fPhydrophilic ).14 (Reproduced from Ref. 12.  Wiley-VCH, 2009.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc152

Self-assembly of polymers

Sphere

Rod

the (3)14b : fhydrophilic ≈ e−

= 0.66)

(3)

However, it should be noted that these rules have not yet been fully verified in many copolymers. The ranges for forming the micellar structures may also be affected by the chemical structure of the copolymer. For instance, copolymers with charged blocks or rod–coil structures may follow the above trend but the range of the composition for the morphology will be different. As an example, linear polystyrene-b-poly(acrylic acid) (PS-b-PAA) formed spheres to rods and then to vesicles as the length of the hydrophilic PAA segment decreased, but the values of fhydrophilic forming the structures did not fall into the ranges indicated above (Table 1).15 Table 1 The nanostructures formed by polystyrene-b-poly (acrylic acid) (PS-b-PAA) at different compositions.15 PS-b-PAA (units)

PAA (mol%)

170-b-33 180-b-28 180-b-14 410-b-20 410-b-16 200-b-4

16.3 13.5 7.2 4.7 3.8 2.0

Vesicle

10

p/β (β

Fusion

Morphology Sphere Sphere Rod Vesicle Lamella LCM (large compound micelle)

Fission

Soluble

Copolymer concentration (wt%)

where v is the volume of the densely packed hydrophobic segment, lc is the chain length of the hydrophobic block normal to the interface, and a0 is the effective crosssectional area of the hydrophilic segment.13 The dimensionless packing parameter, p, can be considered as the ratio of the effective areas of the hydrophobic to hydrophilic segments. On the basis of the effective-area ratio of the hydrophobic to hydrophilic blocks, or p, it can be estimated that copolymers with large hydrophilic blocks (i.e., high curvature: p ≤ 13 ), will form spherical micelles; those with similar sizes of the two blocks (i.e., very low curvature: 12 ≤ p ≤ 1), will yield vesicles (polymersomes); whereas, those with the block size ratios in between will possibly form rod-like or worm-like micelles. For coil–coil block copolymers readily soluble in selected solvents with molecular weights ranging from 2700 to 20 000 g mol−1 , a more easily accessible parameter, the mass fraction of the hydrophilic block in the copolymer (fhydrophilic ) can be used to predicate the formed structures, as proposed by Discher and Eisenberg.14a The copolymer with fhydrophilic > 45% may not only form spherical micelles but also form rod-like micelles (fhydrophilic < 50%). Vesicles may be observed at fhydrophilic of around 35%. Inverted microstructures such as large compound micelles may be observed at fhydrophilic < 25%. On the basis of simulation and experiments for various amphiphilic copolymers, Discher et al. correlated fhydrophilic and p with

3

1

0.1 0

(a)

5

10 15 20 25 30 35 40 45 Water content (wt% in dioxane) Add water

20.0%

24.5%

28.6%

39.4%

50.0%

20.0%

24.5%

28.6%

39.4%

50.0%

66.7%

% Water content 200 nm

(b)

Add THF/dioxane (44/56)

(c)

Figure 3 Morphological phases and vesicle transformations in dilute solutions. (a) Phase diagram of PS310 –PAA52 in dioxane plus water. The colored regions between sphere and rod phases and between rod and vesicle phases correspond to coexistence regions. (b) Reversibility of the vesicle formation and growth process for PS300 –PAA44 , presumably based on part of fusion and fission processes illustrated in (c). (Reproduced with permission from Ref. 14a.  American Association for the Advancement of Science, 2002.) Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc152

4

Soft matter (a)

(b)

(c)

100 nm

(d)

100 nm

(e)

100 nm

(g)

100 nm

(f)

100 nm

100 nm

(h)

(i)

200 nm

200 nm

500 nm

Figure 4 Aggregates from PS410 -b-PAA25 without any additive (a) and with added NaCl to different final concentrations: (b) 1.1 mM (R = 0.20); (c) 2.1 mM (R = 0.40); (d) 3.2 mM (R = 0.60); (e) 4.3 mM (R = 0.80); (f) 5.3 mM (R = 1.0); (g) 10.6 mM (R = 2.0); (h) 16.0 mM (R = 3.0); and (i) 21 mM (R = 4.0). (Reproduced from Ref. 19b.  American Chemical Society, 1996.)

Furthermore, as the formation of a micellar structure is basically determined by its Gibbs free energy, which depends on (i) the stretching of the core-forming chains, (ii) the core–corona interfacial energy, and (iii) the repulsion among coronal chains, while the composition of a block copolymer is a determining parameter for the nanoassembly, changes in any of the three parameters that change the free energy of the micellar structures may induce a transition in their morphology (Figure 3), as observed by Zhang and Eisenberg and Yu and Eisenberg in the morphological transitions of PS–PAA8, 16 and PS–PEO (poly(ethylene oxide)).17 For instance, the starting nonselective solvent18 and adding salts19 (Figures 3 and 4) during the micellization could induce a morphological change.20

3 3.1

THEORETICAL PREDICTION OF THE BASIC STRUCTURAL PARAMETERS Spherical micelles

An AB-diblock copolymer composed of an insoluble block A with repeating units NAi and a sufficiently long soluble block B with repeating units NBs , with a right packing parameter p can self-assemble into micelles in which the insoluble block A forms the core and the soluble block B forms the shell or corona. When NAi < NBs , starlike or hairy micelles are formed, and when NAi > NBs , crew-cut micelles are formed (Figure 5). This process is driven by the free energy of a micelle contributed from the interfacial energy of the core/shell interface,

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc152

Self-assembly of polymers

5

107

Z = Z0 NA2 NB−0.8 106

b

b

105

Rc Rm Rm

ZN B0.8 /Z 0

Rc Dh

Dh (b)

(a)

104 103 102

Figure 5 The key parameters describing a star-like or hairy micelle (a) and a crew-cut micelle (b).

101 100

the stretching energy of the block copolymer chains, and the repulsion among the coronal chains.21 Several important parameters are used to characterize a micelle, the aggregation number or the number of block copolymer chains in one micelle Z, the core radius Rc , the overall radius of the micelle Rm , the micelle radius of gyration Rg , the hydrophilic shell (or corona) thickness Dh , and the distance between neighboring blocks at the core/shell interface or the grafting distance b. Therefore, b2 is the area occupied by one chain at the core/corona interface (Figure 5). The key properties that determine the above parameters are the degree of polymerization of the polymer blocks, NAi and NBs , and the Flory–Huggins interaction parameter χ. In scaling theories, micelles’ Rc , Rm , and Z are directly correlated to NAi and NBs of the block copolymer. For hairy micelles, the number of block copolymers in a micelle, Z, and the graft density, b, can be given using a general relation in (4) and (5). These equations are valid for uncharged diblocks, triblocks, and graft copolymers, as well as for low-molecular weight nonionic, anionic, and cationic surfactants (Figure 6).22, 23 Z=

Z0 NA2 i NB−0.8 s β/

b = b0 NBs6 with b0 ∼ 1 nm and β ∼ 0.8

(4)

1

10

100

1000

NA

Figure 6 Aggregation number Z as function of NAi and NBs for various block copolymer/solvent systems: poly(styrene-b-4vinylpyridine)/toluene (◦), poly(styrene-b-methacrylic acid)/ dioxane–water (
), poly(methacrylic acid-b-styrene-b-methacrylic acid)/dioxane–water (), poly(styrene-co-maleic anhydride-g-ethylene oxide)/water (), alkyl ethylene glycol/ water (•), alkylammonium bromide/water (), alkylsulfonate/ water (s), alkylsulfate/water (). The solid line corresponds to the relation Z = Z0 NA2 i NB−0.8 . (Reproduced from Ref. 2.  Wileys VCH, 1998.)

For hairy or star-like micelles, the micelle total radius Rm , is mostly controlled by the coronal chains (6)21, 24 : 4/

3/

Rm ∼ NAi25 ·NBs5

(6)

Shell thickness depends on both NAi and NBs (7).22 1

3/

Dh ∼ Z /5 NBs5

(7)

For spherical micelles, the core radius Rc and the occupied area per chain b2 are directly related to the number of polymers per micelles, the aggregation number (8):

(5) Z = 4π Rc2 /b2

where NAi and NBs are the degrees of polymerization of the insoluble and soluble blocks, respectively. Z0 is known for many block copolymer and surfactant systems,22 and for many Z0 = 1. Therefore, an increase in the insoluble block chain length leads to an increase in the aggregation number, but an increase in the soluble block chain length decreases the aggregation number. The graft density is only determined by the soluble block and decreases (i.e., the area per chain b2 increases) as its block becomes longer.

(8)

Nagarajan and Ganesh developed MFTs by considering the influence of the soluble block B in the formulation of the free energy of micellization and obtained scaling relations between micelle size characteristics. They found that the solvent-compatible B block played an important role in determining the characteristic features, especially when the B block is more soluble.25 For polystyrene–polybutadiene (PS–PBD) in n-heptane, (9)–(11) give the parameters related to the PS (NSt ) and

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc152

6

Soft matter

PBD (NBD ) chain lengths.25 1.10 −0.24 Z ∼ NSt NBD

(9)

0.70 −0.08 NBD Rc ∼ NSt

(10)

0.07 0.68 Dh ∼ NSt NBD

(11)

For a PEO–PPO (polypropylene oxide)–water system, (12)–(14) give the parameters related to the PPO (NPO ) and PEO (NEO ) chain lengths.25 1.19 −0.51 Z ∼ NPO NEO 0.06 0.74 NEO Dh ∼ NPO

(13)

0.73 −0.17 NEO Rc ∼ NPO

(14)

Rm =

3/

1/

 1/3

3NA2 i (σ A−S l 2 /kT ) + NAi2 + NAi NBs2 (Rm Dh ) − 1/3

[1 + NAi

1/

2/

3 Rc ∼ σ A−S NAi3 a

Z ∼ σ A−S NAi

(18) (19)

where a is the segment length.

(12)

By correlation of the calculated numerical results, Nagarajan and Ganesh further derived a general relation of Rc , Rm , and Z as a function of NAi and NBs (15)–(17), the interaction parameter χ B–S between the B block and the solvent S, the interfacial tension σ A–S between the A block and the solvent, and the characteristic segment length of the solvent. It is proposed to be valid for any block copolymer–solvent systems with intermediate values of the interfacial tension σ A–S and χ B–S . It should be noted that these general relations are valid only for systems that give rise to narrowly dispersed micellar aggregates.25


is calculated from the volume of the solvent molecule as 1/ l = Vs 3 . For crew-cut micelles with NAi  NBs , the micellar parameters only depend on NAi following (18) and (19)7c :

l

+ (NAi /NBs )(Dh Rm )2 ] 1/3

(15)   4π 1/2 4π 1/2 4π NAi (σ A-S l 2 /kT ) + NAi + NBs (Rm /Dh ) 3 3
 Z= − 1/3 1 + NAi + (NAi /NBs )(Dh /Rm )2 (16) The dimensionless shell thickness can be calculated from   1/5 NA2 i NBs 1 Dh 6/ − 8/ = 0.867 − χ B-S NAi 11 NBs7 + Rm 2 (NAi + NBs )3 (17) where χ B–S is the Flory–Huggins interaction parameter between the B block and the solvent S, σ A–S is the interfacial tension between the micelle core block A and the solvent S, k is Boltzmann constant, and T is the temperature. l is the characteristic segment length which

3.2

Cylindrical or worm-like micelles

Block copolymers with packing parameters in the range of 1/3 – 1/2 or hydrophilic block-weight fraction in the range of 45–55% may form cylindrical or worm-like micelles under the right conditions.20 A worm-like micelle may have an overall length, referred to as the contour length L, of a few nanometers to micrometers. Cryo-TEM (transmission electron microscopy)26 or fluorescence video imaging27 (Figure 7a) can be used to directly visualize the micelle and estimate its contour length, while light and neutron scattering gives a more accurate determination. Typical examples of polymer worm-like micelles are made from PEO–PBD (poly(ethylene oxide)–poly(butadiene)),28 poly(ethylene oxide)-block -polycaprolactone.27a Worm-like micelles have very interesting but complicated static and dynamic properties,30 which will not be covered here. The structural parameters describing wormlike micelles of our interest (Figure 7b) include the contour length L, the average contour length LA , the persistence length lp , which are the characteristic lengths governing the decay of tangent–tangent correlations, and the hydrophobic core diameter d. These basic parameters are essential for designing worm micelles for biodelivery. For a linear chain without closed rings, MFT, generally used for surfactants, predicts that the average contour length LA of worm-like micelles is a function of the surfactant volume fraction φ, and the end-cap energy Ec (the energy required to form two hemispherical end caps as a result of rod scission)30a : LA ≈ ϕ

1/2 exp[Ec /2kB T ]

(20)

and the length distribution, c(L) ∝ exp [−L /LA ]

(21)

 where the total volume fraction ϕ = L Lc(L); c(L) is the number density of chains of length L, T is the temperature, and kB is the Boltzmann constant.

Supramolecular Chemistry: From Molecules to Nanomaterials, Online  2012 John Wiley & Sons, Ltd. This article is  2012 John Wiley & Sons, Ltd. This article was published in the Supramolecular Chemistry: From Molecules to Nanomaterials in 2012 by John Wiley & Sons, Ltd. DOI: 10.1002/9780470661345.smc152

Self-assembly of polymers

1s

0s

7

2s

L

3s

6s

lp

(a)

d

(b)

Figure 7 (a) Dynamic snapshots and overlay of 15 backbone traces of a typical PAA45 –PBD107 worm micelle, scale bar is 5 µm. (Reprinted with permission from Ref. 27, Copyright 2005 American Chemical Society) and (b) Schematic representation of a worm-like micelle showing characteristic length-scales: the overall radius of gyration Rg , the contour length L, the persistence length lp , and the hydrophobic core diameter d. (Adapted from Ref. 29.  Royal Society of Chemistry, 2007.)

Worm-like micelles can be very flexible; therefore, the equation for flexible polymers31 can be used to calculate lp for worm-like micelles with good flexibility32 : 

lp  − L/ 1 − e lp R = 2lp L 1 − L 2

 (22)

where R is the end-to-end distance of the worm-like micelle. The persistence length is determined by the bending rigidit