Functional Materials: For Energy, Sustainable Development and Biomedical Sciences 9783110307825, 9783110307818

Table of contents :
Contents
Foreword
Preface
Contributing authors
About the editors
1 Introduction
Part I: Functional materials: Synthesis and applications
2 A primer on polymer colloids: structure, synthesis and colloidal stability
2.1 Introduction
2.2 Polymer colloids inside out
2.2.1 How many polymer chains per particle?
2.2.2 How many particles?
2.2.3 Are the chains immobile within the nanoparticle?
2.2.4 Morphology of polymeric nanoparticles
2.3 Preparation of polymer nanoparticles
2.3.1 Emulsion polymerization
2.3.2 Miniemulsion polymerization
2.3.3 Microemulsion polymerization
2.3.4 Self-assembly in selective solvents
2.4 Colloidal stabilization
2.4.1 Electrostatic stabilization
2.4.2 Steric stabilization
2.4.3 Depletion stabilization
2.4.4 Future directions
3 Synthesis, functionalization and properties of fullerenes and graphene materials
3.1 Introduction
3.2 Fullerenes
3.2.1 General considerations
3.2.2 Synthesis and purification of fullerenes
3.2.3 Chemical and physical properties of C60
3.2.4 Chemical functionalization of C60
3.2.5 Applications
3.3 Graphene
3.3.1 Production of graphene
3.3.2 Graphene in energy conversion devices
4 Ordered mesoporous silica: synthesis and applications
4.1 Introduction
4.2 Ordered mesoporous silica (OMS)
4.2.1 Principle of synthesis
4.2.2 Mesostructure diversity and tailoring
4.3 Functionalization of ordered mesoporous silica
4.4 Morphology control
4.5 Selected applications of functionalized ordered mesoporous silica
4.5.1 Functionalized MSNs as controlled drug delivery platforms
4.5.2 Functionalized mesoporous materials for extraction chromatography (EXC) applications
4.5.3 Mesoporous organic-inorganic hybrid membranes for water desalination
5 Nanoparticles: Properties and applications
5.1 Introduction
5.2 Synthetic methods
5.2.1 Particle nucleation and growth
5.2.2 Synthesis in inverse micelles
5.3 Particle aggregation and stabilization of colloidal suspensions
5.4 Colloidal quantum dots
5.5 Metal nanoparticles
5.6 Metal oxide nanoparticles
5.6.1 Titanium dioxide
5.6.2 Iron oxide
5.6.3 Silica
5.7 Polymeric nanoparticles
5.8 Advanced architectures and hybrid systems
6 Conjugated polymers for organic electronics
6.1 Introduction
6.2 Processable conjugated polymers
6.3 Applications in renewable energy
6.3.1 Organic solar cells
6.3.2 Conjugated polymers for organic solar cells
6.4 Applications in micro-electronics
6.4.1 Field-effect transistors
6.4.2 Conjugated polymers for field-effect transistors
6.5 Applications in lighting
6.5.1 Light-emitting diodes
6.5.2 Conjugated polymers for light-emitting diodes
6.6 Summary
7 Theoretical tools for designing microscopic to macroscopic properties of functional materials
7.1 Methods
7.1.1 The link between microscopic and macroscopic scales
7.1.2 Ab initio methods
7.1.3 Bridging the gap between ab initio and atomistic levels
7.1.4 Atomistic simulation
7.1.5 Bridging the gap between atomistic and mesoscale levels
7.2 Examples
7.2.1 Quantum studies
7.2.2 Atomistic simulation
7.3 Summary
Part II: Development of new materials for energy applications
8 Electrochemical energy storage systems
8.1 Introduction
8.2 Metrics and performance evaluation
8.3 Models and theory of electrochemical charge storage
8.3.1 Battery operation – a Faradaic process
8.3.2 Electrochemical capacitor operation – a non-Faradaic process
8.4 Electrolytes
8.5 Electrode materials
8.5.1 Electrochemical capacitors
8.5.2 Hybrid electrochemical capacitors
8.5.3 Lithium battery electrode materials
8.5.4 Negative (anode) electrode materials
8.5.5 The positive (cathode) electrode
8.5.6 Electrode production
8.6 Summary
9 Functional ionic liquids electrolytes in lithium-ion batteries
9.1 Introduction
9.1.1 Historical overview
9.1.2 What are ionic liquids?
9.1.3 Key properties as electrolytes
9.2 Ionic liquids as Li and Lithium-ion battery electrolytes
9.3 Functional ionic liquid electrolytes
9.3.1 Overview of functional ionic liquids
9.3.2 Solid electrolyte interphase
9.3.3 Transport of lithium ions
9.3.4 Electroactive ionic liquids as redox shuttles
9.3.5 Perspectives
10 Solid polymer proton conducting electrolytes for fuel cells
10.1 Introduction
10.2 Proton exchange membranes
10.2.1 Nafion®
10.2.2 Alternative sulfonated ionomers and membranes
10.3 Characterization of solid polymer electrolytes
10.3.1 Proton conductivity
10.3.2 States of water and water mobility
10.4 Summary
11 Supercritical adsorption of hydrogen on microporous adsorbents
11.1 Introduction
11.2 Fundamentals of supercritical adsorption
11.3 Supercritical adsorption isotherms
11.3.1 Virial expansion of the excess density in terms of pressure
11.3.2 Basic analytic models of the adsorption isotherm
11.3.3 Self-consistent approaches
11.4 The thermodynamics of adsorption
11.4.1 Properties of surface potential
11.5 Microporous adsorbents for hydrogen storage
11.5.1 Activated carbons
11.5.2 Single wall nanotubes
11.5.3 Metal organic frameworks
Part III: New trends in sustainable development and biomedical applications
12 Advanced materials for biomedical applications
12.1 Introduction
12.2 History of biomaterials
12.3 Basics in material science for biomaterial applications
12.3.1 Biomaterial properties
12.3.2 Biometals
12.3.3 Bioceramics
12.3.4 Biosynthetic polymers
12.3.5 Natural polymers
12.4 Biomedical applications
12.4.1 Cardiovascular system
12.4.2 Musculoskeletal system
12.4.3 Visceral organs
12.4.4 Nervous system and sensory organs
12.4.5 Esthetic applications
12.4.6 Skin
12.5 Future trends
12.5.1 Tissue engineering basic concepts
12.5.2 Scaffolds
12.5.3 Surface modification
12.5.4 Stem cells
12.5.5 Bioreactors
12.5.6 Computational models
12.6 Summary
13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine
13.1 The basics of MRI in medicine
13.2 Relaxivity: the performance of MRI contrast agents
13.3 Synthesis and characterization of magnetic nanoparticles
13.3.1 Synthesis of magnetic nanocrystals
13.3.2 Nanoparticle coatings for MRI applications
13.3.3 Physicochemical characterization
13.4 Physical properties of magnetic nanoparticles
13.5 MR relaxation properties of magnetic nanoparticles
13.5.1 Relaxivity of paramagnetic CAs
13.5.2 Relaxivity of superparamagnetic CAs
13.5.3 Relaxometric performance of MRI CAs at clinical magnetic field strengths
13.6 Biological performance of magnetic nanoparticles for MRI
13.6.1 In vivo barriers
13.6.2 Impact of nanoparticle size and surface on colloidal stability and blood retention
13.6.3 Directing nanoparticles in vivo
13.6.4 Toxicity
13.7 Summary
14 Microfluidics for synthesis and biological functional materials: from device fabrication to applications
14.1 Introduction
14.2 A practical introduction to microfluidic reactors for material synthesis
14.2.1 Microfluidic reactor geometries
14.2.2 Device fabrication materials
14.2.3 Fabrication of polymer-based planar microreactors and components
14.3 Manipulating and measuring precursor reagent streams in microchannels
14.3.1 High surface area to volume ratios in microchannels
14.3.2 Rapid heat transfer
14.3.3 Control of concentrations
14.3.4 Controlling “time on chip”
14.3.5 Control of hydrodynamics and mass transfer
14.3.6 Characterization in microchannels
14.4 Microfluidics for polymer microparticles
14.4.1 Manipulating the shaping of liquid precursors
14.4.2 Effect of the channel wall
14.4.3 Emulsification of precursor droplets
14.4.4 Channel geometries to achieve emulsified droplets
14.4.5 Multiple emulsions
14.4.6 Forming linear threads and two-dimensional interfaces
14.4.7 Converting liquid precursors into solid micro-materials
14.4.8 Scale up: a circuit analysis of microfluidic flow in a highly parallelized microreactor
14.5 Microfluidics for synthesis of functional nanoparticles
14.5.1 Microfluidics for highly controlled nanoparticle synthesis
14.6 Biomaterials
14.6.1 Tissue engineering and membranes
14.6.2 Microenvironments for encapsulated cells
14.6.3 Biofilms
14.6.4 Microdevices utilizing functional biomaterials
14.7 Summary
15 Protein- and peptide-based materials: a source of inspiration for innovation
15.1 Introduction
15.2 Basics of proteins, peptides and polypeptides
15.2.1 Polypeptides are sequences of amino acids
15.2.2 Polypeptides can adopt various conformations
15.2.3 Polypeptides possess various levels of structural organization
15.3 Functional materials from fibrous proteins
15.3.1 Resilin & abductin
15.3.2 Byssus (mussel anchoring threads)
15.3.3 Silk
15.4 Functional materials from globular proteins
15.4.1 Natural proteins
15.4.2 Artificial proteins
15.5 Functional materials from synthetic peptides
15.6 Summary
16 Nanocomposite coatings
16.1 Introduction
16.2 Coating formulations
16.2.1 Chemical components
16.2.2 Mixing techniques
16.2.3 Application and curing
16.3 Nanoparticle additives
16.4 Coating characterization
16.4.1 Mechanical properties
16.4.2 Optical properties
16.4.3 X-ray imaging and particle aggregation
16.4.4 Weathering and artificial aging
16.5 Bio-based coatings
16.6 Future developments
16.7 Summary
Index

Citation preview

De Gruyter Graduate Leclerc, Gauvin ∙ Functional Materials

Also of Interest Polymer Surface Characterization Luigia Sabbatini (Ed.), 2014 ISBN 978-3-11-027508-7, e-ISBN 978-3-11-028811-7

Organic and Hybrid Solar Cells – An Introduction Lukas Schmidt-Mende, Jonas Weickert, 2015 ISBN 978-3-11-028318-1, e-ISBN 978-3-11-028320-4

An Introduction to Surfactants Tharwat F. Tadros, 2014 ISBN 978-3-11-031212-6, e-ISBN 978-3-11-031213-3

e-Polymers Seema Agarwal, Andreas Greiner (Editor-in-Chief) ISSN 1618-7229

Journal of the Mechanical Behavior of Materials Aifantis, Elias C. (Editor-in-Chief) ISSN 2191-0243

Functional Materials | For Energy, Sustainable Development and Biomedical Sciences Edited by Mario Leclerc and Robert Gauvin

Editors Prof. Mario Leclerc Department of Chemistry Université Laval Quebec, Canada Email: [email protected]

Dr. Robert Gauvin Scientific Liaison Officer Québec Center for Functional Materials (CQMF) Quebec, Canada Email: [email protected]

ISBN 978-3-11-030781-8 e-ISBN (PDF) 978-3-11-030782-5 e-ISBN (EPUB) 978-3-11-038819-0 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2014 Walter de Gruyter GmbH, Berlin/Boston Cover image: PASCAL GOETGHELUCK/SPL/Agentur Focus Typesetting: PTP-Berlin, Protago TEX-Produktion GmbH, www.ptp-berlin.de Printing and binding: Hubert & Co. GmbH & Co. KG, Göttingen ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Foreword When asked to write a foreword for this new book on functional materials, I initially hesitated to comply with the request for a few simple reasons. First of all, new journals and books on functional materials are sprouting these days like cabbage in the spring. This is even more the case for original research papers printed in the usual disciplinary journals in chemistry, physics, medicine, and engineering. Consequently, it is almost impossible to critically keep track and differentiate important studies from the less significant ones. Moreover, not every original contribution contains completely novel work, as many of them are instead additions to already well-known facts or wellaccepted lines of thought. In addition, the word “functional” implies that the material serves a specific and defined purpose in a unique manner; but many contributors to the field forget or overlook the fact that the material under consideration or subject under study is just one of many components in a device and its function only evolves through interplay with all the other components in the device. Note that even the simplest of all rechargeable batteries as an example of a device contains at least a dozen different functional materials, which all need to comply with the overall performance of the battery. Thus, compatibility of materials is an important issue frequently overlooked by eager researchers. These critical remarks should not be misunderstood. We do not argue against academically motivated research in materials science. On the contrary, we think that high quality work aiming at improved, optimized, and – to some extent – novel functional materials is a necessity. We would merely like to express a warning against exaggerated promises and hopes that have no basis in the world of practice. After all, research on functional materials must envisage practical solutions and is never justified as a purely academic exercise. The clash between academic ideas and the reality of the industry and the markets is nowhere harsher than in this field of science. Despite these critical remarks, it is highly appreciated that this book is not simply a review of the vast body of literature of the recent years, as it holds the focus upon various aspects of application. Moreover, it selects only a few topics in favor of a solid and thorough treatment of the relevant aspects. It is particularly pleasing that polymer colloids receive much attention. Polymer colloids have seen an impressive revival as a subject of academic research. This is a bit surprising at first sight, given the situation that emulsion and suspension polymerization processes have served as major industrial processes in the mass production of polymers, to supply markets with materials for the coatings and adhesives industry, as textile and leather modifiers and as constituent in cosmetics, just to mention a few. However, the widespread interest in nanotechnology has guided the researchers to develop methods which not simply aim at the synthesis of new molecules, but rather target the production of defined objects. These nanosized objects are not only composed of different molecules, each type of molecule contributing different functions to the object, but they also exhibit a defined

VI | Foreword shape and size distribution, have a defined and controllable surface structure, and a structured internal composition. Similar cases are characteristic of the synthesis of mesoporous materials, which have a long history as absorbents and filtering materials in industry, but have also found new attention, with the desire to embed more functions into these materials than was previously possible. This includes better control of pore structure and pore size distribution. Here again, the target of the synthesis is the object, and not the individual molecule which would only later be subject to further processing. Similarly, nanoscale pigments have been available through industrial processes for decades. Nevertheless, academic interest centers on the development of novel pigments with particular size distribution, surface structure, and function, as well as specific electromagnetic properties. The definition of viable conditions for mass production by industrially acceptable processes is a key to the success of such research. All of these aspects are treated in this book. In addition, materials for energy storage (batteries and capacitors) as well as fuel cells are brought to the readers’s attention. Last but not least, materials aiming towards application in the vast field of biomedicine are discussed, described, and critically evaluated. Biocompatibility, safety aspects, and the essential approval by the authorities are important aspects not to be overlooked in this area, requiring extensive knowledge in both materials science and medicine. In summary, this book comes at a good time, when a large body of academic literature has been accumulated and is waiting for a critical inspection in the light of the real demands of application. This book will provide valuable information to a critical readership and should receive wide-spread recognition from the scientific community. September 2014

Professor Gerhard Wegner Emeritus scientific member Max-Planck Institute for Polymer Research Mainz, Germany

Preface Energy, health, and environment are certainly at the top of the list of priorities in the challenges facing society worldwide for the next 50 years. Functional materials are providing new solutions and opportunities to ensure sustainable energy and environment, as well as improved medical care for the future. By its nature, the field of functional materials is highly interdisciplinary, comprising basic sciences such as physics, chemistry, and biology, as well as applied sciences and engineering. In this regard, this book brings together the expertise of multiple experts and gives an overview of emerging trends in this field of research. It allows the reader to associate a multitude of functional materials with their respective application, in a way that has not previously been done in existing books covering smart or functional materials. As one can imagine, this interdependence between synthesis, structure, properties, and performance of functional materials requires applied and theoretical chemistry, physics, biology, and engineering tools in order to ensure the development of novel and efficient technologies (Fig. 1). Synthesis

Properties Structure Performance Fig. 1. The field of functional materials is by definition interdisciplinary, involving scientists from various backgrounds in designing and synthesizing new materials with structure and properties tailored in order to achieve optimal performance.

This book is addressed to experts as well as graduate students interested in the most recent developments in the field of functional materials. It includes 15 chapters which are organized in three sections focusing on (1) the synthesis and applications of functional materials, (2) novel materials for energy applications, and (3) new trends in functional materials for sustainable development and biomedical applications. The chapters have been written by recognized experts in their respective fields and cover topics ranging from fundamental concepts in synthesis to applied research projects. Recent examples of how functional materials are improving many aspects of technology currently available to us on a daily basis are also discussed. The authors are all

VIII | Preface members of the Quebec Center for Functional Materials (CQMF), a research center located in Quebec, Canada, and are internationally known for their work in the development of innovative and state-of-the-art functional materials. In summary, this book provides an overview of the recent advances made in synthesis, characterization, and computer modeling, as well as various applications involving functional materials, and should be of great interest for the scientific community working in this area of research. The authors are grateful to Mrs. Patricia Basque (CQMF), and Mr. Philippe Dufour (NanoULaval) for their collaboration, and to Mrs. Karin Sora (De Gruyter), Mrs. Julia Lauterbach (De Gruyter), and Mrs. Kathleen Prüfer (De Gruyter) for their assistance in editing this book. We also acknowledge financial support from the FRQNT-Strategic Networks Funding Program, which provides the opportunity to bring scientists and engineers together in order to develop novel and innovative technologies. Québec, August 2014

Mario Leclerc Robert Gauvin

Contents Foreword | V Preface | VII Contributing authors | XVII About the editors | XXIII R. Gauvin 1 Introduction | 1

Part I:

Functional materials: Synthesis and applications

A. Al Shboul, F. Pierre, and J. P. Claverie 2 A primer on polymer colloids: structure, synthesis and colloidal stability | 9 2.1 Introduction | 9 2.2 Polymer colloids inside out | 10 2.2.1 How many polymer chains per particle? | 10 2.2.2 How many particles? | 10 2.2.3 Are the chains immobile within the nanoparticle? | 12 2.2.4 Morphology of polymeric nanoparticles | 13 2.3 Preparation of polymer nanoparticles | 17 2.3.1 Emulsion polymerization | 18 2.3.2 Miniemulsion polymerization | 22 2.3.3 Microemulsion polymerization | 24 2.3.4 Self-assembly in selective solvents | 25 2.4 Colloidal stabilization | 26 2.4.1 Electrostatic stabilization | 26 2.4.2 Steric stabilization | 30 2.4.3 Depletion stabilization | 31 2.4.4 Future directions | 33 S. Rondeau-Gagné and J.-F. Morin 3 Synthesis, functionalization and properties of fullerenes and graphene materials | 37 3.1 Introduction | 37 3.2 Fullerenes | 37

X | Contents 3.2.1 3.2.2 3.2.3 3.2.4 3.2.5 3.3 3.3.1 3.3.2

General considerations | 38 Synthesis and purification of fullerenes | 39 Chemical and physical properties of C60 | 40 Chemical functionalization of C60 | 42 Applications | 45 Graphene | 47 Production of graphene | 49 Graphene in energy conversion devices | 52

J. Florek, R. Guillet-Nicolas, and F. Kleitz 4 Ordered mesoporous silica: synthesis and applications | 61 4.1 Introduction | 61 4.2 Ordered mesoporous silica (OMS) | 62 4.2.1 Principle of synthesis | 63 4.2.2 Mesostructure diversity and tailoring | 69 4.3 Functionalization of ordered mesoporous silica | 78 4.4 Morphology control | 80 4.5 Selected applications of functionalized ordered mesoporous silica | 82 4.5.1 Functionalized MSNs as controlled drug delivery platforms | 83 4.5.2 Functionalized mesoporous materials for extraction chromatography (EXC) applications | 88 4.5.3 Mesoporous organic-inorganic hybrid membranes for water desalination | 91 A. Ritcey 5 Nanoparticles: Properties and applications | 101 5.1 Introduction | 101 5.2 Synthetic methods | 101 5.2.1 Particle nucleation and growth | 102 5.2.2 Synthesis in inverse micelles | 104 5.3 Particle aggregation and stabilization of colloidal suspensions | 105 5.4 Colloidal quantum dots | 107 5.5 Metal nanoparticles | 110 5.6 Metal oxide nanoparticles | 112 5.6.1 Titanium dioxide | 112 5.6.2 Iron oxide | 113 5.6.3 Silica | 115 5.7 Polymeric nanoparticles | 115 5.8 Advanced architectures and hybrid systems | 117

Contents

| XI

N. Allard and M. Leclerc 6 Conjugated polymers for organic electronics | 121 6.1 Introduction | 121 6.2 Processable conjugated polymers | 122 6.3 Applications in renewable energy | 126 6.3.1 Organic solar cells | 126 6.3.2 Conjugated polymers for organic solar cells | 128 6.4 Applications in micro-electronics | 130 6.4.1 Field-effect transistors | 130 6.4.2 Conjugated polymers for field-effect transistors | 132 6.5 Applications in lighting | 133 6.5.1 Light-emitting diodes | 133 6.5.2 Conjugated polymers for light-emitting diodes | 135 6.6 Summary | 136 A. Soldera 7 Theoretical tools for designing microscopic to macroscopic properties of functional materials | 139 7.1 Methods | 140 7.1.1 The link between microscopic and macroscopic scales | 140 7.1.2 Ab initio methods | 142 7.1.3 Bridging the gap between ab initio and atomistic levels | 146 7.1.4 Atomistic simulation | 147 7.1.5 Bridging the gap between atomistic and mesoscale levels | 151 7.2 Examples | 151 7.2.1 Quantum studies | 152 7.2.2 Atomistic simulation | 156 7.3 Summary | 164

Part II:

Development of new materials for energy applications

S. B. Schougaard and D. Bélanger 8 Electrochemical energy storage systems | 171 8.1 Introduction | 171 8.2 Metrics and performance evaluation | 171 8.3 Models and theory of electrochemical charge storage | 173 8.3.1 Battery operation – a Faradaic process | 174 8.3.2 Electrochemical capacitor operation – a non-Faradaic process | 175 8.4 Electrolytes | 178 8.5 Electrode materials | 180 8.5.1 Electrochemical capacitors | 180

XII | Contents 8.5.2 8.5.3 8.5.4 8.5.5 8.5.6 8.6

Hybrid electrochemical capacitors | 181 Lithium battery electrode materials | 183 Negative (anode) electrode materials | 184 The positive (cathode) electrode | 185 Electrode production | 186 Summary | 186

D. Rochefort 9 Functional ionic liquids electrolytes in lithium-ion batteries | 189 9.1 Introduction | 189 9.1.1 Historical overview | 190 9.1.2 What are ionic liquids? | 191 9.1.3 Key properties as electrolytes | 192 9.2 Ionic liquids as Li and Lithium-ion battery electrolytes | 193 9.3 Functional ionic liquid electrolytes | 194 9.3.1 Overview of functional ionic liquids | 195 9.3.2 Solid electrolyte interphase | 196 9.3.3 Transport of lithium ions | 197 9.3.4 Electroactive ionic liquids as redox shuttles | 198 9.3.5 Perspectives | 202 C. de Bonis, A. D’Epifanio, B. Mecheri, S. Licoccia, and A. C. Tavares 10 Solid polymer proton conducting electrolytes for fuel cells | 207 10.1 Introduction | 207 10.2 Proton exchange membranes | 209 10.2.1 Nafion® | 210 10.2.2 Alternative sulfonated ionomers and membranes | 213 10.3 Characterization of solid polymer electrolytes | 218 10.3.1 Proton conductivity | 218 10.3.2 States of water and water mobility | 222 10.4 Summary | 233 P. Bénard, A.-M. Beaulieu, D. Durette, and R. Chahine 11 Supercritical adsorption of hydrogen on microporous adsorbents | 241 11.1 Introduction | 241 11.2 Fundamentals of supercritical adsorption | 242 11.3 Supercritical adsorption isotherms | 246 11.3.1 Virial expansion of the excess density in terms of pressure | 246 11.3.2 Basic analytic models of the adsorption isotherm | 252 11.3.3 Self-consistent approaches | 256 11.4 The thermodynamics of adsorption | 257 11.4.1 Properties of surface potential | 259

Contents

| XIII

11.5 11.5.1 11.5.2 11.5.3

Microporous adsorbents for hydrogen storage | 261 Activated carbons | 261 Single wall nanotubes | 262 Metal organic frameworks | 263

Part III:

New trends in sustainable development and biomedical applications

D. Mantovani, L. Levesque, G. Sabbatier, M. Leroy, D. G. Seifu, R. Tolouei, V. Montaño, M. Cloutier, I. Bilem, C. Loy, M. Byad, C. Paternoster, C. A. Hoesli, B. Drouin, G. Laroche 12 Advanced materials for biomedical applications | 277 12.1 Introduction | 277 12.2 History of biomaterials | 278 12.3 Basics in material science for biomaterial applications | 280 12.3.1 Biomaterial properties | 280 12.3.2 Biometals | 280 12.3.3 Bioceramics | 281 12.3.4 Biosynthetic polymers | 282 12.3.5 Natural polymers | 284 12.4 Biomedical applications | 286 12.4.1 Cardiovascular system | 286 12.4.2 Musculoskeletal system | 291 12.4.3 Visceral organs | 300 12.4.4 Nervous system and sensory organs | 304 12.4.5 Esthetic applications | 310 12.4.6 Skin | 312 12.5 Future trends | 319 12.5.1 Tissue engineering basic concepts | 319 12.5.2 Scaffolds | 319 12.5.3 Surface modification | 323 12.5.4 Stem cells | 323 12.5.5 Bioreactors | 324 12.5.6 Computational models | 324 12.6 Summary | 326 M.-A. Fortin 13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine | 333 13.1 The basics of MRI in medicine | 337 13.2 Relaxivity: the performance of MRI contrast agents | 339

XIV | Contents 13.3 13.3.1 13.3.2 13.3.3 13.4 13.5 13.5.1 13.5.2 13.5.3 13.6 13.6.1 13.6.2 13.6.3 13.6.4 13.7

Synthesis and characterization of magnetic nanoparticles | 340 Synthesis of magnetic nanocrystals | 340 Nanoparticle coatings for MRI applications | 344 Physicochemical characterization | 346 Physical properties of magnetic nanoparticles | 347 MR relaxation properties of magnetic nanoparticles | 352 Relaxivity of paramagnetic CAs | 353 Relaxivity of superparamagnetic CAs | 356 Relaxometric performance of MRI CAs at clinical magnetic field strengths | 358 Biological performance of magnetic nanoparticles for MRI | 358 In vivo barriers | 360 Impact of nanoparticle size and surface on colloidal stability and blood retention | 361 Directing nanoparticles in vivo | 362 Toxicity | 363 Summary | 364

J. Greener 14 Microfluidics for synthesis and biological functional materials: from device fabrication to applications | 375 14.1 Introduction | 375 14.2 A practical introduction to microfluidic reactors for material synthesis | 376 14.2.1 Microfluidic reactor geometries | 376 14.2.2 Device fabrication materials | 377 14.2.3 Fabrication of polymer-based planar microreactors and components | 380 14.3 Manipulating and measuring precursor reagent streams in microchannels | 383 14.3.1 High surface area to volume ratios in microchannels | 383 14.3.2 Rapid heat transfer | 384 14.3.3 Control of concentrations | 384 14.3.4 Controlling “time on chip” | 386 14.3.5 Control of hydrodynamics and mass transfer | 386 14.3.6 Characterization in microchannels | 389 14.4 Microfluidics for polymer microparticles | 391 14.4.1 Manipulating the shaping of liquid precursors | 392 14.4.2 Effect of the channel wall | 392 14.4.3 Emulsification of precursor droplets | 393 14.4.4 Channel geometries to achieve emulsified droplets | 393 14.4.5 Multiple emulsions | 395

Contents

14.4.6 14.4.7 14.4.8 14.5 14.5.1 14.6 14.6.1 14.6.2 14.6.3 14.6.4 14.7

| XV

Forming linear threads and two-dimensional interfaces | 395 Converting liquid precursors into solid micro-materials | 397 Scale up: a circuit analysis of microfluidic flow in a highly parallelized microreactor | 397 Microfluidics for synthesis of functional nanoparticles | 400 Microfluidics for highly controlled nanoparticle synthesis | 401 Biomaterials | 402 Tissue engineering and membranes | 403 Microenvironments for encapsulated cells | 404 Biofilms | 406 Microdevices utilizing functional biomaterials | 407 Summary | 410

T. Lefèvre, F. Byette, I. Marcotte, and M. Auger 15 Protein- and peptide-based materials: a source of inspiration for innovation | 415 15.1 Introduction | 415 15.2 Basics of proteins, peptides and polypeptides | 417 15.2.1 Polypeptides are sequences of amino acids | 417 15.2.2 Polypeptides can adopt various conformations | 418 15.2.3 Polypeptides possess various levels of structural organization | 419 15.3 Functional materials from fibrous proteins | 420 15.3.1 Resilin & abductin | 421 15.3.2 Byssus (mussel anchoring threads) | 422 15.3.3 Silk | 425 15.4 Functional materials from globular proteins | 429 15.4.1 Natural proteins | 429 15.4.2 Artificial proteins | 430 15.5 Functional materials from synthetic peptides | 432 15.6 Summary | 435 B. Riedl, V. Vardanyan, W. N. Nkeuwa, A. Kaboorani, V. Landry, B. Poaty, M. Vlad, and C. Sow 16 Nanocomposite coatings | 443 16.1 Introduction | 443 16.2 Coating formulations | 446 16.2.1 Chemical components | 446 16.2.2 Mixing techniques | 447 16.2.3 Application and curing | 449 16.3 Nanoparticle additives | 449 16.4 Coating characterization | 454 16.4.1 Mechanical properties | 454

XVI | Contents 16.4.2 16.4.3 16.4.4 16.5 16.6 16.7

Optical properties | 456 X-ray imaging and particle aggregation | 459 Weathering and artificial aging | 459 Bio-based coatings | 460 Future developments | 462 Summary | 463

Index | 465

Contributing authors Nicolas Allard Department of Chemistry Université Laval Quebec, QC, Canada e-mail: [email protected] Chapter 6 Ahmad Al Shboul Department of Chemistry Université du Québec à Montréal (UQAM) Montréal, QC, Canada e-mail: [email protected] Chapter 2 Michèle Auger Department of Chemistry Université Laval Quebec, QC, Canada e-mail: [email protected] Chapter 15 Ann-Marie Beaulieu Research Institute on Hydrogen Université du Québec à Trois-Rivières (UQTR) Trois-Rivières, QC, Canada e-mail: [email protected] Chapter 11 Daniel Bélanger Department of Chemistry Université du Québec à Montréal (UQAM) Montréal, QC, Canada e-mail: [email protected] Chapter 8 Pierre Bénard Research Institute on Hydrogen Université du Québec à Trois-Rivières (UQTR) Trois-Rivières, QC, Canada e-mail: [email protected] Chapter 11

Ibrahim Bilem Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Michael Byad Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Frédéric Byette Department of Chemistry Université du Québec à Montréal and Université de Montréal Montréal, QC, Canada e-mail: [email protected] Chapter 15 Richard Chahine Department of Electrical Engineering and Computer Engineering Université du Québec à Trois-Rivières Trois-Rivières, QC, Canada e-mail: [email protected] Chapter 11 Jerome P. Claverie Department of Chemistry Université du Québec à Montréal (UQAM) Montréal, QC, Canada e-mail: [email protected] Chapter 2

XVIII | Contributing authors Maxime Cloutier Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Catia de Bonis Department of Chemical Science and Technology & NAST Center University of Rome Tor Vergata Rome, Italy e-mail: [email protected] Chapter 10 Alessandra D’Epifanio Department of Chemical Science and Technology & NAST Center University of Rome Tor Vergata Rome, Italy e-mail: [email protected] Chapter 10 Bernard Drouin College François Xavier Garneau e-mail: [email protected] Chapter 12 David Durette Department of Chemistry, Biochemistry and Physics Université du Québec à Trois-Rivières Trois-Rivières, QC, Canada e-mail: [email protected] Chapter 11 Justyna Florek Department of Chemistry and Centre de Recherche sur les Matériaux Avancés (CERMA) Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 4

Marc-André Fortin Department of Mining, Metallurgical, and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 13 Robert Gauvin Quebec Center for Functional Materials Québec, QC, Canada e-mail: [email protected] Chapter 1 Jesse Greener Department of Chemistry Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 14 Rémy Guillet-Nicolas Department of Chemistry and Centre de Recherche sur les Matériaux Avancés (CERMA) Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 4 Corinne A. Hoesli Department of Mining, Metallurgical and Materials Engineering Université Laval Centre de recherche du CHU de Québec and McGill University Québec, QC, Canada e-mail: [email protected] Chapter 12 Alireza Kaboorani Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16

Contributing authors

Freddy Kleitz Department of Chemistry and Centre de Recherche sur les Matériaux Avancés (CERMA) Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 4 Véronic Landry Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16 Gaétan Laroche Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Mario Leclerc Department of Chemistry Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 6 Thierry Lefèvre Department of Chemistry Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 15 Marie Leroy Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12

| XIX

Lucie Levesque Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Silvia Licoccia Department of Chemical Science and Technology & NAST Center University of Rome Tor Vergata Rome, Italy e-mail: [email protected] Chapter 10 Caroline Loy Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Diego Mantovani Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Isabelle Marcotte Department of Chemistry Université du Québec à Montréal Montréal, QC, Canada e-mail: [email protected] Chapter 15

XX | Contributing authors Barbara Mecheri Department of Chemical Science and Technology & NAST Center University of Rome Tor Vergata Rome, Italy e-mail: [email protected] Chapter 10 Vanessa Montaño-Machado Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Jean-François Morin Department of Chemistry Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 3 William Nguegang Nkeuwa Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16 Carlo Paternoster Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Florian Pierre Department of Chemistry Université du Québec à Montréal (UQAM) Montréal, QC, Canada e-mail: [email protected] Chapter 2

Bouddah Poaty Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16 Bernard Riedl Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16 Anna Ritcey Department of Chemistry Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 5 Dominic Rochefort Department of Chemistry Université de Montréal Montréal, QC, Canada e-mail: [email protected] Chapter 9 Simon Rondeau-Gagné Department of Chemistry Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 3 Gad Sabbatier Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12

Contributing authors

Steen B. Schougaard Department of Chemistry Université du Québec à Montréal (UQAM) Montréal, QC, Canada e-mail: [email protected] Chapter 8 Dawit G. Seifu Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Armand Soldera Department of Chemistry Université de Sherbrooke Sherbrooke, QC, Canada e-mail: [email protected] Chapter 7 Caroline Sow Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16

| XXI

Ana C. Tavares Institut National de la Recherche Scientifique Énergie, Matériaux et Télécommunications (INRS-EMT) Varennes, QC, Canada e-mail: [email protected] Chapter 10 Ranna Tolouei Department of Mining, Metallurgical and Materials Engineering Université Laval and Centre de recherche du CHU de Québec Québec, QC, Canada e-mail: [email protected] Chapter 12 Vahe Vardanyan Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: vahe.vardanyan.1@ ulaval.ca Chapter 16 Mirela Vlad Department of Wood and Forest Sciences Université Laval Québec, QC, Canada e-mail: [email protected] Chapter 16

About the editors Mario Leclerc was awarded a Ph.D. in chemistry from Université Laval, Quebec City, Canada, in 1987, under the guidance of Prof. R.E. Prud’homme. After a short post-doctoral stay at INRSEnergie et Matériaux near Montréal with professor L.H. Dao, he joined the Max-Planck-Institute for Polymer Research, in Mainz, Germany, as a post-doctoral fellow in the research group of Prof. Dr. G. Wegner. In 1989, he accepted a position of professor at the department of chemistry of Université de Montréal. He returned to Université Laval in 1998 where he has held since 2001 the Canada Research Chair for Electroactive and Photoactive Polymers. Prof. Leclerc has co-authored about 250 papers published in leading scientific journals which have been cited more than 16 000 times. According to Science Citation Index, he has an h-index of 64. His current research activities include the synthesis and characterization of new oligomers and polymers for applications in micro-electronics, energy, sensors, and genomics. Robert Gauvin holds a Mechanical Engineering degree from Université Laval (Québec, Qc, Canada) and has worked as R&D Engineer in the medical device industry (AltertekBio, SainteFoy, Qc, Canada) prior to attending graduate school. Following his Ph.D. studies in Biomedical Engineering at Université Laval and at the Georgia Institute of Technology (Atlanta, GA, USA), he was awarded Postdoctoral Research Fellowships at the Massachusetts Institute of Technology (Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA, USA) and at the Wyss Institute for Biologically Inspired Engineering (Harvard Medical School, Boston, MA, USA). After completion of his postdoctoral research, he joined the Personnel Protection Section as Staff Scientist for Defense Research and Development Canada (DRDC-Valcartier, Qc, Canada), before accepting the position of Scientific Liaison Officer at the Quebec Center for Functional Materials (CQMF, Qc, Canada). He also served as the elected Chair of Scientific and Professional Development of TERMIS-North America from 2010 to 2013 and holds a Research Professor position at the Department of Surgery (Faculty of Medicine) of Université Laval. Dr. Gauvin is co-author of over 35 peer-reviewed journal articles, 4 review articles and 3 book chapters. He presented his work at over 50 national and international conferences and is a reviewer for numerous scientific journals in the fields of biomedical engineering and material sciences.

R. Gauvin

1 Introduction Since its beginning, materials science has evolved from the use of inert structural materials to materials with tailored properties which allow reactive capacities, such as the intrinsic ability to respond to stimuli and environmental changes, and activation of specific functions according to these changes. Improvements and innovations in the fields of chemistry, physics, and engineering have allowed better understanding of the structure-property-performance relationships and precise control of the composition of materials. As a result, considerable developments have been made in the engineering of innovative functional materials, and research is currently addressing numerous fields such as photovoltaics, batteries, electrolytes, supercapacitors, energy conversion, biomaterials, tissue engineering, medical imaging, nanotechnology, microfluidics, and computer modeling, to name just a few. Functional materials are therefore expected to have a considerable impact on many aspects of our lives such as energy, transportation, life sciences, and environment. As applied research tends to be a very dynamic and innovation-driven field, writing a book on functional materials is therefore a never-ending enterprise, as a new technology appears on the market every month. This book is designed in such a way that the reader will learn about functional materials and the types of applications they are designed for. Thus, it will be possible to correlate emerging functional materials with the underlying concepts that were involved in their design, although these might still be developed in the future. The book is divided into three sections and covers the synthesis and applications of functional materials (Part I), novel materials for energy applications (Part II), and new trends in functional materials for sustainable development and biomedical applications (Part III). Part I comprises six chapters focusing on the various approaches to preparation and synthesis of organic and inorganic functional materials. Since the most promising technologies for increased efficiency and improved reliability involve controlling the composition and the microstructure of novel materials, it is essential for the reader to understand the underlying concepts directing the synthesis and the generation of their assembly. In Chapter 2, the fundamental concepts of colloids and polymerization in dispersed medium are introduced. A colloid is a dispersion of very fine objects in a fluid, these objects being solids, liquids or gas, and the corresponding colloidal dispersion being referred to as a suspension, an emulsion or foam. Polymer colloids are used for a large number of applications, ranging from coatings, adhesives, inks, impact modifiers, drug-delivery vehicles, etc. The domain of colloidal stabilization is therefore a fascinating and vibrant area of science and this chapter presents important technological advances combining physical chemistry and fluid dynamics related to colloids and polymer chemistry. Chapter 3 illustrates the tremendous influence fullerene and graphene have had on various technologies in recent years. These

2 | Introduction carbon-based materials are undoubtedly amongst the most-studied materials in both academic and industrial laboratories since the discovery of fullerenes in 1985, carbon nanotubes in 1991, and graphene in 2004. Due to their great structural diversity, numerous applications can be envisioned for these materials, and recent synthetic methods for the design of functional mesostructured materials are presented. Some perspectives of these applications in catalysis, selective sorption, and biomedical devices are reviewed, as well as numerous methods of modification available for modulation of the surface properties, introduction of functionalities, and control of size and shape of the particles of these mesoporous solids. Chapter 4 presents an overview of materials of a porous structure with features in the nanometer range which have emerged as key elements in the development of future technologies including miniaturized electronics, magnetic and optical devices, environmentally-friendly catalysts, materials for pollutant removal, biocompatible implants, and drug delivery systems. The nanopore size range offers vast potential for elaborate functional systems with tailored properties, as the size and volume of the pores in a given material have a profound influence on its final properties, such as adsorption/desorption capability, diffusion mechanisms, storage capacity, density, and mechanical stability. A few examples are presented, such as nanoporous materials used as highly selective sorbents, selective membranes, systems for energy storage or energy conversion, recyclable solid catalysts, low k-dielectrics, sensors, biomaterials for drug delivery, and contrast agents for medical imaging. Chapter 5 focuses on the properties of nanomaterials for the synthesis and design of functional materials based on the fundamental structures of these particles. The fabrication of nanostructures using “top-down” and “bottom-up” approaches is described. As their name implies, top-down methods involve the creation of nanosized entities from larger blocks of matter, including mechanical size reduction by crushing and grinding, as well as more sophisticated lithographic techniques. Bottom-up approaches, on the other hand, seek to build nanostructures from smaller components, typically atoms or molecules, and therefore frequently involve elements of self-assembly or supramolecular chemistry. Chapter 6 covers the synthesis of conjugated polymers for organic electronics applications. These polymers have recently received a great deal of attention since they combine the features of metals or inorganic semiconducting materials (excellent electrical and optical properties), with those of synthetic polymers (flexibility, ease of processing and low cost). This synergy makes these functional materials useful in existing optoelectronic devices and creates completely new technological opportunities. For instance, polymeric semiconductors are considered to be one of the most promising materials for lowering the cost of solar energy. The electrical conductivity of these polymer materials has therefore enabled a wide range of technologies such as photovoltaics, field-effect transistors, and light emitting diodes, which are expected to foster the upcoming era of plastic and printable electronics. Chapter 7 covers the computational approaches and theoretical tools which are used to investigate the composition and performance of various functional materials in silica. This chapter demonstrates that a complex structure can be stud-

1 Introduction

| 3

ied as an assembly of a multitude of building blocks, and that optimization of these functional units can lead to improved performance of the macroscopic properties of a material. Moreover, the versatility of the available chemical processes combined with theoretical modeling enables the design of compounds with well-defined and tunable properties. Part II addresses common challenges regarding the development of functional materials for energy applications including improved performance, conversion efficiency, energy density, and sustainability. Since the efficient use of energy is directly linked to its production and conservation, energy must ideally be stored for further use or made available for portable applications once it has been generated from a readily available source. Thus, there is a tremendous need for electrochemical storage devices such as high energy density batteries and supercapacitors. Similarly, the storage of hydrogen is also of great interest as it represents the principal fuel source for electric vehicles. As fuel cells allow electrochemical oxidation of hydrogen-rich fuel into electrical current, they produce clean energy directly in the form of current and do not require recharging. Therefore, the drive for innovation in sustainable, clean, efficient, and portable energy is now stronger than ever. Chapter 8 discusses electrochemical energy storage systems such as batteries and supercapacitors. The ability to efficiently store and retrieve electrical energy is at the heart of the mobile revolution and is currently of great concern globally. Efficient and low-cost energy storage systems will be required to meet this challenge. Two of the candidates most likely to meet these high power, high energy density requirements are optimized lithium-ion batteries and advanced electrochemical capacitors. This chapter presents the fundamental concepts associated with electrochemical energy storage systems using the thermodynamics, kinetics, structure, and mass transport properties of battery materials. In Chapter 9 the design of ionic liquid electrolytes for lithium-ion batteries is explained to highlight the fact that although the redox behaviour and the kinetics of dissolved species are important, the electrolyte plays a crucial role in the electrochemistry of advanced systems such as energy storage devices. Indeed, the design of the electrolyte will affect the transport of lithium ions and the range of operational voltage, thereby affecting the power and energy density values of the system. Ionic liquids have the potential to play an important role in the development of electrolytes, which will increase the overall performance and improve the efficiency of energy storage devices. Continuing with this topic, Chapter 10 focuses on solid electrolytes, which are materials capable of conducting ions used in many electrochemical devices including batteries, fuel cells, sensors, electrolyzers, and electrodialysis systems for water purification and saltwater desalination. These electrochemical cells are devices capable of producing electrical energy from spontaneous chemical reactions. For example, proton exchange membrane fuel cells are considered attractive power sources for portable applications and for the automotive industry. Nevertheless, these systems still suffer from limitations, discussed in this chapter, hindering the competition of this technology with lead-acid batteries, fossil fuels, and internal combustion engines. Chapter 11

4 | Introduction introduces the basic concepts related to supercritical adsorption of hydrogen on microporous adsorbents. Physical adsorption on microporous adsorbents is widely used in a number of industrial physicochemical processes such as gas separation and purification or catalytic support. Materials-based storage of gases such as hydrogen can be achieved through chemical binding or through absorption inside a solid metal hydride matrix. Activated carbons have been proposed as a means of storing gases such as natural gas at room temperature, and have been widely studied for hydrogen storage applications, due to their availability and the possibility of tailoring their porous structures through various chemical and physical treatments. For hydrogen, storage applications require low temperature operation and highly microporous activated carbon matrices with high specific surface areas, due to the small value of the binding energies between carbon structures and molecular hydrogen. This chapter will cover the thermodynamic description of these systems, which regulates the energy scale required for charging and discharging gases through these devices. Part III groups together the design, synthesis and fabrication of novel functional materials according to new trends in sustainable development and biomedical applications. Chapter 12 gives a detailed overview of advanced materials used in biomedical applications. More specifically, it aims to summarize the fields of implants and prostheses from a materials science and engineering point of view. Since biomaterials are at the interface between medicine, life sciences, and materials science, the number of strategies for their design and engineering is almost infinite, spanning from the control of surface roughness, the conjugation of signal peptides for cell adhesion and drug encapsulation, to biodegradable materials which will be replaced during the healing process. The synthetic and biological materials and devices available to repair or replace damaged tissues and organs are reviewed herein. Following on from the previous chapter, an introduction to the synthesis of functionalized nanoparticles for bioimaging and diagnostic applications is provided in Chapter 13. The development of magnetic nanoparticles as contrast agents for vascular, molecular, and cellular magnetic resonance imaging (MRI) applications has been exponential during the last decade. Nanoparticles generating contrast in MRI is one of the most direct applications of nanotechnology. This chapter is an introduction to the principles underlying the performance of nanoparticulate-based MRI contrast agents. It addresses the main considerations guiding the design and physicochemical characterization of magnetic nanoparticles. The fundamental aspects of nanoparticle magnetism and relaxometric characterization are discussed, as well as examples of applications in biological models. While introducing the fundamentals of microfluidics, Chapter 14 reviews important considerations regarding the synthesis of synthetic and biological functional materials on the microscale. The topics covered range from the introduction of key concepts in microfluidics and reactor fabrication to their utilisation in the study and synthesis of new micromaterials for various applications. Chapter 15 provides a concise description of natural and synthetic polypeptides, from peptides to globular and fibrous proteins, for the development of functional materials. Proteins and pep-

1 Introduction

| 5

tides are promising building blocks for the conception of materials, as they can be tailored for a broad variety of functions, structures and properties, possess the ability to self-assemble into complex architectures, respond to specific environmental stimuli, bind to specific receptors and ligands, and resist mechanical stresses and strains. Through several examples, general principles and important areas of research regarding polypeptide-based materials are described, as well as their considerable potential in various, and possibly unanticipated, application fields. Chapter 16 introduces aspects of chemistry and physics of coatings reinforced with nanoparticles, especially for wood substrates intended for common indoor and outdoor use. Nanoparticle reinforcement can improve resistance to UV degradation, water absorption, and wear and tear without affecting the most important characteristics of paints and coatings: brilliance, color and transparency. This applied research example provides a good illustration of how even the simplest everyday paint or varnish requires extensive research and design in order to be competitive and display optimal properties. Functional materials are at the center of most technology advancements. This book covers three important fields in the development of functional materials: energy, environment, and biomedical applications. These topics are explained and discussed from an experimental and theoretical perspective, as an understanding of material properties is fundamental to the development of novel functionalities and technologies. It is therefore believed that this book will be of great interest to any scientist keen to broaden their knowledge in both the fundamentals and applications of functional materials.

| Part I: Functional materials: Synthesis and applications

A. Al Shboul, F. Pierre, and J. P. Claverie

2 A primer on polymer colloids: structure, synthesis and colloidal stability 2.1 Introduction A colloid is a dispersion of very fine objects in a fluid [1]. These objects can be solids, liquids or gas, and the corresponding colloidal dispersion is then referred to as suspension, emulsion or foam. Colloids possess unique characteristics. For example, as their size is smaller than the wavelength of light, they scatter light. They also offer a large interfacial surface area, meaning that interfacial phenomena are of paramount importance in these dispersions. The weight of each dispersed particle being small, gravity and buoyancy forces are not sufficient to counteract the thermal random motion of the particle, named Brownian motion (in tribute to the 19th century botanist Robert Brown who first characterized it). The particles do not remain in a dispersed state indefinitely: they will sooner or later aggregate (phase separation). Thus, the colloidal state is in general metastable and colloidal stability is one of the key features to take into account when working with colloids. Among all colloids, the polymer colloid family is one of the most widely investigated [2]. Polymer colloids are used for a large number of applications, ranging from coatings, adhesives, inks, impact modifiers, drug-delivery vehicles, etc. The particles range in size from about 10 nm to 1 000 nm (1 μm) in diameter. They are usually spherical, but numerous other shapes have been observed. Polymer colloids are not uncommon in nature. For example, natural rubber latex, the secretion of the Hevea brasiliensis tree, is in fact a dispersion of polyisoprene nanoparticles in water. Synthetic polymer colloids, also called synthetic latexes, play a prominent role in industrial chemistry. Interest in synthetic latexes developed during the Second World War, when the Japanese Navy threatened access to natural Hevea, an important raw material for tire manufacturing at that time. It appeared judicious to produce synthetic polymers under the same aspect, so that downstream operations on the elastomer processing units could remain unchanged. This led to the development of the emulsion polymerization process, one of the most versatile polymerization processes [3, 4]. The words latex, polymer colloids, and dispersed polymer nanoparticles are used interchangeably for any kind of stable colloidal submicronic polymer dispersions in a solvent, which in the majority of cases is water. In this chapter we will present a few salient features of polymer colloid structure, followed by data on the synthesis of these colloids, and finally we will give several key points on colloidal stability.

10 | Part I Functional materials: Synthesis and applications

2.2 Polymer colloids inside out 2.2.1 How many polymer chains per particle? Let us consider a polymeric sphere of diameter (dp ) = 20 nm, constituted of polystyrene (molecular weight 105 g/mol). Thus, the weight of a single polymer chain is m = 105 /Na = 1.7 ⋅ 10−19 g, where (Na ) is the Avogadro number. The volume (V) of one spherical nanoparticle is V = π/6 d3p = 4.2 ⋅ 10−21 l. Knowing that polystyrene density (ρ) is ρ = 1 040 g l−1 , the average number of polymer chains per polymer particle is Vρ/m = 26. The average number of chains per particle for representative particle sizes and polymer molecular weights is listed in Table 2.1. Core-shell particles, whereby a core particle of a given material is engulfed in a shell of another component, form an important class of polymer colloids. For example, hybrid nanoparticles made up of an inorganic core and a polymer shell find wide application in various fields of materials science such as optics, catalysis, microelectronics, biology, and medicine [5]. As another example, let us consider a 20 nm diameter silica particle surrounded by a 5 nm polymeric layer (molecular weight 105 g/mol), which contains 61 polymer chains. At first glance, this number may seem large in comparison to the 26 chains constituting a 20 nm particle. However, it should be remembered that most of the weight of a particle resides in its outer layer. For core-shell particles, the volume of the shell becomes larger than the volume of the core when the shell thickness is 12.5 % of the core diameter. Table 2.1. Number of polymer chains per particle for various polymer nanoparticles. Particle

Outer diameter (nm)

Polymer molecular weight (g/mol)

Number of chains per particle

20

105

26

100

5

3 200

4

Spherical, polystyrene Spherical, polystyrene

10

Spherical, polystyrene

20

10

260

Core-shell Core = silica (20 nm) Shell = polystyrene

30

105

61

2.2.2 How many particles? Three main parameters characterize a polymer dispersion, namely the particle diameter in nanometers (dp ), the solid content (SC), i.e., the weight percentage of solid in the dispersion, and the particle number (Np ), which is the number of polymer particles per unit volume of dispersion. These three parameters are interrelated and they

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 11

cannot be varied independently of one another. Simple geometrical arguments can be used to demonstrate that: SC Np = 6 ⋅ 1025 3 , (2.1) dp ρ where ρ is the density in g/l of the polymer particle in water. Usually, Np will consist of between 1013 and 1019 particles/l, which correspond respectively to dilute and concentrated dispersions. The final use of the polymer dispersion dictates what the values of Np , SC, and dp should be. For example, when the polymer dispersion is dried to form a polymer film, the solid content should be as high as possible in order for the film to form rapidly. Dispersions with an SC of 60 % and higher are used in the manufacture of aqueous paints (such as acrylic paints). For polymers with a density close to one, volume fraction and SC are similar. For a monodisperse particle distribution (i.e., particles all have the same diameter), the theoretical maximum volume fraction is 74 %, which corresponds to a cubic close-packed arrangement of spheres (also called face-centered cubic, or FCC) [6]. As the nanoparticle arrangement is similar to that of molecules in a crystal, this state is referred to as a colloidal crystal (Fig. 2.1). The periodic arrangement of particles and the voids in between them acts as a diffraction grating for light, with the resulting effect that colloidal crystals are brightly colored. This explains why they are also called photonic crystals [7]. The majority of colloidal crystals are prepared with inorganic materials such as silica, although polymer colloidal crystals have also been developed [8, 9].

500 nm

Fig. 2.1. Scanning electron microscopy images of crystalline assemblies of 100 nm polystyrene nanoparticles (reprinted with permission from [10]).

When latex is dried, water first evaporates until the particles are in close contact (stage I of film formation, Fig. 2.2). If the particles are monodisperse and hard (high glass transition temperature, Tg ), and if the arrangement is devoid of defects (a condition which is practically difficult to achieve), then a colloidal crystal is formed (see Fig. 2.1). If the particles are deformable (low Tg ), the particles are compressed into a rhombic dodecahedral arrangement (stage II). Since Tg is low, the polymer from one particle will diffuse in adjacent particles, resulting in the formation of a continuous film (stage III). The three stages of film formation represent a simplified view of

12 | Part I Functional materials: Synthesis and applications

Stage I water evaporates Stage II particle deforms

Stage III interparticle diffusion Fig. 2.2. Schematic representation of the film formation process from a latex and freeze-fracture transmission electron micrograph image of a latex film at stage II (reprinted with permission from [12]). The marker bar represents 370 nm.

the formation of a film from latex. The film formation process is in fact much more complex [11], a fact with important repercussions in the domain of coating technology. Returning to a dispersion of polymer nanoparticles, we have seen that for a volume fraction of 74 %, monodisperse latex takes the form of a crystal-like solid. Practically, the latex does not flow anymore and has an infinitely high viscosity as soon as the volume fraction reaches 64 % [13, 14]. At this point, the spheres are randomly packed, and long-range motions are impossible. However, free flowing latexes with volume fractions higher than 64 % can nonetheless be obtained. In this case, the latex must have a polydisperse size distribution, whereby small particles can be lodged in the voids created by large particles [15].

2.2.3 Are the chains immobile within the nanoparticle? When Tg of the polymer is above room temperature, the polymer chains are frozen and the chains can be considered immobile within the time frame of the experiment. When Tg is below room temperature, various models such as the Rouse model or the reptation model, can be used to estimate the self-diffusion coefficient (D) of the chains. Let us consider a case for which the value of D = 10−20 m2 /s [16], which corresponds to a low value of D that would either be encountered for a high molecular weight chain, or at temperatures just above Tg . The diffusion length scale (L) can be approximated as L = (DT)0.5 , where (T) is the diffusion time. In a time interval T = 1 month, the diffusion length scale L will be 161 nm, which is larger than the size of the particles that we will consider in this chapter. Thus, above Tg the chains diffuse freely within the time scale of the experiment, whereas below Tg the chains are immobile. So the remaining question is whether the Tg of the nanoparticle corresponds to the Tg of the bulk polymer. Before answering this question, one should consider the environment of the nanoparticle. The solvent in which the nanoparticle is suspended (such as water, for

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 13

example) can act as a plasticizer, resulting in a lower Tg . Hydroplasticization is generally thought to be responsible for a decrease in Tg of approximately 12°C for styrenic and acrylic polymer nanoparticles suspended in water [17]. As a result of the polymer chains being confined in a small nanoparticle, Tg deviates from the bulk value. This so-called confinement effect is due to the combination of interfacial and size effects. Interfacial effects are influenced by the specific interaction between the particle surface and its environment, whereas size effects occur when the particle size is of the same order of magnitude as the characteristic size of the region in which cooperative motion occurs. Although no exact number can be given, one would expect that size effects are important for nanoparticles smaller than 10 nm. Numerous observations on polymeric thin films also confirm that size effects are only significant for films thinner than approximately 20 nm [18]. However, data on the Tg of polymeric nanoparticles are still very much controversial. For example, recent data indicate that in water, the Tg of polystyrene nanoparticles with a diameter of 90 nm is 58°C lower than bulk Tg [19]. The relatively simple picture presented in the above paragraph only applies to simple nanoparticles composed of a single homopolymer. Usually one would prepare, characterize, and utilize nanoparticles made up of a large number of components, either under the form of blends of homopolymers or of copolymers with various architectures. This point is introduced in the following paragraph.

2.2.4 Morphology of polymeric nanoparticles As polymers are rarely miscible, phase separation often occurs within nanoparticles, leading to the formation of a nanostructured object with a specific morphology – that is to say, with a specific arrangement of each of the polymeric phases. Unraveling the morphology of a polymer nanoparticle is not a simple task. It is often achieved by transmission electron microscopy (TEM) with selective staining of one of the phases, although other techniques such as nuclear magnetic resonance (NMR, [20]) or differential scanning calorimetry (DSC) have also been used [21]. Starting with a simple blend of two non-compatible polymers, numerous morphologies can be obtained such as core-shell particles, hemispherical particles like half-moons, particles with various inclusions, etc. (Fig. 2.3). Among all of these, kinetically trapped morphologies should be distinguished from thermodynamic morphologies. The latter is usually attained when the particle has been exposed to temperatures above Tg , as in this case the chain’s mobility is high enough to reach the lowest energy conformation. Thermodynamically speaking, the system is constituted of three phases: two polymers and the continuous phase. In a seminal study on the coalescence of immiscible droplets [22], Torza and Mason defined spreading coefficients Sij as Si = 𝛾jk − (𝛾ik + 𝛾ij ), (2.2)

14 | Part I Functional materials: Synthesis and applications whereby 𝛾ij defines the interfacial tension between the i and j phases. If the continuous phase is numbered 2 and the polymer phase 1 is the one for which 𝛾12 > 𝛾23 , then S1 < 0. There are only three possible sets of values for Si , corresponding to three different equilibrium morphological configurations (Fig. 2.3).

3

1

S1 < 0 S2 < 0 S3 > 0

S1 < 0 S2 < 0 S3 < 0

S1 < 0 S2 > 0 S3 < 0

Complete engulfing

Partial engulfing

No engulfing

Fig. 2.3. Possible equilibrium configurations corresponding to the three sets of Si . The continuous medium is phase 2.

In fact, the analysis by Torza and Mason relies on simplified scenarios. A more rigorous approach, developed by Sundberg et al. [23], indicates that equilibrium morphologies depend not only on the exact values of the respective surface tensions 𝛾ij , but also on the amount of each polymer and the surface coverage of the particle by a surfactant (Fig. 2.4). We will see below that surfactants are indispensable to impart colloidal stability to the polymer nanoparticles. 0%

10%

15%

25%

50%

100%

Fig. 2.4. Predicted morphologies of a polymer particle consisting of two polymers (polystyrene in black and poly-n-butylmethacrylate in grey) suspended in water with various surface coverage by sodium dodecyl sulfate (% surface saturation). Reprinted with permission from [23].

As mentioned above, the main technique for unraveling the morphological features of polymer nanoparticles is TEM. A selective positive stain is used to discriminate one of the polymer phases. For example, osmium tetroxide can be used to stain polymers containing benzene rings or isolated double bonds. For simple cases (such as core shell particles), it is possible to assess the morphology without sectioning the particles in thin layers with a microtome. However, for more complex cases the particles must be microtomed after embedding in a hard resin or the morphology may be wrongly

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 15

250 nm

Fig. 2.5. Simulated TEM pictures of a single composite particle consisting of polystyrene in black and poly-n-butylmethacrylate in grey, microtomed at a thickness of 90 nm, as predicted by the software UNHLATEXTM_EQMORPH (Advanced Polymer Laboratory, University of New Hampshire). Reprinted with permission from [23].

attributed. TEM pictures must also be analyzed with great care. For example, Fig. 2.5 illustrates all possible TEM pictures of a single type of nanoparticle, depending on its preparation and its orientation under the TEM beam. The morphological features of particles consisting of two homopolymers are very different from those obtained from polymers with more complex architectures. Among these, amphiphilic diblock copolymers, which consist of a hydrophobic and a hydrophilic block, have been the most studied [24, 25]. Such copolymers, comprised of a hydrophilic and a hydrophobic block self-assemble in water in order to shield the hydrophobic units from water. Water is then qualified as a selective solvent, as it is a good solvent for one part of the block and a poor solvent for the other one. Organic solvents can also be selective, resulting in reverse structures with hydrophilic units on the inside and hydrophobic on the outside. Several equilibrium morphologies are possible, the most classical ones being polymeric spherical and cylindrical micelles and vesicles (Fig. 2.6). The self-assembly of amphiphilic polymers is dictated by the value of the critical packing parameter (P), where P = V/(a0 lc ), where V is the volume of the hydrophobic block, lc is the end-to-end distance between both extremities of the hydrophobic block, and a0 is the cross-section between both blocks. This parameter, first introduced by Israelachvili to predict the assembly of non-polymeric surfactants [26], is a nondimensional number which measures the curvature of the cone generated by the surface enveloping both blocks. Low values of P are obtained for asymmetric block copolymers formed by a short hydrophobic and a long hydrophilic block. They self-assemble into spherical polymeric micelles, whereby the hydrophilic block is solvated, whereas the hydrophobic one forms the core of the micelle. When the hydrophobic block becomes longer, packing into a sphere is impossible and the micelles become elongated (so-called cylindrical micelles). Eventually, for a high packing number, lamellar phases are formed, whereby two diblock copolymers in a headto-tail conformation generate a locally planar arrangement. These lamellar structures can fold in on themselves to form a vesicle. Polymer vesicles are often called polymersomes [27]. Importantly, at the length-scale of the polymer, the vesicle interface is flat, and the curvature of the interface is nearly null, thus the diameter of the vesicle is in-

16 | Part I Functional materials: Synthesis and applications Spherical micelles

Cylindrical micelles

‘Polymersomes’

lc a0 v

High curvature P≤⅓

Medium curvature ⅓≤P≤½

Low curvature ½≤P≤1

Fig. 2.6. Various self-assembled structures formed by amphiphilic block copolymers in a blockselective solvent. Reprinted with permission from [28].

dependent of the structure of the block copolymer and mainly depends on preparation conditions (stirring, temperature, ionic strength, etc.). The value of the packing parameter can only be used to predict equilibrium morphology. Practically, this condition is achieved when the amphiphilic chains are able to diffuse through the solvent and probe several conformations until they form the structure of lowest energy. This is the case when the amphiphilic polymer has nonzero water solubility (when the self-assembly is performed in water), thus for polymers with a rather long hydrophilic block. In this case, self-assembly (Fig. 2.6) occurs when the concentration of polymers in water is above a critical concentration, called critical assembly concentration (CAC), in reference to the critical micellar concentration (CMC) observed for small molecule surfactants. It should however be mentioned that CAC are usually very low (frequently in the μmol/l or below), whereas CMC are usually higher. Various techniques have been devised to precisely measure CAC, the most classical being monitoring the change of the fine structure fluorescent emission band of an organic fluorophore (pyrene), for various concentrations of polymer [29]. Below the CAC, the polymer chains are dissolved as individual entities, whereas above CAC, self-assembled have a structured form, but a fraction of the polymer chains (at concentration CAC) is solubilized in water. In the latter case, the soluble polymer chains and self-assembled chains exchange during the experiment. Does this mean that selfassembled structures are always dynamic and exchange with dissolved chains? Not at all, it is in fact possible to prepare self-assembled polymer colloids for polymers with virtually zero solubility in the solvent. Various techniques exist and will be described below, but it is important to remember that the resulting self-assembled object may be in a kinetically trapped conformation.

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 17

Star micelles

Crew-cut micelles

Fig. 2.7. Self-assembly of block-copolymers into star polymeric micelles and crew-cut polymeric micelles.

Polymeric spherical micelles, the most commonly encountered self-assembled polymeric object, can thus be classified into two distinct categories using the terminology first introduced by Eisenberg (Fig. 2.7, [30, 31]). Star micelles in water are composed of a long hydrophilic and a short hydrophobic block, are in dynamic exchange with freely dissolved polymer chains, and adopt a thermodynamically favored conformation. They have a low but nonzero CAC, and they form spontaneously when the polymer is added to water. By contrast, crew-cut micelles, consisting of a long hydrophobic and a short hydrophilic block, are kinetically frozen (Fig. 2.7). Their CAC is virtually zero and no dynamic exchange process occurs within a reasonable time period. Lastly, they can only be formed by an indirect method (see below).

2.3 Preparation of polymer nanoparticles There are so many methods of preparing polymer nanoparticles that an entire book would not be sufficient to cover them all. This section will present two main classes of preparation techniques, the first belonging to the class of heterophase polymerizations, and the second to the class of self-assembly. Within heterophase polymerizations [32], three techniques will be introduced: emulsion, miniemulsion and microemulsion polymerizations. Other techniques, such as suspension and dispersion polymerizations will not be covered as they usually lead to micron-size particles. Heterophase polymerization (Table 2.2) is the class of polymerization reactions in which a polymer is formed in a nonsolvent (typically water). Most of these polymerizations are radical polymerizations because free radicals, unlike carbanions and carbocations, do not react with water or other solvents currently employed for polymerization. Emulsion polymerization is a process whereby a monomer which is mostly water insoluble is polymerized in water using a water-soluble radical initiator. The resulting product is a latex, that is to say a dispersion of polymer nanoparticles in water. This is the main

18 | Part I Functional materials: Synthesis and applications polymerization technique for obtaining polymer nanoparticles. Besides conventional emulsion polymerization, miniemulsion polymerization and microemulsion polymerization also lead to the formation of latexes. If the monomer is soluble in the solvent but not the polymer, the process is referred to as dispersion polymerization. If both monomer and radical initiators are insoluble in water, then large particles are obtained – this process is referred to as suspension polymerization. Table 2.2. Main features of heterophase polymerization. Polymerization

Conventional

Emulsion Microemulsion

Miniemulsion

Dispersion

Suspension

Continuous Phase

water

water

water

organic solvents or mixture alcohol/water

water

Monomer

– sparingly soluble in water – in droplets – in particles

– in micelles

– in preformed stable droplets

– soluble in the continuous phase

– low solubility in water – droplets

Initiator

water-soluble

water-soluble

water-soluble or insoluble

soluble in the continuous phase

soluble in the monomer

Dispersant

ionic/nonionic surfactant

ionic/nonionic surfactant

ionic/nonionic surfactant

soluble polymer

hydrosoluble polymer

Product

stable latex 10 to 500 nm

stable transparent latex 10 to 20 nm

stable latex 50 to 500 nm

stable latex 0.1 to 0.5 μm; dispersions < 20 μm

suspension (> 5 μm)

2.3.1 Emulsion polymerization Emulsion polymerization is a remarkably simple experiment to conduct which usually leads to a high yield of monodisperse nanoparticles [2]. Numerous variants exist, such as, for example, the seeded emulsion polymerization, whereby small nanoparticle seeds are used as nuclei for polymerization, thus superseding the need for nucleation. By using several monomers, a very large number of particle morphologies can be attained. Emulsion polymerization is one of the major processes used in industry for the production of polymers. Among the monomers polymerized in emulsion, let us cite the acrylic monomers, fluorinated and chlorinated monomers such as vinylidene chloride, vinyl chloride or tetrafluoroethylene, vinyl acetate and its copolymers, styrene, and butadiene [4].

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 19

– – –

– –

–

Emulsion polymerization presents numerous advantages: As water has a very high heat capacity, the dispersed system allows good control of the exothermicity of the polymerization reaction. The viscosity in the latex remains low, even when SC reaches 50 % or above, in contrast with solution polymerization, where viscosity skyrockets as SC increases. Emulsion polymerization yields high molecular weight polymers, usually higher than solution or bulk polymerization under similar conditions. This is due to the fact that the radicals are compartmentalized in separate particles. Their life span is usually longer than in bulk. The rates of emulsion polymerizations are usually higher than solution polymerizations. Emulsion polymerization is a very handy way of manufacturing sticky rubbers or other viscoelastic compounds, since the particles are surrounded by a layer of emulsifier that prevents them from coagulating. Last but not least, emulsion polymerization does not require any organic solvent.

There is of course a price to pay for all these advantages: – First, latexes are very fragile and are prone to flocculation (coagulation). Latexes are always stabilized by convenient surfactants, necessary to retain the colloidal state (see below). However, the surfactant can be undesirable in the final product. – One key step during the polymerization process is nucleation, which is often found to be highly variable and difficult to reproduce. This often results in latexes presenting variable particle sizes. Let us consider a typical emulsion polymerization recipe: 400 g of an organic monomer, 1 000 g of water, 4 g of a surfactant, for example sodium dodecyl sulfate (SDS), and 1 g of a water soluble radical initiator such as potassium persulfate. As the cmc of SDS is 2.4 g/l at room temperature, then approximately 4 − 2.4 = 1.6 g of SDS are engaged in forming micelles. Since on average each SDS micelle is constituted of 50 molecules, there are 6.7 ⋅ 1019 micelles in the system. These micelles are swollen by the monomer: each gram of micelle can accommodate around 0.5 g of styrene within its core. Furthermore, a small amount of styrene (0.3 g) is simply dissolved in water (styrene has a low but nonzero aqueous solubility). Thus, from the initial 400 g of styrene monomer, approximately 399 g are neither in the aqueous phase nor in the micelles. If the system is not stirred, a separate organic phase will appear on top of the aqueous phase, but under mechanical stirring large droplets (1–10 microns in diameter) stabilized by a small amount of surfactant will form. If the water-soluble radical polymerization initiator is added at this point, emulsion polymerization is triggered. A series of colloidal events ensue which were first described by Harkins [33], and which is referred to as the mechanism of emulsion polymerization (Fig. 2.8).

20 | Part I Functional materials: Synthesis and applications Stage 1

I : Initiator M : Monomer : Surfactant

I

MM M I

M

M+I I

Monomer swollen micelle 2-10 nm MM M M M M MM MM M

M

M

I

Monomer swollen polymer particle Stage 2 M

M

MM M M M M MM MM M

I

MMM M MM M I

MM M I

I

M

I

MM MM M M M M M M MM I

Monomer droplet 1-10 microns Stage 3 M MM M M M M M MM MM M

I

M

I

M MM MM M M M M M M M M MM I

MMM M MM M I

I

Fig. 2.8. The three stages in Harkins’ model for emulsion polymerization.

Stage I: Nucleation The radical initiator, dissolved in water, decomposes to form free radicals which subsequently form free radical oligomers by reacting with the low amount of monomers dissolved in the aqueous phase. Each time a monomer is inserted, the oligomer be-

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 21

comes less and less water-soluble and eventually diffuses to the organic phase. Assuming that the monomer droplets have an average diameter of 10 microns, and using equation (2.1), one finds that there are 2 ⋅ 1012 droplets in the system. The total surface area provided by all these droplets is 650 m2 . By contrast, the 6.7 ⋅ 1019 micelles, with an average diameter of 4 nm, present a surface area of 3 400 m2 . Statistically, the radicals are captured by the micelles which are swollen by monomers, and on polymerization, the micelle is transformed into a growing polymer particle. The formation of polymer particles is referred to as nucleation which, according to this model, occurs in the micelles and is a case of micellar nucleation. Generally, only a small fraction of the micelles are used for particle nucleation, the rest serving as a surfactant reservoir to stabilize growing particles. When no more micelles are present, the nucleation is terminated. At this point there are typically 1014 –1018 particles per liter of emulsion and the overall conversion is of the order of 1–10 %. Nucleation can also take place when the surfactant concentration is below CMC. This is referred to as homogeneous nucleation. In this case, the radical oligomer grows into the aqueous phase; it eventually precipitates, forming a very small particle (so-called primary particle), which serves as a nucleus for a polymer particle. Homogeneous nucleation leads to latexes with a lower Np , usually consisting of between 1013 –1015 particles per liter.

Stage II: Steady state Once nucleation is terminated, Np remains constant. Polymerization occurs within the particle where monomer concentration is high (the polymer is swollen by the monomer). Each time a monomer is polymerized in a polymer particle, another one diffuses from the aqueous phase to the particle and yet another diffuses from a monomer droplet to the aqueous phase. Thus, the monomer droplets act as monomer reservoirs and the monomer concentration in the particle (Cp ) remains constant during this stage. Two antagonistic factors control Cp : the swelling of the polymer particle by the monomer, which is entropically favorable for the polymer, and the increase of particle surface upon swelling by the monomer, which is associated with an increase in surface energy. These two contributions are balanced in the Morton–Kaizermann–Altier equation: 2V0 𝛾 0 = ln(1 − φp ) + φp + χφ2p + , (2.3) RTr where χ is the Flory—Huggins interaction parameter between the polymer and the monomer, r the radius of the particle, φp the volume fraction of the polymer in the particle, and V° the molar volume of the monomer. From this equation, φp can be calculated, followed by Cp (Table 2.3, [34]). During stage II, Cp and Np remain constant. Therefore, the rate of polymerization is constant; this stage is the steady state period of the polymerization. Similar to the nucleation state, radicals formed in the aqueous phase enter the particles at regular intervals. If the particle already contains a radical, then chances are that the two radi-

22 | Part I Functional materials: Synthesis and applications Table 2.3. Monomer solubility in polymer particles (Cp ) and in water (Cw ) for the emulsion polymerization of common monomers. Monomer Styrene Methyl methacrylate (MMA) Butyl methacrylate (BMA) Butyl acrylate (BUA) Vinyl acetate (VAC)

Cp (mol/l)

Cw (mol/l)

5.5 6.6 3.8 5.0 7.5

0.0043 0.15 0.0025 0.0064 0.5

cals, being in close proximity, will terminate. Thus, except for very large polymer particles, the particle can only contain 0 or 1 radical, and the average number of radicals per particle is n = 12 . In certain cases (for example after a transfer to monomer), radicals can exit the particle before another radical entry occurs, and in this case, n ≺ 12 . Large particles can accommodate two or more radicals without termination, and in this case, n ≻ 12 .

Stage III At this point the monomer droplets have disappeared and the monomer concentration in the particle diminishes. Often, a gel effect (Trommsdorff effect) occurs: in short, the polymerization rate increases significantly due to a drastic increase of n (radical termination is slowed by the high internal viscosity of the particle). Furthermore, as the monomer is consumed in the polymer particles, the particles shrink during this stage. Although emulsion polymerization is by far the most widely practised heterophase emulsion polymerization technique, it also suffers from intrinsic limitations. For example, its mechanism implies that the monomer diffuses through the aqueous phase. Therefore, monomers with extremely low aqueous solubility (Table 2.3) are often difficult to polymerize. In this case miniemulsion polymerization should be used.

2.3.2 Miniemulsion polymerization The use of miniemulsions to form polymer nanoparticles was pioneered by Ugelstad and El-Aasser [35]. Miniemulsions are dispersions of critically stabilized oil in water droplets prepared by shearing a system containing oil, water, a surfactant, and a hydrophobe [36–38]. In miniemulsion polymerization, the monomer is emulsified to form a stable miniemulsion ranging in size from 100 to 500 nm. Each of these droplets becomes an independent nanoreactor for the polymerization, and the resulting polymer particles form an exact replica of the initial nanodroplets. One of the key features of miniemulsion polymerization is the colloidal stability of the initial droplets. In conventional oil-in-water emulsions, droplets of various sizes

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 23

are formed and rapidly phase separate into one separate oil-phase when stirring is stopped. Ostwald ripening, the diffusion of the oil phase through the aqueous phase (Fig. 2.9) is responsible for this gradual coarsening of the emulsion. This process is driven by the difference in Laplace pressure between droplets having different radii: the oil diffuses from the smallest to the largest droplets. Besides Ostwald ripening, mass transfer between droplets can occur when they collide. This collision mechanism is most conspicuous in highly sheared systems and leads to either fusion or fission of the droplets. Ostwald ripening

Diffusion through continuous phase Collisions

fission

+

fusion

Fig. 2.9. Schematic representations of Ostwald ripening and the fission and fusion mechanisms of droplets.

Ostwald ripening can be effectively suppressed by the addition of a small amount of a hydrophobic compound (called hydrophobe) to the oil (Fig. 2.10). The hydrophobe cannot diffuse from one droplet to another and is trapped in each droplet. If the emulsified oil was to diffuse from a small to a larger droplet, the hydrophobe concentration would increase in the smaller droplet (Fig. 2.10), which is not favorable thermodynamically due to osmotic pressure increase. Therefore, the presence of the hydrophobe in each droplet efficiently prevents Ostwald ripening. Hexadecane is the most commonly used hydropobe, but numerous other hydrophobes have been shown to successfully prevent Ostwald ripening [36–37]. Similar to direct miniemulsion, osmotic pressure in reverse miniemulsion (water-in-oil emulsion) can effectively be suppressed by an agent soluble only in the aqueous phase, a so-called “lipophobe”, such as an inorganic salt [39].

24 | Part I Functional materials: Synthesis and applications The formulation of a miniemulsion is relatively simple: nearly any type of oil and surfactant can be miniemulsified to lead to metastable droplets with sizes usually ranging from 50–500 nm. The dispersion requires a high-energy mechanical device, usually a probe sonicator is used, but other devices such as high pressure homogenizers (for example a commercial microfluidizer), can also be employed.

H H

H H

H

H

H

H

H H

H H

Fig. 2.10. Suppression of Ostwald ripening by addition of a hydrophobe to the emulsified oil.

Miniemulsion polymerization has become increasingly popular in recent years because it is an extremely versatile tool for preparing polymer nanoparticles and also for encapsulating hydrophobic compounds within polymer particles. Encapsulation of lipophilic drugs [40], bactericides [41], functional oligomers [38], and even inorganic particles [42] has been achieved by this method. However, miniemulsion is not an ideal technique for small nanoparticles (d < 50 nm). Recently, much effort has been made to prepare miniemulsions which don’t require sonication of high-shear homogenization [43].

2.3.3 Microemulsion polymerization In microemulsion, the monomer is initially confined to micelles and no droplets are present. Polymerization occurs within the micelle. It has been estimated that approximately one out of ten micelles is nucleated and becomes a polymer particle, the other micelles serving as monomer reservoir and surfactant reservoir. The resulting dispersion is thus formed of small polymer particles which are nonetheless larger than the original micelles. Their size ranges from 10–50 nm [44–45]. Microemulsion polymerization prevents several specific characteristics which are worth mentioning. Firstly, due to their very small size (Fig. 2.11), the nanoparticle dispersion is translucent and does not significantly scatter light. Polymer molecular weights are also usually very high; for a system composed of small particles, the number of particles is very high (see equation (2.1)), and once a radical enters a micelle, another radical entry into the same micelle leading to termination is unlikely. As a result, most if not all particles contain one radical during polymerization, and as Np is very high, polymerization is extremely rapid.

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 25

EMA

S

BMA

EHMA

Fig. 2.11. Nanolatexes of polyethylmethacrylate (EMA), polystyrene (S), butyl methacrylate (BMA), and 2-ethylhexylmethacrylate (EHMA) obtained by microemulsion polymerization. For these nanolatexes, a cobalt-based catalyst was used to limit polymer molecular weight. Reprinted with permission from [46].

In contrast to metastable miniemulsions, which are only obtained after a high-energy mixing step, microemulsions are thermodynamically stable and form spontaneously. However, their formulation is particularly difficult. They require large amounts of surfactant, often larger amounts than monomer. In the ternary phase diagram (water, monomer, surfactant), the stability region for the microemulsion is extremely narrow. Thus, a slight change in monomer concentration, temperature or other operating parameters can result in loss of the microemulsion [47, 48].

2.3.4 Self-assembly in selective solvents As mentioned above, block copolymers with long hydrophobic sequences form crewcut micelles upon self-assembly in water. However, due to the lack of exchange mechanism, the self-assembly process is not spontaneous and it requires an indirect preparation route. The method generally employed consists of transferring the block copolymer from a good solvent, in which both blocks are soluble, to a selective solvent, in which only one of the blocks is soluble. The main requirement is that the good solvent and the selective solvent be miscible. Among the usual combinations of good and selective solvents, one finds THF/water or DMF/water. Once the self-assembly step is performed, the good solvent is removed, either by dialysis or by repeated centrifugations, although care should be taken to avoid flocculation of the nanoparticles during centrifugation. During the addition of the selective solvent, the solvation environment changes: it is first a good solvent, then an intermediate quality solvent, whereby micellization and rapid exchange between micellar and free polymer occurs, and finally it becomes a highly selective solvent, whereby the micelle configuration is frozen [30, 31]. Ternary phase diagrams for the polymer and

26 | Part I Functional materials: Synthesis and applications both solvents are often extremely complex. Depending on the experimental design (polymer concentration and rate of selective solvent addition), a given trajectory will be followed on the phase diagram when the selective solvent is added [49], and various zones of predominance will be visited. The result of the self-assembly process corresponds to the first structure for which exchange between micellar and dissolved chains is slow relative to the duration of the experiment.

2.4 Colloidal stabilization As mentioned above, polymer nanoparticles are not colloidally stable unless they are stabilized, for example by a surfactant. Several mechanisms can be used to provide colloidal stability; they are based on (1) electrostatic stabilization, (2) steric stabilization, and (3) depletion mechanism. When the system is not colloidally stable it will coagulate or flocculate, resulting in the formation of coagulum, or floc. Both terms refer to the formation of particular aggregates, although for polymer colloids, the word coagulum is usually employed for compact aggregates which cannot be redispersed, whereas flocs are loosely bound aggregates, which in some cases can be redispersed into free flowing colloids. Although the colloidal stabilization of polymer nanoparticles at low solid content does not present any major difficulty, at high solid content (above 50 %), systems are prone to flocculate and one cannot rely on a single mechanism (electrostatic, steric, depletion) to impart colloidal stabilization. As a last note, several manufacturers commercially produce surfactant-free polystyrene latexes, for example as a tool for calibrating light-scattering instruments. Surfactantfree does not mean charge-free (residual charges from the emulsion polymerization initiator are present on the particle surface), and this low number of charges is sufficient to impart colloidal stability.

2.4.1 Electrostatic stabilization Electrostatic stabilization is the predominant strategy used to stabilize polymer colloids in aqueous media. It is effective as soon as charges are immobilized at the surface of the nanoparticle. This can be achieved in a variety of ways: most commonly a charged surfactant will be added to the colloidal dispersion. In the case of the anionic surfactant sodium dodecyl sulfate (C12 H25 –O–SO−3 Na+ ), the hydrophobic tail (C12 H25 ) adsorbs on the surface of the particle, leaving the polar sulfate head (OSO−3 ) at the interface between the particle and the aqueous phase. The sodium counter ion is ‘free’ in water. Thus, the polymer nanoparticle is decorated by negative charges via the adsorption of the surfactant (sulfate heads, Fig. 2.12). Note that in this case, the adsorbed surfactant is in dynamic equilibrium with the dissolved surfactant, therefore they are not static punctual charges per se. However, the mathematical descrip-

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 27

tion of electrostatic stabilization is not affected by the presence of this dynamic equilibrium. The simplest (and inaccurate) explanation for the electrostatic stabilization mechanism is the presence of an electrostatic repulsion between the two negatively charged nanoparticles (assuming that an anionic surfactant is used). This is far from satisfying and more information can be gleaned by considering the electrical potential of a single charged spherical nanoparticle immersed in water (Fig. 2.12). Due to the presence of surface charges, ions of opposite charge are condensed (adsorbed) at the proximity of the particle surface. The liquid layer where these ions are located is called the Stern layer. Further away from the Stern layer, the ions are attracted by their close neighbors, but they can also exchange with freely diffusing ions in the solution. Far away from the particle surface, the spatial distribution of cations and anions is homogeneous (concentration c∞ ), and the attraction (or repulsion) by the charged particle surface is entirely screened. Thus, an intermediate layer exists between the Stern layer and the free solution containing ions of low (but non-zero) mobility, which feel the electrostatic potential generated by the surface charge. It is referred to as the diffuse layer (Fig. 2.12), and the Stern and diffuse layers combined form the double layer. Let us now consider two charged nanoparticles approaching one another. When the diffuse layers interpenetrate, the concentration of ions in the overlap region becomes larger than the concentration of ions which are free in solution, resulting in an increase of osmotic pressure. The system will thus tend to separate the two charged nanoparticles in order to recover osmotic equilibrium. Therefore, the ‘force’ which repels two charged nanoparticles is principally osmotic in nature and is often called an electro-osmotic force. By solving the Poisson–Boltzmann equation using a number of assumptions [50], one finds that the electrical potential (V(r)) generated from the surface is maximal at the surface and decreases exponentially with r, where r is the distance from the surface of the nanoparticle (Fig. 2.12). The characteristic decay length of this exponential is 1/κ, which is referred to as the Debye length of the diffuse layer. The quantity 1/κ is often used as a measurement of the thickness of the double layer. The electrostatic potential has nearly entirely vanished at a distance 2/κ from the surface and thus, the distance at which the surfaces of two approaching nanoparticles start to electrostatically interact is 4/κ. The mathematical expression of the Debye length (Fig. 2.12) is relatively simple and only takes temperature and ionic strength of the medium into account, with a characteristic exponent of −1/2. Thus, the higher the ionic strength, the faster the surface charges will be screened by other charges and the steeper the decay of the electrostatic potential. In Table 2.4, values of Debye lengths are consigned for various electrolytes and concentrations. For divalent and trivalent cations, the Debye length becomes very small, even at moderate ionic strength; thus nanoparticles can approach each other at very close proximity without any electrostatic repulsion. In these conditions of ionic strength, the electrostatic stabilization is ineffective. It is thus not surprising that Al3+ -containing salts such as aluminum sulfate are used industrially as flocculants. Inversely, at very low ionic strength, the diffuse layer ex-

28 | Part I Functional materials: Synthesis and applications tends very far into the solution, and even at low solid content the diffuse layers of each polymer nanoparticle interpenetrate. As the ions and the surrounding hydrating water in the diffuse layer have low mobility, the viscosity of the latex increases significantly. Such an effect is referred to as an electroviscous effect. A spectacular experiment for demonstrating this effect consists of deionizing the continuous phase of a monodisperse latex with an equimassic mixture of acidic and basic ion exchange resins. When the ionic strength is low enough, the latex looks like an opalescent viscous gel which will immediately return to liquid on addition of a pinch of salt.

V(r)

2 

V(r) = 64 RT c∞ ϒ 0 1 K

RTεε0

= F

2 i

–kr

e

k ≈

0.301

c∞ z

2

–ζ

1

Particle Stern layer Diffuse layer

I

i

Buffer concentration c∞

r 1/k Fig. 2.12. Representation of the double layer around a charged nanoparticle (charge = z) and electrostatic potential V(r) (in absolute value) versus distance from the nanoparticle surface (r). The potential depends on the temperature (T), the concentration of ions in the solution (c∞ ), and the term 𝛾o = tan h (zFΨ/4RT) where (Ψ) is the potential at the surface and (F) the Faraday constant. The Debye length (1/κ) depends on the ionic strength, defined as I = Σ(c∞ z2i ), where zi is the respective charge of each ion in solution. The potential (ζ) corresponds to the value of the electrostatic potential at the interface between the Stern and the diffuse layer.

Table 2.4. Debye length in nm versus electrolyte concentration. Electrolyte concentration (mol/l) −5

10 10−4 10−3 10−2 10−1

Debye length (1/κ) in nm NaCl

MgCl2

Al(NO3 )3

96.2 30.4 9.6 3.1 1

55.5 17.5 5.6 1.8 0.5

39.2 12.4 3.9 1.2 0.3

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 29

So far, only the repulsive electro-osmotic force was considered to describe the interaction of charged nanoparticles. However, other interactions are at stake and a simplified description of all the interactions between two charged nanoparticles was proposed by Derjaguin and Landau [51], Verwey and Overbeek [52], in what constitutes the DLVO theory. The interaction potential (Fig. 2.13) between two polymer particles is given by the summation of an attractive van der Waals potential, Va , the repulsive electro-osmotic potential, Vr described above, and a repulsive very short action Born potential, Vb , which prevents the two nanoparticles from interpenetrating. The overall potential presents two minima (Fig. 2.13). A deep primary minimum is located at a short distance which corresponds to the distance at which the two nanoparticles are in close contact. When the particles fall into this minimum, they are (irreversibly) coagulated. A maximum height (Vm ) is present at intermediate distance. Lastly, a shallow secondary minimum is present at large distance The energy difference ΔVf represents the energy barrier that a particle must cross in order to reach the primary minimum. The higher this barrier, the more stable the colloid. It is generally accepted that under normal conditions a barrier of ΔVf = 15 RT is sufficiently high for the colloid to be stable. As the Debye length is shorter with higher ionic strength, the increase in ionic strength is one of the key factors responsible for a decrease of ΔVf and therefore for the loss of colloidal stabilization. Primary maximum

V

V = Vr + Va + Vb 2 

Vr = 64RT c∞ ϒ 0 Vm Vb

Vr

Va =

–AR

1

–kr

e

k

12r

ΔVf r

Va

Secondary mininum

Primary mininum Fig. 2.13. Interaction energy versus H, where H is the distance separating two spheres. Vb is the Born repulsive potential, Va is the van der Waals attractive potential (lower dashed line), and Vr is the electrostatic potential (upper dashed line). The radius of each sphere is R and the composite Hamaker constant (which takes both the sphere and the continuous medium into account) is A.

30 | Part I Functional materials: Synthesis and applications 2.4.2 Steric stabilization Another method of stabilizing latex particles is to physically prevent their approach by introducing a hairy layer of solvated polymer at their surface. If the hairy layers are sufficiently thick, the two spheres will only be able to approach each other up to a distance at which the van der Waals attraction is small relative to thermal energy (kT). For a particle of 20 nm size (Fig. 2.14), a layer a few nanometers thick will be sufficient. However, for larger particles, very thick layers need to be employed as the reach of the van der Waals attraction increases with particle size. Thus, it is easier to stabilize small particles by a steric mechanism. From Fig. 2.14, one may infer the size of the hairy layer which is necessary to impart steric stabilization. It is possible to convert layer thickness into polymer molecular weight, provided specific information on the polymer conformation in the solvent is available. Such information can be gathered experimentally (using viscosimetric or light scattering experiments) or theoretically (via atomistic calculations). The rule of thumb is that the radius of gyration of the chain is Rg ≈ 0.05 MW0.5 . For a particle of diameter 200 nm, a hairy layer of 6.6 nm provides a stabilization of 2 kT, which corresponds to a molecular weight of 17 000 g/mol. When sterically stabilized particles approach each other, interpenetration of the hairy layers results in a restriction of their accessible configurations, which is unfavorable in terms of entropy. Furthermore, the concentration of hairs in the overlap region increases, which results in a local increase of osmotic pressure. To maintain osmotic equilibrium, the solvent will tend to decrease the local concentration of hairs, which results in separation of the particles. Thus, steric stabilization is osmotic in nature [53]. Steric stabilization can be achieved by several means. For example, non-ionic surfactants, consisting of a hydrophobic unit and a pegylated water-soluble polymer can impart steric stabilization. The reader should be aware that these surfactants exhibit a cloud point, that is to say a temperature above which they are not soluble [54]. Indeed, at low temperature, the pegylated chain is soluble in water due to the presence of hydrogen bonds between water and the oxygen atom of the pegylated chain. This represents an enthalpic contribution to the heat of dissolution of the polymer. At higher temperature, the entropic cost associated with the liberation of the hydrogen bonded water molecules offset the enthalpic gain and the pegylated chain becomes insoluble in water. Cloud points are dependent on the nature and concentration of the surfactant and the nature of the ions present in the continuous phase. Another method of achieving steric stabilization relies on the use of water-soluble polymers which are introduced during polymerization and act as chain transfer agents. For example, polyvinyl pyrollidone (PVP) is susceptible to hydrogen abstraction (chain transfer) in α to the carbonyl group. During an emulsion polymerization, the hydrogen abstraction step is followed by the insertion of hydrophobic monomers, resulting in the formation of a graft amphiphilic copolymer on a PVP backbone. This amphiphilic graft polymer, formed in situ during polymerization, serves as steric stabilizer.

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 31

0 –2 –4

Va/kT

–6 –8 –10

dp = 20 nm dp = 200 nm dp = 500 nm

–12 –14 –16 –18 0

5

10

15

20

r (nm) Fig. 2.14. Effect of the particle diameter on the van der Waals attraction between two particles. The potential of interaction was calculated using the analytical expression for van der Waals potential presented in [55]. The horizontal line corresponds to an attraction potential of 2 kT. A Hamaker constant of 0.95 ⋅ 10−20 J, corresponding to the Hamaker constant of polystyrene in water, was used to create this graph.

Steric stabilization is usually considered to be insensitive to ionic strength but sensitive to temperature. Unlike electrostatic stabilization, steric stabilization is also effective in organic mediums, provided the hairy layer is soluble in the solvent. Finally, in aqueous mediums, a very efficient method of stabilizing a colloid consists of the electrosteric mechanism, whereby the hairy layer is constituted of a polyelectrolyte [56], as it combines the features of electrostatic and steric stabilization.

2.4.3 Depletion stabilization In the depletion stabilization mechanism, the stabilizing polymer does not interact directly with the nanoparticle surface, but is simply dissolved in the continuous medium. Thus, the choice of stabilizer is greatly simplified, as its only structural requirement is that it be soluble in the solvent. However, this mechanism is to be handled with great care, as the same molecule can act as either stabilizer or flocculant [57]. The stabilization mechanism occurs when the medium is concentrated in stabilizer (or dilute in nanoparticles). Inversely, depletion occurs when the medium is dilute in polymer (or concentrated in nanoparticles). Practically, stabilization by depletion occurs at stabilizer concentrations which are so high that it is difficult to use under most experimental conditions. The reverse is unfortunately true: it is a common observation that flocculation of a stable dispersion is triggered upon addition of a soluble polymer.

32 | Part I Functional materials: Synthesis and applications When a polymer is dissolved in the solvent, its center of mass cannot approach the surface of the nanoparticle (Fig. 2.15). Therefore, each nanoparticle is surrounded by an imaginary layer which cannot be visited by the center of mass of the dissolved polymer. The thickness of this depletion layer is equal to Rg [58]. When two nanoparticles approach one another, their depletion layers overlap. In the overlap volume, the depletion layer is shared by two particles, with the consequence that solvent has been released outside the depletion layer. Thus, the volume of solvent accessible to the polymer increases. Lowering the polymer concentration results in a decrease of its chemical potential, which is favorable thermodynamically. This is the basis for depletion flocculation.

Rg

Particle

Particle

Particle

Particle

Depletion zone V Depletion stabilisation d/Rg d : Inter-particle distance

2

1

Depletion flocculation

Fig. 2.15. Sketch of the interaction potential between two spheres immersed in a solvent containing a polymer (radius of gyration Rg ) dissolved in the solvent.

Let us now consider the case where the nanoparticle suspension is very dilute. When two nanoparticles approach each other, they will have to first repulse (or exclude) dissolved polymer chains before the depletion layers start to overlap. Compression of the polymer chains starts to occur when the inter-particle distance is equal to 2 Rg . When the inter-particle spacing is further reduced, the conformation of the polymer chain is changed from a random coil to a conformation which is more geometrically constrained and less entropically favorable. Thus, an energy barrier must be crossed in order for the depletion layers to overlap and the particles to flocculate. According to Napper, if this barrier is higher than 20 kT [53], then the colloid is protected from flocculation. However, this is only the case if the concentration of polymer dissolved in

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 33

the solvent is high and if the polymer molecular weight is high. Indicative calculations are presented in [53] for a polystyrene latex stabilized by a non-ionic surfactant with 17 ethylene oxide units. Depletion stabilization occurs when a 1 000 g/mol polyethylene glycol polymer is added at a concentration of 380 g/l or above. Any lower concentration results in destabilization by depletion flocculation. If a 10 000 g/mol polyethylene glycol is used instead, stabilization occurs for concentration greater than 55 g/l. Obviously such high concentrations are impractical, as they will lead to very high viscosity.

2.4.4 Future directions The field of colloidal stabilization is a fascinating and vibrant area of science which combines physical chemistry and fluid dynamics. For example, recently a new form of particle stabilization, named haloing, was discovered. It occurs when an uncharged nanoparticle is immersed in a medium containing very small, highly charged nanoparticles [59]. The domain also has exceedingly important technological repercussions, ranging from advanced materials to drug delivery. The small chapter presented here only scratches the surface of the field, and the interested reader should consult the references listed below.

References [1]

Morrison ID, Ross S. Colloidal dispersions: suspensions, emulsions, and foams. John Wiley & Sons, New York, USA, 2002. [2] Fitch RM. Polymer Colloids: A comprehensive introduction. Academic Press, San Diego, USA, 1997. [3] Lovell PA, El-Aasser M. Wiley: Emulsion polymerization and emulsion polymers. John Wiley & Sons, Chichester, UK, 1997. [4] van Herk, AM. Chemistry and technology of emulsion polymerization. Blackwell, Oxford, UK, 2005. [5] Bourgeat-Lami E. Organic–inorganic nanostructured colloids. J. Nanosci. Nanotechnol. 2 (2002), 1–24. [6] Manoharan V, Pine D. Building materials by packing spheres. MRS Bull. 2004, 91–5. [7] Joannopoulos J, Johnson S, Winn J, Meade R. Photonic crystals: molding the flow of light. Princeton University Press, Princeton, USA, 2011. [8] Edrington AC, Urbas AM, DeRege P, et al. Polymer-based photonic crystals. Adv. Mat. 13(6) (2001), 421–5. [9] Li Z, Wang J, Song Y. Self-assembly of latex particles for colloidal crystals. Particuology. 9 (2011), 559–65. [10] Gates B, Qin D, Xia Y. Assembly of nanoparticles into opaline structures over large areas. Adv. Mat. 11 (1999), 466–9. [11] Keddie JL, Routh AF. Fundamentals of latex film formation, processes and properties. Springer, Dordrecht, the Netherlands, 2010.

34 | Part I Functional materials: Synthesis and applications [12] Wang Y, Kats A, Juhue D, Winnik MA, Shivers RR, Dinsdale CJ. Freeze-fracture studies of latex films formed in the absence and presence of surfactant. Langmuir. 8 (1992), 1435–42. [13] Greenwood R, Luckham PF, Gregory T. The effect of diameter ratio and volume ratio on the viscosity of bimodal suspensions of polymer lattices. J. Colloid Interface Sci. 191 (1997), 11–21. [14] Torquato S, Truskett T, Debenedetti P. Is random close packing of spheres well defined? Phys. Rev. Lett. 84 (2000), 2064–7. [15] Schneider M, Claverie J, Graillat C, McKenna TF. High solids content emulsions. I. A study of the influence of the particle size distribution and polymer concentration on viscosity. J. Appl. Polym. Sci. 84 (2002), 1878–96. [16] Bartels CR, Crist B, Graessley WW. Self-diffusion coefficient in melts of linear polymers: chain length and temperature dependence for hydrogenated polybutadiene. Macromolecules. 17 (1984), 2702–8. [17] Jiang B, Tsavalas J, Sundberg D. Measuring the glass transition of latex-based polymers in the hydroplasticized state via differential scanning calorimetry. Langmuir. 26 (2010), 9408–15. [18] Forrest J, Dalnoki-Veress K, Stevens J, Dutcher J. Effect of free surfaces on the glass transition temperature of thin polymer films. Phys. Rev. Lett. 77 (1996), 2002–5. [19] Zhang C, Guo Y, Priestley RD. Glass Transition temperature of polymer nanoparticles under soft and hard confinement. Macromolecules. 44 (2011), 4001–6. [20] Landfester K, Dimonie VL, El-Aasser MS. The evaluation of the size and the structure of the interphase in composite particles containing a macromonomer studied by solid-state NMR. Macromol. Chem. Phys. 203 (2002), 1772–80. [21] Stubbs JM, Sundberg DC. Measuring the extent of phase separation during polymerization of composite latex particles using modulated temperature DSC. J. Polym. Sci. Pol. Phys. 43 (2005), 2790–806. [22] Torza S, Mason SG. Coalescence of two immiscible liquid drops. Science. 163 (1969), 813–4. [23] Sundberg DC, Durant YG. Latex Particle morphology, fundamental aspects: A review. Polym. Reac. Eng. 11 (2003), 379–432. [24] Förster S, Antonietti M. Amphiphilic block copolymers in structure-controlled nanomaterial hybrids. Adv. Mat. 10 (1998), 195–217. [25] Riess G. Micellization of block copolymers. Prog. Polym. Sci. 28 (2003), 1107–70. [26] Israelachvili JN. Intermolecular and surface forces, revised 3rd edition. Academic Press, Waltham, USA, 2011. [27] Discher BM. Polymersomes: tough vesicles made from diblock copolymers. Science. 284 (1999), 1143–6. [28] Blanazs A, Armes SP, Ryan AJ. Self-assembled block copolymer aggregates: From micelles to vesicles and their biological applications. Macromol. Rapid Comm. 30 (2009), 267–77. [29] Tsitsilianis C, Iliopoulos I, Ducouret G. An associative polyelectrolyte end-capped with short polystyrene chains. Synthesis and rheological behavior. Macromolecules. 38 (2000), 2936–43. [30] Zhang L, Eisenberg A. Multiple morphologies and characteristics of “Crew-Cut ” micelle-like aggregates of polystyrene-b-poly ( acrylic acid ) diblock copolymers in aqueous solutions. Macromolecules. 24 (1996), 3168–81. [31] Zhang L, Eisenberg A. Multiple morphologies of “crew-cut” aggregates of polystyrene-bpoly(acrylic acid) block copolymers. Science. 268 (1995), 1728–31. [32] Antonietti M, Tauer K. 90 Years of polymer latexes and heterophase polymerization: more vital than ever. Macromol. Chem. Phys. 204 (2003), 207–19. [33] Harkins WD. A general theory of the mechanism of emulsion polymerization. J. Am. Chem. Soc. 69 (1947), 1428–44.

2 A primer on polymer colloids: structure, synthesis and colloidal stability | 35

[34] Gilbert RG. Emulsion polymerization: a mechanistic approach. Academic Press, Oxford, UK, 1995. [35] Ugelstad J, El-Aasser MS, Vanderhoff JW. Emulsion polymerization: initiation of polymerization in monomer droplets. J. Polym. Sci.: Polym. Lett. 11 (1973), 503–13. [36] Asua JM. Miniemulsion polymerization. Prog. Polym. Sci. 27 (2002), 1283–346 [37] Schork FJ, Luo Y, Smulders W, Russum JP, Butté A, Fontenot K. Miniemulsion polymerization. Adv. Polym. Sci. 175 (2005), 129–255. [38] Crespy D, Landfester K. Miniemulsion polymerization as a versatile tool for the synthesis of functionalized polymers. Beilstein. J. Org. Chem. 6 (2010), 1132–48. [39] Landfester K, Willert M, Antonietti M. Preparation of polymer particles in nonaqueous direct and inverse miniemulsions. Macromolecules. 33 (2000), 2370–6. [40] Huang C-Y, Chen C-M, Lee Y-D. Synthesis of high loading and encapsulation efficient paclitaxelloaded poly(n-butyl cyanoacrylate) nanoparticles via miniemulsion. Int. J. Pharm. 338 (2007), 267–75. [41] Zhang M, Cabane E, Claverie J. Transparent antifouling coatings via nanoencapsulation of a biocide. J. Appl. Polym. Sci. 105 (2007), 3824–33. [42] Tiarks F, Landfester K, Antonietti M. Preparation of polymeric nanocapsules by miniemulsion polymerization. Langmuir. 17 (2001), 908–18. [43] Cheng S, Ting SRS, Lucien FP, Zetterlund PB. Size-tunable nanoparticle synthesis by RAFT polymerization in CO2 -induced miniemulsions. Macromolecules. 45 (2012), 1803–10. [44] Capek I. Microemulsion polymerization of styrene in the presence of anionic emulsifier. Adv. Colloid Interface Sci. 82 (1999), 253–73. [45] Chow PY. Microemulsion polymerizations and reactions. Adv. Polym. Sci. 175 (2005), 257–98. [46] Smeets NMB, McKenna TFL. The synthesis of translucent polymer nanolatexes via microemulsion polymerization. J. Colloid Interface Sci. 383 (2012), 28–35. [47] Holtzscherer C, Candau F. Application of the cohesive energy ratio concept (CER) to the formation of polymerizable microemulsions. Colloids Surf. 29 (1988), 411-23. [48] Baglioni M, Rengstl D, Berti D, Bonini M, Giorgi R, Baglioni P. Removal of acrylic coatings from works of art by means of nanofluids: understanding the mechanism at the nanoscale. Nanoscale. 2 (2010), 1723–32. [49] Hanley KJ, Lodge TP, Huang C-I. Phase behavior of a block copolymer in solvents of varying selectivity. Macromolecules. 33 (2000), 5918–31. [50] Lyklema J. Fundamentals of interface and colloid science / Volume IV, Particulate colloids. Academic Press, San Diego, USA, 2005. [51] Derjaguin BV, Landau LD. No Title. Acta Physicochim USSR. 14 (1941), 633. [52] Verwey E, Overbeek JT. Theory of the stability of lyophobic colloids. Dover Publications, Dover USA, 1999. [53] Napper DH. Polymeric stabilization of colloidal dispersions. Academic Press, London, UK, 1983. [54] Nozary S, Modarress H, Eliassi A. Cloud-point measurements for salt + poly(ethylene glycol) + water systems by viscometry and laser beam scattering methods. J. Appl. Polym. Sci. 89 (2003), 1983–90. [55] Hiemenz PC, Raj Rajagopalan. Principles of colloid and surface chemistry, Third edition. Table 10.4, p 486. Marcel Dekker, New York, USA, 1997. [56] Fritz G, Schädler V, Willenbacher N, Wagner NJ. Electrosteric stabilization of colloidal dispersions. Langmuir. 18 (2002), 6381–90. [57] Feigin RI, Napper DH. Depletion stabilization and depletion flocculation. J. Colloid Interface Sci. 75 (1980), 525–41.

36 | Part I Functional materials: Synthesis and applications [58] Lekkerkerker H, Tuinier R. Colloids and the depletion interaction. Springer, Dorchester, UK, 2011. [59] Tohver V, Smay JE, Braem a, Braun PV, Lewis J a. Nanoparticle halos: a new colloid stabilization mechanism. Proc. Natl. Acad. Sci. 98, 8950–4, USA, 2001.

S. Rondeau-Gagné and J.-F. Morin

3 Synthesis, functionalization and properties of fullerenes and graphene materials 3.1 Introduction Carbon-based materials are undoubtedly amongst the most studied materials in both academic and industrial laboratories since the discovery of fullerenes in 1985 [1]. Prior to this discovery, three carbon allotropes were known, namely amorphous carbon, graphite, and diamond. These rather abundant naturally occurring materials have been used by human kind for centuries for different applications, ranging from energy production to fabrication of luxury goods. Although very useful, these carbon materials were perceived as low-tech and not much research regarding their physical or chemical properties was undertaken. However, everything changed very rapidly with the discovery of fullerene in 1985 and carbon nanotubes in 1991 [2], and a renaissance of carbon materials took place in many scientific areas. These new carbon allotropes showed very interesting properties, especially electronic ones, due to their delocalized π-electron system. While the research and development related to fullerenes and carbon nanotubes were at their prime, graphene, a layer of graphite one atom thick, was isolated for the first time in 2004 [3], giving new reasons for scientists to be excited about new forms of carbon materials. With such a diversity of structures, a plethora of applications can be envisioned for these materials, including electronic, biomedical, energy, aerospace, water treatment, and so on. In this chapter, we will discuss the synthesis and chemical modifications of two of these materials, more particularly fullerene-C60 and graphene. A brief discussion of their applications in materials science is also presented.

3.2 Fullerenes Discovered in 1985 by Robert Curl, Harold Kroto, and Richard Smalley, fullerenes have brought about a revolution in materials science, chemistry, and physics [4, 5]. With their amazing properties, these carbon clusters have stimulated research and incorporated into different types of devices, often leading to increased efficiency and unique properties [6, 7]. Among this family, C60 is particularly interesting, since it is symmetrical, reactive towards different kinds of reagents, and can be prepared in relatively large quantities (the most abundant of all fullerenes). Therefore, it is not surprising that an increasing number of scientific papers on C60 synthesis and characterization have been published over the past two decades. The chemistry of this new carbon allotrope is wide-ranging and extends from chemical functionalization to encapsula-

38 | Part I Functional materials: Synthesis and applications tion of atoms and molecules. In this section, the chemistry of fullerene C60 and the properties that make it a major building block in modern chemistry will be looked at in detail.

3.2.1 General considerations Fullerenes are a new family of compounds consisting entirely of sp2 -hybridized carbon atoms. From a structural point of view, fullerene C60 is perfectly spherical and consists of twelve pentagons and twenty hexagons. However, depending on their size, fullerenes can adopt different shapes. For example, the shape of C70 is ovoid rather than spherical. Also, fullerenes can have from twenty to a hundred carbon atoms [8–10]. The first mention of fullerenes in the literature was by Osawa in 1970, as he predicted that the corannulene was, in fact, a sub-structure of C60 [11]. However, it was only fifteen years later that his hypothesis was confirmed by Curl, Kroto, and Smalley in an experiment aiming at “[. . . ] understanding the mechanism by which longchain carbon molecules are formed in interstellar space and circumstellar shells” [1]. By vaporizing a rotating graphite target with a laser beam, fullerenes were produced and detected by mass spectrometry (peaks at m/z = 720.7 and 841.8 for C60 and C70 , respectively). The importance of this discovery for materials science and nanotechnology was rewarded by the Nobel committee in 1996 [12].

(a)

(b)

Fig. 3.1. C60 (a) and its Schlegel representation (b).

Although its single and double bonds alternate, C60 is not aromatic. Despite the presence of 32 aromatic rings, C60 has 60 π-electrons, and therefore does not meet the criteria of the Hückel rule: 2(N + 1)2 with N = 0, 1, 2, 3, . . . , for three-dimensional molecules [13]. As an indication of this anti-aromatic character, C–C bond junctions [6 : 6] are slightly smaller (1.40 Å) than the C–C bonds of the [5 : 6] junctions (1.45 Å), meaning that the [6 : 6] junctions and [5 : 6] junctions have double and single bond character, respectively. The van der Waals diameter of C60 is 1.1 nm. Table 3.1 summarizes some interesting properties of this carbon allotrope [14].

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

39

Table 3.1. Some interesting properties of C60 . Inner diameter Crystalline network Density Thermal conductivity Electronic affinity 1st ionization potential Band gap Solubility o-DCB (25°C) Solubility CS2 (25°C) Solubility toluene (25°C)

3.48 Å FCC 1.72 g/cm 0.4 W/mK 2.65 eV 7.58 eV 1.7 eV 27 mg/mL 12 mg/mL 3.2 mg/mL

3.2.2 Synthesis and purification of fullerenes Synthesis of fullerenes can be achieved using various techniques and processes. Generally, these techniques require sublimation of simple carbon feedstock such as gas (methane, ethylene, and acetylene), solvents (methanol, ethanol, and hydrocarbons) or solid (graphite) at high temperature [15]. The formation mechanism of fullerenes is still a source of debate among the scientific community, but the most plausible hypothesis seems to be nucleation of carbon particles [16]. Production of fullerenes always results in the formation of several types of fullerenes, which differ in size. However, C60 and C70 are the two major isomers formed after pyrolysis. Interestingly, it is possible to control the C60 /C70 ratio by changing various synthetic parameters. Nowadays, the most commonly used production route for the large-scale synthesis of fullerene is the flame method. This method requires installations allowing control of multiple parameters, such as pressure, flame temperature, concentration of carrier, gas and oxygen/carbon ratio. The flame production method consists of the sublimation of carbon rods with a flame fed with benzene or acetylene. On sublimation, carbon grows around a center of nucleation and forms different types of fullerenes within the residual soot from which they can then be isolated. For best results, it is important to introduce a certain amount of carrier gas such as H2 or O2 . In the best production conditions, the mass percentage of fullerene collected is about 20%. It is worth mentioning that some companies have been using this method to produce a total of 40 metric tons of C60 per year [17]. Following the production of fullerenes, a problem arises: purification. Indeed, given that the fullerenes obtained have very similar polarity, size, and low solubility in most common solvents, purification so as to obtain pure fullerene clusters is very difficult. Several purification methods, such as high-performance liquid chromatography (HPLC), have been developed in order to isolate the fullerene C60 [18, 19]. This technique is the most frequently used for the purification of different-sized fullerenes. It is also the only known method for separation of larger fullerenes (> C76 ). However, this method is very expensive when hundreds of milligrams or even grams of mate-

40 | Part I Functional materials: Synthesis and applications

N N (DBU) C60, C70, C>76

C70

DBU

C>76

DBU

+ C60 (99,5%)

Fig. 3.2. Complexation of C≥70 with DBU.

rial are needed. In addition, modern chromatography columns cannot separate more than one gram per hour, which makes it a time-consuming method. Thus, purification of large quantities of fullerenes is very difficult. To overcome this problem, a technique using a chemical agent, 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU), has been developed, enabling the isolation of pure C60 with excellent yields [20]. This technique relies on complexation between large fullerenes (≥C70 ) with low reduction potential and DBU, leading to a precipitation of the C≥70 ⋅ DBU complex and leaving the C60 in solution. Figure 3.2 presents the general complexation of C≥70 with DBU.

3.2.3 Chemical and physical properties of C60 Because of its structure, its incredible electronic and electrochemical properties, and its specific reactivity, C60 is very attractive to the scientific community. Its most remarkable property is undoubtedly its triply degenerated LUMO orbital, which can accommodate up to six electrons upon electrochemical reduction [21]. This property makes C60 a unique electron acceptor for different applications, especially in electronics. Thus, as shown in Fig. 3.3, six reversible redox processes can be observed in controlled electrochemical experiments (cyclic voltammetry) [22]. It is important to mention that C70 has a doubly degenerated LUMO orbital which can accept up to four electrons. However, the level of the nondegenerated LUMO + 1 is sufficiently close in energy to accept two additional electrons, which gives C70 similar electronic properties to those of C60 . This remarkable electron affinity is still not fully understood and continues to be investigated by the scientific community. Studies on bowl-shaped subunits of C60 are particularly useful and may help in gaining insights into the electronic properties of fullerenes [23]. The study of fullerene subunits also allows more understanding of their reactivity. As a simple model, the fullerene C60 can be considered a collection of pyracyclene units, as shown in Fig. 3.4. The pyracyclene unit is anti-aromatic (4n πelectrons). By accepting two electrons, anionic units are formed, each having a 4n +2 electrons structure which leads to aromaticity and thereby enhances stability. The formation of cyclopentadienide units is thought to give to C60 its “n-type” character. The

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

41

c60 at –10°c 10 µA

5 µA

–1.0

–2.0

–3.0

Potential (Volts vs Fc/Fc+) Fig. 3.3. Electrochemical reduction of C60 as observed by cyclic voltammetry (CV) [22].

formation mechanism of cyclopentadienide via direct electron transfer or nucleophilic reaction is shown in Fig. 3.4 [24]. Since the electronic properties of fullerenes are those most exploited, several studies have been conducted to control and modify them, especially the LUMO energy level of C60 , in order to improve its performance in various types of electronic devices. This is particularly important in solar cells, in which C60 derivatives are used as electronaccepting units [24]. Modulation of the LUMO energy level of C60 can be accomplished

+2 e–

Nu–

Nu

Fig. 3.4. Formation of cyclopentadienide unit via direct electron transfer or nucleophilic addition.

42 | Part I Functional materials: Synthesis and applications through various covalent and supramolecular functionalizations [25–30]. Generally, the electronic level of the orbitals in a fullerene can be modified by incorporating different electron-accepting or electron-withdrawing groups. A strong electron-donating group increases the electron density of C60 , which thereby increases the LUMO energy level. An electron-accepting substituent has exactly the opposite effect. In addition, a through-space electronic phenomenon called periconjugation also allows modulation of the electronic properties of the fullerene [31]. A through-space electron transfer is possible by an orbital overlapping between a substituent and C60 , resulting in an increase or decrease of the electron density of C60 . Thus, the direct consequence of this change in electron density is the modulation of electronic levels.

3.2.4 Chemical functionalization of C60 Since the discovery of fullerenes several studies have been conducted to improve their solubility in common solvents. As mentioned previously, a second objective of the functionalization of fullerenes is modulation of their properties, mainly the electronic properties. In order to make these changes, chemists have attempted to use conventional tools of organic chemistry. However, functionalization does not proceed with the same ease on every fullerene. In fact, fullerenes of different sizes have different reactivity toward organic reactions, and C60 is one of the most reactive fullerenes. Reaction on its highly curved structure promotes the release of cyclic strain. Moreover, as a result of being perfectly spherical and symmetric, it is possible to perform regioselective reactions on the fullerene cage, which greatly facilitate product purification. The main functionalization methods are [3 + 2] and [4 + 2] cycloaddition reactions, as well as nucleophilic reactions. One of the most useful functionalization reactions on C60 is the [3 + 2] cycloaddition. The most famous reaction is the Prato reaction [32, 33]. Using C60 in the presence of an amino acid, usually a sarcosine derivative and an aldehyde, cycloaddition takes place at a [6,6] junction of the fullerene core, resulting in the formation of a fulleropyrrolidine. The general Prato reaction and its mechanism are shown in Fig. 3.5. The Prato reaction is a very useful reaction for the formation of several derivatives because it has important advantages over other methods. Essentially, this reaction does not require the use of toxic reagents, catalysts or unstable organometallic reagents. Some side products may be formed during the reaction, which usually leads to a 60–80% conversion of C60 into fulleropyrrolidine (30 to 40% yield overall). Properties of the parent fullerenes can be modulated through two side groups, which can be of different chemical nature. This reaction is mainly used in the synthesis of functionalized polymer [34], donor-acceptor complexes [35], functionalization of peptides [36], and synthesis of water-soluble derivatives of C60 [37]. It is noteworthy that fulleropyrrolidines can also be obtained by thermal ring opening of aziridine [38].

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

43

R2 N

O H

1) R1

R2

R1

COOH

N H

2) C60, toluene, reflux

R2 N

R2

OH

N H

O

H

O

+ H

R1

R2 N +

R1

H

O O

H

R1

N R2 + –

R1

Cycloaddition [3+2]

Fig. 3.5. The Prato reaction.

Another widely used type of fullerene functionalization is the [4 + 2] cycloaddition. Such functionalization can be carried out via three main methods: by generating a nucleophilic carbon from a α-halo ester, by the thermal addition of diazo compounds followed by thermolysis (or photolysis), or by a carbene addition. One of these three methods is particularly interesting, and is called the Bingel reaction [39]. This reaction requires the use of a halogenated diester which, in the presence of a strong base and C60 , forms a fullerene derivative bearing a cyclopropane moiety. This type of derivative is called methanofullerene. The general reaction is shown in Fig. 3.6.

O RO O

O

RO

NaH, C60 OR

–

X

O

O OR

RO

O OR

Toluene, reflux

X

Fig. 3.6. The Bingel reaction.

Functionalization of C60 by this method has several advantages. First, given that the synthesis of diester compounds is relatively easy, the reaction is very versatile and can be applied to a variety of compounds. Another advantage is that the ester groups can easily be post-functionalized to form more complex architectures. In addition, the synthetic yields are generally very good (50% or more), and this reaction occurs regios-

44 | Part I Functional materials: Synthesis and applications electively at the [6,6] junctions of the fullerene core. It is noteworthy that the α-halo ester may be formed in situ from a diester and a dihalogen (Br2 or I2 ) in the presence of DBU. Nucleophilic addition is also another very interesting method of C60 functionalization [40–43]. This reaction is much less common in the literature and is often used strictly for synthetic ease. Nonetheless, this method allows the preparation of several types of useful derivatives whose properties and functions depend on the atom directly attached to C60 . In most cases, the nucleophilic species is an acetylenic or benzylic carbon, but it can also be a heteroatom such as oxygen or nitrogen. An example of a nucleophilic substitution reaction on C60 with an acetylide derivative is shown in Fig. 3.7.

R

Li

R

E

+

R2 E

Fig. 3.7. Nucleophilic substitution of an acetylenic carbon on C60 .

Usually, the nucleophilic moiety is formed on a terminal alkyne in the presence of a strong base, such as the non-nucleophilic lithium hexamethyldisilylamide (LHMDS), or n-butyllithium. The resulting carbanion generated can be added to C60 , thus transferring the negative charge onto the cage, more specifically onto the adjacent carbon of the newly formed bond. An electrophilic species must then be added to quench this anion. The electrophilic species can be of various natures such as a proton, an alkyl chain, a benzoyl group, etc. Similar to functionalization methods previously presented, the nucleophilic addition has several advantages. First, this method allows a wide range of derivatives from readily available terminal alkyne derivatives to be obtained. Given that two different side groups (nucleophiles and electrophiles) can be added to C60 in a single reaction, the number of possible combinations is astonishing. Nucleophilic addition also allows the synthesis of derivatives with very good yields (40–60%). Purification of the reaction media is also simplified, as few side products are formed during the reaction. There is, however, the possibility that polyaddition products will be formed. This side reaction can be minimized by controlling the reaction temperature and the number of equivalents of acetylide in the reaction media. The reaction is also very difficult to control, and a difference of one minute between various additions of reagents may significantly affect the reaction yield. Although nucleophilic addition is a promising class of functionalization reactions for fullerenes, another type of reaction leading to water-soluble polysubstituted derivatives has proven to be very useful for biological applications [44, 45]. Experi-

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

45

mental conditions are found to be relatively harsher than in other types of functionalization, but the products obtained have interesting properties and specific reactivity. In a first step, fullerene is halogenated using high-pressure F2 gas for a long period of time. Thereafter, by adding a nucleophile and an organic catalyst, it is possible to carry out nucleophilic substitution on the fullerene cage. However, the rules of nucleophilic substitution are somewhat changed with a fullerene. In fact, given the shape of the C60 cage, the normal SN 2 reaction is geometrically impossible. Thus, the reaction mechanism is still controversial and many studies have been conducted to determine it precisely. Since the reaction requires long exposure to a gaseous halogen, few manipulations are necessary. It is also possible to introduce a wide variety of nucleophilic species, which allows several types of derivatives with completely different properties, depending on the nucleophile chemical nature, to be obtained. However, the exact number of substituents inserted onto the fullerene cage is very difficult to control and may vary significantly, which is a major drawback when pure materials are needed, as is the case for biological applications. Currently, the best estimate of substitution ratio comes from the study by nuclear magnetic resonance (NMR) spectroscopy. Encapsulation of various metals and molecules in the C60 core is also an increasingly popular functionalization method. This technique leads to endohedral species, called endofullerenes, with interesting properties and rather unusual structures [46, 47]. Generally used with higher fullerenes (C80 and above), encapsulation of small molecules and metals is also feasible with C60 [48]. By opening the C60 cage using laser or chemicals, it is possible to introduce small molecules such as H2 or He. It is also possible to introduce lanthanides and metals. These new endohedric fullerenes have a very interesting advantage. Since the introduction of molecules into the core does not alter the electronic structure, there is no loss of π electrons and, thereby, the electronic properties of pristine C60 are conserved [49]. This last point is the main reason why formation of endofullerenes is an increasingly popular avenue for fullerene functionalization, especially in organic electronics.

3.2.5 Applications As shown previously, fullerenes have a number of interesting properties. It is therefore not surprising to see fullerenes in several complex architectures and in a variety of applications ranging from organic electronics to biotechnology. We present below some of the most interesting applications for which fullerenes are becoming important building blocks and components. Given its strong n-type character, C60 is the building block of choice in organic electronics, particularly in organic photovoltaics [50, 51]. In a bulk heterojunction solar cell (Fig. 3.8), a blend made of a p-type, light-harvesting polymer (electron-donor) and a methanofullerene, the C61 -phenylbutyricmethyl ester (PCBM), which serves as

46 | Part I Functional materials: Synthesis and applications electron-acceptor [52–54]. When the polymer absorbs a photon from the solar spectrum, an exciton is formed. This exciton can migrate into the polymer-PCBM blend by means of various photophysical processes and thereby generate an electric current. This type of solar cell is increasingly used, and their conversion efficiency is constantly increasing. In the field of organic photovoltaics, C60 is also increasingly used as an electron acceptor in artificial photosynthesis systems [55, 56]. These types of system are studied to gain insights into the photophysical processes of charge transfer/separation and energy generation. Understanding these phenomena makes the creation of more efficient and green devices to collect solar energy for generating electricity possible. –

+

Aluminum OCH3 O

PEDOT-PSS

ITO Plastic foil

Ligh

(a)

MDMO-PPV PCBM

(b)

PC61BM

Fig. 3.8. (a) Architecture of a bulk heterojunction solar cell and (b) structure of the PCBM methanofullerene commonly used as electron acceptor.

The fullerene C60 is also widely used in molecular electronics. Given its ability to accept multiple electrons reversibly, and its high stability versus redox processes, C60 is a promising building block for the fabrication of monomolecular transistors and the construction of flash memory [57]. Thus, it has been demonstrated that self-assembled monolayers containing ethynyl-bridged C60 can be formed on gold (Fig. 3.9). These monolayers can store two charges in a stable way, making them excellent surface capacitors, which paved the way to surface molecular electronics. Fullerenes are also found in rather unusual applications in nanoscience. These new applications exploit several structural properties of fullerenes, such as their spherical structure. Due to this spherical structure, C60 is often used as a “wheel” in nanocars and nanovehicles (Fig. 3.10) [58, 59]. By creating different models of these nanocars and studying their behavior on a gold surface with a scanning tunneling microscope (STM), it was recently demonstrated that fullerenes do not slide but literally roll over the surface. This result helps gain understanding of molecular movement and how to control it in order to further develop useful nanomachines.

47

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

–2.9

H3C

H3C

H3C

H3C

H3C

H 3C

F3C

MeO

OMe O2N

SH 1

Orbital energy (eV)

–3.0 –3.1 –3.2 –5.4 –5.6 –5.8 SH 2

SH 3

SH 4

SH 5

–6.0

SH 6

C60

3

5

1

2

4

6

Fig. 3.9. Fullerene/thiol-terminated molecules used in flash memory devices [57]. OC10H21

H

OC10H21 H21C10O

H21C10O H21C10O

H

H OC10H21

OC10H21 H21C10O

H21C10O H

OC10H21

2

OC10H21

OC10H21

H21C10O

H21C10O

H21C10O

H

OC10H21

2

H OC10H21

H21C10O

2

H

Fig. 3.10. Some examples of C60 -containing nanocars.

Finally, applications have even been found in medicine and pharmacology for fullerene C60 [60, 61]. Despite significant solubility problems often resulting in accumulation in the kidneys, C60 and its water-soluble derivatives can be used in medicine for the treatment of particular diseases such as AIDS [62], in medical imaging [63], and as an antioxidant [64]. In the case of diseases like AIDS, it has been shown that C60 can cause inhibition of the protease enzyme, which significantly reduces the replication rate of these viruses. In medical imagery, endohedral derivatives represent an interesting avenue, due to the possibility of including heavy ions such as Gd, currently used in magnetic resonance imaging (MRI), in their core. C60 has also demonstrated interesting antioxidant properties and cancer cytotoxicity, mainly due to its ability to generate singlet oxygen.

3.3 Graphene Unlike carbon nanotubes and fullerenes, graphene, an all-carbon 2D polymer, is a naturally abundant material, which makes it very attractive and has prompted a rapid

48 | Part I Functional materials: Synthesis and applications technology market penetration [65–67]. However, graphene does not exist as such, but rather as stacks of sheets commonly known as graphite. The molecular structure of graphite is represented in Fig. 3.11. Graphene exists in different forms, including one-atom thick layer, double layers, and several-layers, and each of them has particular optical and electronic properties which will not be discussed here. In some ways, a one-atom thick graphene sheet can be seen as an unzipped carbon nanotube, although the “graphene nanoribbon (GNR)” would be more appropriate to describe this material in this case, since the unzipping of a carbon nanotube would lead to a graphene sheet with a finite width.

Individual graphene sheet

0.34 nm

Fig. 3.11. Molecular structure of graphite.

Defect-free graphene is a semimetal, zero gap semiconducting material owing to the alternation of single and double carbon bonds which allow electron delocalization all along the structure of the graphene sheet [68]. The high value of electron mobility measured at room temperature (∼ 15 000 cm2 V−1 s−1 ) is a few orders of magnitude higher than silicon, making graphene highly valuable for electronic applications such as transparent conducting electrodes, energy conversion, circuitry, transistors, capacitors, and so on [69]. In order to be isolated, graphene sheets must be separated from each other through exfoliation. This process is not simple, since rather strong van der Waals interactions between the graphene sheets have to be broken, which requires quite a large amount of energy. Graphite can be exfoliated using mechanical methods, such as the famous “scotch-tape” method [3], or by using solvents whose surface energies match that of graphene [70]. However, none of the methods of producing graphene from graphite yield large amount of materials, thus limiting the scope of this strategy for many applications. Another way of producing graphene is to prepare it from the bottom-up using reactive carbon species as feedstock. The so-called “physical

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

49

method” can yield very high quality graphene with tunable properties such as size, shape, and physical properties, although some issues need to be addressed regarding purity and batch-to-batch variability. In order to make highly pure nanographenes (NG) and nanoribbons (GNR), an “all-organic” approach involving synthetic organic chemistry can be used. In the following section, we will briefly describe the more common methods, both “physical” and “all-organic” approaches, for the production of graphene and GNR. We will also present some of the most useful strategies to functionalize them covalently before we end this chapter by briefly presenting different areas of application for graphene.

3.3.1 Production of graphene Physical methods One of the most popular methods of producing graphene from carbonaceous feedstock is chemical vapour deposition (CVD) [71]. In this technique, graphene is produced when a transition metal surface is heated to 750°C and higher, depending on the gas, and exposed to hot hydrocarbon gases such as methane or acetylene. Using the CVD method, large sheets of highly pure graphene, mostly single-layer, can be produced. When proper metallic substrate is used, the graphene sheets can be transferred to other substrates such as silicon for electronic applications. When liquid precursors such as hexane are used as the carbon feedstock on a polycrystalline copper foil, graphene sheets of a few centimeters can be prepared and transferred easily to other substrates [72]. Inspired by the production of carbon nanotubes, researchers used the arc discharge method to produce graphene. This method of production has the advantage of repairing graphene while it is formed, in addition to decreasing significantly the amount of amorphous carbon produced during the process. In this method, a graphite-based anode and cathode are exposed to a 100-ampere current, leading to a discharge which produces highly reactive carbon species. This process can be achieved under hydrogen (H2 ) [73] or air [74] to produce graphene in optimized conditions. Another way of producing graphene is to use epitaxial growth of metal or silicon carbide (SiC) substrate. In the latter technique, SiC is heated to an elevated temperature (> 1 200°C) under vacuum in order to displace the Si atoms, whose sublimation rate is higher than carbon, thus concentrating carbon atoms at the surface. The carbon atoms rearrange at high temperature to form uniform sheets of graphene [75]. Similarly, metallic surfaces can be use as a seed for the growth of graphene of quality, allowing the study of its intrinsic physical properties [76]. Important methods for the production of graphene do not involve the transformation of a carbon feedstock, but rather the transformation of naturally occurring

50 | Part I Functional materials: Synthesis and applications graphite. When graphite is used as a starting material, the challenge is to break the van der Waals interactions between the graphene sheets in order to individualize them. One of the most efficient techniques is the oxidation of graphite to form graphene oxide, followed by a reduction process to produce individualized graphene sheets [77]. This method allows the low-cost production of a very large quantity of graphene at once, although the quality of the final material is far from that obtained by other physical methods. Basically, natural graphite is heated in a strong oxidizing acidic mixture to produce exfoliated sheets of graphene oxide (GO) which can be solubilized or suspended in different solvents, including water (Fig. 3.12; [78]). Then, GO can be reduced using thermal annealing or a chemical reducing agent such as hydrazine, NaBH4, vitamin C or pyrogallol to produce graphene. The quality of the final material depends on the conditions used, but usually a significant amount of oxygen is still present after the reduction process. Thus, this production method can be used for a limited number of potential applications and is unlikely for the production of electronic-grade graphene. For this reason, discussion of this technique will not be extended further here. However, GO is a very useful intermediate in the covalent functionalization of graphene sheet.

Graphite

Oxidation

Exfoliation

Graphene oxide

Fig. 3.12. Preparation of GO from graphite [78].

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

51

Another method which uses graphite as the starting material is exfoliation. The greatest advantage of this technique over the oxidation/reduction process described above is that graphene sheets are not chemically transformed, so the sp2 carbon atom network is not disrupted, thus keeping the band structure and the electronic properties intact. In this regard, exfoliated graphite opens the way to large-scale production of electronic-grade graphene without a complex production setup. The strategy behind this technique is to use a stabilizer which will force the graphene sheets to separate by forming van der Waals interactions that are at least as strong as the intersheet interactions. Hence, stabilizers which show a high affinity for graphene have to be used. Among the most efficient are stabilizers with π-conjugated moieties such as perylene [79] and tetracyanoquinodimethane [80], and a polar group which enables dispersion in a polar solvent. More classical surfactants such as sodium dodecylbenzene sulfonate (SDBS) also give good results [81].

Stabilizer

(Energy)

Graphite

Stabilized Graphene Fig. 3.13. Dispersion of graphene sheets through exfoliation.

Solvent molecules can also be used to exfoliate graphite efficiently. Solvents such as ortho-dichlorobenzene (ODCB) [82], N-methylpyrrolidone (NMP) [70], and perfluorinated solvents [83] are examples of solvents which have been used to disperse graphene sheets. An external source of energy such as heat or ultrasound is often required to make this process work.

All-organic methods Although graphene with a high level of purity can be obtained via the above-mentioned physical methods, none of them allow perfect control over the size and shape of the resulting sheets [84–86]. For this reason, chemists and materials scientists have been attempting for decades to produce well-defined graphene nanosheets with customized properties using of organic chemistry, although research still needs to be done

52 | Part I Functional materials: Synthesis and applications in this area to make the whole process more efficient in terms of time and quantity of materials produced. One way of producing graphene nanosheets or nanoribbons is by preparing a polyphenylene-type dendrimer or polymer which is rigidified by intermolecular crosslinking reactions known as Scholes reactions, as shown in Fig. 3.14 [87–90]. In this case, owing to its finite width, the resulting GNR is a semiconductor. Besides, because alkyl chains can be installed into the periphery of the nanoribbons, the resulting materials are soluble in common organic solvent, thus enabling their characterization. This good solubility also allows the semiconducting GNR to be processed from solution to form thin films on various kinds of substrates, opening the way for the use of GNRs in organic electronics. Many other types of nanoribbons with various band gap and carbon atom configuration have been synthesized using this intramolecular cross-linking approach. In addition to nanoribbons, finite graphene nanosheets can be constructed using the intramolecular cross-linking approach (Fig. 3.15; [91]). Again, the band gap of these semiconducting materials can be directly modulated by the size of the nanosheets. The addition of functional groups such as ester, alkyne, ether, triarylamine, aryl, and transition metals to the periphery of graphene nanosheets is another efficient strategy for modulating the electronic properties [88].

3.3.2 Graphene in energy conversion devices One of the most promising applications of graphene in electronic devices is its use as transparent electrode to eventually replace commonly used semiconductors such as ITO, which suffers from significant disadvantages: rarity of indium, chemical sensitivity, and poor mechanical properties [92, 93]. The high transparency (97.7% for a single-layer sheet), high electron mobility, excellent chemical stability, and mechanical properties of graphene make it a “near-perfect” material for electrode fabrication. The applications that would benefit the most from all the qualities of graphene are undoubtedly organic solar cells (OSCs) and light-emitting diodes (LEDs), in which transparent electrodes are necessary to let the light in to and out of the devices. For example, Peumans and coworkers replaced ITO by solution-processed graphene in P3HT-based bulk heterojunction (BHJ) solar cells and observed that the devices suffered only a slight decrease in conversion efficiency compared to ITO-based devices (η = 0.4% and 0.8% for graphene and ITO, respectively; [94]). The efficiency can be increased (η = 1.2%) when CVD graphene is used, since the sheet resistance is lower than for solution-processed graphene [95]. As for BHJ solar cells, graphene can also be used as different components of DSSCs. More specifically, graphene can be used as a transparent counter electrode to replace the expensive Pt/ITO often used to inject electrons into the electrolytes. When redox couples other than I−3 /I2 are used, graphene counter electrodes show higher catalytic activity than Pt/ITO [96].

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

53

I Br + (HO)2B

Br

a

Br

b

Br

x

x

I 3

2

1+5

4 x= TMS 5 x= H

c

d

x

y 6 z

e

x

y 7 z

Fig. 3.14. Preparation of a graphene nanoribbon through an all-organic process [90].

Besides electrode fabrication, graphene can be used as electron acceptor (or ntype materials) in BHJ solar cells, although its transparency might be an important drawback to consider. By replacing PCBM with graphene in a blend with poly(3octylthiophene) (P3OT), efficiency of η = 1.4% was obtained [97]. Dai and coworkers showed that covalent functionalization of graphene with C60 led to an electron acceptor with improved properties compared to C60 or graphene alone [98]. Graphene oxide can also be used as hole-transporting material (HTL) on top of the ITO electrode in BHJ solar cells. In addition, to change the work function of ITO to better match that of active materials, GO protects the surface of ITO from chemical deterioration due to ambient conditions [93]. In a P3HT-based device, efficiency of up to η = 3.5% was measured, which is almost the same as that measured for a similar device using PEDOT:PSS as the HTL material [99].

54 | Part I Functional materials: Synthesis and applications

1

2 –12 H

4

3 –56 H

5

–106 H

6

Fig. 3.15. Preparation of a graphene nanosheet through an all-organic process [91].

Summary Although carbon nanomaterials have their place in a huge number of different technological applications, one can expect to witness even more research activities in this area in the years to come. Some challenges still need to be overcome in order for these nanomaterials to be widely used in commodity devices. Purity, uniformity, batch-to-batch consistency, processability, and price are important issues which must be addressed in the near future to go beyond the simple academic object of study. Moreover, real efforts have to be made to better control their properties by developing synthetic methods which will allow controlled introduction of chemical functions. This could give rise to new electronic and optical properties from which unique applications could be developed.

References [1] [2] [3]

Kroto, H. W.; Heath, J. R.; O’brien, S. C.; Curl, R. F.; Smalley, R. E. C60 : Buckminsterfullerene. Nature 318, 1985, 162–163. Ijima, S. Helical microtubules of graphitic carbon. Nature 354, 1991, 56–58. Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Electric field effect in atomically thin carbon films. Science 306, 2004, 666–669.

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

[4] [5] [6] [7] [8] [9] [10]

[11] [12] [13] [14] [15] [16] [17] [18] [19]

[20]

[21] [22] [23]

[24] [25]

[26]

55

Walton, H. W. K. D. R. M. The fullerenes, New Horizons for the Chemistry, Physics and Astrophysics of Carbon; Cambridge University Press, 1993. Curl, R. F. Dawn of the fullerenes: experiment and conjecture. Mod. Phys. 69, 1997, 691–702. Wudl, F. Fullerene materials. J. Mater. Chem. 12, 2002, 1959–1963. Thompson, B. C.; Frechet, J. M. J. Polymer–fullerene composite solar cells. Angew. Chem. Int. Ed. 47, 2008, 58–77. Kroto, H.W. Smaller carbon species in the laboratory and space. Int. J. Mass Spectrom. Ions Process. 138, 1994, 1–15. van Orden, A.; Saykally, R. J. Small carbon clusters: spectroscopy, structure, and energetics. Chem. Rev. 98, 1998, 2313–2357. Prinzbach, H.; Weiler, A.; Landenberger, P.; Wahl, F.; Wörth, J.; Scott, L. T.; Gelmont, M.; Olevano, D.; Issendorff, B.V. Gas-phase production and photoelectron spectroscopy of the smallest fullerene, C20. Nature, 407, 2000, 60–63. Osawa, E. Superaromaticity. Chem. Abstr. 74, 1971, 75698. Kroto, H. W.; Heath, J. R.; Obrien, S. C.; Curl, R. F.; Smalley, R. E. Symmetry, space, stars, and C60 (Nobel lecture). Nature, 318, 1985, 162–163. Hirsch, A.; Chen, Z.; Jiao, H. Spherical aromaticity in Ih symmetrical fullerenes: the 2(N + 1)2 rule. Angew. Chem. Int. Ed. 39, 2000, 3915–3917. Taylor, R. The Chemistry of Fullerenes; World Scientific Publishing: Vol. 4, 1995. Kratschmer, W.; Lamb, L. D.; Fostiropoulos, K.; Huffman, D. R. C60 : a new form of carbon. Nature, 347, 1990, 354–358 Howard, J. B.; Lafleur, A. L.; Makarovsky, Y.; Mitra, S.; Pope, C. J.; Yadav, T. K. Fullerenes synthesis in combustion. Carbon, 30, 1992, 1183–1201. Tremblay, J.-F. Fullerenes by the ton. Chem. Eng. News, 81, 2003, 13–14. Klute, R. C.; Dorn, H. C.; Mcnair, H. M. HPLC separation of higher (C84+ ) fullerenes. J. Chromatogr. Sci. 30, 1992, 438–442. Kimata, K.; Hirose, T.; Moriuchi, K.; Hosoya, K.; Araki, T.; Tanaka, N. High-capacity stationary phases containing heavy atoms for HPLC separation of fullerenes. Anal. Chem. 67, 1995, 2556–2561 Nagata, K. D., E.; Kikuchi, Y. and Hashiguchi, M. Efficient and scalable method for [60] fullerene separation from a fullerene mixture: selective complexation of fullerenes with DBU in the presence of water. Org. Process Res. Dev. 9, 2005, 660–662. Guldi, D. M.; Prato, M. Excited-state properties of C60 fullerene derivatives. Acc. Chem. Res. 33, 2000, 695–703. Echegoyen, L.; Echegoyen, L. E. Electrochemistry of fullerenes and their derivatives. Acc. Chem. Res. 31, 1998, 593–601. Wu, T.-C.; Chen, M.-K.; Lee, Y.-W.; Kuo, M.-Y.; Wu, Y.-T. Fragments of C70 or higher fullerenes: synthesis, structural analysis, and inversion dynamics. Angew. Chem. Int. Ed. 52, 2013, 1289–1293. Wudl, F. The chemical properties of buckminsterfullerene (C60 ) and the birth and infancy of fulleroids. Acc. Chem. Res. 25, 1992, 157–161. Kooistra, F. B.; Knol, J.; Kastenberg, F.; Popescu, L. M.; Verhees, W. J. H.; Kroon, J. M.; Hummelen, J. C. Increasing the open circuit voltage of bulk-heterojunction solar cells by raising the LUMO level of the acceptor. Org. Let. 9, 2007, 551–554 Ross, R. B.; Cardona, C. M.; Guldi, D. M.; Sankaranarayanan, S. G.; Reese, M. O.; Kopidakis, N.; Peet, J.; Walker, B.; Bazan, G. C.; Van Keuren, E.; Holloway, B. C.; Drees, M. Endohedral fullerenes for organic photovoltaic devices. Nat. Mater. 8, 2009, 208–212.

56 | Part I Functional materials: Synthesis and applications [27] Rondeau-Gagné, S.; Curutchet, C.; Grenier, F.; Scholes, G.; Morin, J.-F. Synthesis, characterization and DFT calculations of new ethynyl-bridged C60 derivatives. Tetrahedron, 66, 2010, 4230–4242. [28] Rondeau-Gagné, S.; Lafleur-Lambert, A.; Soldera, A.; Morin, J.-F. Ethynyl-bridged fullerene derivatives: effect of the secondary group on electronic properties. New J. Chem. 35, 2011, 942–947. [29] Lafleur-Lambert, A.; Rondeau-Gagné, S.; Morin, J.-F. Synthesis and characterization of a new ethynyl-bridged C60 derivative bearing a diketopyrrolopyrrole moiety. Tetrahedron Let. 52, 2011, 5008–5011. [30] Backer, S. A.; Sivula, K.; Kavulak, D. F.; Frechet, J. M. J. High efficiency organic photovoltaics incorporating a new family of soluble fullerene derivatives. Chem. Mater. 19, 2007, 2927–2929. [31] Hamasaki, R.; Ito, M.; Lamrani, M.; Mitsuishi, M.; Miyashita, T.; Yamamoto, Y. Nonlinear optical studies of fullerene–arylethyne hybrids. J. Mater. Chem. 13, 2003, 21–26. [32] Wudl, F.; Sukuki, T.; Prato, M. Probing the properties of C60 through fulleroids ABC61 . Synth. Met. 59, 1993, 297–305. [33] Maggini, M.; Scorrano, G.; Prato, M. Addition of azomethine ylides to C60 : synthesis, characterization, and functionalization of fullerene pyrrolidines. J. Am. Chem. Soc. 115, 1993, 9798–9799. [34] Cravino, A.; Zerza, G.; Neugebauer, H.; Maggini, M.; Bucella, S.; Menna, E.; Svensson, M.; Andersson, M. R.; Brabec, C. J.; Sariciftci, N. S. Electrochemical and photophysical properties of a novel polythiophene with pendant fulleropyrrolidine moieties: toward “double cable” polymers for optoelectronic devices. J. Phys. Chem. B. 106, 2002, 70–76. [35] Guldi, D. M.; Maggini, M.; Scorrano, G.; Prato, M. Intramolecular electron transfer in fullerene/ferrocene based donor-bridge-acceptor dyads. J. Am. Chem. Soc. 119, 1997, 974–980. [36] Pantarotto, D.; Bianco, A.; Pellarini, F.; Tossi, A.; Giangaspero, A.; Zelezetsky, I.; Briand, J. P.; Prato, M. Solid-phase synthesis of fullerene-peptides. J. Am. Chem. Soc. 124, 2002, 12543–12549. [37] Da Ros, T.; Prato, M.; Carano, M.; Ceroni, P.; Paolucci, F.; Roffia, S. Enhanced acceptor character in fullerene derivatives. Synthesis and electrochemical properties of fulleropyrrolidinium salts. J. Am. Chem. Soc. 120, 1998, 11645–11648. [38] Kuwashima, S. Y.; Kubota, M.; Kushida, K.; Ishida, T.; Ohashi, M.; Nogami, T. Synthesis and structure of nitrene-C60 adduct C60 NPhth (Phth = Phthalimido). Tet. Let. 35, 1994, 4371–4374. [39] Bingel, K. Cyclopropanierung von Fullerenen. Chem. Ber. 126, 1993, 1957–1959. [40] Shirai, Y.; Zhao, Y. M.; Cheng, L.; Tour, J. M. Facile synthesis of multifullerene-OPE hybrids via in situ ethynylation. Org. Let. 6, 2004, 2129–2132. [41] Anderson, H. L.; Faust, R.; Rubin, Y.; Diederich, F. Fullerene–acetylene hybrids: on the way to synthetic molecular carbon allotropes. Angew. Chem. Int. Ed. 33, 1994, 1366–1368. [42] Komatsu, K.; Murata, Y.; Takimoto, N.; Mori, S.; Sugita, N.; Wan, T. S. M. Synthesis and properties of the first acetylene derivatives of C60 . J. Org. Chem. 59, 1994, 6101–6102. [43] Murata, Y.; Motoyama, K.; Komatsu, K. Synthesis, properties, and reactions of a stable carbanion derived from alkynyldihydrofullerene: 1-octynyl-C60 carbanion. Tetrahedron, 52, 1996, 5077–5090. [44] Taylor, R.; Holloway, J. H.; Hope, E. G.; Avent, A. G.; Langley, G. J.; Dennis, T. J.; Hare, J. P.; Kroto, H. W.; Walton, D. R. M. Nucleophilic substitution of fluorinated C60 . J. Chem. Soc. Chem. Comm. 1992, 665–667. [45] Chiang, L. Y.; Bhonsle, J. B.; Wang, L. Efficient one-flask synthesis of water-soluble [60] fullerenols. Tetrahedron, 52, 1996, 4963–4972.

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

57

[46] Shinohara, H.; Yamaguchi, H.; Hayashi, N.; Sato, H.; Ohkohchi, M.; Ando, Y.; Saito, Y. Isolation and spectroscopic properties of scandium fullerenes (Sc2@ C74, Sc2@ C82, and Sc2@ C84). J. Phys. Chem. 97, 1993, 4259–4261. [47] Bethune, S.; Johnson, J. S.; Salem, J. R.; De Vries, M. S.; Yannoni, C. S. Atoms in carbon cages: the structure and properties of endohedral fullerenes. Nature, 366, 1993, 123–128. [48] Murata, M.; Murata, Y.; Komatsu, K. Surgery of fullerenes. Chem. Comm. 2008, 6083–6094. [49] Ross, R. B.; Cardona, C. M.; Guldi, D. M.; Sankaranarayanan, S. G.; Reese, M. O.; Kopidakis, N.; Peet, J.; Walker, B.; Bazan, G. C.; Van Keuren, E.; Holloway, B. C.; Drees, M. Electronic structures and spectral properties of endohedral fullerenes. Nat. Mater. 2009, 8, 2009, 208–212. [50] S.; Neugebauer, H.; Sariciftci, N. S. Conjugated polymer-based organic solar cells. Chem. Rev. 107, 2007, 1324–1338. [51] Brabec, C. J.; Gowrisanker, S.; Halls, J. J.; Laird, D.; Jia, S.; Williams, S. P. Polymer–fullerene bulk-heterojunction solar cells. Adv. Mater. 22, 2010, 3839–3856. [52] Hummelen, J. C.; Knight, B. W.; Lepeq, F.; Wudl, F.; Yao, J.; Wilkins, C. L. Preparation and characterization of fulleroid and methanofullerene derivatives. J. Org. Chem. 60, 1995, 532–538. [53] Chirvase, D.; Parisi, J.; Hummelen, J. C.; Dyakonov, V. Influence of nanomorphology on the photovoltaic action of polymer–fullerene composites. Nanotech, 15, 2004, 1317–1323. [54] Janssen, R. A. J.; Hummelen, J. C.; Lee, K.; Pakbaz, K.; Sariciftci, N. S.; Heeger, A. J.; Wudl, F. Photoinduced electron transfer from π-conjugated polymers onto Buckminsterfullerene, fulleroids, and methanofullerenes. J. Chem. Phys. 103, 1995, 788–793. [55] Imahori, H.; Sakata, Y. Fullerenes as novel acceptors in photosynthetic electron transfer. Eur. J. Org. Chem. 1999, 2445–2457. [56] Kuciauskas, D.; Lin, S.;Seely, G. R.; Moore, S. L.; Moore, T. A.; Gust, D. Photoinduced electron transfer in carotenoporphyrin-fullerene triads: Temperature and solvent effects. J. Phys. Chem. 100, 1996, 15926–15932. [57] Shirai, Y.; Guerrero, J. M.; Sasaki, T.; He, T.; Ding, H.; Vives, G.; Yu, B. C.; Cheng, L.; Flatt, A. K.; Taylor, P. G.; Gao, Y.; Tour, J. M. J. Org. Chem. 74, 2009, 7885–7897. [58] Shirai, Y.; Morin, J.-F.; Sasaki, T.; Guerrero, J. M.; Tour, J. M. Recent progress on nanovehicles. Chem. Soc. Rev. 35, 2009, 1043–1055. [59] Vive, G.; Tour, J. M. Synthesis of single-molecule nanocars. Acc. Chem. Res. 42, 2009, 473–487. [60] Pantarotto, D.; Tagmatarchis, N.; Bianco, A.; Prato, M. Synthesis and biological properties of fullerene-containing amino acids and peptides. Rev. Med. Chem. 4, 2004, 805–814. [61] Jensen, A. B.; Wilson, S. R.; Schuster, D. I. Biological applications of fullerenes. Bioorg. Med. Chem. 4, 1996, 767–779. [62] Bosi, S.; Da Ros, T.; Spalluto, G.; Prato, M. Fullerene derivatives: an attractive tool for biological applications. Eur. J. Med. Chem. 38, 2003, 913–923. [63] Wharton, T.; Wilson, L. J. Highly-iodinated fullerene as a contrast agent for X-ray imaging. Bioorg. Med. Chem. 10, 2002, 3545–3554. [64] Gharbi, N.; Pressac, M.; Hadchouel, M.; Szwarc, H.; Wilson, S. R.; Moussa, F. [60] Fullerene is a powerful antioxidant in vivo with no acute or subacute toxicity. Nano Lett. 5, 2005, 2578–2585. [65] Guo, S.; Dong, S. Graphene nanosheet: synthesis, molecular engineering, thin film, hybrids, and energy and analytical applications. Chem. Soc. Rev. 40, 2011, 2644–2672. [66] Novoselov, K. S.; Fal’ko, V. I.; Gelbert, P. R.; Schwab, M. G.; Kim, K. A roadmap for graphene. Nature 490, 2012, 192–200. [67] James, D. K.; Tour, J. M. Graphene: Powder, flakes, ribbons, and sheets. Acc. Chem. Res. ASAP Article DOI : 10.1021/ar300127r. [68] Mas-Ballesté, R.; Gomez-Navarro, C.; Gomez-Herrero, J.; Zamora, F. 2D materials: to graphene and beyond. Nanoscale 3, 2011, 20–30.

58 | Part I Functional materials: Synthesis and applications [69] Geim, A. K.; Novoselov, K. S. The rise of graphene. Nature Mat. 6, 2007, 183–191. [70] Hernandez, Y.; Nicolosi, V.; Lotya, M.; Blighe, F. M.; Sun, Z.; De, S.; McGovern, I. T.; Holland, B.; Byrne, M.; Gun’ko, Y. K.; Bolanf, J. J.; Niraj, P.; Duesberg, G.; Krishnamurthy, S.; Goodhue, R.; Hutchinson, J.; Scardaci, V.; Ferrari, A. C.; Coleman, J. N. High-yield production of graphene by liquid-phase exfoliation of graphite. Nature Nanotech. 3, 2008, 563–568. [71] For representative example, see: Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. F. Science, 324, 2009, 1312–1314. [72] Srivastava, A.; Galande, C.; Ci, L.; Song, L.; Rai, C.; Jariwala, D.; Kelly, K. F.; Ajayan, P. M. Novel liquid precursor-based facile synthesis of large-area continuous, single, and few-layer graphene films. Chem. Mater. 22, 2010, 3457–3461. [73] Subrahmanyam, K. S.; Panchakarla, L. S.; Govindaraj, A.; Rao, C. N. R. Simple method of preparing graphene flakes by an arc-discharge method. J. Phys. Chem. C, 113, 2009, 4257–4259. [74] Wang, Z.; Li, N.; Shi, Z.; Gu, Z. Low-cost and large-scale synthesis of graphene nanosheets by arc discharge in air. Nanotechnology, 21, 2010, 175602 (4pp). [75] Shivaraman, S.; Barton, R. A.; Yu, X.; Alden, J.; Herman, L.; Chandrashekhar, M. V. S.; Park, J.; McEuen, P. L.; Parpia, J. M.; Craighead, H. G.; Spencer, M. G. Free-standing epitaxial grapheme. Nano Lett., 9, 2009, 3100–3105. [76] Pletikosic, I.; Kralj, M.; Pervan, P.; Brako, R.; Coraux, J.; N’Diaye, A. T.; Busse, C.; Michely, T. Dirac cones and minigaps for graphene on Ir(111). Phys. Rev. Lett. 102, 2009, 056808 (4pp). [77] Chen, D.; Feng, H.; Li, J. Graphene oxide: preparation, functionalization, and electrochemical applications. Chem. Rev. 112, 2012, 6027–6053. [78] Su, C.; Loh, K. P. Carbocatalysts: graphene oxide and its derivatives. Acc. Chem. Res. ASAP Articles DOI : 10.1021/ar300118v [79] Englert, J. M.; Röhrl, J.; Schmidt, C. D.; Graupner, R.; Hundhausen, M.; Hauke, F.; Hirsch, A. Soluble Graphene: generation of aqueous graphene solutions aided by a perylenebisimidebased bolaamphiphile. Adv. Mater. 21, 2009, 4265–4269. [80] Hao, R.; Qian, W.; Zhang, L.; Hou, Y. Chem. Commun. 2008, 6576–6578. [81] Lotya, M.; Hernandez, Y. Aqueous dispersions of TCNQ-anion-stabilized graphene sheets. King, P. J.; Smith, R. J.; Nicolosi, V.; Karlsson, L. S.; Blighe, F. M.; De, S.; Wang, Z.; McGovern, I. T.; Duesberg, G. S.; Coleman, J. N. J. Am. Chem. Soc. 131, 2009, 3611–3620. [82] Hamilton, C. E.; Lomeda, J. R.; Sun, Z.; Tour, J. M.; Barron, A. R. High-yield organic dispersions of unfunctionalized graphene. Nano Lett., 9, 2009, 3460–3462. [83] Bourlinos, A. B.; Georgakilas, V.; Zboril, R.; Steriotis, T. A.; Stubos, A. K. Liquid-Phase exfoliation of graphite towards solubilized graphenes. Small, 5, 2009, 1841–1845. [84] E. T. Chernick and R. R. Tykwinski. Carbon-rich nanostructures: the conversion of acetylenes into materials. J. Phys. Org. Chem., 26, 2013, 742–749. [85] Rondeau-Gagné, S.; Morin, J.-F. Preparation of carbon nanomaterials from molecular precursors. Chem. Soc. Rev. Advance Article DOI: 10.1039/c3cs60210a. [86] Morin J.-F, Oligoyne derivatives as reactive precursors for the preparation of carbon nanomaterials. Synlett, 2013 DOI: 10.1055/s-0033-1339680. [87] Wu, J.; Pisula, W.; Müllen, K. Graphenes as potential material for electronics. Chem. Rev. 107, 2007, 718–747. [88] Dössel, L.; Gherghel, L.; Feng, X.; Müllen, K. The graphene nanoribbons by chemists– nanometer sized, soluble and defect-free. Angew. Chem. Int. Ed. 50, 2011, 1–5. [89] Schwab, M. G.; Narita, A.; Hernandez, Y.; Balandina, T.; Mali, K. S.; De Feyter, S.; Feng, X.; Müllen, K. Structurally defined graphene nanoribbons with high lateral extension. J. Am. Chem. Soc. 134, 2012, 18169–18172.

3 Synthesis, functionalization and properties of fullerenes and graphene materials |

59

[90] Wu, J.; Gherghel, L.; Watson, M. D.; Li, J.; Wang, Z.; Simpson, C.; Kolb, U.; Müllen, K. From branched polyphenylenes to graphite ribbons. Macromolecules 36, 2003, 7082–7089. [91] Simpson, C. D.; Mattersteig, G.; Martin, K.; Gherghel, L.; Bauer, R. E.; Räder, H. J.; Müllen, K. Nanosized molecular propellers by cyclodehydrogenation of polyphenylene dendrimers. J. Am. Chem. Soc. 126, 2004, 3139–3147. [92] Wan, X.; Huang, Y.; Chen, Y. Focusing on energy and optoelectronic applications: A journey for graphene and graphene oxide at large scale. Acc. Chem. Res. 45, 2012, 598–607. [93] Chen, D.; Zhang, H.; Liu, Y.; Li, J. Graphene and its derivatives for the development of solar cells, photoelectrochemical, and photocatalytic applications. Energy Environ. Sci. 6, 2013, 1362–1387. [94] Wu, J.; Becerril, H. A.; Bao, Z. N.; Liu, Z. F.; Chen, Y.; Peumans, P. Organic solar cells with solution-processed graphene transparent electrodes. Appl. Phys. Lett. 92, 2008, 263302. [95] De Arco, L. G.; Zhang, Y.; Schlenker, C.; Ryu, K.; Thompson, M. E.; Zhou, C. Continuous, highly flexible, and transparent graphene films by chemical vapor deposition for organic photovoltaics. ACS Nano 4, 2010, 2865–2873. [96] Kavan, L.; Yum, J. H.; Grätzel, M. Graphene nanoplatelets outperforming platinum as the electrocatalyst in co-bipyridine-mediated dye-sensitized solar cells. Nano Lett., 11, 2011, 5501–5506. [97] Liu, Z.; Liu, Q.; Huang, Y.; Ma, Y.; Yin, S.; Zhang, X.; Sun, W.; Chen, Y. Organic photovoltaic devices based on a novel acceptor material: grapheme. Adv. Mater. 20, 2008, 3924–3930. [98] Yu, D.; Park, K.; Durstock, M.; Dai, L. Fullerene-grafted graphene for efficient bulk heterojunction polymer photovoltaic devices. J. Phys. Chem. Lett. 2, 2011, 1113–1118. [99] Li, S. S.; Tu, K. H.; Lin, C. C.; Chen, C. W.; Chhowalla, M. Solution-processable graphene oxide as an efficient hole transport layer in polymer solar cells. ACS Nano, 4, 2010, 3169–3174.

J. Florek, R. Guillet-Nicolas, and F. Kleitz

4 Ordered mesoporous silica: synthesis and applications 4.1 Introduction Porous materials were initially defined in terms of their adsorption properties, and thus distinguished by their pore size range. Pore size usually refers to pore width, i.e., the diameter or distance between opposite walls in a solid. According to the IUPAC definition [1], porous solids are then divided into 3 classes: microporous (< 2 nm), mesoporous (2–50 nm) and macroporous (> 50 nm) materials. Additionally, the term “nanoporous”, which refers to pores in the nanometer size range (< 100 nm), is increasingly being used. Materials with pores in the nanometer range have emerged as key elements for the development of future technologies including miniaturized electronics, magnetic and optical devices, environmentally friendly catalysts, materials for pollutant removal, biocompatible implants, and drug delivery systems [2–4]. The nanopore size range offers vast potential for the construction of elaborate functional systems with tailor-made properties. As a few examples, nanoporous materials may be used as highly selective sorbents, permselective membranes, systems for energy storage or energy conversion, recyclable solid catalysts, low k-dielectrics, sensors, biomaterials for drug/gene delivery or medical imaging, etc. Furthermore, nanoscale pores enable confinement effects which can restrict the growth of crystals and quantum objects, shift the phase behavior of fluids, or create compatible hybrid interfaces. Nano-reactors able to perform size- and shape-selective chemical conversions are also being developed on the basis of nanoporous solids. In such systems, cooperative/complementary chemical processes may be realized in the confined space of the nanopores, acting as spatially-functionalized cavities in analogy to enzymatic active sites. However, it is important to keep in mind that the size and volume of the pores in a given material have a profound influence on the final properties of the solid, such as adsorption-desorption processes, diffusion mechanisms, storage capacity, size exclusion, confinement, density, mechanical stability, etc. Many types of synthetic nanoporous materials, such as ordered microporous and mesoporous materials, controlled pore glasses, gels, pillared clays, anodic alumina, porous polymers, carbon nanotubes, and so forth, have been intensively studied [5–7]. In general, the main attribute is that nanoporous materials exhibit high surface area, and the only way to generate such materials with the desired surface area is through a structuring of the solid at the nanometer level. In many cases, this is achieved with procedures which rely on structuring through the addition of porogen species or templates, which can be single organic molecules or supramolecular aggregates, most

62 | Part I Functional materials: Synthesis and applications frequently. In addition, some other methods are based on the use of solid templates, namely the nanocasting techniques [8–10]. All of these templating pathways are used to synthesize nanoporous materials with a high level of control over their structural and textural properties. These categories of solids are generally viewed as high performance materials for applications in catalysis, separation or storage. As discussed by Schüth and Schmidt [11], their regular pore system with its exceedingly high surface area can be used to introduce guests (e.g., molecules, clusters, macromolecules, or particles) that are stabilized by the solid framework and spatially organized, which enables the development of functional materials. In particular, the use of supramolecular assemblies (i.e., micellar aggregates) as structure-directing agents (SDAs) allowed the synthesis of a new family of mesoporous silica and aluminosilicate compounds, which was designated M41S [12, 13]. These solids were the first examples of solids exhibiting ordered arrangements of mesopores with a narrow pore size distribution. This discovery was a major breakthrough, which then opened up a whole field of research, and great new possibilities in many areas of chemistry and materials science. In this chapter, the authors focus first on general aspects related to the synthesis and functionalization of ordered mesoporous silica (OMS) materials. Then, a few recent developments concerning the use of functionalized mesoporous silica will be presented, with special emphasis on modern drug delivery and separation applications.

4.2 Ordered mesoporous silica (OMS) In the area of ordered mesoporous materials, silica-based systems are still the most widely studied. There are several reasons for this: the great variety of possible structures, the precise control of the hydrolysis-condensation reactions possible due to the lower reactivity (compared to transition metals for example), enhanced thermal stability, and a great variety of functionalization methods available. In addition, mesoporous silica is fairly biocompatible, which is of great importance for biomedical applications (see below, Section 4.5). The characteristic approach for the synthesis of ordered mesoporous materials is the use of liquid crystal-forming templates which enable the specific formation of pores with a predetermined size. Among the different materials reported in 1992 by the scientists at Mobil Corporation [12], the one named MCM-41 (Mobil Composition of Matter №41) exhibits a highly ordered hexagonal array of cylindrical mesopores with a relatively narrow pore size distribution. This new type of solid is thus characterized by periodic arrangements of pores, but the framework pore walls are built of amorphous silica. In general, the combination of powder X-ray diffraction (XRD), transmission electron microscopy (TEM), and gas physisorption analysis enables reliable characterization of ordered mesoporous materials (Fig. 4.1). In particular, the hexagonal arrangement of uniform pores of MCM-41 can be clearly visualized by TEM (Fig. 4.1(a)).

4 Ordered mesoporous silica: synthesis and applications

|

63

The Mobil synthesis performed in alkaline medium led to three well-defined structures: MCM-41, MCM-48, and MCM-50. Vartuli et al. [14] showed that the surfactantto-silica mole ratio is a critical variable in the formation of M41S materials. Using tetraethoxysilane (TEOS) with cetyltrimethylammonium chloride (CTAC, C16 H33 (CH3 )3 NCl), they found that progressively increasing the surfactant-to-silica molar ¯ (1–1.5), lamellar (1.2–2), ratio from 0.5 to 2.0 resulted in hexagonal (< 1), cubic Ia3d and uncondensed cubic octamer (> 2) mesoscale structures. The structure of MCM-48 ¯ space group (Fig. 4.1c), which has also been found in the binary belongs to the Ia3d water/cetyltrimethylammonium bromide (CTAB) system [15]. This three-dimensional (3D) porous structure is considered to be bicontinuous with a simplified representation of two 3D mutually intertwined networks of rods [16, 17]. The unit cell parameter measured for cubic MCM-48 usually ranges in between 8 nm and 10 nm. MCM-50 is a lamellar mesostructure in the as-synthesized form. However, removal of the template results in the collapse of this layered structure unless the material has previously been stabilized. Since the discovery of the MCM family in the early 90s, considerable progress has been achieved regarding the synthesis, characterization, and porosity control of ordered mesoporous silica. An impressive diversity of synthesis approaches have been developed, which now enables the formation of various OMS (e.g., MSU [19], SBA-15 [20], SBA-16 [20], FDU [21], KIT-6 [22], etc.) with a great variety of morphological, structural, and textural properties [23, 24]. However, for all the materials, the concepts involved in their synthesis usually remain quite similar and can be seen mainly as a combination of three key features: (1) surfactant, co-surfactant, solvent, and co-solvent types, (2) controlled polymerization of suitable inorganic species, and (3) interactions between the inorganic (silica) precursor and the organic templating agents.

4.2.1 Principle of synthesis For the synthesis of OMS, two main routes are possible. These are based on the interactions between organic moieties, known as structure-directing-agents (SDAs) or templates, and the silica precursors, which ultimately lead to the formation of an ordered hybrid mesophase (as illustrated in Fig. 4.2). After extended polymerization and condensation of the silicates (or poly-silicic acid species for low pH synthesis conditions), mesoporosity is then created through the removal of the template. While the general concept is similar, there are differences between these two pathways: – In pathway A, i.e., the cooperative self-assembly route, the mesophase is created through the addition of the silica precursor to the micellar system. Prior to mixing, no liquid crystalline phase exists; only isotropic micelles are present in the solution. The self-assembly reorganization of the micelles occurs in situ, cooper-

64 | Part I Functional materials: Synthesis and applications 700 Volume adsorbed (cm3/g)

600 500 0.25 Dv(d) (cm3/A/g)

400 300 200

0 0.0

(a)

0.15 0.10 0.05

100

20 nm

0.20

Ads. Des. 0.2

0.00

0 1 2 34 5 678 Pore width (nm)

0.4 0.6 P/P0

0.8

1.0

(b)

2 (c)

3

4 2Θ

5

6

2

3

4 2Θ

5

6

Fig. 4.1. (a) Transmission electron microscopy image of MCM-41. Reprinted with permission from [24]. (b) Nitrogen adsorption-desorption isotherms (−196°C) and corresponding NLDFT pore size distribution for calcined MCM-41 silica (BET surface area: 1 070 m2 g−1 . Total pore volume: 0.92 cm3 g−1 ; NLDFT pore size: 4.1 nm). Reprinted with permission from [18]. (c) Examples of powder X-ray patterns obtained for ordered mesoporous silica mesophases shown with their pore topology. ¯ symmetry. Reprinted On the left: MCM-41 with p6mm symmetry, on the right: MCM-48 with Ia3d with permission from [24].

–

atively with the formation and polymerization of the inorganic network around them, leading ultimately to a highly organized hybrid mesophase. In pathway B, i.e., the true liquid-crystal templating (TLCT), a pre-formed liquid crystalline phase is used as a template for the infiltration of the silica precursors. The polymerization of the inorganic network is then triggered around this micellar organization. This synthesis pathway was successfully used by Attard et al. [25] to synthesize OMS with various mesostructures, and especially mesoporous silica monoliths.

4 Ordered mesoporous silica: synthesis and applications

|

65

Most often, however, the so-called cooperative self-assembly takes place between the templating species and the mineral network precursors with synchronized selfassembly and inorganic network formation, yielding highly organized mesoscopic architectures. This cooperative formation mechanism, which was first proposed by Stucky, Schüth and co-workers, is now largely accepted by most researchers as an explanation for the mesophase formation [24, 26–28].

Liquid solution

Mixture of solution and precipitation

Surfactant

+

Inorganic Species A

Cooperative nucleation

Cooperative aggregation and phase separation

Liquid crystal formation with molecular inorganics

Further polymerization and condensation of inorganics Template

Elimination

Mesoporous framework of final product

Liquid crystal formation

Incorporation of inorganics’ precursor

Transformation of precursors to aimed materials

Template elimination

B

Fig. 4.2. The two main synthesis pathways for the formation of ordered mesoporous materials: A, cooperative self-assembly and B, true liquid-crystal templating. Reprinted with permission from [23].

Cooperative self-assembly and hybrid interfaces In practice, synthesis starts with the dissolution of an amphiphilic molecule, i.e., surfactant that can be ionic or non-ionic, in water. Here, the surfactant-inorganic hybrid mesophase forms cooperatively from the species present in solution which are not in a liquid crystalline state prior to mixing of the precursors. Quaternary cationic surfactants, Cn H2n+1 N(CH3 )3 Br (n = 8–22), are the most common for the synthesis of ordered mesoporous silica materials with pores ≤ 5 nm. The commercially available C16 TAB is widely used for the synthesis of M41S silicas. In order to synthesize materials with larger pores, swelling agents (SAs) can be used, but they generally lead to poorly reproducible syntheses and low-quality materials due to the heterogeneous dispersion of the SAs in the system. Alternatively, amphiphilic non-ionic triblock copolymers were also proposed as self-assembling SDAs, and these have become more and more popular as they allow the design of OMS with larger pores and thicker walls in acidic or neutral media [19, 29]. Nowadays, the commercially available

66 | Part I Functional materials: Synthesis and applications Pluronics™ ((EO)x -(PO)y -(EO)x ) are widely used for the syntheses of large pore ordered mesoporous silicas, e.g., SBA-15, SBA-16 or KIT-6 (vide infra). When dissolved in aqueous solutions, if proper conditions are met, ionic surfactants or block copolymers assemble into isotropic micelles due to their amphiphilic behavior. Alkoxysilanes, e.g., TEOS, are then added to this stock solution. Inorganic silica precursors are then catalytically hydrolyzed producing silanol groups due to the non-neutral nature of the media. Silanols further condense, forming polymeric silicabased species which aggregate around the micellar structures. The cooperative selfassembly process eventually leads to a supramolecular templating, which results in the hybrid mesophase formation (Fig. 4.2A). Achieving a well-defined segregation of the organic and inorganic domains at the nanometric scale plays an essential role in the synthesis of OMS. Indeed, it is the nature of this hybrid interface that governs the assembly process, and thus the overall quality of the resulting material. The key thermodynamic factors affecting the formation of a hybrid interface were identified by Huo et al. [27, 28]. In their model, described as charge density matching, the free energy of the mesostructure formation (ΔGms ) is considered the sum of the free energy contributions of the inorganic-organic interface (ΔGinter ), the inorganic framework (ΔGinorg ), the self-assembly of the organic molecules (ΔGorg ), and a contribution of the solution (ΔGsol ). The final hybrid mesophase consists of the ordered arrangement with the lowest interface energy. As explained previously, in the cooperative assembly route the template concentration may be well below that necessary for obtaining liquid crystalline assemblies or even micelles. Therefore, the creation of a compatible hybrid interface between the inorganic walls and the organic templates (ΔGinter ) is essential for the generation of a well-ordered mesoscopic hybrid structure with appropriate curvature. From the kinetic point of view, the formation of an organized hybrid mesostructure results thus from a well-balanced combination of inorganic polymerization, organization of the SDA, and organic-inorganic phase separation. Hence, two aspects are essential to finetune the mesophase formation: the reactivity of the inorganic precursors (polymerization rate, isoelectric point, etc.) and the interactions involved in generating the hybrid interface. Here, a generalized cooperative mechanism of formation can be described based on the specific (electrostatic) interactions between the inorganic precursor (I) and the surfactant head group (S). Soler-Illia et al. [30] summarized six possible cooperative interaction pathways for the assembly of the hybrid mesophase, which were originally proposed by Huo et al. [27, 28] and later extended by Pinnavaia [19, 31]. As presented in Fig. 4.3, in the simplest case, the ionic surfactant and the inorganic species are oppositely charged, leading to electrostatic interactions (S+ I− or S− I+ ). Since the synthesis of MCM-41 and MCM-48 silicas is carried out in strongly alkaline media, silica oligomers have a negative charge that interacts with the positive charge of the cationic CTAB molecules (S+ I− mechanism). These two direct routes can be completed by two indirect ones, i.e., interactions between surfactants and inorganic moieties with the same charges through the use of a counter-ion [28]. For ex-

4 Ordered mesoporous silica: synthesis and applications

S–I+

M

O

O–

OH M

O H H

O

M

+

O–

O

+ M O H H

M O–

S+X–I+

M

O

S–M+I–

O

O H H

M O

H+ H

O– M

+

M O– O

SI

M

O

O

S (IX)

O

OH

M OH

+

O H O H H+ M O H

M O H+

M OH

M

OH

O

O

O

M

O

O–

O

O O

M

O

M

O

OH M

M

O

O H+

M

O

O

M

O

67

M

O

O H+

M

O

S+I–

|

Fig. 4.3. Schematic representation of the different types of organic-inorganic hybrid interfaces for mesostructure formation. S corresponds to the surfactant species, and I to the inorganic framework. M+ and X− are corresponding counter-ions. Solvent molecules are represented as triangles. Reprinted with permission from [30].

ample, the S+ X− I+ pathway can explain the mesophase formation under acidic conditions and in the presence of halide anions (X− = Cl− , Br− ), whereas the S− M+ I− route is characteristic of base-catalyzed synthesis in the presence of alkaline metal ions (M+ = Na+ , K+ ). Moreover, following the early development of the so-called MSU synthesis which was performed in neutral conditions using non-ionic block copolymers, a novel assembly approach was proposed and denoted No Io [19]. In this pathway, mesophase formation is driven by hydrogen-bonding forces. Later on, Zhao et al. proposed the use of non-ionic triblock copolymers (i.e., the Pluronics) and acidic conditions to synthesize SBA-15 silica and other related large pore materials [20]. However, in this case, as the silicic acid species are positively charged and the block copolymer could also be positively charged, the No Io route was thus derived to a (No H+ )(X− I+ ) pathway. Finally, a last concept, the S0 I0 interaction is associated with materials called HMS [31], where silicate species interact with a neutral SDA through hydrogen-bonding between the hydroxyl groups of hydrolyzed silicate species and polar amine head groups.

Hydrothermal treatment and template removal Most common syntheses are carried out in aqueous media at temperatures close to room temperature (< 60°C), depending on the type of material. After this step, a hydrothermal treatment (HT) or aging can be performed for over a few hours or up to several days in order to achieve a more complete condensation of the silica frame-

68 | Part I Functional materials: Synthesis and applications work. Even if this aging step is not always mandatory, it usually greatly improves the quality and organization of the resulting OMS [32, 33]. Moreover, mesophase tailoring can be performed during this treatment [22, 34, 35]. Even mesophase transitions are possible in some cases, because variations in the charge density of the silicatebased framework may occur upon condensation [26]. Furthermore, various organics (i.e., co-surfactants) can also be added to the mother liquid in order to modulate the “soft” mesostructure of the material during this stage [24]. After aging, the resultant product is cooled, recovered by filtration or centrifugation if the particles are nanosized, and properly dried in air. The mesoporous material is finally obtained after the removal of the organic template. The most common method for the elimination of the organic moieties is calcination (at temperatures usually above 500°C)[36]. Calcination allows the complete removal of organic species from surfactant-templated silicas. However, when as-made materials containing large amounts of organics are calcined, some carbon deposits or coke formation may be observed. To avoid this contamination, calcination should be performed under sufficient air flow with slow heating rates (1°C min−1 ) and an extended period of heating once the 500–550°C plateau is reached (4 to 8 h). Calcination is the most efficient and convenient method for the removal of organic templates, but the silanol condensation which occurs during this process has two major impacts on the resulting mesostructure: (1) a pronounced decrease of the unit cell of the material (shrinking), and (2) an increase in surface hydrophobicity. Framework condensation can be limited by adjusting the temperature and the duration of the hydrothermal treatment [36]. Furthermore, for copolymer-templated materials especially (e.g., SBA-type and KIT-6 silicas), a brief extraction step using an ethanol/HCl mixture can be performed prior to calcination, in order to improve efficiency and limit detrimental exothermic effects [37, 38]. In addition, the surface silanol density can be easily restored after calcination by performing controlled acidic (aqueous) treatments [39, 40], in order to carry out more efficient post-grafting procedures [41]. Note that calcination is often recommended when OMS are designed for biomedical applications because it removes all trapped SDA organic species. This aspect is of tremendous importance because residual amounts of free surfactants (e.g., CTAB), which typically remain after liquid extraction, were found to be toxic [42, 43]. On the other hand, calcined OMS were found to be “reasonably” biocompatible [44, 45]. Other methods for template removal include liquid extraction [46], acid treatments [47], H2 O2 oxidation under microwave irradiation [48], supercritical CO2 extraction [49] and ozone treatment [50]. Each method causes noticeable variations in the final properties of the porogen-free materials. For example, prolonged (12–24 hours) or multiple liquid extraction steps (e.g., Soxhlet extraction) can be used in the case of organic-inorganic hybrid materials in which some organic functionalities must be preserved after template removal. In contrast, H2 O2 oxidation under microwave irradiation cannot be used with functionalized organic–inorganic hybrids, but it enables complete removal of the template while generating porous materials with higher pore

4 Ordered mesoporous silica: synthesis and applications

| 69

volume and a more hydrophilic surface than calcined counterparts, and this in a very short time (< 15 min) [51].

4.2.2 Mesostructure diversity and tailoring Tailoring of the textural and structural properties of the material, i.e., pore size, pore shape, and connectivity, is essential, especially regarding the potential application of OMS in catalysis, selective sorption, sensing technologies, and so on. Various parameters may be tuned, but the most important are the choice and ratios of reactants, synthesis time and temperature, and the use of additives. By carefully adjusting these parameters, a large diversity of materials can be synthesized. The most frequent ordered mesoporous silica materials are compiled in Table 4.1, along with their structural characteristics. Table 4.1. Structural parameters of the most common ordered mesoporous silicas (OMS). OMS

SDA system

Type of interaction

Pore mesostructure

Typical References pore size (nm)

MCM-41

Cn (CH3 )3 N+ Br− or Cl− 8 ≤ n ≤ 18

S+ I−

2D hexagonal p6mm

1.5–10.0

[12–14]

MCM-48

Cn (CH3 )3 N+ Br− or Cl− 12 ≤ n ≤ 18

S+ I−

¯ 3D cubic Ia3d

1.5–4.6

[14, 26, 37]

SBA-1

Cn (C2 H5 )3 N+ Br− or Cl 12 ≤ n ≤ 18

S+ X− I+

¯ 3D cubic Pm3n

1.5–3.0

[27, 28]

SBA-3

Cn (CH3 )3 N+ Br− or Cl 12 ≤ n ≤ 18

S+ X− I+

2D hexagonal p6mm

1.5–3.5

[33, 52]

SBA-12

Brij 76 (C18 EO10 )

(So H+ )(X− I+ )

3D hexagonal (intergrowth)

3.0–5.0

[53, 61]

SBA-15

P123 (EO20 PO70 EO20 )

(No H+ )(X− I+ )

2D hexagonal p6mm

4.0–15.0

[20, 37, 61]

SBA-16

F127 (EO106 PO70 EO106 ) (No H+ )(X− I+ ) F127 + P123 F127 + BuOH

3D cage-like cubic ¯ Im3m

4.7–12.0

[35, 54, 61]

KIT-6

P123 + BuOH

¯ 3D cubic Ia3d

(No H+ )(X− I+ )

4.0–12.0

[22, 55]

FDU-1

B50-6600 (EO39 BO47 EO39 )

(N H )(X I )

3D cage-like cubic ¯ Fm3m

8.0–14.0

[21, 56]

FDU-12 or KIT-5

F127 F127 + TMB

(No H+ )(X− I+ )

3D cage-like cubic ¯ Fm3m

6.0–12.5

[57, 58]

MSU-H

P123

No Io

2D hexagonal p6mm

7.5–12.0

[59, 60]

TMB = trimethylbenzene

o +

− +

70 | Part I Functional materials: Synthesis and applications Triblock copolymer-templated large pore silica SBA-15 (Santa Barbara Acid №15). A breakthrough in the preparation of ordered mesoporous silica was made by Zhao and Stucky in 1998 [20, 61], who used poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) triblock copolymers for the synthesis of the large pore SBA-15 material. The synthesis is simple and based on the use of organic silica sources, such as TEOS or tetramethoxysilane (TMOS), in combination with diluted acidic aqueous solution of a Pluronic-type triblock copolymer (2–7 wt% in water), such as P123 (EO20 –PO70 –EO20 ). As discussed above, the hybrid interface formation is here suggested to follow a (N0 H+ ) (X− I+ ) model, since the block copolymer could be positively charged under the reaction conditions. Evidently, the use of triblock copolymers expands the accessible range of mesopore sizes. Mesoporous silicas obtained with such copolymers usually exhibit uniform large pores with diameters well above 5 nm and quite thick walls, the latter providing high thermal stability and improved hydrothermal stability compared to OMS synthesized with ionic surfactants, e.g., M41S materials [62]. To compare, MCM-41 materials usually showed a wall thickness of about 1 nm. SBA-15 silica can be synthesized with pore sizes ranging between 5 nm and 12 nm and thick walls (3.0–6.5 nm in width), depending on the reagent ratios, pH, and aging temperature (Figs. 4.4(a), (c); [38]). This material exhibits a large surface area of around 800-1 000 m2 g−1 and pore volume up to 1.5 cm3 g−1 . A TEM image showing the hexagonal structure of SBA-15 is presented in Fig. 4.4(a). SBA-15 silica is of growing interest for a wide range of applications (e.g., sorbent, support for catalysts and biomolecules, nanoreactor, solid template, etc.). At first, SBA-15 was thought to be a large pore equivalent of MCM-41, which has unconnected mesoporous cylindrical channels. However, studies showed that the pore size distribution of SBA-15 is rather bimodal, whereby the larger, hexagonally ordered structural mesopores are connected by smaller pores (micropores or small mesopores) located inside the silica walls [37, 63–66]. These pores are not ordered and most probably originate from the penetration of the PEO blocks of the copolymer inside the silica framework. Owing to the interaction of the hydrophilic chains of the P123 copolymer with the polymerizing silica species during the mesophase formation of SBA-15, some EO groups are occluded in the silica walls. After removal of P123, SBA-15 exhibits therefore a secondary pore system in its framework wall (i.e., intra-wall pores). These intra-wall pores are usually in the micropore-small mesopore (∼ 2–3 nm) range, but the actual size and associated volume are highly dependent on the details of the synthesis (see below). SBA-16. In the family of copolymer-templated materials, ordered mesoporous silicas consisting of interconnected large cage-like pores are also of significant interest. A sil¯ symica mesophase related to SBA-15 is the material designated SBA-16 (cubic Im3m metry), which is synthesized in a similar way, but using a different nonionic triblock copolymer, e.g., Pluronic F127 (EO106 –PO70 –EO106 ). This large pore silica consists of

4 Ordered mesoporous silica: synthesis and applications

50 nm

100 nm

(a)

(b)

800 700 600

Dv(d) (cm3/A/g)

Volume adsorbed (cm3/g)

900

500 400

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00

Volume adsorbed (cm3/g)

1000

0 2 4 6 8 10 12 14 Pore width (nm)

300 200 Ads. Des.

100 0 0.0 (c)

| 71

0.2

0.4 0.6 P/P0

0.8

500 400 300 200 100 0 0.0

1.0

130°C 100°C

0.2

0.4 0.6 P/P0

0.8

1.0

(d)

Fig. 4.4. (a) TEM image showing the 2D hexagonal arrangement of the mesopores in SBA-15. Reprinted with permission from [2]. (b) TEM image of cage-like SBA-16 silica aged at 100°C, viewed along the [111] direction. Reprinted with permission from [2]. (c) Nitrogen adsorption-desorption isotherm (−196°C) and corresponding NLDFT pore size distribution for SBA-15 (BET surface area: 875 m2 g−1 ; Total pore volume: 1.26 cm3 g−1 ; NLDFT pore size: 8.5 nm). Reprinted with permission from [18]. (d) Nitrogen adsorption-desorption isotherms (−196°C) of SBA-16 samples aged at 100°C or 130°C, as indicated. Insert shows a scheme of the pore structure. Reprinted with permission from [54].

spherical cavities of 6–11 nm in diameter organized in a body-centered cubic (bcc) array, and the cavities are 3D interconnected through mesoporous openings of 2–4 nm (Figs. 4.4(b), (d)) [54, 61, 67]. Pluronic F127 presents a high hydrophilic to hydrophobic volume ratio (high EO/PO ratio), which is favorable for the formation of highly curved globular micelles under aqueous conditions.

72 | Part I Functional materials: Synthesis and applications KIT-6 (Korea Institute of Technology №6). Another member of the family of triblock copolymer-based silica mesophases is the large pore equivalent of MCM-48 known as KIT-6. One of the easiest methods of generating this interesting mesophase was introduced in 2003 by Kleitz et al. [22, 55], who used a blend of Pluronic P123 and n-butanol for the structure-direction, along with a fine tuning of the acid concentration. This ¯ symmetry, and the pore netKIT-6 silica material exhibits a structure with cubic Ia3d work topology can be described as an interpenetrating bicontinuous network of highly interconnected channels (shown schematically in Fig. 4.5; [68]). The mesopore structure of KIT-6 is thus 3D interconnected and built of two continuous ordered channel systems separated by a silica wall which follows the infinite periodic minimal surface (IPMS) called the Gyroid surface (G) [22, 55, 69]. The porosity of KIT-6 is quite similar in nature to that of SBA-15, although subtle differences have been observed [70]. KIT-6 silica has high pore volume and large accessible pores tailored between 5 and 12 nm, with additional intra-wall pores as well [22, 51, 55, 69, 71]. Several other methods of ¯ silica have been reported [72–74]. producing a large pore cubic Ia3d

[111]

(a)

50 nm (b)

Fig. 4.5. (a) Representative TEM of mesoporous KIT-6 silica. Shown is a view along the [111] direction. (b) Representation of the Gyroid G infinite periodic minimal surface, which is followed by the ¯ structure, as two intersilica walls in KIT-6. Also shown is an alternative representation of the Ia3d woven networks of branched cylindrical channels. The G surface separates the two sub-frameworks of rod-like mesopores. Reprinted with permission from [70].

Some tools for tailoring structure and porosity A method of tuning the pore size of surfactant-directed inorganic materials is to simply change the length of the surfactant carbon chain. Usually, a linear relationship is observed between pore size and length of the carbon chain of a molecular template. With Cn TAB (n = 8–18), the pore size of the as-synthesized MCM-41 material increases by about 0.45 nm when increasing n by two carbon atoms. Kruk et al. [75, 76] also confirmed that the pore size of calcined MCM-41 and MCM-48 materials increases almost linearly using Cn TAB surfactants with chain lengths of 8 to 16 carbons. However, this simple strategy is applicable only as long as the surfactant is soluble and

4 Ordered mesoporous silica: synthesis and applications

| 73

leads to the formation of a mesophase. It seems that the shortest chain surfactant from which a mesophase could be created is with n = 8. On the other hand, long-chain surfactants (n > 20) are not easily available and are practically insoluble in water, and the mesophases obtained are sometimes rather poorly ordered [12, 13, 77]. Similarly, changes in molecular geometry and chain length of nonionic block copolymers permit fine tuning of the pore size of large pore mesoporous silica. There, the adjustment of pore size can be continuously performed by varying the concentration of SDA and changing the composition of the copolymer or the block size [61, 78, 79]. Indeed, the ratio of hydrophilic to hydrophobic blocks (EO/PO) in the block-copolymer can be decisive for the nature of the mesostructure. In general, lowering this ratio results in the formation of lamellar mesostructures, while higher ratios favor the packing of spherical micelles into cubic mesophases [79]. For instance, triblock copolymers possessing long hydrophilic chains (i.e., high EO/PO ratio), such as F127, lead to materials with highly curved cage-like pores (e.g., SBA-16, KIT-5; [58]). In these cases, simultaneous tailoring of the cage dimensions and pore openings of these cage-like silicas is also feasible by using copolymer blends (P123 mixed with F127), both under control of synthesis temperature and time [35]. The EO/PO ratio of the copolymer has a marked influence on pore size and wall thickness of the resulting materials. Alfredsson et al. investigated the influence of the variation in block copolymer composition in the synthesis of SBA-15 [79, 80]. Their results established that, for synthesis conditions where a hexagonal mesostructure is obtained, an increase in PO chain length resulted in larger pores. On the contrary, an increase in EO chain length led to thicker walls. Another convenient way to tailor the pore size of an OMS is to vary the temperature and duration of the HT. Applying aging treatments at different temperatures and for prolonged periods (from 24 hours up to several days) can efficiently modulate the nature of the mesophase. This type of treatment can either be performed directly in the mother liquid or at a different pH in fresh solutions (typically water or alcohol). For instance, MCM-41 silica could be restructured at elevated temperatures in its mother liquid, resulting in pore size expansion from 3.5 to 6 nm [34, 81]. Moreover, this approach usually results in a material with enhanced stability and higher structural quality owing to denser walls [82]. This improved stability could arise from increased condensation of the silanols within the silicate framework, i.e., better silica polymerization, leading to less silanol groups, thus less shrinkage occurring during calcination, and thicker walls. In the case of large pore OMS prepared with triblock copolymers, such as SBA-15 or KIT-6, both synthesis and aging temperature strongly influence the mesostructure formation of these OMS, and this by altering micelle hydrophobicity and silica condensation. The solution reactions leading to mesophase formation are usually performed within a range of 35°C to 45°C depending on the block copolymer. Increasing the synthesis temperature within this range and beyond renders the EO groups more hydrophobic, leading to micelles with larger hydrophobic core volume and smaller hydrophilic regions [78]. The temperature and duration of

74 | Part I Functional materials: Synthesis and applications the hydrothermal aging, which is then applied after this first synthesis step, are also critical. The aging temperature is usually between 40°C to 150°C (often 90–100°C). The mesopore size of SBA-15 is easily tailored from about 5 nm up to 12 nm simply by increasing the aging temperature from 60°C to 140°C. Also, substantially larger pore volumes are obtained and the nature of the intra-wall porosity is drastically modified, depending on the HT temperature applied (Fig. 4.6, [22, 55, 63, 66, 83–89]). It is proposed that SBA-15 prepared with aging between 35°C and 60°C exhibits micropores with no apparent connection between mesopores. In contrast, at 100°C, SBA-15 shows both the presence of micropores and larger connections between the ordered mesopores. At 130°C, the material shows no more micropores, but much larger pore interconnections are present [66]. Note that an aging of 12 hours to several days is normally required to produce silica materials with satisfying quality. Similar to SBA-15, ¯ KIT-6 silica can be varied within a comparable the mesopore size of the cubic Ia3d range of diameters [22]. Also, the size of the spherical mesopores of cage-like materials (e.g., SBA-16, FDU-1, KIT-5) can be tailored by applying different aging temperatures and times. However, in this case, not only the main mesocage is enlarged, but the pore openings of the cages also become wider upon prolonged hydrothermal treatment. Fine tuning of pore size and mesostructure can also be performed by adjusting the solution pH upon addition of given amounts of acid or base during synthesis. This is, for example, well-established for MCM-41 and MCM-48 syntheses [90, 91]. It is usually explained by the strong influence that the solution pH has on the degree of condensation and polymerization of the inorganic oligomeric species, on the charge density of the polyelectrolyte inorganic species involved, and on the micellar organization. Indeed, as reported by Ryoo, adjusting the pH of the reaction mixture in situ during the synthesis of MCM-48 favors the formation of cubic mesophase [92]. Moreover, the materials (MCM-type) prepared with careful monitoring and adjustment of the pH often demonstrate a high degree of long-range order [91]. For block copolymertemplated syntheses, usually performed in acidic media, pH also has great importance. For SBA-15, it is the key parameter to control the overall kinetics of the synthesis. SBA-15 can be synthesized in various acid concentrations [38, 93, 94], however, some noticeable differences in the structural order, particle morphology, and porosity features of the resulting solids have been observed. Syntheses performed with acid concentration ≥ 1.5 mol l−1 led to very rapid precipitation [61]. Furthermore, high acid content may somewhat influence the micellar organization [95], although little effect on the micelle shape has been observed, before addition of TEOS, with [HCl] up to 2 mol l−1 [93]. The addition of electrolyte salts, such as NaCl or KCl, obviously affects surfactant packing and the interactions between surfactant molecules and silica. The surface charge density of the surfactant micelles can be modified by adsorbed counter-ions. The surfactant molecules may then self-assemble into a mesophase with, for example, a lower surface charge density, and mesophase transitions could take place upon modification of the surface curvature. Usually, inorganic salts have a strong influence

4 Ordered mesoporous silica: synthesis and applications

Front view

(a)

|

75

Side view

No connections

(b)

Connections 2–3 nm diameter

(c)

Large mesoporous connections

Fig. 4.6. The effects of hydrothermal aging temperature on the pore structure of SBA-15-type materials: (a) low temperature aging (35–60°C); main mesopores: 5–6 nm - wall thickness: 4 nm – micropore volume: ∼ 0.3 cm3 g−1 ; (b) aging at 80–100°C; main mesopores: 7–9 nm – wall thickness: 3.2 nm – micropore volume ∼ 0.1 cm3 g−1 ; (c) high temperature aging (> 120°C); main mesopores: > 9 nm – wall thickness: 2 nm – no micropores. Reprinted from [70].

on the values of CMC (critical micellar concentration) and CMT (critical micellar temperature) of the triblock copolymer micelles, which can both be decreased or increased upon salt addition. Salting-out electrolytes (lyotropic ions) such as KCl, NaCl or K2 SO4 , are not adsorbed in the copolymer micelles. These salts dehydrate the hydrophilic portion of the block copolymer, inducing a pronounced reduction in the preferential interfacial curvature of the micelles. On the other hand, salting-in electrolytes (hydrotropic ions) are adsorbed in the micelles and tend to inhibit their growth, which could thus increase the preferential interfacial curvature [73, 96]. The Zhao group was the first to use the salting-out effects caused by the addition of electrolytes to triblock copolymer solutions to produce various well-defined mesostructures [97]. It should be kept in mind, however, that not only the aggregation behavior of the copolymer micelles is affected by salt additions, but the presence of electrolytes will influence hydrolysis and condensation, and the kinetics of aggregation of the inorganic species.

76 | Part I Functional materials: Synthesis and applications Dissolving hydrophobic additives inside the core of the micelles is largely exploited to increase the pore size of mesoporous silicas. They can alter the interface energy of the system, ultimately leading to changes in numerous features, e.g., micelle shape and size, mesophase transition, enlargement of mesopores and/or variation in the morphology of the final products [23, 24]. Trimethylbenzene (TMB) has been one of the most widely used additives [33, 98], although aliphatic hydrocarbons such as hexane have been used as well [99]. For example, it was shown that the pore size of MCM-41 can be altered in a controlled manner between 2 nm and 10 nm by addition of TMB [13]. An almost linear relationship was found between TMB concentration and final pore size. Hydrocarbons or hydrophobic aromatics are regarded as swelling agents that are preferentially solubilized in the core of the micelles. In contrast, cosurfactant molecules, such as short-chain n-alcohols or n-amines, are accumulated in the palisade layer of the micellar aggregates and therefore induce more intricate effects, whereby both the mesophase behavior and the d-spacing of the mesoscopically ordered material can be affected [100]. The MCM-48 syntheses performed by Ryoo et al. [101] (addition of EtOH) and by Schumacher et al. [102] (addition of triethylamine) are good illustrations of the complex roles of these additives. The use of TMB to swell the pores of the triblock copolymer-based OMS is also widespread. SBA-15 materials with pore sizes of 15–16 nm can be obtained with this additive. However, the quantity introduced has to be cautiously controlled for retention of the ordered mesoporous mesostructure [103–105]. The silica materials exhibiting pores reaching 30 nm which were obtained by addition of TMB to the synthesis mixture were in fact disordered foam-like structures [105]. Also, additives and HT modulation (time and temperature) can be synergistically combined to tailor porosity of SBA-15 [85]. TMB can also be used to control the phase transition in the triblock copolymer F127 system, leading to large pore OMS [106]. Similarly, for block copolymer-templated OMS synthesized in acidic media, the use of co-surfactant additives can lead to complex systems [107, 108]. The synthesis of KIT-6 silica is a perfect example. This material is uniquely obtained in very high phase purity by carefully controlling both the addition of a co-surfactant (n-butanol) and the acidity of the mother liquid ([HCl] < 0.8 M) [22]. In fact, in these syntheses using n-butanol, the adjustment of the HCl concentration to 0.3–0.7 M is a prerequisite for directing the formation of a given silica-based mesophase. In this way, it actually became possible to synthesize ¯ Im3m, ¯ and Fm3m) ¯ in a wide range of several large pore cubic silica mesophases (Ia3d, reagent compositions (see Fig. 4.7 for the diagram of product phases, as a function of reagent ratios; [109]).

4 Ordered mesoporous silica: synthesis and applications

| 77

3.2 H

H

+

2.8 + + + +

H H

TEOS (molar ratio)

2.4

H

+

2.0

+ H

1.6

H

+ +

H

+ +

H

0.8 0.4 0.0

+

+ + + + + + + Ia-3d + + + H 2-D hex + Disordered Mixed with 2-D hex to Ia-3d Mixed Ia-3d to disordered H

H

0.4

0.8

(a)

1.6 2.0 1.2 BuOH (molar ratio)

2.4

3.0 D

2.5

D

D

D

D

D

D

2.8

Im-3m F Fm-3m D Disordered D

Disordered D

TEOS (molar ratio)

+

H

1.2

D

D

2.0

D

D

D

D

D

D

D

D

1.5

– Im3 m

F F

– Fm3 m

1.0

D

D

D

D

D

D

F

F D D

F

2-D hex

F

D D

D

D

D

Disordered

0.5 0.0 (b)

+

+ + + +

0.5

1.0 1.5 2.0 BuOH (molar ratio)

2.5

3.0

Fig. 4.7. Diagrams of mesophase structures synthesized using blends of triblock copolymer and n-butanol. The diagrams are established according to XRD measurements. (a) Each sample is prepared with a molar ratio of 0.017 P123/x TEOS/y BuOH/1.83 HCl/195 H2 O. Reprinted with permission from [22]. (b) Each sample is prepared with a molar ratio of 0.0035 F127/x TEOS/y BuOH/ 0.91 HCl/117 H2 O. Reprinted with permission from [109].

78 | Part I Functional materials: Synthesis and applications

4.3 Functionalization of ordered mesoporous silica The available methods for the functionalization of ordered mesoporous materials are vast, and a considerable number of studies have been dedicated to potential applications of functionalized mesoporous materials, especially in heterogeneous catalysis. This section is not intended to be a comprehensive survey of the topic; instead it should only provide a brief overview of the various strategies which have been developed to modify mesoporous materials. OMS are promising in many applications because of their unique porous properties. However, in order to be useful, they usually need to be functionalized. Depending on the requirements of the synthesis and/or the targeted application, various strategies to introduce useful functions to OMS can be considered. It is important to note that only the most common methods will be presented in this brief section. As illustrated in Fig. 4.8, there are two main strategies available to integrate (organic or inorganic) functionalities to OMS [41, 110]. – Post-synthetic modification, which is usually performed by grafting or impregnation/adsorption methods on porogen-free OMS. This approach allows functionalization of the pores and the external surface. – Direct addition of the functionalities during the synthesis of OMS. This “one-pot” process is also described as co-condensation. In this method, both the silica walls and pores can be modified. A variation of this method consists of the sequential addition of organosilanes or metal-precursors at different synthetic stages. This approach may allow control of the spatial localization of the functionalities [111, 112]. Both methods have their own advantages and drawbacks [110, 111, 113]. Post-grafting functionalization relies on the reaction of organosilanes (e.g., amino-, thio- , phosphosilanes, etc.) with the free silanol groups of the pore surface. Indeed, even after calcination at 550°C, the pristine OMS is not fully condensed and some silanols are still available for grafting even if their density is quite low (1–2 SiOH nm−2 ; [24]). Postgrafting is usually performed by treating the OMS powder in organic solutions of the organosilane under reflux for a prolonged period. This method has several advantages: ease of implementation, it does not alter the mesostructure, and it also offers a great versatility in the choice of introduced groups. However, in some cases, a preferential reaction of organosilanes at the pore entrance can be observed, leading to an inhomogeneous distribution of functionalities. If very large molecules are grafted, some pore blocking may occur [110]. In contrast, ordered mesoporous organosilicas can be obtained by co-condensation of tetraalkoxysilanes (TEOS or TMOS) and terminal trialkoxyorganosilanes of the type (R󸀠 O)3 Si-R (where R󸀠 is either methyl or ethyl, and R is a non-hydrolyzable organic group). Here, the functional groups are most often placed dangling on the surface, but also partly inside the framework walls [114–117]. In fact, the organoalkoxysilane plays

4 Ordered mesoporous silica: synthesis and applications

|

79

Direct synthesis or post synthesis

Substitution grafting

M+ L L

Immobilization

M

Ion exchange Silyalation

E

Enzyme encapsulation

Nanoparticle Non-silica material Organic-inorganic hybrid framework

Surface coating

Mesoporous materials (MCM–41, –48, SBA–15, etc)

High surface area (1000 m2/g) Narrow pore size distribution Thermal stability

Fig. 4.8. Schematic representation of the different methods available for the functionalization of OMS. Reprinted with permission from [110].

two roles, since it acts as a building block in the inorganic structure, co-condensing with the tetraalkoxysilane precursor, and it supplies organic functionality. A wide variety of functional groups can be incorporated using this method (e.g., vinyl, phenyl, aminopropyl, imidazole, cyanopropyl, mercaptopropyl, etc.). The mercaptopropyl groups are especially interesting since these groups can subsequently be oxidized with nitric acid and/or H2 O2 to yield sulfonic acid groups [118]. However, the choice of a suitable organosilane precursor is usually limited by the conditions of synthesis, and the template removal must be performed by solvent extraction or careful acid treatment. The impregnation/adsorption pathway is also an important post-synthesis method for modifying OMS with organics or inorganic species. This technique is based on the capillary introduction of a volatile solvent containing the precursors or molecules of interest inside the mesopores of the solid. After evaporation of the solvent, the functional group is then chemically linked (grafted) to the OMS by a subsequent thermal treatment. Different techniques can be used [111, 119], but the incipient wetness, which employs a minimal amount of solvent at the limit of the powder wetness, has been shown to be very effective [120]. As a nice example, Choi et al. have successfully applied this technique to confine polymerization of vinyl monomers (e.g., styrene, acrylates) selectively on the mesopore surface of SBA-15 silica [121]. The incipient wetness

80 | Part I Functional materials: Synthesis and applications is a very versatile tool which usually leads to a rather homogeneous distribution of functional species inside the pore network of the OMS [120, 122].

4.4 Morphology control Morphology control is indispensable in many of the advanced applications envisioned for functional mesoporous materials [123]. Perm-selective membranes, microspheres or monoliths are important for sorption, separation and chromatography purposes. Porous thin films or fibrous structures are relevant for electronics, optics, low k-dielectrics, and sensing applications. Colloidal particles or nano-spheres are preferred for biomedical systems to be used in drug delivery or magnetic resonance imaging (MRI) with contrast agents. The first ordered mesoporous materials which were synthesized were typically finely divided powders consisting of small particles (< 10 μm) with no well-defined morphology. Since then, a wide variety of shapes, including thin films, (nano)spheres, fibers, tubes, macroporous-mesoporous monoliths, and many other complex morphologies have been described for ordered mesoporous materials (Fig. 4.9; [124–135]). Mesoporous solids with controlled macroscale morphology can either be designed by processing conditions such as dip-coating, spin-coating or emulsion templating, or alternatively, formed spontaneously through self-organization processes which are mostly based on kinetic regimes. Due to the amorphous nature of the silica walls, simultaneous modulation of both the mesoscale (hybrid mesophase) and macroscale (particle size and shape) is possible during synthesis. However, it has to be kept in mind that formation of the mesophase and growth of the morphology influence one another and cannot be seen as separate aspects [23]. With the ongoing emergence of complex nanostructures, controlling both mesopore structures and morphologies of MSNs at the nanoscale is not straightforward and requires a thorough understanding of the chemistry involved [136]. Mesoporous particles with spherical morphology are easily synthesized under alkaline aqueous conditions. For instance, Huo and Schüth [137] reported in 1997 the preparation of hard transparent spheres from an emulsion at room temperature. Ingeniously, Grün and Unger modified the Stöber synthesis of monodisperse spheres performed in the presence of ethanol and ammonium hydroxide [138] and could successfully prepare almost monodispersed mesoporous MCM-41 and MCM-48 spheres [102]. Also, pseudomorphic transformation of commercially available pre-shaped spherical silica particles (5 to 800 μm) can also be used to produce ordered mesoporous MCM-41 and MCM-48-like particles with a spherical morphology. In this latter method, amorphous silica is progressively and locally dissolved under mild alkaline conditions and re-precipitated at the same rate in the presence of the surfactant, without modifying the global spherical morphology [139, 140]. Finally, the synthesis of hollow particles is achievable, for example, by spray-drying techniques, based on very rapid solvent evaporation and retention of the pre-formed shapes [141].

4 Ordered mesoporous silica: synthesis and applications

(a)

50 nm

81

(b)

50 μm (c)

|

1 μm (d)

50 nm

Fig. 4.9. Illustration of possible OMS morphologies: (a) SEM images of mesoporous silica fibers (image: R. Guillet-Nicolas, F. Kleitz, U Laval), (b) ordered mesoporous silica colloidal spheres (image: R. Guillet-Nicolas, F. Kleitz, U Laval), (c) TEM images of ordered mesoporous MCM-48 silica nanospheres (image: R. Guillet-Nicolas, F. Kleitz, U Laval), and (d) representative TEM image of an ordered mesoporous silica thin film. Reprinted with permission from [126].

Among these, morphologies, spheres, and especially nanospheres, are most interesting for the biomedical world because these objects may interact well with cells and do not exhibit sharp edges or preferential faces [142]. Most of the current mesoporous spherical particle syntheses are actually derived from the seminal work of Grün, which described the synthesis of colloidal spheres of MCM-41 and MCM-48 [102, 143]. In 2001, Cai et al. [144] and Mann et al. [145] both reported the first successful syntheses of individualized MCM-41 nanoparticles. However, the term Mesoporous Silica Nanoparticles (MSNs) was popularized in 2003 by Victor Lin, who published the first facile preparation of functionalized MCM-41 nanoparticles for drug delivery applications [146]. This synthesis provided homogeneous spherical particles with a diameter ≤ 200 nm and good porosity features, making them excellent candidates for cellular applications. Following this breakthrough, many efforts were devoted to developing MSNs with controllable particle and pore sizes [136]. In particular, in 2008, Kim et al. reported the first synthesis of MCM-48 nanospheres with high pore ordering and highly uniform particle size (Fig. 4.10(c); [147]). By using a modified version of the Stöber synthesis, and Pluronic F127 as a “particle-designer” agent in alkaline media again, they successfully formed monodisperse MSNs with controllable sizes within the range of 70–500 nm [148]. However, the main drawback of the most common MSN protocols remains the relatively small pore sizes of the synthesized particles. Indeed, because almost all

82 | Part I Functional materials: Synthesis and applications syntheses are performed with CTAB-like molecules as SDAs, a maximum pore size of 4–4.5 nm is achieved. Swelling agents can be used to reach larger pores, but as for the classical OMS, they lead to a decrease or a loss in mesostructure ordering [149]. However, this latter aspect is not necessarily of critical importance for most biomedical applications. Triblock copolymers may also be suggested as SDAs to produce particles with larger pore size, while keeping mesoscopic ordering. Unfortunately, obtaining MSNs in acidic medium is more difficult than in alkaline medium (more complex interactions occurring during synthesis, and difficulties associated with kinetic control). In this area, He et al. [150] and Kim et al. [151] recently proposed two different pathways leading to pellet-shaped SBA-15 silica (large pores) of 300–600 nm (see Fig. 4.10). Both methods are based on restricting the growth of the SBA-15 particles. The first group used ZrIV multivalent metal ions to “cut” the micellar aggregates in situ, whereas the latter employed Pluronic P104 as SDA, and specific synthesis conditions to favor initial nucleation and growth of primary particles while limiting further aggregation [152]. Alternatively, by mixing fluorocarbon-surfactant (e.g., FC-4) with the classical Pluronics, Han et al. successfully synthesized a new family of large pore MSNs which were called IBN [153]. These new materials could have great potential for biomedical applications as they exhibit spheroidal shapes with required particle dimensions (Fig. 4.10(c)) and fairly large pores (> 8 nm; [154, 155]). (a)

200 nm

(b)

200 nm

(c)

100 nm

Fig. 4.10. TEM images of large pore ordered MSNs: (a) SBA-15 pellets obtained using the protocol described by He et al. [150], (b) Kim et al. [151] (images: R. Guillet-Nicolas, F. Kleitz, U Laval), and (c) IBN-like material obtained by Hartono et al. Reprinted with permission from [155].

4.5 Selected applications of functionalized ordered mesoporous silica Applications of mesoporous materials have been considered in many areas, including catalysis, optoelectronics, sensors, sorption, biomedical materials, environmental remediation, green chemistry, and most recently, energy storage and conversion [5, 156–159]. In this chapter, emphasis is placed on recent developments in the area of sorbents and materials for chromatographic extraction, as well as new innovative drug delivery systems.

4 Ordered mesoporous silica: synthesis and applications

| 83

4.5.1 Functionalized MSNs as controlled drug delivery platforms Targeted drug delivery is one of the greatest challenges in modern medicine. To address the limitations of conventional drug delivery systems, mesoporous silica nanoparticles (MSNs) are considered robust inorganic alternatives to polymeric nanoparticles, owing to their high porosity, biocompatibility, and ease of modification. MSNs have thus captivated a lot of interest worldwide and have emerged as promising carrier materials for controlled and vectorized delivery of drugs [160–175]. Their stable mesoporous structure and well-defined surface properties make mesoporous silicas good matrices to host a wide variety of drugs and biologically active species for local and controlled drug delivery applications [149, 176–179]. Another attractive advantage is that amorphous silica is fairly degradable in aqueous solution, and thus problems related to the removal of the material after use can be avoided. Moreover, there are large numbers of silanol groups covering the mesoporous silica walls which are susceptible of undergoing chemical or biochemical functionalization, which is one of the key aspects for the prospect of biological applications of these materials. An ideal drug delivery system should enable efficient healing at the lowest drug concentration and dosage frequency, while being both patient-friendly and safe. Owing to their outstanding features, which allow both the loading of various drugs or bioactive species and the adequate functionalization needed for in vitro and in vivo purposes, stimuli-responsive MSNs can be seen as truly promising carriers for delivering precise doses of drugs to targeted sites. Moreover, by combining diagnosis and therapeutic tools, theranostic MSN-based platforms may be designed [164, 166]. The concept of controlled drug delivery comprises 4 major steps: (1) The fabrication of a biocompatible device that will efficiently encapsulate high loading of the desired drug(s). (2) No premature release prior to reaching the target location is needed in order to protect the healthy organs and/or cells, i.e., avoidance of side effects due to nonspecific interactions, and prevent the decomposition/denaturing of the drugs. (3) Efficient and sustained release of the drug at the targeted location. (4) Easy clearance of the biocompatible and/or biodegradable carrier by natural pathways. Such a strategy is expected to enhance the drug efficiency while minimizing the required quantities owing to enhanced bioavailability at the key location [180–182]. Nevertheless, the key challenge still remains to simultaneously achieve precise targeting and proper colloidal stability, especially in real physiological media [122, 183]. To achieve these goals, several controlled drug delivery systems (CDDS) have been proposed based on various external or internal stimuli, e.g., temperature, time, chemical reactions, enzymes or pH, to name a few [184–189]. With MSNs, the sequestration of the drug inside the porous network is usually realized through the functionalization of the inner and/or outer pore surface with var-

84 | Part I Functional materials: Synthesis and applications ious barriers acting as “gates”, such as proteins, polymers, macrocycles, or even nanoparticles. These “gatekeepers” respond to a specific chemical or environmental stimulus which will induce their (reversible) removal upon exposure, hence triggering the drug release. Barriers are of prime importance as immediate drug release is commonly observed after administration in sink conditions, when drugs are simply adsorbed into non-modified MSNs [177, 190]. However, the release behavior of pristine MSNs should be considered with great care as it is also dependent on the conditions chosen to simulate the different body fluids [191]. The main MSN-CDDS are summarized in Table 4.2. Among these, the systems based on drug release triggered by pH variations have been widely investigated. Indeed, since the pH variations within the body and/or cells are well-known, they can be advantageously used for target drug delivery. For example, the pH difference between normal and cancer cells may be used for the specific targeting of tumor cells using MSNs [192]. Because of the extreme importance of cancer therapy, most of the research involving MSNs has so far been dedicated to the release of compounds in an acidic environment, i.e., cancer cells where pH is mildly acidic [166, 182, 189]. However, MSN-CDDS could also be synthesized for drug delivery applications where release is triggered at neutral or physiological pH. This feature is, for instance, of high interest for oral delivery applications because the pH in the human gastrointestinal tract naturally varies, i.e., the stomach is highly acidic (pH = 1.2) compared to the small intestine (pH = 6.5–7.0) and colon (pH = 7.0–8.0), making a neutral pH-triggered approach a smart strategy for oral delivery of drugs into the intestine. This is highly desirable as it improves patient compliance and convenience [193, 194]. In 2007, Kawi reported a simple and fast method for encapsulating protein-loaded NH2 –SBA-15 with polyacetic acid, creating a smart pH-responsive protein delivery system for the first time [195]. This material exhibited almost no premature release in pH = 1.2 (less than 2%) during the first five hours, being compatible with the United States Pharmacopeia and the National Formulary (USP–NF) guidelines for gastro-resistant compounds, i.e., less than 10% drug release after 2 hours in gastric conditions [196]. However, in this example, protein release at pH = 7.4 was only 40% after 35 hours, limiting somewhat the efficiency of this system. This low release in physiological conditions was linked to the poor colloidal and chemical stability of the materials. Another method of generating pH-responsive MSN-CDDS is to incorporate positive charges into the mesopore channels of MSNs by means of trimethylammonium (TA) groups [197]. These groups allow efficient adsorption of anionic molecules and minimize their release under acidic pH owing to unfavorable electrostatic interactions. At neutral pH the strong electrostatic repulsions then trigger a sustained release of the loaded drug. This original system showed excellent drug sequestration ability in an acidic environment and appreciable drug release in physiological media. However, if such materials are not coated or properly encapsulated, the drug loaded inside the pores might be denatured by the acidity of the stomach environment, ultimately altering the bioactivity of the molecule once released in the intestine.

4 Ordered mesoporous silica: synthesis and applications

| 85

Table 4.2. Some controlled drug delivery systems (CDDS) using MSNs, reported in the literature. Adapted with permission from [172]. Class

Examples

Structure I Nanoparticles

Au Nps CdS Nps Fe3 O4 Nps ZnO Nps

Structure II Macrocyclic organic molecules

Cyclodextrin (CD) Cucurbit[6]uril (CB[6]) Dibenzo-24-crown-8 (DB24C8)

Structure III Linear molecules

Linear polyamine Saccharide derivative Linear polymer Peptide sequence

Structure IV Multilayer shell coating

Polymer layers Biomolecules Polyelectrolyte multilayers

Structure V Pore modification

Functional molecule Polymer Azobenzene derivatives impeller

Schematic structure

Functiionalized NPs

External stimuli

Drug molecule Modification group α-CD OH O HO OH 6

BD24C8

Mesoporous silica

CB[6] O N N

Stalk

N N 6 O O O O O O O O O

External stimuli

CD

Linear molecules

External stimuli

or

External stimuli Mesoporous silica

External stimuli

External stimuli (a)

Functional molecule or polymer

(b)

Nanoimpeller

Furthermore, this approach can only be used with anionic compounds. In 2011, the Cheng group [198] improved this system by introducing a hydrazone bond to synthesize MSN–hydrazone–TA materials. The TA groups can further be eliminated through the progressive hydrolysis of hydrazone bonds in gastric pH conditions, leading to a rapid and complete release of the drug only once the intestine has been reached. This time-dependent and pH-sensitive MSN-CDDS was shown to enable accurate delivery of therapeutic drugs to the targeted tissue (colon), while limiting premature release during gastric emptying. Here, high biocompatibility was found and no cytotoxicity, even at a high nanoparticle concentration, i.e., 500 μg ml−1 . In another example, Sun et al. [199] also developed a pH-responsive oral delivery system, this time by exploiting coordination between an anti-tumor-active poly-oxo-metalate and amino-functionalized MCM-41 mesoporous silica nanosphere. Their results indicate zero premature release in stomach and small intestine conditions, whereas release was observed in colon conditions due to the pH increase. In this system, the gating effect is believed to be based on the rupture of the coordinate bond between the metal

86 | Part I Functional materials: Synthesis and applications centers and the amino groups in mildly basic conditions. Importantly, no cytotoxicity was observed with healthy human amniotic cells. Unfortunately, the polyoxometalate clusters only showed modest inhibition when tested on human malignant tumor cell lines derived from different tissues. From the perspective of delivery of active compounds through the gastrointestinal tract, the use of nanocapsules or coatings would be highly beneficial to completely isolate and protect the active compound from the acidity of the stomach. As such, one could argue that focus should be placed on the development of simple, efficient, biocompatible and/or biodegradable materials. Furthermore, in vivo colloidal and chemical stability of functionalized MSNs is a critical aspect of this technology. In this case, addressing the issue of premature release of bioactive compounds loaded in the nanocarriers is extremely critical, especially with costly drugs. In most reported cases, the drug is just physically adsorbed into the pores, which may cause undesired leaching of the cargo before it reaches the target site making these systems less effective in therapeutic targeting. In this context, Kleitz and Qiao described a new system in which an azo prodrug is covalently bound to MSNs and the drug is released via reduction of the azo bond by the enzyme azoreductase, which is present in the microflora of the colon. Hence, enzyme-responsive MSNs have been designed for a site-specific delivery to the colon, which is achieved by a combination of passive targeting via the nanoparticles and selectivity of the cargo itself towards the colonic azoreductase. In this system, sulfasalazine (SZ, a prodrug firstline therapy for inflammatory bowel disease (IBD)) was covalently attached to the mesopore surface of MCM-48 silica nanospheres acting as the enzyme-responsive carrier, and the molecule could then be reduced to 5-ASA and sulfapyridine by the azo-reductase bacteria inside a simulated colon medium (see Fig. 4.11). The chemical binding of the prodrug on the pore surface prevents premature release of the drug. Monodisperse MCM-48 nanospheres (∼ 150 nm) were chosen as an inorganic scaffold due to their proven superiority in adsorption and release, owing to their 3D continuous pore network and their very high surface area. This system responded very well to the enzyme and no release was observed before the particles were exposed to the simulated colon medium. The covalent attachment of SZ onto nanoparticles thus clearly ensures zero release in stomach (pH 1.2) or intestine (pH 7.4), and at the same time this system also has great potential for delivering therapeutics to the right locations i.e., inflamed tissues or cancer cells, therefore reducing side effects and improving bioavailability of the drugs. With this type of carrier, one could hope that high concentrations of drug-loaded nanoparticles will reach the inflamed tissues and cells, with high in vivo efficacy ensured by zero premature release. Leading candidates for the next generation therapeutic carriers may also be pHresponsive systems. As discussed above, this approach has received a lot of attention for cancer treatment application, owing to the noticeable pH difference between normal and cancer cells [197, 200, 201]. Particularly interesting are strategies based on the use of proteins as stimuli-responsive gating systems [202, 203]. Among their advan-

| 87

4 Ordered mesoporous silica: synthesis and applications

MCM-48 ̶OH = 50nm

N

Toluene Δ Me O I Me O Si O Me O O Si O

I

N

N N N N N

THF/TEA

N N

N N N

N N

N

M-IP

N N

N

M-IP-SZ

=

0 S 0

N HO

N

N

N

N

N

=

NH

N

SZ

HOOC

(a)

N N

N

Azo reductase

N N N NN

N N N N N N N N N

N N N

N

N N

N N N

N

N N N N N N

Enzyme mediated release

N N

N N N

N

N

NH2 NH2

NH2 NH2 NH2

H2N H2N

NH2

NH2 NH2

NH2

N N

H2N

NH2 NH2

(b)

NH2

NH2

NH2

NH2 NH2

= Sulfapyridine

NH2 NH2 NH2 = 5-ASA

Fig. 4.11. (a) Synthesis of the enzyme-responsive drug delivery system based on mesoporous silica nanoparticles with 3D pore structure. (b) Representation of the process of enzymatic release of 5-ASA and sulfapyridine from the nanoreactors in the presence of azoreductase bacteria. Reprinted with permission from [188].

tages, these materials are characterized by high biocompatibility and biodegradability, abundance of reactive groups which can be used for chemical modifications, i.e., grafting [204, 205]. Similarly, being able to enhance colloidal stability in the desired release media constitutes a major improvement, as it will lead to a more homogeneous and efficient delivery of the active compound to the target location. In this area, our laboratory evaluated the potential of the nutraceutical MSN conjugates as a new pH-responsive oral drug delivery system showing low toxicity and sufficient colloidal stability (Fig. 4.12; [206]). In this work, nutraceuticalfunctionalized mesoporous silica particles have been developed for the protection and site-specific release of gastro-sensitive compounds via the oral route. This innovative approach makes use of the fully biocompatible and biodegradable nutraceutical βlactoglobulin, which serves as a pH-responsive gating device. This nano-conjugation led to significantly enhanced colloidal stability (a prerequisite for most drug delivery or cell tracking applications), while the functionality of β-lactoglobulin was fully preserved, i.e., the gelation and gating effect was observed, and thus it could be used for protecting and/or delivering bioactive components (i.e., via pH-controlled delivery). Nutraceutical proteins, such as β-lactoglobulin (a member of the albumin family, present in cow’s milk) or soy protein isolates, clearly offer the advantages of low cost

88 | Part I Functional materials: Synthesis and applications and excellent biocompatibility. By using bio-conjugation chemistry smartly, it was possible to synthesize a pH-responsive system showing limited premature release in acidic media (mimicking stomach acidity). In contrast, when the particles where suspended in physiological intestine conditions (pH 7.4), sustained release was observed for 24 hours, together with enhanced colloidal stability over 48 hours. Moreover, the differences in drug release which were observed at pH = 5 could also make this system very attractive from the perspective of cancer therapeutics. In addition, since β-lactoglobulin is a major by-product of the dairy industry, its production is quite costeffective compared to other complex and expensive systems. As such, this approach is hence also inscribed in a strategy of smart recycling of industrial waste.

APTES grafting

IBU/ACR loading

Toluene 110°C, overnight

Hexane/DMSO RT, 24 hours

pH responsive release

pH > 5

Succinylated β-lactoglobulin grafting

EDC activation MES/ DMSO-MES RT, 2 hours

pH < 5

: –NH2

: Succynilated β-Lactoglobulin (pH > 5)

: Ibuprofen (IBU)/Acridine (ACR)

: Succynilated β-Lactoglobulin (pH < 5)

Fig. 4.12. Schematic representation of the post grafting, bio-functionalization, and pH-responsive release of a drug/dye from β-lactoglobulin-modified-MSNs. Reprinted with permission from [206].

4.5.2 Functionalized mesoporous materials for extraction chromatography (EXC) applications Historically, the first system for adsorption of heavy metal ions based on mesoporous silica supports was used for the trapping of mercury [207]. It was proven that silica materials containing thiols (or thiol derivatives) were particularly effective as sorbents for the capture of environmentally harmful Hg2+ . Soon after, several other silica materi-

4 Ordered mesoporous silica: synthesis and applications

|

89

als, modified mainly with amino- or phosphonate-groups, were applied for adsorbing metals, such as Cu2+ , Zn2+ , Ni2+ , or Fe3+ [208, 209]. Later it was proposed that silica-based inorganic-organic hybrid sorbents could also be versatile for sequestrating/extracting radionuclides [210, 211]. For instance, functionalized MCM-41 materials grafted with acetamide-phosphonate ligands have shown promising properties for Am3+ and Pu3+ decontamination from acidic aqueous solutions [212]. Evidently, the development of new sequestration materials for the nuclear industry is essential from a health and environmental protection point of view, as they not only provide an appropriate solution to nuclear wastes, but also enable their detection if coupled with the appropriate analytical methodologies. The incident in Fukushima Dai-Ichi, Japan, and the long-term consequences of this event are convincing reminders that sequestration/detection of long-lived radionuclides is critical. The expected consequences of radiological/nuclear events and their aftermath on agriculture, population, and the environment have led the scientific community to rethink its approach to monitoring methods based on radiochemical separation [213]. With this objective in mind, our laboratory described a simple and effective functionalization of large pore 3D cubic mesoporous KIT-6 silica to produce a new family of selective sorbents for radionuclides [214, 215]. Using a simple one-step post-grafting method, phosphonate groups were chemically anchored to the silica surface, providing a highly selective and durable functionalization for the extraction of actinides (see Fig. 4.13). Extraction experiments performed with this material demonstrated extremely rapid kinetics (< 1 min) and high selectivity towards actinide extraction, especially uranium (VI). The unique spatial configuration of the 3D cubic mesoporous hybrid induced much higher selectivity and adsorption capacity than other sorbents, and the material is applicable over a range of conditions (pH = 4, room temperature) which are relevant for real environmental analysis. In addition to these unique extraction performances, synthesis of the mesoporous hybrid EXC chromatography materials is very simple and easily scalable, which represents a true alternative for the replacement of currently non-recyclable commercial EXC resins at an acceptable cost. The high extraction efficiency of the functionalized KIT-6 sorbent for U (VI) at relatively neutral pH could also lead to alternative analytical strategies for uranium wastewater management, and environmental and biological monitoring from the perspective of anthropogenic contamination. In addition, such materials are mandatory for all aspects of the nuclear fuel cycles, from mining (treatment of wastewater) to fuel reprocessing (separation of impurities). Well-defined mesoporous materials can also provide opportunities in the highly valuable area of rare-earth elements [216–218]. Nowadays, the importance of rareearth elements (REEs) in the global economy is booming as they are used in numerous advanced technologies more and more. High purity REEs are needed for the production of magnets, chemical sensors or lasers, computers, plasma screens, cell phones, cameras, and so forth; however, natural REE resources are either very limited or their

90 | Part I Functional materials: Synthesis and applications

CH3 CH3

–OH H3C –OH +

O

Si O

O P

O

Toluene, reflux, 24h N2 atm

O

–OH

CH3

–O

O

–O

Si

O P

–O

O

O

CH3 CH3 SBA-15/KIT-6

CH3

DPTS

SBA-15-P/KIT-6-P

(a)

mg U g–1 sample

SPE cartridge

U/TEVA

(b)

SBA-15-P

KIT-6-P

U extraction capacity

Fig. 4.13. (a) Schematic representation of the synthesis of the extracting agent-functionalized SBA-15-P/KIT-6-P materials. (b) Comparison of the extraction capacities of different phosphonatefunctionalized materials and photograph of an SPE cartridge assembled with the KIT-6-based sorbent. Reprinted with permission from [215].

extraction/separation processes are not acceptable in terms of sustainable development [219, 220]. Therefore, it is certainly of value to develop efficient sorbents for the extraction and valorization of REEs, also from alternative sources, for instance, industrial and mining wastes. Commercially, extraction and purification of REEs is based on multiple liquid-liquid extraction (LLE) or chromatographic-based resin separations. However, these approaches are hampered by several issues. Because of the subtle dif-

4 Ordered mesoporous silica: synthesis and applications

| 91

ferences between the various REEs, extraction and purification of REEs is very timeconsuming and involves multiple extraction/purification steps. In order to substantially improve this process and provide a greener alternative to LLE, novel functional nanoporous hybrid materials were proposed demonstrating enhanced selectivity towards heavier REEs in comparison to commercially available products. In this case, KIT-6 silica was modified with the diglycolylamide (DGA) ligand to generate an efficient sorbent for REEs extraction applications (Fig. 4.14; [221]). For this, the DGA ligand was chemically grafted to the silica surface, which enabled the resulting hybrid materials to be cycled and regenerated, a key feature in the development of sustainable and cost-effective resin-type materials minimizing waste production. The choice of KIT-6 silica was motivated here by the highly interconnected nature of the porous network of this material, which is expected to reduce the risk of pore blocking and to be beneficial for diffusion of liquids through the system, all being obvious advantages in chromatographic processes. Extraction of REEs was tested in a solid-liquid system, and distribution coefficients (named Kd ) from batch extraction tests were obtained. Clearly, these new sorbents showed excellent stability upon recycling and demonstrated greater selectivity than commercially available DGA resins under the extraction conditions tested. Most importantly, some of the new sorbents exhibited a much higher affinity for the separation of heavier lanthanides (i.e., yttric earths), being most relevant for the electronics industry. These hybrid sorbents also showed specificity towards Eu and Gd and low competitive behavior with other non-lanthanide trivalent ions and actinides, which are problematic in the commercial extraction of REEs. The perspective for this new system will be to minimize the number of extraction steps used for the purification of REEs as far as possible. Functionalized nanoporous materials with a high surface area and high pore volume can offer high contact efficiency with solutions and high adsorption capacities while preserving adequate flow and transport properties if properly structured. Using such materials may indeed enable the substantial reduction of the number of steps needed for separation of these critical elements, and thus decrease both the required time and the waste production.

4.5.3 Mesoporous organic-inorganic hybrid membranes for water desalination Nowadays, water scarcity, brought about by population growth and industrialization, is one of the major challenges facing society. Desalination of brackish or seawater is one of the most effectively implemented solutions. However, an alternative process to the conventional desalination technologies, e.g., distillation and reverse osmosis (RO), is the thermally-driven membrane distillation (MD; [222]). Typically, such a membrane maintains water at the pore entrance and allows water evaporation, leaving the non-volatile salts behind [205]. Membrane distillation has some benefits, although its industrial application remains low. This lack of commercial success is mostly the result of the dominance of reverse osmosis processes, mem-

92 | Part I Functional materials: Synthesis and applications

DGA-functionalized KIT-6

50 nm

Fig. 4.14. On the left: high-resolution SEM image of the functional organic-inorganic hybrid KIT-6. On the right: diglycolylamide(DGA)-modification of the surface of KIT-6 silica to generate the mesoporous rare-earth element (REE) sorbents. Reprinted with permission from [221].

brane flux decay, and the use of macroporous (0.2–0.7 mm), hydrophobic polymeric membranes [224, 225]. These hydrophobic polymeric membranes are often plagued by fouling and pore wetting issues, and thus the development of new membrane materials which will overcome these limitations is necessary. In this area, membranes based on nanoporous, inorganic-organic hybrid materials represent a potential alternative, with adequate chemical and thermal stability and appropriate pore structure [226, 227]. Within this context, the use of nanoporous organosilica thin-film membranes (∼ 20 cm2 in size) with highly ordered pores (∼ 2 nm) has been suggested (Fig. 4.15; [228]). The mesoporous hybrid membranes were prepared by the dip-coating technique using 1,2-bis(triethoxysilyl)ethane, as a single bridged silicon source, in the presence of Pluronic F68 (E080 P030 EO80 ). After adequate drying, the films were then calcined in air at 300°C to preserve the organic groups in the framework walls. These mesoporous films exhibit excellent desalination performance in the case of synthetic salt solutions with feed temperatures between 20 and 60°C (using a vacuum MD process). These membranes produced pure water across a large range of salt concentrations (10–150 g l−1 NaCl) at average temperatures ≤ 60°C, without exhibiting the usual degradation. Furthermore, the results revealed excellent salt rejection (> 99.9%) and good water fluxes (up to 13 kg m−2 h−1 at 60°C). Here, it was hypothesized that the organic moieties placed within the siloxane framework conferred enough hydrophobicity to the pore walls to form a liquid/vapor interface at the pore entrance, whilst the small mesopore size was crucial in preventing pore wetting. This most recent development could indeed open up a potentially scalable process for fabricating high-performance membranes for efficient water desalination.

4 Ordered mesoporous silica: synthesis and applications

Salt water

Membrane

Na+

H2O

|

93

Fresh water

Cl–

(a) 100 98

15

96 10 94

Flux(OS) Flux(PS)

5

Salt Rejection(OS)

Salt Rejection, %

Water flux, kg m–2 hr–1

20

92

Salt Rejection(PS) 0 0

20

40

(b)

60 80 Time, min

100

120

90 140

Fig. 4.15. (a) Schematic representation of the water desalination process by membrane distillation using a periodic mesoporous organosilica (the mesoporous structure of the hybrid membrane is shown by TEM). (b) Comparison of organosilica (OS) membrane (square symbols) and pure silica (PS) membrane (cross symbols) in 50 g l−1 feed concentration run at 60°C. Filled symbols represent water fluxes and open symbols represent salt rejection. Reprinted with permission from [228].

Acknowledgments The authors would like to thank NSERC (Canada) and FQRNT (Province of Quebec) for their financial support.

94 | Part I Functional materials: Synthesis and applications

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]

[14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]

[25] [26] [27] [28] [29] [30] [31]

K.S.W. Sing, D.H. Everett, R.H.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 1985, 57, 603. K. Klabunde, R. Richards, (Eds.) Nanoscale Materials in Chemistry, 2nd edition, Wiley-VCH Publishers, Weinheim, Germany, 2009. G. Cao, Nanostructures and Nanomaterials, Synthesis, Properties and Applications, Imperial College Press, London, 2004. C.P. Poole Jr, F.J. Owens, Introduction to Nanotechnology, John Wiley and Sons, New Jersey, 2003. F. Schüth, K.S.W. Sing, J. Weitkamp (Eds.), Handbook of Porous Solids, Wiley-VCH Publishers, Weinheim, Germany, 2002. G.Q. Lu, X.S. Zhao (Eds.), Nanoporous Materials: Science and Engineering (Series on Chemical Engineering), Imperial College Press, London, 2005. G. Ertl, H. Knözinger, F. Schüth, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis, 2nd edition, Wiley-VCH Publishers, Weinheim, Germany, 2008. R. Ryoo, S.H. Joo, S. Jun, J. Phys. Chem. B 1999, 103. A.-H. Lu and F. Schüth, Adv. Mater. 2006, 18, 1793. M. Tiemann, Chem. Mater. 2008, 20, 961. F. Schüth, W. Schmidt, Adv. Mater. 2002, 14, 629. C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli, J. S. Beck, Nature 1992, 359, 710. J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T.-W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.L. Schlenker, J. Am. Chem. Soc. 1992, 114, 10834. J.C. Vartuli, K.D. Schmitt, C.T. Kresge, W.J. Roth, M.E. Leonowicz, S.B. McCullen, S.D. Hellring, J.S. Beck, J.L. Schlenker, D.H. Olson, E.W. Sheppard, Chem. Mater. 1994, 6, 2317. X. Auvray, C. Petipas, R. Anthore, I. Rico, A.J. Lattes, J. Phys. Chem. 1989, 93, 7458. R. Schmidt, M. Stöcker, D. Akporiaye, E.H. Tørstad, A. Olsen, Microporous Mater. 1995, 5, 1. V. Alfredsson, M.W. Anderson, Chem. Mater. 1996, 8, 1141. H. Staub, I. Del Rosal, L. Maron, F. Kleitz, F.-G. Fontaine, J. Phys. Chem. C. 2012, 116, 25919. S.A. Bagshaw, E. Prouzet, T.J. Pinnavaia, Science 1995, 269, 1242. D. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Frederickson, B.F. Chmelka, G.D. Stucky, Science 1998, 279, 548. C.Z. Yu, Y. Yu, D.Y. Zhao, Chem. Commun. 2000, 575. T.W. Kim, F. Kleitz, B. Paul, R. Ryoo, J. Am. Chem. Soc. 2005, 127, 7601. Y. Wan, D.Y. Zhao, Chem. Rev. 2007, 107, 2821. F. Kleitz, Ordered Mesoporous Materials in Handbook of Heterogeneous Catalysis, 2nd edition, Chapter 8, 168. G. Ertl, H. Knötzinger, F. Schüth, J. Weitkamp, (Eds.), VCH-Wiley), Weinheim, 2008. G.S. Attard, J.C. Glyde, C.G. Göltner, Nature 1995, 378, 366. A. Monnier, F. Schüth, Q.S. Huo, D. Kumar, D. Margolese, R.S. Maxwell, G.D. Stucky, M. Krishnamurty, P. Petroff, A. Firouzi, M. Janicke, B.F. Chmelka, Science 1993, 261, 1299. Q.S. Huo, D.I. Margolese, U. Ciesla, P. Feng, T.E. Gier, P. Sieger, R. Leon, P.M. Petroff, F. Schüth, G.D. Stucky, Nature 1994, 368, 317. Q.S. Huo, D.I. Margolese, U. Ciesla, D.G. Demuth, P. Feng, T.E. Gier, P. Sieger, A. Firouzi, B.F. Chmelka, F. Schüth, G.D. Stucky, Chem. Mater. 1994, 6, 1176. G.J. A.A. Soler-Illia, E.L. Crepaldi, D. Grosso, C. Sanchez, Curr. Opin. Colloid In. 2003, 8, 109. G.J. A.A. Soler-Illia, J. Patarin, B. Lebeau, C. Sanchez, Chem. Rev. 2002, 102, 4093. P.T. Tanev, T.J. Pinnavaia, Science 1995, 267, 865.

4 Ordered mesoporous silica: synthesis and applications

[32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68]

|

95

J. Patarin, B. Lebeau, R. Zana, Curr. Opin. Colloid In. 2002, 7, 107. Q.S. Huo, D.I. Margolese, G.D. Stucky, Chem. Mater. 1996, 8, 1147. D. Khushalani, A. Kuperman, G.A. Ozin, K. Tanaka, N. Coombs, M.M. Olken J. Garcés, Adv. Mater. 1995, 7, 842. T.-W. Kim, R. Ryoo, M. Kruk, K.P. Gierszal, M. Jaroniec, S. Kamiya, O. Terasaki, J. Phys. Chem. B 2004, 108, 11480. F. Kleitz, W. Schmidt, F. Schüth, Microporous Mesoporous Mater. 2003, 65, 1. M. Kruk, M. Jaroniec, C.H. Ko, R. Ryoo, Chem. Mater. 2000, 12, 1961. M. Choi, W. Heo, F. Kleitz, R. Ryoo, Chem. Commun. 2003, 1340. E.G. Langeroudi, F. Kleitz, M.C. Iliuta, F. Larachi, J. Phys. Chem. C 2009, 113, 21866. V. Reichhardt, R. Guillet-Nicolas, M. Thommes, B. Klösgen, T. Nylander, F. Kleitz, V. Alfredsson, Phys. Chem. Chem. Phys., 2012, 14, 5651. F. Hoffmann, M. Cornelius, J. Morell, M. Fröba, Angew. Chem. Int. Ed. 2006, 45, 3216. Q. He, J. Shi, F. Chen, M. Zhu, L. Zhang, Biomaterials 2010, 31, 3335. Y.-S. Lin, C.L. Haynes, J. Am. Chem. Soc. 2010, 132, 4834. M. Al Shamsi, M.T. Al Samri, S. Al-Salam, W. Conca, S. Shaban, S. Benedict, S. Tariq, A.V. Biradar, H.S. Penefsky, T. Asefa, A.-K. Souid, Chem. Res. Toxicol. 2010, 23, 1796. S. Al-Salam, G. Balhaj, S. Al-Hammadi, M. Sudhadevi, S. Tariq, A.V. Biradar, T. Asefa, A.-K. Souid, Toxicol. Sci. 2011, 122, 86. S. Hitz, R. Prins, J. Catal. 1997, 168, 194. C.M. Yang, B. Zibrowius, W. Schmidt, F. Schüth, Chem. Mater. 2003, 15, 3739. B. Tian, X. Liu, C.Z. Yu, F. Gao, Q. Luo, S. Xie, B. Tu, D.Y. Zhao, Chem. Commun. 2002, 1186. S. Kawi, M.W. Lai, Chem. Commun. 1998, 1407. M.T.J. Keene, R. Denoyel, P.L. Llewellyn, Chem. Commun. 1998, 2203. A. Rumplecker, F. Kleitz, E.L. Salabas, F. Schüth, Chem. Mater. 2007, 19, 485. Q.S. Huo, R. Leon, P.M. Petroff, G.D. Stucky, Science 1995, 268, 1324. Y. Sakamoto, I. Diaz, O. Terasaki, D.Y. Zhao, J. Perez-Pariente, J.M. Kim, G.D. Stucky, J. Phys. Chem. B 2002, 106, 3118. F. Kleitz, T. Czuryszkiewicz, L.A. Solovyov, M. Lindén, Chem. Mater. 2006, 18, 5070. F. Kleitz, S.H. Choi, R. Ryoo, Chem. Commun. 2003, 2136. J.R. Matos, M. Kruk, L.P. Mercuri, M. Jaroniec, L. Zhao, T. Kamiyama, O. Terasaki, T.J. Pinnavaia, Y. Liu, J. Am. Chem. Soc. 2003, 125, 821. J. Fan, C.Z. Yu, F. Gao, J. Lei, B. Tian, L. Wang, Q. Luo, B. Tu, W. Zhou, D.Y. Zhao, Angew. Chem. Int. Ed. 2003, 42, 3146. F. Kleitz, D. Liu, G.M. Anilkumar, I.S. Park, L.A. Solovyov, A.N. Shmakov, R. Ryoo, J. Phys. Chem. B 2003, 107, 14296. S.S. Kim, T.R. Pauly, T.J. Pinnavaia, Chem. Commun. 2000, 1661. S.S. Kim, A. Karkambar, T.J. Pinnavaia, M. Kruk, M. Jaroniec, J. Phys. Chem. B 2001, 105, 7663. D. Zhao, Q. Huo, J. Feng, B. F. Chmelka, G. D. Stucky, J. Am. Chem. Soc. 1998, 120, 6024. K. Cassiers, T. Linssen, M. Mathieu, M. Benjelloun, K. Schrijnemakers, P. van der Voort, P. Cool, E.F. Vansant, Chem. Mater. 2002, 14, 2317. R. Ryoo, C.H. Ko, M. Kruk, V. Antochsuk, M. Jaroniec, J. Phys. Chem. B 2000, 104, 11465. M. Impéror-Clerc, P. Davidson, A. Davidson, J. Am. Chem. Soc. 2000, 122, 11925. P.I. Ravikovitch, A.V. Neimark, J. Phys. Chem. B 2001, 105, 6817. A. Galarneau, H. Cambon, F. DiRenzo, R. Ryoo, M. Choi, F. Fajula, New J. Chem. 2003, 27, 73. Y. Sakamoto, M. Kaneda, O. Terasaki, D.Y. Zhao, J.M. Kim, G.D. Stucky, H.J. Shin, R. Ryoo, Nature 2000, 408, 449. Y. Sakamoto, T.W. Kim, R. Ryoo, O. Terasaki, Angew. Chem. Int. Ed. 2004, 42, 5231.

96 | Part I Functional materials: Synthesis and applications Y. Doi, A. Takai, Y. Sakamoto, O. Terasaki, Y. Yamauchi, K. Kuroda, Chem. Commun. 2010, 46, 6365. [70] F. Kleitz, F. Bérubé, R. Guillet-Nicolas, C.M. Yang, M. Thommes, J. Phys. Chem. C 2010, 114, 9344. [71] T.-W. Kim, L.A. Solovyov, J. Mater. Chem. 2006, 16, 1445. [72] X. Liu, B. Tian, C. Yu, F. Gao, S. Xie, B. Tu, R. Che, L .-M. Peng, D. Zhao, Angew. Chem. Int. Ed. 2002, 41, 3876. [73] Y.Q. Wang, C.M. Yang, B. Zibrowius, B. Spliethoff, M. Lindén, F. Schüth, Chem. Mater. 2003, 15, 5029. [74] K. Flodström, V. Alfredsson, N. Källrot, J. Am. Chem. Soc. 2003, 125, 4402. [75] M. Kruk, M. Jaroniec, A. Sayari, J. Phys. Chem. B 1997, 101, 583. [76] M. Kruk, M. Jaroniec, R. Ryoo, S.H. Joo, Chem. Mater. 2000, 12, 1414. [77] U. Ciesla, F. Schüth, Microporous Mesoporous Mater. 1999, 27, 131. [78] J.M. Kim, Y. Sakamoto, Y.K. Hwang, Y.U. Kwon, O. Terasaki, S.-E. Park, G.D. Stucky J. Phys. Chem. B 2002, 106, 2552. [79] P. Kipkemboi, A. Fogden, V. Alfredsson, K. Flodström, Langmuir 2001, 17, 5398. [80] K. Flodström, V. Alfredsson, Microporous Mesoporous Mater. 2003, 59, 167. [81] K. Kushalani, A. Kuperman, G. A. Ozin, K. Tanaka, J. Garces, M. M Olken, N. Coombs, Adv. Mater. 1995, 7, 842. [82] M. Grün, K.K. Unger, A. Matsumoto, K. Tsutsumi, Microporous Mesoporous Mater. 1999, 27, 207. [83] A. Galarneau, M. Nader, F. Guenneau, F. DiRenzo, A. Gedeon, J. Phys. Chem. C 2007, 111, 8268. [84] P.F. Fulvio, S. Pikus, M. Jaroniec, J. Mater. Chem. 2005, 15, 5049. [85] J. Fan, C.Z. Yu, L. Wang, B. Tu, D.Y. Zhao, Y. Sakamoto, O. Terasaki, J. Am. Chem. Soc. 2001, 123, 12113. [86] B. Coasne, A. Galarneau, R.J.M. Pellenq, F. Di Renzo, Chem. Soc. Rev. 2013, 42, 4141. [87] C.J. Gommes, H. Friedrich, M. Wolters, P.E. de Jongh, K.P. de Jong, Chem. Mater. 2009, 21, 1311. [88] H. Tüysüz, C.W. Lehmann, H. Bongard, B. Tesche, R. Schmidt, F. Schüth, J. Am. Chem. Soc. 2008, 130, 11510. [89] M. Thommes, Chem. Ing. Tech. 2010, 82, 1059. [90] R. Ryoo, J.M. Kim, Chem. Commun. 1995, 711. [91] K.J. Edler, J.W. White, Chem. Mater. 1997, 9, 1226. [92] R. Ryoo, S.H. Joo, J.M. Kim, J. Phys. Chem. B 1999, 103, 7435. [93] S. Manet, J. Schmitt, M. Impéror-Clerc, V. Zholobenko, D. Durand, C.L.P. Oliviera, J.S. Pedersen, C. Gervais, N. Baccile, F. Babonneau, I. Grillo, F. Meneau, C.J. Rochas, J. Phys. Chem. B 2011, 115, 11330. [94] R. Guillet-Nicolas, F. Bérubé, T.W. Kim, M. Thommes, F. Kleitz, Stud. Surf. Sci. Catal. 2008, 174, 141. [95] B. Yang, C. Guo, S. Chen, J. Ma, J. Wang, X. Liang, L. Zheng, H. Liu, J. Phys. Chem. B 2006, 110, 23068. [96] A. Kabalnov, U. Olsson, H. Wennerström, J. Phys. Chem. 1995, 98, 6220. [97] C.Z. Yu, B. Tian, J. Fan, G.D. Stucky, D.Y. Zhao, J. Am. Chem. Soc. 2002, 124, 4556. [98] C. Boissière, M.A.U. Martines, M. Tokumoto, A. Larbot, E. Prouzet, Chem. Mater. 2003, 15, 509. [99] M. Lindén, P. Ågren, S. Karlsson, P. Bussian, H. Amenitsch, Langmuir 2000, 16, 5831. [100] F. Kleitz, J. Blanchard, B. Zibrowius, F. Schüth, P. Ågren, M. Lindén, Langmuir 2002, 18, 4863. [101] J.M. Kim, S.K. Kim, R. Ryoo, Chem. Commun. 1998, 259. [69]

4 Ordered mesoporous silica: synthesis and applications

|

97

[102] K. Schumacher, M. Grün, K.K. Unger, Microporous Mesoporous Mater. 1999, 27, 201. [103] J.S. Lettow, Y.J. Han, P. Schmidt-Winkel, P. Yang, D.Y. Zhao, Langmuir 2002, 16, 8291. [104] J. Sun, H. Zhang, D. Ma, Y. Chen, X. Bao, A. Klein-Hoffmann, N. Pfander, D.S. Su, Chem. Commun. 2005, 5343. [105] P. Schmidt-Winkel, W.W. Lukens, D.Y. Zhao, P. Yang, B.F. Chmelka, G.D. Stucky, J. Am. Chem. Soc. 1999, 121, 254. [106] D.H. Chen, Z. Li, Y. Wan, X.J. Tu, Y.F. Shi, Z.X. Chen, W. Shen, C.Z. Yu, B. Tu, D.Y. Zhao, J. Mater. Chem. 2006, 16, 1511. [107] P. Holmqvist, P. Alexandridis, B. Lindman, Macromolecules 1997, 30, 6788. [108] P. Holmqvist, P. Alexandridis, B. Lindman, J. Phys. Chem. B 1998, 102, 1149. [109] F. Kleitz, T.W. Kim, R. Ryoo, Langmuir 2006, 22, 440. [110] A. Taguchi, F. Schüth, Microporous Mesoporous Mater. 2005, 77, 1. [111] F. Bérubé, B. Nohair, F. Kleitz, S. Kaliaguine, Chem. Mater. 2010, 22, 1988. [112] V. Clauda, A. Schlossbauer, J. Kecht, A. Zurner, T. Bein, J. Am. Chem. Soc. 2009, 131, 11361. [113] A. Stein, B.J. Melde, R.C. Schroden, Adv. Mater. 2000, 12, 1403. [114] M.H. Lim, A. Stein, Chem. Mater. 1999, 11, 3285. [115] J.A. Melero, R. van Grieken, G. Morales, Chem. Rev. 2006, 106, 3790. [116] E.L. Burkett, S.D. Sims, S. Mann, Chem. Commun. 1996, 1367. [117] D.J. Macquarrie, Chem. Commun. 1996, 1961. [118] D. Margolese, J.A. Melero, S.C. Christiansen, B.F. Chmelka, G.D. Stucky, Chem. Mater. 2000, 12, 2448. [119] H. Yen, Y. Seo, R. Guillet-Nicolas, S. Kaliaguine, F. Kleitz, Chem. Commun. 2011, 47, 10473. [120] Z.H. Luan, E.M. Maes, P.A.W. van der Heide, D.Y. Zhao, R.S. Czernuszewicz, L. Kevan, Chem. Mater. 1999, 11, 3680. [121] M. Choi, F. Kleitz, D.N. Liu, H.Y. Lee, W.S. Ahn, R. Ryoo, J. Am. Chem. Soc. 2005, 127, 1924. [122] R. Guillet-Nicolas, J.-L. Bridot, Y. Seo, M.-A. Fortin, F. Kleitz, Adv. Funct. Mater. 2011, 21, 4653. [123] R.C. Hayward, P. Alberius-Henning, B.F. Chmelka, G.D. Stucky, Microporous Mesoporous Mater. 2001, 44-45, 619. [124] S.H. Tolbert, A. Firouzi, G.D. Stucky, B.F. Chmelka, Science 1997, 278, 264. [125] P. Yang, T. Deng, D. Zhao, P. Feng, D. Pine, B.F. Chmelka, G.M. Whitesides, G.D. Stucky, Science 1998, 282, 2244. [126] H. Miyata, K. Kuroda, Chem. Mater. 2000, 12, 49. [127] H. Fan, S. Reed, T. Bear, R. Schunk, G. P. López, C.J. Brinker, Microporous Mesoporous Mater. 2001, 44-45, 625. [128] J.-H. Smått, S. Schunk, M. Lindén Chem. Mater. 2003, 15, 2354–2361. [129] F. Maddox Sayler, M. G. Bakker, J.-H.Smått, M.Lindén, J. Phys. Chem. C 2010, 114, 8710. [130] P. Yang, D.Y. Zhao, B.F. Chmelka, G.D. Stucky, Chem. Mater. 1998, 10, 2033. [131] N.A. Melosh, P. Lipic, F.S. Bates, F. Wudl, G.D. Stucky, G.H. Fredrickson, B.F. Chmelka, Macromolecules 1999, 32, 4332. [132] K. Yano, Y. Fukushima, J. Mater. Chem. 2004, 14, 1579. [133] C. Boissière, D. Grosso, A. Chaumonnot, L. Nicole, C. Sanchez, Adv. Mater. 2011, 23, 599. [134] C. Sanchez, C. Boissière, D. Grosso, C. Laberty, L. Nicole, Chem. Mater. 2008, 20, 682. [135] F. Kleitz, F. Marlow, G.D. Stucky, F. Schüth, Chem. Mater. 2001, 13, 3587. [136] S.-H. Wu, C.-H. Mou, H.-P. Lin, Chem. Soc. Rev. 2013, 42, 3862. [137] Q. Huo, J. Feng, F. Schüth, G. D. Stucky, Chem. Mater. 1997, 9, 14. [138] W. Stöber, A. Funk, E. Bohn, J. Colloid Interfaces Sci. 1968, 26, 62. [139] T. Martin, A. Galarneau, F. DiRenzo, F. Fajula, D. Plee, Angew. Chem. Int. Ed. 2002, 41, 2590. [140] A. Galarneau, J. Iapichella, K. Bonhomme, F. DiRenzo, P. Kooyman, O. Terasaki, F. Fajula, Adv. Funct. Mater. 2006, 16, 1657.

98 | Part I Functional materials: Synthesis and applications [141] [142] [143] [144] [145] [146] [147] [148] [149] [150] [151] [152] [153] [154] [155] [156] [157] [158] [159] [160] [161] [162] [163] [164] [165] [166] [167] [168] [169] [170] [171] [172] [173] [174] [175] [176] [177] [178] [179]

P.J. Bruinsma, A.Y. Kim., J. Liu, S. Baskaran, Chem. Mater. 1997, 9, 2507. I.I. Slowing, B.G. Trewyn, S. Giri, V.S.-Y. Lin, Adv. Funct. Mater. 2007, 17, 1225. M. Grün, I. Lauer, K.K. Unger, Adv. Mater. 1997, 9, 254. Q. Cai, Z.-S. Luo, W.-Q. Pang, Y.-W. Fan, X.-H. Chen, F.-Z. Cui, Chem. Mater. 2001, 13, 258. C.E. Fowler, D. Khushalani, B. Lebeau, S. Mann, Adv. Funct. Mater. 2001, 13, 649. B.G. Trewyn, I. Slowing, S. Giri, H.-T. Chen, V.S.-Y. Lin, Acc. Chem. Res. 2007, 40, 846. T.-W. Kim, P.-W. Chung, I.I. Slowing, M. Tsunoda, E.S. Yeung, V.S.-Y. Lin, Nano Lett. 2008, 8, 3724. T.-W. Kim, P.-W. Chung, V.S.-Y. Lin, Chem. Mater. 2010, 22, 5093. M.-H. Kim, H.-K. Na, Y.-K. Kim, S.-R. Ryoo, H.S. Cho, K.E. Lee, H. Jeon, R. Ryoo, D.-H. Min, ACS Nano 2011, 5, 3568. Q. He, J. Shi, J. Zhao, Y. Chen, F. Chen, J. Mater. Chem. 2009, 19, 6498. T.W. Kim, I.I. Slowing, P.-W. Chung, V.S.Y. Lin, ACS Nano 2011, 5, 360. P. Linton, V. Alfredsson, Chem. Mater. 2008, 20, 2878. Y. Han, J.Y. Ying, Angew. Chem. Int. Ed. 2005, 44, 288. F. Gao, P. Botella, A. Corma, J. Blesa, L. Dong, J. Phys. Chem. B 2009, 113, 1796. S.B. Hartono, W. Gu, F. Kleitz, J. Liu, L. He, A.P.J. Middelberg, C.Z. Yu, G.Q. Lu, S.Z. Qiao, ACS Nano 2012, 6, 2104. M.E. Davis, Nature 2002 , 417, 813. A. Corma, Chem. Rev. 1997, 97, 2373. C. Perego, R. Millini, Chem. Soc. Rev. 2013, 42, 3956. A. Walcarius, Chem. Soc. Rev. 2013, 42, 4098. A. Bernardos, E. Aznar, M.D. Marcos, R. Martínez-Máñez, F. Sancenón, J. Soto, J.M. Barat, P. Amorós, Angew. Chem. Int. Ed. 2009, 48, 5884. F. Muhammad, M. Guo, W. Qi, F. Sun, A. Wang, Y. Guo, G. Zhu, J. Am. Chem. Soc. 2011, 133, 8778. E. Aznar, M.D. Marcos, R. Martínez-Máñez, F. Sancenón, J. Soto, P. Amorós, C. Guillem, J. Am. Chem. Soc. 2009, 131, 6833. C. Coll, L. Mondragón, R. Martínez-Máñez, F. Sancenón, M. D. Marcos, J. Soto, P. Amorós, E. Pérez-Payá, Angew. Chem. Int. Ed. 2011, 50, 2138. J.M. Rosenholm, C. Sahlgren, M. Lindén, Curr. Drug Targets 2011, 12, 1166. A.E. Garcia-Bennett, Nanomedecine 2011, 6, 867. V. Mamaeva, C. Sahlgren, M. Lindén, Adv. Drug Deliver. Rev. 2013, 65, 689. D. Tarn, C.E. Ashley, M. Xue, E.C. Carnes, J.I. Zink, C.J. Brinker, Accounts Chem. Res. 2013, 46, 792. W.X. Mai, H. Meng, Integr. Biol. 2013, 5, 19. N.-T. Chen, S.-H. Cheng, J.S. Souris, C.-T. Chen, C.-Y. Mou, L.-W. Lo, J. Mater. Chem. B 2013, 1, 3128. F. Tang, L. Li, S. Chen, Adv. Mater. 2012, 24, 1504. Y.-W. Yang, Med. Chem. Commun. 2011, 2, 1033. P. Yang, S. Gai, J. Lin, Chem. Soc. Rev. 2012, 41, 3679. C. Park, K. Oh, S.C. Lee, C. Kim, Angew. Chem. Int. Ed. 2007, 46, 1455. F. Muhammad, M. Guo, W. Qi, F. Sun, A. Wang, Y. Guo, G. Zhu, J. Am. Chem. Soc. 2011, 133, 8778. J.M. Rosenholm, V. Mamaeva, C. Sahlgren, M. Lindén, Nanomedecine 2012, 7, 111. M. Hartmann, Chem. Mater. 2005, 17, 4577. J. Andersson, J. Rosenholm, S. Areva, M. Lindén, Chem. Mater. 2004, 16, 4160. A. Datt, I. El-Maazawi, S. C. Larsen, J. Phys. Chem. C 2012, 116, 18358. J.M. Rosenholm, C. Sahlgren, M. Lindén, Nanoscale 2010, 2, 1870.

4 Ordered mesoporous silica: synthesis and applications

[180] [181] [182] [183] [184] [185] [186] [187] [188] [189] [190] [191] [192] [193] [194] [195] [196] [197] [198] [199] [200] [201] [202] [203] [204] [205] [206] [207] [208] [209] [210] [211] [212] [213] [214] [215] [216] [217]

| 99

R. Langer, Nature 1998, 392, S5. F. Tang, L. Li, D. Chen, Adv. Mater. 2012, 24, 1504. Q. He, J. Shi, Adv. Mater. 2014, 26, 391. L. Pan, Q. He, J. Liu, Y. Chen, M. Ma, L. Zhang, J. Shi, J. Am. Chem. Soc. 2012, 134, 5722. N.K. Mal, M. Fujiwara, Y. Tanaka, Nature 2003, 421, 350. S. Angelos, N.M. Khashab, Y.-W. Yang, A. Trabolsi, H.A. Khatib, J.F. Stoddart, J.I. Zink, J. Am. Chem. Soc. 2009, 131, 12912. J.-H. Park, Y.-H. Lee, S.-G. Oh, Macromol. Chem. Phys. 2007, 208, 2419. A. Popat, J. Liu, G.Q. Lu, S.Z. Qiao, J. Mater. Chem. 2012, 22, 11173. A. Popat, B.P. Ross, J. Liu, S. Jambhrunkar, F. Kleitz, S.Z. Qiao Angew. Chem. Int. Ed. 2012, 51, 12486. A. Popat, S.B. Hartono, F. Stahr, J. Liu, S.Z. Qiao, C.Q. Lu, Nanoscale 2011, 3, 2801. M. Manzano, V. Aina, C.O. Aréan, F. Balas, V. Cauda, M. Colilla, M.R. Delgado, M. Vallet-Regì, Chem. Eng. J. 2008, 137, 30. M. Van Speybroeck, R. Mellaerts, T.D. Thi, J.A. Martens, J. Van Humbeeck, P. Annaert, G. Van Den Mooter, P. Augustijns, J. Pharm. Sci. 2011, 100, 4864. I.-J. Fang, B.G. Trewyn, Nanomedecine 2012, 508, 41. H. Chen, R. Langer, Adv. Drug Deliver. Rev. 1998, 34, 339 L.S. Liu, M.L. Fishman, J. Kost, K.B. Hicks, Biomaterials 2003, 24, 3333. S.W. Song, K. Hidajat, S. Kawi, Chem. Commun. 2007, 4396. US Pharmacopoeia and national formulary, Drug release, delayed-release (enteric-coated) articles and general drug release standard, 2017, Rockville, MD: USP Convention Inc., 2002. C.-H. Lee, L.-W. Lo, C.-Y. Mou, C.-S. Yang, Adv. Funct. Mater. 2008, 18, 3283. S.-H. Cheng, W.-N. Liao, L.-M. Chen, C.-H. Lee, J. Mater. Chem. 2011, 21, 7130. G. Sun, Y. Chang, S. Li, Q. Li, R. Xu, J. Gu, E. Wang, Dalton Trans. 2009, 38, 4481. S. Angelos, E. Choi, F. Vögtle, L. De Cola, J. I. Zink, J. Phys. Chem. C 2007, 111, 6589. Y. Ma, L. Zhou, H. Zheng, L. Xing, C. Li, J. Cui, S. Che, J. Mater. Chem. 2011, 21, 9483. D.M.R. Georget, S.A. Barker, P.S. Belton, Eur. J. Pharm. Biopharm. 2008, 69, 718. X. Liu, Q. Sun, H. Wang, L. Zhang, J.-Y. Wang, Biomaterials 2005, 26, 109. X. Zhang, M. Oulad-Abdelghani, A.N. Zelkin, Y. Wang, Y. Haîkel, D. Mainard, J.-C. Voegel, F. Caruso, N. Benkirane-Jessel, Biomaterials 2010, 31, 1699. M. Subirade, J. Gueguen, K.D. Schwenke, J. Colloid Interface Sci. 1992, 152, 442. R. Guillet-Nicolas, A. Popat, J.-L. Bridot, G. Monteith, S.Z. Qiao, F. Kleitz, Angew. Chem. Int. Ed. 2013, 52, 2318. X. Feng, G.E. Fryxell, L.Q. Wang, A.Y. Kim, J. Liu, K.M. Kemner, Science 1997, 276, 923. L. Bois, A. Bonhommé, A. Ribes, B. Pais, G. Raffin, F. Tessier, Colloids Surf. A 2003, 221, 221. K.F. Lam, K. L. Yeung, G. McKay, Environ. Sci. Technol. 2007, 41, 3329. P. Makowski, X. Deschanels, A. Grandjean, D. Mayer, G. Toquer, F. Goettmann, New. J. Chem. 2012, 36, 531. P. Makowski, A. Thomas, P, Kuhn, F. Goettmann, Energy Environ. Sci. 2009, 2, 480. G.E. Fryxell, Y. Lin, S. Fiskum, J.C. Birnbaum, H. Wu, K. Kemner, S. Kelly, Environ. Sci. Technol. 2005, 39, 1324. D. Butler, L. Stricker, Nature 2011, 472, 274. P.J. Lebed, K. de Souza, F. Bilodeau, D. Larivière, F. Kleitz, Chem. Commun. 2011, 47, 11525. P.J. Lebed, J.-D. Savoie, J. Florek, F. Bilodeau, D. Larivière, F. Kleitz, Chem. Mater. 2012, 24, 4166. X. Kang, C. Li, Z. Cheng, P. Ma, Z. Hou, J. Lin, WIREs Nanomed Nanobiotechnol. 2014, 6, 80. J. Feng, H. Zhang, Chem. Soc. Rev. 2013, 42, 387.

100 | Part I Functional materials: Synthesis and applications [218] M.S. Moorthy, P.K. Tapaswi, S.S. Park, A. Mathew, H.-J. Cho, C.-S. Ha, Microporous Mesoporous Mater. 2013, 180, 162. [219] S. Massari, M. Ruberti, Resources Policy 2013, 38, 36. [220] C.K. Gupta, N. Krishnamurthy, Int. Mater. Rev. 1992, 37, 197. [221] J. Florek, F. Chalifour, F. Bilodeau, D. Larivière, F. Kleitz, Adv. Funct. Mater. 2014, 24, 2668. [222] M. Khayet, Adv. Colloid Interface Sci. 2011, 164, 56. [223] M.S. El-Bourawi, Z. Ding, R. Ma, M. Khayet, J. Membr. Sci. 2006, 285, 4. [224] Z. Steiner, J. Miao and R. Kasher, Chem. Commun. 2011, 47, 2384. [225] J.-P. Mericq, S. Laborie, C. Cabassud, Water Res. 2010, 44, 5260. [226] H.L. Castricum, A. Sah, R. Kreiter, D.H.A. Blank, J.F. Vente, J.E. Ten Elshof, Chem. Commun. 2008, 1103. [227] N. Liu, R. A. Assink, C.J. Brinker, Chem. Commun. 2003, 370. [228] Y.T. Chua, C.X.C. Lin, F. Kleitz, X.S. Zhao, S. Smart, Chem. Commun. 2013, 49, 4534.

A. Ritcey

5 Nanoparticles: Properties and applications 5.1 Introduction The ever-increasing importance of nanoparticles to the development of functional materials is incontestable. A simple keyword search of the scientific literature [1] reveals the accelerating growth of nanoparticle research over the past decade and, as illustrated in Fig. 5.1, nearly 60 000 publications treating this subject appeared in 2012 alone. New scientific journals dedicated to the field of nanoscience have also been created [2] and are rapidly establishing their place among high impact periodicals.

Number of publications

60000 50000 40000 30000 20000 10000 0

2000

2002

2004 2006 2008 Publication year

2010

2012

Fig. 5.1. Annual number of publications retrieved by a keyword search for “nanoparticles” [1].

Within this context, the following pages can only scratch the surface of the title subject and cannot in any way be considered a complete review of the literature. It is hoped, however, that the selected examples will provide the reader with a valid appreciation of the enormous potential of functional nanoparticles in materials science. In addition, this chapter treats some fundamental issues of nanoparticle preparation and handling through a general approach that is relevant to many diverse specific systems.

5.2 Synthetic methods Most introductions to the fabrication of nanostructures begin with a reference to the contrasting “top-down” and “bottom-up” approaches, terms first applied to the field

102 | Part I Functional materials: Synthesis and applications of nanoscience by the Foresight Institute in 1989 [3]. As their name implies, top-down methods involve the creation of nanosized entities from larger blocks of matter. Examples include mechanical size reduction by crushing and grinding as well as more sophisticated lithographic techniques. Bottom-up approaches, on the other hand, seek to build nanostructures from smaller components, typically atoms or molecules, and therefore frequently involve elements of self-assembly or supramolecular chemistry. In the case of nanoparticles, the large majority of preparation methods are based on controlled precipitation, that is, the nucleation and growth of particles from a supersaturated solution. Supersaturation can be achieved by a chemical reaction, such as reduction, hydrolysis, condensation or decomposition, which converts a soluble precursor to the desired, sparingly soluble, material.

5.2.1 Particle nucleation and growth Any solution process that leads to the generation of a sparingly soluble species will result in the eventual nucleation of a second, typically solid, phase. The thermodynamics of nucleation includes two considerations: the decrease in free energy associated with the phase transition (usually crystallization), and the increase in free energy associated with the necessary creation of the solid-liquid interface. The free energy of nucleation can therefore be written as ΔGnucleation = n ⋅ ΔḠ crystallization + ΔGsurface ,

(5.1)

where ΔḠ crystallization is the molar free energy of crystallization and n is the number of moles of the substance i in the nucleus. ΔḠ crystallization depends on the concentration of i, [i], according to ΔḠ crystal = RT ln

[i]sat , [i]

(5.2)

and thus becomes increasingly negative as the concentration exceeds saturation at [i]sat . The second contribution to ΔGnucleation can be expressed as ΔGsurface = σ ⋅ A,

(5.3)

where σ is the interfacial energy and A is the surface area of the nucleus. This component is always positive. Nucleation can only occur if the crystallization process liberates sufficient energy to cover the cost of creating the new interface between the particle seed and the solution. Importantly, although the absolute value of both ΔGsurface and ΔGcrystallization increase with nucleus size, the former is proportional to the surface area, whereas the latter is proportional to the volume. In the simple case of a spherical nucleus of radius r, ΔGsurface varies with r2 and ΔGcrystallization with r3 . These dependencies (dashed lines) are illustrated schematically in Fig. 5.2 and lead to the presence

5 Nanoparticles: Properties and applications

|

103

of a maximum, labeled rcr , in the sum of the two contributions. Nuclei smaller than the critical size rcr are thermodynamically unstable and will spontaneously disappear as indicated by the left-pointing arrow. Nuclei larger than the critical size, however, will spontaneously grow, since in this size regime increasing size corresponds to a decrease in free energy, as indicated by the arrow to the right.

ΔGnucleation

ΔGsurface

rcr

Nucleus size ΔGcrystallization

Fig. 5.2. Free energy of nucleation as a function of nucleus size. Contributions from surface and crystallization energies are shown as dashed lines. rcr denotes the critical nucleus size and the arrows indicate the direction of spontaneous size evolution.

How can particle nucleation occur if small nuclei are unstable and disappear rather than grow? The answer lies in random concentration fluctuations which allow for the spontaneous formation of aggregates that, upon formation, already exceed the critical size. Clearly, the probability of the formation of stable nuclei will therefore decrease with increasing critical size. As the concentration is increased beyond saturation, ΔGcrystallization becomes increasingly negative, permitting the formation of stable nuclei of smaller size. Nucleation, in turn, becomes more probable. These thermodynamic considerations explain the general behavior illustrated by the well-known LaMer diagram provided in Fig. 5.3 [4, 5]. The process of nanoparticle formation can be divided into three phases. In the early stages, immediately following the initiation of the reaction responsible the generation of the particle forming species, the concentration is lower than the solubility and there is thus no thermodynamic driving force for the formation of a second phase. As the reaction proceeds, the concentration increases, eventually reaching saturation. Particle nucleation, however, does not occur immediately because of the free energy cost of creating a new interface. As outlined above, nucleation requires that the solution become super-saturated to the point where ΔḠ crystallization , as given by equation (5.2), is sufficient to cover the cost of the new surface. This point is labeled the critical nucleation concentration in Fig. 5.3. Once the critical concentration is reached, nucleation begins, removing material from solution. As nucleation proceeds, the solution concentration decreases and eventually falls below the critical value for nuclei formation. After this point, no new nuclei are formed and any material remaining in solution or generated by further reaction can only contribute to particle growth. The concentration profile of the LaMer diagram has important implications for the

104 | Part I Functional materials: Synthesis and applications Concentration

Critical nucleation concentration

Solubility

PreNucleation nucleation

Growth

Time

Fig. 5.3. LaMer diagram illustrating the evolution of solution concentration during the nucleation and growth process.

size distribution of the resulting particle population. Conditions which restrict the duration of the nucleation stage favor populations of low polydispersity. Simultaneous nucleation, followed by growth in a common medium, will lead to particles of quasiidentical size.

5.2.2 Synthesis in inverse micelles In the presence of an appropriate surfactant, small amounts of water can be dispersed in an organic medium to form inverse micelles. As illustrated in Fig. 5.4, an inverse micelle can be viewed as a nanometer-sized water droplet surrounded by a surfactant layer. The capacity to dissolve reactants in the water core, combined with the constant exchange of the aqueous phase between micelles during collisions, allows for chemical reactions to be carried out in the confined volume of the enclosed water reservoir. In general, if the reaction leads to a solid product, nanoparticles are obtained. This method has been employed for the preparation of nanoparticles from a diverse variety of materials, including metals, metal oxides, and even organic compounds [6]. The inverse micelle approach typically offers exceptional control of particle size, which can be conveniently modulated through the variation of simple system parameters such as the relative quantities of solvent, surfactant, and reagents. Furthermore, in certain cases, essentially monodisperse size-distributions are obtained. For example, we have employed the inverse micelle approach to prepare single crystal yttrium fluoride nanoparticles with the exquisite size control illustrated in Fig. 5.5 [7, 8].

5 Nanoparticles: Properties and applications

|

105

H2 O

(a)

50 nm

(b)

X300000 1300x1030 pixels 4/29/2004 4.16852 s

Fig. 5.4. Schematic representations of (a) an inverse micelle, and (b) content exchange during micellar collisions.

Fig. 5.5. (H3 O)Y3 F10 nanocrystals prepared by the inverse micelle method [7, 8].

5.3 Particle aggregation and stabilization of colloidal suspensions The free energy of the interface between two immiscible phases is always positive. If this were not the case, the free energy of the system would decrease with increasing interfacial area and the phases would spontaneously divide into progressively smaller and smaller domains, leading to miscibility. Since interfacial energy is positive, suspensions of nanoparticles are necessarily thermodynamically unstable. Kinetically stable colloidal suspensions can, however, be obtained by ensuring sufficient interparticle repulsion. Common stabilization strategies fall into two main categories; electrostatic and steric. Electrostatic stabilization involves the introduction of a surface charge to the nanoparticles and is primarily restricted to aqueous systems, since solvents other than water are unable to support the necessary charge separation. Many inorganic nanoparticles, such as those composed of sparingly soluble salts or metal oxides, are naturally charged when dispersed in water. Surface charge is frequently determined by equilibria controlling ion adsorption and dissolution or the protonation and deprotonation of surface moieties. In other cases, surface charge can be introduced by the adsorption of charged species, such as the citrate ion. Electrostatic repulsion between particles of like charge serves as a barrier to aggregation and, if sufficiently strong with respect to the van der Waals attraction, can result in a stable suspension. The interplay of van der Waals attractions and electrostatic repulsion was quantitatively captured by Derjaguin, Verwey, Landau, and Overbeek in what is commonly

106 | Part I Functional materials: Synthesis and applications known as DVLO theory [9]. The basic outcome of DVLO calculations is illustrated in Fig. 5.6. Both electrostatic repulsion, shown as the dashed line, and van der Waals attractions, represented by the dotted line, become stronger as particles are brought closer together. Importantly, the exact way in which each of the two forces varies with separation distance is different: interparticle van der Waals attractions vary inversely to distance squared, whereas electrostatic repulsion follows an exponential dependency. The net force is weakly attractive at large distances. However, as the particles approach each other, electrostatic repulsion increases faster than the attractive force. This leads to the appearance of a repulsive barrier in the net interaction curve. Particles able to cross this barrier will reach very small separation distances where attractive forces dominate and aggregation will occur.

Electrostatic repulsion

Interaction energy

Repulsive barrier

0

Van der Waals attraction

Interparticle distance Fig. 5.6. Interaction energy between two particles as a function of separation. The net interaction, shown as the solid line, is the sum of electrostatic repulsion and attractive van der Waals forces.

The stability of a colloidal suspension therefore depends on the height of the repulsive barrier with respect to the energy of particle collisions. If particles collide with sufficient energy to pass the repulsive barrier, they will fall into the attractive well and aggregate. Once in contact, there is no driving force for spontaneous re-dispersion, and in this respect, aggregation is irreversible. Collision energy is determined by the kinetic energy of the colliding particles, which depends on temperature and follows a Boltzman distribution centered at 3/2 kT. Importantly, at a given temperature, collisions are not all of the same energy and there is a nonzero probability that some will have sufficient energy to pass the barrier. In fact, given enough time, all of the particles in a suspension will eventually reach the aggregated state. The stability of a colloidal suspension therefore depends on the time scale of observation, and for most applications colloids are typically considered to be kinetically stable if the repulsive barrier

5 Nanoparticles: Properties and applications

|

107

exceeds 10 kT. The thermodynamically stable state remains the aggregated one; the repulsive barrier simply serves to delay the inevitable. The second common strategy for stabilization employs steric repulsion. In this approach, polymer chains are attached to the particle surface, either through chemical grafting or simple adsorption. As two particles come together, a repulsive force will be felt at the point of chain overlap. Although conceptually simple, as illustrated by Fig. 5.7, repulsive contributions from both osmotic pressure and chain deformation must be taken into account, and the quantitative treatment of steric stabilization rapidly becomes complex [10]. Steric stabilization can be employed in both aqueous and organic media, and with the appropriate choice of surface ligand nanoparticles can be dispersed in any carrier liquid.

Fig. 5.7. Schematic illustration of steric stabilization through surface grafting of polymer chains.

5.4 Colloidal quantum dots Among the panoply of functional nanoparticles, it is probably the family known as quantum dots which exhibits the most striking example of size-dependent properties. Quantum dots are nanocrystals of inorganic semi-conductor materials which possess unique optical properties arising from the electronic confinement imposed by the small size of the particles [11, 12]. The electronic structure of bulk semiconductors is characterized by two distinct bands, the valence band which contains the electrons that bond the material together, and, at higher energy, the conduction band, which is essentially devoid of electrons at low temperature. Energies between the valence band and the conduction band, that is, within the so-called band gap, are forbidden. Electrons can, however, be promoted from the valence band across the band gap into the conduction band, if supplied with sufficient energy. For a bulk semiconductor, optical absorption will therefore begin at the band gap energy and, because of the band nature of the excited state, the optical absorption spectrum will appear as a continuum above the band gap. In contrast to the band structure of bulk semiconductors, quantum dots are characterized by a series of discrete electronic quantized levels, and their optical spectra correspondingly exhibit discrete electronic transitions. The promotion of an electron from the valence band to the conduction band results in the creation of an electron-hole pair known as an exciton. The electron-hole pair is bound by Coulombic interactions which are relatively small when compared

108 | Part I Functional materials: Synthesis and applications to kT and, in a bulk semiconductor, considerable charge separation is possible. In the case of a quantum dot, particle size is reduced below the characteristic distance of charge separation for the electron-hole pair and confinement effects become pronounced [13]. The energies of the electron and hole levels are sensitive to the degree of confinement and, as a result, the optical absorption spectra of quantum dots depend strongly on their size. As the exciton becomes increasingly confined, the energy of the primary transition increases and the corresponding absorption band moves to shorter wavelengths, as illustrated for PbSe in Fig. 5.8.

Absorbance (arbitrary units)

8.1 nm

6.5 nm 4.8 nm 4.6 nm 3.6 nm 3.3 nm 800

1200

1600 2000 2400

Wavelength (nm)

Fig. 5.8. Absorption spectra of PbSe quantum dots of various sizes. Reprinted with permission from [13].

Quantum dots have received considerable attention as functional materials because of their luminescence properties. Here, exciton confinement also plays an important role by greatly enhancing the probability of radiative recombination. Radiative recombination is the return of an electron from the conduction band to the valence band accompanied by emission of a photon. The wavelength of emission is thus determined by the band gap and is highly size-dependent. This is illustrated in Fig. 5.9 for a series of CdSe quantum dots, removed from the reaction medium at different times during particle synthesis [14]. As the particles grow larger, the emission spectra systematically shift to longer wavelengths. Early research on quantum dots primarily involved materials exhibiting absorption and emission within the visible region of the spectrum. As recently discussed by Shirasaki et al. [15], quantum dots show great potential as emitters in thin film light-emitting devices. For this application, quantum dots offer several important advantages with respect to organic fluorophores. For example, the characteristic narrow emission band of quantum dots assures high color purity, and the tunability of the

5 Nanoparticles: Properties and applications

|

109

1.0

0.5 min 15 min 60 min 120 min

0.5

280°C 400

450

Normalized PL Intensity

10 min

0.0 500

550

600

650

700

Wavelength (nm) Fig. 5.9. Temporal evolution of the photoluminescence spectra of crude solutions of colloidal CdSe nanocrystals during their growth at 280°C. Adapted with permission from [14].

emission color through particle size allows for the fabrication of multicolor devices from a single material. More recently, research activity has been increasingly focused on the development of narrow band gap materials which emit in infrared [16–18]. Current efforts aim at infrared emission for applications in bioimaging, telecommunications, and solar cells. A solar cell is a device which converts the sun’s energy to electricity. The availability of efficient solar cells, amenable to large scale production at reasonable cost, will clearly have major positive repercussions for the environmental impact of future human activity and economic development. Numerous designs for solar cells have been proposed and all involve the following three basic processes: the absorption of light to generate electron-hole pairs, the separation of charge carriers, and the capture of the charges by external electrodes. One of the primary limitations of solar cell performance is the low efficiency of the primary photoconversion step [13]. The solar spectrum is relatively broad, and only photons absorbed by the solar cell can lead to charge separation. The incorporation of narrow gap quantum dots in hybrid solar cells can thus lead to improvements in light harvesting in the infrared region [19]. Quantum dots are also being developed for novel solar energy conversion strategies. For example, cell performance can be enhanced through a process known as multiple exciton generation [20]. In this process, single photons with energy surpassing twice that of the band gap can produce two or more electron-hole pairs. This leads to improved cell efficiency, since in the absence of multiple exciton generation, the excess energy of high energy photons is lost as heat. Significantly, the confinement of charge carriers

110 | Part I Functional materials: Synthesis and applications greatly enhances the probability of multiple exciton generation, leading to exciting prospects for the application of quantum dots in next generation solar cells [20]. Clearly the role of quantum dots as functional materials extends beyond the limited examples cited here. Furthermore, current research involves particles of increasingly complex composition and structure [18], as well as new paradigms, such as the control of properties through ligand-mediated processes [21]. The interested reader is encouraged to consult more complete treatises on the subject.

5.5 Metal nanoparticles

Electric field

Metal nanoparticles are ubiquitous and have been known since antiquity. The unique optical properties of colloidal noble metals, exploited in ancient times to stain glass, are today at the basis of exciting new developments in both fundamental science and the design of novel optical materials. The interaction of light with metallic nanostructures is dominated by the collective excitation of free electrons, known as the surface plasmon; illustrated in Fig. 5.10. The optical excitation of plasmon is the most efficient process by which light interacts with matter [22] and confers the ability to concentrate and manipulate light at dimensions below the diffraction limit onto metal nanostructures. Plasmonics is a rapidly growing field of research and has recently been identified [23] as a potential pervasive technology, offering opportunities to achieve unprecedented optical functionalities.

Fig. 5.10. Schematic representation of the polarization of metal nanoparticles by excitation of the surface plasmon.

Although light of all frequencies can induce electron oscillations in metals, the interaction is particularly efficient at the characteristic plasmon resonance frequency. Plasmon frequency depends on a variety of factors, including the identity of the metal, the dielectric constant of the surrounding medium, and particle size [24–26]. The size dependence of plasmon frequency is illustrated in Fig. 5.11 for a series of gold particles. The surface plasmons of neighboring nanoparticles in close proximity can couple, and for this reason assemblies of metal nanoparticles have plasmon frequencies which differ from those of the isolated constituent particles. The coupled plasmon resonance

5 Nanoparticles: Properties and applications

|

111

60 nm 8 nm

Absorbance (a.u.)

4.6 nm

300

120 nm

2.5 nm

400

500 600 Wavelength (nm)

700

Fig. 5.11. Plasmon extinction spectra of aqueous suspensions of gold nanoparticles of varying size. Adapted with permission from [26].

depends on the distance of separation in a known way, and can thus be exploited to probe nanosystems through a novel application known as the plasmon ruler [27–29]. Such a ruler allows determination of separation distances in media (solution) which are not accessible with electron or scanning probe microscopy. The surface plasmon is probably best known for its role in surface-enhanced spectroscopies. The surface oscillation of electric charges generated by plasmon excitation leads to significant enhancement in the local electromagnetic field [30], which in turn modifies the probability of optical transitions for molecules located within the affected region. For example, in the case of surface-enhanced Raman scattering, normally weak Raman signals can be enhanced by many orders of magnitude [31], leading to such exceptional sensitivities that single molecule detection becomes possible [32]. The intense local fields generated by excitation of nanoparticle plasmon can also modify the fluorescence properties of nearby molecules through a phenomenon known as metal-enhanced fluorescence [33]. The exact mechanism of fluorescence enhancement is not fully understood and probably involves more than one photophysical process. Firstly, the intense local field associated with the surface plasmon can increase the excitation of flourophores in proximity. Additional enhancement effects are also possible through interactions between the excited state fluorophore and the metal nanoparticle. In particular, if the resonance frequencies are matched, the excited state fluorophore can couple with a neighboring metal nanoparticle to

112 | Part I Functional materials: Synthesis and applications excite the plasmon. The plasmon can then relax radiatively to emit light. With this mechanism, the emission spectrum will be identical to that of the fluorophore, even though the nanoparticle is the emitting species. Typically, metal-enhanced fluorescence has a much shorter lifespan than that of the fluorophore alone, supporting the conclusion that the nanoparticle is the emitting species. It has been suggested that it is best to think of the fluorophore-metal complex as the emitting species [34]. Plasmon-enhanced fluorescence offers exiting possibilities in the development of ultra-sensitive methods, probes, and devices for biomedical detection. The majority of scientific publications in the field of nanoparticle plasmonics involve either silver or gold. This is, in part, because of the ease of particle synthesis, with gold and silver nanoparticles being readily accessible by the reduction of watersoluble salts. Furthermore, nanoparticles of these two metals exhibit resonance plasmon frequencies located within the visible region of the spectrum. Silver has the more intense plasmon of the two [35], but gold is biocompatible and more stable in biological media. Despite the prevalence of silver and gold, current research is looking beyond these two metals for next generation technologies [36, 37]. Metal nanoparticles are currently being employed in applications other than those based on their plasmonic properties. For example, silver nanoparticles have important antibacterial and anti-fungal properties [38] and are being incorporated into a growing number of consumer products such as antiseptic sprays, antimicrobial bandages, and sports clothing. Metals are also known to catalyze a number of important organic reactions. Since surface area plays a predominant role in catalytic activity, the increased specific surface area associated with decreasing particle size can improve efficiency, and supported metal nanoparticles are widely used as solid catalysts [39, 40].

5.6 Metal oxide nanoparticles The wide range of properties and applications related to this diverse class of nanomaterials will be illustrated by consideration of three specific examples; titanium dioxide (TiO2 ), iron oxide (Fe3 O4 and Fe2 O3 ), and silicon dioxide, or silica (SiO2 ).

5.6.1 Titanium dioxide Titanium dioxide is highly reflective in the visible region of the spectrum, and for this reason has been long utilized as a pigment material [41]. Small particles of titanium dioxide are the brightest of all commercial white pigments, and the 2010 global production of nanoscale TiO2 is estimated at 5 000 metric tons [42]. The primary application is in paints and other opaque coatings, but TiO2 is also a frequent component of personal care products such as toothpaste and sunscreen.

5 Nanoparticles: Properties and applications

|

113

Transition metal oxides are semi-conductors, and thus share some of the electronic properties and applications described above for quantum dots. TiO2 is a wide band gap material and absorbs UV light in the 280–400 nm wavelength range [41]. Upon absorption, an electron is elevated from the valence band to the conduction band to generate an electron-hole pair, and TiO2 nanoparticles have been employed to enhance photoconversion in solar cells [43]. The electron-hole pair can also participate in charge transfer to chemical species adsorbed at the semiconductor surface, leading to chemical reactions through a process known as photocatalysis [44]. The excited state conduction band electron can serve to reduce an accepter molecule, whereas the valence level hole can accept an electron from a neighboring donor species. These photoinduced electron transfer processes are governed by the position of the semiconductor bands with respect to the redox potentials of the surface adsorbed molecules. In the case of TiO2 , the position of the valence band leads to a particularly high oxidation potential for the photoinduced holes. The holes can thus readily react with adsorbed water to produce hydroxyl radicals, which in turn are strongly oxidizing and able to convert organic compounds to CO2 and water. For this reason, TiO2 nanoparticles can be used as antibacterial [45] and water treatment agents [46], and have even been demonstrated to split water into hydrogen and oxygen [47]. Since the photocatalytic process requires migration of electrons and holes to the TiO2 surface, nanoparticles, with their characteristically large surface to volume ratio, show greater catalytic activity than the corresponding bulk material.

5.6.2 Iron oxide The second important category of metal oxide nanoparticles is that composed of the iron oxides. These functional materials are primarily of interest for their magnetic properties. A variety of oxides are known, differing in the oxidation state of iron and the crystal structure. The two main forms are magnetite (Fe3 O4 ), which contains both Fe (II) and Fe (III) in a 1 : 2 ratio, and the gamma phase of iron (III) oxide (𝛾-Fe2 O3 ), known as maghemite. As is the case for other transition metals, the magnetic moments of iron atoms and ions arise from the presence of unpaired electrons in the d orbitals. Within small regions of a bulk material, the individual neighboring atomic magnetic moments are generally aligned with one another to form a magnetic domain. The magnetization within each domain points in a uniform direction, but the magnetization of different domains may point in different directions, as illustrated in Fig. 5.12. The domain structure can be modified by the application of an external magnetic field to increase magnetization in a given direction. If the domains do not return to a random state after removal of the external field, a permanent magnet will result.

114 | Part I Functional materials: Synthesis and applications

Fig. 5.12. Schematic representation of a multidomain magnetic material.

The novel properties of magnetic nanoparticles appear when particle size is reduced to below domain size, which is of the order of 100 nm [48]. Small iron oxide nanoparticles are monodomain and exhibit what is known as superparamagnetic behavior. The direction of magnetization within a domain can randomly flip at a characteristic frequency determined by what is known as the Néel relaxation time. At a given temperature, the Néel relaxation time decreases with decreasing particle size and, in the case of small nanoparticles, is generally much shorter than typical observation times. Under these conditions, known as the superparamagnetic state, the magnetization of the nanoparticle averages to zero. However, when an external magnetic field is applied to an assembly of superparamagnetic nanoparticles, their magnetic moments tend to align along the applied field, leading to a net magnetization. Superparamagnetic nanoparticles can be dispersed in a carrier liquid to form what is known as a ferrofluid [49]. When placed in a magnetic field gradient, a ferrofluid moves to the region of highest flux. This means that ferrofluids can be precisely positioned and shaped by an external magnetic field. Ferrofluids are found in a large number of commercial devices, including audio speakers, vacuum seals, switches, and sensors [50]. We have recently demonstrated the fabrication of a ferrofluidic deformable liquid mirror, based on the system shown in Fig. 5.13 [51, 52]. Clearly, the role of magnetic nanoparticles as functional materials is not restricted to their use in ferrofluids. In particular, iron oxide nanoparticles are being increasingly employed in medical applications, including, for example, their use as contrast agents in magnetic resonance imaging, [53] as discussed elsewhere in this volume.

Fig. 5.13. Magnetically deformable mirror prepared by coating an appropriate ferrofluid with a reflective monolayer of silver nanoparticles. The central deformation is generated by the magnetic field from an underlying permanent magnet.

5 Nanoparticles: Properties and applications

|

115

5.6.3 Silica Nanoparticles composed of silica have a large number of applications in materials science. Silica is a very versatile substance offering a number of attractive properties, including, for example, chemical and thermal stability and biocompatibility [54]. Furthermore, the surface of silica can be readily functionalized with organic or biologically active molecules, a key requirement for biological imaging or sensing applications. One important example of the use of silica is in the preparation of luminescent nanoparticles by the encapsulation of fluorophores [55, 56]. Luminescent nanoparticles are used in many fields of application, the most prevalent being bioimaging. Fluorescence organic molecules typically exhibit enhanced properties when dispersed in the silica matrix, which protects them against both collisional quenching and photodegradation [57]. In addition, for applications in living systems, nanoparticles have been reported to offer significant advantages over organic molecules with respect to resistance to metabolic disintegration, low toxicity, and adequate bioavailability [58]. Fluorescent doped silica nanoparticles have been demonstrated as promising probes for the detection of specific ionic species [57], intercellular sensing [55], cellular imaging [56], and even the diagnosis of cancer [59]. One very important class of silica nanoparticles are the mesoporous materials [60, 61]. The high surface area of these materials makes them inherently interesting for catalysis. They are also receiving a great deal of attention as drug delivery platforms [62], and can be designed to offer sophisticated functionalities such as stimulus-responsive gated release [63]. Since the preceding chapter of this volume is dedicated to mesoporous materials, no further details will be provided here.

5.7 Polymeric nanoparticles Polymer microbeads and nanoparticles are used on a very large scale as latex paints. As aqueous suspensions, latexes are environmentally friendly, practically odorless and faster drying than their oil-based counterparts. Latexes are typically obtained by the well-known process of emulsion polymerization [64]. More recently, polymer nanoparticles have been attracting additional interest as host matrices for the preparation of novel hybrid materials. Polymers offer great versatility as host matrices. The large inventory of existing polymers means that a number of properties, such as polarity, elasticity and refractive index, can all be adjusted to meet specific requirements. The formation of polymer nanoparticles is achieved by the dispersion of the monomer into small droplets prior to the initiation of the polymerization reaction. Many considerations influence the exact choice of the method of monomer dispersion, but for the preparation of nanosized particles, miniemulsion polymerization is often the method of choice [65]. The process of miniemulsion polymerization is illustrated in

116 | Part I Functional materials: Synthesis and applications Fig. 5.14. In contrast to emulsion polymerization, in which the monomer must diffuse through the continuous phase to feed particle growth, in the miniemulsion process, all of the monomer is pre-dispersed as nanodroplets prior to the initiation step. Thus, in miniemulsion polymerizations, each droplet acts as an independent preformed nanoreactor. Miniemulsion polymerization is therefore particularly well-suited to the preparation of doped nanoparticles, since dopants can be dissolved in the monomer at the desired concentration before dispersion [65]. For example, we have employed this approach for the preparation of polystyrene nanoparticles doped with a luminescent lanthanide complex, as illustrated in Fig. 5.15 [66]. As reviewed elsewhere [65], this approach can be extended beyond molecular dopants, and miniemulsion polymerization is finding increasing interest for the encapsulation of solid nanomaterials. H2O Ultrasound

Monomer

Emulsion

Surfactant

Miniemulsion

Polymerization

Polymer nanoparticles Fig. 5.14. Schematic depiction of miniemulsion polymerization.

x60 000

0.2 μm

Fig. 5.15. Transmission electron microscope image of polystyrene nanoparticles prepared by miniemulsion polymerization. Reprinted with permission from [66].

5 Nanoparticles: Properties and applications

|

117

5.8 Advanced architectures and hybrid systems The examples provided above illustrate the immense diversity of functional nanoparticles and their role in developing technologies. For the most part, the present discussion has, however, been limited to monolithic particles. In reality, nanoparticle architecture is becoming increasingly sophisticated, and a cursory review of current scientific literature is sufficient to ascertain the growing trend toward multicomponent, multi-function materials, such as core-shell and doped nanoparticles. For example, metal cores have been coated with fluorescent dye-functionalized silica shells to combine plasmonic and luminescence properties in nanoparticles designed for biomedical detection [67]. Core-shell architectures have also been employed to facilitate energy transfer and upconversion luminescence in lathanide-doped materials [68]. These are only two examples of the exciting future of nanoparticles as functional materials. The possibilities are essentially endless. Promising new materials are being reported daily and any attempt to adequately review the scientific literature in this field becomes outdated before even going to press.

References [1]

SciFinder, American Chemical Society, Chemical Abstracts Service, Columbus, Ohio 1999. https://scifinder.cas.org/scifinder consulted June 26, 2013. [2] See for example titles such as ACS Nano, Nano Letters, Small, Nature Nanotechnology and the Journal of Nanoparticle Research. [3] Nanotechnology and Enabling Technologies, Foresight Briefing #2, The Foresight Institute, 1989. http://www.foresight.org/Updates/Briefing2.html, consulted January 7, 2014. [4] Lamer V, Dinegar R. Theory, Production and Mechanism of Formation of Monodispersed Hydrosols. J Amer Chem Soc 1950; 72: 4847–4854. [5] Sugimoto T. Nucleation and Growth of Monodispersed Particles: Mechanisms, Encyclopedia of Surface and Colloid Science, Taylor & Francis, 2006: 1, 4257–4270. [6] Destrée C, Nagy JB. Mechanism of formation of inorganic and organic nanoparticles from microemulsions. Adv Colloid Interface Sci 2006; 123–126: 353–367. [7] Lemyre J-L, Ritcey AM. Synthesis of Lanthanide Fluoride Nanoparticles of Varying Shape and Size. Chem Mater 2005; 17: 3040–3043. [8] Lucier BEG, Johnston KE, Arnold DC, Lemyre J-L, Beaupreì A, Blanchette M, Ritcey AM, Schurko RW. Comprehensive Solid-State Characterization of Rare Earth Fluoride Nanoparticles. J Phys Chem C 2014; 118: 1213–1228. [9] Ohshima H. The DLVO Theory of Colloidal Stability. In Electrical Phenomena at Interfaces and Biointerfaces: Fundamentals and Applications in Nano-Bio- and Environmental Sciences, Chapter 3. First Edition. Ohshima. H, Ed., John Wiley & Sons, Inc. 2012. [10] Israelachvili J. Intermolecular & Surface Forces, Second Edition, pp 293-298, Academic Press, San Diego, 1992. [11] Norris, DJ. Electronic Structure in Semiconductor Nanocrystals: Optical Experiment. In Semiconductor and Metal Nanocrystals, Synthesis and Electronic and Optical Properties; Klimov VI, Ed.; Chapter 2. Marcel Dekker, Inc., New York, 2004.

118 | Part I Functional materials: Synthesis and applications [12] Brus LE. A simple model for the ionization potential, electron affinity, and aqueous redox potentials of small semiconductor crystallites. J Chem Phys 1983;79: 5566–5571. [13] Semonin OE, Luther JM and Beard MC. Quantum dots for next generation Photovoltaics. Mater Today 2012; 15(11): 508–515. [14] de Mello Donega C, Hickey SG, Wuister SF, Vanmaekelbergh D, Meijerink A. J Phys Chem B 2003; 107: 489–496. [15] Shirasaki Y, Supran GJ, Bawendi MG, Bulović V. Emergence of colloidal quantum-dot lightemitting technologies. Nat Photonics 2013; 7: 13–23. [16] Kershaw SV, Susha AS, Rogach AL. Narrow bandgap colloidal metal chalcogenide quantum dots: synthetic methods, heterostructures, assemblies, electronic and infrared optical properties. Chem Soc Rev 2013; 42: 3033–3087. [17] Gaponik N, Hickey SG, Dorfs D, Rogach AL, Eychmüller A. Progress in the Light Emission of Colloidal Semiconductor Nanocrystals. Small 2010; 6(13): 1364–1378. [18] Samokhvalov P, Artemyev M, Nabiev I. Basic Principles and Current Trends in Colloidal Synthesis of Highly Luminescent Semiconductor Nanocrystals. Chem Eur J 2013; 19: 1534–1546. [19] Talapin DV, Lee J-S, Kovalenko MV, Shevchenko EV. Prospects of Colloidal Nanocrystals for Electronic and Optoelectronic Applications. Chem Rev 2010; 110: 389–458. [20] Nozik AJ, Beard MC, Luther JM, Law M, Ellingson RJ, Johnson JC. Semiconductor Quantum Dots and Quantum Dot Arrays and Applications of Multiple Exciton Generation to Third-Generation Photovoltaic Solar Cells. Chem Rev 2010: 110: 6873–6890. [21] Knowles KE, Frederick MT, Tice DB, Morris-Cohen AJ, Weiss EA. Colloidal Quantum Dots: Think Outside the (Particle-in-a-) Box. J Phys Chem Lett 2012; 3: 18–26. [22] Malynych S, Chumanov G. Light-Induced Coherent Interactions between Silver Nanoparticles in Two-Dimensional Arrays. J Amer Chem Soc 2003; 125; 2896–2898. [23] Schuller JA, Barnard ES, Cai W, Jun YC, White JS, Brongersma ML. Plasmonics for extreme light concentration and manipulation. Nat Mater 2010; 9: 193–204. [24] Henry A-I, Bingham JM, Ringe E, Marks LD, Schatz GC, Van Duyne RP. Correlated Structure and Optical Property Studies of Plasmonic Nanoparticles. J Phys Chem C 2011; 115: 9291–9305 [25] Schatz GC, Kelly KL, Coronado E, Zhao LL. The Optical Properties of Metal Nanoparticles: The Influence of Size, Shape, and Dielectric Environment. J Phys Chem B 2003; 107: 668–677. [26] Hodak JH, Henglein A, Hartland GV. Photophysics of Nanometer Sized Metal Particles: ElectronPhonon Coupling and Coherent Excitation of Breathing Vibrational Modes. J Phys Chem B 2000; 104: 9954–9965. [27] Reinhard BM, Siu M, Agarwal H, Alivisatos P, Liphardt J. Calibration of Dynamic Molecular Rulers Based on Plasmon Coupling between Gold Nanoparticles. Nano Lett 2005; 5: 2246–2252. [28] Jain PK, Huang W, El-Sayed MA. On the universal scaling behavior of the distance decay of plasmon coupling in metal nanoparticles pairs: A plasmon ruler equation. Nano Lett 2007; 7: 2080–2088. [29] Yang L, Wang H, Yan B, Reinhard BM. Calibration of Silver Plasmon Rulers in the 1–25 nm Separation Range: Experimental Indications of Distinct Plasmon Coupling Regimes. J Phys Chem C 2010; 114: 4901–4908. [30] Stockman, MI. Nanoplasmonics: The physics behind the applications. Phys Today 2011; 64 (2): 39–44. [31] Rycenga M, Camargo PHC, Li W, Moran CH, XiaY. Understanding the SERS Effects of Single Silver Nanoparticles and Their Dimers, One at a Time. J Phys Chem Lett 2010; 1: 696–703. [32] Nie S, Emory SR. Probing Single Molecules and Single Nanoparticles by Surface-Enhanced Raman Scattering. Science 1997; 275: 1102–1106.

5 Nanoparticles: Properties and applications

|

119

[33] Lakowicz JR, Ray K, Chowdhury M, Szmacinski H, FuY, Zhang J, Nowaczyk K. Plasmoncontrolled fluorescence: a new paradigm in fluorescence spectroscopy. Analyst, 2008; 133: 1308–1346. [34] Lakowicz JR. Principles of Radiative Decay Engineering: Metal-Enhanced Fluorescence. In Fluorescence Spectroscopy, Third Edition, Chapter 25. Springer Science 2006. [35] Mahmoud M, El-Sayed MA. Different Plasmon Sensing Behavior of Silver and Gold Nanorods. J Phys Chem Lett 2013; 4: 1541–1545. [36] DeSantis CJ, Weiner RG, Radmilovic A, Bower MM, Skrabalak SE. Seeding Bimetallic Nanostructures as a New Class of Plasmonic Colloids. J Phys Chem Lett 2013; 4: 3072–3082. [37] Naik GV, Shalaev VM, Boltasseva A. Alternative Plasmonic Materials: Beyond Gold and Silver. Adv Mater 2013; 25(24): 3264–3294. [38] Liu L, Yang J, Xie J, Luo Z, Jiang J, Yang YY, Liu S. The potent antimicrobial properties of cell penetrating peptide-conjugated silver nanoparticles with excellent selectivity for Gram-positive bacteria over erythrocytes. Nanoscale 2013; 5: 3834–3840. [39] Jin R. The impacts of nanotechnology on catalysis by precious metal nanoparticles. Nanotechnol Rev 2012; 1: 31–56. [40] Reetz MT. Size-selective Synthesis of Nanostructured Metal and Metal Oxide Colloids and Their Use as Catalysts. In Nanoparticles and Catalysis, Astruc D, Ed., Chapter 8. Wiley-VCH Weinheim, 2008. [41] Lan Y, Lu Y, Ren Z. Mini review on photocatalysis of titanium dioxide nanoparticles and their solar applications. Nano Energy 2013; 2:1031–1045. [42] Weir A, Westerhoff P, Fabricius L, von Goetz N. Titanium Dioxide Nanoparticles in Food and Personal Care Products. Environ Sci Technol 2012; 46(4): 2242–2250. [43] Genovese MP, Lightcap IV, Kamat PV. Sun-Believable Solar Paint. A Transformative One-Step Approach for Designing Nanocrystalline Solar Cells. ACS Nano 2012; 6(1): 865–872. [44] Linsebigler AL, Lu G, Yates JT Jr. Photocatalysis on TiOn Surfaces: Principles, Mechanisms, and Selected Results. Chem Rev 1995; 95: 735–758. [45] Wolfrum EJ, Huang J, Blake DM, Maness P-C, Huang Z, Nefiest J, Jacoby WA. Photocatalytic Oxidation of Bacteria, Bacterial and Fungal Spores, and Model Biofilm Components to Carbon Dioxide on Titanium Dioxide-Coated Surfaces. Environ Sci Technol 2002; 36: 3412–3419. [46] Reisch MS. Workhorse pigment titanium dioxide is also the stuff that dreams are made of. Chem Eng News 2003; 81(13): 13. [47] Fujishima A, Honda K. Electrochemical Photolysis of Water at a Semiconductor Electrode. Nature 1972; 238: 37–38. [48] Sarkar D, Mandal M, Mandal K. Domain controlled magnetic and electric properties of variable sized magnetite nanohollow spheres. J Appl Phys 2012; 112: 064318. [49] Rosensweig, RE. Ferrohydrodynamics, Cambridge University Press: London, 1985. [50] Raj K, Moskowitz R. Commercial Applications of Ferrofluids. J Magn Magn Mater 1990; 85: 233–245. [51] Déry J-P, Borra EF, Ritcey AM. An ethylene glycol based ferrofluid for the fabrication of magnetically deformable liquid mirrors. Chem Mater 2008; 20: 6420–6426. [52] Ritcey AM, Borra E. Magnetically Deformable Liquid Mirrors from Surface Films of Silver Nanoparticles. Chem Phys Chem 2010; 11: 981–986. [53] Na HB, Song IC, Hyeon T. Inorganic Nanoparticles for MRI Contrast Agents. Adv Mater 2009; 21: 2133–2148. [54] Asefa T, Tao Z. Biocompatibility of Mesoporous Silica Nanoparticles. Chem Res Toxicol 2012; 25: 2265–2284. [55] Ruedas-Rama MJ, Walters JD, Orte A, Hall EAH. Fluorescent nanoparticles for intracellular sensing: A review. Anal Chim Acta 2012; 751: 1–23.

120 | Part I Functional materials: Synthesis and applications [56] Wang K, He X, Yang X, Shi H. Functionalized Silica Nanoparticles: A Platform for Fluorescence Imaging at the Cell and Small Animal Levels. Acc Chem Res 2013; 46(7): 1367–1376. [57] Montalti M, Rampazzo E, Zaccheroni N, Prodi L. Luminescent chemosensors based on silica nanoparticles for the detection of ionic species. New J Chem 2013; 37: 28–34. [58] Sharma P, Brown S, Walter G, Santra S, Moudgil B. Nanoparticles for bioimaging. Adv. Colloid Interface Sci. 2006; 123-126: 471–485. [59] Arap W, Pasqualini R, Montalti M, Petrizza L, Prodi L, Rampazzo E, Zaccheroni N, Marchiò S. Luminescent silica nanoparticles for cancer diagnosis. Curr Med Chem 2013; 20(17): 2195–211. [60] Chen Y, Chen H-R, Shi J-L. Construction of Homogenous/Heterogeneous Hollow Mesoporous Silica Nanostructures by Silica-Etching Chemistry: Principles, Synthesis, and Applications. Acc Chem Res 2014; 47(1): 125–137. [61] Wu S-H, Mou C-Y, Lin H-P. Synthesis of mesoporous silica nanoparticles. Chem Soc Rev 2013; 42(9): 3862–3875. [62] Argyo C, Weiss V, Braeuchle C, Bein T. Multifunctional Mesoporous Silica Nanoparticles as a Universal Platform for Drug Delivery. Chem Mater 2014; 26(1): 435–451. [63] Nadrah P, Planins˘ek O, Gabersč˘ek M. Stimulus-responsive mesoporous silica particles. J Mater Sci 2014: 49: 481–495. [64] Landfester K. Synthesis of colloidal particles in miniemulsions. Annu Rev Mater Res 2006; 36: 231–79. [65] Landfester K, Weiss CK. Encapsulation by Miniemulsion Polymerization. Adv Polym Sci 2010; 229: 1–49. [66] Desbiens J, Bergeron B, Patry M, Ritcey AM. Polystyrene nanoparticles doped with a luminescent europium complex. J Colloid Interface Sci 2012; 376 (1): 12–19. [67] Lessard-Viger M, Rioux M, Rainville L, Boudreau D. Nano Lett 2009; 9(8): 3066–3071. [68] Wang F, Deng R, Wang J, Wang Q, Han Y, Zhu H, Chen X, Liu X. Tuning upconversion through energy migration in core-shell nanoparticles. Nat Mater 2011; 10: 968–973.

N. Allard and M. Leclerc

6 Conjugated polymers for organic electronics Conjugated polymers have received a great deal of attention from both academic and industrial laboratories, since they combine the best features of metals or inorganic semiconducting materials (excellent electrical and optical properties), with those of synthetic polymers (mechanical flexibility, simple processing, low cost). This synergy makes these functional materials useful in existing optoelectronic devices, and creates completely new technological opportunities. For instance, polymeric semiconductors are considered to be one of the most promising materials to lower the cost of solar energy. Since they can be dissolved in common solvents, processing techniques such as inkjet printing, spin coating and large scale roll-to-roll coating techniques, widely used for thermoplastics, are now available for the low-cost production of printed photovoltaic devices. All of these properties will help to establish the socalled plastic electronics era, which includes light-emitting diodes and field-effect transistors.

6.1 Introduction There is hardly an aspect of our lives which is not touched by polymeric materials. In the fields of optics and electronics, synthetic polymers are traditionally used in applications such as packaging, electrical insulators, and photoresists. However, new opportunities have emerged with the development of electroactive and photoactive conjugated polymers. This chapter will describe several examples which demonstrate the tremendous innovation potential of these novel functional materials. As with many other important discoveries, the development of the field of conjugated polymers was somehow accidental. A student of Professor Hideki Shirakawa at the Tokyo Institute of Technology was working on the preparation of polyacetylene (Fig. 6.1) by the Ziegler–Natta polymerization method. By mistake, this student prepared a multimolar concentration of the catalyst, instead of the usual millimolar concentration, and obtained a thin polyacetylene film which looked like a metallic foil instead of the usual dark, powdery material. During a visit to Japan, Professor A.G. MacDiarmid (Department of Chemistry, University of Pennsylvania) met Professor Shirakawa and was fascinated by this shiny polymeric film. He invited him to come to Philadelphia to investigate this new form of polyacetylene in more detail. In 1977, in collaboration with Professor A.J. Heeger, a colleague from the Physics Department, this team reported that upon partial oxidation with iodine or bromine (so-called doping reaction), the conductivity of the intrinsically semiconducting polyacetylene increased more than a millionfold (up to 100 S/cm) [1]. These days, electrical conductivities of up to 105 S/cm are now obtained with some highly crystalline forms of

122 | Part I Functional materials: Synthesis and applications polyacetylene. The delocalized electronic structure (alternation of single and double bonds) of polyacetylene is responsible for the good intrachain and interchain mobility of the charge carriers (radical cations or their negative analogs) created upon doping. This delocalized (conjugated) structure is also responsible for a strong absorption in the UV-visible range. This group (Shirakawa, MacDiarmid, Heeger) eventually received the 2000 Nobel Prize in Chemistry for this discovery [2–4]. Unfortunately, this first conducting polymer is still very difficult to process and unstable in the presence of oxygen. Rapidly, many scientists investigated more stable conjugated polymers. For instance, in the early 1980s, a lot of studies were devoted to electropolymerized polythiophene [5], polypyrrole [5, 6], and polyaniline [7] (Fig. 6.1). This method has the advantage of being capable of preparing thin films of these infusible and insoluble rigid-rod conjugated polymers in one step. Depending upon the electrochemical potential, it is possible to switch between the undoped semiconducting state and the oxidized conducting state. Interestingly, these redox processes are also accompanied by color changes (electrochromism). However, the ultimate goal remains the development of polymeric materials which combine the electrical and optical features of metals or semiconductors with the processing advantages and mechanical properties of traditional polymers.

S n Polyacetylene

n Polythiophene

N H

NH n

Polypyrrole

n Polyaniline

Fig. 6.1. First generation of conjugated polymers.

6.2 Processable conjugated polymers The attempts to fulfill these requirements led to the development of a second generation of processable conjugated polymers. A first example of processable polyacetylene was demonstrated by Edwards and Feast in 1980 [8]. Using ring-opening metathesis polymerization (ROMP), a soluble polymeric non-conjugated precursor could be converted into a polyacetylene thin film and a volatile by-product upon heating (Fig. 6.2). This two-step procedure was not perfect, but it was the first example of a processable and potentially conjugated polymer. A similar approach was developed for the preparation of poly(para-phenylene) (PPP) [9]. A breakthrough occurred in 1985–86 with the syntheses of highly conjugated and processable poly(3-alkylthiophene)s [10]. The fact that these five-membered rings can exhibit an anti co-planar conformation (Fig. 6.3) reduces the steric hindrance developed by the presence of the solubilizing side-chains. Thus, the coexistence of both planarity (important to keep the delocalized and conjugated structure) and processability

6 Conjugated polymers for organic electronics

F3C

CF3

F3C

|

123

CF3

WCI6

F3C

Δ

CF3

+ n

n

Fig. 6.2. Synthesis of polyacetylene from a processable precursor.

was demonstrated for the first time. For instance, it is interesting to note that processable alkyl-substituted polyacetylenes, polyanilines, and poly(para-phenylene)s are non-planar and have poor electrical properties. Some researchers describe these types of substituted conjugated macromolecules as hairy-rod polymers, this configuration facilitating the dispersion and interactions with the solvent. Following these first studies on poly(3-alkylthiophene)s, it became quite clear that the synthesis of well-defined head-to-tail coupled poly(3-alkylthiophene)s, which should yield to the lowest steric hindrance from the side chains and possibly a more efficient threedimensional packing, would lead to a significant improvement in the performance of these polymeric materials [11]. Therefore, in attempts to bring more reliable synthetic procedures to the field of electronic materials, a variety of synthetic tools have been implemented, allowing significant advances in this research field. Among other advantages, these investigations led to the first preparations of well-defined regioregular poly(3-alkylthiophene)s by McCullough and Rieke in 1992 (Fig. 6.3, [12, 13]). Further, these relatively complicated polymerization procedures have been optimized and simplified, leading to the Grignard metathesis method (GRIM, [14]). For instance, this new method eliminates the need for highly reactive metals and can be performed at low temperatures. Among all poly(3-alkylthiophene)s investigated, regioregular poly(3hexylthiophene), or (P3HT), has become the polymer of choice, exhibiting the best electrical and optical properties in addition to adequate processability. The synthesis of air-stable, semi-transparent, highly conducting poly(3,4-ethylenedioxythiophene), or (PEDOT), is another example of rational design of a conjugated polymer with optimized structural and physical properties [15]. Both P3HT and PEDOT are currently the most utilized conjugated polymers and are commercially available from various sources. In the meantime, the focus shifted from the synthesis of highly conductive polymers to the design of stable semiconducting polymers through collaborations between physicists and engineers. This new driving force was based on the aim to initiate the so-called plastic electronics era, where micro-electronic devices could be printed on different substrates using practical processing methods. For this purpose, the electronic properties of the existing polymers have to be chemically tuned to obtain low bandgap semiconducting materials. This bandgap tuning can be achieved in various ways. For instance, the rigidification of the conjugated backbone, the introduction of electron-withdrawing or electron-donating side groups, and the increase of the quinoid (versus aromatic) character of the main chain have been widely used

124 | Part I Functional materials: Synthesis and applications Oxidation or electropolymerization

R

R S S

n

S R

Br

S

R

Kumada (Mc Cullough)

R

S S

[Ni]cat.

n

R R Br

Br

S

R

Negishi (Rieke)

S S

[Ni]cat.

n

R

Br

S

R

Kumada (GRIM)

R Br

S S

[Ni]cat. R

n

Fig. 6.3. Synthesis of processable poly(3-alkylthiophene)s.

in the past few years to modulate the bandgap of such conjugated polymers [16]. However, one of the most efficient approaches involves the utilization of electronrich or electron-poor units leading to alternating push-pull architecture. As shown in Fig. 6.4, when these two different moieties are combined in an alternating copolymer, hybridization occurs between the molecular orbitals of the electron-rich unit and of the electron-poor unit, leading to a significant modulation of the copolymer bandgap. Usually, the energy level of the highest occupied molecular orbital (HOMO) of the resulting copolymer is determined by the electron-rich unit, while the energy level of the lowest unoccupied molecular orbital (LUMO) is mainly influenced by the electron-poor unit. LUMO LUMO

Electron-rich moiety

Eg

Electron-poor moiety

HOMO HOMO D

D-A

A

Fig. 6.4. Hybridization of the HOMO-LUMO energy levels in donor-acceptor (push-pull) copolymers.

6 Conjugated polymers for organic electronics

(RO)2B

B(OR)2 + Br

N R

S N N

S N N S

S

S

Suzuki Br

[Pd]cat.

R

R

S

S

SnR3

O

N

Br

S

O

+

Si R R

Br

n

R

R O

Stille

Br R

S

R

S S

N

+ Br

S

R

N

O

S

DHAP [Pd]cat.

n

R O

R O

O

Si R

O

N

S S

[Pd]cat.

R

S

N

R R3Sn

| 125

S

S

n

R

Fig. 6.5. Typical polymerization reactions for conjugated alternating copolymers.

Over the years, a third generation of semiconducting copolymers have therefore been synthesized to satisfy the needs of various electronic applications [17]. Unfortunately, there are relatively few synthetic methods which allow efficient preparation of such alternating copolymers; most of these copolymers being obtained via Suzuki [18] or Stille [19] cross-coupling polymerization (Fig. 6.5). These state-of-the-art methods generally involve numerous synthetic steps and organometallic reagents which give rise to metal waste and various by-products. Therefore cheaper, environmentally friendly and more efficient synthetic procedures would clearly be a great asset for the sustainable preparation of affordable conjugated polymers. In order to solve these problems, the utilization of the latest synthetic developments in organic chemistry, termed direct (hetero)arylation, is very appealing. These new reactions allow for the formation of carbon-carbon bonds between (hetero)arenes and aryl halides. Hence, they do not require organometallic intermediates, thereby significantly reducing both synthetic steps and cost (see Fig. 6.5). Moreover, this new direct (hetero)arylation polymerization (DHAP) method, which only produces acid as a by-product, has recently attracted a lot of interest [20]. Although these reactions are not easily controlled for compounds having more than one type of C-H bond, they can sometimes lead to the synthesis of branched or crosslinked polymers. A fine-tuning of the reaction conditions and careful choice of the monomers may suppress these unwelcome side reactions. All of these step-growth polymerization methods (Suzuki, Stille, DHAP) can lead to number-average molecular weights of up to 100 kDa/mol in the best case, and must exhibit polydispersity indexes of about 2 according to the Carother equation. Finally, it is important to note that most recently-developed monomers (e.g. carbazoles, dithienosiloles, thienopyrrolodiones, etc.) bear flexible side-chains which are far from the conjugated backbone, reducing the steric hindrance created within these hairy-rod chemical structures.

126 | Part I Functional materials: Synthesis and applications These chemical tools allowed chemists and engineers to synthesize a large number of semiconducting polymers for the manufacture of diverse electronic devices such as light-emitting diodes, field-effect transistors, solar cells, and a continuously growing list of other applications. For instance, water-soluble luminescent conjugated polymers have allowed the specific, rapid and ultra-sensitive detection of unlabeled DNA, RNA, or proteins. This relatively young field of chemical and biological sensors based on semiconducting conjugated polymers is expanding rapidly but will not be further discussed in this chapter. For those interested in these biomedical applications, the authors recommend several reviews and the references therein [21, 22].

6.3 Applications in renewable energy 6.3.1 Organic solar cells Solar energy represents an attractive solution to fulfill our needs regarding green and renewable energy, while protecting the environment at the same time. For example, one should be aware that the sun provides in one hour what humanity consumes during one year. However, the large scale development of this technology is somewhat limited by the relatively high cost of producing and installing a solar panel versus its moderate efficiency. Many studies suggest that the cost of a solar panel must be decreased by a factor of 5 to 10 in order to compete with other already known energy sources such as hydro, nuclear, wind or fossil. Electroactive and photoactive conjugated polymers are considered to be one of the most promising ways to lower the price of solar energy. They can be solubilized in common solvents and processed using techniques such as inkjet printing, spin coating and large scale roll-to-roll coating techniques, which are widely used for thermoplastics and are available for the manufacture of various devices. In particular, bulk heterojunction (usually a blend of a polymeric p-type semiconductor with an n-type fullerene derivative) solar cells offer great opportunities, with power conversion efficiency (PCE) now approaching threshold values (a PCE of 10 % with a lifespan of 5–10 years) for large scale commercialization. Indeed, the most efficient polymeric materials can now reach power conversion efficiencies of over 9 % in a single-layer configuration [23]. These new functional materials should help to initiate new and competitive manufacturing of photovoltaic devices. Different types of device architecture have been utilized for the manufacture of organic photovoltaic devices. As mentioned above, the most common configuration is called the bulk heterojunction (BHJ) approach (Fig. 6.6). Glass is usually employed as common substrate, although plastics can be used to generate flexible devices. A layer of semi-transparent indium-tin oxide (ITO), which acts as the anode, is deposited on the substrate. After deposition of the anode, PEDOT, a semi-transparent and highly conductive polymer (described in Section 6.1), is spin-coated from a water-based so-

6 Conjugated polymers for organic electronics

| 127

lution on top of ITO for the purpose of enhancing the hole transport and improving contacts between the ITO electrode and the organic BHJ layer. Then the active layer, which is a bulk heterojunction of a donor (p-type semiconductor) and an acceptor (ntype semiconductor), usually from an organic solution containing both components, is deposited over the PEDOT layer. Such a structure provides large surface area interfaces which allow the generation of a charge separated state with efficient subsequent charge migration through 3-D percolation networks. The donor is usually a p-type semiconducting conjugated polymer, while the acceptor is generally a processable fullerene derivative such as [6,6]-phenyl-C61 -butyric acid methyl ester (PCBM) or its C71 analog; fullerenes being known to support stable and mobile negative charge carriers (see Chapter 3). Finally, a reflective metallic layer acting as the cathode is generally evaporated or printed on top of the assembly.

Metal Polymer: PCBM PEDOT ITO Glass or plastic

Fig. 6.6. Standard architecture for an organic solar cell.

As shown in Fig. 6.7, upon illumination, the semiconducting BHJ active layer absorbs some of the photons leading to the formation of excitons (Step 1). For instance, these excitons may come from the excitation of an electron from the HOMO of the polymer to its LUMO, creating bound electron-hole pairs (Step 2). These excitons can then migrate about 10–20 nm within the polymer phase (related to their lifespan and mobility) to reach the acceptor phase. The nanoscale BHJ morphology is particularly helpful for this purpose: at this boundary, the electron in the LUMO of the polymer is transferred to the LUMO of the acceptor (Step 3). After this photo-induced electron transfer, the positive charge carriers and the electrons, which are no longer held by Coulombic interactions, travel through their respective percolated phases to reach their respective electrodes (Step 4). It is important to mention that a similar mechanism is observed when light is absorbed by the fullerene derivative or any other n-type semiconductor; the only major difference being the fact that the electron transfer involves the HOMO of the semiconductors. Interestingly, despite the relative complexity of all these interfaces and physical mechanisms, internal quantum efficiency of close to 100 % has been reported, implying that for some optimized nanoscale morphology, essentially every absorbed photon results in a separated pair of charge carriers which are collected at the electrodes [24].

128 | Part I Functional materials: Synthesis and applications LUMO

1.

LUMO –

2.

LUMO

LUMO hv

AI

AI ITO

ITO

HOMO

HOMO

HOMO

HOMO + Polymer

4.

Polymer

PCBM

LUMO –

3.

LUMO LUMO

–

– –

ITO

PCBM

LUMO AI

AI ITO

HOMO

+

HOMO HOMO

HOMO + Polymer

+ PCBM

Polymer

PCBM

Fig. 6.7. General working mechanism of an organic solar cell.

6.3.2 Conjugated polymers for organic solar cells While the manufacture and working mechanism of an organic solar cell may seem relatively simple, finding the optimal conjugated polymer and obtaining the right morphology (which depends upon the blend ratio, solvent, deposition technique, temperature, etc.) is still a big challenge. For instance, as a first requirement, the semiconducting polymer must be stable in air. Consequently, its HOMO energy level should be under −5.2 eV, which is about the air-oxidation threshold. Also, as mentioned previously, there must be an electron transfer between the LUMO of the donor and the LUMO of the acceptor. To make this electron transfer possible, there must be an energetic difference of about 0.3 eV between these LUMO energy levels (the same is true when the electron transfer takes place through the HOMO energy levels). In addition, it is preferable for the HOMO energy level of the polymer to be as low as possible, since the open circuit voltage (Voc ) is mostly defined by the difference between the LUMO of the acceptor and the HOMO of the polymer. Moreover, since the polymer is the main light absorber when blended with fullerene derivatives, it is necessary that the polymer absorbs as much sunlight as possible. Thus, bandgaps between 1.2 and 1.9 eV are desired, since they correspond to the maximum intensity of the solar emission spectra and allow for the best compromise between optimized photocurrent and photovoltage (Fig. 6.8).

6 Conjugated polymers for organic electronics

Energy (eV)

OCH3

4.00 3.00 2.50 2.00 1.75 1.50

2x1018

1x1018

1100

1200

300

900

0

–6.1 eV

1000

Homo

3x1018

800

VOC max

Low VOC

700

Aluminum

4x10

500

Band gap 1.2–1.9 eV

–4.3 eV

1.00

Ideal Zone Low JSC

18

600

Acceptor > 0.3 eV

1.25

5x1018

400

PCBM

Lumo

ITO

Photon flux AM 1.5 (m–2 .s–1 .nm–1)

O Polymer Donor

| 129

Wavelength (nm) Fig. 6.8. Properties needed for conjugated polymers and photon flux spectra of the sun.

Current density (mA/cm2)

Once a good polymer/fullerene composite with the right morphology is found, a device is created and tested to determine the PCE. For this purpose, the electrodes are connected and light is beamed through the ITO electrode using a solar simulator (AM1.5G) imitating the sun irradiation at a tilted angle of 37 degrees (Pin is of 100 mW/cm2 ). At the same time, a potential sweeping is applied. After data processing, a typical J-V curve from which the fill factor (FF), the short-circuit current density (Jsc ), and the open-circuit voltage (Voc ) can be obtained, is used to calculate the PCE of the device (Fig. 6.9). Pmax represents the maximum power which can be produced by such a solar cell. This last parameter is important for the characterization of the fill factor, which describes the overall quality of the device.

VPmax

JPmax JSC

VOC

Pmax

PCE =

FF =

Pout Pin

=

FF x JSC x VOC

JPmax x VPmax Jsc x VOC

Potential (V)

Fig. 6.9. Typical J-V curve obtained for a photovoltaic device.

Pin

130 | Part I Functional materials: Synthesis and applications Hundreds of polymeric structures have been synthesized and investigated in the last few years, but only three main polymers (P3HT [25, 26], PCDTBT [25, 27], and PTB7 [23]) have demonstrated good and reproducible PCEs, with a lifespan of up to 7 years for PCDTBT (Fig. 6.10). However, large, stable, and low-cost modules will have to be developed and tested under “real” conditions in order to reach commercial applications. C4H9 H9C4 C2H5 O S N N

C6H13 S S

n

P3HT PCE = 6.5%

N H17C8 C8H17

O

F

O

C2H5

S

S

S

S

n

S

n

O PCDTBT PCE = 7.9%

H9C4 C2H5

PTB7 PCE = 9.2%

Fig. 6.10. Most widely studied conjugated polymers for organic solar cells.

6.4 Applications in micro-electronics 6.4.1 Field-effect transistors Field-effect transistors (FETs) are electronic switches widely utilized in electronic devices such as displays, computer logics, radio frequency identification tags (RFID), etc. The vast majority of these devices are currently made from monocrystalline, polycrystalline or amorphous silicon. In FETs, one of the main performance criteria is the charge mobility which can be simply related to the on/off switching speed of the device. Higher switching speeds permit applications in high performance electronics. These widely used silicon compounds have shown charge carrier mobility of up to 900 cm2 /V s for monocrystalline silicon [28] used in high performance electronic devices, and up to 1 cm2 /V s for amorphous silicon [28] utilized in standard electronic systems. Although silicon-based devices perform very well, they are relatively difficult to produce and cannot be used in flexible devices, whereas conjugated polymer-based devices can be solution-processed at low temperatures and easily coated onto flexible devices. Despite these great possibilities, polymer-based FETs will probably only compete with amorphous silicon. Indeed, amorphous or semi-crystalline semiconducting polymers do not exhibit the necessary long-range organization to lead to very high charge carrier mobility. In most cases, an FET is constituted of three electrodes: the gate, the source, and the drain. It is also composed of a dielectric layer and of a semiconducting layer. Figure 6.11 shows the different configurations for organic FETs (OFETs). The two most

6 Conjugated polymers for organic electronics

D Semiconductor Dielectric Gate Electrode

Semiconductor

S

D

S Dielectric Gate Electrode

(a)

(b)

Gate Dielectric

Gate Gate dielectric Semiconductor

S

D S

D

Semiconductor Substrate

Substrate (c)

| 131

(d)

Fig. 6.11. Schematic description of OFETs with (a) bottom-gate top-contact (BG/TC), (b) bottom-gate bottom-contact (BG/BC), (c) top-gate bottom-contact (TG/BC), and (d) top-gate top-contact (TG/TC) structures [29].

frequent configurations are bottom-gate top-contact (BG/TC; (a)) and bottom-gate bottom-contact (BG-BC; (b)) due to their relatively simple production. The general working mechanism of an OFET starts by applying a potential to the gate while a potential is constantly applied between the source and the drain (see Fig. 6.12). If a negative potential is applied to the gate (VGS ), positive charge carriers are formed at the interface between the semiconducting polymer and the dielectric layer. Then, due to the potential between the source electrode and the drain electrode, these positive charge carriers travel through the semiconducting layer, forming a ptype OFET. The drain-source current (IDS ) first follows ohmic behavior as a function of the applied drain-source potential (VDS ), then saturates at high voltages. If a positive potential is applied at the gate, negative charge carriers are formed at the polymer/dielectric interface creating n-type OFET. These OFETs are mainly characterized by three values: the charge carrier mobility (μ), the Ion /Ioff ratio, and the threshold voltage (VT ). The charge mobility refers to the speed at which the charges formed in the semiconductor layer travel between the source and the drain electrodes. These values can be determined from I-V curves, similar to those reported in Fig. 6.12. The higher the charge mobility, the faster switching between the on and off states of the OFET can occur. This charge mobility is numbered either in hole mobility or electron mobility depending on the potential applied at the gate. The Ion /Ioff ratio refers to the ratio between the current when the transistor is at the on-state and when it is at the off-state. While multiple switching happens in the device even if there is no potential applied at the gate, residual charges are still in the semiconducting layer, creating an electric current at the off state. Therefore, it is

132 | Part I Functional materials: Synthesis and applications important to have Ion /Ioff ratios as high as possible so that clear switching is always possible. The threshold voltage corresponds to the minimum potential which needs to be applied at the gate to obtain switching. It is necessary from a commercial point of view to have a low and stable threshold voltage. The electronic properties of the polymeric semiconductor and the nature of the dielectric layer play a major role in the values of the threshold voltage.

VGS 0V –20V –40V –60V –80V –100V

–60

IDS/uA

–40

–20

0

0

–20

–40 –60 VDS/V

–80

–100 Fig. 6.12. Typical I-V curve of a p-type OFET.

6.4.2 Conjugated polymers for field-effect transistors Conjugated polymers need to possess some critical properties in order to be good candidates for OFET applications. First, the semiconducting polymers must be stable over time and, as mentioned earlier, must possess a HOMO energy level lower than −5.2 eV to be air-stable. Stability is important, but the main factor which allows conjugated polymers to attain high mobility is their ability to form highly organized films. It has been shown that the closer the π-stacking is, the better the charge carrier mobility. The strategy for obtaining a polymer with good packing properties is to use planar conjugated units which will drive efficient π-stacking. To obtain processable materials, alkyl chains are added along the conjugated backbone. Obviously, adding alkyl chains is not necessarily good for the π-stacking efficiency, but the goal is to find a compromise between the alkyl chains needed for solubility and the π-stacking in order to obtain good mobility when solution-processed. In addition to polymer chain stacking, there is also the point of how these stacks of polymer chains are oriented

6 Conjugated polymers for organic electronics

|

133

relative to the substrate. It has also been shown that the packing of these polymers should be oriented perpendicular to the surface in order to permit good charge circulation through the source and drain electrodes [30]. See Fig. 6.13 for examples of the best polymers reported so far for OFETs. Obviously they exhibit the right combination of processability, π-stacking, and polymer chain orientations onto the substrates. C8H17

C10H21 N O H17C8

N

C8H17 O S

N

S

S

S

S

S

H13C6 O N

N

S

n

n H33C16 C16H33

S

N

C10H21

F

F

F

F

S

n

O C6H13

C8H17 PDTT-C8C10 μh= 10.5 cm2V–1s–1

CDT-BTZ μh= 5.5 cm2V–1s–1

N-CS2DPP-OD-TEG μe= 2.36 cm2V–1s–1

Fig. 6.13. Efficient conjugated polymers for p-type and n-type OFETs [31–33].

6.5 Applications in lighting 6.5.1 Light-emitting diodes Light-emitting diodes (LEDs) are mostly used as lighting sources. Today, they can be found in a large variety of electronic devices acting as electronic displays or light indicators. The first LED was demonstrated in 1936 by George Destriau in the laboratory of Marie Curie. When a potential is applied to ZnS:P powder placed between two electrodes, the emission of a low intensity red light can be observed. That discovery led to the development of a large variety of red, orange, yellow, and green inorganic light-emitting diodes. In 1962, General Electric commercialized the first inorganic LED which was based on phosphorous doped gallium arsenide (GaAs:P) [34]. Light-emission from an organic material was only reported for the first time in 1963 by Martin Pope and his collaborators at New York University. They demonstrated that an anthracene crystal placed between two silver electrodes could emit blue light [35]. Unfortunately, the lifespan of this type of device was far shorter than those obtained with inorganic-based devices. The real interest in organic material for LED came from the research of Ching W. Tang and Steven Van Slyke at Eastman Kodak Research Laboratories in 1987. They invented an organic LED (OLED) based on a thin evaporated layer of tris(8-hydroxyquinoline)aluminum (AlQ3 ), which led to high electroluminescence yield and interesting lifespans [36]. It was only in 1990 that the first polymeric LED (PLED) based on a π-conjugated polymer was demonstrated, when Sir Richard Friend and his collaborators at the University of Cambridge observed a

134 | Part I Functional materials: Synthesis and applications

Emitting layer

Electron transporting layer

Anode

Hole transporting layer

green light emission from a thin film of poly(para-phenylenevinylene), or PPV, placed between two electrodes [37]. Since that discovery, a large number of small molecules, oligomers, and polymers have been synthesized and used in light-emitting devices. It is important to recall here that OLEDS are generally prepared using small vacuumprocessed organic molecules, whereas polymers have the advantage of being solution processable, providing better opportunities for making large-area devices. Today, many electronic devices made of OLEDs such as flat screen televisions, mp3 players, digital cameras, and smartphones are commercially available and a number of companies, such as Universal Display, Cambridge Display Technology (CDT), and Philips produce lighting sources from conjugated small molecules, oligomers or polymers. A large number of different configurations exist for OLEDs and PLEDs resulting in different lifespans and efficiency. The simplest device configuration, and the first which was demonstrated, consists of a light-emitting material sandwiched between two electrodes. Electroluminescence is defined by the emission of light induced by charge recombination. Therefore, in order for LEDs to function, charges must be injected into the material. An electron is removed from the HOMO of the light-emitting material at the cathode, creating a positive charge carrier. Inversely, at the cathode an electron is inserted in the LUMO of the light-emitting material creating a negative charge carrier. The hole and the electron travel to their opposite electrodes under the influence of the applied electric field. They can also recombine to form an excited state commonly called exciton. That exciton can be either in the singlet state, where the spins of the electron/hole pairs are in the opposite direction, or in the triplet state where the spins of the electron/hole pairs are in the same direction. The singlet state is the only one which can lead to light emission in fluorescent conjugated polymers (Fig. 6.14). However, phosphorescent materials can help to increase the efficiency of such devices. It is interesting to note that light-emitting diodes behave essentially in the opposite manner to photovoltaic devices; in other words, they create light from electricity whereas solar cells produce electricity from the light source.

Cathode

Fig. 6.14. Schematic working mechanism for LEDs.

6 Conjugated polymers for organic electronics

|

135

In the case of the simple device made of only two electrodes and the light-emitting material, the efficiency of the charge recombination is very low since the charge can directly travel to the respective opposite electrode, leading to poor electroluminescence. A large number of device architectures have been developed in order to contain the electrons and holes in the light-emitting material. First, the light-emitting layer is “sandwiched” between a hole-transporting layer and an electron-transporting layer to maximize the containment of the charge carriers in the emitting layer (Fig. 6.15). Furthermore, in order to contain even more charge carriers within the emitting layer, the previous structure was optimized by adding a hole and electron blocking layer around the emitting layer (Fig. 6.15). Containing the charges in the light-emitting material allows for better efficiency and stability of the OLEDs and PLEDs. Such a multi-layered structure is more adequate for vacuum-processed molecules than solution-processed polymers and can explain the superior properties of OLEDs versus PLEDs. However, the development of so-called orthogonal solvents (for instance, aqueous versus nonaqueous polymeric solutions) could lead to multi-step depositions and lead to more efficient PLEDs.

Cathode Electron transporting layer Cathode

Hole blocking layer

Electron transporting layer

Light-emitting material

Cathode

Light-emitting material

Electron blocking layer

Light-emitting material

Hole transporting layer

Hole transporting layer

Anode

Anode

Anode

Substrate

Substrate

Substrate

Efficiency and stability Fig. 6.15. Examples of architectures for OLEDs and PLEDs.

6.5.2 Conjugated polymers for light-emitting diodes In order to commercially produce viable PLEDs made of conjugated polymers, they must be soluble enough to be solution-processed and chemically stable over time. A high glass transition temperature is also needed, since heat can be produced in the device during its use and the morphology of the polymer film has to remain intact. From an electronic point of view, the polymer should possess good charge injection and charge transport to produce efficient PLEDs. Dependent upon the color needed, the bandgap of the polymer needs to be tuned. High bandgap polymers can produce

136 | Part I Functional materials: Synthesis and applications blue emission, while low bandgap polymers can produce a red emission. To obtain a clear and clean color, the emission spectra of the polymer should be as narrow as possible. In terms of performance, the device must present high electroluminescence quantum efficiency (φEL ). That quantum efficiency is defined by the ratio of the number of photons emitted over the number of charges injected. This quantum efficiency characterizes the electronic processes in the device but does not measure the light produced. Different approaches have been used to characterize the emitted light. First, there are two units to characterize the light intensity: lumen (lm) and candela (cd). Lumen corresponds to the total quantity of light emitted in all directions, whereas candela represents the quantity of light emitted at a certain angle. In the literature, there are three main parameters to measure in order to fully characterize the light emitted by a LED. These are luminance (Lmax ), which is reported in candela per square meter (cd/m2 ); luminous efficiency (LEmax ), reported in candela per ampere (cd/A), and power efficiency (PE) in lumen per watt (lm/W). The latter is the figure of merit for characterizing LEDs and is widely used by the lighting industry. Traditional lighting technologies mostly rely on incandescent, fluorescent or halogen lamps. While light bulbs show the lowest efficiency (10–35 lm/W), fluorescent or halogen lighting devices are more efficient with power efficiency (PE) ranging from 50 to 100 lm/W. With performance of about 15 lm/W, the best polymer-based light-emitting diodes (PLEDs) are not yet competing with the best lighting sources [38, 39]. Future investigations should find strategies to enhance (by about one order of magnitude) the luminosity of such devices, probably by developing new multi-layered configurations and phosphorescent polymeric materials. However, it is important to repeat here that vacuum-processed OLEDs have already achieved the required electronic and optical properties to make commercial devices.

6.6 Summary Conjugated polymers are an important class of electroactive and photoactive materials. In the last 10–15 years, this research field has literally exploded owing to their applications in energy, micro-electronics, and lighting, to name a few. However, in order to obtain the desired performance with polymeric semiconductors, different parameters at the molecular and supramolecular levels (such as electronic structure, molecular weight, regioregularity, solubility, morphology, etc.) must be carefully controlled. These parameters are different for each application and should be considered during the design of polymer molecular structures, the choice of monomers, and the polymerization reactions. With respect to applications, it is quite evident that conjugated polymers are promised a bright future. Research is not only being conducted in academic institutions, but many companies of all sizes are now involved in R&D projects devoted to semiconducting polymers. On the basis of this tremendous momentum, it is firmly

6 Conjugated polymers for organic electronics

| 137

believed that over the next few years conjugated polymers will continue to address important problems of society, such as the production of renewable energy with minimum detrimental effects on the environment or printed electronic sensors for smart packaging applications. As Benjamin Braddock (Dustin Hoffman) is told in The Graduate: “there is a great future in plastics. . . [or should we now say plastic electronics?] . . . think about it” [40].

References [1]

[2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]

[14]

[15] [16] [17] [18]

Shirakawa H, Louis EJ, MacDiarmid AG, Chiang CK, Heeger AJ. Synthesis of electrically conducting organic polymers : halogen derivatives of polyacetylene, (CHx ). J. Chem. Soc., Chem. Commun. (1977), 578–580. MacDiarmid AG. Synthetic metals: a novel role for organic polymers (Nobel Lecture). Angew. Chem. Int. Ed. 40 (2001), 2581–2590. Shirakawa H. The discovery of polyacetylene film: the dawning of an era of conducting polymers (Nobel Lecture). Angew. Chem. Int. Ed. 40 (2001), 2574–2580. Heeger AJ. Semiconducting and metallic polymers: the fourth generation of polymeric materials (Nobel Lecture). Angew. Chem. Int. Ed. 40 (2001), 2591–2611. Tourillon G, Garnier F. New electrochemically generated organic conducting polymers. J. Electroanal. Chem. 135 (1982),173–178. Diaz AF, Kanazawa KK, Gardini GP. Electrochemical polymeriation of pyrrole. J. Chem. Soc., Chem. Commun. (1979), 635–636. Diaz AF, Logan JA. Electroactive polyaniline films. J. Electroanal. Chem. 111 (1980), 111–114. Edwards JH, Feast WJ. A new synthesis of poly(acetylene). Polymer 21 (1980), 595–596. Ballard DGH, Courtis A, Shirley IM, Taylor SC. A biotech route to polyphenylene. J. Chem. Soc., Chem. Commun. (1983), 954–955. McCullough RD. The chemistry of conducting polythiophenes. Adv. Mater. 10 (1998), 93–116. Leclerc M, Diaz FM, Wegner G. Structural analysis of poly(3-alkylthiophene)s. Makromol. Chem. 190 (1989), 3105–3116. McCullough RD, Lowe RD. Enhanced electrical conductivity in regioselectively synthesized poly(3-alkylthiophenes). J. Chem. Soc., Chem. Commun. (1992), 70–72. Chen TA, Rieke RD. The first regioregular head-to-tail poly(3-hexylthiophene-2,5-diyl) and a regiorandom isopolymer: nickel versus palladium catalysis of 2(5)-bromo-5(2)-(bromozincio)3-hexylthiophene polymerization. J. Am. Chem. Soc. 114 (1992), 10087–10088. Loewe RS, Khersonsky SM, McCullough RD. A simple method to prepare head-to-tail coupled, regioregular poly(3-alkylthiophenes) using Grignard metathesis. Adv. Mater. 11 (1999), 250–253. Kirchmeyer S, Reuter K. Scientific importance, properties and growing applications of poly(3,4-ethylenedioxythiophene). J. Mater. Chem. 15 (2005), 2077–2088. Roncali J. Molecular engineering of the band gap of π-conjugated systems: facing technological applications. Macromol. Rapid Commun. 28, (2007), 1761–1775. Heeger AJ. Semiconducting polymers: the third generation. Chem. Soc. Rev. 39 (2010), 2354–2371. Sakamoto J, Rehahn M, Wegner G, Schlüter AD. Suzuki polycondensation: polyarylenes à la carte. Macromol. Rapid Commun. 30 (2009), 653–687.

138 | Part I Functional materials: Synthesis and applications [19] Carsten B, He F, Son HJ, Xu T, Yu L. Stille polycondensation for synthesis of functional materials. Chem. Rev. 111 (2011), 1493–1528. [20] Mercier LG, Leclerc M. Direct (hetero)arylation: a new tool for polymer chemists. Acc. Chem. Res. 46 (2013), 1597–1605. [21] Ho H-A, Najari A, Leclerc M. Optical detection of DNA and proteins with cationic polythiophenes. Acc. Chem. Res. 41 (2008), 168–178. [22] Rochat S, Swager TM. Conjugated amplifying polymers for optical sensing applications. ACS Applied Materials & Interfaces 5 (2013), 4488–4502. [23] He Z, Zhong C, Su S, Xu M, Wu H, Cao Y. Enhanced power-conversion efficiency in polymer solar cells using an inverted device structure. Nat. Photon 6 (2012), 593–597. [24] Park SH, Roy A, Beaupré S, et al. Bulk heterojunction solar cells with internal quantum efficiency approaching 100 %. Nat. Photon 3 (2009), 297–302. [25] Peters CH, Sachs-Quintana IT, Kastrop JP, Beaupré S, Leclerc M, McGehee MD. High efficiency polymer solar cells with long operating lifetimes. Adv. Energy Mater. 1 (2011), 491–494. [26] Zhao G, He Y, Li Y. 6.5 % efficiency of polymer solar cells based on poly(3-hexylthiophene) and indene-C60 bisadduct by device optimization. Adv. Mater. 22 (2010), 4355–4358. [27] Zhang Y, Zhou H, Seifter J, et al. Molecular doping enhances photoconductivity in polymer bulk heterojunction solar cells. Adv. Mater. 25 (2013), 7038–7044. [28] Shaw JM, Seidler PF. Organic electronics: introduction. IBM J. Res. & Dev. 45 (2001), 3–9. [29] Namdas EB, Saricifti NS, Heeger AJ. Semiconducting and metallic polymers. Oxford University Press, New York, 2010. [30] Sirringhaus H, Brown PJ, Friend RH, et al. Two-dimensional charge transport in self-organized, high-mobility conjugated polymers. Nature 401 (1999), 685–688. [31] Li J, Zhao Y, Tan HS, et al. A stable solution-processed polymer semiconductor with record high-mobility for printed transistors. Sci. Rep. 2 (2012), 754. [32] Wang S, Kappl M, Liebewirth I, et al. Organic field-effect transistors based on highly ordered single polymer fibers. Adv. Mater. 24 (2012), 417–420. [33] Park JH, Jung EH, Jung JW, Jo WH. A fluorinated phenylene unit as a building block for highperformance n-type semiconducting polymer. Adv. Mater. 25 (2013), 2583–2588. [34] Holonyak N, Bevacqua SF. Coherent (visible) light emission from Ga(As1–xPx) junctions. Appl. Phys. Lett. 1 (1962), 82–83. [35] Pope M, Kallmann HP, Magnante P. Electroluminescence in organic crystals. J. Chem. Phys. 38 (1963), 2042–2043. [36] Tang CW, Van Slyke SA. Organic electroluminescent diodes. Appl. Phys. Lett. 51 (1987), 913–915. [37] Burroughes JH, Bradley DDC, Brown AR, et al. Light-emitting diodes based on conjugated polymers. Nature 347 (1990), 539–541. [38] Huang J, Hou W-J, Li J-H, Li G, Yang Y. Improving the power efficiency of white light-emitting diode by doping electron transport material. Appl. Phys. Lett. 89 (2006), 133509. [39] Beaupré S, Boudreault P-LT, Leclerc M. Solar-energy production and energy-efficient lighting: photovoltaic devices and white-light-emitting diodes using poly(2,7-fluorene), poly(2,7-carbazole), and poly(2,7-dibenzosilole) derivatives. Adv. Mater. 22 (2010), E6–E27. [40] http://www.youtube.com/watch?v=PSxihhBzCjk (Accessed January 28, 2014)

A. Soldera

7 Theoretical tools for designing microscopic to macroscopic properties of functional materials The very essence of science is to unveil the huge palette of surrounding mysteries. Deciphering the forces that sustain an object in water, or understanding how an apple falls from a tree are some well-known findings which have made a huge step towards comprehending our world for humanity. Our contemplation of the world thus changes according to the current advancements of science. Thanks to the extensive progress made with experimental techniques and computers, it has become possible to reveal some aspects of the behavior of the microscopic domain. From a computational viewpoint, approaches in solving present-day problems are certainly different than those used 50 years ago. In the realm of functional materials, have we actually reached what Dirac desired in 1929 in the Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character, Vol. 123, No. 792 (6 April 1929): The underlying physical laws necessary for the mathematical theory of a large part of physics and the whole of chemistry are thus completely known, and the difficulty is only that the exact application of these laws leads to equations much too complicated to be soluble. It therefore becomes desirable that approximate practical methods of applying quantum mechanics should be developed, which can lead to an explanation of the main features of complex atomic systems without too much computation.

This chapter intends to address the issue raised by Dirac to a certain extent. More specifically, we will explore whether codes and computers are efficient enough to reveal the behavior of the complex atomic systems which ultimately lead to the world we know. This would mean that, by improving computer performance, such as reaching exascale computing and beyond, solving the “equation of everything” [1] i.e., the famous Schrödinger’s equation, the whole mechanism of the microscopic world would be revealed. At the risk of disappointing the reader, the answer seems to be “No, it cannot”, as claimed by Philip W. Anderson (Nobel prize in Physics in 1977) [2]. Viscosity is an illustrative example. It can be roughly explained by the presence of dissipative forces which cannot be deduced as a result of solving the Schrödinger’s equation. The answer actually divides two wholly opposed viewpoints: reductionism and emergence [3, 4]. Reductionism envisions understanding a system through its “elementary” constituents and their interactions. Emergence can be summarized by the vision of Plato: the whole is greater than and different to the sum of its parts. Between these two opposing perspectives, obtaining a clear response is not straightforward. Molecular simulation is in some sense to illuminate this antagonism. Although this chapter is not intended to discuss this aspect in detail, such a reflection should nevertheless be kept in mind.

140 | Part I Functional materials: Synthesis and applications The first part of the chapter is devoted to those issues ongoing from microscopic to macroscopic scales, and thus the importance of possessing notions in statistical thermodynamics to make meaningful molecular modeling. Examples of this statement are presented in the second part of the chapter, and aim to illustrate the different uses of simulation. These two examples originate from collaborations with my distinguished colleagues from the Quebec Centre for Functional Materials (CQMF), Professors Claverie and Morin, who are also authors of Chapters 2 and 3 in this book.

7.1 Methods 7.1.1 The link between microscopic and macroscopic scales At the heart of molecular simulation it is necessary to establish the link between micro and macro scales. Statistical physics is the branch of science which provides the tools to build this bridge. However, relationships are first established for systems with no or very weak interactions. Molecular simulation can be seen as an experiment performed with a computer using a code [5]. It thus provides the means to apply such tools in complex systems such as functional materials. However, special care is needed with regard to applying these tools. The following example illustrates the nature of the problem when deducing macroscopic properties from the microscopic realm. Without a doubt, it takes some liberties from a rigorous approach which would necessitate a greater discussion of statistical physics. It is actually intended to give a simple overview of the general issues required to relate the micro and the macro states and their properties, as schematically summarized in Fig. 7.1. Consider a glass of water in a room whose ambient temperature is 10°C. Now, a Maxwell demon [6], instead of opening a gate to select molecules with specific velocity, measures the speed of the water molecules inside the glass of water by observing each individual water molecule. Moreover, he clearly ignores any uncertainty principle raised by Heisenberg (see equation (7.2)). Accordingly, he is able to provide the speed (vo ) of one molecule of water. The equipartition theorem indicates that the square of speed is related to the temperature¹. He thus finds that the computed temperature is very high, strictly higher than the ambient temperature. Despite his demon character, he is astonished and says “Hey, there is some nonsense going on here”. Back to reality, the actual incongruity comes from applying the equipartition theorem to the speed of only one molecule. The real explanation is summarized in Fig. 7.2. Distributions need to be considered, and the Maxwell–Boltzmann distribution must be applied in this case in order to compute an average value. By integrating the sta-

1 More rigorously, each degree of freedom described by the square of any coordinate corresponds to 1/2 kB T, where kB is the Boltzmann constant.

7 Theoretical tools for designing functional materials |

°C 5 4 3 2 1 0 1 2

0 0 0 0 0 0 0 0

141

νo

ν2 α T

To >> 10°C

Fig. 7.1. Principle of inconsistency in measuring only properties of a single molecule.

tistical weights over the whole domain, or, in the case of discreteness, by adding up all the possibilities, the result is one. Distributions are thus required to handle microscopic properties. Only then can the comparison between speeds measured at the microscopic level and the value of the macroscopic temperature be undertaken in a meaningful fashion. Applying this rule appropriately is at the heart of molecular simulation. Instead of speed, you can think of any other microscopic property, be it a scalar, a vector or a tensor. Of course, it is not applied when you are only interested in the configuration exhibiting the minimum energy, as in the crystal. Micro

Macro ν2

ρ(ν) Distribution (Maxwell-boltzmann)

α T

Vo

V

Fig. 7.2. Distribution is required when microscopic properties are compared to macroscopic values; vo is the result of the fishing in Fig. 7.1.

In Gibbs’ approach to statistical thermodynamics, the calculation of the average of a distribution depends on the statistical ensemble of systems being applied. Consider a distribution of systems in which the probability is described by pi for the appearance of the ith system. Then ∑i pi = 1. For the sake of clarity, the integral form is not introduced here. To compute the system ensemble average of a property (A), one needs to compute, ⟨A⟩ = ∑i Ai pi where Ai is the value of the property A for the ith system. At this point, the ergodic hypothesis must be introduced briefly. It specifies that the value of A over a huge number of systems at a specific time, ⟨A⟩ equals the average stemming from considering the distribution of the property of a single system over a “long” period of time, A.̄ When Ā = ⟨A⟩ the computed property can be compared to the experimental one. For a more detailed description of the ergodic hypothesis, interested readers are referred to the seminal book of Allen and Tildesley [7]. However, computing Ā is time-consuming and it can be difficult to obtain by simulation. A most accurate representation of ⟨A⟩ is thus mandatory. Experimentally, ergodicity

142 | Part I Functional materials: Synthesis and applications is not attained in certain systems, such as molecular glasses and polymers. However, conceptually, a simulation code must be ergodic. The different scales have thus been exposed. The intent of this chapter is not to provide the theoretical development underlying each scale. A list of books focusing on this topic is available at the end of the chapter. This chapter aims to reveal the “very substance”, as Gargantua said to Pantagruel, of molecular simulation, and to emphasize the main difficulties most students will encounter during this course. The bridge between scales will also be discussed.

7.1.2 Ab initio methods “In the beginning was” Schrödinger’s equation, whose expression when it is not timedependent is: ̂ (r)⃗ = Eψ (r)⃗ Hψ (7.1) This equation indicates that the Hamiltonian operator H,̂ acts on a wave function, ψ, which depends on coordinates of electrons r ⃗ (considering only electronic contribution). This operation leads to the energy (eigenvalue) multiplied by the same wave function which is then known as an eigenfunction. In bounded states² the resulting energies are quantified. A difference in energy between two levels is then related to a frequency which can be measured experimentally. Introducing the Dirac notation, the ensuing equation (Fig. 7.3) can be represented similar to the one shown in Fig. 7.2, and indicates that the resulting energy corresponds to an average. However, there is a clear distinction to be made between these two averages. For the quantum average in Fig. 7.3, the microscopic state, described by ψ, is known; in this case, the ensemble is constituted only of systems in the ψ state, which is not necessarily an eigenstate of the Hamiltonian. In statistical thermodynamics, in addition to the quantum uncertainty, there is uncertainty regarding the state of the system; thus, the ensemble average of Fig. 7.2 is over a distribution of systems in different states, each with its own statistical weight. Micro ψ Ĥ ψ Distribution

E

Macro Fig. 7.3. Schrödinger’s equation in Dirac’s notation.

In the kingdom of electrons, it might seem amazing that no classic apparatus is able to reveal one electron specifically. What is the actual problem? In a first step, we consider our classical world, and try to use our understanding of it to disclose the microscopic

2 The electron confined to the field exerted by the nuclear charge is an example of confined state.

7 Theoretical tools for designing functional materials |

143

environment. The use of the adjective ‘classic’ indicates that the world is governed by the classical laws of physics, such as the Newtonian laws of mechanics, or Maxwell equations in optics. So, if we want to find our way in a room in the middle of the night and avoid any untimely meetings with chairs, we need a flashlight. Does this light change our environment? The answer is clearly no. Walls do not move, nor does their shape change as a result of the beam of our flashlight. Can we use such an experiment at the electron scale? Before answering this question, we have to answer to at least two other ones. First question: is the electron a particle, or a wave? The answer depends on the kind of experiment carried out to reveal the presence of the electron. To symbolize this problem, we use the image of the 2D world proposed by Carl Sagan in his famous book, Cosmos, to represent the fourth dimension [8]. A cylinder is a 3D object. Its base lies on the (x, y) plane. The demon (taking a little break from fishing for water molecules) has only two possibilities for examining it: through the x or y axis, and along the z axis. He actually changes color according to the direction he looks. In the first case, he observes a rectangle while in the second situation, a circle is revealed to him (Fig. 7.4). Accordingly, dependant on the experience of looking at the cylinder from two possible observation points, the cylinder is either a rectangle or a circle. In the same way, the electron is either a particle or a wave, depending on the type of instrument used to observe it. Let us take a brief detour from this particular wave-particle duality of electrons. In the late nineteenth century, the electron was recognized as carrying a negative charge instead of being considered to be embedded in ether. The existence of subatomic corpuscles was evidenced by J.J. Thomson in 1897 [9]. However, De Broglie showed in his Ph.D. thesis published in 1924 that particles, e.g., electrons, behave as waves. This was experimentally confirmed the following year [10].

y Fig. 7.4. In a 2D world, is a cylinder a rectangle or a circle?

X

The second question concerns the fact that the position and momentum (velocity times weight) of the electron cannot be determined at precisely the same time, i.e. Heisenberg’s uncertainty principle: σpx σx ≥

h , 4π

(7.2)

144 | Part I Functional materials: Synthesis and applications where σpx and σx are the standard mean deviation of (respectively) momentum and position along the x axis, and h is the Planck constant. The intrinsic behavior of electrons is thus barely grasped, is even impossible to grasp, by classical methods. Using an analytic tool, i.e. the equivalent of the flashlight but for electrons, the state of the electron will automatically change. How then is the configuration of electrons uncovered? By using mathematics! Very unpopular answer, as I usually hear from the students in my quantum chemistry class! The reader can however be reassured: the purpose of this overview of simulation applied to functional materials is not to develop mathematical models. It is aimed at explaining how mathematics is used to grasp details that govern the microscopic world. Mathematical description can be retrieved from those textbooks suggested at the end of this chapter. The equation in Fig. 7.3 indicates that the state of the system, and accordingly of the electrons, can be described by a mathematical function (ψ), called the wave function. When it is squared, this equation leads to a physically relevant result: the probability of finding an electron somewhere in the space (Born interpretation). However, it can only be solved analytically for one electron. Indeed, adding another electron, combined with the fact that electrons are indistinguishable particles (being fermions), as well as the presence of a spin and Heisenberg’s uncertainty principle make analytical solution of Schrödinger’s equation impossible. Computing Coulomb interactions between these two negative charges becomes thus unfeasible, since their distance of separation remains unknown. Accordingly, approximations are required. The prerequisite in manipulating approximations to solve the Schrödinger equation gives rise to different approaches. They can be roughly classified into three major types: ab initio (Hartree–Fock method and derivatives), density functional theory (DFT), and semiempirical methods. Ab initio is a Latin phrase which means “from the beginning”. In this approach, only fundamental constants are used and no experimental data are introduced to compensate for the use of approximations. These approximations thus lead to numerous commercial or freeware codes such as Gaussian [11, 12], Gamess [13], DMol [3] Accelrys [14], and others. The Hartree–Fock method is the most common ab initio approximation employed by scientists. The usual approximation employed by chemists is to handle hydrogen atomic orbitals³. The intrinsic reason for employing hydrogenoid atomic orbitals comes from the fact that Coulombic interactions between electrons can be reduced to a central core problem. This makes it much easier to calculate, through the introduction of the mean-field concept: each electron undergoes interactions with all other electrons. In order to get the final wave function, a selfconsistency procedure is then applied. Nevertheless, initial wave functions are necessary. Hydrogenoid wave functions are chosen since they correspond to the exact solutions of the Schrödinger’s equation applied to the hydrogen atom. For practical

3 Physicists usually prefer the plane waves approach.

7 Theoretical tools for designing functional materials |

145

reasons, these wave functions are replaced by linear combinations of Gaussian type orbitals (GTO), or Slater orbitals. A notation for GTO was introduced by Pople⁴ et al. to indicate the number of Gaussian functions used to describe the different hydrogenoid orbitals forming a basis set of equations. The essence of the basis set consists of functions which fit the hydrogenoid atomic orbitals, such as 6-31g. 6-31g actually means that 6 Gaussian functions depict the core electrons and the valence electrons are represented by a combination of two functions (Double Zeta), represented by 3 and 1 Gaussian functions respectively; g stands for Gaussian orbitals. Specific functions can then be added, such as polarization functions (d, p) to enhance directional effects, or diffuse functions (+), to take long-range effects due to the presence of electronic delocalization in the studied system. Post-Hartree–Fock methods, such as the Møller–Plesset perturbation theory (MP), are aimed at refining this approach, but require more Central Processing Units (CPU) and are therefore time-consuming. In order to improve computer efficiency, the Density Functional Theory (DFT) method offers a very interesting alternative. Within this approach, total energy is expressed in terms of total electron density, instead of the wave function which depends on atomic position. This functional, i.e., the function of a function is inferred from computations or experimental measurements. The PBE and hybrid B3LYP functionals are the most commonly used, hybrid meaning that it also contains a Hartree–Fock component. However, the choice of functional must be carried out carefully, since a universal functional is not yet available. To further increase computer efficiency, CPU and time consuming methods such as multiplication of overlap matrices can be removed. These omitted data must then be balanced by experimental data such as the heat of formation. The most-employed semi-empirical codes are the Austin model 1 (AM1) and Parameterized Model (PM3). The main drawback in using quantum calculations is the prohibitive demand on CPU. This problem increases with the number of atoms. As a typical example, the computer time for the Hartree–Fock approach is proportional to N4 (which is the total number of integrals to be evaluated), where N is the number of basis functions used to described the studied system. For DFT, the computer time increases to N3 . DFT is generally employed for large systems. In practice, optimization (i.e. finding the structure of lowest energy), can be carried out using a low level basis set. It can then be followed by calculations carried out at a higher level of theory without the optimization step, leading to single point energy. Alternatives also exist to address problems stemming from larger systems such as polymers, where long range and large time scale are relevant factors in computing properties. Thus, instead of considering electrons as wave functions, the atomistic approach depicts atoms as particles with electronic interactions embedded in a force field.

4 Nobel Prize in chemistry in 1998 “for his development of computational methods in quantum chemistry", shared with Kohn “for his development of the density-functional theory”.

146 | Part I Functional materials: Synthesis and applications The Schrödinger’s equation and the different approximations are used when specific interactions involving electrons are of importance: electronic conjugation, proton dissociation, infrared and NMR spectra, hyperpolarizabilities and others. However, it is not always necessary to consider specifically all electrons during calculations. As an atom moves in a medium, interactions with other atoms can be approximated in many cases. This link to another level is discussed in the following section.

7.1.3 Bridging the gap between ab initio and atomistic levels The graph shown in Fig. 7.2 corresponds to the Maxwell–Boltzmann distribution of gas molecules. It may be used to reveal the bridge between the electronic and atomic levels [15]. Rather than debating this 3D distribution, we focus on the distribution of momentums in one direction only (x axis), which has the following form: f(px ) = (

1/2 p2x 1 ) exp (− ), 2πmkB T 2mkB T

(7.3)

where m is the weight of the particle, kB is the Boltzmann constant, and T is the temperature. Again, this equation might seem complex, but it is essential to further understand this approach. The shape of this distribution corresponds to a Gaussian curve that presents the following generic form: g(x) =

1 x2 exp (− 2 ) , 2σ σ√2π

(7.4)

where σ is the standard mean deviation. This Gaussian curve is found in many systems containing purely random events: tossing coins, dice etc. . . For example, let’s consider tossing coins, for which the result can be either heads or tails. One event can be tossing a coin 30 times. What is then the probability that in these 30 flips, the result is 15 times heads, and thus 15 times tails? This event is actually x in equation (7.4), and the probability that this event occurs is g(x). However, in order for the discreteness⁵ (flipping the coins, or the microscopic level) to reach continuity (Gaussian curve, and thus the macro level), the number of events must be important. In fact, there is a theorem, the central limit theorem, which actually specifies that the standard mean deviation varies as 1/√N, where N is the number of events. By comparing equations (7.3) and (7.4), it can be found that σpx = √mkB T. Inserting this value into equation (7.2) leads to the conclusion that uncertainty in the x position of a particle of weight m must be greater than h/(4π√mkB T). Thus at 300 K, the precision in the position of the electron is of the order of 15 Å, while for the hydrogen

5 This discretization of the distribution must be put in parallel with the measuring of the velocity of the water molecules one by one as the demon did.

7 Theoretical tools for designing functional materials |

147

(van der Waals radius rH = 1.2 Å) and carbon atoms (van der Waals radius rC = 1.7 Å), it is 0.35 and 0.1 Å respectively. This confirms that electrons are quantum particles, since we have no idea where they can be located. Moreover, heavy atoms such as carbon atoms can be considered to be governed by classical mechanics. However, the classical character of the hydrogen atom is difficult to establish clearly, although it can generally be considered to be a particle, and can then be described by classical equations.

7.1.4 Atomistic simulation At the heart of any atomistic simulation we find the force field. Originally stemming from vibrational spectroscopy, it was applied to depict interactions between all atoms, including non-bonding terms. For example, consider an atom in a specific environment, for instance two linked sp3 carbon atoms. This environment can be defined by a particular potential energy surface (PES). Such a PES does not change greatly if the same environment is found in another system or molecule. Following the previous example, the number of molecules possessing this C–C link is unlimited. A force field is then aimed at representing the PES as accurately as possible. To achieve this goal, a force field in fact consists of (1) equations in which terms tend to grasp specific interactions, and (2) parameters which stem from the fit of these equations to the different PES. Experiments or quantum calculations are carried out to accurately depict the PES. Returning to the C–C link, we will show how PES associated with interactions stemming from this molecular characteristic is computed, and subsequently reveal the great diversity in force fields. We actually describe the link between two carbon atoms in a simple ethane molecule. 200 Eh = k (d–d0 )2 Harmonic function

180

Energy (kcal/mol)

160

EM = Ediss. [1–exp(–α (d–d0 ))]2

140 120

Morse function

100 80 60

4

40

2

20

0 1.4

0 0

1

1.5

4 2 3 C-C distance (A)

1.6

1.7

5

6

Fig. 7.5. Potential energy with respect to the C–C distance in an ethane molecule. The dotted curve represents the actual energy behavior.

148 | Part I Functional materials: Synthesis and applications The PES associated with the C–C elongation can be computed using quantum methods. It is in fact reduced to a simple curve (Fig. 7.5). The potential energy is thus computed, while the distance between two carbon atoms in an ethane molecule is varied. A dotted line, joining the different data, is displayed. The Morse function is usually employed to fit these data: it is displayed in gray in Fig. 7.5. As can be seen, three parameters are required to fit the curve: the dissociation energy, Ediss. , a parameter which describes the width of the potential well, do , which corresponds to the distance of minimum in potential energy, the actual equilibrium distance between the two carbon atoms, and α which defines the well width. Since a covalent bond is exposed to small fluctuations around do most of the time, the Morse function can be expanded in a Taylor development at d = do until the third term. This leads to a harmonic function (in black in Fig. 7.5), revealing in fact an elastic behavior (Hooke’s law). The three parameters of the Morse function used to fit the data are then replaced by only two fitting parameters: the spring constant, k = Ediss. α2 (stemming from the Taylor development), and the equilibrium distance, do . The link between two atoms can thus be described by either of these two functions. The harmonic function is usually preferred to the Morse function since it needs two fitting parameters instead of three, and more importantly it does not use an exponential function which consumes more CPU time. Nevertheless, as can be observed in the inset of Fig. 7.5, a quadratic function is not sufficient to describe interactions between two atoms properly. Computational data are thus fitted by a 4th or 6th order polynomial. The higher the order, the better the fit, and thus less important errors leading to better transferability (i.e., the availability of a force field for the study of a great number of systems). Depending on the kind of calculations to be carried out, several functions are thus available. In fact, there are other parameters which also differentiate the force fields: – Parameterization: what is the chosen value for do , 1.53 or 1.54 Å, or is greater precision needed? Actually, the equilibrium value depends on the kinds of systems investigated and the type of calculation carried out. – Environment: how many different environments of an atom are needed? Consider the sp2 carbon atom. If only one series of fitting parameters is selected, one will certainly encounter problems when willing to differentiate between alkene and ketone bonding. In the case of second generation force field, carbon atoms can have 20 different environments, making such a force field more transferable. In summary, interactions between two linked atoms require a function and parameters specific to each atom in order to be represented. However, the potential energy to describe a molecule in a medium needs additional terms. They are characterized in Fig. 7.6, where 4 atoms are schematically displayed. The force field equation is in fact separated into two terms: bonding and nonbonding equations.

7 Theoretical tools for designing functional materials | 149

d φ θ

rij

Fig. 7.6. Representation of the variables discussed in equations (7.5–7.9).

Bonding interaction terms The bonding interaction terms depend on the internal coordinates of the molecule, i.e., bonds, valence angles, and dihedral angles. Accordingly, its simplest expression consists of the sum of all the terms that are functions of stretching, bending, and torsion found in a molecule. Thus, in addition to the stretching term previously defined, a simplified force field, usually classified as a class I generation force field, possesses valence and dihedral angle terms to describe the bonding interactions. – Stretching term: Estretching (d) = ks (d − do )2 (7.5)

–

–

where d is the atomic distance at which the stretching potential energy is computed, ks and do are the spring constant and the value of atomic distance for which the stretching potential energy is zero respectively. This actually corresponds to the black curve in Fig. 7.5. Bending term: EBending (θ) = kθ (θ − θo )2 (7.6) where θ is the valence angle at which the bending potential energy is computed, kθ and θo are the spring constant and the valence angle for which the bending potential energy is zero respectively. Dihedral term: 3

Ediheadral (φ) = ∑ Vn [1 − cos(nφ − φ0n )]

(7.7)

n=1

where φ is the torsion angle at which the dihedral potential energy is computed, Vn corresponds to the potential energy barrier, and φ0n the phase. Other forms of bonding terms exist in a class I force field, such as the out-of-plane term, due to variations in the planarity of a group of atoms as in the ester group. Moreover, due to increasing computational efficiency, bonding interaction cross terms, as well as higher terms in Taylor development, can be added to better depict the PES, leading to second generation force fields which cover a greater number of compounds and thus have greater transferability. Accordingly, a great variety of force fields exists: pcff, COMPASS, OPLS, MM2, MM4, UFF, AMBER, ESFF, CVFF, etc. [16–18]. Although not explored in this chapter, interested readers are invited to read about the work of Sun [16].

150 | Part I Functional materials: Synthesis and applications Non-bonding interaction terms Non-bonding terms express the interactions between linked atoms separated by a certain number of atoms (2 in Fig. 7.6), as well as between atoms belonging to different molecules. They generally consist of two terms. – van der Waals interactions Aij Bij EijLJ (rij ) = 12 − 6 (7.8) rij rij

–

where Aij and Bij are fitting parameters, and rij is the actual distance between atoms i and j. The equation displayed in (7.8) corresponds to the Lennard-Jones −6 potential: it is formed by a repulsive (in r−12 ij ) and a dispersive (in rij ) term. While the latter term results from the London–Keesom–Debye interactions [19], the former term can have different forms. The r−12 behavior can be substituted by an ij exponential decay, exp(−ρrij ) to yield the Buckingham potential, or a less abrupt function, r−9 ij [20]. Electrostatic, or Coulomb potential energy: EijCoulomb (rij ) =

qi qj 4πεo rij

(7.9)

where qi , qj , and εo are respectively the partial charges of atoms i, j, and the dielectric constant. The partial charges reproduce the first nonzero multipolar component. They stem from the difference in electronegativity between atoms sharing a bond. Different equations are thus available to portray the whole PES. The approach used to establish the PES from experiments or calculations can lead to different fitting parameters and also contribute to the great diversity of available force fields. The MonteCarlo method [21] and molecular mechanics consist of direct use of a force field. Static properties such as energies [22], infrared spectra [23], and mechanical properties [24] can be computed with the latter approach. Molecular dynamics (MD) is required to reveal dynamic properties, such as autocorrelation function [25, 26] and relaxation times [27]. It consists of integrating equations of motions, usually the second law of Newton, to give mobility to atoms considered to be simple particles with interactions, as previously stated. Most useful information is obtained by using MD. However, the main problem with MD simulation is that the time period covered is usually small. In order to integrate Newton’s equations, a time step (δt) is required for which various types of integrators could be used. For example, a peculiarity of the Verlet algorithm and its derivatives is the use of an “important” time step: δt = 1 fs, or 10−15 s. This means that to produce a trajectory, i.e., a series of conformations during 1 ns, solving the equations would necessitate 1 000 000 integrations! Each integration consists of computing the velocity and position of all atoms, according to the selected force field (second law of Newton). So why would it be necessary to use such a low integration step? The main reason stems from the fact that all of the motions involving atoms

7 Theoretical tools for designing functional materials |

151

must be correctly reproduced. The most rapid motion existing among atoms is the C– H stretching vibration, of the order of 10 fs. Thus, an integration step of 1 fs using the Verlet algorithm makes the use of MD available to accurately describe atoms’ motions. To increase the duration of MD, it is necessary to diminish the number of integration calculations and/or increase the integration step. One method is to replace C–H with a bigger “C” atom whose force field parameters will definitely be different to the whole atom description. Such a procedure is called coarse-grained simulation and it is at the beginning of the mesoscale level simulation.

7.1.5 Bridging the gap between atomistic and mesoscale levels As the duration of MD increases, dissipative effects emerge. The Newton equations of motion occur in systems with constant energy. In order to introduce a temperature term into a system, various algorithms can be used. However, dissipative effects leading to an exchange of energy cannot be easily inserted into such systems, since all these effects take place at the mesoscale level. The purpose is thus to reproduce the general behavior in a more efficient way. The actual problems stemming from atomistic simulation are the large number of degrees of freedom. We must therefore distinguish between those that are of interest for the simulation and those that must reach equilibrium rapidly. The general behavior of big molecular systems must be correctly simulated, while specific details can be neglected at the mesoscale level. Let us consider a polymer mixed in a solvent, where the polymer behavior is the main interest. Details of the solvent molecules can actually be ignored in the simulation of such a system. Similarly, if this polymer corresponds to a very long chain, details of atoms or valence angles and dihedral angles are not of importance in describing how the polymer behaves. On the other hand, the center of mass of a group of atoms whose length can be Kuhn’s length is more important. Dissipative particle dynamics simulation uses such approximations, and is very promising. We recently published a study where we showed that water channels can be formed in fuel cell membranes by applying shearing. They are observed only for long chains and low water content [28]. Despite growing interest, mesoscale simulation is not discussed in this chapter. The focus is on scales where chemical functionality is explicitly pictured, i.e., the quantum computational and atomistic domains. Examples of the applications of these two domains of approximation will be the focus of the next section.

7.2 Examples Molecular simulation is a powerful tool but it must be manipulated with great care. Calculations can be run easily (push-button temptation stemming from the simple handling of commercial codes), providing some data. Conclusions can then be drawn

152 | Part I Functional materials: Synthesis and applications rapidly, but how effective are they exactly? Many questions must be answered in order to run a simulation efficiently. For instance, is the pertinent portion of the phase space correctly explored? Have the right calculation characteristics been chosen, such as the functional in DFT or the force field in atomistic simulation? Instead of exposing a list of criteria which must be fulfilled before extracting any meaningful conclusion from calculations, this section will emphasize the procedure that must be employed. Since molecular simulation is based on approximations, correlation of results stemming from calculations with experimental data must be carried out first. Corrections can then be brought to the type of calculation or to the protocol used. Examples will be discussed in order to illustrate this procedure. The very purpose of simulations performed in the lab is to guide the synthesis or formulation of new compounds through a better understanding of the molecular interactions that give rise to the macroscopic properties. Functional materials particularly suit this approach, since small changes in the molecule can lead to important changes in the final properties. Accordingly, if molecular simulation can deal with such changes, new molecules with enhanced properties can be proposed. Molecular simulation is thus particularly aimed at supporting experimentalists in their studies.

7.2.1 Quantum studies Quantum calculations are of particular interest to guide experimentalists in the development of materials and the optimization of their properties. Two illustrations of this objective are reviewed. The first example is related to collaboration between our laboratory and the laboratory of our colleague Professor Jean-François Morin, author of Chapter 3 in the current book. The other example deals with a study carried out in our laboratory.

Electronic properties [29] Professor Morin and his group study organic and molecular electronics (Chapter 3). Such compounds are very promising materials due to their great ability to tune their properties by modifying their chemical structure. Among these compounds, fullerene (C60 ) is a very interesting n-type candidate, since it can easily accommodate negative charges in a reversible way. The reason for this peculiarity stems from the presence of a triply degenerate lowest unoccupied molecular orbital (LUMO) level which allows the addition of up to six electrons to a single fullerene cage. However, the LUMO of C60 is 4.5 eV, which is too high for electronic applications. Variations of this value can be addressed through the addition of specific groups. Molecular simulation then becomes of great interest, since the effect of this group on the value of the LUMO can be rapidly revealed. Nevertheless, comparison with experimental data must first be carried out. Koopmans theorem is used to compute approximate ioniza-

7 Theoretical tools for designing functional materials |

153

tion energy, considering that it corresponds to the negative of the highest occupied molecular orbital (HOMO). Computing electron affinity, i.e., negative of LUMO, is less straightforward, due mainly to the fact that it is a virtual molecular orbital. A series of new molecules derived from molecules with an ethynyl-bridged function and linked to fullerene have been synthesized by Professor Morin’s group and have also been characterized electrochemically. Selection of the appropriate functional and basis set was first carried out. The ensuing values for the LUMO were compared to computational data (Fig. 7.7, left). Despite an underestimation of the LUMO, a linear relationship was obtained for almost all of the synthesized compounds. However, the LUMO of the molecule possessing a penta-fluoro phenyl group at the α position of the ethynyl-bridge in the fullerene (Fig. 7.7, right), was found out of range, with a very low LUMO. The experimental LUMO was −3.84 eV. Using the PBE functional, a value of −4.05 eV was computed. The chemical structure of the molecule indicates that it mainly possesses one degree of freedom that is the torsion angle (φ) associated with the bond linking the penta-fluoro phenyl group to the fullerene. This dihedral angle was turned in 10° steps. At each step, the LUMO was calculated. The resulting graph is shown in Fig. 7.7 on the right. As observed in this graph, the value of the dihedral angle greatly influences the value of the LUMO. –3.85

–3.70

Lumo simulated/eV

3 8

6

–4.05

7

4

1

–4.00

PC 5 BM

2

–3.95

–4.10 –4.15

–4.25

–3.75

OC8H17

φ

–3.80

F

F

–3.85

F –3.90

F F

–3.95 –4.00 –4.05

C6 0

–4.20

–3.82

–3.84

–3.86

–3.88

–3.90

–3.92

–3.94

–3.96

–4.00

–4.10 –3.98

Lumo simulated/eV

–3.90

10

20

30

40

50

60

70

80

φ (deg.)

Lumo experimental /eV Fig. 7.7. LUMO with respect to the dihedral φ.

When fluorine atoms are in an ortho position relative to the 1-ethynyl-4-(octyloxy)benzene group, φ = 10° (position shown schematically in Fig. 7.7, right), the highest value of LUMO (−3.71 eV) is obtained. Rotating the phenyl group until the fluorine atoms are approximately at the same distance from the fullerene (φ = 80°), results in a much lower LUMO energy level (−4.05 eV). This difference of 10 % between the two values can be directly attributed to the δ− character of fluorine atoms, in agreement with the electrochemical results. In fact, substituting the fluorine atoms with hydrogen atoms in the phenyl group (leading to compound 3 in Fig. 7.7, left) does not have any impact on the value of LUMO by rotating the dihedral angle. Accordingly, calculations show

154 | Part I Functional materials: Synthesis and applications that LUMO can be tuned when a hexa-fluoro phenyl group is linked to C60 . Molecular simulation thus acts as a guide to the synthesis of new compounds. More specifically, it does not give the final molecule, but gives hints for the development of new functional materials. In other words, molecular simulation can be used to guide the synthesis of new and more efficient materials. A different approach, resulting in a proposition of very new molecules based on simulations, will be developed in the next paragraph.

Design of new membranes [30, 31] Polymer electrolyte membrane fuel cells (PEMFC) are the primary system being developed for use in automotive applications. The perfluorinated polymer Nafion® remains extensively used as a PEM. Since it cannot be operated at low humidity and/or at temperatures above 100°C, great efforts have been dedicated to proposing alternatives to this copolymer. They can be roughly classified into two main approaches. The first series of molecules is derived from the structure of Nafion® , i.e., they possess a polytetrafluoroethylene (PTFE) backbone and randomly distributed fluorinated ether side chains, each terminated with a sulfonic acid group. The second type of molecule possesses the sulfonate acid part mentioned above, but the rest of the core is generally polyether ether ketone (PEEK)-like. The limited development of new classes of membranes may stem from the lack of understanding of the factors contributing to the morphology, conductivity, and hydration of Nafion® . Great effort has been made to develop new membranes through trial and error and serendipity, although these strategies are not discussed in this chapter. Nafion® is a long polymer chain with a huge degree of freedom. Its polytetrafluoroethylene (PTFE) backbone is extremely hydrophobic, while the fluorinated side chains have high electron affinity which promotes the dissociation of the proton of the sulfonic acid, making it a superacid. Due to the electronic contribution, quantum method is the appropriate approach for dealing with proton dissociation. Rather than considering all the chains, our group performed calculations focused on trifluorosulfonic acid (TfOH), CF3 SO3 H. Our methodology was then to find conformation of minimum energy as a water molecule is added to the system. Infrared spectra were computed at each step and compared to experimental ones, which allowed us to notice that considering only one TfOH is not sufficient to accurately describe the infrared spectra. Two acidic molecules were found to be sufficient. However, studies with three TfOH molecules involve an important number of degrees of freedom (important configurational space), making access to all potential energy wells hazardous. The agreement between selected computed and experimental frequencies indicated that the structural changes occurring during proton dissociation could be used. This sequence of molecular systems can be seen as a flip book (i.e., a book with a series of pictures that vary gradually from one page to the next, so that when the pages are turned rapidly, the pictures appear to animate by simulating motion or some other change (Wikipedia)). Each image of the sequence corresponds to the addition of one water

7 Theoretical tools for designing functional materials |

155

molecule. Initially, the two sulfonic acid groups form a complex through hydrogen bonding. Analysis of the structural changes as a function of added water molecules reveals that only one sulfonic group, the protogenic group, is affected by the inclusion of water molecules before the first proton dissociation occurs. In fact, the distance bê tween the two sulfur atoms remains relatively constant at 4.3 Å, as well as the HO–H angle ∼ 178°, as shown in Fig. 7.8. It was then argued that new bis-sulfonic acid molecular scaffolds possessing this optimal S–S distance and comparable behavior of the ̂ angle might perform as well as or better than Nafion® . This molecule is shown HO–H in Fig. 7.8: the distance between the two sulfur atoms and the angle correspond to the geometrical characteristics computed in trifluorosulfonic acids. By calculation, it is demonstrated that they exhibit the same structural behavior as the bis-sulfonic system when water molecules are added to the system. Moreover, the rigid structure that supports the two sulfonic groups can possibly help reduce the water content necessary for the first proton dissociation, and lead to a very low pKa. Ultimately, the mechanical properties of the ensuing membrane may be improved. Synthesis of the proposed molecules is currently underway. At the moment of editing this book, the molecule had not yet been synthesized. Its synthesis is clearly not straightforward. 1st proton dissociation 0

2

3 180

dS-S(A)

160 5

4.3 Å θOH-O(deg.)

6

OH-O

S-S

1

140 4 0

1 2 n/2 H2O

3

Fig. 7.8. Some architectural features of the two TfOH molecules and the new molecule are comparable.

Molecular simulation is thus a complementary tool to experiment with in order to guide the design of new molecules with optimal properties. It is the combination of simulation and experiment that yields interesting conclusions, in agreement with the principle of emergence. The next section focuses on the use of atomistic simulation to provide insight into molecular interactions.

156 | Part I Functional materials: Synthesis and applications 7.2.2 Atomistic simulation Thermal transitions of polymers such as melting points (Tm ) and glass transition temperatures (Tg ) can be difficult to measure experimentally. For example, the thermal transition of rigid rod polymers can be so high that the compound starts decomposing when the transition temperature is reached. In other cases, the transition can be very weak and hardly observable with the usual instrumental techniques, such as differential scanning calorimetry (DSC), or differential mechanical analysis (DMA). Recently, nanomaterials have become more important due to their potential industrial applications. However, the measurement of thermal transitions of dispersed nanoparticles in a continuous phase is often problematic, as the amount of material constituting the nanoparticles can be very small. Nevertheless, the knowledge of their thermal transitions is a key parameter for understanding the morphology of the nanoparticle, as well as its ability to aggregate. In this regard, atomistic simulation is well-suited for studying nanomaterials, as it probes comparable dimensions and thus provides a valuable tool for the prediction of thermal properties in functional materials. It has to be pointed out that, due to the great number of atoms required to represent a transition, quantum calculations cannot describe such phenomena. Measurement of Tm and Tg is explained in the following section.

The melting transition [32] Fundamental properties of crystals at the nanoscale are significantly different to those of bulk crystals and melting and crystallization temperatures rarely correspond to those of the bulk. Properties of nanocrystals are also strongly influenced by their shape and size. An efficient way to clarify experimental observations relies on the use of appropriate computational tools to probe the very nature of interactions at stake. The study of Tm stems from a collaboration with Professor Jérôme Claverie, author of Chapter 2 in the current book. Professor Claverie and his group are interested in functional groups containing polyethylene (PE), such as the carboxylic acid group. Recently, they demonstrated that nanoparticles of PE are thermoreversible [33]. Understanding this phenomenon is not straightforward and the use of molecular simulation becomes a reliable asset in unveiling the fundamental nature of this thermoreversibility. Simulation of the melting point of functional materials conforms to this vision. From a simulation viewpoint, the question was: how can interactions inside a PE nanoparticle affect the melting point? A protocol had thus to be carefully designed to model such a transition and consists of a simple but realistic representation of the actual material. A nanocrystal constituted of alkane chains of fixed length was used to mimic the PE nanocrystals. How to compute the melting point? It is known that it is a first order transition (heat capacity tends to infinite) with latent heat. Moreover, in the case of alkane chains, conformations change from all-trans to a Boltzmann distribution of the rotameric states. These experimental facts are then considered in order

7 Theoretical tools for designing functional materials |

157

to envision by simulation the melting transition. The pcff force field was used, since it was created specially to simulate polymers appropriately. First, calculations were performed on a crystal formed with chains 24 atoms long, 4 chains in each x and y direction, as shown in Fig. 7.9. The crystal is in the vacuum.

250

300

Tm

350

400

350

400

σe

x

% Trans

1.0 0.9 0.8 0.7

Cv (J/g-K)

15.6

l

11.7 7.8 3.9 0.0

z

E_VdW

160

y y x

0 –160

–320 250

300 Temp (K)

Fig. 7.9. Simulation of the melting point.

From the original nanocrystal, MD are run at different temperatures, each temperature leading to a trajectory. In fact, positions and velocities of each atom are recorded, making visualization of the melting phenomenon possible, as well as the computation of many properties (temperature, pressure, local dynamics, and so on). In order to observe the melting transition, three properties are reported with respect to temperature: van der Waals energy, heat capacity, and percentage of the trans-rotameric states (Fig. 7.9). It has to be mentioned that the heat capacity is computed using the fluctuation formula, (⟨E2 ⟩ − ⟨E⟩2 / kb T2 ), where E is the energy, kb is the Boltzmann constant, and T is the temperature. The thermodynamic definition (𝜕⟨E⟩/ 𝜕T)V of heat capacity cannot be used since the thermodynamic limit⁶ is not attained. As shown in Fig. 7.9, it is observed that a transition occurs at the same temperature for the three properties reported, which clearly corresponds to the melting temperature of the nanocrystal. All details on the simulation can be found in Metatla et al., as we only present the main conclusions of the study [32]. In Fig. 7.10, melting temperatures are reported with respect to the inverse of the thickness of the crystals. The lines correspond to the linear fit using the Gibbs–

6 The thermodynamic limit is attained when N/V = constant with N → ∞ and V → ∞.

158 | Part I Functional materials: Synthesis and applications 0.95 Functionalized PE experience

0.90

Tm/T°m

0.85 0.80 0.75

Alkanes experience

Alkanes simulation

0.70 0.65 0.2

0.3

0.4 1/l (nm–1)

0.5

0.6

Fig. 7.10. Gibbs–Thomson equation representation reporting experimental (dark blue line) and simulated (blue line) Tm for alkane chains, and experimental Tm for functionalized PE (dotted line).

Thomson equation: 2σe 1 Tm =1− Tom Δhm l

(7.10)

where Tom and Δhm are the melting temperature and the melting enthalpy per unit volume of bulk alkane chain respectively, σe is the interfacial tension of the crystal in the plane normal to z (Fig. 7.9), and l is the crystal thickness in nm. There is excellent agreement in the slope between experimental and simulated data. Such accuracy reveals the precision of the procedure and the description of the σe /Δhm ratio. If we compare simulated data with experimental values stemming from functionalized PE, there is a clear discrepancy in the slope. This difference is in fact of great interest and fostered this collaboration. Clearly, the dissimilarity in the slope relies on the value of the interfacial tension, σe ⋅ Δhm can be considered comparable to alkane and PE chains, while changes in the interfacial tension directly result from alterations at the surface. The occurrence of defects brought by chain loops or functionalized groups modifies σe . The presence of these defects can be definitely simulated, leading to a change in the slope (Fig. 7.10), approaching that displayed by functionalized PE. Such treatment is undoubtedly a molecular viewpoint of the surface problem. Supported by experiments, this result should lead to interesting conclusions. Further simulations are presently being carried out, and first data reveal that the presence of defects on the surface lead to a change in Tm .

7 Theoretical tools for designing functional materials |

159

The glass transition [34–41] The glass transition temperature (Tg ) has been known since ancient Egypt, as craftsmen knew how to process a glassy material by heating it above a certain temperature in order to shape it to a certain design. They may not have handled glass as glassblowers from Murano Island do nowadays, but they had identified that glass is fragile and is a very malleable material above it. More than three thousand years later, in 1995, the Nobel Prize-winning physicist, Philip W. Anderson (already encountered at the beginning of this chapter), wrote in the magazine Science: “The deepest and most interesting unsolved problem in solid state theory is probably the theory of the nature of glass and the glass transition.” It is peculiar that this unique phenomenon remains an unsolved problem. This is due to the fact that the time and length domains it covers are huge: from nanoseconds (10−9 sec) to years (108 sec), highlighting the problems of polymer aging. It is thus very difficult for an experiment or a theory to uncover this domain entirely. Can atomistic simulation possibly reveal such phenomenon? At first glance, the response is clearly negative. But let’s have a look at what exactly glass transition is. Following that, the protocol and some results will be exposed, and it will be left to the reader to decide whether atomistic simulation is capable of dealing with the glass transition of polymers. At high temperature, a polymer chain can reach many states. More particularly, the different rotameric states⁷ associated with a bond along the backbone bonds are reachable, as the chain is not restricted in its movements at all. As the temperature is decreased, crossing the energetic barrier between rotameric bonds in order for the bond to rotate occurs less often. The polymer thus reaches a state which makes it very difficult for the backbone to move. Imagine the 15-block puzzle displayed in Fig. 7.11, in which you want to move the G block. You need to move several blocks to make a space available near G. One motion thus involves several motions. This observation incorporates the concept of cooperation, in fact. This important concept in the study of glass transition corresponds to the fact that positions of neighboring entities (molecules or bonds) are strongly correlated, leading to a modulation of density extending over a certain molecular scale. The glass transition phenomenon is thus a longer-range phenomenon than the length scale associated with several molecules. In some ways, it confirms that atomistic simulation is not the best tool for dealing with this complex transition. However, Roe and Rigby [42] first showed that it is possible to simulate the glass transition of polymers by using atomistic simulated dilatometry. As in experiments, the dilatometric technique consists of reporting the specific volume i.e., the inverse density, with respect to temperature. As the system cools down, the specific volume is reported for different temperatures. The departure from a linear relationship between specific volume and temperature yields the value of Tg , which effectively corresponds to the tran-

7 A rotameric state is defined by a well in the potential energy as a bond is rotated.

160 | Part I Functional materials: Synthesis and applications

G

I

a

s

s

T

r

a

n

s

i

t

i

O

n

Fig. 7.11. A 15-block puzzle representing a polymer chain in the glassy state.

sition from the rubbery to the vitreous phases. Examples are shown in Fig. 7.12. In fact, numerous atomic simulation studies can be found in the literature. The main drawback of this method is that a great disparity is found in the values of Tg higher than, equal to or lower than the experimental values. Questions necessarily arise about the pertinence of and the interest in using all-atomistic simulation to compute Tg . One of the reasons is that atomistic MD simulations mainly inspect the region of space located near the initial portion. Moreover, the length scale associated with the glass transition should exceed the domain available with atomistic simulations; coarse-grained simulations addressing the underlying theories accountable for the occurrence of the glass transition [43]. Nevertheless, meaningful information can be extracted from such studies: the effect of side chain, tacticity, and so on. Atomistic simulation must be considered a very complementary tool, as shown in the next paragraph. Prior to any simulation, great care must be taken to select the appropriate force field. It must be chosen properly in order to specifically address the problem it is required to solve. The glass transition temperature of polymers can be greatly affected by the choice of the force field, as shown in Fig. 7.12. Simulated dilatometries of one polymer, isotactic PMMA, using two force fields, AMBER (black circles) and AMBEROPLS (gray diamonds), are displayed in Fig. 7.12. The difference in the force fields stays in computation of the partial charges in the Coulomb potential energy (equation (7.9)). In the latter force field, partial charges have been computed in order to reproduce liquid density, while in AMBER, charges have been computed using the ESPR method based on quantum calculations. AMBER-OPLS was deliberately chosen, since a glass transition is observed when it is used (black circles in Fig. 7.12). Once a glass transition temperature simulation has been completed, two questions must be raised: what is the degree of trust attached to the conclusions stemming from this analysis, and does this transition correspond to the experimentally observed glass transition temperature? One way to answer these questions is to get reproducible values of Tg and compare the stemming analysis to experimental data and theories. The focus is thus to develop a procedure based on physics and statistical physics concepts to compute Tg . The different steps are summarized in Fig. 7.13 and in the following paragraph.

7 Theoretical tools for designing functional materials |

161

Specific volume (cm3.g–1 )

1.04

1.00

0.96

0.92 0

50 100 150 200 250 300 350 400 450 Temperature (°C)

Fig. 7.12. Simulated dilatometric curves for iso-PMMA according to two first generation force fields: AMBER (⧫), and AMBER-OPLS (∙).

Appropriate exploration of the configurational space is a key issue. It corresponds to Step 1 in Fig. 7.13. The algorithm used for generating polymer chains (100) is very powerful and was built by Theodorou and Suter [44], based on the Self-Avoiding Walk protocol and the Rotational Isomeric States model [45, 46]. For each polymer chain, the radius of gyration (Rg ) is then computed (Step 2). A Maxwell–Boltzmann-like distribution is obtained. Polymer chains which exhibit Rg with the most probable values are selected. An additional criterion based on the lowest energy is then completed to finally select ten configurations (Step 3). This number of configurations is considered a reasonable compromise between accurate representation of the configuration (through Steps 1 to 3), and CPU capability. It is thus a way of circumventing the central limit theorem. A heating-cooling process is then performed to relax the system (Step 4). At this stage, Tg can be computed when the system has cooled down. However, it has been shown that this value is not reproducible. In fact, in order to effectively characterize a small portion of the configurational space representative of the entire domain, the simulated system must approach reality. The conclusion reached was that the simulated polymers in their glassy state were not in mechanical equilibrium, leading to Step 5, which consists of getting 𝜕E/ 𝜕V = 0, where the internal pressure and the external stress are made equal. Entropy and vibrational energy should be introduced, but there are negligible quantities for this kind of simulation. When this step has been completed, numerous properties such as mechanical properties and values of Tg are found to be reproducible. This procedure was applied to get Tg of organic glasses whose values are in very good agreement with experimental data. We can thus suspect that the glass transition has been accurately reproduced, and that local characteristics of this phenomenon can be disclosed. It has to be mentioned that this procedure was transposed to simulate glass transition of molecular glasses [47] and liquid crystalline phases [48] with the same reliable results. As an illustrative example, Fig. 7.14 shows values of Tg for a series of vinyl polymers computed following the procedure described in the previous paragraph with respect

162 | Part I Functional materials: Synthesis and applications

Generation of 100 1 configurations

16

Number of configurations

14 12 10 8

2

6

Computation of the Rg

4 2 0 20

40

60

80

100

120

140

160

1. Higher probability 2. Energetic criteria

3

2 selection criteria

VSp/cm3 .g–1

1.08

Heating-Cooling process 4

1.04

1.00

0.96 100

150

200

250

300

Tg

Temperature/°C –7650 1.08

–7670 –7680

Mechanical 5 Equilibrium

–7690 –7700 –7710

VSp/cm3 .g–1

Energy/kcal mol–1

–7660

1.04

1.00

–7720 –7730

0.96 27500

28000

28500

29000

29500

Volume/Å3

100

150

200

250

300

Temperature/°C

Fig. 7.13. Procedure for computing reproducible values of Tg .

to the experimental values. The linear relationship is excellent, which makes the identification of molecular events that lead to changes in Tg feasible. Thanks to the use of MD, specific local dynamics can be revealed. Two generic vectors have been selected: along C (backbone)-H, and originating from Cα (side chain), shown in Fig. 7.15. Analysis of the movement of these vectors according to temperature yields a correlation time associated with each motion at a specific temperature. This correlation time can be inserted into an Arrhenian diagram: the natural logarithm of this correlation time is displayed with respect to the inverse of the temperature. Surprisingly, a non-linear

560 540 520 500 480 460 440 420 400 380 360 340 320 300 280

CH3

n X

n

R

163

n

N

CH3

n O CH3

O

n

N

n

460

480

420

440

400

360

380

320

300

260

280

220

240

200

n

340

n O CH3

O

180

Simulated Tg/K

7 Theoretical tools for designing functional materials |

Experimental Tg/K Fig. 7.14. Simulated and experimental Tg of a series of vinyl polymers and linear fitting highlighting the correlation.

fit is observed. Arrhenian behavior, i.e., a linear relationship where the slope is related to the activation energy, is expected. The physical sense beneath the non-linear fit is related to the occurrence of cooperation along the polymer chains, as schematically shown in Fig. 7.11. An equation which is better-suited to improving the fit of the traditional Arrhenius equation is the Vogel–Fulcher–Tamman equation: τ = τo exp (

Eeff ), T − To

(7.11)

where the three fitting parameters τo , To , and Eeff correspond respectively to the correlation time for infinite temperature, Vogel temperature, and the effective activation energy. The behavior of the effective activation energy with Tg for the different vinyl polymers studied is of particular interest. Figure 7.15 shows the behavior of Eeff for the two bonds. Values of Eeff related to the C–H bond (in blue) are relatively constant. This means that the carbon backbone, to which C–H movement is related, possesses an intrinsic dynamic that does not influence Tg . Conversely, Eeff associated with the motion of the side chain (dark blue) reveals a linear relationship with Tg . In fact, the dynamics associated with this bond correspond to three motions: librational motion (not considered in the analysis), side chain rotation, and overall motion of the chain. The two latter movements cannot be separated easily. As a consequence, they are studied simultaneously, leading to a molecular analysis of the coupling between the side chain and the backbone. A linear relationship is used to fit the data. The Tg of a hypothetic chain reduced to only a carbon backbone, and for which Eeff = 0 was found to be 250 K. This is in fact close to the actual Tg of PE, which is of the order of 280 K. Thus, side

164 | Part I Functional materials: Synthesis and applications 20

X

R

Backbone/side chain coupling

Effective Ea (kcal/mol)

Backbone dynamics

18 16 14 12 10 8 250

300

350

400 450 Tg /K

500

550

Fig. 7.15. Local dynamics in simulated vinyl polymers.

chains greatly influence the backbone motions which ultimately yield differences in the value of Tg through specific coupling.

7.3 Summary Molecular simulation is becoming a very powerful tool in the lab and is not just for theoreticians. A corollary to this situation is that it has great difficulty finding its niche. It is theory for experimentalists, and experiment for theoreticians. Moreover, extreme caution should be exercised in interpreting ensuing results, as misleading interpretation can be very problematic. The examples presented in this chapter show that meticulous protocols must be established. Clearly, simulations must be used as a tool to complement experiment and theory. Consequently, these studies emphasize the great strength of molecular simulation in unveiling molecular characteristics, thus acting as a real guide in the design of new molecules with improved properties. Moreover, codes and computers are now efficient enough to bring a new perspective to those properties. Local phenomena can be scrutinized and molecular simulation acts as a microscope. Can it solve the antagonism raised by the difference between the two visions of the world, reduction and emergence? We will see in the future! In fact, increasing computer capacity and efficiency of codes, the molecular origin of a property can be definitely disclosed at different length and time scales. The multiscale approach, as highlighted in this chapter, thus becomes more and more attractive. The very purpose is to make the most of conclusions stemming from one scale by linking them to data extracted at another scale. A property can thus be comprehended along the complete spectrum of interactions. This viewpoint has become so important that Martin Karplus, Michael Levitt, and Arieh Warshel were awarded the 2013 Nobel Prize in Chemistry “for the development of multiscale models for complex chemical systems”.

7 Theoretical tools for designing functional materials |

165

Molecular simulation – Young DC. Computational chemistry: a practical guide for applying techniques to real world problems. Wiley, New York, 2001. – Jensen F. Introduction to computational chemistry. 2nd ed. John Wiley & Sons, Chichester, UK, Hoboken, USA, 2007. – Leach AR. Molecular modelling: principles and applications. 2nd ed. Prentice Hall, Harlow, England, New York, 2001. – Cramer CJ. Essentials of computational chemistry: theories and models. 2nd ed. John Wiley & Sons, Chichester, UK, Hoboken, USA, 2004. – Ramachandran KI, Deepa G, Namboori K, SpringerLink (Online service). Computational chemistry and molecular modeling principles and applications. Berlin: Springer; 2008. – Hinchliffe A. Molecular modelling for beginners. 2nd ed. Wiley, Chichester, UK, Hoboken, USA, 2008. – Höltje H-D. Molecular modeling: basic principles and applications. 3rd rev. and expanded ed. Wiley-VCH, Weinheim, 2008. – Leszczynski J. Handbook of computational chemistry. Springer reference. Springer Science+ Business Media B.V., Dordrecht, New York, 2012. – Lewars E, Computational chemistry introduction to the theory and applications of molecular and quantum mechanics. 2nd ed. Springer, Dordrecht, London, 2011. – Reviews in computational chemistry. Editors: Lipkowitz KB, Boyd DB. 26 Volumes. Wiley-VCH, New York. Multiscale simulation – Berendsen HJC. Simulating the physical world: hierarchical modeling from quantum mechanics to fluid dynamics. Cambridge University Press, Cambridge, UK, 2007. Quantum calculation – Parr RG, Yang W. Density-functional theory of atoms and molecules. Oxford University Press, New York, Oxford, 1994. – Koch W, Holthausen MC, Wiley InterScience (Online service). A chemist’s guide to density functional theory. Second ed. Wiley-VCH, Weinheim, New York, 2001. – Sholl DS, Steckel JA. Density functional theory: a practical introduction. Wiley, Hoboken, N.J., 2009. – Hehre WJ. Ab initio molecular orbital theory. Wiley, New York, 1986. Atomistic simulation – Allen MP, Tildesley DJ. Computer simulation of liquids. Oxford University Press, Oxford, 1987. – Frenkel D, Smit B. Understanding molecular simulation: from algorithms to applications. 2nd ed. Academic Press, San Diego USA, London, UK, 2002. – Kotelyanskii M, Theodorou DN. Simulation methods for polymers. Marcel Dekker, New York, 2004. – Haile JM. Molecular Dynamics Simulation. John Wiley & Sons, New York, 1992.

166 | Part I Functional materials: Synthesis and applications

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]

[18] [19] [20]

[21] [22] [23]

[24] [25] [26] [27]

Laughlin RB, Pines D. The theory of everything. Proceedings of the National Academy of Sciences of the United States of America. 97 (2000) 28–31. Anderson PW. More Is Different. Science. 177 ( 1972) 393–6. Laughlin RB. A different universe: reinventing physics from the bottom down. Basic Books, New York, 2005. Scerri E. Reduction and Emergence in Chemistry – Two Recent Approaches. Philosophy of Science. 74 (2007) 920–31. Ramachandran KI, Deepa G, Namboori K, SpringerLink (Online service). Computational chemistry and molecular modeling principles and applications. xxi, 397 p. Springer, Berlin, 2008. Leff HS, Rex AF. Maxwell’s demon : entropy, information, computing. Princeton University Press, Princeton, NJ, 1990. Allen MP, Tildesley DJ. Computer Simulation of Liquids. Clarendon Press, Oxford, 1987. Sagan C. Cosmos. 1st ed. Random House, New York, 1980. Thomson JJ. Cathode Rays. Phil Mag. 44 (1897) 293. Davisson C, Germer LH. Diffraction of Electrons by a Crystal of Nickel. Physical Review. 30 (1927) 705–40. Frisch MJ, Trucks GW, Schlegel HB, et al. Gaussian 09. Wallingford CT, Gaussian Inc, 2009. www.gaussian.com www.msg.ameslab.gov/gamess/ accelrys.com/ Berendsen HJC. Simulating the physical world: hierarchical modeling from quantum mechanics to fluid dynamics. Cambridge University Press, Cambridge, 2007. Sun H. COMPASS: An ab initio force-field optimized for condensed-phase applicationsOverview with details on alkane and benzene compounds. J Phys Chem B. 102 (1998) 7338–64. Jorgensen WL, Maxwell DS, Tirado-rives J. Development and Testing of the OPLS All-Atom Force Field on Conformational Energetics and Properties of Organic Liquids. J Am Chem Soc. 118 (1996) 11225–36. Cornell WD, Cieplak P, Bayly CI, et al. A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J Am Chem Soc. 117 (1995) 5179–97. Levine IN. Physical chemistry. 6th ed. McGraw-Hill, Boston, 2009. Warshel A. Consistent force field calculations. II. Crystal structures, sublimation energies, molecular and lattice vibrations, molecular conformations, and enthalpies of alkanes. J Chem Phys.53 (1970) 582–94. Binder K, Heermann DW, SpringerLink (Service en ligne). Monte Carlo simulation in statistical physics an introduction. 5th ed. Springer-Verlag, Heidelberg , New York, 2010. Soldera A. Energetic analysis of the two PMMA chain tacticities and PMA through molecular dynamics simulations. Polymer. 43 (2002) 4269–75. Soldera A, Monterrat E. Mid-infrared optical properties of a polymer film: comparison between classical molecular simulations: spectrometry, and ellipsometry techniques. Polymer. 43 (2002) 6027–35. Metatla N, Soldera A. Effect of the molar volume on the elastic properties of vinylic polymers: A static molecular modeling approach. Macromol Theor Simul. 20 (2011) 266–74. McQuarrie DA. Statistical mechanics. Harper and Row, New York, 1976. Soldera A, Grohens Y. Local dynamics of stereoregular PMMAs using molecular simulation. Macromolecules. 35 (2002) 722–6. Palato S, Metatla N, Soldera A. Temperature behavior of the Kohlrausch exponent for a series of vinylic polymers modelled by an all-atomistic approach. Eur Phys J E. 34 (2011) 90.

7 Theoretical tools for designing functional materials |

167

[28] Metatla N, Palato S, Soldera A. Change in Morphology of Fuel Cell Membranes under Shearing. Soft Matter. 9, (2013) 11093–7. [29] Rondeau-Gagné S, Lafleur-Lambert A, Soldera A, Morin JF. Ethynyl-bridged fullerene derivatives: Effect of the secondary group on electronic properties. New Journal of Chemistry. 35 (2011) 942–7. [30] Laflamme P, Beaudoin A, Chapaton T, Spino C, Soldera A. Simulated infrared spectra of triflic acid during proton dissociation. J Comput Chem. 33 (2012) 1190–6. [31] Laflamme P, Beaudoin A, Chapaton T, Spino C, Soldera A. Molecular modeling assisted design of new monomers utilized in fuel cell proton exchange membranes. J Membr Sci. 401–402 (2012) 56–60. [32] Metatla N, Palato S, Commarieu B, Claverie JP, Soldera A. Melting of polymer nanocrystals: a comparison between experiments and simulation. Soft Matter. 8 (2012) 347. [33] Kryuchkov VA, Daigle J-C, Skupov KM, Claverie JP, Winnik FM. Amphiphilic Polyethylenes Leading to Surfactant-Free Thermoresponsive Nanoparticles. J Am Chem Soc. 132 (2010) 15573–9. [34] Metatla N, Soldera A. Computation of densities, bulk moduli and glass transition temperatures of vinylic polymers from atomistic simulation. Mol Simul. 32 (2006) 1187-93. [35] Soldera A, Metatla N. Glass transition of polymers: Atomistic simulation versus experiments. Phys Rev E. 74 (2006) 061803. [36] Metatla N, Soldera A. The Vogel–Fulcher–Tamman equation investigated by atomistic simulation with regard to the Adam-Gibbs model. Macromolecules. 40 (2007) 9680–5. [37] Soldera A, Metatla N, Beaudoin A, Said S, Grohens Y. Heat capacities of both PMMA stereomers: Comparison between atomistic simulation and experimental data. Polymer. 51 (2010) 2106–11. [38] Metatla N, Soldera A. Effect of the molar volume on the elastic properties of vinylic polymers: A static molecular modeling approach. Macromolecular Theory and Simulations. 20 (2011) 266–74. [39] Palato S, Metatla N, Soldera A. Temperature behavior of the Kohlrausch exponent for a series of vinylic polymers modelled by an all-atomistic approach. European Physical Journal E. 34 (2011) 90. [40] Soldera A. Atomistic simulations of vinyl polymers. Mol Simul. 38 (2012) 762–71. [41] Plante A, Palato S, Lebel O, Soldera A. Functionalization of molecular glasses: effect on the glass transition temperature. Journal of Materials Chemistry C. 1 (2013) 1037. [42] Rigby D, Roe RJ. Molecular dynamics simulation of polymer liquid and glass. I. Glass transition. J Chem Phys. 87(12), (1987) 7285–92. [43] Baschnagel J, Binder K, Doruker P, et al. Bridging the gap between atomistic and coarsegrained models of polymers: Status and perspectives. Advances in Polymer Science: Viscoelasticity, Atomistic Models, Statistical Chemistry. 152 (2000) 41–156. [44] Theodorou DN, Suter UW. Detailed molecular structure of a vinyl Polymer glass. Macromolecules. 18 (1985) 1467–78. [45] Mattice WL, Suter UW. Conformational Theory of Large Molecules. The Rotational Isomeric State Model in Macromolecular Systems. John Wiley & Sons, New York, 1994. [46] Flory PJ. Statistical Mechanics of Chain Molecules. Hanser Publishers, New York, 1989. [47] Porzio F, Levert E, Vadnais R, Soldera A. New Insights into the Thermal Stability of the Smectic C Phase. Journal of Physical Chemistry B 118 (2014) 4037–43.

| Part II: Development of new materials for energy applications

S. B. Schougaard and D. Bélanger

8 Electrochemical energy storage systems 8.1 Introduction The ability to efficiently store and retrieve electrical energy is at the heart of the mobile revolution which has swept through society since the late 1980s. The development of green energies such as solar and wind, which could ultimately be used for transportation in hybrid and electric cars, began during the same period. Making this “green” transition will require an efficient, low cost energy storage system. Two of the most promising candidates to meet the high power-high energy density storage requirements are lithium-ion batteries and advanced electrochemical capacitors. Yet the current state-of-the art performance requires improvements to meet customer expectations. To this end, the active materials still need to be perfected. Materials used in energy storage systems which convert chemical energy into electrical energy must have unique properties for this specific application. A wide variety of materials such as metal, metal oxides, and even polymers have been used for this purpose over the past two centuries. One of the first batteries developed which has reached widespread commercial use is the lead-acid battery. This battery still occupies a major segment of the energy storage market today, due to its use in cars and other automotives. Other common battery technologies include alkaline and nickelcadmium cells. Materials used in current and future energy storage systems must satisfy design criteria such as low cost, stability, low environmental impact, and toxicity, and must be produced from widely available resources. This is obviously not the case for all batteries thus far developed. Safety is also a major concern due to the quest for higher energy and power density, which require more reactive materials. The aim of this chapter is to present the fundamental concepts associated with electrochemical energy storage systems in such a way that they will be accessible to a broad audience working in the area of functional materials. Firstly, the metrics used to evaluate the performance of batteries will be described. Secondly, the inner workings of electrochemical energy storage will be explained using thermodynamic, kinetic, structure, and mass transport properties of battery materials. The current state-of-the art materials used in these electrochemical energy storage systems will also be presented.

8.2 Metrics and performance evaluation The most fundamental characteristic of any energy storage system is how much energy per unit of mass can be stored (specific energy), and how fast this energy can be retrieved and/or stored (power density). Alternatively, the storage capacity and power

172 | Part II Development of new materials for energy applications density may be expressed in terms of volume, when space constraints are more relevant. Electrochemical energy storage systems may be analyzed using several different protocols, one of the most common being the constant current. Here, the current is sourced (controlled) and the potential (voltage) is measured as a function of time. Importantly, several different units and nomenclatures are used to express the controlled current. Among the more common ones are A/g, A/cm2 and C-rate, where x C represents full discharge in C/x hour, so that a current density of C/10 represents full charge in 10 hours. Batteries and electrochemical capacitors differ in their fundamental charge storage mechanism; as such, it is not surprising that their electrochemical responses are also different (Fig. 8.1). 10A/g

5A/g 1A/g

4.2

0.9

Potential (V)

Potential (V)

1.2

0.6

3.8 3.4 3.0

0.3

0.2A/g 0.05A/g

0

0 (a)

40 80 120 Capacity (mA.h/g)

160

2.6

0

0.005A/g

40 80 120 160 Capacity (mA.h/g)

(b)

Fig. 8.1. Potential vs. capacity curves. (a) electrochemical capacitor, and (b) battery. The stored energy is proportional to the area under the curve when extended to 0 V.

From Fig. 8.1 it is evident that capacity is dependent on the current density used. To put this into a meaningful metric, the power density can be defined as the product of voltage and current density. As this product varies over the charge/discharge cycle, an average is normally cited. This definition makes comparison of different devices according to their power delivery ability and their total energy density in the Ragone Plot possible (Fig. 8.2). Figure 8.2 shows that electrochemical capacitors provide high power while batteries provide high energy density. It is therefore important to understand the required specifications for a specific application before choosing a battery or an electrochemical capacitor. However, for an electric car, it is not a question of battery or electrochemical capacitor, since both serve distinct design parameters; i.e., electrochemical capacitors are ideal for storing the power generated when the car decelerates rapidly (e.g., regenerative braking), and for the high power needed when the car accelerates from a standing position. However, the battery is essential to give the car a significant range of autonomy due to its high energy density.

8 Electrochemical energy storage systems | 173

107 Capacitors Specific power (W/kg)

106 Combustion engine

105 104 103

Electrochemical capacitors

100

Batteries

10 1 0.01

0.1

100 1 10 Specific energy (Wh/kg)

Fuel cells

1000

Fig. 8.2. The Ragone Plot comparing different energy technologies.

For historical reasons, the energy storage capability of the electrochemical capacitor is often expressed as capacitance (units: Farad (F)), using the idealized properties of capacitors i.e., E = Q/C where Q is the stored charge, E is the potential, and C is the capacitance (see below). This relationship leads to W = 1/2 CE2 , so the stored energy (W) is proportional to the capacitance. Caution: Specific energy, specific power, and capacities are highly dependent on the current density. As a rule of thumb, these values should not be extrapolated. Moreover, care should be taken to identify the mass or volume on which the analysis is based (the entire device including housing, the electrode, or the active material).

8.3 Models and theory of electrochemical charge storage Electrochemical energy storage takes place by separating the connection between two electrodes into two different charge transport paths, one for the ions (electrolyte) which blocks the electrons, and one for the electrons (external circuit) which blocks the ions (Fig. 8.3). The external circuit serves as the point where the electric energy is injected or retrieved. It is this electrical energy that is converted into chemical energy in the device. The major difference between the electrochemical capacitor and the battery is the processes which take place at the electrodes. In the battery it is the chemical process of electron transfer, i.e., oxidation/reduction, while in the capacitor the energy is stored electrostatically as “charges” on the electrode surfaces.

174 | Part II Development of new materials for energy applications External circuit

Positive electrode

Li+ + e–+ TiS2 LiTiS2

Li+ + e– Li

Negative electrode

e–

Conducts Ions

Electrons

Electrolyte

Yes

No

External circuit

No

Yes

Li+ Electrolyte Fig. 8.3. Schematic of an electrochemical cell.

8.3.1 Battery operation – a Faradaic process The separation of the charge transport allows the separation of the redox reaction (cell reaction) into two Faradaic reactions (half cell), which occur simultaneously in two different locations. A Faradaic process involves a gain or loss of at least one electron. For example, the relevant reactions during charging are given below for the first lithium metal battery developed by Whittingham [1]: Half cell (positive electrode): Half cell (negative electrode):

LiTiS2 → TiS2 + Li+ + e− Li+ + e− → Li

Cell reaction:

LiTiS2 → Li + TiS2

(8.1)

The operational principle of charging relies on electrons and lithium ions released from the oxidation of LiTiS2 to TiS2 in the positive electrode being transported to the negative electrode by the external circuit (e− ) and electrolyte (Li+ ) where they are used to form Li(0). Therefore, the overall process does not generate or remove electrons or ions; they are only transported by the energy source in the external circuit to a location where they have higher potential energy. Thermodynamics tells us that the work done by the electrons in the external circuit during discharge cannot exceed the chemical energy stored in the chemical reaction of the battery. Moreover, the maximum work possible for an electron in the external circuit is its potential energy, measured as the voltage between the positive and the negative electrode when no current is flowing (E). The maximum work that can be done by the chemical reaction (at constant pressure and temperature) is its Gibbs free energy (Δ r G). This relationship leads to: Δ r G = −nFE,

(8.2)

where F is the Faraday constant 96 485 C/mol, and n is the number of electrons transferred (1 for the Li/TiS2 cell reaction). Δ r G can be determined from either thermodynamic tables, calorimetry, first principles, or empirical considerations, which are especially useful when designing new battery materials.

8 Electrochemical energy storage systems |

175

Armed with a method for approximating the potential of a battery (the y-axis of Fig. 8.1(b)), another for predicting the maximum theoretical value for the x-axis would be useful. The challenge is determining the charge transferred through the external circuit by the reaction of one gram of reactant in (1). Thus: theoretical capacity [in C/g] = nF/M,

(8.3)

where M is the molecular mass of the reactant (g/mol), F is the Faraday constant, n is the number of electrons transferred per formula unit. (Note: 1 C/g = 1/3.6 mAh/g).

8.3.2 Electrochemical capacitor operation – a non-Faradaic process The mechanism behind the electrochemical capacitor differs fundamentally from the battery, as no redox reaction occurs (e.g. no electron transfer). The classical electrochemical double layer capacitor comprises two porous carbon electrodes in the presence of an appropriate electrolyte, separated by a porous and inert separator. At the electrode/electrolyte interface, opposite charges accumulate on the electrode surface (within the porous structure) and in the electrolyte. At equilibrium, the electronic charges in the electrode are counterbalanced by ionic charges from the electrolyte. This leads to the formation of an electrochemical double layer, whose thickness is in the nanometer range. The energy of an electrochemical double layer capacitor is determined by the potential being developed and the capacitance of the electrodes, as discussed previously in the metrics section. The first description of a model for the double layer was proposed by Helmholtz, and is illustrated in Fig. 8.4(a) [2]. In this simple model, which is similar to that of a two-plate capacitor, two layers of opposite charge are present at the electrode/ electrolyte interface and are separated by a distance of molecular order. The Gouy– Chapman model considers the fact that in contrast to the electrode, for which the excess of charge is confined to the surface, the excess of charge required in the electrolyte for charge compensation is spread over a thickness which depends on the concentration of the electrolyte and is referred to as a diffuse layer (Fig. 8.4(b)). A third model, known as the Gouy–Chapman–Stern model, combines the first two models and includes a first compact layer next to the electrode surface, followed by a diffusion layer. In this model, the inner Helmholtz plane (IHP) refers to the plane that crosses the species of the first layer, and the outer Helmholtz plane (OHP) passes through the centres of solvated ions which are close to the electrode surface (Fig. 8.4(c)). The effect of specific adsorption and the predominance of water molecules (in aqueous electrolytes) for which the dipoles are oriented according to the electrode charge was later proposed in a model by Grahame, Bockris, Devananthan, and Muller as illustrated in Fig. 8.4(d) [3].

176 | Part II Development of new materials for energy applications Diffusion layer

Stern layer

Stern layer

Diffusion layer

D

ψ

Positively charged surface

ψ

Positively charged surface

Positively charged surface

Anion

Diffusion layer

Solvated cation

ψ0 Positively charged surface

ψ0

ψ0

Solvent Inner plane Outer plane (a)

(b)

(c)

Primary solvent layer

Secondary solvent layer

(d)

Fig. 8.4. Schematic representation of the: (a) Helmholtz, (b) Gouy–Chapman, (c) Gouy–Chapman– Stern, and (d) Bockris et al. models for the electrochemical double layer [2, 3].

The capacitance (C) of such an interface can be roughly estimated by analogy with a double plate capacitor using the equation: C = εr ε0 A/d,

(8.4)

where εr is the dielectric constant of the electrolyte, ε0 is the dielectric constant of vacuum, A is the area of the electrode, and d is the thickness of the double layer. From this equation, a typical double layer capacitance of 10–15 μF cm−2 can be estimated for metals in pure water by taking a value of 5 for the dielectric constant of a first layer of water molecules and the radius of a water molecule [3]. Since the capacitance of an electrode material is related to its surface area, an obvious approach to improvement has been to develop porous materials with a large surface area. The pores of these materials are classified into four categories: ultramicropores (< 1 nm), micropores (< 2 nm), mesopores (2–50 nm); and macropores (> 50 nm). However, it is now clear that parts of the surface area are not electrochemically accessible when the pore size is smaller than the diameter of the electrolyte ions. Nonetheless, recent studies have demonstrated a significant increase of capacitance for electrodes presenting micropores (< 2 nm), which are smaller than the size of hydrated ions [4]. Another important parameter of an electrochemical capacitor is the cell voltage. Figure 8.5 shows a schematic representation of an ideal electrochemical double layer capacitor, in which both electrode/electrolyte interfaces behave similarly. Initially, the two electrodes have the same potential and therefore the potential of the cell is null. When the electrochemical capacitor is charged, electrons are forced to flow through an external circuit from one electrode to the other. The electrode which accumulates electrons is termed the negative electrode, and cations from the electrolyte move to the

8 Electrochemical energy storage systems | 177

electrode surface under the influence of the electric field. Simultaneously, the second electrode, from which electrons have been removed, becomes the positive electrode, which is accompanied by displacement of anions of the electrolyte to the electrode surface. Note that the terms negative and positive are used instead of anode and cathode. This is because in contrast to the latter, which refers to the electrode where electron transfer occurs and involves a change in the oxidation state of the electrode or solution species (see above), no electron transfer occurs between the electrodes and the electrolyte in a classical electrochemical double layer. In these instances, the two electrodes are said to be blocking electrodes. Porous carbon +

–

–

Current collector

+

+– +

–

+

–

–

+

–

+

Electric charge

+–

Electric double layer capacitor

Liquid (electrolyte)

+

–

+

–

+

–

+

–

+ +

– –

+ +

– –

+

–

+

–

+

–

+

–

Liquid (electrolyte)

ψ0+ψ1 ψ0

ψ0

Discharged state ψ0–ψ1 Charged state Fig. 8.5. Schematic representation of the charges of each component (electrodes and electrolyte) of an electrochemical capacitor at the electrode/electrolyte interface (top) and the potential profile (bottom) for both its discharged and charged states.

Upon charging by application of a constant current, the double layer will become charged to a level that will depend on the applied potential (Fig. 8.6). In the case of an ideal electrochemical capacitor, the potential of each electrode varies linearly but in opposite directions, with the charge passed during the charging process of the device. The potential of each electrode can vary up to a limit, determined by the potential at which oxidation or reduction of electrolytes commences.

Potential/Reference

Voltage

178 | Part II Development of new materials for energy applications

0

+

–

Charge

Fig. 8.6. Variation of the potential of the positive (- - -) and negative (– –) electrodes as a function during constant current charge/discharge together with the evolution of the voltage (—) of the electrochemical capacitor.

8.4 Electrolytes From the operational principles of the battery and electrochemical capacitor, it is clear that the energy required for transportation of charge through the electrolyte and the potential limits outside which the electrolyte is oxidized or reduce, are key physicochemical design parameters. Transport of species in solution (mass transport) is related to three phenomena: diffusion, migration, and convection. Convection, which is forced movement of the liquid including all its components, is only of importance in flow-batteries [5]. Diffusion, which is important in batteries, is the process by which high concentration solutions flow towards low concentration at a speed proportional to the concentration difference until equilibrium has been reached. Migration is the mechanism by which an electric field moves charged species through a solution, and is crucial in both batteries and electrochemical capacitors. Importantly, only migration can carry the current required to close the circuit in Fig. 8.3. The drag force caused by moving ions through the solution has two effects. First, there is a significant resistance to charge transport. This leads to Ohm’s law behavior, where the current (I) is proportional to the potential difference between the electrodes (E), and conductivity (κ) is proportional to the concentration of free ions (C+ , C− ) and their mobility (u+ , u− ), i.e., Iionic = E ∗ κ and κ ∝ u+ C+ + u− C−

(8.5)

The energy lost to resistive heating in the electrolyte should be kept to a minimum, which suggests that C and u should be maximized. Electrolytes should therefore be highly concentrated, low viscosity solutions of free ions. This is a challenge, since as the concentration of electrolyte increases, so does the tendency for ion-pairing. The result is a viscous solution and a significant fraction of ions that are no longer free to engage in migration. The second effect of the drag force is that it is different for ions of different sizes. This leads to strong concentration differences during high current operation, which the diffusion process is slow to counteract. In particular, the lithium

8 Electrochemical energy storage systems |

179

ion in the Li-battery electrolyte typically moves at only about 65 % of the speed of the counter-ion. This leads to major problems during high power operation, since only lithium ions can take part in the charge transfer at the electrodes. The electrochemical potential range of stability of the electrolyte limits the maximum quantity of energy that can be stored per electron. For example, in aqueous electrolytes used for electrochemical capacitors, this value is set at 1.23 V by the thermodynamic standard potentials of the H2 O/O2 and H2 O/H2 redox systems. However, in practice a much lower value of about 1 V can be achieved. A typical nonaqueousbased device using acetonitrile or propylene carbonate as solvent can, however, reach close to 3 V. Even higher cell voltage could be obtained using materials with a wider potential range of stability, such as room temperature ionic liquids, which are salts that are liquid at room temperature. A list of common electrolyte systems is given in Table 8.1. Table 8.1. Various types of electrolytes and relevant properties. Electrolyte type

Example

Practical conductivity (mS/cm)

Potential window

Comment

Aqueous solutions

KCl (aq)

∼ 100

∼ 1.23 V

Cheap, nontoxic, nonflammable, limited potential window.

Nonaqueous solutions

LiPF6 in organic carbonates

∼ 10

∼ 4.3 V

Flammable, large potential window, compatible with low potential anodes.

Ionic liquids

1-ethyl3-methylimidazolium dicyanamide (EMI-DCA)

∼ 0.1–30

> 5V

Nonflammable, large potential window.

Polymer

Polyethylene oxide combined with LiPF6 .

∼ 0.1

> 4V

One of the two ions may be linked to the polymer to improve selective ion transport.

Solid

Lix POy Nz

< 0.1 at room temperature.

Highly dependent on system chosen.

Only high temperature or thin film application due to poor conductivity.

180 | Part II Development of new materials for energy applications

8.5 Electrode materials 8.5.1 Electrochemical capacitors Carbon is the material of choice for the fabrication of electrode materials of electrochemical capacitor. This is primarily because its low cost, its good electrochemical stability, high conductivity, and high specific surface area. A large variety of other types of materials such as conducting polymers and metal nitrides have been used for this application, but arguably the most promising materials are metal oxides such as ruthenium dioxide and manganese oxide. Carbon can be produced in various morphologies and shapes. Initial studies mainly dealt with high specific surface area powders and high area fibers. Activated carbons are the most widely investigated, but other forms such as templated carbons, aerogels, nanotubes, nano-onions, and graphene have also been examined for electrochemical capacitor applications [6]. Carbon materials commonly used in electrochemical capacitors are characterized by a high specific surface area which ranges from 1 000 to 2 500 m2 /g. Depending on the production method, the porous texture of the carbon materials will display a variable pore size distribution. As mentioned previously, it is now well-established that not all of the porous surface area of the carbon material is active in the charge storage process. Indeed, by taking a low value of 10 μF cm−2 for the double layer capacitance, the expected specific capacitance of 250 F/g cannot be reached for the highest surface area materials. Instead, the typical specific capacitance of standard activated carbons in organic electrolyte is about 100 F/g [6]. It is possible to attain a higher value (150 F/g) in aqueous electrolytes (e.g. aqueous KOH) due to the pseudocapacitive contribution of oxygenated surface groups such as quinone. These surface functional groups will also contribute to the wettability of carbon, thereby improving the capacitance of the electrode. Several single metal and mixed metal oxides have been investigated for their application as active electrode materials for electrochemical capacitors. One is ruthenium dioxide (RuO2 ), which is characterized by pseudocapacitive behavior that, in contrast to capacitive materials such as carbon, is Faradaic in origin. Thus, pseudocapacitive electrode materials undergo reversible redox reaction involving change of oxidation states of the metallic species [7]. These redox reactions are distributed over a large potential range, and are consequently characterized by a rectangular cyclic voltammogram shape and a linear charge/discharge profile such as the one shown by carbon electrodes (Fig. 8.7). On the other hand, most other metal oxides are characterized by Faradaic redox processes (in fact they are battery material) which usually encompass a relatively small potential window. Therefore, a symmetric electrochemical capacitor based on such material for both the positive and negative electrodes will be characterized by a relatively small cell voltage, and would be almost useless unless

8 Electrochemical energy storage systems |

181

the Faradaic metal oxide is used with a complementary electrode material in a hybrid electrochemical capacitor, as discussed below. Another pseudocapacitive material that has attracted attention in the past decade is MnO2 . It is obtained in various crystalline forms, including many tunnel and layer structures which allow cation exchange. Amorphous or poorly crystallized MnO2 was also widely investigated as an electrochemical capacitor material. These poorly crystallized compounds consist of a random arrangement of MnO6 octahedra, which leads to an intergrowth of different structures, alternating tunnels of different sizes alongside water molecules and alkaline cations (e.g. Na+ , K+ ). The charge storage processes for the MnO2 electrode can occur by two mechanisms depending on the nature of the oxide, and involves cations (C+ = H+ , Li+ ) from the electrolyte in both cases, the Mn3+ / Mn4+ redox interconversion (equation (8.6)) and the following electrochemical reactions: MnOOC → MnO2 + C+ + e− +

−

(MnO2 )surface + C + e →

(MnO−2 C+ )surface

(8.6) (8.7)

Spectroscopy investigations, typically using thin-film MnO2 electrodes, have confirmed that the Mn3+ / Mn4+ redox couple is primarily responsible for the observed pseudocapacitance in mild aqueous electrolytes, as demonstrated for poorly crystallized manganese dioxides using in-situ Mn K-Edge x-ray absorption and x-ray photoelectron spectroscopy [8].

8.5.2 Hybrid electrochemical capacitors A conventional electrochemical capacitor is usually comprised of two identical capacitive electrodes in a symmetrical configuration. On the other hand, in a hybrid electrochemical capacitor, one of the capacitive electrodes is replaced by a Faradaic electrode. Several types of hybrid electrochemical capacitors have been developed; the most common entails the positive electrode of a symmetric carbon/carbon electrochemical capacitor being replaced with an electrode characterized by a much higher charge-storage capability. These include pseudocapacitive and Faradaic electrode materials, such as conducting polymers and metal oxides such as MnO2 , NiOOH and PbO2 [8]. This type of electrochemical device, also called asymmetric capacitor, takes advantage of the best performance characteristics of electrochemical capacitors and batteries. This can be achieved by the fast charge storage process of the capacitive negative electrode and by the increase of the specific capacity of the cell due to the Faradaic charge storage mechanism of the positive electrode. In addition, by selecting two electrodes characterized by different operating ranges it is possible to extend the voltage of an aqueous hybrid electrochemical capacitor beyond the thermodynamic limit of 1.23 V, as illustrated in Figs. 8.7(b) and (c).

182 | Part II Development of new materials for energy applications 8

Negative electrode 4 (porous carbon)

–1.0 –0.5 E (V) vs NHE

0

I (mA cm–2) Positive electrode (porous carbon)

0.5

1.0

1.5

2.0

–4

–8

(a)

8

Negative electrode 4 (porous carbon)

–1.0

–0.5

0

I (mA cm–2) Positive electrode (MnO2)

0.5

1.0

1.5

2.0

E (V) vs NHE –4

(b)

–8 8

Negative electrode (porous carbon) –1.0 –0.5 E (V) vs NHE

Positive electrode (PbO2)

I (mA cm–2)

4

0

0.5

1.0

1.5

2.0

–4

(c)

–8

Fig. 8.7. Schematic representation of cyclic voltammograms for three different configurations of aqueous-based electrochemical capacitors (ECs), in which areas shaded in black and white represent the potential window of the positive and negative electrode, respectively for: (a) symmetric carbon//carbon device in 1 M H2 SO4 , (b) asymmetric activated carbon//MnO2 device in 0.5 M K2 SO4 , and (c) asymmetric activated carbon//PbO2 device in 1 M H2 SO4 . NHE, normal hydrogen electrode, I, measured current, and E, electrode potential.

8 Electrochemical energy storage systems | 183

The operating range of an electrode material in a given electrolyte can be assessed by cyclic voltammetry (Fig. 8.7). Figure 8.7(c) shows that the electrochemical potential window of an activated carbon electrode ranges from −0.4 to 0.7 V in H2 SO4 (aqueous). Therefore, it can be used as a negative electrode of a hybrid device comprising of a PbO2 positive electrode, with a range of electroactivity between 1.2 and 1.8 V, over which the interconversion of PbO2 to PbSO4 is: − PbO2 + 4H+ + SO2− 4 + 2e → PbSO4 + 2H2 O

(8.8)

Thus, Fig. 8.7(c) indicates that a carbon/PbO2 hybrid electrochemical capacitor can deliver a cell voltage of about 2.3 V, in contrast to a carbon/carbon device, which will be limited to 1.1 V at the most (Fig. 8.7(a)). In a symmetrical device, each carbon electrode operates in a limited electrochemical window of about 0.55 V, which is only 50 % of the full electrochemical potential window of carbon (Fig. 8.7(a), white or black areas). This means that the resulting capacitance (F/g) of the symmetrical device is only 25 % that of a single carbon electrode. 1/CEC = 1/C+ + 1/C−

(8.9)

The use of a positive electrode with a larger capacitance (capacity) and a complementary electrochemical window will result in full utilization of the carbon material and will increase the energy density due to the capacitive and pseudocapacitive/Faradaic behavior of each electrode. Similarly, a carbon/MnO2 hybrid electrochemical capacitor can deliver a slightly larger cell voltage than a carbon/carbon device, since the electrochemical potential window of an MnO2 electrode extends to slightly more positive value than a carbon electrode (Fig. 8.7(b)). The former is arguably the most thoroughly investigated hybrid electrochemical capacitor due to the fact that MnO2 is characterized by low cost, low toxicity, natural abundance, and environmentally friendly use in mild aqueous electrolytes.

8.5.3 Lithium battery¹ electrode materials Charge storage requires that the charge/discharge cycle can be repeated thousands of times. In batteries, this is a severe limitation for the type of chemical reactions which can be employed since high yield is required (for example, a 99.99 % yield per cycle for 1 000 cycles leads to a 10 % total capacity loss). In fact, only three types of reactions are currently being used in lithium batteries: insertion, alloying, and conversion.

1 Batteries where an intercalation electrode is used, both positive and negative electrodes, are known as lithium-ion batteries.

184 | Part II Development of new materials for energy applications 8.5.4 Negative (anode) electrode materials The simplest negative electrode is arguably metallic lithium. However, difficulties in plating lithium uniformly lead to dendrite formation in liquid electrolytes. In turn, this can lead to short-circuiting the electrodes, entailing thermal runaway and even explosions. Graphitic carbon where Li(0) atoms are nominally intercalated between graphene sheets is used to circumvent this problem (Fig. 8.8).

Electrolyte

Fig. 8.8. Lithium (large circles) intercalation into graphite. The reaction leads to a slight increase of the unit cell without destroying the layered structure.

Current Graphite collector

The limited capacity of the graphite lithium insertion reaction (339 mAh/g for LiC6 ) has led to the search for new materials for its replacement. A promising candidate is silicon, which forms an alloy that can store 4.4 Li per Si atom (theoretical capacity 4 200 mAh/g). The accompanying 400 % volume change is however a major design challenge, since it causes not only stress cracking of the silicon particle, but also in the entire composite electrode. Another important property of all low potential negative electrodes is their lack of thermodynamic stability relative to the reduction of the electrolyte. Fortunately, in organic carbonate electrolytes this reaction leads to the formation of a thin layer at the solid electrolyte interface (SEI), which allows Li-ions to pass and limits further reactions between electrode and electrolyte; see Fig. 8.9.

Electrolyte Solvent reduction

Li2CO3, alkoxides Li2O, RCO2Li Li+ Li+

Current Graphite SEI collector

Electrolyte

Fig. 8.9. The solid electrolyte interface (SEI) layer formation (top) and ionic transport (bottom) at the anode.

8 Electrochemical energy storage systems | 185

8.5.5 The positive (cathode) electrode Most practical positive electrode materials are based on redox cycling of 4th row transition metals. These are combined with highly electronegative elements like oxygen and fluorine to form solid-state structures, where elevated oxidation states of the metal are stable. Unique to the structures selected for battery electrodes is that they are electron conductors and allow insertion and extraction of lithium ions to compensate the change in metal ion oxidation state associated with the redox process. c b

b a

a b

c

a

c

(a)

(b)

(c)

Fig. 8.10. Materials with (a) 3D, (b) 2D, and (c) one 1D lithium transport path, exemplified by a spinel, layered α-NaFeO2 , and an olivine structure, respectively.

These intercalation compounds can be classified according to the dimensionality of the lithium ions transport path, i.e. 3D, 2D, and 1D (Fig. 8.10). Arguably, the most well known 3D material is the spinel LiMnTMO4 (TM: transition metal ion), which offers “5V” vs. Li+ /Li. However, problems with Mn dissolution and electrolyte stability have strongly limited its practical application. The 2D materials with α-NaFeO2 structure have not suffered a similar fate. In fact, LiCoO2 , which was used in the first Li-ion battery launched by Sony in 1993, is still used in most cellphones today. More recently Li(Ni0.8 Co0.15 Al0.05 )O2 , which has the same layered structure, has been targeted for electric car batteries due to its vastly improved safety characteristics and lower cost. Surprisingly, even a material with a 1D lithium transportation path, olivine LiFePO4 , has been commercialized. This material, once carbon coated, offers elevated storage capacity, high power, and remarkable safety characteristics. However, its lithium insertion/extraction mechanism is still a subject of intense debate. Yet it is clear that the phase boundary between the lithium-poor heterosite FePO4 phase and the lithiumrich olivine LiFePO4 phase formed during cycling plays a major role in reaction kinetics. While insertion materials currently dominate the market, some transformation

186 | Part II Development of new materials for energy applications materials have also been proposed. One notable example is nanosized FeF3 , which is believed to convert reversibly into LiF and Fe, entailing a remarkable theoretical capacity of 712 mAh/g.

8.5.6 Electrode production The common positive electrode materials are hard ceramics with limited electronic and ionic conductivity. They are therefore produced as small particles which are mixed with conducting carbon particles and polymer binders and cast as thin films (10–100 μm) onto metallic current collectors. This open composite structure allows the electrolyte to penetrate the electrode. As such, it provides improved ionic and electronic conduction compared to the active material itself. Unsurprisingly, the detailed structure and composition of the electrode has a profound effect on the power performance of the battery [9], and must consequently be optimized for the specific application.

8.6 Summary Batteries and electrochemical capacitors are currently the technologies which allow efficient electrical energy recovery. There are further great hopes both in industry and in society as a whole that batteries may become the energy vector which makes the use of solar, wind, and hydro energy in cars and trucks widely possible. Current technology, however, does not fully reach the expectations of the average consumer. Specifically, autonomy, i.e., driving distance per charge, appears to be a concern. The challenge over the coming years will therefore be to improve the energy density of the batteries or the charging times, to make them comparable to refuelling with gasoline or diesel. Chemistries which have been identified as capable of yielding increased energy density include lithium air and Li-sulphur, which pose unique challenges in their own right. The alternative is to design batteries which allow for unprecedented high charging rate, such as 70 % of a complete charge within five minutes. This rapid charge will then be used for emergency charging in the few instances where the average driver needs to travel beyond the range of a single charge within one day. Obtaining such extreme redox kinetics will require that the charge compensating ions involved in the redox process travel over very short distances inside the solid material. As such, the distinction between the hybrid electrochemical capacitor and battery becomes less clear, since both rely on redox process in a thin layer of the solid. It will therefore be fascinating to see if joint research in these fields will be able to combine the high power of the electrochemical capacitor with the high energy density of the battery in a single device.

8 Electrochemical energy storage systems |

187

Further reading Readers interested in methods used for the production of various carbon materials will find valuable information in the monographs by Kinoshita [10] and Conway [11]. Similarly, those interested in information about the inner workings of batteries may consult references [12] and [13].

References [1] [2] [3] [4] [5]

[6] [7] [8]

[9] [10] [11] [12] [13]

Whittingham, M.S. Electrical Energy Storage and Intercalation Chemistry (1976) Science, 192, 1126–1127. Bard, A.J., Faulkner, L.R. Electrochemical methods Fundamentals and applications. 2nd edition. John Wiley & Sons, New York, 2001. Bockris, J.O’M., Khan, S.U.M. Surface electrochemistry A molecular level approach, Plenum Press, New York, 1993. Simon, P., Gogotsi, Y. Materials for electrochemical capacitors (2008) Nature Materials, 7, 845–854. Shin, S.-H., Yun, S.-H., Moon, S.-H. A review of current developments in non-aqueous redox flow batteries: characterization of their membranes for design perspective. (2013) RSC Advances. 3, 9095–9116. Frackowiak, E. Carbon materials for supercapacitor application (2007) Physical Chemistry Chemical Physics, 9, 1774–1785. Conway, B.E. Transition from ’supercapacitor’ to ’battery’ behavior in electrochemical energy storage (1991) Journal of the Electrochemical Society, 138, 1539–1548. Long, J.W., Bélanger, D., Brousse, T., Sugimoto, W., Sassin, M.B., Crosnier, O. Asymmetric electrochemical capacitors-Stretching the limits of aqueous electrolytes (2011) MRS Bulletin, 36, 513–522. Yu, D.Y.W, Donoue, K., Inoue, T., Fujimoto. M., Fujitani, S. Effect of electrode parameters on LiFePO4 cathodes. Journal of the Electrochemical Society (2006), 153, A835–A9. Kinoshita, K. Carbon: Electrochemical and physiochemical properties. John Wiley & Sons, New York, 1988. Conway, B.E. Electrochemical Supercapacitors: Scientific Fundamentals and Technological Applications, Kluwer Academic/Plenum Publishers, New York, 1999. Newman, J., Thomas-Alyea, K.E., Electrochemical Systems. 3rd edition. John Wiley & Sons, New York, 2004. Reddy, T.B., Linden’s Handbook of Batteries. McGraw-Hill Companies. Inc., New York. 2011.

D. Rochefort

9 Functional ionic liquids electrolytes in lithium-ion batteries 9.1 Introduction When thinking about functional materials and their applications, advanced catalysts, light harvesting materials, photovoltaic systems, biosensors and other systems usually come to mind, rather than electrolytes or ionic liquids. The reason is that we are not in the habit of thinking about electrolytes as materials, and even less as materials which can be functionalized. The origin of the interpretation of electrolytes as spectator constituents of an electrochemical system can be traced back to our first encounters with electrochemical cells where the electrolyte, an aqueous solution of an inert salt like KCl for instance, serves the purpose of balancing charges during the more important processes taking place at the electrode. While this might be true for the study of redox behavior and the kinetics of dissolved species and other basic electrochemical systems, the electrolyte plays a crucial role in the electrochemistry of advanced systems such as energy storage devices. Charge is stored from ions forming the electrical double-layer in electrochemical capacitors and carbonates are involved in the formation of the solid electrolyte interface stabilising the anode of lithium-ion batteries. In the latter, the design of the electrolyte will affect the transport of lithium ions and the limits of the voltage of operation, thereby impacting the power and energy density values of the system. These are only a few examples showing that development of improved electrolytes must take place parallel with that of electrode material. For reasons that will be explained, ionic liquids have the potential to play an important role in the development of electrolytes which will increase the overall performance of energy storage devices. The aim of this contribution is to introduce the reader to the use of ionic liquids as electrolytes in lithium-ion batteries. It will become clear that ionic liquids are still far from being successfully widely applied in commercial batteries. This chapter will serve the purpose of stimulating interest by demonstrating the progress and potential developments which could stem from research into ionic liquid electrolytes, rather than simply providing a list of articles on the subject. The more general topic of ionic liquid-based electrolytes has already been the subject of several reviews and books [1–6]. Contributions with a focus on electrolytes for energy storage devices and batteries in particular [7–9] can be consulted as additional sources of information to this chapter, which is dedicated to the discussion of functional ionic liquids (ILs). The first section will cover general knowledge of ionic liquids which is required in order for the reader unacquainted to ILs to understand the origin of the properties that are responsible for their interest as electrolytes. Next, the use of typical ionic liquids

190 | Part II Development of new materials for energy applications as battery electrolytes will be examined to show how their intrinsic properties, mostly of high stability, can be an advantage in the development of safer and more efficient lithium-ion batteries. Finally, the last section will deal with ionic liquids as organic molecules which can be modified to bear a functional group. It will be demonstrated that the modification, by known and accessible synthetic procedures, can provide additional benefits to the use of ionic liquids as electrolytes. This represents the true potential of the use of IL, which is to allow the development of customized solventselectrolytes which can perform tasks that no other conventional solvents can achieve. Examples of ionic liquids designed to act as redox shuttles, to promote the formation of the solid electrolyte interphase, and to increase lithium ion transport will be discussed. Before going into detail about advanced functional ionic liquid materials, a quick glance at what exactly ionic liquids are is necessary.

9.1.1 Historical overview The concept of ionic liquids is not a new one, or one that has spontaneously emerged from a specific discovery. Their development stems from studies on molten salts with a high melting point and, appropriate to this text, their electrochemical reactions [10]. The most famous early example of the study of ionic liquid is attributed to Paul Walden, who reported about salts such as ethylammonium nitrate which “allow the reproduction at ordinary temperatures of the phenomena observed with fused inorganic salts at higher temperature, and permit an approach to the conditions prevailing in ordinary aqueous and non-aqueous solutions” in 1914 [11]. The field of ionic liquids that we know now, a hundred years later, has been defined by several key periods. The 1960s saw the beginning of the “chloroaluminate era”, during which alkali-aluminum chloride systems were studied for the development of low temperature molten salts. An important contribution to this era came from a program in the U.S. Air Force aimed at developing electrolytes for thermal batteries [12]. Low temperature electrolytes were obtained with chloroaluminate anions and organic cations such as alkylpyridinium and dialkylimidazolium, which are now widely used in ionic liquids. Such chloroaluminate ionic liquids are, however, reactive to water, and while this does not represent a huge problem in the field of batteries studied and produced in controlled environments, their reactivity hindered their use in many other applications. The next major period occurred in the 1990s, when air and water stable alkylimidazolium ionic liquids were demonstrated, using anions such as BF−4 , PF−6 , Br− , CN− , and later triflates, mesylates and bis(trifluoromethane)sulfonimide (TFSI) [13–15]. The discovery of water stable ionic liquids resulted in a tremendous increase in interest as depicted by the publication trend shown in Fig. 9.1. Ionic liquids are now the focus of numerous research groups and a very broad range of applications in the fields of electrochemistry [1, 2], catalysis [16–18], and analytical sciences [19, 20], to name but a few. Ionic liquids have also reached beyond the academic world and

9 Functional ionic liquids electrolytes in lithium-ion batteries |

191

2.0

Number of publications

6000 5000

1.5

4000 1.0

3000 2000

0.5 1000 0

1995

2000 2005 Year of publication

2010

0.0

% of publications related to Li-ion batteries

found applications in industry [21]. The BASIL™ process, involving the ionic liquid 1-methylimidazolium chloride which is reused for the synthesis of alkoxyphenylphosphines, is the first and certainly the most famous example of a commercial process using ionic liquids. While there are currently no ionic liquids commercialized as electrolytes in batteries or supercapacitors, there is a huge research effort worldwide to reach this goal. This is exemplified in Fig. 9.1, which shows that if the total amount of publications on ionic liquids continues increasing over the years, the fraction of those related to lithium-ion batteries will become larger, highlighting the relevance of this chapter.

Fig. 9.1. Analysis of the publication trends on the topic of ionic liquids. Only papers written in English were selected for the compilation. The squares show the percentage of total publications (gray bars) that were related to lithium-ion batteries.

9.1.2 What are ionic liquids? Ionic liquids are broadly defined as salts, usually composed of a cation of organic structure and an inorganic anion. When certain conditions relating to the chemical structure are fulfilled (discussed below), such a combination of cation and anion can melt at low temperatures, well below ambient in some cases. Provided with enough thermal energy, any salt will reach a liquid phase composed exclusively of ions. NaCl for instance melts at 800°C, and the resulting liquid phase has a viscosity of 1.03 cP and a conductivity of 3.6 S cm−1 [22]. In order to make a clear distinction between these high temperature molten salts and those able to form a liquid phase around room temperature, a restrictive but generally accepted definition of ionic liquids is “salts with a melting point below 100°C”. It should be noted that the set point of 100°C, used as a comparison to limit the liquid phase of the ubiquitous liquid, water, is purely arbitrary and is not based on any particular property of ILs. However, this definition is very

192 | Part II Development of new materials for energy applications helpful in setting a boundary between the molten salts which can be used in systems operating at moderate temperatures and those requiring high temperatures. This is especially true for electrolytes, such as those developed for Li-ion batteries and other electrochemical energy storage devices which are meant to work at temperatures ranging from −40 to +80°C. When used as electrolytes in their pure form, ionic liquids must be liquid in this temperature range, so the general definition given above will not be an issue in this chapter. At the molecular level, what distinguishes typical ionic liquids (such as those found in Fig. 9.2) from high temperature molten salts, NaCl for example, is a combination of the low level of symmetry of the cation and charge delocalization on the ions. This combination translates into systems for which ion packing in a solid state is very difficult and results in a low lattice enthalpy [23]. In addition to the low symmetry, longer flexible chains on the structure bring conformation flexibility, which further reduces the melting point and renders crystallization more difficult. Ionic liquids generally present a supercooling state in which crystallization does not occur even if below the freezing point [24, 25]. The families of ionic liquids presented in Fig. 9.2 are certainly the most studied, but this list is far from exhaustive. Considering the many different possible substituents on the structures and different combinations of cations and anions, billions of ionic liquid formulations could possibly exist.

9.1.3 Key properties as electrolytes Several key properties of ionic liquids result from the nature of their components and the strong ionic interactions existing between them. These interactions hold the constituents strongly together, making the liquid very difficult to vaporize. Having an electrolyte with a high boiling point and a very small (even immeasurable in some cases) vapour pressure adds to the stability and safety of batteries. This is also a desirable feature which allows the operation of cells at higher temperatures than with conventional organic solvents. However, due to these strong intermolecular interactions, ionic liquids are often more viscous than solvent-solute systems, resulting in a lower conductivity. Since they are entirely composed of ions, the concentration of charge carriers is very high, and ionic liquids should therefore be able to transport high currents. Their conductivity is, however, limited by their viscosity and non-ideal ionicity, meaning that the ions are screened by surrounding counter-ions and do not behave as freely moving charges. Overall, their conductivity can be high enough for electrochemical applications but remains in the same range as organic solvent-solute systems. The excellent electrochemical stability is also a property which makes for a strong case in the application of ionic liquids in batteries. Ionic liquids that begin oxidizing at potentials above those of carbonates or acetonitrile can be useful for the application of high voltage cathode materials. Imidazoliums, by far the most studied family of cations, are reduced, however, at the anode at potentials above those of lithium deposition

9 Functional ionic liquids electrolytes in lithium-ion batteries |

R2 R1

N

N

R4

R3

Alkylimidazolium

R1

R1

R1

+

+

N

R4

P

R2

R3

Tetraalkylammonium

Tetraalkylphosphonium

R1

R2 N+

N+

Dialkylpyrrolidinium

R2

N+

R2 N+

R3

R1

R2

R2

R1

193

R1 S+

O Dialkylpiperidinium

CI–, Br–, I– Halides

(R4)2N

PF6– Hexafluorophosphate O

C

S

C N

Dicyanamide (DCN)

F3 C

O

O

O

CF3

Bis(trifluoromethylsulfonyl)imide (TFSI)

F

CF3SO3– Trifluoromethanesulfonate O

N– S

S

R2

Trialkylsulfonium

CH3SO4– Methyl sulfate

O

N–

N– N

Guanidinium

Dialkylmorpholinium

BF4– Tetrafluoroborate

R3

N(R3)2

S O

O

F

Bis(fluorosulfonyl)imide (FSI)

Fig. 9.2. Cations and anions of common air and water stable organic room temperature ionic liquids.

and intercalation in some other materials. Various types of cations (phosphoniums and pyrrolidiniums), which lack an aromatic system, can be used instead of imidazoliums, however, providing better cycling stability. Other properties such as tunable water miscibility (due to the nature of the anion) and solvating power can be taken advantage of for several applications.

9.2 Ionic liquids as Li and Lithium-ion battery electrolytes Like all batteries and electrochemical cells, Li and Li-ion batteries are composed of two electrodes (the positive cathode and negative anode) separated by an electrolyte. Operation of Li-ion batteries is based on the reversible intercalation of Li-ions in oxides (cathode) and graphite (anode) [26–28]. A Li battery is similar in construction, but metallic lithium is used at the anode instead of graphite or another intercalating material. While Li-ion batteries are secondary (i.e., rechargeable) batteries, lithium batteries are primary devices which cannot usually be recharged, due to the formation of dendrites at the anode surface following the reduction of Li-ions. Controlling

194 | Part II Development of new materials for energy applications Li deposition during recharging is an important issue due to the very high capacity of Li(0) anodes. Ionic liquids have shown promising results in that field, but this topic will not be discussed further here [29, 30]. For both types of battery, the electrolyte has the important role of ionic conduction for the transportation of lithium ions between the electrodes and balancing the charge passing through the external circuit as current. The electrolyte should be sufficiently fluid to allow rapid ion diffusion, should be good at solvating ions (high dielectric constant), and should be electrochemically and thermally stable. The idea behind the use of ionic liquids as battery (and other electrochemical energy storage devices) electrolyte is to take advantage of the intrinsic properties of ILs which increase the thermal and electrochemical stability of the device. Replacing traditional volatile and flammable solvents with ionic liquids would provide safer batteries, decreasing the risk of self-combustion and explosion in the event of abuse or an accident of the system. The increased electrochemical stability would also allow the use of high potential cathodes, thereby increasing the operating voltage and the energy of the battery. On the downside, the significant disadvantages of ionic liquids, mostly their high viscosity and low transport, will decrease the overall cell performance, explaining in part why ionic liquid electrolytes have not yet made it to commercial batteries, cost being another issue. The most studied types of anode and cathode material for lithium and lithium-ion batteries have been tested in cells with ionic liquid electrolytes. An extensive list can be found in a review paper by A. Lewandowski [8]. The ionic liquids which have been the most studied for batteries are based on the alkylimidazolium and pyrrolidinium families. Ionic liquids based on imidazolium have the lowest viscosity, which is highly important to favor charge transport by the Li-ions. Pyrrolidiniums are more stable towards the reduction reaction than imidazoliums, which usually decompose at the Li(0) and graphite anodes. The TFSI anion is the most reported due to the low viscosity of the ionic liquids employing it, but tetrafluoroborate and hexafluorophosphate have been studied extensively due to their wide application in conventional battery electrolytes. There are currently no commercial lithium batteries based on ionic liquid electrolytes, but research is progressing and studies on lab-scale stacked prototypes are showing promise and helping to identify current technological challenges [31, 32]. In addition to the thermal stability they confer to electrochemical systems, ionic liquids might very well make their place in batteries by achieving tasks which no other electrolyte based on solvent-solute systems can do.

9.3 Functional ionic liquid electrolytes The preceding section demonstrated the use of typical ionic liquids as Li-ion battery electrolytes, taking advantage of their intrinsic thermal and electrochemical stability to improve related properties in the battery. In this section, ionic liquids modified with a functional group will be evaluated in the performance of certain tasks essential to

9 Functional ionic liquids electrolytes in lithium-ion batteries |

195

efficient battery operation. Such functionalization of ionic liquids is usually done on organic cations (imidazoliums, phosphoniums or ammoniums) and uses well-known chemistries which are generally accessible to research groups not specialized in organic molecule synthesis. The application of functional ionic liquids for generation of the solid electrolyte interphase (SEI), to improve Li-ion transport and to prevent over-oxidation of cathodes will be discussed.

9.3.1 Overview of functional ionic liquids In a 2001 publication, R. Rogers (University of Alabama, U.S.A.) et al. demonstrated the first example of an ionic liquid with a structure that was developed for a specific purpose [33]. These “task-specific ionic liquids” or TSIL, were obtained by modifying imidazoliums with a functional group based on the chemical structure of extractants used to extract heavy metal ions from non-aqueous solvents (Fig. 9.3). By combining functionalized imidazoliums with the hydrophobic PF−6 anion, the ionic liquids were non-miscible with water, allowing the extraction of Hg2+ and Cd2+ from the ionic liquid from an aqueous phase. The distribution ratios obtained with various TSIL based on the structures shown in Fig. 9.2 were above 300 for both M2+ ions with pure TSIL. In comparison, the distribution ratio for both ions with the unmodified 1-butyl-3-methylimidazolium: PF6 extracting phase were below unity, showing the role of the functional group. This is a strong demonstration of how a properly designed ionic liquid can perform the task of a solute-solvent system without requiring dissolved additives. Several advantages are associated with this concept of adding a functional group directly to the molecular structure of one of the constituents of a liquid phase. The common benefits of ionic liquids such as low volatility, intrinsic conductivity, and thermal stability are retained in TSIL and can facilitate the recyclability of the liquid phase. These advantages motivated the development of numerous functionalized ionic liquids over the years. S R

R N

N

+

PF6–

R

N

N

H

H S

N PF6–

+

N

R

Fig. 9.3. Functional (task-specific) ionic liquids based on alkylimidazoliums modified with thiourea (top) and thioether (bottom) for the extraction of heavy metal ions. R represents a C3−6 alkyl chain.

196 | Part II Development of new materials for energy applications Currently, TSIL are being developed for CO2 capture [34, 35], catalysis [36–39], and extraction or solubilization of insoluble materials [40, 41]. There are only a few examples of functional ionic liquids being designed as electrolytes for energy storage and conversion systems. Apart from specific examples of lithium and lithium-ion batteries (discussed below), dye-sensitized solar cells [42], and lithium-air batteries [43], the majority of papers on the functionalization of ionic liquids used as electrolytes are based on ether modification to decrease viscosity and therefore address one of the major limitations of ionic liquids [44]. In fact, most of the cations presented in Fig. 9.2 were functionalized with an ether group on one of their alkyl chains and their electrochemical behaviour was studied later. While modifying ionic liquids to reduce their viscosity is highly important for their application as electrolytes, this chapter focuses on ionic liquids functionalized with a reactive group designed to be involved in some of the electrochemical reactions taking place during battery operation. The selected examples presented below provide an overview of the possibilities offered by the use of functional ionic liquid electrolytes in lithium and lithium-ion batteries.

9.3.2 Solid electrolyte interphase During the first cycle of operation of Li-ion batteries, a passivation layer is formed at the anode. This layer, called solid electrolyte interphase (SEI), is generated by the reduction of electrolyte (and additives if present), resulting in a complex mixture of inorganic and organic materials with a thickness in the nanometer range [45]. The SEI passivates the anode which prevents further electrolyte decomposition, but allows Li-ions to permeate and ensures good cyclability with sufficient cell performance. Common solvents based on carbonates used in Li-ion battery electrolytes (ethylene, diethyl, dimethyl, propylene carbonates) will typically decompose at the anode, generating Li2 CO3 , polycarbonates, and several other species [45]. Additives like vinylene carbonate (VC) are commonly added to the electrolyte to accelerate SEI formation and prevent excessive generation of gases from the reduction of the carbonate solvents [46]. As it is usually done with electrolytes based on conventional solvents, VC and other carbonates can be dissolved in ionic liquids (5–10 wt %) to accelerate SEI formation [47]. However, some issues related to the reduction of the ionic liquid itself (imidazolium) and limited Li+ intercalation might occur [8]. In the case of lithium batteries, in which the anode is metallic lithium, and where Li deposition-dissolution processes are involved rather than intercalation (as in graphite), the irreversible decomposition of 1,3-substituted imidazolium cations essentially blocks the anode surface and decreases cell output. An approach taken by M. Egashira (Yamaguchi University, Japan) to better control the SEI in Li batteries using ionic liquid electrolytes is to modify imidazolium salts with cyano groups [48–50]. Examples of such functional ionic liquids are presented in Fig. 9.4. The idea that motivated the design of these ionic liquids was to deliberately incorporate a functionality of the cation that, once decomposed, will

9 Functional ionic liquids electrolytes in lithium-ion batteries |

197

enter the composition of the SEI to provide a stable film compatible with cell operation. The chemistry of the SEI is controlled by the solvent itself rather than via the use of additives. The results obtained showed that the cyano group on the ionic liquid accelerated the imidazolium ring opening reaction, leading to SEI formation [48]. The SEI layer formed from CN-modified imidazolium on a stainless steel anode provided a reversible lithium deposition and dissolution. There was no passivation on electrodeposited Li with an electrolyte based on unmodified 1-ethyl-3-methylimidazolium (EMIM) cation electrolyte. The capacity of a LiCoO2 cathode obtained in an electrolyte composed of a mixture of the cyano-substituted quaternary ammonium TFSI (Fig. 9.4) and EMIM TFSI was measured at 110 mAh/g, slightly lower than could be obtained in conventional electrolytes [50]. Functional ionic liquids based on guanidinium cations bearing ether groups have also been studied in lithium batteries to improve Li deposition [44]. Interestingly, the authors reported the plating of Li on a Ni anode even if the cathodic potential limit of their ionic liquids was higher than that of Li+ reduction. This was attributed to the formation of the SEI, which was obtained by decomposition of the functional ionic liquid (no SEI-forming additives were added). Their electrolyte also provided good passivation in a Li/LiFePO4 cell, but the capacity obtained (110– 120 mAh/g) was lower than for conventional electrolytes, most likely due to the higher viscosity of IL-based electrolytes.

N TFSI

+

N

CN

–

TFSI–

N+

CN

Fig. 9.4. Imidazolium and ammonium ionic liquids functionalized by cyano groups used to improve SEI formation in Li batteries.

9.3.3 Transport of lithium ions One of the major limitations arising from the use of ionic liquids as electrolytes for Li-ion batteries is the low transport numbers for Li+ . For a battery to work at high power rates during charging and discharging, the amount of ionic charge by the Li+ ions moving across the electrolyte must be as high as possible. The fraction of charge transported by the lithium ions, represented by the transport number, is necessarily lower in ionic liquids than in typical electrolytes due to the absence of solvent in the former. This is clearly demonstrated by comparison of the equations for the transport

198 | Part II Development of new materials for energy applications number of Li+ (tLi+ ) for the two systems: Ionic liquid C+ A− + Li+ A− salt:

tLi+ =

uLi+ uLi+ + uC+ + 2uA−

Li+ A− salt in solvent:

tLi+ =

uLi+ uLi+ + uA−

In these equations, ui represents the mobility of the ion i. In other words, there is a high concentration of cations other than Li+ in ionic liquids that participates in charge transport, but these other ions will not intercalate and therefore won’t contribute to energy storage. To address this issue, neutral lithium ion-complexing additives such as glymes (oligoethers of CH3 (CH2 CH2 O)n CH3 composition) have been added to ionic liquid electrolytes to reduce the interactions with the anion [51]. In this case, the ionicity of the ionic liquid – LiTFSI electrolyte was found to increase with the addition of glyme. M. Watanabe (Yokohama University, Japan) demonstrated that by selecting the appropriate anion, mixtures of Li salts and glymes formed a liquid phase on their own with ionic liquid properties [52, 53]. Fig. 9.5 shows the structure of such an ionic liquid in which the cation is composed of a Li+ solvated by a molecule of tetraglyme. This family of ionic liquids are hence called solvate ionic liquids. Because of such solvated complexes, the Li ions interact less with the TFSI anion, behave like an ion in a neutral solvent and provide a high transport number. In the Li(tetraglyme)-TFSI, the tLi+ was as high as 0.51 [52], representing a significant improvement to that of mixtures of imidazolium ionic liquid with Li salts, which are found below 0.1 [54, 55].

O S F3C

O

– N

O S O

O

O Li+

CF3 O

O

Fig. 9.5. Structure of a Li(glyme) ionic liquid based on tetraglyme and the TFSI anion.

Lithium batteries employing LiFePO4 and LiNi1/3 Mn1/3 Co1/3 O2 as low (3V) and high (4 V) voltage cathodes have been studied with the pure Li-(CH3 O(CH2 CH2 )3 CH3 )-TFSI ionic liquid as the electrolyte [56]. The cell capacities as a function of cycle number are shown in Fig. 9.6. Good stability was obtained for the LiFePO4 cells for 600 cycles, showing stable electrochemical reactions at both positive and negative electrodes in the ionic liquid. Decomposition of the electrolyte is likely to occur at higher potentials, which could explain the important capacity loss in the 4 V cells (Fig. 9.6(b)).

9.3.4 Electroactive ionic liquids as redox shuttles Redox shuttles are electroactive molecules added to the Li-ion battery electrolytes to prevent oxidation of the cathode over the level required for charging (over-oxidation)

4.0

4.2

3.7

3.9

3.4 3.1 2.8 2.5 0

(a)

Cell Voltage/V

Cell Voltage/V

9 Functional ionic liquids electrolytes in lithium-ion batteries | 199

1st 50th 100th 200th

300th 400th 500th 600th

50

100

3.6 3.3 1st 10th 50th 100th

3.0 2.7

150

200

Capacity/mAhg–1

0 (b)

25

200th 300th 400th

50

75 100 125 150 175

Capacity/mAhg–1

Fig. 9.6. Charge-discharge profiles of (a) Li/LiFePO4 , and (b) Li/LiNi1/3 Mn1/3 Co1/3 O2 cells using Li(glyme)-TFSI ionic liquid electrolyte (C/8 rate at 30°C). Reproduced by permission from [56].

[57–59]. The operating principle is shown in Fig. 9.7. The over-oxidation of cathode material leads not only to the irreversible degradation of the electrode, but will also generate radicals, oxygen, and other reactive species which can cause a rapid thermal runaway leading to explosion of the battery. Under normal operation, the redox shuttle has no function and should not interfere with the processes involved in charging or discharging the battery. When current is passed through a fully charged (oxidized) cathode, the redox shuttle (R-S[R]) will however be oxidized (to R-S[O]), thereby preventing over-oxidation. The now oxidized shuttle will diffuse (D[O]) to the anode to be reduced to complete the cycle and will become available to diffuse back to the cathode (D[R] in Fig. 9.7). The motivation behind the development of functional ionic liquids which can be used as redox shuttle is two-fold. Firstly, there is the very appealing idea of combining the stability inherent to ionic liquids with the cathode protection offered by the redox shuttle principle to develop a bi-functional electrolyte for safer batteries. Secondly, the modification of the ionic liquid could address the general solubility issues of electroactive organic molecules in solvents and ionic liquids, leading to much higher concentrations of redox moieties in the electrolyte. This higher concentration could provide better protection, especially when high currents are applied to charge the battery in a shorter time. The following equation shows the direct relationship between the maximum current (IMAX ) which can be carried out between two electrodes separated by a distance (L) from the redox shuttle, and its concentration (C), in which n is the number of charge per shuttle, F is the Faraday constant, A is the surface area of the electrode, and D is the diffusion coefficient of the shuttle [60]: IMAX =

nFADC L

(9.1)

If a pure electroactive ionic liquid phase can be used (that is, if its viscosity is not too high) the concentration of redox moiety in the liquid could exceed 2 M, which could never be obtained by simply dissolving a redox molecule in an ionic liquid [61].

200 | Part II Development of new materials for energy applications

A

e

e

e

R–S [O]

R–S [O] D[O]

e

R–S [R]

R–S [R] D[R]

Anode

Cathode

Fig. 9.7. Operating principle of a redox shuttle dissolved in a Li-ion battery electrolyte. R–S[O] and R–S[R] are the redox shuttle in its oxidized and reduced states, respectively, D shows the diffusion of the shuttle in the two states between the electrodes. Reproduced by permission from [65].

Link

Side Chain O

N Fe

CF3

N n

O O

O S

S O

+ N

N

S

–N

+

O

F3C CF3

O

– N

O S O

CF3

O

Fig. 9.8. Two types of electroactive ionic liquids studied as redox shuttle for Li-ion batteries, obtained by the modification of imidazolium with ferrocene (left) and 2,5-di-tert-butylmethoxybenzyne (right).

The first demonstration of the use of electroactive ionic liquids as redox shuttle was recently reported by our group [62]. In this report, an imidazolium-based ionic liquid was functionalized with ferrocene as the redox moiety (Fig. 9.8). The 1-ferrocenylmethyl-3-methylimidazolium bis(trifluoromethanesulfonyl) amide (Fc-MIm TFSI; Fig. 9.8, left with n = 3) ionic liquid was dissolved in ethylene carbonate–diethyl carbonate solvent (EC-DEC + 1.5 M LiTFSI) and incorporated into a coin-type cell with a Li4 Ti5 O12 (LTO) cathode and a Li foil anode. LTO has a very low charging potential (1.6 V vs. Li/Li+ ) and while not usable as commercial battery material, it was used to demonstrate the concept. Figure 9.9 shows the specific capacity curves for the charge/discharge cycles of the test coin-type cell, starting with a full charge, followed by a 100 % overcharge for a Li/Li4 Ti5 O12 coin cell at a C/10 rate. In this figure, the unmodified electrolyte (i.e., without electroactive ionic liquid)

9 Functional ionic liquids electrolytes in lithium-ion batteries |

201

is the solid line and the electrolyte containing 10 % of Fc-MIm TFSI is represented by the broken lines. The overcharging situation appears very clearly for the cell without redox shuttle added where the voltage increases sharply to a 4 V cut-off after the charging plateau at 1.6 V for Li4 Ti5 O12 material. Adding the redox ionic liquid TO the electrolyte prevents reaching the cut-off voltage of 4 V. A plateau is rather observed at ∼ 3.36 V, corresponding to the onset potential for oxidation of the ferrocene moiety on the ionic liquid and demonstrating the protection against over-charging by the shuttle mechanism. No Fc-MIm TFSI, all cycles 10% Fc-MIm TFSI, 1st cycles 10% Fc-MIm TFSI, 2nd cycles 10% Fc-MIm TFSI, 3rd cycles

4.0

Cell voltage (V)

3.5 3.0 2.5 2.0 1.5 1.0 –50

0

50

100 150 200 250 300 350 400 Specific capacity (mAh/g)

Fig. 9.9. Capacity curves (C/10) for Li/Li4 Ti5 O12 cells using EC/DEC (1.5 M LiTFSI) electrolyte either pure (solid line), or modified with 10 % Fc-MImTFSI (broken lines). The parameters were set to a full charge followed by a 100 % overcharge and a cut-off at 4 V. Reproduced by permission from [62].

The use of Fc-MIm TFSI presents two important limitations. First, the electroactive ionic liquid is very viscous due to a large cation and additional intermolecular interactions between the ferrocene and imidazolium aromatic systems [63]. The high viscosity prevented the use of an ionic liquid as an undiluted phase, but its dilution at 10 vol % in an ethylene carbonate–diethyl carbonate solvent (EC-DEC + 1.5 M LiTFSI) provided an electrolyte with acceptable viscosity (7.1 cP) and conductivity (5.5 mS/cm) for a use in Li-ion cells [62]. The second challenge is related to the low oxidation potential of ferrocene. The onset for the oxidation of Fc-MIm TFSI in the carbonate electrolyte (at 10 vol %) is 3.4 V vs. Li/Li+ , which is very close to the charging potential of LiFePO4 , a common Li-ion battery cathode material. In order to apply the electroactive ionic liquid shuttle principle to this cathode material, a second series of ionic liquids was designed, based on the successful commercial shuttle 2,5-di-tert-1,4-dimethoxybenzene (DDB) [64]. These ionic liquids, dissolved in the EC-DEC + 1.5 M LiTFSI solution; start being oxidized at potential values between 3.81 and 3.94 V vs. Li/Li+ due to the higher

202 | Part II Development of new materials for energy applications redox potential of the DDB moiety (Fig. 9.8) [64]. Preliminary results in coin-type cells using LiFePO4 cathodes demonstrated cathode protection capabilities against overcharging (Fig. 9.10). Notably, the concentration in the redox shuttle was increased up to 1 M due to the high solubility of imidazolium TFSI ionic liquid in carbonates. In comparison, the poor solubility of DDB limits its working concentration in batteries below 0.1 M.

Potential\V vs Li/Li+

4.4

4

3.6

3.2

2.8 0

50

100

150

200

250

300

350

Time\h Fig. 9.10. Voltage profile of a Li/LiFePO4 cell containing 1 M of the 2,5-di-tert-1,4-dimethoxybenzenebased ionic liquid in 0.7 M LiTFSI EC/DEC (1 : 2 v/v).

9.3.5 Perspectives Ionic liquids with various structures and properties have been applied and are still being studied to replace conventional electrolytes in energy storage devices. Advances in understanding of how the structure of ionic liquid components affects its properties made several of these discoveries possible. The application of functional ionic liquids is likely to follow a similar pattern. The examples provided in this text demonstrate the potential of task-specific ionic liquids to address specific issues of energy storage devices although their application usually decreases the device performance. The challenge is to determine if (and how) some of the most important viscosity, conductivity, and melting point limitations caused by the increase in ion size when adding a large and complex (e.g., aromatic) functional group can be overcome. Therefore, more work remains to be done to better understand the structure and the interactions taking place between molecules in ionic liquids bearing a functional group.

9 Functional ionic liquids electrolytes in lithium-ion batteries |

203

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13]

[14] [15] [16] [17] [18] [19] [20] [21] [22] [23]

[24]

Armand M, Endres F, MacFarlane DR, Ohno H, Scrosati B. Ionic-liquid materials for the electrochemical challenges of the future. Nat Mater, 8 (2009) 621–9. Galinski M, Lewandowski A, Stepniak I. Ionic liquids as electrolytes. Electrochim Acta, 51 (2006) 5567–80. Macfarlane DR, Forsyth M, Howlett PC, et al. Ionic liquids in electrochemical devices and processes: managing interfacial Electrochemistry. Acc Chem Res, 40 (2007)1165–73. Baldelli S. Surface structure at the ionic liquid-electrified metal interface. Acc Chem Res, 41 (2008) 421–31. Fernicola A, Scrosati B, Ohno H. Potentialities of ionic liquids as new electrolyte media in advanced electrochemical devices. Ionics, 12 (2006) 95–102. Ohno H, ed. Electrochemical Aspects of Ionic Liquids. 2nd ed. Hoboken, NJ: John Wiley & Sons; 2011. Garcia B, Lavallee S, Perron G, Michot C, Armand M. Room temperature molten salts as lithium battery electrolyte. Electrochim Acta, 49 (2004) 4583–8. Lewandowski A, Swiderska-Mocek A. Ionic liquids as electrolytes for Li-ion batteries: An overview of electrochemical studies. J Power Sources, 194 (2009) 601–9. Wishart JF. Energy applications of ionic liquids. Energy Environ Sci, 2 (2009) 956–61. Angell CA, Ansari Y, Zhao ZF. Ionic Liquids: Past, present and future. Faraday Discussions, 154 (2012) 9–27. Walden P. Molecular weights and electrical conductivity of several fused salts. Bulletin de l’Académie Impériale des Sciences de St-Petersbourg 1914, 405–22. Wilkes JS. A short history of ionic liquids - from molten salts to neoteric solvents. Green Chemistry, 4 (2002) 73–80. Fuller J, Carlin RT, Delong HC, Haworth D. Structure of 1-ethyl-3-methylimidazolium hexafluorophosphate: model for room temperature molten salts . Journal of the Chemical SocietyChemical Communications 1994, 299–300. Wilkes JS, Zaworotko MJ. Air and water stable 1-ethyl-3-methylimidazolium based ionic liquids. Journal of the Chemical Society-Chemical Communications 1992, 965–7. Cooper EI, O’Sullivan EJ. New, stable, ambient-temperature molten salts. Proceedings of the 8th International Symposium on Molten Salts, 92-16, pp 386–96, 1992. Parvulescu VI, Hardacre C. Catalysis in ionic liquids. Chem Rev, 107 (2007) 2615–65. Sheldon R. Catalytic reactions in ionic liquids. Chem Comm 2001, 2399–407. Welton T. Room-temperature ionic liquids. Solvents for synthesis and catalysis. Chem Rev, 99 (1999) 2071–83. Ho TD, Zhang C, Hantao LW, Anderson JL. Ionic Liquids in Analytical Chemistry: Fundamentals, Advances, and Perspectives. Anal Chem, 86 (2013) 262–85. Koel M, ed. Ionic Liquids in Chemical Analysis. Boca Raton, FL: CRC Press; 2012. Plechkova NV, Seddon KR. Applications of ionic liquids in the chemical industry. Chem Soc Rev, 37 (2008) 123–50. Janz GJ. Molten salts data as reference standards for density, surface tension, viscosity and electrical conductance - KNO3 and NaCl. J Phys Chem Ref Data, 9 (1980) 791–829. Krossing I, Slattery JM, Daguenet C, Dyson PJ, Oleinikova A, Weingartner H. Why are ionic liquids liquid? A simple explanation based on lattice and solvation energies. J Am Chem Soc, 128 (2006) 13427–34. Faria LFO, Matos JR, Ribeiro MCC. Thermal Analysis and Raman Spectra of Different Phases of the Ionic Liquid Butyltrimethylammonium Bis(trifluoromethylsulfonyl)imide. J Phys Chem B, 116 (2012) 9238–45.

204 | Part II Development of new materials for energy applications [25] Ngo HL, LeCompte K, Hargens L, McEwen AB. Thermal properties of imidazolium ionic liquids. Thermochimica Acta, 357 (2000) 97–102. [26] Tarascon JM, Armand M. Issues and challenges facing rechargeable lithium batteries. Nature, 414 (2001) 359–67. [27] Goodenough JB, Park K-S. The Li-Ion Rechargeable Battery: A Perspective. J Am Chem Soc, 135 (2013) 1167–76. [28] Goodenough JB, Kim Y. Challenges for Rechargeable Li Batteries. Chem Mater, 22 (2010) 587–603. [29] Lane GH, Best AS, MacFarlane DR, Hollenkamp AF, Forsyth M. An Azo-Spiro Mixed Ionic Liquid Electrolyte for Lithium Metal-LiFePO4 Batteries. J Electrochem Soc, 157 (2010) A876–A84. [30] Schweikert N, Hofmann A, Schulz M, et al. Suppressed lithium dendrite growth in lithium batteries using ionic liquid electrolytes: Investigation by electrochemical impedance spectroscopy, scanning electron microscopy and in situ 7Li nuclear magnetic resonance spectroscopy. J Power Sources, 228 (2013) 237–43. [31] Balducci A, Jeong SS, Kim GT, et al. Development of safe, green and high performance ionic liquids-based batteries (ILLIBATT project). J Power Sources, 196 (2011) 9719–30. [32] Kim GT, Jeong SS, Xue MZ, et al. Development of ionic liquid-based lithium battery prototypes. J Power Sources, 199 (2012) 239–46. [33] Visser AE, Swatloski RP, Reichert WM, et al. Task-specific ionic liquids for the extraction of metal ions from aqueous solutions. Chem Comm 2001, 135–6. [34] Wang C-M, Mahurin SM, Luo H-M, Baker GA, Li H-R, Dai S. Reversible and robust CO2 capture by equimolar task-specific ionic liquid-superbase mixtures. Green Chem, 12 (2010) 870–4. [35] Zhang Z, Xie Y, Li W, et al. Hydrogenation of carbon dioxide is promoted by a task-specific ionic liquid. Angew Chem, Int Ed, 47 (2008) 1127–9. [36] Sun J, Cheng W, Fan W, Wang Y, Meng Z, Zhang S. Reusable and efficient polymer-supported task-specific ionic liquid catalyst for cycloaddition of epoxide with CO2. Catal Today, 148 (2009) 361–7. [37] Wang L, Li H, Li P. Task-specific ionic liquid as base, ligand and reaction medium for the palladium-catalyzed Heck reaction. Tetrahedron, 65 (2008) 364–8. [38] Xu J-M, Liu B-K, Wu W-B, Qian C, Wu Q, Lin X-F. Basic Ionic Liquid as Catalysis and Reaction Medium: A Novel and Green Protocol for the Markovnikov Addition of N-Heterocycles to Vinyl Esters, Using a Task-Specific Ionic Liquid, [bmIm]OH. J Org Chem, 71 (2006) 3991–3. [39] Kamal A, Chouhan G. A task-specific ionic liquid [bmim]SCN for the conversion of alkyl halides to alkyl thiocyanates at room temperature. Tetrahedron Lett, 46 (2005) 1489–91. [40] Nockemann P, Thijs B, Pittois S, et al. Task-Specific Ionic Liquid for Solubilizing Metal Oxides. J Phys Chem B, 110 (2006) 20978–92. [41] Ohno H, Fukaya Y. Task specific ionic liquids for cellulose technology. Chem Lett, 38 (2009) 2–7. [42] Miao Q, Zhang S, Xu H, Zhang P, Li H. A novel ionic liquid-metal complex electrolyte for a remarkable increase in the efficiency of dye-sensitized solar cells. Chem Comm, 49 (2013) 6980–2. [43] Nakamoto H, Suzuki Y, Shiotsuki T, et al. Ether-functionalized ionic liquid electrolytes for lithium-air batteries. J Power Sources, 243 (2013) 19–23. [44] Jin YD, Fang SH, Yang L, Hirano SI, Tachibana K. Functionalized ionic liquids based on guanidinium cations with two ether groups as new electrolytes for lithium battery. J Power Sources, 196 (2011) 10658–66. [45] Verma P, Maire P, Novak P. A review of the features and analyses of the solid electrolyte interphase in Li-ion batteries. Electrochim Acta, 55 (2010) 6332–41.

9 Functional ionic liquids electrolytes in lithium-ion batteries |

205

[46] Ota H, Sakata Y, Inoue A, Yamaguchi S. Analysis of Vinylene Carbonate Derived SEI Layers on Graphite Anode. J Electrochem Soc, 151 (2004) A1659–A69. [47] Egashira M, Kiyabu T, Watanabe I, Okada S, Yamaki J. The effect of additives in room temperature molten salt - based lithium battery electrolytes. Electrochemistry, 71 (2003) 1114–6. [48] Zhao L, Yamaki J-i, Egashira M. Analysis of SEI formed with cyano-containing imidazoliumbased ionic liquid electrolyte in lithium secondary batteries. J Power Sources, 174 (2007) 352–8. [49] Egashira M, Todo H, Yoshimoto N, Morita M, Yamaki J. Functionalized imidazolium ionic liquids as electrolyte components of lithium batteries. J Power Sources, 174 (2007) 560–4. [50] Egashira M, Tanaka-Nakagawa M, Watanabe I, Okada S, Yamaki JI. Charge-discharge and high temperature reaction of LiCoO2 in ionic liquid electrolytes based on cyano-substituted quaternary ammonium cation. J Power Sources, 160 (2006) 1387–90. [51] Bayley PM, Lane GH, Lyons LJ, MacFarlane DR, Forsyth M. Undoing Lithium Ion Association in Ionic Liquids through the Complexation by Oligoethers†. J Phys Chem C, 114 (2010) 20569–76. [52] Ueno K, Yoshida K, Tsuchiya M, Tachikawa N, Dokko K, Watanabe M. Glyme-Lithium Salt Equimolar Molten Mixtures: Concentrated Solutions or Solvate Ionic Liquids? J Phys Chem B, 116 (2012) 11323–31. [53] Tsuzuki S, Shinoda W, Seki S, et al. Intermolecular Interactions in Li+-glyme and Li+-glymeTFSA(-) Complexes: Relationship with Physicochemical Properties of Li(glyme) TFSA Ionic Liquids. Chemphyschem, 14 (2013) 1993–2001. [54] Wu TY, Hao L, Chen PR, Liao JW. Ionic Conductivity and Transporting Properties in LiTFSI-Doped Bis(trifluoromethanesulfonyl)imide-Based Ionic Liquid Electrolyte. Int J Electrochem Sci, 8 (2013) 2606–24. [55] Hayamizu K, Aihara Y, Nakagawa H, Nukuda T, Price WS. Ionic Conduction and Ion Diffusion in Binary Room-Temperature Ionic Liquids Composed of [emim][BF4] and LiBF4. J Phys Chem B, 108 (2004) 19527–32. [56] Seki S, Serizawa N, Takei K, Dokko K, Watanabe M. Charge/discharge performances of glyme– lithium salt equimolar complex electrolyte for lithium secondary batteries. J Power Sources, 2 43 (2013) 323–7. [57] Balakrishnan PG, Ramesh R, Kumar TP. Safety mechanisms in lithium-ion batteries. J Power Sources, 155 (2006) 401–14. [58] Chen ZH, Qin Y, Amine K. Redox shuttles for safer lithium-ion batteries. Electrochim Acta, 54 (2009) 5605–13. [59] Chen J, Buhrmester C, Dahn JR. Chemical overcharge and overdischarge protection for lithiumion batteries. Electrochem Solid-State Lett, 8 (2005) A59–A62. [60] Dahn JR, Jiang JW, Moshurchak LM, Fleischauer MD, Buhrmester C, Krause LJ. High-rate overcharge protection of LiFePO4-based Li-ion cells using the redox shuttle additive 2,5-ditertbutyl-1,4-dimethoxybenzene. J Electrochem Soc, 152 (2005) A1283–A9. [61] Balasubramanian R, Wang W, Murray RW. Redox Ionic Liquid Phases:? Ferrocenated Imidazoliums. J Am Chem Soc, 128 (2006) 9994–5. [62] Forgie JC, El Khakani S, MacNeil DD, Rochefort D. Electrochemical characterisation of a lithiumion battery electrolyte based on mixtures of carbonates with a ferrocene-functionalised imidazolium electroactive ionic liquid. Phys Chem Chem Phys, 15 (2013) 7713–21. [63] Fontaine O, Lagrost C, Ghilane J, et al. Mass transport and heterogeneous electron transfer of a ferrocene derivative in a room-temperature ionic liquid. J Electroanal Chem, 632 (2009) 88–96. [64] Forgie JC, Rochefort D. Electroactive imidazolium salts based on 1,4-dimethoxybenzene redox groups: synthesis and electrochemical characterisation. RSC Adv, 3 (2013) 12035–8. [65] Park J-K, ed. Principles and Applications of Lithium Secondary Batteries. Weinhem: Wiley-VCH; 2012.

C. de Bonis, A. D’Epifanio, B. Mecheri, S. Licoccia, and A. C. Tavares

10 Solid polymer proton conducting electrolytes for fuel cells 10.1 Introduction Numerous practical applications based on electrochemical cells use solid electrolytes. These include batteries, fuel cells, sensors, electrolyzers, and also water purification, electrodialysis, and seawater desalination [1–7]. An electrochemical cell is a device capable of producing electrical energy from spontaneous chemical reactions (ΔGo < 0), or driving chemical reactions through an external source of electrical energy (ΔGo > 0). Electrochemical cells contain two electrodes (the anode and the cathode) with an electrolyte between them. Reactions at the anode (oxidation) and at the cathode (reduction) are the driving forces of an electrochemical cell. Oxidation refers to the loss of electrons by a chemical species (reductant) and reduction to the gain of electrodes by a chemical species (oxidant). In an electrochemical cell the electronic current flowing outside the cell equals the ionic current flowing within the cell [8]. An electrolyte is a material able to conduct ions which is usually an electrical insulator [8]. Electrolytes can be solutions (e.g., KOH, H2 SO4 ), molten salts (e.g. Li2 CO3 ), solid ion conducting polymers (e.g., perfluorinated polymers bearing sulfonic acid groups or benzyltrimethylammonium groups), and ionic crystals (e.g., ZrO2 :Y2 O3 , Na3 Zr2 PSiO12 ). Solid electrolytes conducting O2− , H+ , Li+ , Na+ , Ag+ , F− , Cl− , OH− ions have been reported for many years now. The conductivity range is typically 10−3 S/cm < σ < 10 S/cm depending on the material structure and operating temperature. For the purpose of comparison, the conductivity of a solid electrolyte at room temperature is inferior to that of a liquid electrolyte. For example, σ(KOH) = 0.6 S cm−1 (30%, 20°C), σ(H2 SO4 ) = 0.82 S cm−1 (5.2 M, 20°C), and σ(Nafion® ) ≈ 0.07 S cm−1 (fully hydrated, 20°C). Solid electrolytes can be used in electrochemical cells as ion exchange membranes to allow the passage of ionic current between the anode and the cathode placed on opposite sides of the electrolyte, or in the electrodes when mixed (electronic and ionic) conductivity is needed [5, 9]. Ion conductivity σi is given by: σi = zi ⋅ Ci ⋅ μi ,

(10.1)

where zi is the ion charge, Ci is the density of mobile ions, and μi the ion mobility. Thus, a solid electrolyte has a large number of mobile ions. Ion conductivity is an activated transport; therefore it increases exponentially as temperature increases: σ = A exp (−

Ea ), RT

(10.2)

208 | Part II Development of new materials for energy applications where A is a proportional constant, Ea is the activation energy, R is the universal gas constant (8.314 J mol−1 K−1 ), and T the temperature (in K). There are two main classes of solid electrolytes: crystalline (or ionic) solids and ion conducting polymers. In crystalline solid electrolytes, ion conductivity occurs by means of ions hopping through energetically equivalent sites in the crystal structure. High conductivity requires a large number of mobile ions, or in other words, a large number of accessible empty sites, either vacancies or interstitial sites. A practical way of increasing the density of mobile ions is by doping the crystalline solid with heterovalent ions forming solid solutions. For example, replacing three Li+ ions with one Al3+ ion in Li4−3x Alx SiO4 to generate cation vacancies, or by replacing Zr4+ ions by Y3+ ions in yttria stabilized zirconia to generate anion vacancies. The activation energy controls ion mobility. The empty and occupied sites should have similar potential energies with a low activation energy barrier for ion hopping between neighboring sites. Ion mobility is thus related to the crystal structure. Ionic solids with a densely packed crystal structure are characterized by large activation energy (1 eV or higher) and low conductivity. Ionic solids such as α-AgI, RbAg4 I5, and Na β-Al2 O3 are known as fast ion conductors, formed by solid frameworks with open conduction pathways, and are characterized by low values of activation energy, for example 0.03 eV for AgI above 420 K. Practical applications of this class of compounds include, for example, ion selective electrodes (Ag2 S for Ag+ and LaF3 for F− ), molten salts electrochemical cells (Na β-alumina in ZEBRA batteries), and oxygen anion conductors (yttria stabilized zirconia) for solid oxide fuel cells and oxygen sensors. Solid polymer electrolytes (SPE) consist of polymer backbones functionalized with a high concentration of fixed ionic charges. The function of the SPE is determined by the charge of the ion exchange groups and the nature of the counter ions, and can be classified as follows [10]: (a) cation exchange membranes have anionic charged groups (-COO− , -SO−3 , etc.) and cations can selectively permeate through them; (b) anion exchange membranes have cationic charged groups (e.g. -NR+3 ) and anions can selectively permeate through them; (c) amphoteric ion exchange membranes contain randomly distributed cationic and anionic functional groups; (d) bipolar ion exchange membranes are bi-layer membranes with both a cation and an anion exchange membrane layer; (e) mosaic ion exchange membranes, which have separate domains with cationic and anionic groups. Types (a) and (b) are those used industrially. Due to the presence of ionic groups, ion exchange membranes adsorb water molecules to an extent dependent on the surrounding relative humidity. The electrical conductivity of ion exchange membranes depends on the concentration, size, and charge of the ions, as well as on the water content, chemical structure, and morphology of the

10 Solid polymer proton conducting electrolytes for fuel cells

|

209

membranes. In particular, ion mobility depends on its charge density and degree of solvation [11]. Ion exchange membranes are used for the dehumidification of gases, in humidity sensors, actuators, pervaporation, facilitated transport, and in electrochemical processes such as electrodialysis, brine electrolysis, redox flow vanadium batteries, and solid polymer electrolyte fuel cells [10]. A significant amount of the work on SPEs is relevant to proton exchange membrane fuel cells, therefore this application has been targeted for this chapter, with emphasis on SPEs functionalized with sulfonic acid groups.

10.2 Proton exchange membranes A fuel cell is an electrochemical device which converts the energy from a fuel into electricity, and those using proton exchange membranes (PEM) are among the most promising because of their potential application in portable electronics, stationary and automotives. They operate between room temperature and 140°C and can use hydrogen, methanol or other liquid fuels. During operation the fuel is oxidized at the anode, generating protons and electrons. The electrons flow from anode to cathode through the external circuit, whereas the protons cross the electrolyte membrane to reach the cathode. An oxygen reduction reaction takes place and water is produced: Anode:

H2 → 2 H+ + 2 e− +

(10.3) −

Cathode:

1/2 O2 + 2 H + 2 e → H2 O

(10.4)

Overall:

H2 + 1/2 O2 → 2 H2 O

(10.5)

The proton exchange membrane is the core component of PEM fuel cells. In order to achieve high efficiency, the membrane must possess the following features: (a) high proton conductivity to support high current with minimal resistive losses, (b) low permeability to reactants, (c) chemical and electrochemical stability under operating conditions, (d) adequate mechanical strength and stability, and (e) production costs compatible with the intended application. PEMs can be classified according to their polymer backbone as hydrocarbon membranes, partially halogenated hydrocarbon membranes, and perfluorocarbon membranes. The most common cation exchange group used for fuel cell applications is the sulfonic acid group because it is a very strong acid (apparent pk −6 for –CF2 SO3 H and 0–1 for alyl/alkyl –SO3 H), although phosphonic acid (pk1 = 2–3, pk2 = 7–8) and imidazole are also protogenic groups of potential interest for operating temperatures above 100°C and low relative humidity (RH) [10, 12]. PEMs rely on the mobility of protons in the aqueous network formed inside the solid polymer [13]. Proton transport proceeds through the membrane following two main mechanisms [14]. The first is the vehicle mechanism, where the proton diffuses

210 | Part II Development of new materials for energy applications with the vehicle water. The counter diffusion of unprotonated water allows the net transport of protons. Therefore, the observed conductivity depends on the rate of vehicle diffusion and can be expressed as a function of the water self-diffusion coefficient (DH2 O ), which represents a measure of the average mobility of water in the membrane. The other mechanism is known as the Grotthuss mechanism, “proton hopping” or “structure diffusion”. In this process, the water molecules show pronounced local dynamics but reside on their sites. The process consists of two steps: (1) proton transfer from one water molecule to the other by hydrogen bonds, (2) consequent reorientation of water dipoles resulting in the formation of an uninterrupted trajectory for proton migration. A schematic description of two typical proton conduction mechanisms is shown in Fig. 10.1 The prevalence of one mechanism or the other depends on the hydration level of the membrane, and it has been suggested that proton hopping is more significant for high water contents [15]. The activation energy for proton conduction in SPEs depends on the water content, and typically decreases from 0.4–0.5 eV for dry membranes (contain only residual water molecules) and 0.1 eV for fully hydrated and swollen membranes [16].

S

S

S

S

S

S

S

S

Fig. 10.1. Simplified scheme of the proton transfer in Nafion® by the Grotthuss mechanism (solid lines) and the vehicle mechanism (dotted lines). Adapted with permission from [17].

10.2.1 Nafion® The key ionomer currently used in PEM fuel cell applications is Nafion® , which is produced by DuPont. Nafion® is a perfluorosulfonic acid polymer (PFSA). It has a polytetrafluoroethylene (PTFE) backbone, which confers high chemical inertness, and side chains consisting of perfluorinated vinyl polyether ending in sulfonic acid groups (–SO3 H), which give proton exchange capability to the polymer. The chemical structure is shown in Fig. 10.2, where the values of n, x, and y can be varied to produce materials with different equivalent weights. Nafion® shows excellent proton conductivity (0.09 to 0.12 S cm−1 at 80°C and RH between 34 and 100% RH [18]) and mechanical strength, as well as high thermal and chemical stability. The structure of Nafion® as a function of its water content has been the topic of many investigations and it has been investigated by swelling studies, infrared spectroscopy, small angle x-ray scattering, and transmission electron

10 Solid polymer proton conducting electrolytes for fuel cells

211

CF2

CF2 CF2

|

CF

x

O

y

CF z

O

CF2

O

CF2

S

CF2 O

OH

Fig. 10.2. Chemical structure of the Nafion® ionomer.

CF3

microscopy to name a few [19]. These studies have shown that a hydrated membrane contains two phases, an ionic phase which is associated with the hydrated sulfonic acid groups, and a non-ionic phase which is the perfluorinated matrix. The actual form of the phases depends on the water content. Since the early 1970s several models have been proposed for prediction of ionic transport properties of Nafion® , describing the way in which ionic groups aggregate. These models include the Mauritz– Hopfinger Model [20], the Yeager Three Phase Model [21], and the Gierke Cluster Network Model [22]. In the cluster network model proposed by Gierke and Hsu, the structure is an inverted micelle in which the ion exchange sites are separated from the fluorocarbon backbone, thus forming spherical clusters, connected by short narrow channels; see Fig. 10.3. Thus, with increasing water content, the clusters grow and form transitory interconnections with one another. This network of collapsed channels leads to a percolation-type phenomenon. Gierke and Hsu also used the percolation theory to correlate electrical conductivity with the water content of the membrane, expressed as λ, i.e., the number of water molecules per sulfonic group. According to this theory, there is a critical amount of water available in the membrane, below which ion transport is extremely difficult due to the absence of extended pathways. The percolation threshold in Nafion® is around λ = 2 [20], as shown in Fig. 10.4, where conductivity data is plotted against λ. At low hydration level, i.e., λ in the range of 1–2, it is rea5.0nm

SO–3

SO–3

SO–3

SO–3 SO–3

SO–3 SO

SO–3

SO–3 – 3

SO

SO–3

– 3

SO SO–3 1.0nm– SO3

4.0nm

– 3

SO–3

– 3

SO

SO–3 SO–3 SO–3 SO–3

SO–3 – 3

SO–3 SO

Fig. 10.3. Gierke’s cluster network model of Nafion® membranes. Reprinted with permission from [22].

212 | Part II Development of new materials for energy applications sonable to consider that all water molecules absorbed by the Nafion® membrane are associated with the sulfonate heads because of the hydrophobic nature of the backbone and the hydrophilic nature of the sulfonic groups. Moreover, hydronium ions will be localized on the sulfonate heads and conductivity will be extremely low, as the amount of water absorbed is insufficient for the formation of a continuous water phase [23]. For λ in the range of 3–5, the counterion clusters continue to grow and, as λ approaches 5, the membrane will become more conductive since some counterion clusters may connect. However, there is still insufficient water for all clusters to coalesce. Molecular dynamics simulations indicate that 5 water molecules form the primary hydration shell for the sulfonic groups and any additional water molecules are not as strongly bound and thus form a free phase [24]. For λ ≥ 6, counterion clusters coalesce to form larger clusters and eventually a continuous phase is formed and the conductivity threshold is overcome. 0.10

Conductivity/Scm–1

0.08 0.06 0.04 0.02 0.00 0

5

15 10 Mole ratio H2O/SO3H

20

25

Fig. 10.4. Variation of the proton conductivity of Nafion® as a function of the water content in the membrane. Reprinted with permission from [4].

Since proton mobility relies on the formation of a continuous aqueous network inside the ionomer, proton conductivity shows a strong dependence on the hydration level of the PEM, as shown in Fig. 10.4 for Nafion® . A humidification system is thus necessary to keep the membrane hydrated during fuel cell operation, which represents a major cost of the fuel cell system [25]. While water sorption improves proton conductivity, it also leads to morphological instability and, at elevated water content, to membrane swelling [26]. The maximum working temperature of all Nafion® -based fuel cells is limited to 80–90°C due to the loss of mechanical strength of the membrane determined, at higher temperature and

10 Solid polymer proton conducting electrolytes for fuel cells

|

213

large hydration, by the plasticizing effect of water. Moreover, under dynamic conditions, swelling cycles contribute to mechanical fatigue. In fact, one of the key challenges in the design of proton exchange membranes is the retention of high conductivity at low water content [3], especially at high temperatures [25]. Overall, operation at high temperature (> 100°C) is desirable to reduce PEM fuel cell costs and to promote large-scale commercialization: it enhances reaction kinetics at both electrodes thus reducing the catalyst loading on both electrodes, it allows more efficient utilization of the waste heat, and simplifies the water and thermal management systems [25]. Nafion® membranes possess an additional major hurdle which inhibits largescale commercialization of fuel cells operating with liquid fuels. The unique microstructure of the Nafion® ionomer results in a high crossover rate of liquid fuels from the anode to the cathode through the membrane [27]. This not only lowers fuel utilization at the anode but also increases the overpotential of the cathode, hence lowering the cell performance. All of these drawbacks essentially imply that the Nafion® membrane cannot be used “as is” for fuel cell applications in a wide temperature range, relative humidity (RH) and liquid fuels. Substantial effort is being made to develop membranes with appropriate electrochemical and other physico-chemical properties under operating conditions [28–30]. Different approaches are being pursued and include: (a) the development of alternative ionomers based on non perfluorinated polymers [31, 32], on polyarylene or on aliphatic main chains [33–38], (b) the modification of existing ionomer membranes through the formation of blends and composites [29], and (c) the synthesis of new hybrid systems [39, 40].

10.2.2 Alternative sulfonated ionomers and membranes Among the alternative polymers to Nafion® , arylene main-chain polymers, such as poly(ether ketone), poly(ether sulfone), poly(benzimidazole), and poly(phenylene sulfone), as shown in Fig. 10.5, have been widely investigated [34–36, 41–44]. Such polymers are inexpensive and possess high chemical and mechanical stability at temperatures higher than 90–100°C [34–36]. Moreover, their aromatic structure offers the possibility of electrophilic and nucleophilic substitutions, to prepare ionomers with the desired features for PEMFC and DMFC applications [45]. The most important modification regards the introduction of sulfonic acid moieties to obtain proton-conducting aromatic polymers. Several methods have been developed for the preparation of proton-conducting electrolytes, including direct sulfonation of a polymer backbone, total synthesis from monomer building blocks, and grafting of functional groups onto a polymer main chain [46]. In general, the larger the number of sulfonic acid groups per structural unit, the larger the membranes’ ionic exchange capacity and water uptake and the higher their conductivity. However, excessive swelling of

214 | Part II Development of new materials for energy applications O O

O

C

(a)

n O

CH3 O

O

C

S O

CH3

(b)

n

H N

N

NH

N n

(c) Fig. 10.5. Chemical structure of (a) Polyetheretherketone; (b) Udel Polysulfone; and (c) Poly(2,2󸀠 -m-(phenylene)-5,5󸀠 -bibenzimidazole).

the membranes could lead to dilution of the charge carriers and to lower proton conductivity [33]. Some variations on Nafion® ’s hydration scheme are expected for sulfonated polyarylene membranes. Sulfonated polyarylenes with sulfonic acid groups bound directly to the aromatic chain have less pronounced hydrophobic/hydrophilic separation with respect to Nafion® because their backbones are less hydrophobic and flexible, and their sulfonic acid groups are less acidic and therefore also less polar. As a consequence, narrower channels and a less-connected network of clusters are present in the microstructure of sulfonated polyetherketones, resulting in a higher dependence of the transport properties on water content due to percolation concepts [27]. A schematic representation of the microstructure of sulfonated polyetherketone compared with that of Nafion® is shown in Fig. 10.6. As illustrated in Fig. 10.7, high conductivity levels are achieved only with a high degree of sulfonation. This results in low mechanical properties and a high rate of methanol crossover due to excessive swelling [47]. Several strategies have been used to overcome the excessive swelling of highly sulfonated polyarylenes. These include the synthesis of aromatic polymer chains crosslinked covalently by organic spacers such as α,ω-dihalogenoalkanes [48], the use of partially fluorinated backbones [31], placing the protogenic groups on short pendant side chains to increase the separation between the polymer main chains and the sulfonic acid groups [49], or building multi-block copolymers using coupling reactions between hydrophilic and hydrophobic macromonomers [29]. Polymers with pendant

10 Solid polymer proton conducting electrolytes for fuel cells

NAFION

|

215

Sulfonated polyetherketone (PEEKK) O

–(CF2–CF2)n–CF–CF2–

O

O–(CF2–CF–O)m–CF2–CF2–SO3H CF3

O O

SO3H

1 nm

: –SO3– : Protonic charge carrier : H2O • Wide channels • More separated • Less branched • Good connectivity • Small –SO3–/–SO3– separation • pKa~–6

• Narrow channels • Less separated • Highly branched • Dead-end channels • Large –SO3–/–SO3– separation • pKa~–1

Fig. 10.6. Schematic representation of the microstructures of Nafion® and a sulfonated polyetherketone. Reprinted with permission from [27].

sulfonic acid groups in side chains are in general more stable against hydrolysis than those with sulfonic acid groups attached directly to the polymer backbone. In addition, sulfonic acid groups on pendant side chains have a higher degree of freedom, which results in better phase separation and higher proton conductivity with respect to random sulfonated analogues [50, 51]. Hydrophilic-hydrophobic multiblock copolymers are considered an interesting step forward in the rational design of PEMs. The ideal morphology has been pursued by controlling the microphase separation in segmented block copolymers where hydrophilic sulfonated polymer segments form an interconnected 3D network responsible for efficient proton transport especially at low relative humidity [52–54], while a complementary network of hydrophobic non-sulfonated segments causes a reinforcing effect, preventing excessive swelling in water and enhancing mechanical

216 | Part II Development of new materials for energy applications

0.15

Dow 800

T = 300 K

σ/S cm–1

S-POP 625

Nafion 1100

0.10 Dow 1000

0.05 S-PEEKK 700

S-PEK 730 S-POP 833

0

10

20

30

40

50

n = [H2O]/[–SO3H] Fig. 10.7. Proton conductivity (measured at room-temperature) of two Dow membranes, Nafion® , two sulfonated poly(arylene ether ketone)s (SPEK and S-PEEKK), and sulfonated poly(phenoxyphosphazene) (S-POP) as a function of the degree of hydration n; the number below the compound acronym/name indicates the equivalent weight of the ionomer. Reprinted with permission from [47].

properties [55, 56]. Proton and water transport increase significantly with increasing block length because the longer block induces a more developed phase separation [57]. However, their synthesis is often complex, thus increasing the materials’ cost. Overall, synthetic approaches based on structure-property relationships of ionomers represent a very promising method of obtaining more efficient proton-conducting membranes with the desired features for fuel cell applications [31, 57, 58]. Anhydrous proton-conducting electrolytes consist of a more or less inert polymer matrix which is swollen with an appropriate proton solvent, usually phosphoric acid. These membranes are appealing for fuel cell operation at temperatures well above 100°C without the need for humidification. One of such membrane is poly(2,2󸀠 -m(phenylene)-5,5󸀠 -bibenzimidazole) (PBI) whose structure is shown in Fig. 10.8. Nonmodified PBI shows very low proton conductivity. It is therefore necessary to dope the polymer with sulfuric or phosphoric acid to increase its proton conductivity [59, 60]. However, acid leaching from the membranes and corrosion of cell components are two of the problems limiting the performance of fuel cell devices based on such membranes. Alternative concepts use amphoteric heterocycles such as imidazole as a proton conducting species “imbibed” in a polymer matrix. Proton transport occurs through heterocyclic hydrogen-bonded networks under both anhydrous and low relative humidity conditions. As for sulfonic acid-based ionomers, ion conductivity depends on the local mobility of the heterocycles within the polymer films and on the effective concentration of mobile protons in the membranes [12, 60–64]. The proton conductiv-

10 Solid polymer proton conducting electrolytes for fuel cells

|

217

ity of these systems increases with the addition of strong acids due to the protonation of some of the heterocycles within the polymer matrix [64, 65]. Recently, ionic liquids have also been proposed for high temperature proton conductors, mainly due to their anhydrous high conductivity and good thermal stability. Nevertheless, the conductivity of ionic liquid-based composite membranes is lower than that of the original ionic liquids. Therefore, only a few groups have reported demonstrations of the ionic liquid-based solid membrane electrolytes in fuel cells [66–68]. The composite strategy, where an inorganic phase is dispersed within the ionomeric host has been demonstrated to be another effective way of improving the transport and mechanical properties of ionomers. Several advantages can be obtained by using composite membranes, such as: (a) improving the self-humidification of the membrane at the anode side, (b) suppressing the fuel crossover, e.g., methanol in DMFC, and (c) improving the mechanical strength of membranes without excessively sacrificing proton conductivity [39, 40, 69]. Solid inorganic compounds can be classified as inert hygroscopic fillers, proton conductive fillers, and hydrophilic and proton conductive fillers. They include: hygroscopic oxides (SiO2 , TiO2 , SnO2 ), clays, zeolites, heteropoly acids, and zirconium phosphonates [39, 70–73]. For example, (a) hygroscopic fillers, e.g., SiO2 , TiO2 , SnO2 , and zeolites improve water retention and dimensional stability of the membranes [74, 75], (b) the operation of fuel cells fed with liquid fuels was sucessfully extended to high temperature [76, 77], and (c) beta and faujasite zeolites improved proton conductivity and the DMFC performance of Nafion® [73, 78]. Composite membranes containing exfoliated layered compounds or 1D structures such as nanotubes or nanorods as fillers are also an effective strategy for improving relevant properties of electrolytes [76, 79]. The presence of one- and two-dimensional nanomaterials, which have substantially different properties to those of nanometric spherical particles, can enhance mechanical strength while acting as a physical barrier to fuel crossover [76, 80, 81]. Performance of composite electrolytic membranes is in fact strongly related to the polymer/inorganic phase interfacial properties. In detail, the higher the interface interaction between the polymer and the dispersed particles, the greater the filler’s influence on the original characteristics of the polymer [82, 83]. Composite membranes are generally prepared by casting the polymer solution with an inorganic component. The main disadvantage of such composite systems is related to the fact that it is very difficult to obtain homogenous systems, where the inorganic particles are well dispersed in the polymeric matrix. Therefore, when applicable, in-situ sol-gel synthesis of the inorganic filler in the hydrophilic clusters of the PEM is a preferred alternative to the nanocasting method [73, 76]. According to recent reports [76, 83, 84], endeavours to identify the conditions under which inorganic-organic membranes provide properties superior to those shown by their polymer-only counterparts have been successful, and there is every reason to be optimistic that MEAs based on nanocomposite membranes have a role to play in liquid feed fuel cells, or in the highly strategic operating conditions of low

218 | Part II Development of new materials for energy applications RH at 110–130°C. Current hurdles persist: membrane electrical resistance and longterm durability under fuel cell operation. Surface functionalization of the inorganic fillers with protogenic groups is being exploited to boost membrane conductivity [73, 85–87]. However, in-depth studies of aging and degradation under realistic operating conditions are still needed to enable the synthesis of more advanced materials and alignment with current targets.

10.3 Characterization of solid polymer electrolytes The performance of H2 /O2 and liquid feed fuel cells is strongly influenced by the proton and water transport properties of the PEM. The fuel cell ohmic loss is proportional to the ionic resistance of the PEM, and high conductivity is essential to assure the required performance. Water molecules in the membrane increase proton mobility according to the vehicle mechanism, but a high water uptake by the membrane decreases the density of sulfonic acid groups or charge carriers [88]. Therefore, changes in water content and water mobility have an impact on the proton conductivity of membranes [78, 88]. This section provides a short description and application of complementary characterization tools (proton conductivity measurements, dynamic vapor sorption, and differential scanning calorimetry) used to assess transport properties of PEMs. Although these are the first properties to be considered when evaluating PEMs for potential use in fuel cells, it should be stressed that other chemical, morphological, mechanical, and thermal properties are also critical for the definition of the “ideal” electrolyte for fuel cell applications and the study of structure–property relationships. These properties can be studied by means of several characterization techniques including bulk chemical analysis and ion exchange capacity, thermal gravimetric analysis, transmission electron microscopy, small-angle x-ray scattering, tensile tests, dynamic mechanical analysis, fuel cell life, and Fenton’s tests [28, 89–91].

10.3.1 Proton conductivity The proton conductivity of a membrane is determined by measuring its resistance against the flow of a direct current or an alternative current at controlled temperature and hydration level. The conductivity σ is calculated by the equation:

10 Solid polymer proton conducting electrolytes for fuel cells

σ=

l , RS

|

219

(10.6)

where l is the distance between the two probe electrodes and S the cross-sectional area of the membrane. In the dc method, the potential difference across two probe electrodes in contact with the membrane follows Ohm’s law over a wide range of current densities and the resistance can be determined from the slope of the line ΔE vs j. In the ac method, a periodic small-amplitude ac signal (voltage or current) is applied and the associated response (current or voltage) coming from the cell is measured [92]. The voltage response to a sinusoidal current signal is a sinusoid, at the same frequency (ω) but shifted in phase (φ): it = i0 sin(ωt)

(10.7)

Et = E0 sin(ωt + φ)

(10.8)

The impedance, Z, is defined as the ratio of the voltage to the current at a given frequency: E E sin(ωt + φ) Z= t = 0 (10.9) it i0 sin(ωt) By applying Euler’s relationship, the impedance can be expressed as a complex function with a real and an imaginary part: Z = Z0 (cos φ + j sin φ)

(10.10)

In an electrochemical system, slow kinetic reactions and diffusion of chemical species can impede the flow of electrons. Electrochemical systems can thus be considered analogous to the resistors, capacitors, and inductors which hinder the flow of electrons in an electrical circuit. In the case of a simple resistor, the phase shift is zero degrees and the current is in phase with the voltage. Thus, according to equation (10.9), the impedance is purely real and independent of the frequency. For an ohmic resistance Zt = R. Fig. 10.8 shows typical impedance data in the form of a Bode plot obtained for a Nafion® 117 membrane using a four-probe cell (see below) at 100% relative humidity. First, the frequency region over which the impedance has a constant value is identified, and the impedance value taken to calculate the membrane’s conductivity using equation (10.6) [93]. Electrochemical impedance spectroscopy is the most commonly used method for measuring the membrane’s resistance and to determine its proton conductivity. It is a rapid and accurate method and is quite suitable for dielectric materials such as PEMs. The dc method has also been used [72, 94], the major advantage of this method being the straightforward analysis of the E-j data. Conductivity of the membrane can be measured perpendicular to the membrane’s thickness (through-plane conductivity) or along the plane of the membrane (in-plane conductivity). In addition, measurements can be made using either the four-probe or

220 | Part II Development of new materials for energy applications 103

Impedance/Ω

4

0

–4

Phase angle/degree

8

Impedance Degree 102 0

1

2

3

4

5

–8

Iog frequency Fig. 10.8. Impedance data recorded for a Nafion® 117 membrane at 100% RH. Data was acquired in the in-plane-plane direction using a four-probe cell (reprinted with permission from [93]).

the two-probe method. In fact, there is no standard method for measuring the proton conductivity of ionomers, and each method/cell configuration has its own advantages and disadvantages. Figure 10.9 illustrates some of the conductivity cells reported in the literature. During fuel cell operation, protons move through the cross-section of the membrane. Thus, measurements made in this direction are more relevant for practical applications. However, in this configuration the area of the electrodes (≈ cm2 ) is much larger than the distance between them (given by the membrane thickness, ≈ micrometers), so the cell constant (l/S) is small and the contribution from the interface formed between the membrane and the electrodes is large. On the contrary, for in-plane measurements the cell constant is larger since the distance between the electrodes is of the order of mm to cm, and the section of interest is the cross-section of the membrane. The bulk conductivity of the membrane is the dominant element contributing to the measurement [95]. Conductivity measurements in both directions are nevertheless important to quantify the effect of morphological anisotropy of PEMs on their proton conductivity [95–97]. In the two-probe method, the voltage drop is measured across the same two electrodes where the current flows. Accordingly, the measured impedance (or resistance) includes the contribution of all components on the current pathway. When determining, for example, the membrane resistance from the total cell impedance, all other contributions such as electrode resistance, lead inductance, and membrane – electrode contact resistance should be subtracted from the total cell impedance [95, 98]. This is done by recording the impedance of the short-circuited and open cells [95]. With the four-probe method, only the bulk membrane resistance is measured because two distinct pairs of electrodes are used, and current flow and voltage sensing are done independently. The current is imposed on the external pair and the voltage drop

10 Solid polymer proton conducting electrolytes for fuel cells

|

221

Teflon block

Teflon block

Contact electrode Contact electrode PEM

PEM

Contact electrode Teflon block

Teflon block Contact electrode (a)

(b)

Membrane Waveform generator

Galvanostat (c) I

High impedance DVM

I

Δ Eref Cation exchange membrane Glass flange 8 cm

Platinum gauze Luggin counter electrode capillary (1.5 cm2) (d)

L

Saturated calomel reference electrode Silicone-rubber gaskets

20 cm

Fig. 10.9. (a) In-plane and (b) through-plane two-probe conductivity cells; (c) In-plane and (d) fourprobe conductivity cells. (a) and (b) reprinted with permission from [94] and (d) reprinted with permission from [95].

along the membrane sectional area is measured using the central pair of electrodes. The effect of contact resistance is clearly seen in Fig. 10.10, which shows the variation of the in-plane proton conductivity of a Nafion® 112 membrane exposed to 95% RH and immersed in liquid water, as a function of the torque applied to the electrodes, for a four-probe and a two-probe configuration. A lower interface resistance between

222 | Part II Development of new materials for energy applications 2x10–1

Proton conductivity (S/cm)

1.5x10–1

Four probe method at RH 95% Two probe method at RH 95% Four probe method in liquid water Two probe method in liquid water

10–1

5x10–2 0.4

0.6

0.8

1.0 1.2 Torque (kgfcm)

1.4

1.6

Fig. 10.10. Effect of applied torque on the measured proton conductivity of Nafion® 112 by fourprobe (◼) and two-probe (∙) configurations at 95% RH and 60°C, and by four-probe (◻) and twoprobe (∘) methods in the liquid-water state at 60°C. Reprinted with permission from [98].

membrane and electrodes corresponds to a higher torque, and the influence of this additional resistance is more important when the conductivity of the membrane is lower (less hydrated membrane) [98]. Measurements of ion conductivity over a wide range of temperature and relative humidity are important to determine the effect of composition and structure of the new ionomers on the proton conduction and operational temperature. An example is given in Fig. 10.11 which illustrates the effect of the filler content and composition on the proton conductivity of Nafion® composite membranes containing TiO2 and propyl sulfonic acid functionalized TiO2 nanoparticles [87]. The loading of an appropriate amount of propylsulfonic-functionalized titania allows the preparation of Nafion® based composite membranes with higher conductivity and dimensional stability than pristine Nafion® up to 140°C.

10.3.2 States of water and water mobility Two critical parameters affecting the performance and proton conduction mechanism of PEMs are their hydration level and water diffusion coefficient as a function of the water content. It is therefore important to study the water sorption and diffusion behavior of electrolytes over a wide range of relative humidity.

10 Solid polymer proton conducting electrolytes for fuel cells

140

120

100

80

60

|

223

40

–1.0

log (σ/S cm–1)

–1.2 –1.4 –1.6 –1.8

N_RC N_5TiO2-RSO3H N_10TiO2-RSO3H N_20TiO2-RSO3H N_5TiO2

–2.0

2.4

2.5

2.6

2.7 2.8 2.9 1000/T (K–1)

3.0

3.1

3.2

3.3

Fig. 10.11. Arrhenius plot of Nafion® , Nafion® -TiO2, and Nafion® propyl sulfonic acid functionalized TiO2 at 100% RH. The numbers in the legend indicate the wt% of filler with respect to Nafion® . Reprinted with permission from [87].

Dynamic vapor sorption Dynamic Vapor Sorption (DVS) is a gravimetric technique which allows fast and accurate determination of vapor sorption isotherms and diffusion kinetics. A simplified scheme of the DVS apparatus is shown in Fig. 10.12(a) the samples are placed in a weighing pan and exposed to partial pressure- and temperature-controlled environment. In order to study water management in solid electrolytes, water is used as sorbate and the electrolyte as sorbent. The vapor partial pressure around the sample is controlled by mixing saturated and dry carrier gas steams using electronic mass flow controllers. A constant temperature is maintained by enclosing the entire system in a temperature-controlled incubator. By measuring the change in mass as a function of time to equilibrium, a typical diagram such as that shown in Fig. 10.12(b) is obtained. The amount of water uptake (WU) by the sample exposed to a defined partial pressure can thus be obtained using equation (10.11): WUaw =

m(eq)aw − mdry mdry

,

(10.11)

where m(eq)aw is the mass of the sample at equilibrium for a defined water activity (aw ), and mdry is the dry mass of the sample. A sorption isotherm is the graphic representation of WU values; it describes the relationship between the water content of the electrolyte and water activity at constant temperature. Water is a small molecule and a polar adsorptive, therefore its adsorption mechanism is influenced by water affinity to the adsorbent surface. The

224 | Part II Development of new materials for energy applications

Dry gas

Microbalance module Humidity regulated sample module

Vapour Vapour

Vapour generator module

Flow control module

Temperature controlled chamber

Sample reference Camera option

7.25

Target aw Actual aw

7.00

1.0 0.8

M/mg

6.75 6.50

Mass

6.25

0.4

6.00

0.2

5.75 0 (b)

0.6

500

1000 t/min

1500

Water activity (aw=P/P0 )

(a)

0.0 2000

Fig. 10.12. (a) Illustration of the DVS apparatus interfaced with a personal computer (reprinted with permission from Surface Measurements Systems), and (b) kinetics of water adsorption of a typical Nafion® membrane at 25°C and different partial pressures.

shape of the isotherm thus also reflects the hydrophilicity/hydrophobicity of the surface. IUPAC proposed a classification for water sorption isotherms as illustrated in Fig. 10.13 [99, 100]. Each isotherm shape is related to a material with specific hydrophilic characteristics. Type I is characteristic of every hydrophilic material. Type II and type IV isotherms are characteristic of moderate hydrophilic materials. Adsorbents showing a type IV isotherm are hydrophilic as well. Adsorbents with low hydrophilicity will give rise to type III and type V isotherms. Type VI is typical of a hydrophilic material with multiple sorbent–water interactions and stepwise sorption, while type VII isotherms are characteristic of very hydrophobic materials.

10 Solid polymer proton conducting electrolytes for fuel cells

|

225

Hydrophilic materials V

V

V

II

I 0

P/PO

V

0

IV 0

P/PO

P/PO

VI 0

P/PO

Hydrophobic materials V

III 0

V

V

P/PO

VII

V 0

P/PO

0

P/PO

Fig. 10.13. IUPAC classification of adsorption isotherms for materials with different hydrophilicity. Adapted with permission from [100].

Determination of the diffusion coefficient from DVS measurements DVS measurements allow evaluation of the water diffusion coefficient through electrolyte materials. Assuming that water sorption can be described by fickian behavior, the water diffusion coefficient (D) can be calculated from the relationship between mass variation and the time of water vapor exposure up to equilibrium [101]. This relationship is obtained by combining Fick’s first law (10.12; describing the transfer of solute atoms per unit area in a one-dimensional flow) and the conservation of mass relationship (10.13) and expressed by Fick’s second law (10.14): 𝜕C 𝜕x 𝜕J =− 𝜕x 𝜕2 J = −D 2 𝜕x

J = −D 𝜕C 𝜕t 𝜕C 𝜕t

(10.12) (10.13) ,

(10.14)

where J is the amount of substance flowing per unit area as a function of time, C is the concentration, and x is the position. Assuming constant diffusivity and that water activity is constant across the membrane/vapor interface (c = c∞ at x ± d/2), solving (10.14) gives the normalized mass change as a function of the time: Mt 4 D⋅t = √ , M∞ d π

(10.15)

where Mt is the amount of water adsorbed at time t, M∞ is the amount of water adsorbed at equilibrium, and d is the sample thickness. D can be obtained by plotting Mt /M∞ for a sample exposed to a certain partial pressure P/P0 (i.e., water activity) as a function of the square root of time (Fig. 10.14)

226 | Part II Development of new materials for energy applications 1.0

M(t)/M (∞)

0.8

0.6

0.4 Equation Adj. R-Squar

y = a + bx 0.9976

0.2

Standard error

Value Intercept Slope

–0.04147

0.01338

0.02505

0.00121

0.0 0

20

40

60 80 t1/2 / S1/2

100

120

140

Fig. 10.14. Typical plot of Mt /M∞ versus t1/2 at a given value of water activity (aw ). Adapted with permission from [78].

and by fitting the curve to equation (10.15). This equation is valid for values of Mt /M∞ < 0.4, where the plot of Mt /M∞ against t1/2 is linear [102, 103]. Figure 10.15 shows the water diffusion coefficient of a Nafion® membrane as a function of water activity, measured by DVS at 25°C. As shown, D increases with water content in the membrane at low aw and reaches a maximum in the 0.3 to 0.4 aw range. The increase in D in this aw range is due to the fact that water is less tightly associated with the sulfonic acid sites of Nafion® as water content increases. At higher water activities, D decreases with increasing aw due to the water aggregation process which occurs and provides kinetic limitations on the adsorption of water on the polymer matrix [78].

Determination of the different states of water It is also possible to obtain information about water mobility (and consequently proton transport) from water sorption measurements by investigating the state of the water in the electrolytes. In fact, specifically designed models can be applied to the sorption isotherms in order to obtain insights into the water transport properties of electrolytes. For instance, conventional dual mode sorption models (Langmuir-type) are effective in describing isotherms which are concave towards the activity axis, while the engaged species induced clustering model (Flory-type) has been highly successful in modeling isotherms in polymers which are convex to the aw axis [104]. Multimode sorption models (Park-type) are particularly suited to sigmoidal isotherms, which are the most common isotherm shapes among ionomers [105].

10 Solid polymer proton conducting electrolytes for fuel cells

|

227

1.75x10–7 1.50x10–7

D/cm2s–1

1.25x10–7 1.00x10–7 7.50x10–8 5.00x10–8 2.50x10–8 0.00 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 aw Fig. 10.15. Water diffusion coefficient (D) values of a recast Nafion® membrane as a function of water activity at 25°C. Adapted with permission from [78].

By applying the multi-mode model proposed by Park to the sorption isotherms, the presence of three different mechanisms in the sorption process can be hypothesized: (a) specific adsorption at low water activity, described by the Langmuir model; (b) nonspecific adsorption, described by Henry’s law; (c) water clustering at high water activity. All of these contributions can be formulated in the following equation: WU =

aL KL aW + KH aw + nKA anW , 1 + KL aW

(10.16)

where aL is the specific site capacity, KL is an affinity constant, KH is Henry’s law coefficient, KA is the aggregation equilibrium constant, and n is the aggregate size. A distinct population of water adsorbed in the membrane can be associated to each adsorption mechanism: specific adsorbed water (WSA ), nonspecific adsorbed water (WNSA ), and clustered water (WC ). Each water population is described by the terms constituting equation (10.16) as follows: WSA =

aL KL aW 1 + KL aW

WNSA = KH aw WC =

nKA anW

(10.17) (10.18) (10.19)

Figure 10.16 depicts the typical result of the curve fitting of a polymer membrane sorption isotherm, where the adsorbed water was separated into three contributions so that the sum of WSA , WNSA , and WC matched the experimental isotherm data.

228 | Part II Development of new materials for energy applications Taking into account that each type of adsorbed water is characterized by different mobility, we correlated the different water population to the water mobility degree in the membrane. Being strongly bound to specific sites, the specific adsorbed water is characterized by low mobility, whereas the dissolved water molecules (Henry population) have higher mobility. The growth of water clusters then reduces the mobility of the water aggregates. As a consequence, among the three types of water population, the nonspecific adsorbed water is characterized by the highest mobility. The amount of each type of adsorbed water was normalized to the total water content in the membranes as follows: θW [SA] = θW [NSA] = θW [C] =

W[SA] × 100 WTOT

(10.20)

W[NSA] × 100 WTOT

(10.21)

W[C] × 100 WTOT

(10.22)

As θ parameters were defined, θW [SA], θW [NSA], and θW [C] represent the “specific adsorbed water degree”, “nonspecific adsorbed water degree”, and “clustered water degree”, respectively. As shown in Fig. 10.16(b), specific adsorbed water dominates at low relative humidity, nonspecific adsorbed water at intermediate values of RH, and clustered water dominates at high relative humidity. These variations are consistent with those found for D and shown in Fig. 10.15. The θW [NSA] parameter thus represents the “water mobility degree” and allows comparison of different electrolytes in terms of water mobility: the higher θW [NSA], the greater the water mobility in the electrolyte is expected to be [72, 78]. As already mentioned, the analysis of water sorption isotherms of ionomers is of paramount importance for the final fuel cell performance and scientific literature in this field is mainly based on adsorption properties of water on perfluorinated polymers, in particular Nafion® , which is the state-of-the-art material [106]. Sorption isotherms of most common perfluorinated ionomers can be indeed described by Park’s model. However, the involvement of five adjustable parameters (see equation (10.16)) makes the chemico-physical interpretation unclear for ionomers whose microstructure is considerably different from that of Nafion® , which is the case for polyaromatic polymers (see Fig. 10.6). The less pronounced hydrophobic/hydrophilic separation in polyaromatic polymers compared to Nafion® makes the distinction between specific adsorbed water, nonspecific adsorbed water and clustered water (WC ) quite difficult. As an alternative to the multimode Park’s model, the sorption behavior of the membranes can be analyzed and interpreted on the basis of the dual mode sorption model proposed by Feng [107]. The model is based on the Guggenheim–Anderson–

10 Solid polymer proton conducting electrolytes for fuel cells

|

229

25

WU/wt.%

20

Equation:

Adj. R-Square: 0.99828

Park model

Red. Chi-Square: 0.0614

Sample N_0_HP

AL KL

2.89 5.85

0.418 0.423

KH n KA

9.38 10.1 1.43

0.729 0.87 0.125

Value

15

10

Standard error

Experimental data Curve fitting WC

5

WNSA WSA

0 0.0

0.2

0.4

(a)

0.6

0.8

1.0

aw

70

θNSA

60 50 θSA

θ/%

40 30 20 10

θC 0 0.0 (b)

0.2

0.4

0.6

0.8

1.0

aw

Fig. 10.16. (a) Typical curve fitting (Park’s model) of experimental sorption isotherm data (Nafion® membrane at T = 25°C) and the corresponding fitting parameters, (b) variation of the three types of water population in the membrane with the water activity. Adapted with permission from [78].

de Boer (GAB) multilayer sorption theory [108–110] and, at variance with the GAB model which considers all sorption sites equivalent, the Feng model is based on the assumption that the sorption sites can be divided into two different types, one being the polymer matrix region and the other the microvoid region (specific sorption sites).

230 | Part II Development of new materials for energy applications According to this model, the water content in the membranes can be described using equation (10.23): WU = Cp

k󸀠 aw (A󸀠 − 1)k󸀠 aw + Cp , 󸀠 1 − k aw 1 + (A󸀠 − 1)k󸀠 aw

(10.23)

where Cp is the weighted mean value of the polymer sorption capacity, k󸀠 and A󸀠 are temperature-dependent constants, k󸀠 provides a measure of the interaction between water and the polymer matrix. Values lower than 1 indicate very weak interactions between water and polymer matrix. The higher the k󸀠 value, the greater the hydrophilicity of the polymer. A󸀠 represents the difference between the interaction of a microvoid and its first molecule adsorbed, and that of a microvoid and the subsequent adsorbed molecules. Thus it provides a measure of the affinity between water and the polymer microvoid. A󸀠 values close to 1 correspond to a polymer in a rubbery state without microvoids; the higher A󸀠 is, the greater the dependence of sorption on microvoids and affinity of specific sites to water. Feng’s model requires only three parameters: Cp, the weighted mean value of the sorption capacity of the polymer to water, k󸀠 , the affinity between water and the polymer matrix (hydrophobic region), and A󸀠 , the affinity between water and the polymer microvoid (hydrophilic domains). Rather than discriminating between the different states of water in the membrane, the values and comparison of Feng’s parameters make deep insight into chemical nature, and also into polymer microstructure possible. Figure 10.17 shows the water adsorption isotherms of Nafion® and sulfonated polysulfone (SPS) with the result of a typical curve fitting Feng’s model. The figure shows the very good match between the experimental data and the fit curves, and the corresponding fitting parameters of all samples are summarized in the inset table. Cp, k󸀠 , and A󸀠 parameters of the unfilled Nafion® membranes are in good agreement with values found in previous papers. Both polymers showed low k󸀠 values, indicating that sorption in the polymer matrix region is negligible. Hence, microvoid sorption is predominant, as expected in the case of ionomer systems in which water associates through the sulfonic acid groups, and Cp represents the monolayer sorption capacity in the microvoid region (specific adsorption) [83, 111]. The comparison between the fitting parameters of unfilled Nafion® and SPS indicated that Cp was higher for SPS than for Nafion® , whereas the A󸀠 parameter shows the opposite trend. Differences in Cp and A󸀠 for Nafion® and SPS are ascribed to differences in the microstructures of the two ionomers. Both Nafion® and SPS phases separate in hydrophilic and hydrophobic domains. The hydrophobic domains consist of the perfluorinated and polyaromatic backbone of Nafion® and SPS respectively. The hydrophilic domains arise from the sulfonic acid groups (–SO3 H) which are responsible for bonding with water molecules. Hydrophobic/hydrophilic separation is more pronounced in the case of Nafion® , as depicted in Fig. 10.6. The greater tortuosity of the hydrophilic domains in SPS than in Nafion®

10 Solid polymer proton conducting electrolytes for fuel cells

|

231

35 30 25

WU/%

20 15

Sample

Parameters Value SE

Nafion

Cp

3.57

0.07

Reduced Chi-Sqr k' 0.04697

0.88

0.03

Adj. R-Square 0.99893

A'

15.8

1.8

S

Cp

7.46

0.37

Reduced Chi-Sqr k' 0.32917

0.83

0.01

Adj. R-Square 0.99697

5.32

0.42

A'

SPS

Nafion

10 5 0 0.0

0.2

0.4

0.6

0.8

1.0

aw Fig. 10.17. Curve fitting of experimental adsorption isotherm data of the unfilled Nafion® and SPS membranes at T = 25°C. SE = Standard Error. Adapted with permission from [83].

makes the water phase in SPS lower interconnected than in Nafion® , thus explaining the higher Cp and lower A󸀠 values of SPS compared to those of Nafion® .

Differential scanning calorimetry Differential Scanning Calorimetry (DSC) is a thermoanalytical technique which monitors heat effects associated with phase transitions and chemical reactions as a function of temperature [112–114]. It consists of measuring the difference in heat flow between the sample and a reference at the same temperature, the temperature of both sample and reference being increased at a constant rate. The heat flow difference between the sample and the reference can be either positive or negative, depending on whether the process is endothermic or exothermic. The result of a DSC experiment is a curve of heat flux versus temperature or time. The area enclosed between the trend line and the base line is a direct measurement of the amount of heat, ΔH, needed for transformation. Useful information can be obtained by DSC analysis of polymer samples, such as degree of crystallinity (from the ratio of the heat of fusion of a polymer sample and the enthalpy of a 100% crystalline sample), specific heat, the purity of the polymer and occurrence of oxidation, cross-linking, and chain breakage. As far as water management of electrolytes is concerned, DSC provides information on states of water and water mobility through the electrolyte material. Focusing on ionomer electrolytes, three different categories of water can be discerned by recording DSC thermograms at subzero temperatures:

232 | Part II Development of new materials for energy applications (1) nonfreezable bound water (WNF ), strongly bound to the ionic groups present in the polymer. This type of water is characterized by the fact that it does not crystallize even when the swollen sample is cooled down to −100°C. These water molecules are in close proximity to an ionic group as in hydration shells, are highly polarized, and are unable to crystallize. WNF does not yield characteristic thermal transition in DSC analysis. (2) Freezable bound water, weakly polarized. This type of water crystallizes at temperatures below than 0°C. (3) Freezable unbound water, which crystallizes at 0°C. Freezable water (WF ), being more loosely bound, has higher mobility than nonfreezable water and is expected to give a more significant contribution to the proton transport mechanism. By performing DSC analysis in the range between −50°C and 10°C, freezable water can be quantified from the endothermic peak below 0°C. An example of DSC thermograms obtained from two different polymer electrolyte membranes showing an endothermic peak ascribed to the melting of freezable water is given in Fig. 10.18. –1.0

Heat flow/Wg-1

–0.8 –0.6

(b)

–0.4 –0.2

ENDO (a)

–0.0 –40 –35 –30 –25 –20 –15 –10 –5 T/°C

0

5

10

Fig. 10.18. DSC thermogram of (a) an unfilled Nafion® membrane, and (b) a composite Nafion® / zeolite membrane. Reprinted with permission from [78].

The percentage of freezable water in the sample can obtained from the following formula: 1 A ) × 100, (10.24) WF (%) = ( ΔHW mdry where A is the area of the endothermic peak, ΔHw is the enthalpy of melting for bulk water (333 J g−1 ), and mdry is the mass of the dried sample. The degree of freezable water, θF , can be defined by normalizing the freezable water content to the total WU, which can be measured gravimetrically (for instance by DVS). W θF = F × 100 (10.25) WU

10 Solid polymer proton conducting electrolytes for fuel cells

|

233

As previously mentioned, a higher degree of mobile water corresponds to higher proton conductivity, and since WF yields to thermal transitions similar to bulk water, its content in the membrane can be discerned from total WU by using DSC [72, 78]. Figure 10.19 shows the variation of θF with the filler content for Nafion® –zeolite composite membranes. Zeolites are aluminosilicates with cations relatively free to move along the cavities of the framework. Moreover, they have a very high specific surface area which results in a high water sorption capacity, further facilitating the ion transport. As the zeolite content increases, qF values increase up to a maximum value and then decrease at highest zeolite content. These findings indicate that the zeolite likely contributes to the enhancement of the water mobility degree in the composite membrane, which is related to its high water sorption capacity and to the introduction of porosities at the polymer/filler interface. However, the reduction of this effect over ca. 4 wt% zeolite content, suggests the formation of dead-end porosities which hinder water mobility [78].

55 50

θF /%

45 40 35 30 25 0

5 10 15 Zeolite content/wt.%

20

Fig. 10.19. Variation of θF as a function of the zeolite content for Nafion® –Faujasite composite membranes. Adapted with permission from [78].

10.4 Summary Solid electrolytes are materials capable of conducting ions. They are used in many electrochemical devices including batteries, sensors, electrolyzers, and fuel cells. Proton exchange membrane fuel cells are considered attractive power sources for portable applications, in-situ power generation, and for automotives. Nevertheless, these systems still suffer from limitations which need to be addressed before they can compete with batteries, fossil fuels, and internal combustion engines. Polymer electrolyte membranes are one of the limiting elements of this technology. Nafion® , a perfluorinated sulphonic acid ionomer, is the most widely used electrolyte for both

234 | Part II Development of new materials for energy applications hydrogen and liquid-fed proton exchange membrane fuel cells due to its high proton conductivity, chemical and mechanical stability. A unique feature of Nafion® is the microphase separation between the hydrophobic backbone and the hydrated sulfonic acid domains, resulting in the formation of wide and well-separated water channels for proton transport. Nafion® membranes show a strong dependence of proton conductivity on the membrane’s hydration level and are permeable to liquid fuels. In the first case, the fuel cell system needs an expensive humidification auxiliary system to keep the membranes hydrated. In the second, fuel cell efficiency is dramatically reduced. Therefore the development of alternative polymer electrolyte membranes with high proton conductivity in a wide range of temperature and hydration conditions, which exhibit mechanical robustness, chemical and electrochemical stability, low cost, and low fuel permeability remains a critical challenge for advancing fuel cell technology. Hydrocarbon membranes are potential candidates to replace Nafion® . Significant efforts are being made to develop novel ionomers consisting of hydrocarbon backbones and pending side chains with terminal sulfonic acid groups to mimic Nafion® ’s unique morphology. Structure–properties relationships are fundamental to learn about the dependence of the transport properties on the membranes’ composition, morphology, and water content, and to design better electrolytes. Proton conductivity is a fundamental property of a proton exchange membrane. When evaluating potential electrolytes for fuel cells, their proton conductivity is usually measured under controlled temperature and relative humidity. There is not yet a standard method for measuring the membranes’ proton conductivity, but measurements on both directions of the membrane (in-plane and through-plane) could provide valuable information on the membranes’ anisotropy. Proton conductivity of electrolytes depends on their hydration level; hence it is important to study water sorption and water diffusion over a wide range of relative humidity. The water states and water diffusion in electrolytes can be assessed by dynamic vapor sorption and differential scanning calorimetry. Excellent correlation has been found between proton conductivity and degree of mobile water determined by the two methods mentioned. Advances in proton exchange membranes for fuel cells will likely contribute to the development of other related fields including electrodialysis (water purification and treatment) and redox flow batteries for energy conversion.

Acknowledgments The authors would like to thank Maria J.V.R. Paulo (INRS-EMT) for helping with the editing of this chapter.

10 Solid polymer proton conducting electrolytes for fuel cells

|

235

References [1] [2] [3] [4] [5]

[6] [7] [8] [9]

[10] [11] [12]

[13] [14] [15]

[16] [17]

[18] [19] [20] [21]

Hayashi A, Noi K, Sakuda A, Tatsumisago M. Superionic glass-ceramic electrolytes for roomtemperature rechargeable sodium batteries. Nat Commun. 3 (2012) 856. Tarascon JM, Armand M. Issues and challenges facing rechargeable lithium batteries. Nature. 414 (2001) 359–67. Hickner MA. Ion-containing polymers: new energy & clean water. Materials Today. 13 (2010) 34–41. Carrette L, Friedrich KA, Stimming U. Fuel Cells – Fundamentals and Applications. Fuel Cells. 1 (2001) 5–39. Chroneos A, Yildiz B, Tarancon A, Parfitt D, Kilner JA. Oxygen diffusion in solid oxide fuel cell cathode and electrolyte materials: mechanistic insights from atomistic simulations. Energy & Environmental Science. 4 (2011) 2774–89. Brett DJL, Atkinson A, Brandon NP, Skinner SJ. Intermediate temperature solid oxide fuel cells. Chemical Society Reviews. 37 (2008) 1568–78. Arico AS, Bruce P, Scrosati B, Tarascon J-M, van Schalkwijk W. Nanostructured materials for advanced energy conversion and storage devices. Nat Mater. 4 (2005) 366–77. Goodenough JB. Ceramic solid electrolytes. Solid State Ionics. 94 (1997) 17–25. Passalacqua E, Lufrano F, Squadrito G, Patti A, Giorgi L. Nafion® content in the catalyst layer of polymer electrolyte fuel cells: effects on structure and performance. Electrochimica Acta. 46 (2001) 799–805. Sata T. Ion Exchange Membranes: Preparation, Characterization, Modification and Application, Royal Society of Chemistry; Oxford, UK, 2004. Hickner MA. Water-Mediated Transport in Ion-Containing Polymers. Journal of Polymer Science Part B: Polymer Physics. 50 (2011) 9–20. Schuster M, Rager T, Noda A, Kreuer KD, Maier J. About the Choice of the Protogenic Group in PEM Separator Materials for Intermediate Temperature, Low Humidity Operation: A Critical Comparison of Sulfonic Acid, Phosphonic Acid and Imidazole Functionalized Model Compounds. Fuel Cells. 5 (2005) 355–65. Eikerling M, Kornyshev A, Spohr E. Proton-Conducting Polymer Electrolyte Membranes: Water and Structure in Charge. Advances in Polymer Science. 215 (2008) 15–54. Kreuer K-D. Proton Conductivity: Materials and Applications. Chemistry of Materials. 8 (1996) 610–41. Zawodzinski TA, Neeman M, Sillerud LO, Gottesfeld S. Determination of water diffusion coefficients in perfluorosulfonate ionomeric membranes. The Journal of Physical Chemistry. 95 (1991) 6040–4. Eikerling M, Kornyshev AA. Proton transfer in a single pore of a polymer electrolyte membrane. Journal of Electroanalytical Chemistry. 502 (2001) 1–14. Choi P, Jalani NH, Datta R. Thermodynamics and Proton Transport in Nafion® : II. Proton Diffusion Mechanisms and Conductivity. Journal of The Electrochemical Society. 152 (2005) E123–E30. Alberti G, Casciola M. Solid state protonic conductors, present main applications and future prospects. Solid State Ionics. 145 (2001) 3–16. McLean RS, Doyle M, Sauer BB. High-Resolution Imaging of Ionic Domains and Crystal Morphology in Ionomers Using AFM Techniques. Macromolecules.33 (2000) 6541–50. Mauritz K A, Hora C J, Hopfinger A J. Theoretical Model for the Structures of Ionomers: Application to Nafion® Materials. Ions in Polymers: American Chemical Society. 1980: 123–44. Yeager HL, Steck A. Cation and Water Diffusion in Nafion® Ion Exchange Membranes: Influence of Polymer Structure. Journal of The Electrochemical Society. 128 (1981) 1880–4.

236 | Part II Development of new materials for energy applications [22] [23] [24]

[25]

[26] [27] [28] [29] [30] [31] [32] [33] [34]

[35] [36]

[37]

[38]

[39] [40] [41] [42]

Hsu WY, Gierke TD. Elastic theory for ionic clustering in perfluorinated ionomers. Macromolecules. 15 (1982) 101–5. Fimrite J, Struchtrup H, Djilali N. Transport Phenomena in Polymer Electrolyte Membranes: I. Modeling Framework. Journal of The Electrochemical Society. 152 (2005) A1804–A14. Elliott JA, Hanna S, Elliott AMS, Cooley GE. Atomistic simulation and molecular dynamics of model systems for perfluorinated ionomer membranes. Physical Chemistry Chemical Physics.1 (1999) 4855–63. Li Q, He R, Jensen JO, Bjerrum NJ. Approaches and Recent Development of Polymer Electrolyte Membranes for Fuel Cells Operating above 100°C. Chemistry of Materials. 15 (2003) 4896–915. Casciola M, Alberti G, Sganappa M, Narducci R. On the decay of Nafion® proton conductivity at high temperature and relative humidity. Journal of Power Sources. 162 (2006) 141–5. Kreuer KD. On the development of proton conducting polymer membranes for hydrogen and methanol fuel cells. Journal of Membrane Science. 185 (2001) 29–39. Peighambardoust SJ, Rowshanzamir S, Amjadi M. Review of the proton exchange membranes for fuel cell applications. International Journal of Hydrogen Energy. 35 (2010) 9349–84. Zhang H, Shen PK. Recent Development of Polymer Electrolyte Membranes for Fuel Cells. Chemical Reviews. 112 (2012) 2780–832. Di Noto V, Zawodzinski TA, Herring AM, Giffin GA, Negro E, Lavina S. Polymer electrolytes for a hydrogen economy. International Journal of Hydrogen Energy. 37 (2012) 6120–31. Peckham TJ, Holdcroft S. Structure-Morphology-Property Relationships of Non-Perfluorinated Proton-Conducting Membranes. Advanced Materials. 22 (2010) 4667–90. Alkan Gürsel S, Gubler L, Gupta B, Scherer G. Radiation Grafted Membranes. Advances in Polymer Science. 215 (2008) 157–217. Yang Y, Siu A, Peckham T, Holdcroft S. Structural and Morphological Features of Acid-Bearing Polymers for PEM Fuel Cells. Advances in Polymer Science. 215 (2008) 55–126. Wang F, Hickner M, Kim YS, Zawodzinski TA, McGrath JE. Direct polymerization of sulfonated poly(arylene ether sulfone) random (statistical) copolymers: candidates for new proton exchange membranes. Journal of Membrane Science. 197 (2002) 231–42. Jannasch P. Recent developments in high-temperature proton conducting polymer electrolyte membranes. Current Opinion in Colloid & Interface Science. 8 (2003) 96–102. Harrison WL, Hickner MA, Kim YS, McGrath JE. Poly(Arylene Ether Sulfone) Copolymers and Related Systems from Disulfonated Monomer Building Blocks: Synthesis, Characterization, and Performance – A Topical Review. Fuel Cells. 5 (2005) 201–12. Daigle JC, Dube-Savoie V, Tavares AC, Claverie JP. Copolymers of ethylene and sulfonated norbornene for proton exchange membranes. Journal of Polymer Science Part A: Polymer Chemistry. 51 (2013) 2669–76. Zaopo A, Lopes Correia Tavares AB, Ballabio O, inventors; PIRELLI & C SPA, Italy, assignee. Fuel Cell and Polymer Electrolyte Membrane ; (2007) WO2007128330 A1; Application date: 2006-05-05; EP 2025031 B1 (2011). Herring AM. Inorganic–Polymer Composite Membranes for Proton Exchange Membrane Fuel Cells. Journal of Macromolecular Science, Part C. 46 (2006) 245–96. Laberty-Robert C, Valle K, Pereira F, Sanchez C. Design and properties of functional hybrid organic-inorganic membranes for fuel cells. Chemical Society Reviews. 40 (2011) 961–1005. Fu T, Cui Z, Zhong S, et al. Sulfonated poly(ether ether ketone)/clay-SO3H hybrid proton exchange membranes for direct methanol fuel cells. Journal of Power Sources. 185 (2008) 32–9. Kerres J. Covalent-Ionically Cross-linked Poly(Etheretherketone)-Basic Polysulfone Blend Ionomer Membranes. Fuel Cells. 6 (2006) 251–60.

10 Solid polymer proton conducting electrolytes for fuel cells

[43] [44]

[45]

[46] [47]

[48] [49]

[50]

[51] [52] [53]

[54]

[55] [56] [57]

[58] [59]

[60] [61]

|

237

Lafitte B, Karlsson LE, Jannasch P. Sulfophenylation of Polysulfones for Proton-Conducting Fuel Cell Membranes. Macromolecular Rapid Communications. 23 (2002) 896–900. Tavares AC, Pedicini R, Gatto I, Dubitsky YA, Zaopo A, Passalacqua E. New Sulfonated Polysulfone Co-Polymer Membrane for Low Temperature Fuel Cells. Journal of New Materials for Electrochemical Systems. 6 (2003) 211–6. de Bonis C, D’Epifanio A, Di Vona ML, et al. Proton Conducting Hybrid Membranes Based on Aromatic Polymers Blends for Direct Methanol Fuel Cell Applications. Fuel Cells. 9 (2009) 387–93. Rozière J, Jones DJ. Non-fluorinated polymer materials for proton exchange membrane fuel cells. Annual Review of Materials Research. 33 (2003) 503–55. Kreuer K-D, Paddison SJ, Spohr E, Schuster M. Transport in Proton Conductors for Fuel-Cell Applications: Simulations, Elementary Reactions, and Phenomenology. Chemical Reviews. 104 (2004) 4637–78. Zhang W, Gogel V, Friedrich KA, Kerres J. Novel covalently cross-linked poly(etheretherketone) ionomer membranes. Journal of Power Sources. 155 (2006) 3–12. Karlsson LE, Jannasch P. Polysulfone ionomers for proton-conducting fuel cell membranes: 2. Sulfophenylated polysulfones and polyphenylsulfones. Electrochimica Acta. 50 (2005) 1939–46. Pang J, Zhang H, Li X, Jiang Z. Novel Wholly Aromatic Sulfonated Poly(arylene ether) Copolymers Containing Sulfonic Acid Groups on the Pendants for Proton Exchange Membrane Materials. Macromolecules. 40 (2007) 9435–42. Lafitte B, Jannasch P. Proton-Conducting Aromatic Polymers Carrying Hypersulfonated Side Chains for Fuel Cell Applications. Advanced Functional Materials. 17 (2007) 2823–34. Markova D, Kumar A, Klapper M, Müllen K. Phosphonic acid-containing homo-, AB and BAB block copolymers via ATRP designed for fuel cell applications. Polymer. 50 (2009) 3411–21. Lee H-S, Badami AS, Roy A, McGrath JE. Segmented sulfonated poly(arylene ether sulfone)b-polyimide copolymers for proton exchange membrane fuel cells. I. Copolymer synthesis and fundamental properties. Journal of Polymer Science Part A: Polymer Chemistry. 45 (2007) 4879–90. Nakabayashi K, Matsumoto K, Ueda M. Synthesis and properties of sulfonated multiblock copoly(ether sulfone)s by a chain extender. Journal of Polymer Science Part A: Polymer Chemistry. 46 (2008) 3947–57. Hickner MA, Ghassemi H, Kim YS, Einsla BR, McGrath JE. Alternative Polymer Systems for Proton Exchange Membranes (PEMs). Chemical Reviews. 104 (2004) 4587–612. Ghassemi H, McGrath JE, Zawodzinski Jr TA. Multiblock sulfonated–fluorinated poly(arylene ether)s for a proton exchange membrane fuel cell. Polymer. 47 (2006) 4132–9. Park MJ, Kim S, Minor AM, Hexemer A, Balsara NP. Control of Domain Orientation in Block Copolymer Electrolyte Membranes at the Interface with Humid Air. Advanced Materials. 21 (2009) 203–8. Ulbricht M. Advanced functional polymer membranes. Polymer. 47 (2006) 2217–62. Mustarelli P, Quartarone E, Grandi S, Carollo A, Magistris A. Polybenzimidazole-Based Membranes as a Real Alternative to Nafion® for Fuel Cells Operating at Low Temperature. Advanced Materials. 20 (2008) 1339–43. Schuster MFH, Meyer WH. Anhydrous Proton-Conducting Polymers. Annual Review of Materials Research 33 (2003) 233–61. Herz HG, Kreuer KD, Maier J, Scharfenberger G, Schuster MFH, Meyer WH. New fully polymeric proton solvents with high proton mobility. Electrochimica Acta. 48( 2003) 2165–71.

238 | Part II Development of new materials for energy applications [62]

[63] [64]

[65]

[66]

[67] [68] [69] [70] [71] [72]

[73] [74]

[75] [76]

[77]

[78]

[79]

[80]

Goward GR, Schuster MFH, Sebastiani D, Schnell I, Spiess HW. High-Resolution Solid-State NMR Studies of Imidazole-Based Proton Conductors: Structure Motifs and Chemical Exchange from 1H NMR. The Journal of Physical Chemistry B. 106 (2002) 9322–34. Persson JC, Jannasch P. Intrinsically Proton-Conducting Benzimidazole Units Tethered to Polysiloxanes. Macromolecules. 38 (2005) 3283–9. Schuster MFH, Meyer WH, Schuster M, Kreuer KD. Toward a New Type of Anhydrous Organic Proton Conductor Based on Immobilized Imidazole. Chemistry of Materials. 16 (2003) 329–37. Granados-Focil S, Woudenberg RC, Yavuzcetin O, Tuominen MT, Coughlin EB. Water-Free Proton-Conducting Polysiloxanes: A Study on the Effect of Heterocycle Structure. Macromolecules. 40 (2007) 8708–13. Che Q, He R, Yang J, Feng L, Savinell RF. Phosphoric acid doped high temperature proton exchange membranes based on sulfonated polyetheretherketone incorporated with ionic liquids. Electrochemistry Communications. 12 (2010) 647–9. Fernicola A, Navarra M, Panero S. Aprotic ionic liquids as electrolyte components in protonic membranes. J Appl Electrochem. 38 (2008) 993–6. Sekhon SS, Krishnan P, Singh B, Yamada K, Kim CS. Proton conducting membrane containing room temperature ionic liquid. Electrochimica Acta. 52 (2006) 1639–44. Alberti G, Casciola M. Composite Membranes for Medium-Temperature PEM Fuel Cells Annual Review of Materials Research. 33 (2003) 129–54. Jones D, Rozière J. Advances in the Development of Inorganic–Organic Membranes for Fuel Cell Applications. Advances in Polymer Science. 215 (2008) 219–64. de Bonis C, D’Epifanio A, Mecheri B, et al. Layered tetratitanate intercalating sulfanilic acid for organic/inorganic proton conductors. Solid State Ionics. 227 (2012) 73–9. Zhang Z, Désilets F, Felice V, Mecheri B, Licoccia S, Tavares AC. On the proton conductivity of Nafion® –Faujasite composite membranes for low temperature direct methanol fuel cells. Journal of Power Sources. 196 (2011) 9176–87. Chen Z, Holmberg B, Li W, et al. Nafion® /Zeolite Nanocomposite Membrane by in Situ Crystallization for a Direct Methanol Fuel Cell. Chemistry of Materials. 18 (2006) 5669–75. Wang Y, Chen KS, Mishler J, Cho SC, Adroher XC. A review of polymer electrolyte membrane fuel cells: Technology, applications, and needs on fundamental research. Applied Energy. 88 (2011) 981–1007. Chen F, Mecheri B, D’Epifanio A, Traversa E, Licoccia S. Development of Nafion® /Tin Oxide Composite MEA for DMFC Applications. Fuel Cells. 10 (2010) 790–7. Matos BR, Isidoro RA, Santiago EI, et al. In Situ Fabrication of Nafion –Titanate Hybrid Electrolytes for High-Temperature Direct Ethanol Fuel Cell. The Journal of Physical Chemistry C. 117 (2013) 16863–70. Aricò AS, Baglio V, Di Blasi A, Creti P, Antonucci PL, Antonucci V. Influence of the acid–base characteristics of inorganic fillers on the high temperature performance of composite membranes in direct methanol fuel cells. Solid State Ionics. 161 (2003) 251–65. Mecheri B, Felice V, Zhang Z, D’Epifanio A, Licoccia S, Tavares AC. DSC and DVS Investigation of Water Mobility in Nafion® /Zeolite Composite Membranes for Fuel Cell Applications. The Journal of Physical Chemistry C. 116 (2012) 20820–9. Alberti G, Casciola M, Pica M, Tarpanelli T, Sganappa M. New Preparation Methods for Composite Membranes for Medium Temperature Fuel Cells Based on Precursor Solutions of Insoluble Inorganic Compounds. Fuel Cells. 5 (2005) 366–74. Rhee CH, Kim Y, Lee JS, Kim HK, Chang H. Nanocomposite membranes of surface-sulfonated titanate and Nafion® for direct methanol fuel cells. Journal of Power Sources 159 (2006) 1015–24.

10 Solid polymer proton conducting electrolytes for fuel cells

[81]

|

239

Thomassin J-M, Pagnoulle C, Bizzari D, Caldarella G, Germain A, Jérôme R. Improvement of the barrier properties of Nafion® by fluoro-modified montmorillonite. Solid State Ionics 177 (2006) 1137–44. [82] Marani D, D’Epifanio A, Traversa E, Miyayama M, Licoccia S. Titania Nanosheets (TNS)/Sulfonated Poly Ether Ether Ketone (SPEEK) Nanocomposite Proton Exchange Membranes for Fuel Cells. Chemistry of Materials 22 (2009) 1126–33. [83] Mecheri B, Felice V, D’Epifanio A, Tavares AC, Licoccia S. Composite Polymer Electrolytes for Fuel Cell Applications: Filler-Induced Effect on Water Sorption and Transport Properties. ChemPhysChem. 14 (2013) 3814–21. [84] Mecheri B, D’Epifanio A, Pisani L, et al. Effect of a Proton Conducting Filler on the PhysicoChemical Properties of SPEEK-Based Membranes. Fuel Cells. 9 (2009) 372–80. [85] Hamoudi S, Royer S, Kaliaguine S. Propyl- and arene-sulfonic acid functionalized periodic mesoporous organosilicas. Microporous and Mesoporous Materials. 71 (2004) 17–25. [86] Felice V, Ntais S, Tavares AC. Propyl sulfonic acid functionalization of faujasite-type zeolites: Effect on water and methanol sorption and on proton conductivity. Microporous and Mesoporous Materials. 169 (2013) 128–36. [87] Cozzi D, de Bonis C, D’Epifanio A, Mecheri B, Tavares AC, Licoccia S. Organically functionalized titanium oxide/Nafion® composite proton exchange membranes for fuel cells applications. Journal of Power Sources. 248 (2014) 1127–32. [88] Peckham TJ, Schmeisser J, Rodgers M, Holdcroft S. Main-chain, statistically sulfonated proton exchange membranes: the relationships of acid concentration and proton mobility to water content and their effect upon proton conductivity. Journal of Materials Chemistry. 17 (2007) 3255–68. [89] Dupuis A-C. Proton exchange membranes for fuel cells operated at medium temperatures: Materials and experimental techniques. Progress in Materials Science. 56 (2011) 289-327. [90] Yang Y, Holdcroft S. Synthetic Strategies for Controlling the Morphology of Proton Conducting Polymer Membranes. Fuel Cells. 5 (2005) 171–86. [91] Hou H, Di Vona ML, Knauth P. Durability of Sulfonated Aromatic Polymers for ProtonExchange-Membrane Fuel Cells. Chem Sus Chem. 4 (2011) 1526–36. [92] Barsoukov E, Macdonald JR. Impedance Spectroscopy. Theory Experiment and Applications John Wiley & Sons, Hoboken NJ, 2005. [93] Sone Y, Ekdunge P, Simonsson D. Proton Conductivity of Nafion® 117 as Measured by a FourElectrode AC Impedance Method. Journal of The Electrochemical Society. 143 (1996) 1254–9. [94] Slade S, Campbell SA, Ralph TR, Walsh FC. Ionic Conductivity of an Extruded Nafion® 1100 EW Series of Membranes. Journal of The Electrochemical Society. 149 (2002) A1556–A64. [95] Soboleva T, Xie Z, Shi Z, Tsang E, Navessin T, Holdcroft S. Investigation of the through-plane impedance technique for evaluation of anisotropy of proton conducting polymer membranes. Journal of Electroanalytical Chemistry. 622 (2008) 145–52. [96] Cooper K. Characterizing Through-Plane and In-Plane Ionic Conductivity of Polymer Electrolyte Membranes. ECS Transactions. 41 (2011) 1371–80. [97] Shuhua Ma ZS, Hirokazu Tanakaa. Anisotropic Conductivity Over In-Plane and Thickness Directions in Nafion® -117. Journal of The Electrochemical Society. 153(12), (2006) A2274–81 . [98] Lee CH, Park HB, Lee YM, Lee RD. Importance of Proton Conductivity Measurement in Polymer Electrolyte Membrane for Fuel Cell Application. Industrial & Engineering Chemistry Research. 44 (2005) 7617–26. [99] Sing KSW, Everett DH, Haul RAW, et al. Reporting Physisorption Data for Gas-Solid systems. Pure and Applied Chemistry. 57 (1985) 603–19. [100] Ng E-P, Mintova S. Nanoporous materials with enhanced hydrophilicity and high water sorption capacity. Microporous and Mesoporous Materials. 114 (2008) 1–26.

240 | Part II Development of new materials for energy applications [101] Crank J, Park GS, (Eds). Diffusion in polymers, Academic Press, London New York, 1968. [102] Morris DR, Sun X. Water-sorption and transport properties of Nafion® 117 H. Journal of Applied Polymer Science. 50 (1993) 1445–52. [103] Takamatsu T, Hashiyama M, Eisenberg A. Sorption phenomena in nafion membranes. Journal of Applied Polymer Science. 24 (1979) 2199–220. [104] Foo KY, Hameed BH. Insights into the modeling of adsorption isotherm systems. Chemical Engineering Journal. 156 (2010) 2–10. [105] Park GS. Synthetic Membranes: Science, Engineering and Applications. Dordrecht, The Netherlands: Reidel Publishing Company; 1986. [106] Takata H, Mizuno N, Nishikawa M, Fukada S, Yoshitake M. Adsorption properties of water vapor on sulfonated perfluoropolymer membranes. International Journal of Hydrogen Energy. 32 (2007) 371–9. [107] Feng H. Modeling of vapor sorption in glassy polymers using a new dual mode sorption model based on multilayer sorption theory. Polymer. 48 (2007) 2988–3002. [108] Guggenheim EA. Applications of Statistical Mechanics. Oxford, U.K: Clarendon Press; 1966. [109] Anderson RB. Modifications of the Brunauer, Emmett and Teller Equation1. Journal of the American Chemical Society. 68 (1946) 686–91. [110] de Boer JH. The Dynamical Character of Adsorption. Oxford, U.K.: Clarendon Press; 1968. [111] Li Y, Nguyen QT, Buquet CL, Langevin D, Legras M, Marais S. Water sorption in Nafion® membranes analyzed with an improved dual-mode sorption model—Structure/property relationships. Journal of Membrane Science. 439 (2013) 1–11. [112] Osswald T, Menges G. Materials Science of Polymers for Engineers. 2nd ed. Cincinnati, USA: Hanser Gardner Publications, Inc.; 2003. [113] Höhne G, Hemminger WF, Flammersheim HJ. Differential Scanning Calorimetry. 2nd ed. Berlin, Heidelberg Germany: Springer; 2003. [114] Wunderlich B. Thermal Analysis of Polymeric Materials. Berlin, Germany: Springer; 2005.

P. Bénard, A.-M. Beaulieu, D. Durette, and R. Chahine

11 Supercritical adsorption of hydrogen on microporous adsorbents 11.1 Introduction Physical adsorption on microporous adsorbents is widely used in industrial physicochemical processes such as gas separation and purification [1, 2]. The adsorption phenomena can be defined as the enrichment or depletion of a species of fluid in an interfacial layer between the fluid and a substrate [3]. The absorption process, in contrast, occurs when the species is integrated into the structure of the substrate. Chemisorption typically involves both phenomena. Physisorption (or physical adsorption) is associated with weak and nonspecific interactions, typically van der Waals forces. Physisorption does not involve the hybridization of electronic orbitals. The distinction between physisorption and chemisorption can be subtle. Typically, adsorption processes with a characteristic energy scale smaller than 15 kJ/mol are considered to be physical. The adsorption process occurs within the porous structure of the adsorbent, which is the volume of the adsorbent externally accessible to the adsorbate molecules. The pores are formally classified by IUPAC as micropores (characteristic size < 2 nm), mesopores (2–50 nm) and macropores (characteristic size > 50 nm) [4]. Gas storage applications of the adsorption phenomena usually require maximizing the micropore volume of an adsorbent. The adsorption process can be used to store light gases at substantially lower pressures, offering the possibility of safer storage conditions with respect to high pressure compression. The storage density of gases such as hydrogen and methane can thus be increased significantly through the interatomic and intermolecular forces between the adsorbate molecules (adatoms) and a solid surface. Materials-based storage of gases such as hydrogen can also be achieved through chemical binding to other elements (chemical hydrides) or through absorption inside a solid metallic matrix (metal hydrides). For materials-based storage, the characteristic binding energy of hydrogen to the material is a determining factor in the thermodynamic description of systems. It sets the energy scale required for charging and discharging gases, regulates its thermal behavior, and determines boil-off rates. Materials with binding energies smaller than 10 kJ/mol require cryogenic operation to achieve acceptable storage densities. Passive heat sources from the environment can then be used to thermally manage outgassing. Low binding energies occur when hydrogen is reversibly bound in molecular form to a substrate through weak, nonspecific interactions. The hard-to-achieve range of 10–20 kJ/mol is believed to be optimal for materials-based hydrogen storage applications closer to room temperature.

242 | Part II Development of new materials for energy applications

11.2 Fundamentals of supercritical adsorption Weak interactions between a gas and a substrate usually result in the adsorption of gas molecules (the adsorbate) onto the surface of the substrate (the adsorbent). The adsorption effect corresponds to a local enhancement of the density of the adsorbate close to the adsorbent’s surface resulting from those interactions, in the region where the adsorbate/adsorbent interaction potential is significant (Fig. 11.1). The total adsorbed density nt is defined as the number of adsorbate molecules present in the porous structure of the adsorbate. The adsorbed density of the adsorbate is usually divided by the quantity of adsorbent, either volumetrically (quantity of adsorbate molecules per unit volume of adsorbent) or gravimetrically (per unit mass of the adsorbent). Both are related to one another by the density of the adsorbent. The density profile of the gas close to the surface can be separated into 3 regions: (1) the solid phase, (2) the interface, and (3) the gas phase with volumes Vs , Va, and Vg , respectively. The volume of the solid phase (Vs ) is also called the volume of the skeleton, as it excludes the contribution of the pores. The total volume of the adsorbate is

V

d

(a) Gas

Solid

Interface ntot

ng 0 d

(b)

Solid

ntot Gibbs surface

ng Gas 0 (c)

X0

d

Fig. 11.1. (a) Interaction potential between an adsorbate molecule and the surface of an adsorbent. (b) Typical density profile of a gas close to an adsorbing surface. The enhanced density in the vicinity of the adsorbent is caused by gas-surface interactions. The boundary indicated on the figure depends on the definition of a cut-off for the interaction potential. (c) Gibbs dividing surface: the adsorbed atoms are smeared onto the solid. The dividing surface is an ideal construct separating the adsorbent and the adsorbate gas.

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 243

the sum of the three contributions. The adsorbate is usually described as belonging to one of two phases: the gaseous phase (where the interaction potential is negligible) or the adsorbed phase located in the interface. The excess adsorbed density (na ) is defined as the number of molecules in a porous structure in excess of the number that would be present in the same volume under the same thermodynamic conditions in the absence of adsorbate-adsorbent interactions. It can be defined as follows: n a = n t − n g Vp . The pore volume (Vp ) is defined as the sum of the volume of the gas phase in the adsorbent and of the volume of the interface. It is clear from Fig. 11.1 that the distinction between adsorbed molecules and the bulk gas is ambiguous for long-range interaction potentials, which do not have a clear cut-off. The boundary between the phases can be conventionally defined using a reference gas that is assumed to be nonadsorptive under ambient conditions. In Gibb’s formulation of adsorption, this is done through the Gibbs dividing surface. The Gibbs surface is chosen in such a way that the excess density of the reference gas (typically helium) is zero under ambient conditions: na (T0 , P0 ) = nt − ng Vp = 0. This equation defines Vp and the location of the dividing surface. The pore volume (Vp ) thus formally depends on the reference gas and the thermodynamic conditions under which it is calculated. In the Gibbs approach, the gas phase density is assumed to be constant and equal to the bulk phase up to the Gibbs dividing surface [5]. The adsorbed gas molecules (sometimes referred to as the adatoms) are assumed to be located on the surface. The adsorbate phase has no volume (Va = 0). When considered as an interface problem, the thermodynamic description of adsorption requires the introduction of the surface area of the interface as a thermodynamic variable [6], coupled to a conjugate force called the spreading pressure. It is coupled to the interfacial surface area in the free energy equations of the system. The spreading pressure is not directly accessible and acts as an intermediate calculation variable. For highly microporous adsorbents (∼ 1.5 nm), the pore size is usually such that from a microscopic point of view, an adatom is affected by adsorbent-adsorbate interactions throughout the pore structure of the adsorbent. As a result, the local adsorbate density remains larger than the free gas value in the pore volume. A macroscopic description of the adsorption process based on interface properties then becomes counterintuitive, as a zero-volume interface has no simple microscopic correspondence. From a macroscopic point of view, however, a consistent thermodynamic description of the adsorption phenomena in microporous adsorbents based on interface thermodynamic variables can still be used, as the latter are intermediate calculation variables not directly accessible experimentally [6].

244 | Part II Development of new materials for energy applications The adsorption isotherm (na (T, P)) expresses the relationship between the excess density defined in the Gibbs sense and the pressure of the bulk gas phase with which it is in equilibrium for a given temperature. It represents the difference between the mass of adsorbate in the porous structure of the adsorbate and the amount that would be present in the same volume in the absence of the adsorbate. The Gibbs adsorption isotherm is also called the excess adsorption isotherm and can be seen as the gain in density over compression at the equilibrium pressure and temperature. The absolute adsorption isotherm refers to the total amount of hydrogen present within the potential field of the adsorbent. Obtaining the absolute adsorption isotherm from the excess requires knowledge of the volume of the adsorbed phase, which is not directly accessible experimentally and depends very much on the definition of a cut-off of the adsorbate/adsorbent interaction potential. For highly microporous adsorbents, in which most of the volume of the pores is subject to the interaction potential, the absolute adsorption isotherm corresponds to the total uptake of adsorbate inside the pores. However, if macropores and mesopores are also present, adsorbate molecules far from their surfaces should be excluded. An unambiguous definition of the absolute adsorption isotherm can therefore be difficult, particularly for adsorbents with a complex pore size distribution. The excess adsorption isotherm represents, however, a clear and experimentally accessible concept. The classification of adsorption isotherms by the International Union of Pure and Applied Chemistry (IUPAC) consists of 6 categories [7]. Types I–III are reversible isotherms which do not exhibit hysteresis. The Type I isotherm is concave with respect to the pressure axis. It is usually associated with microporous solids (e.g., activated carbons, molecular sieve zeolites, and certain porous oxides). The Type II isotherm represents unrestricted monolayer-multilayer adsorption occurring in a macroporous adsorbent. Point B in Fig. 11.2 indicates where multilayer adsorption begins to take place. Type III isotherms are uncommon and characterized by a convex curve with respect to the pressure axis. Type IV and V isotherms exhibit a hysteresis loop. Type IV isotherms are associated with capillary condensation taking place in mesopores. Type V isotherms are related to Type III, with a hysteresis loop as an additional feature. Type VI isotherms are associated with stepwise multilayer adsorption on a uniform nonporous surface. The height of each step is associated with the monolayer capacity for each adsorbed layer. Only adsorption processes associated with Type I isotherms will be addressed in this chapter. We will only consider supercritical adsorption, which occurs above the critical point of the adsorbate. We will thus limit ourselves to physical adsorption (or physisorption) processes, associated with relatively weak adsorbate-adsorbent interactions, and to the study of single component adsorbates, as the focus of this paper is gas storage. As discussed in the introduction, the adsorption phenomena, through gas surface interactions, offers the possibility of substantially lowering the storage pressure of gases [1, 2]. The presence of an interaction potential with a characteristic binding

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 245

energy (EA ) permeating narrow pores can substantially enhance the equilibrium densities of the adsorbed gas inside the pore, compared to the far-field bulk gas, resulting in very high local pressures in the pores of the adsorbing nanostructure. The resulting fluid densities of the adsorbate in the pores can become comparable and sometimes even exceed its liquid density, depending on the pressure and temperature of its bulk phase. In addition to the characteristic energy (EA ), the surface area available for adsorption processes and the bulk density of the adsorbent are important parameters for adsorption-based applications. The specific surface is an important contributor to the saturation density of an adsorbent, whereas the bulk density determines the weight of the storage unit. Both parameters can be combined into the surface available to the adsorbate per unit volume of the adsorbent, which should be maximized to optimize the storage density. This quantity can be expressed in terms of adsorbate mass/adsorbent volume, or adsorbate volume/adsorbent volume, the latter yielding the former when multiplied by the bulk gas density at 298 K and 1 atmosphere of the adsorbate.

I

II

B

Amount odsorbed

III

IV

B

V

VI

Relative pressure

Fig. 11.2. Types of adsorption isotherms. Reprinted with permission from [7].

246 | Part II Development of new materials for energy applications

11.3 Supercritical adsorption isotherms The adsorption isotherm is the equation of the state of the adsorption process and as such, plays a key role in its thermodynamic description. Predictive strategies involve first principles calculations using computer simulations and numerical analysis of its properties in specific limits. Such approaches include ab initio quantum chemistry calculations, grand canonical Monte Carlo simulations (with or without quantum corrections), the virial expansion, and to some extent classical density functional theory. Descriptive approaches aim to describe the isotherms using adsorption models (analytic or not) appropriate for the adsorbent structure and the adsorbate-adsorbent interactions. These rely on approximations which invariably limit their applicability to specific ranges of pressure and temperature.

11.3.1 Virial expansion of the excess density in terms of pressure In the low pressure limit, the excess adsorbed density can be expressed as a virial series expansion in terms of pressure [8]: na (p, T) = BAS

p p2 p3 p4 + CAAS + DAAS + ..., + BAAS 2 3 kT (kT) (kT) (kT)4

(11.1)

where p is the pressure of the gas phase. At the lowest order in pressure, the excessadsorbed density varies linearly with pressure, which corresponds to Henry’s law as applied to the adsorption phenomena. The coefficient BAS , generally called the second virial coefficient, is fully determined by the interaction U (r)⃗ between a single molecule and the surface of the adsorbent: U (r)⃗ 1 BAS (T) = ∫ (exp (− ) − 1) dr,⃗ (11.2) M kT where M is the mass of the adsorbent. The virial coefficient BAS can be obtained experimentally by finding the intercept of a plot of na (p, T) kT/p as a function of the pressure (p). The interaction potential V(r)⃗ between an adsorbate molecule and the adsorbent atoms can be written as: U(r)⃗ = ∑ Uαi (r ⃗ − R⃗ i ),

(11.3)

i

where r ⃗ and R⃗ i are the positions of the adsorbate molecule and the ith atom of the adsorbent, and where αi refers to the atomic species of the atom at R⃗ i . The interaction potential U(r)⃗ is often modeled using the Lennard-Jones potential: 6 12 Uαi (r)⃗ = 4ε [(σ/r) − (σ/r) ] ,

where σ is the distance where V(r)⃗ = 0 and 𝜖 is the minimum interaction energy.

(11.4)

247

11 Supercritical adsorption of hydrogen on microporous adsorbents |

Effective adsorbate-adsorbent potentials based on the Lennard-Jones potential have been developed to model the interaction of a crystalline adsorbent surface with adsorbate molecules. The planar Lennard-Jones potential, for instance, represents an averaged interaction potential resulting from smearing the atoms of a corrugated planar crystal into a uniform, continuous distribution onto an infinite two-dimensional plane. The 10-4 Lennard-Jones planar potential is given by the: V (y) = 2εs (

2 1 1 − ), 5 y10 y4

(11.5)

where εs = πθεσ2 . The parameter θ is the surface density of a graphene plane (0.38 atoms per Å [2]). The variable (y) is dimensionless and rescaled to σ (y = z/σ) and represents the distance (z) along the normal to the plane. The parameters ε = 30.5 K and σ = 3.19 Å represent, respectively, the Lennard-Jones energy and distance parameters between carbon atoms and hydrogen. The pores of an activated carbon are typically modeled using the so-called slit pore approach, in which they consist of two parallel smeared infinite graphene layers separated by the characteristic pore size (d). The resulting interaction potential is shown in Fig. 11.3. As a function of decreasing distance between the graphene layers, the well depth of the potential gets deeper as the overlap of the contributions from each plane increases. Ultimately, the two minima overlap completely, resulting in a single well. Further decreasing the distance between the planes will decrease the depth of the well, until the interaction potential in the pores becomes positive everywhere, quenching the adsorption process in the pore. V є

0.5 z –2

–1

1

2

σ

–0.5

–1.0

Fig. 11.3. The slit pore potential for activated carbons. A pore is modeled by two infinite parallel graphene layers smeared into planes, separated by a distance assumed to be representative of the size of the pore resulting in the above adsorbate–adsorbent interaction potential within the slit pore.

248 | Part II Development of new materials for energy applications Assuming an average noncorrugated planar interaction potential between the adsorbent planes and an adsorbate molecule, the second virial coefficient is then given by: d∗ /2

−1 ∗ ∗ BAS = 2 ∫ [e kT [V(y+d /2)+V(y−d /2)] − 1] dy, SA σ

(11.6)

0

where d∗ = d/σ, y is a dimensonless integration variable, and SA is the specific surface, which is the area of the adsorbent accessible to adsorbate molecules per unit mass of the adsorbent. The second virial coefficient BAS for the activated carbon AX-21 is shown in Fig. 11.4. 12 10

In (B AS/αZo )

8 6 4 2 0 –2 0

2

4

6

8

10

12

14

X = ɛs/kT Fig. 11.4. Normalized second virial coefficient as a function of the rescaled inverse temperature. The line shows the theoretical curve. The black and white dots show experimental data points from the adsorption of methane and hydrogen (respectively) on the activated carbon AX-21. The normalized curve is expected to be universal for nonpolar classical gases. Reprinted with permission from [12].

The distance (d) and the specific surface of the activated carbon are treated as adjustable parameters, the values d = 8.1 Å for the interlayer distance and 2 900 m2 /Å for the specific surface area for hydrogen sorption on the activated carbon AX-21 are obtained. Although the estimated specific surface is close to the experimentally determined BET value (2 800 m2 /Å), DFT analysis of the adsorption isotherm suggests a pore size distribution peaked around 12.5 Å. The second virial coefficient of single-wall nanotubes has also been studied using a cylindrical potential in the continuum approximation [9]: V∗ (r∗ , R∗ ) = 3 [

21 1 10 1 4 ( ∗ ) M11 (r∗ /R∗ ) − ( ∗ ) M5 (r∗ /R∗ )] , 32 R R

(11.7)

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 249

where r∗ = r/σ, R∗ = R/σ, and V∗ (r∗ , R∗ ) = V(r∗ σ, R∗ σ)/εs are respectively the reduced distance from the axis, the reduced radius of the SWNT, and the reduced potential, with π dφ Mn (x) = ∫ . (11.8) 2 − 2x cos φ)n/2 (1 + x 0 The interaction potential inside a nanotube is shown in Fig. 11.5. 10

0

V/ε

–10

–20

–30

–40 0.0

0.2

0.4 r/R

0.6

0.8

Fig. 11.5. Potential of Stan and Cole for SWNT of various diameters R and helium, a function of the distance from the center, divided by R. When the radius R is smaller than a critical value Rc = 1.212 σ, the minimum is at the center. Further decreasing the radius leads to a shallower potential well, and eventually a positive repulsive core [9].

Stan and Cole [9, 10] performed a study of the virial coefficient BAS for the adsorption of rare gases on single wall nanotubes as a function of the radius using the cylindrical Lennard-Jones potential. The adsorbed density of Ne atoms on an SWNT was compared to that obtained using a planar graphene sheet with the same surface area. They found a significantly larger adsorbed density inside the SWNT compared to the graphene sheet due to the curvature of the nanotubes. An estimate of the maximum binding energy of adsorbate molecules in the cylindrical pore can be determined with the cylindrical potential. The value of the potential at the centre of the tube is found by setting normalized distance r∗ to zero in equation: V (x = 0, R) = 3π2 θεσ2 (

21 σ 10 σ 4 ( ) − ( ) ). 32 R R

(11.9)

From this expression, it can be determined that the maximum well depth is Vo = V(r = 0, R) = −12.77 θε (or 12.5 kJ/mol for hydrogen) for a radius R0 = 1.086 σ, or R0 = 3.43 Å for hydrogen.

250 | Part II Development of new materials for energy applications The second virial coefficient, for uncapped SWNTs of length L, is given by the following expression: BAS = ∫ dr ⃗ (exp (−

1 V (r,⃗ R)) − 1) . kT

(11.10)

V

Since the interaction potential is only a function of the normalized radius, the volume integral can be performed using cylindrical coordinates: 2π

∞

L/2

BAS = ∫ dθ ∫ dz ∫ rdr (exp (− 0

which leads to

−L/2

0

1 V (r∗ , R∗ )) − 1), kT

(11.11)

∞

BAS 1 = 2π ∫ r∗ dr∗ (exp (− ∗ V∗ (r∗ , R∗ )) − 1). T Lσ2

(11.12)

0

Capped nanotubes do not adsorb internally, so the volume integral must exclude the internal volume of the nanotube: ∞

BAS 1 = 2π ∫ rdr (exp (− V (r∗ , R∗ )) − 1). T Lσ2 ∗

(11.13)

R

The behavior of the second virial coefficient as a function of the radius of the nanotube is quite complex, and summarized in Fig. 11.6 below for hydrogen. At 77 K, the coefficient BAS reaches its maximum value when the radius of the nanotube corresponds to the maximum well depth of the adsorption potential in the nanotube (full line). The peak value of BAS is 3 orders of magnitude greater than the equivalent graphene sheet. This feature, associated with a strong peak, is absent when the nanotube is closed. The peak is quenched very quickly as a function of temperature. At larger radii, the curves converge towards the value of an equivalent graphene sheet. It is worthwhile to note that the smallest possible SWNT has a radius equal to that of a C20 . These have been observed inside multi-walled carbon nanotubes as their innermost shells. The smallest experimentally accessible SWNTs have radii comparable to C60 (3.5 Å). The quenching of the BAS of open SWNTs observed at higher temperatures can be linked to the presence of repulsive and attractive regions inside the nanotube, which at low temperature usually lead to a positive and large value of the integrant in the expression for BAS , but which can become negative at lower temperature because the exponential, which is sensitive to temperature, is subtracted by a temperatureindependent constant. It is then possible for the virial coefficient to become negative and impair the adsorption process to first order in pressure. This occurs at temperatures above the Boyle temperature.

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 251

107

105

BAS/Lσ2

103

10–1

10–1

10–3

C60

C20 0

1

2 R/σ

3

4

Opened SWNT, T* = 0.2089 (H2: 77.4 K) Closed SWNT, T* = 0.2089 Opened SWNT, T* = 0.8097 (H2: 300 K) Closed SWNT, T* = 0.8097 Opened SWNT, T* = 1.000 (H2: 371 K) Closed SWNT, T* = 1.000 Fig. 11.6. Behavior of the second virial coefficient of the adsorbed density of hydrogen on isolated SWNTs as a function of its radius. The SWNT reduced radii corresponding to the C20 and C60 fullerene cages are indicated by the vertical lines. Reprinted with permission from [11].

The behavior of the virial coefficient in bundles of SWNTs is rich, associated with the grooves of the bundle (peripheral sites in Fig. 11.7(a)) and the interstices between the nanotubes in the bundles (interstitial sites in Fig. 11.7(a)), which lead to additional structures associated with the presence of additional extrema in the interaction potential of the bundles (Fig. 11.7(b)). The virial coefficient BAS is a readily physically accessible quantity which contains a wealth of information about the energy scales of the adsorption process and in some cases, the very structure of an adsorbent, particularly in ordered monoatomic solids such as single wall nanotubes or other nanocrystalline carbon structures. It can be determined experimentally from the slope of the excess adsorption isotherm at low pressure, where the isotherm behaves according to Henry’s Law, from two pressure measurements at a given temperature. The energy scales that can be extracted from BAS require, however, the data from several low pressure isotherms. They can be approximately determined from a semi-logarithmic fit of BAS as a function of inverse temperature.

252 | Part II Development of new materials for energy applications The virial coefficient, however, does not contain much information on the high pressure behavior of the adsorption isotherm, as it is determined in a régime where intermolecular interactions between adatoms have not yet impacted the adsorption phenomena, and provides little information on the eventual saturation of the adsorption isotherm. It can thus be used to determine the temperature scale of the adsorption process, but not its saturation properties. y 15

Tube site

Peripheral site

10 5 0 –5 –10

(a)

–15 –15

Interstitial site –10

–5

x 0

5

10

15

–10 x

0

10 2 0 Vbundle –2 –4 –10

(b)

0 y

10

Fig. 11.7. (a) Adsorption sites and (b) interaction potential of a bundle with 19 SWNTs with R∗ = 2.038 and with a normalized lattice parameter d∗ = 5.235 relative to hydrogen. Reprinted with permission from [11].

11.3.2 Basic analytic models of the adsorption isotherm The Langmuir model The Langmuir isotherm is a local model of the absolute adsorption based on monolayer filling of noninteracting molecules. As such, it represents a model of supercritical adsorption on surfaces. It predicts a monotonically increasing function of pressure, which saturates asymptotically as a function of pressure to a value nm : n = nm

KL (T) P , 1 + KL (T) P

(11.14)

11 Supercritical adsorption of hydrogen on microporous adsorbents |

253

with

1 exp (ΔS0 /R) exp (−ΔH0 /RT), (11.15) P0 where ΔS0 and ΔH0 represent the entropy and enthalpy variations associated with the adsorption process, and where P is pressure [12]. There are several strategies for deriving the Langmuir isotherm. It can be obtained from either kinetic theory or a lattice gas approach. The Langmuir isotherm can be put into the following form: KL (T) =

P P 1 + , = n KL (T)nm nm

(11.16)

from which the parameters nm and KL (T) can be determined from experimental isotherms. The saturation density (nm ) and the coefficient (C) can be obtained from a simple linear fit of the experimental data. Figure 11.8 shows that even excess adsorption data (shown as points) represented in this way yields relatively linear curves. Note, however, that the Langmuir isotherm is not a Gibbs (excess) adsorption isotherm. A more detailed examination of the data represented in Fig. 11.8 would show substantial deviations from linear behavior between the low and high pressure regions. The excess adsorption version of the Langmuir isotherm requires knowledge of the pore volume: na (P󸀠 , T) = n(P󸀠 , T) − Vp ng (P󸀠 , T) = nm

KL (T)P󸀠 P󸀠 − Vp . 󸀠 1 + KL (T)P RT

(11.17)

T=273 K 2.0

P/n (Mpa gr/mmol)

1.6

1.2

0.8

0.4 T=77 K 0.0 0

1

2

3 4 Pressure (Mpa)

5

6

Fig. 11.8. Expression of the excess adsorption isotherms of hydrogen on the activated carbon AX-21 as the pressure over excess density as a function of pressure. Reprinted with permission from [12].

254 | Part II Development of new materials for energy applications The validity of the Langmuir model is typically limited to low pressure, high temperature adsorption of gases in the supercritical regime. As such, it is of limited usefulness to the description of adsorption over the wide temperature and pressure ranges relevant to storage applications.

The Dubinin isotherm The adsorption isotherm of microporous adsorbents has often been modeled by the Dubinin–Astakhov model. In this approach, the adsorption process is viewed as a volumetric process in which a pore of an adsorbent is progressively filled by a subcritical gas (T < Tc ) whose vapor phase can be described by the ideal gas law. If the conditions inside a pore are such that the local pressure is greater than the saturation, then the adsorbate is assumed to condense to a liquid state. The characteristic energy of adsorption of the pores is also assumed to be distributed according to a Weibull distribution. These assumptions lead to an isotherm where total adsorbed density is expressed as: n = n0 exp(−(A/E)m ), (11.18) where n0 is the saturation density of the adsorbate in the pores, m is a structural heterogeneity parameter, E is a characteristic energy of the average gas-solid interaction potential in the pores [13], and A is the adsorption potential: A = RT ln (

Ps ), P

(11.19)

where R is the universal gas constant, and Ps the saturation pressure. The parameter m is fitted in such a way that the characteristic curves of nA as functions of A are independent of temperature. The limit m = 2 reduce to the Dubinin–Raduskevitch isotherm [14, 15]. The excess density is obtained by considering the adsorption volume Va as a fitting parameter: nex = na − Vp ρg , (11.20) where ρg is the bulk gas density of hydrogen. Surprisingly, this isotherm has been found to accurately describe the adsorption process even in the supercritical regime. The pressure parameter Ps cannot in this case be associated with the saturation pressure, which is undefined when T > Tc . Amankwah and Schwarz have proposed the following equation [16] to determine Ps : 𝛾

PS = (T/TC ) PC ,

(11.21)

where Tc and Pc are the critical temperature and pressure of the adsorbate. This pressure, however, should be interpreted as the pressure at which the adsorbed phase behaves like an almost incompressible fluid in the supercritical regime, leading to a very large value of Ps when the isotherm is fitted to experimental data over wide pressure

11 Supercritical adsorption of hydrogen on microporous adsorbents |

255

range. Formally, at P = PS , the isotherm reaches a maximum value as a function of pressure. If the adsorbed phase behaves like a quasi-incompressible fluid, it can be expected that the characteristic energy also depends on temperature to reflect a limited dependence on temperature that should be expected when P is such that the pore is basically saturated. An analysis of the best fit of experimental isotherms of nitrogen on activated carbon over wide ranges of pressure and temperature suggest a linear dependence of the characteristic energy of adsorption as a function of temperature. A modification of the Dubinin isotherm was then proposed in which the characteristic energy of adsorption of the original model is replaced by a temperature-dependent expression E = α + βT.

30 K

Excess Adsorption (mol/kg)

50

35 K 40 K

40

45 K 60 K

30

77 K 93 K

20

113 K 153 K

10

213 K 298 K

0 0

1

2

3 4 Pressure (Mpa)

5

6

Fig. 11.9. Modified D–A isotherm (solid lines) parameterized on the experimental excess adsorption isotherms of hydrogen (shown as points) [19]. The optimal isotherm parameters are n0 = 71.6 mol kg−1 , α = 3.08 kJ mol−1 , β = 18.9 J mol−1 K−1 , P0 = 1 470 MPa, and Vp = 1.43 l kg−1 . The large values of P0 may be seen as coherent with the high density limit of the adsorbed gas. Reprinted with permission from [17]

The model successfully describes the adsorption of hydrogen on several microporous systems [17, 18], such as the activated carbon AX-21™ over the range 0–6 MPa and 30– 298 K, as shown in Fig. 11.9. The limiting density associated with complete filling of the micropores can be estimated by dividing nmax by Vpore . A value of 50.2 mol/l is obtained for hydrogen, which is a value close to the density of solid hydrogen (43.7 mol/l). The model has also been used to describe the adsorption isotherms of hydrogen on the activated carbon CNS-201™ and the metal-organic framework Cu3 (BTC)2 over the range (0–6 MPa and 77–296 K), as well as the adsorption isotherms of nitrogen and methane on single wall nanotubes over wide temperature and pressure ranges [19].

256 | Part II Development of new materials for energy applications The Dubinin model has some issues, such as the absence of a proper Henry’s law régime as a function of pressure. This approach should be viewed as a high density approximation. This can lead to problematic results at low pressure, such as negative excess adsorption isotherms, for the adsorption of hydrogen on the metal organic framework MOF 5. Its validitiy is also limited at high temperature by the fact that the argument of the isotherm depends very little on temperature at large temperatures.

11.3.3 Self-consistent approaches In addition to the examples of analytic isotherms described in the preceding section, nonanalytic isotherm models have been proposed for supercritical gases over wide pressure and temperature ranges.

Lattice gas isotherm Aranovitch et al. proposed a set of self-consistent equations describing the adsorption process on carbon slit pores based on a lattice-gas approach [12, 20], applicable in principle to the supercritical and subcritical regimes. This approach, which can be seen as a lattice-gas version of the Langmuir isotherm, calculated assuming a slit pore geometry, while accounting for proximity interactions between adsorbate molecules in a molecular field approximation. It correctly yields Henry’s law in the low pressure limit but does not, however, describe the porous structure of carbon. The model assumes that the adsorbate molecules form N discrete layers sandwiched by two graphene planes representing a slit pore. Adjacent molecules interact with an Ising site potential E. Molecules adjacent to the two graphene planes are subjected to a uniform potential EA . The coverage of the ith layer, defined by the relative density xi = ni /ns (where ns is the molar density of a completely filled adsorption layer) is obtained from self-consistent equations with the boundary condition x1 = xN . The excess adsorption isotherm is defined as: N

nex = A ∑ xi − ρg Vp .

(11.22)

i=1

The model has been successfully fitted to the experimental data shown in Fig. 11.10 for hydrogen on AX-21 with the following model parameters: Ea = −3.871 kJ/mol, E = 0.519 kJ/mol and ns = 74.0 mol/l. The saturation constant (A) was fitted in the following way: A (T) = A0 − A1 exp (−B/T) , (11.23) with A0 = 30.426 mol/l, A1 = 12.623 mol/l and B = 71.042 K. This isotherm leads to a pore volume Vpore of 1.62 liter/kg [21].

11 Supercritical adsorption of hydrogen on microporous adsorbents

Excess adsorbed amount [mol/kg]

50

257

30 K 35 K 45 K 60 K 77 K 93 K 113 K 133 K 153 K 173 K 193 K 213 K 233 K 253 K 273 K 295 K

40

30

20

10

0 0

|

1

2

3 4 Pressure [Mpa]

5

6

Fig. 11.10. Excess adsorption isotherms of hydrogen on the activated carbon AX-21. The full lines represent a fit to the Ono-Kondo adsorption isotherm. Reprinted with permission from [21].

11.4 The thermodynamics of adsorption Myers et al. [6] proposed a thermodynamic formulation of the adsorption process in which the pore structure is considered to be a bulk property of the adsorbent (assuming with Gibbs a zero volume adsorbed phase). This approach has the advantage of eliminating the need for additional thermodynamic variables and their conjugated forces. The adsorption properties are generally referred to per unit quantity of adsorbent (e.g., mass or volume). The adsorbed density thus represents a quantity of adsorbate (mass, volume, moles or direct number of particles) per unit quantity of the adsorbent. In their approach, all thermodynamic properties (Z) of the total system are considered to consist of three contributions: the adsorbate (a), the adsorbent (s) and the gas phase (g): Zt = Za + Zg + Zs , (11.24) with Zg = ng Vg zg .

(11.25)

The density of the free gas (in the absence of the solid) is ng . The intensive property zg is a thermodynamic property of the free gas which depends on temperature T, pressure P and, in the case of a mixture, the set of mass fractions (yi ). The volume of the gas phase Vg corresponds to the pore volume Vg . The properties of the solid phase are determined through thermodynamic measurements of the solid in the absence of the adsorbate. The properties of the adsorbed phase, on the other hand, intrinsically depend on the interaction between the solid and the adsorbate and are obtained by

258 | Part II Development of new materials for energy applications taking the difference between the thermodynamic properties of the total system and the properties of the free gas and the solid adsorbent: Za = Zt − Zg − Zs ,

(11.26)

which leads to the following expression for the adsorbed density: na = n t − n g .

(11.27)

This expression is also consistent with the excess adsorbed density defined earlier if the system does not contain a free volume external to the porous structure of the adsorbent, in which case Vg = Vp : na = nt − ngaz = nt − ρg Vg = nt − ρg Vp ,

(11.28)

since the volume of the system is the sum of the volume of the solid and the volume occupied by the gas in the adsorbent. At a given pressure and temperature, the thermodynamic properties of the adsorbed phase can be obtained by subtracting the contribution of the gas phase and the adsorbent: U a = Ut − Ug − Us = U − U s ,

(11.29)

Sa = St − Sg − Ss = S − Ss ,

(11.30)

Va = Vt − Vg − Vs = Vt − (Vg + Vs ) = 0.

(11.31)

The internal energy of adsorption can be written: Ua = U − Us = TSa + μg na + (μ − μs )

(11.32)

Ua = TSa + μg na + Φ,

(11.33)

Φ = μ − μs .

(11.34)

or: with The potential Φ is called the surface potential, defined as the difference between the chemical potential of the adsorbent interacting with the adsorbate and its value in the absence of such interactions, but still subjected to hydrostatic pressure P. A description based on the surface potential considers the adsorption process basically as a bulk property. It is as such appropriate from a conceptual point of view to the description of highly microporous adsorbents such as metal organic frameworks, in which a clear interface cannot easily be defined. The Gibbs potential can be found using the standard Legendre transform: Ga = Ua + PVa − TSa = Ua − TSa = μg na + Φ.

(11.35)

The free energies of the adsorbed phase thus depend on the adsorption isotherms (na ) and the surface potential.

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 259

The differential forms of the thermodynamic potentials can be written: dUa = TdSa − μg dna ,

(11.36)

dGa = Sa dT − μg dna ,

(11.37)

dΦ = −Sa dT − na dμg .

(11.38)

The adsorption isotherm and the entropy of the adsorbed phase can be obtained from: 𝜕Φ 󵄨󵄨󵄨󵄨 (11.39) na = − μ 󵄨 𝜕 g 󵄨󵄨󵄨T and Sa = −

𝜕Φ 󵄨󵄨󵄨󵄨 󵄨 . 𝜕T 󵄨󵄨󵄨μg

(11.40)

Using the Gibbs potential it can be shown that thermodynamic equilibrium between the adsorbed molecules and their bulk gas phase is expressed through the equality of their temperatures and their chemical potentials: μa = μgaz .

(11.41)

11.4.1 Properties of surface potential At constant temperature, the difference equation for surface potential becomes: dΦ|T = −na dμg .

(11.42)

Expressing chemical potential in terms of the fugacity (f): f = f0 exp

μg RT

.

(11.43)

Fugacity converges to the pressure (P) as P→0. We obtain the constant temperature contribution to the surface potential: dΦ|T = −RTna d (ln

f df ) = −RTna . f0 f

(11.44)

In terms of the fugacity, the differential of the surface potential becomes: dΦ = −Sa dT − na

RT df. f

(11.45)

Note that the integrated surface potential should have the following form: Φ(f, T) = u(f) + v(T) + w(f, T).

(11.46)

When f → P → 0, however, we expect that Φ(f, T) = μ − μs = 0 because the chemical potential of the system becomes equal to the chemical potential of the adsorbent (μ → μs ). This implies that v(T) = 0, and thus Φ(f, T) = Φ|T .

260 | Part II Development of new materials for energy applications Low pressure limit At low pressures, f ∝ P (this is exact for an ideal gas). Integrating from P󸀠 = 0 to P we find that: P

Φ = −RT ∫ na (P󸀠 , T)

dP󸀠 . P󸀠

(11.47)

P=0

In the low pressure limit, Henry’s law states that the total adsorbed density is proportional to pressure: n(P󸀠 , T) = K(T)P󸀠 , (11.48) where K (T) is Henry’s coefficient. The excess adsorbed density in this limit is: na (P󸀠 , T) = K(T)P󸀠 − Vp ng = K(T)P󸀠 − Vp

P󸀠 . RT

(11.49)

In the low pressure limit, the surface potential behaves as follows: Φ = − RTK (T) P + Vp P = −RTna (P, T) + Vp P.

(11.50)

The Langmuir isotherm As an illustrative example, we consider the excess Langmuir adsorption isotherm, which can be written in the following general form, assuming an ideal gas: na (P󸀠 , T) = n(P󸀠 , T) − Vp ng (P󸀠 , T) = nm with KL (T) =

KL (T)P󸀠 P󸀠 − Vp , 󸀠 1 + KL (T)P RT

1 exp (ΔS0 /R) exp(−ΔH0 /RT), P0

(11.51)

(11.52)

where ΔS0 is the entropy difference relative to the reference pressure P0 (typically one atmosphere), and ΔH0 is the enthalpy difference associated with the adsorption process. The coefficient KL (T) corresponds to Henry’s coefficient. The corresponding surface potential, for an ideal gas, is thus given by: Φ = −RTnm ln(KL (T)P + 1) + Vp P.

(11.53)

The chemical potential of the ideal gas is given by: μg = RT ln (

P ). P0

(11.54)

Using this expression to eliminate the pressure from the surface potential, we can express the latter in terms of its natural variables T and μ: Φ = −RTnm ln (KL (T)P0 exp (

μg RT

) + 1) + Vp P0 exp (

μg RT

).

(11.55)

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 261

It is easily verified that the Langmuir isotherm is recovered by the equation: 󵄨 𝜕Φ 󵄨󵄨󵄨 󵄨󵄨 . na = − 𝜕μg 󵄨󵄨󵄨T

(11.56)

Entropy, internal energy, and Helmholtz free energy can be readily obtained from the expressions for the surface and chemical potentials of the ideal gas.

11.5 Microporous adsorbents for hydrogen storage 11.5.1 Activated carbons Activated carbons (Fig. 11.11) have been proposed as a means of storing gases such as natural gas at room temperature [22]. For hydrogen, storage applications require low temperature operation and highly microporous activated carbons with high specific surface areas due to the small value of the binding energies between carbon structures and molecular hydrogen. Activated carbons have been widely studied for hydrogen storage applications, because of their availability and the possibility of tailoring their porous structures via various chemical and physical treatments. The high pressure, low temperature adsorbed density of hydrogen on activated carbon seems to correlate linearly with the micropore volume and the specific surface area of the adsorbent [23].

IRH-4D

18kV X1, 500 10

Fig. 11.11. SEM image of a sample of the activated carbon IRH-4DD with a specific surface area of 2 000 m2 /g showing the macroporous and mesoporous structure of the carbon [22].

The excess adsorption isotherms of hydrogen over the supercritical temperature range of 35 K to 300 K on the activated carbon AX-21 are type I fully reversible isotherms, as shown in Figs. 11.7 and 11.8. The specific surface and bulk density were respectively 2 800 m2 /g and 0.3 g/cm3 . The maximum excess density at 77 K was 54 g/kg at 35 bars. This maximum occurs when the slope of the absolute adsorbed density becomes equal to the slope of the bulk density curve multiplied by the pore volume. The total amount of hydrogen contained in the adsorbent is the sum of the excess density and the compressed bulk phase in the pore volume of the adsorbent. Assuming that graphite corresponds to the limiting case of an activated carbon with zero pore volume, the latter can be estimated from the ratio of the density of activated carbon to

262 | Part II Development of new materials for energy applications graphite [22]: Vpore = (1 −

dAC dgraphite

) Vadsorbent ,

(11.57)

where dAC is the density of the activated carbon and dgraphite , the density of graphite (2.2 g/cm3 ). Dividing equation (11.57) by the mass of the adsorbent yields an estimate of the gravimetric pore volume, in this case 2.88 ml/g. The micropore volume is estimated to be 1.06 ml/g, or 37% of the measured total pore volume. The macropore volume represents about 10% of the total pore volume. The remaining portion is associated with the external surface of the adsorbent (including the intergranular volume). The contribution of the compressed gravimetric density of the pore structure is obtained by multiplying the total pore volume by the bulk gas density of the adsorbate, resulting in a value of 33 g/kg at 77 K and 35 bars and leading to a total gravimetric density of 8.7% of hydrogen in the activated carbon, most of it corresponding to a gravimetric density of 6.6% in the micropores. The fluid phase density of hydrogen inside the micropores can be calculated by dividing the total density adsorbed by the micropore volume. This yields to a value of 62 kg/m3 , suggesting that at 77 K and 35 bars, the density of hydrogen inside the micropore is already close to that of liquid hydrogen. Similar values (from 61 to 71 mg/ml) are obtained for other activated carbon adsorbents with different pore volumes and specific surfaces [22].

11.5.2 Single wall nanotubes Single wall nanotubes (Fig. 11.12) were first proposed as storage media for hydrogen by Dillon et al. [24]. They are no longer considered viable candidates in their pure form for practical storage applications of hydrogen, due to their cost and the relatively marginal gain they may offer over activated carbons in terms of storage capacity. The adsorption of hydrogen on single wall nanotubes results in type I isotherms (Fig. 11.13). The cylindrical geometry of nanotubes is conducive to deeper adsorbate-adsorbent interaction potential, well inside small diameter single wall nanotube. In a bundle, the interstitial sites could be even more favorable due to the overlap of the molecular force fields of each SWNT. The small pore volume associated with these sites, along with the small diameter of these pores, makes their relative contribution to the overall adsorbed density small. Under ambient conditions, pure SWNT adsorb less than 1 wt% hydrogen. At 77 K and 1 bar, excess adsorbed densities of up to 2.5 wt% have been reported, depending on sample preparation [25–28]. Values of 6 wt% at 77 K and 2 bars [29] and 8 wt% at 40 bars and 80 K have been obtained [30]. The adsorption enthalpy is estimated to be about 4.5 kJ/mol, comparable to activated carbon adsorbents. The adsorption of hydrogen on SWNTs is very sensitive to chemical and heat treatments, as they can facilitate access to internal sites [31].

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 263

Dai18-8 140kx 100 nm

Fig. 11.12. SEM image of SWNTs [22].

3.0

Adsorption (wt%)

2.5 2.0 1.5 1.0 SWNTS-Pr. SWNTS-700 CNI-3P

0.5 0.0 0

10

20

30

40

50

P (atm) Fig. 11.13. Hydrogen adsorption isotherms on single wall nanotubes at 77 K [28].

11.5.3 Metal organic frameworks Metal organic frameworks (MOFs; Fig. 11.14) are hybrid polymers made of organic and inorganic units, bound together during isoreticular synthesis [32, 33]. They are soluble and their configuration depends on the environmental conditions during the reaction. The interactions that occur between metal ions and organic linkers are ionic, covalent, and coordination bonds. There are also hydrogen and π–π bonds [34]. The synthesis conditions are dictated by the chemical properties of the organic linkers and the final desired structure. Generally, MOFs are obtained through solvothermal synthesis, during which the organic linker and the metal salt are mixed in a polar solvent. The temperature is then slowly increased and maintained between 50 to 150°C. The reaction is generally complete within several hours to several

264 | Part II Development of new materials for energy applications Molecular complexes

Extended solids

+

(a)

Expanded framework

+

(b)

Decorated-expanded framework

Fig. 11.14. Assembly of MOFs by copolymerization of metal ions with organic linkers to give (a) flexible metal-bipyridine structures, (b) rigid metal-carboxylate clusters [33].

days [35, 36]. One of the biggest advantages for MOF synthesis stems from the labile nature of the metal-linkers bond, in the sense that if the linker is not stable, it will dissociate by itself and take a configuration that corresponds to the minimum energy of the final structure. Depending on the linker and the metal node geometry, ditopic or tritopic, the building blocks will have two or three dimensions [37]; see Fig. 11.15. MOFs are of particular interest for their adsorption properties, and can be used for purification [38], gas storage, sensing [39, 40], catalysis or ion exchange [41–46]. It is possible to tailor a pore distribution according to a desired application by selecting the proper linker and metal ion. The pore size can even be changed without modifying the global geometry of the network, by simply selecting an organic linker with the same geometry, but with a different size (Fig. 11.16). This property is referred to as the isoreticular principle [47]. MOFs can thus be synthesized to capture larger molecules such as vitamin B12 and proteins [32]. MOFs are the materials with the biggest pore openings (98 Å) and the smallest density (0.13 g/cm3 ). Increasing the pore size generally results in a larger specific surface area and in greater adsorption capacity at higher pressures. Physiosorption inside MOFs is the result of weak van der Waals interactions (typically less than 10 kJ/mol; [48]) between the MOF and an adsorbate gas, which leads to a reversible increase in the gas density

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 265

Donors 0°

60°

90°

109°

120°

180°

Acceptors

0°

60° 90°

109° 120° 180°

(a) Tritopic subunit

Ditopic subunit

60°

90°

109°

90°

109°

Trigonal bipyramid Double square

Trigonal bipyramid

Tetrahedron Cube

Truncated tetrahedron

Adamantanoid Cuboctahedron

180°

(b)

120°

Dodecahedron

Fig. 11.15. (a) Ditopic building blocks generate 2D convex polygons. (b) The combination of ditopic and tritopic building blocks results in 3D polygons [37].

inside the MOFs. A significant hydrogen adsorption capacity in MOFs is only possible for small pores (between 4.5 and 5 Å; [49–53]), which allows increased adsorbentadsorbate interactions due to the proximity of the structures in the unit cell, creating overlaps of the potential energy curves [54]. Table 11.1 shows the effect of temperature on the CO2 adsorption properties of IRMOF-1. Table 11.2 illustrates the adsorption capacity (in wt%) of three MOFs, IRMOF-1 (Zn4 O(BDC)3 ), IRMOF-11 (Zn4 O(HPDC)3 ), and MOF-177 (Zn4 (BTB)2 )¹, for H2 , CO2 , H2 S, and CH4 at different pressure [55]. As expected, uptake decreases with temperature but increases with pressure.

1 BDC = benzene-1,4-dicarboxylic acid; HPDC = 4,5,9,10-tetrahydropyrene–2,7-dicarboxylate; BTB = benzene-1,3,5-tribenzoate; AM = amide

266 | Part II Development of new materials for energy applications Table 11.1. Quantity of CO2 adsorbed (in mg) per gram of IRMOF-1.

Temperature (K) 195 208 218 233 273

IRMOF-1; CO2 Pressure (Torr) wt(%) 74.0 75.0 75.0 91.5 91.2

11.76 7.62 3.89 2.73 0.81

Pressure (Torr)

wt%

751.0 750.2 750.0 735.8 748.4

148.90 140.55 123.24 29.87 6.61

The organic linkers in the MOFs open the door to postsynthetic modifications by organic reactions. Indeed, by targeting the organic part, it is possible to change the chemical and physical properties of an MOF, basically creating a new MOF [56]. This allows the synthesis of MOFs which would have been impossible to create directly, because the desired functional groups would have interfered with crystal growth. The transformation of IRMOF-3 (Zn4 O(BDC-NH2 )3 ) into IRMOF-3-AM1 is a good illustration of this principle. In this synthesis, the amine functional group from the organic linker reacts with acetic anhydride, in CHCl3, to produce a secondary amide functional group [57]; see Fig. 11.17. Table 11.2. Quantity of adsorbed gas (in mg) per gram of MOF, as a function of pressure. Temperature (K) : 77 Gas MOF Pressure (Torr)

H2

IRMOF-1 IRMOF-11 MOF-177

77.5 75.1 69.1

Temperature (K): 298 Gas MOF Pressure (Torr)

CO2

IRMOF-1 IRMOF-11 MOF-177

920 801 770

Temperature (K): 298 Gas MOF Pressure (Torr) H2 S

IRMOF-1 MOF-177

708 657

Temperature (K): 298 Gas MOF Pressure (Torr)

CH4

IRMOF-1 IRMOF-11 MOF-177

889 889 863

wt%

Pressure (Torr)

wt%

0.25 0.49 0.17

752.9 753.2 751.0

1.32 1.62 1.25

wt%

Pressure (Torr)

wt%

4.73 7.90 3.51

1 556 1 572 1 623

wt%

Pressure (Torr)

19.86 16.46

1 458 1 437

8.10 13.74 7.86 wt% 25.00 18.91

wt%

Pressure (Torr)

wt%

0.72 1.02 0.12

2 337 2 507 2 425

1.74 2.79 1.29

11 Supercritical adsorption of hydrogen on microporous adsorbents |

267

‒

OOC

‒

OOC

‒

OOC N N

COO‒

N

COO‒

N N

‒

OOC BTC ‒

OOC

MOF-199 HKUST-1 Cu3(BTC)2

N

COO‒ N

4,4′,4″ –(1,3,4,6,7,9,9bheptaazaphenalene2,5,8-triyl) tribenzoate (HTB)

PCN-HTB′ Cu3(HTB)2

(BBC) ‒

OOC

MOF-399 Cu3(BBC)2

Fig. 11.16. Molecular structures of organic linkers (top). Single crystal structures of MOF-199, PCN-HTB0, and MOF-399 (bottom). The yellow balls indicate space in the cage [47].

NH2

O

H N

O

nO CHCl3, r.t.

n

n O

n = 0 – 18 IRMOF-3

IRMOF-3-AM

Fig. 11.17. Scheme of representative postsynthetic modification reactions with IRMOF-3 and acetic anhydride [57].

Many applications of MOFs expose them to various degrees of heat and pressure. The stability of their structure as a function of temperature and pressure is thus a critical issue which must be taken into account. Some MOFs undergo structural phase transitions as a function of temperature and pressure due to their flexible network [58–60]. Some, under high pressure, lose their crystallinity and become irreversibly amorphous [61]. There may also be a reversible rearrangement of the binding between

268 | Part II Development of new materials for energy applications a

a

b

b

(a)

(b)

Fig. 11.18. Phase transition, from α-phase (a) to β-phase (b), which occurs inside ZnIm at pressure between 0.543 GPa and 0.874 GPa [62].

the metallic centers and the linkers [62] as the pressure is increased (Fig. 11.18). This phenomenon is called “breathing effect”. As for all the nanostructured adsorbents discussed in this chapter, the adsorption isotherm of hydrogen on metal organic framework nanostructures is a type I isotherm. In 2006 Wong-Foy et al. reported hydrogen adsorption results on seven different MOFs: IRMOF-1 (MOF-5), IRMOF-6, IRMOF-11, HKUST-1 (CuBTC), MOF-74 and MOF-177 [63]. The saturation uptake was observed to vary sharply depending on the MOF; MOF-74 exhibited the lowest adsorbed density with a saturation uptake of 2.3 wt% at 26 bar and 3.5 wt% for IRMOF-11 at 34 bar, whereas the highest was observed on MOF-177 and IRMOF-20, where the maximum uptake was measured at 7.5 and 6.7 wt% respectively for pressures ranging from 70 to 80 bar. HKUST-1, IRMOF-1 and IRMOF-6 exhibited maximum uptakes of 3.0, 4.8 and 4.5 wt% for a pressure range of 40–60 bars. The two best adsorbents on a volumetric basis were found to be IRMOF-20 and IRMOF-177 for a pressure range of 70–80 bar. Recent measurements of the excess adsorption densities of hydrogen are summarized in Table 11.3. It can be seen that the excess adsorbed density of hydrogen on MOF nanostructures range from 6.5 to 9.0 (wt%). A total gravimetric density of up to 150 g/kg is possible for MOF 210, the corresponding volumetric density being 44 g/l. MOFs are highly porous nanostructures (void space of 90% for MOF-200) with a low density. At room temperature (295 K), reversible hydrogen uptakes of less than 1 wt% have been obtained. The excess adsorbed density seems to correlate more or less linearly with the specific surface area at high pressure ([22] and references therein). This situation seems to be reversed at very low pressure, where the trend is reversed [63], due to the fact that gravimetric adsorption is determined by the binding energy at low pressure and by the specific surface at high pressure, where saturation becomes an

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 269

Table 11.3. Adsorption properties of hydrogen on various metal organic frameworks at 77 K and 80 bars [65]. MOF

MOF 5 MOF 177 MOF 200 MOF 205 MOF 210 UMCM-2 NU-100

Density (g/cm3 )

BET Surface area (m2 /g)

Excess gravimetric density (wt%)

Total gravimetric density (wt%)

Total volumetric density (g/l)

0.59 0.43 0.22 0.38 0.25 0.40 0.29

3 800 4 500 4 530 4 460 6 240 5 200 6 143

7.1 6.8 6.9 6.5 7.9 6.5 9.0

9.6 10.4 14.0 10.7 15.0 11.0 14.1

63 50 36 46 44 50 41

issue. It is interesting to note that the adsorbed density of hydrogen, on a volumetric basis, correlates with the crystal density of the MOF. This is in agreement with a Grand Canonical Monte Carlo study of the effect of the specific surface area and adsorbent density on the adsorption of hydrogen on contrived carbon structures [64]. The contrived structures were benzene rings networked in such a way as to vary density and surface area systematically, using the same Lennard-Jones interaction parameters between the adsorbate molecules and the adsorbent atoms. The gravimetric density of hydrogen was shown to increase as a function of the density of the carbon nanostructure at low pressure due to narrower pores. The trend was reversed at high pressure, at which point the specific surface of the structures became the determining factor. This behavior seems to be observed in metal organic frameworks. On a volumetric basis, however, the high density materials are always found to be the best adsorbent. A linear correlation between adsorbed density and specific surface was also observed for these structures at 35 bars and 77 K. For physisorbed hydrogen, these results confirm that for a given atomic composition, maximizing the specific surface of the adsorbent seems to remain the optimal strategy for the design of materials for hydrogen storage if volumetric considerations are not critical. All contrived structures were weak hydrogen adsorbents under ambient conditions.

270 | Part II Development of new materials for energy applications

References [1] [2] [3]

[4] [5] [6] [7] [8] [9] [10] [11] [12]

[13] [14]

[15]

[16] [17]

[18]

[19]

Sircar S, Golden TC, Rao MB. Activated carbon for gas separation and storage. Carbon. 34 (1996) 1–12. Chahine R, Bose TK. Low-pressure adsorption storage of hydrogen. International Journal of Hydrogen Energy. 19 (1994) 161–164. Gross K, Carrington K, Barcelo S, Karkamkar A, Purewal J, Parilla P. Recommended Best Practices for the Characterization of Storage Properties of Hydrogen Storage, Materials. Report to the Department of Energy Office of Energy Efficiency and Renewable Energy Hydrogen Storage Program under National Renewable Energy Laboratory Contract No. 147388, H2 Technology Consulting LLC, 2012. Pierotti R, Rouquerol J. Reporting physisorption data for gas/solid systems with special reference to the determination of surface area and porosity. Pure Appl Chem. 57 (1985) 603–619. Myers A. Thermodynamics of adsorption in porous materials. AIChE Journal. 48 (2002) 145–160. Landau L. and Lifchitz E.; Physique Théorique : Physique Statistique; 4ième édition, Ellipse Marketing, Paris, 1998. K.S.W Sing et al. Reporting data for gas/solid systems with special reference to the determination of surface area and the porosity. Pure & Appl Chem, 57(4), (1985), 603–619. Clark A. The theory of adsorption and catalysis: Academic Press, New York; 1970. Stan G, Cole MW. Low coverage adsorption in cylindrical pores. Surface Science. 395 (1998) 280–291. Stan G, Cole MW. Hydrogen Adsorption in Nanotubes. Journal of Low Temperature Physics. 110 (1998) 539–544. Mélançon É, Bénard P. Theoretical study of the contribution of physisorption to the lowpressure adsorption of hydrogen on carbon nanotubes. Langmuir, 20 (2004) 7852–7859. Bénard P, Chahine R. Determination of the adsorption isotherms of hydrogen on activated carbons above the critical temperature of the adsorbate over wide temperature and pressure ranges. Langmuir. 17 (2001) 1950–1955. Chen SG, Yang RT. Theoretical Basis for the Potential Theory Adsorption Isotherms. The Dubinin-Radushkevich and Dubinin-Astakhov Equations. Langmuir. 10 (1994) 4244–4249. Walton KS, LeVan MD. Adsorbed-Phase Heat Capacities: Thermodynamically Consistent Values Determined from Temperature-Dependent Equilibrium Models. Industrial & Engineering Chemistry Research. 44 (2004) 178–182. Dubinin MM. Physical Adsorption of Gases and Vapors in Micropores, In: D.A. Cadenhead, J.F. Danielli and M.D. Rosenberg, (Eds), Volume 9, Pages 1–70. Progress in Surface and Membrane Science, Elsevier, 1975. Amankwah KAG, Schwarz JA. A modified approach for estimating pseudo-vapor pressures in the application of the Dubinin-Astakhov equation. Carbon. 33 (1995) 1313–1319. Richard MA, Bénard P, Chahine R. Gas adsorption process in activated carbon over a wide temperature range above the critical point. Part 1: Modified Dubinin-Astakhov model. Adsorption. 15 (2009) 43–51. Richard MA, Bénard P, Chahine R. Gas adsorption process in activated carbon over a wide temperature range above the critical point. Part 2: Conservation of mass and energy. Adsorption. 15 (2009) 53–63. A. Kleinhammes, B.J. Anderson, Q. Chen and Y. Wu, Characterization of hydrogen adsorption by NMR, poster presented at DOE Annual Merit Reviews Meeting, Arlington, Virginia, May 15–18 2007.

11 Supercritical adsorption of hydrogen on microporous adsorbents

| 271

[20] Aranovich GL, Donohue MD. Adsorption isotherms for microporous adsorbents. Carbon. 33 (1995) 1369–1375. [21] R. Paggiaro, P. Bénard and W. Polifke, Cryo-adsorptive hydrogen storage on activated carbon. I: Thermodynamic analysis of adsorption vessels and comparison with liquid and compressed gas hydrogen storage, International Journal of Hydrogen Energy; 35(2), (2010) 638–647. [22] Bénard P, Chahine R. Carbon nanostructures for H2 storage. In: G. W, ed. Solid State Hydrogen Storage – Mats and Chem, 261–287. Woodhead Publ.; 2008. [23] Chahine R, Bose T. Characterization and optimization of adsorbents for hydrogen storage. Hydrogen Energy Progress. 2 (1996) 1259–1264. [24] Dillon AC, Jones KM, Bekkedahl TA, Kiang CH, Bethune DS, Heben MJ. Storage of hydrogen in single-walled carbon nanotubes. Nature 386 (1997) 377–379. [25] Ansón A, Callejas MA, Benito AM, et al. Hydrogen adsorption studies on single wall carbon nanotubes. Carbon. 42 (2004) 1243–1248. [26] Tarasov B, Maehlen J, Lototsky M, Muradyan V, Yartys V. Hydrogen sorption properties of arc generated single-wall carbon nanotubes. Journal of alloys and compounds. 356 (2003) 510–514. [27] Nishimiya N, Ishigaki K, Takikawa H, et al. Hydrogen sorption by single-walled carbon nanotubes prepared by a torch arc method. Journal of alloys and compounds. 339 (2002) 275–282. [28] Poirier E, Chahine R, Bénard P, Lafi L, Dorval-Douville G, Chandonia P. Hydrogen adsorption measurements and modeling on metal-organic frameworks and single-walled carbon nanotubes. Langmuir.22 (2006) 8784–8789. [29] Pradhan B, Harutyunyan A, Stojkovic D, et al. Large cryogenic storage of hydrogen in carbon nanotubes at low pressures. MRS Proceedings; Cambridge Univ Press, 2001. [30] Ye Y, Ahn C, Witham C, et al. Hydrogen adsorption and cohesive energy of single-walled carbon nanotubes. Applied physics letters. 74 (1999) 2307. [31] Lafi L, Cossement D, Chahine R. Raman spectroscopy and nitrogen vapour adsorption for the study of structural changes during purification of single-wall carbon nanotubes. Carbon. 43 (2005) 1347–1357. [32] Hiroyasu F, Kyle EC, Michael OK, Omar MY. The chemistry and applications of metal-organic frameworks. Science. 341 (2013) 1230444. [33] Eddaoudi M, Moler DB, Li H, et al. Modular Chemistry: Secondary Building Units as a Basis for the Design of Highly Porous and Robust Metal–Organic Carboxylate Frameworks. Accounts of Chemical Research. 34 (2001) 319–330. [34] Rood JA. Metal-organic frameworks as functional, porous materials [Ph.D.]. Ann Arbor: University of Notre Dame; 2009. [35] Tranchemontagne DJ, Hunt JR, Yaghi OM. Room temperature synthesis of metal-organic frameworks: MOF-5, MOF-74, MOF-177, MOF-199, and IRMOF-0. Tetrahedron. 64 (2008) 8553–8557. [36] Rowsell JLC, Yaghi OM. Strategies for hydrogen storage in metal-organic frameworks. Angewandte Chemie – International Edition. 44 (2005) 4670–4679. [37] Cook TR, Zheng Y-R, Stang PJ. Metal–organic frameworks and self-assembled supramolecular coordination complexes: comparing and contrasting the design, synthesis, and functionality of metal–organic materials. Chemical reviews. 113 (2012) 734–777. [38] Li J-R, Sculley J, Zhou H-C. Metal–organic frameworks for separations. Chemical reviews. 112 (2011) 869–932. [39] Resendiz MJ, Noveron JC, Disteldorf H, Fischer S, Stang PJ. A Self-Assembled Supramolecular Optical Sensor for Ni (II), Cd (II), and Cr (III). Organic letters. 6 (2004) 651–853. [40] Vajpayee V, Song YH, Lee MH, et al. Self-assembled arene-ruthenium-based rectangles for the selective sensing of multi-carboxylate anions. Chemistry - A European Journal. 17 (2011) 7837–7444.

272 | Part II Development of new materials for energy applications [41] Hu M. Design, synthesis and applications of Metal Organic Frameworks: Masters Thesis. Worcester Polytechnic Institute; 2011. [42] Lin X, Jia J, Hubberstey P, Schröder M, Champness NR. Hydrogen storage in metal–organic frameworks. Cryst Eng Comm. 9 (2007) 438–448. [43] Kitaura R, Seki K, Akiyama G, Kitagawa S. Porous coordination-polymer crystals with gated channels specific for supercritical gases. Angewandte Chemie - International Edition. 42 (2003) 428–431. [44] Chen B, Liang C, Yang J, et al. A microporous metal-organic framework for gaschromatographic separation of alkanes. Angewandte Chemie - International Edition. 45 (2006) 1390–1393. [45] Millward AR, Yaghi OM. Metal-organic frameworks with exceptionally high capacity for storage of carbon dioxide at room temperature. Journal of the American Chemical Society. 127 (2005) 17998–17999. [46] Suh MP, Park HJ, Prasad TK, Lim D-W. Hydrogen storage in metal–organic frameworks. Chemical reviews. 112 (2011) 782–835. [47] Furukawa H, Go YB, Ko N, et al. Isoreticular expansion of metal-organic frameworks with triangular and square building units and the lowest calculated density for porous crystals. Inorganic Chemistry. 50 (2011) 9147–9152. [48] Collins DJ, Zhou H-C. Hydrogen storage in metal-organic frameworks. Journal of Materials Chemistry. 17 (2007) 3154–3160. [49] Kesanli B, Cui Y, Smith MR, Bittner EW, Bockrath BC, Lin W. Highly interpenetrated metal– organic frameworks for hydrogen storage. Angewandte Chemie International Edition. 44 (2005) 72–75. [50] Chun H, Dybtsev DN, Kim H, Kim K. Synthesis, X-ray Crystal Structures, and Gas Sorption Properties of Pillared Square Grid Nets Based on Paddle-Wheel Motifs: Implications for Hydrogen Storage in Porous Materials. Chemistry-A European Journal 11 (2005) 3521–3529. [51] Zhao X, Xiao B, Fletcher A, Thomas K. Hydrogen adsorption on functionalized nanoporous activated carbons. The Journal of Physical Chemistry B. 109 (2005) 8880–8888. [52] Gogotsi Y, Dash RK, Yushin G, Yildirim T, Laudisio G, Fischer JE. Tailoring of nanoscale porosity in carbide-derived carbons for hydrogen storage. Journal of the American Chemical Society. 127 (2005) 16006–16007. [53] Armandi M, Bonelli B, Areán CO, Garrone E. Role of microporosity in hydrogen adsorption on templated nanoporous carbons. Microporous and Mesoporous Materials. 112 (2008) 411–418. [54] Thomas KM. Adsorption and desorption of hydrogen on metal–organic framework materials for storage applications: comparison with other nanoporous materials. Dalton Transactions (2009) 1487–1505. [55] Millward AR. Adsorption of environmentally significant gases (hydrogen, carbon dioxide, hydrogen sulfide, methane) in metal-organic frameworks [Ph.D.]. Ann Arbor: University of Michigan (2006). [56] Wang Z, Cohen SM. Postsynthetic covalent modification of a neutral metal-organic framework. Journal of the American Chemical Society. 129 (2007) 12368–12369. [57] Wang Z, Cohen SM. Postsynthetic modification of metal-organic frameworks. Chemical Society Reviews. 38 (2009) 1315–1329. [58] Chapman KW, Chupas PJ. Pressure enhancement of negative thermal expansion behavior and induced framework softening in zinc cyanide. Journal of the American Chemical Society. 129 (2007) 10090–10091. [59] Chapman KW, Chupas PJ, Kepert CJ. Direct observation of a transverse vibrational mechanism for negative thermal expansion in Zn (CN) 2: An atomic pair distribution function analysis. Journal of the American Chemical Society. 127 (2005) 15630–15636.

11 Supercritical adsorption of hydrogen on microporous adsorbents

|

273

[60] Biswas MM, Cagin T. High pressure structural transformation of selected metal organic frameworks – A theoretical investigation. Materials Chemistry and Physics. 131 (2011) 44–51. [61] Chapman KW, Halder GJ, Chupas PJ. Pressure-induced amorphization and porosity modification in a metal-organic framework. Journal of the American Chemical Society. 131 (2009) 17546–17547. [62] Spencer EC, Angel RJ, Ross NL, Hanson BE, Howard JAK. Pressure-induced cooperative bond rearrangement in a zinc imidazolate framework: A high-pressure single-crystal X-ray diffraction study. Journal of the American Chemical Society. 131 (2009) 4022–4026. [63] Wong-Foy AG, Matzger AJ, Yaghi OM. Exceptional H2 saturation uptake in microporous metalorganic frameworks. Journal of the American Chemical Society. 128 (2006) 3494–3495. [64] Lachance P, Bénard P. Specific surface effects on the storage of hydrogen on carbon nanostructures. International journal of green energy. 4 (2007) 377–384. [65] O. M. Yaghi. Hydrogen storage in metal-organic frameworks. Annual Merit Review Proceedings, U.S. Department of Energy, Washington, May 9–13 2011.

| Part III: New trends in sustainable development and biomedical applications

D. Mantovani, L. Levesque, G. Sabbatier, M. Leroy, D. G. Seifu, R. Tolouei, V. Montaño, M. Cloutier, I. Bilem, C. Loy, M. Byad, C. Paternoster, C. A. Hoesli, B. Drouin, G. Laroche

12 Advanced materials for biomedical applications 12.1 Introduction Since 1970, biomaterials have saved or improved the lives of millions of humans around the globe. Mostly made of polymers, metals, and ceramics, biomaterials lead to the design, production, development and optimization of implants and prostheses to replace, restore or reconstruct parts of diseased or damaged tissues and organs. Research is very dynamic in all fields of biomaterials worldwide, and is mainly aimed at improving the clinical performance of those implants and prostheses currently on the market. It is a growing multimillion dollar industry, justified by the world’s aging population and by humans’ desire to experience the highest quality of life even when elderly. The main goal of this chapter is to provide a concise and schematic overview of the principal applications of biomaterials for the benefit of graduate and undergraduate students as well as scientists who are curious to explore this challenging and multidisciplinary field. A number of more specialized books are available to those readers interested in broadening their knowledge in the field. For example, sterilization processes will be briefly mentioned, although these represent a potential major bottleneck in material design and development. Similarly, the FDA and other accreditation processes are not addressed in this chapter, despite representing a major challenge for the industrial development of the field. In terms of material engineering, this chapter also has some limitations, as it does not address the flourishing field of surface modifications, which are largely applied to metals, polymers, and ceramics, mainly to modulate the interfacial properties with the surrounding (living) tissues and organs. This chapter aims to set forth a realistic overview of the fields of implants and prostheses, mainly from the materials science and engineering point of view. Since biomaterials science and engineering is a scientific field at the frontier between medicine, biology (life sciences), and materials science, this chapter is divided into six sections. Each of these sections addresses a physiological system and its main pathologies, and presents a schematic view of the leading biomaterials, implants, and prostheses used clinically to treat these pathologies. The focus is consistently on commercially available materials used by clinicians to treat, restore, and replace diseased tissue and organs.

1810

Kidney Esthetical applications

In 1953, the heart-lung machine was invented by John Gibbon which made easier cardiovascular procedures[1, 4].

1800

Skeletal system Others

1840

John Charnley[1, 5, 6] invented the first successful hip prosthesis. He used Teflon acetabular cup at first in 1958.

1830

Between 1957 and 1958, Earl E. Bakken, founder of Madtronic, Inc., developed the first wearable transistor pacemaker[1].

1820

In 1860, an optometrist Since 1809, named Adolf Fick dentists were introduced the using gold anchor glass contact to fastened tooth[1]. lenses[1]

Cardiovascular system

This discovery led to the first prosthetic vascular graft made in Vinyon N, textile fibers, implanted in humans in 1952[1].

1790

In 1790, Volta discovered that electric current stimulated hearing capabilities[2]

Sensorial organs

Throughout, the 1950’s to the 1980’s atrificial heart were implanted in animals and humans, but without satisfactory results (patients living up to 620 days)[1].

Sponges of poly (vinyl alcohol) were used in the 1950’s as breast prosthesis, but it had low success[1]

200

210

An iron dental implant was also found dated 200 A.D. in Europe[1]

600 220

Dental implants were seen as early as 600 A.D. in Mayan people using sea shells[1].

1870

1900

In 1960, it was Dr. Belding Scribner that made routinely dialysis treatment possible. He developed a shunt which gave an easy and safe access to the blood[1, 4].

1890

In 1894, Tetamore isbelieved to have use cellulose nitrate to reconstruct parts of the face[3].

The development of the first leaflet valve was made in 1960 by Warren Hancock[1].

1880

The replacement of maxilla and mandible by alloplastic implants were described by Martin in 1889[3].

The first successful mitral valve replacement in human was made in 1960 and the valve consisted of a silicone ball and poly (methyl methacrylate) cage[1].

1860

The first fully implantable pacemaker was developed in 1959 by engineer Wilson Greatbatch and cardiologist W. M. Chardack[1].

1850

In 1880, artificial parts were used by Kinsgley to reconstruct orbit, nose and palate[3]

Sutures are one of the oldest medical procedures. there is evidence that they have been used as long as 32 000 years ago[1]

278 | Part III New trends in sustainable development and biomedical applications

12.2 History of biomaterials

Injections of substances were first used as breast augmentation, but in the 1960’s California and Utah classified silicone injections as a criminal offense[1]

1930

1940

1960

In 1972 the first channel cochlear implant was introduced[7]

With hip prosthesis technology was developed knee replacements leaded by Drs. Frank Gunston and John Insall in 1968 to 1972[1] Stent were design and developed by Dr. Julio Palmaz to 1978 to 1986. Eventually, Jonhson and Jonhson adopted the project and clinical trials were instituted under the food and drug Administration (FDA)[1]

1970

In 1984, the first multichannel cochlear implant system was introduced by the Cochlear Corporation[7]

1980

In 1998, the first tissue engineering product was approved by the FDA as living skin equivalent[8]

2000

In 1947, Dr. Arthur Voorhees noticed tissue had grown around a silk suture left inside a lab animal[1].

In 1995, Charles Vacanti seeded a polymeric scaffold in the shape of a human ear and implanted it on the back of a nude mouse[8]

1990

In 1945, Willem J. Kolff used a rotating wooden drum around Plastic contact The first successful which a new membrane lenses in poly(methyl dental implants was made of cellophane methacrylate) were made in 1937 by was wrapped to treat developed between Strock at Harvard successfully a 67-year-old the years 1936 using a screw-type patient with acute kidney [1] [1] to 1948 implant of Vitallium . failure[1, 4]

1950

In 1932, Dr. Kazanjian provided the initiative for maxillofacial, dental, and plastic surgeons to work together for the improvement of care for patient[3]

In 1964, Dr. Branemark made the unexpected discovery of osseointegration of titanium. most of dental and orthopedic implants are now made of titanium and its alloy[1].

In 1930-31, two groups invented the portable peacemaker[1].

John Charnley evolved to highmolecularweight polyethylene cup in 1961 to reduced wear debris[1, 5]

1920

In the 1960’s Thomas Cronin and Frank Gerow invented a silicone shell filled with silicone gel[1]

1910

In 1912, John Jacob Abel and It is in 1912 when his colleagues Dr. Alexis Carrel developed methods developed the first to anastomose blood functioning dialysis vessel that vascular machine in order to investigate by grafting would products in rabbit eventually see the [1] blood[4] day .

12 Advanced materials for biomedical applications |

279

280 | Part III New trends in sustainable development and biomedical applications

12.3 Basics in material science for biomaterial applications 12.3.1 Biomaterial properties Biomaterials present different properties according to their biological application. Each site of implantation has its own requirements with respect to anatomy, physiology, geometry, mechanical properties, and bioresponse. Mechanical properties will vary in terms of strength, elasticity, rigidity, flexibility; durability, permeability, and lubricity to name just a few. For example, materials used for hip replacement need to be strong and durable; heart valve leaflets must be flexible and tough. Similarly, bioresponses will also vary depending on the anatomical site involved and the intended use. Biological responses can be classified into three major categories: bioinert, bioactive, and bioresorbable. Bioinertia involves minimal interaction with the surrounding tissues, as a fibrous capsule usually forms around the implant. In contrast, a bioactive material will interact with the surrounding tissue, inducing cell adhesion, proliferation, and differentiation, and leading to integration of the biomaterial. Bioresorbable biomaterial refers to a material which will be degraded and eliminated through the metabolic activity of the human or animal cells. The physical or chemical behavior of a biomaterial can be tuned in order to control the biodegradation time, for example. Tissue engineering using scaffold and drug delivery systems benefits from controlled biodegradation time, to allow the regeneration of functional tissue or eluted drug delivery, for example. Other types of bioresponses may be required depending on the application. For example, hemocompatibility will be necessary for cardiovascular purposes. Furthermore, all biomaterials are required to be non-toxic, to lead to the appropriate host reaction (immune system) and to be sterilizable.

12.3.2 Biometals Metals are commonly used in biomedical applications for load bearing implants and internal fixation devices. When adequately processed, they display high tensile, fatigue, and yield strength properties, as well as relatively good biocompatibility. The properties of these metals depend on the processing method and the purity of the material. Some important physical properties of metals are luster (shininess), good conductivity of heat and electricity, high density, high melting point, ductility, and malleability. Moreover, metals intended for biological applications must have well-known chemical properties such as reactive capabilities, as well as corrosion and oxidation behaviors. Numerous metals are used as biomaterials, however the most common are stainless steel, cobalt chromium alloy (L-605), nickel titanium alloy, silver, titanium, platinum, gold, tungsten, and iridium.

12 Advanced materials for biomedical applications

|

281

Metal machining and forming are the activities and processes used to give a piece of metal a precise shape. In contrast, molding involves melting metal and pouring it into a casting mold to create a particular product with a desired form. The way in which these steps are carried out is very important, since heat and plastic deformation can strongly affect the mechanical properties of metals, thus extra annealing steps are generally required. Corrosion is the gradual degradation of materials by electrochemicals and is of great concern, particularly when a metallic implant is placed in the hostile electrolytic environment of the human body. Implants in the human body face corrosive environments such as blood and other constituents, body fluids, including several components such as water, sodium, chlorine, proteins, plasma, amino acids, along with mucin in the case of saliva. The aqueous medium in the human body consists of various anions such as chloride, phosphate and bicarbonate ions, cations (Na+, K+, Ca2+, Mg2+, etc.), organic substances of low molecular weight species, along with relatively high molecular weight polymeric components, and dissolved oxygen. Biological molecules disrupt the equilibrium of the corrosion reactions of an implant. In spite of the fact that most of the materials used are protected from environmental attack by surface oxide layers, there is clinical evidence for the release of metal ions from implants, this leaching being attributed to the corrosion process. The surface oxide film formed on metallic materials plays an important role as an inhibitor of the release of metallic ions into the surroundings of the implant. Moreover, reactions between the surfaces of metallic materials and living tissues change the composition of the surface oxide film. Metal ion release may also be accelerated by means of low concentration of dissolved oxygen, inorganic ions, proteins, and cells. The regeneration time of the surface oxide film after disruption also determines the number of ions released. The nature of the reactions which take place during the initial stages following implantation determines tissue compatibility. Thus, the success of the implant depends on the reactions which take place between the surface of a metallic material and living tissues directly after the implantation procedure. Therefore, surface oxide films present on metallic materials play a major role, not only in corrosion resistance but also in biocompatibility.

12.3.3 Bioceramics Polycrystalline ceramics are widely used as implant materials in the human body due to their excellent biocompatibility [9]. Subsequently, ceramics used for this purpose are termed “bioceramics”. Bioceramics can have structural functions as joint or hard tissue replacements, or can be used as a coating on metal implants to enhance fixation and bonding with surrounding tissues. They can also be used in bulk, porous or powder form providing temporary or permanent fillers. Applications of bioceramics can be categorized in two groups: load bearing, and non-load bearing applications.

282 | Part III New trends in sustainable development and biomedical applications Alumina (Al2 O3 ), zirconia (ZrO2 ) and their composites are common inert, load bearing bioceramics used in hip and knee prostheses due to their strength and good corrosion resistance. However, these materials are usually bioinert, as they do not form good bonds with living tissues [10]. The non-load bearing application of bioceramics can be classified into resorbable and non-resorbable bioceramics. Non-resorbable bioceramics are bioactive materials which encourage the formation of a biological bond between tissues and the implant without experiencing degradation over time. On the other hand, resorbable bioceramics such as calcium phosphate ceramics (CPC), hydroxyapatite (HA), tricalcium phosphates (TCP), bioactive glasses, bioactive glass-ceramics, corals, and calcium sulphate degrade upon implantation and are replaced by living tissues in the host under physiological conditions. Calcium phosphate bioceramics may be bioactive or bioresorbable depending on the calcium to phosphor ratio [11]. The temperature and the presence of water, both during processing and in vivo, determine the stable phases of calcium phosphate. Like most ceramics, calcium phosphate exhibits low strength when compared to metals, and has a tendency towards brittle fracture. Table 12.1. Characteristics of common ceramics biomaterials. Material

Structure

Young’s modulus (GPa)

Compressive strength (MPa)

Hardness (GPa)

Fracture Density toughness (g/cm3 ) (MPa m1/2 )

Alumina

Octahedral

300–400

2 000–4 000

20–30

5–6

3.9

Zirconia

Monoclinic tetragonal cubic

150–250

2 000

10–30

4–12

6

HA

Hexagonal rhombic

70–120

1 000–1 500

3–9

1–2

3.1

Bioglass

Amorphous

30–35

500

D > 6 mm) – Thrombogenic – Non-compliant – Patency rate: 20% with a microvessel PTFE graft (diameters lower than 6 mm) after four weeks in vivo. – Currently used for large vessel diameters (D > 10 mm) – More thrombogenic than ePTFE – Patency rate: 95% after 5 years in vivo, 40–50% after 10 years for lower limb bypass implantation.

– Inert surface

– Relatively good mechanical properties (tensile stress, compliance. . . ).

Polyethylene terephthalate (PET)

– Woven or knitted crimped textile structure – Implanted by open surgery

– Microporous structure Vascular Expanded prostheses [22] polytetrafluoro- – Implanted by open surgery – Used for most serious cases of ethylene vascular diseases (ePTFE)

– Highly thrombogenic

– Percutaneous implantation

Disadvantages

Stent framed valves

Advantages – Loud opening and closing sound. – Thrombogenic – Ball or disc variability or failure with different degrees of severity – 75% survival rate of the Bjork-Shiley mechanical heart valve implant after 20 years. However, only 18% had conserved their original implant. The remaining patients mainly died of heart valve malfunction.

Caged-ball valves

Aortic and mitral heart valve replacement [17]

Description

– Different geometries available, based on available hemodynamic features. Caged-disc – Implanted by sternotomy and thovalves racotomy surgeries, but newest Bileaflet tilting valves are stent-framed and are disc valves increasingly implanted by percuta(synthetic or neous surgery. natural leaflets)

Material & Devices

Application

12 Advanced materials for biomedical applications |

289

Material & Devices

– Impregnation of natural gels (collagen, gelatin, albumin. . . ) or antibacterial agents (silver, carbon)

Description

– Same as vascular stents.

Endovascular stents

– Can cause arterial wound.

The main problems experienced by vascular implants are due to infections or system malfunctions, whereas failures of vascular prostheses are currently caused by intimal hyperplasia, false anastomosis aneurysm or thrombosis of conduits [26–28].

– Bioresorbable – Based on regenerative medicine principles.

Drug eluding polylactic acid [24, 25]

– Polymeric fabric supported by a stent – Implanted by angioplasty – Used to restore blood flow after an aneurysm.

– Shape-memory alloy.

– Optimal tensile and yield strengths.

– Improved surface properties.

– Optimal elongation properties.

– Thrombogenic – Could cause arterial wound – Restenosis may appear in 30–40% cases after 6 months due to arterial lesions.

– Quickly degraded

– Improved blood permeability. – Improved biocompatibility or infection resistance. – Less thrombogenic during the first days after implantation.

Disadvantages

Advantages

Nickel-titanium alloy (Nitinol)

Vascular stents Stainless steel – Tubular metallic or polymeric [22, 23] (316L) meshed device – Implanted by angioplasty Drug-coated – Used to restore blood flow after an 316L stainless atherosclerosis-induced constricsteel tion. L-605 cobaltchromium alloy

Heparin bounded PET

Vascular Impregnated prostheses [22] PET

Application

290 | Part III New trends in sustainable development and biomedical applications

12 Advanced materials for biomedical applications

|

291

12.4.2 Musculoskeletal system Anatomy and physiology The musculoskeletal system enables movement, weight support and protection of soft tissues and organs from the environmental stresses and strains experienced by the human body. It consists mainly of bones, cartilage, tendons, ligaments, joints, muscles, and teeth (Fig. 12.2).

Cellular and collagen fibre organisation of cartilage Cellular organisation

Collagen fibre architecture Articular surface Superficial tangential zone Middle zone Perimysium Deep zone Calcifield zone Subchondral bone

Collagen hierachical structure in tendons and ligaments Collagen fibril (100 nm) Collagen fibre Primary fibre bundle (subfascicle)

Secondary fibre bundle (20-200 μm)

Tertiary fibre bundle

Bone hierachical structure Osteocyte

Haversian system Primary structure

Nutriment canal (for blood vessels and nerves into and from marrow) Contains yellow marrow Compact bone tissue Spongy bone tissue

α-chains Hydroxyapatite nanocrystals Collagen fibrils Spongy bone tissue

Collagen fibers Compact bone tissue

Fig. 12.2. Anatomy and physiology of the musculoskeletal system.

Outer layer (dense connective tissue)

Lamellae

292 | Part III New trends in sustainable development and biomedical applications Human teeth are composed of three primary tissues: enamel, dentin, and pulp. Enamel forms a protective layer at the anatomical crown of the teeth and is known as the hardest substance in the human body. Dentin is the biggest component of teeth and is a mineralized tissue with an overall composition similar to bone. It is composed of approximately 70 percent in weight (wt%) mineral phase (apatite form), 20 wt% collagen, and 10 wt% water. Bone is a high strength material which serves both as load bearing and non-load bearing structure. In biological terms, bone is described as a connective tissue, and in materials terms bone is a natural composite material which contains both mineral (solid matrix) and organic (soft matrix) phases. The structure of bone biocomposite contains around 65 wt% mineral phase (calcium phosphate as the main inorganic component), 25 wt% collagen, and 10 wt% water. Additionally, it also contains small amounts of organic materials such as proteins, polysaccharides, and lipids. Hyaline articular cartilage acts as a low-friction, wear-resistant, load bearing surface in synovial joints and, combined with synovial fluid, facilitates the absorption of impact and the smooth transmission of loads from the joint to the underlying bone. Cartilage may be considered a porous composite organic solid matrix swollen by water. It is constituted of chondrocytes embedded in a dense extracellular matrix (ECM), mostly composed of a collagen type II fibril network, proteoglycan and water. Their respective concentration varies with depth. Ligaments and tendons are tough, flexible, and highly anisotropic bands of fibrous connective tissue ensuring the transmission of loads between the elements they connect: tendons link muscle to bone, while ligaments connect the extremities of bones. Although their functions are different, the structure and composition of tendons and ligaments are similar from an engineering standpoint. They both consist of relatively few cells (fibroblasts make up around 20% of the total volume) and a significant amount of extracellular matrix. Roughly 70% of this ECM is water, the rest of the solid elements being collagen, elastin, and ground substance (proteoglycans, glycosaminoglycans, and other glycoproteins) which stabilize the collagen organization. The viscoelastic properties of these tissues, highlighted by the creep and stress-relaxation behavior of ligament and tendons, are mainly due to this ground substance. The arrangement of collagen fibers in tendons and ligaments differs: they are bound together in parallel (tendon) or nearly parallel (ligament) bundles arranged along the central axis. In the first case, this arrangement creates a stiff (Achilles tendon Young’s modulus is 0.8–1.5 GPa), highly anisotropic material with high tensile strength (50–125 MPa), but with little resistance to shear and compression. As for the ligament, the slight misorientation, combined with the larger elastin content, allows nonaxial loads to be carried, but also reduces their overall stiffness and tensile strength. This hierarchical structure is also present in muscles, where the building blocks are myofibrils (bundles of actin and myosin proteins) instead of collagen.

12 Advanced materials for biomedical applications

|

293

Table 12.5. Composition of musculoskeletal tissues (%). Components

Hard tissues

Soft tissues

Bone

Enamel

Dentin

Cartilage

Ligament

Tendon

Water

8–10

1–2

8–10

65–80

60–80

60–80

Solids

90–92

98–99

90–92

20–35

20–40

20–40

Collagen

< 30

70

> 98

> 70

–

–

–

15–30

15–25

10–12

>5

2

Proteins

–

–

–

Proteoglycan & ground substance

–

–

–

Elastin

–

–

–

Musculoskeletal traumas and diseases Musculoskeletal disorders have been identified as among the most important health conditions which exist today, according to World Health Organization (WHO) reports. As a result of traumatic injuries such as bone fractures, bone disorders, dislodged teeth, or non-traumatic injury like age related degeneration, tissues may require repair, replacement or reconstruction [29]. Biomaterials designed for applications in the oral and maxillofacial regions are among the most common to have reached the clinical stage. Restorative materials can be separated into direct devices, which can be directly placed in or on the natural dental tissues; and indirect devices, which are produced outside the mouth and then bonded in or onto the remaining natural tissue. In more severe conditions (i.e., missing tooth), common treatments involve the use of partial or full dentures. However, due to their recent success, there has been a dramatic increase in the inclusion of dental implants as a viable treatment option. Design and production of dental implants have been vastly studied and commercialized; parameters such as length, implant diameter, implant geometry, and surface characteristics are customized from case to case [30]. Bone restoration can be achieved via transplantation or implantation. Autograft or allograft tissues can replace damaged bone tissue, but this replacement may lead to several problems (donor site morbidity, limited availability, transmission of diseases, etc.). For these reasons, there is a growing need for the production of artificial bone fixations and bone fillers [31]. Various forms of bone-graft substitutes are available and include allograft bone preparations such as demineralized bone matrix and calciumbased materials. Human synovial joints, such as ankle, fingers, hip and knee, are replaced with artificial joint implants. The most common joint replacement implants are weightbearing joints like the hip and knee [32]. From a materials science point of view, joint

294 | Part III New trends in sustainable development and biomedical applications replacements can be considered devices for bearing implants. A variety of design parameters based on many surgical factors (such as length of stem, diameter of hip head, etc.) governs the choice of weight-bearing joint replacement implants. One of the most important challenges in hip replacement is the need to improve load transfer to the bone and reduce the incidence of loosening and stress shielding which can lead to implant dislocation in long term applications. The majority of current clinical procedures in the case of osteoarthritis (also known as degenerative joint arthritis and involving the degradation of articular cartilage) involve non-biomaterial related approaches. Several therapeutic interventions have been developed to induce cartilage healing [33]. Among the grafting procedures, osteochondral autograft transfer (mosaicplasty) and autologous chondrocyte implantation (ACI) are popular strategies used in clinical situations. While the former only consists of the implantation of multiple osteochondral plugs (forming a mosaic pattern) in the damaged joint, the latter procedure, ACI, involves a three-step procedure of cell harvesting (from non-weight bearing sites of the knee), in vitro chondrocyte culture and cell implantation [34]. In the first generation of ACI, in clinical use since 1987, a periosteal (outer surface of bones) cover was used to seal the wound and the culture-expanded autologous cells were injected via an aqueous solution. However, thanks to the technological advances of biomaterial research, new generations of ACI have initiated the use of purpose-designed biomaterials (see Table 12.6). The high prevalence of anterior cruciate ligament (ACL), rotator cuff, and Achilles tendon injuries as a result of occupational and sporting injuries has become a major economic burden for healthcare systems worldwide [35]. Tendon and ligament regeneration procedures have proven to be a considerable clinical challenge owing to the high mechanical stresses placed on these tissues. Currently, the most successful strategy remains grafting; autografts are the most common procedures (often referred to as the gold standard) used when tissue defects cannot be repaired naturally [36, 37]. However, their use is limited by the availability of healthy tissues and is often accompanied by donor site morbidity and pain. Allografts and xenografts, on the other hand, require intense sterilization procedures which are detrimental to the biomechanical properties of the tissue and can lead to increased graft failure [38]. Furthermore, they are susceptible to disease transmission from the donor, restricting the already limited availability of transplants. The use of synthetic polymers as scaffolds was also attempted for tendon and ligament regeneration, with limited long-term success [35, 39]. Synthetic scaffolds all displayed adequate initial performances due to their superior mechanical properties. However, they showed little host tissue in-growth, inadequate fiber abrasion resistance, and their degradation products often caused important inflammatory responses, which ultimately led to these devices no longer being used. Commercial biological scaffolds did not fare better in clinical applications. While they were able to induce tissue formation, their mechanical properties were significantly lower than native tissues, which led to multiple reported failures in clinical studies [35, 40].

Orthodontic products

Wires Bracket Screws Bands SS CoCr alloy NiTi alloy Plastics Elastomers

Braces and orthodontic treatments used to correct malocclusion.

Effective for large dental defects Better esthetics Better wear resistance Higher biocompatibility

– Good corrosion resistance – Superior properties such as rigidity flexibility, fatigue, and durability.

– – – –

– Metal ion release – Sensitivity to metal traces – Esthetic problems for metallic materials – Low operability, low resilience and fatigue for polymeric materials.

– Improving esthetics – Costly – Require multiple appointments and laboratory time and production personnel.

– Chemical leakage from the material – Matching shade, translucency, and texture with surrounding teeth – Lower biocompatibility – Minor tooth tissue damage due to heat generated during restoration.

– Fast – Cost effective – Relatively good mechanical properties.

Replace diseased, discolored and malformed orodental structures in order to improve their function, speech, comfort, esthetics and integrity.

Direct: Amalgam resin composites, glass inomer silicate cement.

Restorative dental materials

Indirect: inlays, onlays, veneer cast metals (gold) porcelain composites.

Challenges

Advantages

Description

Material & Devices

Application or Trauma

Table 12.6. Materials and devices for musculoskeletal system repair.

12 Advanced materials for biomedical applications |

295

Bone fixation

Pure titanium Ti-based alloys Zirconia composites Alumina composites

Dental implants [41]

Ti-based alloys Stainless steel CoCr alloys

Partial or full dentures/crowns CrCo alloys Gold alloys Porcelain

Material & Devices

Application or Trauma

– Surface modification – Implant loss due to insufficient suitable bone surface – Maintenance of healthy surrounding tissues during the first year of loading. – Osteointegration. – Poor biocompatibility, hypersensitivity – Improving esthetics for metallic implants – Loosening and cleaning – Maintenance of healthy surrounding tissues – Require highly accurate clinical and laboratory processing – Require multiple appointments and laboratory expertise for production. – Low to moderate resistance to fracture – Allergenic sensitivity to base metals. – Second operation required for removal.

– Almost 100% survival rate for metallic implants – Good mechanical properties – Less damaging to the gum – Higher comfort, chewing ability, and quality of life.

– Mimic the optical characteristics of enamel and dentin – Esthetics – Chemical inertia – Moderate resistance to fracture in high-load restorations

– High mechanical properties

Missing teeth due to periodontal disease. Dental implants produced in various shapes and sizes, placed in the jaw bone as artificial roots to support a crown or a denture.

Fractured bones are primarily repaired using fixation devices such as wires, nails, pins, screws, and plates.

Challenges

Advantages

Description

296 | Part III New trends in sustainable development and biomedical applications

The femoral part comprises of stem and head; it can be in one piece or as a modular design. A joint implant replacement usually requires the removal of a significant segment of diseased articular tissue, moreover subject to enormous repetitive stresses from body movement and weight-bearing.

Metal on plastic

Bone joint replacement implants

Ceramic on polymer

Metal on metal

Bone cements and fillers are used in the treatment of benign bone tumors, injuries, and disorders as alternatives to autografts or allografts

Description

Vitroceramics Calcium phosphates Demineralized human bone Ceramic collagen composites Ceramized allografts

Bioresorbable polymers [42] PLA, PLLA

Material & Devices

Bone fillers

Application or Trauma

– Better wear resistance – Lower fracture rate compared to ceramic on ceramic.

– Better wear resistance – Largest head design: higher stability and motion.

– High durability – Less expensive.

Good biocompatibility Unlimited supply Uniform quality High mechanical properties Osteoconductivity Good formability.

– Associated with adverse local and remote tissue responses – Failure diagnosis difficult – High articular wear rates.

– Few cases of device recalls by FDA – Concerns about long term biocompatibility due to chemical composition and biological response to debris.

– Low wear resistance – Osteolysis (bone loss) – Concerns about long term biocompatibility due to chemical composition and biological response to debris.

– Vascularization – Osetoinductivity – Oseteogenicity.

– Low mechanical properties – Long term balance between mechanical strength and material bioabsorption.

– Biodegradation of the whole material.

– – – – – –

Challenges

Advantages

12 Advanced materials for biomedical applications |

297

High wear resistance Durability Reliability Good biocompatibility.

Before implantation the autologous – chondrocytes are seeded on a – biomaterial carrier (i.e., three-dimensional scaffold/matrix or gel) upon which they are grown. The biomaterials with seeded cells are – trimmed to the precise defect size and implanted without the use of a periosteal cover or fixing stitches (fibrin glue is often used for fixation). –

Three-dimensional cellular architecture Higher mechanical stability (compared with 1st and 2nd generation ACI) Scaffolds may act as a barrier to the invasion of the graft by fibroblasts, which may otherwise induce fibrous repair Faster, less extensive procedure, suture-free – Some models (especially gels) can be applied with minimal invasive surgery – High biocompatibility.

– – – –

Advantages

3rd generation – Hyalograft C (hyaluronan ACI (matrixpolymer) assisted) – CaRes (collagen gel) – ArthroMatrix, MACI, MACT (collagen membrane) – Bioseed-C (polymer) – Novocart 3D (polymer) – Cartipatch (natural hydrogel)

Description

– Reduced risk of hypertrophy – No second operation site, leading to shorter operation time and the absence of donor site morbidity – No patient-related material quality issues – Easy to handle and tear-resistant

Ceramic on ceramic

Material & Devices

2nd generation Chondro-gide (collagen Porcine-derived type I/III collagen ACI membrane) bilayer membranes are used to suture defect before injection of a cell suspension. The membrane is composed of a compact and a porous side. The compact side prevents cells from diffusing into the synovial fluid while the other layer favors even cell distribution and attachment.

Application or Trauma

– Three-stage procedure since cell harvesting and culture are still necessary – Ex vivo chondrocyte culture – Lack of long-term follow-up – Few materials are FDA approved – The properties of healthy cartilage tissue are still unmatched by any available substitute.

– Three-stage procedure – Uneven distribution of chondrocytes – Cartilage damaging sutures required for fixation – Low mechanical stability.

– Brittleness – Catastrophic fracture.

Challenges

298 | Part III New trends in sustainable development and biomedical applications

– Mechanical properties are significantly lower than native tissues – High failure rates – Inflammatory responses and edema have been reported with the use of biomaterials derived from porcine/bovine sources.

– Tissue ingrowth varies with the material but is usually very incomplete – Stress shielding of maturing tissues, limiting the transfer of load-bearing responsibilities – Wear debris can induce long-term complications & inflammatory responses – Limited biocompatibility – Mechanical failure caused by swelling and separation of fibers.

– No donor availability problem – Biochemical composition similar to tendons and ligaments – Surface chemistry and structure favorable to tissue ingrowth – Most are FDA-approved.

– Superior mechanical properties – Easy to handle – Effective in treating complex instabilities – Good short-term clinical results.

Derived from mammalian tissues. Processed through cascade steps to isolate an acellular collagen scaffold. Removal of noncollagen components to prevent host rejection. Used as alternative to autografts or allografts.

Synthetic fibers are woven, braided or knitted into a high strength scaffold. Used as alternative to autografts or allografts. Several prostheses have been withdrawn from the market due to unsatisfactory long-term results.

– Restore (porcine submuscosa) – GraftJacket (human cadaver dermis) – Zimmer (porcine dermis) – TissueMend (fetal Bovine Dermis) – OrthADAPT (bovine/porcine pericardium)

Gore-Tex (PTFE) Dacron (PET) Stryker (PET/PP) Leeds-Keio ligament (PET) – Kennedy-LAD (PP)

Tendon and ligament injuries (biological scaffolds)

Tendon and ligament injuries (synthetic scaffolds)

– – – –

Challenges

Advantages

Description

Material & Devices

Application or Trauma

12 Advanced materials for biomedical applications |

299

300 | Part III New trends in sustainable development and biomedical applications 12.4.3 Visceral organs The digestive and excretory systems are comprised of the gastrointestinal tract, solid organs including the liver and pancreas, and the urinary tract. The main function of the gastrointestinal tract is to break down and absorb ingested nutrients. The liver and pancreas participate in the breakdown of carbohydrates, proteins, and fats into smaller molecules such as glucose, amino acids, and fatty acids by secreting digestive enzymes into the small intestine (Fig. 12.3). Oesophagus

Pancreas Acini (digestive enzyme production)

Lumen Mucosa Submucosa Muscularis propria

Islet (hormone production)

Liver (hepatic lobule)

Kidney (glomerulus) Blood flow

Bile flow out

Capsule Glomerulus

Blood flow in

Hepatocytes Blood flow (metabolic reactions out ans detoxification)

Filtrate

Fig. 12.3. Anatomy and physiology of visceral organs.

In addition to their digestive functions, the liver, pancreas, and kidneys play a crucial role in metabolic and hormonal regulation. For example, the liver stores glucose in the form of glycogen and synthesizes proteins from amino acids. The pancreas controls blood glucose levels by releasing hormones such as insulin which increases glucose uptake by the muscles and decreases glucose release by the liver. Waste products produced by liver, muscles and other organs are finally removed from the blood stream by the kidneys. The kidneys first eliminate small molecules non-selectively by size-based filtration in the glomerulus, and then reabsorb useful nutrients and osmolytes. In sum, each organ within the digestive and excretory systems plays a role in the transport, filtration, and transformation of nutrients or waste products. The structural role and the absorption and filtration capacities of these organs can be assisted by medical devices, whereas novel cell-based therapies may be required to fully replace their metabolic and secretory functions.

Composed of a blood glucose sensor, an external insulin reservoir, an open-loop electronic controller and a subcutaneous cannula to deliver rapid-acting insulin.

Salivary bypass stent

External insulin pumps

– Post laryngoesophagectomy fistula – Palliation of advanced pharyngeal and cervical esophageal cancer – Managing traumatic injuries of the cervical esophagus

Pancreas failure

Compared to manual injections, insulin pumps improve blood glucose control [43], which may reduce the life-threatening complications of diabetes.

Soft and pliable stent allowing easy delivery using the push-through technique.

Stents made of medical-grade silicone.

Esophageal selfexpandable plastic stents (SEPS)

– Tracheoesophageal fistulas – Benign esophageal strictures – Esophageal perforations and leaks – Malignant airway disease or silicone.

Stents made of metals such as – Appropriate mechanical nitinol and stainless steel. properties to resist expanThey can be covered, uncovered sion or partially covered with – Porosity promotes gas and polyurethane, polyethylene, nutrient exchange polytetrafluoroethylene (PTFE) – Biocompatible, allowing or silicone. adhesion and proliferation Stents mainly made of polyester of cells as well.

Esophageal selfexpandable metal stents (SEMS)

– Palliation of malignant dysphagia – Gastroduodenal outlet obstructions

Advantages

Description

Material & Devices

Application

Table 12.7. Materials and devices for repair of visceral organs.

Costly, requires human input and prone to catheter-related complications

Clinical complications: – stent migration – acquired tracheoesophageal fistula (TEF) – difficulty in swallowing – inadequate expansion with increased post-procedural dysphagia – variable throat or chest pain, esophageal erosions with bleeding and fistulization – significant reflux

Disadvantages

12 Advanced materials for biomedical applications |

301

Kidney and liver failure

Application

– Medications are needed to control the levels of minerals and replace hormones. – Some patients do not have suitable blood vessels for establishing an access site. – A permanent abdominal catheter in the case of peritoneal dialysis, is required, with the risk of peritonitis. – The difference in composition of dialysis fluid may have deleterious downstream effects and metabolic abnormalities [47].

– Bioartificial kidney can Separation process: mimic glomerular funcpressure-driven ultrafiltration. tions [45]. Made of polysulfone, polyvinylpyrrolidone, modified – Renal function can be partially replaced by dialysis cellulosic membranes, through semi-permeable polyacrylonitrile. membranes with wellSample retains molecules of defined porosities made 8 ∼ 60 kDa: cells and most from natural or synthetic proteins including albumin. polymers. – Membranes can be coated with biomolecules such as heparin to reduce membrane thrombogenicity, or with molecules which increase membrane permeability, such as extracellular matrix Separation process: albumin proteins [46]. dialysis Made of acrylonitrile/ sodium – Certain liver support systems can reduce the incimethallyl sulfonate copolymer. dence of brain dysfunction Sample retains molecules of (hepatic encephalopathy) 35∼40 kDa: cells and most by reducing the levels of proteins including albumin. molecules such as ammonia.

Hemodialysis or hemofiltration

Single pass albumin dialysis (SPAD)

∼ 7 year battery lifespan, surgical complications, need for in-hospital insulin refills, as well as insulin aggregation leading to pump or catheter obstruction [44].

Delivers insulin into the peritoneal cavity through a polyethylene-lined silicone catheter.

Disadvantages

Implantable pumps

Advantages

Description

Material & Devices

302 | Part III New trends in sustainable development and biomedical applications

Application Separation process: albumin dialysis, charcoal and resin adsorption. Made of polysulfone. Sample retains molecules of 50 kDa: cells, large proteins, albumin (with significant albumin losses). Separation process: plasmapheresis, usually charcoal or anion-exchange resins. Made of cellulose acetate or polysulfone. Sample retains molecules of 100 ∼ 400 kDa: Cells, multimeric proteins. Separation process: 1) plasmapheresis, 2) neutral resin, 3) anion exchange. Made of polysulfone. Sample retains molecules of > 100 kDa: cells, large proteins (e.g., antibodies). Separation process: plasmapheresis Made of polypropylene Sample retains molecules of > 1 000 kDa: cells.

Molecular Adsorbent Recirculating System (MARS)

Bioartificial liver

Fractionated plasma separation and adsorption (PROMETHEUS)

Therapeutic plasma exchange

Advantages

Description

Material & Devices

Disadvantages

12 Advanced materials for biomedical applications |

303

304 | Part III New trends in sustainable development and biomedical applications 12.4.4 Nervous system and sensory organs Nervous system The human nervous system is equipped to sense and respond to continuous changes within the body and its external environment. The fluctuations of the internal and external environments are regulated by the nervous system to maintain homeostasis (state of relative stability within the body). Homeostasis is critical for survival because the body can only survive within a narrow range of conditions. The nervous system monitors and controls most body processes, from automatic functions (such as breathing) to activities involving fine motor coordination, learning, and cognition (such as playing a musical instrument). The nervous system has two major divisions: the central nervous system (CNS), and the peripheral nervous system (PNS). – The CNS integrates and processes information sent by nerves through the brain and the spinal cord. – The PNS includes nerves which carry sensory messages to the CNS and nerves which send information from the CNS to muscles and glands. It is further divided into the somatic system and the autonomic system. – The somatic system consists of sensory receptors in the head and extremities, nerves which carry sensory information to the CNS, and nerves which carry instructions from the CNS to the skeletal muscles. – The autonomic system controls glandular secretions and the functioning of the smooth and cardiac muscles. The nervous system is composed of neurons and cells which support the neurons (glial cells). Neurons are the basic structural and functional units of the nervous system. They are specialized to respond to physical and chemical stimuli, to conduct electrochemical signals, and to release chemicals which regulate various body processes. The activity of neurons is supported by another type of cells called glial cells. Glial cells outnumber neurons by about 10 to 1, and they account for about half the volume of the nervous system. Collectively, glial cells nourish neurons, remove their waste, and defend against infection. Glial cells also provide a supporting framework for all nervous-system tissue. Neurons share four common features: dendrites, a cell body (soma), an axon, and branching ends.

Nervous system damage and disorders Nervous system disorders and damage can be caused by a wide range of conditions including infections, hypoxia, poisoning, stroke, chronic degenerative disease, and acute trauma. Some of the most problematic forms of nervous system disorders and damage are those related to chronic neurodegenerative disease or acute trauma caused by injury. After injury to the nervous system, a complex cascade of events proceeds in a spatially and temporally specific manner. The effect on the nervous

12 Advanced materials for biomedical applications

|

305

system will vary depending on the type and extent of the injury [48]. Some examples of damage are: cell death related to lesions and Wallerian degeneration, die-back of axons, glial scarring which inhibits axonal growth, and breakdown of native ECM components such as hyaluronic acid [49].

Ocular system Light waves enter the eye through the cornea and are converged by the cornea and the crystalline lens. These waves continue through the clear vitreous humor (a gel) that accounts for about 80% of the volume of the eye, and are focused on the retina. The retina changes light into electrical signals which are sent through the optic nerve to the occipital cortex at the back of the brain. The eye is comprised of three tissues subject to replacement by biomaterials: the cornea, the intraocular lens, and the retina. Moreover, to correct focus problems, lenses in direct contact with the cornea are used. For example, retinitis pigmentosa (RP), an inherited retinal degenerative disease which often results in almost complete blindness, affects roughly 100,000 US citizens. The worst-affected RP patients, can be helped by a retinal prosthesis, named Argus® II. The system works by converting video images captured by a miniature camera housed in the patient’s glasses into a series of small electrical pulses which are transmitted wirelessly to an array of electrodes on the surface of the retina. These pulses are intended to stimulate the retina’s remaining cells resulting in the corresponding perception of patterns of light in the brain. The patient then learns to interpret these visual patterns, thereby regaining some visual function. Second Sight, manufacturer of Argus® II, gained European approval (CE Mark) for the system in 2011 and FDA approval in 2013. This approval was given under a Humanitarian Device Exemption intended to expedite market introduction of technologies for the treatment of smaller, underserved patient populations. It is the first approved retinal prosthesis in the world.

Auditory system Humans can perceive sounds in two ways, by air and by bone conduction. Air conduction uses the ear canal, eardrum, middle ear, and the inner ear to transduce sound waves to the CNS. In contrast, sound is transmitted via bone conduction directly through the bones, bypassing the outer and middle ear. People with a hearing impairment use hearing aids to amplify sound so that damaged ears can detect it. Cochlear implants bypass damaged portions of the ear and directly stimulate the auditory nerve. A microphone picks up sounds from the environment, a speech processor selectively filters sound, splits it into channels and sends the electrical signals to the transmitter. The processed sound signals are transmitted across the skin by electromagnetic induction. The internal stimulator sends signals to electrodes inserted through the cochlea, which send the impulses to the nerves in

Material & Devices

Pathological disorders

Typically consists of a proximal catheter, which runs from the cerebral ventricle’s subarachnoid spaces to a valve which connects to a distal catheter [50]. Medical grade silicone is almost the only material used to produce CNS shunts.

Description

Stimulating Microwire electrodes are fine wires of electrodes [52] 20–50 cm in diameter made of conductive metals insulated with Teflon or polyimide. The tips of the wires are not insulated and are used to record neuronal signals [53, 54]. Silicon micromachined microprobes have unsurpassed control over electron size, shape, texture, and spacing, and allow for multiple recording sites to be Recording electrodes or placed on a single electrode shank brain computer due to their photolithographic interfaces [56] processability [55].

Hydrocephalus CNS shunt treatment system

Application

Table 12.8. Materials and devices for nervous system repair.

These shunts have high failure rate (40–50%) during the first 2 years following implantation [51]. Complications can be due to obstruction, mechanical failure, infection or excess drainage. The formation of glial scars around the implanted electrodes hinders their long-term performance.

These shunts relieve pressure by draining excess cerebrospinal fluid from the cerebral ventricles or subarachnoid spaces into a less constrained area of the body.

Used to restore hearing and alleviate symptoms of Parkinson’s disease.

Used to control mechanical devices for the control of, for example, prosthetic limbs or motorized wheelchairs for paralyzed patients.

Disadvantages

Advantages

306 | Part III New trends in sustainable development and biomedical applications

Dendrimers

Methylprednisolone

Degradable Local drug delivery to the and nondegradable CNS polymers [60, 61].

Traumatic spinal cord and brain injury

Description

Advantages

The only clinically used therapy.

Local drug delivery circumvents the difficulty of penetrating the BBB, systemic side effects and toxicity, peripheral drug inactivation, and modification of the carrier surface.

Implants at injured site using hydrogels Hydrogel scaffold ability to gel in situ is an important property in order to fill impregnated with therapeutic molethe irregular lesion cavities. cules [64, 65].

Intravenous or intraperipheral injections [62, 63].

Polymeric implants in the CNS are commonly used in local drug delivery.

Repeatedly branched polymer molecules which contain a cascade of branches grown from one or several cores

Small vesicles in which an aqueous inner core is entirely enclosed by unilamellar or multilamellar phospholipid bilayers.

The blood-brain barrier (BBB) is the principal obstacle to delivering drugs into the CNS, as it forms a barrier for the entry of therapeutic agents from the bloodstream. Systemic drug carriers Polymeric and Particles ranging from 10 to 1 000 nm in have been developed to avoid the size, in which therapeutic drugs can be ceramic BBB [59]. nanoparticles adsorbed, dissolved, entrapped, Intravenous or intraperipheral injecencapsulated or covalently attached. tions are used for systemic drug Polymeric Micelles form spontaneously in the delivery, which is non-invasive. micelles aqueous solution of amphiphilic block copolymers, have core-shell architecture and carry drugs in their core.

Material & Devices

Systemic drug Liposomes delivery to the CNS [57, 58]

Application

As yet only in clinical trial stage.

This could have adverse systemic side effects like immune suppression.

Local drug delivery is limited by the invasive nature of surgical implants of polymers and the difficulty of attaining long-term drug delivery at therapeutic levels.

Systemic delivery requires large doses to achieve therapeutic concentrations at the target site, which can affect non-targeted tissues and organs. Inactivation of peripheral drug and modification of carrier surface are common drawbacks while in the bloodstream.

Disadvantages

12 Advanced materials for biomedical applications |

307

Corneal replacements [67]

Poly(methyl methacrylate) prosthesis

– Used in case of transplant failure or organ donor shortage.

– High rate of long-term success

– Results depend on the progress of the disease.

– Dryness of eyes due to hydrophobicity

– Highly oxygen-permeability: Dk = 130 through the center and Dk = 84 through the border

Hybrid contact – A rigid gas-permeable central optilenses cal zone surrounded by a peripheral fitting zone made of a soft contact lens material.

– Less optical features than gaspermeable contact lenses – Dryness of eyes because of hydrophobicity – Not suitable for all patients.

– Absorb significant amounts of water (40 to 75%), allowing oxygen to pass through the lens – More comfortable.

Disadvantages

– Good oxygen permeability: Dk* between 12 and 150 – Superior optical features, provide sharper vision than soft lenses – Chemically inert.

– Silicone hydrogel.

Soft contact lenses

Focus image on the retina [67]

Advantages

Gas-permeable – Poly(methyl methacrylate) rigid contact lenses microporous structure.

Description

Material & Devices

Application

Table 12.9. Materials and devices for ocular repair.

308 | Part III New trends in sustainable development and biomedical applications

– Designed for several refractive, diffractive or both foci.

Multifocal IOL

Accommodating – Designed for enabling IOL displaceIOL ments.

– IOL is injected into a capsular bag through a small incision in the peripheral cornea.

Foldable acrylate IOL

– Optic of 6mm diameter and intraPMMA intraocular lens ocular lens-haptics. (IOL)

Cataracts or myopia [68]

Description

Material & Devices

Application

– Allow accommodation with modulated dioptry.

– Reading glasses no longer necessary

– Surgical implantation has greatly reduced perioperative morbidity. – Disadvantages of PMMA IOL are greatly reduced.

Advantages

– Presbyopia from 45 years of age leads to reoperation.

– Reduces the quality of vision (contrast, creation of halos or glare) – Presbyopia after 45 years old leads to reoperation.

– Material aging induces presbyopia – Presbyopia from 45 years of age leads to reoperation. – Reading glasses required

– Could lead to posterior capsule opacification due to migration of lens epithelial cells – Surgically induced and corneal astigmatism – Spherical aberration – Risk of postoperative refractive errors. – Reading glasses required – Presbyopia from 45 years of age leads to reoperation.

Disadvantages

12 Advanced materials for biomedical applications |

309

310 | Part III New trends in sustainable development and biomedical applications the scala tympani and to the brain through the auditory nerve. The electrode array is usually made from silicone rubber, and the electrodes are made of highly conductive material (e.g., platinum). The Bone Anchored Hearing Aid (BAHA) system stimulates the cochlea by transmitting the sound waves through the bones, bypassing the outer and middle ear [69].

12.4.5 Esthetic applications Breast implants Breast implantation is one of the most widespread cosmetic surgical procedures; this surgery is used both for esthetic reasons and to correct a number of pathologies such as asymmetry, contracture, ptosis, infection, and necrosis, which can occur subsequent to different kinds of diseases such as breast cancer, malformations, and treatments. The number of breast implantation procedures performed in the US has doubled between 1998 and 2005 [70]. The breast contains the mammary gland, which is responsible for the production of milk for infant nutrition. It is composed of several superposed layers of tissues (glandular, fatty, and fibrous). It is positioned on the pectoral muscles of the external chest wall and attached to the chest wall by Cooper’s ligaments. The breast extends horizontally from the edge of the sternum to the mid-axillary line and vertically up to the axilla (axillary tail of Spence). A thin layer of connective tissue called fascia encircles the internal breast tissues; the superficial layer of the fascia is situated just under the skin, while the deep one is on the top of the pectoralis muscle. Usually an implant is inserted into the submuscular or subglandular plane. The glandular tissue of the breast is surrounded by a layer of fat which gives the breast a soft consistency. The glandular tissue houses the lobes, each composed of many milk secreting glands (lobules). Ducts are small vessels connecting lobes; they widen toward the nipple to form small bulbs called ampullas. Ampullas gather the milk produced by the lobules and make it available to the nipple during the lactation period. The two most important vessels in the circulatory system of the breast are the axillary artery (extending from the armpit and supplying the outer part of the breast) and the internal mammary artery (extending from the neck and supplying the inner part of the breast). Modern breast implants are composed of an elastomer containment shell and filler whose consistency should be close to that of the natural breast. Materials used can be separated into 3 groups: saline implants, silicone implants [71], and alternative composite implants.

12 Advanced materials for biomedical applications

|

311

Table 12.10. Materials and devices used in breast implants. Material & devices

Description

Advantages

Disadvantages

Saline implants

Liquid solution encapsulated in a room-temperature vulcanized shell made of a silicone elastomer.

Better performance; look and feels closer to normal breasts; resistant to rupture and gel exposure; completely harmless on rupture

Asymmetry, removal/replacement, malposition, loss of nipple sensation, wrinkling, palpability/visibility more evident than silicone

Silicone implant

Polydimethylsiloxane polymeric compound and gel exhibiting three degrees of cohesiveness.

Lower rates of malposition and asymmetry than their saline counterparts; better overall performance than saline implants

Silicone gel breast implants cost roughly as much as saline implants.

Alternative material implants

Soybean oil (Trilucent™ ) [72] Hyaluronic acid (HA) (Macrolane™ ), semi-liquid material.

Withdrawn from market in 1999. Local non-permanent procedure, low risk of allergic reaction or transmission of infections; does not require surgical intervention since it is injectable

Injection site pain and capsular contracture in some cases, reabsorbed after 12 months.

Maxillofacial prostheses Maxillofacial prosthetics are defined as a branch of prosthodontics concerned with restoration and replacement of both stomatognathic and associated facial structures by permanent or temporary artificial substitutes [73, 74]. Maxillofacial prostheses are used to correct body defects and to restore both appearance and functionality to impaired organs, leading to a relative physical and psychological recovery. Maxillofacial rehabilitation consists of repair of nose, maxilla, mandibula, oral cavity, ears, eyes and ocular orbits. Materials used in the different stages of maxillofacial prostheses creation can be divided into five groups: impression materials (plaster of Paris, plaster bandage, silicone putty, alginate), construction materials (for prosthetic, pattern, mold making, separating medium), pigmentation materials (intrinsic and extrinsic coloration), auxiliary materials (i.e., extrinsic sealant, Aerosil® 130, Silastic® foam, macrocellular foam, Comfeel® and materials for retention (primers, double sided adhesive tape,

312 | Part III New trends in sustainable development and biomedical applications magnets, implants, etc.). The following section of this chapter focuses mainly on materials and their properties. Modern techniques of maxillofacial prosthesis production use laser to measure the defect and computer aided manufacturing for the prosthesis realization.

12.4.6 Skin Anatomy and physiology The skin is the largest organ of the human body. It consists of three principal layers: the hypodermis, the dermis and the epidermis (Fig. 12.4). The skin plays the role of barrier between the organism and the environment, thanks to the stratum corneum which is the outermost layer of the epidermis, consisting of keratinocytes displaying different levels of differentiation through a pluristratified structure. The dermis is the mechanical support of the skin and plays an important role in thermoregulation. Vascular, nervous, and lymphatic networks are found in this layer. Fibroblasts, responsible for the synthesis of the extracellular matrix (collagen, elastin), are the main cell found in the dermis, although some immune cells can also be found in this layer. The hypodermis is an adipose tissue which stores fat. It contributes to the plasticity of the skin and thermoregulation [80].

Skin defects Wounds can be described as a defect or damage to the skin and can be classified as: – Acute: resulting from physical damage. These tissue injuries heal completely, with minimal scarring, usually within 8–12 weeks. Examples are mechanical injuries due to external factors (abrasions), penetrating wounds (surgical wounds, knife and gun shot wounds, tumours), chemical injuries (radiation, electricity, corrosive chemicals, and thermal sources), and burns [81, 82]. – Chronic: resulting from an underlying medical or physiological condition (e.g., diabetes, malignancies). These tissue injuries heal slowly (more than 12 weeks) and often recur (e.g., ulcers) [81, 82]. Wounds are also classified according to the thickness of the skin layers affected [81]: superficial (only the epidermis is affected), partial thickness (both the epidermis and the dermis are affected) and full thickness wounds (underlying subcutaneous fat or deeper is affected, as well as dermis and epidermis). Skin defects may be treated using dressings or substitutes, depending on the severity of the wound and the surface area affected. The main reasons for closing or covering a skin defect are to (a) provide rapid and cosmetically acceptable healing, (b) prevent infection, (c) contain exudate, (d) minimize distress to the patient, (e) reduce pain, and (f) cover a wound for cosmetic reasons [81, 83].

Poor edge strength, degradation due to sunlight, difficult processing and coloration, easy stainability.

Low chemical resistance, low mechanical properties, high degradation rate, short working life. Need for plasticizers, linking agents and metal molds, low dimensional stability.

Soft and elastic, resembling natural Examples of acrylic copolymers are chains of hexanediol diacrylate (HDDA), replaced tissues. hexyl acrylate (HA), 2-ethyl hexyl acrylate (EHA) and dodecyl acrylate (DDA) monomers.

PVC chains are composed of CH2 -chcl– Flexible, adaptable to coloration for suitable appearance, high glass monomers, while plastisols are made of PVC chains and plasticizing additives transition temperature. such as phtalates, phosphates or fatty acids; PVCA copolymers are composed of polyvinyl chloride and polvinyl chloride acetate monomers in different amounts.

Acrylic copolymers [76]

Polyvinyl chloride (PVC), plastisols and polyvinyl chloride acetate (PVCA) co-polymers

Rigidity, discomfort, high thermal conductivity, no possible duplication.

Readily available, ease of coloration, adequate mechanical properties, compatibility with adhesives, long working life, color stability, easily repaired or reshaped.

Thermoplastic or thermosetting plastic materials composed of acrylic acid, methacrylic acid or similar monomers.

Acrylic resins [75]

Disadvantages

Advantages

Resins

Description

Material & devices

Applications

Table 12.11. Materials and devices used in maxillofacial implants.

12 Advanced materials for biomedical applications |

313

Material & devices

Description

Advantages

MDX 4-42103 [78] Silastic Medical Adhesive Silicone, Type A® (# 891)3 A-21864

Siloxanes containing vinyl polymers and hydrides; chloroplatinic acid is often used as a catalyst, different kinds of ethoxysilane are used as cross-linking agents.

Room temperature curing, no shrinking, no presence of byproducts and peroxides. Cures at room temperature when exposed to atmospheric moisture, solvent-free, one-component adhesive/ sealant, humidity sensitive. Translucent and pigmentable, short preparation and curing time.

Long curing time (up to 24 hours), larger parts with thick sections could require longer, catalyzer and degasing procedure needed. Releases acetic acid vapor as a byproduct. Time consuming, especially for low humidity levels; heat not accelerating curing time

Need for heat curing (at least 116°C) and catalyzers (platinum), two components to mix before activating with catalyzer. High temperature curing (127–175°C), need for pigmentation to perfect final appearance.

Disadvantages

1 Momentive Performance Materials Inc. product; 2 Technovent Ltd. product; 3 Dow Corning Corporation product; 4 Factor II Inc. product A more exhaustive list of silicone-based materials used nowadays for prosthodontics can be found in the work by Montgomery et al. [79]. Other siliconeelastomer producers can be found in [74]. Pigments generally used are metal oxides, silicone intrinsic pigment, oil and dry earth pigments [79].

Room Temperature Vulcanized silicones (RTV)

Q7-4720/4780 Polydimethyl vinyl siloxane copolymer – Enhanced tear resistance, possibilHigh ity of hardness control, no residues SE-4524U1 with approx. 0.5% vinyl side chains, Temperature because of thermal curing via addi2,4-dichlorobenzoylperoxide as an Vulcanized M51142 [77] tion chemistry. initiator and a silica filler obtained from silicones (HTV) – High tear resistance, uniform propmethyl silane combustion. erties with different curing agents. – Good edge tear strength, short work and curing time, tunable mechanical, rheological and adhesion properties by specific agent addition.

Applications

314 | Part III New trends in sustainable development and biomedical applications

Description – Non-absorbent dressings; made from a thin sheet of polyurethane coated with a layer of adhesive. – Different types of film differ in their moisture vapor permeability, method of application, extensibility, weight and thickness.

– Colloidal materials (gel forming agents) combined with other materials such as elastomers and adhesives; composed mainly of cellulose and typical gel forming agents including carboxymethyl cellulose (cmc), gelatin, and pectin. – Occur in the form of thin films and sheets or as composite dressings in combination with other materials such as alginates.

Material & Devices

Semi-permeable adhesive films

Hydrocolloid dressing

Application

No exudate to low exudate wounds – Partial-thickness wounds, donor sites, minor burns, pressure ulcers. – As a secondary dressing over primary dressings (gels, alginates and hydrofibers)

Low to moderate exudate wounds – Wide range of exudate from light to very heavy, granulating, sloughing or necrotic wounds – E.g., pressure sores, abrasions, minor burns, surgical wounds, traumatic injuries, pressure ulcers, leg ulcers

Table 12.12. Materials and devices for wound dressing applications [81, 83–86]. Disadvantages Too thin to be packed into deep or cavity wounds; care must be taken when removing film dressings (to ensure atraumatic removal); may promote periwound maceration due to occlusive nature.

Leave residue at the wound bed which may be mistaken for infection; cannot be used in the presence of infection; may encourage the growth of anaerobic bacteria; may roll over certain body areas (friction)

Advantages Can be transparent (no need to be removed to visualize the wound); conform to contours (elastic and flexible nature); do not require additional taping; permeable to moisture, vapor, and gases; impermeable to fluid and microorganisms; can stay in place for up to one week; aid autolytic debridement; prevent friction against the wound bed; keep wound bed dry; prevent bacterial contamination of the wound. Permeable to water and air after application; facilitate rehydration and debridement of the wound; adhere to moist and dry sites; insulate the wound bed; waterproof and impermeable to bacteria, urine or stool; painless removal; allow patients to shower and bath and ideally should be left in place for 3–5 days.

12 Advanced materials for biomedical applications |

315

Foam dressings (recommended for dry wounds.)

– Consist of porous polyurethane foam or polyurethane foam film, sometimes with adhesive borders. – Variety of foams made from different base materials and constructions which have similar but varying performance characteristics; available in a variety of shapes, sizes and thicknesses.

Alginate dressings (not – Produced from calcium salts of alginic acid (polysaccharecommended for dry ride), a component of brown wounds) seaweed. – Ideal for bleed– Occur either in the form of ing wounds; unfreeze-dried porous sheets suitable for dry (foams) or as flexible fibers wounds or those (indicated for packing cavity with dry, hard, wounds); may be woven or necrotic tissues. nonwoven. – Have the ability to form gels upon contact with wound exudates (high absorbency).

Medium to high exudate wounds – Wide range of wound types and cavities which are granulating, with small amounts of slough. – E.g., venous, diabetic and pressure ulcers, wounds with tunneling, wounds with heavy exudate.

Description

Material & Devices

Application

Disadvantages

Less apt to stick to delicate wound May require a secondary dressing; beds; non-occlusive; comfortable; may promote periwound maceration; cannot be used on allow for less frequent dressing changes (depending on the amount wounds with eschar or wounds which are not draining; some foams of exudate); maintain a moist may not be suitable for certain environment around the wound; provide thermal insulation; highly wounds (those that are infected); sheet dressings not suitable as absorbent, absorbency being packs for cavity wounds. controlled by foam properties.

In the form of readily biodegradableAlways require a secondary dressing (adhesive foam dressing fibers: when trapped in a wound and can be rinsed away with saline or a semipermeable film); may cause desiccation of the wound irrigation; do not cause pain on removal; may be used on infected bed, as well as drying exposed wounds; non-adherent; encourage tendon, capsule or bone. autolytic debridement.

Advantages

316 | Part III New trends in sustainable development and biomedical applications

Material & Devices

Dry or sloughy wounds Hydrogel dressings

Application – Insoluble, swellable hydrophilic materials made from synthetic polymers such as poly(methacrylates) and polyvinylpyrrolidine; contains large amounts of water (80% or more) and are combined with a range of other materials (hydrocolloid materials, alginates and starch-based polymers). – Can be applied either as an amorphous gel or as elastic, solid sheet or film.

Description

Disadvantages

Absorb exudate, thereby providing Amorphous hydrogels usually a moist environment; nonadherent; require a secondary covering such nonreactive with biological tissue; as gauze and need to be changed frequently; permeable to metabolites and nonirritant; cool the surface of the Hydrogels have low mechanical strength and are therefore difficult wound (reduction in pain, high to handle (this has been noted to patient acceptability); leave no affect patient compliance). residue; malleable; improve reepithelialization of wounds; control transmission of water vapor through the dressing; sheets do not need a secondary dressing; can be cut to fit around the wound due to flexible nature

Advantages

12 Advanced materials for biomedical applications |

317

318 | Part III New trends in sustainable development and biomedical applications

Minimizes scarring Provides scaffold for cell growth Forms barrier against infection Encourages natural blood clotting

Epidermis

Dermis

Strenghtens new tissue Provides protein for healing Absorb fluid from inflammation Blocks nerve ending to reduce pain

Hypodermis

Fig. 12.4. Anatomy and physiology of the skin and properties of an ideal wound dressing.

Medicated dressings (drug delivery) A new generation of medicated dressings incorporates chemicals which have a therapeutic effect on wound healing. The incorporated drugs can be cleansing or debriding agents (removal of necrotic tissues), antimicrobials (prevent or treat infection) or growth factors (support tissue regeneration) which play an active role in the wound healing process. Modern dressings used to deliver active agents to wounds include hydrocolloids, hydrogels, alginates, polyurethane foam/films, and silicone gels [81]. Controlled delivery dressings can provide an excellent means of delivering drugs to wound sites over long periods of time without the need for frequent dressing changes. Bio-adhesive, synthetic, semisynthetic, and naturally derived polymeric dressings are useful in the treatment of local infections where it may be beneficial to have increased local concentrations of antibiotics, whilst avoiding toxic reactions due to high doses. Moreover, they are readily biodegradable and therefore can be easily washed off the wound surface once they have had the desired effect [81].

12 Advanced materials for biomedical applications

|

319

Bioengineered skin substitutes Bioengineered skin substitutes (biosynthetic skin substitutes and cultured autologous engineered skin) are available in large quantities with negligible risk of infection or immunologic issues, to provide temporary or permanent coverage of wounds. Although one major disadvantage of many of these products remains cost and availability [84, 87, 88], there are advantages in specific cases: they avoid the creation of a donor site (pain and morbidity), reduce autologous skin graft, require fewer postoperative dressing changes and speed up healing [87]. The use of autologous cells in cultured skin products has the advantage of immunological safety [88]. These skin substitutes can be used for permanent wound closure, restoring physiological stability. However, the use of autologous cells requires extended culture time for cells extracted from the skin biopsy [89]. Autologous skin equivalents seem to be the best treatment for chronic wounds and deep and extensive burns. Secretion of growth factors results in faster healing of donor sites and better esthetic results when compared to conventional therapeutics (Biobrane-L) [90] (Please see Table 12.13, p. 320, 321.).

12.5 Future trends 12.5.1 Tissue engineering basic concepts The term “tissue engineering” was first introduced about 25 years ago, but tissue engineering strategies have been employed for skin substitutes since the 1970s. Tissue engineering aims to “restore tissue and organ function by employing biological and engineering strategies to solve clinical problems” [92]. Traditional approaches to tissue engineering are: – implant a combination of an engineered matrix with cells as a tissue replacement; – implant an engineered matrix as a guiding template for inducing tissue regeneration in vivo. 12.5.2 Scaffolds Scaffold design Most strategies in regenerative medicine have focused on 3D mechanical support enhancing biological tissue growth. This degradable support, also known as scaffold, must have tailored properties for creating a specific tissue or organ. Biocompatible and biodegradable synthetic or natural polymers are well-suited for these applications because they offer great versatility and choice of processing and material features for matching tissue requirements. Indeed, scaffold hydrophobicity, porosity or crystallinity can induce various cell behaviors through signaling pathways. For example, a pore size of 20 μm is ideal for fibroblast ingrowth, whereas bone cells need a

May not provide complete coverage (small and circular grafts); multiple grafting procedures may be required for definitive wound closure.

Improves healing of (relatively small) Sheets of cultured autologous chronic leg ulcers. keratinocytes transplanted with a supportive silicone membrane; discs (Ø 1 cm) placed within the wound margins; silicone backing removed at first dressing change.

Epidex

Mechanical fragility (absence of an integrated dermal component); at least 3 weeks required for graft cultivation; hyperkeratosis; contracture; scarring; cost; long-term safety unknown.

FDA approved; valuable for patients with very large burns (60% of the body surface, with poor availability and/or quality of donor sites); autologous tissue can be used as a permanent replacement; can be used with a dermal substitute (Integra), in full-thickness burns.

Sheets of autologous keratinocytes (approximately 50 cm2 , two to eight cell layers thick) attached to a supporting petrolatum gauze backing (removed approximately 1 week after grafting); can be stored for 24 hours at cool room temperature (13 to 23°C).

Epicel

Epidermal grafts

Disadvantages

Advantages

Description

Material & Devices

Application

Table 12.13. Materials and devices for skin substitute applications [84, 87–91].

320 | Part III New trends in sustainable development and biomedical applications

Risk of seroma, infection; training Immediate availability in large quantities; used in complex traumatic required; second procedure typically necessary; expensive. soft tissue reconstruction (over exposed tendons, joints, bone, and wounds from vascular and pressure ulcers); inhibits wound contraction and scar formation. Semi-permeable silicone membrane controls water vapor loss, provides a flexible adherent covering for the wound surface, increased tear strength of the substitute. Minimal clinical data Collagen incorporated into the silicone and nylon components provides a flexible and adherent surface for wound coverage; semipermeable external silicone membrane allows excretion of burn exudates and permeability for topical antimicrobials; controls water vapor loss from the wound; showed improvement in reepithelialization; reduced pain; decreased nursing costs.

Semi-biological bi-layered dressing composed of an inner permanent dermal analogue (porous matrix of cross-linked type I bovine collagen, chondroitin-6-sulfate, a sharkderived glycosaminoglycan), under an external temporary silicone membrane (epidermis-like function); Storage: room temperature; 2-year shelf life; silicone sheet removed as wound heals; thin autograft grafted onto the neodermis to complete wound coverage. Temporary biosynthetic dressing composed of knitted nylon mesh bonded to a thin silicone membrane (acts as an epidermal layer) and coated with porcine polypeptides; immobilized with suture, tape, or other dressing material, later removed after wound healing (7 to 14 days) or when autograft skin is available. Manufactured in different sizes (13 × 13 cm, to 38 × 50 cm); No special storage conditions (at least 3 years at room temperature)

Integra Indicated for excised deep partial- and full-thickness burn wounds.

Biobrane Placed on a fresh, clean wound bed, superficial and partial thickness burns or as coverage for donor sites.

Dermal grafts Severe burns and chronic wounds

Disadvantages

Advantages

Description

Material & Devices

Application

12 Advanced materials for biomedical applications |

321

322 | Part III New trends in sustainable development and biomedical applications pore size of 200–400 μm. Numerous properties of scaffolds can improve cell adhesion, migration, proliferation, orientation, and protein secretion. Physiological stresses are also crucial for organ regeneration; therefore the mechanical properties of scaffolds must be similar to those of the tissue of interest and able to resist physiological duty cycles [93].

Degradation aspects A suitable degradable scaffold must provide a biomimetic matrix allowing cells to proliferate and produce their ECM, while the scaffold gradually vanishes over an appropriate time period. Consequently, understanding and controlling the degradation mechanism of a scaffold is the key to the long-term success of resorbable implants. Indeed, implantable degradable materials and their degradation products must be biocompatible until the degradation process has been completed. The degradation rate is tailored according to the application, with several processing parameters such as composition, molecular weight, polymer chain conformation, polymer structure, and scaffold morphology. However, polymer materials are not able to deliver all signaling properties required for cells to migrate, adhere, proliferate or secrete ECM. Consequently, biological molecules like peptides or growth factors may be added to the scaffold to fulfill these specifications [93, 94].

Scaffolding materials Hydrogel materials are widely used in biomedical applications as space-filling agents and bioactive molecule vehicles, and also for tissue constructs. The tridimensional architecture of hydrogels is similar to ECM, easy to manufacture, and can be doped with molecules of interest to enhance cell proliferation and tissue growth. However, current trends are focusing on mechanical improvements of hydrogels [95, 96]. Nanofiber scaffold strategy provides a 3D framework which mimics the fibrous state and physical scale of the ECM components. Moreover, fibers exhibit exceptional elasticity features. Electrospinning is a common technique used to produce nanofiber scaffolds, but other methods such as blow spinning, air spinning, and force spinning exist. Major challenges in the nanofiber technology field are producing scaffolds with the desired porosity for cellularizing nanofiber scaffolds with optimum requirements and designing complex macrostructures [96, 97]. Another very common technique for generating porous scaffold is thermalinduced phase separation. First, a polymer is totally dissolved in a solvent. Next the polymer solution is cooled at gelation temperature in a mold, and finally the solvent is evaporated via freeze drying in a vacuum. This technique offers very complex and suitable scaffold morphologies but cannot be used for all polymers or scaled up to an industrial level [97, 98].

12 Advanced materials for biomedical applications

|

323

Nanofibers may also be generated by self-assembly of peptide amphiphiles. This approach aims to create supramolecular constructs from specific peptides containing an alkyl chain tail and an antigenic determinant head, which aggregate into fibrils with specific conditions. This spontaneous process is driven by three forces: a hydrophobic attractive interaction between alkyl tails which forms the core, hydrogen bonding which forms the shape, and repulsive electrostatic forces between peptide heads. However, this smart but complex method of forming nanofibers is only rarely applied [98]. Decellularized matrix is another scaffold strategy. This technique consists of removing cellular material from an organ via chemical treatments. The scaffold retains the 3D structure and most of the ECM components of the organ. Nevertheless, detergents have been found to denature or eliminate some important proteins like collagen. On the other hand, most common methods do not achieve decellularization, and DNA or antigen materials remain in the scaffold, which can lead to a host immune response [99].

12.5.3 Surface modification Biological responses to biomaterials are greatly controlled by surface physics and chemistry. The key concepts of the surface modification strategy are to improve biological acceptance and functionalize a medical device or scaffold by changing its surface properties while conserving its bulk properties. Material surfaces can be modified by altering the first atomic layers (chemical modification, etching, roughening), or by adding other components (coating, grafting, deposition). Several surface modifications are aimed at modifying blood compatibility to influence cell adhesion and growth, to control protein adsorption, to produce a nonfouling surface, to improve lubricity, corrosion resistance, antibacterial features, and electrical properties. The main challenges in surface engineering lie in the stability of the treatment, control of molecule delivery, and delamination of the surface [100].

12.5.4 Stem cells Terminally differentiated cells are limited in their proliferative capability and are therefore less likely to completely regenerate an organ. Because of this limitation, the use of stem or progenitor cells has become a major driver in the field of regenerative medicine. Stem cells are able to self-renew, to extensively proliferate, and to differentiate into multiple cell types, making them a promising cell source for regenerating tissues. Stem cells are divided into two categories: embryonic and somatic. Embryonic stem cells (ESC) are pluripotent and proliferate very quickly, but have great limitations. First, ethical issues must be taken into account because embryos

324 | Part III New trends in sustainable development and biomedical applications are destroyed to obtain ESC. Second, ESCs are allogeneic and lead to an immune response or differentiate into malignant tissues. On the other hand, multipotent adult stem cells are able to differentiate into several tissues and mainly come from umbilical cord blood and bone marrow. The advantage when comparing these cells to ESC is that adult stem cells can be obtained from patients themselves, thus limiting immune rejections. However, adult stem cells have a much more limited proliferation potential. Recent research has found the ability to reprogram somatic stem cells to obtain characteristics similar to embryonic stem cells. Thus, finding a better approach to differentiate stem cells into various cell-lineages is definitely the major future trend in this field [98].

12.5.5 Bioreactors A bioreactor is a dynamic in vitro environment which uses both biochemical and mechanical signals to guide and regulate tissue development. The concept of bioreactors is not restricted to tissue engineering, since producing proteins from the culture of microbial or mammalian cells for therapeutic or diagnostic applications is a very wellknown technique [101]. Bioreactors have been developed in response to static culture limitations and are used to distribute cells uniformly in 3D scaffolds. They provide the desired concentrations of gases and nutrients in the culture medium, maintain efficient mass transfer to growing tissue, and apply a physical stimulus to improve tissue development. Several bioreactor designs such as rotating-wall, spinner flasks, and perfusion have been developed to engineer a variety of tissues (Fig. 12.5). In order to properly simulate physiological conditions, tissue engineering requires a system which mimics the hemodynamic forces experienced by tissues. These mechanical forces include shear stress and stresses in the radial, circumferential, and longitudinal directions. Initial attempts to create engineered tissues resulted in the manufacturing of grafts with poor mechanical properties. The new generation of tissue engineered substitutes produced using bioreactors has dramatically improved the mechanical properties of the tissues; however, it takes several months to develop engineered tissues with the desired mechanical properties. Because elastin and collagen are the two main components of ECM which influence the biomechanical properties of tissues, dynamic mechanical conditioning of the construct accelerates the production of these two proteins [102].

12.5.6 Computational models Over the past two decades, computational modeling has emerged as a significant research activity in biomaterials, in particular in the area of polymeric science. Its extended importance is attributable to three factors. First, the development of high

12 Advanced materials for biomedical applications

|

325

Tissue Engineering Bioreactors

In vivo bioreactors

In vitro bioreactors

Static systems Petri dish Multiwell plate T-flask Membrane flask bioreactor Multitray cell culture system Culture bag

Dynamic systems Mechanically driven Hydraulically driven Shaker Shake flask Multiwell plate (shaken) Rotating unit Rotating wall Spinner flask Membrane flask bioreactor

The peritoneal cavity bioreactor The bone bioreactor

Hollow fibre bioreactor Perfusion bioreactor Pulsatile flow bioreactor Compression applied bioreactor

Rocker unit or raising platform Tensile bioreactor Fig. 12.5. Summary of different bioreactors used for tissue fabrication (adapted from [103]).

throughput combinatorial polymer synthesis techniques has enabled the creation of polymer libraries of extraordinary size. For example, the tyrosine-derived library of polyarylates consists of eight diacids and fourteen diphenols for a total of 112 polymers developed by the Kohn lab. In comparison, a large virtual library of polymethacrylates recently designed in the same lab resulted in more than 40 000 polymers. The conventional experimental approach to characterizing bioresponse and materials properties of polymers is clearly unsuitable for such enormous libraries. Second, modeling techniques developed in computational chemistry (e.g., drug discovery) and computer science are readily adaptable to biomaterials science. This technology transfer leverages longstanding research programs in these related fields to provide new modeling capabilities in biomaterials science. Third, computational resources have continued to expand at a dramatic rate. Cluster computing using Linux-based servers is now both commonplace and inexpensive, and has enabled application of detailed physical modeling (e.g., molecular dynamics (MD) simulation) to specific problems in biomaterials science [104]. Recently, computer simulation of tissue differentiation in response to mechanical forces has become an important element in modeling studies. It involves defining algorithms for mechanoregulation of each of the following cell activities: proliferation, apoptosis, migration, and differentiation using a stimulus based on a combination of strain and fluid flow algorithms, which are based on lattice-modeling which also facil-

326 | Part III New trends in sustainable development and biomedical applications itates building algorithms for complex processes such as angiogenesis. The algorithms are designed individually but can be processed together. They can be combined to create a computational simulation method for tissue differentiation, using finite element analysis to compute the mechanical stimuli even in quite complex biomechanical environments [105]. The seeding of a porous scaffold with cells is a fundamental step in engineering sizable tissue constructs that are clinically viable. However, a key problem often encountered is inhomogeneous seeding of the cells, particularly when the cells are delivered through the thickness of the scaffold. A quantitative relationship between cell seeding efficiency and the initial vacuum pressure in a compact perfusion seeding device which uses the effect of differential pressure induced by vacuum to seed cells on a porous scaffold was established with the help of computational models. A transient CFD solution of the fluid flow in the device was used to formulate a 3D computational model which can be employed to design and optimize cell seeding techniques and the corresponding technology [106].

12.6 Summary This chapter briefly summarized the synthetic materials and devices available to surgeons for the treatment, repair or replacement of tissues or organs. It should be understood that these materials and devices are the result of decades of research and development during which strategies evolved with the improvement of our understanding of physiology and disease, as well as our materials science and engineering capabilities. Therefore, knowledge of the biomaterials field requires a historical overview. The very first biomedical devices developed were essentially designed to reproduce the basic functions of the tissue or organ they were intended to replace. For example, the first arterial prostheses were made from materials which could form tubes with adequate mechanical properties and chemical stability to withstand the pulsatile blood pressure for many years without failure. During this early period very few concerns were raised regarding the interaction between the material and the physiological environment. This last issue began to be of interest when researchers realized that the body’s response towards a synthetic implant plays a very important role in its long-term patency. This awareness led to the development of stealth materials which were intended to be hidden from the body and to have very minimal interaction with the biological environment. This was the beginning of the development of non-fouling coatings made with polyethylene glycol, various carbohydrates, phospholipids, etc. Despite the fact that this approach has been demonstrated to be successful in some instances (for example, such coatings are used to decrease protein adsorption on contact lenses), it turned out to be almost impossible to make an antifouling layer free of defects, which in turn, were the sites of nucleation for protein adsorption.

12 Advanced materials for biomedical applications

|

327

Today, the common strategy is to develop materials likely to proactively interact with the physiological environment. The number of strategies are almost limitless and they vary from the control of material surface roughness for appropriate size matching with cell receptors, the conjugation of signal peptides for cell recruitment, cell/drug encapsulation, to biodegradable materials that will be replaced during the tissue/organ healing/reconstruction process. These approaches often require the combined properties of synthetic and biological materials to be taken advantage of. Where do we go from here? If the trend continues, it is likely that the biomaterials of the future will be increasingly biological. Indeed, scientists now have a better understanding of how cells interact to form tissues and organs and accordingly, how to enable cells to form suitable tissue substitutes. Indeed, such expertise is already in use to make skin substitutes which are highly beneficial for severely burnt patients. Very promising experiments in the generation of bone, blood vessels, or even a compete heart from cells are ongoing. In addition, researchers are now able to benefit from cell functions and properties at different levels of differentiation states. Having said that, it should be kept in mind that it took nature several thousands of years to create the human body. In this context, it is unlikely that human replacement parts with the same level of complexity as the original ones will be available in the near future. Therefore, synthetic biomaterials will continue to play a key role in the treatment of patients for many years to come.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

B.D. Ratner, A.S. Hoffman, F.J. Schoen, J.E. Lemons, Biomaterials Science – An Introduction to Materials in Medicine (3rd Edition), Elsevier, 2013. A.R. Kumara, K. Senthila, S. La, P. Kumara, P. Chirakkala, Sri Ramachandra Journal of Medicine, (2010) 1–38. N.A. Malik, Textbook of oral and maxillofacial surgery, Jaypee Brothers Medical Pub, 2008. S.V. Madihally, Principles of biomedical engineering, Artech House Publishers, 2010. V.C. Mow, R. Huiskes, Basic orthopaedic biomechanics and mechano-biology, Lippincott Williams & Wilkins, 2004. M. Navarro, A. Michiardi, O. Castaño, J.A. Planell, Journal of The Royal Society Interface, 5 (2008) 1137–1158. A.S.-L.-H. Association. A Brief History of Cochlear Implants. Available from: http://www.asha.org/policy/TR2004-00041/ [2014-05-27] G. Steinhoff, Regenerative Medicine, Springer, 2011. M. Vallet-Regí, F. Balas, M. Colilla, M. Manzano, Solid State Sciences, 9 (2007) 768–776. E. Clozza, M. Pea, F. Cavalli, L. Moimas, R. Di Lenarda, M. Biasotto, Clinical Implant Dentistry and Related Research, 16(1) (2012). S. Best, A. Porter, E. Thian, J. Huang, Journal of the European Ceramic Society, 28 (2008) 1319–1327. J.E. Mark, Polymer Data Handbook. Oxford University Press, 1999.

328 | Part III New trends in sustainable development and biomedical applications [13]

[14] [15] [16] [17] [18] [19] [20] [21] [22]

[23] [24]

[25]

[26] [27] [28] [29]

[30] [31] [32] [33] [34] [35] [36] [37] [38]

M.C.L. Martins, Appendix B - Properties of Soft Materials, in: D.R. Buddy, S.H. Allan, J.S. Frederick, A.S.H.F.J.S. Jack E. LemonsA2 - Buddy D. Ratner, E.L. Jack (Eds.) Biomaterials Science (Third Edition), pp. 1483–1485. Academic Press, 2013. S.F. Badylak, T.W. Gilbert, Seminars in Immunology, 20 (2008) 109–116. WHO, Cardiovascular Diseases 2013 Available from: http://www.who.int/mediacentre/ factsheets/fs317/en/ [2014-05-27]. P. Libby, M. Aikawa, M.K. Jain, Handbook of experimental pharmacology, (2006) 285–306. P.A. Laizzo, R.W. Bianco, A.J. Hill, J.D. St. Louis, Heart Valves, Springer, Boston, MA, 2013. C. Milano, A. Simeone, Heart Failure Reviews, 18 (2013) 35–53. M.R. Rosen, P.R. Brink, I.S. Cohen, R.B. Robinson, Circulation: Arrhythmia and Electrophysiology, 1 (2008) 54–61. J.G. Copeland, R.G. Smith, F.A. Arabia, P.E. Nolan, G.K. Sethi, P.H. Tsau, D. McClellan, M.J. Slepian, New England Journal of Medicine, 351 (2004) 859–867. M.J. Slepian, Y. Alemu, J.S. Soares, R. G. Smith, S. Einav, D. Bluestein, Journal of Biomechanics, 46 (2013) 266–275. F.J. Schoen, R.F. Padera Jr, B - Endovascular Stents, Vascular Grafts, and Stent Grafts, in: D.R. Buddy, S.H. Allan, J.S. Frederick, A.S.H.F.J.S. Jack E. LemonsA2 - Buddy D. Ratner, E.L. Jack (Eds.) Biomaterials Science (Third Edition) pp. 771–784, Academic Press, 2013. G. Mani, M.D. Feldman, D. Patel, C.M. Agrawal, Biomaterials, 28 (2007) 1689-1710. S. Nishio, K. Kosuga, K. Igaki, M. Okada, E. Kyo, T. Tsuji, E. Takeuchi, Y. Inuzuka, S. Takeda, T. Hata, Y. Takeuchi, Y. Kawada, T. Harita, J. Seki, S. Akamatsu, S. Hasegawa, N. Bruining, S. Brugaletta, S. de Winter, T. Muramatsu, Y. Onuma, P.W. Serruys, S. Ikeguchi, Circulation, 125 (2012) 2343–2353. M. Wu, L. Kleiner, F.-W. Tang, S. Hossainy, M.C. Davies, C.J. Roberts, European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences, 36 (2009) 493–501. R. Fattori, T. Piva, The Lancet, 361 (2003) 247–249. R.Y. Kannan, H.J. Salacinski, P.E. Butler, G. Hamilton, A.M. Seifalian, Journal of biomedical materials research. Part B, Applied biomaterials, 74 (2005) 570–581. H. Oxenham, P. Bloomfield, D.J. Wheatley, R.J. Lee, J. Cunningham, R.J. Prescott, H.C. Miller, Heart, 89 (2003) 715–721. R.B. Salter, Textbook of disorders and injuries of the musculoskeletal system: an introduction to orthopaedics, fractures and joint injuries, rheumatology, metabolic bone disease, and rehabilitation, Williams & Wilkins, 1999. M.M. Ibrahim, C. Thulasingam, K. Nasser, V. Balaji, M. Rajakumar, P. Rupkumar, The Journal of Indian Prosthodontic Society, 11 (2011) 165–171. A.W. Lindberg, B. Samimi, D.C. Allison, L.R. Menendez, R. Mirzayan, (2011). D.E. Shepherd, Biomechanics of Hard Tissues, (2011) 209–216. E.B. Hunziker, Osteoarthritis and Cartilage, 10 (2002) 432–463. S. Marlovits, P. Zeller, P. Singer, C. Resinger, V. Vécsei, European Journal of Radiology, 57 (2006) 24–31. J. Chen, J. Xu, A. Wang, M. Zheng, Expert Review of Medical Devices, 6 (2009) 61–73. S.J. Kew, J.H. Gwynne, D. Enea, M. Abu-Rub, A. Pandit, D. Zeugolis, R.A. Brooks, N. Rushton, S.M. Best, R.E. Cameron, Acta Biomaterialia, 7 (2011) 3237–3247. G. Vunjak-Novakovic, G. Altman, R. Horan, D.L. Kaplan, Ann Rev Biomed Eng, 6 (2004) 131–156. A.J. Krych, J.D. Jackson, T.L. Hoskin, D.L. Dahm, Arthroscopy: the journal of arthroscopic & related surgery: official publication of the Arthroscopy Association of North America and the International Arthroscopy Association, 24 (2008) 292.

12 Advanced materials for biomedical applications

[39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72]

|

329

M.F. Guidoin, Y. Marois, J. Bejui, N. Poddevin, M.W. King, R. Guidoin, Biomaterials, 21 (2000) 2461–2474. K.A. Derwin, A.R. Baker, R.K. Spragg, D.R. Leigh, J.P. Iannotti, The Journal of Bone & Joint Surgery, 88 (2006) 2665–2672. E. Budtz-Jörgensen, Journal of Dentistry, 24 (1996) 237–244. S. Suzuki, Y. Ikada, Devices for Bone Fixation, Biomaterials for Surgical Operation, pp. 131–144. Springer, 2012. J. Pickup, M. Mattock, S. Kerry, BMJ, 324 (2002) 705. P.R. van Dijk, S.J. Logtenberg, K.H. Groenier, J.W. Haveman, N. Kleefstra, H.J. Bilo, World J Diabetes, 3 (2012) 142–148. Z.Y. Oo, R. Deng, M. Hu, M. Ni, K. Kandasamy, M.S. bin Ibrahim, J.Y. Ying, D. Zink, Biomaterials, 32 (2011) 8806–8815. F. Tasnim, R. Deng, M. Hu, S. Liour, Y. Li, M. Ni, J.Y. Ying, D. Zink, Fibrogenesis Tissue Repair, 3 (2010) 14. J.M. Bargman, Semin Dial, 25 (2012) 545–549. G. Orive, E. Anitua, J.L. Pedraz, D.F. Emerich, Nature Reviews Neuroscience, 10 (2009) 682–U647. Y.H. Zhong, R.V. Bellamkonda, Journal of the Royal Society Interface, 5 (2008) 957–975. X.M. Liang, A.F. Wang, T. Cao, H.Y. Tang, J.P. McAllister, S.O. Salley, K.Y.S. Ng, J Biomed Mater Res A, 76A (2006) 580–588. S.R. Browd, B.T. Ragel, O.N. Gottfried, J.R.W. Kestle, Pediatr Neurol, 34 (2006) 83–92. R.A. Andersen, E.J. Hwang, G.H. Mulliken, Annu Rev Psychol, 61 (2010) 169–190. T.L. Hanson, B. Omarsson, J.E. O’Doherty, I.D. Peikon, M.A. Lebedev, M.A.L. Nicolelis, Ieee T Neur Sys Reh, 20 (2012) 331–340. M.A.L. Nicolelis, A.A. Ghazanfar, C.R. Stambaugh, L.M.O. Oliveira, M. Laubach, J.K. Chapin, R.J. Nelson, J.H. Kaas, Nat Neurosci, 1 (1998) 621–630. D.H. Kim, D.C. Martin, Biomaterials, 27 (2006) 3031–3037. K.A. Ludwig, J.D. Uram, J.Y. Yang, D.C. Martin, D.R. Kipke, J Neural Eng, 3 (2006) 59–70. E. Garcia-Garcia, M.J. Pino-Barrio, L. Lopez-Medina, A. Martinez-Serrano, Mol Biol Cell, 23 (2012) 1167–1180. I.P. Kaur, R. Bhandari, S. Bhandari, V. Kakkar, J Control Release, 127 (2008) 97–109. E. Garcia-Garcia, K. Andrieux, S. Gil, P. Couvreur, Int J Pharm, 298 (2005) 274–292. A. Jain, Y.T. Kim, R.J. McKeon, R.V. Bellamkonda, Biomaterials, 27 (2006) 497–504. M. Yemisci, I. Vural, S. Bozdag, M. Cetin, F. Soylemezoglu, Y. Capan, T. Dalkara, Neurosurgery, 59 (2006) 1296–1302. D. Gupta, C.H. Tator, M.S. Shoichet, Biomaterials, 27 (2006) 2370–2379. R.D. Johnson, H.K. Chadha, V.P. Dugan, D.S. Gupta, S.L. Ferrero, C.H. Hubscher, J Neurotraum, 28 (2011) 595–605. S.E. Stabenfeldt, A.J. Garcia, M.C. LaPlaca, J Biomed Mater Res A, 77A (2006) 718–725. J. Tan, R.A. Gemeinhart, M. Ma, W.M. Saltzman, Biomaterials, 26 (2005) 3663–3671. Z.Z. Khaing, C.E. Schmidt, Neuroscience Letters, 519 (2012) 103–114. A.W. Lloyd, R.G. Faragher, S.P. Denyer, Biomaterials, 22 (2001) 769–785. D. Kook, A. Kampik, A.K. Dexl, N. Zimmermann, A. Glasser, M. Baumeister, T. Kohnen, F1000 medicine reports, 5 (2013) 3–3. F.A. Spelman, Chapter II.5.11 Cochlear prostheses, pp. 967–980, Elsevier, 1998. M.A. Shiffman, Breast Augmentation - Principles and Practice, Springer Berlin Heidelberg, 2009. S.L.S.a.M.R. Jespersen, Aesthetic Surgery Journal, 30 (2010) 14. A.P. Armstrong, B.M. Jones, British Journal of Plastic Surgery, 53 (2000) 479–483.

330 | Part III New trends in sustainable development and biomedical applications [73] [74] [75] [76] [77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92]

[93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104]

J. Beumer, Maxillofacial rehabilitation: prosthodontics and surgical considerations., Medico Dental Media International Inc, 1996. G.H. Jonathan Black, Handbook of Biomaterial Properties, 1st ed., Springer, 1998. J.H. Jorge, E.T. Giampaolo, A.L. Machado, C.E. Vergani, The Journal of Prosthetic Dentistry, 90 (2003) 190–193. H. Lu, J.W. Stansbury, J. Nie, K.A. Berchtold, C.N. Bowman, Biomaterials, 26 (2005) 1329–1336. M.M. Hatamleh, D.C. Watts, Dental Materials, 26 (2010) 185–191. W.T. Bell, V.A. Chalian, B.K. Moore, The Journal of Prosthetic Dentistry, 54 (1985) 404–410. P.C. Montgomery, S. Kiat-Amnuay, Journal of Prosthodontics, 19 (2010) 482–490. A. Shai, Maibach, H.I., Baran, R., Handbook of cosmetic skin care (second edition), CRC Press; 2nd edition , 2009. J.S. Boateng, K.H. Matthews, H.N.E. Stevens, G.M. Eccleston, J Pharm Sci-Us, 97 (2008) 2892–2923. P. Zahedi, I. Rezaeian, S.O. Ranaei-Siadat, S.H. Jafari, P. Supaphol, Polym Adv Technol, 21 (2010) 77–95. C. Morris, Wound Essentials, 1, 178–183, 2006. P.S. Murphy, Ewans, G.R.D., Plastic Surgery International, 2012 (2012). A. Saarai, T. Sedlacek, V. Kasparkova, T. Kitano, P. Saha, J Appl Polym Sci, 126 (2012) E79–E88. S. Thomas, Int Wound J, 5 (2008) 602–613. P.L. Chern, C.L. Baum, C.J. Arpey, Dermatol Surg, 35 (2009) 891–906. C. Pham, J. Greenwood, H. Cleland, P. Woodruff, G. Maddern, Burns, 33 (2007) 946–957. D.M. Supp, S.T. Boyce, Clin Dermatol, 23 (2005) 403–412. C. Auxenfans, J. Fradette, C. Lequeux, L. Germain, B. Kinikoglu, N. Bechetoille, F. Braye, F.A. Auger, O. Damour, Eur J Dermatol, 19 (2009) 107-113. E. Lineen, N. Namias, J Craniofac Surg, 19 (2008) 923–928. D. Sarkar, W. Zhao, S. Schaefer, J.A. Ankrum, G.S.L. Teo, M.N. Pereira, L. Ferreira, J.M. Karp, Chapter II.6.2 Overview of tissue engineering concepts and goals of tissue engineering, pp. 1122–1137. Elsevier, 2013. M. Singh, F.K. Kasper, A.G. Mikos, Chapter II.6.3 – Tissue Engineering Scaffolds, pp. 1138–1159. Elsevier, 2013. C.-c. Lin, K.S. Anseth, Chapter II.4.3 The biodegradation of biodegradable polymeric biomaterials, pp. 716–728. Elsevier, 1996. J.L. Drury, D.J. Mooney, Biomaterials, 24 (2003) 4337–4351. H. Shin, S. Jo, A.G. Mikos, Biomaterials, 24 (2003) 4353–4364. S. Ramakrishna, K. Fujihara, W.-E.T. Ma, T.-C. Lim, Z. Ma, An introduction to electrospinning and nanofibers, 2005. A.C. Bean, R.S. Tuan, Stem cells and nanotechnology in tissue engineering and regenerative medicine, in: M. Ramalingam, E. Jabbari, S. Ramakrishna, A. Khademhosseini (Eds.), pp. 1-26 J.J. Song, H.C. Ott, Trends Mol Med, 17 (2011 [2013]) 424–432. B.D. Ratner, A.S. Hoffman, Chapter I.2.12 Physicochemical surface modification of material used in medicine, pp. 259-276. Elsevier, 2013. S. Nilsang, V. Nehru, F.M. Plieva, K.S. Nandakumar, S.K. Rakshit, R. Holmdahl, B. Mattiasson, A. Kumar, Biotechnology Progress, 24 (2008) 1122–1131. K. Bilodeau, D. Mantovani, Tissue Engineering, 12 (2006) 2367–2383. D.G. Seifu, Chemical and Biochemical Engineering, The University of Western Ontario, 2012. A.D. Costache, J. Ghosh, D.D. Knight, J. Kohn, Advanced Engineering Materials, 12 (2010) B3–B17.

12 Advanced materials for biomedical applications

|

[105] H.B. Henninger, S.P. Reese, A.E. Anderson, J.A. Weiss, Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine, 224 (2010) 801–812. [106] A.A. Adebiyi, M.E. Taslim, K.D. Crawford, Biomaterials, 32 (2011) 8753–8770. [107] Y. Chen, L. Liu, Advanced drug delivery reviews, 64 (2012) 640–665.

331

M.-A. Fortin

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine Magnetic resonance imaging (MRI) has developed at an exponential rate over the last decades, and is now widely used as an anatomical and functional medical imaging modality. The development of magnetic nanoparticles as contrast agents for vascular, molecular, and cellular MRI applications has followed this trend. One of the most important direct applications of nanotechnology is nanoparticles which generate either “positive” or “negative” contrast in MRI. The present chapter is an introduction to the principles underlying the performance of nanoparticulate-based MRI contrast agents. It addresses the main considerations guiding the design, synthesis and the physicochemical characterization of magnetic nanoparticles based on the elements iron, manganese; and gadolinium (Fe, Mn, Gd). The fundamental aspects of nanoparticle magnetism and relaxometric characterization are introduced, as well as examples of applications in biological models.

Introduction to nanoparticles for MRI applications Nanoparticles (NPs) made of the elements iron (Fe), gadolinium (Gd) or manganese (Mn) are currently used in many diagnostic applications performed under magnetic resonance imaging (MRI). In fact, MRI scanners do not directly detect intrinsic signals from magnetic nanoparticles (MNPs); instead, they reconstruct images from the electromagnetic signals generated by the stimulation and relaxation of the large pool of hydrogen (1 H) protons found in biological tissues. MNPs influence the relaxation time of hydrogen protons contained in small, mobile molecules such as water by interactions taking place at the atomic and molecular level. Thereby, MNPs induce contrast enhancement effects on the reconstructed MR images, either as a signal increase, or brightening (“positive” contrast agents, CAs), or a signal decrease, or darkening (“negative” CAs). Modern medical diagnostics now rely on hybrid imaging modalities which are capable of merging excellent anatomical resolution (MRI, CT) with the sensitive detection of molecular and cellular events (e.g., nuclear medicine and luminescence/fluorescence techniques). In fact, MRI is one of the most reliable, high resolution (75–300μm), and multi-functional imaging modalities of modern medicine. In particular, it is the imaging modality of choice for soft tissues (brain, liver, spine). Compared to optical imaging and echography, it allows the acquisition of in-depth, whole-body images, from the mouse model to the human. Contrary to computer tomography (CT), another anatomical imaging modality, MRI does not rely on ionizing

334 | Part III New trends in sustainable development and biomedical applications radiation. It also enables the tracking of cells, molecules, and drug delivery vehicles in the human body. These applications require the development of appropriate biomedical imaging probes (CAs or tracers). Such probes must allow the efficient detection of a reasonable amount of molecules, or cells, in the body (nanomolar or micromolar concentration range) [1–3]. Compared to MRI, positron emission tomography (PET) and luminescence/fluorescence imaging do not provide anatomical images, and neither are they considered high resolution imaging modalities in general. However, both are truly efficient “molecular” imaging modalities (nanomolar detection). Unfortunately, MRI has relatively poor sensitivity compared to such techniques. In order to fully exploit the wide range of advantages MRI has to offer in molecular and cellular imaging applications, it has been necessary to design CAs capable of very efficient interaction with hydrogen protons. Contrast agents (CAs) have been developed since the inception of MRI, to selectively change the longitudinal and transverse relaxation times (T1 and T2 ) of 1 H in biological tissues. Such energy transfers mainly occur through interactions between the magnetic elements and the spins of hydrogen protons in their vicinity. Gadolinium has 7 unpaired electrons in its 4f orbitals, giving it a very large magnetic moment. This translates into a relatively slow electronic relaxation rate compared to other paramagnetic elements, which enhances its proton relaxation properties [4]. Manganese can also be exploited in MRI applications, although its magnetic moment is weaker than that of Gd3+ (Mn2+ has 5 unpaired electrons on its 3d orbital). Most clinically approved CAs are based on small molecules which sequestrate the paramagnetic ion Gd3+ [5, 6]. They are mainly used as nonspecific agents to enhance the general contrast of organs, thereby enabling better identification of anatomical changes occurring in the body. They are also applied to blood-pool and blood perfusion procedures. After the first marketing authorization of Gd-DTPA (Magnevist™ ) in the US, Europe and Japan in 1988, other Gd-based chelates were introduced to the market: Dotarem™ , based on the DOTA chelator, as well as Omniscan™ , Prohance™ , Optimark™ , and Gadovist™ [7]. Today, contrast media are used in approximately 30 to 40% of all MRI procedures. Parallel to the development of paramagnetic CAs, progress also occurred in the field of MRI CAs through the unique properties of iron oxide NPs. In contrast to paramagnetic molecules, which produce “positive” contrast in MRI, iron oxide NPs are well known for their signal decrease. Iron oxide NPs are generally divided into two classes: small particles of iron oxide (SPIO) and ultra-small particles of iron oxide (USPIO). SPIOs consist of iron oxide cores with mean diameters of between 3 and 20 nm; however they form agglomerates of hydrodynamic diameter typically greater than 50 nm. The concept of hydrodynamic diameter (Fig. 13.1) refers to the total effective diameter of a particle suspended in a fluid and forming a colloid. The hydrodynamic diameter is generally measured by means of laser analysis (dynamic light scattering, DLS), which is an indirect measurement of the Brownian motion of NPs in aqueous media. USPIOs, on the other hand, refer to the class of iron oxide NPs which is made of iron

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 335

Water in the hydrodynamic corona Ligand or organic coating molecule

Fe2O3, MnO, Gd2O3, NaGdF4

Link (e.g. covalent) between the particles and the coating molecules

+ – -

+ – + –

+ – + – +

+

Functional anchor (attachment of targeted molecules, or imaging functionalities)

Fe2O3, MnO, Gd2O3, NaGdF4

– – – + – +

–

+ –+ – +

rH : hydrodynamic diameter

(a)

– + – – – +

+

Organic molecules adsorbed in the corona

(b)

Fig. 13.1. Schematic representations of (a) the structure of functional magnetic nanoparticles (MNPs) for MRI applications, and (b) a colloidal nanoparticle suspended in biological media.

oxide cores (3–20 nm diameter) surrounded by a coating of molecules that preserve their individuality (mean diameter < 50 nm). SPIOs and USPIOs have very different mean hydrodynamic diameters, as well as different “relaxometric” properties. Such properties can be advantageously exploited either in cell labeling (preferentially with SPIOs), liver cancer diagnostic (preferentially with SPIOs), molecular targeted imaging (preferentially with USPIOs), or blood pool/angiography procedures (preferentially with USPIOs). In fact, a very large fraction of molecular and cellular MRI applications are based on the design and production of MNPs consisting of (a) an inorganic nanocrystal core (e.g. Fe2 O3 /Fe3 O4 , Gd2 O3 , MnO, NaGdF4 ), (b) a coating made of small ligands, or biocompatible organic molecules, and (c) a functional anchor used to graft specific molecules with complementary imaging functionalities (e.g., radioactive atoms or fluorescent molecules), or medicinal compounds for drug delivery. Injected intravenously, SPIOs are captured by macrophages. They end up in the Kupffer cells of the liver (part of the reticuloendothelial system, RES), where they are expected to degrade. As a result, an important shortening of the transverse relaxation time (T2 /T2∗ ) is observed in liver tissue, which is reflected by a strong signal loss[8–10]. An example of this is provided in Fig. 13.2. By this mechanism, iron oxide NPs “passively”, but rather homogenously accumulate in the liver, where they are found to lower the MR signal. However, tumors and

336 | Part III New trends in sustainable development and biomedical applications

t = 0 min

t = 8 min

t = 24 h

(b)

0 20 40 60 80 100 120 140 Time (min)

(c)

1,20 1,10 1,00 0,90 0,80 0,70 0,60 0,50 0,40

S1/S0

1,60 1,50 1,40 1,30 1,20 1,10 1,00 0,90 0,80

S1/S0

S1/S0

(a)

0

8

16

24 32 40 48 Time (h)

(d)

1,60 1,50 1,40 1,30 1,20 1,10 1,00 0,90 0,80

0

8

16

24 32 40 48 Time (h)

Fig. 13.2. Intravenous injections of polymer-coated iron oxide NPs in the mouse model (unpublished data, Fortin et al.). After injection (t = 24 h compared to t = 0), arrows in (a)), the liver appears darker. The presence of USPIOs in the blood enhances the vascular signal during at least 20 minutes (t = 8 in (a), and (b) blood signal-enhancement ratio): USPIOs can be used as bloodpool agents. After injection, NPs are gradually sequestrated by the macrophages and follow the reticuloendothelial (RES) route, mainly in the liver ((a) t = 24 h, arrow; and (c)). After several hours, a significant signal enhancement appears in the gall bladder (d), an indication of the hepatobiliary excretion of NPs.

metastases are void of Kupffer cells and they do not internalize iron oxide NPs. This strategy has been used in the diagnosis of liver cancer and liver metastasis, as well as for imaging the spleen and the lymph nodes [11, 12]. Because of their elimination by the liver, SPIOs are not efficient in applications requiring longer blood half-lives (MR angiography, tissue perfusion imaging, functional imaging of the brain). USPIOs are better candidates for vascular applications, since they remain in the blood for longer. Their relaxation properties also allow them to be used as positive CAs in T1 -weighted imaging (see next section) and angiography [13–17]. In recent years, research into targeted MRI contrast agents has focused on the development of targeted USPIOs which could enable the efficient, sensitive and selective detection of atherosclerosis, apopto-

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 337

sis, and amyloid deposition in Alzheimer’s disease [18–25]. This concept is frequently referred to as molecular targeted imaging. Parallel to iron oxide NPs, a new generation of paramagnetic NPs has been developed, based on Gd- and Mn-containing inorganic nanocrystals [26, 27]. These are based on the synthesis of Gd2 O3 [28–32], NaGdF4 [33–36], and MnO [37–39] nanocrystals. When adequately covered with biocompatible ligands and suspended in aqueous solutions, these are “positive” CAs in MRI. Since the core of inorganic nanoparticles is very small (< 20 nm), and because they contain hundreds or thousands of paramagnetic atoms, they can be used as molecular or cellular probes. Thereby labeled molecules are then much more sensitively detected than if labeled by traditional MRI contrast agents which contain only one paramagnetic atom per unit of contrast agent (e.g., commercially available Gd-DTPA, or Gd-DOTA) [5, 6]. Although they are not as sensitively detected as their “negative” CA counterparts based on iron oxide NPs, they provide contrast enhancement effects which are largely devoid of magnetic susceptibility image artifacts. In addition, the detection of signal-enhancement effects generated by “positive” CAs could enable more quantitative molecular and cellular imaging studies. Until now, CAs based on Gd and Mn nanocrystals have been largely constrained to pre-clinical studies due to the inherent risks associated with the leaching of potentially toxic Mn2+ and Gd3+ ions. They nonetheless represent very useful contrast agents for research purposes in the field of cell labeling and tracking, as well as for molecular imaging with MRI. This chapter first describes the general considerations which must be taken into account in the design and preparation of MNPs for MRI in vivo applications. Synthesis and coating procedures to prepare suspensions of MNPs of narrow particle size, appropriate magnetic and relaxometric properties, optimal colloidal stability, and good biocompatibility are presented. The contrast-enhancement characteristics of superparamagnetic and paramagnetic NPs are discussed, in particular their respective use either as “negative” or “positive” CAs in T2 /T2∗ -weighted or T1 -weighted MRI sequences.

13.1 The basics of MRI in medicine A brief introduction to the basic principles of MRI will clarify important concepts guiding the design and use of MNP for MRI applications in molecular and cellular imaging. The reader is invited to expand his knowledge on this technology by referring to a selection of readings on the general topic of MRI [40, 41]. In brief, the strength of the magnetic field in common MRI scanners ranges from 1 to 3 Tesla, which is 20 000 to 60 000 times stronger than the Earth’s magnetic field. When a patient is placed in the gantry, the hydrogen protons (1 H) align their spins along the direction of the magnetic field (B0 , Fig. 13.3(a)). The sum of each of the magnetic moments of these spins repre⃗ of the biological tissue. This vector sents the “macroscopic magnetization vector” (M) is globally oriented along the main magnetic field of the scanner. In their initial state,

338 | Part III New trends in sustainable development and biomedical applications the spins “precess” at a certain frequency (ω0 ; the Larmor frequency) and are not coherent in phase. Then, using a transmitter coil, a radiofrequency (RF) wave tuned to Larmor conditions is applied to the biological tissue. This causes the excitation of 1 H spins making them lose their preferential orientation along the main magnetic field (Fig. 13.3(b)). After application of RF excitation, M⃗ oscillates around the main magnetic field (the “z” axis of the MRI scanner), and the spins are phase-coherent in the x-y plane (Fig. 13.3(b)). RF pulse off: – Mz recovery (according to T1 ) – Loss of Mxy phase coherence (according to T2 )

RF pulse on: – Excitation of 1H: loss of Mz – Spin phase coherence in xy

B0

z

B0

Signal

z

B0

z Mz

ω0

ω0

63%

M M

ω0

x

x

T2

37%

T1 Mxy

x

Relaxation time M y

y (a)

(b)

y (c)

(d)

Fig. 13.3. Schematic representation of the macroscopic magnetization vector generated by MR excitation.

From this moment, the oscillation motion along the x-y plane is detected by a receiver coil and recorded. This represents the “MRI signal”. Then the “x-y” phase coherence (Mxy ) is gradually lost (within milliseconds) as the magnetic moments of neighboring 1 H protons exert a mutual influence on each other (Fig. 13.3(c)). The time constant used to quantify this loss of phase coherence is called the transversal relaxation time (T2 ; Fig. 13.3(d)). Independent of this mechanism, the excited spins progressively release their energy and recover their initial orientation along the main magnetic field of the scanner (Mz recovery; Fig. 13.3(c)). This return to the initial macroscopic magnetization state occurs within a time constant that is referred to as the longitudinal relaxation time (T1 ; Fig. 13.3(d)). Both T1 and T2 are intrinsic characteristics of any biological tissue and, together with the density of 1 H spins in the tissue (ρ), are the most important parameters influencing the signal. For instance, the signal recorded for a given tissue (S), using a basic spin-echo sequence, is given by the following equa-

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 339

tion: S = ρ (1 − e(−TR/T1 ) ) (e−TR/T2 ) ,

(13.1)

where TR and TE are the repetition and echo times (parameters in the spin-echo sequence [41]). Interactions between the paramagnetic elements Fe3+ , Gd3+ , and Mn2+ take place at the molecular level, causing T1 and T2 to decrease and, in turn, to modify the MR signal according to equation (13.1). Such interactions accelerate the release of energy communicated to 1 H protons during RF excitation.

13.2 Relaxivity: the performance of MRI contrast agents MNPs made of Fe, Gd, or Mn, influence the macroscopic relaxation times (T1 and T2 ) of 1 H contained in neighboring mobile molecules. In turn, this effect modulates the MRI signal (equation (13.1)), which translates into contrast effects in anatomical MR images. The efficiency of MRI CAs to decrease both T1 and T2 of hydrogen protons contained in the solution, is referred to as the “relaxivity”. Relaxivity depends on the concentration, as well as on the physicochemical characteristics of the CAs: hydrodynamic diameter, number of water binding sites at the magnetic ions, diameter of the inner core, magnetization, high specific surface, etc. The relaxivity of water protons in CAs is also dependent on pH, temperature, and magnetic-field. Some suggested readings on the topic of MRI CA relaxivity are [5, 6, 42–44]. In brief, the effect of CAs on the relaxation time of protons (1 H protons) is usually measured by relaxometric analysis, which is the technical term which refers to the measurement of T1 and T2 of mobile 1 H species in aqueous suspensions or in biological tissues. In MRI, the impact of CAs on the relaxation rate of protons, measured in fixed conditions of magnetic strength and temperature, is described by the following equation: 1 1 Ri = = ( ) + ri C , (13.2) Ti Ti0 where Ri=1,2 is the relaxation rate of the aqueous solution, Ti0 is the relaxation time of the aqueous media in the absence of the CA, ri=1,2 is the relaxivity (usually at T = 20 or 37°C, pH = 7, and B0 = 1.5 or 3.0 Tesla), and C is the CA concentration (in mM of Gd, Fe, or Mn). The concentration of magnetic elements is measured by spectroscopic and spectrometric elemental analysis techniques, for example atomic absorption spectroscopy – AAS, inductively coupled plasma optical emission spectroscopy (ICP-OES), and mass spectrometry, ICP-MS. Therefore, the relaxometric performance of MRI CAs is assessed first by measuring their relaxation rates (1/T1 and 1/T2 ), followed by normalizing the data to the paramagnetic elemental concentration. Relaxivity values (r1 and r2 ) are extracted from the slope of the graph given by equation (13.2). These are often referred to as relaxivity curves, and they figure among the most fundamental aspects of MRI CA quantification. A more detailed explanation of CA relaxivity

340 | Part III New trends in sustainable development and biomedical applications mechanisms is found in Section 13.5. Finally, CAs can be divided into two categories, based on the r2 /r1 ratio. First, CAs having a similar impact on both T1 and T2 , result in low r2 /r1 ratios (close to 1), and a capacity to enhance the MR signal. This is in agreement with equations (13.1) and (13.2). Such CAs are referred to as “positive”. On the other hand, CAs that more preferentially decrease T2 , with r2 /r1 ratios superior to 5 and often as high as 100, are called “negative”.

13.3 Synthesis and characterization of magnetic nanoparticles Comprehensive reviews have been written describing the different ways of synthesizing MNPs. These papers mainly report on iron oxide NPs, and most of these colloidal synthesis routes can also be adapted to other metal ions such as Mn2+ and Gd3+ . The most important criteria guiding the selection and optimization of a particular colloidal NP synthesis route, with MRI applications as an objective, is good control over nanocrystal size, shape, and NP size distribution (as narrow as possible). A reasonable synthesis yield is also critical: CA products must be concentrated enough to produce efficient contrast enhancement effects in MRI procedures; the colloidal synthesis technique must also minimize the loss of paramagnetic materials and surfactants used during the production, in order to enable industrial upscale of the process.

13.3.1 Synthesis of magnetic nanocrystals Until now, NP cores have been made from different materials and with varying sizes, shapes, uniformities, and magnetic properties [45–49]. Apart from MRI applications, MNPs have been formed from iron and cobalt [50]. Procedures enabling the synthesis of CoPt3 [51] and FePt [52], as well as oxides [44] such as magnetite (Fe3 O4 ) and maghemite (𝛾-Fe2 O3 ), have been successfully developed and mastered [44, 53, 54]. Iron oxide NPs have also been doped to enhance their magnetic properties to form MFe2 O4 structures, where M is a +2 cation such as Mn, Fe, Co or Ni [55, 56]. These MNPs make excellent MR contrast agents; their magnetic susceptibility is relatively high, they can also be manipulated by external magnetic fields. However, NPs containing Co and Ni are relatively toxic, making them poor candidates for clinical use. To a lesser extent, this also the case for Gd- and Mn-based nanocrystals. However, these have their own advantages and, as preclinical CAs, can be advantageously applied to molecular and cellular imaging in animal models. For clinical use, the biocompatibility profile of iron oxide NPs is well established, and adequately documented with a large range of clinical studies. Iron oxide NPs injected in vivo eventually degrade to their non-toxic iron and oxygen components, making them particularly attractive as clinical MRI CAs [57].

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

|

341

The present section reports the major colloidal nanoparticle synthesis procedures which must be selected to enable efficient, rapid, and high-yield production of small particles of well-controlled and narrow size distributions. For both paramagnetic and superparamagnetic nanocrystals, core size dictates both the magnetic and the relaxometric performance of CAs. Good control over this parameter is therefore critical in all aspects, and bottom-up approaches using chemical colloidal synthesis techniques are almost invariably preferred. In the interest of brevity, and because nanomaterials made of Fe, Gd or Mn account for the huge majority of MRI NPs in pre-clinical and clinical applications, only synthesis routes enabling the production of nanocrystals containing these three elements are described here. A selection of references is provided for the reader who wishes to read about variants to these colloidal synthesis techniques. Iron oxide nanoparticles: Among the most important colloidal synthesis methods used to make iron oxide NPs figure co-precipitation, co-precipitation in constrained environments, thermal decomposition and/or reduction, hydrothermal synthesis, and polyol synthesis [44, 49, 58–61]. Each has its own specific advantages. The most common, the simplest and possibly the most efficient is co-precipitation. It is based on the use of an aging stoichiometric mixture of ferrous and ferric ions (Fe2+ /Fe3+ ) in aqueous solutions. The chemical reaction for the formation of magnetite (Fe3 O4 ) is: Fe2+ + 2Fe3+ + 8OH− ⇒ Fe3 O4 + 4H2 O. According to the thermodynamics of this reaction, complete precipitation of Fe3 O4 occurs at basic pH, with a stoichiometric ratio of 2 : 1 (Fe3+ : Fe2+ ) [62]. Therefore the reaction takes place with the addition of base under an inert atmosphere. In the presence of oxygen, magnetite oxidizes into maghemite, as follows: Fe3 O4 + 2H+ ⇒ 𝛾–Fe2 O3 + Fe2+ + H2 O. In the co-precipitation technique, NPs are formed by a nucleation and growth mechanism. NPs of relatively narrow size distributions can be synthesized, provided a short nucleation event takes place followed by a slower growth phase. The types of salts employed (e.g., chlorides, sulfates, nitrates), the ratio between ferrous and ferric ions, the temperature, the pH, and ionic strength are all parameters which must be finely tuned to yield NPs of the desired size and narrow distribution [49]. Although they are totally appropriate for the synthesis of large amounts of MNPs, standard co-precipitation methods do not consistently deliver the same products (size, shape, and polydispersity). The presence of impurities and surface defects also affects the magnetic properties of such NPs [46]. Adaptations to the co-precipitation approach to improve the uniformity and stability of USPIOs and SPIOs have been investigated. These were made through addition of polymers or polyelectrolytes to the ferric/ferrous ion solutions, with various improvements to size, shape, and crystallinity [46, 63–66]. The addition of chelating

342 | Part III New trends in sustainable development and biomedical applications organic ions (carboxylate or α hydroxyl carboxylate ions such as citric acid, gluconic, or oleic acid) or polymer surface complexing agents (dextran, carboxydextran, starch, or polyvinyl alcohol) during the formation of the magnetite crystals help to control NP size. Indeed, these charged organic molecules bind at the surface of iron oxide, and this strategy can be used to modulate nanoparticle growth. Adding polymers such as poly (acrylic acid) directly into the synthesis solution and in various concentrations, allows tuning of the particle diameter between 7 and 14 nm [67]. Alternatively, polyethylene glycol (PEG) compounds such as PEG-g-poly(glycerol monoacrylate) are also used to modulate the size of USPIOs [68] during the synthesis stage. Importantly, these polymers may act as surface coatings when nucleation and growth processes are complete. These are called in situ coating processes. In order to achieve smaller sizes, narrower particle distributions and higher magnetic properties, new synthesis techniques have been developed, based on hightemperature decomposition methods using organic iron precursors [69]. For instance, a high-temperature reaction of iron (III) acetylacetonate, Fe(acac)3 , in phenyl ether in the presence of alcohol, oleic acid, and oleylamine, yielded monodisperse, hydrophobic magnetite NPs with tunable sizes of 4–20 nm [70]. One of the major pre-requisites of MNPs for MRI applications is that they disperse well in aqueous solvents. Therefore, an additional step must be introduced to the synthesis procedure: the replacement of the hydrophobic coating with an amphiphilic and biocompatible surfactant. The polyol process is an alternative to thermal decomposition methods for the synthesis of NPs with well-defined shapes and controlled sizes [71, 72]. Owing to their high dielectric constants, solvents such as polyethylene glycol (PEG) are able to dissolve inorganic compounds in a wide range of temperatures. Polyols also serve as stabilizers to control particle growth and prevent particle aggregation. In polyol synthesis, the precursor compound is suspended in a liquid polyol, and then heated to a given temperature. During this reaction, the solubilized metal precursor forms an intermediate, and is then reduced into metal nuclei which lead to NP growth. Good examples of iron oxide NP synthesis in different polyols (di-,tri-,tetra-) ethylene glycol, can be found in a selection of recent articles [73–75]. Overall, the nanoparticles synthesized by polyol routes have the smallest and narrowest size distributions, high water dispersion rates, and higher magnetization compared with particles produced by more conventional methods. Paramagnetic nanocrystals (Gd2 O3 and NaGdF4 ): the main advantage of paramagnetic CAs compared to superparamagnetic NPs is the ability to generate a “positive” signal where the labeled molecules or labeled cells are accumulated. Signal enhancement is generally easier to quantify than signal voids resulting from T2 /T2∗ (e.g., magnetic susceptibility) effects (generated by iron oxide NPs). A good example of this is illustrated in Fig. 13.4 [76]. PEG-coated Gd2 O3 NPs are used to label different types of cells [77–83] and, in this specific example, F98 brain cancer cells. A total of 3 × 105 labeled F98 cells were injected into mice brains (caudoputamen), to pro-

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 343

duce a brain tumor. This animal model is frequently used in the field of brain cancer and oncology research. The same amount of SPIO-labeled cells was injected, to enable comparison of the capacity of both contrast media to allow cell tracking in MRI. Figure 13.4(a) reveals the ability to efficiently visualize the area of brain cancer cell implantation at least 48 hours after injection. Because this is a “positive” CA scanned with a T1 -weighted MR sequence, the anatomical information surrounding the area of cell implantation is preserved. After one week, the tumor contours can be efficiently delineated without the use of a CA. On the other hand, SPIO-labeled cells are much more sensitively detected than Gd2 O3 -labeled ones (Fig. 13.4(b)). However, the susceptibility artifact characteristic of iron oxide NPs largely exceeds the exact location of the implanted cells. Even after one week, the image artifact still obliterates important anatomical information directly in the region of the growing tumor. This is a good example of the potential of paramagnetic NPs to replace SPIOs and UPSIOs in niche applications where the preservation of anatomical details is an issue (such as in cell implantation and tracking studies). Day 2

Day 7

Gd Labeled

Day 1

Fe Labeled

(a)

(b) Fig. 13.4. Mice brains implanted with 300,000 brain cancer cells (F98) labeled with (a) PEG-Gd2O3 (T 1 -weighted MR imaging), and (b) SPIOs, (T 2 -weighted MR imaging). Adapted with permission from [76].

The very high density of Gd atoms per unit of contrast agent is an advantage over macromolecules containing Gd chelates: Gd2 O3 and NaGdF4 ultra-small nanoparticles (2−5 nm diameter) have narrow particle size distributions, and contain hundreds of Gd atoms per CA unit [31, 76, 84]. The production of Gd2 O3 NPs, a pre-clinical CA, has required the development of advanced colloidal synthesis techniques in high boiling point alcohols (e.g., di-, tri-, poly-ethylene glycol) [31, 85–88]. The surface of Gd2 O3 nanocrystals forms hydroxide in contact with water and this form is at risk of leaching potentially toxic Gd3+ ions, as demonstrated in prolonged dialysis procedures [30]. Concerns related to the potential toxicity of Gd2 O3 nanoparticles, even for pre-clinical

344 | Part III New trends in sustainable development and biomedical applications applications in animal models, have led to the development of other forms of potentially more stable paramagnetic nanocrystals. NaGdF4 , in particular, is an attractive material due to its high concentration of Gd atoms in a crystalline form which is less susceptible to degradation in water. Sodium fluoride NPs are synthesized by thermal decomposition in oleic acid and octadecene, leading to hydrophobic surfaces which must subsequently be transferred to aqueous suspensions by using appropriate ligand exchange procedures [89–91]. Rare-earth fluorides doped with series of lanthanides have very promising luminescent up-conversion properties. In particular, it has been demonstrated that ultra-small NaGdF4 nanocrystals (3 nm diam.) doped with Tm and Tb can be used for dual MRI and near-infra-red optical imaging, with a wide array of applications in biomedical research [34]. Antiferromagnetic MnO nanocrystals: Although Mn2+ ions are paramagnetic, MnO NPs express antiferromagnetic behavior [92]. They behave overall as “positive” CAs. Thermal decomposition has been one of the most widely used and reliable methods of producing relatively large NP batches of small and narrow particle size distributions [38, 93–95]. One-pot synthesis techniques have also been developed in high boiling point solvents, enabling the production of 1–3 nm diameter MnO particles [87].

13.3.2 Nanoparticle coatings for MRI applications NPs must form stable colloids in physiological media (blood, plasma, lymph, urine) in order to be considered for MRI applications. They must also demonstrate good biocompatibility and prolonged vascular retention. This is particularly critical in the case of targeted CAs, which must provide sufficient contrast at the molecular site, where they are expected to bind to molecular biomarkers (e.g., of atherosclerosis or Alzheimer’s disease). The ligands and polymers used as particle coatings must efficiently cover the particles and promote their individualization. They must also limit the adsorption of plasma proteins, which is the phenomenon that is a precursor to their retrieval from blood. Indeed, “opsonized” NPs are quickly recognized and ingested by macrophages. NPs must be coated with a surface ligand or a polymer which induces either an electrostatic or a strong steric repulsion between the particles. Otherwise, particles are likely to agglomerate when they are submitted to strongly ionic conditions. Hence, any injection of magnetic nanoparticles in vivo implies their preparation in a milieu close to the osmolality conditions of biological media. In general, small hydrophilic particles of neutral charge demonstrate long plasma half-life [96]. The selected biocompatible coatings should not influence cell viability. They must be grafted with a functionality (e.g. carboxylates, phosphates, and sulfates) which strongly binds to the NP surface. The coatings should also be stable under various pH conditions. For instance, acidic functions (–COOH) can bind to metal oxide surfaces. However, if the bonding is only monodendate, their attachment at the particle surface is relatively weak and not strong enough for many applications. Citric acid is a

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 345

small and very effective ligand which binds to the surface of iron oxide NPs through carboxylic binding [97]. In particular, this strategy was used to synthesize the commercial product VSOP C184 (4 nm core size) [98]. However, citric acid causes a strong surface degradation, which affects the magnetic properties of iron oxide NPs [99]. It is also very rapidly and strongly diverted to the liver due to its high surface charge. Other ligand molecules can be used for the stabilization of iron oxide NPs in aqueous medium (e.g., gluconic acid, dimercaptosuccinic acid, phosphorylcholine, as well as phosphate and phosphonates) [100, 101]. Unfortunately many polymer coatings result in an unacceptable increase of the hydrodynamic diameter, which can be detrimental to the overall performance of the contrast agent (blood retention, relaxivity). In summary, the selection and application of an appropriate coating for MNPs is a critical step which has a large influence on the overall performance of the nanoconstructs.

n=335 8.4±0.8 nm

0

50 nm (a)

4

6

8

10 12 14 16

(b)

25 Arbitrary units (a.u.)

2

Particle core size by TEM (nm)

2 hours 4 days 7 days

20 15 10 5 0 1

10

100

Hydrodynamic size by DLS (nm) (c) Fig. 13.5. (a) Electron microscopy study of MnO nanoparticles produced by thermal decomposition, with (b) particle-size distribution, and (c) hydrodynamic size measurements. Adapted with permission from [39].

346 | Part III New trends in sustainable development and biomedical applications Among the polymer coatings which most efficiently enhance the blood retention of SPIOs and USPIOs figure dextran and (carboxy, carboxymethyl)dextran [102, 103]. In particular, commercial USPIO products such as ferumoxtran-10 (AMI-227), ferumoxytol, and Supravist (SHU-555C), are all based on iron oxide of core diameters in the range of 4 to 8 nm, and of hydrodynamic diameters not larger than 30 nm [104]. Dextran and starch-coated iron oxide nanoparticles have shown very good relaxometric properties; however, such coatings are relatively unstable when submitted to biological media. In order to improve colloidal stability, biocompatibility, and blood retention, PEG is used at the surface of iron oxide nanoparticles [105–107]. Feruglose (Clariscan) is a commercial product which was developed using a PEGylated starch coating [108]. Such coatings are less prone to being recognized by the macrophagemonocytic system. Paramagnetic Gd2 O3 and NaGdF4 , as well as MnO nanoparticles, have also been coated using similar strategies (e.g., citric acid, dimercaptosuccinic acid, glucoronic acid, PEG) [28, 30, 34, 109–112]. PEG is widely applied as coating for paramagnetic nanoparticles, through –OH [86, 113], –COOH [31, 114], -silane [30, 109, 115], and -phosphate grafting [116]. Silane coatings, for instance, could delay the degradation of Gd2 O3 nanoparticles submitted to acidic environments. Unfortunately, silane coatings restrict the optimal water exchange with paramagnetic ions at the surface of NPs, which is the most important relaxation mechanism of “positive” CAs (see Section 13.5). PEG-phosphate molecules can be used instead, with the significant advantage that they are not susceptible to homocondensation such as for silane-based products [116]. Recently, PEGylated phosphonate dendrons were successfully used to cover MnO and iron oxide nanoparticles, and this coating strategy provided enhanced NP excretion profiles through the urinary and gastrointestinal pathways [117, 118].

13.3.3 Physicochemical characterization After synthesis and ligand exchange, NPs must be carefully cleaned of the residual magnetic ions (Fe2+ , Fe3+ , Gd3+ , Mn2+ and other) which could contaminate the physicochemical, magnetometric, and relaxometric measurements. This procedure can be achieved either by dialysis in saline water (e.g., 10–154 mM NaCl), by centrifugationfiltration cycles, or by size-exclusion chromatography. Dynamic light scattering (DLS) measurements are performed directly after the purification process, in order to assess the overall colloidal stability and to demonstrate the absence of large-size aggregates. No drastic divergence should be noted between “intensity”- and “number”-weighted results, as diverging results point to the presence of large-size agglomerates which must be eliminated prior to further use of the particles. The NPs must then ideally demonstrate the presence of only one major hydrodynamic diameter peak for their magnetic and relaxometric properties to be adequately predicted and controlled. To demonstrate the colloidal stability of the particles, a weeklong DLS assay experiment is necessary, in parallel with zeta potential measurements (electrostatic charge

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 347

of NPs). The particles are then submitted to a comprehensive set of physicochemical characterization measurements using high-resolution electron microscopy (HRTEM: particle size, morphology, crystallographic parameters), x-ray photoelectron spectroscopy (XPS: surface elemental analysis), Fourier transform infrared spectroscopy (FTIR: molecular groups at the surface of particles; molecular grafting assessment), thermogravimetric analysis (TGA: mass ratio between inorganic cores and organic coating), 1 H-NMR relaxometry (measurement of T1 and T2 ), and magnetometric measurements, to name the most important techniques. Finally, the elemental concentration (Fe, Gd, Mn) of colloidal NPs is measured with spectrometric or spectroscopic instruments by AAS, ICP-OES, ICP-MS (mentioned in Section 13.2), after careful digestion in appropriate acidic conditions.

13.4 Physical properties of magnetic nanoparticles NPs based on SPIOs, USPIOs, Gd2 O3 , NaGdF4 , and MnO, have core diameters typically in the range of 2–20 nm. However, superparamagnetic and paramagnetic nanoparticles have very different magnetic behaviors, significantly influencing their relaxometric performance. USPIOs and SPIOs are said to be superparamagnetic because they have no remanence after being introduced into and retracted from the strong magnetic field. Upon introduction in the scanner, the global magnetic moment of superparamagnetic NPs aligns in the direction of this magnetic field. As soon as the magnetic field is set back to zero (e.g., when the patient is removed from the scanner), the magnetic moment of the NP also goes down to zero. This is not the same behavior as for “bulk” ferromagnetic magnetite/maghemite, materials which clearly show a strong remanence. The absence of residual magnetization is a very critical and useful aspect of superparamagnetism applied to biomedicine. Macroscopically, the magnetic behavior of superparamagnetic particles is similar to paramagnetism (e.g., Gd2 O3 , NaGdF4 ), except that they feature an exceptionally high magnetic moment per unit of CA (the “core” of iron oxide particles). This strong magnetization is largely responsible for the remarkable “negative” CA properties of iron oxide NPs. In fact, the absence of magnetic remanence for USPIOs and SPIOs is due to the return to equilibrium of the magnetic moments through Néel relaxation. On the other hand, paramagnetic NPs, and to a lesser extent antiferromagnetic MnO nanocrystals, do not develop strong magnetization at clinical magnetic field strengths. Instead, they generate signal enhancement (“positive” contrast), mainly through direct interactions taking place between Gd and Mn ions and mobile 1 H protons in their surroundings, as described in Section 13.5. The basic differences in magnetism between superparamagnetic and paramagnetic NPs are introduced below. (a) Superperparamagnetic NPs are very efficient “negative” CAs in MRI [119]. Due to their strong impact on the transverse relaxation times (T2 /T2∗ ) of aqueous solutions, they have been considered very useful products for molecular and cellular

348 | Part III New trends in sustainable development and biomedical applications MRI [103]. As mentioned previously, each superparamagnetic iron oxide NP features a high magnetic moment, and much higher Curie constants (material-dependent magnetic susceptibility constants) than for paramagnetic NPs. As a result, they respond quickly to the application of an external magnetic field and their magnetization quickly becomes saturated at relatively low magnetic field strengths (Fig. 13.6). 80

Magnetization (Am2 Kg‒1 ferrite)

60

IONs (USPIO), 5.59 nm IONs (USPIO), 5.67 nm IONs (USPIO), 4.84 nm

40 20 0 ‒20 ‒40 ‒60 ‒80 ‒1000

‒500

0 Field (mT)

500

1000

Fig. 13.6. Magnetometric measurements of USPIOs and Gd2O3 nanoparticles Adapted with permission from [44, 120].

Superparamagnetism only occurs when NPs are small enough to belong to single magnetic domains. It is worth mentioning that suspensions of iron NPs (and not iron oxide nanoparticles) would have a much higher magnetization than magnetite/maghemite (about 5 times higher); however, iron NPs are very quickly oxidized into iron oxide NPs in aqueous media, and until now, this potential technology could not be applied in biomedicine. Magnetite (Fe3 O4 ) and maghemite (𝛾-Fe2 O3 ) are two relatively similar forms of iron oxide (crystal structure and magnetic properties) [121, 122]. Both are present in superparamagnetic iron oxide NPs, and it is often difficult to distinguish between them using common x-ray diffraction analysis techniques. Magnetite is typically preferred due to its superior magnetic properties [44]. Maghemite (Fe3+ [Fe2+ Fe3+ ]O4 ) often results from the oxidation of magnetite (Fe3+ [Fe3+ 5/3 V1/3 ]O4 , where V represents a cation vacancy). Bulk magnetite is ferromagnetic. The occurrence of an oxygen-mediated coupling mechanism aligns all

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 349

the magnetic moments of the iron ions located in the tetrahedral sites of the crystal (8 crystallographic sites per unit structure), whereas all the magnetic moments of the octahedral ions (16 crystallographic sites per unit structure) are aligned in the opposite direction. It is assumed that the magnetic properties of magnetite are provided by uncompensated Fe2+ ions, whereas for maghemite they are provided by Fe3+ ions [123]. The magnetic energy of iron oxide NPs depends upon the direction of their magnetization vector, and this vector in turn depends on the crystallographic directions (the magneto-crystalline anisotropy field) [121]. The directions which minimize the magnetic energy are called anisotropy directions, or easy axes (Fig. 13.7(a), from [121]). The resulting magnetic moment of a magnetite/maghemite crystal is preferentially aligned along such specific directions. Magnetic energy increases with the tilt angle between the magnetic vector of the easy directions [124]. For the sake of simplicity, the anisotropy of magnetite particles is often assumed to be uniaxial, with a single anisotropy axis. In fact, there are several anisotropy axes dictated by the crystallographic structure of the oxide. The anisotropy energy (the amplitude of the curve), is given by the product of the crystal volume (V) times a constant (Ka : the anisotropy constant): Ea = Ka V. (13.3) The anisotropy energy, proportional to V, determines the Néel relaxation time. Large samples of bulk ferrimagnetic magnetite/maghemite, are divided into Weiss domains (represented in Fig. 13.7(b)) Inside each of these volumes, the magnetic moments are aligned into one preferential direction; however, between each of these domains, the magnetic moment is not oriented along the same direction. As iron oxide nanocores (such as in USPIOs) are smaller than one of these domains, each NP is composed of a single domain whose magnetic moment is oriented in a specific direction. In these single domains, the direction of the magnetic moment can flip from one orientation to the other. When the thermal energy, given by kT (k: Boltzman constant; T: absolute temperature) is sufficient to overcome this anisotropy energy barrier, the magnetization fluctuates between the different anisotropy directions, according to a characteristic time: the Néel relaxation time (τN ) [126]. Although τN relaxation influences the hydrogen relaxation times by inducing changes to the magnetic moment of MNPs, it is a phenomenon entirely distinct from the nuclear relaxation mechanisms of hydrogen protons (1 H) described in Section 13.5. Néel relaxation refers to the relaxation of the global electronic moment of a superparamagnetic crystal constituted by a ferri-, ferro-, or antiferromagnetic compound. For dry powders of monodomain iron oxide NPs, τN indicates the time it takes for magnetization to return to a state of equilibrium after being submitted to a strong magnetic field. For highly anisotropic crystals, the crystal magnetization is “locked” in the easy axes. Néel relaxation defines the rate of fluctuations arising from the jumps of the magnetic moment between the different easy axes (Fig. 13.7(c)). In order to flip from one easy direction to the other, the magnetization of an NP must jump over an anisotropy energy hump. For a super-

350 | Part III New trends in sustainable development and biomedical applications 0.2 μm

0.012 μ Anisotropy axis

0.010

Probability

0.008

0.006

θ

0.004

0.002

(b)

Typical magnetite contrast agent

0.000 20

0

40

60

(a)

80

100

120

140

160

180

θ (degrees)

Brownian relaxation

After application of a magnetic field B

After application of a magnetic field B τB

τB

τN B

B τN

τB

1e-3

τB

Relaxation time constant, τ(s)

Neel relaxation

(d)

τB

1e-5

τ

1e-6 1e-7 1e-8 4

(c)

τN

1e-4

5

6

7

8

9

10

11

Particle radius, R(nm)

Fig. 13.7. Magnetic behavior of iron oxide NPs (radius = 5 nm): (a) uniaxial anisotropy for magnetite/maghemite nanoparticles (i.e., the probability of alignment of the magnetic moment in one direction with respect to the angle between this direction and the anisotropy axis); (b) representation of Weiss domains in a large magnetite/maghemite crystal, compared to the dimensions of a typical NP (the small circle), much smaller than a Weiss domain; (c) schematic representation of Néel relaxation and Brownian relaxation; (d) relaxation time values plotted as a function of magnetite/maghemite NP size. Reprinted with permission from [100, 121, 125].

paramagnetic NP of specific V and Ka , the Néel relaxation time (τN ) is given by an Arrhenius law which is similar to the one describing the activation energy for a chemical reaction [127]: Ea τN = τ0 (Ea )e kT , (13.4) where τ0 (Ea ) is the pre-exponential factor of the Néel relaxation time expression, which depends on factors such as volume (V), specific magnetization of the nanocrystal and gyromagnetic ratio of the electron [44, 128, 129]. Whereas the pre-exponential factor decreases as the value of anisotropy energy increases, τN increases as an exponential function of V because of the second factor of equation (13.4). For small

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

|

351

values of the anisotropy energy and at high temperatures, Ea ≪ kT (the exponential term tends to equal 1), and τN is mainly determined by the preexponential term. These conditions are fulfilled, for instance, with individual ultra-small NPs of iron oxide of r < 4 nm. On the other hand, for high anisotropy energies, when Ea ≫ kT, the evolution of τN is mainly dictated by the exponential factor (rapid increase with Ea ). According to equation (13.4), the flipping of the magnetic moment of magnetite/ maghemite crystals is observed only for NPs of size r < 12 nm. Indeed, for magnetite (τ0 ≈ 10−9 s, K ≈ 13 500 J m−3 ), τN goes from ∼ 500 years for particles of r = 15 nm, down to the ms for particles of about r = 10 nm. Practically, this means that for particles with a Néel relaxation time (τN ) longer than the measurement time, the magnetization curve of the NP system is irreversible and shows a hysteresis loop. These are referred to as “frozen single domain” NPs. NPs which demonstrate Néel relaxation times of several years or centuries, can be used in computer hard disks. Indeed, such longterm data storage applications require as little magnetic material as possible, with a maximal anisotropy constant (Ea ). However, such NPs cannot be used in MRI applications. Firstly, the Néel relaxation of larger nanoparticles showing high Ea has no effect on the nuclear relaxation of neighboring water protons. Indeed, the diffusive rotation time of the particles in water, and the diffusion time of water around the particles are both much shorter than 1 ms. This is incompatible with the long Néel relaxation times of large particles. In order for magnetic iron oxide NPs to provide efficient properties for use as MRI CAs, the condition τN < 1 s should be respected. Then, for NPs dispersed in a liquid media (a colloid), the return of magnetization to equilibrium after application of a strong magnetic field is determined by both τN and the Brownian relaxation τB of the particles. The latter characterizes the rotation of the particle in the fluid, and takes the viscosity of the solvent into account (Fig. 13.7(c)). The global magnetic relaxation rate is a sum of two processes: 1 1 1 + , (13.5) = τ τN τB where τ is the global magnetization time and τB is the Brownian relaxation time, given by: 3Vη τB = , (13.6) kT where η is the viscosity of the solvent. For large particles, τB < τN because the Brownian component of the magnetic relaxation is proportional to the crystal volume (equation (13.6)), and the Néel relaxation is an exponential function of the volume (equation (13.4)). Therefore, in suspensions of large NPs, the viscous rotation of the particles in the liquid becomes the dominant process determining the global magnetic relaxation properties (Fig. 13.7(d)). As a result, the magnetic relaxation of suspensions of magnetic nanoparticles is much faster than for dry powders of iron oxide NPs. In summary, superparamagnetism refers to a specific magnetic condition for which ultra-small particles of a size well inferior to typical Weiss domains of mag-

352 | Part III New trends in sustainable development and biomedical applications netite/maghemite materials can be submitted to high magnetic field strengths, without evidence of magnetic remanence once they are retrieved (Fig. 13.6). Macroscopically, the magnetization of iron oxide NPs suspensions is described by a Langevin function, whose shape depends on the saturation magnetization (Msat ) and the size of the magnetite crystals: M(B0 ) = Msat L(x), (13.7) where Msat is the magnetization at saturation, and L(x) is the Langevin function as: L(x) = [coth(x) −

1 ] x

(13.8)

with :

Ms (T)VB0 . (13.9) kT Magnetization curves of ultra-small iron oxide NPs are perfectly reversible because the fast magnetic relaxation allows the system to remain consistently at thermodynamic equilibrium [130]. Finally, because iron oxide superparamagnetic NPs obey this law, it is possible to estimate the NP core size from the fit of the magnetization curves (Fig. 13.6). (b) Paramagnetic nanoparticles also respond to the application of an external magnetic field by developing a magnetization vector oriented along the direction of this field which slightly increases the local magnetic field strength [131]. Paramagnetic nanocrystals follow Curie’s law: x=

M = χH =

C H, T

(13.10)

where χ is the magnetic susceptibility, H is the applied magnetic field (e.g., that of the MRI scanner), T is the absolute temperature, and C is a material-specific Curie constant. Unlike ferromagnets, paramagnetic materials do not retain magnetization in the absence of an external magnetic field, and the thermal energy is sufficient to randomize the induced magnetization. At similar concentrations of metal elements (Fe, Gd, or Mn), the “positive contrast effect” of paramagnetic NPs is less efficiently detected than the “negative contrast” effect generated by superparamagnetic agents. The more limited magnetization response of the rare-earth ions compared with USPIOs and SPIOs is due to their magnetic moment, which is not saturated at magnetic field strengths typically used in MRI [4, 132]. The difference of magnetization response between iron oxide (USPIOs) and paramagnetic Gd2 O3 nanoparticles is shown in Fig. 13.6.

13.5 MR relaxation properties of magnetic nanoparticles For a given concentration of magnetic atoms, the magnetization of paramagnetic CAs is much lower than that of superparamagnetic nanocrystals. However, Gd chelates are

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 353

still by far the most widely used CAs in routine MRI. This is mainly due to their exceptional relaxometric properties. Both paramagnetic and superparamagnetic CAs interact with the small, hydrogen-containing mobile molecules found in biological media. In fact, one of the most important factors influencing signal enhancement in MRI is the binding of water molecules to paramagnetic ions. The motion of contrast agents in the fluid also has an impact, as well as the magnetic field strength-dependent electronic relaxation of the paramagnetic elements. Finally, Gd3+ chelates are, in general, very efficiently excreted by the kidneys, this also accounting for their widespread use in clinical MRI. This section does not aim to provide a comprehensive explanation of the complex relaxometric mechanisms taking place between paramagnetic – or superparamagnetic – CAs, and the water molecules. The reader is directed towards several books and review articles which have extensively described this subject; see [4, 6, 44, 133–135]. A summary follows of the most important aspects to take into account when designing optimal nanoparticulate contrast agents for MRI applications.

13.5.1 Relaxivity of paramagnetic CAs Theories explaining the relaxivity of paramagnetic CAs have been developed since the inception of MRI. In fact, the efficiency of CAs is linked to molecular motions of the CA unit, as well as to the motions of small molecules containing the “spins” (1 H protons). It is also linked to intrinsic properties of the nuclei (magnetic moment, gyromagnetic ratio, spin). The relaxation of paramagnetic solutions is mainly explained by two mechanisms: the “inner sphere” (IS) and “outer sphere” (OS) contributions [4]. Using equation (13.2) it is possible to dissociate two contributions for Ti , one from the water protons in contact with the paramagnetic elements (p), and one for the rest of the water protons in the matrix (diamagnetic contribution, d): (

1 1 1 )=( )+( ) Ti, obs Ti,d Ti,p

(13.11)

where i = 1, 2. From the paramagnetic contribution it is possible to dissociate the IS and OS contributions as follows: IS

(

1 1 1 )=( ) +( ) Ti,p Ti,d Ti,p

OS

(13.12)

The IS relaxation relies on the exchange of energy between the spins and the electrons of the paramagnetic elements, which is facilitated when water molecules bind to the paramagnetic ions (Fig. 13.8). The water molecules which bind to the paramagnetic center (Gd3+ , Mn2+ ) ions, and water molecules denoted “p” (red circles in Fig. 13.8) rapidly leave the first coordination sphere and are immediately replaced by “fresh” molecules from the matrix (“d” water molecules: orange circles in Fig. 13.8). Hydrogen relaxes faster on contact with the paramagnetic ions. The water residence time in the

354 | Part III New trends in sustainable development and biomedical applications inner sphere (τM ) is in the order of ∼ 1 ns, which means that the relaxation effect propagates very fast to the rest of the solution (to “d” protons). Each of the water protons which relaxes energy participates in the decrease of the overall longitudinal relaxation time of the water solvent. The IS model is described by the Solomon–Bloembergen– Morgan theory (SBM) [136, 137].

OS

Gd3+ τR

τM

T1,2e

Fig. 13.8. Schematic representation of the inner sphere (IS), outer sphere (OS), chemical exchange, and rotational correlation times guiding paramagnetic relaxation.

IS

From the inner sphere contribution, T1 in the first coordination sphere is: (

1 IS 1 ) = fq( ), T1 T1M + τM

(13.13)

where f is the relative concentration of paramagnetic complex over water molecules, and q is the number of water molecules in the first coordination sphere. The calculation of T1M is based on a model including the amplitude of the magnetic interaction, its temporal modulation, and the effect of the external magnetic field strength as follows: 7τc2 3τc1 1 2 μ0 2 2 2 2 1 ], ( ) 𝛾H 𝛾S ℎ S(S + 1) 6 [ = + T1M 15 4π r 1 + (ωS τc2 )2 1 + (ωH τc1 )2

(13.14)

where: 1 1 1 1 = + + τci τR τM τsi

(13.15)

1 1 1 4 = [ + ] 2 2 τs1 5τSO 1 + ωS τV 1 + 4ω2S τ2V

(13.16)

5 1 2 1 [3 + ]. = + τs2 10τSO 1 + ω2S τ2V 1 + 4ω2S τ2V

(13.17)

Further, 𝛾S and 𝛾H are the gyromagnetic ratio of the electron (S) and the proton (H) respectively, ωS,H is the angular frequency of the electron and the proton, r is the distance

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 355

between coordinated water protons and unpaired electron spins, τc1,2 is the correlation time modulating the interaction (defined by equation (13.15)), τR is the rotational correlation time of the hydrated complex, τs1,2 is the longitudinal and transverse relaxation of the electron, τS0 is the value of τs1,2 at zero field, and τv is the correlation time characteristic of the electronic relaxation times. The second term of equation (13.12), the outer sphere relaxation (OS), is explained by the dipolar interaction at long-distance between the magnetic moment of the paramagnetic substance and the nuclear spin of hydrogen protons. In fact, paramagnetic centers influence the local magnetic field around the 1 H protons flowing in their vicinity. The complete equations explaining the outer sphere contribution of the longitudiOS

nal relaxation rate ( T1 ) , are beyond the scope of this chapter and will not be detailed 1 here. The reader is invited to refer to the description of the OS model by Freed [44, 138]. The most important aspect to retain regarding the OS contribution in the case of paramagnetic substances is the fact that the dipolar intermolecular mechanism is modulated by the translational correlation time (τD ) which takes the relative diffusion (D) of the paramagnetic center and the solvent molecule into account, as well as their distance of closest approach (d) as follows[138]: τD =

d2 D

(13.18)

This expression indicates that viscous solvents and large particles lead to high translational correlation times. In summary, the equations describing IS and OS contributions to the relaxivity of paramagnetic CAs are rather complex, and a large number of parameters influence the overall relaxometric performance of CAs: τM , q, τR , D, r, d, τV , τS0 . Because of this high number of parameters it is often difficult to perform an accurate theoretical estimation of the performance of MRI CAs at different magnetic field strengths. Relaxometric measurements must be performed experimentally, using a technique called proton nuclear magnetic relaxation dispersion (NMRD). NMRD curves characterize the efficiency of CAs at different magnetic fields. For more information regarding the interpretation of NMRD curves, the reader is invited to read a selection of references on the subject [4, 44, 134]. A few points which are important to know prior to the interpretation of NMRD profiles for paramagnetic solutions are presented below. – The rotational correlation time (τR ) characterizes the reorientation of the vector between the paramagnetic ions and the protons of the water molecule. For a low molecular weight complex, τR limits the relaxivity of paramagnetic CAs at magnetic field strengths used in clinical MRI (0.5–3 Tesla). The value of τR cannot be measured by proton NMRD, but instead by 17 O NMR measurements (longitudinal relaxation of the nucleus 17 O), and other methods [4]. – Electronic relaxation times (τS1 and τS2 ). Longitudinal and transverse electronic relaxation times describe the process of return to equilibrium of the magnetization associated with electrons which transit between electronic levels of

356 | Part III New trends in sustainable development and biomedical applications

–

–

–

the paramagnetic center. These transitions produce fluctuations which allow the relaxation of protons; τS1 and τS2 are magnetic field-dependent. Number of coordinated water molecules (q) strongly influences the IS contribution. For small complexes such as Gd-DTPA, the number of coordinated water molecules is equal to 1. This means that, in general, only one water molecule can bind to the paramagnetic Gd3+ ion sequestrated in the DTPA molecule. The value of q can be estimated either in solid phase (x-rays or neutron diffraction) or in solution (fluorescence of Eu or Yb complexes, LIS (lanthanide-induced shift) method in 17 O-NMR). Proton-metal distance (r). The efficiency of the IS dipolar mechanism is proportional to 1/r6 , where r is the metal-proton distance. Small changes to this distance have a considerable impact on relaxivity. Coordinated water residence time (τM ). The mechanism of IS relaxation is based on an exchange between water molecules surrounding the complex and the water molecules coordinated to the lanthanide ion. Consequently, the exchange rate (kex = 1/τM ) is an essential parameter for transmitting the relaxation effect to protons in the water matrix [4].

13.5.2 Relaxivity of superparamagnetic CAs For superparamagnetic particles, the IS contribution to relaxation is minor and often completely negligible compared to the dominant OS contribution. As mentioned in the previous section, this contribution is largely dependent on the movement of water molecules near the local magnetic field gradients generated by the superparamagnetic nanoparticles. The relaxation of superparamagnetic NP suspensions is generally governed by Freed’s equations when τS1 is the Néel relaxation time [139]. When τD is much shorter than the Néel relaxation time, Freed’s equations are simplified. In fact, the ability of a fluctuation to relax the 1H proton spins depends upon whether its correlation time is longer or shorter than the precession period of the spins within the exter−1 −1 nal magnetic field B0 . If the global correlation time τC (τ−1 C = τD + τN ) is longer than this period, the fluctuation is averaged by the precession and it is inefficient. Other parameters such as electron polarization and crystal anisotropy also have a considerable influence on the relaxation times of water protons submitted to different magnetic fields. As in the case of paramagnetic CAs, the theoretical models explaining the relaxometry of aqueous suspensions of iron oxide NPs must be validated by NMRD profiles (Fig. 13.9). First, it is possible to precisely measure, with NMRD, the relaxometric potential of NP as contrast agents for MRI [140]. NMRD profile analysis then also represents a powerful tool to control the reproducibility of synthetic MNP suspensions, as well as for optimizing the parameters of nanomagnet synthesis [141]. Finally, the fitting of NMRD profiles to adequate theories provides information about the average

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 357

30 Theoretical fit Magnetite (experimental points) Rmax ~ C • M2s • τD

Proton relaxivity (s‒1 • mM‒1)

25

20

Low field relaxation rate: depends on the anisotropy energy

ωI • τD ~ 1 τD = r2/D

15

10 Low field dispersion: an indication of anisotropy energy

5

ωi

0 10‒2

10‒1 101 100 Proton Larmor frequency (MHz)

102

103

Fig. 13.9. NMRD profile for magnetite particles in colloidal solution. Reproduced with permission from [134].

radius (r) of the nanocrystals, their specific magnetization (Ms ), their anisotropy energy (Ea ), as well as their Néel relaxation time (τN ) [119]. The main information extracted from NMRD profiles for superparamagnetic NP suspensions is as given below. – The average radius (r): at high magnetic fields, the relaxation rate depends only on τD and the inflection point corresponds to the condition ωI ⋅ τD ∼ 1 (Fig. 13.14). According to equation (13.18), the determination of τD gives the crystal size r (good complement to HRTEM measurements). – The specific magnetization (Ms ): at high magnetic fields, Ms can be obtained from the equation Ms ≈ [(Rmax /(C ⋅ τD )]1/2 , where C is a constant and Rmax the maximal relaxation rate. – The crystal anisotropy energy (Ea ): the absence or presence of an inflection point at low magnetic field strengths (10−2 − 1) is an indication of the anisotropy energy. For crystals characterized by a high Ea compared to the thermal agitation, the low field dispersion disappears. This has been confirmed in previous work with cobalt ferrites, a high anisotropy energy material [142]. – The Néel relaxation time (τN ): the relaxation rate at very low fields (R0 ) is governed by a “zero magnetic field” correlation time τc0 , which is equal to τN if τN ≪ τD . However, this situation is often not given, and in this case τN is only reported as qualitative information.

358 | Part III New trends in sustainable development and biomedical applications 13.5.3 Relaxometric performance of MRI CAs at clinical magnetic field strengths As mentioned in the previous sections, the relaxometric ratio at clinical magnetic strengths (typically at 1.5 T) is used to classify the behavior of CAs between “positive” (i.e., r2 /r1 < 5), and “negative” ones (i.e., r2 /r1 ≫ 10). The different parameters described in the last section indicate that, at fixed magnetic field strength, the relaxometric ratio depends on many factors such as the crystal core diameter (influence Ea and τD ), the specific surface and physicochemical characteristics of NP surfaces (influence τD ; particularly critical for inner-sphere mechanisms), as well as the hydrodynamic diameter of NPs (influence τR ). Because they have the capacity to individualize the particles, can change both their hydrodynamic diameter and their surface charge, and finally because they can considerably modify the accessibility of water molecules to the nanoparticle surfaces, coatings play a major role in the modulation of r1 and r2 relaxivities. Dextran and (carboxy, carboxymethyl)dextran figure among the polymer coatings which have been successfully used to enhance both the relaxometric properties and the blood retention of USPIOs. In particular, commercial products such as Ferumoxtran-10 (AMI-227), Ferumoxytol, and Supravist are all based on iron oxide of core diameters in the range of 4–8 nm, and of hydrodynamic diameters not larger than 30 nm (Table 13.1). At 1.5 Tesla, the longitudinal relaxivities (r1 ) of those particles range between 9 and 15 mM−1 s−1 [143]. The relaxometric ratio of dextran and carboxy/carboxylmethyldextran-coated USPIOs is found between 2 and 5 (1.5 T). Finally, the blood half-lives of these products, ranging from 6 to 36 hours, figure among the longest of all iron oxide NP systems. For such reasons, and in spite of the potential instability of dextran grafting at their surface, Ferumoxtran, Feromoxytol, and Supravist particles have been widely applied to MR molecular imaging. Table 13.1 summarizes the performance of a selection of commercial and pre-clinical products, based on both superparamagnetic and paramagnetic NPs.

13.6 Biological performance of magnetic nanoparticles for MRI MNPs for MRI applications are complex pharmaceutical constructs which must navigate the body either to provide a general contrast enhancement effect, or in search of a target. They are made of at least two, if not four or five, different components (Fig. 13.1): a central magnetically active core, a stabilizing shell or coating, made of one or many types of biocompatible molecules, to which targeting ligands and additional imaging modalities are anchored. Therapeutic agents can also be embedded in the structure. MNPs must be biocompatible and should not harm the patient. The behaviour of the nanoconstruct in vivo, as well as that of each of the different components (blood retention, clearance kinetics, possible degradation resulting in metal ion leaching, polymer or drug elution, etc.), must be comprehensively investigated prior

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 359

Table 13.1. Relaxometric properties of MNPs measured at clinical MRI field strengths. Product, commercial name, coating

r2 /r1 Blood half- Use life, T1/2 (species)

Refs.

NP core diam., TEM (nm)

Hydrodyn. size, DLS (nm)

r1 (1.5T, 37∘ C)

∼5

120–180

10.1

12

2 h LI, CL (humans)

[144, 145]

SPIO, Resovist, Ferucarbotran (SHU-555A), carboxydextran

∼ 10

60

9.7

19.5

2.4–3.6 h LI, CL (humans)

[146]

USPIO, Ferumoxtran-10 (AMI-227), dextran T10

4.5

15–30

9.9

6.57

24–36 h MLNI, [102] (humans) MI, BPA, CL

USPIO, Ferumoxytol-7228, carboxlymethyldextran

6.7

30–35

15

5.93

10–14 h MI, (humans) BPA, CL

[144, 147]

USPIO, Supravist (SHU-555C), carboxydextran

3–5

21

10.7

3.55

6 h BPA, (humans) CL

[148]

USPIO, Feruglose NC100150, Clariscan, PEGylated starch111

6.43

11.9

∼ 18 (∗ at 0.5T)

n. a.

6 h BPA (humans)

[17, 149]

VSOP (iron oxide), C184 citrate

7

14

14

2.4

0.6–1.3 h BPA, (humans) CL

[150]

ESION (iron oxide), PO-PEG

3

4.77 (∗ at 3T)

6.12

> 10 min. BPA (rat)

[151]

< 40 min. BPA (mouse)

[152]

SPIO (Ferumoxides, Endorem, Feridex), (AMI-25), dextran T10

5.5 ±0.6

24 ±3

9.5 (∗ at 3T, RT)

2.97

2

9–11

14.2

1.2

– CL

NaGdF4 citrate

90 min. BPA, [34] (mouse) MRI-OI

MnO bis-phosphonatePEG dendrons

6–8

13.4–16.2

4.4

8.6

USPIO bis-phosphonatePEG Gd2 O3 PEG-diacid

< 20 min. BPA (mouse)

[76]

[118]

* Use: LI: liver imaging; CL: cellular labeling; MLNI: metastatic lymph node imaging; MI: macrophage imaging; BPA: blood pool agent; MRI-OI: MRI-optical imaging

360 | Part III New trends in sustainable development and biomedical applications to approval by the health authorities. Finally, coated and functionalized MNPs should have their relaxometric properties preserved.

13.6.1 In vivo barriers MNPs injected in vivo must overcome several biological barriers either to reach their target or to remain in the blood for a prolonged period. Most MRI CAs are administered through intravascular (i.v.) injections. MNPs immediately encounter blood, a highly ionic and heterogeneous solution containing high concentrations of organic molecules. Chemical binding and electrostatic interactions occurring between these molecules and the MNPs can lead to dramatic changes in their hydrodynamic diameter, relaxometric properties, and colloidal stability. Agglomeration and surface charge effects can also accelerate the sequestration of MNPs by the immune system, resulting in stronger and more rapid uptake by the macrophages. Depending on their molecular coating, MNP can interact more or less strongly with the extracellular matrix and, in the case of binding, this can cause the MNPs to be taken up by cells prematurely before they reach the targeted tissue [153]. The NPs must also overcome different anatomical size restrictions which limit their access to target tissues (e.g., extravasation of lymph node-targeting NPs from the blood) [154]. These size limitations are very stringent in the case of certain organs such as the brain and kidneys [155]. For instance, only NPs of sufficiently small size and appropriate physicochemical properties may pass the blood-brain barrier, a structural and metabolic barrier consisting of endothelial cells and reinforcing astrocyte cells that protect the brain [156]. In addition to biological barriers present in the extracellular space (blood vessels, lymphatic conducts), intracellular barriers may also restrict the function of several biomedical NP systems. NPs which bind at the surface of targeted cells are typically taken up by such cells through receptor-mediated endocytosis mechanisms. Upon ingestion by the cells, NPs are trapped and “trafficked” through endosome compartments where they progressively degrade by acidification. The endosomes are progressively translocated into lysosomes, compartments in which hydrolytic and enzymatic reactions metabolize or evacuate macromolecules and NP debris [157]. Different strategies, more or less complex or potentially cytotoxic, have been developed to facilitate the escape of NPs from the endosomes [158]. Overall, each of these biological obstacles illustrates the different levels of complexity which must be addressed when designing optimal MNPs as CAs for blood pool, cell labeling and tracking, targeted imaging, and/or drug delivery procedures.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

|

361

13.6.2 Impact of nanoparticle size and surface on colloidal stability and blood retention The ability of MNPs to remain in the blood for prolonged periods and to pass biological barriers is largely related to their main physicochemical characteristics: hydrodynamic size, surface charge, and type of coating [155, 159, 160]. The retention of NPs in the blood vessels, the mechanisms of NP clearance, and the permeability of NPs from the vasculature all strongly depend on these three parameters [161–163]. In particular, NP charge and hydrophobicity can affect NP biodistribution by limiting or enhancing interactions of NPs with the adaptive immune system, plasma proteins, extracellular matrices, and non-targeted cells [153]. Specifically, hydrophobic and strongly charged NPs have short circulation times due to the adsorption of plasma proteins such as opsonins, which are responsible for the recognition of MNPs by the reticuloendothelial system (RES). As a consequence, the macrophages recognize opsonins and remove MNPs from the bloodstream [160]. Most particles end up in the liver and spleen, where they eventually degrade. The sequestration of MNPs by the RES can be advantageously used in the diagnosis of hepatic lesions (e.g., liver cancer). However, particles which are too rapidly sequestered and removed from the blood stream cannot optimally fulfill applications such as targeted imaging (e.g., for cancer, atherosclerotic plaque, and Alzheimer’s disease), and drug delivery to specific organs. Hydrodynamic size strongly affects the clearance of NPs from vascular circulation [37, 47–51]. In general, small NPs (< 20 nm) can be more efficiently excreted by the kidneys [164–166], whereas large NPs (> 50 nm) are mostly found in the liver and spleen. However, strongly charged particles such as citrate-capped iron oxide and paramagnetic nanocrystals follow the RES route and are cleared in the liver [98]. As a consequence, a minor fraction of them make their way through the urinary tract [150]. Most endothelial barriers allow nanoparticles of < 150 nm hydrodynamic diameter to pass; however, other barriers, such as the BBB, are far more restrictive. Because the liver can metabolize very large amounts of iron, USPIOs and SPIOs are generally well-tolerated products. As a matter of fact, Ferumoxides, Feromoxtran, Ferucarbotran, Supravist, Clariscan, and a few other iron oxide-based NP systems have either been commercialized, or at least investigated in clinics (see Table 13.1). In the case of paramagnetic nanoparticles not based on Gd or Mn chelates (MnO, Gd2 O3 , NaGdF4 ), the toxicity risk represented by the massive injection of Mn and Gd-based crystals potentially leaching toxic Mn2+ and Gd3+ ions greatly restricts their eventual transfer to clinical applications. However, emerging coating strategies, in particular those based on silane-PEG or phosphate/phosphonate-PEG (Fig. 13.10), have been shown to promote the rapid excretion of paramagnetic nanocrystals via the gastrointestinal and urinairy routes [118, 166]. It is still too soon to predict whether these strategies might allow the clinical use of paramagnetic nanocrystals in the future. However, they facilitate the development of more complex and specific “positive” CAs for molecular and cellular pre-clinical research.

362 | Part III New trends in sustainable development and biomedical applications

t=0

t=10 min

t=1 h

(a)

(b)

(c) Fig. 13.10. Injections of MnO NPs PEGylated with phosphonate dendrons, and scanned in MRI; (a) strong evidence of CA elimination is found in the gall bladder (hepatobiliary way) after 1 h; (b) blood signal-enhancement persists at least 20 minutes after injection, while evidence of CA clearance is found in the kidneys (b), and in the urine (c). Adapted with permission from [118].

In order to be truly efficient for targeted molecular imaging applications, it is necessary for the MNP CAs to reach blood half-lives of many hours (such as for Feromuxtran, Ferumoxytol, and Feruglose; Table 13.1). In this way, CAs have more chances to effectively bind to molecular epitopes and receptors expressed at the surface of vascular cells, which the targeted CAs are designed to reach. Surface modification with the hydrophilic molecules dextran and polyethylene glycol (PEG) have been shown to reduce opsonization, leading to prolonged NP circulation times (Table 13.1, and [167]).

13.6.3 Directing nanoparticles in vivo The specificity of NPs for selected tissues is critical in MRI-based diagnostic imaging [16, 74, 75]. NPs can be engineered to have an affinity for target tissues through passive, active, and magnetic targeting approaches. Passive targeting uses the predetermined physicochemical properties (size and surface charge) of a given NP to specifically migrate to a given tissue region. In particular, the enhanced permeation and retention effect (EPR) can be used to target solid tumour tissue [76]. In an attempt to grow rapidly, tumor cells stimulate the production of new blood vessels (termed “neovasculature”). Such vessels are poorly organized and have leaky fenestrations. This enables extravasation of NPs out of the vasculature into the tumor tissue [77, 78]. Then, because lymphatic drainage is relatively inefficient in solid tumors, NPs tend to accumulate at this site [79, 80]. However, the EPR effect is limited to specific metastatic solid tumors and the successful implementation of CA systems relying on this effect is dependent upon a number of factors including the degree of capillary disorder, blood flow, and lymphatic drainage rate. As a result, it is not possible to base a therapeutic treatment or an MR diagnostic on the EPR effect.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 363

Passive targeting works exclusively for very specific biomedical applications, such as for solid tumors of leaky vasculature, since it does not guarantee their internalization by targeted cells. In order to achieve cellular internalization it is necessary to modify NPs with molecular targeting ligands to provide an “active” approach. NPs functionalized with molecules which specifically bind to molecular epitopes expressed at the surface of cancer cells or other disease tissues allow a more efficient selective concentration of them in the tissue to be diagnosed or treated (81, 82). In particular, a number of iron oxide NP systems have been developed and tested in vivo with varying success. Among targeting molecules which have been used so far to achieve active targeting with iron oxide NPs figure small organic molecules [81, 83, 84], peptides [71, 85–88], proteins [89], antibodies [90–92], and aptamers [93–95]. In addition to engineering NPs for tissue targeting, external magnetic fields can be used to assist the diffusion of MNPs to given organs. This strategy is referred to as “magnetic targeting” [18, 104]. It consists of focusing high fields, high gradients, or strong rare earth magnets on the target organ or biological site. This is a good strategy to accumulate high-susceptibility MNPs at specific sites, to conduct hyperthermia treatments [100]. This technique was successfully implemented in a clinical trial to deliver the chemotherapeutic doxorubicin to hepatocarcinoma cells [105]. The effectiveness of magnetic targeting is unfortunately limited to target tissues located close to the surface of the body (rapid loss of magnetic field strength away from the magnets).

13.6.4 Toxicity In order to generate significant signal-enhancement effects, NPs used as MRI CAs must be injected at relatively high doses compared to common tracers used in nuclear medicine (e.g., PET). Therefore NPs must be demonstrated safe for cells and different tissues, in particular when high quantities find their way to critical organs (liver, kidneys, spleen). The impact of MNPs on the proliferation and viability of different cell types is always demonstrated in vitro prior to injection in vivo [32, 76, 112, 168]. Depending on the concentration, type of particle, surface charge, and class of coating ligand, the presence of a high concentration of nanoparticles in the vicinity of cells can have a transitory effect on their cell division cycle, and sometimes influence their viability. Apoptosis measurements can also be performed at high doses of MNPs to evaluate the damage induced by high concentrations of MNPs to cells. The cells selected for in vitro tests should ideally represent either the tissues expected to receive the highest concentration of MNPs, or the cells most likely to be affected by MNPs. As an example, in order to assess the biocompatibility of ultra-small MNPs for targeted vascular imaging, epithelial vascular, kidney, and hepatic cell lines could be used. Finally, MNPs are not intact when they are excreted from the body. They are confronted with different biological mechanisms which impact on their integrity. Validating the clinical use of nanosystems is challenging, as the toxicity of both the intact

364 | Part III New trends in sustainable development and biomedical applications products and their different components must be rigorously studied. For instance, the potential leaching of metal ions from MNPs in different organs must be carefully and comprehensively quantified. Also, the impact of different NP coatings and their degradation products on specific cells and organs in which they could potentially accumulate is also an important aspect of the toxicity evaluation of MNPs. The nature of the degradation by-products must also be addressed [169]. Nanotoxicology is an emerging and expanding research area and a selection of works specifically address this topic [155, 170, 171].

13.7 Summary The recent advances in synthesis and characterization of MNPs as MRI CAs has allowed the emergence of a variety of new biomedical applications: stem cell labeling and tracking in vivo, imaging-assisted drug and gene delivery, molecular targeting of chronic diseases such as atherosclerosis and cancer. Because of their very strong impact on the transverse relaxivity, superparamagnetic NPs in particular have been used in a variety of clinical applications (liver, spleen, lymph node imaging). In this chapter the basic principles of MNPs for MRI applications were reviewed. The main parameters and conditions guiding their optimal design, use, and performance in biological applications were presented. Because the relaxometric potential of MNPs is very dependent on their size, fine particles of narrow size distributions are developed and then coated with biocompatible molecules which provide enhanced colloidal stability in physiological environments through electrostatic and steric repulsion mechanisms. The biocompatibility of nanoparticles must be assessed in vitro, after careful measurement of their magnetic and relaxometric properties. Finally, the biological kinetics (blood retention, organ uptake, clearance) and contrast-enhancement effects of each new nanoparticulate system must be carefully studied in vivo. NPs are complex systems, and the medical regulatory authorities enforce very strict requirements on design, production reproducibility, potential toxicity, and pharmacokinetics performance of such injectable products. Finally, the expansion of hybrid imaging modalities (MRI/PET, MRI/luminescence, MRI/SPECT, MRI/echography), calls for the development of multifunctional and increasingly complex imaging tracers. For instance, stem cell therapies could be conducted advantageously using MRI/PET, which would enable more sensitive detection and quantitation of areas of implanted stem cells. The delivery of targeted drugs through nanovectors could also be performed under MRI/PET guidance, to provide quantitative measurements of the residual concentration of drug delivery vehicles still present in the blood, then accumulating into specific organs.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 365

References [1]

[2] [3] [4] [5] [6] [7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15] [16]

[17]

Morawski AM, Winter PM, Crowder KC, Caruthers SD, Fuhrhop RW, Scott MJ, et al. Targeted nanoparticles for quantitative imaging of sparse molecular epitopes with MRI. Magnet Reson Med. 51 (2004) 480–6. Tu C, Louie AY. Nanoformulations for molecular MRI. Wiley interdisciplinary reviews Nanomedicine and nanobiotechnology. 4 (2012) 448–57. Gallo J, Long NJ, Aboagye EO. Magnetic nanoparticles as contrast agents in the diagnosis and treatment of cancer. Chemical Society reviews. 42 (2013) 7816–33. Merbach AS, Helm L, Tóth E. The Chemistry of Contrast Agents in Medical Magnetic Resonance Imaging. 2nd ed. Hoboken: Wiley; 2013. Caravan P. Strategies for increasing the sensitivity of gadolinium based MRI contrast agents. Chemical Society reviews. 35 (2006) 512–23. Caravan P, Ellison JJ, McMurry TJ, Lauffer RB. Gadolinium(III) Chelates as MRI Contrast Agents: Structure, Dynamics, and Applications. Chemical reviews. 99 (1999) 2293–352. Rohrer M, Bauer H, Mintorovitch J, Requardt M, Weinmann HJ. Comparison of magnetic properties of MRI contrast media solutions at different magnetic field strengths. Invest Radiol. 40 (2005) 715–24. Ros PR, Freeny PC, Harms SE, Seltzer SE, Davis PL, Chan TW, et al. Hepatic MR imaging with ferumoxides: A multicenter clinical trial of the safety and efficacy in the detection of focal hepatic lesions. Radiology. 196 (1995) 481–8. Arnold P, Ward J, Wilson D, Guthrie JA, Robinson PJ. Superparamagnetic iron oxide (SPIO) enhancement in the cirrhotic liver: A comparison of two doses of ferumoxides in patients with advanced disease. Magnetic Resonance Imaging. 21 (2003) 695–700. Kehagias DT, Gouliamos AD, Smyrniotis V, Vlahos LJ. Diagnostic efficacy and safety of MRI of the liver with superparamagnetic iron oxide particles (SH U 555 A). Journal of Magnetic Resonance Imaging. 14 (2001) 595–601. Kreft BP, Tanimoto A, Leffler S, Finn JP, Oksendal AN, Stark DD. Contrast-enhanced MR imaging of diffuse and focal splenic disease with use of magnetic starch microspheres. Journal of magnetic resonance imaging : JMRI. 4 (1994) 373–9. Anzai Y, Blackwell KE, Hirschowitz SL, Rogers JW, Sato Y, Yuh WTC, et al. Initial clinical experience with dextran-coated superparamagnetic iron oxide for detection of lymph node metastases in patients with head and neck cancer. Radiology. 192 (1994) 709–15. Bjørnerud A, Johansson L. The utility of superparamagnetic contrast agents in MRI: Theoretical consideration and applications in the cardiovascular system. NMR in Biomedicine. 17 (2004) 465–77. Bjerner T, Johansson L, Wikström G, Ericsson A, Briley-Scebo K, Bjørnerud A, et al. In and ex vivo MR evaluation of acute myocardial ischemia in pigs by determining R1 in steady state after the administration of the intravascular contrast agent NC100150 injection. Invest Radiol. 39 (2004) 479–86. Johansson LO, Bjerner T, Bjornerud A, Ahlstrom H, Tarlo KS, Lorenz CH. Utility of NC100150 injection in cardiac MRI. Academic Radiology. 9 (2002) S79–S81. Kellar KE, Fujii DK, Gunther WHH, Briley-Sabo K, Bjornerod A, Spiller M, et al. Important considerations in the design of iron oxide nanoparticles as contrast agents for T1-weighted MRI and MRA. Academic Radiology. 9 (2002) S34–S7. Kellar KE, Fujii DK, Gunther WHH, Briley-Sæbø K, Bjørnerud A, Spiller M, et al. NC 100150 injection, a preparation of optimized iron oxide nanoparticles for positive-contrast MR angiography. Journal of Magnetic Resonance Imaging. 11 (2000) 488–94.

366 | Part III New trends in sustainable development and biomedical applications [18] [19]

[20]

[21]

[22]

[23]

[24]

[25]

[26] [27] [28]

[29] [30]

[31]

[32]

[33]

[34]

[35]

Sosnovik DE. Molecular imaging in cardiovascular magnetic resonance imaging: Current perspective and future potential. Topics in Magnetic Resonance Imaging. 19 (2008) 59–68. Shen T, Weissleder R, Papisov M, Bogdanov Jr A, Brady TJ. Monocrystalline iron oxide nanocompounds (MION): Physicochemical properties. Magnet Reson Med. 29 (1993) 599–604. Wunderbaldinger P, Josephson L, Weissleder R. Crosslinked iron oxides (CLIO): A new platform for the development of targeted MR contrast agents. Academic Radiology. 9 (2002) S304–S6. Nahrendorf M, Jaffer FA, Kelly KA, Sosnovik DE, Aikawa E, Libby P, et al. Noninvasive vascular cell adhesion molecule-1 imaging identifies inflammatory activation of cells in atherosclerosis. Circulation. 114 (2006) 1504–11. Kelly KA, Allport JR, Tsourkas A, Shinde-Patil VR, Josephson L, Weissleder R. Detection of vascular adhesion molecule-1 expression using a novel multimodal nanoparticle. Circulation Research. 96 (2005) 327–36. Jaffer FA, Nahrendorf M, Sosnovik D, Kelly KA, Aikawa E, Weissleder R. Cellular imaging of inflammation in atherosclerosis using magnetofluorescent nanomaterials. Molecular Imaging. 5 (2006) 85–92. Zhao M, Beauregard DA, Loizou L, Davletov B, Brindle KM. Non-invasive detection of apoptosis using magnetic resonance imaging and a targeted contrast agent. Nature medicine. 7 (2001) 1241–4. Wadghiri YZ, Sigurdsson EM, Sadowski M, Elliott JI, Li Y, Scholtzova H, et al. Detection of Alzheimer’s amyloid in transgenic mice using magnetic resonance microimaging. Magnet Reson Med. 50 (2003) 293–302. Na HB, Song IC, Hyeon T. Inorganic Nanoparticles for MRI Contrast Agents. Adv Mater. 21 (2009) 2133–48. Na HB, Hyeon T. Nanostructured T1 MRI contrast agents. J Mater Chem. 19 (2009) 6267–73. Bridot JL, Faure AC, Laurent S, Riviere C, Billotey C, Hiba B, et al. Hybrid gadolinium oxide nanoparticles: multimodal contrast agents for in vivo imaging. Journal of the American Chemical Society. 129 ( 2007) 5076–84. Engström M, Klasson A, Pedersen H, Vahlberg C, Käll PO, Uvdal K. High proton relaxivity for gadolinium oxide nanoparticles. Magma. 19 (2006) 180–6. Ahren M, Selegard L, Klasson A, Soderlind F, Abrikossova N, Skoglund C, et al. Synthesis and Characterization of PEGylated Gd2O3 Nanoparticles for MRI Contrast Enhancement. Langmuir. 26 (2010) 5753–62. Park JY, Baek MJ, Choi ES, Woo S, Kim JH, Kim TJ, et al. Paramagnetic Ultrasmall Gadolinium Oxide Nanoparticles as Advanced T-1 MR1 Contrast Agent: Account for Large Longitudinal Relaxivity, Optimal Particle Diameter, and In Vivo T-1 MR Images. Acs Nano. 3 (2009) 3663–9. Faucher L, Guay-Bégin AA, Lagueux J, Côté MF, Petitclerc E, Fortin MA. Ultra-small gadolinium oxide nanoparticles to image brain cancer cells in vivo with MRI. Contrast Media and Molecular Imaging. 6 (2011) 209–18. Zhou J, Yu MX, Sun Y, Zhang XZ, Zhu XJ, Wu ZH, et al. Fluorine-18-labeled Gd3+/Yb3+/Er3+ co-doped NaYF4 nanophosphors for multimodality PET/MR/UCL imaging. Biomaterials. 32 (2011) 1148–56. Naccache R, Chevallier P, Lagueux J, Gossuin Y, Laurent S, Vander Elst L, et al. High relaxivities and strong vascular signal enhancement for NaGdF4 nanoparticles designed for dual MR/optical imaging. Advanced Healthcare Materials. 2 (2013) 1478–88. Cheung ENM, Alvares RDA, Oakden W, Chaudhary R, Hill ML, Pichaandi J, et al. PolymerStabilized Lanthanide Fluoride Nanoparticle Aggregates as Contrast Agents for Magnetic Resonance Imaging and Computed Tomography. Chemistry of Materials. 22 (2010) 4728–39.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

[36]

[37]

[38]

[39]

[40] [41] [42]

[43]

[44]

[45] [46] [47]

[48] [49] [50] [51]

[52] [53]

[54]

| 367

Chen GY, Ohulchanskyy TY, Law WC, Agren H, Prasad PN. Monodisperse NaYbF4: Tm3+/NaGdF4 core/shell nanocrystals with near-infrared to near-infrared upconversion photoluminescence and magnetic resonance properties. Nanoscale. 3 (2011) 2003–8. Na HB, Lee JH, An K, Park YI, Park M, Lee IS, et al. Development of a T1 contrast agent for magnetic resonance imaging using MnO nanoparticles. Angewandte Chemie – International Edition. 46 (2007) 5397–401. Schladt TD, Schneider K, Shukoor MI, Natalio F, Bauer H, Tahir MN, et al. Highly soluble multifunctional MnO nanoparticles for simultaneous optical and MRI imaging and cancer treatment using photodynamic therapy. J Mater Chem. 20 (2010) 8297–304. Letourneau M, Tremblay M, Faucher L, Rojas D, Chevallier P, Gossuin Y, et al. MnO-Labeled Cells: Positive Contrast Enhancement in MRI. The journal of physical chemistry B. 116 (2012) 13228–38. Hashemi RH, Bradley WG, Lisanti CJ, Ovid Technologies Inc. MRI the basics. 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2010. Bushberg JT. The essential physics of medical imaging. 3rd ed. Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins; 2012. Morawski AM, Winter PM, Yu X, Fuhrhop RW, Scott MJ, Hockett F, et al. Quantitative “magnetic resonance immunohistochemistry” with ligand-targeted F-19 nanoparticles. Magnet Reson Med. 52 (2004) 1255–62. Gossuin Y, Gillis P, Hocq A, Vuong QL, Roch A. Magnetic resonance relaxation properties of superparamagnetic particles. Wiley interdisciplinary reviews Nanomedicine and nanobiotechnology. 1 (2009) 299–310. Laurent S, Forge D, Port M, Roch A, Robic C, Vander Elst L, et al. Magnetic iron oxide nanoparticles: synthesis, stabilization, vectorization, physicochemical characterizations, and biological applications. Chemical reviews. 108 (2008) 2064–110. Sun C, Lee JS, Zhang M. Magnetic nanoparticles in MR imaging and drug delivery. Advanced drug delivery reviews. 60 (2008) 1252–65. Gupta AK, Gupta M. Synthesis and surface engineering of iron oxide nanoparticles for biomedical applications. Biomaterials. 26 (2005) 3995–4021. Gupta AK, Naregalkar RR, Vaidya VD, Gupta M. Recent advances on surface engineering of magnetic iron oxide nanoparticles and their biomedical applications. Nanomedicine (Lond). 2 (2007) 23–39. Ravi Kumar MNV. Handbook of particulate drug delivery. Stevenson Ranch, Ca.: American Scientific Publishers; 2008. Lu AH, Salabas EL, Schuth F. Magnetic nanoparticles: synthesis, protection, functionalization, and application. Angewandte Chemie. 46 (2007) 1222–44. Puntes VF, Krishnan KM, Alivisatos AP. Colloidal nanocrystal shape and size control: the case of cobalt. Science. 291 (2001) 2115–7. Shevchenko EV, Talapin DV, Rogach AL, Kornowski A, Haase M, Weller H. Colloidal synthesis and self-assembly of CoPt(3) nanocrystals. Journal of the American Chemical Society. 124 (2002) 11480–5. Sun S, Murray CB, Weller D, Folks L, Moser A. Monodisperse FePt nanoparticles and ferromagnetic FePt nanocrystal superlattices. Science. 287 (2002) 1989–92. Di Marco M, Sadun C, Port M, Guilbert I, Couvreur P, Dubernet C. Physicochemical characterization of ultrasmall superparamagnetic iron oxide particles (USPIO) for biomedical application as MRI contrast agents. International journal of nanomedicine. 2 (2007) 609–22. Sun S, Zeng H. Size-controlled synthesis of magnetite nanoparticles. Journal of the American Chemical Society. 124 (2002) 8204–5.

368 | Part III New trends in sustainable development and biomedical applications [55] [56] [57]

[58] [59] [60] [61] [62] [63] [64] [65]

[66] [67]

[68]

[69] [70] [71]

[72]

[73] [74] [75]

Lee JH, Huh YM, Jun YW, Seo JW, Jang JT, Song HT, et al. Artificially engineered magnetic nanoparticles for ultra-sensitive molecular imaging. Nature medicine. 13 (2007) 95–9. Park J, An K, Hwang Y, Park JG, Noh HJ, Kim JY, et al. Ultra-large-scale syntheses of monodisperse nanocrystals. Nature materials. 3 (2004) 891–5. Weissleder R, Stark DD, Engelstad BL, Bacon BR, Compton CC, White DL, et al. Superparamagnetic iron oxide: pharmacokinetics and toxicity. AJR American journal of roentgenology. 152 (1989) 167–73. Stephen ZR, Kievit FM, Zhang M. Magnetite nanoparticles for medical MR imaging. Materials Today. 14 (2011) 330–8. Wu W, He Q, Jiang C. Magnetic iron oxide nanoparticles: synthesis and surface functionalization strategies. Nanoscale research letters. 3 (2008) 397–415. Schladt TD, Schneider K, Schild H, Tremel W. Synthesis and bio-functionalization of magnetic nanoparticles for medical diagnosis and treatment. Dalton Transactions. 40 (2011) 6315–43. Rui H, Xing R, Xu Z, Hou Y, Goo S, Sun S. Synthesis, functionalization, and biomedical applications of multifunctional magnetic nanoparticles. Adv Mater. 22 (2010) 2729–42. Jolivet JP, Chanéac C, Tronc E. Iron oxide chemistry. From molecular clusters to extended solid networks. Chemical Communications. 10 (2004) 481–7. Bee A, Massart R, Neveu S. Synthesis of very fine maghemite particles. Journal of Magnetism and Magnetic Materials. 149 (1995) 6–9. Wormuth K. Superparamagnetic latex via inverse emulsion polymerization. Journal of Colloid and Interface Science. 241 (2001) 366–77. Si S, Kotal A, Mandal TK, Giri S, Nakamura H, Kohara T. Size-controlled synthesis of magnetite nanoparticles in the presence of polyelectrolytes. Chemistry of Materials. 16 (2004) 3489–96. Wan S, Huang J, Yan H, Liu K. Size-controlled preparation of magnetite nanoparticles in the presence of graft copolymers. J Mater Chem. 16 (2006) 298–303. Gonzales M, Krishnan KM. Synthesis of magnetoliposomes with monodisperse iron oxide nanocrystal cores for hyperthermia. Journal of Magnetism and Magnetic Materials. 293 (2005) 265–70. Giri J, Guha Thakurta S, Bellare J, Kumar Nigam A, Bahadur D. Preparation and characterization of phospholipid stabilized uniform sized magnetite nanoparticles. Journal of Magnetism and Magnetic Materials. 293 (2005) 62–8. Sun YK, Ma M, Zhang Y, Gu N. Synthesis of nanometer-size maghemite particles from magnetite. Colloids and Surfaces A: Physicochemical and Engineering Aspects. 245 (2004) 15–19. De Cuyper M, Joniau M. Magnetoliposomes. Formation and structural characterization. European Biophysics Journal. 15 (1988) 311–9. Fievet F, Lagier JP, Blin B, Beaudoin B, Figlarz M. Homogeneous and heterogeneous nucleations in the polyol process for the preparation of micron and submicron size metal particles. Solid State Ionics. 32–33 (1989) 198–205. Tzitzios VK, Petridis D, Zafiropoulou I, Hadjipanayis G, Niarchos D. Synthesis and characterization of L10 FePt nanoparticles from Pt-Fe3O4 core-shell nanoparticles. Journal of Magnetism and Magnetic Materials. 294 (2005) e95–e8. Sra AK, Ewers TD, Schaak RE. Direct solution synthesis of intermetallic AuCu and AuCu3 nanocrystals and nanowire networks. Chemistry of Materials. 17 (2005) 758–66. Joseyphus RJ, Kodama D, Matsumoto T, Sato Y, Jeyadevan B, Tohji K. Role of polyol in the synthesis of Fe particles. Journal of Magnetism and Magnetic Materials. 310 (2007) 2393–5. Hu F, MacRenaris KW, Waters EA, Liang T, Schultz-Sikma EA, Eckermann AL, et al. Ultrasmall, water-soluble magnetite nanoparticles with high relaxivity for magnetic resonance imaging. J Phys Chem C. 113 (2009) 20855–60.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

[76]

[77]

[78]

[79]

[80]

[81]

[82]

[83] [84]

[85]

[86] [87]

[88]

[89]

[90]

[91]

[92]

| 369

Faucher L, Tremblay M, Lagueux J, Gossuin Y, Fortin MA. Rapid synthesis of PEGylated ultrasmall gadolinium oxide nanoparticles for cell labeling and tracking with MRI. ACS Applied Materials and Interfaces. 4 (2012) 4506–15. Faucher L, Guay-Begin AA, Lagueux J, Cote MF, Petitclerc E, Fortin MA. Ultra-small gadolinium oxide nanoparticles to image brain cancer cells in vivo with MRI. Contrast media & molecular imaging. 6 (2011) 209–18. Fizet J, Rivière C, Bridot JL, Charvet N, Louis C, Billotey C, et al. Multi-luminescent hybrid gadolinium oxide nanoparticles as potential cell labeling. Journal of Nanoscience and Nanotechnology. 9 (2009) 5717–25. Ahrén M, Selegård L, Klasson A, Söderlind F, Abrikossova N, Skoglund C, et al. Synthesis and characterization of PEGylated Gd2O3 nanoparticles for MRI contrast enhancement. Langmuir. 26 (2010) 5753–62. Shi Z, Neoh KG, Kang ET, Shuter B, Wang SC. Bifunctional Eu3+-doped Gd2O3 nanoparticles as a luminescent and T1 contrast agent for stem cell labeling. Contrast Media and Molecular Imaging. 5 (2010) 105–11. Klasson A, Ahrén M, Hellqvist E, Söderlind F, Rosén A, Käll PO, et al. Positive MRI contrast enhancement in THP-1 cells with Gd2O3 nanoparticles. Contrast Media and Molecular Imaging. 3 (2008) 106–11. Petoral Jr RM, Söderlind F, Klasson A, Suska A, Fortin MA, Abrikossova N, et al. Synthesis and characterization of Tb3+-doped Gd 2O3 nanocrystals: A bifunctional material with combined fluorescent labeling and MRI contrast agent properties. J Phys Chem C. 113 (2009) 6913–20. Bridot JL, Dayde D, Rivière C, Mandon C, Billotey C, Lerondel S, et al. Hybrid gadolinium oxide nanoparticles combining imaging and therapy. J Mater Chem. 19 ( 2009) 2328–35. Faucher L, Gossuin Y, Hocq A, Fortin MA. Impact of agglomeration on the relaxometric properties of paramagnetic ultra-small gadolinium oxide nanoparticles. Nanotechnology. 22(29) (2011) 295103. Bazzi R, Flores-Gonzalez MA, Louis C, Lebbou K, Dujardin C, Brenier A, et al. Synthesis and luminescent properties of sub-5-nm lanthanide oxides nanoparticles. J Lumin. 102 (2003) 445–50. Söderlind F, Pedersen H, Petoral RM, Käll PO, Uvdal K. Synthesis and characterisation of Gd2O3 nanocrystals functionalised by organic acids. J Colloid Interf Sci. 288 (2005) 140–8. Park JY, Choi ES, Baek MJ, Lee GH, Woo S, Chang Y. Water-Soluble Ultra Small Paramagnetic or Superparamagnetic Metal Oxide Nanoparticles for Molecular MR Imaging. European Journal of Inorganic Chemistry. (2009) 2477–81. Faucher L, Tremblay M, Lagueux J, Gossuin Y, Fortin MA. Rapid synthesis of PEGylated ultrasmall gadolinium oxide nanoparticles for cell labeling and tracking with MRI. ACS Appl Mater Interfaces. 4 (2012) 4506–15. Naccache R, Vetrone F, Mahalingam V, Cuccia LA, Capobianco JA. Controlled Synthesis and Water Dispersibility of Hexagonal Phase NaGdF4:Ho3+/Yb3+ Nanoparticles. Chemistry of Materials. 21 (2009) 717–23. Vetrone F, Naccache R, Mahalingam V, Morgan CG, Capobianco JA. The Active-Core/ActiveShell Approach: A Strategy to Enhance the Upconversion Luminescence in Lanthanide-Doped Nanoparticles. Adv Funct Mater. 19 (2009) 2924–9. Wong HT, Vetrone F, Naccache R, Chan HLW, Hao JH, Capobianco JA. Water dispersible ultrasmall multifunctional KGdF4:Tm3+, Yb3+ nanoparticles with near-infrared to near-infrared upconversion. J Mater Chem. 21 (2011) 16589–96. Morales MA, Skomski R, Fritz S, Shelburne G, Shield JE, Yin M, et al. Surface anisotropy and magnetic freezing of MnO nanoparticles. Phys Rev B. (2007) 75.

370 | Part III New trends in sustainable development and biomedical applications [93]

[94]

[95] [96] [97]

[98]

[99] [100]

[101] [102]

[103] [104] [105] [106] [107] [108]

[109]

[110]

[111]

Na HB, Lee J, H., An K, Park YI, Park M, Lee IS, et al. Development of a T1 Contrast Agent for Magnetic Resonance Imaging Using MnO Nanoparticles. Angewandte Chemie Int Ed. 46 (2007) 5397–401. Létourneau M, Tremblay M, Faucher L, Rojas D, Chevallier P, Gossuin Y, et al. MnO-labeled cells: Positive contrast enhancement in MRI. Journal of Physical Chemistry B. 116 (2012) 13228–38. Park J, An KJ, Hwang YS, Park JG, Noh HJ, Kim JY, et al. Ultra-large-scale syntheses of monodisperse nanocrystals. Nature materials. 3 (2004) 891–5. Mornet S, Vasseur S, Grasset F, Duguet E. Magnetic nanoparticle design for medical diagnosis and therapy. J Mater Chem. 14 (2004) 2161–75. Sahoo Y, Goodarzi A, Swihart MT, Ohulchanskyy TY, Kaur N, Furlani EP, et al. Aqueous ferrofluid of magnetite nanoparticles: Fluorescence labeling and magnetophoretic control. Journal of Physical Chemistry B. 109 (2005) 3879–85. Wagner S, Schnorr J, Pilgrimm H, Hamm B, Taupitz M. Monomer-coated very small superparamagnetic iron oxide particles as contrast medium for magnetic resonance imaging Preclinical in vivo characterization. Invest Radiol. 37 (2002) 167–77. Liu C, Huang PM. Atomic force microscopy and surface characteristics of iron oxides formed in citrate solutions. Soil Science Society of America Journal. 63 (1999) 65–72. Laurent S, Forge D, Port M, Roch A, Robic C, Elst LV, et al. Magnetic iron oxide nanoparticles: Synthesis, stabilization, vectorization, physicochemical characterizations, and biological applications. Chemical reviews. 108 (2008) 2064–110. Daou TJ, Pourroy G, Greneche JM, Bertin A, Felder-Flesch D, Begin-Colin S. Water soluble dendronized iron oxide nanoparticles. Dalton Transactions. 0 (2009) 4442–9. McLachlan SJ, Morris MR, Lucas MA, Fisco RA, Eakins MN, Fowler DR, et al. Phase I clinical evaluation of a new iron oxide MR contrast agent. Journal of magnetic resonance imaging : JMRI. 4 (1994) 301–7. Bulte JWM, Kraitchman DL. Iron oxide MR contrast agents for molecular and cellular imaging. NMR in Biomedicine. 17 (2004) 484–99. Corot C, Robert P, Idée JM, Port M. Recent advances in iron oxide nanocrystal technology for medical imaging. Advanced drug delivery reviews. 58 (2006) 1471–504. Paul KG, Frigo TB, Groman JY, Groman EV. Synthesis of ultrasmall superparamagnetic iron oxides using reduced polysaccharides. Bioconjug Chem. 15 (2004) 394–401. Tiefenauer LX, Tschirky A, Kühne G, Andres RY. In vivo evaluation of magnetite nanoparticles for use as a tumor contrast agent in MRI. Magnetic Resonance Imaging. 14 (1996) 391–402. Moghimi SM, Hunter AC, Murray JC. Long-circulating and target-specific nanoparticles: Theory to practice. Pharmacological Reviews. 53 (2001) 283–318. Papisov MI, Bogdanov Jr A, Schaffer B, Nossiff N, Shen T, Weissleder R, et al. Colloidal magnetic resonance contrast agents: effect of particle surface on biodistribution. Journal of Magnetism and Magnetic Materials. 122 (1993) 383–6. Fortin MA, Petoral RM, Söderlind F, Klasson A, Engström M, Veres T, et al. Polyethylene glycol-covered ultra-small Gd2O3 nanoparticles for positive contrast at 1.5 T magnetic resonance clinical scanning. Nanotechnology. 18 (2007) 395501 (1–9). Faure AC, Dufort S, Josserand V, Perriat P, Coll JL, Roux S, et al. Control of the in vivo Biodistribution of Hybrid Nanoparticles with Different Poly(ethylene glycol) Coatings. Small. 5 (2009) 2565–75. Shi ZL, Neoh KG, Kang ET, Shuter B, Wang SC. Bifunctional Eu3+-doped Gd2O3 nanoparticles as a luminescent and T-1 contrast agent for stem cell labeling. Contrast Media & Molecular Imaging. 5 (2010) 105–11.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 371

[112] Letourneau M, Tremblay M, Faucher L, Rojas D, Chevallier P, Gossuin Y, et al. MnO-Labeled Cells: Positive Contrast Enhancement in MRI. Journal of Physical Chemistry B. 116 (2012) 13228–38. [113] Klasson A, Ahren M, Hällquist E, Rosén A, Käll PO, Uvdal K, et al. Positive MRI contrast enhancement in THP-1 cells with Gd2O3 nanoparticles. ContrastMedia and Molecular Imaging. 3 (2008) 106–11. [114] Faucher L, M.A. F, M. T, J. L, Lacroix S. Ultra-small nanoclusters of GdOx : a new, efficient contrast agent for in vivo cell tracking studies in T1-w. MRI. In: Society WMI, editor. World Molecular Imaging Congress. San Diego, 2011. [115] Bridot JL, Faure AC, Laurent S, Riviere C, Billotey C, Hiba B, et al. Hybrid gadolinium oxide nanoparticles: Multimodal contrast agents for in vivo imaging. Journal of the American Chemical Society. 129 (2007) 5076–84. [116] Guay-Begin AA, Chevallier P, Faucher L, Turgeon S, Fortin MA. Surface Modification of Gadolinium Oxide Thin Films and Nanoparticles using Polyethylene glycol-phosphate. Langmuir : the ACS journal of surfaces and colloids. 2011. [117] Lamanna G, Kueny-Stotz M, Mamlouk-Chaouachi H, Ghobril C, Basly B, Bertin A, et al. Dendronized iron oxide nanoparticles for multimodal imaging. Biomaterials. 32 (2011) 8562–73. [118] Chevallier P, A. W, A. G, I. V, J. L, S. B-C, et al. Tailored biological retention and efficient clearance of pegylated ultra-small MnO nanoparticles as positive MRI contrast agents for molecular imaging. Journal of Materials Chemistry B. 2014;(accepted). [119] Muller RN, Vander Elst L, Roch A, Peters JA, Csajbok E, Gillis P, et al. Relaxation by metalcontaining nanosystems. (2006) 239–92. [120] Fortin MA, Petoral Jr RM, Söoderlind F, Klasson A, Engströom M, Veres T, et al. Polyethylene glycol-covered ultra-small Gd2O3 nanoparticles for positive contrast at 1.5 T magnetic resonance clinical scanning. Nanotechnology. (2007) 18. [121] Gossuin Y, Gillis P, Hocq A, Vuong QL, Roch A. Magnetic resonance relaxation properties of superparamagnetic particles. Wiley Interdisciplinary Reviews: Nanomedicine and Nanobiotechnology. 1 (2009) 299–310. [122] Cornell RM, Schwertmann U. The iron oxides : structure, properties, reactions, occurrence, and uses. Weinheim ; New York: VCH, 1996. [123] Neel L. Magnetic properties of ferrites: ferrimagnetism and antiferromagnetism. Ann Phys Paris. 3 (1948) 137–98. [124] Crangle J. Solid-state magnetism. New York: Van Nostrand Reinhold, 1991. [125] Rosensweig RE. Heating magnetic fluid with alternating magnetic field. Journal of Magnetism and Magnetic Materials. 252 (2002) 370–4. [126] Dormann JL. Superparamagnetism Phenomenon. RevPhysAppl. 16 (1981) 275–301. [127] Chantrell RW, Lyberatos A, El-Hilo M, O’Grady K. Models of slow relaxation in particulate and thin film materials (invited). Journal of Applied Physics. 76 (1994) 6407–12. [128] Dormann JL, Spinu L, Tronc E, Jolivet JP, Lucari F, D’Orazio F, et al. Effect of interparticle interactions on the dynamical properties of 𝛾-Fe2O3 nanoparticles. Journal of Magnetism and Magnetic Materials. 183 (1998) L255-L60. [129] Dormann JL, D’Orazio F, Lucari F, Tronc E, Prené P, Jolivet JP, et al. Thermal variation of the relaxation time of the magnetic moment of 𝛾-Fe2O3 nanoparticles with interparticle interactions of various strengths. Physical Review B - Condensed Matter and Materials Physics. 53 (1996) 14291–7. [130] Bean CPL, J.D. Superparamagnetism. Journal of Applied Physics. 30 (1959) 120S. [131] O’Handley RC. Modern magnetic materials : principles and applications. New York: Wiley, 2000.

372 | Part III New trends in sustainable development and biomedical applications [132] Gossuin Y, Hocq A, Vuong QL, Disch S, Hermann RP, Gillis P. Physico-chemical and NMR relaxometric characterization of gadolinium hydroxide and dysprosium oxide nanoparticles. Nanotechnology. (2008) 19. [133] Caravan P. Strategies for increasing the sensitivity of gadolinium based MRI contrast agents. Chemical Society reviews. 35 (2006) 512–23. [134] Burtea C, Laurent S, Vander Elst L, Muller RN. Contrast agents: magnetic resonance. Handbook of experimental pharmacology. (2008) 135–65. [135] Gossuin Y, Gillis P, Hocq A, Vuong QL, Roch A. Magnetic resonance relaxation properties of superparamagnetic particles. Wires Nanomed Nanobi. 1 (2009) 299–310. [136] Solomon I. Relaxation Processes in a System of Two Spins. Physical Review. 99 (1955) 559–65. [137] Bloembergen NJ. Proton relaxation times in paramagnetic solutions. The Journal of Chemical Physics. 27 (1957) 573. [138] Freed JH. Dynamic effects of pair correlation functions on spin relaxation by translational diffusion in liquids. II. Finite jumps and independent T1 processes. Journal of The Journal of Chemical Physics. 68 (1978) 4034–7. [139] Hwang JS, Rao KVS, Freed JH. An electron spin resonance study of the pressure dependence of ordering and spin relaxation in a liquid crystalline solvent. Journal of Physical Chemistry. 80 (1976) 1490–501. [140] Muller RN, Vallet P, Maton F, Roch A, Goudemant JF, Vander Elst L, et al. Recent developments in design, characterization, and understanding of MRI and MRS contrast media. Invest Radiol. 25 (1990) S34–S6. [141] Ouakssim A, Fastrez S, Roch A, Laurent S, Gossuin Y, Piérart C, et al. Control of the synthesis of magnetic fluids by relaxometry and magnetometry. Journal of Magnetism and Magnetic Materials. 272–276 (2004) e1711–e3. [142] Roch A, Muller RN, Gillis P. Theory of proton relaxation induced by superparamagnetic particles. Journal of Chemical Physics. 110 (1999) 5403–11. [143] Corot C, Robert P, Idee JM, Port M. Recent advances in iron oxide nanocrystal technology for medical imaging. Advanced drug delivery reviews. 58 (2006) 1471–504. [144] Li W, Tutton S, Vu AT, Pierchala L, Li BSY, Lewis JM, et al. First-pass contrast-enhanced magnetic resonance angiography in humans using ferumoxytol, a novel ultrasmall superparamagnetic iron oxide (USPIO)-based blood pool agent. Journal of Magnetic Resonance Imaging. 21 (2005) 46–52. [145] Jung CW, Jacobs P. Physical and chemical properties of superparamagnetic iron oxide MR contrast agents: Ferumoxides, Ferumoxtran, ferumoxsil. Magnetic Resonance Imaging. 13 (1995) 661–74. [146] Reimer P, Marx C, Rummeny EJ, Müller M, Lentschig M, Balzer T, et al. SPIO-enhanced 2DTOF MR angiography of the portal venous system: Results of an intraindividual comparison. Journal of Magnetic Resonance Imaging. 7 (1997) 945–9. [147] Modo MMJJ, Bulte JWM. Molecular and cellular MR imaging. Boca Raton: CRC Press; 2007. [148] Simon GH, von Vopelius-Feldt J, Fu Y, Schlegel J, Pinotek G, Wendland MF, et al. Ultrasmall supraparamagnetic iron oxide-enhanced magnetic resonance imaging of antigen-induced arthritis: a comparative study between SHU 555 C, ferumoxtran-10, and ferumoxytol. Invest Radiol. 41 (2006) 45–51. [149] Daldrup-Link HE, Kaiser A, Helbich T, Werner M, Bjørnerud A, Link TM, et al. Macromolecular Contrast Medium (Feruglose) versus Small Molecular Contrast Medium (Gadopentetate) Enhanced Magnetic Resonance Imaging: Differentiation of Benign and Malignant Breast Lesions. Academic Radiology. 10 (2003) 1237–46.

13 Nanoparticles for magnetic resonance imaging (MRI) applications in medicine

| 373

[150] Taupitz M, Wagner S, Schnorr J, Kravec I, Pilgrimm H, Bergmann-Fritsch H, et al. Phase I clinical evaluation of citrate-coated monocrystalline very small superparamagnetic iron oxide particles as a new contrast medium for magnetic resonance imaging. Invest Radiol. 39 (2004) 394–405. [151] Kim BH, Lee N, Kim H, An K, Park YI, Choi Y, et al. Large-scale synthesis of uniform and extremely small-sized iron oxide nanoparticles for high-resolution T 1 magnetic resonance imaging contrast agents. Journal of the American Chemical Society. 133 (2011) 12624–31. [152] Sandiford L, Phinikaridou A, Protti A, Meszaros LK, Cui X, Yan Y, et al. Bisphosphonateanchored PEGylation and radiolabeling of superparamagnetic iron oxide: long-circulating nanoparticles for in vivo multimodal (T1 MRI-SPECT) imaging. ACS Nano. 7 (2013) 500–12. [153] Davis ME. Non-viral gene delivery systems. Current opinion in biotechnology. 13 (2002) 128–31. [154] Ferrari M. Cancer nanotechnology: opportunities and challenges. Nature reviews Cancer. 5 (2005) 161–71. [155] Longmire M, Choyke PL, Kobayashi H. Clearance properties of nano-sized particles and molecules as imaging agents: considerations and caveats. Nanomedicine (Lond). 3 (2008) 703–17. [156] Begley DJ. Delivery of therapeutic agents to the central nervous system: the problems and the possibilities. Pharmacology & Therapeutics. 104 (2004) 29–45. [157] Bareford LM, Swaan PW. Endocytic mechanisms for targeted drug delivery. Advanced drug delivery reviews. 59 (2007) 748–58. [158] Belting M, Sandgren S, Wittrup A. Nuclear delivery of macromolecules: barriers and carriers. Advanced drug delivery reviews. 57 (2005) 505–27. [159] Dobrovolskaia MA, Aggarwal P, Hall JB, McNeil SE. Preclinical studies to understand nanoparticle interaction with the immune system and its potential effects on nanoparticle biodistribution. Molecular pharmaceutics. 5 (2008) 487–95. [160] Chouly C, Pouliquen D, Lucet I, Jeune JJ, Jallet P. Development of superparamagnetic nanoparticles for MRI: effect of particle size, charge and surface nature on biodistribution. Journal of microencapsulation. 13 (1996) 245–55. [161] Decuzzi P, Causa F, Ferrari M, Netti PA. The effective dispersion of nanovectors within the tumor microvasculature. Annals of biomedical engineering. 34 (2006) 633–41. [162] Decuzzi P, Ferrari M. The adhesive strength of non-spherical particles mediated by specific interactions. Biomaterials. 27 (2006) 5307–14. [163] Decuzzi P, Lee S, Bhushan B, Ferrari M. A theoretical model for the margination of particles within blood vessels. Annals of biomedical engineering. 33 (2005) 179–90. [164] Choi HS, Liu W, Misra P, Tanaka E, Zimmer JP, Itty Ipe B, et al. Renal clearance of quantum dots. Nature biotechnology. 25 (2007) 1165–70. [165] Banerjee T, Mitra S, Kumar Singh A, Kumar Sharma R, Maitra A. Preparation, characterization and biodistribution of ultrafine chitosan nanoparticles. International journal of pharmaceutics. 243 (2002) 93–105. [166] Kryza D, Taleb J, Janier M, Marmuse L, Miladi I, Bonazza P, et al. Biodistribution Study of Nanometric Hybrid Gadolinium Oxide Particles as a Multimodal SPECT/MR/Optical Imaging and Theragnostic Agent. Bioconjugate Chemistry. 22 (2011) 1145–52. [167] Harris JM, Chess RB. Effect of pegylation on pharmaceuticals. Nature reviews Drug discovery. 2 (2003) 214–21. [168] Guillet-Nicolas R, Laprise-Pelletier M, Nair MM, Chevallier P, Lagueux J, Gossuin Y, et al. Manganese-impregnated mesoporous silica nanoparticles for signal enhancement in MRI cell labelling studies. Nanoscale. 5 (2013) 11499–511. [169] Lewinski N, Colvin V, Drezek R. Cytotoxicity of nanoparticles. Small. 4 (2008) 26–49.

374 | Part III New trends in sustainable development and biomedical applications [170] Vega-Villa KR, Takemoto JK, Yanez JA, Remsberg CM, Forrest ML, Davies NM. Clinical toxicities of nanocarrier systems. Advanced drug delivery reviews. 60 (2008) 929–38. [171] Zhao YL. Nanotoxicology. Los Angeles, California, CA: American Scientific Publishers, 2006.

J. Greener

14 Microfluidics for synthesis and biological functional materials: from device fabrication to applications 14.1 Introduction Microfluidics (MF) is a general term for the new science and technology relating to the manipulation of small volumes of liquid using small channels. It is considered a green technology due to the small volumes of liquid used, which greatly reduces material consumption and waste, thus enhancing safety when working with toxic or exothermic reactions [1]. Waste and energy associated with post-synthesis fractionation of side-products from target products are further reduced because side reactions are minimized in microreactors. Rapid advancement of this technology is opening up new opportunities in many areas of research and development. Microfluidics enables strong control over both the physical and chemical reaction environments, thereby benefiting applications involving materials synthesis. For example, control of chemical concentrations and their gradients within microchannels opens up new possibilities for controlling polymer chain lengths or embedding gradients in bulk material properties. On the other hand, control of physical conditions such as channel dimensions and shear forces enables control over material shape and size. Together, MF provides the opportunity to create materials at the microscale and smaller with controlled size and shape, and allows separate control of surface chemistry and internal properties. The growth of MF as a platform for microscale materials synthesis supports a growing range of applications such as optics, microelectromechanical systems (MEMS), biomaterials, self-assembly, catalysis, drug delivery and more. In this chapter we review a range of important considerations regarding the synthesis of synthetic and biological functional materials at the micro- and nanoscale. It is designed to give a post-graduate level overview of MF and applications related to functional materials. It covers a broad range of topics, from the introduction of key concepts in MF and reactor fabrication to its utilization in the study and synthesis of new materials. It provides valuable information for both graduate students and professional researchers who are new to the areas of MF and functional micro-materials. Practical considerations such as a review of the different types of microreactors, the materials and fabrication methods for their manufacture are summarized in Section 2. In Section 3 we discuss the special properties of liquids flowing through microchannels and how to manipulate and monitor reaction solutions. In Section 4 we discuss how MF can be used to synthesize polymer micro particles, 1D threads, and 2D microsurfaces by shaping precursor liquids into 1-, 2-, and 3D structures. This discussion includes relevant concepts in MF channel geometries, their surface properties,

376 | Part III New trends in sustainable development and biomedical applications and macroscopic properties of the precursor liquids. Section 5 discusses synthesis of functional nanoparticles in microchannels. Finally, Section 6 reviews state-of-the-art applications of MF for the synthesis and study of microscale functional biomaterials for tissue engineering, cellular microenvironments, and biofilms (BFs). At the end of Section 6 we provide two illustrative examples: microscale microbial fuel cells, and devices for clinical diagnostics related to exposure to nanoparticle materials.

14.2 A practical introduction to microfluidic reactors for material synthesis 14.2.1 Microfluidic reactor geometries For the purpose of this chapter, we define MF channels as possessing dimensions of 1 mm or less along one of their cross-sectional axes. In general, there are three types of MF reactor platforms: capillary tubing, planar devices and digital MF. The first two are reviewed here, but we focus on planar devices due to their versatility and wide usage in various applications.

Capillary tubing Using off-the-shelf polymer capillary tubing is an attractive choice from the perspective of cost and simplicity. In addition, capillary tubing can be acquired in metal, plastic and glass, which provides many opportunities to find chemical compatibility with reaction solutions (Section 2.2). Capillary tubing is generally robust, enabling high pressures and flow rates, as well as high temperatures, particularly for metal tubing. Another major benefit is that their outer dimensions are highly standardized, which has resulted in many commercially available connectors for the introduction of functional elements such as junctions, mixers, filters, pressure regulators, and also for interfacing with standard instrumentation for characterization and fluid delivery. Generally, metal and plastic capillary tubing is opaque, limiting optical imaging and spectroscopic characterization. Opacity can be addressed by introducing a window stage between two opaque capillary tubes. Another solution is the use of capillaries with optical cladding on their inner surface, which can guide light in internal reflection mode, thereby enabling some forms of spectroscopic characterization [2]. Fused silica tubing enables light transmission in the visible wavelengths, but images are generally distorted due to optical index of refraction mismatch at the curved capillary/air and capillary/liquid interfaces. Capillary tubing with square cross-sectional dimensions is available for specific cases, thus addressing the problem of image distortion. The main drawback of capillary tubing is that channel designs are limited to long, straight channel sections. Due to the simplicity of capillary tubing-based reactors, there are limited opportunities in MF. However, capillary tubing is particularly useful for the

14 Microfluidics for synthesis and biological functional materials |

377

emulsification of precursor materials via co-flow geometry (see Section 14.4 of this chapter).

Planar “chips” The majority of microfluidic research and development is conducted on planar “chips”, which are typically two-level planar devices. The first layer is a planar substrate with a patterned series of trenches on its surface, which define 3 walls of the microchannels. Approaches to fabricating this layer from polymer materials are discussed in Section 2.3. The second layer is a sealing layer, which covers the open face of the trenches, thereby defining the fourth wall (the ceiling) of the microchannel. Multilevel devices and 3D fabrication techniques are currently opening the way for more complex chip-based MF reactors [3]. Fabrication techniques are different based on material selection (Section 2.2), but generally involve the replication of channel geometries from a computer-aided design (CAD), enabling high complexity and reproducibility. A range of designs can be implemented to control flow profiles to synthesize functional materials at and below the microscale. Fabrication of microreactors in polymers and elastomers is generally straightforward in addition, an emerging industry providing MF in glass or hard plastic is opening the way for a wider range of MF applications and uses.

14.2.2 Device fabrication materials Microfluidic devices feature very large surface area to volume ratios, enhancing the interaction between the microchannel wall and the solution environment. The constrained physical dimensions of microchannels often result in the need to position optical characterization equipment outside optical microchannels. In cases where probes and other functional elements can be interfaced directly with the microreactor, the mechanical properties of the device should be robust. A review of materials for MF manufacture and their suitability for functional material synthesis and characterization is presented below.

Polymers Polymers play a dominant role as MF device fabrication materials due to their low cost, the wide range of formulations available, the ease of processing, and the potential for mass production. For these reasons, sales of polymer MF devices currently account for approximately 50% of the world market. Moreover, polymer materials can be transparent, enabling optical characterization. Popular polymer materials include thermoreactive (also known as curable) polymers, elastomers, and thermoplastics. One of the most commonly used plastics is an elastomeric material called

378 | Part III New trends in sustainable development and biomedical applications poly(dimethyl siloxane) (PDMS). Its wide use, particularly in academic research and for developmental purposes, results from its low cost, easy demolding and its good self-adhesive properties following surface activation using O2 plasma or UV light exposure. Enhanced bonding techniques for PDMS can result in stronger bonds and seals between layers [4], but they are still more prone to debonding and channel deformation than devices made from more robust materials such as glass or metal. Other major drawbacks of PDMS include its permeability to small molecules and gases, limited chemical compatibility and its propensity to swell in some solvents. As a result, there has been a recent focus on developing new hard plastic-based microreactors, which could enable operation at high pressures and allow interconnectivity with other instrumentation. The rise of polymer-based MF platforms has paved the way for customized MF and thus maximized the flexibility of these systems. Micropatterning of polymer materials can be accomplished by a number of techniques such as curing, micromachining, laser ablation, hot embossing, and injection molding. Bonding the patterned layer to a planar layer for chip-based devices can be accomplished using heat, epoxies, and surface activation techniques. With a few exceptions, most polymers are hydrophobic, which can present challenges in forming oil droplets in water phase (see Section 4) due to the preference of the emulsified monomer-phase precursor liquids which wet the channel walls over the aqueous carrier phase. Surface treatment techniques can change the wetting properties of microchannel walls, making it more suitable for emulsification of a range of liquid precursors. For example, exposure to O2 or air plasma can add OH groups to exposed surfaces, thereby converting the wall to hydrophilic. Chemically treated thermoplastics have also been demonstrated to alter the wetting properties and to enhance resistance to organic solvents, as demonstrated by a series of papers involving surface modification of polycarbonate [5–7]. Generally, surface treatments of elastomers are less effective than for rigid polymers due to the higher mobility of surface groups.

Glass Glasses, quartz, and fused silica are widely used materials for MF reactors [8]. Their advantages include excellent resistance to most chemical environments (with the notable exceptions of HF and other acids), high temperatures and pressures. Glass is typically hydrophilic, making it an excellent choice for emulsifying oil in water (Section 4), and has excellent transparency in the visible region, enabling optical observation and measurements. However, glass is extremely brittle. In addition, it is costly and time-consuming to manufacture custom geometries as compared to plastic materials. Acid (wet) etching through a chemically resistant mask is usually used in fabrication. Typically, hydrofluoric acid (HF) is used for rapid glass etching, but the etching solution often includes a blend of acids to solubilize all by-products. Wet etching of amorphous glass materials results in isotropic, which produces hemispherical chan-

14 Microfluidics for synthesis and biological functional materials |

379

nel cross-sections. A planar coverslip can be bonded to an etched glass device using high temperature, pressure, or an intervening layer of adhesive. Creating throughholes in glass is relatively difficult, requiring sandblasting, laser drilling, diamond drill bits or deep wet etching, making interface of fluidic connections and other peripheral devices challenging. Commercially available glass-based microreactor chips usually have standard designs, as customization adds cost and complexity. Commercially available capillary tubing may help to expand utilization of glass materials and enable different connection and interface options.

Silicon Silicon devices have similar physical properties to glass, but suffer from the major drawback that they are only transparent in the infrared region, making optical characterization methods impossible without complexities such as integrated windows. Silicon also has excellent temperature and corrosion resistance. Untreated silicon surfaces are moderately hydrophilic, but can be made strongly hydrophilic or hydrophobic using various surface treatments [9]. Patterning techniques are similar to glass, but primarily involve dry etching techniques using reactive ions and can achieve anisotropic etching with high aspect ratios [8].

Metal Metal microchannels are an attractive option due to their robust chemical, thermal, and mechanical properties. For example, metals are generally resistant to chemicals, with the notable exception of acidic solutions. The advantages of metal materials over glass for MF include superior thermal properties and strength, as well as easier processing due to the possibility of direct machining. Nevertheless, metals are opaque and sealing microfabricated metallic devices is challenging. Therefore, metallic tubing and connector assemblies are widely available and more frequently used than planar metallic MF chips.

Ceramic Ceramic based MF devices are relatively new, benefiting from high temperature synthesis resistivity and a wide range of chemical compatibilities. A particularly attractive and versatile approach involves low-temperature cofired ceramics capable of incorporating built-in electronics for control and characterization [10, 11]. The drawbacks to using ceramic materials include its brittleness and the lack of optical transparency.

380 | Part III New trends in sustainable development and biomedical applications 14.2.3 Fabrication of polymer-based planar microreactors and components As the field of MF has grown, so too has the range of techniques for production of MF devices, particularly in various polymer materials. In addition, the production of various functional elements such as microvalves and electrical components is expanding the functionality of MF [12–15]. This section focuses on fabrication of polymer chipbased MF reactors largely because they constitute the fastest growing and most versatile type of architecture. We will start with the different approaches for producing the feature layer (2.3.1–2.3.5) and then discuss the bonding step (2.3.6).

Casting Casting is the process by which a mold is coated with a liquid pre-precursor which can be solidified by heat curing, UV exposure, chemical crosslink or some combination thereof [16]. Figure 1 shows the three step process for casting, which includes (a) coating the mold with the precursor materials, (b) curing, and finally (c) demolding. The molded surface is then bonded to a planar substrate to form enclosed MF channels (Fig. 14.1). Elastomeric materials, such as PDMS, are widely used because they can form patterned surfaces rapidly, are inexpensive and can be easily demolded. Molds for casting MF devices can be produced by many means including machining, ablation, electroforming and etching, but the most common process is photolithography. Casting

(a)

(b)

(c)

(d)

Fig. 14.1. Microfabrication by casting. (a) A liquid pre-polymer is poured on top of a mold containing the inverse design of the desired microchannel geometry. (b) The process of solidification (curing) occurs while precursor material is in contact with the mold. (c) The solidified polymer is removed from the mold. (d) Model of elastomeric PDMS being peeled from a silicon wafer mold.

14 Microfluidics for synthesis and biological functional materials |

381

Photolithography Photolithography is a widely used technique for MF development since it is rapid, inexpensive, and can result in good spatial resolution. A photoresist is spin-coated onto a flat substrate and hardened. Then, using a 2D photo mask, portions of the resist are exposed to light. In negative photoresists, light selectively crosslinks exposed portions through the optical mask, while in positive photoresists, the light breaks bonds. In either case, the topography of the photoresist layer is formed following the removal of the uncrosslinked photoresist. The heights of the mold features are defined by the photoresist thickness. Therefore, care must be taken during the spin-coating process to ensure that the photoresist layer has an even thickness across the entire substrate surface, particularly in the middle and near the edges, where material buildup tends to occur. The patterned surface can then be used to form a casting mold or embossing imprint template for patterning target materials with the desired microreactor channel geometry. In some cases the patterned photoresist is used directly as the MF device.

Soft-lithography Soft lithography is defined as the use of a soft surface as an imprint template, mold, or mask [13]. Very often, the PDMS replicas made in the casting process described above are used as templates for subsequent molding steps when the mold needs to be compliant, for example, to generate features on curved surfaces. In other cases, the PDMS replicas are used as “soft stamps”, enabling good contact between the stamp features and the target surface, and the transfer of thin layers of material from the stamp feature to the target surface.

Nanoimprint lithography Nanoimprint lithography (NIL), also known as hot embossing, is a promising technique for large-area topographic polymer micropatterning by transferring topographical features from an imprint template to the heated target substrate [17]. It is similar to injection molding, as elevated temperatures are used to enable production. However, NIL is more straightforward to implement, requires shorter fabrication times and lower temperatures. This process occurs in three steps as shown in Fig. 14.2. First the system is heated to the embossing temperature (Te ), which is higher than the glass transition temperature (Tg ) of the thermoplastic. Following heating, the embossing pressure (Tp ) is applied, which forces the imprint template features to penetrate the compliant polymer. In the third step, the thermoplastic is cooled to the de-embossing temperature (Td ), which is lower than Tg . This results in the solidification of the thermoplastic, thereby trapping the inverse surface topography of the imprint template. The imprinted substrate is then removed from the imprint template and bonded to a planar material, usually, but not necessarily the same material. Sometimes lubri-

382 | Part III New trends in sustainable development and biomedical applications cating layers are applied to the imprint template to facilitate the de-embossing stage. Typically, features are an exact replication of the imprint template. The main hurdle for the implementation of NIL as a tool for fabrication of customized MF geometries is the imprint template. The imprint template defines the geometry, resolution, and quality of the imprinted substrate. Commercially available imprint templates are usually produced from metals, e.g., nickel, by using mechanical machining, laser ablation, or electroforming [18–20]. The disadvantages of these methods are that they are expensive and production is slow. Moreover, imprint templates based on patterned etched silicon wafers are fragile. A method has recently been developed to overcome these limitations by generating inexpensive and robust masters manufactured via photolithography [12]. The addition of a metallic layer to evenly distribute heat and functional surface groups also helps the de-embossing process [21].

(a) (d)

(b)

(ii) (i)

(c)

(e)

Fig. 14.2. Microfabrication by embossing. (a–c) The steps for embossing a thermoplastic sheet. (d) An image of an imprint template with the geometry of a flow focusing device. Features are highlighted in black. (e) An embossed device with channels highlighted in green. Critical features such as the flow focusing emulsification point (i) and the polymerization compartment (ii) are highlighted (d and e are reprinted with permission from [12]).

Other fabrication techniques Other approaches to microfabrication include etching techniques (usually for glass and silicon) [8, 22–24] and computer numeric controlled (CNC) techniques such as micromilling [25, 26], and laser ablation [27]. Each of these has advantages and limitations. For example, wet etching is expensive and unsuitable for the fabrication of isotropic features in amorphous materials. Micromilling and laser ablation can be ap-

14 Microfluidics for synthesis and biological functional materials |

383

plied to various materials in addition to polymers; however, the generation of waste can be problematic. Additionally, these methods are challenging to implement in the manufacture of high aspect ratio microchannels and channel intersections, and often result in rough surface finish. In each of these cases, prototyping of complex geometries is either expensive or slow, and they are therefore not considered in detail here.

Bonding step Following the fabrication of the microstructured layer containing the channel geometries, fluidic access holes are drilled or punched into the patterned side of the device, which is then bonded to an unpatterned planar layer to seal the channels. Usually the bonding layer is the same material as the microfabricated layer, but in some cases it is advantageous to bond a different material as the sealing layer, for example when optical transparency is required. Bonding can be accomplished by heating, surface activation, or the use of epoxies. In the case of plastics, surface activation is usually accomplished by exposure to plasma gases or UV radiation (Fig. 14.3).

(a)

(b)

Fig. 14.3. Bonding of a microfabricated device. (a) The microfabricated layer (blue) and the planar layer (cross-hatched) are heated, surface activated or coated with epoxy. (b) The activated layers are brought into contact with each other and pressure applied.

14.3 Manipulating and measuring precursor reagent streams in microchannels In this section we look at both the fundamental concepts and practical approaches which can affect the manipulation and measurement of precursor liquids in microchannels.

14.3.1 High surface area to volume ratios in microchannels We first consider the large surface area to volume ratios in microchannels compared to bulk reactors. Surface area (SA) to volume (V) ratios can be described by equation (14.1a) for a spherical reactor vessel, and equations (14.1b) and (14.1c) for channel-

384 | Part III New trends in sustainable development and biomedical applications based reactors with circular and rectangular cross-sections, respectively. SA 4πr2 3 = = VB 4/3πr3 r

(14.1a)

SA 2πrl 2 = 2 = Vcirc r πr l

(14.1b)

(2h + 2w)l (2h + 2w) SA = = , Vrec hwl hw

(14.1c)

where r is the radius of the spherical reactor (14.1a) or the channel radius (14.1b) and h and w are the height and width of a channel with rectangular cross-section (14.1c), respectively. In equations (14.1b) and (14.1c), l is the channel length. Given the differences in the radial dimensions between microreactors (hundreds of micrometers or less) and bulk reactors (meters), SA/V in MF devices can be over 4 orders of magnitude greater than that for batch reactors. As discussed below, large SA/V has important effects on the flow conditions within microreactors.

14.3.2 Rapid heat transfer Large SA/V has a strong effect on the temperature profile of liquid in microchannels due to large thermal flux between the liquid and the microchannel walls [28]. As a result, the liquid rapidly reaches equilibrium temperatures and thus, thermal gradients are strongly suppressed throughout the reactor environment. This has the effect of limiting side reactions and eliminating Arrhenius-related distribution to reaction rate constants in the microreactor. In addition, safety is enhanced by effectively eliminating the chance of runaway exothermic reactions which can occur in batch reactors. Since liquid temperatures quickly equilibrate after entering the microchannels, temperature control of the microreactor environment is an effective way of accurately maintaining temperature throughout the reactor [29]. Enhancing thermal flux for faster equilibration times can be achieved by using fabrication materials with better thermal conductivity.

14.3.3 Control of concentrations In general, MF channels are isolated from ambient conditions, thereby preventing chemical and biological contamination of the reaction solution. The notable exception is for those reactors fabricated in PDMS, which are sufficiently porous to allow small molecules such as O2 and CO2 to diffuse into the reaction solution. A further benefit of the MF approach is the ability to vary the concentration of reagents in a highthroughput way by modulating the relative flow rates of individual reagent streams.

14 Microfluidics for synthesis and biological functional materials |

385

This enables exploration of a large parameter space, which is useful for optimization of formulations. As an example, a study involving a multicomponent polymerization included solutions containing a monomer (M), initiator (I), and chemical accelerator (A) had initial concentrations CM,i , CI,i , and CA,i , respectively [30]. After introduction into an MF reactor and mutual dilution, the reagent concentrations became CM,d = CM,i × QM /QT

(14.2a)

CI,d = CI,i × QI /QT

(14.2b)

CA,d = CA,i × QA /QT ,

(14.2c)

where CM,d , CI,d , and CA,d are the diluted concentrations; QM , QI , QA are the flow rates of the streams containing the monomer, initiator and accelerator, respectively; QT is the total flow rate, which is the sum of all flow rates including a dilution stream of water (QW ), given by QT = QM + QI + QA + QW .

(ii) (iii)

(i)

(iv)

(v)

P1

(vi) P2 P3

Fig. 14.4. Schematic of an MF reactor for the study of free radical polymerization of a complex reaction. Solutions of monomer, a chemical accelerator, a monomer, and water were supplied to the MF reactor by tubing connected via inlets (i–iv). Four small wavy channels following the inlets are used to increase hydrodynamic resistance in order to stabilize flow. Mixing of reagents occurred in a step-wise manner before entering the large serpentine channel (reaction chamber (v)), where mixing was enhanced and the polymerization reaction took place. The composition of the reaction mixture was characterized by ATR-FTIR using a probe placed at point P1 . A temperature probe was located at P2 , and a pH probe was located at P3 . The reaction solution left the MF reactor via the outlet (vi). The scale bar is 1 cm. Reprinted with permission from [30].

386 | Part III New trends in sustainable development and biomedical applications Another example includes the flow-based modulation of reagent concentrations in an acid/base titration experiment to find critical values such as the pKa , the pH at the equivalence point (pHe ) [31]. In this work, a strong base was used as a titrant against strong, weak, and multiprotic acids. As shown in Fig. 14.5, the concentration of the reagents was carefully modulated by flow control of the base and acid solutions (QB /QT ) to achieve accurate titration curves.

14.3.4 Controlling “time on chip” Synthesis in an MF reactor occurs continuously as the reagents flow through the microchannel. Therefore, given a particular flow velocity, there is a relationship between the time of reaction and the distance downstream from the point of initial mixing. This relationship is given by the so-called distance-to-time transformation t = d × A/QT ,

(14.3)

where d is the distance downstream following the point of initial mixing. The velocity is given by the term A/QT , where A is the cross-sectional area of the microchannel, and QT is the total flow rate. Therefore, changing QT enables precise timing of reaction steps or time-delayed measurements for studies of reaction kinetics.

14.3.5 Control of hydrodynamics and mass transfer Control over turbulence The most fundamental difference between macro- and microflows is that there exists no turbulence for flow through a microchannel. This is quantified by a low Reynolds number (Re), which is given by: ρvDh Re = , (14.4) μ where ρ is the liquid density, v is the liquid velocity, Dh is the hydraulic diameter of a channel, and μ is the dynamic viscosity. For channels with rectangular inner crosssections, which are typically fabricated from photolithographic master molds and imprint templates, the hydraulic diameter is given by Dh = 2wh/(w + h), where w is the channel width and h is the channel height. Values of Re which are below the critical threshold of 2 300 describe a laminar flow environment. A typical approach to achieving laminar conditions in macro flow systems is to use high viscosity liquids. However, this is usually not desirable, particularly for biological systems in aqueous media or reaction environments for chemical synthesis of micro materials. The very small values of Dh for microchannels naturally result in laminar flow, even for very low viscous solutions flowing at high velocities. The lack of convective flow in a direction other than collinear along the microchannel means that mass transport in the

14 Microfluidics for synthesis and biological functional materials |

387

13 12 11 10 9 pH

8 7 6 5 4 3 2 1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 QKOH/QT

(a) 14 12 10 pH

pHe 8 6 pKa

4

2 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 (b)

QKOH/QT

Fig. 14.5. Flow-based control of acid and base concentrations for accurate titration experiments in microchannels. (a) Microfluidic titration of a strong acid with a strong base. Titration curves for [KOH]i = 0.05 M, and [HCl]i equal to 0.025 M (◼), 0.035 M (󳵳), and 0.055 M (∙). The total volumetric flow rate was QT = 2.0 ml h−1 . The dashed lines show theoretical curves. (b) Microfluidic titration of a weak acid with a strong base. Titration curve for [CH3 COOH] = 1.00 M, [KOH] = 1.00 M. Data acquisition was conducted 2–3 min after changing the flow rates of the liquids. The pKa is taken from the first inflection point (where the change in pH is minimal) and the pHe is measured from the second inflection point (where the change in pH is maximal). Reprinted with permission from [31].

388 | Part III New trends in sustainable development and biomedical applications lateral direction within the microchannel is controlled strictly by diffusion, which is a very slow process. Diffusion-dominated mass transfer has strong implications for mixing. In addition, the laminar flow environment supports interesting and useful phenomena, such as co-flow of miscible liquids side-by-side for long distances without significant intermixing.

Control of mixing In general, mixing in MF reactors is highly controllable [32]. In inefficient mixing environments, where diffusion is the only mass transfer mechanism, very strong chemical gradients can be established at liquid-liquid interfaces. Certain device designs have also been demonstrated, which establish long-range chemical gradients across the entire microchannel [33]. These devices have had a number of applications including cell culture under continuously varying chemical environments [34, 35]. Due to the long-term stability of laminar flow, concentration gradients are indefinitely stable. As will be discussed in Sections 4.6 and 6.1, this enables the formation of 1D microthreads and 2D membranes. In addition, we will discuss the control over chemical gradients as a means of synthesizing materials with built-in gradients in their properties, thereby enabling further functionality. Conversely, there are a range of techniques for achieving excellent mixing in microchannels. This can be achieved passively, that is without any external actuation, by exploiting the control over MF channel designs featuring chaotic mixing elements [36], two- and three-dimensional switchbacks, inchannel obstacles [37, 38], and hydrodynamic focusing of the stream into very thin interdigitated coflowing streams where mixing by diffusion can occur rapidly. Active mixing is also applicable using alternate-injection or pulsed-flow, electrokinetic mixing, or coflowing droplets or particles. Good mixing improves the polymer polydispersity index [39, 40] and enables studies of reaction kinetics without diffusion limitations [30].

Control of shear forces Due to the non-slip condition at the microfluidic wall, there is always a gradient in flow velocity perpendicular to the channel wall. This results in an average wall shear stress, which is measured in N/m2 and is directed perpendicular to the channel. The average shear stress is given by τav = Δv/L ⋅ μ,

(14.5)

where τav is the wall shear stress vector, μ is the liquid dynamic viscosity, Δv is the difference in flow velocities between minimum velocity, which is zero at the channel wall, and maximum velocity (vmax ), which is typically taken at the centre of the channel, and L is the distance between the wall and the perpendicular distance from the channel wall where flow velocity is vmax (also usually the centre of the channel). The

14 Microfluidics for synthesis and biological functional materials |

389

laminar flow environment results in highly predictable shear stresses which are stable for long periods. As will be discussed shortly, this is an advantage in micromaterial synthesis.

14.3.6 Characterization in microchannels With the notable exception of optical microscopy, analytical techniques are traditionally applied to liquid emerging from the MF device (off-chip). In the last 10 years, however, the MF community has made important advances in the implementation of in situ characterization. This avoids potential contamination from the ambient environment outside the MF device and enables high-throughput screening of the reaction variable space, feedback control of on-chip processes, monitoring of transient species, and kinetic studies with good time resolution [41–45]. Some characterization techniques from the perspective of the development of functional materials are reviewed in the following section.

Microscopy Microscopy is the most well-established in situ characterization method for MF. It is typically achieved via transmission-mode microscopy using bench-top microscopes. In addition, a range of new compact microscopic techniques are becoming available for MF [46]. Optical imaging of the liquid phase gives nondestructive, positiondependant measurements of color and color density with high spatial- and timeresolution [47]. Optical microscopy is usually used for multiphase systems (liquid/ liquid emulsions and bubbles or solid particles suspended in continuous phase liquids), because it is a convenient and versatile tool for measuring size and shape. In this case, optical microscopy relies on differences in optical properties between continuous phase and the micromaterials or their precursors. It is also convenient for collecting statistical information, for example to determine coefficient of variance (CV) of precursor droplet sizes (see Section 4.3). Standard enhancements to sensitivity can be gained using fluorescence, phase contrast or differential interference contrast microscopy in microchannels [48–51]. For example, transparent materials such as cells and low density hydrogels can be measured in phase contrast mode or using fluorescent labels in fluorescent mode. Optical microscopy also enables simultaneous measurements throughout the entire field of view within the microreactor. Using equation (14.3), measurements at multiple times of reaction in a single image can be obtained.

390 | Part III New trends in sustainable development and biomedical applications Spectroscopy There is currently a gap between the high level of control that MFs provides over reaction conditions and the ability to accurately make in situ measurements. Implementation of on-chip spectroscopy holds much promise to fill this niche and bring MF reactors closer to their potential as versatile platforms for discovery [52, 53]. The development of in situ spectroscopic tools has enabled direct measurements of a much larger portion of the relevant variable space for MF material synthesis. Tools include NMR [54, 55], fluorescence [56], infrared (IR) [57, 58], UV-visible, Raman spectroscopy [59] and surface-enhanced Raman spectroscopy (SERS) [3, 60]. In addition, multimodal characterization has been achieved by the combination of spectroscopy with pH and temperature probes [29]. In this work, the authors demonstrated the ability to use the temperature probe as a feedback system for temperature control. This system was used to study the reaction kinetics of N-isopropylacrylamide at different pH values and temperatures. Examples, including vibrational spectroscopy via Raman spectroscopy and attenuated total reflection Fourier transform Infrared (ATR-FTIR) spectroscopies have been used to characterize the degree of conversion of monomer to polymers in microreactors [30, 59]. A Raman-based system with built-in SERS surfaces was used to monitor the growth of BFs under well-controlled hydrodynamic conditions [3]. In Section 5 we review applications of UV-visible spectroscopy for in situ measurements of nanoparticles during synthesis, enabling real-time control of reaction conditions for optimization of formulations. A priority area for material synthesis in microchannels is spectroscopic imaging, which will enable measurements of multiphase systems, due to the ability of the technique to resolve both chemical and spatial information at the same time [61, 62]. The MF reactor must generally be transparent in some operational frequency window in order to achieve light-based measurement. This usually limits the reactor materials to glass or plastic, although silicon could be used for infrared (IR) observations, as it is optically transparent in the IR, whereas materials that are transparent in the visible region of the spectrum may be opaque in other regions of interest.

Other characterization modes Many examples of characterization modes which are conducted directly at the outlet of the MF reactor exist. These are usually destructive, such as mass spectrometry, or disruptive, such as chromatography. These techniques have the benefit of being in direct contact with the flow environment, therefore not placing requirements on the fabrication material of the MF reactor to be optically transmissive, for example. Chromatographic tools such as continuous online size exclusion chromatography, which measures both the molecular weight and its distribution for polymers synthesized by nitroxide-mediated polymerisation have proven to be effective [40]. Multidetection of polymer molecules via gel permission chromatography at the outlet of

14 Microfluidics for synthesis and biological functional materials |

391

a microreactor enabled characterization of linear and branched polymers by combining a concentration- and mass-sensitive detection. Light scattering was also used to characterise the onset and size distribution of multilamellar vesicles of diblock copolymers [63]. Indeed, chromatographic techniques have been used very effectively with MF. However, they are destructive and therefore not easily applicable to real-time optimization of formulations and quality control. A drawback of localized probes is their inability to address different regions onchip simultaneously, unlike imaging-based measurements. Innovative solutions to this problem include internal redirection of flow from different positions on-chip to the probe. For example, a multiplexed microreactor capable of sample fractionation enabled the pre-concentration and elution of protein samples from individually addressable, solid phase extraction (SPE) channels at different locations without crosscontamination between channels. The samples from different SPE channels could be individually eluted and directed to a common outlet for electrospray ionization mass spectroscopy (ESI-MS) [64]. Figure 6 demonstrates this technique by the sequential flow of a fluorescent tracer from different feeder channels into the common outlet. Other approaches to flow redirection have been achieved using physical valves which, however, require complicated fabrication processes.

(a)

(b)

(c)

(d)

Fig. 14.6. A device with a series of downstream channels (8), with SPE being sequentially loaded with fluorescein isothiocyanate-labelled bovine serum albumin (BSA-FITC). Following this step, the protein BSA-FITC sample was sequentially eluted into a common outlet, where ESI-MS was conducted. Figures (a)–(d) show the sequential redirection of flow through four separate channels while maintaining flow through one common channel. Reprinted with permission from [64].

14.4 Microfluidics for polymer microparticles Microfluidics for polymer microparticle synthesis has become a mature field with many methods and applications [65]. In this section we review important concepts in microparticle synthesis using MF reactors.

392 | Part III New trends in sustainable development and biomedical applications 14.4.1 Manipulating the shaping of liquid precursors This section discusses various methods and factors that affect the shaping of precursor liquids. Typically MF synthesis of functional micromaterials begins with the precursor solution, which is manipulated and geometrically confined by a sheath flow. In cases where separate droplets are formed, the precursor solution is called the dispersed phase and the sheath flow is called the continuous phase. A reaction is started, which results in solidification via polymerization, gelation, or crosslinking (Sections 4.7 and 6.2). Dimensions of microscale materials are largely determined by the dimensions of precursor liquid droplets, which are controlled by the channel geometry, the mechanism of droplet formation, the physical properties of the precursor liquid phase and the continuous phases, the volumetric flow rates of the continuous and precursor phases, and the channel wall properties. These issues are discussed in detail in Sections 4.2, 4.3, and 4.4.

14.4.2 Effect of the channel wall There are two considerations in shaping the precursor liquid related to channel walls. The first is their geometry downstream of the droplet formation. In addition to guiding the shear forces which control emulsification, droplets which are solidified within geometrically confined regions will retain their shape even after the confinement conditions are removed [66]. The second important consideration is the channel wall hydrophobicity. In forming and sustaining emulsified precursor droplets, the continuous phase must preferentially wet the channel wall, otherwise emulsions can become unstable, collapse or cause a phase inversion, whereby the intended continuous phase liquid becomes emulsified into segmented droplets (discontinuous). Hydrophilic glass devices resist the wetting of oily precursor phase droplets, whereas hydrophobic plastic devices generally do not. Wetting properties can be changed using a surfactant. Surfactants work by moving to the interface between aqueous and oil phases and creating a buffer layer which resists wetting. For example, surfactant dissolved in the hydrophobic droplet of monomer solution surrounded by a continuous water phase will prefer to form a layer at the water/oil interface. The hydrophilic head group faces away from the monomer solution providing a barrier to wetting the hydrophobic wall. Similarly, a surfactant in the aqueous continuous phase can increase its ability to wet a hydrophobic wall. However, surfactants can block mass transfer across the liquid/ liquid or gas/liquid boundaries and can affect chemical reaction kinetics. Also, surfactant trapped at the interface can result in undesired properties in the final material. It is preferable to choose the appropriate fabrication materials, thus avoiding the need for surfactants. Polymer materials are typically hydrophobic, although some exist which are hydrophilic. In addition, there are a growing number of ways of modify-

14 Microfluidics for synthesis and biological functional materials |

393

ing the surface microchannel wall to change its hydrophobicity to suit the application. The reader is referred to Section 2.2 where this is discussed for polymer materials.

14.4.3 Emulsification of precursor droplets A droplet is defined as a discrete segment of liquid surrounded on all sides by the socalled continuous phase. Usually the most important physical properties of droplets are their size and polydispersity, otherwise known as the coefficient of variance (CV). The CV is given by δ/dm × 100%, where δ and dm are the standard deviation and the average droplet diameter, respectively. According to the standards of the National Institute of Standards and Technology (NIST): “a particle distribution may be considered monodisperse if at least 90% of the distribution lies within 5% of the median size” [67, 68]. Typically, the continuous phase is unreactive, although in some cases it may contain reagents which react at the interface with the precursor phase. Droplets which are larger than the channel cross-sectional dimensions are called plugs. Plugs occupy the entire cross-section of the channel, thereby causing discontinuities in the bulk flow of the continuous phase fluid. Nevertheless, the continuous phase will form a thin layer between plug and wall due to its preference for wetting the inner surface of the channel. Forming particles from droplets is preferable for micromaterial synthesis because they are not in contact with the walls and can freely travel with the continuous liquid phase after solidification. However, there are some applications where plugs are desirable because their shapes can be templated by the channel dimensions (Section 4.2). Control over the size of an emulsion can be gained by changing the relative flow rates of the precursor (Qp ) and continuous phases (Qc ). Generally, increasing the flow rate ratio Qp /Qc results in larger droplets, but the correlation is complex and requires calibration. In addition, increasing the viscosity of the continuous phase will increase shear against the precursor phase, resulting in a similar effect to that of increasing Qc . Increasing the viscosity of the precursor phase resists droplet breakup necessitating higher Qd . In addition, the breakup of viscous precursor phase can result in small unwanted satellite particles. In addition, increasing the interfacial tension between the precursor and continuous phases results in smaller droplets.

14.4.4 Channel geometries to achieve emulsified droplets In the next sections we discuss the different channel geometries used for droplet breakup, which include (a) flow focusing, (b) cross flow, or (c) co-flow devices [69].

394 | Part III New trends in sustainable development and biomedical applications Flow focusing A flow focusing device confines the precursor phase between two continuous phase liquid streams (Fig. 14.7(a)). Breakup of the precursor stream into discrete droplets is caused by flow through a constriction in the channel. In the so-called “dripping mode”, competitive flow through the constriction between the precursors and the continuous phase causes one stream to flow while the other is temporarily blocked. Pressure build-up in the blocked stream results in its eventual flow through the constriction, thereby temporarily blocking the other stream. In this way, the precursor phase is periodically pinched off and forms droplets which are separated from each other and the walls by the continuous phase liquid.

Cross flow T-junction geometry supports the flow of the continuous phase at 90 degrees from the flow of the precursor phase (Fig. 14.7(b)). Droplet formation of the precursor phase is caused by shear stress being applied by the continuous phase. In a second droplet formation mode, the precursor phase enters and fully blocks the downstream channel, resulting in plug formation. This causes a rapid increase in the pressure in the continuous stream channel, which causes the breakup of the precursor phase.

Co-flow A co-flow system is typically used in capillary tube MF systems, as it requires the alignment of an inner capillary within an outer capillary [70]. The precursor liquid flows through the inner capillary, exiting it through a nozzle to meet the continuous liquid phase, which is flowing through the outer capillary (Fig. 14.7(c)). The shear forces applied to the precursor phase liquid by the continuous liquid phase cause droplet breakup. The co-flow approach has the advantage of being an inherently 3D technique, confining the precursor phase on all sides, thereby limiting the precursor phase from touching the capillary walls. In practical terms, however, alignment of the inner and outer capillary is difficult and often the precursor phase is directed toward the capillary wall.

(a)

(b)

(c)

Fig. 14.7. Geometries for microfluidic emulsification based on (a) co-flow, (b) T-junction and (c) flow focusing devices. Reprinted with permission from [69].

14 Microfluidics for synthesis and biological functional materials |

395

14.4.5 Multiple emulsions Multiple emulsions are emulsions within emulsions. They can be beneficial for applications requiring encapsulation, such as drug delivery. The generation of multiple emulsions using MFis conducted via sequential emulsification [71–73]. For example in Fig. 14.8(a), an aqueous phase is first emulsified in oil using a T-junction. In order for the aqueous (dispersed) phase not to wet the channel walls, they must be hydrophobic after the first emulsification junction. The output stream of the first T-junction then feeds into the precursor phase of the second T-junction, where the emulsions are emulsified by an aqueous continuous phase. This is called a water-in-oil-in-water (W/O/W) emulsion, that is to say an oil droplet with an aqueous core. It is important for the channel walls to be hydrophilic after the second T-junction in order to ensure that the outer (oily) shell of the emulsion does not wet the walls. This process can be continued to synthesize triple and higher-order emulsions. Similar strategies have been employed with flow focusing and co-flow geometries [74, 75]. Multiple cores can also be trapped in an outer shell (Fig. 14.8(b)). Synchronizing the formation of the droplet generation at the first and second droplet generator is important to control the filling of the second droplet. The microfluidic approach for sequential emulsions has the dual benefit of narrow polydispersity, while offering a means of generating an arbitrary level of complexity in synthesis which can enable a range of new formulations.

14.4.6 Forming linear threads and two-dimensional interfaces Templating linear threads: An area of growing interest involves continuous synthesis of micro diameter threads, which can have applications in areas including waveguides, drug delivery systems, additives for suspension rheology, and cell culture environments [76, 77]. The liquid thread phase can be formed using flow focusing devices, co-flow systems, T-junctions and other architectures [78, 79] to confine a core precursor phase by a sheath flow. Miscible fluids are usually used in order to suppress the tendency to form isolated emulsified droplets as discussed previously. However, researchers have recently demonstrated the potential of using immiscible liquids for forming linear threads, when droplet emulsification can be prevented. Advantages include true cylindrical confinement of the inner thread phase, low diameter threads, and smooth surfaces. Each of these effects is a result of the strong capillary forces between immiscible fluids, which tend to minimize the thread surface area. The use of immiscible liquids which do not breakup into droplets is achieved by introducing a sheath flow, which is comprised of a non-Newtonian elastic liquid phase, such as a semi-dilute polymer solution. It has been shown that if this elasticity is greater than the elasticity of the precursor solution, the normal forces directed against the precursor thread actually stabilize the liquid thread, thereby overcoming capillary instability which typically causes the thread phase to break up. Using this method, threads with

396 | Part III New trends in sustainable development and biomedical applications Qaq1 Qoil

Hydrophobic surface

Qaq2 (a)

(b) Fig. 14.8. Multiple emulsions. (a) A two-stage T-junction emulsification device forming first water-inoil (W/O) emulsions in a hydrophobic channel due to the shearing action of oil flowing at flow rate Qoil against flow of an aqueous solution Qaq1 . Next a second stream of water (Qaq2 ) applies shearing force against the W/O emulsion to create a W/O/W double emulsion in a hydrophilic channel. (b) An arrangement of different third order O/W/O/W emulsions with precise control over the number and ratio of co-encapsulated droplets using a complex hierarchical and scalable MF encapsulation device. Scale bar is 400 μm. Fig. 14.8(b) is reprinted with permission from [73]

diameters ranging from 1–14 μm were formed. Multi-phase threads have also been used to template hollow tubes and microfibers with controlled lengths [80–82]. Formation of threads of biofilm, called streamers, have been demonstrated to occur at sharp corners due to the templating effect of counter rotating vortices which exist at these locations [83, 84]. Templating planar materials: There are a number of ways of forming interfaces which can serve as templates for planar 2D microstructures. For example, the liquidliquid interface between two co-flowing phases will persist for a long time in the diffusion limited environment within a microchannel. If one of the streams is a precursor phase (containing, for example, monomer and initiator molecules), and the other a trigger phase containing molecular species which react with the initiator to form radicals, the formation of a thin film will occur at the interface between them. Another approach is to cause pH-triggered precipitation in the precursor phase by co-flowing a second phase with a pH that does not support material dissolution, as discussed in Section 6 [85]. Liquid-solid interfaces within microchannels offer another conve-

14 Microfluidics for synthesis and biological functional materials |

397

nient approach for templating materials in 2D planes. For example, biomaterials such as biofilms can be grown on the microchannel wall and subjected to different shear forces, temperatures or chemical concentrations (Section 6.3).

14.4.7 Converting liquid precursors into solid micro-materials Converting the liquid precursor to a solid is typically the result of a polymerization step which traps the liquid in its state at the time of reaction. This can result in certain shapes or preserve concentration gradients in the droplet, leading to highly functional Janus particles, that is, particles with different properties on opposite sides. First, we consider a simple case where free radical polymerization is used as a method for converting monomer precursor solutions into solids. The monomer solution should include initiator molecules. Initiation can be triggered by temperature, photons or an appropriate chemical reaction. The rate of change of a typical free radical polymer product is given by dCP /dt = k󸀠 C1M C1/2 (14.6) I , where k󸀠 is the effective rate constant, with contributions from rates for chain propagation, decomposition, and termination, and CP , CM , and CI are the concentrations of the product, monomer and initiator species, respectively. Photopolymerization, initiated using a photoinitiator, is typically used because relatively low thermal energy is added to the system and dosages can be well-regulated, as well as spatially constrained, in order to confine the reaction to designated locations. Monomer droplets, linear thread phases or films can be exposed to photons of the proper wavelength, resulting in polymerization at the rate given by equation (14.6). Rapid polymerization may be preferable in order to suppress coalescence due to collisions between unsolidified droplets, in which case high concentrations or intense illumination is helpful. In cases where the device is not transparent in the spectral window required for photoinitiation (usually in UV), other means of initiation can be undertaken such as chemical initiators. Section 6.2 reviews examples of biomaterials for cell encapsulation which involve particle solidification via redox polymerization and gelation.

14.4.8 Scale up: a circuit analysis of microfluidic flow in a highly parallelized microreactor In order to produce micromaterials at industrially relevant quantities, the throughput of MF synthesis should be increased to kilograms per day or higher. Unlike scale-up procedures in bulk reactors, MF reactors cannot change their dimensions. Therefore, the preferred approach is to “number up” microreactors; that is to introduce many parallel reactors with a common inlet and outlet [86–88]. This approach avoids the iterative process of batch reaction scale-up which requires continuous re-optimization due

398 | Part III New trends in sustainable development and biomedical applications to changes in reaction conditions affected by reactor volume. However, there are challenges in the numbering up approach. A set of design rules has been determined to prevent flow redirection to neighboring reactor channels [88]. First, it is critical to ensure that all parallel reactor channel dimensions are identical. This minimizes differences in hydrodynamic resistance, enabling uniform flow rates in each parallel reactor. We take this opportunity to introduce the electrical circuit analog to MF systems. Consider an electrical circuit with three parallel resistors (Fig. 14.9) [89, 90]. Using Ohm’s law (V = iR), the current through each parallel path can be determined from the equations i1 = Vi /R1 , i2 = Vi /R2 , i3 = Vi /R3 , if the applied voltage at the branching point (Vi ) and the electrical resistance of each parallel path (R1 , R2 , R3 ) are known. In addition, we can invoke tools such as conservation of charge (iT = i1 + i2 + i3 ). Similarly, the flow rate of a liquid through parallel MF paths can be determined using the Hagen-Poiseuille law (dp = QRH ) to solve the flow rate through different parallel paths given knowledge of the applied pressure at the branching point (pi ), and the hydrodynamic resistance of each parallel path (RH1 , RH2 , RH3 ). In addition, we can invoke conservation of mass (QT = Q1 + Q2 + Q3 ). The hydrodynamic resistance of a single channel is given by: RH = RH =

8μL πR4

(14.7a) 12μL

wh3 (1

− 0.63 ⋅

h ) w

,

(14.7b)

where the dimensions of the channels are given by the radius (R) for circular channels, and w and h for rectangular channels. In either case, L is the channel length and μ is the liquid viscosity. Since the hydrodynamic resistance is very sensitive to the cross-sectional dimensions of the channel, the first rule to numbering up for parallel synthesis is to ensure that all parallel microchannels are fabricated with high fidelity. This is very difficult to achieve with micromachining techniques, and therefore replicate molding is preferable. As discussed in Section 4.4, the design of an emulsification device includes the intersection of two or more channels, thereby increasing the complexity of modeling and the possibility of destabilizing events. Figure 14.9(c) illustrates a paralyzed system with T-intersection points in each parallel reactor. As discussed, the formation of emulsions results in well-defined variations in pressure with the production of each droplet from each emulsification compartment. This can result in repeated pressure spikes with frequencies in the kHz range emanating from each emulsion point. This causes “cross-talk” throughout the rest of the fluidic circuit and results in a complicated superposition of pressure spikes, ultimately increasing droplet polydispersity. Therefore, the second rule to numbering up for parallel synthesis involves decoupling liquid streams from each other by increasing the hydrodynamic resistance of both the monomer and continuous stream phases prior to emulsification. In Fig. 14.9(c) the hydrodynamic resistances of the channel segment between the upstream branching

14 Microfluidics for synthesis and biological functional materials |

iT

R1

i1

R2

i2

Vi

Q1

QT

iT

pi

Vo R3

399

RH1

RH2 Q2

QT po

RH3

i3

Q3

(b)

(a)

RH1a

RH1b p1

RH2a

RH2b p2

pi

po

RH3a

RH3b p3

(c) Fig. 14.9. (a) A parallel circuit diagram modeling the fluidic circuit of a parallelized MF device. Electrical current iT is supplied to the inlet side of the circuit (left) and is split into i1 , i2 , i3 over 3 parallel branches. Current flow due to a driving voltage, which is the difference between Vi and Vo , and is applied against resistances R1 , R2 , and R3 in each parallel arm. Conservation of charge implies that i1 + i2 + i3 = iT , and that electrical current flowing into the circuit must flow out. Outlet voltage (Vo ) can be calculated using iT in Ohm’s law, and by reducing the parallel resistor to a single equivalent resistance using 1/Req = 1/R1 + 1/R2 + 1/R3 . (b) Fluidic circuit with three parallel arms with hydrodynamic resistances of RH1 , RH2 , and RH3 . Total flow rate QT is supplied to the channels by an external pressure differential (not shown) and is split into flow rates Q1 , Q2 , and Q3 , through 3 three parallel branches due to the pressure differential between pi and po . Conservation of mass implies that Q1 + Q2 + Q3 = QT , and that flow rate into and out of this segment of the fluidic circuit are the same. Outlet voltage (Vo ) can be calculated using QT in Hagen-Poiseuille law and by reducing the parallel hydraulic resistance to a single equivalent resistance using 1/Qeq = 1/Q1 + 1/Q2 + 1/Q3 . (c) Parallel microemulsification units featuring the dispersed phase (gray) intersecting the continuous phase (white) in a T-junction geometry. Not shown are the resistance values associated with the common inlet and outlet sections in (a), (b), and (c).

point and the emulsification points in the continuous-phase streams are denoted by RH1a , RH2a , and RH3a . Not shown, but equally important to consider, are the hydrodynamic resistances of the segments between the branching point and the T-junction in the monomer-phase channels. In cases where a channel becomes blocked or its dimensions are significantly changed due to material deposition on the channel walls, a monolithic reactor will have to be completely de-commissioned for cleaning or replacement. Therefore, the third rule involves a strategy to implement modular synthesis platforms so that malfunctioning components can be removed without impacting the rest of the system.

400 | Part III New trends in sustainable development and biomedical applications Using pressure-driven flow, it is possible to disconnect one module while the others continue to operate. Figure 14.10 shows a multiple module MF (M3 ) system fabricated in PDMS [88]. Emulsification in this system is accomplished using flow focusing geometry (Fig. 14.10(a)). Each module (Fig. 14.10(b)) contains 16 separate flow focusing devices and the entire system is comprised of 8 separate modules. A module consists of a reactor level and a manifold level. The possibility of droplet coalescence must be avoided before the monomer droplets become sufficiently polymerized. Therefore, the fourth design rule requires that (a) monomer emulsions be quickly polymerized after formation, and that (b) the recombination points are designed to prevent droplet/droplet interaction. Figure 14.10(c) shows the M3 system used in photopolymerization mode, whereby emulsified monomer droplets formed in all 8 modules are immediately irradiated by UV light. Figure 14.10(d) shows that the formation of monomer droplets occurs between them with well-defined spacing. This behavior continues into the first recombination point (Fig. 14.10(e)), because the droplets and continuous-phase liquids do not change their velocity after recombination. However, after the second (Fig. 14.10(f)) and third recombination points (Fig. 14.10(g)), the change in volume of the downstream channel causes a reduction in velocity and collisions can occur. This behavior can be minimized by ensuring the sum of the cross-sectional area of all inlets and the outlet at each recombination point is always the same. In addition, monomer droplets should be immediately exposed to UV radiation (or another polymer initiation process) after formation so that they are solidified before reaching convergence points, yielding monodispersed particles (h). If solidification happens too slowly, or is initiated off-chip, the particles will coalesce and be polydispersed (i).

14.5 Microfluidics for synthesis of functional nanoparticles Nanoparticles (NPs) are an important emerging class of materials with applications in medicine, optics, energy, electronics, sensing and consumer products. Their versatility is enhanced because of their very high surface area to volume ratio, meaning that chemical modification of their surface can provide control over their properties. From the perspective of drug delivery, changes to surface chemistry have been exploited to target, prevent or control aggregation, and for drug release. NPs can also have coreshell or porous architectures, which has been exploited as caches for drug payloads. They are also used as sensitivity enhancement agents for vibrational spectroscopy and for visualization using fluorescence and magnetic resonance imaging tools. One of the reasons NPs have been slow to find commercial applications results from inconsistent batch-to-batch properties and slow approaches to their characterization. As discussed below, MF offer an exciting opportunity for high quality NP synthesis.

14 Microfluidics for synthesis and biological functional materials |

(d)

(a) (a)

401

(e)

B A B

a–1

a–1 Droplet generator

Polymerization compartment

a–2

(b) Inlet A (f) Inlet B

(g)

a–1–2–3–4

a–1–2 Outlet C a–3–4

(c) UV irradiation

(h)

(i)

Fig. 14.10. Schematics of a multiple modular MF (M3 ) reactor. (a) A single reactor comprising of flow focusing emulsification and polymerization compartments. (b) A single module containing 16 individual reactors described in (a) connected by liquid distribution manifolds. (c) The entire M3 reactor consisting of 8 separate modules described in (b). (d) Monomer droplets immediately after generation from a single reactor. (e–g) Microscope images of droplets emerging from two, four, and eight droplet generators respectively. (h) Particles are monodispersed following polymerization within the M3 system. (i) Particles are polydispersed following polymerization off-chip. All scale bars are 500 mm. (d–i) reprinted with permission from [88].

14.5.1 Microfluidics for highly controlled nanoparticle synthesis Synthesis of NPs with tunable, highly monodispered size distributions is critical due to the tight correlation between NP size and function. In this respect, MF synthesis of NPs is powerful due to the ability to precisely control reaction conditions and reaction times [91, 92]. In addition, MF offer the opportunity for on-line characterization and real time optimization [93]. Real-time optimization involves the control of synthesis variables, in situ characterization, and real-time optimization algorithms. Readers are directed to a very good overview of these concepts in the synthesis of CdSe NPs using an automated MF system described extensively elsewhere [94]. Real-time optimization can enable point-of-use synthesis of high quality NPs. This is a common goal in the field because NPs are known to rapidly change their properties, leading to problems associated with centralized production/shipping models for NP manufacture.

402 | Part III New trends in sustainable development and biomedical applications For example, changes to surface charge can result in loss of colloidal stability and aggregation. Therefore, fresh and locally synthesized NPs using a low cost MF synthesis system would be highly advantageous. High quality inorganic NPs require rapid nucleation and growth in well-specified reaction conditions. The reaction conditions in bulk often suffer from temperature and concentration gradients, as well as from poorly controlled reaction times compared to MF synthesis environments [95–97]. As seen in Fig. 14.11, the stronger control of the reaction environment in MF channels can result in NPs with well-defined sizes and absorption properties and the ability to tune their properties by changing the reaction time [98]. In addition to addressing these problems, MF have been used to synthesize high quality inorganic NPs and inorganic core-shell NPs (e.g., CdS, and CdS/CdSe) in an aqueous solution by implementing on-chip quenching steps with millisecond time resolution by the addition of reagents at different locations downstream of the initial point of mixing [99]. This process generated NPs with well-defined optical properties. In the same work, the authors addressed the problem of NP aggregation on microchannel walls by conducting synthesis in water plugs, which were shielded from the hydrophobic PDMS walls by a thin layer of non-polar continuous-phase liquid as discussed in Section 4.3. Disadvantages of NP synthesis in MF channels include limited solvent compatibilities and temperature resistance for polymer fabrication materials. However, silicon, glass and ceramic devices can extend the range of compatible chemical reaction environments for MF synthesis.

14.6 Biomaterials

(a)

1.0 0.8 0.6 0.4 0.2 0.0 250

16

On chip Benchtop

300

350 400 λ(nm)

Normalized fluorescence

Absorbance

Increasingly, microscale biomaterials are servicing sophisticated life-science and technology applications including tissue engineering, drug delivery, and even energy production. Microfluidic platforms have been demonstrated to be efficient for both

450

12 8

3 min 10 min 20 min

4 0 420 470 520 570 620 670 720 λ (nm)

500 (b)

Fig. 14.11. Improvements in NP products using MF synthesis. (a) Sharp vs. broad absorption peak, in MF and bulk synthesis, respectively. (b) Precise timing of reaction for MF synthesis enables excellent tunabilty of NP optical properties. Reprinted with permission from [98].

14 Microfluidics for synthesis and biological functional materials |

403

the synthesis and study of highly functional biomaterials. They are also natural platforms for the development of miniaturized devices. With the help of MF technology and other peripheral tools, researchers have been able to precisely determine biomaterial properties such as chemical composition, mechanical properties, physical dimensions, and porosity, to name a few. In this section we review some areas where MF is poised to make important advances in the synthesis of functional microscale biomaterials.

14.6.1 Tissue engineering and membranes Tissue engineering is an area where MFs is anticipated to make important impacts. For example, the growth of biologically relevant membranes has been demonstrated due to the ability to form stable chemical gradients. Researchers have mixed chitosan with a low pH aqueous solution which protonated amide groups, thereby enhancing the polarity of the molecule, rendering it soluble (Fig. 14.12) [85]. The chitosan solution was then passed through an “X channel”, where it co-flowed beside another aqueous solution of high pH. The absence of turbulence in the microchannel supported a very sharp liquid/liquid interface, where a strong pH gradient was established. The chitosan molecules became insoluble after deprotonation in this region. Selective precipitation at this location caused the growth of a film. The thickness of the film grew reproducibly from microns in size to nearly 100 μm in 10 minutes and could therefore be controlled. Furthermore, the semi-permeable membranes featured pore sizes in the nm range, similar to the size of antibodies. In another study, an electrochemically generated OH− concentration near a cathode resulted in chitosan film deposition at the cathode/solution interface within an MF device [100–102]. The device enabled studies which resolved the electrogelling mechanism of chitosan, helped to determine the dominant factors driving deposition, and characterized the density distribution within the resulting hydrogel. Other work using MF-based devices for chitosan deposition have investigated factors affecting chemical adhesion of chitosan and strategies for its modification. Further studies using mass transport control in 3D tissue cultures have been made with the MF platform giving precise control of the Péclet number (ratio of convective to diffusive transport) over a range of nearly five orders of magnitude [103]. The ability to synthesize tissue membranes and reproduce critical functionality using MF has resulted in a wave of advancements in tissue engineering. For example, an MF model of the human lung was demonstrated for the study of cellular response to foreign particles and pathogenic bacteria (Section 6.4) [104, 105]. The tissue model consisted of a membrane formed from a synthetic elastomeric film patterned with an array of oversized (15 μm) pores, onto which cells were cultured. The model introduced 2D pulmonary tissue stretching as an approximation of the 3D stretching motion in the human lung. Using this model, it was determined that tissue stretching was critical in

404 | Part III New trends in sustainable development and biomedical applications Low pH, soluble + OH HO NH3 O O O O HO +NH3 OH

pKa=6.3 n

High pH, insoluble OH HO NH2 O O + 2n H+ O O NH HO 2 OH n

(a) Chitosan (pH=5) pH=6.3

Chitosan Aperture

Membrane Buffer Buffer (pH=10)

(b)

(c)

Fig. 14.12. (a) Schematic for the conversion of chitosan from water-soluble to water-insoluble using a pH responsive system. (b) Close-up and 3D views of the pH gradient area where the membrane is formed. Reprinted with permission from [85].

replicating NP crossings into the blood stream observed in animal models, while enabling easier characterization and reducing ethical concerns regarding this type of study. Other examples of the synthesis of in vitro tissue models with organ-level functionality include blood vessels [106–108], muscles [109], bones [110], airways [111], liver tissue [112–114], brain components [115, 116], the gut [117, 118], kidneys [119, 120], and others [121, 122]. Complementary to tissue engineering, real biological samples can be extracted from the body and integrated into an MF platform in order to observe their behavior under precise hydrodynamic conditions. For example, an MF device was developed with vacuum microchannels for on-chip fixation of mouse arterial segments. The samples were cultured under a well-defined chemical environment and subjected to a drug solution for contraction-expansion measurements (Fig. 14.13) [108]. This work featured an impressive device design, which included on-chip temperature control, pressure channels, and fully automated acquisition of up to ten dose sequences, which were followed with real-time microscopic inspection of the changes to the physical dimensions of the arterial segment.

14.6.2 Microenvironments for encapsulated cells Typically, cell cultures are conducted on two-dimensional surfaces within Petri dishes or in multi-well plates. However, in their native environments, the majority of cells are exposed to three-dimensional environments. The rise of MF as a platform for the

14 Microfluidics for synthesis and biological functional materials |

(a)

(b)

405

(c)

Fig. 14.13. An artery on a chip. (a) Artery segment, consisting of endothelial cells on the inner side, and smooth muscle cells on the exterior. Pressurized flow stream is directed through the artery (green). (b) Channel geometry in the vicinity of the artery. Yellow channels are vacuum channels used for artery fixation and red channels supply the drug solution to the exterior wall of the artery segment. (c) The entire planar MF device which enables microscopic inspection of the artery segment. Figure reprinted with permission from [108].

synthesis of three-dimensional microenvironments has been important in providing enhanced control over forces imposed on cells, facilitating visualization, easing combination or reconfiguration of the cell-laden droplets or hydrogels, and enabling the control over transport of oxygen, nutrients, growth factors, and waste [123]. Cell encapsulation by aqueous droplets in a non-polar continuous phase (usually a mineral, vegetable, or fluorinated oil) can be achieved using one of the MF channel geometries discussed in Section 4.4. Encapsulation in hydrogels is advantageous because it offers the possibility of transferring to an aqueous culture phase, in addition to the ability to vary the chemical and physical properties of the substrate to mimic the natural microenvironment. Chemical gelation consists of crosslinking and polymerization. This is ideally not carried out using UV photoinitiation due to the stresses it places on the cells. Other biomaterials used for encapsulation include synthetic and biopolymers. Redox polymerization of synthetic polymers in mild conditions is a promising alternative to UV photoinitiated reactions. For example, yeast cell-laden droplets containing low concentrations of glycerol decaacrylate and ethylene glycol diacrylate were transformed into hydrogels due to the redox initiation from ammonium persulfate [124]. Compared to other chemical gelled approaches, this method resulted in cell viability of 30% after overnight polymerization. Alternatively, non-synthetic biopolymers such as proteins and polysaccharides have been used for MF cell encapsulation, because they are biocompatible and can form gels under mild conditions. For example, the polysaccharide alginate can achieve gelation via coordination of the carboxylic acid groups with divalent ions such as Ca2+ . This is often accomplished by introducing the droplets containing alginate into a solution of CaCl2 via a downstream channel. The large diffusivity of the small cations results in rapid gelation and rapid increases in viscosity. In order to prevent related problems, low concentration alginate solutions must be used, which is not ideal. A solution to this problem is internal gelation, whereby CaCO3 NPs are added to the alginate precursor droplets, which are triggered to dis-

406 | Part III New trends in sustainable development and biomedical applications solve and release Ca2+ by decreasing the pH of the surrounding solution [125]. This resulted in better control over gelation when compared to external gelation, and in turn resulted in a more narrow size distribution of hydrogels. Cell viability also increased to nearly 75%, which was attributed, in part, to the buffering effect of the carbonate ions. Agarose is another polysaccharide which has been used for on-chip encapsulation. Agarose can be triggered to gel by reducing the temperature below 20°C, but it can maintain its gelled state even when warmed to a physiological temperature of 37°C [126, 127]. Other polysaccharides which have been used for MF cell encapsulation include pectin and chitosan. Proteins used for gelation include gelatin, which gels via a cold-setting mechanism but maintains is gel form at 37°C. Puramatrix™ is a mixture of 16 peptides which gel upon exposure to solutions containing salt. The MF synthesis of Puramatrix™ gels containing bovine carotid artery endothelial cells displayed excellent cell viability, with 93% of cells demonstrating growth and movement within the gel. One interesting potential application of cellular microenvironments has been the effect of the mechanical properties of extracellular matrix on cells. The elastic modulus in tissues in the human body ranges from 0.1 kPa in the brain to 40 kPa in the osteoid matrix, which can, for example, guide the differentiation of stem cells. A proofof-principle MF platform was developed which enabled the high-throughput synthesis of nearly monodispersed microgels with elastic moduli ranging from 15 Pa to 520 Pa and the loading of the microgels with murine embryonic stem cells, studied by microscopy. The study of intercellular effects through quorum sensing is another area where MF can play a role. By controlling the relative flow rates of the cell-laden dispersed phase and the continuous-phase it is possible to control both the size of the emulsified microenvironment and the number of cells it contains. It is not possible to fully control cell loading, as this is a stochastic process, with loading being determined by Poisson statistics. Therefore, MF encapsulation in this manner can control the average number of cells per droplet by varying the concentration of cells in the feed supply or the size of the droplet. However, there will always be a statistical distribution in cell loading, necessitating post encapsulation fractionation, by flow cytometry, for example.

14.6.3 Biofilms Here we highlight a new area of functional biomaterial development: using cultured biofilms in microchannels. Biofilm (BF) growth occurs due to the transformation of nutrient molecules into biopolymer building blocks which can adhere to surfaces or form free-standing structures. Biofilms are an aggregation of microbes living within a self-produced protective matrix of natural polymers called an extracellular polymeric substance (EPS), which protects from detachment and against harsh chemical environments. Many types of bacteria form BFs, and natural BFs are usually com-

14 Microfluidics for synthesis and biological functional materials |

407

prised of a community of bacteria as well as other life forms. The typical thickness of a BF is between 10 and 1 000 μm, thus on the same scale as MF channels themselves. Biofilms are initiated in a microchannel by bacterial adhesion to the wall followed by aggregation into colonies. After this aggregation, they begin to produce EPS until the film forms a 3D structure in which the bacteria can live. Important factors affecting the formation and proliferation of BFs are known to include hydrodynamic shear force, temperature, nutrient concentration, and quorum sensing signals. The ability to apply relevant external stimuli to an MF device while supporting advanced characterization makes MF a promising tool for the study of BFs. For example, recently flow-templated nutrient streams enabled micropatterening of BFs on a single microfluidic wall. Authors were able to isolate growth to a segment of the microchannel that experienced uniform shear stress, thereby enabling well-specified growth conditions and highly qualifiable growth kinetic measurements [62, 128]. Tools for studying BFs in microchannels include optical methods, electrochemical measurements, and mechanical techniques [129–131]. The use of MF for the formation and long-term culture of BF has been demonstrated in studies of BF properties such as Young’s modulus, cell morphology, adhesion and proliferation, as well as diffusivity [132]. With a better understanding and tools to control BF properties, the door is open for applications including energy production, water remediation and catalysis.

14.6.4 Microdevices utilizing functional biomaterials Microbial fuel cells In addition to being an excellent platform for the study of biomaterials, MF serve as an ideal format for miniaturized devices. To exemplify this application, we briefly consider microbial fuel cells (MFCs), which have the potential of purifying waste water while producing energy. Typical MFCs use BFs adhered to an electrode surface to catalyze the breakdown of a broad range of molecules into an oxygen-deprived solution via an oxidative pathway. The result is the generation of electrons, which flow out of the MFC via the electrode. For example, as seen in Fig. 14.14, a molecule of sucrose enters the BF and is oxidised by a bacteria, resulting in the generation of protons, CO2 and electrons. The protons will travel to the cathode where they complete the reaction by being reduced in the presence of O2 to form water. One of the main benefits of an MF-based MFC is that the proton exchange membrane, which separates the anolyte liquid from the catholyte liquid, is not necessary, since the diffusion-limited environment in the microchannel strongly limits mixing [133]. Moreover, the ability to cultivate, study, and optimize BFs for MFCs has been conducted by screening bacterial consortia [130]. Although some evidence points to the possibility of enhancing electrical output by tuning the shear force applied to the BF, this effect needs to be conducted in microchannels in order to be studied more carefully [134].

408 | Part III New trends in sustainable development and biomedical applications 48e‒

48e‒ 12O2

C12H22O11

24H2O

O2 saturated solution

48H+ 12CO 2 Sucrose solution

48e‒ Fig. 14.14. Schematic of the critical components of a microfluidic MFC. A biofilm (brown) is attached to an anode (cross-hatched). A sucrose molecule enters the biofilm from the solution phase and is oxidized by a bacterium, producing electrons (which travel through an external circuit before entering the cathode), protons (which diffuse away from the anode toward the cathode and participate in O2 reduction), and CO2 . The laminar flow environment results in the separation of the anolyte and catholyte solutions without the need for a proton exchange membrane, thereby reducing internal resistance.

Clinical diagnostics Clinical diagnostics is one of the strongest drivers for MF development. Tissue development using MF is inspiring new diagnostic tools which are inexpensive, robust, accurate, and furthermore can improve clinical phase trials for new in vitro technologies. For example, rapid development of NP technology has been accompanied by enhanced production and utilization of new nanomaterials by industry, with relatively little research into the effect on humans and other life forms. Nevertheless, the rapid expansion of NP synthesis has resulted in the unprecedented – and rapidly growing – exposure of workers and the general public to these new materials. The effect of NPs on human health raises several important questions related to long-term storage and toxicity. At the same time, NPs are being purposefully introduced to the body for new medical purposes such as drug carriers, targeting, and imaging. In any case, rapid clinical testing techniques for NPs are needed to keep pace with their development. The corresponding control of the hydrodynamic environment of the confined flow enables more realistic pre-clinical trials of NP functionality in the human body [135–137], and can accelerate the clinical translation of NPs [98]. Typically, in vitro pre-clinical testing is conducted on cells cultured at the bottom of a culture plate. The two-dimensional environment is static and leads to gravitationally-driven NP sedimentation, resulting in tests far from in vivo conditions. On the other hand, new MF-based in vitro platforms lead to strong control of cell culture environments which more closely mimic physiological geometries and organlevel functionality. For example, Figure 15 shows the functional elements of a “lung-

14 Microfluidics for synthesis and biological functional materials |

409

on-a-chip”, which replicates the mechanical and cellular properties of the tissue that separates the alveolar air space and blood capillaries in the lungs [104]. A porous membrane was coated with an extracellular matrix material on which epithelial cells were cultured on the air side and endothelium cells on the blood side. The resulting membrane model was periodically stretched using two vacuum channels on either side to replicate the regular mechanical strain experienced in vivo during breathing. This platform was used to model the translocation of inhaled NPs across pulmonary

Epithelium

Air

Endothelium Membrane Nanoparticles

Flow

Side chambers (b)

(a) 6

10% strain No strain Transwell

% translocation

5

Gut epithelium

4

Porous membrane

3 2 Vacuum chamber

1

Vacuum chamber

0 0

1

2

3

4

Time (hr) (c)

(d)

Fig. 14.15. Organs on chips. (a) Schematic of a “lung-on-a-chip” made of elastomeric material. A tissue model was formed by suspending a thin porous elastomeric film across the channel which was coated in situ by epithelial cells on the air side of the lung and endothelial cells on the blood side. Two side chambers were periodically depressurized to induce lateral stretching motion. (b) Image of an alveolar air space surrounded by blood capillaries containing red blood cells. (c) Introduction of NPs to the epithelial layer causes response to the cell layers and NP translocation into the bloodstream at a higher rate than for traditional membrane suspension multiwell plates or MF platforms with static membranes. (d) A schematic for a gut-on-a-chip device with a porous membrane coated with endothelium on both sides with similar capabilities. Figures (a) and (b) reprinted with permission from [104]. Figures (c) and (d) reprinted with permission from [118].

410 | Part III New trends in sustainable development and biomedical applications tissue (Fig. 14.15). The dynamic stretching motion of membranes was a key feature which resulted in higher NP translocation into the bloodstream and a stronger inflammatory response due to toxins. This key feature, along with realistic blood flow, is not possible in static culture plate in vitro and can give a more realistic assessment of hazards associated with exposure to airborne NPs. A similar platform has been developed for a ‘gut-on-a-chip’, which mimics the intestinal wall, thereby enabling uptake of ingested NPs [118].

14.7 Summary In this chapter a range of important issues regarding the synthesis of synthetic and biological functional materials at the nano- and microscale using MFs have been reviewed. We began Section 2 with a categorization of different types of microreactors and the materials and fabrication methods to make microreactors. In Section 3 we discussed the special properties of liquids flowing through microchannels and how to manipulate and monitor reaction solutions. In Section 4 we discussed how MFs can be used to synthesize polymer particles, 1D threads and 2D microsurfaces. This discussion included relevant concepts in MF channel geometries, their surface properties and macroscopic properties of the precursor liquids. Section 5 discussed synthesis of NPs in microchannels. Finally, Section 6 concluded with state-of-the-art applications of MFs in the synthesis and study of microscale functional biomaterials, such as tissue engineering, cellular microenvironments, and BFs. We finished the chapter with two illustrative examples of miniaturized devices that use MF with functional biomaterial components: microscale microbial fuel cells and devices for clinical diagnostics related to exposure to NP materials. This chapter covered many MF applications, as well as a diverse range of practical issues regarding MFs.

Acknowledgments JG thanks Ms Nahid Babaei Aznaveh for the image in Fig. 14.1(d) and Mr Mohamed Larbi Gharib for the image in Fig. 14.8(a).

References [1] [2] [3] [4] [5]

B.P. Mason, K.E. Price, J.L. Steinbacher, A.R. Bogdan, D.T. McQuade, Chem. Rev., 107,(2007) 2300. M. Ooms, L. Bajin, D. Sinton, Appl. Phys. Lett. 101, (2012) 253701. F. Paquet-Mercier, N. Babaei Aznaveh, M. Safdar, J. Greener, Sensors, 2013. D. Cai, A. Neyer, Microfluid. and Nanofluid., 9, (2010) 855. L. Derzsi, P. Jankowski, W. Lisowski, P. Garstecki, Lab Chip, 11, (2011) 1151.

14 Microfluidics for synthesis and biological functional materials |

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41]

411

P. Jankowski , D. Ogończyk, W. Lisowski, P. Garstecki, Lab Chip, 2012, 12, 2580. P. Jankowski, D. Ogonczyk, A. Kosinski, W. Lisowski, P. Garstecki, Lab Chip, 11, (2011) 748. C. Iliescu, H. Taylor, M. Avram, J. Miao, S. Franssila, Biomicrofluidics, 6 (2012) 016505. K. Hermansson, U. Lindberg, B. Hok, G. Palmskog, Solid-State Sensors and Actuators, Digest of Technical Papers, TRANSDUCERS , Wetting properties of silicon surfaces, (1991) 193–196. S. Gómez-de Pedro, M. Puyol, J. Alonso-Chamarro, Nanotech., 21, (2010) 415603. Z.M. da Rocha, C.S. Martinez-Cisneros, A.C. Seabra, F. Valdés, M.R. Gongora-Rubio, J. AlonsoChamarro Lab Chip, 12, (2012) 109. J. Greener, W. Li, J. Ren, D. Voicu, V. Pakharenko, T. Tang, E. Kumacheva, Lab Chip, 10, (2010) 522–524. Y. Xia, G. M. Whitesides, Annu. Rev. Mater. Sci., 28, (1998) 153. J. Huft, D. J. Da Costa , D. Walker, C. L. Hansen Lab Chip, 10, (2010) 2358. J. Steigert, S. Haeberle, T. Brenner, C. Muller, C. P. Steinert, P Koltay, N. Gottschlich, H. Reinecke, J. Ruhe, R. Zengerle, J. Ducree, J. Micromech. Microeng., 17, (2007) 333. J. S. Kuo, L. Y. Ng, G. S. Yen, R. M. Lorenz, P. G. Schiro, J. S. Edgar, Y. X. Zhao, D. S. W. Lim, P. B. Allen, G. D. M. Jeffries and D. T. Chiu, Lab Chip, 9, (2009) 870. S.Y. Chou, P. R. Krauss, P. J. Renstrom, Science, 272, (1996) 85. T. Schaller, L. Bohn, J. Mayer and K. Schubert, Precision Eng., 23, (1999) 229. P. P. Shiu, G. K. Knopf, M. Ostojic and S. Nikumb, J. Micromech. Microeng., 18, (2008) 025012. T. Nielsen, D. Nilsson, F. Bundgaard, P. Shi, P. Szabo, O. Geschke, A. Kristensen, J. Vac. Sci. Technol. B, 22, (2004) 1770. Y. Fan, T. Li, W-M. Lau, J. Yang, J. Microelectromec. Sys., 21, (2012) 875. I. Rodriguez, P. Spicar-Mihalic, C. L. Kuyper, G. S. Fiorini, D. T. Chiu, Anal. Chim. Acta, 496, (2003) 205. A. Grosse, M. Grewe, H. Fouckhardt, J. Micromech. Microeng., 11, (2001) 257. L. Ceriotti, K. Weible, N. F. de Rooij, E. Verpoorte, Microelectron. Eng., 67–8, (2003) 865. C. R. Friedrich, M. J. Vasile, J. Microelectromech Syst., 5, (1996) 33. J. S. Mecomber, D. Hurd and P. A. Limbach, Int. J. Mach. Tools Manuf., 45, (2005) 1542. M. A. Roberts, J. S. Rossier, P. Bercier and H. Girault, Anal. Chem., 69, (1997) 2035. A. Mohammed Adham,N. Mohd-Ghazali, R. Ahmad, Renew. Sust. Energ. Rev., 21, (2013) 614. J. Greener, E. Tumarkin, M. Debono, C-H Kwan, M. Abolhasani, A. Guenther, E. Kumacheva, Analyst, 137, (2012) 444. D. Voicu, C. Scholl, W. Li, D. Jagadeesan, I. Nasimova, J. Greener, E. Kumacheva, Macromolecules, 45, (2012) 4469. J. Greener, E. Tumarkin, M. Debono, A.P. Dicks, E. Kumacheva, Lab Chip.12, (2012) 696. Y. K. Suh, S. Kang, Micromachines, 1, (2010) 82. S. K. W. Dertinger, D. T. Chiu , N. L.i Jeon, G. M. Whitesides, Anal. Chem., 73, (2001) 1240. Š. Selimović, W. Y. Sim, S. B. Kim, Y. H. Jang, W. G. Lee, M. Khabiry, H. Bae, S. Jambovane, J. Wook Hong, A. Khademhosseini, Anal. Chem., 83, (2011) 2020. E. Cimetta, C. Cannizzaro, R. James, T. Biechele, R. T. Moon, N. Elvassore, G. VunjakNovakovic, Lab Chip, 10, (2010) 327. A. D. Stroock, S. K. W. Dertinger, A. Ajdari, I. Mezić, H. A. Stone, G. M. Whitesides, Science, 295, (2002) 647. F. Ballya, C. A. Serra, V. Hessel, G. Hadziioannou, Chem. Eng. Sci., 2011, 66, 1449. C. J. Campbell, B. A. Grzybowski, Phil. Trans. R. Soc. Lond. A, 362, (2004) 1069. T. Iwasaki, J. I. Yoshida, Macromol., 2005, 38, 1159. C. Rosenfeld, C. Serra, G. Hadziioannou, Lab Chip, 8, (2008) 1682. H. B. Mao, T. L. Yang and P. S. Cremer, J. Am. Chem. Soc., 124, (2002) 4432.

412 | Part III New trends in sustainable development and biomedical applications [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54]

[55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76]

N. L. Jeon, H. Baskaran, S. K. W. Dertinger, G. M. Whitesides, L. Van de Water and M. Toner, Nat. Biotechnol., 20, (2002) 826. C. R. Cabrera, B. Finlayson and P. Yager, Anal. Chem., 73, (2001) 658. J. I. Park, Z. Nie, A. Kumachev, A. I. Abdelrahman, B. R. Binks, H. A. Stone and E. Kumacheva, Angew. Chem., Int. Ed., 48, (2009) 5300–5304. C. B. Elias and J. B. Joshi, Adv. Biochem. Eng. Biot., 59, (1998) 47. J. Wu, G. Zheng, L. M. Le, Lab Chip, 12, (2012) 3566. A. Günthera, K. F. Jensen, Lab Chip, 6, (2006) 1487. X.C. Solvas, X. Niu, K. Leeper, S. Cho, S.I. Chang, J. B. Edel, et al., J. Vis. Exp., 58, (2011) e3437. M. Paturzo, A. Finizio, P. Memmolo, R. Puglisi, D. Balduzzi, A. Galli, P. Ferraro, Lab Chip, 12, (2012) 3073–3076. F. Holtmann, M. Eversloh, C. Denz, J. Opt. A: Pure Appl. Opt., 11, (2009) 034014. P. Panorchan, D. Wirtz, Y. Tseng, Phys. Rev. E Stat. Nonlin. Soft. Matter. Phys., 70, (2004) 041906. E. L. May, A. C. Hillier, Anal. Chem., 77, (2005) 6487–6493. J. Yue, J. C. Schouten, T. Alexander Nijhuis, Ind. Eng. Chem. Res., 51, (2012) 14583. J. Bart, A. J. Kolkman, A. J. Oosthoek-de Vries, K. Koch, P. J. Nieuwland, H. Janssen, P. J. M. van Bentum, K. A. M. Ampt, F. Rutjes, S. S. Wijmenga, H. Gardeniers, A. P. M. Kentgens, J. Am. Chem. Soc., 131, (2009) 5014. M. Utz and R. Monazami, J. Magn. Reson., 198, (2009) 132. G. Ryu, J. Huang, O. Hofmann, C.A. Walshe, J.Y.Y. Sze, G. D. McClean, A. Mosley, S.J. Rattle, J.C. deMello, A.J. deMello, D.D.C. Bradley, Lab Chip, 11, (2011) 1664. J. Greener, B. Abbasi and E. Kumacheva, Lab Chip, 10, (2010) 1561–1566. K.L.A. Chan, S. Gulati, J.B. Edel, A. J. de Mello, S.G. Kazarian, Lab Chip, 9, (2009) 2909. S. E. Barnes, Z. T. Cygan, J. K. Yates, K. L. Beers, E. J. Amis, Analyst, 131, (2006) 1027. M. P. Cecchini, J. Hong, C. Lim, J. Choo, T. Albrecht, A. J. deMello and J. B. Edel, Anal. Chem., 83, (2011) 3076–3081. K.L.A. Chan, S.G. Kazarian, Anal. Chem., 84, (2012) 4052. F. Paquet-Mercier, N. Babaei Aznaveh, M. Safdar, J. Greener, Sensors, 2013. A. I. Norman, Q. Zhang, K. L. Beers, E. J. Amis, J. Colloid Interf. Sci., 299, (2006) 580–588. Y. Hua, A.B. Jemere, J. Dragoljic, D.J. Harrison, Lab Chip, 13, (2013) 2651. D. Dendukuri, P.S. Doyle, Adv. Mater., 21, (2009) 1. S.Q. Xu, Z.H. Nie, M. Seo, P.C. Lewis., E. Kumacheva, H.A. Stone, P. Garstecki, D.B. Weibel, I. Gitlin, G.M. Whitesides, Angew. Chem. Int. Ed., 44, (2005) 724. T. M. Squires and S. R. Quake, Rev. Mod. Phys., 77, (2005) 977–1026. Particle Size Characterization, eds. A. Jillavenkatesa, S. J. Dapkunas, L.-S. H. Lum, Special Publication, 2001, 960. G. F. Christopher, S. L. Anna, J. Phys. D: Appl. Phys., 40, (2007) R319. D. Wenzlik, C. Ohm, C. Serra, R. Zentel, Soft Matter, 7, (2011) 2340. T. Nisisako , S. Okushima, T. Torii, Soft Matter, 1, (2005) 23. E.K. Fleischmann, H.L. Liang, N. Kapernaum, F. Giesselmann, J. Lagerwall, R. Zentel, Nat Commun., 3, (2012) 1178. W. Wang, R. Xie, X.-J. Ju, T. Luo, L. Liu, D.A. Weitz, L.-Y. Chu, Lab Chip, ,11, (2011) 1587. A.S. Utada, E. Lorenceau, D.R. Link, P.D. Kaplan, H.A. Stone, D.A. Weitz, Science, 308, (2005) 537. M. Seo, C. Paquet, Z.H. Nie, S.Q. Xu, E. Kumacheva, Soft Mater, 3, (2007) 986–992. J. Champion and S. Mitragotri, Pharm. Res., 26, (2008) 244.

14 Microfluidics for synthesis and biological functional materials |

[77] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115]

413

Y. Geng, P. Dalhaimer, S. Cai, R. Tsai,M. Tewari, T. Minko, D. E. Discher, Nat. Nanotechnol., 2, (2007) 249. A. Duboin, R. Middleton, F. Malloggi, F. Montia, P. Tabeling, Soft Matter, 9, (2013) 3041. D.A. Boyd , A.R. Shields , J. Naciri , F.S. Ligler, ACS Appl. Mater. Interfaces, 5, (2013) 114. C.-H. Choi, H. Yi, S. Hwang, D. A. Weitz, C.-S. Lee, Lab Chip, 11, (2011) 1477. W. Jeong, J. Kim, S. Kim, S. Lee, G. Mensing, D.J. Beebe, Lab Chip., 4, (2004) 576. J.K. Nunes, K. Sadlej, J.I. Tam, H.A. Stone, Lab Chip, 12, (2012) 2301. R. Rusconi, S. Lecuyer, N. Autrusson, L. Guglielmini, H.A. Stone, Biophys J. 100, (2011) 1392. K. Dreschera, Y. Shen, B.L. Bassler, H.A. Stone, Proc. Natl. Acad. Sci. USA, 110, (2013) 4149. X. Luo, D. L. Berlin, J. Betz, G.F F. Payne, W. E. Bentley, G. W. Rubloff, Lab Chip, 10, (2010) 59. T. Nisisako, T. Torii, T. Takahashi and Y. Takizawa, Adv. Mater., 18, (2006) 1152. T. Nisisako and T. Torii, Lab Chip, 8, (2008) 287. W. Li, J. Greener, D. Voicu, E. Kumacheva, Lab Chip, 9, (2009) 2715. K.W. Oh, K. Lee, B. Ahn, E.P. Furlani, Lab Chip, 12, (2012) 515. V. Labrot, M. Schindler, P. Guillot, A. Colin, M. Joanicot, Biomicrofluid., 3, (2009) 012804. S. Marre, K. J. Jensen, K. F. Chem. Soc. Rev., 39, (2010) 1183. C. X. Zhao, L. Z. He, S. Z. Qiao, A. P. J. Middelberg, Chem. Eng. Sci., 66, (2011) 1463. S. G. Pedro, M. Puyol, D. Izquierdo, I. Salinas, J. M. Fuentec, J. Alonso-Chamarro, Nanoscale, 4, (2012) 1328. A.J. deMello, Microfluidic Reactors for Nanomaterial Synthesis, in Advances in Chemical Engineering (Vol. 38), Ed. J.C. Schouten. Elsevier, 2010 B.K. Johnson, R.K. Prud’homme, Phys. Rev. Lett., 91, (2003) 118302. R. Karnik, F. Gu, P. Basto, C. Cannizzaro, L. Dean, W. Kyei-Manu, R. Langer, O. C. Farokhzad, Nano Lett., 8, (2008) 2906. R. Kikkeri, P. Laurino, A. Odedra, P.H. Seeberger, Angew. Chem. Int. Ed., 49, (2010) 2054. P. M. Valencia, O.C. Farokhzad, R. Karnik, R. Langer, Nat. Nanotech., 7, (2012) 623. I. Shestopalov,a J. D. Ticea, R. F. Ismagilov, Lab Chip, 4, (2004) 316. Y. Cheng, X. Luo, J. Betz, S. Buckhout-White, O. Bekdash, G.F. Payne, W.E. Bentley, G. W. Rubloff, Soft Matter, 6, (2010) 3177–3183. Y. Cheng, X. Luo, G. F. Payne, G. W. Rubloff, J. Mater. Chem., 22, (2012) 7659. S. T. Koev, P. H. Dykstra, X. Luo, G. W. Rubloff, W. E. Bentley,G. F. Payne, R. Ghodssi, Lab Chip, 10, (2010) 3026–3042. Y.-H. Hsu, M. L. Moya, P. Abiri, C. C.W. Hughes, S. C. George, A.P. Lee, Lab Chip, 13, (2013) 81–89 D. Huh, B.D. Matthews, A. Mammoto, M. Montoya-Zavala, H. Yuan Hsin, D.E. Ingber, Science, 25, (2010) 1662. S. R. Khetani, S.N. Bhatia, Nat. Biotech., 26, (2008) 120. J. W. Song , W. Gu ,N. Futai, K.A. Warner, J.E. Nor, S. Takayama, Anal. Chem., 77, (2005) 3993. Huh, D., G.A. Hamilton, D.E. Ingber, Trends Cell Biol., 21, (2011) 745–54. A. Guenther, S. Yasotharan, A. Vagaon, C. Lochovsky, S. Pinto, J. Yang, C. Lau, J. VoigtlaenderBolz, S.-S. Bolz, Lab Chip., 10, (2010) 2341. M. T. Lam, Y. C. Huang, R. K. Birla, S. Takayama, Biomaterials, 30, (2009) 1150. K. Jang, K. Sato, K. Igawa, U. I. Chung, T. Kitamori, Anal. Bioanal. Chem., 390, (2008) 825. D. Huh, H. Fujioka, Y.-C. Tung, N. Futai, R. Paine III, J. B. Grotberg, S.Takayama, Proc. Natl. Acad. Sci. USA, 104, (2007) 18886 S. R. Khetani, S. N. Bhatia, Nat. Biotechnol., 26, (2008) 120. P. J. Lee, P. J. Hung, L. P. Lee, Biotechnol. Bioeng., 97, (2007) 1340. Y. C. Toh, T.C. Lim, D. Tai, G. Xiao, D. van Noort, H. Yu, Lab Chip, 9, (2009) 2026. J. W. Park, B. Vahidi, A. M. Taylor, S. W. Rhee, N. L. Jeon, Nat. Protoc., 1, (2006) 2128.

414 | Part III New trends in sustainable development and biomedical applications [116] [117] [118] [119] [120] [121] [122] [123] [124] [125] [126] [127] [128] [129] [130] [131] [132] [133] [134] [135] [136] [137]

R. Baudoin, L. Griscom, M. Monge, C. Legallais, E. Leclerc, Biotechnol. Prog., 23, (2007) 1245. G. J. Mahler, M. B. Esch, R. P. Glahn, M. L. Shuler, Biotechnol. Bioeng., 104, (2009) 193. H.J. Kim, D. Huh, G. Hamilton, D.E. Ingber, Lab Chip, 12, (2012) 2165. R. Baudoin, L. Griscom, M. Monge, C. Legallais, E. Leclerc, Biotechnol. Prog., 23, (2007) 1245. K. J. Jang, K. Y. Suh, Lab Chip, 10, (2010) 36. K. Ziolkowska, R. Kwapiszewski, Z. Brzozka, New J. Chem., 35, (2011) 979. D. Huh, Y.S. Torisawa, G.A. Hamilton, H.J. Kim, D.E. Ingber, Lab Chip, 12, (2012) 2156. D. Velasco, E. Tumarkin, E. Kumacheva, Small, 8, (2012) 1633–1642. D. Steinhilber , S. Seiffert , J. A. Heyman , F. Paulus , D.A. Weitz, R. Haag, Biomaterials, 32, (2011) 1311 W. H. Tan, S. Takeuchi, Adv. Mater., 19, (2007) 2696. A. Kumachev, J. Greener, E. Tumarkin, E. Eiser, P.W. Zandstra, E. Kumacheva, Biomaterials, 32 (2011) 1477. E. Tumarkin, L. Tzadu, E. Csaszar, M. Seo, H. Zhang, A. Lee, R. Peerani, K. Purpura, P.W. Zandstra, E. Kumacheva, Integr. Biol., 3, (2011) 653. N. Babaei Aznaveh, M. Safdar, G. Wolfaardt, J. Greener, Lab Chip, (2014), DOI:10.1039/C4LC00084F. M.T. Meyer, V. Roy, W.E. Bentley, R. GhodssiJ, Micromech. Microeng, 21, (2011) 054023. H. Hou, L. Li, C.Ü. Ceylan, A. Haynes, J. Cope, H.H. Wilkinson, C. Erbay, P. de Figueiredo, A. Han, Lab Chip, 12, (2012) 4151. D.N. Hohne, J.G. Younger, M.J. Solomon, Langmuir, 25, (2009) 7743. E. Bester, G.M. Wolfaardt, N.B. Aznaveh, J. Greener, Biofilms and their role in planktonic cell proliferation, Int. J. Mol. Sci. 2013. Z. Li, Y. Zhang, P.R. LeDuc, K.B Gregory, Biotechnol Bioeng., 108, (2011) 2061. H.T. Pham, N. Boon, K. Verbeken, K. Rabaey, W. Verstraete, Microbiol. Biotech., 1, (2008) 487. H.K. Keith, J.M. Chan, R.D. Kamm, J. Tien, Annu. Rev. Biomed. Eng., 14, (2012) 205. I. Meyvantsson, D.J. Beebe. Annu Rev Anal Chem, 1, (2008) 423. E. C. Cho, Q. Zhang, Y. Xia, Nature Nanotech., 6, (2011) 385.

T. Lefèvre, F. Byette, I. Marcotte, and M. Auger

15 Protein- and peptide-based materials: a source of inspiration for innovation 15.1 Introduction It is now widely recognized that, due to its pronounced and global footprint and its excessive use of materials and energy resources, humanity is the cause of various environmental degradations (land degradation and biodiversity loss, climate change, ocean acidification, water shortages, chemical pollution, etc.) which have reached a point where they affect the functioning of the planetary system itself [1–5]. This situation requires drastic transformation of our socioeconomic system and way of life. Integration of sustainability into human activities is thus one of the main challenges humanity faces. Although the required transformations are essentially sociological, political, and economic, science and technology can greatly contribute to making societies sustainable. There is henceforth an imperative need for novel useful products and devices with a low ecological footprint, made of renewable raw materials, produced by eco-friendly processes and totally recyclable and/or biodegradable. Besides these attributes, products which also combine other attractive and useful properties such as mechanical resistance, optical or electrical characteristics or biological activity are also desired for applications in various fields. Because evolution has solved several environmental and technological problems which humans need to address, nature is becoming a source of inspiration for materials scientists. Living organisms exhibit a variety of organic and hybrid materials (composites) which are self-assembled and display complex hierarchical levels of organization and miniaturization, and possess a broad diversity of functions and properties [6, 7]. A renowned and classic hallmark of biomimicry is the remarkable resistance and architecture of silicic skeletons of unicellular organisms (diatoms). A second concerns silk, a micrometer-sized nanostructured material produced by arthropods which exhibits a combination of strength and extensibility unmatched by any industrial material. In this context, proteins and peptides are promising building blocks for the manufacture of materials, as they encompass in nature a broad variety of functions, structures, and properties. They have the ability to self-assemble into complex architectures, respond to specific environmental stimuli, bind to specific receptors and ligands or resist mechanical stresses [8]. Potential advantages of peptide- and protein-based materials are as follows:

416 | Part III New trends in sustainable development and biomedical applications (i) Raw materials may be obtained in large amounts from renewable resources, either from the biomass (or by-products) [9], or by using biotechnology (recombinant DNA technology). The production of recombinant proteins in bacteria or other expression organisms indeed routinely yields reasonable quantities of proteins with high purity, precise amino acid sequence, and a virtually monodisperse molecular weight [8]. (ii) Polypeptides (proteins and peptides) are biodegradable or bioresorbable and potentially biocompatible. For example, the RGD (arginine-glycine-aspartic acid) motif, the cell-binding domain of fibronectin [10], is indeed commonly used to mediate cell adhesion. (iii) The natural solvent of polypeptides is water, and the processes underlying the production of biological systems occur in mild temperature and pressure conditions, which may potentially make it possible to avoid the use of harmful and environmentally costly organic solvents and physicochemical processes. (iv) As proteins naturally exhibit diverse and hierarchical levels of organization, a vast array of structures organized at different length scales can be considered, thus providing design flexibility. By optimal design of the protein sequence and through control of the aqueous environment, protein-based systems may be selfassembled into hierarchically organized structures, while the process may be directed and (reversibly) triggered upon application of controlled constraints (smart materials). (v) Proteins and peptides are naturally subjected to conformational change and their reversible assembly/disassembly property provides the capacity to build “molecular switches”, i.e., molecular systems which can be reversibly shifted between two or more stable states in response to stimuli such as light, temperature or pH changes. (vi) Another strategy is the production of artificial or chimeric proteins and peptides which contain specific chemical groups intended to fulfil particular functions such as molecule recognition or cell growth, or that are made of a combination of sequences originating from two or more natural proteins in order to integrate and combine different specific properties. (vii)Proteins can be processed in various colloidal and physical states such as films, capsules, gels, emulsions, foams, porous systems, fibers, and non-woven fiber mats. Therefore, many possibilities for harnessing the advantageous characteristics of natural protein materials and systems exist. The potential applications are thus widespread and may contribute to various sectors of society, although medicine seems actually to be the main domain of applications. For example, proteinous matrices can be developed for the effective encapsulation of drugs or bioactive molecules and their delivery at the appropriate site of action, in particular in response to a specific stimulus. They can also be used as scaffolds for 3D cellular growth, migration, and differentiation. These templates may find applications

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

417

in tissue engineering, tissue repair, implants, and wound healing. They may also be useful for biomineralization, the process by which organic–inorganic hybrid materials such as bones, teeth, and shells are constructed, allowing the production of innovative material composites. Polypeptide molecules have great potential for the development of biosensors, in particular for more efficient diagnostics. In the food industry, they may be interesting for new packaging or for gastrointestinal delivery systems of nutrients or bioactive molecules. They may also prove to be advantageous in the areas of membrane filters, textiles, optical fibers, and in nanotechnology in general. As can be seen, proteins and peptides may lead to useful and efficient functional materials in the future. This chapter is devoted to a brief description of natural and synthetic polypeptides, from peptides to globular and fibrous proteins, for the development of functional materials. General principles and important areas of research regarding polypeptide-based materials will be described, as well as their benefits and future applications with the help of several examples. Necessary knowledge regarding some protein characteristics will first be recalled.

15.2 Basics of proteins, peptides and polypeptides Proteins exhibit different levels of structural organization called primary, secondary, tertiary, and quaternary structures. This intrinsic characteristic makes polypeptide molecules potentially interesting building blocks for the production of hierarchically organized and stimuli-responsive materials.

15.2.1 Polypeptides are sequences of amino acids Polypeptide is a generic term for molecules formed by a succession of amino acids. It covers short peptides to large proteins and synthetic homopolypeptides. Peptides are formed by anything from several to ∼ 50 amino acids, for a molecular weight of 1 to 5–6 kDa (1 Da = 1 g/mol). Proteins contain hundreds to thousands of amino acids. They can be divided into three main families: globular, membrane, and fibrous proteins. The first two are constituted of about 100–700 amino acids (10–80 kDa) and fulfil biological functions such as enzymatic activity, catalysis, recognition, and transport of ions or larger molecules. Fibrous proteins are very large biopolymers with a basic structural role in tissues and cells. Synthetic homopolypeptides exhibit the same diversity as polymers [11] and encompass a large range of molecular weights, up to ∼ 300 000 kDa. 20 natural amino acids exist, all with different properties, such as steric hindrance, polarity, charge, hydrophobicity, etc. The polypeptide backbone is constituted of a succession of peptide bonds, each carrying a specific chemical moiety called a residue or side-chain. The composition and linear arrangement of the residues in the

418 | Part III New trends in sustainable development and biomedical applications chain form the so-called primary structure or sequence. It determines the folding and association of the proteins, although the environment also strongly contributes to these behaviors. As a matter of fact, intramolecular and intermolecular interactions depend on pH, temperature, ionic strength, type of salt, solvents, presence of an interface, etc. The diversity of amino acids makes the physicochemical properties of proteins highly diversified.

15.2.2 Polypeptides can adopt various conformations The sequence partly determines the chain conformation, also called secondary structure, in physiological or other conditions. It can be described by two rotational angles around the C–N and C–C bonds, the dihedral angles, which define the angular orientation of the plane formed by the CONH amide groups. Due to steric hindrance, certain secondary structures are more often encountered: α-helix, β-sheet, turns, 31 helix and 310 -helix (Fig. 15.1).

(a)

(b)

(c)

Fig. 15.1. Representation of typical secondary structures: (a) α-helix, (b) β-strand (middle), (c) and 31 -helix or PPII helix.

The α-helix is a widespread structural element, especially in membrane proteins. This helix is right-handed, with 3.6 amino acid residues per turn, and a rise of 1.5 Å per residue. It is stabilized by intramolecular H-bonds along the chain between the carbonyl oxygen atoms of ith residues with the amide hydrogen atoms of the (i + 4)th residues [12]. The β-strand is also a very common secondary structure [13]. It is linear with a rise of ∼ 3.3 Å per residue [12]. It most often associates with one or many β-strands to form β-sheets. Two types of β-sheets exist: antiparallel β-sheets, where the two neighboring chains are aligned in opposite directions, and parallel β-sheets, where the chains are aligned in the same direction. The 310 -helix is right-handed. It has three residues per turn and a translation of 2.0 Å along the helical axis. The N-H group of ith amino acids form an intramolecular

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

419

hydrogen bond with the C=O group of the (i + 3)th amino acid. There are 10 atoms in the ring formed by the hydrogen bond [12]. The 31 -helix, or polyproline II (PPII) helix, is a left-handed helix, with three residues per turn and a rise of ∼ 3.1 Å per residue [14]. Its dihedral angles are close to those of the β-strand. It has no intramolecular H-bonds, but forms H-bonds with water molecules, which appears to be important for structural stabilization [15]. This secondary structure is present for only 2% of amino acids of protein sequences. However, more than half of the proteins contain at least one region of PPII helix which is longer than three amino acids [13]. Due to the chemical structure of their side-chain, each amino acid has particular intrinsic propensities to form the regular secondary structures. Moreover, these propensities are context-dependent, as they are affected by the environment (pH, ionic strength, temperature, etc.) and by neighboring amino acids. Thus, the specific conformation adopted by a given polypeptide depends on a complex combination of various physicochemical parameters. Polypeptide chain segments can lack structure. Some proteins can be even almost entirely unordered in their native state and are thus named “intrinsically unstructured (or disordered) proteins” (IUP or IDP) or “natively disordered proteins” (NDP). Despite their low level of order, these proteins can exhibit biological activity [16–18]. IDPs are generally favored by a combination of high net charge and low hydrophobicity [19]. It is recognized that the PPII helix is frequently observed in this type of proteins [14, 20], suggesting that they retain a certain amount of structural order.

15.2.3 Polypeptides possess various levels of structural organization Tertiary structure, the third level of organization, refers to the spatial arrangement of the secondary structure elements. The organization into secondary and tertiary structures constitutes the folding of the polypeptide chain. Due to the hydrophobic effect, proteins generally tend to expose hydrophilic residues and to segregate apolar ones from the aqueous solvent. For example, several β-sheets can arrange themselves to form barrels with a hydrophobic pocket in the interior. Similarly, α-helices often exhibit a distinct amphiphilicity which can promote their self-association and stabilize the helical structure. α-Helix bundles and “coiled coils” constitute two examples of such assemblies. In the latter case, two or three α-helices are intertwined such that the hydrophobic surfaces are hidden from the aqueous phase [21]. The so-called leucine zipper is a well-known sequence motif which forms a coiled coil [12]. Finally, a last level of protein organization exists, the quaternary structure, which is related to the reversible association of proteins into oligomeric forms (dimers, trimers, etc.) [12]. A distinctive mode of association, widespread among proteins, consists of the self-aggregation of proteins. The microstructure of these aggregates can be very diverse, from well-ordered filaments (protofibrils (monofilaments), straight or rod-like

420 | Part III New trends in sustainable development and biomedical applications fibrils, worm-like fibrils and branched fibrils) to amorphous particles [22–26]. The common point of this phenomenon is the formation of ordered intermolecular β-sheets between polypeptide chains. This β-aggregation can be used advantageously by nature to form fibers such as silk, or it may lead to pathological outcomes as seen in neurodegenerative diseases such as Alzheimer’s and Parkinson’s diseases, type II diabetes, etc. These neurodegenerative disorders, also called amyloidoses, are characterized by the presence of fibrils which form amyloid plaques in tissues. These fibrils are constituted of the assembly of specific proteins into cross β-sheets stacked perpendicularly to the main fibril axis. It has been realized that many proteins unrelated to amyloid pathologies, if not all proteins, can associate and form cross-β fibrils if the conditions are appropriate. The formation of amyloid fibrils thus appears to be an intrinsic property of proteins and peptides [27]. Consequently, this generic and intrinsic aggregation propensity of polypeptides can be used to produce functional materials as will be shown below.

15.3 Functional materials from fibrous proteins Fibrous proteins are a broad family of high molecular weight proteins which play a major role in the organ structure of many living organisms. They are constituted of successions of repeat sequences rich in glycine and proline residues. These repeat sequences are often constituted of small well-defined motifs, which make the sequences of fibrous proteins interesting candidates for the examination of the relationship between sequence and structure. Some general rules driving protein assembly are starting to emerge thanks to studies in this field. Specific sequence motifs are known to be involved in the formation of stiff β-sheets or deformable disordered structures [28]. Moreover, the content of proline and glycine residues seems to determine the existence of elastomeric properties of biological materials and the formation of amyloid fibrils [29]. Each protein adopts a particular architecture in vivo. Collagen, a structural constituent of bones and connective tissues such as skin, tendons, and ligaments, forms triple PPII helices which associate and form fibrils [30]. Elastin, on the other hand, is a rubber-like polymeric protein which exhibits large extensibility under mechanical stress, and almost complete recoil recovery once the stress is removed. This protein can be found in tissues requiring strong elastic behavior such as arteries, lung parenchyma, and ligaments [31]. Similarly, keratin makes coiled coils which assemble to become filaments which form stiff structures like hair, feathers, and nails [32]. A last example is that of actin fibers, an intracellular protein which constitutes the essential component of the contractile apparatus of muscle cells [30]. The diverse examples found in nature provide instruction on how to assemble proteins to innovative functional systems. As it exceeds the scope of this chapter to cover all fibrous proteins

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

421

and properties, the focus will be devoted mainly to resilin/abductin, byssus, and silk, and we will review some recent applications of these proteins as functional materials.

15.3.1 Resilin & abductin Resilin and abductin, two elastic fibrous proteins similar to elastin, are found in the joints of arthropods and hinges of bivalve molluscs. They play an essential role in the abductor ligaments of animals whose physiology requires high strain as well as storage of mechanical energy and its instantaneous and total recovery (the name resilin comes from high resilience, the percentage of recovery after a mechanical deformation). These proteins form filaments and organize as tendons which act as elastic springs against muscle contraction. Resilin, for instance, plays a role in the outstanding jumping capabilities of froghoppers and cat fleas, the wing flapping of dragonflies, and the membrane vibration (vocalization) of cicadas [33]. Abductin allows the swimming scallop Placopecten magellanicus to move in water, thanks to the propulsion created by the rapid and repetitive opening and closing of the shell [34]. The elasticity efficiency and long-term fatigue resistance of these proteins are remarkable. Less than 5% of energy is lost by viscous processes during organ movement [35]. They can extend and relax millions of times without structure alterations, thus constituting almost perfect elastics. The primary structure of resilin and abductin contains repeat motifs such as GGRPSDSYGAPGGGN and FGGMG sequences, respectively. Resilin in particular is rich in tyrosine residues which associate in nature to form di- and tri-tyrosine. Like disulfide bonds in rubber, the covalent bonds seem to play the role of crosslinks in the resilin network. These remarkable properties have attracted the interest of scientists attempting to produce new, elastic biomaterials. For example, a resilin-like protein called pro-resilin, constituted of 17 copies of the putative elastic repeat motif GGRPSDSYGAPGGGN has been expressed in E. coli. Using a photochemical cross-linking process, this protein allows the production of solid hydrogels, i.e., tridimensional cross-linked self-supported networks in which water is immobilized. These gels exhibit an extensibility of more than 300% and resilience superior to synthetic rubbers [36]. In order to take advantage of the properties of resilin, a modular recombinant protein made of the same repeat motif and biologically active domain has been used to produce cross-linked hydrogels and films. These materials allowed successful adhesion and proliferation of viable cells and possessed good mechanical properties (resilience of 90%) [37]. Tensile testing methods indeed indicate excellent resilience (higher than 90%), while studies at frequencies close to human phonation indicated elastic modulus values within the range of experimental mechanical performances collected on excised porcine and human vocal fold tissues [38]. These materials may thus have high interest for future medical applications, in particular in tissue regeneration such as blood vessels, cardiovascular tissues, and vocal folds.

422 | Part III New trends in sustainable development and biomedical applications 15.3.2 Byssus (mussel anchoring threads) The mussel holdfast, also known as “byssus”, is a biological fiber which has recently attracted the attention of researchers for the conceptualization and design of renewed biomimetic materials. Byssal threads present a unique combination of extensibility, stiffness, and toughness only surpassed by silk [39]. Under environmental stress such as waves and water current, a mussel produces a series of anchoring byssal threads through what is believed to be similar to an injection-molding process. Indeed, this mollusk uses different glands to secrete a mixture of proteins into the groove of a retractile organ called “foot”. The process is repeated several times and results in a bundle of 20 to 60 threads, approximately 3 cm in length and 100 μm in diameter. The fibers are attached to the stem of the foot and glued to a substrate, such as rocks or other mussels, by the plaque located at the other end of the thread (Fig. 15.2). Between these two regions, the fiber is composed of a corrugated elastic (proximal) portion as well as a smooth and stiffer (distal) part (Fig. 15.2). A 2 to 5 μm thick layer of proteins covers the fibrillar core. Approximately 95% (dry weight) of the byssus is made of proteins, the rest being mainly hexoses and inorganic content [40, 41]. Stem Proximal

Plaque

Distal

Sheath

(a)

Core

(b)

Fig. 15.2. (a) Mussel dangling from another mussel shell using its byssus. (b) Scheme of the different parts of a byssal thread. The stem is linked to the animal. The corrugated proximal part is extensible while the smooth distal part is stiff. The plaque anchors the threads to solid surfaces. The insert illustrates the fibrous core of the thread, which is entirely covered by a thin sheath.

It is noteworthy to mention that byssus represents a mechanical mode of protection. Indeed, although these sessile organisms are depleted of any anatomic parts which may allow for easy displacement around the costal reefs, they can adhere easily and strongly to almost any surface. Once anchored, it becomes very difficult to dislodge the animal only by the lift and drag forces of the waves and tides. The glue, which works underwater, is made of the so-called mussel foot proteins (mfps), which are intrinsic components of the plaque and the cuticle of the fiber. The mfps are rich in 3,4-

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

423

dihydroxyphenylalanine (DOPA), which contributes to the adhesion properties [42] and self-healing behaviour of the thread [43] when subjected to elongation. DOPA is believed to form complexes using coordination metals from the sea (e.g. Cu2+ , Zn2+ , Fe3+ ). The resulting metal coordinate acts as a sacrificial bond, thus sparing the covalent bonds from collapsing when submitted to external stress and promoting the recovery of the initial mechanical properties when regenerated [43]. In addition, radical coupling through an oxidation mechanism can lead to di-DOPA cross-linking, further increasing the mechanical performance. The radical generation of DOPA is also believed to be involved in the adhesion mechanism of the mfps [44]. The core of the fiber also plays an important role in the self-healing behavior and outstanding mechanical properties of byssus [40]. It is made up of a peculiar arrangement of block copolymer-like proteins named preCols. As shown in Fig. 15.3, these preCols are made of a central collagenous portion flanked by either elastin-like (preCol-P), silk-like (preCol-D), or plant cell wall-like (preCol-NG) domains at both ends, as determined by amino acid sequence homology [45]. The flanking domains are followed by histidine- and DOPA-rich regions. The preCols are distributed in different ratios along the axis of the fiber. PreCol-P is mostly abundant in the proximal region near the stem (first third of the fiber), while preCol-D occurrence gradually increases from the stem to the plaque. The concentration of preCol-NG is constant along the entire thread [46]. An atomic force microscopy (AFM) study of the byssus from Mytilus galloprovinciallis confirmed the presence and arrangement of the preCols in a 6 + 1 banana-like rod-shape [47]. The banana shape arises from the kinked central collagen, as a result of an amino acid (generally glycine) deletion or exchange in the canonical Gly-X-Y repeats found in the primary sequence. The bent, rod-like structures finally assemble in a head-to-head tail-to-tail array in the core of the fiber via metal cross-linking of adjacent histidine-rich domains [47]. Considering the high complexity and organization of the different proteins found in the byssus, determination of the molecular structure of these fibers represents a great challenge. Using 1D solid-state (SS-) NMR, FTIR, and x-ray diffraction studies, Hagenau et al. demonstrated conformational differences between the proximal and distal parts of the fiber [48]. Their results suggest that the distal region is well-oriented and rich in β-sheet structures, while the proximal section is isotropic and rich in αhelices. Using 2D 13 C solid-state NMR and 13 C-enriched byssus as well as chemical shift prediction, Arnold et al. identified most of the amino acids found in the primary sequences of the byssal proteins and determined the conformation in which they were found [49, 50]. Additionally, FTIR spectromicroscopy of a fiber split through its long axis revealed structural differences between the core and the sheath in the distal portion of the byssus. A combination of 2D 13 C NMR and FTIR confirmed the presence of the collagen triple helix, β-sheets (parallel and anti-parallel), and of β-turns structures in the core of the fiber. Moreover, unordered structure was found to be the major conformation in the cuticle. These studies progressively led to a more in-depth understanding of the molecular structure of the entire fiber assembly.

424 | Part III New trends in sustainable development and biomedical applications Flanking domains Histidine-rich domains

Collagen domain 6 + 1 preCol bundle Side view

Top View

Fig. 15.3. Representation of preCol units and their arrangement in a 6 + 1 banana-like bundle (side and top view). All preCols are made of a kinked collagen portion flanked by elastin-like (preCol-P), silk-like (preCol-D) or plant cell wall-like (preCol-NG) domains and terminated by histidine-rich regions.

Byssus is currently a waste product of the mussel farming industry. To give an idea of the scale of the loss, about 200 tons of byssus were discarded across Canada in 2010 prior to mussel commercialization [51]. Because of the known biocompatibility and high occurrence of collagen in mammals, materials scientists are trying to take advantage of this protein for various applications. The byssus preCols constitute a new type of collagen waiting to be exploited. Despite the potential biocompatibility associated with the protein content and the global interest in underwater adhesives, materials scientists have so far failed to efficiently extract any byssal protein for the preparation of new materials. The main reason is the complexity of the protein content relative to its high degree of cross-linking. Therefore, the ability to recycle byssus for collagen extraction would be a commercial asset for both the biotechnological industry and mussel farmers. Fibers have been prepared by drawing solutions of purified rod-like preCols extracted from the mussel foot [52]. It was demonstrated that preCols tend to spontaneously self-assemble into higher order structures, i.e., banana rods. Since the drawing process cannot be compared to injection-molding, it was suggested that selfassembly might be regulated by a predisposition of the preCols to form an anisotropic liquid-crystalline (meso)phase when submitted to stress. In that specific case, being able to extract or produce a large amount of these block copolymer-like proteins could be a very efficient way of preparing self-assembling and potentially biocompatible collagen-based oriented structures. The self-healing behavior and DOPA-related chemistry of the byssus inspired researchers in the development of functional materials designed for coatings or adhesives. For example, polyethylene glycol (PEG) based dendrimers were functionalized using either DOPA or histidine as chain-end moieties [53–55]. The gelation properties

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

425

of the resulting materials depend on both the presence of metals and pH conditions during chelation. Gelation of histidine-based polymers was shown to be efficient in the presence of divalent cations (Zn2+ , Cu2+ , Co2+ , Ni2+ ) at a pH where histidines are completely deprotonated (pH ≥ 7). The strength and self-healing properties of the gels were linked to the relaxation rate of the bonds between the metals and the histidines [55]. For the PEG-DOPA polymer, Fe3+ was specifically used as cross-linking metal, and the mechanical properties were compared to oxidation-induced crosslinking (covalent bonding through radical coupling). Higher pH (8 to 12) led to a more rigid cross-linked gel, with elastic moduli nearly matching the covalently cross-linked material. This was attributable to the formation of tris-cathecol-Fe3+ complexes, similar to those found within the mfps, constituting the cuticle of the byssus [53]. The major difference between the two sets of gels was the self-healing behavior. When submitted to shear stress until physical rupture, the PEG-DOPA-Fe3+ gels recovered their elastic modulus and cohesiveness within minutes, whereas the covalently crosslinked gels did not. The last application covered in this section is the production of synthetic mussel adhesive mimicking materials. Wilker et al. synthesized different poly[(3,4-dihydroxystyrene-co-styrene] to better understand the relationship between the number of pendant groups (i.e., catechol) and the adhesion performance of the material [56]. The catechol-functionalized polymer was submitted to periodate (IO4 )− oxidation to produce radicals. This reaction enhanced the adhesive properties of the polymer, but also produced cross-linking within the bulk material (between two cathecol groups). It was found that a molar ratio of 33% catechol and 67% styrene in the presence of periodate was optimal for adhesion. A higher content of pendant catechol groups gave rise to too much effective cross-linking in the polymer, thus increasing cohesion within the bulk material at the expense of surface attachment. The types of substrate to which adhesives were applied were also investigated and benchmarked against commercial glues. The adhesive performance of the biomimetic polymer was found to be comparable and, in some cases, better than commercial products made from cyanoacrylate or epoxy. Even though the investigation of mussel mimicking synthetic polymers only began very recently, these studies provide good insight into how byssal thread attachment and mechanical properties are modulated in their environment. Moreover, these materials offer promising applications in wet and dry conditions where there is a need for adhesion or tunable mechanical properties with self-healing behavior, such as material coatings or glues.

15.3.3 Silk Besides the fascination with spiders, the popularity of silk is mainly due to the striking tensile properties of the spider dragline fiber which is as strong as steel and five

426 | Part III New trends in sustainable development and biomedical applications times more resistant than Kevlar [57, 58]. These peculiar properties result from an excellent combination of high strength and extensibility. The cocoon silk of the domestic silkworm Bombyx mori (B. mori) is also a remarkable material, although more brittle. It is an “old” material which has been used for centuries by humans on a large scale as a textile fiber and suture material. It has the advantage of being available in large amounts from sericulture, while farming is not viable for spiders due to their territorial and cannibalistic nature. Silk actually includes a wide diversity of structures and properties. By definition, silk is a proteinous fiber excreted by arthropods such as moths, acarids, butterflies, bees, spiders, etc. Each type of silk is produced by specialized labial or abdominal glands and is made of one or multiple specific protein(s). For example, orb-weaving spiders can produce six types of silk, each with a specific biological function and adapted properties. The major ampullate (MA) glands produce the dragline silk which is used as lifeline, web radii, and frame. The flagelliform (Flag) glands produce the thread used to make the spiral of the web in which prey is caught. This fiber has low tensile strength and a very high failure strain (extensibility), so that its overall toughness is close to that of dragline silk [58]. Another example is provided by aciniform silk, which allows spiders to wrap their prey. This fiber exhibits an even better compromise between strength and extensibility than MA silk, providing greater toughness [59]. Overall, silk in general and spider silk in particular, exhibits very diverse properties. Thus, from a fundamental point of view, silk represents an attractive system to investigate structure-function relationships and, from an application point of view, suggests that the mechanical properties of silk-inspired biomaterials could be tuned to be adapted to specific applications by designing the appropriate proteins and applying adequate physicochemical processes. The properties of biological materials are totally determined by their structure at all length scales. A typical model of silk fiber is shown in Fig. 15.4. This structure is representative of a family of silk fibers, including MA silk and cocoon silk from B. mori. It is composed of highly oriented nanocrystalline β-sheets dispersed within an amorphous matrix constituted of more or less disordered polypeptide chains (random segments, turns, 31 -helices). As a first approximation, the stiff β-sheets are responsible for the strength of the fiber, whereas the unordered domains would be at the origin of its extensibility. The β-sheets of MA silk are formed by 4–7 amino acid-long polyalanine stretches [60], whereas more disordered segments are constituted of small glycine-rich motifs such as GGX (where X, Q, Y, L or R) or GPGQQ and GPGGY. For B. mori, the β-sheets are basically formed by GAGAGS motifs, the rest of the amino acids being involved in disordered domains [61, 62]. Other spider silk fibers do not correspond to this scheme. The wrapping silk of the spider Nephila clavipes (N. clavipes), for example, is marked by a mixture of mod-

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

427

Nephila clavipes MaSp1 AAAAAA GGAGQ GGY GGL GGQ GAGQ GGY GGL GSQ GAGR GGL GGQ GAG Major ampullate MaSp2 SAAAAAAAAS GPGQQ GPGGY GPGQQ GPGGY GPGQQ GLSGPG (spider) Bombyx mori (silkworm)

Heavy chain

GAGAGSGAAS(GAGAGS)n GAGAGYGAGVGAGYGAGYGAGAGAGY

Fig. 15.4. A schematized model of the structure of the MA fiber of the spider N. clavipes and the cocoon fiber of B. mori with the protein consensus repeat sequences. MA silk is composed of two proteins called MaSpI and MaSpII, whereas the main component of B. mori silk is called the heavy chain. The dark blue amino acids are involved in β-sheets, the light blue ones in disordered structures.

erately oriented β-sheets and α-helices, and distributed within more disordered regions (Fig. 15.5) [63]. Finally, the core fiber of the spiral of N. clavipes is constituted of almost randomly oriented proteins with a very heterogeneous and disordered secondary structure (Fig. 15.5), although a very low amount (about 7%) of slightly oriented β-sheets has been detected [64].

Aciniform, piriform

Flagelliform

Fig. 15.5. Other typical models of silk structures produced by the spider N. clavipes.

However, for other spider species such as Araneus diadematus and Argiope aurentia, Flag silk exhibits a significant amount of moderately oriented β-sheets. A relationship seems to exist between the proportion of β-sheets and the failure stresses and strains. A similar function-structure relationship appears to exist with the level of orientation of these β-sheets [64]. It was found that these structural elements are made up of a sequence segment called “spacer”, thus clarifying the role of this amino acid segment in the organization of Flag silk [64]. Overall, the results show that the same type of silk can exhibit differences between species, which emphasizes the potential of biomimetic silk to fulfil desirable properties.

428 | Part III New trends in sustainable development and biomedical applications One of the main incentives for the use of silk is to capitalize on its mechanical properties and ability to produce different materials with various shapes, states, and textures. After dissolution in water or in an organic solvent, silk can indeed be reprocessed into various types of matrices including hydrogels, films, foams, non-woven fiber mats, capsules, and spheres [28, 65]. Due to the complexity in the production of large amounts of spider silk proteins, the most advanced applications are currently based on B. mori silk proteins. B. mori fibers still arouse interest in the textile sector, especially for new properties (dyeing and grating of polymers), and for efficient suture applications, in particular regarding biocompatibility and immunogenicity [28]. Silk protein may be modified chemically to provide biocompatible coating for the surface of these systems, for example to impart coagulant activity or to promote cell adhesion and growth [28]. Various silk-based systems have successfully been used in vitro to support cell development, induce biomineralization, and control drug delivery [28, 65–68]. More specific 3D microperiodic scaffolds (square lattices, web-like circular lattices) have been prepared by direct ink printing (Fig. 15.6). The ink consisted of a silk fibroin solution from B. mori deposited in layers through a fine deposition nozzle to produce under mild ambient conditions a 3D array of filaments of 5 μm diameter. This precisely controlled architecture allowed the adhesion and growth of bone marrowderived stem cells and resulted in chondrogenic differentiation, an important character for the development of cartilage [69]. Wireless passive food sensors have been produced using B. mori silk. They consist of a microfabricated antenna or an array of antennae/resonators made of gold deposited on a silk substrate. These radio frequency identification (RFID)-like silk tag sensors are flexible and adapt to the non-planar shape of fruits or vegetables [70]. The response of these antennae is affected by the dielectric properties of the object to be probed. Their resonant responses were successfully tested during the ripening process to assess the potential for monitoring changes due to food spoilage. Further applications are envisaged for human health and environmental quality monitoring. Silkworm silk also exhibits interesting optical properties. Thanks to surface nanopatterning, a technology for fabricating defined structures on surfaces at a nanometer scale, optically transparent silk protein materials have been developed to form bioactive devices for optical diffractive applications such as diffraction gratings, pattern generators, and lenses [71] via the control of β-sheet crystallinity. The “direct ink writing” method has been used to produce optical waveguides, which offers new possibilities for creating biophotonic elements that can be readily doped or functionalized with biologically active agents [72]. Recently, the pristine spider dragline silk fiber has been tested as an optical fiber [73]. In this experiment, light is transmitted in a straight or bent filament. The fiber can also be integrated into a photonic chip made of polymer microstructures fabricated by UV-lithography, leading to efficient micro-optical coupling between silk and synthetic optical structures [73].

15 Protein- and peptide-based materials: a source of inspiration for innovation

(a)

|

429

(b)

200μm (c)

(d)

200μm

2μm

Fig. 15.6. (a) Schematic representation of 3D direct ink writing method using a silk fibroin solution. Typical 3D structures obtained: (b) square lattice and (c) circular web. (d) Magnified image of direct write silk fiber. Reprinted with permission from [69].

15.4 Functional materials from globular proteins The production of protein-based materials can rely on two strategies: the first one consists of the engineering of natural proteins and the second is based on the design of synthetic ones.

15.4.1 Natural proteins As discussed previously for byssus, the use of biomass constituents to make innovative and functional materials may represent an interesting method of valorizing waste by-products. For example, whey is a milk by-product of the cheese industry. Its main

430 | Part III New trends in sustainable development and biomedical applications component is β-lactoglobulin (β-lg), a small water-soluble protein of 164 amino acids which exists in the dimeric form. Under particular treatments, this protein exhibits technofunctional properties, including the ability to stabilize interfaces such as emulsions and foams, or a gelation capacity. Gelation of β-lg can be achieved by heating. Increasing temperature to ∼ 75°C induces the protein unfolding (denaturation), i.e., the breakage of intramolecular bonds which stabilize the native globular structure. As a result of unfolding, hydrophobic moieties initially buried inside the native structure are exposed, leading proteins to aggregate, the particles or fibers formed being rich in β-structures. If the concentration is sufficiently high, a gel can be obtained. Interestingly, β-lg can generate two types of gels depending on pH [22, 25]. Elastic and transparent gels are formed near the isoelectric point of the protein, i.e., between pH 4 and 6, whereas stiff and opaque gels occur for low electrostatic repulsions. The former gels are constituted of fine filamentous strands (fine-stranded gels), whereas the latter are made of coarse, roughly spherical particles and have a lower water-holding capacity (particulate gels). These differences can be accounted for by the different amplitude of the electrostatic interactions depending on pH. Another procedure for forming gels consists of the application of a pre-denaturation step consisting of moderate heating of a protein solution, followed by cold gelation induced by the addition of a divalent cation such as Ca2+ or Fe2+ . Depending on the cation concentration, fine-stranded or particulate gels can be generated. Capitalizing on such matrices to develop new encapsulation materials for nutrients, the gel structure appears to influence the extent of iron release in vitro in gastrointestinal conditions [74]. Taking further advantage of this method, an emulsification-cold gelation procedure has been developed to make edible beads [75] or microspheres [76], intended for effective and selective delivery of bioactive agents such as retinol to the site of action. According to in vitro studies, these beads are not susceptible to enzymatic attack during rapid transit of the stomach (gastroresistant) and form good matrices to protect fat-soluble bioactive molecules for specific absorption in the intestine.

15.4.2 Artificial proteins Artificial proteins offer different advantages to natural ones, as they allow control of the amino acid composition (non-natural amino acids [32], hydrophobic-hydrophilic pattern) and secondary structure [77]. Many artificial (chimeric) proteins have been inspired by fibrous proteins [66, 78–81], but other developments have been influenced by globular proteins. In nature, the function of some exogenous polypeptides is to self-assemble into pores on biological membranes and function as transporters, toxins, ion channels, and antibiotics. Trying to mimic nature, the design of protein-based stimuli-

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

431

responsive pores in membranes is a very active field of research. The control of gating, the ability of pores or channels to open and close in response to a stimulus, may potentially lead to applications such as sensors and drug controlled release. One strategy in this area consists of using existing pore-forming protein complexes such as α-hemolysin. This protein is a 293-amino-acid which naturally forms heptameric pores in lipid bilayers. Using genetic engineering and chemical modifications, biochemical, chemical, and physical stresses can trigger or switch on and off α-hemolysin poreforming activity [82, 83]. Furthermore, activation of the pore can also be reversibly mediated by light by grafting an azobenzene group onto the protein which undergoes a reversible trans-cis isomerization by illumination with UV-visible light (Fig. 15.7) [82, 84]. MAL-AZO-QA O

O H N

N

H N

N

O

O

N N

O

N

O

380nm N H

~17Å

N

500nm

O

HN O

~10Å

N+

N +

(a)

N S

Out

N N

N

S

+

+

380nm

K+

500nm

In (b) Fig. 15.7. Principle of a light-activated pore based on α-hemolysin. A mutant of the protein with a cysteine residue at position 422 is grafted through a disulfide bond with the molecule MAL-AZO-QA. This “gate” consists of a maleimide (MAL) group, an azobenzene (AZO) group and a quaternary ammonium (QA) group. The azobenzene isomerizes from the trans to the cis configuration with light, which shortens the molecule by 7 Å, opens the pore and allows ion transfer. The reverse transition for pore closing is triggered at another wavelength. Reprinted with permission from [84].

432 | Part III New trends in sustainable development and biomedical applications Another strategy is based on the capacity of proteins to undergo conformational change from folded to unfolded conformations (molecular switches). Inspired by works on polymer brushes, IDPs have been grafted to porous polymer membranes. When the protein is unfolded, the pores are closed due to the extended and interpenetrating conformation of the chains, whereas upon protein folding, the pores open. This simple switching mechanism, induced by change in pH or ionic strength, allows the selective transfer of molecules through the pores [8, 85, 86]. Recombinant DNA methods have also been used to create artificial proteins which undergo reversible gelation in response to changes in pH or temperature. An example is given by a protein composed of two terminal leucine zipper domains flanking a central, flexible, water-soluble segment [87]. The assembly of the terminal domains into coiled-coil structures at neutral pH leads to the formation of a three-dimensional network, with the central segment retaining solvent and preventing complete precipitation of the chains. Dissociation of the coiled-coil assemblies through elevation of pH or temperature causes the return to the solution state. The control of gel formation (near-neutral pH and near-ambient temperature) suggests that these switchable hydrogels have potential applications in encapsulation or controlled release of bioactive molecules.

15.5 Functional materials from synthetic peptides Control of material organization at the nanoscale is crucial for the development of functional materials, and peptides can be rationally designed to fulfill this aim. Short peptides have several advantages, such as control of the self-assembly process, numerous possible final structures, various textures and forms of the final materials, ease of synthesis by solid-phase synthesis, and potential for large scale production, as already demonstrated for aspartame [88]. Previous work has shown, for example, that binary patterning of polar and nonpolar amino acids arranged with periodicity can direct protein sequences to form fibrils resembling amyloids [89]. This type of patterned sequence seems to have been disfavored by evolutionary selection [90, 91]. A strategy based on this pattern has been used to develop peptides that self-assemble in a reversible manner [92], suggesting that assembly is dictated by thermodynamic stability and not by kinetic trapping. The same strategy with a template-directed method has been used to form left-handed helical ribbons [93] or to direct the assembly of peptides into β-sheets with a three-fold symmetry [94]. The propensity of polypeptides to aggregate into amyloid cross-β fibrils has been exploited to serve as a template to generate metal nanowires [95]. The peptide is a diphenylalanine that forms typical amyloid fibrils with a hollow tubular structure. Upon incorporation into the solution, ionic silver settles within the tubes. After subsequent reduction with citric acid to form metal silver, and after enzymatic degrada-

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

433

tion of the peptide backbone, discrete nanowires are produced with a long persistence length. Generally speaking, the literature shows that the design of biomaterials based on peptide amyloid fibrils appears as a widespread strategy [96, 97]. For twenty years, various peptides have been engineered to produce diverse small building blocks which can self-assemble into hierarchical structures with potential relevant applications, especially in the biomedical area [21, 98–100]. Pioneering works in this field use alternation of natural L- and non-natural D-amino acid enantiomers to form ring-shaped cyclic peptides which stack one-dimensionally, via an antiparallel β-sheet-like arrangement, into hollow tubular structures nanometers long (Fig. 15.8) [101, 102]. The alternation of L- and D- amino acids forces the residues to lie at the outside of the ring, leading to nanotubes with uniform internal diameters of typically 10 Å. A peptide forming β-hairpin (two β-strands joined by a β-turn) has been designed to assemble into fibrils, which ultimately forms a temperature-responsive gel [78, 103]. Interestingly, hydrogels prepared from enantiomeric mixtures of these β-hairpin peptides exhibit synergistic, although unexplained, increase in rigidity compared to gels prepared from pure peptide enantiomers [104]. In another series of studies, an 11 amino acid-long peptide has been shown to form a chiral rod-like unit as a β-sheet nanotape. This unit can self-assemble into hierarchical organizations (Fig. 15.9) such as helical tapes and, with increasing concentration, further associate into twisted ribbons (double tapes), fibrils (twisted stacks of ribbons), and fibers (entwined fibrils) [105]. By controlling the charged amino acids in the design of this type of peptides, transition from a nematic gel to an isotropic solution can be triggered by acidic or basic pH conditions, or can be completely independent of pH [106]. Using peptides, one can form a vast array of structures, including nanofibers [99], nanoropes, and nanofilaments [107]. Interestingly, surfactant-like peptides, i.e., with two distinctive polar and apolar regions, have been shown to form nanotubes and nanovesicles [108]. The design of peptides can also be rationalized using a library of known sequences and functions. Using this approach, a rational combinatorial library based on the structural principles of known membrane-spanning β-sheets, has allowed the design of a pore-forming β-sheet peptide. The chosen peptide exhibits a pore-forming activity with a mechanism of action very similar to that of natural pore-forming peptides. This peptide is characterized by aromatic residues at the lipidexposed interfacial positions and basic residues in the pore-lining portion of the sequence [109]. Hybrid peptide-amphiphiles can also be used to build soft materials, in particular scaffolds for cell growth. For example, peptides equipped with an alkyl chain have been constructed to form cylindrical micelles which can be considered fibers with lengths of up to several micrometers and nearly uniform diameters of 7–8 ±1 nm (Fig. 15.10). Nanofiber formation transforms the liquid solution into three-dimensional reversible gels. The amphiphile-peptide molecules also contain consecutive cysteine residues which, when oxidized, form disulfide bonds to stabilize the self-assembled structure. Depending on the amino acid composition, the assembly process can be

434 | Part III New trends in sustainable development and biomedical applications

O

NH2

H CH3 O H N N D

O

O HN L

HO O O L

NH

H3C HN

CH3 O NH

D

D

O L

HO O

O H3C

O D

D

HN

L

O

N H

Self-Assembly H2N

O

OH R

R

HO N N

N N O H OH O HO H N N N N HO O H OH

R

R

NH2

L

HN

HO N N

N N O H OH O HO H N N N N HO O H

D

N O CH H O 3 OH N

N HO O H N N HO OH N N HO O H N N HO

HO N N H O HO N N H O HO N N OH HO N N OH

CH3 O

L N H O

O

NH

OH OH N

HO N

N N HO OH O H HO N N N N H O OH OH N

HO N

N N HO OH O H HO N N N N H O OH

R

R

R

R

Fig. 15.8. Model of hierarchical assembly of a peptide based on the alternation of L- and D-amino acid enantiomers. This peptide forms a ring-shaped structure which stacks one-dimensionally, via an antiparallel β-sheet-like hydrogen bonded arrangement, into hollow tubules nanometers long. Reprinted with permission from [102].

triggered by a change in pH, addition of divalent cations [110], or control of electrostatic interactions between polypeptides [111]. Some functionalized moieties can also be added to the peptide. For example, the acidic amino acids aspartic acid and phosphorylated serine (phosphoserine) were specifically used in the above peptides since they are abundant in the proteins of mineralized tissues such as bones and are known to initiate hydroxyapatite (HA) crystal growth [112]. As a matter of fact, at the lowest level of their hierarchical organization bones are formed by the self-assembly of collagen triple helices with HA crys-

15 Protein- and peptide-based materials: a source of inspiration for innovation

(c’)

(d’)

htape

(e’)

hribbon

|

435

(f’)

hfibril

(a) Ɛtape

Ɛfibril Ɛribbon

Rod-like monomer

Ɛfibre

Ɛtrans b2

a bI

Monomer

Tape

Ribbon

Fibril

Fibre

(b)

(c)

(d)

(e)

(f)

Concentration Fig. 15.9. Model of hierarchical assembly of a chiral rod-like β-sheet nanotape. This unit can selfassemble into helical tape, twisted ribbons (double tapes), fibrils (twisted stacks of ribbons), and fibers (entwined fibrils). Reprinted with permission from [105].

tals grown within these fibrils. Incorporating phosphoserine residue into the peptide sequence allows the fiber to display a phosphorylated surface able to direct mineralization of HA in which the c axes of the crystals are mainly aligned with the long axes of the fibers, similar to what is observed in collagen fibrils of bones [112]. This biomimetic strategy was thus successful in forming a nanostructured composite, with potential use in bone tissue engineering.

15.6 Summary Peptide- and protein-based materials offer numerous potential advantages in various and also possibly unanticipated application fields. Their ability to change conformation and self-assemble paves the way for the fabrication of stimuli-controlled, nanostructured, and hierarchically organized functionalized materials. Various strategies may be adopted: from natural proteins and peptides to artificial and de novo ones. The polypeptide world exhibits such a diversity and richness that it undoubtedly constitutes a fertile ground for innovation in the field of sustainable functional materials.

436 | Part III New trends in sustainable development and biomedical applications 4 O

2 HO H N O

1

(a)

O N H SH

SH H N O

O N H SH

SH H N O

O

H N

N H

O

P

OH

O

O N H

H N

O

H N

N H

O

O

NH

3 H2N

O OH O OH

NH

5

(b)

(c) Fig. 15.10. (a) Chemical structure of the peptide-amphiphile with its key structural features. Part 1 is an alkyl chain which provides the molecule with a hydrophobic character, the rest of the molecule being hydrophilic. Part 2 is a four cysteine residue motif which polymerizes through disulfide bonds when oxidized, in order to stabilize the final self-assembled structure. Part 3 is a linker made of three glycine residues which provides some flexibility to the hydrophilic head group. Part 4 is a single phosphorylated serine residue whose role is to interact with calcium ions and then help mineralization of hydroxyapatite. Part 5 represents the cell adhesion binding site RGD. (b) Molecular model of the peptide-amphiphile to highlight the overall conical shape of the molecule. (c) Schematic showing the self-assembly of PA molecules into a cylindrical micelle. Reprinted with permission from [112].

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

437

References [1]

[2] [3] [4] [5] [6] [7] [8] [9] [10] [11]

[12] [13] [14] [15] [16] [17] [18]

[19] [20] [21] [22]

State of the Planet Declaration - Planet under pressure: New knowledge towards solutions: Planet under Pressure Conference. 2012. (Accessed May 13, 2013, at http://www.essp.org/ fileadmin/redakteure/pdf/others/PUP_declaration.pdf.) United Nations Environment Programme. Global Environment Outlook 5 (GEO5). Valleta, Malta, UNEP. 2012. Barnosky AD, Hadly EA, Bascompte J, et al. Approaching a state shift in Earth’s biosphere. Nature, 486, (2012) 52–8. Crutzen P, Stoermer EF. The “Anthropocene”. IGPB Newsletter, 41, (2000) 17–18. Rockström J, Steffen W, Noone K, et al. A safe operating space for humanity. Nature, 461, (2009) 472–5. Benyus J. Biomimicry: Innovation inspired by nature. William Morrow & Company, New York USA, 1997. Sanchez C, Arribart H, Giraud Guille M-M. Biomimetism and bioinspiration as tools for the design of innovative materials and systems. Nature Mater, 4, (2005) 277–88. Srinivasan N, Kumar S. Ordered and disordered proteins as nanomaterial building blocks. WIREs Nanomed Nanobiotechnol, 4, (2012) 204–18. Andrew JP, Jeffrey SC, Mickey GH. Environmentally sustainable fibers from regenerated protein. Biomacromolecules, 10, (2009) 1–8. Pierschbacher MD, Ruoslahti E. Cell attachment activity of fibronectin can be duplicated by small synthetic fragments of the molecule. Nature, 309, (1984) 30–3. Abbrebviated nomenclature of synthetic polypeptides (polymerized amino acids). International Union of Pure and Applied Chemistry and International Union of Biochemistry ed. Butterworths, London UK, 1971. Voet D, Voet JG, Pratt CW. Fundamentals of biochemistry: life a the molecular level. 4th ed. John Wiley and Sons, Hoboken, NJ, USA, 2013. Stapley B, Creamer T. A survey of left-handed polyproline II helices. Protein Sci 1999, 8, 587–95. Adzhubei AA, Sternberg MJE, Makarov AA. Polyproline-II helix in proteins: Structure and function. J Mol Biol, 425, (2013) 2100–32. Eker F, Griebenow K, Schweitzer-Stenner R. Stable conformations of tripeptides in aqueous solution studied by UV circular dichroism spectroscopy. J Am Chem Soc, 125, (2003) 8178–85. Peter T. The interplay between structure and function in intrinsically unstructured proteins. FEBS Letters, 579, (2005) 3346–54. Cortese MS, Uversky VN, Dunker AK. Intrinsic disorder in scaffold proteins: Getting more from less. Prog Biophys Mol Biol, 98, (2008) 85–106. Turoverov KK, Kuznetsova IM, Uversky VN. The protein kingdom extended: Ordered and intrinsically disordered proteins, their folding, supramolecular complex formation, and aggregation. Prog Biophys Mol Biol, 102, (2010) 73–84. Uversky VN, Gillespie JR, Fink AL. Why are “natively unfolded” proteins unstructured under physiologic conditions? Proteins, 41, (2000) 415–27. Shi Z, Woody RW, Kallenbach NR. Is polyproline II a major backbone conformation in unfolded proteins? Adv Protein Chem, 62, (2002) 163–240. Woolfson DN, Ryadnov MG. Peptide-based fibrous biomaterials: some things old, new and borrowed. Curr Opinion Chem Biol, 10, (2006) 559–67. Clark AH, Judge FJ, Richards JB, Stubbs JM, Suggett A. Electron microscopy of network structures in the thermally-induced globular protein gels. Int J Peptide Protein Res, 17, (1981) 380–92.

438 | Part III New trends in sustainable development and biomedical applications [23] Pedersen JS, Otzen DE. Amyloid—a state in many guises: Survival of the fittest fibril fold. Protein Sci, 17, (2008) 2–10. [24] Yagi H, Ban T, Morigaki K, Naiki H, Goto Y. Visualization and classification of amyloid beta supramolecular assemblies. Biochemistry, 46, (2007) 15009–17. [25] Stading m, Hermansson A-M. Large deformation properties of β-lactoglobulin gel structures. Food Hydrocolloids, 5, (1991) 339–52. [26] Lefèvre T, Subirade M. Molecular differences in the formation and structure of fine-stranded and particulate β-lactoglobulin gels. Biopolymers, 54, (2000) 578–86. [27] Dobson CM. The structural basis of protein folding and its links with human disease. Philos Trans Biol Sci, 356, (2001) 133–45. [28] Hardy JG, Römer LM, Scheibel TR. Polymeric materials based on silk proteins. Polymer, 49, (2008) 4309–27. [29] Rauscher S, Baud S, Miao M, Keeley FW, Pomès R. Proline and glycine control protein selforganization into elastomeric or amyloid fibrils. Structure, 14, (2006) 1667–76. [30] Scheibel T. Protein fibers as performance proteins: new technologies and applications. Curr Opinion Biotechnol, 16, (2005) 427–33. [31] Bellingham CM, Keeley FW. Self-ordered polymerization of elastin-based biomaterials. Curr Opinion Solid State Mater Sci, 8, (2004) 135–9. [32] Renuart E, Viney C. Biological fibrous materials: self-assembled structures and optimised properties. In: Structural biological materials Design and structure-property relationships. Elices M, ed. 233–67. Elsevier Science Ltd., Amsterdam, 2000. [33] Hu X, Qin G, Cebe P, Kaplan DL. Structure and elasticity mechanism of full length resilin proteins. In: IEEE 36th Annual Northeast, Bioengineering Conference 26–28 March 2010, 1–2. New York USA, 2010. [34] Ehrlich H. Abductin. In: Gorb SN, ed. Biological materials of marine origin. 319–22. Springer, Dordrecht, 2010. [35] Jensen M, Weis-Fogh T. Biology and physics of locust flight. V. Strength and elasticity of locust cuticle. Philos Trans R Soc London B, 245, (1962) 137–69. [36] Elvin CM, Carr AG, Huson MG, et al. Synthesis and properties of crosslinked recombinant proresilin. Nature, 437, (2005) 999–1002. [37] Charati MB, Ifkovits JL, Burdick JA, Linhardtc JG, Kiick KL. Hydrophilic elastomeric biomaterials based on resilin-like polypeptides. Soft Matter, 5, (2009) 3412–6. [38] Li L, Teller S, Clifton RJ, Jia X, Kiick KL. Tunable mechanical stability and deformation response of a resilin-based elastomer. Biomacromolecules, 12, (2011) 2302–10. [39] Gosline J, Lillie M, Carrington E, Guerette P, Ortlepp C, Savage K. Elastic proteins: biological roles and mechanical properties. Philos Trans R Soc London B, 357, (2002) 121–32. [40] Waite JH, Vaccaro E, Sun C, Lucas JM. Elastomeric gradients: a hedge against stress concentration in marine holdfasts? Philos Trans R Soc London B, 357, (2002) 143–53. [41] Hennebicq R, Fabra G, Pellerin C, Marcotte I, Myrand B, Tremblay R. The effect of spawning of cultured mussels (Mytilus edulis) on mechanical properties, chemical and biochemical composition of byssal threads. Aquaculture, 410–411, (2013) 11–17. [42] Sever MJ, Weisser JT, Monahan J, Srinivasan S, Wilker JJ. Metal-mediated cross-linking in the generation of a marine-mussel adhesive. Angew Chem Int Ed Engl, 43, (2004) 448–50. [43] Harrington MJ, Masic A, Holten-Andersen N, Waite JH, Fratzl P. Iron-clad fibers: a metal-based biological strategy for hard flexible coatings. Science, 328, (2010) 216–20. [44] Hight L, Wilker J. Synergistic effects of metals and oxidants in the curing of marine mussel adhesive. J Mater Sci, 42, (2007) 8934–42. [45] Coyne KJ, Qin X-X, Waite JH. Extensible collagen in mussel byssus: a natural block copolymer. Science, 277, (1997) 1830–2.

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

439

[46] Waite JH, Qin X-X, Coyne KJ. The peculiar collagens of mussel byssus. Matrix Biol, 17, (1998) 93–106. [47] Hassenkam T, Gutsmann T, Hansma P, Sagert J, Waite JH. Giant bent-core mesogens in the thread forming process of marine mussels. Biomacromolecules, 5, (2004) 1351–5. [48] Hagenau A, Scheidt HA, Serpell L, Huster D, Scheibel T. Structural analysis of proteinaceous components in byssal threads of the mussel Mytilus galloprovincialis. Macromol Biosci, 9, (2009) 162–8. [49] LeBlanc A, Arnold AA, Genard B, et al. Determination of isotopic labeling of proteins by precursor ion scanning liquid chromatography/tandem mass spectrometry of derivatized amino acids applied to nuclear magnetic resonance studies. Rapid Comm Mass Spectrom, 26, (2012) 1165–74. [50] Arnold AA, Byette F, Séguin-Heine M-O, et al. Solid-state NMR structure determination of whole anchoring threads from the blue mussel Mytilus edulis. Biomacromolecules, 14, (2013) 132–41. [51] Aquaculture Statistics 2010. Statistics Canada, 2011. (Accessed June 20, 2013, at http://www.statcan.gc.ca/pub/23-222-x/23-222-x2011000-eng.htm.) [52] Harrington MJ, Waite JH. pH-dependent locking of giant mesogens in fibers drawn from mussel byssal collagens. Biomacromolecules, 9, (2008) 1480–6. [53] Holten-Andersen N, Harrington MJ, Birkedal H, et al. pH-induced metal-ligand cross-links inspired by mussel yield self-healing polymer networks with near-covalent elastic moduli. Proc Nat Acad Sci, 2011. [54] Fullenkamp DE, Rivera JG, Gong Y-k, et al. Mussel-inspired silver-releasing antibacterial hydrogels. Biomaterials, 33, (2012) 3783–91. [55] Fullenkamp DE, He L, Barrett DG, Burghardt WR, Messersmith PB. Mussel-inspired histidinebased transient network metal coordination hydrogels. Macromolecules, 46, (2013) 1167–74. [56] Matos-Pérez CR, White JD, Wilker JJ. Polymer composition and substrate influences on the adhesive bonding of a biomimetic, cross-linking polymer. J Am Chem Soc, 134, (2012) 9498–505. [57] Knight D, Vollrath F. Liquid crystalline spinning of spider silk. Nature 2001, 410. [58] Gosline JM, DeMont ME, Denny MW. The structure and properties of spider silk. Endeavour, 10, (1986) 37–43. [59] Blackledge TA, Hayashi CY. Silken toolkits: biomechanics of silk fibers spun by the orb web spider Argiope argentata (Fabricius 1775). J Exp Biol, 209, (2006) 2452–61. [60] Simmons A, Ray E, Jelinski LW. Solid-State 13 C NMR of Nephila clavipes dragline silk establishes structure and identity of crystalline regions. Macromolecules, 27, (1994) 5235–7. [61] Marsh RE, Corey RB, Pauling L. An investigation of the structure of silk fibroin. Biochim Biophys Acta, 16, (1955) 1–33. [62] Lotz B, Cesari FC. Chemical structure and the crystalline structures of Bombyx mori silk fibroin. Biochimie, 61, (1979) 205–14. [63] Rousseau M-E, Lefèvre T, Pézolet M. Conformation and orientation of proteins in various types of silk fibers produced by Nephila clavipes spiders. Biomacromolecules, 10, (2009) 2945–53. [64] Lefèvre T, Pézolet M. Unexpected β-sheets and molecular orientation in flagelliform spider silk as revealed by Raman spectromicroscopy. Soft Matter, 8, (2012) 6350–7. [65] Vepari C. Silk as a biomaterial. Prog Polym Sci, 32, (2007) 991–1007. [66] Wang X, Kim HJ, Wong C, Vepari C, Matsumoto A, Kaplan DL. Fibrous proteins and tissue engineering. Mater Today, 9, (2006) 44–53. [67] Hardy JG, Scheibel T. Composite materials based on silk proteins. Prog Polym Sci, 35, (2010) 1093–115. [68] Lammel AS, Hu X, Park S-H, Kaplan DL, Scheibel T. Controlling silk fibroin particle features for drug delivery. Biomaterials, 31, (2010) 4583–91.

440 | Part III New trends in sustainable development and biomedical applications [69] Ghosh S, Parker ST, Wang X, Kaplan DL, Lewis JA. Direct-write assembly of microperiodic silk fibroin scaffolds for tissue engineering applications. Adv Funct Mater, 18, (2008) 1883–9. [70] Tao H, Brenckle MA, Yang M, et al. Silk-based conformal, adhesive, edible food densors. Adv Mater, 24, (2012) 1067–72. [71] Lawrence BD, Cronin-Golomb M, Georgakoudi I, Kaplan DL, Omenetto FG. Bioactive silk protein biomaterial systems for optical devices. Biomacromolecules, 9, (2008) 1214–20. [72] Parker ST, Domachuk P, Amsden J, et al. Biocompatible silk printed optical waveguides. Adv Mater, 21, (2009) 2411–5. [73] Huby N, Vié V, Beaufils S, et al. Native spider silk as a biological optical fiber. Appl Phys Letter, 102, (2013) 123702. [74] Remondetto G, Beyssac É, Subirade M. Iron availability from whey protein hydrogels: An in vitro study. J Agric Food Chem, 52, (2004) 8137–43. [75] Beaulieu L, Savoie L, Paquin P, Subirade M. Elaboration and characterization of whey protein beads by an emulsification/cold gelation process: application for the protection of retinol. Biomacromolecules, 3, (2002) 239–48. [76] Chen L, Subirade M. Alginate–whey protein granular microspheres as oral delivery vehicles for bioactive compounds. Biomaterials, 27, (2006) 4646–54. [77] Rabotyagova OS, Cebe P, Kaplan DL. Protein-based block copolymers. Biomacromolecules, 12, (2011) 269–89. [78] Grove TZ, Regan L. New materials from proteins and peptides. Curr Opinion Chem Biol, 22, (2012) 451–6. [79] Gomes S, Numata K, Isabel B. Leonor, Mano JF, Reis RL, Kaplan DL. AFM study of morphology and mechanical properties of a chimeric spider silk and bone sialoprotein protein for bone regeneration. Biomacromolecules, 12, (2011) 1675–85. [80] Currie HA, Deschaume O, Naik RR, Perry CC, Kaplan DL. Genetically engineered chimeric silk– silver binding proteins. Adv Funct Mater, 21, (2011) 2889–95. [81] Bracalello A, Santopietro V, Vassalli M, et al. Design and production of a chimeric resilin-, elastin-, and collagen-like engineered polypeptide. Biomacromolecules, 12, (2011) 2957–65. [82] Astier Y, Bayley H, Howorka S. Protein components for nanodevices. Curr Opinion Chem Biol, 9, (2005) 576–84. [83] Bayley H. Pore-forming proteins with built-in triggers and switches. Bioorg Chem, 23, (1995) 340–54. [84] Banghart M, Borges K, Isacoff E, Trauner D, Kramer RH. Light-activated ion channels for remote control of neuronal firing. Nature Neurosci, 7, (2004) 1381–6. [85] Jovanovic-Talisman T, Tetenbaum-Novatt J, McKenney AS, et al. Artificial nanopores that mimic the transport selectivity of the nuclear pore complex. Nature, 457, (2009) 1023–7. [86] Ito Y, Ochiai Y, Park YS, Imanishi Y. pH-Sensitive gating by conformational change of a polypeptide brush grafted onto a porous polymer membrane. J Am Chem Soc, 119, (1997) 1619–23. [87] Petka WA, Harden JL, McGrath KP, Wirtz D, Tirrell DA. Reversible hydrogels from selfassembling artificial proteins. Science, (1998) 389–92. [88] Colombo G, Soto P, Gazit E. Peptide self-assembly at the nanoscale: a challenging target for computational and experimental biotechnology. Trends Biotechnol, 25, (2007) 211–8. [89] Xiong H, Buckwalter bL, Shieh H-M, Hecht MH. Periodicity of polar and nonpolar amino acids is the major determinant of secondary structure in self-assembling oligomeric peptides. Proc Nat Acad Sci, 92, (1995) 6349–53. [90] Broome BM, Hecht MH. Nature disfavors sequences of alternating polar and non-polar amino Acids: Implications for amyloidogenesis. J Mol Biol, 296, (2000) 961–8. [91] West MW, Hecht MH. Binary patterning of polar and nonpolar amino acids in the sequences and structures of native proteins. Protein Sci, 4, (1995) 2032–9.

15 Protein- and peptide-based materials: a source of inspiration for innovation

|

441

[92] West MW, Wang W, Patterson J, Mancias JD, Beasley JR, Hecht MH. De novo amyloid proteins from designed combinatorial libraries. Proc Nat Acad Sci, 96, (1999) 11211–6. [93] Marini DM, Hwang W, Lauffenburger DA, Zhang S, Kamm RD. Left-handed helical ribbon intermediates in the self-assembly of a β-sheet peptide. Nanoletter, 2, (2002) 295–9. [94] Brown CL, Aksay IA, Saville DA. Template-directed assembly of a de novo designed protein. J Am Chem Soc, 124, (2002) 6846–8. [95] Reches M, Gazit E. Casting metal nanowires within discrete self-assembled peptide nanotubes. Science, 300, (2003) 625–7. [96] Channon K, MacPhee CE. Possibilities for ‘smart’ materials exploiting the self-assembly of polypeptides into fibrils. Soft Matter, 4, (2008) 647–52. [97] MacPhee CE, Woolfson DN. Engineered and designed peptide-based fibrous biomaterials. Curr Opinion Solid State Mater Sci, 8, (2004) 141–9. [98] Zhang S. Emerging biological materials through molecular self-assembly. Biotechnol Adv, 20, (2002) 321–39. [99] Zhang S, Marini DM, Hwang W, Santoso S. Design of nanostructured biological materials through self-assembly of peptides and proteins. Curr Opinion Chem Biol, 6, (2002) 865–71. [100] Fairman R, Åkerfeldt KS. Peptides as novel smart materials. Curr Opinion struct Biol, 15, (2005) 453–63. [101] Ghadiri MR, Ganja JR, Milligan RA, McRee DE, Khazanovitch. Self-assembling organic nanotubes based on a cyclic peptide architecture. Nature, 366, (1993) 324–7. [102] Khazanovich N, Granja JR, McRee DE, Milligan RA, Ghadiri MR. Nanoscale tubular ensembles with specified internal diameters. Design of a self-assembled nanotube with a 13-A pore. J Am Chem Soc, 116, (1994) 6011–6012. [103] Pochan DJ, Schneider JP, Kretsinger J, Ozbas B, Rajagopal K, Haines L. Thermally reversible hydrogels via intramolecular folding and consequent self-assembly of a de novo designed peptide. J Am Chem Soc, 125, (2003) 11802–3. [104] Nagy KJ, Giano MC, Jin A, Pochan DJ, Schneider JP. Enhanced mechanical rigidity of hydrogels formed from enantiomeric peptide assemblies. J Am Chem Soc, 133, (2011) 14975–7. [105] Aggeli A, Nyrkova IA, Bell M, et al. Hierarchical self-assembly of chiral rod-like molecules as a model for peptide β-sheet tapes, ribbons, fibrils, and fibers. Proc Nat Acad Sci, 98, (2001) 11857–62. [106] Aggeli A, Mark Bell, Carrick LM, et al. pH as a trigger of peptide β-sheet self-assembly and reversible switching between nematic and isotropic phases. J Am Chem Soc, (2003) 125. [107] Wagner DE, Phillips CL, Ali WM, et al. Toward the development of peptide nanofilaments and nanoropes as smart materials. Proc Nat Acad Sci, 102, (2005) 12656–61. [108] Santoso S, Hwang W, Hartman H, Zhang J. Self-assembly of surfactant-like peptides with variable glycine tails to form nanotubes and nanovesicles. Nanoletter, 2, (2002) 687–91. [109] Raush JM, Marks JR, Wimley WC. Rational combinatorial design of pore-forming β-sheet peptides. Proc Nat Acad Sci, 102, (2005) 10511–5. [110] Hartgerink JD, Beniash E, Stupp SI. Peptide-amphiphile nanofibers: A versatile scaffoldfor the preparation of self-assembling materials. Proc Nat Acad Sci, (2002) 99. [111] Niece KL, Hartgerink JD, Donners JJJM, Stupp SI. Self-assembly combining two bioactive peptide-amphiphile molecules into nanofibers by electrostatic attraction. J Am Chem Soc, 125, (2003) 7146–7. [112] Hartgerink JD, Beniash E, Stupp SI. Self-assembly and mineralization of peptide-amphiphile nanofibers. Science, 295, (2001) 1684–8.

B. Riedl, V. Vardanyan, W. N. Nkeuwa, A. Kaboorani, V. Landry, B. Poaty, M. Vlad, and C. Sow

16 Nanocomposite coatings 16.1 Introduction This chapter will review the techniques for synthesis and characterization of coatings containing nanoparticles for various substrates (i.e., nanocoatings and bio-based nanocoatings). The discussion will focus mainly on coatings developed for wood substrates, although coatings can also be applied to other types of substrates requiring functional surface properties. There are several terms to describe different types of coatings: stains, varnishes, paints, lacquers, and glazes. Here we will use the general term coating and adapt the specific terms to address specific situations. This chapter will introduce some aspects of the chemistry and physics of coatings reinforced with nanoparticles. There is a large variety of coatings for wood and one can categorize them as a function of their chemical, physical, and optical properties (Figs. 16.1, 16.2). According to the formulation, these can be water-based or solvent-borne, high solid contents, UV-cured or even powders. Considering a particular application, they can be designed for exterior or interior use; they can also be transparent or opaque and can incorporate several combinations of these characteristics. The main objective of coatings incorporating nanoparticles reinforcement is better resistance to wear, UV degradation and water ingress, all of which must be done without affecting visual characteristics such as brilliance, color and transparency. The starting hypothesis is that nanoparticles, when combined with existing coating formulation, may display improved performance when compared to a similar coating containing microparticles. Overall, the global coatings market is estimated at 30 million tons per year and is worth about 120 billion US dollars [1]. Any additive which makes an inroad into this huge market and very competitive environment is thus very welcome. Moreover, technology in this field is changing rapidly due to market and environmental pressures. In recent years there has been significant emphasis on the development of water-based and so-called ‘green’ formulations and coatings [1]. There is, especially in Europe, a massive shift towards waterborne coatings, as they are perceived as environmentally friendly. This is even further encouraged by government regulations related to issues such as volatile organic compound (VOC) emissions [2]. In the field of thermoplastic composites, nanoparticle additives have been the subject of active research for many years, whereas this is a rather new avenue for coatings. Due to their relatively low thickness (around 30 microns), paints and coatings are not thought of as composites, even in the presence of filler microparticles such as pigments. Pigments may reinforce the coatings, but their main intended contributions are opacity and UV-absorption functions, properties for which they are designed in the first place.

444 | Part III New trends in sustainable development and biomedical applications

Fig. 16.1. Parquet floor with Al2 O3 UV-cured coatings resistant to wear and scratching.

Fig. 16.2. Building with UV-resistant coatings (Maibec Inc., Québec, Canada).

Wood surface characteristics Wood is a porous, hydrophilic, and anisotropic material which is widely used because of its availability in a great variety of species and surface textures (grain), elegant appearance, and ease of processing. It is by far the most widely used biosourced material in the world. It is eco-friendly, is an efficient carbon sink and comes from a renewable resource. Most wood products, except those used for structural purposes, are coated

16 Nanocomposite coatings

| 445

during processing for protection against humidity and wear. Wood is especially sensitive to ultraviolet light (UV) degradation, as one of its main polymeric constituents is lignin, an unsaturated molecule which readily absorbs light in the UV range, resulting in its degradation. In its natural configuration, wood is shielded and protected from such degradation by the layer of bark surrounding the soft core of the tree. When used indoors, wood must also be protected against wear by a protective coating, especially when used in flooring applications. Coatings are usually more hydrophobic than the wood itself. Initial roughness of the substrate, even after careful sanding, is in the sub-millimeter (