Handbook of Sol-Gel Science and Technology 9783319194547

Citation preview

The Synthesis and Solution Stability of Alkoxide Precursors Vadim G. Kessler

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis of Homometallic Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reactions of Metals with Alcohols (Method 1.1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anodic Oxidation of Metals (Method 1.2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reactions of Metal Oxides or Hydroxides with Alcohols (Method 1.3) . . . . . . . . . . . . . . . . . . . . Alcoholysis of Metal Derivatives of Weak or Volatile Acids (Method 1.4) . . . . . . . . . . . . . . . . . Metathesis Reactions with Alkali Alkoxides and Ammonia or Amines (Method 1.5) . . . . . Alcohol Interchange Reactions (Method 1.6) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Redox Processes in Approach to Alkoxide Precursors (Method 1.7) . . . . . . . . . . . . . . . . . . . . . . . Synthesis of Heterometallic Alkoxide Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heterometallic Alkoxides Formed via Lewis Acid-Base Interaction . . . . . . . . . . . . . . . . . . . . . . . . Heterometallic Alkoxides Formed via Formation of Heteronuclear Metal-Metal Bonds or Isomorphous Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solution Stability with Respect to Formation of Oxoalkoxides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Partial Hydrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oxidation by Oxygen from Atmosphere or Dissolved in Solvents . . . . . . . . . . . . . . . . . . . . . . . . . . Ether and Ester Elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . β-Hydrogen Elimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Desolvation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solution Stability with Respect to Solvolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Individual Alkoxide Complexes Applied as Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Homometallic Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Heterometallic Precursors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Chemistry of Alkoxide Precursors’ Transformation into Oxide Materials in Sol-Gel Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 3 5 6 7 9 10 12 12 13 17 18 18 19 20 21 22 22 24 25 25 34 39

V.G. Kessler (*) Department of Chemistry and Biotechnology, Swedish University of Agricultural Sciences (SLU), Uppsala, Sweden e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_1-1

1

2

V.G. Kessler

Abstract

The aim of this chapter is to serve as a guide in understanding the principles in the chemical approaches to and stability of the metal and silicon alkoxide precursors. The major accent is made on the overview of the accessible precursor compounds applicable in solgel technology, focusing on their synthesis and reactivity including the chemistry behind the solgel process. The chapter provides an insight into the synthesis of homometallic precursors (see section “Synthesis of Homometallic Precursors”), synthesis of heterometallic precursors (see section “Synthesis of Heterometallic Alkoxide Precursors”), solution stability with respect to formation of oxoalkoxides (see section “Solution Stability with Respect to Formation of Oxoalkoxides”), solution stability with respect to solvolysis (see section “Solution Stability with Respect to Solvolysis”), and, finally, a short review summarizing the literature data on individual alkoxide complexes applied as precursors (see section “Individual Alkoxide Complexes Applied as Precursors”) considering separately the homometallic species (see section “Homometallic Precursors”) and the heterometallic ones – so-called singlesource precursors (see section “Heterometallic Precursors”). Finally, a brief overview of the modern concepts treating the transformation of precursors into materials is also provided (see section “The Chemistry of Alkoxide Precursors’ Transformation into Oxide Materials in Sol-Gel Technology”).

Introduction The development of solgel technology has at a very early step put forward a request on development of precursor compounds – chemical substances that have high solubility in organic solvents are easily transformed into chemically reactive forms of hydrated oxides on hydrolysis. They should display considerable stability in solution to guarantee the reproducibility of the materials preparation and, last but not the least, be easy to be purified to provide sufficient chemical quality of the final products. Metal alkoxides, M(OR)n, are derivatives of alcohols, ROH, which are usually easily accessible and inexpensive organic compounds and are extremely weak as acids, easily removable via hydrolysis and thermal treatment, leaving high purity hydrated oxides. This circumstance made metal alkoxides the most common candidates for the role of molecular precursors (Veith 2002; Jones 2002; HubertPfalzgraf 2003; Kessler 2003). The works in this field during the last 20 years, including both the studies of the molecular and crystal structure and the reactivity of these compounds, have considerably changed their image in the eyes of both chemists and the materials scientists: it turned out that sometimes the compounds that are the most stable products in the reactions of synthesis of metal alkoxides and that were earlier considered to be M(OR)n are in fact oxoalkoxides MOx(OR)y. In many cases, especially for the preparation of complex solutions, including derivatives of several metals, it turned out impossible to use only the derivatives of aliphatic alcohols, CnH2n+1OH, because of their poor solubility, stability, or reactivity. This gave

The Synthesis and Solution Stability of Alkoxide Precursors

3

rise to the development of two new types of alkoxide precursors – derivatives of functional alcohols (alkoxyalcohols and aminoalcohols), on one hand, and heteroleptic alkoxides, including other ligands (such as β-diketonate, carboxylate, and aminoalkoxide ones) in addition to common aliphatic alkoxide groups, on the other. This change in the understanding of nature of alkoxide complexes has been reflected even by the titles of the modern textbooks on this topic called “Alkoxide and Phenoxide Derivatives of Metals” (Bradley 1965; Bradley et al. 2001) and “The Chemistry of Metal Alkoxides” (Turova et al. 2002). The complexity of situation has been increased even more by the rise of a still quite small but quickly growing family of alcoholates – highly soluble complexes of metal carboxylates or β-diketonates with functional alcohols. The latter do not contain formally the alkoxide ligands but are related to metal alkoxides in many of their properties and find the increasing application in solgel technology. The aim of this chapter is to serve as a guide in understanding the principles in the chemical approaches to and stability of the metal alkoxide precursors known already in fact for all the elements of the periodic table, excluding only the highly radioactive ones. The major accent is made on the laboratory approaches to the soluble and chemically reactive alkoxide derivatives, applicable in solgel technology. The chapter includes synthesis of homometallic precursors (see section “Synthesis of Homometallic Precursors”), synthesis of heterometallic precursors (see section “Synthesis of Heterometallic Alkoxide Precursors”), solution stability with respect to formation of oxoalkoxides (see section “Solution Stability with Respect to Formation of Oxoalkoxides”), solution stability with respect to solvolysis (see section “Solution Stability with Respect to Solvolysis”), and, finally, a short review summarizing the literature data on individual alkoxide complexes applied as precursors (see section “Individual Alkoxide Complexes Applied as Precursors”) considering separately the homometallic species (see section “Homometallic Precursors”) and the heterometallic ones – so-called single-source precursors (see section “Heterometallic Precursors”). The custom synthesis is a challenging task and requires application of anhydrous solvents and, what is absolutely crucial, an inert moisture-free atmosphere. A proper infrastructure including Schlenk lines of glove boxes needs to be available for success of this work. It has to be noted that a broad variety of precursors are commercially available nowadays (for details, please, see section “Individual Alkoxide Complexes Applied as Precursors”), but need to be stored and handled properly to succeed with their application in solgel preparation of materials.

Synthesis of Homometallic Precursors Reactions of Metals with Alcohols (Method 1.1) Direct Reaction in Inert Atmosphere (Argon or Nitrogen) Direct reaction with alcohols with evolution of hydrogen gas and formation of metal alkoxides is possible only for the most electropositive metals such as alkali, magnesium and alkaline earth, rare earth metals, and aluminum:

4

V.G. Kessler

M þ nROH ! MðORÞn þ n=2 H2 ðgÞ

(1)

In fact the reaction proceeds readily enough at room temperature only for the alkali metals and most acidic alcohols such as MeOH and EtOH or the functional ones such as 2-methoxyethanol. Heating is usually indispensable to lead reaction to the completion even in the case of alkaline and alkaline earth metals. The readiness of an alcohol to react with metals may depend strongly on its purity. Thus, comparably small water contents, such as less than 0.5 wt%, may, for example, leave ethanol almost inert in reaction with magnesium or calcium even on reflux, while ethanol with less than 0.01 wt% water has an observable reactivity toward these metals even at room temperature. To increase the reaction temperature in the case of magnesium or rare earth metals, the reaction is often carried out in a mixture of a parent alcohol (usually iPrOH in the latter case) and a high-boiling point hydrocarbon (toluene or even xylene, in 1:1 or 1:2 volume ratio to alcohol). The reaction requires the use of a catalyst for the alkaline earth metals, rare earth metals, and aluminum. The most common approaches are the use of (in the laboratory practice only) the salts of mercury(II) such as HgCl2 or Hg(OAc)2. Very small portions of these salts cause amalgamation of the metal surface (and thus clean it from the oxide layer) and facilitate the reaction with alcohols. The larger-scale synthesis (and thus the industrial one – in the scope of pollution danger) uses the initial addition of solid iodine (1 g or less per 100 g of alkoxide to be prepared). Formation of metal iodide serves both for cleaning the surface and increases also slightly the acidity of alcohols via formation of solvate complexes. In the case of barium, the application of dry ammonia gas has been reported for this purpose (Caulton et al. 1990; Drake et al. 1992). The major factor facilitating the reaction of metals with alcohols is the solubility of the alkoxides formed. Insoluble alkoxides form a protective layer on the surface of the metal and it hinders the reaction. Even the reaction of sodium with tBuOH in toluene may be almost stopped by the formation of poorly soluble NaOtBu. It is to be mentioned that the reaction of metals with an excess of alcohol leads normally, except for aluminum, to formation of solvates with corresponding alcohols, such as Li(OEt)2EtOH, Mg(OMe)23.5MeOH, Ca(OEt)24EtOH, [Nd(OiPr)3 (iPrOH)]4, etc. (see 1.5). To obtain non-solvated species, the reaction should be carried out with a stoichiometric amount of alcohol in a different solvent (most often, toluene) (Fisher and McElvain 1934). It is also important to notice that the reaction products quite often contain impurities of oxoalkoxides resulting from the presence of residual oxygen in the solvents or quite complicated redox side reactions. If such oxoalkoxides possess considerable thermodynamic stability, as, for example, Ba6O (OC2H4OMe)10 (MeOC2H4OH)4 (Caulton et al. 1993) or Ln5O(OiPr)13, where Ln = Sc, Y, or lanthanides (Hubert-Pfalzgraf et al. 1997a), their formation cannot be avoided, and they will in any case be isolated as the major reaction product and may be purified further by recrystallization. The reaction of metals with alcohols in inert atmosphere (except for the alkali ones) leaves very often a dark residue of unreacted small particles of metal or metal suboxides. This kind of residue is almost inevitable for aluminum and rare earth metals and can be simply removed by decantation at the end of the reaction.

The Synthesis and Solution Stability of Alkoxide Precursors

5

Oxidation of Metals by Oxygen Gas in Alcohol Media This approach offers only extremely highly soluble and stable alkoxide complexes with rather high resistance to hydrolysis. It was first applied to the preparation of thallium(I) ethoxide, carried out in a Soxhlet filter: 2Tl þ EtOH þ 1=2 O2 ! TlOEt þ TlOH

(2)

The reaction results in formation of a double-layer system, where the bottom layer is a 95 wt% solution of TlOEt in EtOH, while the upper one is containing almost all TlOH (Turevskaya et al. 1975). Oxidation by oxygen in air and solvents turned also to be a useful tool in approach to the copper(II) derivatives of aminoalcohols. The hydrolytic stability of Cu (II) aminoalkoxides appears to be its driving force. It is also apparently the same for the reaction of copper metal with alcohols in the presence of N-donor ligands, L, and considerable excess of alcohol (Kovbasyuk et al. 1998): Cu þ excess ROH þ nL þ 1=2 O2 ! CuðORÞ2  nL  mROH þ H2 O

(3)

Anodic Oxidation of Metals (Method 1.2) Considered initially as, in general, a simple extension of the direct reaction with alcohols for less active metals by application of an anodic potential, the anodic oxidation of metals turned to be a much more complicated process. At present, at least three different oxidation mechanisms have been proposed for different groups of metals: – The most active metals, such as lanthanides, receive really just a support for the direct interaction with alcohols (2-propanol in this case) from the applied anodic potential supposedly via the elimination of the oxide barrier. The electric current yields (the ratio of the alkoxide obtained to the total charge that passed through the system) often exceed 100 %. High concentrations of soluble conductive additives (LiX or R4NX, where X = Cl, Br), which contaminate the product have to be removed by repeated recrystallization from hydrocarbon solvents. – The late transition and main group metals follow the anodic oxidation pathway analogous to that in aqueous solutions. The minimal oxidation potentials in these cases can in fact be very low (up to max. 3.0 V), while the higher ones are readily applied to accelerate the process. The anodic reaction consists of dissolution of metal ions in the form of anionic halide complexes, which are later transformed into insoluble alkoxides by reaction with alkoxide anions generated at the cathode, for example (Lehmkuhl et al. 1975): Cathode : ROH þ e ! RO þ 1=2 H2 ðgÞ Anode : Cu þ 4Cl  2e ! CuCl4 2



(4) (5)

6

V.G. Kessler 

Solution : CuCl4 2 þ 2RO ! CuðORÞ2 ðsÞ þ 4Cl

(6)

Only insoluble alkoxides can be obtained by this method because the soluble ones are normally reduced at the cathode, transforming the process into the electrochemical transport of the metal from anode to cathode. The products again are usually heavily polluted by halide admixtures and should be then washed repeatedly with alcohols to remove adsorbed conductive additives (Hubert-Pfalzgraf et al. 1997b). It has, however, been reported that application of amines (such as dipyridyl, phenantroline), giving rather stable insoluble complexes with Cd and Cu alkoxides, allows alkoxides free from halide admixtures to be isolated (Banait and Pahil 1986). – The early transition metals are dissolved via a complex mechanism involving oxidation of alkoxide ligands with formation of extremely reactive alkoxoradicals that in turn attack the metal, forming soluble alkoxide complexes already at the anode: M þ n RO  n e ! MðORÞn

(7)

The reaction has highest speed in the alcohols displaying highest electric conductivity, such as MeOH or MeOC2H4OH. Low concentrations of conductive additives applied in this case assure high purity of the final product. It is in fact very important to keep the concentrations of the additives in the interval 0.01–0.05 M as the high potentials applied cause the formation of free halogens that oxidize the alcohols and provide finally water as by-product, leading to the formation of oxoalkoxide impurities. The other impurity formed simultaneously is the alkoxide derived from the conductive additive, for example, lithium alkoxides from lithium halides. On interaction with the metal alkoxide, they provide heterometallic complexes. Thus, a whole series of different bimetallic Li-Mo and Li-W alkoxides have been isolated and characterized as by-products of the electrochemical syntheses of M(VI) alkoxides (Kessler et al. 1998b). Another source of oxoalkoxide impurities is the cathodic reduction, which transforms low oxidation state impurities into oxoalkoxides via subsequent re-oxidation by oxygen dissolved in solvents. Following the optimized procedures, it is possible, however, to produce rather high-quality methoxide derivatives of Nb (Turevskaya et al. 1995a, b, c), Ta (Turova et al. 1996), Mo (Kessler et al. 1993), W (Seisenbaeva et al. 2001a), and Re (Seisenbaeva et al. 2001b).

Reactions of Metal Oxides or Hydroxides with Alcohols (Method 1.3) This reaction is useful for preparation of alkoxides from most basic or most acidic oxides and hydroxides. The alkoxides obtained should have quite high hydrolytic and thermal stability, because water formed during the reaction is removed by distillation as an azeotrope with an aromatic hydrocarbon solvent (usually toluene).

The Synthesis and Solution Stability of Alkoxide Precursors

7

In the laboratory practice, it can be applied for the preparation of phenoxides of alkali or alkaline earth metals, for example: NaOH þ PhOH ! NaOPh þ H2 O

(8)

Recent crystal structure studies have shown that the interaction of basic hydroxides with aliphatic alcohols does not lead to metal alkoxides but to alcohol solvates of the hydroxides. For example, the reaction of Ba(OH)2 with MeOH was found to provide Ba(OH)2(MeOH)2MeOH (Turova priv. comm.). This reaction has been in contrast successfully applied for the synthesis of alkoxide derivatives of acidic oxides, as the whole homologous series of vanadium alkoxides (Orlov and Voronkov 1959; Prandtl and Hess 1913), and for the preparation of a number of hydrocarbon-soluble complexes with diols of molybdenum (VI) (Bishop et al. 1979), rhenium(VI) (Edwards et al. 1998), and osmium (VI) (Lehtonen et al. 1999): V2 O5 þ 6 ROH ! 2VOðORÞ3 þ 3H2 O, R ¼ Et  C5 H11 MoO3 þ 3½MeCHðOHÞ2 ! ðMeCHOÞ2 MoO2  2RðOHÞ2 þ H2 O

(9) (10)

Re2 O7 þ 4HOCMe2 CMe2 OH ! 2ReO3 ðOCMe2 CMe2 OHÞðHOCMe2 CMe2 OHÞ (11) The reaction achieves completeness, when the alkoxides thus formed display quite considerable resistance to hydrolysis and can then be purified by some efficient technique (distillation for vanadium derivatives and recrystallization for the diolates).

Alcoholysis of Metal Derivatives of Weak or Volatile Acids (Method 1.4) The action of alcohols on the metal derivatives of extremely weak and, which is of special importance, highly volatile acids, for example, alkyls, carbides, nitrides, amides, alkyl amides, silazides, hydrosulfides, hydrides, etc., provides an approach to high purity samples of metal alkoxides, usually under extremely mild conditions. The reaction MXn þ nROH ! MðORÞn þ nHX X ¼ H, Alk, C= 2C, 2 = N, NH2 , NR2 , NðSiR3 Þ2 , SH

(12)

is usually carried out in a volatile hydrocarbon solvent (such as hexanes or pentane) and the products are purified by evaporation of the byproducts and the solvent in vacuum, leaving the target alkoxide as the residue. Hydrides can be used as sources of alkaline metal alkoxides (LiH, NaH) in the reactions with halogenated alcohols, such as, (CF3)3COH, to avoid the danger of condensation of Wurtz type (Dear

8

V.G. Kessler

et al. 1970). Metal alkyls have been applied earlier for the preparation of a number of early transition metal derivatives, for example, CuOMe (Costa et al. 1965), Cr(OR)2 (Chisholm et al. 1979), and V(OtBu)4 (Razuvaev and Drobotenko 1977), but are themselves extremely unstable and normally not commercially available, which precludes their application in common laboratory practice. The most broadly applied laboratory approach of this type is the reaction of metal alkyl amides, usually bis-(trimethylsilyl)-amides with the stoichiometric amounts of alcohols. The starting reagents even in this case are not available commercially but can be obtained more or less easily by reaction of the corresponding metal chlorides with commercially available LiN(SiR3)2 in anhydrous diethyl ether:
 MCln þ nLiNðSiR3 Þ2 ! M NðSiR3 Þ2 n þ nLiCl

(13)

M[N(SiR3)2]n can then be purified – for the main group derivatives (for application in the synthesis of alkoxides, see Zn (Goel et al. 1990a), Cd (Boulmaaz et al. 1992), Pb (Matchett et al. 1990; Papiernik et al. 1989), and Bi (Massiani et al. 1990; Goel et al. 1990a)) – by sublimation direct from the reaction mixture, after removal of the Et2O in vacuum, and (for the early transition metal compounds (Cr(II), Mn (II) (Horvath et al. 1979)), after the removal of ether) by the extraction from the residue with pentane or hexanes, separating LiCl by decantation. It should be mentioned that this approach is hardly practically applicable for the synthesis of the derivatives of late transition metals such as Co, Ni, or Cu because of poor stability of their amide derivatives (Bryndza and Tam 1988). It should be mentioned that the reaction of metal chlorides with alcohols could not be applied for the synthesis of metal alkoxides – precursors of oxide materials. Its products are usually quite complex mixtures of alkoxide chlorides and alcohol solvates of metal oxochlorides (Turova et al. 2002; Turevskaya et al. 1989). Formation of alcoholates – solvate complexes with functional alcohols – can be considered as a variety of this synthetic approach. Metal β-diketonates or carboxylates are reacted with amino- or alkoxy-alcohols in stoichiometric amounts in organic solvents (both nonpolar, such as toluene or hexane, or polar, such as methanol or ethanol, can be applied (Williams et al. 2001; Seisenbaeva to be published): NiðOAcÞ2  4H2 O þ 2HOC2 H4 NMe2 ! NiðOAcÞ2 ðHOC2 H4 NMe2 Þ2 þ 4H2 O

(14)

MnðacacÞ2  xH2 O þ 2HOCHMeCH2 NMe2 ! MnðacacÞ2 ðHOCHMeCH2 NMe2 Þ

(15)

The advantage of the alcoholate complexes lies in their high solubility in organic solvents. They provide also a possibility to avoid more complicated dehydration procedures necessary for the derivatives of late transition metals to be used for the preparation of complex solutions together with metal alkoxides.

The Synthesis and Solution Stability of Alkoxide Precursors

9

Metathesis Reactions with Alkali Alkoxides and Ammonia or Amines (Method 1.5) This approach remains the most commonly applied in the synthesis of metal alkoxides. The starting reagents are the anhydrous metal halides, most often chlorides, or other anhydrous metal salts, such as nitrates or acetates: MXn þ nROH þ nR3 N ! MðORÞn þ nR3 NHX

(16)

MXn þ nMI OR ! MðORÞn þ nMI X

(17)

The traditional technique using ammonia gas has been applied for the preparation of the alkoxides of titanium (Demarcay 1875), zirconium and hafnium (Bradley et al. 1952), cerium (IV) (Bradley and Holloway 1962), and niobium and tantalum (Bradley et al. 1956a, b). In this approach, a halide or a pyridinium halogenometallate salt, for example, (PyH)2ZrCl6, is dissolved in a mixture of toluene with the parent alcohol, and ammonia gas is bubbled through the solution for several hours. The voluminous precipitate of NH4Cl is removed by filtration and washed with the alcohol on the filter to improve the yield of the soluble alkoxide. To avoid the use of ammonia gas and simplify the procedure as a whole and specifically the separation of the ammonium salts, there has been proposed to use the amines such as triethylamine or pyridine in the same purpose. This route provided access, for example, to the stable samples of MoO(OiPr)4 from MoOCl4 (Chisholm et al. 1984) and those of Re2O3(OMe)6 from ReOCl4 (Edwards et al. 1980). It is necessary to mention that neither ammonia nor amines can be applied for the preparation of pronouncedly basic alkoxides – derivatives of alkali, alkaline earth, or rare earth metals (their formation is impossible in the presence of acidic ammonium salts). A specific problem in application of ammonia or amines lies in the need of introducing a metal halide into this reaction as a solution in a solvent mixture including the parent alcohol. Strong Lewis acids such as metal halides are at room temperature prone to convert the alcohols, especially the ramified ones, into alkyl halides and transforming themselves into oxohalides (Turova et al. 2002). This side reaction decreases the yield of the target products and, when the tertiary (Bradley et al. 1978) or aromatic (Niederberger et al. 2002) alcohols are used, can lead (in not completely anhydrous conditions) even to formation of oxides or hydrated oxides. This effect can be avoided if the halides are introduced as solutions in aprotic solvents (toluene, ether, THF) into the solutions of alkali alkoxides, for example: FeCl3 =toluene þ 3NaOEt=EtOH ! FeðOEtÞ3 þ 3NaCl

(18)

BiCl3 =ether þ 3NaOEt=EtOH ! BiðOEtÞ3 þ 3NaCl

(19)

Strong cooling is always recommended at the initial step of this process. Then the reaction mixtures are usually warmed to room temperature after the complete

10

V.G. Kessler

addition of the halide and then even often subjected to reflux in order to destroy the possible heterometallic impurities. The heterometallic impurities are sometimes so stable, for example, NaZr2(OR)9, that they can even be distilled in vacuum without decomposition (Bartley and Wardlaw 1958). It is important to avoid the possible deviations in the reaction stoichiometry, but even the perfect one does not guarantee the purity of the obtained samples because the alkoxide chlorides or bimetallic alkoxide chlorides can sometimes display really high stability. For example, Y3(OtBu)8Cl2THF or Nd6(OiPr)17Cl have been isolated as the major products in the reaction of the corresponding trichlorides with three equivalents of NaOR (Evans et al. 1988; Andersen et al. 1978). In many cases, larger halide ligands (Br or I instead of Cl) or a larger alkaline metal atoms (K instead of Na or Li) can help to avoid the side reactions of this kind, for example, in Kritikos et al. (2001),   5LnCl3 þ 15KOi Pr þ H2 O ! Ln5 O Oi Pr 13 þ 2iPrOH

(20)

Metal acetates have been used as reagents in the metathesis with alkali alkoxide mainly in order to produce the main group metal derivatives, for example, the homologous series of lead alkoxides (Papiernik et al. 1989; Turevskaya et al. 1982). The reaction produces the insoluble sodium acetate that can be removed by filtration or decantation. It should be mentioned that, when carried out in toluene (on reflux), it could very easily lead to oxoalkoxide derivatives via ester or ether elimination side reactions (see section “Synthesis of Heterometallic Alkoxide Precursors”). Application of the nitrate complexes has been proposed in the metathesis-based approaches to the derivatives of Ce(IV) in the view of their much higher stability and commercial availability (Gradeff et al. 1985). It is necessary to mention that during the development of metathetic approaches, a number of alkoxylating agents other than the alkali alkoxides have been tested in this purpose. For example, the gas phase co-condensation of volatile metal fluorides or chlorides with alkylsiliconalkoxides has been reported for the preparation of M (OMe)6, M = Mo, W, and Re (Jacob 1982; Bryan et al. 1991). This technique requires a special equipment and provides rather small quantities of these products that can be obtained much more easily by the anodic oxidation of the corresponding metals. Another example of a different alkoxylating agent is the soluble Mg(OMe)2, which has been used to produce the methoxides from the metal fluorides (Bryan et al. 1991).

Alcohol Interchange Reactions (Method 1.6) The equilibrium reactions of metal alkoxides with alcohols: MðORÞn þ R’OH Ð MðORÞn1 ðOR’Þ þ ROH;

(21)

The Synthesis and Solution Stability of Alkoxide Precursors

11

are often very flexible. It is important to keep this fact in mind in developing the procedures for solgel applications of the alkoxide precursors: when dissolved in other alcohols than the parent one, they would undergo a ligand exchange that can change their molecular structure and hydrolytic properties and, in case of the heterometallic compounds (to be used as single-source precursors), even their chemical compositions. This method is useful as a synthetic procedure, when an efficient synthetic approach or commercial availability provides a different homologue than the one desired for further applications. The completeness of the reaction is achieved more easily if the desired product possesses much lower solubility in the new parent alcohol, for example:     ZrðOn PrÞ4 þ excessi PrOH ! ZrðOn PrÞ Oi Pr 3 i PrOH

(22)

or if the alcohol to be introduced has a considerably higher boiling point, which facilitates the removal of the other alcohol, formed in the reaction, by vacuum distillation, for example (Johansson et al. 2000):   MoOðOMeÞ4 þ excessi PrOH ! MoO Oi Pr 4 þ 4MeOH

(23)

The treatment with the new alcohol must be repeated several times (with complete dissolution of the crude intermediate product) to insure the completeness of transformation. It should be mentioned that the completeness in the exchange of the alkoxide groups might not always be achieved: the stability of molecular structures including small bridging ligands or functional chelating alkoxide ligands prevent in many cases the possibility to replace them, for example (Bradley et al. 1978; Johansson et al. 2000):   TaðOMeÞ5 þ excessi PrOH ! TaðOMeÞ Oi Pr 4 þ 4MeOH MoO2 ðOC2 H4 OMeÞ2 þ excess EtOH ! recrystallization without substitution (24) Replacement can also be achieved by other sources than alcohols, for example, esters. This is of interest if the boiling points of the two alcohols are very close (e.g., i PrOH and tBuOH, b.p. 82.4  C) or when the alcohol to be applied is highly unstable (e.g., silanols, unsaturated alcohols, etc.):   Zr Oi Pr 4 þ 4AcOt Bu ! ZrðOt BuÞ4 þ 4AcOi Pr

(25)

  Ti Oi Pr 4 þ 4AcOSiR3 ! TiðOSiR3 Þ4 þ 4AcOi Pr

(26)

The reactions with esters are most often carried out in aromatic hydrocarbon solvents to decrease the reaction temperature by removing a more volatile azeotropic mixture of a new ester with, for example, toluene or (in the past) benzene (Bradley et al. 1978; Bradley and Thomas 1959).

12

V.G. Kessler

A specific class of the ligand exchange reactions and, in some cases, of alcohol interchange improves the solubility and behavior in the hydrolysis and subsequent gelation processes. The reaction can in general be written as follows: MðORÞn þ mHZ ! MðORÞnm Zm þ mROH

(27)

where HZ represents aminoalcohols or other functional alcohols, β-diketones, or carboxylic acids. These reactions have been rather thoroughly studied for the derivatives of M(IV) such as titanium, zirconium, and cerium and are described in a number of recent review articles (Ribot et al. 1991; Hubert-Pfalzgraf 2003; Jones 2002).

Redox Processes in Approach to Alkoxide Precursors (Method 1.7) The redox reactions do not in fact belong to the common approaches in preparation of precursors for the solgel technology. The only example worth noting here is the oxidation of low-valent chromium derivatives (dibenzene-chromium, Cr(OR)3) by the t-butylperoxide, providing access to chromium(IV) alkoxides – highly soluble and volatile compounds (Krauss and Munster 1967). On the other hand, the redox reactions are in many cases responsible for the transformation of metal alkoxides in solutions leading to formation of oxoalkoxides and will be discussed below in section “Solution Stability with Respect to Formation of Oxoalkoxides.”

Synthesis of Heterometallic Alkoxide Precursors The special interest in heterometallic alkoxide complexes is due to the possibility of their application as single-source precursors in the preparation of complex inorganic materials (oxides, sulfides, metal alloys, and even nanocomposites) (Veith 2002). The single-source precursor represents a compound containing the necessary elements in desired stoichiometric ratio. Synthesis and properties of the heterometallic alkoxides have been described in detail in a number of recent reviews (Veith 2002; Jones 2002; Hubert-Pfalzgraf 2003; Caulton et al. 1990; Kessler 2003). The formation of heterometallic complexes in general can occur due to one of the three following factors: 1. Lewis acid-base interaction (exploiting the difference between two or several metal atoms in electronegativity, which permits to consider one metal center as a stronger acceptor of the electron density and the alkoxide or other ligands at the other as a better donor of it) 2. Formation of a heterometallic metal-metal bond, which in this case should also provide a donor-acceptor interaction

The Synthesis and Solution Stability of Alkoxide Precursors

13

3. Isomorphous substitution, which might in some cases not lead to formation of the true heterometallic species, but provides in any case the homogenization at the molecular level The synthetic approaches to heterometallic complexes will be classified here below according to these three principles providing their formation.

Heterometallic Alkoxides Formed via Lewis Acid-Base Interaction Complex Formation Between Two Alkoxides (Method 2.1) The pronounced Lewis basicity of the alkoxide ligands of the alkali and alkaline earth metal alkoxides explains their capacity to form heterometallic complexes in solution with the vast majority of high-valent transition or main group metal alkoxides, for example: LiOR þ NbðORÞ5 ! LiNbðORÞ6

(28)

The chemical composition of the products is determined by a number of important factors, such as the nature of metal atoms involved, the nature of alkoxide groups, the ratio of homometallic reactants applied, and, in certain cases, even on the solvent in which the reaction is carried out. For example, the reaction of barium ethoxide and titanium ethoxide in 1: x (where x  2) ratio in oxygen-free solutions provides different products in alcohol and in hydrocarbon media (Yanovsky et al. 1995; Kessler et al. 1994a): EtOH

BaðOEtÞ2 þ > 2TiðOEtÞ4 ! BaTi2 ðOEtÞ10 ðEtOHÞ5 toluene

BaðOEtÞ2 þ > 2TiðOEtÞ4 ! BaTi4 ðOEtÞ18

(29) (30)

Formation of different complexes at different ratios of reactants in the same solvent (THF in this case) can be illustrated by the examples from barium-zirconium isopropoxide system (Vaartstra et al. 1991) (the authors used barium metal in the presence of solvating alcohol, but application of the ready alkoxide gives the same result (Turevskaya et al. 1995a)): h       i Ba þ Zr2 Oi Pr 8 i PrOH 2 ! BaZr2 Oi Pr 10 þ H2 ðgÞ 2

h       i Ba þ 2Zr2 Oi Pr 8 i PrOH 2 ! Ba Zr2 Oi Pr 9 þ H2 ðgÞ þ 2i PrOH 2

(31) (32)

Synthesis of heterometallic complexes by direct interaction of homometallic alkoxides has been reported in many cases also for the rare earth metals, but in this case, it is necessary to keep in mind that the commercial “Ln(OiPr)3” usually

14

V.G. Kessler

contain the oxoalkoxide complex, Ln5O(OiPr)13, as their major component. The reactivity of the latter toward other alkoxides is comparably low, and prolonged refluxing in toluene or the reaction in a melt is recommended to insure the completeness of transformation (Poncelet et al. 1989a, b). The only reaction between the two high-valent metal alkoxides, not involving specific mechanisms with formation of oxoalkoxides, is the formation of the aluminum and hafnium isopropoxide (Turevskaya et al. 1997):         Hf Oi Pr 4 i PrOH þ 2Al Oi Pr 3 ! HfAl2 Oi Pr 8 þ 2i PrOH

(33)

It is also important to notice that this reaction takes place even in the alcohol media, but gives in this case far not quantitative yields of the product, whose formation took several days.

Metathesis of a Metal Halide with a Bimetallic Alkoxide of Another Metal and an Alkali Metal (Method 2.2) This approach has been proposed for the case, where the alkoxide of one of the metals is not easily accessible or is an insoluble and inert solid as, for example, the alkoxides of late transition and some main group metals (Mn(II), Fe(II), Fe(III), Co, Ni, Cu, Zn, Cd, Sn(II), Pb(II)). It has been applied also for the preparation of heterometallic derivatives of rare earth elements (see Bradley et al. 1978, 2001; Turova et al. 2002). It is necessary to take into account that the reaction:
 MXn þ MI M’ðORÞmþ1 ! M M’ðORÞmþ1 n þ nMI X;

(34)

not always follows the simplified reaction formula given by the Eq. 34. The so-called alkoxometallate ligands, existing in the structures of heterometallic alkoxides of alkali metals, such as [Al(OR)4], [Zr2(OR)9], or [Nb(OR)6], are not the ultimately thermodynamically stable ionic units. They are therefore very rarely just transferred by this reaction from an alkali metal cation to a less electropositive metal cation. The deviations may occur both due to formation of more stable oxocomplexes (Boulmaaz et al. 1994) and even more stable homoleptic alkoxide complexes (Kessler et al. 1994b; Yanovsky et al. 1994), for example:     ZnI2 þ 2NaTa Oi Pr 6 ! Zn2 Ta4 O4 Oi Pr 16       LaCl3 þ 3NaNb Oi Pr 6 ! LaNb2 Oi Pr 13 þ Nb Oi Pr 5 þ 3NaCl

(35) (36)

A serious problem that can also be associated with application of this technique lies in the possibilities of formation of by-products including the halides or alkali metals (or even both as in case of formation of [NaPb2Ti2O(OiPr)10]+Cl (Hubert-Pfalzgraf private comm.). A family of the techniques described below is based on the introduction of new ligands such as oxo-groups or different organic residues such as β-diketonate,

The Synthesis and Solution Stability of Alkoxide Precursors

15

carboxylate, or aminoalkoxides groups. Successful development of new approaches of this kind needs prediction of the chemical composition and even the structure of the new complexes to be prepared and choice of the proper reaction stoichiometry. This prediction can be done using the Molecular Structure Design Concept, described in detail by Kessler (2003), and including the following steps: 1. Choice of the structure type to be used (see Fig. 1) 2. Calculation of the necessary number of the donor atoms 3. Choice of the ligands with proper composition and sterical requirements that provide both the right number of donor functions and the protection of the chosen core of metal and donor atoms (placing the metal atoms into the thermodynamically preferred coordination polyhedra) The most complete classification of the stable structure types for the alkoxide complexes can be found in Turova et al. (2002). The practical principles in application of this concept lie either in decreasing the number of donor atoms by replacing two OR groups with an oxo-ligand and thus increasing the strength of the Lewis acids involved or by providing some additional donor atoms from bidentate (usually) chelating ligands, which are necessary to support the chosen structure type (Bradley and Holloway 1965).

Microhydrolysis of Alkoxides of Different Metals in Solutions (Method 2.3) This approach has been first applied to access to the heterometallic alkoxides of bismuth, because the homoleptic Bi(OR)3 usually does not form any heterometallic complexes with the alkoxides of other high-valent metals (Parola et al. 1997): BiðORÞ3 þ 2TiðORÞ4 þ H2 O ! BiTi2 OðORÞ9 , R ¼ Et, i Pr

(37)

This reaction was carried out successfully in an alcohol media. It is necessary to mention, however, that alcohols very often destroy the heterometallic complexes derived from elements with close Lewis acidity. The efficiency of microhydrolysis as an approach to heterometallic species is much better in inert media (hydrocarbon solvent).

Micropyrolysis of Alkoxides of Different Metals in Solutions (Method 2.4) When an alkoxide derivative can easily decompose thermally in solution forming an oxoalkoxide (see section “Solution Stability with Respect to Formation of Oxoalkoxides”), this reaction might be exploited in approach to heterometallic species. This reaction is useful especially for the synthesis of heterometallic derivatives of molybdenum (Johansson et al. 2000; Johansson and Kessler 2000a, b) and zirconium (Kessler et al. 1998a) (see also in 2.5), for example:

16

V.G. Kessler

Fig. 1 Schematic views of the most important building blocks in the structures of metal alkoxide aggregates

OR

OR M

OR

M

M OR

M

OR

M

M

OR

OR

OR

OR X

M

X

M

OR

OR

M

M

OR

OR

OR

M M

M

OR OR

OR OR

M OR

M

M

OR

OR

M

M OR M

OR

OR

      4MoO Oi Pr 4 þ 2Ta Oi Pr 5 ! Mo4 Ta2 O8 Oi Pr 14 þ 2ðCH3 Þ2 CO   þ 2i PrOH þ 4 i Pr 2 O         MoO Oi Pr 4 þ ”Zr Oi Pr 4 ” ! Zr3 Mo8 O24 Oi Pr 12 i PrOH 4

(38) (39)

It is important to underline that this approach is not applicable in the preparation of temperature-sensitive alkoxide derivatives such as those of Ni, Cu, Zn, Cd, Pb, or Bi, where the heating results in formation of metals or oxides (see section “β-Hydrogen Elimination”).

Interaction of Metal Complexes with Organic Ligands and Metal Alkoxides or Chemical Modification of Complexes in Solutions (Method 2.5) This approach provides an important alternative to the method 2.2 in the preparation of derivatives of late transition elements (the homometallic alkoxides of those being insoluble and not reactive polymeric solids). The reaction stoichiometry and conditions are dependent on the nature of reactants and on the composition of the product to be obtained. In some cases, the reaction is facile and provides the desired products

The Synthesis and Solution Stability of Alkoxide Precursors

17

as the result of mixing the reactants in proper ratio (usually in toluene) and subjecting them to short reflux, for example (Boulmaaz et al. 1997; Kessler 2003):     CdðOAcÞ2 þ 2Nb Oi Pr 5 ! CdNb2 ðOAcÞ2 Oi Pr 10

(40)

    CuðacacÞ2 þ 2Al Oi Pr 3 ! CuAl2 ðacacÞ2 Oi Pr 6

(41)

Preparation of the heterometallic complexes of zirconium from Zr(OiPr)4(iPrOH) requires prolonged refluxing in order to generate the reactive oxo-species (see section “Synthesis of Heterometallic Alkoxide Precursors”) (Hubert-Pfalzgraf 1994):       PbðOAcÞ2 þ Zr Oi Pr 4 i PrOH ! PbZr3 O Oi Pr 10 ðOAcÞ2

(42)

When the number of additional donor atoms of chelating ligands remains insufficient for the stabilization of the proper structure, an additional modification by chelating ligands is required (Kessler 2003; Kessler et al. 2003):     MII ðacacÞ2 þ 2Al Oi Pr 3 þ 2Hacac ! MII Al2 ðacacÞ4 Oi Pr 4 þ 2i PrOH (43) MII ¼ Co, Ni, Mg The procedure should be carried out in separate steps including mixing of the homometallic reagents, refluxing them in toluene, cooling down below the room temperature, and, only then, the addition of the necessary extra amount of acidic chelating agent (β-diketone or carboxylic acid). Introduction of an organic acid into a warm solution would result in an instant gelation (due to the condensation with released alcohols producing water in situ) instead of formation of heterometallic mixed-ligand complexes.

Heterometallic Alkoxides Formed via Formation of Heteronuclear Metal-Metal Bonds or Isomorphous Substitution No specific techniques have been elaborated for these particular cases. The majority of compounds of these two classes are obtained by complex formation between homometallic species (most often in hydrocarbon solvents) at ambient conditions or via short-term reflux (method 2.1). The formation of a metal-metal bond requires interaction of electron-rich low-valent derivatives of one metal with electron-deficient, high-valent derivatives of the other, for example (Chisholm et al. 1981; Kessler et al. 1995):       W2 Oi Pr 6 þ MoO Oi Pr 4 ! W2 MoO Oi Pr 10

(44)

ReOðOMeÞ3 þ MoOðOMeÞ4 ! ReMoO2 ðOMeÞ7

(45)

18

V.G. Kessler

In the simplest cases, the isomorphous substitution can be achieved via mixing the isostructural but chemically different species in solution (Hubert-Pfalzgarf et al. 1978): Nb2 ðOMeÞ10 þ Ta2 ðOMeÞ10 ! 2NbTaðOMeÞ10

(46)

When the bonding parameters of two metal atoms are analogous, but the molecular aggregates observed for the homometallic species are different, the formation of heterometallic complexes, following the structure type observed for only one of two elements, can be achieved, sometimes applying the solution thermolysis (method 2.4), for example (Seisenbaeva et al. 2001b): Re2 O7 þ MoOðOMeÞ4 ! Re4x Mox O6 ðOMeÞ12

(47)

It is important to mention that the same reactants (even in the same ratio) can provide different heterometallic products if different reaction temperatures are applied. For example, the interaction between NbTa(OMe)10 and Re2O7 in toluene at room temperature provides NbTa(OMe)8(ReO4)2, while the reflux of the same reaction mixture gives Nb2Ta2O2(OMe)14(ReO4)2 (Shcheglov et al. 2002).

Solution Stability with Respect to Formation of Oxoalkoxides One of the most important requirements put on application of molecular precursors in different technological procedures, and in solgel technology in particular, is the demand of stability during the application procedure itself and on storage. It is, therefore, very important to know what mechanisms can lead to the changes in the properties and molecular structures of precursors and what measures should be undertaken to support the stability of solutions and to achieve reproducibility in their application. We distinguish here between the reactions leading to transformation of alkoxide ligands resulting in formation of oxoalkoxides discussed in this part and the solvent-supported reactions of ligand redistribution (solvolysis), presented in the next one. The most important reaction pathways leading to oxoalkoxides are partial hydrolysis, oxidation by oxygen from the atmosphere and dissolved in solvents, ether and ester elimination, β-hydrogen elimination, and thermal desolvation.

Partial Hydrolysis Almost all metal alkoxides (with the exception of the kinetically rendered derivatives of precious metals and the most sterically hindered complexes) are extremely moisture sensitive. Interaction with water molecules from moist atmosphere or not properly dried solvents results in drastic changes in molecular complexity and chemical composition, for example (Ibers 1963; Bradley and Holloway 1962):

The Synthesis and Solution Stability of Alkoxide Precursors




TiðOEtÞ4



toluene

19

þ H2 O ! Ti7 O4 ðOEtÞ20

(48)

Nb2 ðOEtÞ10 þ H2 O!Nb8 O10 ðOEtÞ20

(49)

3

Different hydrolysis ratios, h (number of water molecules per correspondent alkoxide formula unit, M(OR)n), provide different aggregates. For example, for the titanium ethoxide, different conditions of partial hydrolysis have also provided such aggregates as Ti8O6(OEt)20 (Day et al. 1991), Ti10O8(OEt)24 (Day et al. 1991), and Ti16O16(OEt)32 (Mosset and Galy 1988). It is not always pointed out directly, but in the complex solutions, the microhydrolysis can turn out rather selective, transforming into oxoalkoxide species only one or few of the components and changing the stoichiometry of molecular precursors. The risk of uncontrolled hydrolysis should be then eliminated as thoroughly as possible: all the operations in the preparation and weighing the samples of alkoxides are to be carried out in dry atmosphere using a Schlenk line or a dry box. The solvents dried according to reliable techniques (see Errington 1997) have to be applied. It is necessary to take into account that the water molecules can appear not only due to improper drying of the system but can even be products of different side reactions. For example, they are formed on modification of (warm) alkoxide solutions with carboxylic acids (Steunou et al. 1998): RCOOH þ R’OH ! RCOOR’ þ H2 O    Ti Oi Pr 4 þ HOAc ! Ti6 O4 Oi Pr 8 ðOAcÞ8 þ . . . 

(50) (51)

Strict control of the reaction temperature and stoichiometry (modification ratio, x, – number of modifying ligand molecules per alkoxide formula unit) is very important to insure the reproducibility of further application of such solutions.

Oxidation by Oxygen from Atmosphere or Dissolved in Solvents The alkoxide groups possessing a hydrogen atom in α-position, i.e., at the first carbon atom connected to the oxygen – all primary and secondary ones – react with oxygen in basic media, forming the products of oxidation such as carbonyl compounds and water (Turova et al. 2002): RCH2 O þ O2 ! RCHðOOHÞO ! RCHO þ OH þ . . . (52) mMðOCH2 RÞn þ OH ! Mm OðOCH2 RÞnm2 þ RCH2 O þ RCH2 OH This means that the homoleptic (alkoxide-only) derivatives of alkali, alkaline earth, and rare earth metals are very sensitive in solution to the presence of even very small traces of oxygen. The reaction is proceeding with a radical mechanism, which results in a very intensive yellowish orange (in case of high concentrations of both basic alkoxide and oxygen – even brown) coloration of solutions. The reaction speed

20

V.G. Kessler

increases with the basicity of media (alkali > alkaline earth >> rare earth elements). It is much higher in alcohols than in hydrocarbon solvents and much higher for homometallic than for the heterometallic derivatives of these elements. Really rigorous precautions can (under laboratory conditions) provide formation of the samples free from oxidation products. For the alkaline earth or rare earth elements, these are always solvates with O-donor ligands such as alcohols, THF, or dme (dimethoxyethane) (Turova et al. 2002) as desolvation itself produces the oxo-species (see below). In order to provide the samples more stable in solutions, there have been reported numerous attempts of their chemical modification using acidic ligands such as β-diketonates (Arunasalam et al. 1995) or aminoalkoxides (Poncelet et al. 1991). One of the major trends in the recently reported solgel preparations of the derivatives of these elements lies in application of other organic precursors than alkoxides (β-diketonates, 2-ethylhexanoates) for the preparation of solutions or application of stable heterometallic alkoxide or heteroleptic complexes (Kessler 2003; Veith 2002; Hubert-Pfalzgraf 2003).

Ether and Ester Elimination The ether elimination reaction is a spontaneous decomposition process characteristic of, in the first hand, high-valent early transition elements, such as Mo, W, Re, Nb, and possibly Ta. The reaction mechanism involves at the first step a redistribution of electron density with a heterolytic cleavage of an O-C bond as a result. The liberated alkyl-cation is transferred to a neighboring terminal alkoxide group, forming an ether molecule: Od−_Rd+ M

O –...R+ M

OR

M=O + R2O OR

The reaction speed increases in the series of homologues Me < Et 3nm

a

b Amplification

Single crystal

Mesoscale assembly

c

Mesocrystal

Oriented Aggregation Iso-oriented crystal

Fusion

Fusion

Fig. 1 Schematic representation of classical and nonclassical crystallization. (a) Classical crystallization pathway, (b) oriented attachment of primary nanoparticles forming an iso-oriented crystal upon fusing, (c) mesocrystal formation via self-assembly of primary nanoparticles covered with organics (Reproduced with permission from reference Wohlrab et al. (2005). Copyright 2005 Wiley-VCH)

energy of metal oxide clusters, thus favoring nucleation; they provide a nonreactive shell around the particles, which favors the formation of nanosized crystallites and improve their colloidal stability. Moreover, they can orientate the crystal growth by capping specific crystal faces, which gives rise to anisotropic shapes (Djerdj et al. 2007; Garnweitner and Niederberger 2008). Accordingly, the size and shape (spheres, rods, platelets, multipods, etc.) of final crystalline nanoparticles depend on the choice of both molecular precursor and oxygen donor, providing an attractive versatility. In some cases, the presence of organics on the crystal faces has been shown to be able to move the crystal growth mechanism from classical nucleationand-growth process to nonclassical crystallization processes. These processes are driven by the self-assembly of organic groups through bonding interactions (e.g., Van der Waals or π-π interactions). The oriented attachment process involves the spontaneous aggregation of adjacent nanoparticles into superstructures. The arrangements of individual nanocrystals in a common crystallographic orientation result in mesocrystal formation (Fig. 1) (Antonietti et al. 2008; Niederberger and Cölfen 2006). Self-assembly of nanoparticles arising from nonhydrolytic methods, either in solution or on a substrate, gives access to functional solid superstructures, films, and patterned coatings.

10

A. Vioux and P. Hubert Mutin

Nonhydrolytic Chemical Routes to Metal Oxide Nanoparticles Benzyl Alcohol Route This method is straightforward and produces in good yields high-quality crystalline metal oxide nanoparticles with morphologies closely related to the starting metal reactant, without any surfactant additive. It is very versatile and may involve the reaction of benzyl alcohol with metal chlorides, alkoxides, acetates, or acetylacetonates. Depending on the reaction system, different mechanisms have been postulated (Niederberger et al. 2006a, b). Metal halide-benzyl alcohol system offers the advantage of a low operating temperature (typically between 40  C and 120  C), but residual halide impurities in the final oxide materials may be detrimental for applications such as catalysis or gas sensing. The solvothermal reaction of metal alkoxides in an autoclave at 180–250  C (the boiling point of benzyl alcohol is 205  C) offers an alternative halide-free route to nanocrystalline binary and ternary metal oxides. As a matter of fact, the reaction of tungsten chloride with benzyl alcohol yielded tungstite nanoplatelets (Polleux et al. 2005), whereas the reaction of tungsten isopropoxide yielded tungsten oxide nanowires self-assembled into bundles (Polleux et al. 2006). The uniform distance between the nanowires indicated that the attachment originated from intercalated organic molecules (most probably benzaldehyde based on FTIR spectroscopy, which arose from the oxidation of benzyl alcohol). The removal of these organic molecules by adding formamide resulted in the dispersion of individual nanowires in ethanol. Taking advantage of the in-situ formation of organic-inorganic hybrid structures, Pinna and coworkers prepared lanthanide-based lamellar hybrids with outstanding emission properties, in which oxide layers were regularly separated from each other by organic layers of intercalated benzoate molecules arising from in-situ oxidation of benzyl alcohol (Ferreira et al. 2006; Karmaoui et al. 2006). Note that the reducing ability of benzyl alcohol was applied to the synthesis of pure zero-valent metal instead of metal oxide nanostructures, by reacting nickel (Jia et al. 2008) or copper (Dar et al. 2012) acetylacetonates with benzyl alcohol under thermal or microwave activation. Actually, high boiling point and high dielectric loss factor make benzyl alcohol an appropriate medium for microwave irradiation. Microwave activation is able to reduce reaction times from days to hours, making the benzyl alcohol route energy and time efficient for continuous synthesis of large quantities of products (Bilecka et al. 2008). Control of the crystal size through the initial precursor concentration and the irradiation time was demonstrated in the synthesis of various single and mixed oxide nanoparticles, such as ZnO, Fe3O4, CoO, MnO, Mn3O4, NiFe2O4, and BaTiO3 (Bilecka et al. 2011a, b; Kubli et al. 2010). Tert-Butyl Alcohol Route The benzyl alcohol route generally results in the presence of benzyl alcoholate or benzoate residues on the particles surface, which influence the physical (typically optical properties) and chemical properties of the interface (Pucci et al. 2012). The tert-butanol route offers an alternative route to nanoparticles free of strongly

Nonhydrolytic Sol-Gel Technology

11

Fig. 2 HRTEM analyses and simulated three-dimensional shape of TiO2 nanocrystals prepared from TiCl4 and Ti(OiPr)4 in the presence of TOPO and lauric acid: (a) bullet, (b) diamond, (c) short rod, (d) long rod, and (e) branched rod. Scale bar 3 nm (Reproduced with permission from reference Jun et al. (2003). Copyright 2003 American Chemical Society)

chelating surface ligands. The reactivity of tert-butanol with metal precursors can be explained by the strong inductive stabilization of an intermediate carbocation (via SN1 mechanism in Eq. 8).

Alkoxide and Ether Routes The alkoxide route was initially used for the production of TiO2 (Koo et al. 2006; Trentler et al. 1999), ZrO2 (Joo et al. 2003), or HfxZr1-xO2 nanocrystals (Tang et al. 2004) by reaction at high temperature (300  C or more) of metal chloride and alkoxide precursors in a surfactant (trioctyl phosphine oxide, oleic acid, oleylamine). The use of carboxylic acids which bind very strongly to anatase 001 provides a powerful tool for tailoring nanocrystal shape (Fig. 2) (Jun et al. 2003). Playing on the injection rate also allowed to control the phase and shape of TiO2 nanorods (Koo et al. 2006). The ether route was shown to be suitable for the synthesis of amorphous silicabased nanoparticles (SiO2, SiO2-TiO2) as well as crystalline metal oxide nanoparticles (TiO2, SnO2) (Aboulaich et al. 2009, 2010, 2011). The syntheses were performed in dilute CH2Cl2 solutions, using chloride precursors and a stoichiometric amount of iPr2O, at mild temperatures (80–150  C), in the absence of any surfactant or coordinating solvent. The reactions conditions are thus quite different from those previously reported, which involve much higher temperatures and surfactant molecules (alkoxide route) or are performed in the alcohol acting as both an oxygen donor and a coordinating solvent (benzyl alcohol route).

12

A. Vioux and P. Hubert Mutin

Carboxylic Acid Route The reaction of titanium isoproxide in oleic acid at 270  C generated TiO2 nanorods (Joo et al. 2005). The diameter of the nanorods could be controlled by adding 1-hexadecylamine as a cosurfactant. The reaction of acetic acid with titanium n-butoxide at 100  C, without any cosolvent or additive, was successfully used in the synthesis of anatase nanoparticles (Jiang et al. 2008), while a subsequent solvothermal treatment at 200  C in the presence of in-situ produced butyl acetate, followed by calcination at 400  C, resulted in the oriented aggregation of nanocrystals (see below) into spindle-shaped mesocrystals with a single-crystal-like structure (Ye et al. 2011). The obtained nanoporous anatase mesocrystals exhibited remarkable crystalline stability and improved performances as anode materials for lithium ion batteries. Acetophenone Route Ketones and aldehydes have been shown to be able to act as alternative oxygen donors. Typically, highly crystalline anatase nanoparticles of 5–20 nm in size were obtained by reacting titanium tetraisopropoxide with common ketones and aldehydes under solvothermal conditions (Garnweitner et al. 2005). Niederberger, Kessler, and Rivas groups developed an acetophenone route from metal alkoxide or acetylacetonate precursors, which was successfully applied to the synthesis of perovskite nanophosphors and magnetic ferrite spinel nanoparticles (Abtmeyer et al. 2014; Pazik et al. 2009, 2010, 2013; Vazquez-Vazquez et al. 2008, 2011; Zhou et al. 2007). Benzylamine Route Nonhydrolytic sol-gel reactions can be carried out in an organic solvent that do not act as oxygen donors but as coordinating agents, providing control over the size, shape, and surface groups of the nanoparticles (Niederberger et al. 2006a). In those cases, the oxygen is supplied by the molecular precursors themselves. The thermal reaction of metal acetylacetonates in benzylamine belongs to this category, even though benzylamine takes part as a nucleophilic reagent. Actually, a complex mechanism was demonstrated, which involved C–C bond cleavage of the acetylacetonate ligand, followed by ketimine and aldol-like condensation reactions (Pinna et al. 2005). This route was successfully used to synthesize nanocrystalline simple oxides (e.g., aluminum, gallium, indium, iron, zinc oxides) as well as a mixed oxide (ZnGaO4) (Cao et al. 2007; Pinna et al. 2005; Zhou et al. 2007). The simple one-pot solvothermal reaction of titanium isopropoxide in benzylamine led to highly ordered hybrid structures (Garnweitner et al. 2008). These structures consisted of anatase nanoplatelets that were stacked in a lamellar fashion with a small organic layer of benzylamine molecules in between. Thus, benzylamine is involved both in the reaction mechanism leading to the transformation of the titanium isopropoxide into anatase and in the shape control of the crystals by selective capping of the (001) crystal face, which favors their growth in the [001] direction and leads to the formation of nanoplatelets. Then, the benzylamine

Nonhydrolytic Sol-Gel Technology

13

molecules bound to the (001) surfaces interact with each other through pi-pi interactions, driving the stacking of the nanoplatelets into highly ordered lamellar superstructures.

Oriented Attachment and Mesocrystals Oriented attachment mechanism refers to the self-organization of nanoparticles into oriented assemblies (Niederberger and Cölfen 2006). Among these ordered nanoparticle superstructures, mesocrystals, which are assemblies of crystallographically oriented nanocrystals, differ from nanocrystal superlattices, which are periodic arrangements of nanocrystals, irrespective of their mutual orientation (Cölfen and Antonietti 2005). Different examples have been given in the previous sections yet. A further illustration of the influence of organics (solvent and additional coordinating agents) not only on the size and shape of nanoparticles but also on their self-assembling behavior is provided by the synthesis of tungstite nanostructures by benzyl alcohol route (Fig. 3). The reaction of tungsten chloride without any additive yielded tungstite nanoplatelets with a relatively broad size distribution of 30–100 nm. A comprehensive study combining ex situ and in situ techniques (Olliges-Stadler et al. 2013) evidenced a nonclassical crystallization pathway, which involved the formation of polydisperse spherical particles, their arrangement into rod-like assemblies, internal reorganization into stacked platelets, and exfoliation into shorter stacks and individual platelets. Addition of the bioligand deferoxamine mesylate, prone to providing intermolecular amide-amide interactions similar to proteins, led to nanowire bundles. These nanowires were single-crystalline and exhibited a uniform diameter of 1.3 nm. On the other hand, addition of a small amount of 4-tert-butylcatechol led to anisotropic rod-like architectures with diameters between 35 and 40 nm. The rods consisted of a highly ordered stacking of organic–inorganic hybrid nanoplatelets. However, the reaction between tungsten chloride and 4-tert-butylbenzyl alcohol (which differs from benzyl alcohol only in the presence of a bulky group in para position on the aromatic ring) resulted in the formation of ribbon-like structures consisting of laterally assembled columns of stacked nanoplatelets about 1 nm thick and 4 nm width. Another example of nonclassical crystallization mechanism is given by the crystallization pathway of indium-tin-oxide nanoparticles during solvothermal synthesis in benzyl alcohol (Ba et al. 2007). A two-step process is involved. First, a sheet-like superstructures formed, consisting of an organic matrix in which small crystallites (3–6 nm) are aligned but without any crystallographic orientation. It is assumed that the organic matrix stabilizes (both kinetically and thermodynamically) the intermediate nanoparticles. However, when the nanoparticles reach a certain size, they lost their organic protection and undergo a sudden phase transformation into the final bixbyite structure.

14

A. Vioux and P. Hubert Mutin

Fig. 3 Time-dependent TEM and HRTEM images of the tungstite nanostructures after different reaction times of (a) and (b) 10 min, (c) and (d) 20 min, (e) 60 min and (f) 240 min (Reprinted with permission from reference Olliges-Stadler et al. (2013). Copyright 2013 The Royal Society of Chemistry)

Nonhydrolytic Sol-Gel Technology

15

Processing of Metal Oxide Nanoparticles Arising from Nonhydrolytic Routes Metal oxide nanoparticles arising from nonhydrolytic routes are suitable for numerous applications, not only because of the specific organophilic properties (typically as nanofillers in polymer composites) but also owing to the new collective properties arising from their macroscopic self-assembly as films or patterned coatings.

Nanofillers Polymer nanocomposites, which intimately associate organic polymers and inorganic nanofillers, have attracted considerable attention because of their high mechanical performances and their transparency. However, oxide nanoparticles that arise from flame pyrolysis or aqueous syntheses cannot be easily dispersed in organic media, due to their hydrophilic character. These processing issues can be circumvented by using oxide nanoparticles prepared by nonhydrolytic routes. Thus, monodisperse highly crystalline ZrO2 nanoparticles were prepared by benzyl alcohol route from zirconium isopropoxide isopropanol complex as a precursor, then subjected to a simple postsynthesis treatment consisting in stirring in organic solutions of fatty-acid stabilizers at room temperature. It was shown that low amounts of these stabilizers, resulting in zirconia nanoparticles containing less than 25 wt% of organics, led to completely transparent dispersions in organic media, enabling their subsequent transfer into photo-polymerisable organic monomer phases. Transmission holographic gratings based on the prepared nanocomposites demonstrated an outstanding refractive index contrast (Garnweitner et al. 2007). Polymer nanocomposites such as epoxy and polymethacrylate (PMMA) resins were also prepared by using TiO2 nanofillers arising from benzyl alcohol and tertbutanol routes (Koziej et al. 2009; Morselli et al. 2012a, b). Recently, epoxy nanocomposites were prepared from suspensions of magnetite nanocrystals arising from benzyl alcohol route. Above blocking temperature, the magnetite nanoparticles dispersed in the epoxy resin gave rise to an interacting superparamagnetic system (Sciancalepore et al. 2015). It is worth mentioning that the ether route was shown to be suitable for the synthesis of amorphous silica-based nanoparticles (SiO2, SiO2-TiO2) (Aboulaich et al. 2009) as well as crystalline metal oxide nanoparticles (TiO2, SnO2) (Aboulaich et al. 2010, 2011) in the absence of any surfactant or coordinating solvent. Typically, dispersions of amorphous silica-based nanoparticles in CH2Cl2 were stable for months at room temperature in the absence of water and could be dispersed in an hydrophobic organic polymer without any further surface modification treatment (Aboulaich et al. 2009). Evaporation-Induced Self-Assembly Typically, evaporated-induced self-assembly (EISA) process can be implemented in the presence of surfactant-stabilized nanoparticle dispersions. As a matter of fact, ultrasmall (around 3 nm size) and highly soluble anatase nanoparticles were

16

A. Vioux and P. Hubert Mutin

synthesized from TiCl4 using tert-butyl alcohol under microwave heating, with reaction times of less than 1 h at temperatures as low as 50  C (Szeifert et al. 2010). Mesoporous titania coatings were obtained in a one-pot procedure using sols synthesized in the presence of commercial Pluronic surfactants. These coatings could be converted into anatase upon calcination at 450  C, due to a seeding effect of the previously formed crystalline nanoparticles. The high surface-to-bulk ratio of the nanocrystals and the easily accessible mesoporous structures with extremely thin walls led to a drastic acceleration of the electrochemical Li insertion and showed high maximum capacitance in Li-ion batteries. Mesoporous materials with large mesopores of about 20 nm ordered in a cubiclike arrangement were obtained via the block-copolymer-assisted assembly of crystalline tin oxide nanoparticles arising from the reaction of SnCl4 with benzyl alcohol and redispersed in tetrahydrofuran (Ba et al. 2005). In another work, combining diblock copolymer micellar lithography with the benzyl alcohol route enabled the fabrication of patterned arrays made of quasi-hexagonally organized TiO2 nanoparticles or parallel nanowires (Polleux et al. 2011). EISA processes without any surfactant additive were successfully applied to suspensions arising from nonhydrolytic reactions. Typically, the nanoparticles prepared by the ether route, which were terminated by isopropoxide and chloride groups instead of hydroxyl groups, could be concentrated and redispersed in organic solvent; they were found to bind strongly to hydroxylated surfaces, leading to the self-limiting deposition of monolayers (in the absence of water) (Aboulaich et al. 2009). The above mentioned tungsten oxide nanobundles prepared by benzyl alcohol route from tungsten alkoxide could be dispersed in ethanol and deposited onto alumina substrates by drop-coating; a subsequent calcination step at 500  C in air removed the organics from between the nanowires without changing the macroscopic fibrous morphology (Polleux et al. 2006). The resulting highly porous coating demonstrated promising gas-sensing properties.

Microwave-Assisted Deposition One great potential of microwave heating is the possibility to implement both the synthesis of nanoparticles and their deposition on various supports (Bilecka et al. 2011a). Moreover, it is possible to selectively activate the surface of a substrate with suitable microwave-absorbing properties and the deposition of one material on top of another one (Bilecka and Niederberger 2010a). Typically, the reaction of metal acetate or acetylacetonate precursors with benzyl alcohol under microwave irradiation in the presence of immersed flat or curved glass substrates resulted in the deposition of homogeneous metal ferrite films (Bilecka et al. 2011a). The microwave-assisted benzyl alcohol route was also successfully applied to Zn (II) acetate mixed with Al(III) isopropoxide or Sn(IV) tert-butoxide, yielding Al: ZnO and Sn:ZnO nanoparticles with different doping levels (Luo et al. 2013a, b). The nanoparticle dispersions could be subsequently processed into dense transparent conducting films by repeated dip-coating and microwave-assisted densification steps before annealing under air or nitrogen atmosphere.

Nonhydrolytic Sol-Gel Technology

17

Fig. 4 (a) TEM image of an amorphous SiO2-Al2O3-MoO3 mixed oxide, after (Debecker et al. 2009); (b) SEM image of a nanocrystalline mixed TiO2-V2O5 oxide, after (Debecker et al. 2010b)

Nonhydrolytic Synthesis of Mesoporous (Mixed) Oxides and Their Application as Catalytic Systems Mesoporous oxides and mixed oxides are used in a wide range of applications, such as catalysis, photocatalysis, sorption, sensing, or energy storage to cite a few. Using conventional sol-gel process, the simultaneous control of the composition, structure, and texture may be problematic, or require elaborate synthetic procedures such as prehydrolysis, chemical modification, templating, or supercritical drying. Conversely, nonhydrolytic routes (particularly the ether and alkoxide routes) can provide simple and effective methods to prepare mesoporous oxide and mixed oxide xerogels. In most cases, the use of a structure-directing agent or of supercritical drying is not necessary. As in conventional nontemplated sol-gel processes, the porosity of the xerogel (the gel dried by evaporation) results from the removal of the liquid phase (solvent plus byproducts) from the gel, as far as the gel does not completely collapse under the capillary stresses that develop during the evaporation of the liquid phase. In nonhydrolytic processes, the formation of mesoporous xerogels, with sometimes outstanding textures, has been ascribed to several factors. First, the degree of condensation of the gels can be particularly high, leading to tough solid networks, able to withstand the capillary stresses. In addition, the surface tension of the liquid phase is low compared to water, and its interaction with the nonhydroxylated pore surface is weaker. However, in the absence of a templating agent the pore structure is disordered, and the texture depends on the crystallinity of the oxide, going from interconnected pores (sponge-like texture) for amorphous materials to interparticle porosity for nanocrystalline materials (Fig. 4).

18

A. Vioux and P. Hubert Mutin

Nonhydrolytic Routes to Mesoporous (Mixed) Oxides Alkoxide and Ether Route Mesoporous xerogels can be obtained by the strictly aprotic alkoxide and ether routes in the absence of any structure-directing agent and without using supercritical drying. After calcination, the nonordered mesoporous mixed oxides obtained by these routes maintained specific surface areas and pore volumes similar to (or higher than) those reported for solids derived from (ordered) mesoporous xerogels or aerogels prepared by conventional sol-gel (Fig. 3). A wide range of oxides and mixed oxide systems have been investigated, including SiO2, TiO2, Al2O3, WO3, SiO2-ZrO2, SiO2-TiO2, SiO2-Al2O3, TiO2-Al2O3, TiO2-V2O5, SiO2-Al2O3-MoO3, Al2O3-Ag2O-Nb2O5 (Debecker and Mutin 2012; Mutin and Vioux 2009). The textural properties of the solids depend on the crystallinity of the materials and on the reaction parameters. Thus, the porosity of amorphous SiO2-TiO2 xerogels made by these routes was found to depend on the degree of condensation of the gel and on the volume fraction of liquid phase (solvent and iPrCl by-product) in the gel; accordingly, the texture of the xerogels could be tuned simply by changing the reaction time, the reaction temperature, or the volume of solvent (Lafond et al. 2004). Optimization of these parameters led to silica–titania xerogels exhibiting outstanding textures, with specific surface areas as high as 1200 m2 g1 and pore volumes up to 2.4 cm3 g1 after calcination (Cojocariu et al. 2010). Alumina xerogels prepared by the ether or the alkoxide route led to amorphous Al2O3 with very high specific surface area (up to 400 m2 g1) after calcination at 650 C; mesoporous γ-alumina with specific surface area up to 220 m2 g1 was obtained after calcination at 850  C (Acosta et al. 1994). Mesoporous crystalline TiO2-V2O5 catalysts with specific surface areas up to 90 m2 g1 after calcination at 500  C were prepared by the ether route. The narrow mesopore distribution found in these crystalline materials results from the interspace between aggregated well-calibrated nanocrystals, as shown by scanning electron microscopy (SEM) images (Debecker et al. 2010a; Mutin et al. 2006) (Fig. 4b). Interestingly, hybrid SiO2-TiO2-MeSiO1.5 or SiO2-TiO2-Me3SiO0.5 xerogels with outstanding mesoporous textures compared to conventional sol-gel were readily obtained in one step by the ether route, from the reaction of iPr2O with SiCl4 and MeSiCl3 or Me3SiCl (Lorret et al. 2006). The incorporation of organic groups offers further control on the properties of the xerogels (in this case the hydrophobicity, which led to increased catalytic performances). Acetamide Elimination Route Quite recently, Pinkas and coworkers in Brno investigated the synthesis of SiO2TiO2 and SiO2-ZrO2 mixed oxides by a nonhydrolytic route involving acetamide elimination from silicon acetate, Si(OAc)4, and titanium or zirconium diethylamide, M(NEt2)4, at 80  C (Skoda et al. 2015a, b). In these cases, the addition of Pluronic P123 as a structure-directing agent was needed to obtain after calcination homogeneous mixed oxides with a “wormhole” mesopore structure and high specific surface area. Interestingly, when an excess of silicon acetate was used to reach Si:M ratios

Nonhydrolytic Sol-Gel Technology

19

higher than 1, the formation of acetic anydride was detected, suggesting that homocondensation of excess Si–OAc groups took place, probably catalyzed by metal Lewis acidic species.

Ester Elimination Route Pinkas and coworkers recently described the nonhydrolytic synthesis of silicon orthophosphate xerogels by elimination of trimethylsilylacetate, AcOSiMe3, from silicon acetate, Si(OAc)4, and the tris(trimethylsilyl)phosphate, PO(OSiMe3)3 (Styskalik et al. 2014, 2015b). The nonhydrolytic polycondensation led to the formation of an inorganic network with a high content of Si–O–P bonds and hexacoordinated SiO6 moieties. In the absence of structure-directing agent, the silicophosphate xerogels exhibited significant specific surface area but were largely microporous (Styskalik et al. 2014). When a Pluronic P123 template was added (Styskalik et al. 2015b), mesoporous silicophosphate materials with specific surface area up to 128 m2 g1 and pore diameters around 20 nm were obtained after calcination. This method has been extended to the synthesis of hybrid “silicophosphonate,” starting from organotriacetoxysilanes, RSi(OAc)3, with trimethylsilyl esters of phosphonic acid, RPO(OSiMe3)2 (Styskalik et al. 2015a). R2Si(OAc)2 precursors were not incorporated in the network but acted as templates. Bridged acetoxysilanes, (AcO)3Si-X-Si(OAc)3, and bridged trimethylsilylphosphonates, (SiMe3O)2PO-XPO(OSiMe3)2, have also been used, leading to hybrids with large specific surface area (up to 700 m2 g1) and pore volumes (up to 1.6 cm3 g1). Interestingly, the microporosity of these bridged hybrid materials is correlated to the presence of hexacoordinated SiO6 structural units. Carboxylic Acid Route The reaction of silicon alkoxides with formic acid, originally reported by Sharp (1994), leads to low porosity silicas. However, mesoporous silicas with a surface area of 720 m2 g1 and pore volume of 1.4 cm3 g1 could be obtained via the formic acid route by using an ionic liquid as solvent and long aging times. In this case, the ionic liquid was then washed off with a polar solvent (Dai et al. 2000). This route was also applied to the synthesis of crack-free ionogel monoliths, consisting of an ionic liquid phase confined in a mesoporous silica (Neouze et al. 2006; Viau et al. 2012). Ionogels are well suited for the in situ fabrication of temperature-resistant electrolytes for energy storage devices, including thin film electrochemical double layer capacitors (Horowitz and Panzer 2012).

Application to Heterogeneous Catalysis The mesoporous oxides and mixed oxides prepared by nonhydrolytic routes have been successfully applied to the design of heterogeneous catalysts and photocatalysts, as recently reviewed (Debecker et al. 2013; Debecker and Mutin 2012).

20

A. Vioux and P. Hubert Mutin

In particular, the ether route provides a simple, one-step method to prepare mesoporous mixed oxide catalysts with well-controlled compositions and textures, starting from cheap chloride precursors and avoiding the use of reactivity modifier, templating agent, or supercritical drying step (Debecker et al. 2013). When solubilization of a chloride precursor is problematic, the alkoxide route may be more indicated (Helena Kaper et al. 2012). The nonordered texture of the catalysts prepared by these routes is well-suited to catalytic applications. The large, interconnected mesopores featured by these materials are more favorable than ordered mesopores. These routes have been used for the synthesis of a wide variety of nonordered mesoporous catalysts (Debecker et al. 2013), such as SiO2-TiO2 (Cojocariu et al. 2008), SiO2-ZrO2 (Helena Kaper et al. 2012), SiO2-WO3 (Maksasithorn et al. 2014), TiO2-V2O5 (Mutin et al. 2006), Ag-Nb2O5-Al2O3 (Petitto et al. 2013), SiO2-Al2O3-MoO3 (Debecker et al. 2009), and SiO2-Al2O3Re2O7 (Bouchmella et al. 2013), which were tested in various reactions, including mild and total oxidation, alkylation, selective catalytic reduction of NOx by NH3 or decane, and olefin metathesis. In the case of mixed oxides, the catalytic properties strongly depend on the degree of homogeneity. Nonhydrolytic methods often lead to highly homogeneous mixed oxide xerogels, owing to their easily controllable kinetics and, in the case of metal silicates, to the levelling of reactivities at silicon and metal centers. When the oxide components feature high Tammann temperatures, as for instance in the case of SiO2TiO2, SiO2-ZrO2, or SiO2-Al2O3, homogeneity is maintained after calcination (typically at 500  C). For instance, SiO2-TiO2 mixed oxides obtained by the ether route show excellent performances in the mild oxidation of organic compounds, ascribed to the formation of well-dispersed Ti species linked to the SiO2 matrix by Si–O–Ti bonds and to their very high specific surface area (Cojocariu et al. 2010; Lafond et al. 2002). On the other hand, when the active oxide phase has a low Tammann temperature (cases of VOx, MoOx, ReOx, WOx for instance), appropriate thermal treatments can provoke the migration of active oxide species toward the surface, leading to an increase of their surface concentration (Debecker et al. 2013). For instance, highly active MoO3-SiO2-Al2O3 and Re2O7-SiO2-Al2O3 olefin metathesis catalysts have been obtained by the ether route (Bouchmella et al. 2013; Debecker et al. 2009). Migration of Mo or Re oxide species toward the surface occurred during the calcination at 500 C, as evidenced by XPS and ToF-SIMS, leading to high concentration of well-dispersed Mo or Re surface species. This, together with the mesoporous texture and acidic character of the catalysts, accounted for their excellent catalytic performances in the metathesis of ethene and butene to propene (Bouchmella et al. 2013; Debecker et al. 2009). Recently, mesoporous SiO2-ZrO2 obtained by the templated acetamide elimination route were found to display good activity and selectivity in the Meerwein–Ponndorf–Verley reduction of 4-tert-butylcyclohexanone and for aminolysis of styrene oxide with aniline (Skoda et al. 2015b). Mesoporous SiO2TiO2 xerogels prepared by the same route were tested in cyclohexene epoxidation with cumyl hydroperoxide in toluene. They displayed a good catalytic activity in cyclohexene epoxidation with cumyl hydroperoxide (Skoda et al. 2015a).

Nonhydrolytic Sol-Gel Technology

21

Conclusion Nonhydrolytic sol-gel is now well established as a powerful methodology for the synthesis of oxide-based materials, notably mixed-oxide xerogels with wellcontrolled composition and mesoporosity, highly crystalline oxide nanoparticles with control over size and shape, as well as superstructures and films resulting from their assembly. One major asset of this low-temperature process is to offer simple synthesis protocols avoiding the use of chemical additives or templates. Actually, the control over composition, texture, structure, and morphology of the final materials arises from the intrinsic nature of the organic reactions involved. Typically, in the synthesis of mixed oxides, nonhydrolytic condensation reactions are slower than hydrolytic ones, while leading to higher condensation degrees and hydroxyl-free surfaces. As the reaction rates depend more on the nature of the carbon center than on the nature of the metal center, the reactivity of the different metal precursors is leveled. As for the reactivity of silicon precursors, which is much lower than that of metal precursors, the reactions around the silicon atom are catalyzed by Lewis acidic metal species, which again levels the kinetics when metal-silicates are prepared. In the synthesis of oxide nanoparticles, the presence in the nonaqueous reaction mixture of organic derivatives endowed with complexing ability strongly influences the size, shape, and structure of crystallites and also their assembly into superstructures. Even though the various organic reactions involved in nonhydrolytic processes may make it difficult to predict the characteristics of the final material, in practice, this complexity offers wide possibilities to tailor the reaction system, by varying the precursor-oxygen donor combination. Recently, the use of microwave chemistry has even considerably enlarged this potential. As a matter of fact, over the last decade, nonhydrolytic reactions have been extended to other processes (solvothermal synthesis, chemical solution deposition, atomic layer deposition), solvents (ionic liquids), O-donors (including cellulose), activation modes (microwaves), and materials (metal chalcogenides or phosphates, carbon nanocomposites) (Bilecka and Niederberger 2010b; Mutin and Vioux 2013). Such a development in innovative chemical methods should be continued, owing to the increasing demand for advanced nanostructured materials, for such applications as energy storage, catalysis, sensing, and optics.

References Aboulaich A, Lorret O, Boury B, Mutin PH. Surfactant-free organo-soluble silica-titania and silica nanoparticles. Chem Mater. 2009;21:2577–9. Aboulaich A, Boury B, Mutin PH. Reactive and organosoluble anatase nanoparticles by a surfactant-free nonhydrolytic synthesis. Chem Mater. 2010;22:4519–21. Aboulaich A, Boury B, Mutin PH. Reactive and organosoluble SnO2 nanoparticles by a surfactantfree non-hydrolytic sol-gel route. Eur J Inorg Chem. 2011;2011:3644–9. Abtmeyer S, Pazik R, Wiglusz RJ, Malecka M, Seisenbaeva GA, Kessler VG. Lanthanum molybdate nanoparticles from the Bradley reaction: factors influencing their composition, structure,

22

A. Vioux and P. Hubert Mutin

and functional characteristics as potential matrixes for luminescent phosphors. Inorg Chem. 2014;53:943–51. Acosta S, Corriu RJP, Leclercq D, Lefevre P, Mutin PH, Vioux A. Preparation of alumina gels by a non-hydrolytic sol gel processing method. J Non-Cryst Solids. 1994;170:234–42. Andrianainarivelo M, Corriu R, Leclercq D, Mutin PH, Vioux A. Mixed oxides SiO2-ZrO2 and SiO2-TiO2 By a non-hydrolytic sol-gel route. J Mater Chem. 1996;6:1665–71. Antonietti M, Niederberger M, Smarsly B. Self-assembly in inorganic and hybrid systems: beyond the molecular scale. Dalton Trans. 2008:18–24. Arnal P, Corriu RJP, Leclercq D, Mutin PH, Vioux A. Preparation of anatase, brookite and rutile at low temperature by non-hydrolytic sol-gel methods. J Mater Chem. 1996;6:1925–32. Avci N, Smet PF, Poelman H, Velde N, Buysser K, Driessche I, Poelman D. Characterization of TiO2 powders and thin films prepared by non-aqueous sol–gel techniques. J Sol-Gel Sci Technol. 2009;52:424–31. Ba J, Polleux J, Antonietti M, Niederberger M. Non-aqueous synthesis of tin oxide nanocrystals and their assembly into ordered porous mesostructures. Adv Mater. 2005;17:2509–12. Ba J, Feldhoff A, Rohlfing DF, Wark M, Antonietti M, Niederberger M. Crystallization of indium tin oxide nanoparticles. From cooperative behavior to individuality. Small. 2007;3:310–7. Bilecka I, Niederberger M. Microwave chemistry for inorganic nanomaterials synthesis. Nanoscale. 2010a;2:1358–74. Bilecka I, Niederberger M. New developments in the nonaqueous and/or non-hydrolytic sol-gel synthesis of inorganic nanoparticles. Electrochim Acta. 2010b;55:7717–25. Bilecka I, Djerdj I, Niederberger M. One-minute synthesis of crystalline binary and ternary metal oxide nanoparticles. Chem Commun. 2008:886–8. Bilecka I, Elser P, Niederberger M. Kinetic and thermodynamic aspects in the microwave-assisted synthesis of ZnO nanoparticles in benzyl alcohol. ACS Nano. 2009;3:467–77. Bilecka I, Kubli M, Amstad E, Niederberger M. Simultaneous formation of ferrite nanocrystals and deposition of thin films via a microwave-assisted nonaqueous sol-gel process. J Sol-Gel Sci Technol. 2011a;57:313–22. Bilecka I, Luo L, Djerdj I, Rossell MD, Jagodic M, Jaglicic Z, Masubuchi Y, Kikkawa S, Niederberger M. Microwave-assisted nonaqueous sol-gel chemistry for highly concentrated ZnO-based magnetic semiconductor nanocrystals. J Phys Chem C. 2011b;115:1484–95. Bouchmella K, Mutin PH, Stoyanova M, Poleunis C, Eloy P, Rodemerck U, Gaigneaux EM, Debecker DP. Olefin metathesis with mesoporous rhenium-silicium-aluminum mixed oxides obtained via a one-step non-hydrolytic sol-gel route. J Catal. 2013;301:233–41. Bourget L, Corriu RJP, Leclercq D, Mutin PH, Vioux A. Non-hydrolytic sol-gel routes to silica. J Non-Cryst Solids. 1998a;242:81–91. Bourget L, Mutin PH, Vioux A, Frances JM. Nonhydrolytic synthesis and structural study of methoxyl-terminated polysiloxane D/Q resins. J Polym Sci A. 1998b;36:2415–25. Bourget L, Leclercq D, Vioux A. Catalyzed nonhydrolytic sol-gel route to organosilsesquioxane gels. J Sol-Gel Sci Technol. 1999;14:137–47. Cao M, Djerdj I, Antonietti M, Niederberger M. Nonaqueous synthesis of colloidal ZnGa2O4 nanocrystals and their photoluminescence properties. Chem Mater. 2007;19:5830–2. Caruso J, Hampden-Smith MJ. Ester elimination: a general solvent dependent non-hydrolytic route to metal and mixed-metal oxides. J Sol-Gel Sci Technol. 1997;8:35–9. Clavel G, Marichy C, Pinna N. Sol-gel chemistry and atomic layer deposition. In: Pinna N, Knez M, editors. Atomic layer deposition of nanostructured materials. Weinheim: Wiley-VCH; 2012. p. 61–82. Cojocariu AM, Mutin PH, Dumitriu E, Fajula F, Vioux A, Hulea V. Non-hydrolytic synthesis of mesoporous silica-titania catalysts for the mild oxidation of sulfur compounds with hydrogen peroxide. Chem Commun. 2008: 5357–9. Cojocariu AM, Mutin PH, Dumitriu E, Fajula F, Vioux A, Hulea V. Mild oxidation of bulky organic compounds with hydrogen peroxide over mesoporous TiO2-SiO2 xerogels prepared by non-hydrolytic sol-gel. Appl Catal B. 2010;97:407–13.

Nonhydrolytic Sol-Gel Technology

23

Cölfen H, Antonietti M. Mesocrystals: inorganic superstructures made by highly parallel crystallization and controlled alignment. Angew Chem Int Ed. 2005;44:5576–91. Corriu R, Leclercq D, Lefevre P, Mutin PH, Vioux A. Preparation of monolithic binary oxide gels by a nonhydrolytic sol-gel process. Chem Mater. 1992a;4:961–3. Corriu RJP, Leclercq D, Lefèvre P, Mutin PH, Vioux A. Preparation of monolithic gels from silicon halides by a non-hydrolytic sol-gel process. J Non-Cryst Solids. 1992b;146:301–3. Corriu RJP, Leclercq D, Mutin PH, Sarlin L, Vioux A. Nonhydrolyticc sol-gel routes to layered metal(IV) and silicon phosphonates. J Mater Chem. 1998;8:1827–33. Dai S, Ju YH, Gao HJ, Lin JS, Pennycook SJ, Barnes CE. Preparation of silica aerogel using ionic liquids as solvents. Chem Commun. 2000;3:243–4. Dar MI, Sampath S, Shivashankar SA. Microwave-assisted, surfactant-free synthesis of air-stable copper nanostructures and their SERS study. J Mater Chem. 2012;22:22418–23. Debecker DP, Mutin PH. Non-hydrolytic sol-gel routes to heterogeneous catalysts. Chem Soc Rev. 2012;41:3624–50. Debecker DP, Bouchmella K, Poleunis C, Eloy P, Bertrand P, Gaigneaux EM, Mutin PH. Design of SiO2-Al2O3-MoO3 metathesis catalysts by nonhydrolytic sol-gel. Chem Mater. 2009;21:2817–24. Debecker DP, Bouchmella K, Delaigle R, Eloy P, Poleunis C, Bertrand P, Gaigneaux EM, Mutin PH. One-step non-hydrolytic sol-gel preparation of efficient V2O5-TiO2 catalysts for VOC total oxidation. Appl Catal B. 2010a;94:38–45. Debecker DP, Delaigle R, Bouchmella K, Eloy P, Gaigneaux EM, Mutin PH. Total oxidation of benzene and chlorobenzene with MoO3- and WO3-promoted V2O5/TiO2 catalysts prepared by a nonhydrolytic sol-gel route. Catal Today. 2010b;157:125–30. Debecker DP, Hulea V, Mutin PH. Mesoporous mixed oxide catalysts via non-hydrolytic sol-gel: a review. Appl Catal A. 2013;451:192–206. Djerdj I, Garnweitner G, Su DS, Niederberger M. Morphology-controlled nonaqueous synthesis of anisotropic lanthanum hydroxide nanoparticles. J Solid State Chem. 2007;180:2154–65. Ferreira RAS, Karmaoui M, Nobre SS, Carlos LD, Pinna N. Optical properties of lanthanide-doped lamellar nanohybrids. ChemPhysChem. 2006;7:2215–22. Garnweitner G, Niederberger M. Nonaqueous and surfactant-free synthesis routes to metal oxide nanoparticles. J Am Ceram Soc. 2006;89:1801–8. Garnweitner G, Niederberger M. Organic chemistry in inorganic nanomaterials synthesis. J Mater Chem. 2008;18:1171–82. Garnweitner G, Antonietti M, Niederberger M. Nonaqueous synthesis of crystalline anatase nanoparticles in simple ketones and aldehydes as oxygen-supplying agents. Chem Commun. 2005:397–9. Garnweitner G, Goldenberg LM, Sakhno OV, Antonietti M, Niederberger M, Stumpe J. Large-scale synthesis of organophilic zirconia nanoparticles and their application in organic-inorganic nanocomposites for efficient volume holography. Small. 2007;3:1626–32. Garnweitner G, Tsedev N, Dierke H, Niederberger M. Benzylamines as versatile agents for the one-pot synthesis and highly ordered stacking of anatase nanoplatelets. Eur J Inorg Chem. 2008;2008:890–5. Horowitz AI, Panzer MJ. High-performance, mechanically compliant silica-based ionogels for electrical energy storage applications. J Mater Chem. 2012;22:16534–9. Iwasaki M, Yasumori A, Shibata S, Yamane M. Preparation of high homogeneity BaO-TiO2-SiO2 gel. J Sol-Gel Sci Technol. 1994;2:387–91. Jansen M, Guenther E. Oxide gels and ceramics prepared by a nonhydrolytic sol-gel process. Chem Mater. 1995;7:2110–4. Jansen M, Guenther E. Water- and hydroxyl group-free gels and xerogels based on a network of oxygen-bridged metal and/or semimetal atoms, and their manufacture and use. Eur Pat Appl. (Cerdec Aktiengesellschaft Keramische Farben, Germany). 1996. p. 11. Jia F, Zhang L, Shang X, Yang Y. Non-aqueous sol-gel approach towards the controllable synthesis of nickel nanospheres, nanowires, and nanoflowers. Adv Mater. 2008;20:1050–4.

24

A. Vioux and P. Hubert Mutin

Jiang D, Xu Y, Hou B, Wu D, Sun Y. A simple non-aqueous route to anatase TiO2. Eur J Inorg Chem. 2008;1236–40. Joo J, Yu T, Kim YW, Park HM, Wu F, Zhang JZ, Hyeon T. Multigram scale synthesis and characterization of monodisperse tetragonal zirconia nanocrystals. J Am Chem Soc. 2003;125:6553–7. Joo J, Kwon SG, Yu T, Cho M, Lee J, Yoon J, Hyeon T. Large-scale synthesis of TiO2 nanorods via non-hydrolytic sol-gel ester elimination reaction and their application to photocatalytic inactivation of E. coli. J Phys Chem B. 2005;109:15297–302. Jun YW, Casula MF, Sim J-H, Kim SY, Cheon J, Alivisatos AP. Surfactant-assisted elimination of a high energy facet as a means of controlling the shapes of TiO2 nanocrystals. J Am Chem Soc. 2003;125:15981–5. Kaper H, Karim B, Hubert Mutin P, Goettmann F. High surface area SiO2-ZrO2 mixed oxides as catalysts for Friedel-Crafts-type alkylation of arenes with alcohols and tandem cyclopropanation. ChemCatChem. 2012;4:1813–8. Karmaoui M, Sa Ferreira RA, Mane AT, Carlos LD, Pinna N. Lanthanide-based lamellar nanohybrids: synthesis, structural characterization, and optical properties. Chem Mater. 2006;18:4493–9. Koo B, Park J, Kim Y, Choi S-H, Sung Y-E, Hyeon T. Simultaneous phase- and size-controlled synthesis of TiO2 nanorods via non-hydrolytic sol-gel reaction of syringe pump delivered precursors. J Phys Chem B. 2006;110:24318–23. Koziej D, Fischer F, Kranzlin N, Caseri WR, Niederberger M. Nonaqueous TiO2 nanoparticle synthesis: a versatile basis for the fabrication of self-supporting, transparent, and UV-absorbing composite films. ACS Appl Mater Interfaces. 2009;1:1097–104. Kubli M, Luo L, Bilecka I, Niederberger M. Microwave-assisted nonaqueous sol-gel deposition of different spinel ferrites and barium titanate perovskite thin films. Chimia. 2010;64:170–2. Lafond V, Mutin PH, Vioux A. Non-hydrolytic sol-gel routes based on alkyl halide elimination: toward better mixed oxide catalysts and new supports application to the preparation of a SiO2TiO2 epoxidation catalyst. J Mol Catal A. 2002;182–183:81–8. Lafond V, Mutin PH, Vioux A. Control of the texture of titania-silica mixed oxides prepared by nonhydrolytic sol-gel. Chem Mater. 2004;16:5380–6. Lorret O, Lafond V, Mutin PH, Vioux A. One-step synthesis of mesoporous hybrid titania-silica xerogels for the epoxidation of alkenes. Chem Mater. 2006;18:4707–9. Ludi B, Olliges-Stadler I, Rossell MD, Niederberger M. Extension of the benzyl alcohol route to metal sulfides: “nonhydrolytic” thio sol-gel synthesis of ZnS and SnS2. Chem Commun. 2011;47:5280–2. Luo L, Haefliger K, Xie D, Niederberger M. Transparent conducting Sn:ZnO films deposited from nanoparticles. J Sol-Gel Sci Technol. 2013a;65:28–35. Luo L, Rossell MD, Xie D, Erni R, Niederberger M. Microwave-assisted nonaqueous sol-gel synthesis: from Al:ZnO nanoparticles to transparent conducting films. ACS Sustain Chem Eng. 2013b;1:152–60. Maksasithorn S, Praserthdam P, Suriye K, Devillers M, Debecker DP. WO3-based catalysts prepared by non-hydrolytic sol-gel for the production of propene by cross-metathesis of ethene and 2-butene. Appl Catal A. 2014;488:200–7. Mizuno M, Takahashi M, Tokuda Y, Yoko T. Organic-inorganic hybrid material of phenyl-modified polysilicophosphate prepared through nonaqueous acid-base reaction. Chem Mater. 2006;18:2075–80. Morselli D, Bondioli F, Fiorini M, Messori M. Poly(methyl methacrylate)-TiO2 nanocomposites obtained by non-hydrolytic sol-gel synthesis: the innovative tert-butyl alcohol route. J Mater Sci. 2012a;47:7003–12. Morselli D, Bondioli F, Sangermano M, Messori M. Photo-cured epoxy networks reinforced with TiO2 in-situ generated by means of non-hydrolytic sol-gel process. Polymer. 2012b;53:283–90. Mutin PH, Vioux A. Nonhydrolytic processing of oxide-based materials: simple routes to control homogeneity, morphology, and nanostructure. Chem Mater. 2009;21:582–96.

Nonhydrolytic Sol-Gel Technology

25

Mutin PH, Vioux A. Recent advances in the synthesis of inorganic materials via non-hydrolytic condensation and related low-temperature routes. J Mater Chem A. 2013;1:11504–12. Mutin PH, Popa AF, Vioux A, Delahay G, Coq B. Nonhydrolytic vanadia-titania xerogels: synthesis, characterization, and behavior in the selective catalytic reduction of NO by NH3. Appl Catal B Environ. 2006;69:49–57. Neouze M-A, Le Bideau J, Gaveau P, Bellayer S, Vioux A. Ionogels, new materials arising from the confinement of ionic liquids within silica-derived networks. Chem Mater. 2006;18:3931–6. Niederberger M. Nonaqueous sol-gel routes to metal oxide nanoparticles. Acc Chem Res. 2007;40:793–800. Niederberger M, Cölfen H. Oriented attachment and mesocrystals: non-classical crystallization mechanisms based on nanoparticle assembly. Phys Chem Chem Phys. 2006;8:3271–87. Niederberger M, Garnweitner G. Organic reaction pathways in the nonaqueous synthesis of metal oxide nanoparticles. Chem Eur J. 2006;12:7282–302. Niederberger M, Bartl MH, Stucky GD. Benzyl alcohol and titanium tetrachloride-A versatile reaction system for the nonaqueous and low-temperature preparation of crystalline and luminescent titania nanoparticles. Chem Mater. 2002a;14:4364–70. Niederberger M, Bartl MH, Stucky GD. Benzyl alcohol and transition metal chlorides as a versatile reaction system for the nonaqueous and low-temperature synthesis of crystalline nano-objects with controlled dimensionality. J Am Chem Soc. 2002b;124:13642–3. Niederberger M, Garnweitner G, Buha J, Polleux J, Ba J, Pinna N. Nonaqueous synthesis of metal oxide nanoparticles: review and indium oxide as case study for the dependence of particle morphology on precursors and solvents. J Sol-Gel Sci Technol. 2006a;40:259–66. Niederberger M, Garnweitner G, Pinna N, Neri G. Non-aqueous routes to crystalline metal oxide nanoparticles: formation mechanisms and applications. Prog Solid State Chem. 2006b;33:59–70. Niederberger M, Garnweitner G, Ba J, Polleux J, Pinna N. Nonaqueous synthesis, assembly and formation mechanisms of metal oxide nanocrystals. Int J Nanotechnol. 2007;4:263–81. Olliges-Stadler I, Rossell MD, Sueess MJ, Ludi B, Bunk O, Pedersen JS, Birkedal H, Niederberger M. A comprehensive study of the crystallization mechanism involved in the nonaqueous formation of tungstite. Nanoscale. 2013;5:8517–25. Pazik R, Tekoriute R, Hakansson S, Wiglusz R, Strek W, Seisenbaeva GA, Gun’ko YK, Kessler VG. Precursor and solvent effects in the nonhydrolytic synthesis of complex oxide nanoparticles for bioimaging applications by the ether elimination (Bradley) reaction. Chem Eur J. 2009;15:6820–6 .S6820/6821-S6820/6811 Pazik R, Seisenbaeva GA, Gohil S, Wiglusz R, Kepinski L, Strek W, Kessler VG. Simple and efficient synthesis of a Nd:LaAlO3 NIR nanophosphor from rare earth alkoxo-monoaluminates Ln2Al2(OiPr)12(iPrOH)2 single source precursors by Bradley reaction. Inorg Chem. 2010;49:2684–91. Pazik R, Piasecka E, Malecka M, Kessler VG, Idzikowski B, Sniadecki Z, Wiglusz RJ. Facile non-hydrolytic synthesis of highly water dispersible, surfactant free nanoparticles of synthetic MFe2O4 (M-Mn2+, Fe2+, Co2+, Ni2+) ferrite spinel by a modified Bradley reaction. RSC Adv. 2013;3:12230–43. Pereira PFS, Matos MG, Ferreira CMA, De Faria EH, Calefi PS, Rocha LA, Ciuffi KJ, Nassar EJ. Aluminate matrix doped with Tm3+/Tb3+/Eu3+ obtained by non-hydrolytic sol-gel route: white light emission. J Lumin. 2014;146:394–7. Petitto C, Mutin HP, Gr D. Hydrothermal activation of silver supported alumina catalysts prepared by sol-gel method: application to the selective catalytic reduction (SCR) of NOx by n-decane. Appl Catal B Environ. 2013;134–135:258–64. Pinna N, Niederberger M. Surfactant-free nonaqueous synthesis of metal oxide nanostructures. Angew Chem Int Ed. 2008;47:5292–304. Pinna N, Garnweitner G, Antonietti M, Niederberger M. Non-aqueous synthesis of high-purity metal oxide nanopowders using an ether elimination process. Adv Mater. 2004;16:2196–200.

26

A. Vioux and P. Hubert Mutin

Pinna N, Garnweitner G, Antonietti M, Niederberger M. A general nonaqueous route to binary metal oxide nanocrystals involving a C–C bond cleavage. J Am Chem Soc. 2005;127:5608–12. Polleux J, Pinna N, Antonietti M, Niederberger M. Growth and assembly of crystalline tungsten oxide nanostructures assisted by bioligation. J Am Chem Soc. 2005;127:15595–601. Polleux J, Gurlo A, Barsan N, Weimar U, Antonietti M, Niederberger M. Template-free synthesis and assembly of single-crystalline tungsten oxide nanowires and their gas-sensing properties. Angew Chem Int Ed. 2006;45:261–5. Polleux J, Rasp M, Louban I, Plath N, Feldhoff A, Spatz JP. Benzyl alcohol and block copolymer micellar lithography: a versatile route to assembling gold and in situ generated titania nanoparticles into uniform binary nanoarrays. ACS Nano. 2011;5:6355–64. Pucci A, Willinger M-G, Liu F, Zeng X, Rebuttini V, Clavel G, Bai X, Ungar G, Pinna N. One-step synthesis and self-assembly of metal oxide nanoparticles into 3D superlattices. ACS Nano. 2012;6:4382–91. Sciancalepore C, Bondioli F, Messori M, Barrera G, Tiberto P, Allia P. Epoxy nanocomposites functionalized with in situ generated magnetite nanocrystals: microstructure, magnetic properties, interaction among magnetic particles. Polymer. 2015;59:278–89. Sharp KG. A two-component, non-aqueous route to silica gel. J Sol-Gel Sci Technol. 1994;2:35–41. Skoda D, Styskalik A, Moravec Z, Bezdicka P, Barnes CE, Pinkas J. Mesoporous titanosilicates by templated non-hydrolytic sol-gel reactions. J Sol-Gel Sci Technol. 2015a;74:810–22. Skoda D, Styskalik A, Moravec Z, Bezdicka P, Pinkas J. Templated non-hydrolytic synthesis of mesoporous zirconium silicates and their catalytic properties. J Mater Sci. 2015b;50:3371–82. Steunou N, Ribot F, Boubekeur K, Maquet J, Sanchez C. Ketones as an oxolation source for the synthesis of titanium-oxo-organoclusters. New J Chem. 1999;23:1079–86. Styskalik A, Skoda D, Pinkas J, Mathur S. Non-hydrolytic synthesis of titanosilicate xerogels by acetamide elimination and their use as epoxidation catalysts. J Sol-Gel Sci Technol. 2012;63:463–72. Styskalik A, Skoda D, Moravec Z, Abbott JG, Barnes CE, Pinkas J. Synthesis of homogeneous silicophosphate xerogels by non-hydrolytic condensation reactions. Microporous Mesoporous Mater. 2014;197:204–12. Styskalik A, Skoda D, Moravec Z, Babiak M, Barnes CE, Pinkas J. Control of micro/mesoporosity in non-hydrolytic hybrid silicophosphate xerogels. J Mater Chem A. 2015a;3:7477–87. Styskalik A, Skoda D, Moravec Z, Roupcova P, Barnes CE, Pinkas J. Non-aqueous templateassisted synthesis of mesoporous nanocrystalline silicon orthophosphate. RSC Adv. 2015b;5:73670–6. Szeifert JM, Feckl JM, Fattakhova-Rohlfing D, Liu Y, Kalousek V, Rathousky J, Bein T. Ultrasmall titania nanocrystals and their direct assembly into mesoporous structures showing fast lithium insertion. J Am Chem Soc. 2010;132:12605–11. Tang J, Fabbri J, Robinson RD, Zhu Y, Herman IP, Steigerwald ML, Brus LE. Solid-solution nanoparticles: use of a nonhydrolytic sol-gel synthesis to prepare HfO2 and HfxZr1xO2 nanocrystals. Chem Mater. 2004;16:1336–42. Trentler TJ, Denler TE, Bertone JF, Agrawal A, Colvin VL. Synthesis of TiO2 nanocrystals by nonhydrolytic solution-based reactions. J Am Chem Soc. 1999;121:1613–4. Vazquez-Vazquez C, Lovelle M, Mateo C, Lopez-Quintela MA, Bujan-Nunez MC, Serantes D, Baldomir D, Rivas J. Magnetocaloric effect and size-dependent study of the magnetic properties of cobalt ferrite nanoparticles prepared by solvothermal synthesis. Phys Status Solidi A. 2008;205:1358–62. Vazquez-Vazquez C, Lopez-Quintela MA, Bujan-Nunez MC, Rivas J. Finite size and surface effects on the magnetic properties of cobalt ferrite nanoparticles. J Nanopart Res. 2011;13:1663–76. Viau L, Neouze M-A, Biolley C, Volland S, Brevet D, Gaveau P, Dieudonne P, Galarneau A, Vioux A. Ionic liquid mediated sol-gel synthesis in the presence of water or formic acid: which synthesis for which material? Chem Mater. 2012;24:3128–34. Vioux A. Nonhydrolytic sol-gel routes to oxides. Chem Mater. 1997;9:2292–9.

Nonhydrolytic Sol-Gel Technology

27

Wohlrab S, Pinna N, Antonietti M, Cölfen H. Polymer-induced alignment of dl-alanine nanocrystals to crystalline mesostructures. Chem Eur J. 2005;11:2903–13. Ye J, Liu W, Cai J, Chen S, Zhao X, Zhou H, Qi L. Nanoporous anatase TiO2 mesocrystals: additive-free synthesis, remarkable crystalline-phase stability, and improved lithium insertion behavior. J Am Chem Soc. 2011;133:933–40. Zhang L, Garnweitner G, Djerdj I, Antonietti M, Niederberger M. Generalized nonaqueous sol-gel synthesis of different transition-metal niobate nanocrystals and analysis of the growth mechanism. Chem Asian J. 2008;3:746–52. Zhou S, Antonietti M, Niederberger M. Low-temperature synthesis of gamma-alumina nanocrystals from aluminum acetylacetonate in nonaqueous media. Small. 2007;3:763–7.

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel Route Francesco d’Acapito

Contents Introduction to X-ray Absorption Spectroscopy (XAS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Collection and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data Analysis Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of XAS to Sol-Gel Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials for Optics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials for Catalysts and Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrodes for Batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2 7 9 11 11 15 17 20 22 22

Abstract

This chapter reports on the use of X-ray Absorption Spectroscopy in the characterization of materials obtained by the Sol-Gel method. Firstly, an introduction to the theoretical bases of the technique is given followed by a brief description of the most relevant data collection schemes and data analysis. Successively, a collection of recent relevant experiments is presented putting in evidence, case by case, the peculiar information retrieved thanks to this technique. These studies are classed following the kind of material or its use: materials for optics,

F. d’Acapito (*) CNR-IOM-OGG c/o European Synchrotron Radiation Facility, LISA CRG, Grenoble, France e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_31-1

1

2

F. d’Acapito

nanoparticles, materials for catalysis and sensors, and materials for batteries. Eventually, a brief overview is given on the future perspectives of the technique in terms of new data collection methods and data analysis. A rich list of bibliographic references completes this contribution.

Introduction to X-ray Absorption Spectroscopy (XAS) Theory

Absorption Coefficient µ (Arb. Units)

X-ray Absorption spectroscopy (XAS) (Lee et al. 1981) is an experimental technique that permits the quantitative determination of structural parameters around chosen atomic species and to obtain a description of the electronic structure. In particular, it derives this information from the small oscillations that appear in the absorption coefficient as a function of energy just above the edge step from a deep core state (1s or 2s, 2p). These oscillations appear only when the absorbing atom is regularly surrounded by other atoms as shown in Fig. 1. XAS permits the determination of the number of neighbors N and distance R from the atom that absorbs the X-ray photon (hereafter defined as the absorber) with an accuracy of about 10 % for N and 1 % for R. Moreover, it is chemical selective as the absorption edges of elements are well spaced in energy from each other (for edge energy values refer to data in Bearden and Burr (1967)). The presence of small

10800 11000 11200 11400 11600 11800 12000 12200 12400 Energy (ev)

Fig. 1 Example of a XAS spectrum: the Ge-K edge of crystalline Ge (Data collected at the GILDA beamline, European Synchrotron Radiation Facility)

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

3

oscillations above the absorption edge was reported much earlier but only in the early 1970s (Sayers et al. 1971), and with the successive advent of synchrotron radiation facilities (Kincaid and Eisenberger 1975), XAS theory has been fully understood and developed (Rehr and Albers 2000), and the technique has been used in a variety of fields like solid state physics, materials science, environmental science, chemistry, structural biology, and archaeometry as can be seen in a series of publications on this technique (Köningsberger 1988; Bunker 2010). The physical origin of the μ oscillations above the edge is the modification, due to the neighboring atoms, of the final state of the electron emitted by the atom that absorbs the incoming X-ray photon (hereafter called “Central” or “Absorber” atom) (Lee et al. 1981; Stern 1988). The X-ray absorption cross-section, Σ, of an atom !

interacting with a photon described by a dipole transition operator D is given by (Natoli and Benfatto 1986): Σ ¼ 4π2 Eα

X D  ! ! E2  i p  D f  δðE  Ef þ E0 Þ

(1)

f

where E is the photon energy, α the fine structure constant, EF is the photoelectron energy, and E0 is the threshold   energy. hii and hfi are, respectively, the initial and !

final electron states, and p is the electron momentum operator. The initial state hii

is usually an atomic core state (K or L) so it is strongly localized around the central atom site: This means that the space integrals in Eq. 1 are calculated in a restricted region around the absorber. In the case of the isolated atom, h f | is an outgoing qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0Þ spherical wave where k is the photoelectron wavevector k ¼ 2m ðEE ℏ . In this case, hfi is a smooth and structureless function of E, and a similar behavior will be found in the absorption cross section Σ0. When a neighbor (Backscattering atom) is placed near to the central atom, a part of the outcoming electron wave will be backscattered by the neighbor: hf i will be the sum of the original outgoing wave plus a modified part hΔf i, coming from the waves backscattered by the neighbors. The backscattered wave will have an amplitude A(k, Rj) and a phase φ(k, Rj) which depend on the details of the scattering potential. The interference between these two terms in the absorber site gives rise to the oscillations observed in the experimental data. The oscillatory part χ of the signal is defined considering Σ0 (containing hfi) and 0Þ Σ (containing hf þ Δf i) and taking χ ¼ ðΣΣ Σ0 . The explicit expression for χ in the case of an “S” state excitation and considering only single scattering events is (Lee et al. 1981): χ¼

X j

χj ¼

X j

3S20

2 2       Nj  ^ j 2 (2) A k, Rj sin 2kRj þ φ k, Rj þ 2δc ðkÞ ek σj e2Rj λ ^e  R kR2j

4

F. d’Acapito

Here the sum is extended over the various coordination shells indexed by j, each containing Nj identical atoms at a distance Rj from the central one. The backscattering amplitude and phase, A and φ, are expressed in terms of the bond length Rj ! (curved wave approximation). In Eq. 2, e is the field polarization unit vector and the related term expresses the dependence of χ on the beam polarization. This depen   ! ! dence comes from the p  D operator and is of particular interest for single crystal samples, where the absorption coefficient depends on the sample orientation. This effect is different if {K, LI} or {LII, LIII} edges are considered. In the former case, the relation between the amplitude N of a single bond making an angle δ with the polarization vector is NK, LI ¼ 3  cos2 δ, whereas in the latter case, it is NLII , LIII ¼ 0:7 þ 0:9  cos2 δ (Iwasawa 1997). Conversely, in the case of isotropic   ! ! 2 samples (polycrystalline powders, amorphous, liquids,. . .), the e  R j term is replaced by its angular average, 1/3. In the χ(k) formula shown above, other terms are added to the simple scattering approximation in order to account for two damping processes. One is the limited lifetime of both the photoelectron and the core-hole. Indeed, the excited atom keeps this state for a limited time, after which an electron from the upper state fills the hole in the core state, destroying the extended X-ray absorption fine structure (EXAFS) signal. This problem is particularly severe when working on edges of heavy elements at high (E > 40 keV) energies (Borowski et al. 1999) as the broadening due to the core-hole limited lifetime can be several eV (Krause and Oliver 1979). Also, when the emitted photoelectron undergoes momentum transfer or inelastic scattering with other electrons, the EXAFS signal is cancelled. All this can be accounted for by adding an imaginary part to the scattering potential (Chou et al. 1987) and is contained in the exp(rj/λ) term. As λ is of the order of a few Å, EXAFS signals coming from long paths are rapidly damped and this makes the technique sensitive only to the local atomic structure around the absorber. Simultaneously with the electron emission, other electrons can also be excited in the central atom, both to bound states or continuum states (Lee and Beni 1977), as shown by photoemission experiments. This reduction of the overall amplitude is accounted for by the S02 term. The physical process at the origin of the EXAFS signal is interference, so it strongly depends on the spatial disorder of the back-scattering centers. The disorder can be originated by thermal vibrations or by random positions of the neighbor atoms in amorphous systems. Considering this, the χj function should be calculated as the integral of χj(R), weighted by the Pair Distribution Function (PDF) relative to the i-th shell (Benfatto et al. 1989). If we suppose a small disorder and a gaussian pair distribution function with mean square displacement σ2j, the result of the integral is the exp(k2σj2) term in Eq. 2 (Beni and Platzmann 1976). Note that this term, although similar to the Debye-Waller factor in X-ray diffraction, has here a different meaning. Whereas in the latter case, σ is related to the variation of the atomic position with respect to the ideal lattice site, in EXAFS it means the relative displacement of the neighbor relative to the absorber position. If thermal disorder

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

5

dominates, the values of σj can be calculated from the dynamic properties of the lattice under study, as shown in Beni and Platzmann (1976) and Sevillano et al. (1979). When the atomic oscillations become large, and this is the case, namely, for lattices at temperatures near or above their Debye temperature or in case of strongly disordered systems, the pair distribution functions also acquires a nonsymmetric character with respect to the distribution maximum, i.e., they become skewed at higher r values. This is due to the atomic interaction potential, which generally exhibits a Lennard-Jones-like behavior. The asymmetry of the PDF leads to unphysical results if data are treated in the gaussian approximation, typically a contraction of the distances and a drop in the coordination numbers, as shown in Eisenberger and Brown (1979). In this case, an analysis based on a more detailed model for the PDF is needed, by taking into account higher cumulants of the distribution (Tranquada and Ingalls 1983) or introducing ad hoc PDFs (Filipponi et al. 1995; Filipponi and DiCicco 1995; Kuzmin 1997) for the construction of the χ function. So far we have considered the photoelectron interactions in the solid as a series of single scattering events with the surrounding atoms. In a more general view, we should take into account processes where the photoexcited electron is scattered by several neighbors before returning to the central atom. While the scattering amplitude dies with the path length, the number of possible paths increases and it comes out that multiple scattering (MS) phenomena are not negligible in the interpretation of an X-ray Absorption Spectrum (Lee and Beni 1977) and extend up to several tens of eV above the edge (Benfatto et al. 1986). The calculation of the photoelectron final state hf i, accounting for multiple scattering events, can be carried out in several ways, like in the scattering approach, the band-structure approach, or the Greenfunction approach (Natoli and Benfatto 1986), all leading exactly to the same result. The X-ray cross section is expressed in matrix form by Natoli and Benfatto (1986): "

# i0, 0 X h 1 1 σ ¼ σ0 ℑ ðI  TGÞ T  lm, lm ð2L0 þ 1Þ sin2 δ0l m " # X X 1 0, 0 ℑ ½ðTGÞn Tlm, lm σ0 ℑ 2 0 ð2L0 þ 1Þ sin δl m n

(3)

Here, G is the so-called Propagator matrix and contains all the geometrical details of the cluster surrounding the absorber, whereas T is the “Scattering matrix” and it contains the details of the scattering potentials. L0 is the angular moment of the initial state, and σat is the (structureless) atomic cross section. The (1 + GT)1 term can, in some cases, be replaced by its series expansion, giving rise to the so-called multiple scattering series approximation of σ (right side of Eq. 3). The terms with n = 2 in the expansion correspond to single scattering events (photoelectron scattered by the neighbor and back to the absorber), whereas terms with n = 3, n = 4, . . . correspond to events where the photoelectron is scattered 2, 3, . . . times by other neighbors, before reaching back the absorber. Examples of single and multiple scattering paths are given in Fig. 4. The multiple scattering expansion is valid provided that the

6

F. d’Acapito

related series is convergent. As the G matrix elements depend on 1/k or, equivalently, pffiffiffiffiffiffiffiffiffiffiffiffiffiffi on 1= E  E0 (whereas the elements of the T matrix depend on the chemical nature of the scatterer), the convergence criterium is usually satisfied for photoelectron energies above a few tens of eV and depends on the system under investigation (Natoli and Benfatto 1986). The standard EXAFS formula shown in the first paragraph can be obtained from the MS expansion, taking L0 = 0 and cutting the series at the n = 2 term. For the higher terms, it can be demonstrated (Zabinsky et al. 1995) that, for any given path i, they can be reduced to expressions similar to that shown in Eq. 2 with suitable amplitudes Ain(k, Ri) and phases φin(k, Ri), where now Ri is half the total path length. In the example of Fig. 4, the path in (c) is associated to a given χi and the path in (d) to a different χj. By summing all these partial χi, the total signal can be reproduced. Debye Waller factors related to Multiple Scattering paths have been defined via a generalized formula as presented in Poiarkova and Rehr (1999). Using a different approach (Filipponi and DiCicco 1995), the contribution to the total χ of a given atomic arrangement, called γi, can also be parametrized similarly to Eq. 2. When approaching the absorption edge, the photoelectron mean free path becomes appreciably large (as well as its wavelength) and a larger number of paths contribute to the total signal. There exists a limit beyond which the MS expansion no longer converges, so it is not possible to calculate the cross section

Fig. 2 Partition of a typical XAS spectrum in the EXAFS and XANES regions, here for the spectrum of hematite (Fe2O3) at the Fe-K edge. In the EXAFS region (roughly above 50 eV from the edge), only Single Scattering (SS) or a mixture of Single and a limited number of Multiple Scattering paths contribute to the spectrum. In the XANES region, all the possible paths contribute to the spectrum and the Full Multiple Scattering (FMS) approach is needed to reproduce the spectrum

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

7

in this way. The 1/(1  GT) matrix has to be calculated explicitly and this is the so-called Full Multiple Scattering method (Tyson et al. 1992). This region is called XANES (X-ray Absorption Near Edge Structure) and it contains in principle threedimensional information on the absorber site (Fig. 2). Apart from the complex spectrum simulations, XANES can be easily used for the determination of the valence state and local symmetry of the element under investigation prior to comparison with known compounds. The valence state is derived from the energy position of the edge (Cramer et al. 1976), the more oxidized states corresponding to higher edge energy values. The local symmetry can be derived from the intensity of the small peak appearing before the edge (that in the case of K edges of the 3d metals are due to partially forbidden 1s-3d transitions) with tetrahedral environments exhibiting more intense peaks (Galoisy et al. 2001).

Data Collection and Analysis The absorption coefficient, μ, of a sample as a function of the energy is defined as:

 Φ 1 ð EÞ μðEÞ ¼ ln Φ 0 ð EÞ

Fig. 3 Experimental setup for the direct (a) and indirect (b) measurement modes of the X-ray absorption coefficient of a sample

a

Ic 1

(4)

Sample

Ic 0

Transmission Mode

b

X-ray or Electron Detector

Ic 0

Sample Fluorescence (or Electron detection) Mode

8

F. d’Acapito

where Φ0 and Φ1 are the photon fluxes before and after the sample, respectively (Fig. 3a). The fluxes are normally measured by an ionization chamber, usually filled with low-pressure gas. There are different methods to measure the absorption coefficient, depending on the nature of the sample being investigated. As a thumbrule, when the atomic species of interest contributes significantly to the total absorption of the sample, it is convenient measuring the intensity transmitted after the sample (transmission mode). This is commonly the case of systems with a high number of absorbing centers, like bulk, heavy (high atomic number Z) elements. The measure of the transmitted intensity is carried out by a second ionization chamber (Lee et al. 1981). In some cases, e.g., samples with a low concentration of absorbers, it is better to measure indirectly μexp, looking at the processes usually coupled to the X-ray absorption. Indeed, the photoionization process creates a hole in the core state that is suddenly filled by another electron. This occurs through a couple of competitive processes: a radiative one, where an electron “falls” from a higher energy state to the ionized one emitting a photon (fluorescence). Alternatively, hole filling can be accompanied by the emission of fast electrons to balance the energy (Auger processes). The higher the Z of the involved atom, the more probable is the fluorescence process, with equal probability occurring at roughly Z = 30. It must be noted that the emission of the photon occurs at a well-determined energy, depending on the atom and on the level involved. It is thus easy to separate the desired signal from the background. Exploiting the first effect described above is called the fluorescence collection mode (Jaklevich et al. 1977) (Fig. 3b). In this case, an energy-selective detector (namely, solid state detectors such Li:Si, Si Drift, High Purity Ge, . . .) is used to separate the fluorescence from the background (coherent and incoherent scattering), or fluorescence from other elements. This method is particularly well adapted for the analysis of diluted samples but leads to spectrum distortions when applied to concentrated samples. In this case either data-correction routines have to be applied (Troger et al. 1992) or data have to be collected at grazing exit angle (Pfalzer et al. 1999). Recently, a novel detection scheme for fluorescence XAS on concentrated samples has been proposed (Achkar et al. 2011) where the absorption signal of an atom A is collected by recording the modulations of the fluorescence yield of a lighter atom B while scanning the energy through the A absorption edge. A method that gained considerable popularity in the latest years is the High Energy Resolution Fluorescence Detection (HERFD). Indeed, as shown in H€am€al€ainen et al. (1991), collecting the fluorescence yield from a sample with an energy resolution lower than the core-hole linewidth produces XANES spectra no longer broadened by the initial state, so permitting to evidence structures in the spectrum barely visible with the conventional data collection. This technique is called High Energy Resolution Fluorescence Detection (HERFD) and results to be particularly effective in the analysis of the L edges of 5d metals where the typical broadening of the states is of the order of 5–10 eV. The detector is usually a multiple crystal analyzer with a limited solid angle acceptance, so this technique can be carried out only on very intense sources. A description of a complete spectrometer

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

9

can be found in Glatzel et al. (2005) and Rovezzi and Glatzel (2014). The drawback is that using crystal analyzers greatly reduces the effective collection solid angle for the fluorescence so the total efficiency (flux on the detector) results to be greatly reduced with respect to the conventional method with a considerable increase of the minimum concentration limits. Alternatively to fluorescence, the total electron yield from the sample due to cascades initiated by the Auger processes can be detected (Citrin et al. 1978). The signal is measured with a channeltron detector or simply by measuring the photoionization current from the sample. This method has the peculiarity of probing a few thousand Å under the sample surface (due to the limited electron escape depth) and can be useful in studying macroscopically layered structures, e.g., ion-implanted materials. The other way to study surfaces is based on the x-ray total reflection: When the x-ray beam impinges with a sufficiently small angle on the sample, the x-ray penetration is limited to a few Å (Parratt 1954). In this case, the signal can be collected both from the fluorescence yield from the sample or from the reflected beam, using detectors described above such as photodiodes or scintillation detectors (Heald et al. 1988). In this setup, the ultra-high vacuum environment is not needed and it can be exploited to investigate liquid surfaces buried interfaces or gas-solid interfaces. An example of a typical experimental apparatus for total reflection EXAFS can be found in d’Acapito et al. (2003 and references therein) and consists of an accurate sample positioning stage with detectors for the impinging and reflected beam as well as for the fluorescence from the sample. The main limitation of this procedure is the need of sufficiently long (a few cm) and smooth (microscopeslide grade) samples. This method, enhancing the signal from the surface thin film with respect to the substrate, is particularly indicated in the study of thin layers like those obtained by sol-gel (d’Acapito et al. 2008). By exploiting the luminescent properties of the samples, it is possible to derive interesting results using the X-ray excited optical luminescence (XEOL) technique. XEOL is a particular data collection technique (Rogalev and Goulon 2002; Sham and Rosenberg 2007) which derives the absorption coefficient μ from the optical luminescence that follows the absorption of an X-ray photon in particular materials. The electrons emitted in each absorption event produce a shower of lower energy electrons (that extends far from the X-ray absorbing atom) in the conduction band that excites the luminescent centers contained in the material with consequent emission of a low energy photon. The intensity of this signal is proportional to the absorption coefficient and will contain the XAS signal related to the environment of the absorbing atoms. The site selectivity between luminescent and nonluminescent zones under investigation can be achieved provided that the two families are sufficiently spatially separated (10 nm (Rogalev and Goulon 2002) for soft X-rays, an order of magnitude more for hard X-rays). In practical terms, the data collection is realized by collecting the optical emission from the sample with a lens focusing on an optic fiber leading to the entrance slit of a wavelength dispersive optical monochromator. Different portions of the emission spectrum can be thus used for the data collection permitting the selection of the luminescent sites.

F. d’Acapito

a

1.8

Absorption Coeff. (Arb. Units)

10

1.6

b 3

1.4

2

1.2

1 k2 * x(k)

1 0.8 0.6 0.4 0.2

-3 -4

d

4

150

3

100

2.5

50

2 1.5

-100 -150

0 3

4

5

R [Å]

20

15

20

0

1

2

15

-50

0.5 1

10

200

3.5

0

5

k [Å 1]

q2 * x(q)

-3

0

-

Energy (eV)

Magnitude of the F T [Å ]

-1 -2

0 10500 11000 11500 12000 12500 13000 13500

c

0

6

7

8

-200

0

5

10 q [Å -1]

Fig. 4 Example of EXAFS data treatment relative to crystalline Ge. (a) raw absorption spectrum, (b) extracted EXAFS spectrum χ(k), (c) Fourier Transform, (d) Back Fourier Transform

Data Analysis Procedures It can be seen in Eq. 2 that each particular coordination shell contributes to the total signal with a sine oscillation in k space with frequency 2R (plus a term ϕ(k, Rj) weakly depending on k). Here we provide a description of the standard data analysis procedure (Lee et al. 1981): the raw absorption spectrum (Fig. 4a) the oscillating part χ is isolated by subtraction of a structureless atomic background (Fig. 4b). When the χ signal is Fourier transformed (FT), peaks appear in the spectrum each corresponding to a different coordination shell (Fig. 4c). By windowing the FT in a way to leave only one peak and by applying to this function a back Fourier transform, the contribution from only one coordination shell is obtained (Fig. 4d). The basic principle of EXAFS data analysis is to reproduce the filtered experimental data (Fig. 4d) with a model based on the expression shown in Eq. 2. In that expression, two kinds of variables are present: • Parameters linked to the photoelectron interaction with the medium, like the backscattered amplitude A(k, Rj), phase φ(k, Rj), photoelectron mean free path λ, and S02. • Parameters linked to the local atomic structure, like the number of neighbors Nj, the bond length Rj, and the Debye-Waller factor σj.

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

11

The former parameters can be calculated from ab initio methods. Several codes, like FEFF (Zabinsky et al. 1995; Ankudinov et al. 1998), GNXAS (Filipponi et al. 1995), and EXCURVE (Binsted) are available for this purpose. These functions can be used in a multiparameter fitting procedure with variable atomic structure parameters, to reproduce the experimental data. Different strategies of data fitting are used, the fit being done on Fourier-filtered EXAFS data (Lee et al. 1981) or directly on the FT (Zabinsky et al. 1995; Newille 2005). In both cases, the analysis can be done on a part of the spectrum (namely, the first shell) or on the whole. Other codes fit directly the absorption coefficient, making no use of Fourier Transform (Filipponi and DiCicco 1995), but in this case, a model accurately describing the entire structure needs to be considered. In general, bond lengths are determined with relatively high accuracy (0.02 Å, or better) whereas amplitude parameters (N and σ) are determined within 10 %. A different approach is made on the XANES region. Here, since the calculation of the absorption coefficient is quite time consuming, the spectrum is reproduced at a qualitative level starting from a supposed three-dimensional model of the structure. Different codes based on a real space approach, like CONTINUUM (Tyson et al. 1992), FDMNES (Joly 2001), FEFF9 (Rehr et al. 2009) permit a modeling of the XANES part starting form a given structure and using the FMS approach. Other codes based on band structure methods (WIEN2K (Blaha et al. 1990), SPRKKR (Ebert 1998), XSPECTRA (Gougoussis et al. 2009)) are also available. Quite recently a new code called MXAN has been presented (Benfatto et al. 2001), permitting the structural refinement using the XANES part of the spectrum, with interesting results in biological applications (Della Longa et al. 2001).

Applications of XAS to Sol-Gel Materials A variety of examples on the use of XAS on materials obtained by sol-gel route will be presented here, with particular attention to materials for optical applications, catalysis, nanoparticles, and batteries.

Materials for Optics A considerable work has been carried out on Rare-Earth doped phosphors for applications in white Light Emitting Diodes. Indeed, these devices emit light in the blue or near ultraviolet (UV) portion of the spectrum and need a conversion agent to obtain white emission. Sol-Gel revealed to be a powerful technique for the realization of phosphors via treatments using low temperatures. Potdevin et al. (2010a, b) have studied the effect of Acetylacetone (acacH) as a chemical modifier in the synthesis of Tb-doped Y3Al5O12 (Yttrium Aluminum Garnet, YAG) phosphor in powder form. Indeed, sol-gel procedures revealed to be interesting with respect to the preparation via solid-state reaction involving high temperatures (T 1500 C) as this process can lead to byproducts spoiling the luminescent

12

F. d’Acapito

properties. In the procedure of production of YAG, it was found that using acacH instead of hydrolyzing the solution with water (Potdevin et al. 2010a), it was possible to obtain better materials using lower temperatures for treating the xerogels. A XAS investigation at the Y-K edge carried out on xerogels obtained with or without acacH revealed that in the former case the realization of a crystalline order around Y started already at 600 C whereas 800 C were necessary in the case of the materials obtained with the standard hydrolysis method (Fig. 5). The same behavior was evidenced in the environment of Tb dopants, confirming that the crystalline order was realized at lower temperatures in the matrix obtained with acacH and that dopants are incorporated in crystalline nuclei since the early stages of the process. Nakajima et al. (2013) have studied the incorporation of Ca-α-SiAlON phosphor doped with Eu (again for applications in white LEDs) in TiO2-SiO2 matrices obtained by sol-gel. Analyzing the XANES spectra, the authors have found that Ti was 5-coordinated for low Ti content (10 mol%) and turned predominantly 6-coordinated for high content (30 mol %). Eu contained in the phosphor was affected by the incorporation in the sol-gel matrix as the ratio Eu2+/Eu3+ was altered with a reducing effect (higher Eu2+ content) following the incorporation. Ye et al. (2013) have demonstrated the possibility of doping Mg2TiO4 with Mn via a sol-gel method with the aim of obtaining a red phosphor to be used in the wavelength conversion in white LEDs. In this case, X-ray Diffraction (XRD) revealed the formation of the Mg2TiO4 phase after annealing the xerogel at 1300 C. The chemical state and location of Mn dopants was elucidated by XAS at the Mn-K edge. By comparison with model compounds, the valence state of Mn was defined at Mn4+ and the detailed analysis of the XAS data (3 coordination shells) permitted to recognize MnTi as the incorporation site of this metal as required for the formation of the phosphor. Light conversion is needed not only in LED technology but also in solar cells. Indeed, the mismatch between the gap of Si and the solar spectrum leads to a limited efficiency of this kind of cells. To overcome this problem, particular coatings for solar cells have been proposed, called Quantum Cutting materials, capable to convert energy from the UV-visible (VIS) region into Near Infra-Red (NIR) where Si cells have a high efficiency. Sol-gel is an interesting method to produce these materials for the capability of covering large areas and the limited cost of the process. Terra et al. (2013) have investigated the formation of Tb3+-Yb3+ doped ZrO2 nanocrystals obtained by sol-gel. The two rare-earth ions have different roles: Tb3+ is the sensitizer that absorbs light in the VIS range and then transfers the excitation to Yb3+ (activator) which emits in the NIR. During the production of the material, it is of paramount importance that the dopants enter the matrix in the Zr-substitutional site and keeping the 3+ valence state. Materials were obtained with the one-step sol-gel method and contained different amounts of Yb (0–20 mol%) and fixed (1 mol%) amount of Tb. XRD showed that Yb favors the formation of cubic zirconia for content values above about 5 mol%. XAS at the Tb-L3 edge elucidated the valence state of Tb in the xerogel and calcined materials. In the former case, Tb is 3+ for any value of Yb content, presumably due to the fact that Tb nitrate was used as

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

13

F(R) /arb.units

1100 °C 800 °C 700 °C 600 °C 400 °C 200 °C xerogel 0

2

4

6

8

R/Å −1

k/A

F(R) /arb.units

1100 °C 800 °C 700 °C 600 °C 400 °C 200 °C xerogel 0

2

4

6

8

R/Å

Fig. 5 Comparison between the Fourier Transforms of XAS data at the Y-K edge in xerogels obtained with standard hydrolysis method (left) or with acetylacetone (right) (Figure Reproduced from Ref. (Potdevin et al. 2010a) with permission from The Royal Society of Chemistry). The peaks appearing around 3A are due to the second coordination (cationic) shell and indicate the formation of a crystalline nucleus. In the right panel (relative to samples prepared with acacH the second shell peak start to appear already at 600 C whereas the same peaks appear at 800 C in the case (left panel) of samples prepared with the standard hydrolysis method

precursor. Upon calcination, formation of Tb4+ was observed for low-Yb containing materials, whereas Tb resulted to keep the 3+ state for Yb at 20 mol%. The inhibition of the unwanted oxidation of Tb was explained by the effect of oxygen vacancies generated by Yb (aliovalent with respect to Zr) and permitted to define the optimum conditions for obtaining an effective QC material. Similar effects linked to the

14

F. d’Acapito

valence state of rare earths have been also reported in Carvalho et al. (2016) where the valence state of Pr in Gd-doped ZrO2 obtained by a standard sol-gel procedure was controlled via the Gd content as revealed by XAS at the Pr-L3 edge. The authors have shown that with Gd at 10 mol % the structure of ZrO2 is cubic and that Pr is exclusively in the Pr3+ state. Codoping can be used also to improve the behavior of photocatalysts as shown in the case of TiO2 by Majeed et al. (2015). TiO2 is known to be an excellent photocatalyst but it suffers from the fact that its gap is 3.15 eV so it can use only about 3–5 % of the solar spectrum. Doping TiO2 with metals can contribute to an improvement of its properties as in the case of Mo doping. TiO2 was prepared via sol-gel with the addition of molybdic acid as Mo precursor. Mo was added at different concentration values from 1 % to 10 % and the obtained samples were analyzed by XAS at the Mo-K edge. The XAS data revealed that Mo is present in two phases: substitutional phase in TiO2 and a separated MoO3 phase. Mo is totally substitutional for the samples at 1 % and the amount of this phase decays with Mo content, being only 7 % in the highly doped sample. Tests of efficiency were carried out measuring the photodegradation of Methylene-Blue pigment in presence of the catalyst under visible irradiation. The result was that the better performing sample was that containing 5 % Mo out of which 30 % is in substitutional phase, demonstrating that the interplay of all three components (TiO2, Mo: TiO2, and MoO3) contribute to the improvement of the efficiency of the catalyst. Studying luminescent materials permits the exploitation of nonconventional XAS techniques like X-ray Excited Optical Luminescence (XEOL). This technique has been used to characterize Ge nanoparticles (NP) (Little et al. 2014) in free-standing form (comparing in this case oxygen-terminated particles with hydrogen-terminated particles) and Ge nanoparticles embedded in silica produced by the sol-gel route. The spectra at the Ge-K edge were collected in transmission mode (sensitive to all the Ge species) and XEOL mode (sensitive to Ge species in luminescent regions) for comparison. In the case of Ge nanoparticles surrounded by O species (O-terminated free standing NPs or NPs embedded in silica), both spectra revealed a predominance of Ge-O bonds meaning that the luminescence comes from the oxidized zones. In the case of free NPs terminated with hydrogen, the XEOL spectrum exhibited a higher disorder and a comparison with Molecular Dynamics simulations permitted to establish that the luminescent regions were the most disordered one at the boundaries of the particle. XEOL detection was also used to characterize Rare-Earth doped nanospheres for medical applications (Fortes et al. 2014). Hollow and bulk silica particles coated with Er2O3 and Yb2O3 were characterized by XAS in fluorescence and optically excited mode. The comparison between the spectra obtained in the two ways showed no differences meaning that all the rare-earth species were present in luminescent portions of the samples. Further uses of XAS for the characterization of materials for optical applications produced by sol-gel can be found in the case of Er-doped Silica-Hafnia waveguides for optical amplifiers (d’Acapito et al. 2008; Afify et al. 2007), glass ceramics (Van et al. 2015), and scintillator for detectors (Liu et al. 2015b).

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

15

Nanoparticles The sol-gel process permits to produce nanocrystalline materials with a good control of composition and size that reveals to be invaluable for a large variety of systems. Metal oxide nanoparticles, namely, find applications in several fields like gas sensors, catalysts, electrodes for batteries and transparent electrodes. Caetano et al. (2014) have studied in detail the structural evolution from the gel to the densified material of SnO2 nanoparticles using time-resolved XAS at the Sn-K edge. The synthesis was based on Tin Chloride (SnCl4 • 5H2O) dissolved in ethanol that was subsequently hydrolyzed with water. Quick-XAS data were collected at a frequency of 2 Hz during hydrolyzation and XAS, and Raman spectra were collected at the same time in a specially conceived cell. The first 30 min of the process have been described as a progressive loss of chlorine ligands for Sn substituted by oxygen ligands. The XAS data as a function of time could be reproduced as a linear combination of different complexes in solution [SnClx(H2O)6x]4x (with x = 3, 4 and 5) and the spectrum of SnO2 nanoparticles. With the support of Raman data, it was possible to describe the evolution of the various complexes involving Sn as shown in Fig. 6. Time-resolved XAS was also used in the description of the growth of TiO2 nanoparticles in sol-gel as reported in Stotzel et al. (2010). In an ad hoc conceived reaction, cell Titanium tetraisopropoxide was solved in isopropanol and successively

Fig. 6 Evolution of different species [SnClx(H2O)6x]4x and SnO2 in solution during hydrolyzation and annealing. Results from XAS and Raman data collected in situ at the same time (Reprinted with permission from Ref. (Caetano et al. 2014). Copyright 2014 American Chemical Society)

16

F. d’Acapito

a water-isopropanol solution was added to carry out the hydrolysis of the solution. At the same time, XAS spectra were collected with a duration time of 5 s (10 successive spectra were averaged to improve the signal to noise ratio) up a total process time of 1500–2000s. The analysis of the XAS data was carried out by linear combination of the spectra at different times with model compounds. The formation of TiO2 particles involves different steps: polynuclear titanium species are created and, after a few minutes, oligomers of the type Ti11O13 and Ti12O16. The consumption of the precursor is particularly evident as it exhibits a particularly strong pre-edge resonance at about 4970 eV that is considerably damped in the oligomers and TiO2 species due to the different site symmetry in the two cases. In an intermediate phase, the oligomers start to form titania nanoparticles and, at the end, they are totally converted in TiO2. The doping of SnO2 nanoparticles with Sb, obtained by sol-gel processing, has been studied by V. Geraldo et al. (2010). In this case, gels at different Sb content (3–16 %) were prepared from SbF3 and (SnCl4 • 5H2O) precursors. The obtained xerogels were treated at 200 C or 500 C to study the evolution of the valence state and incorporation site of Sb. It was evidenced that after treating at 200 C, Sb is pentavalent if present at low concentration ( 4 %) and a trivalent fraction is present for higher Sb content. After firing at 500 C, Sb is predominantly pentavalent. Sb5+ species enter the SnO2 matrix in a substitutional site. Conversely Sb3+ species, more abundant at high concentration values, remain grafted at the surface of SnO2 particles. The formation of Cooxide nanoparticles in porous silica has been studied by Liu et al. (2015a). In this case, Cobalt doped silica at different Co/Si molar ratio (5 %, 10 %, 25 %) was obtained by sol-gel using Tetraethoxysilane (TEOS) for silicon and [Co(NO3) • 6 H2O] as precursor for Cobalt. The sols were first dried at 60 C and then ground in fine powder and calcined at 630 C. The materials were successively tested with a hydrothermal treatment (HT) consisting in annealing the samples at 550 C for 40 h in an atmosphere containing water vapor at 75 mol%. XAS measurements at the Co-K edge were carried out to reveal the chemical status of the metal prior or after the HT. All samples revealed to be stable under HT; no remarkable differences were present in the XAS spectra with respect to the non-HT ones. In samples containing 5 % Co, Cobalt was found tetracoordinated to oxygen with a faint signal from a Co second shell. On the other samples with higher Co content, a well-defined second shell signal was visible that was attributed to the formation of Co3O4 nanoparticles. Again on the topic of metallic nanoparticles, Takao et al. (2012) have demonstrated the production of metallic Pd and PdO particles in hollow silica nanoparticles. This method possesses the advantage of producing metallic clusters with an extremely reduced number of atoms (4–60) that is difficult to be obtained in other ways. The hollow nanoparticles were produced starting from a spherical template of Pd12L24 where L is a suitable ligand (triethoxysil group). The silica is deposited on the surface of this template by using tetramethoxysilane with the standard procedure. Successive calcination at 400 C in air eliminated the ligands leaving small clusters with 12 Pd atoms inside the hollows of the spheres. XAS at the Pd-K edge showed that the clusters are made of PdO and that further treatment with H2 leads to the formation of metallic Pd.

X-ray Absorption Spectroscopy Studies on Materials Obtained by the Sol-Gel. . .

17

Further studies on nanoparticles obtained by sol-gel and characterized by XAS have been carried out for the production of magnetic semiconductors (Bilecka et al. 2011; Kumar et al. 2014; Hu et al. 2011; Liu et al. 2010) and catalysts (Santos et al. 2012; Prieto et al. 2010; Mathew et al. 2011).

Materials for Catalysts and Sensors Metallic nanoparticles (MNP) embedded in a porous material are largely used in heterogeneous catalysis but these systems easily undergo deactivation by sintering. A way to limit this problem is to host them in a suitable host to maintain their dispersion. Among others, silica Aerogels have been proposed but their use remains limited due to their high production costs. Recently, Kristiansen et al. (2011, 2012) have succeeded in incorporating Cu in silica aerogels produced by Ambient Pressure Drying, a method that greatly simplifies the production of this kind of material. In the first paper (Kristiansen et al. 2011), the group has demonstrated the dispersion of Cu species in the aerogel by looking to the XAS spectra collected at the Cu-K edge and verified that no Cu-Cu bonds were present. In a subsequent study (Kristiansen et al. 2012), the group has used XAS to study the formation of Cu nanoparticles in presence of a reductive atmosphere (5 % H2 in He) and at high temperature. The XAS spectra presented metal-metal bonds with a reduction of the first shell coordination number due to the reduced size of the particles (estimated to contain about 80 atoms). Data were collected at increasing temperature values and differences in the spectra were evidenced suggesting a change in shape and local symmetry of the clusters. In order to interpret the XAS experimental data, XAS spectra were simulated based on the results of atomistic simulations (in the framework of the DFT theory) of clusters of different shape. The results of the study are collected in Fig. 7. At about 300 C, Cu forms nanoclusters that, from the number of Cu first neighbors observed (8) and comparison with DFT theoretical data, are identified as clusters with Td symmetry (fcc structure) containing about 80 atoms. The size of these particles (about 1.4 nm) is well matched with that of the typical pores ( 1/2 are additionally affected by the electric quadrupole interaction, an interaction between the nuclear electric quadrupole moment and the gradient in the electric field. Although this is an electrical interaction, it depends on the nuclear spin quantum number and therefore its effect is also observed in the NMR spectrum.

Atomic-Scale Structure of Gel Materials by Solid-State NMR

5

All of these interactions are spatially dependent and are described by three components, which are typically the average or isotropic value, an anisotropy, and an asymmetry. In NMR papers there are several ways of defining these values, and the reader should carefully note which convention has been adopted. It is this anisotropy that causes broadening of the NMR spectrum from an isotropic powder sample, where all particle orientations are present with respect to the static magnetic field. Fortunately, to first-order, the energy of interaction can be written in a similar manner for all the major interactions
 Frequency ¼ Aiso þ Δ A 3 cos2 θ  1 þ η sin2 θ cos 2φ

(2)

where Aiso is the isotropic value, ΔA is a measure of the anisotropy, and η is the asymmetry parameter of the interaction. Lineshapes can provide very important information about the local symmetry of the interaction which can often be related to some details of the local structural symmetry. A complication arises for quadrupole nuclei which constitute around three quarters of the NMR-active nuclei in the Periodic Table (MacKenzie and Smith 2002). For non-integer spin quadrupole nuclei, the central (1/2, 1/2) transition does not experience any first-order quadrupole effects, then the interaction has to be considered to higher-order of perturbation. The higher-order perturbation terms have more complex angular dependencies, which has consequences for the line narrowing techniques that are employed. The quadrupole interaction is described by the quadrupole coupling constant χQ = e2qQ/h and the asymmetry parameter (η) (MacKenzie and Smith 2002).

Basic Experimental Considerations The NMR experiment involves measurement of the energy level separation by application of a time-varying magnetic field B1 orthogonal to B0. B1 excites transitions when its frequency (ω) is close to ω0 in eq. (1). There is simultaneous excitation of a wide range of frequencies by a short rf pulse of duration Tp. The magnetization is consequently tipped by an angle θp (=γB1Tp) away from the direction defined by B0. After a π/2-pulse, all the net magnetization is in a plane transverse to the direction to B0 and hence is termed transverse magnetization. B0 exerts a torque on the transverse magnetization, which will consequently Larmor precess about B0. The rotating magnetization is then providing an alternating flux linkage with the NMR coil that, through Faraday’s Law of Electromagnetic Induction, will cause a voltage in the NMR coil, which is recorded as a function of time, decaying through spin–spin relaxation processes characterized by a time constant T2. An important experimental consideration is the time that must be allowed to elapse before the experiment can be repeated. After the pulse the spin distribution between energy levels has been significantly perturbed away from equilibrium, and the time

6

M.E. Smith and D. Holland

constant for recovery to equilibrium is called the spin–lattice relaxation time (T1). One of the major advantages of NMR is that spectral integration directly gives the populations of the different species present, but this is only true provided that all species are fully relaxed. Intrinsic signal sensitivity arising from the Boltzmann distribution is relatively weak for a low-frequency spectroscopy such as NMR, so to improve signal-to-noise the experiment is repeated n times and the results summed to produce a factor of √n improvement. The averaged time domain signal is then Fourier transformed to produce the spectrum as a function of frequency. The basic components of a pulse FT NMR spectrometer are shown schematically in Fig. 1. It can be seen that in concept, an NMR spectrometer is quite simple. There is a high field magnet that provides the basic Zeeman states on which to carry out the NMR experiment. The probe circuit containing the sample in the NMR coil is placed in the magnetic field. The probe is connected to the transmitter which provides the pulses that produce the excitation. The probe is also connected to the receiver, and it requires some careful design to ensure that the receiver, that is sensitive to μV, does not see any of the large excitation voltages produced by the transmitter. Despite the relative simplicity, there is a lot of technology behind NMR spectrometers. Key advances in the last decade have included much lower noise electronic systems improving the sensitivity, high power (1 kW) linear amplifiers so that more sophisticated shaped pulses can be used and much more use of digital rather than analogue technology. Typical spectrometers for solid-state NMR studies of sol–gel materials are equipped with medium field magnets of 7.05–11.7 T. For some applications, much higher or much lower magnetic fields may be advantageous. Persistent high field magnet technology now sees wide bore (89 mm) magnets at 20 T commercially available, with a few higher field solids-enabled facilities up 23.5 T, but with standard bore (57 mm) magnets.

ADC

Frequency Synthesiser

Magnet

Receiver Probe

Computer

Gate

Pre-amplifier

Amplifier

Fig. 1 Basic components of a pulsed FT NMR spectrometer

Atomic-Scale Structure of Gel Materials by Solid-State NMR

7

Experimental Formats Static Broad-Line Experiments. The simplest experiment to carry out is one-pulse acquisition (Fig. 2a). However, with broad lines that decay rapidly in the time domain (i.e., a short T2), there is often a problem with the start of the signal being corrupted by so-called deadtime problems although this problem has reduced as the applied magnetic field and observation frequency have increased. These can be overcome by forming a signal with an effective time zero point outside the deadtime, i.e., an echo (Kunwar et al. 1986). There is a large variety of methods to form such echoes. Most echo methods are two-pulse sequences, with the classic spin-echo consisting of 90 –τ–180 , which refocuses at time τ after the second pulse (Fig. 2b). The echo decay shape is a good replica of the original FID, so that the spectrum is usually produced by FT from the echo maximum. This is an example of where technology has changed the approach, since to determine broad spectral lineshapes accurately from echoes, short (termed hard) rf pulses were formerly preferred for uniform excitation. Alternatively to allow unambiguous simulation with undistorted lines, relatively soft pulses collecting partial spectral which are then summed using frequency offset techniques, such as variable offset cumulative spectroscopy (VOCS), can be used to collect the broad-line spectra (Massiot et al. 1995).

Fig. 2 Pulse sequences commonly used for solid-state NMR: (a) one pulse, (b) spin echo, and (c) cross-polarization

8

M.E. Smith and D. Holland

Increased sensitivity can be provided by summed echo trains such as in the quadrupolar Carr–Purcell–Meiboom–Gill (QCPMG) pulse sequence (Siegel et al. 2005a, b). FT of the complete echo train results in a spectrum consisting of narrow “spikelets” whose envelope is the same spectrum that would be obtained by FT of one-half echo. In addition polarization transfer (PT) experiments have become popular for enhancing the signal of half-integer spin quadrupolar nuclei by transferring polarization from outer (satellite) transitions to the observed central transition. The new electronics of spectrometers allow sophisticated pulse shapes to saturate or invert the outer transitions to effect PT. These PT sequences can be combined with the QCPMG pulse train to provide further signal enhancement (Schurko 2013). Magic-Angle Spinning. The most commonly adopted experimental approach for solid-state NMR is to place the powder sample in a container (rotor) which is inclined at 54.7 to the magnetic field and rapidly rotated about its axis. This is magic-angle spinning (MAS). This causes considerable narrowing of the spectrum (Fig. 3) since the anisotropic first-order interactions described by Eq. 2 pick up an additional modulation of 3 cos2 β – 1 under MAS, where β is the angle between the rotor axis and the magnetic field. There is a usually a compromise necessary between maximum rotation rate and sensitivity, since the safe maximum rate scales inversely with the diameter of the rotor. Typical spinning speeds used are in the range 5–25 kHz, but over the last decade with much smaller diameter rotors, down to 1 mm, MAS in excess of 100 kHz is now available. The rate that is necessary to cause line narrowing depends on the interaction to be narrowed. For example, chemical shift anisotropy breaks up into the central line and spinning sidebands, Fig. 3 Effect of magic-angle spinning on the 27Al lineshape of a sintered Y2O3:Al2O3 mixture showing the great improvement in resolution from static (upper) to spinning (lower)

200

150

100

50 27Al

0

−50 −100 −150 −200

shift (ppm)

Atomic-Scale Structure of Gel Materials by Solid-State NMR

9

separated by the spinning frequency, at frequencies well below the static linewidth. Strongly dipolar coupled lines, such as between interacting protons, need the spinning rate to approach the static linewidth which can be 50 kHz. For amorphous materials, often a consideration is the intrinsic linewidth caused by the isotropic interaction being slightly different from site to site. This is termed chemical shift dispersion and produces a width that must be exceeded for line narrowing to occur. For some nuclei in sol–gel materials (e.g., 29Si), this condition is readily met, whereas for others (e.g., 119Sn, 207Pb) this often requires much faster rotation, but also can now usually be met with the ultrafast MAS available. MAS can also be used in the PT experiments to enhance the signal from quadrupolar nuclei with sequences such as rotor-assisted polarization transfer (RAPT) (Yao et al. 2000) which uses the changing frequency of the STs during rotation to aid the saturation process. High-Resolution One-Dimensional Techniques for Quadrupolar Nuclei. Simple MAS about a single axis cannot produce a completely averaged isotropic spectrum from a non-integer spin quadrupole nucleus, because the second-order terms (/35 cos4 θ – 30 cos2 θ + 3) that are particularly important for the central transition do not transform to zero under MAS at 54.7 . This can leave substantial broadening, and sometimes a characteristic well-defined lineshape (Fig. 4) is observed from which the NMR interaction parameters can be extracted. In structurally disordered materials, the distribution of interactions means that the well-defined

Fig. 4 (a) The second-order quadrupole lineshape from the central transition with the site on average experiencing χQ = 2 MHz and η = 0 observed at a frequency of 80 MHz. The lineshapes are then modified to take into account a distribution of interactions where (b) the isotropic chemical shift has a Gaussian distribution of 0.17A, (c) the quadrupole interaction has a Gaussian distribution of 340A, and (d) both are distributed (A = 2344 Hz)

a

b

c

d

1

0

−1

−2

−3

10

M.E. Smith and D. Holland

lineshape becomes smoothed and MAS spectra typically show a characteristic tail to negative shift (Fig. 4). This is seen in spectra from sol–gel materials for nuclei such as 27Al. An important feature of this residual second-order quadrupole interaction is that it scales inversely with magnetic field so that the residual linewidths are greater at lower applied magnetic fields. There is also a residual isotropic second-order quadrupole shift that varies with field so that, unlike spin-1/2 nuclei, the peak position is field dependent. The availability of higher magnetic fields has reduced second-order quadrupolar effects, meaning higher-quality spectra can be observed from nuclei such as 27Al, as well as being the major driving force behind the extension of MAS to low-γ nuclei. Low-γ nuclei will be observed at relatively low frequencies, even at high magnetic fields so that quadrupolar effects can still be pronounced for them. Other approaches have been developed where second-order quadrupole effects are smaller, such as MAS of the non-central or satellite transitions that can produce narrower spectra for some combinations of I and mz. Other schemes such as double angle rotation (DOR) and dynamic angle spinning (DAS) produce more complete averaging of the interactions by imposing more complex time dependencies on the spatial part of the quadrupole interaction, but their application to sol–gel materials has been very limited (see Smith and van Eck 1999; MacKenzie and Smith 2002 and references therein). DOR probe technology has become more reliable and widely available in recent years. Two-Dimensional Multiple-Quantum MAS. For non-integer spin quadrupolar nuclei, a two-dimensional (2D) multiple-quantum magic-angle spinning (2D MQMAS) technique emerged in 1995 that has had a profound effect on their study and has meant that much higher resolution spectra can be produced from these nuclei in sol–gel materials. This experiment correlates the (m, –m) multiple-quantum transition to the (1/2, 1/2) transition. The resolution enhancement stems from the fact that the quadrupole frequencies for both transitions are correlated. At specific times, the anisotropic parts of the quadrupole interaction are refocused and an echo forms. The amplitude of the echo is determined only by isotropic interactions, so that the experiment is repeated at different echo times which forms the 2D data set where the echo amplitude is modulated only by isotropic effects. FT of the echo amplitude then produces an isotropic spectrum that has much higher resolution than the MAS spectrum. The 2D data set also contains the anisotropic information (see Smith and van Eck 1999; MacKenzie and Smith 2002 and references therein). Multiple Resonance Experiments. There is a range of approaches where, during different parts of the experiment, different nuclei are manipulated which means the nuclei can influence one another, often providing structural information. The most widely used multiple resonance method is cross-polarization (CP) where usually both abundant (I) and dilute (S) nuclei coexist, then transverse I-magnetization is first created and then spin-locked (Fig. 2c). Mutual magnetization exchange between the I and S spins is then allowed by simultaneously applying rf fields to both spin systems that satisfy the Hartmann–Hahn condition (γ I B1I = γ S B1S). This was

Atomic-Scale Structure of Gel Materials by Solid-State NMR

11

developed for improved sensitivity for weak signals, but, for example, CP between 1 H and 29Si can also be used to provide additional structural information, since silicon atoms in closest proximity to protons are preferentially enhanced. For example, when there are differences between CP and non-CP spectra, as is the case for the hydrated phases of pure silica, the intensities of the Q2(OH) and Q3(OH) peaks bonded to hydroxyl groups are usually significantly increased in intensity by comparison with the Q4 peaks. The dynamics of polarization transfer are determined by spin thermodynamics and the strength of the coupling. The magnetization of the spin being polarized should build up exponentially, with a characteristic buildup time of TIS, the cross-relaxation time. As this depends on the dipolar coupling strength, it provides information about the relative spatial proximity of the different sites. This data should never be used too quantitatively as motion can produce some complicating effects, but is a good guide as to how close different species are. A range of more sophisticated double-resonance experiments that exploit heteronuclear dipolar coupling have been developed. Qualitative information regarding the proximity of spins I and S and even quantitative information regarding the I–S distance can be derived. Although MAS will readily average away the heteronuclear dipolar coupling, it can be reintroduced in various sequences, with the spinning greatly enhancing the sensitivity and resolution, allows the echo to be measured for longer times. The presence of quadrupolar nuclei may prohibit straightforward application of these experiments and can alter the interpretation of the results. transfer of populations by double-resonance (TRAPDOR) and rotationalecho adiabatic passage double-resonance (REAPDOR) experiments are designed specifically for quadrupolar nuclei and do not work on systems containing only spin1/2 nuclei (see Smith and van Eck 1999; MacKenzie and Smith 2002 for a detailed discussion and the primary references). Correlation Experiments. In looking at spatial proximity of different units, 2D heteronuclear correlation can be a useful approach (Babonneau and Maquet 2000). In systems co-hydrolyzed to produce mixed, differently functionalized siloxane systems, the 1H resonances from the attached –H and –CH3 groups can be distinguished. The buildup of correlation peaks with the mixing time shows that the two networks are usually intimately mixed (Babonneau and Maquet 2000; Peeters et al. 1995). When there are spin-1/2 nuclei in the system especially if the nuclei have high natural abundance, then, using either the dipolar or J-coupling, various correlation experiments exist between combinations of single-quantum (SQ), double-quantum (DQ), and triple-quantum (TQ) signals. The key is that 1D spectra simply reveal which units are present and their relative disposition is only available via indirect observations such as the influence on the isotropic chemical shift. In correlation spectra cross-peaks occur when there is an interaction, with sequences based around the J-coupling depending on a chemical bond between the two species.

12

M.E. Smith and D. Holland

Computational Approaches One of the major strides in methodology associated with solid-state NMR since 2005 is in computational work to simulate spectra and the ability to calculate the NMR interaction parameters from a proposed structure. The following sections look at these two key areas. Spectral Analysis and Simulation. Several computer programs currently exist for fitting solid-state NMR spectra and for taking into account the effects of the pulse sequences on lineshapes. QuadFit is a program tailored toward fitting of NMR spectra from amorphous materials, but is also equally applicable to spectra from crystalline materials (http://go.warwick.ac.uk/quadfit) (Kemp and Smith 2009). This paper makes reference to other widely used platforms such as SIMPSON (Bak et al. 2000), WSolids (Eichele and Wasylishen 2001), and DMFit (Massiot et al. 2002) and would certainly recommend them all for the different specific problems they are tailored for. For spin-1/2 nuclei in disordered samples such as sol–gels, the lineshape is usually closely related to a Gaussian. As shown in Fig. 4, the sharp features of the second-order quadrupolar lineshapes are usually smoothed by distributions in the quadrupolar parameters. Then to characterize the interactions as well as the average values, distributions of the interaction parameters are needed. It has been shown that a Czjzek distribution or a modified version (Le Caer et al. 2010) provides a way to describe the disorder present in the material. Analyzing disordered inorganic materials in recent years has thereby greatly improved for quadrupolar nuclei through a combination of data from the 1D spectra (especially if constrained at several magnetic fields) and the 2D contours of MQ data. First-Principles Calculations of NMR Parameters. The benefits of combining solid-state NMR and first-principles quantum mechanical calculations for structural elucidation are widely recognized. In recent years the increasing availability of firstprinciples quantum calculations of the NMR parameters from the structure has provided an additional dimension to NMR as a probe technique of the intricacies of the atomic-scale structure (Charpentier 2011; Bonhomme et al. 2012). A range of approaches for the calculation of the NMR parameters have been advanced, but ab initio density functional theory (DFT) calculations using the GIPAW approach has been implemented as the CASTEP code (see primary references in the two reviews above). Reports of calculations of chemical shielding, quadrupolar, and J-coupling parameters all now exist. The ability to calculate these parameters depends on having a suitable structural model, and this adds to the usefulness of NMR in pinning down structural detail. The structure can itself be calculated using DFT or molecular dynamics (MD) approaches. Hence, either one may use structure calculations to identify and quantify spectral features or the interaction parameters can be used as a constraint on the structure. In structurally disordered systems, the former approach is the most common way to combine calculations with the NMR data as DFT calculations of the NMR parameters are more straightforward in periodic (ordered) systems. MD calculations of a sol–gel-derived CaO–SiO2 showed the effect of OH on the silicate network connectivity and the coordination of the Ca2+ ions (Mead and Mountjoy 2006).

Atomic-Scale Structure of Gel Materials by Solid-State NMR

13

Silicon-29 This spin-1/2 nucleus has a natural abundance of 4.7 %. Despite its relatively low signal strength, the ability of the chemical shift to readily distinguish different local structural units has made 29Si NMR an ideal probe of sol–gel materials. Careful consideration needs to be given to the recycle time since 29Si can exhibit a wide range of relaxation times. When there are protons present T1 is typically a few seconds, whereas in proton-free, dense gels and glasses T1 can rise to many minutes. To use the quantitative advantage of NMR, it is necessary to ensure that all resonances are fully relaxed. In silicate-based gels, the structure contains cornersharing SiO4 units, and the number of corners (n) shared with other SiO4 units leads to the designation Qn for the central SiO4 unit. These Qn units have well-defined shift ranges. In systems where carbon is directly attached to the central silicon, there is also the possibility of SiCxO4–x, where x can have values of 0, 1, 2, 3, and 4, which gives rise to Qn, Tn, Dn, Mn, and X units, and although there is some overlap of their shift ranges, they can usually be distinguished from NMR (Fig. 5a). Although the distribution of these units provides a picture of the structure, it is often helpful to reduce this to a number giving the connectivity of the network (Dc). When there are only Qn species present, Dc is Dc ¼

Xn¼4 n¼0

n  fractionQn

(3)

Silicates The simplest gels studied by 29Si NMR are those consisting only of SiO2, in which the Qn distribution can be followed through condensation and gelation and used to determine the kinetics of these processes (Fyfe and Aroca 1995). Silica xerogels have been studied by 29Si MAS NMR (Abidi et al. 1998) along with the effect of Mn2+ ions on the 29Si relaxation time. 129Xe NMR of adsorbed xenon gas suggested that the paramagnetic ions were localized at the surface (Abidi et al. 1998). In SiO2 aerogels the local structural changes with pH have been followed, and with increasing pH more Q4 form in the gel which is attributed to fewer H+ ions being present. There was also an increase in the linewidth associated with each component that indicated the more acidic precursors are more ordered (Damrau et al. 1992). Many technologically important sol–gel materials involve the addition of second oxides to SiO2, and the properties of these materials often depend on how the second oxide is distributed in the structure. 29Si MAS NMR has been extensively applied to this problem since the Qn distribution is taken to indicate the effect of the added metal. A note of caution about interpretation of the different peaks in silicon spectra is

14

M.E. Smith and D. Holland 0

a

4 Q

0

3 T

0

2 D

0

1 M

X 20

10

0

−10

−20

−30

−40 −50

29Si

−60 −70

−80

−90

−100 −110 −120

chemical shift (ppm)

b

Q4

Q3 750⬚C

Q3 Q4

Q2 25⬚C

−70

−80

−90

−100 −110 ppm

−120

−130

−140

Fig. 5 (a) Chemical shift ranges for 29 Si from different local structural units in silicate and related materials for all different silicon functionalizations with the numbers indicating the possible variations for each unit (e.g., Q0–Q4) and (b) typical MAS spectra from a sol–gel-produced TiO2–SiO2 (18 mol% TiO2) sample showing the change of Qn distribution with heat treatment

Atomic-Scale Structure of Gel Materials by Solid-State NMR

15

required. In the initial gels, the lower Qn species are often related to the hydroxyls present (e.g., Q3(OH)) and should not be confused with genuine Q3 associated with non-bridging oxygens –Si–O– +M. Hence, in gels produced at low temperature, the Qn distribution usually is dominated by the residual groups present and not the metal distribution. It is only after quite high temperature heat treatment (e.g., 750  C) that the effects of metals begin to dominate (Fig. 5b). In comparisons of melt-quenched and sol–gel glasses, the possibility of residual protons, even at quite high temperatures, can explain differences in structure. A system that has attracted significant attention in recent years is CaO–SiO2 because of its significant potential as a biomaterial and sol–gel approaches to formation have been investigated in detail. Ca(NO3)2 is a common calcium precursor in such sol–gel materials. The usual formation process is to age (60  C), dry (120/130  C), and then stabilize (500–800  C). Interestingly 29Si MAS NMR data shows that Dc initially increases and was much higher than the composition predicted. In the 5–8 nm secondary silicate particles formed during this process at

300  C, except the broad band at 400–800 cm
1, whose intensity increases with the temperature of the heat treatment performed (Alam and Cameron 2002). After a heat treatment at 500  C, there is the formation of a well-defined band around 434 cm
1, assigned to the Ti–O–Ti stretching vibration in the anatase phase, which tends to increase as the temperature increases up to 600  C. For T > 700  C, the band at 434 cm
1 decreases and eventually disappears, for T > 800  C. In the meantime, a new band at 485 cm
1, ascribed to Ti–O–Ti stretching vibrations in the rutile phase, appears (Alam and Cameron 2002; Djaoued et al. 2002b). Despite the fact that all the studied films start to crystallize at 400  C, Djaoued et al. (2002b) found a significant influence of complexing agents, present in the

26

R.M. Almeida and A.C. Marques

coating solution, on the phase transition temperature. For instance, acetylacetone (AcAc) and its mixture with acetic acid were found to stabilize the anatase phase, even up to temperatures as high as 1000  C. Furthermore, Djaoued et al. (2002a) reported a low-temperature method (T  100  C) which, by removal of OH groups, allows a rearrangement of the Ti–O network and promotes the crystallization of titania at very low temperature. This method consists of the deposition of the films, followed by a hot-water treatment for selected times. By using IR absorption spectroscopy, it was possible to study the evolution of the bands assigned to vOH, δOH, Ti–OH and Ti–O, before and after the hot-water treatment. They showed that this treatment promotes a decrease of the intensity of the vOH band and the early appearance of the Ti–O band around 453 cm
1, even for temperatures as low as 100  C. IR absorption spectroscopy has also been used with the aim of studying the in vitro bioactivity of SG-derived TiO2 films deposited by dip coating (Jokinen et al. 1998). The IR spectra of the TiO2 films showed an increase of the absorption peaks related to vibrational modes of the carbonate ion (1650–1300 cm
1, 873 cm
1) and phosphate ion (962 cm
1, 556–597 cm
1, 1096 cm
1, 1112 cm–1), after their immersion in a simulated body fluid (SBF). This indicates the existence of calcium phosphate phases and, therefore the formation of bone-like hydroxyapatite, with the immersion of the TiO2 films in the SBF, evidencing their in vitro bioactivity. The processing of mesoporous TiO2 films by EISA has been studied by (in situ) time-resolved IR spectroscopy in order to follow two of the main stages of selfassembly: evaporation and post-deposition drying. The nanostructured TiO2 cast films have been prepared by using a triblock copolymer (Pluronic F127) and TiCl4 as the titania precursor, subjected to a heat treatment up to 165  C. The evaporation phenomena can be directly observed by FTIR because there is no overlap of the IR signals from the Ti precursor and surfactant. After drying at 165  C, there was evidence from the IR spectra for the complete removal of water and for triblock copolymer crystallization as soon as the water evaporates, which was evidenced by transformation of the typical broad band within the range of 1200–1000 cm
1 into the triplet signature of the Pluronic F127 crystalline phase (Fig. 16). Crystalline micelles are therefore detected within the TiO2 mesopores. Such crystallization was found to be reversible when cooling to room temperature (Innocenzi et al. 2010). Alumina is a well-known high-temperature insulating material, and therefore, it is of value the use of FTIR in the study of the sequence of transformations which occur during heat treatment of the gel material, at different temperatures. On the other hand, the low-temperature phases of alumina are important in catalysis, due to their high specific surface areas and the large number of defects in their crystalline structure (Wang et al. 1999). There are several Al2O3 crystalline phases (η, γ, δ, κ, θ, and α), but since the transition temperature range between them is sharp, most of them appear always mixed. The transformation between the different Al2O3 phases strongly depends on the precursors and the heat treatment performed for their stabilization. Wang et al. reported the study of the low-temperature phases of alumina, in particular the evolution of hydroxyl species with temperature, measured by FTIR spectroscopy. They found that, for the fresh gel sample dried at 70  C, there

Characterization of Sol–Gel Materials by Infrared Spectroscopy

27

Fig. 16 Time-resolved IR spectra of a titania mesostructured film as a function of drying temperature. The spectra have been recorded during in situ measurements from 25 up to 165  C, with steps of 10  C (Reprinted with permission from Innocenzi P., Malfatti L., Kidchob T., Grosso D. Controlling the processing of mesoporous titania films by in-situ FTIR spectroscopy: getting crystalline micelles into the mesopores, J. Phys. Chem. C 2010; 114: 10806–10811. Copyright 2010 American Chemical Society)

was an intense and wide absorption band at 3455 cm
1, assigned to OH groups on the γ-AlO(OH) surface. With the increase of the heat treatment temperature up to 800  C, this band tends to decrease in intensity, due to the removal of water molecules and the boehmite (AlO(OH)) transformation into γ-Al2O3. At T = 800  C, OH bands still remain in the spectra, which shows the high capacity of the low-temperature alumina phase for retaining hydroxyls in its structure, indicating a high specific surface area, an important property in catalysis. O–Al–O bonds also contribute significantly to the IR spectra. Stangar et al. (2002) reported the IR absorption spectra of alumina xerogel powders (in KBr pellets), heat treated at temperatures ranging from 450–550  C, which exhibited two bands, at 640 and 850 cm
1, attributed to vibrational modes of O–Al–O groups. At 600  C, the splitting of these bands decreased (670 and 780 cm
1), and both bands tend to degenerate into a single broad band at 710 cm
1, when the heat treatment temperature was above 650  C. SG-derived ceria–titania films are of importance due to their application as antireflective coatings and electrochromic layers and in self-cleaning glasses. In particular, thermal treatment of ceria–titania films in air at high temperature up to 700  C could lead to the formation of cerium titanate phases such as CeTi2O6. The combination of far IR spectroscopy and X-ray diffraction yields a powerful tool to investigate the crystallization process and identify the phases formed as a function of temperature and composition (Kidchob et al. 2009). However, it should be stressed that a far IR absorption band peaking around 250 cm
1 starts to be observed after annealing at 300  C, increasing its intensity for higher temperatures, which suggests

28

R.M. Almeida and A.C. Marques

that small crystallites of ceria and cerium titanate, which are not yet detectable by X-ray diffraction, have already formed at these lower temperatures, revealing the sensitivity of the FTIR technique for detecting early stage crystallization phenomena. Information on SG processing and characterization of non-oxides is very scarce. An example of reported IR absorption spectra of non-oxide SG materials is for sulfides, namely, GeS2, with characteristic Ge–S IR peaks at 375 and 430 cm
1, assigned to asymmetric stretching vibrations of GeS4 and S3Ge–S–GeS3 units (Frumarová et al. 1999). The synthesis of GeS2 without contamination by oxide (GeO2) is not trivial. Middle and far IR absorption spectra were reported and discussed by Martins et al. (1999), in order to detect the presence of GeO2 in GeS2-based films, after using different Ge precursors, such as germanium tetrachloride (GeCl4) and germanium tetraethoxide and different sources of S, such as thiourea, thioacetamide, and hydrogen sulfide (H2S). For films prepared by the reaction of thioacetamide and germanium tetrachloride, dissolved in ethanol, the absorption bands of Ge–O bonds at 870 and 560 cm
1 were not detected, and therefore, only Ge–S bonds were found in the IR spectrum (Martins et al. 1999). Xu and Almeida (2000) reported a successful preparation of GeSx SG films (with a content of GeO2 of only 6 %), using GeCl4 and H2S as the main precursors and toluene as the solvent. In this work, the concentration of Ge–O bonds in the sulfide films was estimated based on Beer’s law: A ¼ ebc

(10)

where A is the absorbance, e is the molar absorptivity, b is the film thickness, and c is the concentration of absorbing species, in this case Ge–O bonds. The values found, for the concentration of Ge–O bonds, were 100 %, for a GeO2 film and 6 %, for a GeSx film.

Conclusions The IR spectra of SG materials contain considerable information about their composition, structure, and properties. Therefore, IR spectroscopy has been widely applied for silicate materials, but also for other oxides, as well as some non-oxide materials. Some examples of the rich information provided by IR spectroscopy, described in this chapter, include the structural evolution during the different stages of SG material processing, including complex phenomena occurring during EISA, the semiquantitative detection of OH species and porosity, the analysis of structural homogeneity, phase separation and densification degree, and the study of structural differences between thin films, thicker multilayer films, and bulk SG samples, as well as between SG-derived purely inorganic and hybrid materials. In addition, in situ and time-resolved IR spectroscopy analysis enhances the ability to follow the reactions and physical phenomena occurring at the various stages of SG processing.

Characterization of Sol–Gel Materials by Infrared Spectroscopy

29

References Alam MJ, Cameron DC. Preparation and characterization of TiO2 thin films by sol–gel method. J Sol–gel Sci Technol. 2002;25:137–45. Almeida RM. Vibrational spectroscopy studies of halide glass structure. In: Almeida RM, editor. Halide glasses for infrared fiberoptics. Dordrecht: Martinus Nijhoff; 1987. Almeida RM. Vibrational spectroscopy of glasses. J Non-Cryst Solids. 1988;106:347–58. Almeida RM. Detection of LO modes in glass by infrared reflection spectroscopy at oblique incidence. Phys Rev B. 1992;45:161–70. Almeida RM. Sol–gel silica films on silicon substrates. Int J Optoelectron. 1994;9:135–42. Almeida RM. Spectroscopy and structure of sol–gel systems. J Sol–gel Sci Technol. 1998;13:51–9. Almeida RM, Pantano CG. Structural investigation of silica gel films by infrared spectroscopy. J Appl Phys. 1990a;68:4225–32. Almeida RM, Pantano CG. Vibrational spectra and structure of silica gel films spun on c-Si substrates. SPIE-Sol–gel Optics. 1990b;1328:329–37. Almeida RM, Guiton TA, Pantano CG. Detection of LO mode in v-SiO2 by infrared diffuse reflectance spectroscopy. J Non-Cryst Solids. 1990a;119:238–41. Almeida RM, Guiton TA, Pantano CG. Characterization of silica gels by infrared reflection spectroscopy. J Non-Cryst Solids. 1990b;121:193–7. Almeida RM, Vasconcelos HC, Ilharco LM. Relationship between infrared absorption and porosity in silica-based sol–gel films. SPIE Sol–gel Optics III. 1994;2288:678–87. Almeida RM, Du XM, Barbier D, Orignac X. Er3+-doped multicomponent silicate glass planar waveguides prepared by sol–gel processing. J Sol–gel Sci Technol. 1999;14:209–16. Almeida RM, Morais PJ, Marques AC. Planar waveguides for integrated optics prepared by sol–gel methods. Philos Mag B. 2002;82:707–19. Available at http://www.tandf.co.uk/journals/titles/ 14786435.html Almeida RM, Marques AC, Pelli S, Righini GC, Chiasera A, Matarelli M, Montagna M, Tosello C, Gonçalves RR, Portales H, Chaussedent S, Ferrari M, Zampedri L. Spectroscopic assessment of silica–titania and silica–hafnia planar waveguides. Philos Mag A. 2004;84:1659–64. Atik M, Kha C, Neto P, Avaca L, Aegerter M, Zarzycki J. Protection of 316L stainless steel by zirconia sol–gel coatings in 15% H2SO4 solutions. J Mater Sci Lett. 1995;14:178–81. Brinker CJ, Scherer GW. Sol–gel science: the physics and chemistry of sol–gel processing. Boston: Academic Press; 1990. Burgos M, Langlet M. The sol–gel transformation of TIPT coatings: a FTIR study. Thin Solid Films. 1999;349:19–23. Chalmers JM, Dent G. Industrial analysis with vibrational spectroscopy. Cambridge: Royal Society Of Chemistry; 1997. Colomban P, Bruneton E. Influence of hydrolysis conditions on crystallization, phase transitions and sintering of zirconia gels prepared by alkoxide hydrolysis. J Non-Cryst Solids. 1992;147&148:201–5. Dawnay EJC, Fardad MA, Green M, Horowitz F, Yeatman EM, Almeida RM, Vasconcelos HC, Guglielmi M, Martucci A. Control and characterization of microstructure in sol–gel films for optical device applications. In Vincenzini P, editor. Advanced materials in optics, electro-optics and communication technologies. Faenza (Italy): Techna; 1995. Djaoued Y, Badilescu S, Ashrit P, Bersani D, Lottici P, Bruning R. Low temperature sol–gel preparation of nanocrystalline TiO2 thin films. J Sol–gel Sci Technol. 2002a;24:247–54. Djaoued Y, Badilescu S, Ashrit P, Bersani D, Lottici P, Robichaud J. Study of anatase to rutile phase transformation in nanocrystalline titania films. J Sol–gel Sci Technol. 2002b;24:255–64. Du XM, Almeida RM. Effects of thermal treatment on the structure and properties of SiO2-TiO2 gel films on silicon substrates. J Sol–gel Sci Technol. 1997;8:377–80. Fardad MA, Yeatman EM, Dawnay EJC, Green M, Horowitz F. Effects of H2O on structure of acidcatalysed SiO2 sol–gel films. J Non-Cryst Solids. 1995;183:260–7.

30

R.M. Almeida and A.C. Marques

Fidalgo A, Ilharco LM. The defect structure of sol–gel-derived silica/polytetrahydrofuran hybrid films by FTIR. J Non-Cryst Solids. 2001;283:144–54. Frumarová B, Nemec P, Frumar M, Oswald J, Vlcek M. Synthesis and optical properties of Ge–SbS: PrCl3 system glasses. J Non-Cryst Solids. 1999;256&257:266–70. Galeener FL. Band limits and the vibrational spectra of tetrahedral glasses. Phys Rev B. 1979;19:4292–7. Gallardo J, Durán A, Martino DD, Almeida RM. Structure of inorganic and hybrid SiO2 sol–gel coatings studied by variable incidence infrared spectroscopy. J Non-Cryst Solids. 2002;298:219–25. Gnado J, Dhamelincourt P, Pélégris C, Traisnel M, Mayot ALM. Raman spectra of oligomeric species obtained by tetraethoxysilane hydrolysis-polycondensation process. J Non-Cryst Solids. 1996;208:247–58. Guglielmi M, Martuccci A, Almeida RM, Vasconcelos HC, Yeatman EM, Dawnay EJC, Fardad MA. Spinning deposition of silica and silica–titania optical coatings: A round robin test. J Mater Res. 1998;13:731–8. Hu SM. Infrared absorption spectra of SiO2 precipitates of various shapes in silicon: calculated and experimental. J Appl Phys. 1980;51:5945–8. Innocenzi P. Infrared spectroscopy of sol–gel derived silica-based films: a spectra-microstructure overview. J Non-Cryst Solids. 2003;316:309–19. Innocenzi P, Abdirashid MO, Guglielmi M. Structure and properties of sol–gel coatings from methyltriethoxysilane and tetraethoxysilane. J Sol–gel Sci Technol. 1994;3:47–55. Innocenzi P, Malfatti L, Kidchob T, Grosso D. Controlling the processing of mesoporous titania films by in-situ FTIR spectroscopy: getting crystalline micelles into the mesopores. J Phys Chem C. 2010;114:10806–11. Innocenzi P, Figus C, Takahashi M, Piccinini M, Malfatti L. Structural evolution during evaporation of a 3-glycidoxypropyltrimethoxysilane film studied in situ by time resolved infrared spectroscopy. J Phys Chem A. 2011;115:10438–44. Izumi K, Murakami M, Deguchi T, Morita A. Zirconia coating on stainless steel sheets from organozirconium compounds. J Am Ceram Soc. 1989;72:1465–8. Jokinen M, Patsi M, Rahiala H, Peltola T, Ritala M, Rosenholm J. Influence of sol and surface properties on in vitro bioactivity of sol–gel-derived TiO2 and TiO2–SiO2 films deposited by dip coating method. Biomed Mater Res. 1998;42:295–302. Kasgoz A, Yoshimura K, Misono T, Abe Y. Preparation and properties of SiO2-TiO2 thin films from silicic acid and titanium tetrachloride. J Sol–gel Sci Technol. 1994;1:185–91. Kidchob T, Malfatti L, Marongiu D, Enzo S. Innocenzi formation of cerium titanate, CeTi2O6, in sol–gel films studied by XRD and Far infrared spectroscopy. J Sol–gel Sci Technol. 2009;52:356–61. Marcelli A, Innocenzi P, Malfatti L, Newton M, Rau J, Ritter E, Schade U, Xu W. IR and X-ray time-resolved simultaneous experiments: an opportunity to investigate the dynamics of complex systems and non-equilibrium phenomena using third-generation synchrotron radiation sources. J Synchrotron Radiat. 2012;19:892–904. Marques AC, Almeida RM, Chiasera A, Ferrari M. Reversible photoluminescence quenching in Er3+-doped silica–titania planar waveguides prepared by sol–gel. J Non-Cryst Solids. 2003;322:272–7. Martins O, Almeida RM. Sintering anomaly in silica-titania sol–gel films. J Sol–gel Sci Technol. 2000;19:651–5. Martins O, Xu J, Almeida RM. Sol–gel processing of germanium sulfide based films. J Non-Cryst Solids. 1999;256&257:25–30. Matos MC, Ilharco LM, Almeida RM. The evolution of TEOS to silica gel and glass by vibrational spectroscopy. J Non-Cryst Solids. 1992;147&148:232–7. Matsuda A, Kogure T, Matsuno Y, Katayama S, Tsuno T, Tohge N, Minami T. Structural changes of sol–gel derived TiO2-SiO2 coatings in a environment of high temperature and high humidity. J Am Ceram Soc. 1993;76:2899–903.

Characterization of Sol–Gel Materials by Infrared Spectroscopy

31

Mukherjee SP. Ultrapure glasses from sol–gel processes. In: Klein LC, editor. Sol–gel technology for thin films, fibres, preforms, electronics and specialty shapes. Park Ridge: N.J. Noyes; 1988. Nakamoto K. Infrared and Raman spectra of inorganic and coordination compounds. New York: John Wiley & Sons, Inc; 1978. Neumayer DA, Carrier E. Materials characterization of ZrO2–SiO2 and HfO2–SiO2 binary oxides deposited by chemical solution deposition. J Appl Phys. 2001;90:1801–8. Niznansky D, Rehspringer JL. Infrared study of SiO2 sol to gel evolution and gel aging. J Non-Cryst Solids. 1995;180:191–6. Okasaka K, Nasu H, Kamiya K. Investigation of coordination state of Zr4+ ions in the sol–gelderived ZrO2-SiO2 glasses by EXAFS. J Non-Cryst Solids. 1991;136:103–10. Orignac X, Almeida RM. Silica-based sol–gel optical waveguides on silicon. IEE Proc Optoelectron. 1996;143:287–92. Popescu R, Zaharescu M, Vasilescu A, Catana G, Manaila R. Intermediate-range order in basecatalysed sol–gel silica. J Non-Cryst Solids. 1995;193:137–9. Primeau N, Vautey C, Langlet M. The effect of thermal annealing on aerosol-gel deposited SiO2 films: a FTIR deconvolution study. Thin Solid Films. 1997;310:47–56. Ricol S, Vernaz E, Barboux P. Synthesis of gels in the system Na2O–ZrO2–SiO2. J Sol–gel Sci Technol. 1997;8:229–33. Saha SK, Pramanik P. Aqueous sol–gel synthesis of powders in the ZrO2-SiO2 system using zirconium formate and tetraethoxisilane. J Non-Cryst Solids. 1993;159:31–7. Salvado IM, Serna CJ, Navarro JM. ZrO2–SiO2 materials prepared by sol–gel. J Non-Cryst Solids. 1988;100:330–8. Seco AM, Goncalves MC, Almeida RM. Densification of hybrid silica–titania sol–gel films studied by ellipsometry and FTIR. Mater Sci Eng B: Solid State Mater Adv Technol. 2000;76:193–9. Sen PN, Thorpe MF. Phonons in AX2 glasses: from molecular to band-like modes. Phys Rev. 1977;15:4030–8. Stangar U, Orel B, Krajnc M, Korosec R, Bukovec P. Sol–gel derived thin ceramic COA12O4 coatings for optical applications. MTAEC. 2002;36:387–93. Stuart B, George B, McIntyre P. Modern Infrared Spectroscopy. Chichester: John Wiley & Sons, Inc; 1996. Thomas IM. Multicomponent glasses from the sol–gel process. In: Klein LC, editor. Sol–gel technology for thin films, fibres, preforms, electronics and specialty shapes. Park Ridge: Noyes; 1988. van der Vis MGM, Konings RJM, Oskam A, Snoeck TL. The vibrational spectra of gaseous and liquid tetraethoxysilane. J Mol Struct. 1992;274:47–57. Wang JA, Biokhimi X, Morales A, Novaro O, López T, Gómez R. Aluminium local environment and defects in the crystalline structure of sol–gel alumina catalyst. J Phys Chem B. 1999;103:299–303. Weinsenbach L, Zelinski BJ, O’Kelly J, Roncone R, Burke J. The influence of processing variables on the optical properties of SiO2-TiO2 planar waveguides. SPIE Submolec Glass Chem Phys. 1991;1591:50–8. Xu J, Almeida RM. Preparation and characterization of germanium sulfide based sol–gel planar waveguides. J Sol–gel Sci Technol. 2000;19:243–8. Yeatman EM, Ahmad MM, McCarthy O, Martucci A, Guglielmi M. Sol–gel fabrication of rareearth doped photonic components. J Sol–gel Sci Technol. 2000;19:231–6.

Characterization of Sol–Gel Materials by Raman and Brillouin Spectroscopies Maurizio Montagna

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inelastic Light Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raman and Brillouin Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Vibrational Dynamics of Aerogels: Fractons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Densification of Silica Xerogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characterization by Waveguided Brillouin Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raman Spectroscopy of Glass Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Raman Spectroscopy of Nanocrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2 6 6 14 18 21 22 27 27

Abstract

The use of Raman and Brillouin spectroscopies in sol–gel-derived materials is reviewed. It covers a quite vast domain of investigation, from the basic glass science to the characterization of materials produced for many different applications. The theory of inelastic light scattering is briefly presented and the basic physical mechanisms involved in Raman and Brillouin spectroscopies are discussed. The instrumentation for measurements by visible, ultraviolet, and X-ray excitation is described. The vibrational dynamics in low-density aerogel is discussed in terms of a random fractal model, where the low-frequency acoustic vibrations are distorted phonon-like extended propagating modes, whereas highfrequency modes can be spatially localized. As a function of the vibrational frequency, the phonon wavelength and mean free path depend on the size of the porosity. The dependence of the sound velocity and attenuation on the densities is M. Montagna (*) Disordered System Laboratory, Department of Physics, Università degli Studi di Trento, Trento, Italy e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_34-1

1

2

M. Montagna

measured by Brillouin spectroscopy. Attenuation in low-density sol–gel-derived solids is nearly temperature independent and is due to structural disorder instead than to anharmonicity, as it is in crystals and compact glasses. It is shown that Raman and Brillouin spectroscopy can be used to follow the different steps of densifications from the wet-gel to the compact glass during thermal annealing. Waveguided spectroscopies for dip- and spin-coated films are described. Low-frequency Raman spectra from the localized acoustic vibration in nanocrystals are presented in sol–gel-derived glass ceramics.

Introduction Raman spectroscopy is a powerful characterization technique in materials science, because the vibrational dynamics gives rich information on the molecular structure. Furthermore, it is a nondestructive technique of low cost and simple analysis of the data. It is widely used for the study of inorganic, organic, and hybrid materials processed by sol–gel technology in all the steps of the process: from the starting solution through hydrolysis and polycondensation, gelling, drying of water and solvents, and high-temperature annealing to form glass and ceramics. About 100 papers per year on studies of sol–gel materials by Raman scattering are now appearing. Bulk samples but also films are characterized by the Raman technique. In the latter case, the exciting laser light is usually coupled to the waveguide by butt coupling, prism coupling, or writing a grating in the film. Brillouin scattering, even if less extensively used as a characterizing technique, gives important structural information on materials produced by sol–gel, since the sound velocity strongly depends on the porosity. By prism coupling, Brillouin scattering in planar waveguides can be measured. Inelastic light scattering has been also extensively used to study the vibrational properties of fractal systems: silica aerogels, with their porosity extending on a wide range of sizes, are very nice examples of fractal systems. The peculiar vibrational dynamics of these systems, with spatially localized vibrational modes, is at the origin of the extremely low-measured thermal conductivity. In this chapter, the application of inelastic light scattering to the study of the vibrational dynamics of sol–gel materials in bulk and planar waveguide forms will be described.

Inelastic Light Scattering Raman and Brillouin scattering are two photon processes. The electromagnetic field exchanges an energy ℏω = hυin
hυout and a wavevector q = kin
kout with the system under study. υin and υout are the frequencies of the incoming and scattered photons, respectively. kin and kout are the wavevectors of the incoming and outgoing

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

3

photon, respectively, with k = 2π/λ = 2πn/λ0, where n is the refractive index and λ and λ0 are the wavelengths in the scattering system and in vacuum, respectively. The units which form the system, atoms, ions, and bonds are polarized by the exciting field and by the fields produced by the other dipoles, through a dipole-induced dipole mechanism (DID). These units irradiate fields at the exciting frequency (elastic or Rayleigh scattering), but also inelastic scattering is produced by the thermal fluctuations of the system. The scattered intensity is proportional to the fourth power of the frequency and to the square of the sum of the electric fields irradiated by all the polarized units. In a quantum mechanical approach to the polarizability of the basic units, the inelastic scattering process is the destruction of a photon, due to an electronic transition to an excited level and the creation of a photon of different energy, assisted by the creation or destruction of one or more vibrational quanta. If the system is transparent to the exciting and scattered light, a virtual transition to the intermediate electronic level occurs (normal Raman scattering). On the contrary, when the photon energy is resonant or near resonance with an electronic transition, one has resonant Raman scattering. The intensity in the various scattering configurations is given by the Fourier transform of the correlation functions (Benassi et al. 1993): Gαβγδ ðtÞ ¼

XD   E Πiαβ ðtÞΠjγδ ð0Þexp iq  ri ðtÞ
rj ð0Þ ;

(1)

ij

where α, β, γ, and δ are Cartesian indices and Πiαβ are the instantaneous atomic polarizability tensor components of the ith atom, at time t and position ri(t). Equation 1 accounts for the elastic (Rayleigh) and inelastic (Brillouin and Raman) light scattering. After expansion in the atomic displacements, ui(t), from the equilibrium positions, Ri, ri ðtÞ ¼ Ri þ ui ðtÞ;

(2)

the Brillouin and one-phonon Raman contributions are given by the linear terms in the displacements u (Benassi et al. 1993): X i

   Πiαβ ðtÞeiqri ðtÞ  

¼ l ph

XX i

e

iqRi

μ



iqμ Πiαβ

 eq

þ



Qiαβ, μ



eq

uiμ ðtÞ;

(3)

where α and β are the direction of polarization of the incoming and outgoing photons, respectively, and (Πiαβ)eq is the polarizability of the ith atom, for all atoms at the equilibrium position, and   X @Πiαβ ¼ Qiαβ, μ eq @uiμ l

! exp½iq  ðRl
Ri Þ: eq

(4)

4

M. Montagna

The first term in the sum of Eq. 3 describes the polarizability modulation on the system by a vibrational mode that displaces the masses from the equilibrium positions. The second term in the sum in Eq. 3 accounts for the polarizability modulation of the ith atom caused by the displacements of the surrounding atoms. For molecules with a small size, a, (qa  1), the fields scattered by all the polarizable units are in phase. In this case a single quantity, the polarizability Пαβ of the molecule determines the Raman spectra. The symmetry of the molecule and of the Пαβ tensor is fundamental for the calculation of the symmetry of the vibrational modes and of the intensity of the relative Raman and IR absorption bands (Long 1977). In harmonic crystals, the vibrations are phonons, i.e., plane waves involving the motion of all atoms in the sample. Phonons are described by a frequency ωph and a wavevector kph. The light is scattered by the whole illuminated volume seen by the detector, and fields irradiated by all polarizable units interfere. As a consequence, one-phonon scattering processes (first-order Raman scattering) are subjected to the selection rules: ωph ¼ 2π ðυin
υout Þ,

kph ¼ q;

(5)

where the plus and minus signs refer to Stokes and anti-Stokes scattering, respectively. In light scattering experiments, the exchanged q is much smaller than phonon wavevectors at the boundary of the Brillouin zone. Since the dispersion curves of optical phonons are flat at the zone center, Raman scattering is q independent, and q = 0 can be taken in Eq. 3. Therefore the analysis of the Raman scattering by optical phonon in crystals is based on the structure and symmetry of the unit cell and on the symmetry of its Παβ, in a way very similar to that of molecules. On the contrary, Brillouin scattering of acoustic phonons is q dependent. By combining the two above selection rules, one obtains ωLph, T ¼ 2kin υL, T sin ðϑ=2Þ;

(6)

where ϑ is the scattering angle, within the sample, and υL and υT are the longitudinal and the (two different, in general) transverse sound velocities, having noted that kin ffi kout. In harmonic glasses, the vibrational normal modes are no longer phonons, due to disorder and lack of the translational symmetry of crystals. However, the Brillouin spectra of glasses are very similar to those of crystals, with sharp peaks, indicating that the low-frequency acoustical modes are nearly plane waves. The Brillouin peaks are found at about 10–40 GHz with a line width, which is a measure of the phonon mean free path, of the order of 100 MHz in simple glasses at room temperature. The temperature-dependent dynamical effects, as anharmonicity and glass relaxation, are comparable to the structural effects, which produce scattering of the plane acoustic waves by the disordered structure. In X-ray inelastic scattering experiments, where the Brillouin peak is in the THz range, the line width is an important fraction of the

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

5

peak frequency. Furthermore, it is nearly temperature independent, indicating that acoustic waves in the THz range are strongly scattered by the glass-disordered structure. Aerogels and most sol–gel-derived materials, having a residual porosity, are strong scatterers of the acoustic modes. The Brillouin line widths in these systems are larger than in the homogeneous bulk glasses and are nearly temperature independent. This subject is discussed in “Vibrational Dynamics of Aerogels: Fractons.” The Raman selection rule on the wavevectors in Eq. 5, valid for crystalline systems, does not hold for glasses, since the vibrational modes of glasses are different from planar waves. Instead of the sharp peaks of crystals, due to the optical phonons, broad bands appear in the Raman spectra of glasses. All vibrational modes are active, with different extents, in Raman scattering. Therefore, it is common to write the Raman intensity in VV-polarized (α = β) and HV-polarized (α 6¼ β) as (Shuker and Gammon 1970) I αβ ðωÞ ¼

nðω, T Þ þ 1 Cαβ ðωÞρðωÞ; ω

(7)

where n(ω, T ) is the Bose–Einstein population factor, C(ω) is the Raman coupling coefficient, and ρ(ω) is the density of vibrational states. For the high-frequency modes, the analogous of the optical modes of the crystals, the Raman activity is mainly related to the symmetry, as for molecules. Also in glasses, it is possible to deduce the symmetry of optical modes from the activity in Raman and IR absorption spectra. In particular, an important quantity is the depolarization ratio, i.e., the ratio of intensity of the HV- and VV-polarized spectra. In any case, there are not strong selection rules, due to disorder. The activity of the low-frequency modes, of acoustic-like nature or mixed acoustical and optical nature, has a more subtle origin, since it is intrinsically related to the presence of electrical and mechanical disorder (Martin and Brenig 1974). It is the disordered structure that does not allow a complete destructive interference of the scattered fields, as it occurs in crystals, where the acoustical phonons do not contribute to the Raman scattering (Benassi et al. 1995). The electrical disorder is caused by the disordered space distribution of the polarizability, as in the case of heavy ions in a silicate glass (Benassi et al. 1991). Mechanical disorder is the deviation of the vibrational mode patterns from the plane wave shape of phonons (Martin and Brenig 1974). In particular, a depolarized broad peak is present in the Raman spectra of all glasses, the boson peak, at frequencies in the range 20–60 cm
1. It corresponds to an excess in the density of states with respect to the ω2 Debye value, as measured in inelastic neutron scattering (INS) experiments. The nature of the vibrational modes at the boson peak frequencies has been extensively debated and many different models have been proposed (Fontana and Viliani 1998). Finally, of particular interest for the study of sol–gel-derived materials is the low-frequency Raman scattering of the acoustic vibrations of nanoparticles embedded in a glass matrix. This subject is discussed in “Raman Spectroscopy of Nanocrystals.”

6

M. Montagna

Raman and Brillouin Instrumentation The evolution of Raman instrumentation has been recently reviewed (Lewis and Edwards Howell 2001), from the first measurements by C. V. Raman, when the spectra were excited with a mercury lamp and recorded with a small prism spectroscope equipped with a photographic plate (Raman 1928), to the new high-resolution microscopes, in Raman scanning near-field optical microscopy (Adar 2001). This review includes Raman microscopy (Baldwin et al. 2001), Raman imaging (Treado Patrick and Nelson Matthew 2001), the adaptation of Raman spectrometry to industrial environment (Slater Joseph et al. 2001), Raman spectroscopy of catalysts (Wachs 2001), and process of Raman spectroscopy (Lewis 2001). Therefore, we restrict the discussion here to Brillouin equipments. Brillouin spectrometry requires setups with high resolution, contrast, and luminosity, since Brillouin lines are very close to the laser excitation, with typical shifts of 10–30 GHz, and very weak, about 10
10–10
12 times weaker than the laser line. On the other hand, a spectral range limited to a few GHz is required. Therefore, Brillouin spectrometers are usually multipass Fabry–Perot (FP) interferometers. Multipass tandem interferometers, based on two-plane FP (Sandercock 1970, 1976, 1978) or a plane and a confocal FP (Sussner and Vacher 1979; Vacher et al. 1980) in sequence, are used to fulfill the request of high contrast and resolution together with a sufficient spectral range. A light-modulation technique at microwave frequencies by a LiNbO3 crystal is usually employed for accurate measurements (Δυ/υ  10
4) of Brillouin line shifts (Sussner and Vacher 1979). A double spectrometer with a focal length of 2 m, and a resolution of about 0.015 cm
1 at 514.5 nm in double pass, is also used. It operates at the 11th order of high-quality ruled gratings with 300 lines/mm. It allows one to measure together Brillouin and low-frequency Raman spectra (Mazzacurati et al. 1988). A new spectrometer for high-resolution and high-contrast scattering spectroscopy in the ultraviolet has been constructed in L’Aquila, Italy. The instrument has two coupled 4 m-focal grating monochromators, with large-dimension gratings (204 408 mm2) ruled with 31.6 grooves/mm, working at the 230th order at λ = 266 nm and at the 115th order at λ = 532 nm, with a spectral range of 160 cm
1 (Caponi et al. 2004). Unfortunately, the L’Aquila earthquake in 2009 damaged it seriously. Brillouin spectroscopy in the THz frequency range is now possible by inelastic X-ray scattering at the very-high-energy-resolution IXS beamlines at the European Synchrotron Radiation Facility, Grenoble, France (Ruocco and Sette 2001; Scopigno et al. 2002).

Vibrational Dynamics of Aerogels: Fractons Silica aerogels have been considered as a model system for the study of the vibrational dynamics of random fractals. This, in turn, provided a deep knowledge of the structure of silica aerogels in terms of the preparation process. Fractals are selfsimilar systems described by a fractional space dimension D (Mandelbrot 1982).

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

7

Scaling is the basic concept for dealing with disordered media having self-similar symmetry (Alexander and Orbach 1982; Rammal and Toulouse 1983). The mass density of a fractal scale is M(R) / RD. This means that, starting from an arbitrary point on the fractal, the mass included in a sphere of radius R around the point increases as RD, instead of as Rd, d being the Euclidean dimension (d = 3 for the three-dimensional space). The structure of silica aerogels, with porosity on a wide range of distances, is well described by a fractal model (Vacher et al. 1988). The system is built by elementary grains of dimension a, which form a fractal structure that becomes statistically homogeneous at a crossover length, ξco, of the order of the maximum pore size. Small-angle X-ray scattering (SAXS) (Schaefer and Keefer 1986) and neutron scattering (SANS) (Vacher et al. 1989) have revealed welldeveloped fractal behavior over several orders of magnitude in length. In particular, samples with identical basic structure have been prepared with different densities and thus different crossover lengths, ξco (Vacher et al. 1988). In these systems, the acoustic vibrations are distorted phonon-like propagating modes for wavelengths λ larger than ξco and frequencies smaller than a first crossover frequency ωco1. In fact, the disordered porous structure is statistically homogeneous on these long-range scales. As in compact glasses, these phonon-like modes can be characterized by a wavevector q(q = 2π/λ), and a linear relation, ω = υL,T q, is observed between frequency and wavevector. The transverse, υT, and longitudinal, υL, sound velocities depend on the mass and bond distribution in the structure. They are smaller than those of bulk silica and decrease as the density of the aerogel decreases. The long-wavelength sound velocities in the phonon regime scale with ~ the density as υL, T / ρðD=d
1Þ=ðd
DÞ (Alexander et al. 1993). For wavelengths smaller than ξco and for the corresponding frequencies higher than ωco1, acoustic vibrations are fractons, i.e., highly disordered modes, localized on a volume of the order of λ3. It is still possible to define a (mean) wavelength λ, but a wide q distribution is present in the spatial Fourier transform of the modes (Montagna et al. 1990; Mazzacurati et al. 1992). Other two lengths useful for describing fractons are (i) the localization length LL, which describes the exponential or super-exponential decay of the envelope of the squared displacements from the center of localization of the mode (Petri and Pietronero 1992), and (ii) the scattering length LS, defined as the mean free path of a planar wave of frequency ω. LS can be obtained by measuring the width in q of the dynamical structure factor S(q, ω) in an inelastic scattering experiment, such as optical, X-ray, or neutron Brillouin scattering. In fractals, the three lengths λ, LL, LS follow the same scaling law (Alexander and Orbach 1982; Aharony et al. 1987): λ LL LS / ω
ðd =DÞ : ~

(8) ~

The fracton density of states scales with frequency as GðωÞ  ωd
1 ; where d~ is the fracton or spectral dimension. It describes the dynamics of the fractal and is different from the fractal dimension D. ωco1 and ξco depend on the density and are
ðd~=DÞ related by the relationship ξco / ωco1 :

8 8000 Intensity (Counts/0.5 sec)

Fig. 1 Backscattering Brillouin spectra for six silica aerogel densities (in kg/m3). The relative intensities, not adjusted for sample turbidity, are otherwise significant. IW is the full instrumental width at half-height. The central portion of the spectra, affected by the elastic line, was removed for clarity (Reprinted figure with permission from Courtens et al. (1987); copyright (1987) by the American Physical Society)

M. Montagna

Q = 180°

407

6000 360

1W

103

4000

330 186

228

2000

0

0

1

2 3 Frequency w/2p (GHz)

4

5

The nature of the vibrational modes of aerogels in the frequency range around ωco1 was well assessed by Brillouin scattering measurements (Courtens et al. 1987, 1988). By varying the sample density and the exchanged q vector, Brillouin scattering covers a quite broad range of acoustic behavior, from disordered phonon-like modes with qξco  1 to fractons with qξco > 1. Figure 1 shows the backscattering Brillouin spectra of samples with different densities. Figure 2 shows the spectra with different exchanged q values for a sample with a density ρ = 186 kg/m3. The samples are part of a series of unoxidized neutrally reacted gels characterized by SANS (Vacher et al. 1988). They were obtained by hydrolysis of tetramethyl orthosilicate (TMOS) dissolved in water–methanol mixtures under initially neutral conditions. After gelling and long aging, the material was hypercritically dried. The spectra, excited with the 514.5 nm argon-laser line, were measured with a six-pass tandem interferometer of about 3 cm spacing (Sandercock 1978). Figure 1 shows that the longitudinal phonon velocity υL(ρ) decreases with decreasing ρ (increasing ξco). Note that the frequency shift of the peak is not completely due to the change of the sound velocity. In fact, also the refractive index changes with the density. Its value is well approximated by an interpolation between the refractive index of bulk silica and that of vacuum: nðρÞ
1 ¼ ðnSiO2
1Þ

ρ ; ρSiO2

(9)

with nSiO2 ¼ 1:46 at 514.5 nm and ρSiO2 ¼ 2200 kg=m3 : With decreasing density, υL decreases, but the attenuation, measured by the line width of the Brillouin peak, increases. The high-density samples show a phonon-like spectrum with a line width Γ smaller than the peak frequency υP. The crossover to fracton dynamics appears as a strong increase of the attenuation. For the sample with ρ = 186 kg/m3, the fracton regime has been almost completely reached, since Γ ffi υP. The presence of a phonon–fracton crossover in the dynamics of aerogels appears even more evident in Fig. 2. At small scattering angles, Brillouin scattering from acoustic phonon is observed. As the scattering angle increases, the exchanged q and the phonon

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

Q = 180°

Intensity (Arbitrary Units)

Fig. 2 Brillouin spectra for a silica aerogel with ρ = 186 kg/ m3 at four internal scattering angles θ, corresponding to k/2π ranging from 17,900 to 40,300 cm
1, for increasing θ (Reprinted figure with permission from Courtens et al. (1987); copyright (1987) by the American Physical Society)

9

Q = 91°

Q = 67°

Q = 53° 0

3 1 2 Frequency w /2p (GHz)

4

frequency increase, approaching the crossover to the fracton regime. A detailed analysis allows one to obtain the dependence of the sound velocity, the attenuation, and the crossover frequency, on the sample density. The density of states of aerogels has been studied by low-frequency Raman spectroscopy (Boukenter et al. 1986; Tsujimi et al. 1988). As shown in Eq. 7, the Raman intensity depends on a coupling coefficient C(ω), which measures the Raman activity of each mode, preventing the possibility of a direct determination of the density of states. For fractal systems, the coupling coefficient is expected to have a scaling law CðωÞ  ωx ; since all physical quantities should scale in a self-similar system. In fact, the fractal model is confirmed by the Raman spectra (Boukenter et al. 1986; Tsujimi et al. 1988), which show a power law for the frequency dependence of the intensity. Figure 3 reports the Raman spectra of four neutrally reacted oxidized aerogels with densities between 158 and 357 kg/m3. The spectra are depolarized, i.e., the polarizations of the exciting and scattering light are crossed. The measurements were performed with a six-pass tandem Sandercock Fabry–Perot interferometer (Sandercock 1978), by exciting with the 514.5 nm line of an Ar+ laser, operating on a single mode at 250 mW. The data were taken by using different mirror

10

M. Montagna

RAMAN SUSCEPTIBILITIES (ARB. UNITS)

NEUTRALLY REACTED OXIDIZED

10

−0.39

−0.35

158

−0.36

−0.36

201

260

357 1 1

10 FREQUENCY (cm−1)

Fig. 3 The Raman susceptibilities I(ω)/n(ω) for four silica aerogel samples designated by their densities in kilograms per cubic meter. The corresponding acoustic correlation lengths ξac are 750, 480, 300, and 170 Å, in order of increasing densities (Courtens et al. 1988). The straight lines are fits with the indicated slopes, while the thin curves are guides to the eye. The different symbols correspond to the four different spacings L of the FP interferometer (Reprinted figure with permission from Tsujimi et al. (1988); copyright (1988) by the American Physical Society)

spacings (0.015 cm < L < 0.165 cm). Note that logarithmic scales are employed both for frequency (from 0.3 to 50 cm
1) and intensity, allowing a direct check of the scaling properties. The peak at about 30 cm
1 is attributed to the acoustic vibrations of the basic units, a spheroidal cluster of porous silica with a size (diameter), a, of 1.2–1.6 nm (see “Raman Spectroscopy of Nanocrystals”). The higher limit, ξco, and the related ωco1 depend on the density of the aerogel. At frequencies ω < ωco1, phonon-like acoustic vibrations are present and a steeper slope is observed in the frequency dependence of the Raman intensity. For the lightest aerogel, the linear behavior extends over 1.5 decades of frequency. Its low-frequency extension is limited by the measured frequency range, in agreement with the value ωco1  0.02 cm
1, derived from Brillouin data (Courtens et al. 1987, 1988). For the heaviest aerogel, Brillouin data gave ωco1  0.3 cm
1, a value which roughly corresponds to the onset of the fractal behavior. Similar results were obtained for base-catalyzed

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

11

aerogels, the main difference being the larger size of the building blocks, a ffi 3.2 nm, and a particle peak at about 15 cm
1. In Fig. 3, the susceptibility I(ω)/(n + 1) is reported. Therefore, from Eq. 7 and from the hypothesis that C(ω)  ωx, the observed slope y =
0.36 should be given by y ¼ x þ d~
2: The independent measurement of the fractal dimension from SANS experiments, D = 2.4 (Vacher et al. 1988), and of the spectral dimension by INS experiments, d~ ¼ 1:3  0:1 (Vacher et al. 1989, 1990; Courtens et al. 1990), allowed the determination of the scaling exponent of C(ω): x = 0.34. Different models have been proposed for the calculation of C(ω), based on the scaling properties of the local strain in fractals (Boukenter et al. 1986; Tsujimi et al. 1988; Alexander 1989; Alexander et al. 1993; Mazzacurati et al. 1992; Duval et al. 1993). Molecular dynamics simulations on model systems, n-dimensional percolators, were also performed by using different Raman scattering mechanisms, DID or bond polarizability (BP) (Montagna et al. 1990; Stoll et al. 1992; Mazzacurati et al. 1992). Also, due to the size limitations of these simulations, a definitive assessment of the scaling properties of C(ω) was not reached. The situation is particularly complicated for DID scattering mechanisms, since the vibrations and the associated strains can propagate only along the disordered connections of the fractal, but electric fields can propagate also in the free space of the porous structure (Alexander 1989; Montagna et al. 1990; Mazzacurati et al. 1992). The microscopic structure of the system and the actual scattering mechanism, DID or BP, seem to determine the shape of the Raman spectrum at least at the same extent as do the macroscopic properties described by the fractal parameters (Mazzacurati et al. 1992). Raman and Brillouin studies were extended to silica aerogels produced with different procedures, having different microstructures, connectivities, and thus different vibrational dynamics. In particular, a series of samples were prepared with different densities ρ and thus different crossover lengths ξco, but with nearly identical structures on fractal scales, i.e., for distances smaller than ξco. SANS measurements showed a fractal behavior, allowing one to obtain the fractal dimension D and the fractal range between a and ξco (Vacher et al. 1988). Observations in direct space, by electron microscopy, were in good agreement with SANS results (Rousset et al. 1990; Courtens and Vacher 1992). Furthermore, INS experiments and neutron spin echo spectroscopy allowed the measurement of the density of vibrational states and thus d~ (Schaefer et al. 1990; Conrad et al. 1990; Courtens et al. 1990; Vacher et al. 1990). A full agreement was found between the values of D, d~ , and ωco1, derived from Brillouin and Raman scattering experiments, and those obtained by SANS, INS, and electron microscopy experiments. The structure strongly depends on the preparation conditions. Silica gels obtained with basic catalysis produce fractal structures composed of connected spheroidal silica units of nanometric size (a = 1–2 nm). In gels obtained with acid catalysis, the structural units are the SiO4 tetrahedra. Gels reacted without the addition of a catalyst produce aerogels with a higher fractal dimension (D  2.4) than gels obtained with basic catalysis (D  1.8) (Courtens and Vacher 1992). The first measurement of d~ by Brillouin and INS experiments in a neutrally reacted series of aerogels with D = 2.46, d~ = 1.3, was very

12

M. Montagna

close to the universal value d~ = 4/3 suggested for percolating networks (Alexander and Orbach 1982). However, further measurements in base-catalyzed series provided d~ ¼ 1:4, D = 1.89, showing that the fracton dimension d~ is not a universal quantity, but is an additional dimension, which describes the connectivity within the fractal system (Courtens and Vacher 1992). Furthermore, the density of vibrational states obtained by INS experiments showed that d~ is not really constant for frequencies between ωco1 and ωco2, but is higher at high frequencies (Vacher et al. 1989, 1990; Courtens et al. 1990). These two different regimes were associated to bending in the low frequency range and to stretching in the high frequency range. This observation confirmed the prediction of Feng that the elasticity of tenuous materials is dominated by stretching at short scales and by bending at longer scales (Feng 1985). The scaling of the depolarized Raman spectra with a single power y in the two frequency ranges, where bending or stretching dominate, can probably be explained by a stronger Raman activity of bending over stretching vibrations (Courtens and Vacher 1992). Low-temperature specific heat and thermal conductivity measurements, between 0.05 and 20 K, confirmed that in silica aerogels three different temperature regimes are present. These three regimes correspond to phonon, fracton, and particle mode vibrations, in increasing order of temperature (Bernasconi et al. 1992). These studies were subsequently extended to partially densified aerogels, prepared by heat treatment and hydrostatic or uniaxial pressure. Changes in the spectra were correlated with densification-induced changes in structure and connectivity of the aerogels. By combining SAXS and SANS, INS, Raman, and Brillouin results, it ~ ωco1 and ωco2, ξco, and a and was possible to obtain the values of the parameters D, d, to study their evolution with densification. Heat treatment induces matter displacements at all scales including the particle scale, a, whereas the effect of pressure essentially affects the long-range structure of fractal aggregates. In base-catalyzed aerogels, the mean size, a, of the spheroidal silica particles increases with thermal treatment. This is shown by the shift of the Raman peak, the peak in the highfrequency side of Fig. 3, to low frequencies, and is confirmed by SAXS measurements (Anglaret et al. 1998). Progressive densification by thermal treatment reduces ξco (increases ωco1), while maintaining constant D and d~ . In these base-catalyzed silica aerogels, D  1.75 and d~  1:1, values quite lower than those of neutrally reacted aerogels (Anglaret et al. 1995). Brillouin measurements show that the elastic modulus increases strongly during sintering, while the attenuation decreases, which is coherent with a larger connectivity in the solid network. Viscous flow sintering creates new siloxane bonds, eliminates pores, and, as expected, stiffens the aerogel (Calas et al. 1998; Caponi et al. 2003). Partially densified aerogels, prepared by hydrostatic or uniaxial pressure, have a different behavior. Brillouin measurements show that the elastic modulus decreases and the attenuation increases in the low-pressure regime (Calas et al. 1998). This is attributed to a plastic deformation with breakage of links between clusters during compression. For higher pressure, the density increase is accompanied by stiffening, suggesting that condensation occurs more than link breakage. Raman and Brillouin data show that d~ increases with

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

13

pressure-induced densification. This is attributed to a strengthening of the bonds at the particle scale, which produces an increase of the density of vibrational states at high frequencies (Anglaret et al. 1998). The application of uniaxial pressure produces elastically isotropic samples, very similar to those obtained by hydrostatic pressure (Levelut et al. 1998). The macroscopic effect of compaction is highly anisotropic. However, the new siloxane bonds created by the asymmetric compaction are not oriented. They are created in all directions by structural rearrangements at the microscopic scale. In aerogels, the measured attenuation, given by the Brillouin line width, is mainly caused by the structural disorder present in the amorphous structure (Courtens et al. 1987). This process is almost temperature independent. On the contrary, in vitreous silica, the attenuation is mainly due to the presence of thermally activated processes, such as relaxations and two-level systems (Vacher et al. 1997). These mechanisms have been proposed to explain the strong increase, about one order of magnitude, of the Brillouin line width in v-SiO2, from liquid helium to room temperature, where Γ  75 MHz  0.0025 cm
1 (Vacher and Pelous 1976). This occurs for Brillouin scattering of visible light, where the phonon frequency is of the order of 1 cm
1 and the exchanged q vector is of the order of 0.04 nm
1. In inelastic X-ray scattering experiments in silica, where ωph  30 cm
1 and q  1 nm
1, the Brillouin line width is quite large, with Γ comparable to the peak frequency (Foret et al. 1996; Benassi et al. 1996; Pilla et al. 2000, 2002; Ruocco and Sette 2001). It is temperature independent and thus attributed to the structural disorder, which strongly attenuates the phonon propagation at these frequencies. It increases with the q vector as Γ / qα, with α  2 (Ruocco and Sette 2001). Therefore, in silica, phonon attenuation is dominated by the disordered structure (static attenuation mechanisms) at high frequency and by glass relaxation (dynamic mechanism) at ωph  1 cm
1. The frequency region in between, where the two mechanisms should give comparable effects, is not covered by experimental Brillouin facilities. However, phonon attenuation produced by structural disorder is higher in porous systems, as partially densified aerogels and xerogels. In these systems, the two mechanisms of phonon attenuation are comparable in visible and UV Brillouin scattering (Caponi et al. 2003, 2004). It has been found that for pore sizes smaller than about 8 nm, the acoustical attenuation is the same as in v-SiO2 at room temperature. The attenuation was therefore attributed to dynamical processes. For larger pore sizes, the Brillouin line width is larger and it becomes nearly temperature independent and strongly increases with the mean pore size. The main mechanism of attenuation is now phonon scattering by the disordered porous structure. Dealing with these porous hygroscopic systems, one needs to take care of removing water by high-temperature thermal treatment in order to isolate the dynamics of dry systems. In samples left in air for days, water adsorbed in the porous structure produces a decrease of the sound velocity and an increase of attenuation (Terki et al. 1999; Fontana et al. 1999). Furthermore, water dynamics gives strong low-frequency Raman scattering (Terki et al. 1999; Cicognani et al. 1999).

14

M. Montagna

Densification of Silica Xerogels Figure 4 shows the VV-polarized Raman spectra of three xerogels, heated at different temperatures, together with that of v-SiO2. The three samples were prepared using as starting solution a mixture of tetramethyl orthosilicate (TMOS), methanol, deionized water, and nitric acid, in the molar ratio 0.06:0.35:0.55:0.04, respectively. The solvent was removed by evaporation in air, at room temperature, for several weeks. The progressive densification of silica gels was obtained by thermal treatment in air, for 72 h, at several temperatures (Tt), reached with a rate of 0.1 C/min (Caponi et al. 2002). The spectra of Fig. 4 cover the frequency range of the O–H stretching vibrations. The porous samples heated at 600 C, with a density of 1300 kg/cm3, show a broad band, extending from about 3200 to about 3800 cm
1. This band is due to the O–H stretching vibration of the “physical” and “chemical” water in the pores of the gel. Physical water, i.e., the residual water in the pores of the structure, produces a broad band, which is typical of a liquid system. Chemical water, i.e., a residual layer of water on the pore surface, produces the weak and sharper structures superimposed on the broad band. Different vibrational frequencies correspond to different configurations of bonds of the chemical water with the Si–OH units and with the neighbor water molecules at the pore surface. The sharp peak, centered at about 3740 cm
1, is attributed to the O–H stretching vibrations of free silanol groups on the pore surface (Bertoluzza et al. 1982; Gottardi et al. 1984). It is much sharper than the bands of physical and chemical water, because the Si–OH groups at the pore surface are only weakly perturbed by site-sensitive interactions. After annealing at 860 C (ρ = 1800 kg/m3), the broad band disappears. The sharp peak of the free silanol groups dominates the spectrum. Residual weak bands, due to chemical water, are also present. The thermal treatment produced a nearly complete evaporation of water molecules, but the temperature was not sufficiently high to produce densification by viscous sintering (Brinker and Scherer 1990). Therefore, the structure maintained a porous structure. The sample annealed at 900 C is a densified glass with negligible porosity. The peak of free silanol groups is no longer present, and a broader band at about 3650 cm
1, also present in the spectrum of v-SiO2, appears. It is attributed to hydrogen-bonded silanol groups (Gottardi et al. 1984). The O–H groups, now embedded in the silica matrix, have sitedependent vibrations, and a Raman peak broader than that of the free silanol groups is observed. Figure 5 shows the VV spectra of some xerogels heated at different temperatures, together with that of v-SiO2, in the whole frequency range of vibration of the silica network. We can see that the major features present in v-SiO2 are also observed in the xerogels. The intense broad band at about 400 cm
1 is attributed to stretching–bending vibrations of the silica network (Galeener 1985). Its width is related to the disorder-induced distribution of angles in the Si–O–Si units, which connect the SiO4 tetrahedra. The band is sharper in the xerogels than in v-SiO2, since the open porous structure has a lower number of constraints than the harder compact silica network. The two sharp peaks of silica at about 490 and 605 cm
1, known as D1 and D2 defect bands, are attributed to local defects with four- and threefold rings,

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

15

Fig. 4 VV-polarized Raman spectra of xerogels annealed at different temperatures Tt in the high-frequency region of the O–H stretching vibrations. For comparison, the spectrum of vitreous silica is also reported (Caponi et al. 2002)

instead than the normal sixfold rings of the silica network (Galeener 1985). In the spectra of xerogels, these bands appear more intense than in silica. Their line shapes are different from those of silica, since the two peaks are broader and slightly shifted in frequency, the D1 band being centered at 478 cm
1 (Walrafen et al. 1985, 1986; Mulder and Damen 1987; Brinker et al. 1990). The Raman spectra clearly show that three- and fourfold rings are favored on the pore surface. Note that the densified xerogel, annealed at 900 C, shows D1 and D2 bands less intense than v-SiO2, indicating a less defective structure. The band at about 800 cm
1 and the two weak bands at 1050 and 1200 cm
1 are attributed to symmetric and antisymmetric Si–O stretching vibrations, respectively. These bands have practically the same shape in the xerogels and in v-SiO2, indicating that the building blocks, i.e., the SiO4 tetrahedral structures, are the same. The relatively sharp band at 970 cm
1, whose intensity decreases with thermal treatment, is due to Si–OH stretching vibrations of silanol groups (Bertoluzza et al. 1982; Gottardi et al. 1984). The peak at about 50 cm
1, present in the spectra of v-SiO2 and of the densified xerogel, is the boson peak, a common feature of the room temperature Raman spectra of all glasses. Its

16

M. Montagna

Fig. 5 VV-polarized Raman spectra of xerogels annealed at different temperatures Tt. For comparison, the spectrum of vitreous silica is also reported (Caponi et al. 2002)

shape appears more clearly in Fig. 6, which shows the HV Raman spectra in the frequency region below 100 cm
1. HV Raman spectra are more useful for the study of the boson peak, which is depolarized with comparable intensities in HV and VV polarizations. In fact, the intense band of Fig. 5, centered at 400 cm
1, but having a long tail toward low frequencies, is almost completely polarized and thus does not appear in HV polarization. Figure 6 shows that the structure of these xerogels abruptly changes in the 860 C < Tt < 875 C temperature range, with the sudden appearance of the boson peak. This result is attributed to the peculiarity of the densification process. The viscous sintering, with the elimination of pores and dangling Si–OH bonds, occurs at a well-defined temperature, producing a glass with a structure close to that of v-SiO2. In the non-densified xerogels, the Raman spectra show a relatively sharp band centered at about 16 cm
1, for the xerogel with Tt = 800 C, and at about 18 cm
1, for the xerogel with Tt = 860 C. In the samples treated below 800 C, it is not possible to observe the low-frequency bump. The strong quasi-elastic scattering is due to vibrational dynamics of the porous fractal-like system, but probably also to

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

17

Fig. 6 HV-polarized Raman spectra in the low-frequency region of xerogels annealed at different temperatures, Tt, and of vitreous silica: (a) several heat treatment temperatures and vitreous silica, with spectra normalized to the intensity of the band at 800 cm
1; (b) spectra normalized to the intensity of the low-frequency band (Caponi et al. 2002)

residual water inside the pores. The low-frequency bumps are attributed to surface vibrations of the particle–pore structure. The peak frequency increases with the annealing temperature, as the mean pore size, measured by SAXS experiments, decreases (see “Raman Spectroscopy of Nanocrystals”). The Raman data in the low frequency range of Fig. 6 show that the vibrational dynamics are almost the same for the densified xerogels and v-SiO2; the boson peak is centered at the same frequency and has a similar spectral shape. An excess of scattering in the low-frequency part of the boson peak is present in the densified xerogel samples heated at 875 C and 900 C, indicating the presence of a small amount of residual porosity, to which low-frequency Raman scattering is very sensitive. The density and SAXS measurements on the sample treated at 900 C show a small residual porosity with large pore size (6 nm) and indeed small internal surface area. Low-temperature specific heat and Brillouin scattering measurements confirm the Raman results. Important specific heat excess is observed for temperatures lower than about 5 Kin the non-densified xerogels, whereas the specific heat of the sample annealed at 900 C is identical to that of v-SiO2 (Caponi et al. 2002). The sound velocity of the xerogel increases with the annealing temperature, suddenly reaching that of v-SiO2 for Tt 875 C (Bartolotta et al. 2001). The appearance of the boson peak in the Raman spectra is now considered a signature of

18

M. Montagna

the occurred densification (Mariotto et al. 1988a; Armellini et al. 1998). This is particularly important when densification is needed, but at the lowest possible annealing temperature, in order to avoid or reduce crystallization effects (Bouajaj et al. 1997; Montagna et al. 2003; Zampedri et al. 2003). It has been observed that silica xerogels activated with rare earth ions have densification temperatures that depend on the content and nature of the doping ions. Doping with Tb3+ increases the transition temperature, whereas the opposite occurs by doping with Pr3+ (Pucker et al. 1998; Armellini et al. 1998).

Characterization by Waveguided Brillouin Spectroscopy As discussed in “Inelastic Light Scattering,” Brillouin scattering in bulk glasses produces sharp peaks, whose line width is determined by the dynamical mechanism of phonon attenuation. In fact, the exchanged q is precisely determined by the geometry of the experiment, since the scattered wave is collected from a large volume, with a typical size of 50 50 5000 μm3, uniformly illuminated by the plane wave of the laser beam. On the contrary, an exciting light beam, confined in a planar waveguide, does not have a single well-defined wavevector. In a thick waveguide, a ray-optic approach can be used for the description of wave propagation, as shown in Fig. 7. The light propagates in the z direction and the plane waves have a zigzag path in the x–z plane, undergoing total internal reflection at the boundary interfaces of the waveguide. The laser beam, polarized along the y direction, can be injected into the waveguide by prism coupling and propagates only at discrete values of the angle ϑ in one of the transverse electric (TEm) modes. The m value (m = 0, 1, 2,. . ., mmax) gives the number of nodes of the electric field, a stationary wave, along the x direction. As shown in Fig. 7, when the scattered light is collected from the front surface of the waveguides, two exchanged q vectors are present. A simple model, which takes into account the relative phases of the two scattered waves, shows that two main peaks due to longitudinal phonons are present, apart from the case of the TE0 excitation, where a single peak is observed (Montagna et al. 1998). The energy separation between the two peaks increases with the mode index. This model neglects the contribution to the scattering coming from the evanescent field in the substrate and considers the waveguide as a homogeneous film with constant refractive index. A single fit parameter, i.e., the longitudinal sound velocity, is used to calculate the m + 1 spectra obtained by exciting the different modes of the waveguide. In graded-index waveguides, m – 1 peaks of comparable intensities are observed. Figure 8 shows the Brillouin spectra of a silica–titania graded-index waveguide. The sample was deposited by a dip-coating technique (Zampedri et al. 2003). The graded-index SiO2–TiO2 planar waveguide was obtained by subsequently depositing 35 layers of TiO2(0.08)–SiO2(0.92) composition, 25 layers of TiO2(0.16)–SiO2(0.84), and finally 23 layers of TiO2(0.24)–SiO2(0.76) composition (molar fractions in parenthesis), on a silica substrate. After each dip, the films were annealed in air at 900 C for 30 s. After every ten dipping cycles, the films were heat treated at 900 C for 2 min. Finally, the

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

19

Fig. 7 Wave propagation in the planar waveguide. q1 and q2 are the exchanged wavevectors of the scattered light in the zigzag paths. In TE modes, the electric field is along the y direction, the light propagates along the z direction, and x is perpendicular to the plane of the waveguide in the direction of the scattered wave. ns and ng stand for the refractive indices of the substrate and waveguide, respectively (Chiasera et al. 2003a)

waveguides were submitted to a further heat treatment at 1000 C, with a heating rate of 20 C/min from 600 C to 1000 C. The waveguide was characterized by m-line spectroscopy. It supports five TE and TM modes at 543.5 nm and four at 632.8 nm. The refractive index profile, calculated at 514.5 nm by extrapolation of the data at 543.5 and at 632.8 nm, is shown on the upper left side of Fig. 8. The calculated profiles of the squared electric field show that the different modes selectively excite the different layers of the waveguide; the TE0 mode is practically completely confined in the third (from the substrate) layer with TiO2(0.24)–SiO2(0.76) composition, while the TE4 mainly occupies the first layer, with TiO2(0.08)–SiO2(0.92) composition. The Brillouin spectra in the longitudinal phonon spectral region are shown on the right-hand side of Fig. 8, together with the results of a numerical model, which considers the spatial distribution of the exciting field in the mode, a simple spatial dependence of the elasto-optic coefficients, through the value of the refractive index, and neglects the refraction of phonons (Chiasera et al. 2003a). A single fit parameter, i.e., the sound velocity, is necessary to obtain the calculated spectra. For the TE0, TE1, and TE2 spectra, which show one, two, and three peaks, respectively, the agreement is good, even if the observed intensity of the higherfrequency components in TE1 and TE2 spectra is lower than expected and the observed line width is slightly larger than expected. In the TE3 spectrum, only three peaks are observed, instead of the four calculated ones. TE4 shows a single strong peak with shoulders, partially reproduced by the calculations. In any case, a general agreement is present in all spectra, sufficient to determine with high accuracy the longitudinal sound velocity. Within the experimental error, we used the same value υL = 5.9 km/s for the four excitations with m  3. These four modes involve, with different weights, the two external layers with the compositions TiO2(0.24)–SiO2(0.76) and TiO2(0.16)–SiO2(0.84). We should conclude that the longitudinal sound velocity is nearly the same for the two compositions. A quite weak dependence of the sound velocity on the titania content was already observed (Montagna et al. 1998). On the contrary, the TE4 spectrum, with most excitation in

20

M. Montagna

Fig. 8 Upper left-hand side frame: refractive index profile at 514.5 nm of a three-layered SiO2–TiO2 planar waveguide. Left-hand side column: calculated squared electric-field patterns of the five TEm modes. Right-hand side column: Brillouin experimental spectra (open circles), calculated spectra (dotted line), and convolution of the calculated spectra with the instrumental response (solid line). The longitudinal sound velocity used in the fit is υL = 5.9 km/s, for m = 0, 1, 2, and 3, and υL = 5.75 km/s, for m = 4 (Chiasera et al. 2003b)

the internal layer with TiO2(0.08)–SiO2(0.92) composition, shows a lower value of the longitudinal sound velocity, υL = 5.75 km/s. This is attributed to a residual porosity of this internal layer, which did not undergo full densification (Chiasera et al. 2003b). The result was confirmed by waveguided Raman spectroscopy. The Raman spectrum obtained by exciting the waveguide in the TE4 mode shows residual porosity. In particular, a weak band due the stretching vibrations of free Si–OH groups appears. On the contrary, the Raman spectra obtained by exciting in

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

21

the TE0, TE1, and TE2 modes show that the two external layers are completely densified. This was probably due to densification of the external layers occurring at a lower temperature than that of the internal first one, which was indeed not allowed to freely expel its reaction products.

Raman Spectroscopy of Glass Ceramics Raman spectroscopy is a very powerful technique to follow the devitrification process induced by thermal treatments, which produce glass ceramics. A good example is that of silica–titania xerogels (Almeida and Christensen 1997; 1998; Strohhöfer et al. 1988; Bersani et al. 1998a, b; Bersani et al. 1988; Karthikeyan and Almeida 2000; Montagna et al. 2003; Zampedri et al. 2003). Figure 9 shows the Raman spectra of a series of TiO2(x)–SiO2(1 – x) waveguides obtained by dip coating and annealing at 900 C. The waveguides are activated with 1 mol% Er3+, they have a thickness between 1.4 and 2.2 μm, and they support a TE and a TM mode at 1.5 μm, plus two or three modes in the visible. The waveguiding configuration was used for the measurements. The Raman spectra were collected in VV polarization, by exciting by prism coupling the TE0 mode with an Ar+-ion laser operating at 457.9 nm and detecting the scattered light from the front of the waveguide (Zampedri et al. 2003). All spectra show the characteristic peaks of the silica. The bands at about 950 and 1100 cm
1 are assigned to the vibrations of mixed Si–O–Ti linkages (Best and Condrate 1985). All spectra show an initial crystallization process, evidenced by the structure in the region between 150 and 350 cm
1, attributed to optical vibrations of TiO2 crystals (Moret et al. 2000), and by the low frequency peak, due to the acoustic vibrations of titania nanocrystals (Montagna et al. 2003), which partially overlaps with the boson peak of the glass at about 40 cm
1. They show that devitrification of the silica–titania film occurs and this effect is more important for the waveguides with 15, 20, and 24 mol% of TiO2. Pure glassy films are obtained by annealing at 700 C after each dip. In any case, the successive annealing processes at higher temperature, necessary for a full densification of the xerogel, produced some degree of devitrification (Zampedri et al. 2003). Figure 10 shows the VV Raman spectra of thinner SiO2(0.8)–TiO2(0.2) waveguides, with a thickness of about 0.4 μm, treated at different temperatures (Montagna et al. 2003). The spectrum of the waveguide heated at 700 C shows the typical bands of silica–titania amorphous network, with no visible contributions from a crystalline phase. After annealing at 800 C, sharp peaks appear, superimposed on the broadband spectrum of the glass. These peaks become more and more intense with increasing annealing temperature and time. They are attributed to optical vibrations of crystalline TiO2. For Ta 1200 C, only the characteristic peaks of the anatase phase are observed (Haro-Poniatowski et al. 1994; Moret et al. 2000), with a sharp and intense peak at about 141 cm
1, attributed to Eg vibration and three other peaks at about 394 cm
1 (B1g), 513 cm
1 (A1g + B1g), and 635 cm
1 (Eg). The frequencies of the peaks are smaller than those of the bulk crystal (Haro-Poniatowski et al. 1994), because of phonon confinement

22

M. Montagna

Fig. 9 VV-polarized Raman spectra of TiO2(x)–SiO2(1
x): Er3+ waveguides with different TiO2 contents. The VV spectrum of SiO2 glass is also shown (Zampedri et al. 2003)

in the nanoparticles (see “Raman Spectroscopy of Nanocrystals”). For Ta in the range 800–1200 C, the Raman spectra indicate that the nanocrystals are a mixture of anatase and brookite phases (Moret et al. 2000). The anatase content seems to increase progressively with the annealing temperature. At very low frequency, a structured band appears in Fig. 10. It increases in intensity and progressively shifts toward low frequency with the annealing temperature. It is assigned to the acoustic vibrations of the titania nanocrystals, and its frequency position allows the determination of the particle size, as we will discuss in detail in the next section.

Raman Spectroscopy of Nanocrystals Conventional methods for measuring nanoparticle size are transmission electron microscopy, in which the particles are directly imaged, and X-ray diffraction, in which the particle size is inferred from the width of the diffraction lines, using the Sherrer method. Raman scattering of optical and, in particular, of acoustic vibrations is a simple, fast technique for obtaining the size distribution of nanoparticles. Phonon confinement in the nanoparticles produces shift and broadening of the Raman lines of optical phonons (Richter et al. 1981; Tiong et al. 1984; Campbell and Fauchet 1986). The momentum conservation rule of Eq. 5 does not hold for small

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

23

Fig. 10 VV-polarized Raman spectra of SiO2(0.8)–TiO2(0.2) waveguides. The labels indicate the temperature ( C) and time (min) of annealing. Excitation was at 514.5 nm, by prism coupling, in the TE0 mode (Montagna et al. 2003)

crystalline particles with qa < 1, where a is the particle size. A phenomenological model, i.e., the spatial correlation model, can account for the frequency shift. Confined optical phonons with any kph contribute to the Raman scattering (Campbell and Fauchet 1986):    C q, kph 2 dkph ; I ð ωÞ /  2 ω
ωph þ ðΓ0 =2Þ2 ð

(10)

where Γ0 is the width of the Raman line in the bulk crystal. Since the q vector of visible light is much smaller than the kph of phonons in the nanocrystal (q  0), a Gaussian function is usually taken for the Fourier coefficient of the confinement function: !    k 2 a2 C q, kph 2 ¼ exp
ph : 16π 2

(11)

The phonon dispersion relation ω(kph) of the bulk modes is usually taken. For most crystalline solids, the phonon dispersion relation has a maximum at the center

24

M. Montagna

of the Brillouin zone, and one observes a low-frequency shift of the Raman line with decreasing crystalline size, but the opposite behavior can also occur (Nemanich et al. 1981). The technique has been applied to the size determination of TiO2 and PbTiO3 nanoparticles in films deposited by sol–gel (Bersani et al. 1998a, b; Bersani et al. 1988). The mechanisms that give rise to the broadening and the shifts of the Raman peaks of titanium dioxide and lead titanate nanocrystals prepared by sol–gel are different. Phonon confinement and oxygen deficiency are competitive mechanisms in TiO2 obtained by different sol–gel preparations, whereas pressure effects on the nanocrystals predominate in ferroelectric PbTiO3. For CdS nanocrystals in silica xerogels, no shift of the Raman line was observed (Capoen et al. 2001). It was suggested that the theoretically predicted redshift due to the phonon confinement may be hindered by a blueshift. This second effect would be caused by strain acting on the surface of the nanocrystals (Shiang et al. 1993). In fact, phonon confinement is one possible cause of shift and broadening of the Raman line, but there are other causes, as the presence of a size distribution. Stress also causes a shift of the line. For these reasons, a reliable measurement of the size of the nanocrystals is rarely obtained from the line shape of optical phonons. On the contrary, a much more precise measurement of the nanocrystal size can be obtained from Raman scattering of the acoustic vibrations. After the first works on spinel nanocrystals in cordierite glasses (Boukenter et al. 1986) and on silver colloids in alkali halides (Mariotto et al. 1988b), low-frequency Raman scattering from symmetric and quadrupolar acoustic vibrations of nanoparticles has become a nondestructive method to determine the size of the particles. A peak in the range 5–50 cm
1 was observed in many composite systems containing metallic, insulator, or semiconductor nanoparticles (Fujii et al. 1991, 1996; Ferrari et al. 1996, 1999; Ceccato et al. 2001; Tikhomirov et al. 2002; Montagna et al. 2003; Ivanda et al. 2003). The size of the nanoparticles is derived from the energy of the peak, since the frequency of all modes scales as the inverse of the linear dimension of the particles. The acoustic vibrations of an elastic homogeneous sphere with a free surface are classified as spheroidal and torsional modes (Lamb 1882). Torsional vibrations involve only shear deformations and are not Raman active (Duval 1992). Spheroidal modes involve both shear and stretching motions and produce radial displacements. They are characterized, following the symmetry of the sphere, by three labels (l, m, p). The symmetric l = 0 (m = 0) spheroidal modes are purely radial with spherical symmetry. At higher l values, angular corrugations appear. l measures the number of wavelengths along a circle on the surface. A third index, p = 1, 2, . . . labels the sequence of modes in increasing order of frequency and radial wavevector at fixed angular shape (l, m). The quantity p
1 measures the number of nodes of the vibrations in the radial direction. The fundamental p = 1 mode is called surface mode, its overtones ( p > 1) being called inner modes. Only the symmetric (l = 0, m = 0) and the quadrupolar (l = 2,
2 < m < 2), with fivefold degeneracy, are Raman active (Duval 1992). The frequencies of all modes scale as the inverse radius of the sphere, allowing the introduction of the dimensional quantities:

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

hR ¼ ω0 R=υL ,

kR ¼ ω2 R=υT ;

25

(12)

for the symmetric and for the quadrupolar modes, where υL and υT are the longitudinal and transverse sound velocities, respectively. hR and kR depend only on the ratio of the longitudinal and transverse sound velocities υL/υT. This dependence is shown in Fig. 11, for the surface modes (p = 1). If the sound velocities of the particles are known, one can obtain a first rough value of the particle size from the position of the l = 0 and/or l = 2 peak, by using the relations (12). The symmetric l = 0 modes give polarized Raman spectra (IVH = 0); the quadrupolar l = 2 modes give depolarized spectra. Therefore, on the basis of the depolarization ratio, IVH/IVV, the Raman peaks can be assigned to symmetrical or quadrupolar vibrations. The depolarization ratio IVH/IVV of the quadrupolar modes and the relative efficiencies of the quadrupolar and symmetric modes are system dependent, since they depend on the microscopic structure and scattering mechanism (Montagna and Dusi 1995). In silver nanoparticles, only the depolarized quadrupolar l = 2 vibrations are Raman active (Fujii et al. 1991; Ferrari et al. 1995, 1996, 1999; Palpant et al. 1999). This occurs because the symmetric l = 0 vibrations are not Raman active in crystals with a cubic Bravais lattice (Montagna and Dusi 1995). On the contrary, for CdS nanocrystals, the symmetric l = 0 modes dominate the Raman spectra, the contribution of the quadrupolar vibrations being relatively weak (Saviot et al. 1998; Ivanda et al. 2003). In other systems, as Ga2O3 and TiO2, quadrupolar and symmetric vibrations have comparable intensities (Ceccato et al. 2001; Montagna et al. 2003). Figure 12 shows the spectra of the acoustic vibrations of TiO2 nanoparticles, grown by thermal treatment of silica–titania waveguides obtained by dip coating (Montagna et al. 2003). The l = 2 surface vibrations, which are active both in VV and VH polarizations, produce the lowest frequency peak. The intense peak at higher frequency, present only in VV spectra, is attributed to the l = 0 surface mode. The weaker peak at higher frequency, at about 35 cm
1 in sample s1200(60) (annealed at 1200 C for 60 min) Fig. 11 Dimensional frequencies of the l = 0 (hR) and l = 2 (kR) surface modes of a free sphere, as a function of the ratio between the longitudinal and transverse sound velocities, υL/υT

26

M. Montagna

and at about 23 cm
1 in sample s1300(30), is due to an inner l = 0 mode (p = 2), a shorter wavelength symmetric mode with a node in the radial wave function. In general, the intensities of the peaks of a sequence ( p = 1, 2, . . .) rapidly decrease with p, especially in the l = 2 case (Montagna and Dusi 1995), so that only the surface p = 1 mode can be easily detected. All peaks shift toward lower frequencies as the annealing temperature increases, showing the progressive increase of the mean size of the nanocrystals. The Raman spectra were fitted by considering that the line width of the peaks has two main sources: the homogeneous broadening, due to the interaction of the vibrating particle with the surrounding glass (Montagna and Dusi 1995), and the inhomogeneous broadening, due to a distribution of the particle sizes. For annealing temperatures between 900 C and 1300 C, the mean size increases from about 4 to 20 nm in diameter. Crystallites are present even after annealing at 800 C, but their size cannot

Fig. 12 Low-frequency VV and HV Raman spectra of SiO2 (0.8)–TiO2 (0.2) waveguides. The labels indicate the temperature ( C) and time (min) of annealing. Excitation was at 514.5 nm, by prism coupling in the TE0 mode (Montagna et al. 2003)

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

27

be well evaluated from the low-frequency Raman spectra, because the relative scattering is weak and not well resolvable from the boson peak of the glass. The mean particle size obtained by the Raman spectra compare well with those obtained from the line width in X-ray diffraction measurements, for particle sizes larger than about 8 nm. For smaller particles, Raman data give sizes larger than those obtained by X-ray measurements (Montagna et al. 2003). Low-frequency Raman scattering from acoustic vibrations was also employed to determine the size of CdS nanocrystals in silica xerogels (Othmani et al. 1992).

Conclusions The survey presented in this chapter shows that Raman and Brillouin spectroscopies in sol–gel-derived materials cover a quite vast domain of investigation, from the basic glass science to the characterization of materials produced for many different applications. The evolution of Raman instrumentation is now allowing extending Raman spectroscopy from the laboratory to industry, as monitoring process for quality control.

References Adar F. Evolution and revolution of Raman instrumentation-application of available technologies to spectroscopy and microscopy. In: Lewis IR, Edwards HGM, editors. Handbook of raman spectroscopy: from the research laboratory to the process line. New York: Marcel Dekker Inc; 2001. Aharony A, Alexander S, Enti-Wohlman O, Orbach R. Scattering of fractons, the Ioffe–Regel criterion, and the (4/3) conjecture. Phys Rev Lett. 1987;58:132–5. Alexander S. Vibrations of fractals and scattering of light from aerogels. Phys Rev B. 1989;40:7953–65. Alexander S, Orbach R. Density of states on fractals: “fractons”. J Phys (Paris) Lett. 1982;43: L625–32. Alexander S, Courtens E, Vacher R. Vibrations of fractals: dynamic scaling, correlation functions and inelastic light scattering. Physica A. 1993;195:286–318. Almeida RM. Spectroscopy and structure of sol–gel systems. J Sol–Gel Sci Technol. 1998;13:51–9. Almeida RM, Christensen EE. Crystallization behavior of SiO2–TiO2 sol–gel thin films. J Sol–Gel Sci Technol. 1997;8:409–13. Anglaret E, Hasmy A, Courtens E, Pelous J, Vacher R. Fracton dimension of mutually self-similar series of base-catalyzed aerogels. J Non Cryst Solids. 1995;186:131–6. Anglaret E, Beurroies I, Duffours L, Levelut C, Foret M, Delord P, Woignier T, Phalippou J, Pelous J. A low frequency Raman study of fractons in partially densified silica aerogels. J Non Cryst Solids. 1998;225:248–53. Armellini C, Del Longo L, Ferrari M, Montagna M, Pucker G, Sagoo P. Effect of Pr3+ doping on the OH content of silica xerogels. J Sol–Gel Sci Technol. 1998;13:599–603. Baldwin KJ, Batchelder DN, Webster S. Raman microscopy: confocal and scanning near-field. In: Lewis IR, Edwards HGM, editors. Handbook of Raman spectroscopy: from the research laboratory to the process line. New York/Basel: Marcel Dekker Inc; 2001.

28

M. Montagna

Bartolotta A, Carini G, D’Angelo G, Ferrari M, Fontana A, Montagna M, Rossi F, Tripodo G. A study of Raman spectroscopy and low temperature specific heat in gel-synthesized amorphous silica. J Non-Cryst Solids. 2001;280:249–54. Benassi P, Pilla O, Mazzacurati V, Montagna M, Ruocco G, Signorelli G. Disorder-induced light scattering in solids: microscopic theory and applications to some model systems. Phys Rev B. 1991;44:11734–42. Benassi P, Mazzacurati V, Ruocco G, Signorelli G. Elasto-optic constants in silicate-glasses: experiment and theory. Phys Rev B. 1993;48:5987–96. Benassi P, Frizzera W, Montagna M, Viliani G, Mazzacurati V, Ruocco G, Signorelli G. Origin of light scattering from disordered systems. Physica A. 1995;216:32–44. Benassi P, Krisch M, Masciovecchio C, Mazzacurati V, Monaco G, Ruocco G, Sette F, Verbeni R. Evidence of high frequency propagating modes in vitreous silica. Phys Rev Lett. 1996;77:3835–8. Bernasconi A, Sleator T, Posselt D, Kjems JK, Ott HR. Dynamic properties of silica aerogels as deduced from specific-heat and thermal-conductivity measurements. Phys Rev B. 1992;45:10363–76. Bersani D, Lottici PP, Lopez T, Ding X-Z. A Raman scattering study of PbTiO3 and TiO2 obtained by Sol–Gel. J Sol–Gel Sci Technol. 1988;13:849–53. Bersani D, Antonioli G, Lottici PP, Lopez T. Raman study of nanosized titania prepared by sol–gel route. J Non-Cryst Solids. 1998a;234:175–81. Bersani D, Lottici PP, Ding XZ. Phonon confinement effects in the Raman scattering by TiO2 nanocrystals. Appl Phys Lett. 1998b;72:73–5. Bertoluzza A, Fagnano C, Morelli MA, Gottardi V, Guglielmi M. Raman and infrared spectra of silica gel evolving toward glass. J Non-Cryst Solids. 1982;48:117–28. Best MF, Condrate RA. A Raman study of TiO2–SiO2 glasses prepared by sol–gel processes. J Mater Sci Lett. 1985;4:994–8. Bouajaj A, Ferrari M, Montagna M. Crystallization of silica xerogels: a study by Raman and fluorescence spectroscopy. J Sol–Gel Sci Technol. 1997;8:391–5. Boukenter A, Champagnon B, Duval E, Dumas J, Quinson JF, Serughetti J. Low-frequency Raman scattering from fractal vibrational modes in a silica gel. Phys Rev Lett. 1986;57: 2391–4. Brinker CJ, Scherer GW. Sol–Gel science – the physics and chemistry of the Sol–Gel processing. New York: Academic; 1990. Brinker CJ, Brow RK, Tallant DR, Kirkpatrick RJ. Surface structure and chemistry of high surface area silica gels. J Non-Cryst Solids. 1990;120:26–33. Calas S, Levelut C, Woignier T, Pelous J. Brillouin scattering study of sintered and compressed aerogels. J Non-Cryst Solids. 1998;225:244–7. Campbell IH, Fauchet PM. The effects of microcrystal size and shape on the one phonon Ramanspectra of crystalline semiconductors. Solid-State Commun. 1986;58:739–41. Capoen B, Gacoin T, Nédélec JM, Turrell S, Bouazaoui M. Spectroscopic investigation of CdS nanoparticles in sol–gel derived polymeric thin films and bulk silica matrices. J Mater Sci. 2001;36:2565–70. Caponi S, Ferrari M, Fontana A, Masciovecchio C, Mermet A, Montagna M, Rossi F, Ruocco G, Sette F. X-ray diffraction and Raman scattering measurements on silica xerogels. J Non-Cryst Solids. 2002;307–310:135–41. Caponi S, Fontana A, Montagna M, Pilla O, Rossi F, Terki F, Woignier T. Acoustic attenuation in silica porous systems. J Non-Cryst Solids. 2003;322:29–34. Caponi S, Benassi P, Eramo R, Giugni A, Nardone M, Fontana A, Sampoli M, Terki F, Woignier T. Phonon attenuation in vitreous silica and silica porous systems. Philos Mag B. 2004;84:1423 (in press). Ceccato R, Dal Maschio R, Gialanella S, Mariotto G, Montagna M, Rossi F, Ferrari M, LipinskaKalita KE, Ohki Y. Nucleation of Ga2O3 nanocrystals in the K2O–Ga2O3–SiO2 glass system. J Appl Phys. 2001;90:2522–7.

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

29

Chiasera A, Montagna M, Rossi F, Ferrari M. Brillouin scattering in planar waveguides. I. Numerical model. J Appl Phys. 2003a;94:4876–81. Chiasera A, Montagna M, Moser E, Rossi F, Tosello C, Ferrari M, Zampedri L, Caponi S, Gonçalves RR, Chaussedent S, Monteil A, Fioretto D, Battaglin G, Gonella F, Mazzoldi P, Righini GC. Brillouin scattering in planar waveguides. II. Experiments. J Appl Phys. 2003b;94:4882–9. Cicognani G, Dianoux AJ, Fontana A, Rossi F, Montagna M, Scopigno T, Pelous J, Terki F, Pilliez JN, Woignier T. Low frequency dynamics of silica xerogel porous system. Philos Mag B. 1999;79:2091–102. Conrad H, Buchenau U, Sch€atzler R, Reichenauer G, Fricke J. Crossover in the vibrational density of states of silica aerogels studied by high-resolution neutron spectroscopy. Phys Rev B. 1990;41:2573–6. Courtens E, Vacher R. Structure and dynamics of silica aerogels. Philos Mag B. 1992;65:347–55. Courtens E, Pelous J, Phalippou J, Vacher R, Woignier T. Brillouin-scattering measurements of phonon–fracton crossover in silica aerogels. Phys Rev Lett. 1987;58:128–31. Courtens E, Vacher R, Pelous J, Woignier T. Observation of fractons in silica aerogels. Europhys Lett. 1988;6:245–50. Courtens E, Lartigue C, Mezei F, Vacher R, Coddens G, Foret M, Pelous J, Woignier T. Measurement of the phonon–fracton crossover in the density of states of silica aerogels. Z Phys B. 1990;79:1–2. Duval E. Far-infrared and Raman vibrational transitions of a solid sphere: selection rules. Phys Rev B. 1992;46:5795–7. Duval E, Garcia N, Boukenter A, Serughetti J. Correlation-effect on Raman-scattering from low-energy vibrational modes in fractal and disordered systems. 1. Theory. J Chem Phys. 1993;99:2040–5. Feng S. Crossover in spectral dimensionality of elastic percolation systems. Phys Rev B. 1985;32:5793–7. Ferrari M, Gratton LM, Maddalena A, Montagna M, Tosello C. Preparation of silver nanoparticles in silica films by combined thermal and electron-beam deposition. J Non-Cryst Solids. 1995;191:101–6. Ferrari M, Gonella F, Montagna M, Tosello C. Detection and size determination of Ag nanoclusters in ion-exchanged soda-lime glasses by waveguided Raman spectroscopy. J Appl Phys. 1996;79:2055–9. Ferrari M, Montagna M, Ronchin S, Rossi F, Righini GC. Waveguide luminescence and Raman spectroscopy: characterization of an inhomogeneous film at different depths. Appl Phys Lett. 1999;75:1529–31. Fontana A, Viliani G, editors. Proceedings of the 6th international workshops on disorder systems, 1997, Andalo. Philos Mag B. 1998;77(2 special issue):901. Fontana A, Montagna M, Rossi F, Ferrari M, Pelous J, Terki F, Woigner T. Low frequency light scattering in silica xerogels. J Phys: Condens Mater. 1999;11:A207–11. Foret M, Courtens E, Vacher R, Suck JB. Scattering investigation of acoustic localization in fused silica. Phys Rev Lett. 1996;77:3831–4. Fujii M, Nagareda T, Hayashi S, Yamamoto K. Low-frequency Raman scattering from small silver particles embedded in SiO2 thin films. Phys Rev B. 1991;44:6243–8. Fujii M, Kanzawa Y, Hayashi S, Yamamoto K. Raman scattering from acoustic phonons confined in Si nanocrystals. Phys Rev B. 1996;54:R8373–6. Galeener FL. Raman and ESR studies of the thermal history of amorphous SiO2. J Non-Cryst Solids. 1985;71:373–86. Gottardi V, Guglielmi M, Bertoluzza A, Fagnano C, Morelli MA. Further investigations on Raman spectra of silica gel evolving towards glass. J Non-Cryst Solids. 1984;63:71–80. Haro-Poniatowski E, Rodriguez-Talavera R, de la Cruz Heredia M, Cano-Corona O, ArroyoMurillo R. Crystallization of nanosized titania particles prepared by sol–gel process. J Mater Res. 1994;9:2102–8.

30

M. Montagna

Ivanda M, Babosci K, Dem C, Schmitt M, Montagna M, Kiefer W. Low-wavenumber Raman scattering from CdSxSe1–x quantum dots embedded in a glass matrix. Phys Rev B. 2003;67(235329):1–8. Karthikeyan A, Almeida RM. Crystallization of SiO2–TiO2 glassy films studied by atomic force microscopy. J Non-Cryst Solids. 2000;274:169–74. Lamb H. Proc Lon Math Soc. 1882;13:187. Levelut C, Anglaret E, Pelous J. Brillouin scattering of aerogels densified under uniaxial pressure. J Non-Cryst Solids. 1998;225:272–6. Lewis IR. Process Raman spectroscopy. In: Lewis IR, Edwards HGM, editors. Handbook of Raman spectroscopy: from the research laboratory to the process line. New York: Marcel Dekker Inc; 2001. Lewis IR, Edwards Howell GM, editors. Handbook of Raman spectroscopy: from the research laboratory to the process line. New York: Marcel Dekker Inc; 2001. Long DA. Raman spectroscopy. New York: McGraw-Hill; 1977. Mandelbrot B. The fractal geometry of nature. San Francisco: Freeman; 1982. Mariotto G, Montagna M, Viliani G, Campostrini R, Carturan G. Low-frequency Raman scattering in thermally treated silica gels: observation of phonon-fracton crossover. J Phys C. 1988a;21:L797–801. Mariotto G, Montagna M, Viliani G, Duval E, Lefrant S, Rzepka E, Mai C. Low energy Raman scattering from silver particles in alkali halides. Europhys Lett. 1988b;6:239–44. Martin AJ, Brenig W. Model for Brillouin scattering in amorphous solids. Phys Stat Sol (b). 1974;64:163–72. Mazzacurati V, Benassi P, Ruocco G. A new class of multiple dispersion grating spectrometers. J Phys E: Sci Instrum. 1988;21:798–804. Mazzacurati V, Montagna M, Pilla O, Viliani G, Ruocco G, Signorelli G. Vibrational dynamics and Raman scattering in fractals: a numerical study. Phys Rev B. 1992;45:2126–37. Montagna M, Dusi R. Raman scattering from small spherical particles. Phys Rev B. 1995;52:10080–9. Montagna M, Pilla O, Viliani G, Mazzacurati V, Ruocco G, Signorelli G. Numerical study of Raman scattering from fractals. Phys Rev Lett. 1990;65:1136–9. Montagna M, Ferrari M, Rossi F, Tonelli F, Tosello C. Brillouin scattering in planar waveguides. Phys Rev B. 1998;58:R547–50. Montagna M, Moser E, Visintainer F, Ferrari M, Zampedri L, Martucci A, Guglielmi M, Ivanda M. Nucleation of titania nanocrystals in silica titania waveguides. J Sol–Gel Sci Technol. 2003;26:241–4. Moret MP, Zallen R, Vijay DP, Desu SB. Brookite-rich titania films made by pulsed laser deposition. Thin Solid Films. 2000;366:8–10. Mulder CAM, Damen AAJM. The origin of the “defect” 490 cm
1 Raman peak in silica gel. J Non-Cryst Solids. 1987;93:387–94. Nemanich RJ, Solin SA, Martin RM. Light scattering study of boron nitride microcrystals. Phys Rev B. 1981;23:6348–56. Othmani A, Bovier C, Dumas J, Champagnon B. Raman scattering in high concentration CdS-doped sol–gel silica glass. J Phys IV. 1992;2:C2-275–8. Palpant B, Portales H, Saviot L, Lerme J, Prevel B, Pellarin M, Duval E, Perez A, Broyer M. Quadrupolar vibrational mode of silver clusters from plasmon-assisted Raman scattering. Phys Rev B. 1999;60:17107–11. Petri A, Pietronero L. Multifractal nature of fractons on the percolating cluster. Phys Rev B. 1992;45:12864–72. Pilla O, Cunsolo A, Fontana A, Masciovecchio C, Monaco G, Montagna M, Ruocco G, Scopigno T, Sette F. Nature of the short wavelength excitations in vitreous silica: an X-ray Brillouin scattering study. Phys Rev Lett. 2000;85:2136–9. Pilla O, Caponi S, Fontana A, Montagna M, Righetti L, Rossi F, Viliani G, Ruocco G, Monaco G, Sette F, Verbeni R, Cicognani G, Dianoux AJ. X-ray and neutron scattering studies in vitreous

Characterization of Sol–Gel Materials by Raman and Brillouin. . .

31

silica: acoustic nature of vibrational dynamics in the mesoscopic range. Philos Mag B. 2002;82:223–32. Pucker G, Parolin S, Moser E, Montagna M, Ferrari M, Del Longo L. Raman and luminescence studies of Tb3+ doped monolithic silica xerogels. Spectrochim Acta A. 1998;54:2133–42. Raman CV. A change of wavelength in light scattering. Nature. 1928;121:619. Rammal R, Toulouse GJ. Random walks on fractal structures and percolation clusters. J Phys Lett. 1983;44:L13–22. Richter H, Wang ZP, Ley L. The one phonon Raman spectrum in microcrystalline silicon. Solid State Commun. 1981;39:625–9. Rousset JL, Boukenter A, Champagnon B, Dumas J, Duval E, Quinson JF, Serughetti J. Antigranulocytes structure and fractal domains of silica aerogels. J Phys Condens Matter. 1990;2:8445–55. Ruocco G, Sette F. High-frequency vibrational dynamics in glasses. J Phys Condens Matter. 2001;13:9141–64. Sandercock JR. Brillouin scattering study of SbSI using a double-passed, stabilised scanning interferometer. Opt Commun. 1970;2:73–6. Sandercock JR. Simple stabilization scheme for maintenance of mirror alignment in a scanning Fabry–Perot interferometer. J Phys E. 1976;9:566–9. Sandercock JR. Light scattering from surface acoustic phonons in metal and semiconductors. Solid State Commun. 1978;26:547–51. Saviot L, Champagnon B, Duval E, Ekimov AI. Size-selective resonant Raman scattering in CdS doped glasses. Phys Rev B. 1998;57:341–6. Schaefer DW, Keefer KD. Structure of random porous materials: silica aerogel. Phys Rev Lett. 1986;56:2199–202. Schaefer DW, Brinker CJ, Richter D, Farago B, Frick B. Dynamics of weakly connected solids: silica aerogels. Phys Rev Lett. 1990;64:2316–9. Scopigno T, Balucani U, Ruocco G, Sette F. Inelastic X-ray scattering and the high-frequency dynamics of disordered systems. Phys B. 2002;318:341–9. Shiang JJ, Risbud SH, Alivisatos AP. Resonance Raman studies of the ground and lowest electronic excited state in CdS nanocrystals. J Chem Phys. 1993;98:8432–42. Shuker R, Gammon RW. Raman-scattering selection-rule breaking and the density of states in amorphous materials. Phys Rev Lett. 1970;25:222–5. Slater Joseph B, Tedesco Jams M, Fairchild Ronald C, Lewis Ian R. Raman spectrometry and its adaptation to the industrial environment. In: Lewis IR, Edwards HGM, editors. Handbook of Raman spectroscopy: from the research laboratory to the process line. New York: Marcel Dekker Inc; 2001. Stoll E, Kolb M, Courtens E. Numerical verification of scaling for scattering from fractons. Phys Rev Lett. 1992;16:2472–5. Strohhöfer C, Fick J, Vasconcelos HC, Almeida RM. Active optical properties of Er-containing crystallites in sol–gel derived glass films. J Non-Cryst Solids. 1988;226:182–91. Sussner H, Vacher R. High precision measurements of Brillouin scattering frequencies. Appl Opt. 1979;18:3815–8. Terki F, Pilliez JN, Woignier T, Pelous J, Fontana A, Rossi F, Montagna M, Ferrari M, Cicognani C, Dianoux AJ. Low-frequency light scattering in silica xerogels: influence of the heat treatment. Philos Mag B. 1999;79:2081–9. Tikhomirov VK, Furniss D, Seddon AB, Reaney IM, Beggiora M, Ferrari M, Montagna M, Rolli R. Fabrication and characterization of nanoscale, Er3+-doped, ultratransparent oxy-fluoride glass ceramics. Appl Phys Lett. 2002;8:1937–9. Tiong KK, Amirtharaj PM, Pollak FH, Aspnes DE. Effects of As+ ion implantation on the Raman spectra of GaAs: “spatial correlation” interpretation. Appl Phys Lett. 1984;44:122–4. Treado Patrick J, Nelson Matthew P. Raman imaging. In: Lewis IR, Edwards HGM, editors. Handbook of Raman spectroscopy: from the research laboratory to the process line. New York: Marcel Dekker Inc; 2001.

32

M. Montagna

Tsujimi Y, Courtens E, Pelous J, Vacher R. Raman-scattering measurements of acoustic superlocalization in silica aerogels. Phys Rev Lett. 1988;60:2757–60. Vacher R, Pelous J. Behavior of thermal phonons in amorphous media from 4 to 300 K. Phys Rev B. 1976;14:823–8. Vacher R, Sussner H, Schichfus MV. A fully stabilized Brillouin spectrometer with high contrast and high resolution. Rev Sci Instrum. 1980;51:288–91. Vacher R, Woignier T, Pelous J, Courtens E. Structure and self-similarity of silica aerogels. Phys Rev B. 1988;37:6500–3. Vacher R, Courtens E, Coddens G, Pelous J, Woignier T. Neutron-spectroscopy measurement of a fracton density of states. Phys Rev B. 1989;39:7384–7. Vacher R, Courtens E, Coddens G, Heidemann A, Tsujimi Y, Pelous J, Foret M. Crossovers in the density of states of fractal silica aerogels. Phys Rev Lett. 1990;65:1008–11. Vacher R, Pelous J, Courtens E. Mean free path of high-frequency acoustic excitations in glasses with application to vitreous silica. Phys Rev B. 1997;56:R481–4. Wachs IE. Raman spectroscopy of catalysts. In: Lewis IR, Edwards HGM, editors. Handbook of Raman spectroscopy: from the research laboratory to the process line. New York: Marcel Dekker Inc; 2001. Walrafen GE, Hokmabadi MS, Holmes NC, Nellis WJ, Henning S. Raman spectrum and structure of silica aerogel. J Chem Phys. 1985;82:2472–6. Walrafen GE, Hokmabadi MS, Holmes NC. Raman spectrum and structure of thermally treated silica aerogel. J Chem Phys. 1986;85:771–6. Zampedri L, Ferrari M, Armellini C, Visintainer F, Tosello C, Ronchin S, Rolli R, Montagna M, Chiasera A, Pelli S, Righini GC, Monteil A, Duverger C, Gonçalves RR. Erbium-activated silica–titania planar waveguides. J Sol–Gel Sci Technol. 2003;26:1033–6.

Surface Structure of Sol–Gel-Derived Materials Using X-ray Photoelectron Spectroscopy (XPS) Diane Holland

Abstract

The sol–gel technique is frequently used to prepare materials where surface properties are of major importance. These might be films, powders, high-surface area catalyst supports, and catalysts themselves. A surface-specific technique is therefore essential to determine quality and reproducibility of surfaces and to correlate properties to chemistry and structure. X-ray photoelectron spectroscopy has become a routine technique for many workers in this field, and this chapter serves to inform the sol–gel community about the benefits and pitfalls of this structural and analytical tool. Areas of application are described with some examples from the literature.

Contents Principles of XPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Arrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inelastic Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surface Specificity and Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantitative Analysis Using XPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peak Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Additional Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Auger Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oxidation States of Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 3 4 5 6 6 9 9 11 11 12 12 13 14

D. Holland (*) Department of Physics, University of Warwick, Coventry, UK e-mail: [email protected] # Springer International Publishing AG 2017 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_35-1

1

2

D. Holland

Coordination Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Neighbors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surface Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15 17 17 18 21 21

Principles of XPS In the 1960s, Siegbahn et al. (1967) showed that the binding energies of core level electrons exhibited small, environment-specific, chemical shifts. Measurement of these binding energies enabled the development of a structural analytical probe which they termed Electron Spectroscopy for Chemical Analysis (ESCA). X-ray photoelectron spectroscopy (XPS) is one variant of ESCA where X-ray photons are used to induce the emission of electrons from atoms (Fig. 1a), at or near the surface of a solid, and the kinetic energies of these photoelectrons are measured. If the X-ray photon is of energy hv and the measured kinetic energy of the photoelectron is Ek, then we can deduce that the electron was originally held to the atom with binding energy Eb, i.e., Eb ¼ hv  Ek

(1)

This ability to measure electron-binding energies in the various atomic energy levels (core and valence) within an atom in a solid can give us much information about that solid since Eb will be very sensitive to changes in the environment of the atom. As well as being characteristic of the particular element, it can also indicate its coordination number and oxidation state and the nature of the nearest neighbors. The excited state resulting from the photoemission process consists of a core hole. This hole can be filled by an electron from a higher energy level, with the energy balance being achieved by X-ray emission (fluorescence – a minor event at these energies) or through Auger emission of a less tightly bound electron. In Fig. 1b, an L1-shell electron moves to fill a K-shell hole, and an L2,3-shell electron is emitted with kinetic energy EKLL ¼ EK  EL1  EL2, 3 þ δ

(2)

where EK, EL1, and EL2,3 are the binding energies of the K, L1, and L2,3 electron shells, respectively, and δ is the modification arising from the charge on the ion and relaxation effects. Note that EKLL is independent of X-ray photon energy.

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

3

Fig. 1 Schematics of the process of generating (a) a photoelectron (K or 1s) and (b) a KLL Auger electron

Experimental Arrangement Equipment A schematic representation of a conventional XPS spectrometer is shown in Fig. 2. An X-ray beam (preferably monochromated) is focused, by means of slits, onto the sample surface, and the photoelectrons emitted are collected by an electron energy analyzer such that a display of signal intensity versus energy is obtained. Although it is electron kinetic energy which is measured, equipment software usually outputs the data in the form of intensity versus binding energy, having employed Eq. (3), in which Eq. (1) is modified to take account of the various work functions (specimen, spectrometer) encountered during emission of the photoelectron and which are specific to a given spectrometer: Eb ¼ hv  Ek þ ϕ

(3)

When studying insulating surfaces, there is an additional static charge which depends on the sample and its surface since this is no longer in electrical contact with the spectrometer. The details of this are discussed later (section “Sample Charging”), but one consequence is that each sample will have an unknown contribution to Eb which can only be eliminated by the use of charge referencing – the inclusion of a photoelectron peak in the spectrum whose energy is known absolutely. The most common method of charge referencing takes advantage of the carbon impurity deposited on the sample surface from the vacuum. The C 1s photoelectron peak is assigned an energy of 284.6 eV, and any difference from the spectrum value is then used to correct the energies of the other photoelectron peaks in the spectrum. This

4

D. Holland

Fig. 2 Schematic diagram of the operation of an XPS spectrometer (based on the Scienta ESCA 300)

does assume that the carbon deposit is uniformly in electrical contact with the sample surface and that the static charge is independent of photoelectron energy, i.e., is constant across the entire spectrum. Ultra high vacuum (UHV) conditions are essential in the analysis chamber to reduce the scattering of the photoelectrons by gas molecules which would prevent them reaching the analyzer. In addition, the escape depth for electrons in the range 0–1500 eV is very short (see section “Surface Specificity and Sample Preparation”), and thus only the first few surface atomic layers contribute, making it necessary to avoid contamination of the analyzed surface during the experiment. The residual gases found in an ultrahigh vacuum system are mainly O2, CO, CO2, H2O, and N2. At a pressure of 3  1010 mbar, one monolayer of contamination is formed in approximately 3–4 h.

Nomenclature The photoelectron peaks are referred to by the electron energy state, usually as described by the j–j coupling scheme, giving the label nlj, e.g., 2p1/2 and 4d3/2. A different convention is used for Auger electrons such that the three different

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

5

Fig. 3 A wide-region “survey” scan from SnO2 powder, showing peaks due to photoelectrons and Auger electrons

energy levels, described using X-ray nomenclature, involved in the transition, are used in the label. For example, Fig. 1b shows the emission of a KLL Auger electron resulting from the transition of an L electron to fill a K-shell vacancy (formed by the normal photoemission process) and the removal of the remaining excitation energy by emission of another L electron. Figure 3 shows an XPS spectrum from SnO2 powder obtained using Al Kα radiation and covering the entire energy range. The various photoelectron and Auger peaks are identified.

Inelastic Processes The major peaks in the spectrum in Fig. 3 can be seen to be associated with step-like features to the high-binding energy side. These are due to photoelectrons which take part in inelastic processes before they finally leave the surface. The associated energy losses mean that the kinetic energies of the photoelectrons are reduced which gives an apparent increase in binding energy. The intensity resulting from these inelastic processes can frequently provide a very steep background to the photoelectron peak which must be removed before peak-fitting routines can be applied (section “Peak Fitting”).

6

D. Holland

Surface Specificity and Sample Preparation Use of conventional X-ray sources to induce photoemission produces electrons with energies in the range ~0–1500 eV. These interact strongly with the sample atoms, having a high probability of inelastic collision and therefore very short inelastic mean free paths of the order of a few nanometers. The escape depth can be defined as the distance, normal to the surface, over which the chance of an electron escaping the solid without inelastic scattering is 1/e. Figure 4 gives an indication of the variation of escape depth with electron energy – an experimentally derived dependence (Seah and Dench 1979). In the range of conventional XPS photoelectron energies, it can be seen that it is the first 1–3 nm of surface which are analyzed; hence, the sensitivity of this technique to the near surface. This sensitivity has implications for preparation of the surface to be examined. It is very easy for the features of interest to be obscured by deposited impurities or by the consequence of surface reactions. If the bulk properties of a sample are of interest, then a suitable surface for examination can be prepared by fracture of a bulk sample under UHV conditions. Some surfaces can be prepared in a controlled environment to avoid contamination and transferred directly to the spectrometer analysis chamber or preparation chamber. Even at the low background pressures in the analysis chamber, gas molecules will adsorb onto and react with surfaces, particularly the often highly reactive surfaces formed by sol–gel. Kim et al. recorded changes to the O 1s spectrum from a reduced NiCo2O4 sample as a function of time exposed under UHV conditions. The changes to the spectrum (Fig. 5) result from reaction with atmospheric H2O and CO. The fit to the O 1s spectrum shows increased peak intensity at 531.5 and 533.0 eV. Cobalt and nickel hydroxides have reported binding energies of 531.7 eV, and the O 1s 533.0 eV peak is consistent with chemisorbed CO and coincides with the appearance of a corresponding peak in the carbon 1s spectrum at 285.2 eV (Kim et al. 2000). If samples are loaded as prepared then due consideration must be given to atmosphere induced changes. Ion etching can be used to remove surface contamination though this can lead to problems of selective sputtering and contamination of the surrounding area. Brenier et al. showed that Ar+ ion sputtering of sol–gel ZrO2 films at 5 keV preferentially removed C and O from the films (Fig. 6) and also resulted in a shift of the Zr 3d5/2 peak as a result of the changes in surface charging characteristics of the defect films (Brenier et al. 1999).

Quantitative Analysis Using XPS One of the frequent uses made of XPS is for compositional analysis of a surface – either to check stoichiometry or to follow changes in the surface during some process. The intensity I of a given photoelectron peak from an element in a sample is proportional to several variables:

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

7

Fig. 4 Escape depth as a function of electron energy

– The X-ray flux – The area of sample corresponding to the volume from which the detected electrons originate – N – the number of atoms of the element in that same volume – σ – the photoionization cross section for the electron energy level at the X-ray energy used – The take-off angle – The attenuation length of the photoelectrons (escape depth corrected for angle) – Other loss mechanisms – Detector efficiency at the photoelectron energy The product of all factors, other than N, is referred to as the sensitivity factor S for the photoevent and we can write I / SN:

(4)

S can be determined experimentally, using standards measured under identical conditions to the unknown sample. Alternatively, if a comparison is being made of two species in a surface using the same spectrum, most of the factors controlling intensity will be constant (if the two photoelectron peaks are close in energy) and the equation can be written as I / σ N:

(5)

Values for σ have been tabulated by Yeh and Lindau (1985). Modern XPS spectrometers are provided with analysis software which includes values for the sensitivity factors for different spectrometer arrangements.

8 Fig. 5 O 1s XPS spectra from a NiCo2O4 sample: (a) after reduction and after (b) 12 h, (c) 24 h, and (d) 48 h under UHV conditions. The arrows show the positions of the three peaks used for fitting the spectrum in the lower panel. A linear background subtraction was used (Kim et al. 2000).

Fig. 6 O/Zr elemental ratios obtained by Ar+ ion sputtering of sol–gel ZrO2 films. Values are shown for different heat treatments and sputtering times. The values for zero sputtering time were obtained by Rutherford backscattering (Brenier et al. 1999)

D. Holland

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

9

Peak Fitting When peaks from different elements or different oxidation states overlap, then a peak fitting procedure is required before accurate binding energies and relative quantities can be measured. The first requirement of the procedure (after shift referencing has been performed) is removal of the background contribution. The background mostly arises from the inelastically scattered electrons such that the photoelectron peaks sit on an edge. The most popular procedure for modeling this edge is due to Shirley (1972). The function applied performs a subtraction which is proportional to the intensity of the peak at that point in binding energy. This is illustrated in Fig. 7, where a Shirley background is plotted for an Fe 3p peak containing contributions from Fe3+ and Fe2+ (Gee 2000). The line shape of the photoelectron peaks is pseudo-voigt, i.e., a Lorentzian line with Gaussian broadening and can be fitted with a function such as "

2 w þ ð1  μÞ y¼A μ π 4ðx  xc Þ2 þ w2

rffiffiffiffiffiffiffiffiffiffiffi  # 4 ln 2 4 ln 2 2 exp  2 ðx  xc Þ πw2 w

(6)

where A is the amplitude of the peak, xc is its center, w is the full width at half maximum, and μ is the fraction of Lorentzian contribution.

Additional Structure The processes of photoelectron emission and Auger electron emission can result in the generation of more than a single peak for each event. Additional structure is generated by a variety of processes, and it is essential to recognize that this can happen so that peaks are not misidentified. 1. X-ray satellites. If the X-ray source is not monochromated, then photoelectron peaks produced by X-ray photons of wavelength corresponding to other characteristic X-ray lines can be observed These are generally very weak but can be the source of otherwise unidentified peaks. Fig. 8 shows part of a spectrum from a sol–gel alumina matrix obtained using non-monochromated Mg Kα radiation (Wannaparhun et al. 2002). The X-ray satellite peaks from the Al 2s and 2p are indicated. 2. Multiplet splitting of core level peaks can occur when the photoemitting atom has unpaired valence electrons. For example, consider the photoemission of an electron from an s-level. After photoionization, the remaining unpaired electron will either have spin parallel or antiparallel to the unpaired valence electrons. This results in two slightly different final state energies, and hence the s-level photoelectron peak will be split into two components. This splitting becomes more complicated for orbitals of higher angular momentum, and, since the magnitudes of the splittings are too small to resolve, this can lead to broadened, asymmetric line shapes.

10

D. Holland

Fig. 7 Example of peak fitting for an Fe 3p peak containing contributions from Fe3+ and Fe2+ (Gee 2000)

AI (2p) AI (2s)

a

140

120

b

d

c

100

80

60

40

20

0

B . E. (eV)

Fig. 8 XPS survey scan of a sol–gel alumina matrix illustrating the presence of X-ray satellite peaks. The scan was obtained using non-monochromated Mg Kα radiation; (a) first satellite of Al 2s; (b) Si 2p region; (c) second satellite of Al 2s; (d) first satellite of Al 2p (Wannaparhun et al. 2002)

3. Shake-up and shake-off satellites. Photoemission of a core level electron results in increased effective nuclear charge Zeff being experienced by the valence electrons. The consequent electron screening rearrangement may enable one of the valence electrons to move to a higher valence level (shake-up) or leave the sample altogether (shake-off). This results in less kinetic energy being available to the original photoelectron, yielding low-intensity structure (in the case of shake-up satellites) at higher binding energies than the main photoelectron peak. Some transition metal and rare earth ions which have unpaired 3d or 4f electrons show

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

11

very pronounced satellite structure. For example, Cu2+ (3d10) shows no shake-up satellites, whereas Cu+ (3d9) exhibits strong shake-up satellites (Frost et al. 1972). This behavior can be used to distinguish the oxidation state(s) of elements in samples. An example of shake-up satellites associated with Fe 2p photoelectrons is shown in Fig. 9 (Gee 2000).

Auger Parameters Auger electron peaks are characteristic of the element and can be used to infer chemical state information. However, more information can be obtained by combining Auger and photoelectron data to obtain the Auger parameter, α (Wagner 1975a, b), defined as α ¼ Ek ðAugerÞ  Ek ðphotoelectronÞ

(7)

A change in the Auger parameter of an element is directly proportional to the polarizability of the electrons associated with that element (Wagner et al. 1982; Woodbridge et al. 1999). The usefulness of the Auger parameter for insulating materials is that, since it is an energy difference, between features from the same element in the same XPS spectrum, absolute binding energies and hence static charge referencing are not necessary. The modified Auger parameter α0 = α + hv (the photon energy) is frequently used. Campostrini et al. quote the unchanging value (1711.8 eV) of the Si Auger parameter as evidence that a silicate network is maintained in oxyfluoride silica gels (Campostrini et al. 2002).

Sample Charging The photoemission of electrons will inevitably result in the formation of a positively charged surface on insulating materials. This in turn reduces the kinetic energy of the photoelectron, leading to a shift of the observed spectrum on the binding energy scale. In addition, if the surface charge is nonuniform (differential charging) – for example, on a fragmented or heterogeneous surface – the spectrum obtained can be broadened and distorted and even split (Fig. 10). The surface charge can be moderated using a low-energy electron flood gun (electrons typically with 2–4 eV energy) (Hunt et al. 1980). However, this may not eliminate the effects of differential charging, and care must be taken to avoid false interpretation of a spectrum such as that in Fig. 10 in terms of multiple sites. Examination of other photoelectron peaks in the spectrum will generally confirm if there is peak distortion because of differential charging.

12

D. Holland

Fig. 9 Photoelectron peaks from sodium iron silicate glasses 0.3Na2OxFe2O3 (0.7 – x)SiO2: (a) Fe 2p and (b) Fe 3p (Gee 2000)

Applications Capabilities XPS peaks can be detected for all elements except hydrogen and helium, where the photoionization cross sections for the 1s electrons are extremely small. This might be considered a drawback since hydrogen is an element of considerable significance for sol–gel prepared materials and surfaces in general. However, the presence of H can often be inferred from the changes to the binding energies of elements to which it is bonded. In addition to compositional analysis of the surface, the features of a photoelectron spectrum, particularly the binding energies, can be used to determine oxidation states, coordination numbers, and the nature of the nearest neighbor atoms. Some examples of this can be seen by examining some of the applications of XPS to sol–gel materials.

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

13

Fig. 10 Multiple spectra produced by a fractured sample in which the different segments charge differentially. The sets of arrows indicate the individual spectra

Catalysts The surface sensitivity of the technique is of great value here, and it is commonly used to detect changes to a surface during a catalyzed process. Because of the UHV requirements of the technique, the process cannot be followed in real time, and the surface state has to be “frozen-in” at specific stages. Allowance must be made for the changes which can occur during this sampling stage. Mitome et al. examined lanthanide-doped Pd/TiO2 catalysts before and after exposure to oxygen at 500  C or H2 at 200  C. The samples were transferred from the reaction chamber to the analysis chamber without exposure to ambient atmosphere. Pd 3d5/2 and Pd 3d3/2 peaks of reduced Pd/TiO2, Ce–Pd/TiO2, and Gd–Pd/TiO2 catalysts at zero oxidation state were present at 334.1 and 339.5 eV, respectively. Completely oxidized catalysts gave peaks at 335.3 and 340.5 eV, respectively, due to PdO. On the reduced Pd/TiO2 catalyst, some Pd was also present as oxide. The extent of oxide conversion was then reflected in the efficiency for methane reduction of NO (Mitome et al. 1999).

14

D. Holland

Figure 11 shows the extent of formation of graphitic carbon on two NiO/γ-Al2O3 catalyst surfaces used for the partial oxidation of methane (POM) prepared by impregnation (IT) and sol–gel (ML) (Zhang et al. 2000). Much greater deposition of graphite, which results in pore closure, was seen for the impregnation sample. Pecchi et al. showed how the oxidation state of Pd in a sol–gel-derived Pd–SiO2 methane combustion catalyst changed through the cycle of sintering, reduction, and combustion (Pecchi et al. 1998). Schroeder et al. found that pre- and post-reaction XPS data indicated that MoO3–V2O5 catalyst surfaces were reduced by the 1,3-butadiene-in-air feeds. After reaction, approximately 20% of the intensity of the V 2p signal could be attributed to V4+ rather than V5+ (Schroeder et al. 2001).

Oxidation States of Species The process of oxidation, removal of a valence electron, means that the remaining electrons experience a greater effective nuclear charge, and thus binding energies are increased. For example, the Mo 3d binding energies for various oxidation states of molybdenum in an oxide environment have been reported by Barath et al. (1999) studying a reduced MoO3/α-Al2O3 catalyst surface (Table 1). The relative amounts of each oxidation state can be quantified, but the accuracy with which this can be done depends on the intensity of the particular photoelectron peak, balanced by the resolution of the contribution from the different oxidation states, and also the nature of the background at that point in the spectrum. For example, Fig. 9 shows Fe 2p and Fe 3p spectra from a series of sodium iron silicate glasses. The intensity of the Fe 2p photoelectron peak is much greater than that of the Fe 3p. However, the former lies on a pronounced edge from the inelastic scattering background, and there are also significant shake-up satellites present. This makes background subtraction and subsequent curve fitting rather inaccurate. Nasser showed that the oxidation state of Ti in BaTiO3 sol–gel thin films varies with film depth (as revealed by Ar+ ion sputtering) and heat treatment. Figure 12 shows a Ti 2p spectrum which contains contributions from Ti4+, Ti3+, and Ti2+, with the lower oxidation species increasing in intensity with length of sputtering (Nasser 2000). Kim et al. used vacuum heating to reduce the Co oxidation state in NiCo2O4 sol–gel films. The reduction of Co3+ to Co2+ could be seen from the appearance in the Co 2p spectrum of the satellite lines which are observed only for high-spin Co2+ in an octahedral site (Kim et al. 2000). Figure 13 shows the effect of oxidation on the S 2p spectrum from mesoporous, thiol functionalized silicas where sulfonic acid sites are required for the catalysis of esterification and condensation reactions (Wilson et al. 2002). The reduction of Rh(III) to Rh(0) is an essential step in the formation of Rh–SiO2 catalysts, and XPS was used to follow the process of salt decomposition and reduction for a range of rhodium precursors (Campostrini et al. 2000). Figure 14 shows the various oxidation states of copper which could be produced on CuO–SiO2 catalysts by reduction treatments. It also illustrates the use of a satellite peak to confirm the presence of Cu2+ (Diaz et al. 1999). Davydov et al. showed that the

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

15

Fig. 11 Formation of graphitic carbon during partial oxidation of methane on NiO/γ-Al2O3 catalyst surfaces prepared by (a) impregnation (IT) and (b) sol–gel (ML) (Zhang et al. 2000)

Table 1 Binding energies for the 3d5/2 and 3d3/2 electrons on molybdenum in various oxidation states (Barath et al. 1999) Oxidation state Binding energy (eV), 3d5/2/3d3/2

Mo0 227.9/231.1

Mo2+ 228.5/231.7

Mo4+ 229.7/232.9

Mo5+ 231.5/234.7

Mo6+ 232.6/235.8

relative amounts of CrIII and CrVI on titania-loaded MCM-41 were dependent on the titania concentration which also influenced the migration of Cr to the catalyst surface (Davydov et al. 2001). The oxidation state of catalysts can depend on the type of substrate being used, and Pouilloux et al. demonstrated that the Co and Sn oxidation states of a reduced CoSn catalyst depended on whether the catalyst was deposited on Al2O3, ZnO, or SiO2. Oxidation states remained higher on alumina (Pouilloux et al. 2000). The conditions of the sol–gel reactions will also influence the oxidation state of the final species. Noh et al. showed that a base catalyst in a sol–gel process is highly important in enhancing the thermal stability of active metal species by anchoring. Such catalysts had Pd 3d5/2 peaks in the range of 336.3–336.5 eV, assigned to PdO, whereas Pd–Al2O3 prepared in acidic conditions gave a Pd 3d5/2 peak at 335.1 eV and a shoulder at 337.3 eV which were ascribed to metallic Pd and less active PdO, respectively (Noh et al. 1999).

Coordination Number Change in coordination number has usually a much smaller effect than a change in oxidation state, but there are still measureable shifts in binding energies for some

Volume

Intensity (arb. units)

Fig. 12 Ti 2p spectra from BaTiO3 films annealed at 600  C for 2 h. Ar+ ion sputtering has been used to profile the films and show the change in oxidation state of the Ti species: (a) as-introduced surface; (b) after 5 min Ar+ sputtering; (c) after 20 min Ar+ sputtering; (d) after 40 min Ar+ sputtering (Nasser 2000)

D. Holland

a

Intensity (arb. units)

16

b

Ti 2p3/2 Ti+4 Ti 2p1/2 Ti+4

Ti+4 Ti+3

Ti+4 Ti+3

Ti+2

Ti+2

Intensity (arb. units)

c

Intensity (arb. units)

468 466 464 462 460 458 456 454 452

d

468 466 464 462 460 458 456 454 452 BINDING ENERGY , eV

elements. For example, literature reports show that there is a change in the binding energy of the Ti 2p3/2 photoelectron from about 458.5 eV for Ti in octahedral coordination (Galan-Fereres et al. 1995) to about 460 eV for Ti in tetrahedral coordination in the silica lattice (Stakheev et al. 1993). Using this information, Beck et al., studying TiO2–SiO2 epoxidation catalysts, showed that titanium underwent a change in coordination from tetrahedral to octahedral with increasing Ti content in the aerogel (Beck et al. 2001).

S 2p XP Signal

Fig. 13 S 2p XPS spectra from mercaptopropyl silica before and after oxidation. The inset shows the surface S content varies with bulk S content (Wilson et al. 2002)

Normalised Intensity

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

17

S 2p S 2p

UnOxidised

0

2

4

6

8

10

Bulk S loading (%)

S SOx Oxidised

170

166

162

158

Binding Energy (eV)

Neighbors The effect of neighboring atoms on the binding energies of electrons on an atom will depend on their distance and also on whether there is through bonding. Hence, nearest neighbors have the greatest effect, but the effects of next nearest neighbors can also be resolved. Almeida et al., investigating titano-phospho-silicate sol–gel films, could resolve the O 1s peak into contributions from Si–O–Si, Si–O–P, Si–O–Ti, and Ti–O–Ti (Fig. 15) The assignments were based on the electronegativities of the different elements involved such that the higher the ionic character of the bonds, the lower the O 1s binding energy. Thus the peak at 530.7 eV was assigned to Ti–O–Ti bonds; that at 531.5 eV to Si–O–Ti bonds; that at 533.8 eV to Si–O–Si bonds; and that at 534.4 eV, to Si–O–P bonds (Almeida et al. 1998).

Profiling In addition to its use as a cleaning tool, Ar+ ion etching is also used as a means of profiling through surface layers and films (Traversa et al. 2001; Armelao et al. 2000). XPS scans are taken after various etching times to build up elemental profiles through the film. Caution must be exercised to avoid misinterpretation of peaks which may result from reduction processes associated with the Ar+ etch itself, as well as the preferential sputtering referred to earlier. Yu et al. produced depth profiles of thin films of Pb-doped TiO2 using argon ions at an acceleration voltage of 5 keV

18

D. Holland

Fig. 14 Cu 2p3/2 core-level spectra from CuO–SiO2 sol–gel-derived catalysts showing the effect of reduction treatments. The shake-up satellite from the Cu2+ is indicated (Diaz et al. 1999)

Cuo / Cu +

Cu 2p3/2 Cu 2sa+

2+

Cu

Counts per second (au)

CSS8, vac

CSS4, vac

CSS8, red

CSS4, red 945

941

937

933

929

BE (eV)

and a beam current of 20–40 mA. The high-resolution XPS spectra from Ti 2p and Pb 4f showed an increase in the amount of reduced species (TiII and Pb0), compared to the oxidized species (TiIV and TiIII and PbIV and PbII), with increased length of time of etching (Fig. 16). They explain that, while this could be due to some reduction process involving carbon, the most probable cause is preferential sputtering of oxygen, i.e., reduction (Yu et al. 2002).

Surface Reactions Gels are also formed by reactions at surfaces, particularly of silicate glasses, where the exchange of mobile cations in the glass for H+ from solutions can result in a silica gel layer which is many tens of microns thick. The formation of such a layer is thought to be an essential precursor to developing pH response in certain glasses. This cation–proton exchange can be followed by XPS but only after the gel layer has

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

19

Fig. 15 O 1s XPS spectra from 80 SiO2 20 TiO2 XP2O5 films: (—) individual fitted peaks; (- - - -) sum of fitted peaks (Almeida et al. 1998)

Si-O-Si Si-O-Ti Si-O-P

Ti-O-Ti

80S20T15P

Intensity (a. u.)

80S20T10P

80S20T5P Si-O-Si

Ti-O-Ti Si-O-Ti 528

532

80S20T 536

Binding energy / eV

been dried sufficiently that it does not compromise the ultrahigh vacuum of the analysis chamber. The drying process generally produces the reaction 2Si  O  H ! Si  O  Si þ H2 O Figure 17 shows the change in the O 1s photoelectron peak from Corning 015 glass, a model pH system, after exposure to various environments. The surface obtained by fracture of the bulk glass in vacuo shows the typical spectrum of bridging oxygen (Si–O–Si) at high binding energy and non-bridging oxygen (Si–O R+) at low binding energy. Exposure to air already results in some cation–proton exchange and exposure to deionized water (pH < 7) removes most surface cations. However, exposure to 1M NaOH results in the reestablishment of some Si–O R+. The intensity of the non-bridging oxygen peak increases with time

20

D. Holland

Fig. 16 Ti 2p XPS spectra showing the effect of Ar+ ion sputtering on the oxidation of Ti in Pb-doped TiO2 films (Yu et al. 2002) (a) surface and after sputtering for (b) 300 s and (c) 600 s

TiIV2p3/2

a

TiIII2p3/2

Relative intensity / c-s–1

TiII2p3/2

TiIV2p3/2

TiII2p3/2

TiIII2p3/2

TiIV2p3/2 TiII2p3/2

b

c

TiIII2p3/2

452 454 456 458 460 462 464 466 468 Binding energy / eV

of exposure. This phenomenon could be the result of migration of Na+ into the gel layer but is more likely due to the dissolution of the gel layer into the alkaline solution, revealing underlying material of more bulk-like composition. Then and Pantano used XPS to characterize changes in the surface of an alkali lead silicate glass on reduction with hydrogen at ~500  C. The reduction in the intensity of the non-bridging oxygen O 1s peak was correlated with the formation of metallic lead which could evaporate from the surface leaving a silica rich layer (Then and Pantano 1990).

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

21

Fig. 17 O 1s spectra from Corning 015 after different periods of immersion in various electrolytes

Conclusions X-ray photoelectron spectroscopy (XPS) is only one of a battery of surface analytical techniques, but it has found regular application in helping to solve many of the chemical and structural questions which are raised by sol–gel science. Acknowledgments The author wishes to thank the Engineering and Sciences Research Council of the UK for providing access to the Scienta spectrometer in the RUSTI facility at Daresbury Laboratory, Warrington, UK. The invaluable help from Dr. Danny D-S Law and Dr. Graham Beamson of that facility is also much appreciated.

References Almeida RM, Vasconcelos HC, Goncalves MC, Santos LF. XPS and NEXAFS studies of rare-earth doped amorphous sol–gel films. J Non-Cryst Solids. 1998;232–4:65 Armelao L, Fabrizio M, Gross S, Martucci A, Tondello E. Molecularly interconnected SiO2-GeO2 thin films: sol–gel synthesis and characterization. J Mater Chem. 2000;10:1147. Barath F, Turki M, Keller V, Maire G. Catalytic activity of reduced MoO3/alpha-Al2O3 for hexanes reforming I. Preparation, characterization, and X-ray photoelectron spectroscopy studies. J Catal. 1999;185:1. Beck C, Mallat T, Burgi T, Baiker A. Nature of active sites in sol–gel TiO2-SiO2 epoxidation catalysts. J Catal. 2001;204:428. Brenier R, Mugnier J, Mirica E. XPS study of amorphous zirconium oxide films prepared by sol–gel. Appl Surf Sci. 1999;143:85. Campostrini R, Ischia M, Carturan G, Gialanella S, Armelao L. Rh Inclusion in sol–gel SiO2. Effects of Rh precursors on metal dispersion and SiO2-Rh thermal behavior. J Sol–Gel Sci Technol. 2000;18:61. Campostrini R, Ischia M, Carturan G, Armelao L. Sol–gel synthesis and pyrolysis study of oxyfluoride silica gels. J Sol–Gel Sci Technol. 2002;23:107.

22

D. Holland

Davydov L, Reddy EP, France P, Smirniotis PG. Transition-metal-substituted titania-loaded MCM41 as photocatalysts for the degradation of aqueous organics in visible light. J Catal. 2001;203:157. Diaz G, Perez-Hernandez R, Gomez-Cortes A, Benaissa M, Mariscal R, Fierro JLG. CuO–SiO2 sol–gel catalysts: characterization and catalytic properties for NO reduction. J Catal. 1999;187:1. Frost DC, McDowell CA, Woolsey IS. Evidence for multiplet splitting of 2p photoelectron lines of transition-metal complexes. Chem Phys Lett. 1972;17:320. Galan-Fereres M, Alemany LJ, Mariscal R, Banares MA, Anderson JA, Fierro JLG. Surface-acidity and properties of titania-silica catalysts. Chem Mater. 1995;7:1342. Gee IA. X-ray photoelectron spectroscopy studies of oxide based glass systems. PhD thesis, Warwick; 2000. Hunt CP, Anthony MT, Stoddart CTH, Seah MP. NPL Chem. Report 108; 1980. Kim JG, Pugmire DL, Battaglia D, Langell MA. Analysis of the NiCo2O4 spinel surface with auger and X-ray photoelectron spectroscopy. Appl Surf Sci. 2000;165:70. Mitome J, Aceves E, Ozkan US. Role of lanthanide elements on the catalytic behavior of supported Pd catalysts in the reduction of NO with methane. Catal Today. 1999;53:597. Nasser SA. X-ray photoelectron spectroscopy study on the composition and structure of BaTiO3 thin films deposited on silicon. Appl Surf Sci. 2000;157:14. Noh J, Yang OB, Kim DH, Woo SI. Characteristics of the Pd-only three-way catalysts prepared by sol–gel method. Catal Today. 1999;53:575. Pecchi G, Reyes P, Concha I, Fierro JLG. Methane combustion on Pd/SiO2 sol gel catalysts. J. Catal. 1998;179:309. Pouilloux Y, Autin F, Barrault J. Selective hydrogenation of methyl oleate into unsaturated alcohols: relationships between catalytic properties and composition of cobalt–tin catalysts. Catal Today. 2000;63:87. Schroeder WD, Fontenot CJ, Schrader GL. 1,3-Butadiene selective oxidation over VMoO catalysts: new insights into the reaction pathway. J Catal. 2001;203:382. Seah MP, Dench WA. Quantitative electron spectroscopy of surfaces: a standard data base for electron inelastic mean free paths in solids. Surf Interface Anal. 1979;1:2. Shirley DA. High-resolution x-ray photoemission spectrum of valence bands of gold. Phys Rev B. 1972;5:4709. Siegbahn K, Nordling CN, Fahlman A, Nordberg R, Hamrin K, Hedman J, Johanssen G, Bergmark T, Karlsson SE, Lindgren I, Lindberg B. ESCA; atomic, molecular and solid state structure studied by means of electron spectroscopy. Uppsala: Almquist and Wiksells; 1967. Stakheev AY, Shpiro ES, Apijok J. XPS and XAES study of Tio2-SiO2 mixed-oxide system. J Phys Chem. 1993;97:202. Then AM, Pantano CG. Formation and behavior of surface-layers on electron-emission glasses. J Non-Cryst Solids. 1990;120:178. Traversa E, Di Vona ML, Nunziante P, Licoccia S, Yoon JW, Sasaki T, Koshizaki N. Photoelectrochemical properties of sol–gel processed Ag-TiO2 nanocomposite thin films. J. Sol–gel Sci Technol. 2001;22:115. Wagner CD. Auger parameter in electron-spectroscopy for identification of chemical species. Anal Chem. 1975a;47:1201 Wagner CD. Chemical-shifts of auger lines, and auger parameter. Farad Discuss Chem Soc. 1975b;60:291. Wagner CD, Passoja DE, Hillery HF, Kinisky TG, Six HA, Jansen WT, Taylor JA. Auger and photoelectron line energy relationships in aluminum-oxygen and silicon-oxygen compounds. J Vac Sci Technol. 1982;21:933. Wannaparhun S, Seal S, Desai V. Surface chemistry of Nextel-720, alumina and Nextel-720/ alumina ceramic matrix composite (CMC) using XPS–A tool for nano-spectroscopy. Appl Surf Sci. 2002;185:183.

Surface Structure of Sol–Gel-Derived Materials Using X-ray. . .

23

Wilson K, Lee AF, Macquarrie DJ, Clark JH. Structure and reactivity of sol–gel sulphonic acid silicas. Appl Catal A Gen. 2002;228:127. Woodbridge CM, Gu XJ, Langell MA. Extra-atomic relaxation energies and auger parameters of titanium compounds. Surf Interface Anal. 1999;27:936. Yeh JJ, Lindau I. Atomic subshell photoionization cross sections and asymmetry parameters: 1 ⩽ Z ⩽ 103. At Nucl Data Tables. 1985;32:1. Yu JG, Yu JC, Cheng B, Zhao XJ. Photocatalytic activity and characterization of the sol–gel derived Pb-doped TiO2 thin films. J Sol–Gel Sci Technol. 2002;24:39. Zhang YH, Xiong GX, Sheng SS, Yang WS. Deactivation studies over NiO/g-Al2O3 catalysts for partial oxidation of methane to syngas. Catal Today. 2000;63:517.

Recommended Reading and Data Handbooks Briggs D, Seah MP. Practical surface analysis: Auger and X-ray photoelectron spectroscopy. New York: Wiley; 1990. Hochella MF. Chapter 13: Auger electron and X-ray photoelectron spectroscopies: in Reviews in mineralogy 18:. In: FCM H, editor. Spectroscopic methods in mineralogy and geology. 18 Washington, DC: Mineralogical Society of America; 1988. Moulder JF, Stickle WF, Sobol PE, Bomber KD. Photoelectron spectroscopy. Standard spectra for identification and interpretation of XPS data. In: Chastain J, editor. Handbook of X-ray photoelectron spectroscopy. Perkin-Elmer Corporation; 1992.

Structural Characterization of Hybrid Organic–Inorganic Materials Plinio Innocenzi, Giovanna Brusatin, Massimo Guglielmi, and Florence Babonneau

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Raman and FT-Raman Spectroscopies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Infrared Spectroscopies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Multinuclear Solid-State NMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Abstract

Organic–inorganic sol–gel hybrid materials (OIHM) possess a high variability in configurations and arrangements of the organic and inorganic components, whose interaction can be mainly regulated through different weak or strong ionocovalent bonds. They can be prepared employing organically modified alkoxides which contain one or more covalent bonds, not cleaved during the sol–gel process and whose organic groups modify the inorganic network or undergo polymerization. Also, organic–inorganic mesostructured and mesoporous hybrid materials P. Innocenzi (*) Laboratory of Materials Science and Nanotechnology, LMNT – D.A.D.U, University of Sassari and CR-INSTM, Alghero, Sassari, Italy e-mail: [email protected]; [email protected] G. Brusatin Department of Industrial Engineering, University of Padova, Padova, Italy e-mail: [email protected] M. Guglielmi University of Padova, Padova, Italy e-mail: [email protected] F. Babonneau l’Institut des Matériaux de Paris-Centre, Paris, France e-mail: fl[email protected] # Springer International Publishing AG 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_36-1

1

2

P. Innocenzi et al.

are synthesized by cooperative self-assembly in the presence of amphiphilic molecules acting as a supramolecular template. As these materials can have a broad possible structural configuration, the combined use of different analytical techniques is essential to given a quite deep understanding of their structure: solid state NMR spectroscopy and the vibrational spectroscopies will be considered hereafter, whose complementary use provided has been demonstrated highly significant to describe HOIM structure.

Introduction The definition of hybrid sol–gel material is quite broad, and different classifications have been proposed on the basis of their structural properties and interactions. A classification that is generally accepted has been proposed by Sanchez and Ribot (1994), who divided the organic–inorganic hybrid materials (OIHM) in two main classes. In the first group, “class I”, included are the OIHM whose interactions between the organic and inorganic parts are mainly regulated through hydrogen bonding, van der Waals forces. Porous oxides doped with organic dyes not grafted to the pore surface fall, for instance, in this first class of hybrids. In the second group of OIHM, the organic and inorganic components are strongly linked through ionocovalent bonds to form the “class II” of hybrids. This is a first broad division of the OIHM, but within each class, because of the so high variability in configurations and arrangements of the organic and inorganic components, large differences in the structure can be envisaged. It is important to observe more closely some of the OIHM included in class II. Hybrids of this class are generally prepared employing organically modified alkoxides which contain one or more covalent bonds that are not cleaved during the sol–gel process (Hoebbel et al. 1998; Schottner 2001). These alkoxides can be divided in two types: alkoxides bearing polymerizable organic functional groups, such as epoxides, methacrylates, and vinyl, or alkoxides whose organic groups are only modifying the inorganic network and don’t contribute to its cross-linking. This is very similar to that of inorganic glasses, where alkali ions act as network modifiers inducing network depolymerization. Another special class of nanostructured OIHM is obtained by sol–gel polycondensation of bridging silsesquioxane precursors, whose general formula is (RO)3–Si–R–Si–(OR)3 (Cerveau et al. 2001; Shea and Loy 2001). In class II, also included are ail the OIHM where organic dyes are covalently grafted to an inorganic oxide network. Mesostructured and mesoporous materials synthesized by cooperative self-assembly in the presence of amphiphilic molecules acting as a supramolecular template are also falling in the broad classification of OIHM and can both fall within class I or II. Interestingly, mesostructured OIHM show the same type of structural architectures that are observed in nonporous “classic” OIHM. Somehow, the basic strategy to obtain a hybrid material is the same, even if the synthesis route is apparently quite different. In this very large group of hybrid materials, with their such large possible structural configurations, the characterization of their structure is still an open

Structural Characterization of Hybrid Organic–Inorganic Materials

3

question. Important contributions have been received from some spectroscopic analytical techniques and from the extension to hybrids of some well-established characterizations, generally developed for oxides or organic polymers. In general, the combination of different types of characterization tools has revealed to be the winning approach to obtain a deeper understanding of the OIHM structure. In particular, hereafter will be considered solid-state NMR spectroscopy and the vibrational spectroscopies, whose complementary use provided very significant structural details to describe in depth the hybrid structure.

Raman and FT-Raman Spectroscopies Fourier transform Raman (FT-Raman) spectroscopy with near-infrared excitation has been widely applied to investigate the inorganic polycondensation and organic cross-linking reactions that occur during the synthesis of OIHM. Raman spectroscopy is generally applied in combination with Fourier transform infrared spectroscopy (FTIR), as a complementary technique, to investigate the structure of oxides and organic polymers. In sol–gel processing, Raman and FT-Raman spectroscopies have been largely used to study the reactions in solution and the structure in the solid state. The various steps, from precursor hydrolysis through condensation and film deposition to the production of optical components or protective coatings, have been investigated by Raman spectroscopy (Sassi et al. 2002, De Ferri et al. 2016). Cyclic species from the inorganic oxide component, in particular, which are hardly detectable by FTIR, can be more easily identified by Raman. Raman spectroscopy has also been applied to hybrid material characterization (Riegel et al. 1997) and can provide very useful information on the chemical structure of components and products (Harreld et al. 1997). A limit to the application of the technique is represented by the possibility to investigate the structure of coating layers, because of the low intrinsic sensitivity of Raman due to the small scattering cross-section. Some special Raman configuration must be adopted, such as confocal Raman spectroscopy (Baia et al. 2002) or waveguide Raman spectroscopy (WRS) (Urlacher and Mugnier 1996; Urlacher et al. 1997; Le Luyer et al. 2003), but this last possibility is restricted to light waveguiding films. In general, FT-Raman spectroscopy, because the excitation wavelength is longer (1050 nm), will not excite the large fluorescence that is typical of silica gels and is in general preferred to analyze OIHM. FT-Raman is particularly useful to characterize the structure, when the hybrid materials have been obtained employing organically modified alkoxides containing polymerizable organic functional groups. Some of the most popular molecules used to synthesize the most successful hybrid materials (Fig. 1) belong to this class of precursors. Examples are 3-glycidoxypropyltrimethoxysilane (GPTMS), with an epoxy group; 3-methacryloxypropyltrimethoxysilane (MPTMS), with a methacrylate function; and vinyltrimethoxysilane (VTMS), which contains vinyl groups (Fig. 1). It is important to follow the extent of the reactions involving these functional groups, both in the liquid and in the solid state.

4

P. Innocenzi et al.

Raman Intensity

1035

1256

643

Fig. 1 Examples of organically modified alkoxides bearing polymerizable functional groups

A

965 922 888

B

1500

1300

1100

900

700

500

300

100

Wavenumber / cm–1

Fig. 2 Raman spectra of GPTMS initially (A) and after 9 days (B) of reaction in NaOCH3/MeOH (From Riegel et al. 2002)

The identification of the epoxy ring is much easier by FT-Raman than by FTIR spectroscopy. The Raman epoxy ring breathing mode (~1256 cm
1) is, in fact, generally sharp and intense, and its assignment is generally straightforward; the opening of the epoxy ring can be easily followed in solution (Riegel et al. 1998) and in the solid state.

Structural Characterization of Hybrid Organic–Inorganic Materials

5

Table 1 Assignment of the Raman bands in GPTMS and EHTMS; SH = shoulder Wave number (cm
1) EHTMS 613 643 834 915 1132 1256/1262 1412 1481 (sh)

GPTMS 613 643 763 842/852 908 1135 1256 1414 1478

Mode v(SiO)? (SiO3) symmetric stretching v(COC) aliphatic Epoxy symmetric deformation in phase Epoxy antisymmetric deformation in phase Epoxy CH2 wagging Epoxy ring breathing Epoxy CH2 twist Epoxy CH2 bending

Adapted from Riegel et al. (2002)

The doublet at 643 and 611 cm
1 (see Fig. 2 and Table 1), assigned to v[Si–(O–CH3)3], is characteristic of non-hydrolized trimethoxysilane groups. The evolution of this band during hydrolysis gives a direct indication of the evolution of the reactions (Riegel et al. 1998). Another important precursor molecule which is very largely used in several applications of hybrids is 3-methacryloxypropyltrimethoxysilane (MPTMS). This organically modified alkoxide is characterized by a C=C double bond that can be opened to form an organic network; the cross-linking reactions are generally thermally or photoinduced by using proper initiators. FT-Raman can be used, in combination with FTIR, to follow the polymerization. The assignments of the most important modes in MPTMS are reported in Table 2. The doublet at 1700 and 1720 cm
1 (see Fig. 2 and Table 2) is assigned to carbonyl ester groups hydrogen bonded and non-hydrogen bonded to silanol groups, respectively, while the band at 1640 cm
1 provides indication of the degree of opening of the double carbon bond. Similarly to GPTMS, the doublet at 607 and 641 cm
1 can be used to follow the progress of the sol–gel reactions. The third example is given by the application of FT-Raman spectroscopy to the materials derived from organically modified alkoxides containing vinylic functional groups, such as vinyltriethoxysilane (VTES). The Raman spectrum of vinyltriethoxysilane shows a strong polarized band at 633 cm
1, due to in-phase stretching (breathing) of CSiO3. This mode progressively decreases and fïnally disappears, with completion of the hydrolysis (Gigant et al. 2003). Partially hydrolyzed species of VTES give bands in the 640–672 cm
1 region (Table 3). The silanol groups of these partially reacted species will form, by polycondensation reactions, dimers, trimers, and higher oligomers, whose Raman modes appear in the 596–460 cm
1 region (Table 3). Characteristic vinyl bands are observed at 1603 cm
1 (C=C stretching), 1410 cm
1 (CH2 = scissoring), and 1275 cm
1 (
CH= rocking).

6

P. Innocenzi et al.

Table 2 Attributions of Raman modes in MPTMS Wavenumber (cm
1) 607 641 1404 1455 1640 1700 1720

Mode v(Si–O–CH3)3 antisymmetric stretching v(Si–O–CH3)3 symmetric stretching δ(CH2) or δ(CH) δ(CH3) C=C stretching mode C=O hydrogen bonded to silanols C=O non-hydrogen bonded to silanols

Table 3 Assignment of the most relevant Raman modes of VTES hydrolysis products Wavenumber (cm
1) 1293 791 672 650 640 633 596 313

Attribution (
CH2–) twisting in ethoxy groups vasym, (CSiO3) in VTES vsym (CSiO3) in ViSi(OH)3 vsym (CSiO3) in ViSi(OEt)(OH)2 vsym (CSiO3) in ViSi(OEt)2(OH) vsym (CSiO3) in VTES vsym (CSiO3) in ViSi(OR)2(OSi) dimers δsym (CSiO3) deformation mode in VTES

From Gigant et al. (2003)

Infrared Spectroscopies Infrared spectroscopy has been one of the most extensively applied analytical tools to investigate the different stages of the reactions in the various steps of the sol–gel process (Yoshino et al. 1990). Several works have appeared on this subject, and a good knowledge on the relationship between IR spectra of sol–gel-derived oxides and their structure has been achieved (Almeida and Pantano 1990; Innocenzi 2003). Analysis of the IR spectra has given important indications on the structural evolution with temperature and on the influence of processing parameters on the oxide structure. This technique has been extended to hybrid materials and has been used, in combination with Raman and solid state NMR, to investigate the structure. One of the main advantages of this technique is the fast acquisition time and the possibility to collect the spectra from films, which is, on the contrary, more difficult with Raman and, at the moment, impossible by NMR. The sol-to-gel conversion in hybrid systems has been studied by FTIR (MendezVivar and Mendoza-Bandala 2000), for tetraethyl orthosilicate (TEOS)–alkyl alkoxysilane (RSi(OR03)) sols. The main assignments of methyltriethoxysilane (MTES)–TEOS systems in fresh sols are reported in Table 4. FTIR spectroscopy can be used, in particular, to follow the inorganic polymerization reactions and, when organic functional groups are present in the hybrids, to

Structural Characterization of Hybrid Organic–Inorganic Materials

7

Table 4 Assignment of the vibrational modes observed in MTES–TEOS fresh sols Wave number (cm
1) 2974 2892–2886 1456–1448 1389–1368 1275–1274 1190 1089–1080 881–880 798–794

Assignment C–H symmetric stretching C–H symmetric stretching CH2 deformation C–H bending Si–CH3 deformation Si–O–CH3 rocking C–O symmetric stretching in EtOH,
C–O– antisymmetric stretching in MTMS CH2 and CH3 in EtOH Si–O–C stretching

Adapted from Mendez-Vivar and Mendoza-Bandala (2000) The bands relative to Si–O modes have been omitted, while the bands due to residual ethanol are reported

observe the organic polymerization and the organic thermal degradation at different temperatures. The observation is generally done by monitoring the variation in intensity with the temperature of a reference vibrational mode. An example is reported by Almeida et al. (Maia et al. 2000), who studied the densification of hybrid silica–titania sol–gel films (methyltriethoxysilane–titania), by ellipsometry and FTIR. The 1275 cm
1 band of Si–CH3 was used to observe the thermal evolution of the structure. The residual Si–CH3 groups, whose presence has been identified in the hybrid silica–titania films until a temperature near 500  C, were shown to transform into Si–OH (silanol) groups as they were eliminated, following the reaction: Si
CH3 + 2O2 !  Si
OH + H2O " + CO2 ". Also, thermal evolution of bridged silsesquioxanes derived hybrids studied by FTIR in combination of ellipsometry allowed to engineer completely inorganic porous thin films of low refractive index, with distinctiveness of extremely controlled and uniform dispersion of the porogen at a molecular level (Brigo et al. 2016). As mentioned, another important application of FTIR spectroscopy is the observation of the organic polymerization reactions, even if the technique cannot give a real information on the extent of polymerization (the length of the organic chains), but it is only a direct observation of the reactions, for instance, the opening of an epoxy ring or the saturation of a C=C bond. Coupling this analysis with solid-state NMR is therefore necessary, in order to obtain a clearer picture of the structure upon polymerization of the organic groups. 3-Methacryloxypropyltrimethoxysilane (MPTMS) is one of the most important organically modifïed alkoxides used in the synthesis of OIHM. The methacryloxy functional groups allow the growth of polymethacrylate chains directly bonded to the silica backbone, formed through polycondensation reactions of alkoxy groups. FTIR spectroscopy has been used to follow directly the degree of organic polymerization through the changes in intensity of the =CH2 vibrational mode (Li and King 1996; Medda et al. 2003; Soppera et al. 2001). This is a simple and effective method to be used in films, even if, as previously observed, it does not give a true information on

8

P. Innocenzi et al.

Table 5 Assignments of the vibrational modes in MPTMS-derived hybrids Wave number (cm
1) 1719 1703 1730

Vibrational mode C=O stretching C=O C=O

1637 1440–1410

C=C CH2 wagging =CH2 deformation

1327, 1300, 1180 1300–1250 1200 939 847 816

Note Conjugated with C=C Conjugated with C=C and hydrogen bonded Non-conjugated with C=C and non-hydrogen bonded

Skeletal vibration from acetate groups Skeletal vibration from acetates coupled with methacrylate unsaturation Si–CH2 =CH2 wagging Si–C =CH2 twisting

After polymerization (methacryloxy unsaturation)

Adapted from Medda et al. (2003)

the degree of polymerization but only on the concentration of the unsaturated C=C bond within the material. In situ FTIR spectroscopy has been also used (Soppera et al. 2001), with the advantage of following directly the changes in the vibrational modes during UV photopolymerization. Table 5 reports the assignments of the main absorption bands in OIHM prepared from MPTMS (Medda et al. 2003). To evaluate the conversion of acrylate double bonds, the changes in intensity of the v(C=C) absorption band at ~1637 cm
1 and v(C=O) vibration at ~1719 cm
1 with the curing treatments are generally monitored. Upon curing a decrease in the v(C=C) band, intensity is followed by a decrease in intensity and broadening of the C=O band, due to the loss of conjugation with C=C with the increase of the polymerization. An example is reported by Innocenzi et al. (2003) for hybrid materials synthesized from MPTMS and 3-aminopropyltriethoxysilane (APTES). The changes in the shape and intensity of the absorption spectra, in the range 1600–1850 cm
1, during thermal curing, are shown in Fig. 3. The FTIR spectra after curing at 25 and 140  C were simulated using three Gaussian components, labeled 1, 2, and 3 in Fig. 4. Band 3 (~1700 cm
1) is assigned to carbonyl groups which are hydrogen bonded to silanols (Li and King 1996), and this mode is only weakly detected in the 140  C cured films. Component 2 is assigned to stretching vibrations of C=O groups that are conjugated to C=C double bonds; this appears as the main component in the 25  C films before thermal curing (see Table 5). Component 1 is assigned to the vibrations of C=O groups located in a more cross-linked structure, as revealed by the shift to higher wave numbers, plus the broadening and decrease in intensity of the band. Bands 1 and 2 are, therefore, correlated to the same C=O mode, but in two different local structural environments. The application of this technique yields detailed information on the changes due to thermal or UV curing in MPTMS hybrids (Innocenzi and Brusatin 2004). An

Structural Characterization of Hybrid Organic–Inorganic Materials

9

Fig. 3 3D image of the FTIR absorption spectra, in the range 1850–1600 cm
1, of MPTMS and 3-aminopropyltriethoxysilane hybrid films thermally cured at different temperatures (From Innocenzi et al. 2003, with permission of the American Chemical Society)

Fig. 4 Gaussian deconvolution of FTIR absorption spectra of MPTMS and 3-aminopropyltriethoxysilane hybrid films thermally cured at 25  C (a) and 140  C (b) (From Innocenzi et al. 2003, with permission of the American Chemical Society)

advantage is also that FTIR spectra give simultaneous information on the state of the organic polymerization reactions together with inorganic polycondensation (Innocenzi 2003), and a broader picture of the structure of the material can be obtained with a few FTIR spectra.

10

P. Innocenzi et al.

Another important organic reaction that must be monitored during the synthesis of hybrid materials containing polymerizable functionalities is the opening of epoxy rings. The most popular organically modified alkoxide bearing an epoxide is 3-glycidoxypropyltrimethoxysilane (GPTMS); several of the reported applications of hybrid materials are based on this compound. The opening reaction is generally more complicated to control and to follow than the polymerization of acrylate groups in MPTMS, because several reactions are possible during sol–gel processing of this class of hybrid materials, as shown further below. The epoxides have characteristic IR absorption bands (Grasselli and Ritchey 1975) at 1260–1240 cm
1 (ring breathing), 950–810 cm
1 (antisymmetrical ring stretching), 865–785 cm
1, and at 3050–2995 cm
1, due to the C–H stretch in the epoxide. A decrease in intensity of these bands is an indication of the opening of the epoxide and can be observed under different polymerization conditions in the sol (Brusatin et al. 2001; Innocenzi et al. 2001) and/or during radiation polymerization in solid state (Brusatin et al. 2006, 2008) or curing with cross-linking moieties as amine (Brusatin et al. 2003, Innocenzi et al. 2005). A combination of different techniques is, however, necessary for the exact identification of the reactions (Innocenzi et al. 1999, 2000, 2001). The method has shown to be reliable to obtain semiquantitative evaluations of the residual epoxy groups within the hybrid material, even if the results are very much dependent on the processing of FTIR spectra: normalization, quality of deconvolutions, and choice of the baseline. An example is reported in Fig. 5 (from Innocenzi et al. 2000), where the C–H epoxy stretching vibration around 3050 cm
1 is used to evaluate the influence of increasing amounts of BF3 on the epoxy opening.

Fig. 5 FTIR normalized spectra of hybrid samples from GPTMS–TMOS–BF3 with an increasing amount of BF3. The samples are indicated with a number to express the content of BF3 (molar ratio with respect to SiO2). GTB1 (a), GTB2 (b), GTB3 (c), GTB5 (d ), and GTB10 (e) (C–H epoxy stretch, around 3060 cm
1). When a larger amount of BF3 is added, spectra (e) and (d ), the epoxy ring result completely opened (From Innocenzi et al. 2000; with permission of the American Chemical Society)

Structural Characterization of Hybrid Organic–Inorganic Materials

11

Table 6 Assignment of the C–H, C–C and Sl–C modes in TMOS–VTMS (30–70) and TEOS–VTES (50–50) hybrid systems Wave number (cm
1) TMOS–VTMS 3059 3024 2975 2956 ~1939 1597 – 1409 1272 1000 577

TEOS–VTES 3060 3024 2977 – ~1951 1604 1452 1408 1275 1000 581

Vibrational mode vas(CH2) Second overtone of (C–H) t.o. wagging v(C–H) vs(C–H2) First overtone of (C–H) t.o. wagging vs(C=C) 2v(Si–C) δ(CH2) in plane δ(C–H) (C–H) trans oop wagging C=CH2 twist

Adapted from Ou and Seddon (1997)

Other authors selected different modes (Lee et al. 2003), for instance, the 910 cm
1 absorption band. The quality of the analysis will depend on the extent of overlap between the various bands. A detailed IR characterization of phenyl and vinyl OIHM has been done by Ou and Seddon (Ou and Seddon 1997). In particular, vibrational modes in tetramethoxysilane (TMOS)–vinyltrimethoxysilane (VTMS) and, comparatively, in TEOS–vinyltriethoxysilane (VTES), have been fully assigned (Table 6). In the case of OIHM derived from phenyltriethoxysilane (PhTES) and TEOS (Table 7), the C–H vibrations result only weakly shifted with respect to the values reported in the literature for organic molecules with an alkyl chain replacing the silicon, R–CH=CH2 instead of Si–CH=CH2 (Ou and Seddon 1997). The analysis has been extended to the near-infrared (NIR) range, with interesting results, because the combination and overtone bands in the NIR leave a free window around the telecommunication wavelengths 1.3 and 1.55 μm (Ou and Seddon 1997). Concerning phenyl, their degradation with radiation exposure is properly monitored by FTIR, a key reaction alternative to polymerization used to densify OIHM films and to produce inorganic oxide structure avoiding the need of thermal curing. Examples are reported on hybrid compositions based on phenyl-modified alkoxides (Falcaro et al. 2011; Brigo et al. 2011; Della et al. 2015, Zanchetta et al. 2013).

Multinuclear Solid-State NMR Vibrational spectroscopies, such as FTIR and FT-Raman, represent important tools to investigate the structure of OIHM. As we have described in the previous paragraphs, the application of these techniques is very effective to identify some of the reactions and their extent, such as the epoxy opening, but show some limitations in the investigation of the local structure. Multinuclear NMR spectroscopy is another

12

P. Innocenzi et al.

Table 7 Assignment of the C–H and C–C bands in the TEOS–PHTES hybrids Wavenumber (cm
1) 3059 3047 2971 2000–1650 1590

Vibrational mode v(C–H) v(C–H) v(C–H) Summation bands from oop C–H bend bands at 1000–700 cm
1 In-plane ring stretching

1566

Ring stretching

Si

1486

Ring stretching with some C–H bend coupling

Si

1423

Ring stretching with some C–H bend coupling 2v(Si–C) 2 (oop ring bend)

1391

1070

In-plane δ(C–H) Ring stretching mixed with ring bend and Si–C stretching Ring stretching with some C–H bend coupling

1030

Ring stretching with some C–H bend coupling

736

In-phase, oop 5 adjacent H wagging

1282 1117

Structural unit C=C=H C=C=H C=C=H C=C=H Si

Overtone of 696 band C=C–H

Si Si H

H

H

H

Si

696

Oop ring bend

Si

490

Oop ring bend

Si

H

Adapted from Ou and Seddon (1997)

technique that has been largely applied to characterize, several times in combination with vibrational spectroscopy, the OIHM’s structure (Babonneau, 1994; Peeters et al. 1995). The main advantage of NMR is to be element selective, allowing one to probe in a very distinct way the organic, as well the inorganic, part in OIHM. Also the availability of new advanced NMR equipments (Massiot 2002) has lead a significant improvement of the understanding of hybrid local structure. NMR with magic angle spinning (MAS) has allowed probing of the local chemical environments of different nuclei and the local dynamics, with a high degree of precision (Spiess 1997). Even if the sensitivity represents in some cases a limitation in the application of solid-state NMR spectroscopy, the possibility to probe the inorganic and organic local structures with their chemical correlations makes NMR, at moment, the best tool available to characterize OIHM structure. Several types of information can be obtained by multinuclear solid-state NMR, such as structural homogeneity, degree of condensation of the inorganic species, presence of mixed X–O–Y linkages in case of mixed oxide compositions, phase separation, and polymerization of the organic functional groups.

Structural Characterization of Hybrid Organic–Inorganic Materials Table 8 Assignment of the 13c NMR signals in GPTMS–TMOS–Al (Osbu)3 SOLS

Chemical shift (ppm) 4.84 22.45 73.05 71.03 50.38 43.69 18.1 57.7 49.5 50.4 18.5 58.4 64.1 67.1 15.3 71.3 Above 74

13

Structural units CH2CHOCH2OCH2CH2CH2–Si CH2CHOCH2OCH2CH2CH2–Si CH2CHOCH2OCH2CH2CH2–Si CH2CHOCH2OCH2CH2CH2–Si CH2CHOCH2OCH2CH2CH2–Si CH2CHOCH2OCH2CH2CH2–Si CH3–CH2–OH CH3–CH2–OH CH3OH CH3–O–Si CH3–CH2–O–Si CH3–CH2–O–Si CH2(OH)–CH(OH)–CH2– CH2(OCH2–CH3)–CH(OH)–CH2– CH2(OCH2–CH3)–CH(OH)–CH2– CH2(OH)–CH(OH)–CH2– Poly(ethylenoxide)

Adapted from Lee et al. (2003)

The characterization of the organic part by solid-state 13C cross-polarization (CP) MAS NMR spectroscopy allows a detailed identification of the reactions in polymerizable OIHM’s, while these reactions are generally more difficult to analyze by vibrational spectroscopies. An example of the potential of the technique has been given by Templin et al. (1997), who observed how the addition of small amounts of Al(OsBu)3 efficiently catalyzes the opening of epoxy ring in GPTMS based sols. A full assignment of signals in 13C NMR spectra, from sols of GPTMS–TMOS–Al (OsBu)3, is reported in Table 8 from Lee et al. (2003). On the other hand, the ring opening can follow different reaction pathways (Fig. 6): (1) hydrolysis of the epoxy with formation of diols, (2) alcoholysis with formation of methyl ether terminal groups, and (3) step polymerization to give oligo- or poly (ethylene) oxide groups. The experiments have clarified that the products of the epoxy polymerization are mainly oligomers, rather than long chains of poly(ethylene) oxide. The extent of condensation of the inorganic network of OIHM with polymerizable functionalities has a major role to model the final hybrid structure; the extension of the organic polymerization is, in fact, affected by the presence of a more or less condensed inorganic network. This effect has been clearly observed in GPTMS–TMOS hybrids catalyzed by BF3 (Innocenzi et al. 2000). This catalyst can efficiently open the epoxy ring in GPTMS, but at the same time, it can also catalyze the polycondensation of the silica network. Larger amounts of BF3 open the epoxide but also strongly catalyze the inorganic polycondensation, leaving a lower free volume within the material for the growth of the organic chains. Figure 7 shows the 13C CP MAS NMR spectra of GPTMS–TMOS–BF3 samples, prepared with increasing amounts of BF3.

14 Fig. 6 Reaction pathways for the hydrolysis of GPTMS based hybrids, under acidic conditions

Fig. 7 13C CP MAS NMR spectra of the OIHM with compositions GPTMS–TMOS–BF3, prepared with increasing amounts of BF3. The sample GTB0 is used as reference and is prepared without adding the catalyst BF3. In the other samples, the BF3 amount increases from 1% (molar ratio) in GTB1, to 10% in GTB10 (From Innocenzi et al. 2000, with permission of the American Chemical Society)

P. Innocenzi et al.

Structural Characterization of Hybrid Organic–Inorganic Materials

15

Fig. 8 Schematic representation of the thermalinduced polymerization in MPTMS-based hybrids. The labels are the same used in Fig. 9 and Table 10

The decrease in intensity of the peaks related to epoxy (peaks 5 and 6 in Fig. 7) is directly related to the appearance of a broad and intense peak around 74 ppm. This chemical shift is typical of C atoms in oligo- or poly(ethylene) oxide derivative species formed from the epoxy opening. Larger amounts of BF3 (samples GTB5 and GTB10 in Fig. 7) induce instead a change in the signal, with the presence of two new sharp and well-resolved peaks (species C and D). The peak sharpening, due to CHx–O species (x = 1 or 2), is attributed to the presence of shorter poly(ethylene oxide) chains. A coupled analysis by 29Si MAS NMR confirmed that, for larger amounts of BF3, only T3 and Q4 units are present. The combination of 13C and 29Si NMR spectra has shown, therefore, that the length of chains formed upon organic polymerization is strongly dependent on the conditions of synthesis and within certain limits can be monitored by solid-state NMR spectroscopy. Another important group of OIHM is derived from MPTMS. The observations about the competition between organic polymerization and inorganic polycondensation are of general value and can be extended also to this type of hybrids. 13C CP MAS NMR has been extensively used to characterize these materials and is quite informative to study organic polymerization. After reaction of acrylate C=C double bonds, the rise of a signal due to quaternary C atoms is an indication of the polymerization (Fig. 8). Similarly to what is observed in FTIR spectra, the change in the chemical environment upon polymerization induces a chemical shift of the signal of the C atom in C=O (peaks 4 and A in Fig. 9). An example of the application of 13C CP MAS NMR to MPTMS–aminopropyltriethoxysilane (APTES) is shown in Fig. 9 (Innocenzi 2003), and the assignment of the chemical shifts is given in Table 10. It is important to note that secondary cross-linking reactions, in this specific case due to the secondary amine in APTES, can be easily recognized by 13C CP MAS NMR (Innocenzi 2003).

16

P. Innocenzi et al.

Fig. 9 13C CP MAS NMR spectra of MPTMS–TMOS–APTES hybrids, thermally cured at different temperatures (From Innocenzi 2003) Table 9 Assignment of the signals in GPTMS–TMOS–BF3 hybrids Chemical shift (ppm) 10 24

Structural unit Si–CH2–CH2– Si–CH2–CH2–

75

–CH2–CH2–O–CH2–

73

–CH2–CH2–O–CH2–

51 44 59

CH2(
O–)–CH– CH2(
O–)–CH– –O–CH2–C (OH)–CH2–O–CH3 –O–CH2–[
CH2–CH2–]n–O–

70–75

Adapted from Innocenzi et al. (2001) The labels are referred to Fig. 7

Comments First carbon atom close to silicon Second carbon atom close to silicon Carbon in the ether bridge og GPTMS Carbon in the ether bridge og GPTMS Carbon in the epoxy ring Carbon in the epoxy ring Therminal methyl ether group

Label ① ②

⑤ ⑥ A

Carbon in the poly(ethylene oxide) chain

B, C, D

③ ④

Structural Characterization of Hybrid Organic–Inorganic Materials

17

Table 10 Assignment of the signals in the 13c CP MAS NMR spectra of MPTMS–TMOS–APTES hybrids Chemical shift (ppm) 9 22 67 167

Units –CH2–CH2–Si –CH2–CH2–Si –O–CH2–CH2–CH2–Si –C(=O)–

137 18 126 177

C=C CH3–C– CH2=C– –C(=O)–

45 55

–(
C–)– –CH2–

Comments First carbon linked to the silicon Second carbon close to the silicon Third carbon close to the silicon Carbon close to vinylic carbons, before the C=C opening Vinylic carbon atom Carbon in the methyl terminal group Vinylic carbon atom Carbon close to aliphatic carbons, after the C=C opening Quaternary carbon Carbon in the polymerized hydrocarbon chain

Label ① ② ③ ④ ⑤ ⑦ ⑥ A B C

Innocenzi (2003), with permission of the American chemical society The numbers from 1 to 7 are used as labels for the carbons in MPTMS, while the letters are used for carbons in MPTMS after reaction of the C=C double bond

The relative amounts of the different Si units and the degree of condensation of the siloxane network in the hybrids are easily obtained by single-pulse 29Si MAS experiments (Cerveau et al. 2002). This technique is generally applied, even if, to obtain reliable quantitative information; the long repetition delays must be used due to the long relaxation times (10–50 s). Information on the structural homogeneity can also be extracted in some cases. The application of the technique to gels prepared from methyldiethoxysilane (MDES)–TEOS and MTES–TEOS is an interesting example. Upon sol–gel reaction, the partially condensed species are observed as DH, T, or Q units. The DH and T units exhibit a continuous low-field shift, after incorporation in a network with an increasing number of Q units (Babonneau et al. 2000) (Fig. 10). This effect can be related to the change in the nature of the neighboring sites of the observed units and is an indication of a large tendency of the two systems to form co-condensed species. On the other hand, to extract information on the spatial distribution from 29Si CP MAS NMR is more difficult. The CP technique can be extended to investigate the correlation of the chemical species and the local structure, using bidimensional heterocorrelations, 1H–29Si CP MAS and 1H–13C CP MAS. These experiments give information on the proximity of the 1H and 29Si or 1H and 13C nuclei. 1H–29Si CP MAS heteronuclear correlation has been applied to MDES–triethoxysilane (TREOS) (Babonneau et al. 2000), to MTES–TEOS (Fyfe et al. 1992) and PhTES–TEOS (Peeters et al. 1995), and the homogeneous distribution of siloxane and silica units within the hybrid network has been derived. In Fig. 11, an example of 1H–29Si heterocorrelated spectra is reported. At short contact times (tc = 250 μs), only correlations between the two types of Si

P. Innocenzi et al. –60

–29

–61

–30

–62

–31

–63

–32

–64

–33

–65

–34

–66

–35

Chemical shift (ppm) - DH units

Chemical shift (ppm) - T units

18

–36

–67 0

2

4 6 8 T/Q (or DH/Q) molar ratio

10

Fig. 10 Variation of chemical shift for the T and DH units in dried gels prepared from co-hydrolysis of MDES/TEOS and MTES/TEOS, in various molar ratios (From Babonneau et al. 2000, with permission of Elsevier)

nuclei (DH and TH) and the protons in the Si–H units are observed. The polarization transfer to 29Si is realized only via directly bonded 1H protons. Increasing the contact time (2.5 ms), new types of correlations between the DH and TH units with CH3 are observed. The similar cross-relaxation times for the Si–H and Si–CH3 peaks suggest a close spatial proximity between the DH and TH units (Babonneau et al. 2000). The identification of the various oxo-bridges in a hybrid network is also a very important feature. The homogeneity of the OIHM with mixed oxides (Gervais et al. 2001) is very much dependent on the possibility of forming mixed bonds, avoiding phase separation and clustering. Siloxane–titania (Gervais et al. 2000; Crouzet et al. 2003), siloxane zirconia and borosiloxane systems are the most commonly used mixed oxides in hybrid materials where the control of the structural homogeneity is critical. 17 O solid state NMR has shown to be quite effective to identify the various oxo-bridges within the hybrid networks (Gervais et al. 2001). In general, for these systems, 17O NMR results more informative with respect to 29Si NMR, because of the large 17O chemical shift range. The possibility to resolve the local configuration in OIHM has been widened by very fast lH MAS NMR (Babonneau et al. 2000; Azais et al. 2002). The homogeneous character of the strong 1H–1H homonuclear dipolar interactions requires, in fact, high fields and very high rotation frequencies (Samoson et al. 2001). The lH

Structural Characterization of Hybrid Organic–Inorganic Materials

19

Fig. 11 Contour plot of the two-dimensional1 H–29 Si correlation experiment on a gel prepared from co-hydrolysis of MDES and TREOS (Babonneau et al. 2000)

MAS NMR spectra result broadened and not informative, because slow MAS is not effective to remove the broadening effect due to strong homonuclear coupling, if the MAS frequency is not larger than the interaction magnitude. The recent availability of new spectrometers with very high fields and very high rotation frequencies opens the road to a more detailed investigation of OIHM structure. An example of the increased resolution in 1H NMR spectra is shown in Fig. 12, where the spectra of MDES–TREOS hybrids were recorded, using a combination of high magnetic fields (14.1–18.8 T) and fast spinning rates (up to 35 kHz) (Babonneau et al. 2000). The resolution of the spectra is clearly influenced by the strength of the magnetic field (spectra a and b) and the spinning rate (spectra b and c). At 14.10 T, the broad signal around 5 ppm is resolved in two distinct peaks at 4.4 and 4.8 ppm, due to Si–H groups in TH and DH units, respectively. At faster spinning rates (30 kHz, spectrum c), an even better resolution is achieved, with lower broadening of the signals.

Conclusions The combined use of different analytical techniques has given a quite good picture of the structure of hybrid materials, even if several features still remain to be fully addressed. High-resolution NMR, together with vibrational spectroscopies, have allowed to recognize that the structure of hybrid materials, especially when

20

P. Innocenzi et al.

Fig. 12 1H MAS NMR spectra of MDES–TREOS hybrids (From Babonneau et al. 2000, with permission of Kluwer)

polymerizable organic groups are present in the material, can be finely controlled through synthesis and processing to tune the properties as a function of the final application.

References Almeida R, Pantano C. Structural investigation of silica gel films by infrared spectroscopy. J Appl Phys. 1990;68:4225–32. Azais T, Bonhomme-Coury L, Vaissermann J, Maquet J, Bonhomme C. The first aluminophosphonate cluster analogue of the 4=1 SBU of zeolites: structure and multinuclear solid-state NMR study, including 1H NMR. Eur J Inorg Chem. 2002;11:2838–43. Babonneau F. 29Si, 17O Liquid NMR and 29Si CP-MAS NMR characterization of siloxane-oxide materials, [(CH3)2SiO/TiO2, (CH3)2SiO/ZrO2]. Mater Res Symp Proc. 1994;346:949–60. Babonneau F, Gualandris V, Maquet J, Massiot D, Janicke MT, Chmelka BF. Newly applied two-dimensional solid state NMR correlation techniques for the characterization of organically modified silicates. J Sol-Gel Sci Technol. 2000;19:113–7. Baia L, Gigant K, Posset U, Petry R, Schottner G, Kiefer W, Popp J. Confocal Raman investigation on hybrid polymer coatings. Vib Spectrosc. 2002;29:245–9. Brigo L, Grenci G, Carpentiero A, Pistore A, Tormen M, Guglielmi M, Brusatin G. Positive resist for UV and X-ray lithography synthesized through sol–gel chemistry. J Sol-Gel Sci Technol. 2011;60:400–7. Brigo L, Faustini M, Pistore A, Kang HK, Ferraris C, Schutzmann S, Brusatin G. Porous inorganic thin films from bridged silsesquioxane sol–gel precursors. J Non-Cryst Solids. 2016;423:399–405.

Structural Characterization of Hybrid Organic–Inorganic Materials

21

Brusatin G, Innocenzi P, Guglielmi M, Signorini R, Bozio R. Hybrid sol–gel materials for optical limiting with increased laser damage resistanc. Nonlinear Opt. 2001;27:259–67. Brusatin G, Innocenzi P, Guglielmi M, Babonneau F. Basic catalyzed synthesis of hybrid sol–gel materials based on 3-glycidoxypropyltrimethoxysilane. J Sol-Gel Sci Technol. 2003;26:303–6. Brusatin G, Della GG, Guglielmi M, Innocenzi P. Photocurable glycidoxypropyltrimethoxysilane based sol-gel hybrid materials. Prog Solid State Chem. 2006;34:223–9. Brusatin G, Della GG, Romanato F, Guglielmi M. Design of hybrid sol–gel films for direct x-ray and electron beam nanopatterning. Nanotechnology. 2008;19:175306 (7 pp). Cerveau J, Corriu RJP, Framery E. Nanostructured organic–inorganic hybrid materials: kinetic control of the texture. Chem Mater. 2001;13:3373–88. Cerveau J, Corriu RJP, Framery E, Ghosh S, Mutin HP. Hybrid materials and silica: drastic control of surfaces and porosity of xerogels via ageing temperature, and influence of drying step on polycondensation at silicon. J Mater Chem. 2002;12:3021–6. Crouzet L, Leclercq D, Hubert MP, Vioux A. Organosilsesquioxane–titanium oxide hybrids by nonhydrolytic sol–gel processes. Study of the rearrangement of Si–O–Ti bonds. Chem Mater. 2003;15:1530–4. De Ferri L, Lorenzi A, Lottici P. OctTES/TEOS system for hybrid coatings: real-time monitoring of the hydrolysis and condensation by Raman spectroscopy. J Raman Spectrosc. 2016;47:699–705. Della GG, Zambon A, Lamberti F, Elvassore N, Brusatin G. Straightforward micropatterning of oligonucleotides in microfluidics by novel spin-on ZrO2 surfaces. ACS Appl Mater Interfaces. 2015;7:13280–8. Falcaro P, Costacurta S, Malfatti L, Buso D, Patelli A, Schiavuta P, Piccinini M, Grenci G, Marmiroli B, Amenitsch H, Innocenzi P. Chemical tailoring of hybrid sol-gel thick coatings as hosting matrix for functional patterned microstructures. ACS Appl Mater Interfaces. 2011;3:245–51. Fyfe CA, Zhang Y, Aroca P. An alternative preparation of organofunctionalized silica gels and their characterization by two-dimensional high-resolution solid-state heteronuclear NMR correlation spectroscopy. J Am Chem Soc. 1992;114:3252–3255. Gervais C, Babonneau F, Hoebbel D, Smith ME. Solid state NMR interaction parameters of oxygens linking titanium and silicon in crystalline cyclic titanodiphenylsiloxanes. Solid State Nucl Magn Reson. 2000;17:2–14. Gervais C, Babonneau F, Smith ME. Detection, quantification, and magnetic field dependence of solid-state 17O NMR of X–O–Y (X, Y = Si, Ti) linkages: implications for characterizing amorphous titania–silica-based materials. J Phys Chem B. 2001;105:1971–7. Gigant K, Posset U, Schottner G, Baia L, Kiefer W, Popp J. Inorganic–organic cross-linking in UV curable hard coats based upon vinyltriethoxysilane–tetraethoxysilane–polyfunctional acrylate hybrid polymers: a Raman spectroscopic study. J Sol–Gel Sci Technol. 2003;26:369–73. Grasselli JG, Ritchey WM (Eds). Atlas of spectral data and physical constants for organic compounds, vol. I. Cleveland: CRC Press Inc.; 1975. Harreld JH, Dunn B, Zink JI. Effects of organic and inorganic network development on the optical properties of ORMOSILs. J Mater Chem. 1997;7:1511–7. Hoebbel D, Nacken M, Schmidt HJ. A NMR Study on the hydrolysis, condensation and epoxide ring-opening reaction in sols and gels of the system glycidoxypropyltrimethoxysilane-watertitaniumtetraethoxide. Sol–Gel Sci Technol. 1998;12:169. Innocenzi P, Brusatin G, Guglielmi M, Bertani R. New synthetic route to (3-glycidoxypropyl) trimethoxysilane-based hybrid organic–inorganic materials. Chem Mater. 1999;11:1672–9. Innocenzi P, Brusatin G, Babonneau F. Competitive polymerization between organic and inorganic networks in hybrid materials. Chem Mater. 2000;12:3726–32. Innocenzi P, Sassi A, Brusatin G, Guglielmi M, Favretto D, Bertani R, Venzo A, Babonneau F. A novel synthesis of sol–gel hybrid materials by a nonhydrolytic/hydrolytic reaction of (3-glycidoxypropyl)trimethoxysilane with TiCl4. Chem Mater. 2001;13:3635–43.

22

P. Innocenzi et al.

Innocenzi P, Brusatin G, Licoccia S, Di Vona L, Babonneau F, Alonso B. Controlling the thermal polymerization process of hybrid organic–inorganic films synthesised from 3-methacryloxypropyltrimethoxysilane and 3-aminopropyltriethoxysilane. Chem Mater. 2003;15:4790–7. Innocenzi P. Infrared spectroscopy of silica sol–gel films: a spectra-microstructure overview. J Non-Cryst Solids. 2003;316:309–19. Innocenzi P, Brusatin G. A comparative FTIR study of thermal and photo-polymerization processes in hybrid sol–gel films. J Non-Cryst Solids. 2004;333:137–42. Innocenzi P, Kidchob T, Yoko O. Hybrid organic-inorganic sol-gel materials based on epoxy-amine systems. J Sol-Gel Sci Techn. 2005;35(3):225–35. Lee T-H, Kang E-S, Bae B-S. Catalytic effects of aluminum buthoxyethoxide in sol–gel hybrid hard coatings. J Sol-Gel Sci Technol. 2003;27:23–9. Le Luyer C, Garcýa-Murillo A, Bernstein E, Mugnier J. Waveguide Raman spectroscopy of sol–gel Gd2O3 thin films. J Raman Spectrosc. 2003;34:234–9. Li X, King TA. Spectroscopic studies of sol–gel derived organically modified silicates. J Non-Cryst Solids. 1996;204:235–42. Maia Seco A, Clara Goncalves AM, Almeida RM. Densification of hybrid silica–titania sol–gel films studied by ellipsometry and FTIR. Mater Sci Eng B. 2000;76:193–9. Massiot D. High resolution solid state NMR. In: Berthier C, Lévy LP, Martinez G, editors. High magnetic fields: applications in condensed matter physics and spectroscopy, LNP, 595. Berlin: Sprienger; 2002. ISBN:3-540-43979-X. Medda SK, Kundu D, De G. Inorganic–organic hybrid coatings on polycarbonate. Spectroscopic studies on the simultaneous polymerizations of methacrylate and silica networks. J Non-Cryst Solids. 2003;318:149–56. Mendez-Vivar J, Mendoza-Bandala A. Spectroscopic study on the early stages of the polymerization of hybrid TEOS–RSi(OR0 )3 sols. J Non-Cryst Solids. 2000;261:127–36. Ou DL, Seddon AB. Near- and mid-infrared spectroscopy of sol–gel derived ormosils: vinyl and phenyl silicates. J Non-Cryst Solids. 1997;210:187–203. Peeters MPJ, Wakelkamp WJJ, Kentgens APMJ. A 29Si solid-state magic angle spinning nuclear magnetic resonance study of TEOS-based hybrid materials. Non-Cryst. Solids. 1995;189:77. Riegel B, Plittersdorf S, Husing N, Kiefer W, Schubert U. Raman spectroscopy analysis of the sol–gel processing of RSi(OMe)3/Si(OMe)4 mixtures. J Mol Struct. 1997;410–411:157–60. Riegel B, Plittersdorf S, Kiefer W, Hofacker S, Muller M, Schottner G. Kinetic investigations of hydrolysis and condensation of the glycidoxypropyltrimethoxysilaneraminopropyltriethoxy–silane system by means of FT-Raman spectroscopy I. J Non-Cryst Solids. 1998;226:76–84. Riegel B, Kiefer W, Hofacker S, Schottner G. FT-Raman spectroscopic investigation on the organic crosslinking in hybrid polymers. Part II: reactions of epoxy silanes. J Sol-Gel Sci Technol. 2002;24:139–45. Samoson A, Tuherm T, Gan Z. High-field high-speed MAS resolution enhancement in 1H NMR spectroscopy of solids. Solid State NMR. 2001;20:130–6. Sanchez C, Ribot F. Design of hybrid organic-inorganic materials synthesized via sol-gel chemistry. New J Chem. 1994;18:1007. Sassi Z, Bureau JC, Bakkali A. Structural characterization of the organic/inorganic networks in the hybrid material (TMOS–TMSM–MMA). Vib Spectrosc. 2002;28:251–62. Schottner G. Hybrid sol–gel derived polymers: applications of multifunctional materials. Chem Mater. 2001;13:3422–35. Shea KJ, Loy DA. Bridged polysilsesquioxanes molecular-engineered hybrid organic–inorganic materials. Chem Mater. 2001;13:3306–30. Soppera O, Croutxeç-Barghorn C, Lougnot DJ. New insights into photoinduced processes in hybrid sol–gel glasses containing modified titanium alkoxides. New J Chem. 2001;25:1006–14. Spiess HW. Multidimensional solid state NMR: a unique tool for the characterisation of complex materials. Ber Bunsen-Ges Phys Chem. 1997;101:153.

Structural Characterization of Hybrid Organic–Inorganic Materials

23

Templin M, Wiesner U, Spiess H. Multinuclear solid-state-NMR studies of hybrid organicinorganic materials. Adv Mater. 1997;9:814–7. Urlacher C, Mugnier J. Waveguide Raman spectroscopy used for structural investigations of ZrO2 sol–gel waveguiding layers. J Raman Spectrosc. 1996;27:785–92. Urlacher C, DeLucas CM, Mugnie J. Chemical and physical aspects of sol–gel process for planar waveguides elaboration: application to zirconia waveguides. Synth Met. 1997;90:199–204. Yoshino H, Kamiya K, Nasu H. IR study on the structural evolution of sol–gel derived SiO2 gels in the early stage of conversion to glasses. J Non-Cryst Solids. 1990;126:68–78. Zanchetta E, Della GG, Grenci G, Pozzato A, Tormen M, Brusatin G. Novel hybrid organicinorganic spin-on resist for electron- or photon-based nanolithography with outstanding resistance to dry etching. Adv Mater. 2013;25:6261–5.

Small-Angle X-ray Scattering by Nanostructured Materials Aldo F. Craievich

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dilute Sets of Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concentrated Sets of Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fractal Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanophase Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Grazing Incidence Small-Angle X-Ray Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Final Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix: Experimental Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 13 17 22 28 32 39 39 44

Abstract

This chapter contains the basic theory of small-angle X-ray scattering (SAXS) and its applications to low-resolution studies of nanostructured materials. The primary purpose is to explain how to obtain structural information from simple systems whose low-resolution structure can be described by a two-electron density model, consisting of either homogeneous nanoparticles embedded in a (solid or liquid) medium with constant electron density or two-phase bicontinuous systems. The presented SAXS theory and the examples of applications refer to different procedures for determinations of geometrical parameters associated to nanoparticles or clusters in dilute solution, spatially correlated nanoparticles, and more general two-phase systems, namely, particle radius of gyration, interface area, size distribution, fractal dimension, and interparticle average distance. Other described applications are in situ SAXS studies of mechanisms involved in transformation processes leading to nanostructured A.F. Craievich (*) Institute of Physics, University of São Paulo, São Paulo, Brasil e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_37-1

1

2

A.F. Craievich

materials such as those occurring in nanophase separation and along the successive steps of sol–gel routes. One section is dedicated to present the basic concepts and describes an application of grazing incidence small-angle scattering (GISAXS), which allows for studying nanostructured thin films and thin layers located close to the external surface of solid substrates. Most of the reported applications refer to nanostructured materials obtained by sol–gel processing and are based on experimental results published by the author and collaborators.

Introduction This chapter describes the basic theory of small-angle X-ray scattering (SAXS) and reports a number of examples of application of this experimental technique to low-resolution structural investigations. Several examples also show how SAXS is applied to the characterization of transformation mechanisms in different nanomaterials. Sol–gel processing starts from colloidal particles in liquid solution and often leads to solid materials with interesting properties. Along all steps of sol–gel transformations the nanoscopic nature of the structure is preserved. This chapter includes several applications of SAXS technique to in situ characterizations of precursor systems starting from liquid sols up to final solid nanostructured materials. The basic process of the scattering of X-rays by materials is the photon–electron interaction. As it will be seen along this chapter, the complex amplitude (or modulus and phase) of the electromagnetic wavelets associated to photons elastically scattered in all directions
 by any material, is related to the tridimensional electron !

density function ρ r

through a Fourier transformation.
 ! The electron density function ρ r fully describes the structure of materials; thus

the ultimate goal of crystallographers and materials scientists is to determine this function, starting from experimental X-ray scattering patterns. Although this detailed information is not in practice fully obtained, relevant and useful structural features can generally be inferred. A typical SAXS setup is schematically shown in Fig. 1a. This technique provides useful structural information about heterogeneities in electron density sized within the range ~5 to 500 Å, these limits depending on the photon energy, sample-todetector distance, size of the beam-stopper and geometry of the X-ray detector. Very large objects as compared to the X-ray wavelength (with a size above, say, 1 μm) produce noticeable scattering intensity only within an extremely small angular domain close to the direction of the incident beam. Thus in this case the scattered photons hit the incident beam-stopper and are not recorded by the X-ray detector. Notice that the X-ray scattering intensity patterns within the “small-angle” range do not contain any information about the very short wavelength modulations in electron density associated to the atomic nature of the material, the effects from them only appearing in the scattering intensity profiles recorded at wide angles.

Small-Angle X-ray Scattering by Nanostructured Materials

3

Fig. 1 (a) Schematic SAXS setup. (b) X-ray beam paths from the source (left) to the detector (right), both located far away from the sample. The total segment Δs ¼ AB þ BC is the optical path ! difference associated to the X-ray scattering by electrons in two elements of volume d r , from which the phase shift is determined

Basic Theory General Equations The intensity associated to electromagnetic waves elastically scattered by an electron was derived by Thompson. Since the amplitude of the X-ray wave scattered by an electron has a well-defined phase relation with the amplitude of the incident wave, interference between scattered wavelets occurs. For a nonpolarized incident X-ray beam with intensity I0, the intensity associated to the wavelets scattered by one electron per unit solid angle Ω, is I e ð2θÞ ¼ I 0 ½ð1 þ cos2 2θÞ=2 :r 2e , where 2θ is the scattering angle (i.e., the angle between the wave-vectors of incident and scattered

4

A.F. Craievich

photons), and re is the classical electron radius. The X-ray intensity associated to the elastic scattering by an electron at small angles per unit of incident beam intensity and per unit of solid angle can be considered as a constant, I e ¼ r 2e . Using this approximated value, the relative error in Ie for 2θ up to 8 is indicating the average over
the analyzed sample volume V. !

As indicated by Eq. 6 the function γ r

– named correlation function (Debye

and Bueche 1949) – is the volume average of the product of
Δρ in two volume ! ! elements connected by the vector r . The correlation function γ r is determined
 ! from the experimental scattering function I q by inverse Fourier transformation:
 ! γ r ¼

1 ð2π Þ3 V

ð
 ! ! ! ! I q ei q : r d q

(7)



 ! ! Provided that ρ r is known, the correlation function γ r can be determined

 ! ! by applying Eq. 6. But, inversely, from a known γ r function, ρ r cannot be unambiguously inferred.

Small-Angle X-ray Scattering by Nanostructured Materials

7

For isotropic systems,
 the correlation function is independent of the direction of !

!

the vector r , i.e., γ r becomes γ(r) and, consequently, the scattering intensity, I(q), ! !

i q  r is replaced in Eq. 5 by its is also isotropic. For D isotropic E systems the function e ! !

spherical average ei q : r

Ω

¼ sin qr=qr. Thus Eqs. 5 and 7, respectively, become

I ð qÞ ¼ V

ð1

4πr 2 γ ðr Þ

0

γ ðr Þ ¼

ð1

1 ð2π Þ3 V

0

sin qr dr qr

4πq2 I ðqÞ

sin qr dq qr

(8) (9)

A useful procedure that is often applied to characterize low-resolution structures, circumventing the phase problem,
is to  begin with an initial model described by a !

guessed electron density function ρ r . The scattering amplitude is thus determined

 ! ! by using Eq. 1 and then the resulting scattering intensity I q ¼ jA q j2 is

compared to the experimental intensity function. The use of ad hoc computer programs allows for many iterations and modifications of the structure model, until a good fit of the calculated function to the experimental curve is achieved. This procedure is, for example, applied to the determination of low-resolution structures (envelope functions) of proteins in dilute solution (Svergun 1999). For materials consisting of isolated (in general nonidentical) nanoparticles embedded in a homogeneous matrix, the scattering intensity I(q) is often modeled under the assumption of simple shapes and taking also into account eventual effects from spatial correlation. The model function is then fitted to the experimental scattering curves. An eventual good fitting justifies a posteriori the proposed model and yields the adjusted parameters that characterize the structure of the studied material. In another procedure, which is often applied to study structural transformations in materials subjected to isothermal annealing, the isotropic correlation function γ(r) is theoretically determined starting from basic thermodynamic and/or statistical concepts (Cahn 1965; Lebowitz et al. 1982). This is followed by the determination of I(q) for increasing periods of time using Eq. 8 and further comparison of the series of model functions with the sequence of experimental SAXS curves determined in situ along the structural transformation. This procedure is applied, for example, to verify the correctness of theoretical models of particle growth and structure coarsening.

Small-Angle Scattering by Nanoscopic Two-Phase Systems: Porod’s Law This section deals with isotropic biphasic materials, i.e., isotropic two-electron density systems with sharp interfaces, such as those schematically drawn in Fig. 2a, b. In this model the relevant parameters are the electron densities ρ1 and

8

A.F. Craievich

Fig. 2 Schematic examples of two types of biphasic structures or two-electron density systems. (a) Set of isolated spherical nano-objects with a constant electron density ρ1 embedded in a homogeneous matrix with electron density ρ2. (b) Bicontinuous structure, both phases with constant electron densities ρ1 and ρ2

ρ2 and the volume fractions φ1 and φ2=1φ1. This model is applied to characterize different nanostructured materials such as nanoporous solids, nanocrystals, or disordered nanoclusters embedded in solid or liquid media, etc. The general properties of Fourier analysis tell us that the asymptotic trend, at high q, of the scattering intensity I(q) is connected to the behavior of the γ(r) function at small r. For isotropic two-electron density systems, the correlation function γ(r) can be approximated at small r by (Porod 1982):  γ ðr Þ ¼ ðρ1  ρ2 Þ2 φ1 ð1  φ1 Þ 1 

S r 4Vφ1 ð1  φ1 Þ

 (10)

where S/V is the area of the interface per unit sample volume. Replacing γ(r) given by Eq. 10 in Eq. 8 and solving the integral, the leading term of the asymptotic intensity I(q), at high q, is given by (Porod 1982) I ð qÞ ¼

2π ðρ1  ρ2 Þ2 S q4

ð q ! 1Þ

(11)

Equation 11, named Porod law, applies to isotropic two-electron density systems with sharp interfaces, such as disordered porous materials and other two-phase systems whose relevant structural feature is their interface surface area. Porod’s law applies to either dilute or concentrated systems of isolated nanoobjects, provided they are not very thin sheets or very narrow cylinders, for which the asymptotic intensities are proportional to 1/q2 and to 1/q, respectively (Shull and Roess 1947). Equation 11 does not hold for sets of identical spherical or cylindrical nano-objects, because in these cases the SAXS intensity exhibits oscillations even at very high q. By analyzing the features of such oscillations, it is possible to determine the distance between the parallel portions of the interfaces (Ciccariello 1991).

Small-Angle X-ray Scattering by Nanostructured Materials

9

However, if the spherical or cylindrical nano-objects have a wide size distribution, the oscillations smear out and the asymptotic Porod’s law holds. For anisotropic ! two-electron density systems, Porod’s law still applies along all q directions, but the parameter S in Eq. 11 has a different meaning (Ciccariello et al. 2002). The behavior of I(q) at high q is often analyzed using I(q)q4 vs. q4 plots. Equation 11 implies that I(q)q4 becomes asymptotically constant in the high-q limit but, for many materials, the SAXS intensity also contains an additional and q-independent contribution from short-range density fluctuations in their phases (Ruland 1971). For these materials, the asymptotic I(q)q4 vs. q4 plot, at high q, is expected to exhibit a linear dependence (i.e., I(q)q4 = a + b q4) with a positive slope (b > 0). Extrapolation of the linear portion of the I(q)q4 function toward q4 = 0 yields I(q)q4(q = 0) = a. By substituting this value in Eq. 11, the interface area between both phases, S, is determined. On the other hand, Ruland (1971) demonstrated for two-phase systems with a smooth transition in electron density between both phases, that the asymptotic I(q)q4 vs. q4 plot at high q also exhibits linear dependence but in this case the slope is negative (b < 0). FromðEq. 9 it can be verified that γ(0) = Q/(2π 2V ), where Q is the integral defined as Q ¼

1

q2 :I ðqÞdq. On the other hand, for two-electron density systems, γ(0) is

0

equal to ðρ1  ρ2 Þ2 φ1 ð1  φ1 Þ (Eq. 10), so as the integral Q becomes Q¼

ð1

q2 I ðqÞdq ¼2π 2 ðρ1  ρ2 Þ2 Vφ1 ð1  φ1 Þ

(12)

0

The integral Q depends on the electron density contrast factor (ρ1 – ρ2)2 and volume fractions of both phases but not on the specific features of their geometrical configuration. For example, along structural transformations that preserve both electron densities and phase volume fractions, even though the structure and, consequently, the shape of the scattering intensity curves change, the integral Q is expected to remain constant. Therefore, the integral Q is named “Porod invariant.” Examples of transformations that occur without significantly affecting the value of the integral Q are the processes of growth of homogeneous nanoclusters by mechanisms of coarsening or coalescence. For the determination of the interface surface area S by applying Eq. 11 the measurement of the scattering intensity in absolute units is required (See “Appendix: Experimental Issues”). Moreover, from Eqs. 11 and 12, the following equation is derived: ½I ðqÞq q!1 S ¼ π:φ1 ð1  φ1 Þ V Q 4

(13)

Thus, if the scattering intensity is only known in relative scale and provided the phase volume fractions are known, Eq. 13 allows for the determination of the specific interface surface area (S/V). Equation 13 is often applied to powdered

10

A.F. Craievich

samples, for which the precise measurement of the scattering intensity in absolute units is difficult.

Small-Angle Scattering by Spatially Uncorrelated Nanoparticles: Guinier’s Law The wavelets associated to the X-ray scattering by a dilute set of spatially uncorrelated nano-objects do not interfere. Under this condition and provided the objects are identical and centrosymmetric, the total scattering intensity I(q) is expressed as I ðqÞ ¼ NI 1 ðqÞ

(14)

where N is the number of nanoparticles and I1(q) is the SAXS intensity produced by a single nanoparticle. By solving Eq. 8 for an arbitrary correlation function γ(r) associated to a single nano-object, it can be demonstrated (Guinier and Fournet 1955) that the SAXS intensity at small q is given by I ðqÞ ¼ N ðΔnÞ2 eRg q

2 2

=3

ð q ! 0Þ

(15)

where Δn is the excess in number of electrons inside the nano-objects and Rg their radius of gyration. For nano-objects with volume V1 and constant electron density ρ1, embedded in a homogeneous matrix with electron density ρ2, the number of electrons in excess is Δn ¼ ðρ1  ρ2 ÞV 1 and the radius of gyration is ð Rg ¼

!

r d r =V 1 2

1=2 (16)

V1

Equation 15 is named Guinier law. In order to derive the radius of gyration Rg of nano-objects from results of SAXS measurements, the Guinier plot (log I vs. q2) is applied. In this plot, a straight line is expected to be observed at small q, within a more or less wide q range depending on the size and shape of the objects (Guinier and Fournet 1955). From the slope αG of the straight line in Guinier plots, the radius of gyration is determined; Rg ¼ ½3ðαG =log:eÞ1=2 ¼ 2:628:jαG j1=2 . For example, the radius of gyration of homogeneous spherical objects is related to their radius R by Rg = (3/5)1/2R and that of homogeneous cylinders with radius R and height H by     1=2 Rg ¼ R2 =2 þ H2 =12 . Guinier plots are also applied to determine the SAXS intensity at q = 0, I(0), by linear extrapolation of log I(q2) to q2 = 0. The total SAXS intensity produced by a dilute set of nano-objects with a distribution of radii of gyration N(Rg) is given by the sum of the individual

Small-Angle X-ray Scattering by Nanostructured Materials

11

contributions of each object. For this system Guinier’s law also holds but the derived parameters are weighted averages. For example, for two-electron density systems consisting of a isotropic and polydisperse set of N spatially uncorrelated nanoobjects, Eq. 15 becomes

 2 2 I ðqÞ ¼ N ðρ1  ρ2 Þ2 V 21 ehRgiG q =3 ðq ! 0Þ

(17)

where is the average of V21 and < Rg >G is a weighted average (named Guinier average) defined as 2ð

31=2   2 2 R dR N R V g g 1 g 6 7

 7 Rg G ¼ 6 4 ð   2 5 N Rg V 1 dRg

(18)

ð   with N Rg dR ¼ N. Notice that the G averaging weights more large objects than small ones. For a polydisperse set of spherical nano-objects, Eq. 18 becomes 2ð

31=2 " #1=2 8 N ð R ÞR dR 6 7 R8 7 ¼  ð hRiG ¼ 6 4 5 R6 N ðRÞR6 dR

(19)

Guinier law is usually applied to determine the radius of gyration of nano-objects with narrow size distribution. For highly polydisperse systems, the q range over which Guinier law holds is small and Guinier plot yields a weighted average of the radius of gyration far from the arithmetic average and strongly biased toward those of the largest objects. This effect is schematically described for two sets of spherical objects with same arithmetic average radius = 40 Å but different widths of radius distribution, as shown in Fig. 3a. For these two systems the slopes of the linear portion of Guinier plots and, consequently, the average radius < R>G derived by applying Guinier law are different (Fig. 3b). The extrapolated intensity I(0) for polydisperse systems, being proportional to the average , also depends on the shape of the radius distribution. From Eq. 15, it can be inferred that the experimental SAXS intensity extrapolated to q = 0 corresponding to a dilute set of N identical objects is given by I(0) = N (Δn)2. For two-electron density systems composed of a dilute set of N nano-objects, each of them with volume V1 and electron density ρ1, embedded in a matrix with electron density ρ2, the SAXS intensity at q = 0 is I ð0Þ ¼ N ðρ1  ρ2 Þ2 V 21

(20)

12

A.F. Craievich

Fig. 3 (a) Narrow N1(R) and wide N2(R) radius distributions of spheres with same arithmetic radius average = 40 Å. (b) Guinier plots of SAXS intensities G1 and G2, at small q, corresponding to the radius distributions N1(R) and N2(R), respectively. The magnitude of the slope of the linear part of the log I vs. q2 plot at small q and the extrapolated intensity I (0) associated to the radius distribution N2(R) are both higher than for N1(R)

For dilute solutions we have φ1 V ¼ NV 1 and ð1  φ1 Þ  1 so as Eq. 12 becomes Q ¼ 2π 2 N ðρ1  ρ2 Þ2 V 1 . Thus, regardless the object shape, its volume V1 can be determined from the quotient I(0)/Q as follows: V 1 ¼ 2π 2

I ð 0Þ Q

(21)

Equation 21 can also be applied to polydisperse systems, the result being in this case the quotient between averages, /. Identical anisotropic objects with same orientation produce anisotropic scattering ! patterns, i.e., the scattering intensity depends on the direction of the vector q . In the ! limit of small q , Guinier law becomes (Guinier and Fournet 1955) I ðqD Þ ¼ N ðρ1  ρ2 Þ2 V 21 eRD qD 2

!

2

(22)

where qD refers to the component of q in the direction along which the scattering intensity is measured and RD is the inertia distance of the object in the same

Small-Angle X-ray Scattering by Nanostructured Materials

13

direction, from a perpendicular plane containing the center of “mass” of the electron density function. If the system is composed of identical and anisotropic nano-objects that are randomly oriented, the resulting scattering intensity is isotropic. In this case, the structural parameter determined by applying Guinier law (Eq. 15) is the radius of gyration of the nano-objects.

Dilute Sets of Nanoparticles Spherical Nanoparticles Schematic views of monodisperse and polydisperse sets of spherical nano-objects are shown in Fig. 4a, b, respectively. The scattering intensity associated to a single spherical and homogeneous nano-object embedded in a homogeneous matrix, with spatially constant electron densities ρ1 and ρ2, respectively, is derived from the amplitude A1(q) defined by Eq. 3. For a spherical nano-object with radius R the scattering intensity is given by  2

I 1 ðqÞ ¼ jA1 ðqÞj ¼ ðρ1  ρ2 Þ

ðR

sin qr dr 4πr qr 2

0

2 (23)

By solving the integral, Eq. 23 becomes  2 I 1 ðqÞ ¼ ðρ1  ρ2 Þð4π=3ÞR3 Φðq, RÞ

(24)

where Φ(q, R) is Φðq, RÞ ¼ 3

sin qR  qR cos qR ðqRÞ3

Thus the total scattering intensity produced by a dilute (spatially uncorrelated) set of N identical spheres is I(q) = NI1(q), i.e., I ðqÞ ¼ N ðρ1  ρ2 Þ2



4π 3: 2 R ½Φðq, RÞ2 3

(25)

The scattering intensity given by Eq. 25 is plotted in Fig. 4a, for identical spheres with radius R = 40 Å. At high q the intensity function exhibits several secondary maxima and zeros, the different zeros being located at qR = 4.50, 7.72, 10.90 . . . The scattering intensity related to a dilute set of N spherical nano-objects with a radius distribution defined by N(R) is calculated by

14

A.F. Craievich

Fig. 4 Schematic views of systems composed of dilute sets of (a) monodisperse and (b) polydisperse sets of spherical nano-objects. (c) Scattering intensities corresponding to three samples containing spherical objects with the same average radius = 40 Å and a Gaussian distribution N(R) with three different standard deviations: σ = 0, σ = 5 and σ = 15 Å

ð I ðqÞ ¼ N ðRÞI 1 ðq, RÞdR

(26)

where I1(q, R) is the scattering intensity produced by a single sphere (Eq. 24). The scattering intensity curves related to three dilute sets of spherical objects, with different Gaussian radius distributions and same arithmetic average radius = 40 Å, are plotted in Fig. 4c. The standard deviations of the Gaussians are σ = 0 (monodisperse system), 5 Å and 15 Å. It can be noticed in Fig. 4 that, for increasing polydispersivity, the secondary maxima and zeros progressively smear out. On the other hand, in this case the intensity I(0), being proportional to or , is higher for wider radius distributions. The radius distribution N(R) of a dilute and polydisperse set of spherical nanoparticles can be derived from the measured I(q) functions by solving the integral Eq. 26. For this purpose, the program package named GNOM (Svergun 1992) is often used. The output of GNOM yields the volume weighted distribution function, D(R), related to N(R) for spheres by D(R) = (4π/3)R3N(R). GNOM is also applied to determine the volume distribution function of nano-objects with other simple shapes.

Small-Angle X-ray Scattering by Nanostructured Materials

15

Moreover the intensity function I1(q) related to objects with complex shapes can be independently determined and used as an input file in GNOM program.

Application (Example 1): PbTe Nanocrystals Embedded in a Silicate Glass An experimental SAXS study of a system composed of PbTe nanocrystals embedded in a silicate glass was performed by Craievich et al. (1997). This nanostructured material exhibits interesting nonlinear optical properties in the infrared, making it potentially useful for applications to telecommunication devices. A silicate glass doped with Pb and Te was held at high temperature, quenched by splat-cooling down to room temperature and then submitted to an isothermal annealing at 650  C. Initially, isolated Pb and Te atomic species diffuse through the supersaturated glass and nucleate PbTe nanocrystals which progressively grow. A number of SAXS intensity curves were successively recorded in situ, along the whole annealing process. The experimental results are displayed in Fig. 5. The SAXS intensity progressively increases for increasing annealing time. At high q, the intensity curves exhibit satellite peaks or secondary maxima that are characteristic of the scattering function associated to a set of spheres with nearly identical

Log intensity (arbitrary units)

10

8

6

4

2

0 –0.2

–0.1

0.1

0.0

0.2

0.3

q (Å–1)

Fig. 5 Scattering intensity curves recorded in situ, corresponding to a dilute set of spherical PbTe nanocrystals embedded in a homogeneous silicate glass, during isothermal growth at T = 650  C. The period of time for nanocrystal growth increases from 19 up to 119 min from bottom to top. The continuous line is the best fits of Eq. 26 using a Gaussian N(R) function with a time-varying radius average and a constant relative standard deviation σ/ = 0.08. The curves are vertically displaced for clarity (Reprinted with permission from Craievich et al. (1997). Copyright 1997 by the International Union of Crystallography)

16

A.F. Craievich

radius. The secondary maxima progressively shift toward smaller q, as expected for a set of growing nanospheres (Eq. 25). Because of the high statistical dispersion in the scattering intensities at high q, the secondary maxima are not clearly apparent in the curves corresponding to early stages of nanocrystal growth. The positive deviation of the experimental points from the theoretical modeled curve, at very small q, indicates the existence of additional and rather large heterogeneities in electron density in the glass matrix. The experimental SAXS curves displayed in Fig. 5 were well fitted by model functions defined by Eq. 26, which applies to dilute sets of spherical objects, assuming a time-varying average nanocrystal radius and a Gaussian radius distribution, N(R), with a time-independent relative standard deviation σ/ = 0.08. For the sample held 2 h at 650  C, the best fit of the model scattering curve led to = 32.5 Å and σ = 2.6 Å. The time dependence of the average radius agrees with the prediction of the classical theory for nucleation and growth of spherical precipitates in a homogeneous matrix.

Application (Example 2): Clustering of Colloidal ZnO Nanoparticles Powders consisting of ZnO nanoparticles produced by sol–gel route are used as precursors for developments of new materials with interesting properties. The first step of the sol–gel route leading to ZnO solid nanoparticles is the formation of a liquid suspension of zinc acetate in ethanol, to which LiOH is added under ultrasound treatment. An in situ SAXS study was performed in order to characterize the first steps of aggregation of ZnO nanoparticles in liquid solution (Tokumoto et al. 1999). The different experimental scattering functions, recorded after increasing periods of time at 40  C, were analyzed by assuming that the system is dilute and that the colloidal nano-objects are spherical. In order to determine the radius distribution of the particles, the integral Eq. 26 was solved by using GNOM program (Svergun and Semenyuk 1991; Svergun 1992). GNOM was applied to all experimental scattering curves of the studied ZnO-based suspension corresponding to different aggregation times, thus yielding the set of volume weighted radius distribution functions D(R) plotted in Fig. 6. The shape of the D(R) function and its time variation (Fig. 6) suggested that the kinetics of formation of ZnO clusters is characterized by two main stages. During the first stage, a growing peak centered at R = 17 Å is apparent, indicating a continuous formation of small clusters. The number of clusters increases monotonously for increasing reaction time, while their average radius, = 17 Å, remains constant. During the second stage, the volume weighted distribution exhibits a still growing peak at 17 Å, while the formation and growth of a second peak corresponding to an initial average particle radius = 60 Å is also apparent. This peak shifts continuously toward higher R values, up to 110 Å, along a period of time of 2 h.

Small-Angle X-ray Scattering by Nanostructured Materials

17

e

Tim

Fig. 6 Time-dependent volume weighted radius distribution, D(R), for ZnO-based colloidal particles in liquid suspension maintained inside a sealed cell during SAXS measurements. The time of growth increases from 10 up to 120 min. The D(R) functions were derived from the set of experimental SAXS curves by applying the GNOM program (Reprinted with permission from Tokumoto et al.(1999). Copyright 1999 by Elsevier)

150

100

50

0

R (Å)

The described time variation of the volume weighted distribution function clearly evidences the continuous formation of colloidal primary clusters and their simultaneous aggregation and growth.

Concentrated Sets of Nanoparticles Spatially Correlated Spherical Nanoparticles Many sol–gel based isotropic nanomaterials consist of spatially correlated nanoparticles embedded in a homogeneous matrix. Examples are concentrated colloidal sols (solid nanoclusters embedded in a liquid medium) and solid hybrid nanomaterials (inorganic clusters embedded in a solid polymeric matrix). Two models of SAXS functions associated to different types of systems composed of spatially correlated nano-objects will be described, one of them containing identical nanoclusters and another consisting of a two-level hierarchical structure. The total scattering intensity produced by a set of identical and spatially correlated nano-objects is affected by interference effects, thus Eq. 14 does not hold. For isotropic systems composed of a set of N spatially correlated spherical (or more generally centrosymmetrical) nano-objects, the SAXS intensity is given by I ðqÞ ¼ NI 1 ðqÞSðqÞ

(27)

18

A.F. Craievich

where S(q) is the structure function that accounts for interference effects produced by spatial correlation. For a set of nano-objects without long-range order, the structure function S(q) tends asymptotically to 1 at high q. For a set of spatially uncorrelated nano-objects SðqÞ ¼ 1 over the whole q domain, and thus Eq. 27 becomes equivalent to Eq. 14. A semiempirical structure function that is often applied to describe spatial correlation in isotropic systems composed of spherical nano-objects embedded in a homogeneous matrix, derived using the Born–Green approximation, is given by (Guinier and Fournet 1955): Sð qÞ ¼

1 1 þ kΦS ðqÞ

(28)

where k, named “packing factor,” is associated to the degree of compactness of the local structure (for the closest packing of spheres kmax is equal to 5.92) and Φs(q) is Φ S ð qÞ ¼ 3

sin qd  qd cos qd ðqdÞ3

(29)

where d is the average distance between the spatially correlated nano-objects. Several examples of models for scattering intensity functions are displayed in Fig. 7a, b. These functions are determined by Eq. 27 with S(q) given by Eq. 28 for a set of spheres with same radius, R = 10 Å, and different d and k values. The intensity curves displayed in Fig. 7a show that the q value corresponding to the maximum of the scattering curves, qmax, decreases for increasing average distances. On the other hand, the different curves plotted in Fig. 7b indicate that increasing values of packing factor k yield more pronounced and well-defined scattering peaks. A rough estimate of the average distance between particles is usually inferred by applying the simple equation d = 2π/qmax. However, by analyzing the curves plotted in Fig. 7a, it can be verified that the equation d = 5.6/qmax yields a better estimate of the average distance. Anyway, the determination of a more precise average distance between particles requires the fitting of a model intensity function to the whole experimental intensity curve. Even for nano-objects that are not spherical but instead exhibit a globular shape, the structure function given by Eq. 28 is usually applied as a good approximation. This structure function is also applied to model scattering intensity curves associated to materials composed of polydispersed nano-objects with narrow radius distributions. Many hybrid materials prepared by the sol–gel process were studied by SAXS. Some of these hybrid materials are composed of a isotropic set of inorganic nanoclusters embedded in a polymeric matrix. The heterogeneous nature of these nanostructured materials is characterized by using a simple two-electron density model consisting of high electron density clusters embedded in a low electron

Small-Angle X-ray Scattering by Nanostructured Materials

19

Fig. 7 Model scattering intensity curves corresponding to different sets of spatially correlated spheres, all of them with same radius, R = 10 Å. (a) Packing factor k = 3 and average interparticle distances: d = 30 Å, d = 50 Å, and d = 70 Å. (b) Average distance d = 50 Å and packing factors k = 1, k = 3 and k = 5. The normalized scattering intensity curve for a dilute set of particles with same radius is displayed as a black line in (a) and (b)

density matrix (Dahmouche et al. 1999). Certainly, the polymeric phase exhibits electron density fluctuations at molecular level that also produces small-angle scattering, but their contribution to the total scattering intensity is assumed to be weak and/or not strongly varying with q. The basic assumption here is that the dominant contribution to small-angle scattering intensity comes from the electron density contrast between inorganic nanoclusters and polymeric matrix. Some materials are heterogeneous at multiple scale levels. For example, nanometric clusters may segregate and form cluster-rich domains embedded in a cluster-depleted matrix. For this particular two-level system, the effects on the SAXS intensity produced by a coarse structural level and another fine level are expected to be dominant at low and high q, respectively. For the example to be described in the next section corresponding to a isotropic two-level structure – with its fine level consisting of spatially correlated nano-objects – the scattering intensity can be modeled by the following semiempirical equation (Beaucage et al. 1995):

20

A.F. Craievich

  n   oP1 ð1=3ÞR2g1 q2 1=2 3 ð1=3ÞR2c q2 I ðqÞ ¼ G1  e þ B1  e erf qRg1 =6 =q   n o P2    3 2 2 þ G2  eð1=3ÞRg2 q þ B2  erf qRg2 =61=2 =q  Sð qÞ

(30)

where sub-indexes 1 and 2 refer to the coarse and fine structure levels, respectively. The factors Gi are equal to Ni(Δni)2 (Eq. 15) and Bi are related to Gi by specific equations that depend on the object geometry, and Pi are Porod exponents that are equal to 4 for simple two-electron density systems and may have other values depending on the geometry of the objects. The second term in Eq. 30 corresponding to the fine level also includes the structure function S(q) accounting for spatial correlation of the small clusters inside the volume defining the coarse level. In the first 2term, associated to the coarse structure, the Gaussian function given by exp Rc q2 =3 is a high-q cutoff factor in which Rc = Rg2 (Beaucage et al. 1995). Provided the X-ray scattering experiment covers a wide q range, hierarchical structures consisting of more than two structure levels can also be characterized. In order to model SAXS intensity curves associated to these complex materials, additional terms are included in Eq. 30. Since the q range to be covered for the study of many-level structures is rather wide, several SAXS measurements with the same sample but using different collimation conditions, sample-to-detector distances and/or X-ray wavelengths, are required. Examples of fittings of model functions assuming multilevel structures to a number of experimental SAXS curves were reported by Beaucage et al. (1995). In order to characterize coarse structures composed of very large (micrometric) particles, the use of ultra-small-angle X-ray scattering (USAXS) – q range below 0.001 Å1 – or light scattering techniques is required.

Application: Fe-Doped Organic–Inorganic Hybrid Nanomaterials Many organic–inorganic composite materials exhibit interesting properties that can be tailored by an adequate control of the preparation conditions (Dahmouche et al. 1999). Moreover, the structural characterization of these materials is needed in order to explain their magnetic behavior. The structure of a number of hybrids composed of spatially correlated siliceous nanoparticles or clusters embedded in a matrix consisting of grafted polymeric chains were well described by a two-electron density model. For these systems the SAXS patterns exhibit a correlation peak located at decreasing q values for increasing molecular weight of the polymer molecule (Dahmouche et al. 1999). A SAXS study of hybrid organic–inorganic nanomaterials composed of Fe(II)doped di-ureasils was carried out by Silva et al. (2003). Figure 8a displays the scattering intensity produced by a di-ureasil hybrid doped with 0.76 wt% Fe(II). In order to characterize the structure of Fe(II)-doped nanohybrids, the two-level model described in the precedent section (Beaucage et al. 1995) was applied. The SAXS

Small-Angle X-ray Scattering by Nanostructured Materials

21

a

Intensity (a.u.)

10

1

0.1 0.02

0.04

0.06 0.08 0.1

0.2

0.4

−1

q (A )

b Level 2 Level 1

Fig. 8 (a) Experimental scattering intensity produced by siliceous clusters containing 0.76 wt% Fe(II) embedded in a polymeric matrix. The continuous line is the best fit of Eq. 30 to the experimental curve. The dashed lines indicate the Guinier and Porod contributions to the scattering intensity produced by siliceous clusters and the structure function (oscillatory curve). The dotted lines are the Guinier and Porod contributions to the scattering intensity associated to the coarse domains. (b) Schematic view of the proposed two-level model. The small circles correspond to siliceous clusters (Figure 8a reprinted with permission from Silva et al. (2003). Copyright 2003 by the International Union of Crystallography)

intensity corresponding to the fine structure level displays a peak associated to cluster-cluster correlations, centered at q = 0.15 Å1, which is also observed for undoped samples. For Fe(II)-doped hybrids, this peak is slightly shifted toward higher q. For q < 0.1 Å1, the scattering intensity is mainly related to the coarse structural level. The model scattering curve defined by Eq. 30 for two structural levels, including the structure function S(q) for the fine level given by Eq. 28, is displayed in Fig. 8a. This figure also shows the Guinier and Porod contributions to

22

A.F. Craievich

the total scattering intensity corresponding to both levels. The radii of gyration Rg obtained by the best fit procedure are 7.5 Å for the small clusters and 54 Å for the coarse domains. Similar analyses of SAXS curves for different Fe(II) doping levels, up to 4.5 wt%, revealed a decreasing average distance between siliceous clusters for increasing Fe(II) content. This result suggests that Fe(II) ions are dispersed in the polymeric matrix, these ions promoting a shrinkage effect that leads to the observed decrease in average cluster-cluster distance. The model structure consisting of large domains containing spatially correlated siliceous particles embedded in a depleted matrix is schematically shown in Fig. 8b. The reported results indicate that the formation of coarse silicide-rich domains is promoted by the addition of Fe(II) ions (Silva et al. 2003).

Fractal Structures Small-Angle Scattering by Fractal Structures The SAXS method is applied to structural characterization of a number of materials which exhibit a self-similar or fractal structure, and also to the determination of the mechanisms involved in aggregation processes, either in precursor sols or after the sol–gel transition. Fractal materials are characterized by three relevant structural parameters: (i) a radius r0 corresponding to the size of the individual primary particles (basic nanoobjects that build up the fractal structure), (ii) a fractal dimension D that depends on the nature of the mechanism of aggregation, and (iii) a correlation length ξ that defines the size of isolated aggregates or the cutoff distance of the fractal structure for percolated systems such as fractal gels. A homogeneous object and another with fractal structure – built up by N small primary particles – are schematically shown in Fig. 9a, b, respectively. The number of primary particles inside a sphere of radius r, measured from the center of mass of the fractal aggregate, is given by  D N ðr Þ ¼ r=r 0

(31)

where r0 is of order of the size of the primary basic units that build up the fractal object. Thus the mass M(r) inside a sphere with radius r, for both (homogeneous and fractal) objects, is proportional to rD, the exponent being D = 3 for homogeneous objects and D < 3 for fractal aggregates. The SAXS intensity associated to a correlated set of primary nanoparticles building up a fractal structure is defined by Eq. 27, which involves the scattering intensity by single primary particles, I1(q), and the structure function, S(q), associated to the nature of their spatial correlation.

Small-Angle X-ray Scattering by Nanostructured Materials

23

Fig. 9 (a) Homogeneous object and (b) fractal object. The red curve in (c) is the scattering intensity, I(q) = I1(q).S(q), associated to a fractal object with size of primary building blocks ro = 5 Å, correlation length ξ = 5000 Å, and fractal dimension D = 1.80. The scattering intensity corresponding to the primary particles, I1(q) (olive), and the structure function, S(q) (blue), are also plotted in (c). Notice that for large fractal aggregates the determination of the correlation length requires SAXS measurements down to very low minimum q value (qmin  1/ξ)

Different simple functions have been used for I1(q), such as the intensity produced by spherical particles (Eq. 25) or the Debye–Bueche function, defined by I 1 ð qÞ ¼ 

A 1 þ r 20 q2

2

(32)

where A is a constant. The structure function S(q) corresponding to a fractal object is derived from the radial distribution function for primary particles inferred from Eq. 31. This

24

A.F. Craievich

distribution function is multiplied by a cutting function that defines a structural correlation length ξ. This analysis finally leads to the following structure function (Teixeira 1988): Sð qÞ ¼ 1 þ

1 ðqr 0 Þ

D

h

 D  Γ ðD  1 Þ 1 iðD1Þ=2 sin ðD  1Þ tan ðqξÞ 1 þ 1=ðqξÞ2

(33)

where Γ is the gamma function. Thus, by selecting I1(q) defined by Eq. 32 and the structure function S(q) given by Eq. 33, the scattering intensity produced by a fractal aggregate, or by a set of spatially uncorrelated fractal aggregates, is I ð qÞ / 

1

2 1 þ r 20 q2 9 8 > > <  = 1 D  Γ ð D  1Þ 1  1þ sin ðD  1Þ tan ðqξÞ h i > > ðqr 0 ÞD 1 þ 1=ðqξÞ2 ðD1Þ=2 ; :

(34)

A scattering intensity function defined by Eq. 34 for particular values of the three structural parameters (ro, ξ, D) is plotted in log–log scale in Fig. 9c. Since the size of the primary particles is much smaller than the correlation length, I1(q) is constant within a rather wide low-q range, thus the variation of the scattering intensity at small q’s is dominated by the structure function. At high q, S(q) becomes a constant (S(q) = 1) and thus the variation in the scattering intensity in this q range is governed by I1(q). Figure 9c displays a log I vs. log q plot associated to a fractal object with correlation length much larger than the size of the primary particles (ξ r0). We notice in this log–log plot the presence of three q ranges over which linear dependences with different slopes are apparent: (i) Over the small q range (q  l/ξ) the slope is zero. In this q range the scattering intensity behaves as expected from Guinier’s law, its value extrapolated to q = 0, I(0), being related to the fractal dimension D by I ð 0Þ / ξ D

or

I ð0Þ / RD g

(35)

with Rg ¼ f2=½DðD þ 1Þg1=2 ξ: (ii) Over the intermediate q range, i.e., for 1/ξ  q  1/r0, the magnitude of the slope is equal to the fractal dimension D. This implies that the scattering intensity exhibits a simple power q-dependence, I(q) / q–D. (iii) Over the high q range (q 1/r0) the slope is 4, this implying that Porod’s law (I(q)/ q4) holds.

Small-Angle X-ray Scattering by Nanostructured Materials

25

Two crossovers of the different linear parts in log I vs. log q plots, at q = q1 and q = q2 (q2 > q1), are shown in Fig. 9c. The radius of the primary particles r0 is simply related to q2 by r0 = 1/q2 and the size parameter of the fractal aggregate or correlation length is given by ξ = 1/q1. Thus, if ξ r0, the relevant structure parameters ξ, and r0 can be directly determined from log–log plots of the scattering intensity. If the condition ξ r0 is not satisfied no well-defined crossovers are apparent. In this case, the parameters ξ, D, and r0, are determined by fitting the I(q) function defined by Eq. 34 to the whole experimental curve. The fractal dimension D can also be determined by applying Eq. 35 to a set of experimental SAXS curves determined in situ, during an aggregation process. The values of I(0) and Rg are determined from Guinier plots (Log I(q) vs. q2) for all successive SAXS curves. Since I ð0Þ / RD G the plot of log I(0)–log Rg is expected to be linear, the slope of the straight line yielding the fractal dimension D. If the condition ξ r0 is not fulfilled the “fractal” model cannot be safely applied. It is a general consensus that, in order to establish the fractal nature of an aggregate, the quotient ξ/r0 should be of the order of or larger than 10. In addition, it must be remembered that power q-dependences leading to D values smaller than 3 are also expected for nonfractal objects such as, for example, narrow linear chains or thin platelets. Therefore, independent evidences supporting the use of fractal models are often required. Many mechanisms involved in aggregation processes were analyzed and the respective fractal dimensions of the resulting structures were theoretically determined (Meakin 1986). By associating these theoretical results with experimental determinations of the dimension D, the mechanisms that govern aggregation processes leading to fractal structures can be established.

Applications: Aggregation in Zirconia-Based Sols and Gels The formation of zirconia-based gels promoted by the aggregation of colloidal particles in sol state was investigated in situ by SAXS (Lecomte et al. 2000). All experimental scattering curves, plotted as log I(q) vs. log q in Fig. 10, exhibit a wide q range with well-defined linear behavior. Following the procedure described in the precedent section, the magnitude of the slope of the straight line was assigned to the fractal dimension of the growing aggregates, D being equal to 1.7 along the whole aggregation process. The low-q limit of the linear portion of the scattering curves displayed in Fig. 10, and thus the crossover q1, progressively shifts toward lower q for increasing periods of time. This indicates that the aggregate size (ξ = 1/q1) continuously grows. The crossover q2 is not visible in the main set of curves displayed in Fig. 10 but, in the inset, corresponding to a SAXS curve determined up to a higher q value, this crossover toward a Porod behavior (I(q) / q4) is apparent. This suggests that the primary subunits have a smooth and well-defined external surface.

26

I(q) (arb. units) 2 000

1 000

500

−1.7

3 000 1 000 INTENSITY (a.u)

Fig. 10 Log I vs. log q plots corresponding to a zirconiabased sol held at room temperature for increasing periods of time, from 4 h (bottom) up to 742 h (top). The inset is the scattering intensity curve of the final gel obtained after a period of about twice the gelling time (Reprinted with permission from Lecomte et al.(2000). Copyright 2000 by the International Union of Crystallography)

A.F. Craievich

300 100

−4

30 10 0.1

200

0.5 1 2 3 q (nm –1)

100

50

20 0.1

0.5

1

2

3

q (nm –1)

The results reported by Lecomte et al. (2000) indicate that the fractal clusters in the studied zirconia-based sols are formed by aggregation of very small colloidal particles already existing at the beginning of the hydrolysis and condensation reactions. On the other hand, the maximum observed in the scattering curves for q 6¼ 0 is related to the existence of spatial correlations between the fractal aggregates, which could analytically be described by an inter-aggregate structure function S0 (q) defined in the same way as S(q), by Eq. 28, and included as another factor in Eq. 34. A fractal dimension close to that experimentally determined (D = 1.7) has been derived by computer simulation (Meakin 1986) for the mechanism of growth named diffusion-limited cluster-cluster aggregation (DLCA). Since the slope of all scattering curves displayed in log–log scale does not exhibit any variation with time, it could be concluded that the fractal dimension D and, consequently, the mechanism of aggregation remains invariant during the whole aggregation process. Another SAXS study of sulfate-zirconia sols with several compositions (varying HNO3, H2O and H2SO4 contents) was reported by Riello et al. (2003). In order to characterize the aggregation mechanism, these authors determined successive SAXS curves after progressively increasing time periods keeping the sols in open cells. The values I(0) and Rg – determined by applying Guinier law (Eq. 15) to every scattering curve – were plotted as log I(0) vs. log Rg. This plot was analyzed by applying Eq. 35, which predicts for fractal objects a linear behavior with a slope equal to the fractal dimension D of the growing aggregates.

Small-Angle X-ray Scattering by Nanostructured Materials

27

a

b

I(q) (arb. units)

I(0) (arb. units)

slope 1.78±0.06

0.1

c

b a

0.01 0.1

slope 0.98±0.06

1E−3 1

q (nm−1)

1

10 Rg(nm)

Fig. 11 Scattering intensity curves from sulfate-zirconia sols with different HNO3, H2SO4, and H2O contents. (a) Log–log plots of the scattering intensity produced by a few selected samples maintained inside a sealed cell at the end of their aggregation process. (b) Plot of I(0) vs. Rg, in log–log scale, corresponding to the final states of a number of sols with different compositions (Reprinted with permission from (Riello et al. 2003). Copyright 2003 by the American Chemical Society)

The process of cluster growth in sulfate-zirconia sols with different compositions in sealed cells was also studied. Since, under sealed condition, the reactions in sols are very fast, only the scattering curves corresponding to the final states could be determined (Fig. 11a). The log I(0) vs. log Rg plot corresponding to the final states of all studied samples is displayed in Fig. 11b. Notice that the experimental points lie on two different straight lines, each of them with a slope similar to those observed by the same authors in previous in situ studies during the cluster growth in open cells. The slope of the straight line for Rg < 20 Å in the log I vs. log Rg plot displayed in Fig. 11b is close to D = 1.0 thus suggesting that the aggregation process starts by the formation of short 1D linear chains. This initial regime is followed by another one involving the cross-linking of the precursor linear chains which build up a threedimensional fractal structure. The fractal dimension experimentally determined for Rg > 20 Å is 1.8, which is close to the expected theoretical value for diffusionlimited cluster-cluster aggregation. It was then concluded that, even though the sizes of the final aggregates in a number of sols, with very different compositions, vary from 0.5 nm up to 10 nm, the mechanism of growth of all of them is essentially the same. In the SAXS study reported by Riello et al. (2003) the mechanism of growth of the aggregates is theoretically characterized by an exponent D from the early stages

28

A.F. Craievich

of the clustering process, when the aggregates are still rather small and the condition ξ r0 for a fractal object is not yet fulfilled. Therefore, in these early stages, the exponent D – derived from in situ SAXS experiments by applying Eq. 35 – should not be assigned to a fractal dimension, but instead it must be considered as a useful parameter that characterizes the mechanism of growth.

Nanophase Separation General Considerations A number of nanoheterogeneous materials are formed by phase separation processes starting from a homogeneous solid solution at high temperature brought by fast cooling into a miscibility gap. In supersaturated and initially homogeneous (quenched) solid solutions with a composition close to the binodal curve (which defines the solubility limits), phase separation occurs by nucleation and growth of a minor new phase. This leads to a final two-phase material consisting of isolated and initially nanoscopic particles embedded in a homogeneous matrix (Fig. 2a). The growth of the second-phase particles can be characterized by in situ SAXS, using in this case a model consisting of a dilute or concentrated set of spherical particles surrounded by a solute-depleted shell. On the other hand, the final structure – after long periods of heat treatment – of initially homogeneous solid solutions brought, by quenching, close to the central part of a miscibility gap is described by a two-phase bicontinuous model, both phases occupying nearly the same volume fraction (Fig. 2b). For the first stages of phase separation occurring near the central part of the miscibility gap, a theoretical model named spinodal decomposition was proposed by Cahn (1965). At advanced stages of phase separation, even after having reached the equilibrium concentrations, both phases still exhibit a structural evolution driven by a pure coarsening mechanism.

Phase Separation and Dynamical Scaling Property In order to describe the advanced stages of nanophase separation (i.e., the coarsening regime) in binary materials, a statistical model was proposed by Marro et al. (1975) and Lebowitz et al. (1982). This model assumes that the material contains atoms A and B arranged in a simple cubic lattice with an occupation function η(ri), which takes values +1 or 1 for sites ri occupied by atoms A or B, respectively. A probability function for atom exchanges and a simple equation for the energy of the system was proposed. This model is analogous to that applied to ferromagnetic Ising spin systems. Finally, the theoretical isotropic and time-dependent structure function, S(q, t), was determined by computer simulation.

Small-Angle X-ray Scattering by Nanostructured Materials

29

In the proposed model the primary particles are spatially correlated atoms whose scattering intensity I1(q) at small q is constant. Consequently, the SAXS intensity (Eq. 27) can be written as I ðq, tÞ / Sðq, tÞ

(36)

Different moments Sn(t) and normalized moments qn(t) of the structure function, S(q,t), are defined as Sn ð t Þ ¼

ð1 ð01

qn ð t Þ ¼

ð0

Sðq, tÞqn dq Sðq, tÞqn dq

1

(37)

Sðq, tÞdq

0

Marro et al. (1975) determined the time variation of the structure function S(q,t) and its associated moments at advanced stages of phase separation, after both phases having reached their final compositions. Their results of computer simulations demonstrated that the structure function and its moments exhibit the following properties: (i) The second moment remains invariant, S2(t) = S2. Since S2 is proportional to the integral Q (Eq. 12), its time invariance implies that the advanced stage of phase separation is governed by a pure coarsening process. (ii) The time variation of the structure function S(q,t) exhibits a dynamical scaling property, evidenced by the existence of a time-independent function F(x) given by Fð x Þ ¼

Sðq, tÞ ½q1 ðtÞds S2

(38)

where the coordinate x is equal to (q/q1) and ds is the dimension of the space in which the process of phase separation occurs (ds = 3 for classical 3D processes). (iii) The normalized first moment of the structure function, q1(t), exhibits a power time-dependence q1(t)/ t–a, the parameter a depending on the detailed mechanism of the aggregation of atoms. (iv) The time dependence of the maximum of the structure function S(qm, t) is given 0 by S(qm, t) / ta with a0 = a.ds. All other moments and normalized moments of the structure function are also related by simple mathematical relations. A number of experimental investigations using small-angle (X-ray or neutron) scattering have demonstrated that the described dynamical scaling property also holds for phase separation processes occurring in many nanostructured materials, including glasses (Craievich and Sanchez 1981) and nanoporous xerogels (Santilli et al. 1995).

30

A.F. Craievich

Since the scattering intensity produced by very small primary particles (atoms) I1(q) is essentially constant within the small q range, all properties related to the time dependence of the structure function S(q) also apply to the time dependence of the experimental SAXS intensity function I(q).

Application (Example 1): Sintering of SnO2-Based Xerogels The theory described in the precedent section referring to phase separation processes in binary materials was applied to understand the structural evolution during isothermal treatment of nanoporous SnO2 xerogels studied by SAXS (Santilli et al. 1995). These nanoporous materials, after a short transient period, preserve their apparent density thus suggesting that the total fraction of porous volume remains constant during isothermal annealing. The series of SAXS curves displayed in Fig. 12a, corresponding to a SnO2-based xerogel isothermally annealed during increasing time periods at 400  C, exhibit a peak located at progressively decreasing q values. This feature is predicted by the statistical model described in the precedent section. The coincidence of all curves plotted as [S(q, t)q31/S2] vs. (q/q1) in Fig. 12b demonstrates that the dynamical scaling property (Eq. 38), theoretically derived for phase separation in simple binary systems, also applies to more complex processes such as the sintering of nanoporous xerogels.

a

b [S(q).q13]/S2 (arb. units)

I(q) (arb. units) 8

1

4

0

0 0

0.05

0.10

q(Å−1)

0

1

2 q/q1

Fig. 12 (a) Scattering intensity curves corresponding to a nanoporous SnO2 based xerogel held at 400  C after increasing periods of time, from 4.5 min (bottom) up to 62 min (top). (b) Scaled structure functions [S(q, t)q31/S2] vs. q/q1 (Reprinted with permission from Santilli et al. (1995). Copyright 1995 by the American Physical Society)

Small-Angle X-ray Scattering by Nanostructured Materials Fig. 13 The same scattering intensity curves displayed in Fig. 10 replotted here as I(q)q1.7 m vs. q/qm, qm being the q value corresponding to the maximum of the scattering curves (Reprinted with permission from Lecomte et al. (2000). Copyright 2000 by the International Union of Crystallography)

31

I(q/qm).qm1.7 100 50 30 20 10 5 3 2 1

0.5

1

2

5

10

20

q/qm

Application (Example 2): Dynamical Scaling of Zirconia-Based Fractal Structures A demonstration of the dynamical scaling property for a system consisting of fractal zirconia-based aggregates embedded in a liquid matrix was reported by Lecomte et al. (2000). These authors analyzed the set of SAXS curves displayed in Fig. 10, which exhibit a maximum shifting progressively toward lower q for increasing periods of time. As pointed out before, the fractal dimension derived from the linear portions of the log I(q) vs. log q plots results D = 1.7. The same set of curves displayed in Fig. 10 was plotted in Fig. 13 using a [I(q/qm)  qds] vs. (q/qm) scale and setting ds = 1.7. In this analysis, the authors assumed that the first normalized moments q1 can be replaced as a reasonable approximation by the q-values associated to the maximum of the scattering curves qm. As it can be seen in Fig. 13 all scattering curves merge into a single scaled curve, this clearly demonstrating that the dynamical scaling property also applies to structural transformations of fractal aggregates. The results reported by Lecomte et al. (2000) referring to fractal structures demonstrated that the quotient of exponents a0 and a associated to the time dependences of the functions Sm(qm,t) and q1(t), respectively, would not be equal to the space dimension, ds = 3, but instead equal to the fractal dimension D. The described experimental results together with those mentioned in the preceding sections and others reported in the literature suggest that the statistical model derived for nanophase separation and particularly the dynamical scaling property of the structure function (Marro et al. (1975)) exhibit universal features, which provide a unified description of processes of structural coarsening in a wide variety of materials.

32

A.F. Craievich

Grazing Incidence Small-Angle X-Ray Scattering Basic Concepts Thin films deposited on solid substrates such as those prepared by spin or dip coating and involving sol–gel transitions deserved the attention of many scientists because of their often interesting technological applications. These films usually have thicknesses ranging from about one nanometer up to a few microns. Since the structure of thin films supported by thick solid substrates cannot be studied by classical transmission SAXS, they are characterized by combining X-ray reflectivity and grazing incidence small-angle X-ray scattering (GISAXS). X-ray reflectivity measurements allow one to determine the thickness, average mass density, and surface roughness of thin films. Details of this experimental technique are not presented here. Readers interested on the basic concepts and applications of X-ray reflectivity are encouraged to consult the existing bibliography (for example Tolan 1999). Some thin films are heterogeneous at the nanometric scale. For example, thin films may be composed of a homogeneous matrix containing nanoclusters and/or nanopores, spatially correlated or not. Other materials consist of a homogeneous bulk volume with a thin layer close to their external surface containing buried nanoparticles. These nanoparticles are implanted by sputtering or plasma treatment or formed by nucleation and growth followed by atomic diffusion from supported thin films. GISAXS is usually applied to characterize nanostructured supported thin films and surface layers. Classical GISAXS experiments are performed using a flat sample, the incident beam hitting the sample surface at grazing incidence angles, αi, typically ~0.3 to 0.6 . The scattering patterns at small angles are recorded by a two-dimensional X-ray detector located at rather long distances from the sample, typically 1–3 m in synchrotron beam lines. Schematic views of the geometry of a GISAXS setup are shown in Fig. 14a, b. Notice that relevant angles in X-ray optics are measured with respect to the sample surface and not with respect to its normal as usual in classical optics. !

!

!

The three components of the scattering vector, q ¼ k  k 0 , associated to a scattered beam hitting a given detector pixel in GISAXS measurements (Fig. 14), are   qx ¼ ð2π=λÞ cos ψ cos αf  cos αi qy ¼ ð2π=λÞ sin ψ cos αf   qz ¼ ð2π=λÞ sin αf þ sin αi

(39)

The angles αf and ψ are determined from the vertical (dz) and horizontal (dy) distances (Fig. 14) as follows αf ¼ tg1½ðdz =LÞ  αi ψ ¼ tg1 d y =L

(40)

Small-Angle X-ray Scattering by Nanostructured Materials

33

Fig. 14 Schematic GISAXS setup. (a) Frontal view of the 2D X-ray detector taken along the direction of the incident beam. The distances that are measured in order to determine the three components of the scattering vector, qx, qy, and qz, associated to each detector pixel (ny, nz) (Eq. 39) are indicated. A narrow beam-stopper is usually vertically placed to avoid detector damaging by the strong reflected X-ray beam. (b) Lateral view indicating all relevant directions and angles. The GISAXS patterns can be recorded only above the sample horizon (dashed line)

Notice that the detection plane of the 2D detector in real experiments is perpendicular to the incident X-ray beam and not parallel to the normal to sample surface, as schematically shown in Fig. 14. Anyway, since the incidence angle αi is very small, the error in the angle αf associated to the use of this geometry can be neglected. For a X-ray beam propagating through a medium with refraction index n0 and hitting a flat interface with another medium with refraction index n, the angle of the refracted beam αr is determined by applying Snell law, ðn0 =nÞ ¼ ð cos αr = cos αi Þ. For an incident beam in vacuum ðno ¼ 1Þ or in a standard gas medium ðno  1Þ hitting a flat material surface, the refraction angle results: αr ¼ cos 1 ð cos αi =nÞ For typical (small) incidence angles, αr is given as a good approximation by

(41)

34

A.F. Craievich

 1=2 αr ¼ α2i  2δ

(42)

where δ ¼ 1  n. Since the refraction index of any material for X-rays is slightly lower than 1(δ ~ 105), there is a critical value of the incidence angle, αc, for which the refracted beam propagates parallel to the sample flat surface. Substituting αr ¼ 0 in Eq. 42 the critical angle results: αc ¼ ð2δÞ1=2

(43)

Values of δ for any material composition and X-ray photon energy up to 30 KeV were reported by Henke et al. (1993). Let us now to describe the main features associated to specular reflection and refraction of an incident monochromatic X-ray beam hitting a flat sample surface under grazing incidence. For different incidence angles the following types of effects occur: (i) For αi < αc, the incident beam undergoes specular reflection at an exit angle αe = αi. (ii) For αi = αc, the refracted beam propagates parallel to the sample surface, i.e., αr = 0. (iii) For αi > αc, the refracted beam propagates inside the sample in a direction defined by the angle αr and amplitude t(αi) given by Snell law and Fresnel transmission function, respectively. The absorption of X-rays penetrating a flat sample produces a decrease in intensity of the incident beam described by the following basic equation: I ðdÞ ¼ I 0 eðμρd= sin αr Þ

(44)

where I0 is the intensity of the incident beam, μ is the mass absorption coefficient, ρ is the mass density, and d is the distance from the sample surface. The attenuation length (also named penetration depth) is the distance from the surface for which the intensity of the X-ray beam – penetrating into a given material – becomes equal to I0/e. This distance is considered to be the approximate thickness of the layer probed in GISAXS measurements. Tables published by Henke et al. (1993) and a program accessible online in their web page yield the attenuation length as a function of the incidence angle, photon energy or wavelength, and sample chemical composition and mass density. If the incidence angle of the incoming X-ray beam is equal to or lower than the critical angle, an evanescent wave is formed whose penetration depth is only ~5 nm for typical materials. When a thicker layer is desired to be probed using an incident beam with same photon energy, the incidence angle αi should be set higher than αc. As an example, Fig. 15a displays the penetration depth of photons in three selected materials, SiO2, Al, and Ti, for a photon energy E = 8.04 KeV

Small-Angle X-ray Scattering by Nanostructured Materials

35

Fig. 15 X-ray attenuation length or penetration depth as function of the incidence angle αi for three selected materials (SiO2, Al and Ti) for incident X-ray beams with two different wavelengths: (a) Cu λKα (1.542 Å) and (b) MoλKα (0.7071 Å). For an incidence angle below the critical angle the attenuation length is very short (a few nanometers)

(corresponding to λCuKα = 1.542 Å) as a function of the incidence angle. Figure 15b shows the same functions for a photon energy E = 17.44 KeV (λMoKα = 0.707 Å). The curves displayed in Fig. 15a, b indicate that the attenuation length below the critical angle αc (a few nanometers) is very small for the three selected materials, while for α > αc the attenuation length exhibits a monotonous increase for increasing incidence angles. Notice that the attenuation length is higher for less dense and lower Z materials. Thus, by adequate choices of incidence angle and photon energy, and depending on material composition and density, a wide range of thicknesses of nanostructured surface layers can be probed. An additional feature that is apparent in GISAXS patterns is named Yoneda peak (Yoneda 1963), which is associated to interference effects between the reflected and refracted waves. This peak appears at an exit angle αY ¼ αc with respect to the sample surface (Fig. 14a, b).

36

A.F. Craievich

The analysis of GISAXS results associated to nanostructured thin films and/or to surface layers is performed by fitting model functions to 2D experimental patterns. A reasonable model of a GISAXS function requires an initial guess of particle shape, size distribution, and structure function and should include the Fresnel transmission and reflection functions and the effects associated to Snell law. More detailed descriptions of GISAXS theory and applications were reported by Kutsch et al. (1997). Moreover, a recent review of modern applications of GISAXS, GISANS (grazing incidence neutron scattering) and grazing incidence X-ray and neutron wide angle scattering was published by Hexemer and MullerBuschbaum (2015).

Example of application: Nanostructure of Thin Films Supported by Si Wafers A simple method for obtaining arrays of CoSi2 nanoplates endotaxially buried in a Si(001) single-crystalline wafer was reported by Kellermann et al. (2012). These authors demonstrated that thermally activated diffusion of Co atoms embedded in a Co-doped SiO2 thin film deposited on the (001) flat surface of a Si wafer promotes the formation of CoSi2 nanoplates buried inside the Si host. A transmission electron microscopy (TEM) study of this material indicated that the CoSi2 nanoplates exhibit a hexagonal lateral shape, are parallel to Si{111} crystallographic planes, have remarkably uniform sizes, and their lattices are coherently related to the host Si lattice. On the other hand, complementary analyses of TEM images showed the additional presence of a polydisperse set of spherical Co nanoparticles embedded in the supported SiO2 thin film. The model GISAXS function for a supported thin film containing an isotropic and dilute set of spherical nanoparticles with a radius distribution Nsph(R) is given by ð
   2   Isph qy , qz / jtðαi Þj2 t αf  N sph ðRÞ:I 1 qx , qy , q~z , R dR

(45)

where t(αi) and t(αf) are Fresnel transmission coefficients (Tolan 1999), I1,(qx, qy, qz, R) is given by Eq. 24 and refers to the SAXS intensity produced by spherical cobalt nanoparticles with radius R embedded in the silica thin film, and q~z is the z-component of the scattering vector considering that the incident beam scattered by nanoparticles is the refracted beam inside the sample. On the other hand, the GISAXS function associated to thin hexagonal CoSi2 nanoplates endotaxially buried in Si wafer, with their faces parallel to all Si{111} crystallographic planes, was modeled as (Kellermann et al. 2012; Kellermann et al. 2015):

Small-Angle X-ray Scattering by Nanostructured Materials

X   2     Ahex ðhklÞ αi , φ, qx , qy , q~z , L, T 2 I hex qy , qz / jtðαi Þj2 t αf  ðN hex =4Þ

37

(46)

hkl

where Nhex is the number of hexagonal nanoplates, Ahex(hkl) are the scattering amplitudes associated to regular hexagons oriented parallel to Si{111} crystallographic planes, with thickness T and lateral side L, φ is the azimuthal angle, and q~z is the component of the scattering vector in z direction inside the sample. The total GISAXS function associated to Co nanospheres embedded in the SiO2 thin film and CoSi2 nanohexagons buried in the Si wafer was modeled assuming independent contributions from both types of nano-objects, i.e.,       I total qy , q~z / ½I sph qy , q~z þ C:I hex qy , q~z 

(47)

where Isph and Ihex are given by Eqs. 45 and 46, respectively and C is an adjustable factor. In modeling the SAXS function, it was assumed that refraction effects are only produced at the interface between air and the SiO2 thin film. Because of the relatively low difference in density between the SiO2 thin film containing Co nanoparticles and the Si substrate, refraction effects associated to this interface were neglected. Kellermann et al. (2015) studied Co-doped SiO2 thin films deposited on silicon wafers with different surface orientations, namely, Si(001), Si(011), and Si(111), all of them previously heat treated at 750  C under identical conditions. Experimental 2D GISAXS patterns corresponding to different wafer orientations and the associated theoretical curves modeled by applying Eq. 47 are displayed in Fig. 16. The analysis of the experimental GISAXS results demonstrated that the sizes of the CoSi2 nanohexagons are functions of the crystallographic orientation of the Si substrate, the lateral size of the nanohexagons buried in Si(111) wafers being remarkably (~50 %) larger than those grown inside the other two substrates, Si(011) and Si(001). The thickness of the platelets also varies for different Si substrate orientations from 2.8 nm for Si(001) up to 5.7 nm for Si(111). On the other hand, the spherical Co nanoparticles embedded in the SiO2 thin film exhibit average radii ranging from 0.6 nm for Si(011) up to 1.5 nm for Si(001). In conclusion, the GISAXS study reported by Kellermann et al. (2015) led to a complete low-resolution characterization of the nanostructures developed in Co-doped SiO2 thin films deposited on Si(001), Si(011), and Si(111) substrates.

Final Remarks A relevant issue omitted in this chapter is the experimental method that uses the properties of “anomalous” (or resonant) small-angle X-ray scattering (ASAXS), which is today widely applied thanks to the availability of tunable synchrotron X-ray sources (Goerigk et al. 2003). ASAXS is particularly useful for structural

Fig. 16 Experimental 2D GISAXS patterns corresponding to Co-doped SiO2 thin films deposited on (a) Si(001), (b) Si(111), (c) Si(011) flat wafers. Pictures (d–f) are the calculated patterns (Eq. 47) that best fit the patterns (a–c), respectively. The main axes of all elongated lobes in GISAXS patterns are perpendicular to Si{111} crystallographic planes (Reprinted from Kellermann et al., Phys. Chem. Chem. Phys. 2015; 17: 4945–4951)

38 A.F. Craievich

Small-Angle X-ray Scattering by Nanostructured Materials

39

studies of biphasic materials with low contrast in electron density and also for analyses of complex multiphase systems. Additional information about SAXS theory is presented in the classical book authored by Guinier and Fournet (1955) and the book edited by Glatter and Kratky (1982). Another book dealing with SAXS, SANS, and light scattering was edited by Lindner and Zemb (1991). Instrumentation issues mainly focusing on SAXS using synchrotron radiation were described by Russell (1991), and SAXS/SANS studies of the structure and structural changes of biological macromolecules in solution were reviewed by Koch et al. (2003). A useful booklet for beginners was written by Schnablegger and Singh (2013). Besides the already mentioned GNOM software for SAXS data analysis (Svergun and Semenyuk 1991), new packages such as the recently developed SASFit (Kohlbrecher and Bressler 2014) are available to interested users. The amount of published articles based on the use of SAXS and SANS exhibited a fast increase during the past three decades (Craievich and Fischer 2010). This fast growth was primarily due to the increasing interest of scientists for studies of structural and physicochemical properties of nanomaterials. Other reasons that explain the observed strong growth in the annual number of published articles are: (i) the commercial availability of modern SAXS setups equipped with novel X-ray sources, focusing and collimating optics and fast high-resolution 2D detectors, (ii) the development of new theoretical approaches and numerical methods for data analysis, (iii) the increasing availability of powerful computers, and (iv) the opening of new small-angle scattering beam lines in many synchrotron and neutron laboratories around the world. Acknowledgments The author thanks the staff of the National Synchrotron Radiation Laboratory (LNLS), Campinas, Brazil, where the experimental parts of most of the SAXS investigations reported in this chapter were conducted; G. Kellermann and C. Huck-Iriart for their useful remarks; and H. Fischer for his help for figures preparation.

Appendix: Experimental Issues Basic Comments Monochromatic X-ray beams are characterized by their photon energy E or wavelength λ, both related by λ = hc/Ε, where h is the Plank constant and c is the speed of light in vacuum, i.e., λ(Å) = 12.398/E(KeV). The wavelengths of typical monochromatic beams used in SAXS experiments are within the range 0.6–2.0 Å circa (i.e., photon energies ranging from ~6 to ~20 KeV). The X-ray beams produced by synchrotrons or typical commercial sources are usually monochromatized by quartz, germanium, or silicon single crystals, which yield incident beams with very narrow pass-bands (Δλ/λ < 103). Considering, for example, a typical SAXS experiment with an X-ray wavelength λ = 1.542 Å (λCuKα), a sample-to-detector distance D = 1 m, a beam-stopper with a

40

A.F. Craievich

diameter ϕ1 = 5 mm and a circular 2D detector with a diameter ϕ2 = 150 mm, and remembering that q ¼ ð4π=λÞ sin θ  ð2π=λÞ:2θ for low q, the range of scattering angles to be covered results 0.14 < 2θ < 4.3 , and the corresponding minimum and maximum q values are 0.01 Å1 and 0.30 Å1, respectively. Different lower and upper q limits can be reached by selecting adequate beam collimation, sample-todetector distances and/or X-ray wavelengths. The choice of the experimental q range depends on the sizes of the nanoparticles to be studied. X-ray beams for SAXS experiments are produced by classical sealed X-ray tubes, rotating anode X-ray generators and synchrotron sources. Synchrotron radiation sources are often preferred because they provide powerful, continuously tunable and well collimated X-ray beams. Another closely related experimental technique often used for same or similar purposes is small-angle neutron scattering (SANS), its basic theory being essentially the same as that developed for the SAXS technique.

Choice of Sample Thickness Classical SAXS experiments are performed in transmission mode and usually under normal incidence. The first step for planning SAXS experiments is to determine the sample thickness that maximizes the scattering intensity for a given material and photon energy. The SAXS intensity produced by any material with arbitrary structure, as a function of sample thickness t, is given by I ðtÞ / teμρt

(48)

where ρ is the mass density and μ the mass X-ray absorption coefficient, which is a function of chemical composition of the material and photon energy. The absorption coefficient can be obtained from tables published by Henke et al. (1993) or by using an online program accessible in their web page. Examples of the function defined by Eq. 48 are plotted in Fig. 17 for three different materials. The optimum thickness tmax corresponding to the maximum of the I(t) function is tmax ¼ ðρμÞ1

(49)

This implies that the transmittance of samples with optimum thickness is T ¼ ðI transmitted =I incident Þ ¼ e1 ¼ 0:37:

(50)

Notice that the tmax values determined by Eq. 49 are just a guide for a convenient choice of sample thickness. However, it is always advisable to avoid the use of very thick or very thin samples which would lead to high absorption and low probed volumes, respectively, both yielding weak scattering intensities. For samples containing large fractions of high Z atoms the optimum thicknesses could be extremely low using CuλKα photons (E = 8.04 KeV). For these

Small-Angle X-ray Scattering by Nanostructured Materials

41

Fig. 17 (a) Examples of scattering intensities in arbitrary units as functions of sample thickness for an incident X-ray beam with a wavelength λCuKα = 1.542 Å, corresponding to different selected materials: Cu, SiO2, and H2O, whose optimum thicknesses tmax are 22 μm, 132 μm, and 1.00 mm, respectively

materials X-ray beams with higher photon energy should be employed. On the other hand, in order to minimize fluorescence effects, the use of beams with photon energies above and close to absorption edges of sample elements should be avoided.

Subtraction of Parasitic Scattering Before further analysis of experimental SAXS results, a pretreatment of rough data is ! required. For anisotropic 2D SAXS patterns, the vector q associated to each detector pixel is calculated. For isotropic 2D SAXS patterns, the scattering intensity is defined as a function of the modulus of the scattering vector, which is determined by circular averaging. In order to subtract the parasitic scattering intensity produced by slits, cell windows, and air, two SAXS patterns should be recorded: (i) the total scattering
 !

intensity (from sample plus parasitic scattering) defined by the counting rate RT q
 ! and (ii) the parasitic scattering intensity given by the counting rate RP q recorded under same experimental conditions but without sample. The scattering intensity exclusively related to the sample is given by

42

A.F. Craievich


 h

  i h

  i ! ! ! R q ¼ RT q  RD =T SþW  RP q  RD =T W

(51)

where RD is the counting rate associated to the detector noise, TS+W is the transmittance of the sample and cell thin windows, and TW is the transmittance of the empty sample cell. For solid samples placed in a windowless holder, we have TW = 1. Often the counting rate associated to parasitic scattering for macromolecules in dilute solution is determined with the sample cell filled with same buffer, thus under this condition the scattering intensity due to statistical density fluctuations in the solvent is also subtracted. When SAXS experiments are conducted using synchrotron beam lines with continuously decreasing electronic current, the effects of time variation of the intensity of the incident X-ray beam should be properly accounted for.

Correction of Smearing Effects The use of X-ray incident beam with rather large cross-section and/or X-ray detectors with large pixel size may produce serious smearing effects on the SAXS curves. However, most of the modern commercial setups and synchrotron beam lines provide an incident beam with pinhole-like cross-section and use X-ray detectors with very small pixel size, thus often making mathematical desmearing procedures unnecessary. When using commercial setups yielding an incident X-ray beam with large cross section (for example a beam with linear cross-section), two approaches can be applied for quantitative analyses: (i) fitting the theoretical model of SAXS curve to previously dismeared experimental functions or (ii) fitting the previously smeared theoretical model of SAXS curve to the experimental function. Since mathematical desmearing of experimental SAXS patterns leads to results with rather high statistical noise, the second procedure is generally preferred.

Determinations of SAXS Intensity in Relative and Absolute Units For pin-hole collimation of the incident beam, the counting rate


 ! R q

corresponding to the X-ray photons scattered by the sample is proportional to the
 ! I q function used along this chapter. Thus Eq. 51 directly yields the scattering intensity in relative scale or arbitrary units, to which model functions are fitted after adequate scaling. However, the SAXS intensity given in absolute scale provides additional information that is often useful for detailed structural characterization. The typical scattering intensity function in absolute scale is the differential scattering cross-section per unit volume (dΣ/dΩ). This function is
related
  to
theSAXS intensity !

!

I q , which was used along this chapter, by ðdΣ=dΩÞ q

!

¼ I q :r 2e =V.

Small-Angle X-ray Scattering by Nanostructured Materials

43

For SAXS measurements using pin-hole collimation (i.e., with a point-like incident beam cross-section), the differential scattering cross-section per unit volume is given by
 !
 R q =η dΣ ! q ¼ dΩ I 0 V:ΔΩ

(52)


 ! where R q is the photon counting rate, η is the detector efficiency, I0 is the photon flux of the incident X-ray beam (number of photons per unit cross-section.second), V is the probed sample volume, and ΔΩ is the solid angle associated to the surface area of the detector pixel. The usual unit for the differential scattering cross-section per unit volume is cm1. Equation 52 can also be written as
 !
 R q :L2 dΣ ! q ¼ dΩ R0 ts Δa

(53)

where R0 is the counting rate (number of photons/second) corresponding to the total incident beam, ts is the sample thickness, Δa is the surface area of the detector pixel, and L is the sample-to-detector It is assumed in Eq. 53 that the efficiency of
distance.  !

the detectors that records R q and Ro are identical. When different detectors are
 ! used for the measurements R q and Ro, the counting rates should be properly normalized to equivalent efficiencies. Equation 53 is usually applied to plate-shaped solid samples or to liquids contained in cells with parallel thin windows for entrance of the incident X-ray beam and exit of the scattered photons. Determinations of SAXS intensity in absolute units associated to powdered samples or liquid samples contained in cylindrical capillaries are also possible but their evaluation is less precise (Fan et al. 2010). Since the measurement of Ro is in practice difficult using standard detectors, the differential scattering cross-section per unit volume of solid materials is generally determined by means of an independently calibrated sample, such as Lupolen or glassy carbon (Fan et al. 2010). In order to determine the differential scattering cross-section per unit volume associated to colloidal particles embedded in a liquid medium, it is also recorded – under the same experimental conditions – the SAXS intensity produced by statistical density fluctuations in water. The differential scattering cross-section per unit volume of water ðdΣ=dΩÞH2 O – which is a isotropic and constant function at small q – is given by (Guinier and Fournet 1955):

44

A.F. Craievich



dΣ dΩ

ðq ! 0Þ ¼ ðN H2 O ne r e Þ2 kTβ

(54)

H2 O

where N H2 O is the number of water molecules per unit volume, ne is the number of electrons per water molecule, k is the Boltzmann constant, T is the absolute temperature, and β is the isothermal compressibility of water at room temperature. Since all parameters in Eq. 54 are known, the differential scattering cross-section per unit volume of water can be written as

dΣ dΩ



ðq ! 0Þ ¼ 1:65:102 cm1

(55)

H2 O

If the counting rate associated to a isotropic liquid sample (for example proteins in liquid buffer), [R(q)]sample, and that corresponding to water, RðqÞH2 O, are determined under same experimental conditions, the differential scattering cross section per unit volume of the studied sample is given by ½RðqÞsample dΣ ðqÞ ¼ 1:65:102 cm1 dΩ < Rð q Þ H 2 O >

(56)

where < RðqÞH2 O > is an average value taken within the small q range over which the counting rate is approximately constant. If SAXS measurements corresponding to sample and water are conducted under different experimental conditions, adequate corrections should be applied. Additional details on this matter were reported by Fan et al. (2010).

References Beaucage G, Ulibarri T, Black EP, Shaeffer DW. Chapter 9. Multiple size scale structures in silicasiloxane composites studied by small-angle scattering. In: Mark JE, Lee CYC, Bianconi PA, editors. Hybrids organic–inorganic composites, vol. ACS series 585. Washington, DC: American Chemical Society; 1995. p. 97–111. Cahn JW. Phase separation by spinodal decomposition in isotropic systems. J Chem Phys. 1965;42:93. Ciccariello S. The leading asymptotic term of the small-angle intensities scattered by some idealized systems. J Appl Crystallogr. 1991;24:509–15. Ciccariello S, Schneider JM, Schonfeld B, Kostorz G. Illustration of the anisotropic Porod law. J Appl Crystallogr. 2002;35:304–13. Craievich AF, Alves OL, Barbosa LC. Formation and growth of semiconductor PbTe nanocrystals in a borosilicate glass matrix. J Appl Crystallogr. 1997;30:623–7. Craievich AF, Fischer H. Quantitative analysis and relevant features of the scientific literature related to SAXS and SANS. J Phys Conf Ser. 2010;247:012003. Craievich AF, Sanchez JM. Dynamical scaling in the glass system B2O3–PbO–Al2O3. Phys Rev Lett. 1981;47:1301311.

Small-Angle X-ray Scattering by Nanostructured Materials

45

Dahmouche K, Santilli CV, Pulcinelli SH, Craievich AF. Small-angle X-ray scattering study of sol–gel-derived siloxane-PEG and siloxane-PPG hybrid materials. J Phys Chem B. 1999;103:4937–42. Debye P, Bueche AM. Scattering by an inhomogeneous solid. J Appl Phys. 1949;20:51525. Fan L, Degen M, Bendle S, Grupido N, Ilavsky J. The absolute calibration of a small-angle scattering instrument with a laboratory X-ray source. J Phys Conf Ser. 2010;247:012005. Glatter O, Kratky O, editors. Small-angle X-ray scattering. London: Academic; 1982. Goerigk G, Haubold HG, Lyon O, Simon JP. Anomalous small-angle X-ray scattering in materials science. J Appl Crystallogr. 2003;36:425. Guinier A, Fournet G. Small-angle scattering of X-rays. New York: Wiley; 1955. Henke BL, Gullikson EM, Davis JC. X-ray interactions: photoabsorption, scattering, transmission and reflection at E = 50-30000 eV and Z = 1-92. Atomic Data and Nuclear Data Tables. 1993;54:181–342. http://henke.lbl.gov/optical_constants/ Hexemer A, Muller-Buschbaum P. Advanced grazing incidence techniques for modern soft-matter materials analysis. IUCrJ. 2015;2:106–25. Kellermann G, Montoro LA, Giovanetti LJ, Santos Claro PC, Zhang L, Ramirez AJ, Requejo FG, Craievich AF. Formation of an extended CoSi2 thin nanohexagons array coherently buried in silicon single crystal. Appl Phys Lett. 2012;100:063116. Kellermann G, Montoro LA, Giovanetti LJ, dos Santos Claro PC, Zhang L, Ramirez AJ, Requejo FG, Craievich AF. Controlled growth of extended arrays of CoSi2 hexagonal nanoplatelets buried in buried in Si(001), Si(011) and Si(111) wafers. Phys Chem Chem Phys. 2015;17:4945–51. Koch MHJ, Vachette P, Svergun DI. Small-angle scattering: a view on the properties, structures and structural changes of biological macromolecules in solution. Q Rev Biophys. 2003;36:147–227. Kohlbrecher J, Bressler I. Software package SASfit for fitting small-angle scattering curves. 2014. http://kurweb.psi.ch/sans1/SANSSoft/sasfit.html Kustch B, Lyon O, Schmitt M, Mennig M, Schmidt H. Small-angle X-ray scattering experiments in grazing incidence on sol–gel coatings containing nano-scaled gold colloids: A new technique for investigating thin coatings and films. J Appl Crystallogr. 1997;30:94956. Lebowitz JL, Marro J, Kalos MK. Dynamical scaling of structure-function in quenched binaryalloys. Acta Metall. 1982;30:297–310. Lecomte A, Dauger A, Lenormand P. Dynamical scaling property of colloidal aggregation in a zirconia-based precursor sol during gelation. J Appl Crystallogr. 2000;33:496–9. Lindner P, Zemb T, editors. Neutron, X-ray and light scattering. Amsterdam: North Holland; 1991. Marro J, Boltz AB, Kalos MH, Lebowitz JL. Time evolution of a quenched binary alloy. II. Computer simulation of a three-dimensional model system. Phys Rev B. 1975;12:2000–11. Meakin P. In: Stanley HE, Ostrowsky N, editors. On growth and form. Boston: Martinus Nijhoff; 1986. p. 111–35. Porod G. Chapter 2: General theory. In: Glatter O, Kratky O, editors. Small-angle X-ray scattering. London: Academic; 1982. Riello P, Minesso A, Craievich AF, Benedetti A. Synchrotron SAXS study of the mechanisms of aggregation of sulfate zirconia sols. J Phys Chem B. 2003;107:3390–9. Ruland W. Small-angle scattering of 2-phase systems. Determination and significance of systematic deviations from Porod’s law. J Appl Crystallogr. 1971;4:70. Russell TP. Chapter 11: Small-angle scattering in synchrotron radiation sources. In: Brown GS, Moncton DE, editors. Handbook on synchrotron radiation, vol. 3. Amsterdam: North Holland; 1991. Santilli CV, Pulcinelli SH, Craievich AF. Porosity evolution in SnO2 xerogels during sintering under isothermal conditions. Phys Rev B. 1995;51:8801–9. Schnablegger H, Singh Y. The SAXS guide. Graz: Anton Paar GmbH; 2013. Shull CG, Roess LC. X-ray scattering at small angles by finely-divided solids. I. General approximate theory and applications. J Appl Phys. 1947;18:295–307.

46

A.F. Craievich

Silva NJO, Dahmouche K, Santilli CV, Amaral VS, Carlos LD, V. BZ, Craievich AF. Structure of magnetic poly(oxyethylene)-siloxane nanohybrids doped with Fe-II and Fe-III. J Appl Crystallogr. 2003;36:961–6. Svergun DI. Determination of the regularization parameter in indirect-transform methods using perceptual criteria. J Appl Crystallogr. 1992;25:495–503. Svergun DI. Restoring low resolution structure of biological macromolecules from solution scattering using simulated annealing. Biophys J. 1999;77:2879–86. Svergun DI, Semenyuk A. Small-angle scattering data processing using the regularization technique.1991. www.embl-hamburg.de/biosaxs/gnom.html Teixeira J. Small-angle scattering by fractal systems. J Appl Crystallogr. 1988;21:781–5. Tokumoto MS, Pulcinelli SH, Santilli CV, Craievich AF. SAXS study of the kinetics of formation of ZnO colloidal suspensions. J Non-Cryst Solids. 1999;247:176–82. Tolan M. X-ray scattering from soft-matter thin films. Berlin: Springer; 1999. Yoneda Y. Anomalous surface reflection of X-rays. Phys Rev. 1963;131:2010–3.

Porosity Measurement Kazuki Nakanishi

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Density Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Determination of Pore Volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Pycnometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Liquid Intrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Capillary Condensation of Gases (Gas Adsorption) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Three-Dimensional Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Abstract

Methods of determining porosity of solid materials are described. Since the spectrum of pore size spreads from sub-nanometers to millimeters, appropriate methods should be adopted to adequately determine the volume and size of the pores contained in the solid materials. Representative methods including liquid intrusion and gas adsorption are concisely explained assuming the samples are obtained via sol–gel routes. Recent direct method of determining the threedimensional configurations of pores and solid parts will also be introduced.

Introduction Since spaces of various amount and size are left within the gel network when gels are dried from their “wet” state (Brinker and Scherer 1990), gels maintain their inherent porous structure even after the solvent removal. In the case of polymeric or organic K. Nakanishi (*) Department of Chemistry, Graduate School of Science, Kyoto University, Kyoto, Japan e-mail: [email protected] # Springer International Publishing AG 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_38-1

1

2

K. Nakanishi

gels, it is sometimes true that all the porosity is eliminated during the drying process, because the network is flexible enough to deform to fill any spaces as the solvent escapes from the network. In the case of inorganic gels, however, the network becomes rigid enough to resist the drying stresses in the course of solvent evaporation, so that interstices left within the network become empty pore spaces. For this reason, dried gels, generally called “xerogels,” usually retain only open pores that are accessible from outside. On heat-treating the xerogels, further internal reactions occur to densify the network. These reactions occasionally result in local closure or isolation of the pore spaces, thus producing “closed” pores in the network. Similarly to ordinary ceramic materials, the porosity of gel materials can be determined by density measurement where accurate measurements of mass and volume are required. In addition, independent determination of open pores by gas adsorption or liquid intrusion can reveal more detailed information on the porosity, such as pore size distribution and shape of pores.

Density Measurement For substances with virtually no pores, the accuracy of density measurement depends primarily on that of volume measurement. The volume of a sample with irregular shape can be determined by the so-called Archimedes method. In this method, as shown in Fig. 1, the sample specimen is immersed in a liquid with known density at a constant temperature, and by measuring the buoyancy, the volume of the specimen is calculated. The practical instrumentation consists of a mechanical or electronic balance, where one can hang and immerse the specimen into a liquid reservoir. The wire hanging the specimen from the balance arm should be thin enough, and in each experiment, the same length of wire should be immersed in the liquid. First, the sample specimen is weighed in air and then in the liquid. The hanging wire, immersed to the same depth as in the above measurement without the sample specimen, is also weighed. Consequently, the density of the sample can be calculated by the equation Fig. 1 Experimental setup of density measurement by Archimedes method

Porosity Measurement

3

ρ ¼ ρl W = ðW  ðWl  Ww ÞÞ

(1)

where W is the weight of the specimen in air, Wl is that of the specimen plus wire immersed in the liquid, Ww is that of the wire in the liquid, and ρl is the density of the liquid at the temperature of the measurement. Pores contained in solid samples can be classified into “closed” and “open” pores. The meaning of “closed” and “open” depends on the size of pores into which a probe fluid (gas or liquid) can diffuse. In any case, when the probe fluid cannot penetrate into certain spaces, they are denoted as “closed” pores. Materials such as foamed metals or polymers have a considerable amount of isolated voids, individually surrounded by a dense matrix. Most of their porosities are denoted as “closed” pores. Gels prepared from a relatively soft network tend to retain molecular-scale pores. They are open pores in nature but virtually inaccessible even by small-sized molecules, so that they are regarded as “closed” pores. On the contrary, any space which can be accessed by the probe within the timescale of the measurement will be regarded as “open” pores (see Fig. 2). Considering the contributions of open and closed pores to the total sample volume, two types of density of a porous sample specimen can be defined. The bulk density, ρbulk, takes contributions both from open and closed pores into account. The apparent density, ρapp, considers only the contribution of closed porosity. The definitions of these densities are ρbulk ¼ W=ðV s þ vo þ vc Þ

(2)

ρapp ¼ W=ðV s þ vc Þ

(3)

where W denotes the weight of the sample, Vs is the true volume of the dense (porefree) matrix, and vo and vc are the volumes of open and closed pores, respectively. The total porosity, et, of a porous sample is defined as et ¼ ðvo þ vc Þ=ðV s þ vo þ vc Þ ¼ 1  ρbulk =ρs

Fig. 2 Schematic illustration of open and closed pores in porous materials

(4)

4

K. Nakanishi

where ρs = W/Vs. In ordinary cases, the weight of the sample can be determined with higher accuracy than any of the volumes. When the open pores are small enough to completely reject the penetration of a liquid probe, a simple Archimedes method can be applied to determine the bulk volume, as described above.

Determination of Pore Volume When dealing with gel samples, it often becomes necessary to evaluate the open porosity independently of that of the matrix or skeleton which may include closed porosity. In evaluating the open porosity of real samples, the accessibility of the probe fluid to the pores depends not only on the fluid molecular size but also on the physical transport properties inside the gel network. With a liquid probe, if the probe has considerably higher surface energy than the gel network, the probe fluid cannot go into the pores of smaller size under atmospheric hydrostatic pressure (corresponding to the simple immersion of the gel piece into the probe fluid). Evacuation and external pressurization, to introduce the probe fluid into pores, will help to some extent. Nevertheless, it is generally difficult to confirm if the probe fluid has completely filled all the accessible spaces. In the case of mercury intrusion, due to its exceptionally high surface energy, mercury cannot enter pores smaller than 15 μm in diameter, under atmospheric pressure. On applying additional pressure, mercury gradually penetrates into smaller pores. With commercially available instruments, pores as small as 7 nm in diameter can be measured with over 200 MPa of applied pressure. This means that pores smaller than 7 nm are regarded as “closed” pores by mercury intrusion, and they are not included in the overall open pore volume. As described below, mercury intrusion and gas adsorption are representative methods to access macropores (larger than 50 nm in diameter) and mesopores (2–50 nm in diameter), respectively. Pores smaller than 2 nm in diameter, termed micropores, can be accessed only partly by nitrogen. In analyzing microporosity, the monolayer thickness of nitrogen molecules is usually set at 0.354 nm (Lippens et al. 1964), which means that pores smaller than 0.7 nm in diameter are by no means occupied by condensed nitrogen. With hypercritical gaseous probes, one can detect open pores nearly as small as the size of the probe molecule. In order to detect as small pores as possible, helium gas is usually used as a probe. Under constant temperature, a volumetric determination of the sample volume inaccessible by helium gas is made. Although the density of real silica gels, determined by helium pycnometry, does not necessarily agree with the ideal density of amorphous silica (2.2 g/cm3), possibly due to the closed porosity or microporosity, density values measured with helium are usually regarded as the “skeletal” density of gel materials (Brinker et al. 1986). This is illustrated in Fig. 3. Including the methods partly depicted above, porosity is determined typically by the following techniques:

Porosity Measurement

5

Fig. 3 An example of measured skeletal density of a silica gel heat-treated at various temperatures

1. 2. 3. 4.

Pycnometry Liquid intrusion Capillary condensation of gases (gas adsorption) Three-dimensional imaging

Pycnometry Pycnometry is a method designed to determine the precise mass–volume relationships of liquids or solids. The measurement is performed using a device called a pycnometer, a glass bulb fitted with a stopper through which runs a capillary bore. By filling the inside volume of the pycnometer completely with a liquid, the mass of the liquid corresponding to that volume may be readily determined. To explore the mass–volume relationship of a solid substance requires a few more steps, although the principle is the same. First, a certain amount of solid sample, usually in the form of powder or granules, is placed in a precisely calibrated pycnometer and is weighed (W0 + Ws, where W0 and Ws denote the weights of the empty pycnometer and of the sample, respectively). Then, the remaining space is filled with a liquid of known mass–volume relationships and is weighed (W0 + Ws + Wl, where Wl denotes the weight of filled liquid). For better accuracy and reproducibility, the liquid has to have low enough surface energy to wet the solid surface completely. Using the known density of the liquid, ρl, and the volume of the pycnometer, V0, the volume of the solid can be calculated by Vs ¼ V0  W1 =ρl

(5)

6

K. Nakanishi

Fig. 4 Schematic experimental setup of gas pycnometry (volume measurement)

If the sample has only small pores where the liquid cannot diffuse into, the apparent volume of the sample can be thus determined. Furthermore, when the true density of the gel skeleton is known, the porosity inaccessible by the liquid can be calculated. One can utilize gas pycnometry, typically measured with helium, to determine the skeleton density, as shown in Fig. 4. In this case, the helium gas is assumed to penetrate into all the pores, except those cannot be accessed even by the size of helium. Commercial instrumentation consists of a constant temperature sample chamber and a helium gas reservoir. Into the evacuated sample chamber with a loaded solid sample, helium gas is introduced from the reservoir and equilibrated at known temperature and pressure. Using the equilibrium pressure after the helium introduction, the dead volume of the chamber can be calculated based on the ideal gas law. For higher accuracy, the sample volume should not be much smaller than that of the sample chamber. In some cases, a solid block with known temperature–volume relationship is placed, for the purpose of reducing the dead volume in the sample chamber.

Liquid Intrusion Liquid intrusion and gas adsorption methods both give information not only on the volume but also on the size distribution of pores in a sample specimen. In liquid intrusion method, a liquid with relatively high surface energy (typically and nearly exclusively mercury) is pressurized and penetrates into the open pores. In the case of mercury, the angle of contact of mercury with solids is ~140 (more than 90 ), and

Porosity Measurement

7

Fig. 5 Definitions of physical parameters used in mercury porosimetry. rp pore radius, r1 radius of curvature of mercury meniscus, θ contact angle

therefore an excess pressure Δp is required to force liquid mercury into the pore of a solid. According to Washburn, who first suggested the use of mercury intrusion to measure pore size, the basic equation between applied pressure and the pore radius which was penetrated can be expressed as follows (Washburn 1921): r p ¼ 2γcosθ=Δp

(6)

where rp is the radius of the pore, assumed to be cylindrical, as shown in Fig. 5. This equation is often termed the Washburn equation. The technique of mercury porosimetry consists essentially in measuring the extent of mercury penetration into an evacuated solid as a function of the applied hydrostatic pressure. As described above, from atmospheric pressure to 200 MPa, a corresponding pore diameter from 15 μm to 7 nm can be covered. Although the surface tension and contact angle values are typically set constant at γ = 480 mN m2 and θ = 140 , respectively, relatively large deviations are recognized in θ, varying from 130 to 150 , depending on the surface of the solid samples. Using the value θ = 140 , a 10–20 % potential error in the calculated pore size should be anticipated. The high pressure of mercury intrusion sometimes causes irreversible changes in the pore structure of the sample specimen. There are experiments reporting both increase and decrease in the pore volume, depending on the nature of the sample solid, due to the intrusion. One example of silica gels by Unger and Fischer reports significant decrease in the pore volume, in every type of silica gel particles, after the first intrusion with mercury. Microscopic examination confirmed that the spherical

8

K. Nakanishi

particles were compacted tighter due to the pressure during the intrusion process (Unger and Fischer 1974).

Capillary Condensation of Gases (Gas Adsorption) Multilayer adsorption and capillary condensation of gases within the narrow diameter pores below a few tens of nanometers can also be used to determine the pore size distribution and pore volume. It is believed that nitrogen cannot “see” the pores larger than ca. 50 nm in diameter in ordinary equipments. Therefore, the determination of total porosity should be done by combining the results of other methods, such as mercury intrusion, which can accurately detect larger pores. The IUPAC recommendation on the nomenclature of the “mesopores,” from 2 to 50 nm in diameter, reasonably corresponds to the size range that nitrogen adsorption can reliably cover. Depending on the type of adsorption–desorption isotherms, a care should be taken to estimate true porosity accessed by nitrogen molecules. The five typical isotherms originally classified by Brunauer, Deming, Deming, and Teller (BDDT) (Brunauer et al. 1940), together with type VI stepped isotherm, are shown in Fig. 6 (Gregg and Sing 1982; Rouquerol et al. 1999). When the porosity exists only in the mesopore range (not extending into the macropore range), isotherms at higher pressure tend to be flattened. Especially in type I isotherm, indicative of the microporosity only, the capillary condensation (or possibly, monolayer adsorption) of the adsorbate is completed within a low relative pressure range. When the sample has mesopores, type IV isotherm is obtained. The uptake at lower relative pressure is attributed to the monolayer adsorption, and the second uptake at higher relative pressure corresponds to the capillary condensation within the mesopores. The flattened portion of the isotherm can be reliably used to determine the pore volume of the mesopores. Type V isotherm also indicates the presence of mesopores in the sample; however, due to relatively weak attractive interaction between the solid surface and adsorbed molecules, the uptake at lower relative pressure is gradual. This does not harm the determination of the volume of mesopores. Isotherms with infinitely increasing portion at higher relative pressure range, types II, III, and VI, should be interpreted with care. These isotherms generally indicate the presence of pores extending into the macropore range, where nitrogen adsorption cannot be correctly applied. In commercial instrumentations, the total pore volume may be automatically reported, setting the final data point at the highest experimental relative pressure, within the accuracy of the instrument. The evaluation of the total porosity, therefore, should be done with caution.

Three-Dimensional Imaging The methods depicted in sections “Pycnometry,” “Liquid Intrusion,” and “Capillary Condensation of Gases (Gas Adsorption)” require so-called “bulk” samples, where

Porosity Measurement

9

Fig. 6 Various kinds of adsorption isotherms classified by Brunauer et al. (1940)

the amount of probe molecule is in the mmol to μmol range. When the thickness of films or fibers becomes very small, reliable measurements cannot be carried out, due to the limit of absolute detection. In addition, these methods give only averaged information over the whole loaded samples, and no local deviation or heterogeneity, such as that across the film thickness, can be detected. For such purposes, methods utilizing 3-D imaging have become recently available (Jinnai et al. 2000; Koster et al. 2000). Several kinds of microscopic methods which can construct a 3-D image of the solid sample have been developed. Depending on the spatial resolution obtained in the 2-D observation, the reliable size of pores and pore volumes vary to a large extent. Typically, the finest resolution is a few micrometers for X-ray tomography, a few hundred nanometers for laser scanning confocal microscope (LSCM), and below 10 nm for high-resolution transmission electron microscopy (HR-TEM) (Midgley and Weyland 2003). In addition, there are special requirements, depending on the method of detecting the interfacial information. The X-ray tomography offers nondestructive observation of the 3-D sample, with its maximum sample size depending only on the size of the detecting device (imaging plates, in advanced instrumentation). Since the contrast is mainly determined by the electron density

10

K. Nakanishi

Fig. 7 Structural analysis of macroporous materials by laser scanning confocal microscopy (LSCM)

differences in the sample, the method is more suitable for ceramics or metallic materials than for polymers. LSCM requires transparency of the whole sample for accurate imaging in the depth direction. Contrast matching can be carried out, by impregnating the sample with a fluid having appropriate refractive index, only when the pores are fully open. TEM requires sample preparation into very thin slices. A practical example of 3-D imaging, using LSCM and analysis of the interfacial structure, including porosity, is presented below.

Example of Macroporous Silica One of the 2-D sliced images of the macroporous siloxane gel is shown in Fig. 7 together with the image processing scheme to reconstruct the 3-D image. In steps 1 and 2, each raw 2-D image is treated with 3-D median filter to remove noise and then digitized into a black-and-white image. Several tens of 2-D sliced images thus obtained are stacked and reconstructed into the 3-D image of, typically, several tens of micrometers in thickness. Once the 3-D reconstruction is completed, that is, every point in the specified sample is attributed to either black or white, one can geometrically define the pore structure three-dimensionally. The volume fractions of pores and skeleton are simply determined as the number fractions of black-and-white points. The measurement of detailed geometrical parameters, such as surface curvature distribution over the sample, also becomes possible (Jinnai et al. 2001). It is also beneficial to simulate the structure function, using Fourier transformation, to assess the statistical heterogeneity of the structure. Since the digitization step critically determines the pore surface geometry in this method, the calibration of digitization conditions using porosimetry data is recommended for the purpose of obtaining reliable absolute porosity values.

Porosity Measurement

11

Conclusion Porosity measurements include conventional techniques to assess pore volume and pore size distribution of mainly bulk gel samples. Although several physical properties, e.g., refractive index, show simple dependence on averaged porosity of the sample, it will become increasingly important to analyze the local pore structure of thin or small amount of samples. In the near future, the three-dimensional imaging with HR-TEM observation will become an important tool for researchers dealing with porous gels.

References Brinker CJ, Scherer GW. Sol–gel science: the physics and chemistry of sol–gel processing. San Diego: Academic; 1990. Brinker CJ, Roth EP, Tallant DR, Scherer GW. In: Hench LL, Ulrich DR, (eds) Science of ceramic chemical processing. New York: Relationships Between Sol to Gel to Glass Conversions: Structure of Gels During Densification. Wiley; 1986. p. 37–51. Brunauer S, Deming LS, Deming WS, Teller EJ. On a Theory of the van der Waals Adsorption of Gases. Am Chem Soc. 1940;62:1723. Gregg SJ, Sing KSW. Adsorption, surface area and porosity. 2nd ed. London: Acadmic; 1982. Jinnai H, Nishikawa Y, Spontak RJ, Smith SD, Agard DA, Hashimoto T. Direct measurement of interfacial curvature distributions in a bicontinuous block copolymer morphology. Phys Rev Lett. 2000;84:518–521. Jinnai H, Nakanishi K, Nishikawa Y, Yamanaka J, Hashimoto T. Three-dimensional structure of a sintered macroporous silica gel. Langmuir. 2001;17:619–25. Koster AJ, Ziese U, Verkleij AJ, Janssen AH, de Jong KP. Three-dimensional transmission electron microscopy: a novel imaging and characterization technique with nanometer scale resolution for materials science. J Phys Chem B. 2000;94:9368. Lippens BC, Linsen BG, de Boer JH. Studies on pore systems in catalysts I. The adsorption of nitrogen; apparatus and calculation. J Catal. 1964;3:32. Midgley PA, Weyland M. 3D electron microscopy in the physical sciences: the development of Z-contrast and EFTEM tomography. Ultramicroscopy. 2003;96:413. Rouquerol F, Rouquerol J, Sing KSW, editors. Adsorption by powders and porous solids: principles, methodology and applications. London: Academic; 1999. Unger K, Fischer H. In: Modry S, Svata M, editors. Proceedings of RILEM/IUPAC international symposium on pore structure and properties of materials, D-127, vol. 5. Prague: Academia; 1974. Washburn EW. Note on a method of determining the distribution of pore sizes in a porous material. Proc Natl Acad Sci U S A. 1921;7:115–116.

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials Kiyoharu Tadanaga and Tsutomu Minami

Contents Gas Adsorption Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Physical Gas Adsorption Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Chemical Gas Adsorption Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Identification of Adsorbed Surface Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Permeation Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Mechanism of Permeation in Porous Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Mechanisms for Nonporous Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Abstract

Many sol–gel-derived materials are porous. Surface area, pore size, and porosity are very important properties for the application of porous materials, and those properties are studied by gas adsorption measurements. The sol–gel-derived materials are sometimes used as gas separation or gas barrier membranes. In this chapter, background and measurement techniques of gas adsorption for

K. Tadanaga (*) Faculty of Engineering, Hokkaido University, Hokkaido, Japan e-mail: [email protected] T. Minami (*) Osaka Prefecture University, Osaka, Japan e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_39-1

1

2

K. Tadanaga and T. Minami

porous sol–gel-derived materials are firstly described. Then mechanism of permeation in porous and nonporous media and gas permeability measurement techniques for membranes prepared by sol–gel processes are described.

Gas Adsorption Measurements Introduction Since sol–gel-derived materials are usually porous, determination of surface area or pore size distribution by gas adsorption is very important to characterize the materials obtained. On the other hand, identifying the surface species formed by molecular adsorption and the species generated by surface reactions is also important in the field of catalysts. In this section, the phenomenon of gas adsorption is briefly reviewed, and measurement techniques of gas adsorption are introduced.

Background When a gas or vapor comes in contact with a solid, some molecules attach to the surface of the solid. This phenomenon is termed “adsorption.” The solid that takes up the gas is called the adsorbent and the gas or vapor taken up on the surface is called the adsorbate. When adsorbed species detached from the surface of the solid and return into the gas phase, this phenomenon is called “desorption.” “Physical adsorption” is caused by van der Waals force and electrostatic forces between adsorbate molecules and the atoms which compose the adsorbent surface (Suzuki 1990). Thus, adsorbents are conveniently characterized by surface properties such as surface area and polarity. In “chemical adsorption,” the adsorbate molecules attach to the surface by forming a chemical (usually covalent) bond. The amount of adsorption or desorption depends on the pressure of the gas, and the relation between the amount of adsorption/desorption and the pressure, at a certain temperature, is called “adsorption/desorption isotherm.” An adsorption/ desorption isotherm is usually recorded as volume of gas adsorbed versus relative pressure (i.e., sample pressure/saturation vapor pressure). As shown below, the shape of adsorption/desorption isotherm varies with the combination of adsorbent and adsorbate. “Pore” is defined as a hole, in which the depth of the hole is larger than its diameter. Materials without pores are called “nonporous” materials and those with pores are called “porous.” Pores are classified according to their diameter, as shown in Table 1. Braunauer et al. classified the adsorption isotherms typically into five types (Brunauer et al. 1938; Adamson 1990), which are shown in Fig. 1. Type I is the Langmuir-type adsorption and corresponds to a monolayer formation. This isotherm

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials Table 1 Classification of pores

50 nm

Ultramicropore Supermicropore Mesopore Macropore

Type II

Type III

Vads

Type I

3

p / p0

p / p0

Type V

Vads

Type IV

p / p0

p / p0

p / p0

Fig. 1 Brunauer’s five types of adsorption isotherms

shape, with very high adsorption at low relative pressures, is typically observed for microporous solids (pore diameter 0.4 region. In the nitrogen adsorption measurements on sol–gel-derived materials with meso- and macropores, type IV is often observed. To determine the surface area and pore size distribution by gas adsorption/ desorption measurements, nitrogen is most commonly used as the adsorbate, since liquid nitrogen is readily available and inexpensive. Total surface area can be determined from the isotherm between 0.05 and 0.3 relative partial pressure, at which point a complete monolayer has been formed on the sample surface. At higher partial pressures, the pores start to be filled by capillary condensation. From an analysis of the shape of the full isotherm, the distribution of pore sizes and total pore volume can be determined. The adsorption process is generally taken as completely

4

K. Tadanaga and T. Minami

reversible, but under some conditions the isotherm may exhibit a different shape in desorption, as compared to absorption. This is called hysteresis. Sometimes, hysteresis data can be used to determine the structure and size of pores in the absorbent. Determinations of pore size distribution using adsorption/desorption isotherms are described in previous “Porosity Measurement” chapter.

Physical Gas Adsorption Measurements In physical gas adsorption measurements, an inert gas such as N2 (or Ar, Kr, CO2, etc.) is used for adsorption measurements on a solid material. Before the measurement, the sample is pretreated at an elevated temperature, in vacuum or a flowing gas, to remove any contaminations. In sol–gel-derived materials, this process is quite important, since the materials are usually porous and adsorb water and other vapors from ambient atmosphere. Thus, pretreatment at an elevated temperature in a vacuum is highly recommended, rather than with a flowing gas. Two practical methods are often used for gas adsorption studies, the dynamic flow and static–volumetric method. In the dynamic flow technique, a gas mixture of the desired concentration of adsorbate gas, such as nitrogen balanced with a non-adsorbing gas, usually helium, is flown across the sample. Nitrogen is adsorbed on the surface and the total amount is detected by a thermal conductivity detector. For each point of the isotherm, a different gas mixture and measurement cycle will be conducted. In this technique, a single-point measurement can be made to determine BET surface area, in a very short time, with certain assumptions (ASTM D4567, D5604). In the static–volumetric technique, the sample chamber is evacuated, and the pure adsorbate gas is introduced. The gas will be adsorbed on the sample surface. As more gas is added, more of it remains available to create pressure within the sample cell. The pressure in the sample cell is measured with a sensitive pressure transducer, and the pressures are compared to the total amount of gas added, in order to generate the isotherm. With a knowledge of the average molecular cross-sectional area of the adsorbent, it is possible to calculate the total and specific surface area. The nitrogen BET method is convenient for the measurement of larger specific surface areas, i.e., larger than about 10 m2/g. Besides, other gases with a much lower saturation vapor pressure than nitrogen at the temperature of adsorption may be applied. Krypton is frequently used for the measurement of specific surface areas in the range of about 1 m2/g. The determination of surface areas of catalyst and catalyst carriers, which have type II or IV nitrogen adsorption isotherms and at least 1 m2/g of surface area, is described in ASTM D3663 and the determination of adsorption and desorption isotherms in ASTM D4222. For the precipitated silica, determination of BET surface area is described in ASTM D1993. For calibrating and checking instruments used to determine the specific surface area, standard reference materials for surface area can be obtained from the National Institute of Standards and Technology, USA.

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials

5

Adsorption measurements of several gases on a surface give information about the microstructure of the solid, according to their difference in average molecular cross-sectional area or affinities between surface and gas molecules. Brinker et al. (1993, 1994) report the measurements of N2 and CO2 adsorption/desorption isotherms on thin films, by using a surface acoustic wave (SAW) technique. Thin films were directly deposited on a piezoelectric SAW substrate, and the volume of adsorbed gas was estimated by the change of SAW frequency at different relative pressures. Okada et al. (2002) report the study of acidic and basic gas adsorption properties of Mg–Al-layered double hydroxide and aluminosilicate gels, using a thermal programmed desorption technique for CO2 and NH3 gases. The fractal analysis of microporous solids, such as sol–gel-derived silica, by gas adsorption of n-alkanes of varying molecular sizes, has been proposed (Sermon et al. 1994; Venkatraman et al. 1996; Tatlier and Erdem-Senatalar 1999). Surface adsorption properties of silica xerogels were also investigated using N2, SF6, and CO2 (Meixner and Dyer 1999), where adsorption isotherms of these gases were measured and surface adsorption properties of silica xerogels were discussed.

Chemical Gas Adsorption Characterization In chemical gas adsorption (or chemisorption), the adsorbate molecules attach to the adsorbent surface by forming a chemical bond. For example, a reactive gas as hydrogen or carbon monoxide is used to obtain information on the active properties of the metal phase of a metal-supported catalyst. The sample is first reduced in hydrogen and then evacuated to retrieve the active metal phase. In the volumetric method, known amounts of hydrogen (Pt, Ni, Rh, Ru) or CO (Pd, Pt) are dosed and subsequently adsorbed at different partial pressures, resulting in a chemisorption isotherm. This isotherm measurement is repeated after applying an evacuation step at the analysis temperature, to remove weakly adsorbed species (back-sorption method). The difference between the two isotherms represents the chemically bonded reactive gas and is used in the calculations.

Identification of Adsorbed Surface Species To identify the surface species generated with molecular adsorption and the species generated by surface reactions, vibrational spectroscopy provides the most definitive method. In principle, any technique that can be used to obtain vibrational data from solid-state or gas-phase samples can be applied to the study of surfaces. Infrared (IR) spectroscopy is often used for the identification of surface species. Transmission IR spectroscopy is often used for studies on supported metal catalysts where the large metallic surface area allows a high concentration of adsorbed species to be examined. For powdered sample, diffuse reflectance IR spectroscopy is also used. In this experiment, the diffusely scattered IR radiation from a sample is collected and analyzed. This modification of the IR technique can be employed

6

K. Tadanaga and T. Minami

with high surface area catalyst samples which are not sufficiently transparent in the IR region. For solid samples with a low surface area, such as thin films or single crystals, attenuated total reflection (ATR) method is often used. In this technique, a plate of a single-crystal waveguide such as Ge is placed in contact with the sample, and the IR beam is passed through the IR-transmitting crystal in a manner such that it alternately undergoes total internal reflection from its front and rear faces. At each reflection, some of the IR radiation may be absorbed by species adsorbed on the solid sample surface. For example, the adsorption behavior of NO on a sol–gel-derived thin film, examined by diffuse reflectance IR spectroscopy, has been reported (Fujimoto et al. 2000). In the measurement, a furnace with a KBr window was placed in the Fourier transform infrared (FTIR) spectrometer, and the thin film was set on a small pan in the furnace. The preheated sample was exposed to a NO/He stream and, after purging with a N2 and O2 mixture gas, the spectra were taken at the elevated temperatures. In another example, the type of acid sites (Bronsted and Lewis) on sol–gel-derived tungstophosphoric acid/ZrO2 catalysts was determined with a Fourier transform infrared (FTIR) spectrometer by means of pyridine adsorption (Hernandez-Cortez et al. 2003).

Permeation Measurements Introduction Many sol–gel products are porous materials and these are often studied as gas separation membranes (Klein and Giszpenc 1990; Yoshioka et al. 2001). A very good review on microporous inorganic membranes, including sol–gel-derived ones, was reported (Lin et al. 2002). In porous membranes, the gas transport mechanism can be derived from gas-phase transport and transport through the solid. However, in some organic–inorganic hybrid materials, the hybrid is dense enough and no pore space is available for gas diffusion. Thus, transport in these membranes is achieved by a solution–diffusion mechanism, due to the fact that gasses are soluble in the membrane matrix. Organic–inorganic hybrid materials were studied as a gas barrier coating on polymer substrates (Tohge et al. 1996; Tadanaga et al. 1996; Iwashita et al. 1997; Amberg-Schwab 1998, 2000; Azuta et al. 1999; Leterrier 2003; Jang and Kim 2007). In this section, mechanisms of permeation of gas in porous media and nonporous media are firstly discussed, and evaluation methods for permeation are then introduced. The permeation rate is basically defined as the volume of the gas permeated in a unit period of time, per unit area of the membrane. The permeation coefficient is obtained by normalizing the permeation rate by the thickness of the membrane and the difference of pressure between both sides of the membrane. In the studies of gas

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials

7

separation using membranes, separation efficiency aij, based on the different rates of permeation of the gas components, is often reported: a
ij ¼

Pi Pj

(1)

Mechanism of Permeation in Porous Media In porous media, the mechanism of permeation of fluids is dependent on the pore size. If the pore diameter is much larger than the mean free path of a gas, the mechanism of flow is governed by viscous flow. In this mechanism, permeability is inversely proportional to the pressure and proportional to the temperature. Contrary, in the case of the pore diameter being much smaller than the mean free path, the mechanism is governed by Knudsen flow. In sol–gel-derived materials, the pore size is usually in the range of the Knudsen flow. In the Knudsen flow mechanism, the permeability (P) is described by the following equation: P¼

C ðMRT Þ1=2

(2)

where C is a parameter depending on the pore geometry and M is the molecular weight of gas. This equation shows that the permeability is independent of the pressure difference between both sides and is proportional to T1/2 and M1/2. Thus, gas separation between two gases A and B can be achieved according to their molecular mass: PoA ¼ PoB



MB MA

1=2

(3)

For microporous materials that have a pore size nearly equal to the diameter of the permeating molecules, a molecular sieving-type separation, which is due to the difference in molecular size and shape, should be expected. In this mechanism, diffusion on the surface is very important, and the permeation coefficient has a tendency to increase with an increase in the temperature. Increasing temperature enhances the mobility (diffusivity) of permeating species. The permeation coefficient (Pa) in this system can be expressed by the following equation:   Ea Pa ¼ Ca exp RT

(4)

where Ca is a constant depending on the system and Ea is the apparent activation energy for surface diffusion.

8 3.0

P×106 [cm3(STP) cm/(cm2 s cmHg)]

Fig. 2 Relationship between permeability and gas molecular weights of an alumina layer at 25  C (Okui et al. 1995)

K. Tadanaga and T. Minami

He

2.0

CH4 1.0

N2 O2

CO2 Ar

0.0 0.0

0.1

0.2

0.3

0.4

0.5

0.6

1/M

When a gas is condensable (i.e., a vapor), the capillary condensation flow takes place in the Knudsen flow media, and the permeability coefficient varies with pressure. In the case where surface flow or capillary condensation flow occurs, the affinity of gas molecules with the surface of pores becomes important. Figure 2 is an example of permeability of a membrane of porous alumina (Okui et al. 1995), where the relationship between permeability and gas molecular weight is shown. Nitrogen adsorption isotherms showed that the pore size of the alumina film was about 3–4 nm and the permeability is inversely proportional to the square root of the gas molecular weight, suggesting that the permeation is governed by Knudsen flow. Figure 3 shows the relationship between permeability and gas molecular weight of organic–inorganic hybrid film membrane from phenyltrimethoxysilane. In this case, the pore size of the hybrid film was around 2 nm, and the permeability of carbon dioxide and oxygen was higher than that of nitrogen, suggesting that the permeation was not governed by Knudsen flow, but by surface flow and that CO2 and oxygen should have a strong affinity for the surface of the hybrid.

Mechanisms for Nonporous Materials In nonporous materials, gas permeation is described by a solution–diffusion mechanism. Each gaseous component being transported through a nonporous material has a characteristic permeation rate that is a function of its ability to dissolve and diffuse through the membrane material. The two relationships are expressed by Fick’s law (diffusion) and Henry’s law (solubility). Thus, permeation (P) through the membrane is a function of solubility (S) and diffusivity (D): Pi ¼ Di Si

(5)

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials 8.0

P×108 [cm3(STP) cm/(cm2 s cmHg)]

Fig. 3 Relationship between permeability and gas molecular weight of a hybrid membrane with phenyl groups at 25  C (Okui et al. 1995)

9

CO2 6.0 He 4.0

2.0 O2

CH4 N2

0.0 0.0

0.1

0.2

0.3

0.4

0.5

0.6

1/M

Measurements To measure the permeation of a gas, the apparatus usually consists of a sample holder, a gas supplying system, a pressure measurement system, and a measurement for the flow rate of the gas passed through the sample. A standard test method for determining gas permeability of polymer films is described in ASTM D1434, or DIN 53380. The apparatus used in our laboratory to measure water vapor permeability is schematically shown in Fig. 4 (Tohge et al. 1996; Tadanaga et al. 1996) as an example. The sample to be evaluated is placed between two parts of the cell and clamped. The temperature of the sample can be controlled by the thermostat or an electronic heater. Both sides of the sample in the permeation cell are evacuated to 103 Torr, and the sample is degassed in the permeation apparatus for several hours. A saturated water vapor of at 0  C (4.58 mmHg) is then introduced to the upstream side, and the pressure change on the downstream side due to the permeated water vapor is measured with a pressure meter. The water permeation coefficient (P) is evaluated from the increasing rate of the pressure (dp/dt) at a steady state and normalized with the thickness of the sample. When the vacuum line is not used, a test gas is supplied to the upstream side of the sample. The pressure difference between the upstream side and downstream side is determined, and the flow rate is also measured, by, for example, soap-bubble displacement (Klein and Giszpenc 1990). Typical water vapor permeation curves are shown in Fig. 5 (Tohge et al. 1996). This figure is for uncoated and 40CH3SiO3/2  60SiO2-coated polyimide (Kapton ®) films. The ordinate indicates the increase in pressure due to water vapor permeation through the films, and the abscissa indicates the time after the introduction of water vapor to the cell. Both curves increase almost linearly with time, after an introduction period which varies from film to film. The slope of the linear part of the curve

10

K. Tadanaga and T. Minami

Geissler

Diffusion Pump

Cell Cold Trap

Rotary Pump

Thermostat Water

Pressure Gauge

Ice / Water Fig. 4 An example of gas permeation measurement apparatus Fig. 5 Time dependence of pressure on the downstream side of the permeation cell

800 a uncoated b coated

a

Pressure / 10–3 Torr

600

400

dp dt

200

b 0

0

10

20 30 40 Time / min

50

60

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials Fig. 6 Illustration of the dish method

11

Sealing wax

sample

CaCl2

(dp/dt) gives a measure of the water vapor permeability of the film. As obvious from the figure, the slope of the coated film is smaller than that for the uncoated film by roughly one order of magnitude, indicating that coating with a silica-based film effectively suppresses the permeation of water vapor through polyimide (Kapton ®) films. As a practical method, the dish method is often used to measure the water vapor transmission rate of sheet materials (ISO 2528; JIS Z0208). In this technique, a dish is covered by the sample film, and the film is sealed by wax, as shown in Fig. 6. Inside of the dish, a desiccant agent such as calcium chloride is placed, to adsorb the permeated water vapor. The total weight of the cup is measured and the dish is placed in a temperature and humidity controlled chamber. An example of the chamber conditions is 40  C and 90 % relative humidity. By measuring the weight of the dish at constant intervals, the increment of the weight is regarded as the weight of permeated water vapor through the sample film, and, thus, the permeation rate can be calculated in terms of grams of permeated water vapor in 24 h (1 day), per square meter (g/(m2 day)). For more information on gas permeation measurements, please refer to a book on permeability of plastics (McKeen 2011).

Conclusions Gas adsorption and permeation in sol–gel-derived materials were described. Since the pore size can be controlled in the preparation of porous sol–gel materials, characterization of porosity and permeation of gas will be very important areas of research.

References Adamson AW. Adsorption of gases and vapors on solids. In physical chemistry of surfaces. 5th ed. New York: Wiley; 1990.

12

K. Tadanaga and T. Minami

Amberg-Schwab S, Hoffmann M, Bader H, Gessler M. Inorganic–organic polymers with barrier properties for water vapor, oxygen and flavors. J Sol–Gel Sci Technol. 1998;13:141–6. Amberg-Schwab S, Katschorek H, Weber U, Hoffmann M, Burger A. Barrier properties of inorganic–organic polymers: influence of starting compounds, curing conditions and storage – scaling-up to industrial application. J Sol–Gel Sci Technol. 2000;19:125–9. ASTM D1434. Standard test method for determining gas permeability characteristics of plastic film, and sheeting. ASTM book of standards volume: 15.10. Pennsylvania: ASTM International. ASTM D1993. Standard test method for precipitated silica-surface area by multipoint BET nitrogen adsorption. ASTM book of standards volume: 09.01. Pennsylvania: ASTM International. ASTM D3663. Standard test method for surface area of catalysts and catalyst carriers. ASTM book of standards volume: 05.06. Pennsylvania: ASTM International. ASTM D4222. Standard test method for determination of nitrogen adsorption and desorption isotherms of catalysts by static volumetric measurements. ASTM book of standards volume: 05.06. Pennsylvania: ASTM International. ASTM D4567. Standard test method for single-point determination of specific surface area of catalysts and catalyst carriers using nitrogen adsorption by continuous flow method. ASTM book of standards volume: 05.06. Pennsylvania: ASTM International. ASTM D5604. Standard test methods for precipitated silica-surface area by single point B.E.T. nitrogen adsorption. Book of standards volume: 09.01. Pennsylvania: ASTM International. Azuta K, Tadanaga K, Minami T. Water and oxygen permeability of silica thin films containing organic polymers coated on poly (ethylene terephtalate) by the sol–gel method. J Ceram Soc Jpn. 1999;107:293–6. Brinker CJ, Ward TL, Sehgal R, Raman NK, Hietala SL, Smith DM, Hua D-W, Headley TJ. “Ultramicroporous” silica-based supported inorganic membranes. J Membr Sci. 1993;77:165–79. Brinker CJ, Sehgal R, Hietala SL, Deshpande RD, Smith DM, Loy D, Ashley CS. Sol–gel strategies for controlled porosity inorganic materials. J Membr Sci. 1994;94:85–102. Brunauer S, Emmett PH, Teller E. Adsorption of gases in multi molecular layers. J Am Chem Soc. 1938;60:309. DIN 53380-1~5. Testing of plastics – determination of gas transmission rate – part 1 ~ 5. Deutsches Institut für Normung e.V. Fujimoto K, Suzuki J, Mori T, Watanabe M. Adsorption behavior of nitrogen monoxide on KxGaxSn8–xO16 hollandite. J Sol–Gel Sci Technol. 2000;19:377–81. Hernandez-Cortez JG, Lopez T, Manriquez ME, Gomez R, Navarrete J. Thermally induced phase transformation on 12-tungstophosphoric acid/ZrO2 sol–gel. J Sol–Gel Sci Technol. 2003;26:213–6. ISO 2528. Sheet materials – determination of water vapour transmission rate – gravimetric (dish) method. International Organization for Standardization. Iwashita K, Tadanaga K, Minami T. Water permeation of SiO2–CH3SiO3/2 thin films modified with trimethylsilyl groups on nylon-6 substrate. J Sol–Gel Sci Technol. 1997;10:301–7. Jang K, Kim H. The gas barrier coating of 3-aminopropyltriethoxysilane on polypropylene film. J Sol-Gel Sci Technol. 2007;41:19–24. JIS Z 0208. Testing methods for determination of the water vapor transmission rate of moistureproof packaging materials (dish method). Japanese Standards Association. Klein LC, Giszpenc N. Sol–gel processing for gas separation membranes. Ceram Bull. 1990;69:1821–30. Leterrier Y. Durability of nanosized oxygen-barrier coatings on polymers. Progress Mater Sci. 2003;48:1–55. Lin YS, Kumakiri I, Nair BN, Alsyouri H. Microporous inorganic membranes. Sep Purif Method. 2002;31:229–379. McKeen LW. Permeability properties of plastics and elastomers. 3rd ed. Oxford: William Andrew; 2011.

Measurements of Gas Adsorption and Permeability of Sol–Gel Materials

13

Meixner DL, Dyer PN. Influence of sol–gel synthesis parameters on the microstructure of particulate silica xerogels. J Sol–Gel Sci Technol. 1999;14:223–32. Okada K, Kaneda A, Kameshima Y, Yasumori A. Acidic and basic gas adsorption properties in composites of layered double hydroxide/aluminosilicate xerogels. Mater Res Bull. 2002;37:209–19. Okui T, Saito Y, Okubo T, Sadakata M. Gas permeation of porous organic/inorganic hybrid membranes. J Sol–Gel Sci Technol. 1995;5:124–34. Sermon PA, Wang Y, Vong MSW. Fractal analysis of sol–gel derived silica by adsorption. J Colloid Interface Sci. 1994;168:327–32. Suzuki M. Adsorption engineering. Tokyo: Kodansha; 1990. Tadanaga K, Iwashita K, Minami T, Tohge N. Coating and water permeation properties of SiO2 thin films prepared by the sol–gel method on nylon-6 substrates. J Sol–Gel Sci Technol. 1996;6:107–11. Tatlier M, Erdem-Senatalar A. Method to evaluate the fractal dimensions of solid adsorbents. J Phys Chem B. 1999;103:4360–5. Tohge N, Tadanaga K, Sakatani H, Minami T. Formation of SiO2-based coatings by the sol–gel method and their effects on water vapour permeability of polyimide films. J Mater Sci Lett. 1996;15:1517–9. Venkatraman A, Fan LT, Walawender WP. Fractal analysis of a sol–gel-derived silica by adsorption revisited. J Colloid Interface Sci. 1996;183:289–90. Yoshioka T, Nakanishi E, Tsuru T, Asaeda M. Experimental studies of gas permeation through microporous silica membranes. AIChE J. 2001;47:2052–63.

Viscosity and Spinnability of Gelling Solutions Sumio Sakka

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Brief Description of the Rheological Behavior of Liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elastic Deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscous Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flow Behaviors of Real Liquids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods for Measuring Viscosity of Sol–Gel Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Falling Sphere Viscometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pulling-Up-Sphere Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Capillary Viscometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rotating Cylinder Viscometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cone and Plate Viscometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of Viscosity Measurements to Sol–Gel Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Change of Viscosity with Time During the Sol-to-Gel Transition . . . . . . . . . . . . . . . . . . . . . . . . . . . Relation of Viscosity to Molecular Weight and Shape of the Particles in Sols . . . . . . . . . . . . . Dependence of the Viscosity on Shear Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rheological Behavior Based on Dynamic Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spinnability and Viscosity of Sols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spinnability and Formation of Fibers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relation Between Rheological Behavior and Spinnability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusive Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2 3 4 6 7 7 8 9 11 13 14 14 17 20 24 25 25 26 31 32

Abstract

Gelling solutions of a certain composition range in the alkoxysilane-water-alcohol systems catalyzed with an acid are spinnable at the viscosity of 10
100 poise and subject to fiber drawing. In order to understand such behavior of sol–gel solutions, rheological properties of solutions, especially, the viscosity S. Sakka (*) Sakka Laboratory, Kuzuha-Asahi, Osaka-fu, Japan e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_41-1

1

2

S. Sakka

change with time and shear rate, and methods of measurement of viscosity of solutions were reviewed. Measurements of viscosity applied to tetraethoxysilane solutions indicated that both spinnable and nonspinnable solutions increase in viscosity with time in a similar manner and that spinnable solutions are Newtonian in flow behavior in the viscosity range of 10
100 poise where fibers can be drawn, while nonspinnable solutions are characterized by non-Newtonian, thixotropic flow behavior. On the basis of the abovementioned results of viscosity measurements, together with the analysis of flow behavior and mechanism of hydrolysis-condensation of tetraethoxysilanes, the occurrence of spinnability was related to the linear shape of particles in the solution. Further, the same discussion was found to be valid for solutions designed for drawing of Al2O3, TiO2, ZrO2, and Y-Ba-Cu-O superconducting fibers.

Introduction In the typical sol–gel method, we start from a solution, which consists of metal compounds, such as metal alkoxides and acetylacetonates, as source of oxides, alcohols as solvent, water as hydrolysis agent, and an acid or base catalyst. Metal compounds undergo hydrolysis and polycondensation near room temperature, giving rise to a sol, in which polymers or fine particles of oxides are dispersed. Further reaction connects the particles, solidifying the sol into a wet gel, eventually. Vaporization of water and solvents produces a dry gel. As the reaction proceeds during sol to gel conversion, the viscosity of the solution gradually increases until the solution is solidified to a gel. For this reason, the measurements of viscous or viscoelastic behavior of the samples are expected to give important information on the degree of polymerization reaction in the solution and the sizes and shapes of the particles. In this chapter, a description will be made of the following: 1. The rheological behavior of solutions 2. The measurements of viscosity of the sol–gel solutions 3. The relation between the viscous behavior and spinnability

Brief Description of the Rheological Behavior of Liquids In this section, minimum information on the rheological behavior of materials will be described (Kanbara 1982; Onogi 1982). The general theories of rheology will not be introduced. A special emphasis is placed on the flow behavior of the liquids. The rheological behavior of materials is based on the following three simple and fundamental deformation and flow patterns:

Viscosity and Spinnability of Gelling Solutions

3

a

b

σ

γ

0

time

0

time

c σ

γ

0

Fig. 1 Stress σ and strain γ in an elastic body. (a) Change of σ with time, (b) change of γ with time, and (c) σ versus γ relation

1. Elastic deformation 2. Viscous flow 3. Plastic flow The rheological behavior of a real material is expressed by a combination of the above three behaviors.

Elastic Deformation When a stress is applied to an elastic body, the corresponding strain occurs instantaneously. Upon removal of the stress, the strain instantaneously disappears. These are demonstrated in Fig. 1a, b. As seen in Fig. 1c, the stress is proportional to the strain. When a solid is deformed by a shear stress σ, the following formula holds: σ γ¼G

(1)

Here γ is the shear strain and G is a constant called shear modulus. A perfect elasticity occurs only in an ideal solid.

4

S. Sakka

y

v A

F

x z

Fig. 2 Concept of viscous flow. Solution fills the space between two planes. One plane is fixed and the upper plane moves with a velocity v in the direction x

a

b

σ

γ

0

to off

t 0

c

to off

t

d

log η

σ

γ (dγ /dt)

log g

Fig. 3 Time change of the shear stress σ (a) and the shear strain γ (b), the flow curve (c), and η versus γ_ relationship for a Newtonian liquid (d)

Viscous Flow Viscous flow occurs when a viscous liquid is subjected to a shear stress. Shear strain caused by the viscous flow is not recovered, even if the stress is removed. The strain varies as a function of time. In viscous flow, viscosity acts as a resistance to the stress. Figure 2 shows the concept of viscous flow. One of the two planes with area A is fixed and the other plane (upper one) moves with a velocity v in the direction of the x-axis. There is no movement in directions y and z. F is the force required for that movement. For a uniform flow, the stress σ is equal to F/A and the gradient of

Viscosity and Spinnability of Gelling Solutions

a

5

b Newtonian

σ Shear thinning

log η

Newtonian Shear thinning

Shear thickening Shear thickening . γ

. log γ

Flow curves

Apparent viscosity

Fig. 4 Flow curves (a) and η versus γ_ relations (b) for non-Newtonian liquids

velocity dv/dy is equal to shear rate γ_ð¼ dγ=dtÞ. For a purely viscous liquid, the following formula holds: σ ¼ ηγ_

(2)

Here, η is a constant and is called viscosity. Liquids for which η is a constant, regardless of the values of σ and γ, _ are called Newtonian liquids and such a flow is called Newtonian flow. Figure 3 shows the time change of shear stress σ and shear strain γ, the flow curve, and η versus γ_ relationship for Newtonian liquids. It is seen that for Newtonian liquids, the σ versus γ_ relationship passes through the origin of the coordinate system and that the viscosity η does not change with the shear rate.

Non-Newtonian Flow As seen in Fig. 4a, some liquids show flow curves whose behavior is called shear thinning and shear thickening. In these flows, the viscosity of the liquid as expressed by Eq. 2 is not constant. Therefore, these flow behaviors are called non-Newtonian and the liquids are called non-Newtonian liquids. The viscosity, defined as the ratio of shear stress to shear rate, is called the apparent viscosity, ηα: ηα ¼

σ γ_

(3)

_ as shown in Fig. 4b. ηα changes with γ, The plastic flow is defined as the flow which takes place in the region of stress higher than the yield stress σ y. Such a flow is also called Bingham flow. The curve

6 Fig. 5 Flow curves for plastic and pseudo-plastic liquids

S. Sakka pseudo-plastic σ

plastic

σy

. g

does not pass through the origin of the coordinate system, as shown in Fig. 5. Plastic flow is expressed by the following equation: σ  σ y ¼ ηp γ_

(4)

where ηp is a constant called plastic viscosity.

Flow Behaviors of Real Liquids Flow behavior other than viscous flow (Newtonian flow) is called non-Newtonian in the broad sense. Many real liquids exhibit flow behavior possessing both viscous and elastic nature, that is, viscoelastic flow or deformation. Concerning viscoelasticity of materials, refer to reference (Doi 1993). Here, a general outline of the flow behavior of liquids obtained from experience by many researchers is given. 1. A material tends to show elastic behavior at low temperatures or for a short time stress application. 2. A material tends to show viscous behavior at high temperatures or for a long duration of applied stress. 3. A solution containing low molecular weight polymers or fine particles shows viscous behavior when the concentration is low. 4. Solutions and melts consisting of high molecular weight polymers or large particles show viscoelastic properties.

Viscosity and Spinnability of Gelling Solutions Fig. 6 Falling sphere viscometer. The cylinder is immersed in a thermostat to control the temperature of the liquid

7

Guiding tube Falling sphere Liquid sample Upper line

Lower line Glass container

Methods for Measuring Viscosity of Sol–Gel Solution The viscous and viscoelastic properties of sol–gel solutions are characterized by measuring viscosity under different conditions. In this section, different methods for measuring viscosity are described (Kanbara 1982). The following five kinds of viscometers can be employed: 1. 2. 3. 4. 5.

Falling sphere viscometer Pulling-up-sphere viscometer Capillary viscometer Rotating cylinder viscometer Cone and plate viscometer

Falling Sphere Viscometer The falling sphere viscometer is employed to estimate the viscosity of a liquid by measuring the stationary falling velocity of a sphere in the liquid. The falling velocity is determined by measuring the time required for the sphere to pass between the upper and lower lines in the container shown in Fig. 6. When a sphere of diameter d is falling down at a velocity v in the liquid, the sphere undergoes a resistance fu:

8

S. Sakka

f u ¼ 3πηvd

(5)

The downward force acting on the falling sphere, fd, is obtained by subtracting the floating force from the weight of the sphere: 1 1 f d ¼ πd 3 ρo g  πd3 ρg 6 6

(6)

ρo, density of the sphere; ρ, density of the liquid; g, acceleration of gravity. For stationary falling, fu = fd. Then, the viscosity of the liquid is expressed by the formula: η¼

d2 ðρ0  ρÞg 18v

(7)

Usually, a steel ball from ball bearing is employed as the sphere. For the validity of Eq. 7, the Reynolds number: Rd ¼

ρvd η

(8)

should be small. In other words, it is important to use a small sphere of low density when the viscosity of the liquid is low. Generally, spheres with diameters less than several millimeters are used in the measurements. Generally speaking, this method is suitable for higher viscosities than, for example, 0.1 Pa  s. It should be noted that this method cannot be applied to the measurement of the viscosity of non-Newtonian liquids. For a precise measurement, the positions of the upper and lower reading lines have to be adjusted. In order for the flow to become stationary, several centimeters are necessary. The effect of the bottom is very small, if the lower reading line is located several centimeters above the bottom. The falling sphere method was applied to study the progress of sol–gel transition by measuring the viscosity (Mizuno et al. 1985).

Pulling-Up-Sphere Method The pulling-up-sphere method has been frequently employed to measure the viscosity of glass melts, instead of the falling sphere method. This is because many repeated runs are possible in the pulling-up-sphere method, when the viscosity of the melt is measured as a function of temperature. In this method, a sphere is connected to a balance, as shown in Fig. 7. In the measurement, the sphere is kept still by placing a weight W0 on the balance dish. Then, the weight is increased to W, in order to pull up the sphere. The sphere moves upward and the velocity becomes constant in a short time. We measure the time t required for the sphere to move a distance Δx. By equating the force applied to the

Viscosity and Spinnability of Gelling Solutions

9

Fig. 7 Pulling-up-sphere method

Connected to balance

Container

Sphere Liquid sample

sphere to the resistance obtained from Stokes’ rule on the fluid resistance, the formula for the viscosity η is derived: η ¼ K ðW  W 0 Þ

1 ðΔx=ΔtÞ

(9)

Here, K is a constant depending on the apparatus employed, Δx/Δt is the velocity of the sphere, and (W – W0) is the upward force. K is obtained by carrying out the measurement of Δx/Δt with a liquid of known viscosity. The method seems to lack precision due to the use of a hook and wire for pulling up the sphere and possible nonstationary movement of the sphere. Takahashi et al. made an extensive study of this problem and succeeded in markedly increasing the precision of the method (Takahashi and Tanioka 1966). This method makes it possible to measure the viscosities in a broad range between 102 and 10+2 Pa  s. The spheres can be made of platinum, steel, carbon, etc. This method was applied to sol–gel solutions by Sakka and Kamiya (1982), Sakka and Kozuka (1988).

Capillary Viscometer The capillary viscometer is an apparatus which is employed to estimate the viscosity of a liquid by measuring the time required for a certain amount of the liquid to pass

10

S. Sakka

Fig. 8 Ostwald viscometer

h

Sphere (Volume V)

Capillary (length l)

through the capillary. A typical capillary viscometer is the Ostwald viscometer. In the Ostwald viscometer, the force which makes the liquid flow through the capillary tube is the weight of the liquid. The Ostwald viscometer is illustrated in Fig. 8. In order to estimate the viscosity with the Ostwald viscometer, the time required for liquid of volume V to flow through the capillary is measured. The viscosity of Newtonian liquids which flow through a capillary of uniform diameter in the Ostwald viscometer is expressed by the following formula: η¼

πr 4 ρght 8 IV

(10)

r, radius of the capillary; l, length of the capillary; g, acceleration of gravity; h, average effective height of liquid column; t, time required for the flow of the liquid; V, volume of the sphere containing liquid. When the values depending only on the apparatus are combined and expressed by c, formula (10) becomes: η ¼ cpt

(11)

c, constant depending on the parameters related to the viscometer; ρ, density of the liquid sample; t, time required for a volume V of liquid to flow through the capillary tube.

Viscosity and Spinnability of Gelling Solutions

11

Fig. 9 Rotating cylinder viscometer

Reading device Torsion wire

Inner cylinder

Outer cylinder

Liquid sample

Once the value c is determined by measuring the time t for a liquid of known viscosity and density, the viscosity η of the sample can be obtained by measuring the time t for the liquid sample. The Ostwald viscometer cannot be employed for high viscosity liquids, because the force available to make the liquid flow through the capillary is small since it is due only to the weight of the liquid itself. Actually, we have used the Ostwald viscometer only for sols of viscosities lower than 0.05 Pa  s Sakka and Kamiya (1982), Sakka and Kozuka (1988). For higher viscosities, the pulling-up-sphere method was employed. It is also noted that the sample to be measured should be a Newtonian liquid.

Rotating Cylinder Viscometer The rotating cylinder viscometer is the most popular type of viscometer, because it gives accurate and precise data. This viscometer can measure non-Newtonian as well as Newtonian flow characteristics. The rotating cylinder viscometer consists of two concentric cylinders, outer and inner, as shown in Fig. 9. The inner cylinder is hung by a torsion wire or spring. The outer cylinder can be rotated at a desired angular velocity. In the measurements, the liquid sample is placed between the outer cylinder and inner cylinder. When the outer cylinder is rotated, the inner cylinder is subjected to a torque, with the wire being distorted by a certain angle. The angle of distortion is read by a pointer or

12

S. Sakka

Fig. 10 Cone and plate viscometer

Cone

Plate

recorded by electrical or optical means. The torque T is estimated as the product of the torsion angle and the torsion constant of the wire. For Newtonian liquids, the viscosity η is expressed by the following Margules formula: η¼

  T 1 1  4πLΩ R21 R22

(12)

Here, T, torque acting on the inner cylinder; L, effective length of the inner cylinder (dipped length); Ω, angular velocity of the outer cylinder; R1 and R2, diameters of the inner and outer cylinders, respectively. The viscosity can be obtained from formula (12) by measuring T as a function of Ω. The measurement by the rotating cylinder viscometer provides the flow curve, σ versus γ. _ The shear stress, σ, acting on the cylindrical layer located at a distance r from the central axis is expressed by the formula: σ¼

T 2πr 2 L

(13)

The shear rate (= velocity gradient), γ, _ is expressed by: γ_ ¼

dv 2Ω R21 R22 ¼ 2 2 dr r R2  R21

(14)

This method can be applied to the analysis of the viscoelastic behavior of non-Newtonian liquids.

Viscosity and Spinnability of Gelling Solutions

13

Cone and Plate Viscometer Apparatus and Use of the Method. The cone and plate viscometer consists of a cone with a very large aperture and a plate, as shown in Fig. 10. They are arranged so that the apex may be touched to the center of the plate. The angle α between the generating line of the cone and the plate is very small, less than 3–4 . That is, the surface of the cone is close to the plate surface. The liquid sample is filled into the space between the cone and plate. The measurement is carried out by rotating the cone with a fixed plate or rotating the plate with a fixed cone. If either the cone or the plate is vibrated, the dynamic measurement is possible. This method proved very useful for characterizing the flow behavior of sols for sol–gel materials. Some of the reasons for that are the following: 1. 2. 3. 4. 5. 6.

A small amount of sample is sufficient for the measurement. Exchange of samples is simple. Measurement of the viscosity in the stationary state is possible. A broad range of viscosities can be measured by changing the angle α. Dynamic measurements are also possible. Measurement of viscoelastic flow as well as viscous flow is possible.

Measurement for Newtonian Liquid. The liquid is filled into the gap between the cone and plate and the plate is rotated at a given rotational speed n. The angular velocity, Ω, is: Ω ¼ 2πn

(15)

A torque, T, is applied to the cone, which is calculated as the product of the torsion rigidity of the wire supporting the cone and the equilibrium torsion angle. The linear velocity at a point on the plate at a distance r from its center of the plane is rΩ and the thickness of the sample at that point is r tan α  rα accordingly, the shear rate γ_ is: γ_ ¼

rΩ Ω ¼ rα α

(16)

It is seen that the shear rate γ_ is constant at any position in the sample, that is, independent from r. The torque, T, on the cone is equal to: T¼ shear stress, σ, is:

ηΩ 2π R3 α 3

(17)

14

S. Sakka

σ ¼ ηγ ¼ η

Ω 3T ¼ α 2π R3

(18)

Since η ¼ ðσ=γÞ _, η¼

3αT 2π R3 Ω

(19)

The viscosity η can be calculated by Eq. 19 from measured torque T and angular velocity Ω. Measurement for a Non-Newtonian Liquid. It is known that when the angle α is very small, Eqs. 16 and 18 can be applied to non-Newtonian liquids. As seen from Eq. 16, very large shear rates can be employed in the measurements.

Application of Viscosity Measurements to Sol–Gel Processing In this section, examples of the change of viscosity during the sol to gel transition are given.

Change of Viscosity with Time During the Sol-to-Gel Transition The gelation process can be followed by measuring the viscosity of a sol as a function of time. The effect of various factors on the gelation of the sol will also be discussed. Figure 11 shows the change of the viscosity of a solution containing Si(OC2H5)4, H2O, C2H5OH, and HCl with molar ratio [H2O]/[Si(OC2H5)4] = 2 and [HCl]/[Si(OC2H5)4] = 0.01 as a function of time (Sakka 1984). The solution was kept at 25  C, 35  C, and 80  C. The viscosity was measured with an Ostwald viscometer for the viscosities lower than about 0.05 Pa  s and by the pulling-upsphere method for viscosities higher than about 0.05 Pa  s. For the three temperatures, the viscosity increases very slowly in the early stage of hydrolysis and condensation of tetraethoxysilane. After a certain time, an abrupt increase in viscosity is seen with time. It is also seen that the increase in viscosity and the gelation take place more quickly at higher temperatures. The progress of the sol to gel conversion for three solutions containing Si(OCH3)4, CH3OH, H2O, and NH4OH with different Si(OCH3)4 contents was monitored by measuring the viscosity as a function of time (Debsikdar 1986). A rotating cylinder viscometer was employed. For all the three solutions, the viscosity increased very slowly in the early stages of reaction and increased rapidly after a certain time. This time point is reached in a shorter time and the rate of increase after this point was higher for higher Si(OCH3)4 contents. The progress of hydrolysis-condensation reaction in solutions containing Si(OCH3)4, CH3OH, H2O, and HCl with different water contents was followed by

Viscosity and Spinnability of Gelling Solutions

15

1000 100 80°C

Viscosity / Pa.s

10

30°C

25°C

1 0.1 0.01 0.001 0.001 10

100

1000 Time / min

10000

100000

Fig. 11 Change of the viscosity of a solution containing Si(OC2 H5)4, H2O, C2H5OH, and HCl with molar ratio [H2O]/[Si(OC2H5)4] = 2 and [HCl]/[Si(OC2H5)4] = 0.01, as a function of time (Sakka 1984)

the viscosity (Mizuno et al. 1985). The viscosity was measured with a falling sphere viscometer. The time required for gelation and viscosity of the solution were compared. For acid catalyzed solutions, the gelation time was shortest; the viscosity was highest for the solution with the intermediate water content (2–4 mol per mole of Si(OCH3)4). This may be due to the highest hydrolysis-condensation rate for that solution. For lower water contents, the gelation time was longer. This is attributed to a lower hydrolysis rate caused by lack of water. For higher water contents, the gelation time is longer, probably because the polymerization rate is low, due to lower silicon alkoxide concentration. The viscosity of three solutions containing Si(OC2H5)4, H2O, C2H5OH, and HCl was measured at 25  C as a function of time of reaction at 80  C. The results are shown in Figure 12 (Sakka et al. 1982). The three solutions are characterized by different [H2O]/[Si(OC2H5)4] ratio, r. The viscosity was measured by Ostwald viscometer for lower viscosities than 0.005 Pa  s and by pulling-up-sphere method for higher viscosities. Different behaviors are found for the three solutions. The solution with r = 2.0 increases in viscosity rapidly and gels in about 3 h. The solution with r = 1.0 increases in viscosity very slowly and levels off at about 0.01 Pa  s. Viscous state and gelling are not observed for this solution even if it is kept at 80  C for a long time. This is probably due to the lack of water for hydrolysis. It is also possible that water evaporates out before it is consumed for the hydrolysis, and polymerization of the alkoxide is stopped. The viscosity of the solution with r = 1.7 increases up to several Pa  s, then its increase becomes slower above this value. This may be due to insufficient water. The slower increase in viscosity in the later stage may be caused by absorption of water from the ambient atmosphere.

16

S. Sakka 10000 Reacted at 80°C 1000

Measured at 25°C

Viscosity / Pa.s

100 1.7

r = 2.0

10 1 0.1

1.0 0.01 0.001 0

100

200 300 Time / min

400

500

Fig. 12 Change of the viscosity measured at 25  C for three solutions containing Si(OC2H5)4, H2O, C2H5OH, and HCl with time of reaction at 80  C. The three solutions are characterized by different [H2O]/[Si(OC2H5)4] ratio of 2.0, 1.7, and 1.0 (From Sakka et al. (1982), with permission of Japan Ceramic Society)

I

Viscosity 25°C

II

3 III

IV

2 log η , poise

Fig. 13 Change of the viscosity and gelling point for solutions containing Si(OC2H5)4, H2O, and C2H5OH. HCl (I, II, and III) or NH4OH (IV and V) is used as catalyst (From Sakka and Kamiya (1982), with permission of Japan Ceramic Society)

V

1

0 −1 −2

0

200

400

600

Time, hr

Figure 13 compares the change of viscosity and the gelling points for solutions containing HCl or NH4OH as catalyst. All the viscosity versus time curves appear similar in shape. However, the difference can be seen in the viscosity at the gelling

Viscosity and Spinnability of Gelling Solutions

17

point. This is lower for NH4OH catalyzed solutions (solutions IV and V) or the high water content solution (solution III) than the HCl catalyzed and low water content solutions (solutions I and II). This is attributed to the difference in the sol structure. It is assumed that sols which gel at lower viscosity have granular particles and become viscoelastic at lower viscosities. The viscosity–time relationship for a solution containing Hf(OC8H17)4, CH3COOH, and C2H5OH with 9 wt% HfO2 and reacted at 65  C was measured with a rotating cylinder viscometer by Larbot et al. (1992). The results show that the viscosity remains low until t = 0.9tg, where tg is the time required for gelation. Then, it rapidly increases after t = 0.9tg. The molecular weight of the particles in the sol obtained by measurement of light scattering was 23,300 at t = 0.90tg and 117,000 at t = 0.95tg. The authors explained that particles rapidly grow at a time near the gelation time as a result of polycondensation of hydrolyzed products.

Relation of Viscosity to Molecular Weight and Shape of the Particles in Sols The viscosity plotted as a function of time provides qualitative information on the growth of particles and formation of cross-linking between the particles. In order to relate the shape of particles to the flow behavior of the sol, determination of the reduced viscosity, ηsp/C, and the intrinsic viscosity, [η], is needed (Kanbara 1982; Onogi 1982; Okamura et al. 1981). The specific viscosity, ηsp is defined as: ηsp ¼

ðη  η0 Þ η0

(20)

where η is the viscosity of the solution and η0 is that of the solvent. The reduced viscosity is calculated by dividing ηsp by the concentration of particles, C, of the sol: ηsp =C ¼

η  η0 1 η0 C

(21)

The reduced viscosity represents the increase in viscosity assigned to a single particle. The intrinsic viscosity [η] can be obtained by extrapolating the ηsp/C versus C curve to C = 0: η ¼ lim ηsp =C C!0

(22)

The intrinsic viscosity represents the increase in viscosity attributed to one particle in the solution of very low concentrations. In the following, we show some examples of viscosity measurements related to the structure of sols. Measurement of molecular weights of particles and viscosity of

18

S. Sakka

sols were carried out for solutions containing Si(OC2H5)4, H2O, C2H5OH, and HCl with r = [H2O]/[Si(OC2H5)4] of 1.0, 2.0, 5.0, and 20.0 (Kamiya et al. 1984). In order to avoid further reaction during the measurement, the samples were subjected to trimethylsilylation with siloxane polymers. The number average molecular weight was determined by the freezing point depression method. The viscosity was measured on benzene solutions containing different concentrations of siloxane polymers. Figure 14 shows the number-average molecular weight, Mn, as a function of the reduced time, t/tg. Mn increases with the reaction time. Figure 15 shows plots of the reduced viscosity, ηsp/C, against the concentration, C, for solutions with r = 1.0 at reaction times, t/tg, of 0.015, 0.71, 0.83, 0.91, and 0.96. The reduced viscosity, ηsp/C, of a solution containing spherical particles may be expressed by the following equation (Einstein 1906): ηsp 0:025 ¼ ρ C

(23)

Molecular weight , Mn

r = 20 5

104

2

1 103

0

0.5

1.0

Relative time, t/tg Fig. 14 Change of number average molecular weight, Mn, of trimethyl-silylated siloxane polymer as a function of the relative time, t/tg, for solutions containing Si(OC2H5)4, H2O, C2H5OH, and HCl with different molar ratio r = [H2O]/[Si(OC2H5)4] (From Kamiya et al. (1984), with permission of Japan Ceramic Society)

Viscosity and Spinnability of Gelling Solutions

t/tg = 0.96

0.04 Reduced viscosity ηsp/C (100 ml/g)

Fig. 15 Plots of the reduced viscosity, ηsp/C, against the concentration, C, for a solution containing Si(OC2H5)4, H2O, C2H5OH, and HCl with molar ratio [H2O]/[Si(OC2H5)4] of 1.0 at the reaction time t/tg of 0.015, 0.71, 0.83, 0.91, and 0.96 (From Kamiya et al. (1984), with permission of Japan Ceramic Society)

19

0.03 0.91 0.02

0.83 0.71 0.015

0.01

5

0

10

Concentration (g/100 ml)

where C is the concentration of particles and ρ is the density of the particles. This equation indicates that ηsp/C is independent of the concentration and size of spherical particles. In Fig. 15, a straight line parallel to the x-axis is obtained for solutions of t/tg = 0.015 and 0.71. This indicates that the solutions contain spherical particles in the early stage of reaction. It is known that the reduced viscosity of a solution can be expressed by Huggins’ equation (Huggins 1942), if particles in the solution are chain-like or linear: ηsp ¼ ½η þ k½η2 C C

(24)

where k is a proportionality constant. Solutions of t/tg = 0.83, 0.91, and 0.96 show definitely a gradient, that is, viscosity–concentration plots follow Eq. 17, indicating that these solutions contain particles with chain-like shape. Figure 16 shows plots of the intrinsic viscosity [η] versus the reaction time, t/tg, for four water contents, r. [η] increases with increasing reaction time. Figure 17 shows plots of log[η] versus log Mn for the four solutions. It is known that the Mark–Houwink–Sakurada relation holds for [η] and Mn (Badgley and Mark 1949; Tsuchida 1975). ½η ¼ K Mnα

(25)

K is a constant depending upon the type of polymers, solvent, and temperature. The slope, α, of the log[η] versus log Mn plot depends on the type of polymer: α = 0 for spheric particles, α = 0.5–1.0 for flexible linear polymers, and α = 1.0–2.0 for rigid rod-like polymers. As indicated in Fig. 17, α is 0.75 for the solution with r = 1 and 0.64 for the solution with r = 2, meaning that these two solutions contain flexible linear siloxane polymers. On the other hand, α ranges from 0.2 to 0.5 for the other

20

1.0

Intrinsic viscosity, [η]

Fig. 16 Change of the intrinsic viscosity [η] with the relative reaction time, t/tg, for solutions containing Si(OC2H5)4, H2O, C2H5OH, and HCl with four different water content ratio r = [H2O]/ [Si(OC2H5)4] (From Kamiya et al. (1984), with permission of Japan Ceramic Society)

S. Sakka

r=20.0

0.05

5.0

2.0 1.0 0

0

0.5

1.0

t / tg

two solutions with r = 5 and 20. This means these two solutions contain spherically grown particles. Actually, however, those conclusions have to be checked by other techniques, in addition to viscosity measurements.

Dependence of the Viscosity on Shear Rate Usually, sols behave as Newtonian liquids, when the viscosity is low due to low concentration of particles, small particle size, and/or separated particles. On the other hand, when the viscosity becomes high due to growth or connection of the particles, the sols behave as non-Newtonian liquids, exhibiting viscoelastic properties, such as shear thinning or shear thickening. These behaviors can be found by measuring the viscosity as a function of shear rate, using the rotating viscometer or the cone and plate viscometer. In this case, the viscosity is the apparent viscosity, previously defined by Eq. 3. However, the word viscosity and the symbol η are used for simplicity. Examples of non-Newtonian flow will now be given for solutions used in sol–gel preparation of various oxides. A solution of composition Si(OC2H5)4:H2O:C2H5OH:HCl 1:2:1:0.01 in mole ratio was prepared (Kozuka et al. 1988). A part of the solution was reacted for gelation in the open state (solution OP: reacted in a beaker with a perforated cover) and the other part was gelled in a closed state (solution CL: reacted in a sealed beaker). After reaction for various time periods, the viscosity was measured by a cone and plate viscometer as a function of shear rate. Figures 18 and 19 show the viscosity versus shear-rate relationships, respectively, for the open state and closed

Viscosity and Spinnability of Gelling Solutions Fig. 17 Plots of log[η] versus log Mn for four solutions containing Si(OC2H5)4, H2O, C2H5OH, and HCl with different water contents (From Kamiya et al. (1984), with permission of Japan Ceramic Society)

−1.5

21

r = 1.0 a = 0.75

−2.0 2.0 −1.5

log [η]

a = 0.64

−2.0

5.0 −1.5 a = 0.5∼0.2 −2.0 −1.5

20.0

a = 0.34 −2.0 log Mn

state, respectively (Kozuka et al. 1988). For every measurement, shear rate was first increased and then decreased. It is seen in Fig. 18 that the viscosity of solution OP does not change with shear rate, remaining Newtonian up to a highly viscous state (η > 10 Pa  s). In contrast, solution CL (Fig. 19) shows shear thinning, even when the viscosity is lower than 1 Pa  s. Casson derived an equation relating the shear stress to the shear rate, called D, for pigment-oil suspensions: σ 1=2 ¼ k0 þ k1 D1=2

(26)

where k0 and k1 are constant values (Casson 1959). k20 is the shear stress F for a shear rate 0 and is called yield value of the suspension. A plot of σ 1/2 against D1/2 should be a straight line with an intercept σ 1/2 = k0 on the σ 1/2 axis. Therefore, the Casson plot gives information as to whether the solution has a yield value or not. Figures 20 and 21 show Casson plots for solutions OP and CL, respectively. All the plots for solution OP shown in Fig. 20 are straight lines, passing through the origin, because

22 Fig. 18 Viscosity versus shear-rate relation for a silicon alkoxide solution reacted in open state (OP) (From Kozuka et al. (1988), with permission of Elsevier)

S. Sakka

a

100 OP-1 24.0 h 10

Viscosity / Pa • s

23.5 h 1 22.0 h 0.1 19.0 h

12.1 h

0.01

5.0 h 0.001

Fig. 19 Viscosity versus shear-rate relation for a silicon alkoxide solution reacted in closed state (CL) (From Kozuka et al. (1988), with permission of Elsevier)

1

10 100 Shear rate / s-1

1000

100 CL-1 10

Viscosity / Pa • s

191 h 1 190 h 188 h 184 h 179 h

0.1

156 h 0.01 50 h

0.001

1

10

100

Shear rate / s-1

1000

Viscosity and Spinnability of Gelling Solutions

20 (Shear stress)1/2 / dyne1/2 cm−1

Fig. 20 Casson plot for the silicon alkoxide solution reacted in open state (OP) (From Kozuka et al. (1988), with permission of Elsevier)

23

23.5 h

OP-1

15

22.0 h

10

19.0 h

12.1 h

5

5.0 h 0 0

Fig. 21 Casson plot for the silicon alkoxide solution reacted in closed state (CL) (From Kozuka et al. (1988), with permission of Elsevier)

5 10 (Shear rate)1/2 / sec−1/2

15

CL-1 20

(Shear stress)1/2 / dyne1/2 cm−1

184 h

15

179 h 10

156 h

5

5.0 h 0

0

5

10

15

(Shear rate)1/2 / sec−1/2

the solution is of Newtonian nature up to the highly viscous state. The Casson plot for solution CL (Fig. 21) reacted for 184 h has an intercept on the vertical axis at about 2.5, indicating that the sol has a yield value and the solution CL is a non-Newtonian liquid in the highly viscous state.

24

S. Sakka

Sacks and Sheu (1987) measured the viscosity of solutions containing Si(OC2H5)4, C2H5OH, H2O, and HNO3 (or NH4OH), aged for various time periods, as a function of shear rate. The steady flow viscosity was determined by a rotating cylinder viscometer. When the aging was short, the viscosity did not change with the shear rate, that is, the solution behaved as a Newtonian liquid. After long-time aging, the solution became viscous and showed shear thinning. That is, the viscosity lowered with increasing shear rate. Upon further aging, the viscosity became higher and the solution showed thixotropy (Okamura et al. 1981). Rabinovich and Kopylov (1988) showed that particulate silica sols exhibited strong shear thinning. Brenna et al. (1991) showed that a typical sol for preparing SiO2–ZrO2 materials was Newtonian at lower viscosities and showed shear thinning behavior at higher viscosities. Both Newtonian and non-Newtonian behaviors were observed for alumina sols. Maki and Sakks determined viscosity versus shear-rate relationships for two kinds of alumina sols reacted for various time periods (Maki and Sakks 1988). The sol containing ribbon-like particles was Newtonian up to high viscosities of about 40 Pa  s; the viscosity did not change with the shear rate. On the other hand, the sol containing granular particles showed a shear thinning behavior at viscosities higher than 0.5 Pa  s Keysar et al. (1999) prepared alumina sols showing shear thinning behavior. Wolf and R€ ussel made rheological measurements of sols of the system consisting of Zr(O–nC3H7)4, CH3COOH, and H2O for preparing zirconia gels (Wolf and R€ ussel 1992). It was found that sols prepared by different methods exhibit different rheological behavior, such as Newtonian, shear thinning, or thixotropic flow.

Rheological Behavior Based on Dynamic Measurements Dynamic rheological measurements will be briefly described. In dynamic measurements, an oscillatory microscopic strain is given and the corresponding stress is measured. This method gives information on viscoelastic behavior even in the region where no viscous flow takes place. The measurements can be carried out by using the rotating cylinder viscometer or the cone and plate viscometer. In the measurements, an oscillatory microscopic shear strain of sine form is imposed on the liquid (Kanbara 1982; Doi 1993; Guizard et al. 1992). The sinusoidal strain is expressed as: γ ¼ γ 0 expðiωtÞ

(27)

Here, γ 0 is the maximum amplitude of the strain and ω is the angular velocity. Then, the frequency is v = ω/2π. Corresponding to the strain, the following sinusoidal stress σ is caused by a phase difference angle δ. σ ¼ σ 0 exp½iðωτ þ δÞ

(28)

Viscosity and Spinnability of Gelling Solutions

25

Here, σ 0 is the maximum amplitude of the stress. σ is related to γ as follows: σ ¼Gγ

(29)

G* is called complex shear modulus and written as: G ¼ G0 þ iG00

(30)

G0 is called storage modulus or elastic modulus, representing the elastic nature of the material. G00 is called loss modulus or viscous modulus, representing the viscoelastic nature of the material. A useful parameter for sol–gel materials is the loss tangent tan δ. tan δ ¼

G00 G0

(31)

The loss tangent is a measure of the ratio of the energy lost to the energy stored. Sacks and Sheu (1987) and Drabarek et al. (2000) made dynamic measurements of the viscoelastic behavior of silica sols as a function of reaction time. Guizard et al. (1992) made measurements on a titania gel. In all cases, both G0 and G00 values of the sol increase as the reaction time passes, that is, as the sol-to-gel conversion progresses, although the shapes of the curves are different from each other. More marked results can be seen in the change of the loss tangent, tan δ. In all cases, a maximum is found in the tan δ versus time curves. Generally, the time of the maximum in these curves is regarded as the gel point.

Spinnability and Viscosity of Sols In this section, the relation between the spinnability and rheological behavior of sols will be discussed.

Spinnability and Formation of Fibers So far, several kinds of inorganic fibers prepared by sol–gel method have been commercialized. Examples include fibers in the Al2O3–SiO2–B2O3 system (Sowman 1988) and heat-resistant silica fibers (Taneda et al. 1988). These fibers are drawn directly from viscous sols, in contrast to optical fibers which are drawn from reheated silica glass preform rods. Optical fibers are not discussed in this section. For possible fiber formation through drawing from a sol, this has to satisfy the following two conditions: 1. The sol should be viscous and spinnable so that it might be elongated into fiber form. A sol with no spinnability cannot be elongated, even if it is viscous.

26

S. Sakka

2. Gelling (solidification) of a drawn fiber should quickly occur in ambient atmosphere so that the fiber might be able to keep a certain diameter. Although both the above conditions are required to form fibers, we will mainly pay attention to the spinnability, discussing the relation of rheological behavior of the sol to the occurrence of spinnability (Sakka and Yoko 1992, Sakka 1984, Sakka and Kamiya 1984).

Relation Between Rheological Behavior and Spinnability The spinnability can be confirmed by immersing the tip of a glass rod and drawing up the rod. If a long, continuous fiber can be drawn, the spinnability is “Good” and judgment is “Yes.” If no fiber can be drawn, we judge “No.” “Poor” means only short fibers can be drawn. In the previous section “Change of Viscosity with Time During the Sol-to-Gel Transition” we showed a figure (Fig. 13) exhibiting the viscosity versus time relationships for five solutions of different compositions (Sakka et al. 1982). Compositions I and II are characterized by HCl catalyst and a low water content, composition III by HCl catalyst and high water content, and compositions IV and V by an ammonia catalyst and low water content. It was shown that compositions I and II showed spinnability, while compositions III, IV, and V did not. We speculated that sols consisting of linear particles become spinnable at high viscosities, but sols consisting of granular particles do not show spinnability, even at high viscosities. Plots of ηsp/C versus SiO2 concentration, shown in Fig. 22 (Sakka and Kamiya 1982), support the above speculation. In Fig. 22a for composition I, the straight lines for the sols with t/tg larger than 0.48 have slopes. As stated in 3.2.2, this indicates that the solution contains chain-like linear particles. Since solution I becomes spinnable, it can be said that the sol is spinnable, when it contains linear particles. In contrast, the straight lines for solution V shown in Fig. 22b are parallel to C axis up to aging time t/tg = 0.75, indicating that nonspinnable sols contain granular particles. Toyoda et al. (1997) made a similar observation with sols for preparing PbTiO3 fibers by sol–gel method. The reduced viscosity, ηsp/C, of sols containing the Pb–Ti double alkoxide, PbTi(OC2H4OCH3)6, was plotted against concentration C for sols aged for various time periods. For the spinnable sol, the ηsp/C versus C plots were straight lines with a certain slope at a later stage of aging, indicating that the sol contained linear particles. On the other hand, for nonspinnable sols, the plots were parallel to the C axis, indicating that the sols contained granular particles. In “pulling-up-sphere method,” we also showed that the value of the exponent α of the Mark–Houwink–Sakurada relation (Eq. 25) is indicative of the shape of polymers in the solution. The α value estimated from log[η] versus log Mn plots in Fig. 17 (Kamiya et al. 1984) is 0.75 for a spinnable solution of water to silicon alkoxide mole ratio r = 1, 0.64 for a spinnable solution of r = 2, 0.2–0.5 for a solution of r = 5 with poor spinnability, and 0.34 for a solution of r = 20 with no

Viscosity and Spinnability of Gelling Solutions

27

Fig. 22 Change of the reduced viscosity with concentration C, for solutions containing Si(OC2 H5)4, H2O, C2H5OH, and HCl with HCl catalyst (a) and ammonia catalyst (b), at different t/tg (From Sakka and Kamiya (1982) with permission of Elsevier)

spinnability. This indicates that spinnable sols contain linear flexible particles, while sols containing granular particles show no spinnability. In the previous section, “Pulling-up-sphere Method,” the shear-rate dependence of the viscosity of silicon alkoxide solutions aged in the open and the closed state (Kozuka et al. 1988) was shown in Figs. 18 and 19. As seen from the figures, the former solution shows the Newtonian flow behavior up to high viscosities and is spinnable, while the latter shows shear thinning and thixotropic flow behavior, exhibiting no spinnability. These observations led to the idea that good spinnability appears in the Newtonian solutions, whereas non-Newtonian, shear thinning solutions do not show spinnability. In order to generalize the above idea, the viscosity versus shear-rate relation was measured for silica sols derived from alkoxide solutions of composition Si(OC2H5)4: H2O:acid:C2H5OH = 1:2:0.01:5 (molar ratio) reacted in the closed state (Sakka and Kozuka 1987). The acids used were CH3COOH, HCOOH, HNO3, (H2SO4)0.5, and HCl. The result with the sol containing CH3COOH as catalyst is shown in Fig. 23. The results with other solutions are similar. Figure 23 shows that the flow property changes from Newtonian to shear thinning and then to thixotropic with reaction time. No spinnability appears in those sols. The sols contain granular particles. It

28

S. Sakka

Fig. 23 Viscosity versus shear rate relation for CH3COOH-containing sols of composition Si(OC2H5)4:2H2 O: C2 H5 OH:CH3 COOH = 1:2:5:0.01 (in molar ratio) reacted in closed state

100 CH3COOH−catalyzed

Viscosity / Pa•s

10

1

0.1

0.01

0.001 1

10 Shear rate /

100

1000

s−1

should be noted that the HCl-containing solutions become spinnable if they react in the open state. In a work on the spinnability and rheological behavior (Sakka and Kozuka 1987), viscosity versus shear-rate relationships and spinnability were measured for sols of composition Si(OR)4:H2O:ROH:HCl = 1:2:2:0.01 and 1:2:7:0.01 in mole ratio by a cone and plate viscometer. The Si(OR)4 used are Si(OCH3)4, Si(OC2H5)4, Si(iOC3H7)4, and Si(n-OC4H9)4. All the alkoxide solutions investigated in the study showed spinnability just before gelation in the high viscosity range. Viscosity versus shear-rate relationships for two sols containing Si(OC2H5)4 (TEOS) and Si(nOC4H9)9 (TBOS), respectively, are shown in Fig. 24a, b, respectively. The viscosity has no dependence on the shear rate up to the viscosity range of 0.1–10 Pa  s, where spinnability is observed. This indicates that these sols behave as Newtonian liquids. In other words, the Newtonian flow is an important character of the spinnable sols, irrespective of the starting silicon alkoxides. The measurements of the change of viscosity with shear rate were extended to sols other than silica sols. Figure 25 shows the viscosity versus shear-rate relations for alumina sols (Maki and Sakka 1988). Transmission electron microscopic observation shows that sol A contains granular particles, while sol B contains ribbon-like particles. Sol A is not spinnable, while sol B is spinnable. It is seen in Fig. 25 that both sols are Newtonian at viscosities lower than about 1 Pa  s. At higher viscosities, sol A shows a shear thinning behavior. On the other hand, sol B shows quasiNewtonian flow throughout the viscosity measurement range. These results indicate

Viscosity and Spinnability of Gelling Solutions

a 100

TEOS soln.

29

b 100 10 Viscosity / Pa•s

Viscosity / Pa•s

10

TBOS soln. 204 h

34.0 h 33.7 h

1

33.0 h

0.1

28.0 h

0.01

0h

0.001 1

10

100

201 h 1

198 h

0.1 185 h 0.01 0h

0.001 1

1000

10

100

1000

Shear rate / s–1

Shear rate / s–1

Fig. 24 Viscosity measured at various shear rates in a TEOS solution with molar ratio C2H5OH/Si (OC2 H5)4 = 7 (a) and TBOS solution with molar ratio n-C4H9 OH/Si(n-OC4 H9)4 = 7 (b). The solutions were reacted in open state (Sakka and Kozuka 1987), with permission of Chemical Society of Japan)

a 100

b

100 Sol B

Viscosity / Pa•s

Viscosity / Pa•s

Sol A

10

1

0.1

10

1

0.1 1

10

100

Shear rate / s

−1

1000

1

10

100

1000

Shear rate / s−1

Fig. 25 Plots of viscosity versus shear-rate for alumina sols. Arrows show the direction of the change of shear rate. The points (●) and cross-marks ( ) denote the values obtained when the shear rate is increased and decreased, respectively. Sol A contains granular particles and Sol B contains ribbon-like particles (From Maki and Sakka (1988), with permission of Elsevier)

that alumina sols which contain ribbon-like linear particles show Newtonian flow behavior up to a high viscosity range and show spinnability, just like silica sols which are Newtonian at high viscosities. Actually, alumina fibers were obtained. Kamiya et al. drew titanium oxide fibers from high viscosity sols containing titanium alkoxide (Kamiya et al. 1986), suggesting that the sol might show Newtonian flow behavior at high viscosities. This was confirmed by measuring the viscosity versus shear-rate relationships for a TiO2 sol of composition Ti(O-iC3H7):H2O:C2H5OH:HCl = 1:2:1:0.53 (in molar ratio) as a function of shear

30 1000

TiO2 sol 100 39.7 h 38.0 h 10 Viscosity / Pa•s

Fig. 26 Viscosity at 30  C as a function of shear rate for a TiO2 sol prepared from a starting solution of molar composition Ti(i-OC3H7)4: H2O:C2H5OH:HCl = 1:2:1:0.53. The viscosity measurement was carried out for the solution kept at 40  C in an uncovered beaker for various time periods

S. Sakka

33.7 h 32.5 h

1

0.1 2.5 h

0.01

0.001 0.0001

1

10 100 Shear rate / s−1

1000

rate by a cone and plate viscometer. As shown in Fig. 26, the sols are Newtonian at high apparent viscosities, where they are spinnable. Sakka et al. made YBa2Cu3O7–x superconducting oxide fibers by heating gel fibers drawn from the acetate sol of Y:Ba:Cu = 1:2:3 (in mole ratio) (Sakka et al. 1988; Umeda et al 1988). The viscosity versus shear-rate relationships shown in Fig. 27 indicate that the acetate sol is quasi-Newtonian, although, strictly, this sol is still slightly shear thinning. Besides those mentioned above, many other fibers were prepared. Fibers in the systems ZrO2–SiO2 and Na2O–ZrO2–SiO2 were drawn from viscous sols obtained by exposing solutions containing Zr(OC3H7)4, C3H7OH, NaOCH3, and Si(OC2H5)4 to air at room temperature (Kamiya et al. 1980). This corresponds to the hydrolysis by a small amount of water. Brenna drew fibers of the system ZrO2–SiO2 from alkoxide-solvent solutions with a small amount of HCl or HNO3 as catalyst (Brenna et al. 1991). No spinnability appeared when NH4OH was used as catalyst or a large amount of water was used with HCl (Shin and Han 1994). Mullite fibers were drawn from the sol mixture of alumina sol of pH = 2 and silica sol of pH = 3 by Chatterjee et al. (2002). Toyoda et al. prepared PbTiO3 fibers by drawing from solutions containing PbTi(OC2H4OCH3)6 and H2O with HCl as catalyst (Toyoda et al. 1997). Park et al. showed that the alkoxide sols for PbTiO3 gels are spinnable, when the catalyst is HNO3 and not spinnable, when the catalyst is NH4OH (Park et al. 1999). The above experimental results and discussion concerning the sol structure–rheological behavior-spinnability relationships can be summarized as follows.

Viscosity and Spinnability of Gelling Solutions 1000 Y,Ba,Cu acetate system 100 145.5 h,15 ml 10 Viscosity / Pa•s

Fig. 27 Viscosity as a _ of the function of shear rate γ, acetate sol of Υ:Ba:Cu = 1:2:3 (atomic ratio). Here, 0.0188 mol of Y(CH3COO)3  4HO, 0.0367 mol of Ba (CH3COO)2, and 0.0564 mol of Cu(CH3COO)2 were dissolved in 300 ml water

31

120 h, 20 ml 1

0.1 99.5 h, 30 ml 0.01 73.5 h, 70 ml 0.001 26 h, 230 ml 0.0001 1

10

100

1000

Shear rate / s−1

1. Spinnable sols contain long-shaped, linear particles, while most particles are granular in nonspinnable sols. 2. Spinnable sols behave as Newtonian liquids at high viscosities where the fibers are drawn from the sol. On the other hand, nonspinnable gels become non-Newtonian at high viscosities, showing shear thinning and thixotropic behavior. 3. In the spinnable sols, the viscosity increase is caused by one-dimensional growth of linear particles, while in nonspinnable sols, the viscosity increase in the final stage is caused by the formation of bonding between granular particles. 4. Spinnable sols can be prepared from solutions with an acid catalyst and small amounts of water.

Conclusive Remarks In this chapter, the author discussed various methods for measuring the viscosity of liquids and the change of the viscosity of sol–gel solutions during the sol to gel transition. Then, the occurrence of spinnability was discussed, based on the viscosity behavior of solutions. It should be pointed out that viscosity and rheological behavior are very important in sol–gel science and technology, because most sol–gel technologies involve the change of rheological properties during the gelation of the sol.

32

S. Sakka

References Badgley WJ, Mark H. High molecular weight organic compounds. New York: Interscience; 1949. Brenna U, Carturan G, Sorarù GD. Rheological behavior of solutions affording SiO2 and SiO2/ZrO2 fibers. J Non-Cryst Solids. 1991;124:191–8. Casson N. In: Mill CC, editor. Rheology of disperse systems. London: Pergamon; 1959. p. 84. Chatterjee M, Naskar MM, Chakravorty PK, Ganguli D. Mullite fibre mats by a sol–gel spining technique. J Sol–Gel Sci Technol. 2002;25:169–74. Debsikdar JC. Effect of the nature of the sol–gel transition on the oxide content and microstructure of silica gel. Adv Ceram Mater. 1986;1:93–8. Doi M. Viscoelastic and rheological properties. In: Thomas EL, editor. Materials science and technology, vol. 12, structure and properties of polymers. Weinheim/New York/Basel/Cambridge: VCH; 1993. p. 391–425. Drabarek E, Bartlett JR, Hanby HJM, Woolfrey JL, Muzny CD, Butler BD. Shear-induced restructuring of colloidal silica gels. J Sol–Gel Sci Technol. 2000;19:279–83. Einstein A. Eine neue Bestimmung der Molek€ uldimensionen. Ann Phys j Non-cyst solids 1906;19:289–306. Guizard C, Achddou JC, Larbot A, Cot L. Sol-to-gel transition in reversal micelle microemusions: III. Rheology. 1992; 147–148:681–685. Huggins ML. The viscosity of dilute solutions of long-chain molecules: IV. Dependence on concentration. J Am Chem Soc. 1942;64:2716–8. Kamiya K, Sakka S, Tatemichi Y. Preparation of glass fibre of the ZrO2–SiO2 and Na2O–ZrO2–SiO2 systems from metal alkoxide and their resistance to alkaline solution. J Mater Sci. 1980;15:1765–71. Kamiya K, Yoko T, Sakka S. Preparation of oxide glasses from metal alkoxides by sol–gel method – investigation on the type of the siloxane polymers produced in the course of hydrolysis of Si (OC2H5)4. J Ceram Soc Jpn. 1984;91:242–7. Kamiya K, Taniomoto K, Yoko T. Preparation of TiO2 fibres by hydrolysis and polycondensation of Ti(O–i–C3H7)4. J Mater Sci Lett. 1986;5:402–4. Kanbara C, editor. Experimental study on high polymers. Mechanical properties I. Tokyo: KyoritsuShuppan; 1982. Keysar S, Cohen Y, Shagal S, Slobodisnsky S, Grader GS. Effect of aging on alumina gels rheology and aerogels surface area. J Sol–Gel Sci Technol. 1999;14:131–6. Kozuka H, Kuroki H, Sakka S. Flow characteristics and spinnability of sols prepared from silicon alkoxide solution. J Non-Cryst Solids. 1988;100:226–30. Larbot A, Hours T, Berger P, Charpin J, Cot L. Study of sol–gel transition during hafnium alkoxide hydrolysis. J Non-Cryst Solids. 1992;147–148:85–91. Maki T, Sakka S. Flow properties and fiber formation of alumina sols. J Non-Cryst Solids. 1988;100:303–8. Mizuno T, Phalippou J, Zarzycki J. Evolution of the viscosity of solutions containing metal alkoxides. Glass Technol. 1985;26:39–45. Okamura S, Nakajima A, Onogi S, Kawai H, Nishijima N, Higashimura T, Ise N. Various properties of polymeric materials, Chapter 4. In: Introduction to polymer chemistry. Tokyo: Kagakudonin; 1981. Onogi S. Rheology for chemists. Kyoto: Kagakudonin; 1982. Park YI, Kim CE, Lee HW. Effects of catalyst and solvent on PbTiO3 fibers prepared from triethanolamine complicated titanium isopropoxide. J Sol–Gel Sci Technol. 1999;14:149–62. Rabinovich EM, Kopylov NJ. Rheological behavior of low-surface-area-particulate silica sols in the presence of F ions. In: Mackenzie JD, Ulrich DR, editors. Ultrastructure processing of advanced ceramics. New York: Wiley; 1988. p. 285–93. Sacks MD, Sheu R-S. Rheological properties of silica sol–gel materials. J Non-Cryst Solids. 1987;92:383–96.

Viscosity and Spinnability of Gelling Solutions

33

Sakka S, Kamiya K. The sol–gel transition in the hydrolysis of metal alkoxides in relation to the formation of glass fibers and films. J Non-Cryst Solids. 1982;48:31–6. Sakka S, Kamiya K, Kato T. Viscosity change and spinnability of Si(OC2H5)4–H2O–C2H5OH solutions on hydrolysis. Yogyo-Kyokai-Shi. 1982;90:555–6. Sakka S. Formation of glass and amorphous oxide fibers from solutions. Mater Res Soc Symp Proc. 1984;32:91–9. Sakka S, Kamiya K. Preparation of shaped glasses through sol–gel method. In: Davis RF, Palmour III H, Porter RL, editors. Emergent process methods for the high technology ceramics. New York: Plenum; 1984. p. 83–94. Sakka S, Kozuka H. Fiber drawing from silicon alkoxide solutions. Chem Lett. 1987; 16:1763–1766. Sakka S, Kozuka H. Rheology of sols and fiber drawing. J Non-Cryst Solids. 1988;16:142–53. Sakka S, Kozuka H, Umeda T. Fabrication of YBa2Cu3Oy fibers through sol–gel method. J Ceram Soc Jpn. 1988;96:468–70. Sakka S, Yoko T. Fibers from gels. J Non-Cryst Solids. 1992;147&148:394–403. Shin DY, Han S-M. Spinnability and rheological properties of sols derived from Si(OC2H5)4 and Zr(O–nC3H7)7 solutions. J Sol-Gel Sci Technol. 1994;1:267–73. Sowman HG. Alumina–baria–silica ceramic fibers from the sol–gel process. In: Klein LC, editor. Sol–gel technology for thin films, fibers, preforms, electronics and specialty shapes. Park Ridge: Noyes; 1988. p. 162–83. Takahashi K, Tanioka M. Studies on the drawing sphere viscometer. Appl Phys (Japan). 1966;35:786–96. Taneda N, Matsusaki K, Arai T, Mukoyama T, Ikemura M. Properties of silica fibers prepared by sol–gel method. Asahi-Glass Res Rep. 1988;38:309–18. Toyoda M, Hamaji Y, Tomono K. Fabrication of PbTiO3 ceramic fibers by sol–gel processing. J Sol–Gel Sci Technol. 1997;9:71–84. Tsuchida H. Science of high polymers. Tokyo: Baihukan; 1975. p. 85–7. Umeda T, Kozuka H, Sakka S. Fabrication of YBa2Cu3O7–δ superconducting fibers by the sol–gel method. Adv Ceram Mater. 1988;3:520–2. Wolf C, R€ussel C. Sol–gel formation of zirconia: preparation, structure and rheology of sols. J Mater Sci. 1992;27:3749–55.

Evolution of the Mechanical Properties During the Gel–Glass Process Thierry Woignier, F. Despetis, P. Etienne, A. Alaoui, L. Duffours, and J. Phalippou

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanical Behavior of Alcogels and Aerogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elastic Moduli and Mechanical Rupture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Toughness and Critical Flaw Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Weibull Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stress Corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aerogel to Glass Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sintering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Plastic Densification in Silica Aerogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cellular Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Percolation Gelation Analogy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fractal Dimension Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Blobs and Links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 4 4 7 11 13 15 15 16 20 21 21 22 22 23 24

T. Woignier (*) IMBE, CNRS, IRD, Aix Marseille Université, Avignon Université, CAEC, Le Lamentin, Martinique, France e-mail: [email protected] F. Despetis • P. Etienne • J. Phalippou Laboratoire Charles Coulomb, Montpellier Université, Montpellier, France A. Alaoui Faculté des Sciences et Techniques de Tanger, Tangier, Morocco L. Duffours PRIME Verre, Montpellier, France Le Lamentin, Martinique, France # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_43-1

1

2

T. Woignier et al.

Abstract

Different kinds of structure in alcogels and aerogels (fractal or not fractal) can be synthesized by a control of the chemical parameters and also by different steps in the preparation such as sintering and plastic compaction. The porosity of the gels is affected either by the adjustment of the gelifying concentration, by a precise control of the viscous flow sintering process, or by an isostatic pressure deformation. The different kinds of gels cover the whole range of porosity between 99 % and 0 %, and their mechanical properties (elastic modulus, strength, toughness) are strongly dependent on the porosity but also on their structure. We follow the mechanical properties of the over the whole process alcogel – aerogel – glass. They vary by five orders of magnitude as a function of the density, and for the same relative density, the elastic modulus and strength can increase by one order of magnitude due to a change in connectivity. The influence of the sintering process compared to isostatic pressure on the mechanical properties is explained by the associated structural changes.

Introduction The most fascinating features of gels (low sound velocity, high specific surface area, low thermal and electric conductivity, etc.) are generally due to their very large porosity, which can be as high as 99 %. However, the counterpart of this huge porosity is poor mechanical properties and the consequence is that gels tend to crack during drying. So, one of the most difficult problems in sol–gel science is to make large bodies of dried gels (xerogels or aerogels). The drying stresses are attributed to capillary phenomena and differential strain which result from a pressure gradient in the pore liquid (Brinker and Scherer 1990). Different ways have been studied to solve this problem, for example, supercritical drying (SD) allows one to avoid the capillary stresses and monolithic aerogel can be obtained (Kistler 1932). The potentialities of this material (Cerenkov detector, acoustic or thermal insulator, host matrix for catalysts) are increased if considered not only as an end product but as a precursor. By a set of sintering treatments, the silica aerogels can be easily transformed into pure silica glass (Woignier et al. 1990), and appropriate heat treatments lead to Partially Dense Aerogel (PDA) which can be used as a host matrix for the synthesis of doped glasses or composites (Reynes et al. 2001; Woignier et al. 2011). The mechanical behavior of gels, xerogels and aerogels, is generally described in terms of brittle and elastic materials, like glass or ceramics (West et al. 1988; Zarzycki 1988; Woignier and Phalippou 1988a; Hafidi-Alaoui et al. 2000; Despetis et al. 2004; Woignier et al. 2015). During mechanical testing even for very porous material (99 %), the stress–strain curve shows a perfect elastic behavior and the conchoidal fracture morphology indicates that the material is brittle, like a conventional glass. The main difference, compared to silica glass, is the order of magnitude

Evolution of the Mechanical Properties During the Gel–Glass Process

3

of the elastic and mechanical moduli which are 104 times lower. However, if this analogy is pertinent when gels are under a tension stress (bending test), they exhibit a more complicated response when the structure is compressed (compression test). The network is linearly elastic under small strains, then exhibits yield, followed by densification and plastic hardening (Pirard et al. 1995; Scherer et al. 1995; Duffours et al. 1995; Perrin et al. 2004; Phalippou et al. 2004). As a consequence of the plastic shrinkage, it is possible to eliminate the pores and stiffen the gel at room temperature. We will see that these opposite behaviors (elastic and plastic) are surprisingly related to the same two kinds of gel features: the silanol content and the pore size distribution. Different kinds of gel structure (fractal or not fractal) have been synthesized, through a control of the different steps of transformation, such as sintering and plastic compaction and were studied as a function of their structure. The influence of the sintering process, compared to the plastic transformation, on the elastic properties and mechanical behavior is explained by the associated structural changes. The relationships between structural and mechanical properties will be discussed in terms of the cellular model, percolation theory, fractal structure, and the blobs and links model.

Experimental Procedure Different families of gels have been elaborated for these studies. The main are silica alcogels and aerogels. The alcogels are prepared by hydrolysis and polycondensation reactions of tetramethoxysilane (TMOS). The TMOS is dissolved in various amounts of ethanol, thereby adjusting the oxide content of the sol (and the final bulk density of the material). The solutions are hydrolyzed under neutral, basic (NH4OH, 5
102 N), or acidic (HNO3, 104 N) conditions. The alcogels are transformed into aerogels by supercritical evacuation of the solvent. For several samples, the supercritical heat treatment is not followed by the evacuation of the superfluid, so the solvent invades the gel during cooling. The interest of this procedure is to prepare materials full of solvent for which the solid network has undergone the same heat treatment as for classical aerogels. These samples are called “rewetted aerogels.” The samples are labeled N, B, or A (with respect to the catalyst), followed by the TMOS weight percent. The densification of the aerogels is obtained by heat treatment at a temperature of 1050  C and, as a function of the sintering time, the bulk density increases. The samples are labeled PDAxx (Partially Dense Aerogel) where xx is the bulk density expressed in g/cm3. The PCAxx samples (Partially Compressed Aerogel) correspond to gels whose density has been increased by room temperature compression. The elastic moduli and the modulus of rupture of the samples were measured by the three-point bending technique, using an Instron testing machine with a 20 N load cell and by sound velocity. The fracture toughness was measured using the singleedge notched beam (SENB) method in the three-point bending. Generally, the beam bending data allow calculation of Young’s modulus, E. That is true for aerogels, but, in the case of the alcogel samples, the beam bending test yields the shear modulus, G.

4

T. Woignier et al.

Due to the low permeability of the alcogels, the fluid forces the network to behave as if it were incompressible. Then the volume of the sample is unchanged by the deformation, and the fluid exerts a transverse stress on the gel network, providing a measurement of the shear modulus G (Scherer et al. 1988). In order to compare the elastic properties of the aerogels and the alcogels, the measurements have been made on alcogels and rewetted aerogels. Their structure will be characterized by small-angle scattering techniques (SANS and SAXS). Scattering vectors q, ranging from 0.0018 to 0.3 Å1, were explored to allow a determination of the density–density correlation function in the length scale from 3 to 500 Å.

Mechanical Behavior of Alcogels and Aerogels Elastic Moduli and Mechanical Rupture Whatever the goal of the aerogel synthesis, it is important to know how the supercritical drying (SD) can modify the physical and chemical features of the parent alcogel. The object of this part is to characterize the evolution of the mechanical properties. We will study the influence of synthesis parameters such as the concentration of gelifying precursors, pH of the hydrolysis solution and aging, on the physicochemical transformations observed during the supercritical fluid extraction. In the literature, mechanical testing has been made either on alcogels (West et al. 1988; Zarzycki 1988; Scherer et al. 1988) or on aerogels (Calemczuck et al. 1987; Dumas et al. 1990; Woignier et al. 1988b; Gross and Fricke 1992; Woignier et al. 2009), and comparison of the two sets of data suggests that supercritical drying enhances the mechanical features of the samples (Hafidi-Alaoui et al. 2000). The application of linear elastic fracture mechanics to wet gels is questionable, because it is necessary to justify that the material has an elastic behavior and can be treated as a continuum. Scherer (1992) has discussed this problem and has shown that it seems reasonable to apply fracture mechanics, because the elastic region near the tip of the crack is much larger than the plastic deformation at the crack tip. As previously mentioned, the beam bending test yields different elastic moduli for alcogels (G) and for aerogels (E). If we want to follow the influence of SD on the mechanical features, the shear modulus has been measured on alcogels and rewetted aerogels. G can also be calculated from E and the Poisson’s ratio v. The shear modulus is related to E by E = 2(1 + v)G. The Poisson’s ratio, measured by Brillouin scattering and sound velocity propagation, is close to 0.2 over the whole range of aerogel density and for different catalysts. In fact G calculated from E and v and G measured on rewetted aerogels give similar results (Woignier et al. 1992). Fig. 1 shows the influence of SD on the mechanical properties G and the mechanical strength σ of a neutral set of samples. It is clear that G and σ increase by a factor of 10–20. The effect of concentration is also shown and obviously the mechanical properties improve with the TMOS concentration, over almost two

Evolution of the Mechanical Properties During the Gel–Glass Process 108

5 107

107

106

106

105

105

σ (Pa)

G (Pa)

Neutral set

104

G alcogels ” R.aerogels σ alcogels ” R.aerogels

104 10

30 TMOS (%)

50

103

Fig. 1 Evolution of the elastic and mechanical properties G and σ as a function of the TMOS content for the neutral sets of alcogels and rewetted gels

orders of magnitude. However, during SD, a shrinkage is observed and the change of the mechanical properties could be attributed to this shrinkage, which increases the load-bearing fraction of solid. In Fig. 2, the shear modulus of the different sets of samples (neutral, basic, and acid) has been plotted as a function of the fraction of solids. G values of rewetted aerogels are still higher by a factor of 4–5. This figure shows also that, if the acidic and neutral samples have nearly the same mechanical properties, for the basic set, the values of G are lower. To explain the strengthening of the materials during SD, we can invoke two kinds of processes occurring in the autoclave. The first must be related to the formation of siloxane bonds between “dangling bonds” in the alcogel. These bonds contribute to the mass, but not to the connectivity of the network. When two branches come into contact, condensation reactions of silanol groups take place, increasing the connectivity. This process would increase the stiffness and the strength, but would also impose stresses on the alcogel network which could explain the shrinkage. The second mechanism of strengthening is due to the growth of the necks between particles. This growth occurs by a mechanism of dissolution–reprecipitation (driven

6

T. Woignier et al.

Fig. 2 Evolution of G as a function of the bulk density for neutral, acid, and basic sets of alcogels and rewetted gels

G of acid alcogel ” ” ”

108

”

G (Pa)

”

R. aerogels ” neutral alcogels R. aerogel ” basic arogels ”

R. aerogels

107

106

105

0

0.2

0.4

ρ (g/cm3)

by a difference in solubility), which transfers silica from the particle surfaces to the necks. This process is accelerated by high temperature and pressure (Iler 1979). In the case of acid and neutral sets which exhibit large shrinkages, the first mechanism is more likely. Acid gels are described as made of small primary particles, forming polymeric chains (Brinker and Scherer 1990). The flexibility of the chains allows contact and condensation reactions. For basic catalysis, the network is built of larger particles (Brinker and Scherer 1990) and the branched parts are more widely spaced, limiting the possibility of new bond formation by the first mechanism. For these reasons, we propose that the strengthening of the basic set is related to the second mechanism. In Fig. 3, the effect of long aging times is shown. We observe that between 15 and 75 days, the samples continue to shrink slightly (the density increases), but the mechanical properties are almost constant. Contrary to a common belief, a very long aging is not a guarantee of gel strengthening. The curves obtained by SANS show the structural differences between aged (75 days) and unaged (1 day) aerogels (Fig. 4). The principal parameter that we can derive from the SANS curves is the fractal dimension, Df, which is related to the mass distribution in the cluster. Df is calculated from the slope of the linear part. The size of the fractal cluster (ξ) and of the elementary particles (a) which built the cluster can be estimated from the two

Evolution of the Mechanical Properties During the Gel–Glass Process 109

7 108

N46 N40 N33

107

106

σ (Pa)

E (Pa)

N26

10D_15D_75D

N26

6D 1D

105

0

0.2

0.4

104

ρ (g / cm3)

Fig. 3 Evolution of E and σ as a function of the bulk density and aging, for the neutral aerogels. N26, N33, N40, and N46 correspond, respectively, to aerogels with 26, 33, 40, and 46 TMOS weight percent. 1D, 6D, 15D, and 75D correspond, respectively, to 1 day, 6 days, 15 days, and 75 days of aging

crossovers at low and high q [19]. For different samples, the fractal range ξ/a is strongly reduced both by the decrease of ξ and the increase of a. If we compare the effects of increasing concentration of TMOS and a longer aging, the effects on ξ and a are not identical. Both parameters decrease the fractal range ξ/a, but the decrease in ξ is more pronounced by the effect of concentration and a long aging leads to the growth of a. As the decrease in ξ corresponds to a lower macroporous volume and the growth of a to the elimination of the microporosity, we will see later that these two factors have different influences on the mechanical properties.

Toughness and Critical Flaw Size For brittle materials, the mechanical strength (σ R) is strongly dependent on the presence of flaws, which act as stress concentrators (Griffith 1920). The most relevant feature of brittle materials is the toughness (KIC), which characterizes the

8

T. Woignier et al.

Fig. 4 Evolution of SANS curves, as a function of aging, for the neutral aerogels

104 Unaged

N10

s ( q ) (a.u.)

Aged 75 days

102

N18

1

N26

10–2

N46

10–4

10–3

10–2

10–1 q

(A°–1)

ability of the material to resist the propagation of flaw. Pores could be considered as flaws or as integral parts of flaw responsible for the failure of this brittle material. If such an assumption is valid, the critical flaw size should depend on the porosity and pore size (Hafidi-Alaoui et al. 2000). Then, we have measured the toughness for alcogels and aerogels with different porosities in order to analyze the toughness evolution during the alcogel–aerogel transformation. The critical flaw size, aC, will be derived from KIC data and discussed in terms of pore size evolution. Figure 5a, b shows the evolution of KIC for the two sets of materials (basic and neutral) as a function of the TMOS vol%. This figure also shows the toughness change upon supercritical drying. KIC increases by a factor of 10 over the TMOS concentration range. The toughening of the two sets of materials is directly related to the decrease of pore volume. But clearly, the reported values are 103 lower than those measured on dense silica glass (0.7–0.8 MPa
m1/2) (Chermant et al. 1980). Taking into account the textural and structural transformations of a gel during SD, a toughness increase is expected. Figure 5 confirms this assumption and the toughening by a factor close to 2 is observed for the two types of samples. Toughening is attributed to the syneresis effect and to the growth of necks between particles.

Evolution of the Mechanical Properties During the Gel–Glass Process

9

Fig. 5 Evolution of KIC as a function of TMOS content: (a) for the basic alcogels and aerogels and (b) for the neutral alcogels and aerogels

The knowledge of KIC and σ R allows one to calculate the critical flaw size, aC (Evans and Tappin 1972). Assuming a Griffith flaw (as for silica glass), the critical size of the fracture initiating flaw may be obtained from aC = l/1.2 l π (KIC/σ)2. The aC evolution as a function of the TMOS% is reported in Fig. 6a, b. The aC data range within 80 and 700 μm and the general trend is an aC decrease with the TMOS content and also during the supercritical drying. The comparison between the two kinds of catalysts (basic and neutral) shows that the aC values for the neutral set are much more affected by SD. A correlation could exist between aC and the evolution of the pore size. The total porous volume in a gel consists of the macroporosity (R > 50 nm), the mesoporosity (R = 2–50 nm), and the microporosity (R < 2 nm). A previous study (Pauthe 1991) has shown that, for the two sets of material (alcogel and aerogel), the microporous and mesoporous volumes are not strongly affected by the TMOS%. On the other

10

T. Woignier et al.

Fig. 6 Evolution of aC as a function of TMOS content: (a) for the basic alcogels and aerogels and (b) for the neutral alcogels and aerogels

Table 1 Macro-, meso-, and micropore volumes for alcogels and aerogels

Vmacro (mm3/g) Vmeso (mm3/g) Vmicro (mm3/g)

Alcogel N26 5610 430 125

N33 4850 420 100

N40 3720 460 80

B33 4380 1750 80

Aerogel N26 N33 2590 780 2460 2720 0 0

N40 380 2290 0

B33 3840 2320 0

hand, the macroporous volume decreases strongly and the pore size distribution changes. The micropores disappear and the mesoporous volume increases at the expense of the macroporous volume (Table 1). This transformation is related to the syneresis process, shrinkage, and dissolution–reprecipitation of silica, favored by the supercritical treatment conditions (high pressure and temperature).

Evolution of the Mechanical Properties During the Gel–Glass Process

11

The ac values suggest that, during SD, the critical flaw size decreases. This can be associated to the transformation occurring during the autoclave treatment, which produces both a macroscopic shrinkage of the material and the evolution of the pore size distribution (Table 1). Consequently the pore could be considered as flaw or as an integral part of flaw responsible for the failure. For the different parameters studied (TMOS%, catalyst, SD), aC decreases with the macropore size. We note that the aC values are much larger than the pore size, so the scale of critical crack extends on a large number of pores. To satisfy this statement, it is necessary to consider that the flaws, which lead to failure, might be created during the test. We can suppose that the failure occurs by progressively breaking bonds, following the minimum solid area and collapsing a large number of pores located between “clusters.” The macropores link into a macroscopic flaw and catastrophic failure occurs, when the size of the flaw becomes critical. These aC values suggest that, during the supercritical drying, the flaw size changes due to the thermal treatment and the resulting shrinkage. However, if this assumption is valid, the flaw size distribution could be modified and the Weibull modulus (m), which characterizes the width of the distribution, has every chance to change. The goal of the next section is to determine the Weibull parameters of both sets of gels, alcogels and aerogels.

Weibull Statistics In Weibull statistics, the failure probability function, P(σ) is given as  ð Pðσ Þ ¼ 1  exp 

V or

  m σ dV or dS S σ0

(1)

where m is the shape parameter (Weibull modulus), σ 0 is the scale parameter (a normalizing constant), and the integral is taken over the volume (V) or surface (S) under tension. When all the specimens have the same volume, the expression can be simplified (Sullivan and Lauzon 1986), resulting in   m  σ Pðσ Þ ¼ 1  exp  σ0

(2)

The analysis consists of converting a set of fracture stresses into an experimental probability distribution. This is done by arranging the results, from the lowest fracture stress to the highest. The jth result in a set of N specimens is assigned a cumulative probability of failure Pj. The estimator is chosen as Pj ¼

j  0:5 N

(3)

12

T. Woignier et al.

P (σ)

1,0

0,8

0,6

0,4

0,2 σ (KPa) 0,0

0

200

400

600

800

1000

Fig. 7 Experimental and calculated cumulated failure probability distribution function, from left to right (N18,x, N26,~, N33,■, N40,-, and N46,♦)

which already allows a good determination of the m value, for sets within the range 20–30 specimens. Sets of 25 specimens (for each type of material tested, alcogels and aerogels) have been studied. Weibull modulus (m) and scale parameter (σ 0) values were derived from the calculations with 90 % confidence limit. Figure 7 shows the large scattering in the data due to the statistical nature of the mechanical strengths. On this figure is reported the experimental and calculated cumulated failure probability distribution function using the Weibull statistical analysis. The small value of m (Table 2) is characteristic of the wide distribution of the strength. While m varies slightly, σ 0 increases with TMOS %. These results show that critical flaw sizes are probably less dispersed for aerogels with higher TMOS content which corresponds to aerogels with a narrow pore size distribution. Weibull modulus (m) and σ 0 deduced from the calculations for the different aerogels are plotted in Table 2. For the alcogel sets (Table 3), the smaller value of m is a characteristic of a wider distribution of strengths, m, and σ 0 increases with SD. These results show that critical flaw sizes are more dispersed for alcogels than for aerogels and confirm the strengthening of the structure attributed to the polycondensation of “dangling bonds.” Since the samples of both materials have approximately the same size, the probability of finding the largest flaw in the same effective volume decreases significantly from alcogels to aerogels. This result can rather be explained by the difference between their texture and particularly assuming that critical flaw is bound to the connectivity in both materials. Alcogels have a low connectivity as was shown previously; as a result, the probability of finding the weakest link in the stressed volume of alcogels is rather high.

Evolution of the Mechanical Properties During the Gel–Glass Process Table 2 Weibull parameters for different aerogels

Sample m σ0

Table 3 Weibull parameters for alcogels and aerogels

Sample Alcogel N40 Aerogel N40 Alcogel B40 Aerogel B40 Aerogel B40 Ox

N18 4 80

N26 5  0.5 175.

N33 5.5  0.5 380

m 5.0 5.9 5.1 6.4 10

    

0.4 0.6 0.5 0.6 0.9

13

N40 6  0.5 500.

N46 7  0.5 720. σ 0 (KPa) 210  2 500  5 120  1 150  2 180  2

The last line of Table 3 gives m and σ 0 for a sample aerogel N40 after an oxidation heat treatment. The oxidation heat treatment replaces the organic groups (generally present at the surface of the solid phase) by OH groups; the surface becomes hydrophilic. The experimental results show a small increase of σ 0 and a large increase of m. This preliminary result shows the clear influence of the chemical nature of the solid surface, on the mechanical behavior of the gels. It is a good introduction to the “stress corrosion effect” developed in the following section.

Stress Corrosion We have stated that careful drying and, specially, supercritical drying allow the preparation of large monolithic gels. However, several months after drying some of the gels and aerogels can crack without any stress applied. A first analysis shows that the reason of these cracks could be the “stress corrosion effect.” This phenomenon is well known in the case of glasses, which can crack under a constant stress lower than the mechanical resistance. It is due to the joint effect of a stress and a corrosion mechanism at the tip of the flaws, by chemical species such as water or alcohol. It is known that the strength and fatigue lifetime of vitreous silica decrease in humid environments (Michalske and Freiman 1983). In the case of gels and aerogels, the internal stresses coming from the shrinkage, the syneresis, and/or the thermal treatment would be responsible for this effect. Moreover, the stress corrosion induced by the supercritical fluid could also explain some cracking in the autoclave (Woignier et al. 1994). Understanding the whole fracture behavior of silica gels implies the study of the stress corrosion effect, because mechanical fatigue will limit their technological applications if the materials are under stress (Crichton et al. 1999; Woignier et al. 2009). The objective of this section is to give additional information to confirm this hypothesis. One of the important corrosion parameters is the stress corrosion susceptibility factor, n. The exponent n characterizes the evolution of the crack velocity as a function of the stress intensity factor. It is determined by fatigue experiments, under the corrosive environment of interest (Despetis et al. 2004).

14

T. Woignier et al.

We have shown that failure is the result of the stress concentration at the flaw tip. When stressed, a brittle material is characterized by a given value of the stress intensity factor KI. Fracture occurs when either the applied stress or the flaw length or both increase in such a way that KI reaches the value, KIC (the critical stress intensity factor or fracture toughness). However, under a corrosive environment, the flaw can grow with time at a rate usually expressed by the following empirical relationship (Evans 1974): V ¼ AK nI

(4)

where A and n are constants and n is the stress corrosion susceptibility factor. Figure 8 shows the evolution of the crack velocity as a function of the stress intensity factor for two kinds of aerogels (as prepared and oxidized). As explained below, the main difference between these two aerogels is the groups bonded to the surface. The surface of the as-prepared aerogels is covered by organic groups which are replaced by OH groups after the oxidation heat treatment. We can say that these porous silica networks are identical except for their hydroxyl contents. At low KI, V is two orders of magnitude higher for the hydrophilic gel. We note that n is identical in this KI range for the two samples, which means that the mechanism associated to crack propagation is the same; n is close to 15 and corresponds to a corrosion mechanism by OH (Despetis et al. 2001). At higher KI, n changes and becomes equal to 35, the classical n value found for vitreous silica, associated to a corrosion mechanism by water molecules (Suratwala and Steele 2003). To explain crack growth in dense silica under air moisture, it was assumed that water molecules close to a strained siloxane bond transform the bridging oxygen in two silanol bonds. The difference in crack velocity between hydrophobic and oxidized aerogels is

Fig. 8 Evolution of the crack velocity as a function of the stress intensity factor, for as-prepared and oxidized aerogel

Air moisture

Crack velocity (m/s)

10–2 B022 oxidized

10–4 n =15 –6

10

B022 as prepared 10–8 0.5

1

1.5 1/2

K1 (kPa m

)

2

Evolution of the Mechanical Properties During the Gel–Glass Process

15

related to the surrounding chemical species. For hydrophilic aerogels, the pores are covered by OH groups and highly structured water molecules, which play the role of a basic catalyst. In the case of hydrophobic materials, OH groups are fewer and the hydrolysis is screened by the presence of organic groups. Because hydrophobic aerogel surfaces have a lower relative humidity, basicity is low and the crack propagation is slower. These results put in evidence the influence of the gel OH content of the gel on its mechanical behavior. Because of the stress corrosion effect, a monolithic gel can crack several days or months later, even under a low stress (e.g., an internal stress), if its OH content favors the progressive hydrolysis of the siloxane bonds leading to the formation of a critical flaw.

Aerogel to Glass Transformation Sintering Aerogels are easily transformed into dense silica glass by oxidation and sintering (Woignier et al. 1990, 2011). During these treatments, the structure of the aerogel is modified and the mechanical properties are improved (Phalippou et al. 2004). Figure 9a, b shows the evolution on a log–log plot, of Young’s modulus (E), the fracture strength (σ), and the toughness, KIC, as a function of the density, produced by different kinds of catalyst (Fig. 9a) and by sintering (Fig. 9b). The two sets of materials, aerogels and Partially Dense Aerogels (PDA), cover the whole range of porosity. During sintering, the PDA strengthens and finally the mechanical features of the fully dense material are identical to those of conventional silica glass. The

b 1.7 107

0.1

106

107

3.7

105 2.6

103

1011

109 1.7

102

1010

108

E (Pa)

1

K1C (KPa.m1/2)

N B

10

σ (Pa)

109

σ (Pa)

10

E (Pa)

K1C (KPa.m1/2)

a

109

2.3

104

105

10-1

ρ (g/cm3)

1

107

3.2

108 2.10-1

1

106 6

ρ (g/cm3)

Fig. 9 Evolution of the elastic and mechanical properties E, σ, and KIC as a function of the bulk density (a) for the neutral and base-catalyzed gels and (b) for sintered gels

16

T. Woignier et al.

Fig. 10 Evolution of critical flaw size (aC) and the larger pore size (LPS), as a function of the bulk density, for sintered gels

100

ac

ac (μm) L.P.S. (nm)

L.P.S.

10

1

0

0.4

0.8

1.2

1.6

2

2.4

ρ (g/cm3)

strengthening of the material is directly related to the decrease of pore volume, but also, at a given density, to the increase of the connectivity of the network. During the oxidizing and sintering treatments, organic species and silanol groups are replaced by siloxane bonds, increasing the connectivity and thus improving the mechanical properties. The two sets of curves show also the power law dependence, with exponents 3.2–3.7, 2.3–2.6, and 1.7, respectively, for E, σ, and KIC. Figure 10 shows the evolution of the critical flaw size as a function of the bulk density, for the sintered aerogels. It appears that aC is within the range of 5–100 μm and decreases over the bulk density range. In Fig. 10, it is also plotted the evolution of the larger pore size (LPS), measured on the sintered material by porosimetry (Beurroies 1995; Woignier et al. 1994). In the first part, we have shown that, during supercritical drying, the aC evolution can be associated to transformations occurring in the autoclave: a macroporous volume decrease and a decrease of the mean pore size. It has been shown that aC follows the size change of the largest pores (macropores). For all the parameters studied (TMOS%, catalyst, supercritical drying), the critical flaw size decreases with the macroporous volume. During sintering the macroporous volume is progressively reduced and, when all the macropores have disappeared, the largest mesopores begin to collapse. Like during supercritical drying, we can associate the aC decrease during sintering to the larger pore size (LPS) and macroporous volume decrease. As already mentioned, the aC values are much larger than the pore size, so the scale of critical crack extends over a large number of pores. The failure occurs by bond breaking, collapsing a large number of pores.

Plastic Densification in Silica Aerogels In the preceding section, silica gels are described as purely elastic materials and the stress–strain relation evolves like for a common elastic material toward a

Evolution of the Mechanical Properties During the Gel–Glass Process

17

“catastrophic” fracture. No plastic deformation has been reported, when a gel is subjected to a tensile load. However, previous works (Pirard et al. 1995; Scherer et al. 1995; Duffours et al. 1995) have shown that if an aerogel is subjected to a compressive load, the solid network initially behaves elastically, until the strain is no longer proportional to the stress. After the stress release, an irreversible strain is observed, characteristic of a plastic behavior. The yield stress and the magnitude of the plastic shrinkage are strongly dependent on the textural and structural features of the silica aerogel (Duffours et al. 1995; Woignier et al. 2010). Owing to the plastic shrinkage, the material shrinks progressively by pore collapse and volumic shrinkage higher than 50 % can be observed. Obviously, one is tempted to draw a parallel between this densification process and the high temperature sintering mechanism which also leads to pore elimination. Moreover, this unexpected result should be explained, in order to understand the whole mechanical behavior of the gels. In this section, this plastic shrinkage is investigated and its mechanism is compared to that of viscous flow sintering. Two kinds of information will be analyzed: the evolution of the elastic properties, due to plastic shrinkage (or sintering), and the associated structural changes by small-angle X-ray scattering. The elastic properties will give insight into the evolution of the connectivity, and the X-ray scattering data will help quantify the eventual transformation of the cluster, internal compactness (fractal dimension), and size of the constitutive particles. Figure 11 shows the response behavior of a compliant gel, during a compression run. If the applied pressure is lower than the yield strength, the sample deforms elastically; the volume strain is proportional to the stress applied and, after the pressure release, the sample recovers its initial volume. When the pressure applied is higher than the yield strength (10 MPa in this case), the behavior is no longer elastic and a part of the volume strain is irreversible. Thus, as a function of the 60 50

ΔV/V0 (%)

40 30 20 (ΔV/V0)pl 10 0 0

5

10

15

20

25

30

35

Pressure P (MPa)

Fig. 11 Typical volume strain (ΔV/V0) versus applied pressure curves

40

45

T. Woignier et al.

Longitudinal modulus H (MPa)

18

800

PDA PCA Sintering

400

Compression

0 0.3

0.4 Density ρ (g/cm

0.5

0.6

3)

Fig. 12 Evolution of the longitudinal elastic modulus H, as a function of the bulk density, for sintered and compressed gels

pressure applied during the run, the sample bulk density increases. Macroscopically, pressure and temperature apparently result in the same effect; they induce pore collapse and a density increase. Generally, such a densification is accompanied by a stiffening of the solid and the elastic properties of the porous material improve with the bulk density. Figure 12 shows that the elastic longitudinal modulus H (calculated from the sound velocity, V: H = ρV2), plotted as a function of the bulk density, exhibits two different behaviors. For the PDA set, the sintering is accompanied by stiffening of the material, the sample of 0.55 g/cm3 being 10 times stiffer than the one of 0.33 g/cm3. On the other hand, the PCA set shows a lowering of the elastic modulus H, between 0.33 and 0.45 g/cm3, characterizing a loss in the network connectivity. In the second part of the curve, the PCA set shows an increase of H above ρ  0.5 g/cm3 leading to the conclusion that after the connectivity loss, compression induces the formation of siloxane bonds. The structural details obtained by small-angle X-ray scattering are used to follow the respective structural evolution of the PDA and PCA sets (Fig. 13a–c). During sintering the fractal dimension and the particle size increase, while the compression runs do not significantly change Df or a. On the other hand, the cluster size is more affected by compression than by sintering. The interpretation of SAXS data suggests that, for sintered samples, the densification is induced by viscous flow, which tends to contract the clusters and consequently reduces the whole sample volume. The densification proceeds by coalescence of small particles into larger ones. Df tends toward 3, indicating cluster densification related to particle coalescence. The local sintering has two effects: it pulls the network, shrinking the clusters, and it increases the connectivity in the whole material. On the other hand, in the case of densification by compression, the main restructuring is due to a new spatial arrangement of the clusters. The clusters

Evolution of the Mechanical Properties During the Gel–Glass Process

a

b 2.3

PCA Cluster size ξ (Å)

2.4 Fractal dimension Df

19

PDA PDA

PCA 2.1 2 0.15

0.25

0.35

150

0.45

PDA 90 PDA PCA 30 0.15

0.25

0.35

0.45

Density ρ (g/cm3)

Density ρ (g/cm ) 16 c 3

PDA 12

a

Primary particles size q–1(Å)

PCA

PCA

8 4 0.15

PDA PCA 0.25

0.35

0.45

Density ρ (g/cm3)

Fig. 13 Evolution of structural features: (a) fractal dimension, (b) cluster size, (c) particle size, as a function of the bulk density for sintered and compressed gels

interpenetrate each other, their periphery is changed, but their internal structure is not affected. This indicates that the pore volume between clusters is reduced. This rearrangement is reasonable, taking into account the process which stresses the samples by isostatic external pressure. During compression, because the solid is not viscous, such a restructuring should induce important strain and local disconnection in the network. To allow the motion of the clusters, a part of the links at the boundaries between aggregates should be broken and the whole connectivity of the network is lowered. However, because the clusters touch and interpenetrate, silanol groups (SiOH) can polycondense and the formation of siloxane bonds would increase the network connectivity. These two opposing effects, breakage and formation of links, occur simultaneously. These assumptions explain the H changes observed in Fig. 12. To check if the plastic shrinkage is also due to the formation of siloxane, Fig. 14 compares the shrinkage measured on as-received basic material and oxidized basic material (large OH content). These two samples have same bulk density and pore size distribution but differ by the OH content. As expected, the OH content favors clearly the shrinkage at low pressure. In oxidized samples, the clusters are chemically active. They may react if silanol groups are close enough to condense, leading to freezing of the strained structure. To confirm this effect, on the same figure is also reported the elastic bulk modulus of the compressed material, as a function the pressure. The large amount of SiOH in the oxidized samples favors the formation of siloxane bonds and stiffens the

20

T. Woignier et al. 400 B 0.2 as prepared B 0.2 oxidized 300

60

200 30 100

0

Bulk modulus K (MPa)

VOLUME SHRINKAGE (%)

90

100

40 P (MPa)

Fig. 14 Plastic shrinkage (ΔV/V0)pl and bulk modulus, versus pressure, for as-prepared and oxidized aerogels

network. A further shrinkage is then more difficult. The change in network connectivity (measured by the K evolution) is the result of the competition between two opposite mechanisms: the breakage of siloxane bonds and the formation of links by silanol condensation. For as-prepared aerogels, densification is due to the action of pressure, which forces clusters to interpenetrate, but only a few siloxane bonds are created. On the other hand, for oxidized samples, because of the large silanol content, the formation of new siloxane bonds largely compensates the cleavage of a part of them.

Models In the literature, many empirical relationships have been used to relate elastic and mechanical properties to porosity of brittle solids. However, these relationships are generally valid for a restricted range of porosity and their exponent is empirical and depends on the nature of the solid phase. Different approaches (cellular models, percolation analogy, fractal structure, blobs and links) have also been proposed, to account for the porous volume and calculate the evolution of the mechanical properties as a function of the structural characteristics. Such models seem attractive to describe the mechanical properties of gels for several reasons. In contrast with the empirical relationships, they try to relate the physical properties to a description of the mean structure or to the aggregation process; they also predict that the Poisson ratio is constant with the fraction of the solid phase, which is an experimental result demonstrated for aerogels and PDA. Another interesting feature of those models is that they predict a power law evolution

Evolution of the Mechanical Properties During the Gel–Glass Process

21

of the mechanical properties as a function of the fraction of solid phase, and the log–log plot of Fig. 8a, b demonstrates such a behavior for E (and for G), σ, and KIC.

Cellular Model In the cellular model (Gibson and Ashby 1988), the porous solid is defined as a material built up of struts or plates, which form the edges and the faces of the cells. The most important structural characteristic of the cellular models is their relative density, which can be expressed as a function of the cell edge length, l, and the size of the cross section, t. For an open-cell foam made of cubic polyhedra, ρ/ρs is proportional to (t/l )2. Then, the elastic and the mechanical features have been calculated, respectively: E / (ρ/ρs)2 and KlC / (ρ/ρs)3/2. The analysis of the results obtained with PDA shows that E varies as (ρ/ρs)3.2 and KIC as (ρ/ρs)1.7. To explain why the model fails, we have to consider that it assumes that all the solid phase is connected and participates in the load bearing. In fact, the bond area between particles is not taken into account and is an important parameter in the case of a real material. The mechanical properties of the PDA must be modeled by a cell having a lower t/l value than those derived from ρ/ρs. In fact, the material behaves as a cellular solid, with an effective bulk density lower than ρ/ρs. The larger experimental scaling exponent can be explained if we consider that, during sintering, the effective bulk density increases rapidly by the elimination of porosity and increase of the bond area.

Percolation Gelation Analogy To explain the power law evolution of the elastic modulus and the exponent a close to 4, it has been proposed that an analogy could exist between a gel and a percolation cluster (Stauffer 1976). In this theory, elastic properties are expected to scale as E / (P  PC)τ where P is the probability for a site to be occupied (or a bond to be created) and PC is the percolation threshold (defined as the magnitude of P above which an infinite cluster exists). Analytical work and simulations have calculated a critical exponent characteristic of the elastic modulus) and τ is close to 4 (Feng and Sen 1984; Kantor and Webman 1984). Experimental studies on gels (Tokita et al. 1984; Adam et al. 1985) have been analyzed using this analogy. The most important difficulty arises from the choice of the physical variable which must be related to the unknown mathematical variable (P  PC). Several variables, such as the concentration of monomer or density, have been proposed, assuming implicitly a proportionality between the variables and the P scale. In fact, in the straightforward gelation/percolation analogy, the gel fraction (and thus the density) is associated with the percolation probability, ℘(P) (the probability for a site to belong to the infinite cluster) and scales with an exponent, β: ρ / ℘(P) / (P  PC)β, then E / ρτ/β. The experimentally determined α in Fig. 8a, b would beτ/

22

T. Woignier et al.

β. Using a β value equal to 0.4 (the theoretical prediction in a three dimension model), we find that τ is equal to 1.5, far from the predicted exponent.

Fractal Dimension Approach Another approach found in the literature tries to relate the exponent, α, to the fractal dimension of the network (Emmerling and Fricke 1997), and a relation is proposed: α¼

5  Df 3  Df

(4)

However, this approach is not valid for the complete sets of data in the literature. For example, it cannot explain why α in Fig. 8a is similar for the different sets of aerogels, N or B, which have different fractal dimensions (2.4 and 1.8, respectively). Moreover, during sintering, Df changes and tends toward 3. The α value should increase between the aerogel sets and the sintered sets. Such an assumption disagrees with the results in Fig. 8a, b (α decreases from 3.7 to 3.2). Finally, for samples having a density >1, the fractal description is meaningless, but, in the whole range of density of 0.4–2.2, the log–log plot still leads to an α close to 3.2. It is clear that the exponent value is not related to the fractal dimension and to percolation theory. The structure and the connections of a gel are the result of a sequence of different processes: gelation, aging, and shrinkage. The α value should describe the way the clusters are connected between them and not the structure inside the clusters.

Blobs and Links In a more recent work (Ma et al. 2001), the authors proposed to use the diffusionlimited cluster–cluster aggregation (DLCA) algorithm, to generate 3-D lattice gels models. The bulk moduli were calculated by the finite element method and the result shows that DLCA models exhibit the same scaling relationship. However, the simulations indicated that α  7.6, overestimating the experimental data. The DLCA model contains excessive dangling bonds which lead to wrong estimation of the stiffness, and simulation shows that loops in the structure are important to account for the elastic behavior. A new algorithm called “dangling bond deflection” has been developed to transform the dangling bonds into loops. This algorithm simulates the motion of dangling bonds and the associated dangling branches. A bond is formed when a deflected dangling branch collides with another cluster and a loop is closed when the bonding particles belong to the same cluster. The exponent α  3.6, calculated using this gel model, is in the range of α values observed in experiments. The compliant nature of aerogels is a result of the fact that stresses localizes on a few tenuous chains. Fewer bonds share the strain when density

Evolution of the Mechanical Properties During the Gel–Glass Process

23

decreases, so the modulus of the network drops faster than for the cellular model. The structure generated by the combination of the DLCA and the dangling bond deflection algorithm can successfully reproduce the scaling relationship of aerogels. More recently, Anez et al. (2010) introduce a method for a computational calculus of the elasticity modulus (E) of simulated porous media using the Monte Carlo technique. The porous media of known geometry is simulated as an elastic network of central forces, to which a known deformation is applied. The minimum strain energy is calculated by applying the Monte Carlo technique.

Conclusion The mechanical behavior of common glasses has been analyzed for a long time and is now well known: they are described as brittle and elastic materials. But what about porous glasses? The general agreement is that porous glasses are brittle and elastic too, but the mechanical characteristics (strength, elastic constants, toughness) are lower, due to the pore volume and the associated lower connectivity of the network. The possibility to sinter gels and aerogels into dense silica glass allows us to follow the evolution of the mechanical properties over the whole range of porosity 99 % to 0 %. At each step of the process, the increase of the elastic moduli and mechanical properties is related to the gel structure and several mechanisms are involved. During supercritical drying, flexible dangling bonds can condense and form new links, but the increase of the mechanical properties is also due to a dissolution–reprecipitation phenomenon. Because of their large pore volume, the mechanical properties of dried gels are several orders of magnitude lower than those of the dense silica, but the mechanical behavior seems identical to that of brittle materials like glasses. The bulk density is the main parameter which defines the elastic and brittle features, but two others parameters (pore size distribution and OH content) are also significant to describe and understand the whole mechanical behavior. Toughness measurements have shown that the flaw size responsible for fracture seems to be related to the size of the largest pores. Failure occurs by progressive breaking bonds collapsing a large number of pores. The macropores link into a macroscopic flaw and catastrophic failure occurs when the size of the flaw becomes critical. Moreover, the porous network is sensitive to the stress corrosion effect, which can lead to failure after several months, under a low stress (monolithic aerogels have cracked because of the internal stresses, only). This effect is favored by the OH content of the gel. Besides the brittle elastic behavior, when a gel is subjected to a tensile load, under a compressive load, the porous network can be irreversibly transformed. This “plasticity” effect depends strongly on the volume fraction of pores, but is also clearly affected by macropores and by the OH content. In fact, either under tension or compression, the gel material is not stable and its structure and mechanical features evolve.

24

T. Woignier et al.

Extrapolation of the densification curves leads to the conclusion that dense glasses could be obtained by applying a pressure in the range 500–1000 MPa. This new densification process could be an interesting way to prepare glasses at room temperature. Acknowledgments The authors would like to thank the editors of Journal of Non-Crystalline Solids, Journal of Sol–Gel Science and Technology and the European Physical Journal for their permission to publish figures here reported.

References Adam M, Delsanti M, Durand D. Mechanical measurements in the reaction bath during polycondensation reaction near the gelation threshold. Macromolecules. 1985;18(11):2285–90. Anez L, Primera J, Hasmy A, Fransceni P, Sanchez N, Woignier T. A method for elasticity modulus calculation in porous media using the Monte Carlo technique. Key Eng Mater. 2010;423:75–82. Brinker CJ, Scherer GW. Sol–gel science. New York: Academic; 1990. Calemczuck R, de Goer AM, Salce B, Maynard R, Zarembowitch A. Low temperature properties of silica aerogels. Europhys Lett. 1987;3(11):1205–11. Chermant JL, Osterstock F, Vadam G. Etude critique de la mesure de Kic dans le cas de quelques matériaux verriers. Verres Refract. 1980;34(5):624–36. Crichton SN, Tomozawa M, Hayden JS, Suratwala TI, Campbell JH. Subcritical crack growth in a phosphate laser glass. J Am Ceram Soc. 1999;82:3097–104. Despetis F, Calas S, Etienne P, Phalippou J. Effect of oxidation treatment on the crack propagation rate of aerogels. J Non-Cryst Solids. 2001;285:251–5. Despetis F, Etienne P, Etienne-Calas S. Subcritical crack growth in silica aerogel. J Non-Cryst Solids. 2004;344:22–5. Duffours L, Woignier T, Phalippou J. Plasticity of aerogels under isostatic pressure. J Non-Cryst Solids. 1995;186:321–7. Dumas J, Quinson JF, Serughetti J. Hierarchy of pores and mechanical behavior of wet silica gels. J Non-Cryst Solids. 1990;125:244–9. Emmerling A, Fricke J. Scaling properties and structure of aerogels. J Sol-Gel Sci Technol. 1997;8:781–8. Evans AG, Tappin G. Effects of microstructure on the stress propagate inherent flaws. Proc Br Ceram Soc. 1972;23:275–96. Evans AG. Slow crack in brittle materials under dynamic loading conditions. Int J Fract. 1974;10:251–61. Feng S, Sen P. Percolation on elastic networks: New exponent and threshold. Phys Rev Lett. 1984;52(3):216–9. Gibson LJ, Ashby MF. Cellular solids structure and properties. Oxford, UK: Pergamon; 1988. Griffith AA. The phenomenon of rupture and flow in solids. Philos Trans R Soc London, Ser A. 1920;221:168–98. Gross J, Fricke J. Ultrasonic velocity measurements in silica, carbon and organic aerogels. J Non-Cryst Solids. 1992;145:217–22. Hafidi-Alaoui A, Woignier T, Pernot F, Phalippou J. Stress intensity factor in silica alcogels and aerogels. J Non-Cryst Solids. 2000;265:29–35. Iler RK. The chemistry of silica. NewYork: Wiley; 1979. Kantor Y, Webman I. Elastic properties of random percolating systems. Phys Rev Lett. 1984;52 (21):1891–4. Kistler SS. Coherent expanded aerogels. J Phys Chem. 1932;34:52–64. Ma HS, Prevost JH, Jullien R, Scherer GW. Computer simulation of mechanical structure–property relationship of aerogels. J Non-Cryst Solids. 2001;285:216–21.

Evolution of the Mechanical Properties During the Gel–Glass Process

25

Michalske TA, Freiman SW. A molecular mechanism for stress corrosion in vitreous silica. J Am Ceram Soc. 1983;66(4):284–8. Perrin L, Faivre AL, Calas S, Woignier T. Nano structural damage associated with isostatic compression of silica aerogels. J Non-Cryst Solids. 2004;333:68–72. Phalippou J, Despetis F, Calas S, Faivre AL, Dieudonné P, Woignier T. Comparison between sintered and compressed aerogels. Opt Mater. 2004;26:167–74. Pauthe M, Quinson JF, Hdach H, Woignier T, Phalippou J, Scherer GW. Autoclave treatment effect on silica alcogel texture. J Non-Cryst Solids. 1991;130:1–7. Pirard R, Blacher S, Brouers F, Pirard JP. Interpretation of mercury porosimetry applied to aerogels. J Mater Res. 1995;10(8):2114–9. Reynes J, Woignier T, Phalippou J. Permeability measurement in composite aerogels: application to nuclear waste storage. J Non-Cryst Solids. 2001;285:323–7. Scherer GW, Pardenec SA, Swiateck RM. Viscoelasticity in silica. J Non-Cryst Solids. 1988;107:14–22. Scherer GW, Smith DM, Qiu X, Anderson JM. Compression of aerogels. J Non-Cryst Solids. 1995;186:316–20. Scherer GW. Crack tip stress in gels. J Non-Cryst Solids. 1992;144:210–4. Stauffer D. Gelation in concentrated branched polymer solution. J Chem Soc Faraday Trans. 1976;2:1354–64. Suratwala TI, Steele RA. Anomalous temperature dependence of sub-critical crack growth in silica glass. J Non-Cryst Solids. 2003;16:174–82. Sullivan JD, Lauzon PH. Experimental probability estimators for Weibull plots. J Mater Sci Lett. 1986;5:1245–7. Tokita M, Niki R, Hikichi K. Percolation theory and elastic modulus of gel. J Phys Soc Jpn. 1984;53 (2):480–2. West JK, Nicles R, Latorre G. Correlations between processing parameters. Ultrastructure and strength in gel-silica. In: Brinker CJ, Clark DE, Ulrich DR, editors. Materials research society symposia processing, vol. 121. Pittsburgh: Materials Research Society; 1988. p. 219–24. Woignier T, Phalippou J. Mechanical strength of silica aerogels. J Non-Cryst Solids. 1988;100:404–8. Woignier T, Phalippou J, Sempere R, Pelous J. Analysis of the elastic behavior of silica aerogels taken as a percolative system. J Phys Fr. 1988;49:289–93. Woignier T, Phalippou J, Prassas M. Glasses from aerogels. J Mater Sci. 1990;25:3118–26. Woignier T, Phalippou J, Hdach H, Larnac G, Pernot F, Scherer GW. Evolution of mechanical properties during the alcogel-aerogel–glass process. J Non-Cryst Solids. 1992;147–148:672–80. Woignier T, Scherer GW, Alaoui A. Stress in aerogel during depressurization of autoclave: II silica gels. J Sol-Gel Sci Technol. 1994;3:141–50. Woignier T, Alaoui A, Primera J, Phalippou J, Scherer GW. Mechanical properties of aerogels: elastic or plastic materials? Key Eng Mater. 2009;391:27–44. Woignier T, Alaoui A, Primera J, Scherer GW. Structural effect on the plastic behavior in highly porous glasses. Key Eng Mater. 2010;423:15–24. Woignier T, Calas S, Reynes J. From nano composites aerogels to glass ceramics. Solid State Phenom. 2011;172–173:791–6. Woignier T, Primera J, Alaoui A, Etienne P, Despestis F, Calas-Etienne S. Mechanical properties and brittle behavior of silica aerogels. Gels. 2015;1(2):256–75. Zarzycki J. Critical stress intensity factors of wet gels. J Non-Cryst Solids. 1988;100:359–63.

Characterization of the Mechanical Properties of Sol–Gel Coatings Michel A. Aegerter

Abstract

This chapter is dedicated to the characterization of the mechanical properties of coatings obtained by the sol–gel process taken in a broad sense, including layers prepared by conventional processes, organic–inorganic or hybrid layers, Ormosil or ® ® Ormocer coatings, and nanocomposite layers, Nanomer coatings. In addition, a brief summary of the most often used techniques to gather data on stress, hardness, fracture toughness, adhesion, and abrasion (rubbing test and tuber test) is presented.

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Some Results on Sol–Gel Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pencil Hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fracture Toughness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adhesion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2 3 4 8 8 10 11 11 13 13 13 15

M.A. Aegerter (*) Former Director at the Institute for New Materials (INM), Saarbruecken, Germany e-mail: [email protected] # Springer International Publishing AG 2017 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_44-1

1

2

M.A. Aegerter

Rubbing Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Taber Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sand Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15 16 16 17 17

Introduction Coatings with thickness from several nanometers to several micrometers play today a very important role for the fabrication of many products. The mechanical properties of such coatings are therefore of utmost importance, since their functionalities may be highly disturbed or even lost if the coatings are deteriorated or fail. The mechanical properties are strongly dependent on their structure, microstructure, chemical composition, plus the incorporation of impurities and are therefore influenced by the production technology and the parameters chosen. The most relevant mechanical properties are stress, residual stress, hardness, elastic modulus, fracture toughness, abrasive resistance, and adhesion (interfacial fracture toughness). To obtain data on these quantities, several techniques and related equipments, some of them highly sophisticated, are today available. Several overviews (not necessarily on sol–gel coatings) on such properties have been published, and the reader could consult them to obtain more detailed information (Ohring 1992; Pharr and Oliver 1992; Pulker 1999; Bhushan 1999; Malzbender et al. 2002; Strauss 2003; Bange 2004; Oliver and Pharr 2004).

Stress Thin films are generally in a state of elastic mechanical stress during and after deposition. This property is very important with respect to their durability, stability, and usability. The total stress is a property determined by the material, the deposition, and the annealing processes and can be tensile or compressive. A tensile stress (positive sign) arises when a film contracts parallel to the substrate, while a compressive stress (negative sign) occurs when the film expands parallel to its surface. The total value consists usually of three components σ f ¼ σ ext þ σ therm þ σ int

(1)

The external stress (σ ext) arises from applied external forces; the thermal stress (σ therm) arises from the difference between the thermal expansion coefficient of the film (αf) and substrate (αs) materials, which, when it is small, can be calculated as σ therm ¼

Ef ðαf  αs Þ ΔT 1  νf

(2)

Characterization of the Mechanical Properties of Sol–Gel Coatings

3

where Ef and νf are the Young’s modulus and Poisson’s ratio of the film, respectively. The intrinsic stress (σ int) is usually the dominant part and is related to the film itself, i.e., a property highly dependent on the microstructure, the thickness, and the presence of defects or impurities in the coatings. It is given as σ int ¼

Ef  e ¼ Mf e 1  νf

(3)

where e = Δl/l (relative change in length) is the strain of the film and Mf = Ef/(1  vf) is the biaxial modulus of the film material. Several models have been proposed to explain the origin of the intrinsic film stresses (see, e.g., Strauss 2003 and enclosed references), but none, to our knowledge, has been used to characterize sol–gel coatings.

Measurements The forces originating from film stress are superimposed on the whole film/substrate system and bend the substrate elastically. The resulting change in the substrate curvature depends on its geometrical shape; two types are especially interesting and widely used: the beam (or cantilever, a long narrow strip) and the disk (see (a), below). Additionally, the strain of the film due to the interaction with the substrate also changes the strength of the interatomic bonds in the film material and the lattice parameters of crystalline materials. These changes in the film and/or substrate properties can also be used for stress determination, using Raman spectroscopy (see (b), below) and X-ray diffraction (XRD) (see (c), below). However, the most popular technique is the bending-substrate method (beam or disk). (a) A disk will adopt the shape of a spherical cap, if the film stress and the elastic properties of the round-shaped substrate are isotropic. As the film thickness, d, is much smaller than the substrate thickness, ts, a relationship between the biaxial stress, σ f, in the film and the resulting radius of curvature, Rs, of the film/ substrate system can be derived by a simple biaxial bending formalism (Stoney’s equation)
 2 Es ts t2 ¼ Ms s (4) σf ¼ 1  νs 6Rs d 6Rs d where Ms. = Es/(1  vs) is the biaxial modulus of the substrate material. Ms is generally well known or easy to measure, and no information about the elastic properties of the film material is required in this relation. If the stress varies along the film depth, then σ f represents an average film stress. For coatings consisting of several layers, σ f is the thickness-weighted average of the stresses in all layers, as long as the total coating thickness is still small compared to ts. For the bending of a beam-shaped substrate, a similar relationship can be derived, with the same dependencies on the film and substrate properties.

4

M.A. Aegerter

Several techniques are used for the detection of the substrate bending and the evaluation of its radius of curvature. They consist either of mechanical pickup systems or optical and electrical pickup systems, which are advantageous because they combine touchless measurements with high sensitivity and enable in situ investigations of film stresses. Therefore, these techniques are widely used for detecting capacitive changes or deviations in the direction of reflected laser beams or for evaluating surface topographies by interference measurements. (b) Changes induced in the lattice potential energy of crystalline films by pressure or temperature are manifested by variations of vibrational frequencies, resulting from lattice expansion or contraction. The temperature-dependent variations in the vibrational frequencies of thin films often differ from those observed in the respective crystalline bulk material. Changes in the microstructure of the film also influence the intrinsic vibrational frequencies under an applied stress. Raman spectroscopy is the most adequate investigation, as it possesses the sensitivity and dynamics to measure these influences on vibrational frequencies. But extensive temperature-dependent and pressure-dependent measurements on bulk materials are needed in order to calibrate the observed frequency shifts. (c) The strain in thin crystalline films can also be detected by X-ray diffraction. A deviation of the lattice parameter from the respective bulk value, a0, establishes the strain. The stress is then calculated from the elastic constants of the film and the geometry of the experiment. For example, the usual diffractometer geometry is widely employed to measure the spacing of planes parallel to the substrate. The stress can be calculated from
 Ef a0  a σf ¼ (5) a0 1  νf where Ef and νf are the Young’s modulus and the Poisson’s ratio of the film material and a0 and a are the unstrained and strained lattice parameters. The X-ray technique is also capable of measuring anisotropic and triaxial stress distributions in crystalline thin films.

Some Results on Sol–Gel Coatings Although the determination of the stress in coatings is an important task, the sol–gel literature is very scarce. Sol–gel coatings for optical applications are often amorphous and have thicknesses between 40 and 1000 nm. In this case, the stress determination by XRD or Raman spectroscopy is quite difficult. The investigation of thick layers is easier, but it has to be taken into account that for oxide layers, the stress depends on the total film thickness. Therefore, bending-substrate techniques, combined with interference-optical measurements of the substrate curvature, are superior to other techniques.

Characterization of the Mechanical Properties of Sol–Gel Coatings

5

Fig. 1 Typical results for substrate bending (fused silica, 1 mm thick, 48 mm in diameter), due to (a) compressive stress and (b) tensile stress on a 100 nm thick TiO2 film (From Bange 2004)

Figure 1 gives typical results for TiO2 films (disk shape) under tensile and compressive stresses. The film stress can be determined according to (4). The uncertainties in stress determination, which are derived from reproducibility and stability studies, are smaller than 10 MPa for a 100 nm thick film on fused silica substrates with a thickness of 1 mm (Ottermann et al. 1993). The stress of a sol–gel layer may not stay constant in time. Reorientations, diffusion processes in the film material, and water absorption from the ambient humidity will cause a relaxation of the stress. Sol–gel films deposited on a substrate usually exhibit an interaction stress which is inherent to the film/glass system and exhibit remarkably different stresses, depending upon the coating technique, the respective deposition parameters, and the film thickness. The annealing of amorphous layers also causes variations in the film stress due to changes in morphology (Exarhos and Hees 1992). Stress is also strongly related to film density (Ottermann and Bange 1996). TiO2 films with densities far lower than that of crystalline anatase are under tensile stress. The mean distances between the atoms are believed to be the origin of this phenomenon. Films with densities equal to or higher than that of their corresponding crystalline phase possess repulsive forces, causing compressive stress. A tensile force is therefore usually obtained in sol–gel films of lower density, showing attractive interactions between the atoms. The tensile stress will increase with decreasing density. But a stronger decrease in film density leads to a more porous film structure, with reduced strength. Some typical results for stress in TiO2 and SiO2 films deposited by spin coating on glass, with a thickness of approximately 100 nm, are summarized in Fig. 2. For TiO2 films, the value of the tensile stress ranged between approximately 150 and 250 MPa, while SiO2 films exhibited values between about 100 and 300 MPa. Both stress regions suggest low film densities.

6

M.A. Aegerter

Fig. 2 Stress of 100 nm thick SiO2 and TiO2 films deposited by spin coating. Ranges of stress variation with respect to different production conditions are depicted by the gray bars. The open bars represent typical changes in stress for distinct samples, within 1 month after production; the directions of the changes are indicated by arrows (From Bange 2004)

Fig. 3 Variation of stress as a function of storage time after production, for SiO2 and TiO2 layers deposited by spin coating, as a function of the relative humidity (r.h.) of the environment (From Bange 2004)

Results on the relaxation of film stress with time after production are also depicted in Fig. 2. It is obvious that the films possess a tendency to decrease their stress after the deposition. The changes in stress as a function of time are different for SiO2 and TiO2 films (Bange 2004). This is illustrated in Fig. 3. While for TiO2 films, the relaxation effect is almost independent of the relative humidity (r.h.) of the surrounding atmosphere, SiO2 layers exhibit a strong dependence on the relative humidity. For crystalline TiO2 films, the relaxation effect is not as strong as in SiO2 films. Several processing parameters affect the level of stresses in sol–gel films. One of the most important is the sintering temperature and has been studied in some detail only recently for SiO2 and TiO2 coatings (Kozuka et al. 2003). The total stress was found to be tensile and to increase with the heat-treatment temperature (Fig. 4). For SiO2 coatings, the residual stress measured after sintering at 600  C was also

Characterization of the Mechanical Properties of Sol–Gel Coatings

7

Fig. 4 Evolution of the residual stress and thickness with the sintering temperature for SiO2 and TiO2 films deposited by spin coating on silica glass substrates (From Kozuka et al. 2003)

strongly affected by the H2O/TEOS ratio, varying from 120  14 MPa (ratio 2), to 290  10 MPa (ratio 4), or 350  20 MPa (ratio 8). The addition of chelating agents in the titania sol also leads to an increase of the residual stress (Ts = 700  C), from 340  70 MPa (no agent) to 490  60 MPa (acetic acid), to 730  70 MPa (acetylacetone), or to 1090  70 MPa (diethanolamine). Too high a stress leads to the formation of cracks in the coatings (see also section “Fracture Toughness”). In order to determine at which heat-treatment steps such cracking occurs, their formation was observed in situ up to 800  C (Kozuka et al. 2003). Cracking was found to occur during the heating up stage, between 100  C and 400  C, depending on the heating rate and the film thickness, thicker films having a tendency to crack at lower temperature, while the onset temperature was found to increase with the heating rate. Cracking appears therefore to be due to the development of intrinsic tensile stresses during the heating process and not to a thermal stress effect. The strategy to suppress the building up of a too high tensile stress is therefore to promote structural relaxation or plastic flow during the coating processing. One successful approach is to add an organic polymer, such as polyvinylpyrrolidone (PVP), to the sols. This allowed to drastically increase the initial thickness for crack formation (Kozuka et al. 2000, 2003). The tensile stress was reduced, for BaTiO3  coatings sintered at 700 about 650 MPa down to about 20 MPa, by  C, ifrom  increasing the PVP=Ti OC3 H7 4 ratio up to 1, with a sharp variation for a ratio of 0.6. It was suggested that PVP retards the condensation reaction and promotes structural relaxation. The atmosphere composition during sintering also seems to play a role. The tensile stress development of ZrO2 (Brenier 2002) was found to rise to higher values when the films were annealed in ozone-enriched oxygen, instead of pure oxygen.

8

M.A. Aegerter

The presence of ozone was found to enhance the rate of organic compound transformation and oxidation and to deccelerate the OH condensation reaction responsible for the ZrO2 network formation. ZnO coatings (Brenier and Ortéga 2004) annealed in pure oxygen, on the other hand, were found in a compressive stress state due to a stuffing effect during grain growth, whereas the same films, when annealed in an ozone-enriched oxygen atmosphere, had a tensile stress state due to pore collapsing. This astonishing difference has been attributed to the building of a thinner skeleton in the latter film compared to the former ones, the acidifying action of ozone on alcohols being thought to be responsible for slowing down grain growth during annealing. Tensile and compressive moisture-induced stress changes during film exposure to ambient air have also been observed for ZrO2 coatings annealed up to 400  C (Brenier 2000). Organic–inorganic coatings find every day more applications, as they have better performances than organic polymers and can be processed at much lower temperatures than inorganic sol–gel coatings. Very few works have been dedicated to the study of stress in such coatings, using the curvature-induced method. The tensile residual stress was found to increase from 40 to 110 MPa, as the curing temperature increased monotonically from 150 C up to 350  C (for coatings derived from 80% TEOS and 20% GPTMS (glycidoxypropyltrimethoxy silane)) and the critical thickness for cracking was found at about 2 μm.

Hardness Hardness is an important material property of concern in films utilized for optical, electronic, and mechanical functions. It affects also the wear resistance and plays an important role in the friction and lubrication of film surfaces in contact. It is a complex property, related to the strength of interatomic forces and complicated interactions caused by deformation mechanisms in the material. This property, therefore, depends on many variables. Hardness is usually defined as the resistance of the material to local plastic deformation. It is not a fundamental property of materials, but it is related to material properties, in particular the yield strength and the elastic modulus.

Measurements Indentation testing, using a sharp diamond indenter, at the nanometer scale, has become one of the most widely used techniques (Pharr and Oliver 1992; Malzbender et al. 2002; Oliver and Pharr 2004). The material surface is indented by a tip load with a force P, resulting in a penetration depth, h, of the indenter. Both parameters are recorded as a function of time. The experiment usually consists of a single loading–unloading cycle. As the specimen is loaded to a maximum force Pmax, the indentation depth increases to a maximum, hmax. If plastic deformation occurs, a

Characterization of the Mechanical Properties of Sol–Gel Coatings

9

different curve is obtained during the unloading cycle, and the final depth is some finite value hf 6¼ 0. Several types of indenters can be used (pyramid, sphere, cone), but, for coatings, sharp indenters are preferred. The influence of indenter geometries on sol–gel coatings was discussed by Malzbender et al. (2000) and Oliver and Pharr (2004). The most used is the Berkovich indenter, a three-sided pyramidal shape with a tip radius of the order of 50–100 nm. Vickers indenters (four-sided pyramids) are less popular. Figure 5 shows a typical indentation load–displacement curve. The key measured quantities are Pmax, hmax, the initial unloading contact stiffness S = dP/dh, and the displacement found by linearly extrapolating the initial portion of the unloading curve to zero load, hc 6¼ 0. For axis-symmetric indenters, the following parameters are obtained: S¼

pffiffiffi dP 2 ¼ pffiffiffi Er A dh π

(6)

where A is the projected area of contact between the indenter and the material and Er is the reduced modulus, which accounts for the fact that the measured elastic displacement includes contributions from both the specimen and the indenter     1  v2f 1  v2i 1 ¼ þ Er Ef Ei

(7)

where Ef and νf are the elastic modulus and Poisson’s ratio of the film, respectively, and Ei and νi are the same quantities for the indenter (for diamond, Ei = 1.141 GPa, vf = 0.07). If the geometry of the indenter is known, the relationship between A and hc is known, and the problem is reduced to determining hc from the measurement hc ¼ hmax  e

Pmax Smax

(8)

For a perfect, sharp Berkovich indenter, Aðhc Þ ¼ 24:5h2c . For real indenters, this function has to be determined by carrying out measurements on a reference material, usually fused silica. From the relations (6) and (7) and a knowledge of νf, the elastic modulus, Ef, can be determined. The hardness is given by H¼

Pmax A

(9)

It is worth to mention that during such measurements, the combined responses of the coating and the substrate are measured. Several relationships have been proposed for modeling this effect (see Pharr and Oliver 1992; Fischer-Cripps 2001; Malzbender et al. 2002; Oliver and Pharr 2004). The contribution of the substrate becomes important when the indentation depth exceeds 10–25% of the film thickness and, therefore, may become critical when very thin films are measured.

10

M.A. Aegerter

Fig. 5 Schematic representation of a typical indentation load displacement curve (from Malzbender et al. 2002)

Pmax

P

Loading curve

Unloading curve

S Possible range for hc

hc for ε = 1

hmax

h

hc for ε = 0.72

Results Many papers appeared in the sol–gel literature, but most of them are related to the measurements of hard inorganic–organic coatings (Ormocer ®, Nanomer ®). As already reported (Mackenzie and Bescher 2000), the hardness of oxide coatings (as well as other physical properties) is inadequate, unless the heattreatment temperatures are in excess of about 400  C. The hardness of a silica coating sintered at 620  C was improved by addition of a small amount of TiO2 (Akamatsu et al. 2000). The increase in the hardness of SiO2-based optical coatings sintered at 500  C, based on GLYMO, MPES, and other silanes, was also found to increase with the titania content, and the influence of carbon contents up to 6.6 GPa was also noted (Que and Huy 2003). The hardness was also found to increase with the heat-treatment temperature in MAPS coatings, but a decrease was observed with increasing H2O/MAPS molar ratio (Gunji et al 2001). The hardness of an aluminum alloy (ADC12) was drastically improved by coating with SiO2, ZrO2, or SiO2–ZrO2 layers containing 1 wt% poly(vinyl butyral) (Tsuge et al. 2001). On the other hand, hybrid coatings processed at low temperatures (1 H (Al-Dahoudi and Aegerter 2002), and for antireflex sol–gel coatings, with values of 7H–8H (Peeters and Bohner 2003).

Fracture Toughness The fracture toughness can also be evaluated using indentation processes (Malzbender et al. 2002). For sufficiently large indentation loads, brittle materials exhibit cracking during indentation (see section “Hardness”). Radial mechanical cracks emerging from the edges of the indenter and lateral cracks are common features for pyramidal indentation. The initiation and subsequent growth of the cracks is determined by the elastic and plastic properties and by the residual stress present in the coatings. The critical load, P*, at which the radial crack initiates is given by

12

M.A. Aegerter

P ¼ A 

K 4IC H3

(10)

where KIC is the fracture toughness, H is the hardness, and A is a constant dependent upon the indenter. Another equation is also used: K IC ¼ χr

P þ Zσ r c1=2 c3=2

(11)

where c is the crack length and Z is a crack shape factor, difficult to measure, but usually taken as equal to 1.26, χr = 0.16(E/H )l/2 for a Berkovich or Vickers indenter and σ r is the residual stress. KIC and σ r can be therefore determined by plotting χr P/ c3/2 versus Zc1/2. Using a concept proposed by Griffith, the fracture toughness can be transformed into the fracture energy Γ Γ¼

K 2IC E

(12)

Such model was developed for monolithic materials, and a suitable extension to coatings on substrates is not yet available, but it can nevertheless be applied, if the stress intensity is not influenced by the substrate, i.e., if the radial cracks are confined to the surface of the coating. Other models are discussed in Malzbender et al. (2002). Another way to assess the residual stress in thin films is to utilize their critical thickness, tc. If the thickness of the coating exceeds this particular value, the coating fractures during the deposition or the thermal treatment, due to an excess of tensile intrinsic or thermal stresses. Several other failure modes may be present, such as surface cracks, channeling, spalling, debonding, etc., which of these actually occurs depends on the relative fracture toughness of the coating, interface, and substrate. The critical thickness, tc, of a particular coating–substrate system, can be determined by: tc ¼

I Ecf Γcf λ σ 2r

(13)

where λ is a configuration-dependent dimensionless cracking number and Γcf and Ecf are the relevant fracture energies and elastic moduli, respectively, of the coating, interface, or substrate. This expression gives only a lower bound for cracking and expresses that failurecannot occur, as long as the elastic strain energy stored in the  coating per unit area σ 2r t=2Ec does not exceed the fracture energy Γ, multiplied by the factor λ/2. The value of λ depends on the failure mode and on the elastic moduli of the coating and substrate; it is 3.39 for surface cracks and substrate damage, 1.98 for channels, 0.34 for spalling, and 0.5 for steady-state debonding (Malzbender et al. 2002).

Characterization of the Mechanical Properties of Sol–Gel Coatings

13

Results The interfacial toughness of sol–gel alumina coatings deposited on sand-blasted steel substrates was found to increase with the process temperature (Hawthorne et al. 2004). In ZrO2–SiO2 coatings deposited on soda lime silicate glass slides, the ratio of the Vickers hardness to the toughness (HV/KIC) increased as the amount of ZrO2 decreased (Garcia-Heras et al. 2003). Sol–gel coatings have also been developed to coat powder, fibers, and whiskers, in order to vary the interfacial strength between them and different materials. For example, carbide powder (TiC, (W,Ti)C), coated with an alumina ceramic sol–gel coating and then sintered by hot pressing to obtain a ceramic tool to machine mild steel, was found to have an improved fracture toughness by up to 33% and superior ability in maintaining the wear resistance (Huang et al. 2000). A BaZrO3 coating applied to Al2O3 fibers, used to fabricate alumina fiber/alumina matrix composites, was also found to improve the toughness of such systems (Chen et al. 2002). A TiO2 coating on SiC whiskers, imbedded in a Si3N4 matrix, also increased the interfacial strength and fracture toughness, but MgO had the opposite behavior (Kato and Goto 1997).

Adhesion Adhesion of coatings is difficult to assess. It is defined by ASTM as the “condition in which two surfaces (e.g. in our case, a coating and a substrate) are held together either by valence forces, or by mechanical anchoring, or by both together” (Pulker 1999). From thermodynamics, it is the work required to separate a unit area of two phases forming an interface. Using a fracture mechanics approach, the adhesion energy of a coating/substrate system may be defined as the energy density needed to propagate a crack along the interface. However, in practical systems, it is not always clear whether the failure is truly interfacial or is fully or partially cohesive in its nature. Many techniques have been developed to determine the degree of adhesion, and they are classified according to various criteria.

Measurements (a) Indentation technique: If interfacial cracking (delamination) occurs during indentation, it is possible to extract information about the fracture energy from such measurements (see section “Fracture Toughness,” as well as den Toonder et al. 2002; Xie and Hawthome 2003). (b) Scratch test: Another very often used technique is scratch testing. In this case, a well-defined tip is drawn over the surface of the coating while applying a normal load, P. This load is usually increased during the displacement of the coating at a

14

M.A. Aegerter

given velocity. Lateral force transducers are used to measure the lateral force acting on the scratch tip, allowing the determination of the friction coefficient μ = μA + μP, where μA and μP are the adhesion and plowing friction coefficients. At a critical load, Pc, a well-defined failure event occurs, which can be either visually observed or can be determined by an acoustic detector. In modern systems, the load–displacement characteristic is recorded automatically. The critical load at which failure occurs can be used as a qualitative measure of the coating–substrate adhesion (Malzbender et al. 2002). By analyzing images of the cracks at failure, one can usually distinguish the different types of failure (cohesive, interfacial, conformal, chipping). Moreover, the elastic modulus can also be estimated. However, the value obtained is influenced by both the coating and the substrate properties. Various models have been developed in order to determine the fracture energy of the interface from scratch testing, but they usually give only a rough estimation. A detailed discussion about models proposed by Burnett et al., Bull et al., Atter et al., and Thouless can be found in Blees et al. (2000) and Malzbender et al. (2002). They all basically assume that failure occurs due to chipping in front of the indenter and that the elastic strain energy stored in the coating above the chipped area, just before chipping, is decreased by interfacial fracture at the critical load. However, fracture may also occur in the coating, so that the values determined are only qualitative. In hybrid coatings deposited on plastic substrates, the critical load for which failure occurred was found to pronouncedly decrease when the friction coefficient increased. Microwave oxygen plasma modification of polypropylene substrates was found to favor the adhesion more than wet-chemical modification (Blees et al 2000). Hybrid sol–gel coatings deposited on brass were also found to be the best compromise in terms of tarnishing, corrosion, and scratching resistance, when compared to hard coatings (TiN, TiZrN, or acrylic varnish) (Dumont et al. 2001). The adhesion measured by the scratch testing of sol–gel coatings deposited on a glass substrate is markedly improved by the heat-treatment temperature (Katayama et al. 1992). α-Al2O3 layers were found to improve by a factor of 50 the damage resistance of uncoated glass surfaces (Hauk et al. 1999). When deposited on Al alloys, it was found that a rapid degradation in the wear resistance arose when the indentation depth was larger than 20% of the coating thickness (Wilson et al. 2000). On the other hand, little correlation was found between coating processing and microstructure on the scratch hardness in phosphate-bonded sol–gel composite alumina coatings deposited on steel substrates (Hawthorne et al. 2004). Inorganic–organic composite materials, containing cross-linked inorganic nanoscaled particles (e.g., boehmite), usually show high scratch resistance. Two-step catalysis and ammonia/water vapor treatment also improve the mechanical strength of SiO2 nanoporous networks developed as antireflective coatings (Wu et al. 2003), as well as of SnO2 coatings. Many works also proposed special procedures to be applied to the substrates in order to improve the adhesion of sol–gel coatings. They involve, e.g., the deposition of a primer solution for polymeric substrates (Li and Wilkes 1998) or oxygen plasma

Characterization of the Mechanical Properties of Sol–Gel Coatings

15

treatment (Gilberts et al. 1998) plus the deposition of TiO2, calcium titanate, or silica, to improve the adhesion of hydroxy-apatite on metal, for orthopedic or dental prosthesis (Kaciulis et al. 1998). (c) Scotch tape test: The adhesion of coatings can be easily tested with a cheap and fast test, using a pressure-sensitive cellophane or cellulose acetate adhesive, pressed onto the surface of the coating. It is a standard test developed especially for optical components (DIN 58196a-Part 6, L-T-90 (USA)). The tape is firmly pressed, at room temperature, against the coated substrate and is then quickly removed at normal angle in about 1 s (K2) or 2–3 s (K1). After the removal of the tape, the coated surface is evaluated, preferentially by visual reflection under white light, for evidence of coating removal. This simple test does not give any scientific information but is helpful to screen the development of coatings for industrial use. Although very simple, this test has rarely been reported in the sol–gel scientific literature (Guglielmi et al. 1992; Winkler et al. 1999; Al-Dahoudi and Aegerter 2002). (d) Crosscut test: This test is a simple empirical technique to analyze the adhesion of simple and multilayer coatings on substrates, as well as between the coatings (DIN 53151, ASTM D 3359). A cutting tool, having one to six special blades with a separation of 1 or 2 mm, is scanned on the coatings to form a grid. The visual inspection is graded from Gt0 (clean cut without removal of the coating) to Gt5 (65% of the coating removed). Equipments are available commercially. This procedure was also rarely used to test sol–gel coatings. Excellent results (Gt0–Gt1) have been obtained for nanocomposite coatings (Winkler et al. 1999; Langenfeld et al. 1998), nanoparticulate coatings developed for interference films (Mennig et al. 1999b), and ITO nanocomposite coatings on plastic substrates (Aegerter and Al-Dahoudi 2003).

Abrasion Several techniques have been developed in order to measure the resistance of coatings against abrasion.

Rubbing Test Rubbing Tests with a Cotton Cloth or a Hard Rubber A simple test to measure the resistance against abrasion of optical coatings is described in the DIN norm 58196-5. It consists of rubbing on the coating a flat stamp of 10 mm diameter covered with a four-ply cotton cloth (DIN 61631 Part 1), with a normal force of 4.5 N, along a minimum length of 20 mm. The number of cycles is 25 or 50. The coating is then observed under white light, in reflection, and transmission. The result is graded from H25-1 or H50-1, no damage, to H25-5 or H50-5, coating fully removed from the substrate.

16

M.A. Aegerter

A similar but much more severe test is to rub a hard rubber eraser (Shore-Ahardness of 75) of 7 mm diameter, during 10 or 20 complete cycles, with a normal force of 10 N, over a distance of 20 mm. The physical damage is then observed visually (flaking, feeling, cracking, or blistering) and graded from G10-1 or G20-1, no damage, to G10-5 or G20-5, coating fully removed from the substrate. A similar process is given in the MIL-C-675C norm. Commercial equipments exist. To our knowledge, the use of this simple test was only reported to characterize ITO nanocomposite (ITO + MPTS) coatings, deposited on plastic substrates, and cured under UV light (Aegerter and Al-Dahoudi 2003). Graded H-25-class 1 and G10-class 1 as well as interference Nanomer coatings graded with the value G20-class 1 (Mennig et al. 1999a). Similar tests have been realized to confirm the good scratch resistance of broadband antireflective coatings made of Ta and Si oxide-based layers, for amplifier blast shields of the French LIL laser (Prene et al. 2000).

Taber Test This simple procedure (available commercially) uses special rotating abrasion wheels (Ø 45–50 mm, 12.5 mm wide, with a well-defined fine abrasion medium imbedded in a rubber), to proof the behavior of thick coatings. The results are mainly analyzed by measuring the light scattered by the treated coating (haze) (DIN 52347, ISO 3537-1875). The probe is placed on a rotating plate and is submitted to abrasion by two rotating wheels, which exert a force of 5.4 or 2.7 N on the sample. The haze of the samples is analyzed, typically after 100 revolutions (or more) of the table, according to ISO 3537-1975. This procedure was used to test thick, hard inorganic–organic coatings, designed to protect non-ferrous alloys (Langenfeld et al. 1998), organic polymers (Sepeur et al. 1999; Gilberts et al. 1998), multilayer interference thin hybrid coatings (Mennig et al. 1999a, b), as well as coatings made with Al2O3, SiO2, or ZrO2 nanoparticles whose surface was modified by epoxy functionalized alkoxy silanes. Such coatings showed only a 2% loss of transmission after 1000 cycles (Winkler et al. 1999; Becker-Willinger 2004).

Sand Test One sand test is used to determine the resistance of transparent coatings to the abrasion of sand (ASTM F 755-81). It consists of measuring and recording the haze and light transmission of samples placed on the bottom of a tray (sand cradle), covering the specimen with an abrading medium, and subjecting the cradle to a specific number of to-and-fro motions. After exposure to the abrasion, the haze and light transmission are determined. Another test (DIN 52348) measures the haze after letting fall 3 kg of 0.5/0.71 mm grain size sand on a rotating specimen, placed at an angle of 45 below the tube. The falling height is 165 cm. Such tests have been used to compare the behavior of hard Nanomer ® coatings, containing boehmite nanoparticles, deposited on plastic

Characterization of the Mechanical Properties of Sol–Gel Coatings

17

substrates, with uncoated glass and plastic substrates (Becker-Willinger 2004). The appearance of the coated substrates was found to be better than that of an automobile front windshield glass.

Conclusion This chapter was dedicated to an overview of the results obtained on the determination of the mechanical properties of sol–gel coatings. It gave as well a brief summary of the most often used techniques to gather data on stress, hardness, fracture toughness, adhesion, and abrasion. Most of the published data referred to just single measurements, and very few systematic works have been so far realized, in order to characterize the influence of some physical and chemical process parameters on such properties. The use of two or three different techniques to assess and compare the same property is even less frequent. This reflects without any doubt the very large diversity of composition and process parameters which are used today in the sol–gel field in order to obtain coatings with functional properties, particularly with hybrid and nanocomposite compositions, which render such an evaluation very difficult.

References Aegerter MA, Al-Dahoudi N. Wet-chemical processing of transparent and antiglare conducting ITO coating on plastic substrates. J Sol–Gel Sci Technol. 2003;27:81–9. Akamatsu Y, Makita K, Inaka H, Minami T. Effect of the addition of TiO2 to SiO2 films with concave–convex surface derived by the sol–gel method on film hardness and application to water-repellent glasses for automotive windows. J Cerma Soc Jpn. 2000;108:365–9. Al-Dahoudi N, Aegerter MA. Conducting, antistatic and antistatic–antiglare coatings made with hybrid sols. Mol Cryst Liquid Cryst. 2002;374:91–100. Bange K. Characterization techniques applied to sol–gel derived coatings and products. In: Aegerter MA, Mennig M, editors. Sol–Gel techniques for glass producers and users. Boston: Kluwer Academic Publishers; 2004. Becker-Willinger C. Private communication. 2004. Bhushan B. Handbook of micro/nano tribology. 2nd ed. Boca Raton: CRC Press; 1999. Blees MH, Winkelman GB, Balkenende AR, den Toonder JMJ. The effect of friction on cratch adhesion testing: application to a sol–gel coating on polypropylene. Thin Solid Films. 2000;359:1–13. Brenier R. Stress and moisture – sorption in ozone – annealed filmes of zirconium oxide obtained from sol–gel. J Sol–Gel Sci Technol. 2002;25:57–63. Brenier R, Ortéga L. Structural properties and stress in ZnO Films obtained from a nanocolloidal sol. J Sol–Gel Sci Technol. 2004;29:137–145. Chen ZC, Duncan S, Chawla KK, Koopman M, Janowski GM. Characterization of interfacial reaction products in alumina fiber/barium zirconate coating/alumina matrix composite. Mater Charact. 2002;48:305–14. den Toonder J, Malzbender H, de With G, Balkenende AR. Fracture toughness and adhesion energy of sol–gel coatings on glass. J Mater Res. 2002;17:224–33. Dumont B, Thiery R, Welter JM, Duterne JP. Hybrid sol–gel clear coatings for decoration brass profiles. Surf Eng. 2001;17:254–8.

18

M.A. Aegerter

Exarkos GJ, Hess NJ. Spectroscopic measurements of stress relaxation during thermally induced crystallization of amorphous titania films. Thin Solid Films. 1992;220:254. Fischer-Cripps AC. A review of analyzing methods for sub-micron indentation testing. Vacuum. 2001;58:569–85. Frings S, Meinema H, van Nostrum C, van der Linde R. Organic–inorganic hybrid coatings for coil coatings application based on polyesters and tetraethoxysilane. Progr Inorg Coat. 1998;33:126–30. Garcia-Heras M, Rincon JM, Romero M, Villegas MA. Indentation properties of ZrO2–SiO2 coatings on glass substrates. Mater Res Bull. 2003;38:1635–44. Gilberts J, Tinnemans AHA, Hoberheide MP, Koster TPM. UV curable hard transparent hybrid coating materials on polycarbonate prepared by the sol–gel method. J Sol–Gel Sci Technol. 1998;11:153–9. Guglielmi M, Testa D, Innocenzi PC, Colombo P, Gobbin M. Borosilicate coatings on mild steel. J Non Cryst Solids. 1992;147:474–7. Gunji T, Makabey Y, Takamara N, Abe Y. Preparation and characterization of organic–inorganic hybrids and coating films from 3-methacryloxypropylpolysilsesquioxane. Appl Organomet Chem. 2001;15:683–92. Gupta N, Sinha TJM, Varma IK. Development of an abrasion resistant coating from organic–inorganic polymeric network by sol–gel process. Indian J Chem Technol. 1997;4:130–4. Hauk R, Frischat GH, Ruppert K. Sol–gel preparation of scratch-resistant A12O3 coatings on float glass. Glass Sci Technol. 1999;72:386–293. Hawthorne HM, Neville A, Troczynski T, Hu X, Thammachart M, Xie Y, Fu J, Yang Q. Characterization of chemically bonded composite sol–gel based alumina coatings on steel substrates. Surf Coat Technol. 2004;176:243–52. Huang CZ, Wang J, Ai X. Development of new ceramic cutting tools with alumina coated carbide powders. Int J Machine Tools Manuf. 2000;40:826–32. Kaciulis S, Mattogno G, Napoli A, Bemporad E, Ferrari F, Montenero A, Gnappi G. Surface analysis of biocompatible coatings on titanium. J Electron Spectrosc Relat Phenomena. 1998;95:61–9. Katayama Y, Ando E, Kawaguchi T. Characterization of SiO2-films on glass substrate by sol–gel and vacuum deposition methods. J Non Cryst Solids. 1992;147:437–41. Kato M, Goto Y. Mechanical properties of Si3N4 matrix composites reinforced with SiC whiskers with oxide coatings. Adv Compos Mater. 1997;6:227–37. Kozuka H, Kajimara M, Hirano T, Katayama K. Crack-free thick ceramic coating films via non-repetitive dip-coating using polyvinylpyrrolidone as stress-relaxing agent. J Sol–Gel Sci Technol. 2000;19:205–9. Kozuka H, Takenaka S, Tokita H, Hirano T, Higashi Y, Hamatani T. Stress and cracks in gel-derived ceramic coatings and thick film formation. J Sol–Gel Sci Technol. 2003;26:681–6. Langenfeld S, Jonschker G, Schmidt H. New sol–gel based coatings as corrosion and wear protection on non-ferrous metals. Mater Werkst. 1998;23–29 (in German). Li CH, Wilkes GL. The mechanism for 3-aminopropyltriethoxysilane to strengthen the interface of polycarbonate substrates with hybrid organic–inorganic sol–gel coatings. J Inorg Organomet Polym. 1998;8:33–45. Mackenzie JD, Bescher EP. Physical properties of sol–gel coatings. J Sol–Gel Sci Technol. 2000;19:23–9. Mackenzie JD, Bescher E. Some factors governing the coating of organic polymers by sol–gel derived hybrid materials. J Sol–Gel Sci Technol. 2003;27:7–14. Makishima A, Mackenzie JM. Hardness equation for Ormosils. J Sol–Gel Sci Technol. 2000;19:627–30. Malzbender J, de With G. Energy dissipation, fracture toughness and the indentation load–displacement curve of coated materials. Surf Coat Technol. 2000;135:60–88. Malzbender J, de With G, den Toonder JMJ. Determination of the elastic modulus and hardness of sol–gel coatings on glass: influence of indenter geometry. Thin Solid Films. 2000;372:134–43.

Characterization of the Mechanical Properties of Sol–Gel Coatings

19

Malzbender J, den Toonder JMJ, Balkenende AR, de With G. Measuring mechanical properties of coatings: a methodology applied to nano-particle-filled sol–gel coatings on glass. Mater Sci Eng. 2002;R36(2–3):47–103. Mennig M, Oliveira PW, Frantzen A, Schmidt H. Multilayer NIR reflective coatings on transparent plastic substrates from photopolymerizable nanoparticulate sols. Thin Solid Films. 1999a;351:225–9. Mennig M, Oliveira PW, Schmidt H. Interference coatings on glass based on photopolymerizable nanomer material. Thin Solid Films. 1999b;351:99–102. Nakajima A, Abe K, Hashimoto K, Watanabe T. Preparation of hard super-hydrophobic films with visible light transmission. Thin Solid Films. 2000;376:140–3. Ohring M. The materials science of thin films. San Diego: Academic; 1992. Oliver WC, Pharr GM. Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology. J Mater Res. 2004;19:3–20. Ottermann C, Bange K. Correlation between the density of TiO2 films and their properties. Thin Solid Films. 1996;296:32. Ottermann G, Otto J, Jeschkowski U, Anderson O, Henning M, Bange K. Stress of TiO2 thin films produced by different deposition techniques. MRS Proc. 1993;308:69. Peeters MPJ, Bohner MR. Optical Application of pigmented sol–gel coatings. J Sol–Gel Sci Technol. 2003;26:57–62. Pharr GM, Oliver WC. Measurement of thin-film mechanical-properties using nanoindentation. MRS Bull. 1992;17:28–33. Prene P, Pritton JJ, Beaurain L, Belleville P. Preparation of a sol–gel broadband antireflective and scratch-resistant coating for amplifier blastshields of the French laser LJL. J Sol–Gel Sci Technol. 2000;19:533–7. Pulker HK. Coatings on glass. 2nd revised ed. Amsterdam: Elsevier; 1999. Que WX, Huy X. Effects of titanium content on sol–gel hard optical films prepared in an organic–inorganic hybrid system. J Vacuum Sci Technol A. 2003;21:1809–13. Sepeur S, Kunze N, Werner B, Schmidt H. UV curable hard coatings on plastics. Thin Solid Films. 1999;351:216–9. Strauss GN. Mechanical stress in optical coatings. In: Kaiser N, Pulker HK, editors. Optical interference coatings. Berlin: Springer; 2003. Tadanaga K, Azuta K, Minami T. Preparation of inorganic–inorganic hybrid coating films from vinyltriethoxysilane–tetraethoxysilane by the sol–gel method. J Cerma Soc Jpn. 1997;105:555–8. Tsuge H, Nishi Y, Kume M, Ono S. Improvement of hardness and corrosion resistance of aluminum alloy by ceramic coating using sol–gel method. Nippon Kagaku Kaishi. 2001;12:715–720 (in Japanese). Wilson S, Hawthorne HM, Yang R, Troczynski T. Sliding and abrasive wear of composite sol–gel alumina coated Al alloy. Surf Coat Technol. 2000;133:389–96. Winkler RP, Arpac E, Schirra H, Sepeur S, Wegner I, Schmidt H. Aqueous wet coatings for transparent plastic glazing. Thin Solid Films. 1999;351:209–11. Wu GM, Shen J, Yang TH, Zhon B, Wang J. Reparation of scratch-resistant nano-porous silica films derived by sol–gel process and their antireflective properties. J Mater Sci Technol. 2003;19:299–302. Xie Y, Hawthorne HM. Measuring the adhesion of sol–gel derived coatings to a ductile substrateby an indentation-based method. Surf Coat Technol. 2003;172:42–50.

Mechanical Properties of Organic–Inorganic Hybrids Jon D. Mackenzie and Eric P. Bescher

Abstract

With the sol–gel technique, mixing organic and inorganic components has progressed beyond the simple mixing of two separate phases with different properties, as in the “polymer-filler” approach of traditional composite synthesis. Rather, “molecular” composites with unique characteristics can now be fabricated. This novel approach not only offers exciting prospects for the fabrication of novel materials, it also presents many challenges for the modelization for such complex materials. Among the many new and unique characteristics that organic–inorganic hybrids exhibit, mechanical properties are particularly important. These novel materials clearly establish a bridge between the properties of inorganic brittle oxides and those of flexible polymers. In this chapter, we review some aspects of the mechanical properties of bulk organic–inorganic hybrids, such as the relationship between structure and properties, as well as various synthesis routes.

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Pdms–SiO2 System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Young’s Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hardness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermal Stabilization of PDMS–SiO2 Hybrids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Aeromosils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sono-Ormosils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 4 5 7 8 9

J.D. Mackenzie (*) • E.P. Bescher Department of Materials Science and Engineering, University of California Los Angeles, Los Angeles, CA, USA e-mail: [email protected]; [email protected]; [email protected] # Springer International Publishing AG 2017 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_45-1

1

2

J.D. Mackenzie and E.P. Bescher

Other Ormosil Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Introduction Incorporating organic polymers or molecules within the structure of ceramics or glasses is a unique feature of the sol–gel process. Organic–inorganic hybrids have emerged recently as a novel family of materials with new and unique properties (Coltrain et al. 1996; Laine et al. 1998; Mackenzie and Ulrich 1990; Mackenzie 1992, 1994; Mackenzie and Bescher 1998). While many novel applications have been proposed, the mechanical properties of these hybrids have thus far attracted a great deal of interest. This is in part due to the growing interest in such hybrids as coatings materials (Mackenzie and Bescher 2000). Already, abrasion-resistant hybrid coatings on organic substrates have been the subject of many patents and have led to several commercial products. Most abrasion-resistant coatings used are based on organosiloxanes, such as methyl trimethoxysilane, phenyl trimethoxysilane, and glycidoxypropyltrimethoxysilane hydrolyzed in the presence of colloidal silica (Baney 1995; Habsuda 2002). It is clear that the mechanical properties and the rheology of coatings may differ significantly from those of bulk materials, because of issues such as substrate/coating interactions, for example. Therefore, a separate chapter has been devoted to the mechanical properties of organic/inorganic coatings, and the specific topic or organic/inorganic coatings will not be discussed here. We shall limit the scope of the discussion in the present chapter to the mechanical properties of bulk organic–inorganic hybrids. Ceramics are well known to exhibit high Young’s modulus, high hardness, and high temperature stability compared to polymers. However, they also have many limitations such as minimal strain at fracture, high brittleness, and, in most cases, low toughness. Many attempts have been made at improving such shortcomings, but the inherent causes of high hardness and modulus, namely the ionic and iono-covalent nature of bonding in ceramics, are also responsible for the shortcomings of brittleness. Therefore, such limitations are not easily overcome. The structure of inorganic oxides lacks the mechanisms that would allow them to deform extensively before fracture. On the other hand, many polymers exhibit large strain before failure and excellent toughness. Their elastic modulus is, however, typically very low. Heretofore, the high temperatures required by traditional ceramic processing precluded the incorporation of organic components within a ceramic structure. The sol–gel process, however, allows processing at or near room temperature, and it is now possible to alter the structure of a ceramic and modify its properties so that the ceramic no longer behaves like a pure inorganic brittle material but may acquire some characteristics of a polymer. A continuous variation in properties from a purely inorganic brittle material to a completely rubbery organic one is now possible, as will be discussed in this chapter. But beyond the simple modification in properties, it is also the possibility of building novel structures, including organized nanostructures, that has excited many

Mechanical Properties of Organic–Inorganic Hybrids

3

scientists. Using the sol–gel process, new synthesis strategies have been developed using novel precursors that allow a more precise control of the organic–inorganic interface, or control over the size of the organic or inorganic phases. Such materials also offer a significant challenge to theoretical modeling. For example, as the size of the organic and inorganic phases reaches the nanometer scale, what is the role of the interface versus the role of the bulk in controlling the mechanical properties? There are two main types of hybrids, depending on whether a chemical bond exists between organic and inorganic components of the hybrid (Sanchez and Ribot 1994). In so-called type I, the organic and inorganic phases are not chemically bonded. In “type II” hybrids, the phases are covalently connected. An example of the type I materials is the PMMA (polymethylmetacrylate)–SiO2 materials, in which PMMA fills the pores of a silica gel (Pope et al. 1989). In this typical example, a polymeric solution of an acrylate was impregnated into the pore of a silica xerogel and polymerized in situ. The relative amount of each phase is adjusted by modifying the porosity of the silica gel prior to impregnation. The compressive strength of the hybrids increases from 10
103 psi, for 100% PMMA, to 48
103 psi for a composite containing 25% PMMA by volume. The variation in compressive strength is linear with increasing PPMA volume fraction. The mechanical properties of such macroscopic systems are more likely to follow a simple rule of mixtures, because the nature of the interface between organic and inorganic phases does not control the properties. Most hybrids studied to date belong to the “type II” category. Our discussion will focus first on the polydimethylsiloxane (PDMS)–SiO2 system, and we shall discuss some of the other hybrids studied for their mechanical properties in a following section.

The Pdms–SiO2 System Because it is one of the oldest systems investigated, the PDMS–SiO2 systems are probably one of the most well-characterized hybrid system. The compatibility between the silanol-terminated PDMS and TEOS has also made the mixture particularly attractive to many researchers (Wilkes et al. 1985; Haung and Wilkes 1987; Guo et al. 1999; Teowee 1996). This compatibility, as shown in the reaction below between a silanol-terminated PDMS chain and a molecule of silicic acid, is typical of the kind of reaction found in so-called type II hybrids where cross-linking between organic and inorganic components can take place: CH3 HO

Si CH3

CH3

OH O H + HO n

Si OH

OH

HO

Si CH3

OH O

Si OH + H2O n OH

(1)

The advantages of this system include the similarity between the silica network and the siloxane structure and the high temperature stability of PDMS compared to other elastomers. Many organically modified silicates (Ormosils) can be prepared by the sol–gel method with very different mechanical properties, by varying the ratio of

4

J.D. Mackenzie and E.P. Bescher

polydimethylsiloxane to tetraethoxysilane (TEOS) and the processing conditions, or even the molecular weight of PDMS. At lower PDMS content, the Ormosils are harder, stiffer, and stronger than those with higher concentration of PDMS. Even harder Ormosils are possible when ultrasonic irradiation is used during synthesis, as will be discussed later. As the PDMS concentration is increased, the Ormosils take on a more flexible nature, and over a critical concentration, they actually become rubbery. These rubbery materials can contain as much as 75% inorganic component and have more stable mechanical properties than conventional rubbers at elevated temperatures. Due to their unique characteristics, rubbery Ormosils have potential applications in areas where materials need to be rubbery, light weight, and resistant to high temperatures and environmental degradation.

Young’s Modulus The Young’s modulus of these hybrids exhibits a continuous variation with increasing organic content (Fig. 1). It is noteworthy that this variation is not linear and does not follow a rule of mixtures. A significant shift in the properties of the hybrids appears to take place near 35% by weight of PDMS. Hard Ormosils have a Young’s modulus of approximately 5 GPa, and the rubbery ones of 10 MPa. Concurrently, the elongation at failure increases from about 1% for a 10% PDMS hybrid to higher than 25% for a hybrid containing 50% PDMS. It is possible that this nonlinearity in the change of Young’s modulus with concentration may be due to the threshold concentration over which the PDMS phase is continuous throughout the material, as shown in the proposed structures in Fig. 2. Below this threshold concentration, the PDMS phase is discontinuous and, therefore, the tensile properties are controlled by the brittle inorganic phase. This variation in mechanical properties of hybrids is similar to the variation in mechanical properties as a function of alkali content in silicate glasses. The properties of PDMS–SiO2 materials can be further modified using several types of reinforcements, such as fumed silica or glass fibers. Using silica fume, the properties of the composites become similar to those of some rubbers (Minervini 1994). A tensile strength of 1–2 MPa and elongation of 25–51% is obtained. This causes a decrease in the elastic modulus from 19.2 to 3 MPa, as shown in Table 1. It is noteworthy that increasing the molecular weight of the PDMS from 1700 to 4200 g mol1 increases the elongation at failure of the rubbery Ormosils to 51%. Reinforcement of PDMS–SiO2 Ormosils with colloidal silica also improves the resilience of the hybrids from 47 to 69% (Table 2). Aburatani et al. (2002) have also modified bioactive Ormosils with colloidal silica. They found that as the colloidal silica content increased from 10 to 25 wt%, the compressive strength of the PDMS–SiO2 Ormosils increased from 0.1 to 3.5 MPa, while the maximum strain remained practically constant near 30%. These bioactive rubbery Ormosils saw their Tg remain almost constant (98  C to 101  C), as the colloidal silica load increased from 14 to 25%. Their storage modulus was 10 MPa above Tg and on the order of 1 GPa below Tg.

Mechanical Properties of Organic–Inorganic Hybrids

5

Fig. 1 Mechanical properties of PDMS–SiO2 Ormosils

It is also possible to reinforce the hybrids with glass fibers. Whereas the glass fiber reinforcement does not appear to provide any benefits in strength, it allows for a drastic reduction in the drying shrinkage of the gel (Huang and Mackenzie 1995).

Hardness Ormosils, in the broadest sense, may be considered an extension of the family of amorphous inorganic glasses. Therefore, some of the models that have been developed for glasses might be of some use in furthering our understanding of the mechanical properties of hybrids. Hardness is a complex property in the context of brittle materials, and even more so for rubbery materials. But there is general agreement that for brittle materials, hardness is directly related to Young’s modulus and packing density. This relationship had been clearly established in the case of amorphous inorganic glasses. Makishima and Mackenzie initially developed a mathematical model for the dependence of elastic modulus on the composition inorganic glass (Makishima and Mackenzie 1973). This model was further refined to derive an equation for the calculation of the Vickers hardness of glass (Yamane and Mackenzie 1974). Based on both models, Makishima and Mackenzie later developed a mathematical model for the hardness of Ormosils (Makishima and Mackenzie 2000). The equation developed for the hardness of Ormosils is

6

J.D. Mackenzie and E.P. Bescher

Fig. 2 Structure of PDMS–SiO2 Ormosils Table 1 Some mechanical properties of colloidal silica-doped PDMS–TEOS Ormosils Ormosil TEOS/PDMS1700 Aerosil ox 50-TEOS/ PDMS1700 Aerosil ox 50-TEOS/ PDMS4200

Density (g cm3) 0.46 0.79

Tensile strength (MPa) 2.15 0.86

Elongation (%) 24.5 42.6

0.79

1.15

51.3

Elastic modulus (MPa) 19.2 3.0 4

Mechanical Properties of Organic–Inorganic Hybrids

7

Table 2 Resilience of PDMS/TEOS hybrids Material PDMS1700/TEOS Aerosil ox 50-PDMS1700/TEOS Aerosil 150-PDMS4200/TEOS Silicone rubber Neoprene

Resilience (%) 47 69 76 49 26



β H v ¼ 1:99 10:8 Vt  1

1=2

Vt3=2 G

(2)

where β = 1–0.5 X (X is the mole fraction of dimethylsiloxane), G = 14.4 (1–0.5 X) in kcal/cm3 is the bonding energy per unit volume, and Vt is the atomic packing density. The agreement between hardness values calculated by this equation and those measured was good. As with all sol–gel systems, the properties of the final gels are strongly influenced by the characteristics of the solution, type of precursor, and sequence of mixing. Oh et al. (2002) found that the hardness of PDMS–SiO2 hybrids decreased from 70 to 40 Hv when the H2O/TEOS ratio increased from 4 to 16. This was ascribed to the higher porosity and lower density of the high water Ormosils. Also, using lower molecular weight, short chain PDMS resulted in more compact and harder materials. Using a PDMS with a molecular weight of 400–700 g mol1, the hardness was 41.9 Hv versus 40.6 Hv for PDMS of 1700 g mol1.

Thermal Stabilization of PDMS–SiO2 Hybrids Because of their inorganic content, Ormosils have the potential of being more thermally stable than organic polymers. Nevertheless, their thermal stability is still limited by the thermal stability of their organic constituent(s). For some applications, it might be desirable to retain some polymer-related properties, such as toughness, rubberiness, or elongation at failure at elevated temperatures. Pure PDMS begins to decompose in air near 350  C. Several strategies have been explored to improve the thermal stability of hybrids. The stabilizing effect of antioxidants (Rao et al. 1991) or iron additions (Nielsen 1973) on the high temperature properties of siloxanes has been known for many years and has been used extensively in the rubber industry. Studies carried out on our laboratory confirm that a similar effect is observed in PDMS–SiO2 hybrids by the sol–gel method. Ormosils see their thermal stability enhanced by the addition of a few weight percent of iron chloride in the solution prior to gelation (Fig. 3) (Mackenzie et al. 1996). A PDMS–SiO2 Ormosil hybrid with approximately 1% FeCl3 begins to decompose at about 500  C (Bescher et al. 2003). The mechanism of stabilization might be more complex than in pure PDMS. It appears to involve a delay in the onset of the oxidation of PDMS, which is consistent with previous observations in silicones. However, it also appears that Fe plays a

8

J.D. Mackenzie and E.P. Bescher 70 60 50 40 30 20

SBR Rubber Rubbery Ormosil

10

Rubbery Ormosil w/Fe 0 50

100

150

200

250

300

350

400

450

Temperature (°C)

Fig. 3 Thermal stabilization of PDMS–SiO2 Ormosils

significant role in altering the molecular structure of the solution during processing. Also, this mechanism is now found to occur in the solid Ormosils, as opposed to the liquid siloxanes. The viscosity of a Fe-doped solution is significantly affected, with iron additions drastically increasing the viscosity of the solution prior to gelation, indicating a modification of the PDMS at the solution stage. The experimental observations point to the important role played by oxygen in the stabilization process. It is possible that the Fe2+/Fe3+ couple plays a role in the stabilization, as the Ormosils are not as strongly stabilized under nitrogen atmosphere. Furthermore, the DMA moduli of gels made with and without irons are appreciably different: a Fe-doped PDMS gel exhibits higher Tg (325  C versus 220  C) and higher modulus (1.5 GPa versus 0.8 GPa) (Fig. 4).

Aeromosils Aerogels are highly porous materials produced by sol–gel processes and usually dried supercritically. They exhibit superior thermal insulation properties. A shortcoming of traditional aerogels is their extreme brittleness, which prevents them from maintaining physical integrity when being handled. Their very small strength is due to their high porosity (typically in excess of 90%) and the brittle nature of the oxide framework. Using the sol–gel process, it is possible to fabricate organic–inorganic aerogels with improved mechanical properties. SiO2–PDMS aerogels have been fabricated by Kramer et al. (1996). The term “Aeromosil” was coined to describe such materials. Table 3 shows that a 10% PDMS–90% SiO2 aerogel exhibits twice

Mechanical Properties of Organic–Inorganic Hybrids

9

1000000

Modulus [Pa]

LOSS Modulus

Storage Modulus

100000

50

100

150

200

250

Temperature [C] 10000000

Modulus [Pa]

Loss Modulus

Storage Modulus

1000000

100000 0

50

100

150

200

250

300

350

400

Temperature [C]

Fig. 4 Dynamic modulus of PDMS–SiO2 hybrids with and without Fe doping

the strength and four times the elongation of a pure inorganic SiO2 gel, for a comparable density. These aerogels also exhibit some rubberiness, as shown in Fig. 5.

Sono-Ormosils PDMS–SiO2 is usually processed using several solvents such as isopropanol and tetrahydrofuran in order to prevent phase separation between PDMS and water at the early stages of sol–gel processing. These solvents are undesirable in the final material and lead to increased porosity and large drying shrinkage. However, it is

10

J.D. Mackenzie and E.P. Bescher

Table 3 Properties of aeromosils Material Silica aerogel 10% PDMS aeromosil

Density(g cm3) 0.10 0.09–0.1

σ (kPa) 10–20 40–74

Elongation (%) 7) for the examined Gd(NO3)3 concentrations, and this pH value decreased when the Gd(NO3)3 concentration was increased. Because the QD/SiO2/GdC particle surface was covered with GdC, properties of the QD/SiO2/GdC particle surfaces were expected to have similar characteristics to those of GdC particle surfaces. ζ-Potentials of GdC particles prepared with the homogeneous precipitation method increased with a decrease in pH. Accordingly, an increase in Gd(NO3)3 concentration could increase the ζ-potential of QD/SiO2/GdC particles. It is possible that the increased ζ-potential at high Gd(NO3)3 concentrations created a larger electric repulsion among the QD/SiO2/GdC particles than at low Gd (NO3)3 concentrations. Thus, highly dispersed QD/SiO2/GdC particles were generated at high Gd(NO3)3 concentrations. Silica coating of the QD/SiO2/GdC particles was performed using the modified Stöber method with TEOS and NaOH in a water/ethanol solution at room temperature for 24 h. The QD/SiO2/GdC/SiO2 particles were washed by centrifugation, removal of the supernatant, addition of water, and sonication and by repeating this procedure three

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

19

Fig. 8 TEM images of QD/SiO2/GdC nanoparticles prepared with homogeneous precipitation method at Gd(NO3)3 concentrations of (a) 3
105, (b) 3
104, and (c) 3
103 M. Initial concentrations of QD, urea, and PVP were 3.2
109 M, 0.5 M, and 1 g/L, respectively (Reprinted from Journal of Materials Science 47 (2012) 1852–1859)

Fig. 9 TEM images of QD/SiO2/GdC/SiO2 nanoparticles prepared at TEOS concentrations of (a) 5
104, (b) 1
103, and (c) 5
103 M. Initial concentrations of QD, Gd(NO3)3, urea, and PVP were 3.2
109 M, 3
103 M, 0.5 M, and 1 g/L, respectively (Reprinted from Journal of Materials Science 47 (2012) 1852–1859)

times. Figure 9 shows TEM images of QD/SiO2/GdC/SiO2 particles prepared with various TEOS concentrations. The QD/SiO2/GdC particles were coated with silica shells in all the TEOS concentrations examined. However, the QD/SiO2/GdC/SiO2 particles aggregated and connected with other QD/SiO2/GdC/SiO2 particles at low concentrations. Additionally, the resulting silica shells were so thin that the QD/SiO2/ GdC/SiO2 particles were not colloidally stable. An increase in TEOS concentration increased the silica shell thickness, which controlled particle aggregation and produced quasi-perfect and highly dispersed QD/SiO2/GdC/SiO2 particles.

Quantum Dot/Silica Core–Shell Particles with Immobilized Au Nanoparticles Materials with high X-ray absorption properties can be applied to X-ray imaging techniques. Iodinated contrast agents are usually used in imaging to obtain clear

20

Y. Kobayashi and K. Gonda

images. Typical X-ray contrast agents that are commercially available are iodine compounds that are homogeneously dissolved in solvents. However, iodine-based X-ray contrast agents are problematic because they can provoke adverse events such as allergic reactions (Zhao et al. 2011; Thomsen 2011) and cannot be administered to some patients. Consequently, Au has become a promising candidate for use in imaging because Au has high X-ray absorption and is less toxic than iodine compounds. Researchers have examined the use of Au nanoparticles as contrast agents to take nanometer images of tissues in living bodies (Menk et al. 2011; Peng et al. 2012; Wang et al. 2013). Taking advantage of the multi-functionalization of materials, particles containing QDs and Au can act as both fluorescent and X-ray contrast agents. This section introduces a proposed method to fabricate composite particles composed of QD/SiO2 core–shell particles and Au nanoparticles or QD/SiO2 core–shell particles with immobilized Au nanoparticles (QD/SiO2/Au). Immobilization facilitates the chemical interaction between metallic Au and amino groups. A freshly prepared Na-cit aqueous solution was added to a HAuCl4 aqueous solution at a constant temperature of 80  C under vigorous stirring, which produced a colloidal solution of Au nanoparticles. For efficient immobilization of Au nanoparticles on the QD/SiO2 particle surface, or the production of QD/SiO2/Au particles, amino groups were first introduced on the QD/SiO2 particle surface using (3-aminopropyl)triethoxysilane (APES) (QD/SiO2-NH2) because the alkoxide groups of the APES were expected to react with the OH groups on the silica surface of the QD/SiO2 particles. APES was added to the QD/SiO2 colloidal solution at room temperature. To immobilize the Au nanoparticles, ethanol, water, and the Au nanoparticle colloid solution were added in turn to the QD/SiO2-NH2 particle colloid solution at 35  C, allowing the amino groups on the particle surface to coordinate with the surface of the Au nanoparticles. The QD/SiO2-NH2 nanoparticles dispersed well in water and were colloidally stable, which indicated that the addition of APES had no serious effect on their colloidal stability in water. TEM observations (Fig. 10c) revealed that some particles appeared to form aggregates that were likely to precipitate. However, no precipitates were found in the particle colloid solution. These aggregates were thought to be produced during the preparation of TEM samples due to evaporation of solvent on the TEM grid. Similar to the QD/SiO2 nanoparticles, particles containing a few QD cores were observed. Their particle sizes were similar to those of the QD/SiO2 nanoparticles, which indicated that their core–shell structures were chemically stable even after the amination process. ELS measurements and TEM observations (Fig. 10b) indicated that the QD surfaces were covered with silica. The IEP shifted to 9.5 upon amination. Acid dissociation constants for amino groups in many types of amines, such as ammonia, alkylamine, and dialkylamine, range between 9 and 11, and the IEP of 9.5 was within the range. These results confirmed that the amino groups were successfully introduced onto the particle surfaces. Figure 10d shows a TEM micrograph of the QD/SiO2/Au nanoparticles. All the Au nanoparticles were immobilized on the surface of particles, indicating that the Au nanoparticle immobilization was successful using the described method. Several Au

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

21

Fig. 10 TEM images of (a) QD, (b) QD/SiO2 particles, (c) QD/SiO2-NH2 particles, and (d) QD/SiO2/Au particles (Reprinted from Applied Nanoscience 6 (2016) 301–307)

nanoparticle-free particles were also observed, but the number ratio of Au nanoparticle/QD/SiO2-NH2 nanoparticles was less than 1/1. Further optimization of fabrication conditions such as concentrations of Au nanoparticles and QD/SiO2-NH2 nanoparticles in the final colloid solutions, reaction temperatures, and stirring rates is required to improve the efficiency of Au nanoparticle immobilization.

The Effect of Silica on the Photostability of Quantum Dots One of the major difficulties in implementing the use of metal chalcogenide semiconductor nanoparticles in devices for practical applications is the prevention of light-induced surface reactions that can lead to the total oxidation of the nanoparticles. In the case of CdS nanoparticles, Henglein showed that CdS degraded under the influence of light in the presence of dissolved oxygen (Henglein 1982). Oxygen facilitates the oxidation of sulfide radicals that are generated through the formation of hole-anion pairs at particle surfaces through the following reaction:  S_ S þ O2 ! SS þ O_ 2

22

Y. Kobayashi and K. Gonda

where the subscript S indicates a surface atom. This process can be inhibited by preventing the CdS surface from contacting oxygen. The silica shells of particles should be useful in preventing this contact. Thus, silica shells may have an additional function as a physical barrier to prevent oxidation in addition to decreasing particle toxicity and increasing colloidal stability of particles. Prior to investigations on the fluorescence imaging properties of the nanoparticles, the effect of silica on the photostability of QDs was investigated. Two types of investigations were performed, including the effects of silica shells and a comparison between silica using sodium silicate solutions and silicon alkoxide.

Effect of Silica Shells The QD/SiO2 used for this investigation was the same as in Fig. 7 (Zhou et al. 2005). UV–vis spectra revealed that extinction of the uncoated CdSe nanoparticle colloid solution decreased markedly with time, while that of the CdSe QD/SiO2 particle colloid solution remained constant. This indicated that the silica shell was rigid enough to prevent O2 from reaching the surfaces of the CdSe QDs. Figure 11 shows changes in photoluminescence (PL) intensity at different illumination times after exposure to an Ar laser (532 nm). The PL intensity of rhodamine 560 gradually decreased upon illumination, which reconfirmed instability of the organic dye. The PL intensity of the bare CdSe QDs increased rapidly in the initial stage of illumination. This seemed to result from photo annealing, which decreased the number of defect sites on the surface of the CdSe QDs. With a long duration of strong illumination, the PL intensity greatly decreased. This likely resulted from the CdSe QDs absorbing so many high-energy photons, which resulted in decomposition into elemental Cd and Se under the extreme conditions. Similar observations on CdS have been previously noted by Spanhel et al. (1987). In contrast, the PL intensity of CdSe QD/SiO2 remained mostly constant except for a slight increase during early stages of illumination. Here, the silica shell was likely rigid enough to confine the Cd and Se atoms within a fixed space and site, which maintained the structure of CdSe QDs.

Comparison of Silica Produced Using a Sodium Silicate Solution and Silicon Alkoxide Silica can be fabricated from both silicon alkoxide and sodium silicate solutions. The two types of silica are compared in this section (Correa-Duarte et al. 2001). This investigation used CdS QD nanoparticles produced by mixing Cd(NO3)2 and Na2S aqueous solutions in the presence of Na-cit. Then, a freshly prepared aqueous solution of MPS and sodium silicate solution were added to the CdS QD particle colloid solution to form thin silica shells that colloidally stabilized the CdS QD particles. The pH of the resulting dispersion was adjusted to 8.5 and was allowed to stand for several days in a nitrogen atmosphere such that the silica could slowly

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . . 2.5 CdSe QD 2.0

Relative Intensity

Fig. 11 Changes in photoluminescence intensity at different illumination (700 mW, Ar laser) times. (a) CdSe QD, (b) CdSe QD/SiO2 particles, (c) rhodamine 560 (Originated from Applied Surface Science 242 (2005) 281–286)

23

CdSe QD/SiO2 Rhodamine 560 1.5

1.0

0.5

0.0

0

1

2

3

4

5

6

7

Time (min)

polymerize onto the modified CdS QD particle surfaces. Two different methods were used to prepare silica gels loaded with nanoparticles. In method 1, the silica-coated CdS particle colloid solution was added to tetramethyl orthosilicate/methanol under vigorous magnetic stirring. Then, the final mixture was poured into a plastic cuvette, which was sealed with a rubber sheet to prevent evaporation. Formation of the silica gel was complete within hours. In method 2, a solution of sodium silicate was added to the CdS QD/SiO2 particle colloid solution with magnetic stirring. Once the addition of silicate was complete, the pH was adjusted to 7 with an aqueous HCl solution. The sample was then poured into a cuvette and quickly sealed with transparent plastic film. Formation of silica gel occurred over several hours. In method 1, optical features of the silica such as color and luminescence were retained for weeks and even months. However, those produced using method 2 suffered from photodegradation in an unusual fashion. Although the entire sample was illuminated and the open end of the cuvette was sealed, the sample’s characteristic pale yellow color and its red luminescence gradually decreased from top to bottom in a time scale of weeks. These phenomena are closely related to the porosity of the silica. To characterize the porosity of the two samples, nitrogen sorption experiments were carried out to gain information on the pore size and specific surface areas of the silica. Surface area values were obtained from the isotherms via the Brunauer-Emmett-Teller expression, and average pore diameters were derived using the Barrett-Joyner-Halenda method and desorption branches of nitrogen sorption isotherms. The surface area obtained for method 1 was 674 m2/g, which was almost twice the value of 321 m2/g obtained for method 2. The calculated pore size for the silica generated using method 1 was 3.0 nm, which was significantly smaller than the 4.5 nm pore size that was calculated for method 2. This result indicated that the silica gel produced using sodium silicate had a much more open structure, such that the particle cores were more accessible to oxygen, which resulted

24

Y. Kobayashi and K. Gonda

in photodegradation. Accordingly, the silica gel produced from silicon alkoxide is expected to stabilize photo-properties of metal chalcogenide QDs.

Fluorescence Imaging Quantum Dot/Silica Core–shell Particles Fluorescence imaging of particles as small as molecules and nanometers in size will need to be developed to improve imaging abilities in the future. Our research group has developed a technique to measure fluorescence intensities of individual particles, using an instrument composed of an optical system equipped with a confocal microscope (Gonda et al. 2010; Hikage et al. 2010). The fluorescence intensity of each QD/SiO2 particle was measured using this system (Kobayashi et al. 2012c). The sample solution was pipetted onto a glass slide and covered with a glass cover slip. Particles were illuminated with a blue laser (488 nm), and the fluorescence images of the particles were obtained using an optical system consisting of an epi-fluorescent microscope with an oil immersion objective lens, a Nipkow disktype confocal unit, and an electron multiplier type charge-coupled device camera. Figure 12a, b shows the fluorescence images of the QDs and QD/SiO2 particles, respectively. The fluorescence of the QDs contained in particles, confirmed as bright spots, was successfully detected using this setup. An in vivo imaging system (IVIS) fluorimeter was also used for fluorescence imaging. The QD/silica particles used were fabricated following the method of producing multiple QD/SiO2 core–shell particles (Kobayashi et al. 2010b). The QD/SiO2 particles, to which poly(ethylene glycol) (PEG) was immobilized, were also used for IVIS imaging. The immune system recognizes hydrophobic materials as foreign, which limits blood circulation when such particles are used in vivo. PEGylation is a process where PEG is immobilized to a material; this process is often performed to make surfaces hydrophilic (Niidome et al. 2010; Basile et al. 2012; Yoshino et al. 2012; Ma et al. 2012; Otsuka et al. 2012; Lo et al. 2013). Our QD/SiO2 particles were also PEGylated. An aqueous solution of methoxy PEG silane (M-SLN-5000) was added to the QD/SiO2 particle colloid solution, where PEG groups were expected to be introduced to particle surface through a reaction between the silanol groups on surfaces of QD/SiO2 particles and the alkoxide groups of the M-SLN-5000. The QD/SiO2/PEG colloid solution was concentrated through the process of centrifugation, removal of supernatant, addition of solvent, and redispersion. Colloid solutions of QD particles, QD/SiO2 particles, and QD/SiO2/ PEG particles successfully produced IVIS images. For all the colloid solutions, radiant efficiencies (RE) increased with an increase in QD concentration. The RE values tended to level off with an increase in QD concentration, which was probably due to concentration quenching. The RE values of the QD/SiO2 particle colloid solution were slightly larger than those of the QD particle colloid solution. The QD nanoparticles are apt to aggregate due to nature of the nanoparticles. Such aggregation probably decreased the fluorescence intensity of the QD

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

25

Fig. 12 Fluorescence images of (a) QD, (b) QD/SiO2, (c) QD/SiO2/GdC, and (d) QD/SiO2/GdC/ SiO2 particles. Journal of Materials Science 47 (2012) 1852–1859

nanoparticles. In contrast, formation of a core–shell structure prevented the QD nanoparticles from aggregating because of the physical barrier created by the silica shell, which also helped retain their fluorescence intensity. Thus, it was noted that the silica coating had the ability to retain efficient fluorescence emissions of QDs. There was no large difference in RE values between the colloid solutions of QD/SiO2 particles and QD/SiO2/PEG particles. This indicated that PEGylation did not deteriorate fluorescence properties of the QDs. Figure 13a, c shows photographs and IVIS images of a mouse taken prior to and after injection of a QD/SiO2/PEG particle colloid solution. Fluorescence emissions increased with time after the injection. Figure 13d shows a photograph and an IVIS image of the mouse after injection and a subsequent laparotomy. Fluorescence emissions from the tissue were clearly observed, which indicated that the injected QD/SiO2/PEG particles were able to reach the tissues from the vein. Figure 14 shows IVIS images of various tissues taken prior to and after an injection. There were no large differences in fluorescence emissions between the kidney and heart prior to and after the injection. Emissions from the lung, liver, and spleen became strong after the injection. These results indicate that the QD/SiO2/PEG particles efficiently reached these tissues by flowing through the veins but were probably recognized as alien substances and contained within them, subsequently emitting strong fluorescence.

26

Y. Kobayashi and K. Gonda

Fig. 13 Photographs (A) and IVIS images (B) of mouse taken (a) prior to (b) at 5 min after and (c) at 1 h after injection of QD/SiO2/PEG particle colloid solution. The mouse (d ) was the mouse (c), for which a laparotomy was performed (Reprinted from Journal of Sol–gel Science and Technology 66 (2013) 31–37)

The effect of particle size on fluorescence imaging was also investigated (Kobayashi et al. 2015b). Figure 15 shows IVIS images for the QD, S-QD/SiO2, and L-QD/SiO2 colloid solutions. All the colloid solutions were successfully imaged, and based on fluorescence intensities or RE values, the images were improved in terms of visual clarity with increasing QD concentrations. For all the colloid solutions, the RE values increased linearly with increasing QD concentrations. Concentration quenching did not occur when we used the colloid solutions of interest. There was no significant difference in RE values of the colloid solutions. These results indicate that the silica-coating process and increased shell thickness did not degrade the QD fluorescence properties. Prior to the IVIS imaging of the mouse, the QD/SiO2 particle surfaces were PEGylated using M-SLN-5000. Figures 16 and 17 show the IVIS images for various mouse tissues extracted after particle injection followed by laparotomy. These tissues exhibited clear fluorescence for both QD/SiO2/PEG particle colloid solutions, which indicated that the injected particle colloid solutions reached the tissues from the vein. Figure 18 shows bar graphs summarizing Figs. 16 and 17. For comparison, the fluorescence emissions of various tissues prior to injection are shown in Fig. 18. For the S-QD/SiO2/PEG particles, the emissions from the lung and liver were strong after the injection. Pulmonary embolism occurred due increased florescence in the lung. A large increase in RE in the liver indicated that the S-QD/SiO2/PEG particles efficiently reached these tissues via blood flow after passing through the lungs. In the liver region, we speculate that the particles were most likely recognized as foreign substances and trapped by the liver, which resulted in strong

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

27

Fig. 14 IVIS images (B) of (a) heart, (b) lung, (c) liver, (d ) spleen, and (e) kidney of mouse after injection of QD/SiO2/PEG particle colloid solution. These tissues were taken from the mouse (d ) in Fig. 12. IVIS images (A) were taken as control (no injection) (Reprinted from Journal of Sol–gel Science and Technology 66 (2013) 31–37)

particle fluorescence emissions. The RE of the heart did not change significantly after the injection, which indicated that the colloid solution did not circulate through the body for a prolonged period. The RE values for the heart, lung, liver, and spleen for the L-QD/SiO2/PEG particles were approximately 4, 3, 94, and 8 times larger, respectively, than the corresponding RE values prior to the injection. The RE for the heart increased in contrast to the levels when the S-QD/SiO2/PEG particles were used, which implied the L-QD/SiO2/PEG particles circulated through the blood for a longer period of time than the S-QD/SiO2/PEG particles. The observed RE increases for the lung, the liver, and the spleen were smaller and larger, respectively, compared with the use of S-QD/SiO2/PEG particles. This indicated that the L-QD/SiO2/PEG particles passed through the lungs more efficiently and were more effectively trapped by the liver and spleen. Small particles tend to aggregate due to their large surface energy. The S-QD/SiO2/PEG particles aggregated and were trapped by the lungs, which caused pulmonary embolism. In contrast, the L-QD/SiO2/PEG particles did not aggregate significantly. This lack of aggregation allowed the particles to pass through the lungs and efficiently reach not only the liver but also the spleen, where they were trapped, and serves as another benefit in the form of smooth particle flow. This smooth flow promoted particle circulation in the blood vessels, which increased the RE for the

28

Y. Kobayashi and K. Gonda

Fig. 15 IVIS images of the (A) QD, (B) S-QD/SiO2, and (C) L-QD/SiO2 colloid solutions with QD concentrations of (a) 6.25
109, (b) 1.25
108, (c) 2.5
108, (d) 5.0
108, and (e) 1.0
107 M (Reprinted from Journal of Chemical Engineering of Japan 48 (2015) 112–117)

Fig. 16 IVIS images of the (a) heart, (b) lung, (c) liver, and (d) spleen of a mouse after injection with the S-QD/SiO2/PEG particle colloid solution (Reprinted from Journal of Chemical Engineering of Japan 48 (2015) 112–117)

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

29

Fig. 17 IVIS images of the (a) heart, (b) lung, (c) liver, and (d) spleen of a mouse after injection with the L-QD/SiO2/PEG particle colloid solution (Reprinted from Journal of Chemical Engineering of Japan 48 (2015) 112–117) Fig. 18 Radiant efficiencies for the various mouse tissues. Red bars and blue bars, 1 h after injecting the S-QD/SiO2/ PEG and L-QD/SiO2/PEG colloid solutions, respectively. Black bars, control (no injection) (Reprinted from Journal of Chemical Engineering of Japan 48 (2015) 112–117)

heart. Thus, different radiant efficiencies were obtained for QD/SiO2/PEG particles of different sizes.

Multilayered Core–Shell Particles Composed of Quantum Dots, Gadolinium Compounds, and Silica The fluorescence intensities of QD/SiO2/GdC and QD/SiO2/GdC/SiO2 particles were also measured using the optical system with a confocal microscope (Kobayashi et al. 2012c). The QD/SiO2/GdC and QD/SiO2/GdC/SiO2 particles used to fabricate the multilayered core–shell particles were the same as in Figs. 7c and 8a, respectively. Figure 11c, d shows fluorescence images of the QD/SiO2/GdC and QD/SiO2/ GdC/SiO2 particles on glass plates. Bright spots were clearly detected, which derived from the fluorescence of the QDs contained in the particles. Fluorescence

30

Y. Kobayashi and K. Gonda

intensities were estimated from the brightness of fluorescence. As the number of layers increased, the fluorescence intensity of single particles tended to decrease. In particular, GdC coating decreased fluorescence intensities significantly. It was found that excitation and emission of QDs were prevented by optical absorption and scatter caused by the GdC layer, although silica layer did not affect the fluorescence intensity of QDs compared to the GdC layer. Our previous studies demonstrated that multilayered SiO2/GdC/SiO2 core–shell particles exhibited MRI abilities (Kobayashi et al. 2007). This implies that the multilayered core–shell particles obtained in this study will also function as an MRI contrast agent. This study indicated that the multilayered core–shell particles emitted strong fluorescence. In summary, the multilayered core–shell particles were harmless to living bodies and could simultaneously function as a fluorescent marker and high-contrast MRI agent.

Quantum Dots/Silica Core–Shell Particles Immobilizing Au Nanoparticles IVIS imaging was performed using a QD/SiO2/Au particle colloid solution (Kobayashi et al. 2016), which was the same as observed in Fig. 9. Figure 19a shows an IVIS image of the QD/SiO2/Au particle colloid solution, which was successfully imaged. Its RE value was 2.23
107 (p/s/cm2/sr)/(μW/cm2). This value was converted into a value with respect to QD concentration to compare it with that of commercially available QDs. Because the colloid solution had a QD concentration of 8
107 M, the converted RE was 2.79
1013 (p/s/cm2/sr)/(μW/ cm2)/M. According to our previous work (Kobayashi et al. 2013e), the RE of QD was 2.93
109 (p/s/cm2/sr)/(μW/cm2) at a QD concentration of 1
107 M, which corresponded to a converted RE value of 2.93
1016 (p/s/cm2/sr)/(μW/cm2)/M. This converted RE for the QD/SiO2/Au particle colloid solution was smaller than that of the commercial QD. The Au nanoparticle colloid solution absorbed visible light in a wide range around the wavelength of 521 nm. The fluorescence from the QD/SiO2/ Au particles was probably diminished by this absorption, although the mechanism is still unclear. Although the converted RE for the QD/SiO2/Au particle colloid solution was small, the intensity of the RE was high enough to image a mouse, based on our previous experience with animal experiments. Figure 19b shows an IVIS image of a mouse after injection of the QD/SiO2/Au particle colloid solution. Fluorescence was clearly observed on the chest after injection into the chest wall, which indicated that fluorescence penetrated through the skin from inside the mouse. Its RE value was 3.30
107 (p/s/cm2/sr)/(μW/cm2). This value was approximately 1.5 times larger than the value of 2.23
107 (p/s/cm2/sr)/(μW/cm2) for the QD/SiO2/Au particle colloid solution, which indicated that the QD/SiO2/Au particle colloid solution could emit strong fluorescence even inside living bodies without quenching. The detection of large RE implied that the QD/SiO2/Au particles accumulated in the chest wall, although the reason behind the large RE is still unclear. Figure 20b shows an X-ray image of the QD/SiO2/Au particle colloid solution. For comparison, an X-ray image of water is shown in Fig. 20a. The white contrast

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

31

Fig. 19 IVIS images of (a) the QD/SiO2/Au particle colloid solution and (b) a mouse after the colloid solution was injected into its chest wall (Reprinted from Applied Nanoscience 6 (2016) 301–307)

in the image of the QD/SiO2/Au particle colloid solution was lighter than that of water. The computed tomography (CT) value of the QD/SiO2/Au particle colloid solution was 1180  314 Hounsfield units (HU) at a Au concentration of 5.4
102 M. This value was converted into a value with respect to the molar concentration of subject materials such as Au and I (i.e., converted CT value) to compare it with that of a commercial contrast agent. The converted CT value was estimated to be 2.19
104 HU/M with respect to the Au concentration. According to our previous work (Kobayashi et al. 2013d), the converted CT value with respect to the iodine concentration of Iopamiron 300 is 4.76
103 HU/M. The converted CT value of the QD/SiO2/Au particle colloid solution was larger than that of Iopamiron 300. Gold absorbs X-rays more than iodine on an atomic level because of its large atomic number, resulting in a larger converted CT value for the QD/SiO2/Au particle colloid solution. Thus, it was confirmed that the QD/SiO2/ Au particle colloid solution had the ability to function as a highly sensitive X-ray contrast agent. Figure 20c shows an X-ray image of a mouse after it was injected with the QD/SiO2/Au particle colloid solution. The location of the particle colloid solution could be recognized clearly at the chest wall because of its light contrast. Its CT value was 1060  374 HU, which was as high as the value obtained for the QD/SiO2/Au particle colloid solution (1180  314 HU). Similar to the fluorescence images, the QD/SiO2/Au particle colloid solution could be clearly observed even inside live bodies without quenching. Another result worthy of noting is that when the mouse injected with the QD/SiO2/Au particle colloid solution was imaged simultaneously using IVIS and X-ray, the positioning of images by IVIS was identical to that of the X-ray images. This suggests that the colloid solution can function as a contrast agent for dual imaging processes.

32

Y. Kobayashi and K. Gonda

Fig. 20 CT images of (a) water, (b) the QD/SiO2/Au particle colloid solution, and (c) a mouse after the colloid solution was injected into its chest wall. The QD/SiO2/Au particle colloid solution was observed in the chest wall, as shown by the red circle (Reprinted from Applied Nanoscience 6 (2016) 301–307)

Conclusion This chapter introduced our recent studies on the development of methods to fabricate particles coated with silica shells using the sol–gel reaction, generate multilayered core–shell particles containing QDs and GdC, and immobilize Au nanoparticles on core–shell particles containing QDs. Successful silica coating of nanoparticles was confirmed by TEM observation and UV–vis spectroscopy of the Au/SiO2 core–shell particles. The silica-coating method was extended to fabricate QD/SiO2 core–shell particles. The QD/SiO2 core–shell particles were coated with GdC shells using a homogeneous precipitation method. The Au nanoparticles were immobilized on the surface of QD/SiO2 core–shell particles by modifying their surfaces with amino groups and then attaching the Au nanoparticles on the surface using amino groups. Fluorescence imaging processes using the particles containing the QDs were also introduced in this chapter. The particle colloid solution was applied to image mouse tissues using the IVIS technique. The multilayered core–shell particles composed of QD, GdC, and silica and the Au nanoparticle-immobilized QD/SiO2 core–shell particles are expected to have dual imaging functions and be able to function not only as a fluorescent marker but also as an MRI and X-ray contrast agent. The studies described above indicate that the QD/SiO2 core–shell particles were able to be applied for use in various imaging techniques. Our previous studies have focused mainly on their imaging abilities. Further studies on subjects such as the toxicity of particle colloid solutions and precise mechanisms of surface coating and surface modification will be necessary for the practical implementation of such particles. Acknowledgments This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas “Nanomedicine Molecular Science” (No. 2306) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, JSPS KAKENHI Grant Number 24310085, and the A3 Foresight Program from JSPS.

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

33

References Ayame T, Kobayashi Y, Nakagawa T, Gonda K, Takeda M, Ohuchi N. Preparation of silica-coated AgI nanoparticles by an amine-free process and their X-ray imaging properties. J Ceram Soc Jpn. 2011;119:397–401. Bahadur NM, Furusawa T, Sato M, Kurayama F, Siddiquey IA, Suzuki N. Fast and facile synthesis of silica coated silver nanoparticles by microwave irradiation. J Colloid Interface Sci. 2011;355:312–20. Basile L, Passirani C, Huynh NT, Bejaud J, Benoit JP, Puglisi G, Pignatello R. Serum-stable, longcirculating paclitaxel-loaded colloidal carriers decorated with a new amphiphilic PEG derivative. Int J Pharm. 2012;426:231–8. Bogush GH, Tracy MH, Zukoski CF. Preparation of monodisperse silica particles: control of size and mass fraction. J Non-Cryst Solids. 1988;104:95–106. Bogush GH, Zukoski CF. Studies of the kinetics of the precipitation of uniform silica particles through the hydrolysis and condensation of silicon alkoxides. J Colloid Interface Sci. 1991;142:1–18. Brus L. Electronic wave functions in semiconductor clusters: experiment and theory. J Phys Chem. 1986;90:2555–60. Chen C, Takezako T, Yamamoto K, Serizawa T, Akashi M. Poly(N-vinylisobutyramide)-stabilized platinum nanoparticles: synthesis and temperature-responsive behavior in aqueous solution. Colloids Surf A. 2000;169:107–16. Cho G, Fung BM, Glatzhofer DT, Lee J-S, Shul Y-G. Preparation and characterization of polypyrrole-coated nanosized novel ceramics. Langmuir. 2001;17:456–61. Correa-Duarte MA, Giersig M, Liz-Marzán LM. Stabilization of CdS semiconductor nanoparticles against photodegradation by a silica coating procedure. Chem Phys Lett. 1998;286:497–501. Correa-Duarte MA, Kobayashi Y, Caruso RA, Liz-Marzán LM. Photodegradation of SiO2-coated CdS nanoparticles within silica gels. J Nanosci Nanotechnol. 2001;1:95–9. Cui W, Li J, Zhang Y, Rong H, Lu W, Jiang L. Effects of aggregation and the surface properties of gold nanoparticles on cytotoxicity and cell growth. Nanomed Nanotechnol Biol Med. 2012;8:46–53. Doremus RH. Optical properties of small gold particles. J Chem Phys. 1964;40:2389–96. Dubus S, Gravel JF, Drogoff BL, Nobert P, Veres T, Boudreau D. PCR-free DNA detection using a magnetic bead-supported polymeric transducer and microelectromagnetic traps. Anal Chem. 2006;78:4457–64. Ekimov AI, Efros AL, Onushchenko AA. Quantum size effect in semiconductor microcrystals. Solid State Commun. 1993;88:947–50. Farbman I, Lev O, Efrima S. Optical response of concentrated colloids of coinage metals in the near-ultraviolet, visible, and infrared regions. J Chem Phys. 1992;96:6477–85. García-Santamaría F, Salgueiriño-Maceira V, López C, Liz-Marzán LM. Synthetic opals based on silica-coated gold nanoparticles. Langmuir. 2002;18:4519–22. Gauden AJ, Phal PM, Drummond KJ. MRI safety; nephrogenic systemic fibrosis and other risks. J Clin Neurosci. 2010;17:1097–104. Gonda K, Watanabe TM, Ohuchi N, Higuchi H. In vivo nano-imaging of membrane dynamics in metastatic tumor cells using quantum dots. J Biol Chem. 2010;285:2750–7. Guo J, Yang W, Deng Y, Wang C, Fu S. Organic-dye-coupled magnetic nanoparticles encaged inside thermoresponsive PNIPAM microcapsules. Small. 2005;1:737–43. Guo J, Yang W, Wang C, He J, Chen J. Poly(N-isopropylacrylamide)-coated luminescent/magnetic silica microspheres: preparation, characterization, and biomedical applications. Chem Mater. 2006;18:5554–62. Hall SR, Davis SA, Mann S. Cocondensation of organosilica hybrid shells on nanoparticle templates: a direct synthetic route to functionalized core-shell colloids. Langmuir. 2000;16:1454–6.

34

Y. Kobayashi and K. Gonda

Hardikar VV, Matijević E. Coating of nanosize silver particles with silica. J Colloid Interface Sci. 2000;221:133–6. Hayat MA. Colloidal gold. San Diego: Academic; 1989. Henglein A. Photo-degradation and fluorescence of colloidal-cadmium sulfide in aqueous solution. Ber Bunsen-Ges Phys Chem. 1982;86:301–5. Henglein A. Small-particle research: physicochemical properties of extremely small colloidal metal and semiconductor particles. Chem Rev. 1989;89:1861–73. Hikage M, Gonda K, Takeda M, Kamei T, Kobayashi M, Kumasaka M, Watanabe M, Satomi S, Ohuchi N. Nano-imaging of the lymph network structure with quantum dots. Nanotechnology. 2010;21:185103. Huang YF, Ma KH, Kang KB, Zhao M, Zhang ZL, Liu YX, Wen T, Wang Q, Qiu WY, Qiu D. Coreshell plasmonic nanostructures to fine-tune long “Au nanoparticle-fluorophore” distance and radiative dynamics. Colloids Surf A. 2013;421:101–8. Jung SH, Kim KI, Ryu JH, Choi SH, Kim JB, Moon JH, Jin JH. Preparation of radio active coreshell type 198Au@SiO2 nanoparticles as a radiotracer for industrial process applications. Appl Radiat Isot. 2010;68:1025–9. Kalantari M, Kazemeini M, Arpanaei A. Facile fabrication and characterization of aminofunctionalized Fe3O4 cluster@SiO2 core-shell nanocomposite spheres Mater. Res Bull. 2013;48:2023–8. Kim EY, Kim TS, Choi JY, Han J, Kim H. Multiple tracheal metastases of lung cancer: CT and integrated PET/CT findings. Clin Radiol. 2010;65:493–5. Kimura K. Shape effect and quantum size effect on small metal particles. Z Phys D. 1989;11:327–32. Kitajima K, Ueno Y, Suzuki K, Kita M, Ebina Y, Yamada H, Senda M, Maeda T, Sugimura K. Low-dose non-enhanced CT versus full-dose contrast-enhanced CT in integrated PET/CT scans for diagnosing ovarian cancer recurrence. Eur J Radiol. 2012;81:3557–62. Kobayashi Y, Correa-Duarte MA, Liz-Marzán LM. Sol–gel processing of silica-coated gold nanoparticles. Langmuir. 2001;17:6375–9. Kobayashi Y, Imai J, Nagao D, Takeda M, Ohuchi N, Kasuya A, Konno M. Preparation of multilayered silica-Gd-silica core-shell particles and their magnetic resonance images. Colloids Surf A. 2007;308:14–9. Kobayashi Y, Nozawa T, Takeda M, Ohuchi N, Kasuya A. Direct silica-coating of quantum dots. J Chem Eng Jpn. 2010a;43:490–3. Kobayashi Y, Nozawa T, Nakagawa T, Gonda K, Takeda M, Ohuchi N, Kasuya A. Direct coating of quantum dots with silica shell. J Sol–Gel Sci Technol. 2010b;55:79–85. Kobayashi Y, Inose H, Nakagawa T, Gonda K, Takeda M, Ohuchi N, Kasuya A. Control of shell thickness in silica-coating of Au nanoparticles and their X-ray imaging properties. J Colloid Interface Sci. 2011;358:329–33. Kobayashi Y, Ayame T, Nakagawa T, Gonda K, Ohuchi N. X-ray imaging technique using colloid solution of AgI/silica/poly(ethylene glycol) nanoparticles Mater. Focus. 2012a;1:127–30. Kobayashi Y, Inose H, Nakagawa T, Gonda K, Takeda M, Ohuchi N, Kasuya A. Synthesis of Au-silica core-shell particles by a sol–gel process. Surf Eng. 2012b;28:129–33. Kobayashi Y, Nozawa T, Nakagawa T, Gonda K, Takeda M, Ohuchi N. Fabrication and fluorescence properties of multilayered core-shell particles composed of quantum dot, Gadolinium Compound, and silica. J Mater Sci. 2012c;47:1852–9. Kobayashi Y, Morimoto H, Nakagawa T, Gonda K, Ohuchi N. Preparation of silica-coated Gadolinium Compound particle colloid solution and its application in imaging. Adv Nano Res. 2013a;1:159–69. Kobayashi Y, Ayame T, Nakagawa T, Kubota Y, Gonda K, Ohuchi N. Preparation of AgI/silica/poly (ethylene glycol) nanoparticle colloid solution and X-ray imaging using it. ISRN Nanomater. 2013b; 2013, Article ID 670402.

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

35

Kobayashi Y, Inose H, Nagasu R, Nakagawa T, Kubota Y, Gonda K, Ohuchi N. X-ray imaging technique using colloid solution of Au/silica/poly(ethylene glycol) nanoparticles. Mater Res Innov. 2013c;17:507–14. Kobayashi Y, Inose H, Nakagawa T, Kubota Y, Gonda K, Ohuchi N. X-ray imaging technique using colloid solution of Au/silica core-shell nanoparticles. J Nanostruct Chem. 2013d;3:62. Kobayashi Y, Matsudo H, Nakagawa T, Kubota Y, Gonda K, Takeda M, Ohuchi N. In-vivo fluorescence imaging technique using colloid solution of multiple quantum dots/silica/poly (ethylene glycol) nanoparticles. J Sol–Gel Sci Technol. 2013e;66:31–7. Kobayashi Y, Nagasu R, Shibuya K, Nakagawa T, Kubota Y, Gonda K, Ohuchi N. Synthesis of a colloid solution of silica-coated gold nanoparticles for X-ray imaging applications. J Nanopart Res. 2014;16:2551. Kobayashi Y, Nagasu R, Nakagawa T, Kubota Y, Gonda K, Ohuchi N. Preparation of Au/silica/ply (ethylene glycol) nanoparticle colloid solution and its use in X-ray imaging process. Nanocomp. 2015a;2:83–8. Kobayashi Y, Matsudo H, Kubota Y, Nakagawa T, Gonda K, Ohuchi N. Preparation of silica-coated quantum dot nanoparticle colloid solutions and their application in in-vivo fluorescence imaging. J Chem Eng Jpn. 2015b;48:112–7. Kobayashi Y, Matsudo H, Li T, Shibuya K, Kubota Y, Oikawa T, Oikawa T, Nakagawa T, Gonda K. Fabrication of quantum dot/silica core-shell particles immobilizing Au nanoparticles and their dual imaging functions. Appl Nanosci. 2016;6:301–307. Kreibig U, Genzel L. Optical absorption of small metallic particles. Surf Sci. 1985;156:678–700. Kunzmann A, Andersson B, Vogt C, Feliu N, Ye F, Gabrielsson S, Toprak MS, Buerki-Thurnherr T, Laurent S, Vahter M, Krug H, Muhammed M, Scheynius A, Fadeel B. Efficient internalization of silica-coated iron oxide nanoparticles of different sizes by primary human macrophages and dendritic cells. Toxicol Appl Pharmacol. 2011;253:81–93. Lasagna-Reeves C, Gonzalez-Romero D, Barria MA, Olmedo I, Clos A, Ramanujam VMS, Urayama A, Vergara L, Kogan MJ, Soto C. Bioaccumulation and toxicity of gold nanoparticles after repeated administration in mice. Biochem Biophys Res Commun. 2010;393:649–55. Lee JW, Othman MR, Eom Y, Lee TG, Kim WS, Kim J. The effects of sonification and TiO2 deposition on the microcharacteristics of the thermally treated SiO2/TiO2 spherical core-shell particles for photo-catalysis of methyl orange. Microporous Mesoporous Mater. 2008;116:561–8. Lee Y-C, Chen D-Y, Dodd SJ, Bouraoud N, Koretsky AP, Krishnan KM. The use of silica coated MnO nanoparticles to control MRI relaxivity in response to specific physiological changes. Biomaterials. 2012;33:3560–7. Li S-Z, Xu R-K. Electrical double layers’ interaction between oppositely charged particles as related to surface charge density and ionic strength. Colloids Surf A. 2008;326:157–61. Li C, Zhu J. Metal-enhanced fluorescence of OG-488 doped in Au@SiO2 core-shell nanoparticles. Mater Lett. 2013;112:169–72. Lifshitz E, Dag I, Litvin I, Hodes G, Gorer S, Reisfeld R, Zelner M, Miuti H. Optical properties of CdSe nanoparticle films prepared by chemical deposition and sol–gel methods. Chem Phys Lett. 1998;288:188–96. Lin YS, Wu SH, Hung Y, Chou YH, Chang C, Lin ML, Tsai CP, Mou CY. Multifunctional composite nanoparticles: magnetic, luminescent, and mesoporous. Chem Mater. 2006;18:5170–2. Liu T, Li D, Zou Y, Yang D, Li H, Wu Y, Jiang M. Preparation of metal@silica core-shell particle films by interfacial self-assembly. J Colloid Interface Sci. 2010;350:58–62. Liz-Marzán LM, Giersig M, Mulvaney P. Synthesis of nanosized gold-silica core-shell particles. Langmuir. 1996;12:4329–35. Lo CL, Chou MH, Lu PL, Lo IW, Chiang YT, Hung SY, Yang CY, Lin SY, Wey SP, Lo JM, Hsiue GH. The effect of PEG-5 K grafting level and particle size on tumor accumulation and cellular uptake. Int J Pharmaceut. 2013;456:424–31.

36

Y. Kobayashi and K. Gonda

Ma Y, Sadoqi M, Shao J. Biodistribution of indocyanine green-loaded nanoparticles with surface modifications of PEG and folic acid. Int J Pharmaceut. 2012;436:25–31. Marinakos SM, Shultz DA, Feldheim DL. Gold nanoparticles as templates for the synthesis of hollow nanometer-sized conductive polymer capsules. Adv Mater. 1999;11:34–7. Marshall G, Kasap C. Adverse events caused by MRI contrast agents: implications for radiographers who inject. Radiography. 2012;18:132–6. Melendez-Ramirez G, Castillo-Castellon F, Espinola-Zavaleta N, Meave A, Kimura-Hayama ET. Left ventricular noncompaction: a proposal of new diagnostic criteria by multidetector computed tomography. J Cardiovasc Comput Tomogr. 2012;6:346–54. Menk RH, Sch€ultke E, Hall C, Arfelli F, Astolfo A, Rigon L, Round A, Ataelmannan K, MacDonald SR, Juurlink BHJ. Gold nanoparticle labeling of cells is a sensitive method to investigate cell distribution and migration in animal models of human disease. Nanomed Nanotechnol Biol Med. 2011;7:647–54. Mine E, Konno M. Secondary particle generation at low monomer concentrations in seeded growth reaction of tetraethyl orthosilicate. J Chem Eng Jpn. 2001;34:545–8. Mine E, Yamada A, Kobayashi Y, Konno M, Liz-Marzán LM. Direct coating of gold nanoparticles with silica by a seeded polymerization technique. J Colloid Interface Sci. 2003;264:385–90. Mojić B, Giannakopoulos KP, Cvejić Ž, Srdić VV. Silica coated ferrite nanoparticles: influence of citrate functionalization procedure on final particle morphology. Ceram Int. 2012;38:6635–41. Morimoto H, Minato M, Nakagawa T, Sato M, Kobayashi Y, Gonda K, Takeda M, Ohuchi N, Suzuki N. X-ray imaging of newly-developed Gadolinium Compound/silica core-shell particles. J Sol–Gel Sci Technol. 2011;59:650–7. Mulvaney P, Liz-Marzán LM, Giersig M, Ung T. Silica encapsulation of quantum dots and metal clusters. J Mater Chem. 2000;10:1259–70. Nagao D, Satoh T, Konno M. A generalized model for describing particle formation in the synthesis of monodisperse oxide particles based on the hydrolysis and condensation of tetraethyl orthosilicate. J Colloid Interface Sci. 2000;232:102–10. Niidome T, Ohga A, Akiyama Y, Watanabe K, Niidome Y, Mori T, Katayama Y. Controlled release of PEG chain from gold nanorods: targeted delivery to tumor. Bioorg Med Chem. 2010;18:4453–8. Oh J-G, Kim H. Synthesis of core-shell nanoparticles with a Pt nanoparticle core and a silica shell. Curr Appl Phys. 2013;13:130–6. Otsuka H, Nagasaki Y, Kataoka K. PEGylated nanoparticles for biological and pharmaceutical applications. Adv Drug Deliver Rev. 2012;64:246–55. Pang G, Chen S, Zhu Y, Palchik O, Koltypin Y, Zaban A, Gedanken A. Preparation and characterization of monodispersed YSZ nanocrystals. J Phys Chem B. 2001;105:4647–52. Peng C, Zheng L, Chen Q, Shen M, Guo R, Wang H, Cao X, Zhang G, Shi X. PEGylated dendrimer-entrapped gold nanoparticles for in vivo blood pool and tumor imaging by computed tomography. Biomaterials. 2012;33:1107–19. Rosenholm JM, Zhang J, Sun W, Gu H. Large-pore mesoporous silica-coated magnetite core-shell nanocomposites and their relevance for biomedical applications. Microporous Mesoporous Mater. 2011;145:14–20. Salgueiriño-Maceira V, Correa-Duarte MA. Increasing the complexity of magnetic core-shell structured nanocomposites for biological applications. Adv Mater. 2007;19:4131–44. Schulz M, Ma-Hock L, Brill S, Strauss V, Treumann S, Gröters S, van Ravenzwaay B, Landsiedel R. Investigation on the genotoxicity of different sizes of gold nanoparticles administered to the lungs of rats. Mutat Res. 2012;745:51–7. Schulzendorf M, Cavelius C, Born P, Murray E, Kraus T. Biphasic synthesis of Au@SiO2 core-shell particles with stepwise ligand exchange. Langmuir. 2011;27:727–32. Secchi F, Leo GD, Papini GDE, Giacomazzi F, Donato MD, Sardanelli F. Optimizing dose and administration regimen of a high-relaxivity contrast agent for myocardial MRI late gadolinium enhancement. Eur J Radiol. 2011;80:96–102.

The Development of Quantum Dot/Silica Particles for Fluorescence Imaging and. . .

37

Selvan ST, Patra PK, Ang CY, Ying JY. Synthesis of silica-coated semiconductor and magnetic quantum dots and their use in the imaging of live cells. Angew Chem Int Ed. 2007;46:2448–52. Singh G, Song L. Experimental correlations of pH and ionic strength effects on the colloidal fouling potential of silica nanoparticles in crossflow ultrafiltration. J Memb Sci. 2007;303:112–8. Song X, Jiang N, Li Y, Xu D, Qiu G. Synthesis of CeO2-coated SiO2 nanoparticle and dispersion stability of its suspension. Mater Chem Phys. 2008;110:128–35. Spanhel L, Haase M, Weller H, Henglein A. Photochemistry of colloidal semiconductors. 20. Surface modification and stability of strong luminescing CdS particles. J Am Chem Soc. 1987;109:5649–55. Tago T, Hatsuta T, Nagase R, Kishida M, Wakabayashi K. Preparation of silica-coated Co-Fe3O4 nanoparticles and their magnetic properties. Kagaku Kogaku Ronbunshu. 2001;27:288–90 (in Japanese). Telgmann L, Sperling M, Karst U. Determination of gadolinium-based MRI contrast agents in biological and environmental samples: a review. Anal Chim Acta. 2013;764:1–16. Thomsen HS. Contrast media safety – an update. Eur J Radiol. 2011;80:77–82. Torimoto T, Horibe H, Kameyama T, Okazaki K, Ikeda S, Matsumura M, Ishikawa A, Ishihara H. Plasmon-enhanced photocatalytic activity of cadmium sulfide nanoparticle immobilized on silica-coated gold particles. J Phys Chem Lett. 2011;2:2057–62. Underwood S, Mulvaney P. Effect of the solution refractive index on the color of gold colloids. Langmuir. 1994;10:3427–30. Ung T, Liz-Marzán LM, Mulvaney P. Controlled method for silica coating of silver colloids. Influence of coating on the rate of chemical reactions. Langmuir. 1998;14:3740–8. Vaz AM, Serrano-Ruiz D, Laurenti M, Alonso-Cristobal P, Lopez-Cabarcos E, Rubio-Retama J. Synthesis and characterization of biocatalytic γ-Fe2O3@SiO2 particles as recoverable bioreactors. Colloids Surf B. 2014;114:11–9. Verhaegh NAM, van Blaaderen A. Dispersions of rhodamine-labeled silica spheres: synthesis, characterization, and fluorescence confocal scanning laser microscopy. Langmuir. 1994;10:1427–38. Vuu K, Xie J, McDonald MA, Bernardo M, Hunter F, Zhang Y, Li K, Bednarski M, Guccione S. Gadolinium-rhodamine nanoparticles for cell labeling and tracking via magnetic resonance and optical imaging. Bioconj Chem. 2005;16:995–9. Wang H, Nakamura H, Yao Y, Maeda H, Abe E. Effect of solvents on the preparation of silicacoated magnetic particles. Chem Lett. 2001;30:1168–9. Wang H, Zheng L, Peng C, Shen M, Shi X, Zhang G. Folic acid-modified dendrimer-entrapped gold nanoparticles as nanoprobes for targeted CT imaging of human lung adenocarcinoma. Biomaterials. 2013;34:470–80. Wu Z, Xiang H, Kim T, Chun M-S, Lee K. Surface properties of submicrometer silica spheres modified with aminopropyltriethoxysilane and phenyltriethoxysilane. J Colloid Interface Sci. 2006;304:119–24. Wu Z, Liang J, Ji X, Yang W. Preparation of uniform Au@SiO2 particles by direct silica coating on citrate-capped Au nanoparticles. Colloids Surf A. 2011;392:220–4. Yilmaz H, Sato K, Watari K. AFM interaction study of α-alumina particle and c-sapphire surfaces at high-ionic strength electrolyte solutions. J Colloid Interface Sci. 2007;307:116–23. Yoshino K, Nakamura K, Terajima Y, Kurita A, Matsuzaki T, Yamashita K, Isozaki M, Kasukawa H. Comparative studies of irinotecan-loaded polyethylene glycol-modified liposomes prepared using different PEG-modification methods. Biochim Biophys Acta. 2012;1818:2901–7. Yu SM, Choi SH, Kim SS, Goo EH, Ji YS, Choe BY. Correlation of the R1 and R2 values of gadolinium-based MRI contrast media with the ΔHounsfield unit of CT contrast media of identical concentration. Curr Appl Phys. 2013;13:857–63. Zhang Z, Patel RC, Kothari R, Johnson CP, Friberg SE, Aikens PA. Stable silver clusters and nanoparticles prepared in polyacrylate and inverse micellar solutions. J Phys Chem B. 2000;104:1176–82.

38

Y. Kobayashi and K. Gonda

Zhang L, Wang H, Zhang Z, Qin F, Liu W, Song Z. Preparation of monodisperse polystyrene/silica core-shell nano-composite abrasive with controllable size and its chemical mechanical polishing performance on copper. Appl Surf Sci. 2011;258:1217–24. Zhao Y, Tao Z, Xu Z, Tao Z, Chen B, Wang L, Li C, Chen L, Jia Q, Jia E, Zhu T, Yang Z. Toxic effects of a high dose of non-ionic iodinated contrast media on renal glomerular and aortic endothelial cells in aged rats in vivo. Toxicol Lett. 2011;202:253–60. Zhou X, Kobayashi Y, Romanyuk V, Ohuchi N, Takeda M, Tsunekawa S, Kasuya A. Preparation of silica encapsulated CdSe quantum dots in aqueous solution with the improved optical properties. Appl Surf Sci. 2005;242:281–6.

Hierarchical Organization in Monolithic Sol–Gel Materials €sing Andrea Feinle, Michael S. Elsaesser, and Nicola Hu

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sol–Gel Processing to Yield Porous Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phase Separation and Templating Strategies Combined With Sol–Gel Processing . . . . . . . . . Emulsions and Foams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ice Templating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hierarchically Organized Porous Materials: Selected Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Silica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Non-siliceous Monoliths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Monoliths With Hierarchical Pore Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymerization of Organic Monomers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phase Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emulsion Templating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Colloidal Crystal Templating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ice Templating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 4 6 8 10 11 11 18 31 34 35 38 39 40 40 42

Abstract

The development of synthetic routes to hierarchically organized porous materials containing multiple, discrete sets of pores having disparate length scales is of high interest for a wide range of applications. One possible route toward the formation of multilevel porous architectures relies on the processing of condensable, network-forming precursors (sol–gel processes) in the presence of molecular porogens, lyotropic mesophases, supramolecular architectures, emulsions, A. Feinle • M.S. Elsaesser • N. H€ using (*) Department Chemistry and Physics of Materials, Paris Lodron University Salzburg, Salzburg, Austria e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_127-1

1

2

A. Feinle et al.

organic polymers, or ice. In this review the focus is on sol–gel processing of inorganic and organic precursors with concurrently occurring microscopic and/or macroscopic phase separation for the formation of self-supporting monoliths. The potential and the limitations of the solution-based approaches are presented with special emphasis to recent examples of hierarchically organized silica, metal oxides, and phosphates as well as carbon monoliths.

Introduction Research on complex and hierarchically organized porous materials has seen tremendous progress in the last decades, and the field is still rapidly evolving (Su et al. 2012). As a result huge progress has been made in the development of synthetic approaches toward porous materials that exhibit interconnected pore dimensions on several length scales, from molecular (100 nm). Pores smaller than 2 nm are typically termed micropores, pores with sizes between 2 and 50 nm mesopores, and pores larger than 50 nm are macropores (McCusker et al. 2001). Such multilevel porous architectures confer unique properties to materials depending on the combination of pore sizes, e.g., micro- and mesopores impart high surface areas and pore volumes providing size and shape selectivity and large interfacial areas, while larger pores (>50 nm) reduce transport limitations in the material and facilitate mass transport to the active sites. A variety of preparation techniques have already been reported for the preparation of micro-/macroporous, micro-/mesoporous, meso-/macroporous, or micro-/meso-/macroporous materials (Dong et al. 2002; Sen et al. 2003; Kuang et al. 2004; Gawel et al. 2010; Lopez-Orozco et al. 2011; Sun et al. 2011; Triantafillidis et al. 2013; Depardieu et al. 2015) with great potential for applications in the fields of catalysis, sorption, separation, energy storage and conversion, sensing, and biomedicine, i.e., medical diagnostics or therapies (Su et al. 2012). However, the applicability of a material depends not only on its pore sizes and size distributions but also on structural characteristics, such as the total amount of pores, the accessibility of the pores (ratio of closed to open pores), tortuosity and interconnectivity, gradients, etc., and, very importantly, the chemical composition as well as the processability in terms of shaping (films, fibers, monoliths, etc.). Shaping of the material is for many applications an inevitable requirement. To name just some of the advantages of highly porous, macroscopic monolithic materials, they can give lower back pressures, a higher permeability, and better performance in flow-through catalytic or separation systems (Siouffi 2006). Well-controlled top-down and bottom-up self-assembly techniques providing a high level of structural control have been reported as very elegant approaches for the synthesis of hierarchically organized porous powders, particles, and monoliths (Yuan and Su 2006; Colombo et al. 2010; Su et al. 2012; Inayat et al. 2013). All these methods can in principle roughly be divided in the following categories: (1) posttreatment, starting from porous/nonporous objects and introducing a first, second, or third level of porosity by, e.g., selective leaching processes (Inayat

Hierarchical Organization in Monolithic Sol–Gel Materials

3

et al. 2013); (2) typical ceramic processing, e.g., starting from powders, including sintering, foaming, and/or leaching (Galassi 2006); (3) synergetic solution-based processes, such as co-assembly of molecular precursors with or without “soft” templates, such as polymers, surfactants, emulsions, etc. (Colombo et al. 2010; Triantafillidis et al. 2013); and (4) transcriptive processes using pre-organized or self-assembled molecular, supramolecular, or solid molds (organic, biological, or inorganic) as templates, also termed “nanocasting” or exotemplating (Sch€uth 2003; Sanchez et al. 2005; Petkovich and Stein 2013). Due to the nature of this review, we primarily limit the discussion to solutionbased processes toward the preparation of monolithic materials of macroscopic dimensions with well-controlled pore sizes and pore orientations over multiple length scales. As a central topic, the simultaneous processing of condensable precursors in the presence of molecular porogens, lyotropic mesophases, supramolecular architectures, emulsions, organic polymers, ice, etc. is discussed. The possibilities in tailoring such multilevel porous architectures arising from concurrent microscopic and macroscopic phase separation in sol–gel systems with the ongoing competition between the timing of gelation versus phase separation will be covered. With that a detailed description of nanocasting or “hard” templating approaches will be beyond the scope of this Chapter, and only a few examples can be found throughout the text. The next sections will first briefly highlight the fundamentals of sol–gel processing, phase separation, and other often applied templating schemes, followed by more recent key examples for silica-based materials as well as non-silica oxides or phosphates and carbon-based structures. These examples have been chosen to illustrate the general applicability of the synthetic routes toward these hierarchically organized materials and give an overview over the range of accessible materials. Many more examples can be found in the literature, and the list of materials is extended every day. We hope this chapter will not only be useful for experienced researchers already working in the field but also for encouragement of others to enter this exciting area of research with new ideas.

Fundamentals Before beginning the exploration of combined phase separation and templating strategies with sol–gel processing, a few remarks on the materials under discussion and the network-forming processes need to be made. As mentioned above, we will focus the discussion on monolithic materials. As defined by IUPAC, “a monolith is a shaped, fabricated, intractable article with a homogeneous microstructure that does not exhibit any structural components distinguishable by optical microscopy” (McNaught and Wilkinson 1997). In this chapter all materials have an interconnected porosity on multiple length scales within the monolithic structure as a second feature in common. Depending on the synthetic strategies, the pores might be arranged and connected in very different ways as schematically shown in Fig. 1.

4

A. Feinle et al.

Fig. 1 Concept of hierarchy in a porous material: Schematics of a hierarchical porous buildup from the micrometer scale (inner blue circle), via the mesoscopic regime (orange circle) to the macroscopic porous dimension (light blue circle) within a monolithic material: In the macroporous regime, the different pore arrangements, such as inverse opal-like structures (a), isolated pores (b), co-continuous porosity (c), and a cellular buildup (d), are shown (in the scheme, the blue part represents the solid network, and the white part is the pore space). For the mesoporous regime, either well-organized pores with monomodal character as shown for a 2D hex structure or disordered arrangements are possible

Sol–Gel Processing to Yield Porous Materials The sol–gel process is a method to produce a solid material (the gel) from molecular precursors via the formation of colloidal particles (the sol) (Brinker and Scherer 1990). Condensation reactions of hydrolyzable precursors, e.g., metal or semimetal alkoxides, but also salts, induced by the controlled addition of water represent the key steps in the synthesis of monolithic materials. The network evolves via nucleation and growth of nanometer-sized sol particles as well as their aggregation. A large variety of parameters, such as the choice of the precursor, its concentration, pH value, temperature, solvent, etc., can be adjusted to deliberately tailor the final network morphology, the network chemistry, and the pore structure (ranging from polymeric to particulate network buildup of either small or large particles, etc.). With

Hierarchical Organization in Monolithic Sol–Gel Materials

5

Fig. 2 Analogy in the condensation-based network formation reactions starting from silicon alkoxides or resorcinol/formaldehyde

that the homogeneity or even heterogeneity of the network is adjusted. This process is routinely used in the formation of oxidic and even hybrid inorganic–organic oxidic networks (Hench and West 1990; Schubert et al. 1995; Corriu and Leclercq 1996) but can also be found in an analogous manner in purely organic systems, i.e., resorcinol–formaldehyde polymers (Pekala et al. 1992; Al‐Muhtaseb and Ritter 2003; Elkhatat and Al‐Muhtaseb 2011). Figure 2 schematically shows the similarity of the basic chemical reactions for metal alkoxides as well as resorcinol–formaldehyde and the resulting network formation. Micro- and mesoporosity is an inherent feature of amorphous gels prepared by sol–gel processing (Brinker and Scherer 1990; Al‐Muhtaseb and Ritter 2003; Antonietti et al. 2014). As described above, the network is build up from aggregated particles, whose size, number and density in the given volume are adjusted by the synthetic conditions. The solvent space between the solid network represents the potential pore space after drying. Thus, the critical step determining the porous character of the final dried material is the removal of the solvent, which is especially true when monoliths are prepared. Drying of large monolithic pieces is often difficult, since surface tension and evolution of capillary pressures can result in large shrinkages or even destruction of the whole gel body. One typical procedure to

6

A. Feinle et al.

prevent cracking and collapse of the gel body is drying with supercritical fluids (scf), e.g., carbon dioxide or alcohols, since the building up of a gas–liquid interface is avoided; hence no capillary pressures evolve (Kistler 1931; H€using and Schubert 1998). This process is routinely used in the preparation of mesoporous materials, such as aerogels of variable composition with porosities as high as 97 % and statistically distributed pore sizes in the upper mesoporous range. This process can also be applied for monolithic systems with hierarchical organization of pores. However, scf extraction is expensive and time-consuming and requires high pressures, sometimes even combined with high temperatures. This process can also be applied for monolithic systems with a hierarchical organization of pores. Another approach for drying hierarchically organized porous monoliths has been presented by Mukai et al. via freeze gelation and freeze-drying (Mukai et al. 2004; Nishihara et al. 2006). Freeze-drying can also be applied to purely mesoporous bodies; however in many cases the monolithic structures cannot be fully retained. A very promising procedure for drying large silica gels relying on a simple surface modification treatment with trimethylchlorosilane was presented in the mid-1990s as an alternative to supercritical drying (Smith et al. 1995). The capillary pressure, Pc, generated during drying is a function of the pore fluid–vapor surface tension, γLV, the contact angle, θ, between the fluid–vapor interface and pore wall, and the pore radius, a, as follows: Pc = (2γLVcosΘ)/a. For a wetting fluid (θ < 90 ), Pc is negative, thus indicating that the fluid is in tension. The presence of organic methyl groups on the surface of the silica gel in combination with a proper selection of the final pore fluid, e.g., hexane, allowed to change the contact angle to lower capillary pressures and thus dry monolithic wet silica gels without cracking (Smith et al. 1995). However, due to the requirement of surface silylation, this process is mostly limited to silica-based monoliths.

Phase Separation and Templating Strategies Combined With Sol–Gel Processing Only when the porogen shows a univocal relationship between its own structure and the final porous structure, it can be termed a template. In most synthetic approaches occurring in solution and relying on phase separation on different length scales, this clear relationship is not given. Even in zeolite synthesis, the porogen typically is not a true template, but more a structure-directing agent. In other words, a template is mostly a “hard” object which does not significantly alter its shape when the solid counterpart is being formed. Because of the strong kinetic control and cooperative nature of the processes that result in mesopore and/or macropore formation via “soft”-templating routes as discussed in this chapter, the term “structure-directing agent” is to be preferred to “template.” The sol–gel process is a dynamic process, in which the ongoing condensation reactions (cross-linking) of the mostly hydrophilic precursor molecules/oligomers result in solidification of the network. In principle, any kind of structure-directing agent can be added as porogen in this solidification process to induce some kind of

Hierarchical Organization in Monolithic Sol–Gel Materials

7

phase separation. Already the nucleation and growth of the sol particles can be considered as a type of phase separation forming two heterogeneous phases: a solid network and a solution phase. This can be extended on the microscopic, but also macroscopic length scale by the addition of molecules, polymers, or supramolecular arrangements that enforce demixing. With the progress in understanding the hydrolysis and condensation reactions, the deliberate design of the porous structures by different phase separation strategies advanced more and more, e.g., by increasing the relative volume of the hydrophobic components within the porogens, the characteristic size of the pore dimensions can be increased from less than 1 nm to tens of micrometers and larger. Examples of suitable structure-directing agents/templates include molecular species as used in zeolite synthesis (Cundy and Cox 2003), lowmolecular-weight and block copolymer surfactants (Wan and Zhao 2007), emulsions (Studart et al. 2006), and/or solid particles (Petkovich and Stein 2013). For materials comprising multiple levels of pore sizes (bimodal such as micro–meso, meso–macro, or even trimodal such as micro–meso–macro, meso–meso–macro, etc.), in principle, a combination of the above mentioned templating strategies is possible. This would mean that mixtures of, i.e., molecules, polymers, or supramolecular arrays are added to the gelling solution, with the intrinsic difficulty in the preservation of the existing levels of organization upon introducing another one. The major challenges arising in the preparation of such bi- or multimodal micro-/meso-/macroporous materials are: (1) to avoid macroscopic demixing of the components, (2) to avoid the formation of significant proportions of closed pores, (3) to control the different pore sizes independently, and (4) to manage shrinkage of the whole structure while retaining the macroscopic shape. As one example for such a phase separation process: Monolithic materials with well-defined, co-continuous porous structures on multiple levels can be obtained by combining liquid–liquid phase separation and sol–gel processing. The phase separation process is induced by the presence of a porogen, which is in many cases a water-soluble polymer such as poly(ethylene oxide) or poly(acrylic acid). Nakanishi and Soga were the first to prepare monolithic silica with interconnected macropores and textural mesoporosity by the addition of poly (sodium styrene sulfonate) to a silica sol–gel mixture, and they could clearly show that the formation of different biphasic morphologies (isolated pores, particle aggregates, interconnected continuous pores) is induced by the polycondensation reaction of the network-forming silica species and is finally irreversibly frozen by the sol–gel transition (Nakanishi and Soga 1991). Therefore, all parameters that change the relative rates of phase separation versus gelation will have a profound influence on the architectural properties of the final gel, including mesoporosity, interconnected macroporosity, and the degree of macroscopic phase separation. A very detailed discussion on the topic of phase separation would be beyond the scope of this chapter, but the reader is referred to some excellent review articles, the original work of Cahn and Hilliard or the Handbook of Porous Solids, all providing a profound summary in the context of porous materials (Cahn and Hilliard 1958; Weitkamp et al. 2002; Nakanishi and Tanaka 2007; Kanamori and Nakanishi 2011).

8

A. Feinle et al.

Fig. 3 Time evolution of phase-separated domains (Reprinted with permission from reference Konishi et al. (2006). Copyright (2006) American Chemical Society)

As has been shown by Nakanishi et al., if a phase separation process is occurring concomitantly to the gelation process, a certain structure is irreversibly frozen in (just like a “snapshot” in time of the heterogeneity). Depending on the timing between phase separation and sol–gel transition as well as the stability of the different heterogeneous phases, different structures will be obtained as shown in Fig. 3. As typical for phase separation phenomena relying on spinodal decomposition of a system, the larger the time difference between phase separation and gelation, the coarser the structure will become, sometimes even breaking up into fragmented particles (in this case no monolithic structure is obtained). Thus, all parameters resulting in faster sol–gel transitions, e.g., higher temperatures, pH value changes, water–precursor ratio, etc., influence the final structure. This phase separation strategy can be complemented by variations on the sol–gel precursor side. Not only alkoxides but also inorganic salts, tailor-made precursor molecules, or even nanoscale building blocks can be used to influence the phase separation tendency in a sol–gel solution. Depending on the desired chemical composition of the final material, an almost unlimited choice of precursors is available and will be discussed in more detail below. To mention just one example, the formation of hierarchical structures involving the assembly of preformed inorganic nanoparticles into materials with higher-order architectures has been applied for oxides derived from highly reactive molecular precursors, e.g., alumina (Davis et al. 2001). These methods are often also found in the literature by using the term “nanotectonics.”

Emulsions and Foams The basic idea behind emulsion templating is to use sol–gel processing to deposit an inorganic material at the exterior of emulsion droplets. Imhof and Pine were the first to show the applicability of this method in the formation of macroporous oxides (Imhof and Pine 1997).

Hierarchical Organization in Monolithic Sol–Gel Materials

9

Fig. 4 Schematic diagram of emulsions and high internal phase emulsions (HIPEs). (top) Normal oil-in-water (O/W) and the reverse (W/O) emulsion; (bottom) high internal phase emulsion

An emulsion is a two-phase mixture (droplet and continuous) of immiscible fluid phases in which one is dispersed in a second in the form of droplets (Fig. 4). Generally there are two types of emulsion: oil-in-water (O/W) emulsion where the droplet phase is an organic solvent while the continuous phase is water and water-in-oil (W/O) emulsion where water or an aqueous solution is the droplet phase with an organic continuous phase. To form an emulsion, a suitable surfactant (emulsifier) is generally required to stabilize the droplets dispersed in the continuous phase. In order to use emulsions as templates in the preparation of porous oxides, the ceramic precursor sol is added to the continuous phase (mostly aqueous for the synthesis of monoliths), and gelation is induced by, i.e., changing the pH value. Upon polymerization the oxide structure is frozen while the emulsion structure is maintained, and, upon removal of the emulsion phase, a macroporous, mostly closed cell, structure is produced. Mesopores can be generated when a structure-directing agent is co-added to the ceramic precursor sol. In this case, hierarchically organized meso-/macroporous objects are obtained as nicely reviewed recently for organic polymers by Silverstein et al. (Silverstein 2014). Emulsion templating takes advantage of the fact that oil droplets – compared to solid particulate templates – are: (1) highly deformable to allow

10

A. Feinle et al.

the inorganic gel to accommodate large shrinkages and thus prevents cracking during drying, (2) can yield architectures on a scale ranging from 5 to 600 μm, and (3) the emulsion droplets are easily removable by evaporation, extraction, or calcination. Compared with the often applied hard-templating approach, the combination of using mesostructure-directing agents together with an emulsion has the advantage that the emulsion droplet size can be adjusted by changing the emulsification conditions, and the use of block copolymer species, which selfassemble to a large extent independently of the emulsion formation, allows tailoring the macro- and mesopore size. The number of droplets in an emulsion can be varied, normally expressed as the volume ratio of the droplet phase to the continuous phase or the volume percentage of the droplet phase in the emulsion. The droplet volume fraction can even exceed the close-packing limit of 74 %, which corresponds to the most compact arrangement of uniform undistorted spherical droplets; this type of emulsion is called high internal phase emulsion (HIPE) in which the structure consequently consists of deformed and/or polydispersed droplets (Fig. 4). The droplets are separated by a thin continuous phase, and a structure resembling gas–liquid foams is formed. Consequently interconnected macroporous structures can be prepared with HIPE systems. Emulsion templating, especially using HIPEs, is typically situated at the borderline between soft and hard templating. Two cases have to be distinguished: On one hand the purely liquid state, in which the emulsion-forming species are in a dynamic equilibrium, and on the other, emulsions, in which the continuous or droplet phase has been polymerized prior to the actual solidification of the continuous network, thus forming a solid foam (Studart et al. 2006; Alvarez and Fuertes 2007).

Ice Templating Freeze-casting or ice templating takes advantage of the growth of ice crystals to template molecular, high-molecular-weight precursors, or colloidal suspensions in either a water solution, a suspension, or a hydrogel followed by sublimation of the ice phase (Deville 2008, 2013; Gutiérrez et al. 2008). The formation of crystalline ice causes the substances originally dispersed in the aqueous medium to be expelled to the boundaries between adjacent ice crystals. High-vacuum sublimation of the ice by subsequent freeze-drying gives rise to cryogels (even monolithic ones) with macroporous structures, which are replicas of the original ice structure. When compared with conventional methods, ice templating has many advantages, such as: Flawless components with monolithic shape can be produced; it does not require the addition of special templates which usually lead to high production costs and require severe removal processes (e.g., calcination and chemical etching, using a strong base). Since the solidification is often directional, the porous channels run from the bottom to the top of the samples (Fig. 5). The final porosity content can be tuned by varying the particle content within the slurry, and the size of the pores is affected by the freezing kinetics.

Hierarchical Organization in Monolithic Sol–Gel Materials

11

Fig. 5 The four processing steps of freeze casting: slurry preparation, solidification, sublimation, and sintering (Reprinted with permission from reference Deville (2008). Copyright (2008) WileyVCH Verlag GmbH & Co. KGaA, Weinheim)

The variety of materials processed by ice templating suggests that the underlying principles of the technique are not strongly dependent on the materials but rely more on physical rather than chemical interactions.

Hierarchically Organized Porous Materials: Selected Examples Silica The formation of silica monoliths comprising either, micro-, meso-, or macropores is well known for several years. The most prominent macroporous example is Vycor glass, which is prepared by phase separation of a melt-quenched metastable glass that is reheated close to its glass-transition temperature followed by a leaching process (Enke et al. 2003). Mesoporous silica monoliths are well known from silica aerogels (H€ using and Schubert 1998), and zeolite monoliths exhibiting micro- and macropores are accessible by, i.e., pseudomorphic transformation reactions to name only some examples (Sachse et al. 2011). In the early nineties two pioneering discoveries resulted in a boost in the development of high surface area materials with controlled porosity. On one hand,

12

A. Feinle et al.

two groups in the USA and Japan independently published work on porous silica templated with surfactants with a periodic ordering in the mesoscopic regime (Yanagisawa et al. 1990; Kresge et al. 1992), and on the other in 1991 Nakanishi and Soga published the first paper on a meso-/macroporous material prepared via a sol–gel route accompanied by a phase separation process (Nakanishi and Soga 1991). Both processes have seen extensive progress in the last decades, and in many cases combinations of both – soft templating with surfactants or block copolymers and phase separation – in sol–gel system result in the desired pore structure.

Phase Separation by Addition of Polymers In a typical synthesis, a tetraalkoxysilane is mixed with water in ratios of water/Si > 4 (in a solvent), and a polymeric phase separation agent is added, e.g., poly(ethylene oxide), poly(acrylic acid), etc. In simple terms, this phase separation process – and with that the macroporous structure – is governed by the interaction between the condensing precursor molecules, the polymeric species, and the solvent. This has been well investigated for silica-based systems, showing that polymers not having any specific attractive interactions with silanols, e.g., poly(acrylic acid) or poly (sodium styrene sulfonate), stay in the solvent phase and thus directly relate to the volume fraction of macropores in the dried material (Nakanishi and Soga 1991, 1992). Polymers that strongly interact with the growing silica network by, i.e., hydrogen bonding interactions, such as poly(ethylene oxide) or cationic surfactants, are typically distributed in the gel phase. In this case, the macropore volume fraction is more or less only determined by the amount of solvent, but the domain size and tendency of phase separation are influenced by the polymeric additive (Nakanishi 1997). If the timing and the dynamics driven by the interfacial energy between phase separation and sol–gel transition are chosen properly, bicontinuous gels constituted by two interconnected phases on the micrometer length scale are formed, one being rich in silica, the other one being rich in solvent. After removal of the solvent and drying, structures with macropores resembling the solvent phase and solid architectures comprising textural mesoporosity in the 10–20 nm range and high specific surface areas are obtained. Keeping in mind that hydrogen bonding results in polymers strongly interacting with the gel phase, a natural extension of the described process is the addition of polyether-based or cationic surfactants/block copolymers to the sol–gel mixture to tailor the mesoporous structure of the materials. Smått et al. (Smått et al. 2003) as well as the group of Nakanishi (Sato et al. 2001; Nakanishi et al. 2003) successfully added different kinds of surfactants (triblock copolymers based on poly(ethylene oxide) and poly(propylene oxide) units or cetyltrimethylammonium bromide) in a double-templating approach. While Smått et al. (2003) relied on a combination of a homopolymeric phase separation agent (PEO) and a cationic surfactant (CTAB) to obtain a hierarchically organized silica monolith with small mesopores (~3 nm) and macropores of 0.5–35 μm, Nakanishi solely used the amphiphilic block copolymer or cationic surfactant as phase separation and supramolecular templating agent. In the latter cases, the mesopores exhibit a certain degree of long-range ordering.

Hierarchical Organization in Monolithic Sol–Gel Materials

13

Fig. 6 TEM image (left), SEM image (middle), and photograph of a silica monolith prepared from EGMS in the presence of a nonionic block copolymer (P123) at pH = 1

A key problem in the combined sol–gel processing and phase separation/ supramolecular templating strategy toward materials with a periodic arrangement of the mesopores lies in the presence of the low-molecular-weight alcohol that is released upon hydrolysis of, e.g., tetraethoxy- or tetramethoxysilanes. Many of the supramolecular arrangement of block copolymers or surfactants are not compatible to higher concentrations of these alcohols as shown by Alexandridis et al. (Ivanova et al. 2000b). H€ using et al. avoided this problem by applying tetrakis(2-hydroxyethyl) orthosilicate as the silica source instead of TEOS or TMOS (H€ using et al. 2003). Here, ethylene glycol is released upon hydrolysis of the silane, which has been proven to be compatible with a variety of lyotropic surfactant phases (Ivanova et al. 2000a, 2001). Processing of this glycol-based silane in the presence of an amphiphilic triblock copolymer surfactant (Pluronic P123 ®) gave hierarchical macro-/mesoporous silica monoliths with ordered mesopore organization after supercritical extraction with carbon dioxide. This work has been extended to a variety of diol- and polyol-modified silanes and typically networks with macropores between 500 nm to 5 μm, and periodically arranged, uniform mesopores of about 5–10 nm are obtained (Fig. 6; Brandhuber et al. 2005; Hartmann et al. 2007; Triantafillidis et al. 2013). It is again noteworthy to mention that the key point in the preparation of these hierarchically organized silica monoliths lies in the timing of the concurrently occurring phase separation and gel formation processes. Thus, the network structure is not only influenced by the type of diol that has been used to modify the silane (Brandhuber et al. 2005) but also by the choice of acid that is used to start the sol–gel reactions (Flaig et al. 2015) or the amphiphilic molecule that is added as phase separation agent (Hartmann et al. 2014).

14

A. Feinle et al.

Other precursor systems that have been applied in the synthesis of silica monoliths with bimodal pore systems are based on triethanolamine solution with the corresponding silatrane silane derivatives (El Haskouri et al. 2002). Monoliths with 4 nm mesopores and a second level of pores with sizes of 30–60 nm are obtained in the presence of the cationic surfactant CTAB. Instead of liquid precursors, preformed, high surface area, mesoporous silica nanoparticles could also be used to prepare hierarchically organized materials (Huerta et al. 2007). Many more examples can be found by applying combinations of supramolecular arrangements and hard templates to achieve a hierarchical organization (Petkovich and Stein 2013). In principle, dual or even multiple micellar-templating approaches based on two or more different structure-directing agents, such as surfactants, block copolymers, or ionic liquids, could be used in the formation of monoliths with multimodal pore sizes (Sel et al. 2006). However, the different structure-directing agents will show a rather complex mixing behavior, thus becoming very difficult to control, and the delicate interaction behavior between the different species will determine whether hierarchical structures are formed or not. So far, the formation of monolithic silica has not explicitly be mentioned; however, the process allows for the formation of different morphologies as presented by Yuan et al. (Yuan et al. 2010). Not only purely silica-based monoliths can be prepared by the abovementioned approaches but also inorganic–organic hybrid materials are accessible (H€using et al. 2006; Hartmann et al. 2007; Kanamori and Nakanishi 2011). Co-condensation reactions of tetraalkoxysilane or tetrakis(2-hydroxyethyl) orthosilicate with organo-functional trialkoxysilanes, such as methyl-, phenyl-, vinyl-, aminopropyl, methacryloxypropyl-, or glycidoxypropyl-derivatized ones to name only a few, are routinely used in the preparation of functionalized silica-based materials. However, one has to keep in mind that changes in polarity in the sol might result in completely different phase separation behaviors, thus resulting in different porous network structures. A very detailed investigation on the formation of hybrid chloroalkyl-modified, meso-/macroporous silica monoliths and the structural changes observed due to the presence of the organofunctional silane and its organic chain, as well as due to post-synthetic processes, such as azide–alkyne Click reactions, has been presented by Keppeler et al. (Keppeler et al. 2011, 2015; Keppeler and H€ using 2011). In addition to co-condensation reactions, pure silsesquioxane-based hybrids are accessible via condensation reactions of the sole organotrialkoxysilane or bis(trialkoxysilyl) precursors, such as 1,4-bis[tris(2-hydroxyethoxy)silyl]benzene or the corresponding alkoxy derivatives (Brandhuber et al. 2006; Kanamori and Nakanishi 2011). Even dendrimeric silanes as well as cyclic preceramic precursors, such as a glycol-modified 1,3,5-trisilacyclohexane carbosilane, have been used in the formation of meso-/macroporous hybrid monoliths (Weinberger et al. 2008, 2010).

Emulsions and Foams Dual meso–macroporous silica monoliths from polymer foams have been presented by the group of Chmelka in 2003 (Maekawa et al. 2003). However, this is an

Hierarchical Organization in Monolithic Sol–Gel Materials

15

Fig. 7 SEM visualization of the inorganic monolith-type material macrostructure. (a) and (b) 1Si-HIPE, (c) and (d) 2Si-HIPE, (e) and (f) 3Si-HIPE (Image taken from reference Carn et al. 2004a)

example, in which styrene first was prepolymerized in an emulsion to give a foam that was soaked in a second step with an acidic silane-based sol–gel solution containing amphiphilic block copolymer species as structure-directing agents. Silica monoliths with cellular macropores (0.3–2 μm) comprising 0.2–0.5 μm cell windows and highly ordered mesopores (5.1 nm) were obtained. A true emulsion-based approach was used by Sen et al., who reported the formation of mesoporous silica, meso-cellular silica foams (MCFs), macro-cellular silica foams (UMCFs) and ordered macroporous silica in a one-pot synthesis at room temperature. However, no comment on the macroscopic morphology of the material was made. At very low oil concentration with slow stirring mesoporous silica was obtained, whereas meso-cellular silica foams were formed with faster stirring. Syntheses using intermediate-to-high oil concentration produced macroporous solids with various pore sizes and wall thicknesses. Upon increasing the pH from acidic to neutral to basic, the macroporous structure starts to disappear, and a mesoporous solid is formed (Sen et al. 2005). The group of Backov applied concentrated emulsions, so-called HIPE systems prepared from dodecane in the presence of an aqueous tetradecyltrimethylammonium bromide (TTAB)–tetraethoxysilane mixture for the preparation of highly porous silica monoliths (Fig. 7; Carn et al. 2004a). TTAB serves as a mesoscopic texturing agent. They obtained hierarchically organized materials with very low densities, vermicular-type mesoporosity, and macropore sizes in the range

16

A. Feinle et al.

Fig. 8 Scanning electron micrographs (SEMs) of meso-/macroporous silica spongelike networks as well as a photograph of the spongelike network formed in the presence of hydrogen peroxide (With kind permission of Springer Science+Business Media; reference Vuong et al. 2008)

of 1–100 μm that they labeled as Si (HIPE). The macroscopic void space could be varied by varying the starting oil volume fraction of the O/W concentrated emulsion; however the texture always resembled hollow spheres. Even air–liquid interfaces in foam structures have been applied in the formation of monolithic materials comprising multiple levels of pore sizes. Carn et al. (2004b) could show that foams give a high level of control over macropore characteristics, such as size, topology (open versus closed cell), and morphology (spherical versus polygonal cell structures). The foaming solution consisted of colloidal silica that has been prepared via the Stöber method and a cationic surfactant in an aqueous medium (pH = 9); the foam was generated by continuous bubbling of perfluorohexanesaturated nitrogen through a porous glass disk. Another unusual approach toward meso-/macroporous silica monoliths has been presented by Vuong et al., who used the release of oxygen gas from hydrogen peroxide decomposition through a silica gel with low viscosity that has been prepared in the presence of a nonionic surfactant. The escape of the oxygen bubbles results in a significant and rapid expansion of the gel body and a spongelike silica monolith exhibiting periodically ordered mesopores within grains of 10–20 μm in diameter, and wall thicknesses of 0.5 μm was obtained (Fig. 8; Vuong et al. 2008).

Ice Templating As described above, ice crystals can be used as macrotemplates, as demonstrated by Nishihara et al. who prepared ordered macroporous materials with micro–mesoporosity by thermally induced phase separation (Nishihara et al. 2005). By this method, it is possible to precisely control the macroporosity, wall thickness, and micro-/mesoporosity of silica materials via very simple procedures.

Hierarchical Organization in Monolithic Sol–Gel Materials

17

Fig. 9 Morphology and structure of silica monoliths exhibiting a micro-honeycomb structure. Upper: (a) An overall image. SEM images of (b) cross section, (c) microchannel structure, and (d) longitudinal section. Lower (a) Detail of a cross section. Nitrogen isotherms (b) are also represented. The inset shows the mesopore size distribution in the desorption branches (Reprinted with permission from reference Mukai et al. 2004. Copyright (2004) The Royal Society of Chemistry)

Tamon and co-workers successfully produced silica monoliths that were not only macroporous (the cell size of the micro-honeycomb structure ranges between 10 and 15 μm) but also meso- and microporous (the BET surface area ranges between 400 and 700 m2/g). Faster immersion rates of the gel in the cold bath produces

18

A. Feinle et al.

smaller macropores. It is noted that the microporosity is simply a consequence of the voids left between silica colloids packaged at the boundaries of adjacent ice crystals (Fig. 9) and can be adjusted by the pH of the parent silica sol.

Non-siliceous Monoliths The methods for preparing non-siliceous materials are in principle similar to those for preparing silica monoliths. However, a one-to-one transfer is not possible due to the typically higher reaction rates of the metal precursors (alkoxides and metal salts) and the tendency of metal oxides to crystallize at relatively low temperatures. The first mesoporous non-siliceous materials were reported by Ying et al. in 1995 (Antonelli and Ying 1995). Since then, the synthesis of mesoporous non-siliceous metal oxides and mixed oxides has seen major progress. The most common synthetic procedure to these materials is based on hard templating (nanocasting) or colloidal crystal templating by using preformed silica or carbon materials. While some excellent reviews and papers concerning these synthetic approaches are available (Lu and Sch€ uth 2005; Smått et al. 2006, 2012; Ren et al. 2012; Petkovich and Stein 2013), only few publications focusing on the preparation of non-siliceous monoliths with hierarchical porosity via sol–gel processing directly from solution can be found. In this chapter we give an overview of the progress in this topic in the last years and introduce a few recent examples. The main difficulties in the synthesis of porous, non-siliceous, hierarchically organized monoliths are: (i) to control the high reactivity of the metal precursors, e. g., metal alkoxides or salts, (ii) to preserve the monolithic form during calcination processes, and (iii) to simultaneously control the resulting crystallinity and porosity. With respect to the first point, metal alkoxides are stronger Lewis acids than silicon alkoxides and thus facilitate the nucleophilic attack of water or other molecules. Furthermore, most metals have several stable coordination numbers and are therefore often present with a coordinatively unsaturated valence state. Both effects dramatically increase the reactivity and thus make it difficult to control homogeneous gelation. As a result, precipitates instead of monolithic materials are often obtained. Today, various approaches to moderate the reactivity of metal alkoxides have been reported. One possibility is to add chelating agents, e.g., acetylacetone (Dutoit et al. 1995) or carboxylic acids (Takenaka et al. 2000) to attenuate the reactivity of the titanium precursor by replacing part of the alkoxy groups. Therefore, the coordination state of titanium is stabilized, and the reactivity toward nucleophilic attacks is lowered. Another possibility to tackle this problem is the addition of strong acids. At low pH, the particle surfaces are positively charged, and condensation is slowed down due to electrostatic repulsion between equally charged particles. If the pH is raised gradually, e.g., by the addition of formamide, the electrostatic repulsion is reduced, and the particles can aggregate until the formation of a gel (Konishi et al. 2006a). Gash and co-workers developed a similar approach for the synthesis of stable non-siliceous aerogel monoliths without the need of strong acids (Gash et al. 2001).

Hierarchical Organization in Monolithic Sol–Gel Materials

19

Fig. 10 Propylene oxide as acid scavenger (Adapted with permission from reference Kido et al. (2012). Copyright (2012) American Chemical Society)

Fig. 11 Methods to control the reaction rate of metal precursors

In their alkoxide-free sol–gel approach, they raised the pH gradually by the addition of epoxides to aqueous and/or ethanolic solutions of metal salts. The effect of the epoxide as an acid scavenger is illustrated in Fig. 10. First, the oxygen atom of an epoxide, e.g., propylene oxide, is protonated by an acid and subsequently undergoes a ring-opening reaction by a nucleophilic attack of an acid anion. This epoxide method allows the formation of various metal oxides in different morphologies, e.g., monoliths, powders, or thin films. The amount of water in the sol represents another possibility to influence the hydrolysis rate of the precursor molecules. Water can either be directly added to the system or it can be released during the reaction, e.g., by esterification reactions. A summary of different possibilities to control the reaction rate of metal precursors is shown in Fig. 11.

Material TiO2

Polymer Cry. – Anatase – – PEO

PEO PEO PEG

PEO

PEO

PVP

Precursors/solvent/additives Ti(OiPr)4/HCl/HAc/H2O

Ti(OiPr)4/HCl/HAc/H2O

Ti(OnPr)4/HCl/FA/H2O

Ti(OnPr)4/HCl/NFA/H2O

TiO2/FA/HNO3/H2O

TiO2/FA/HNO3/H2O Ti(OnPr)4/PrOH/AcAc/EDA

Ti(OnPr)4/EtAcAc/PrOH/ NH4NO3

Ti(OiPr)4/glycerol/H2O

TiOSO4*xH2O/H2O/EG/FA

Pore dimensions Meso, macro Anatase Meso, macro Anatase Micro, macro Anatase Micro/ meso, macro Anatase Meso, macro Anatase Macro Anatase, Micro/ rutile meso, macro Anatase Meso, macro Rutile Macro Amorphous Meso, macro Anatase Meso, macro Amorphous Meso, Macro Anatase Meso, macro 4–9 n.s.

228 73

~15

~3

0.1–0.5 6–13

5–22

–

–

–

– –

n.s. – 0.2–0.4 2–8

Li et al. (2013a)

Chen et al. (2006)

Hasegawa et al. (2010b)

0.6 12–371

Fujita et al. (2004) Hasegawa et al. (2013)

– n.s.

~0.5–5 – ~0.5–5 n.s.

–

Konishi et al. (2006b)

–

~0.5–5 10–50

0.2–5.4 2–5

(Konishi et al. (2009)

Konishi et al. (2006a)

Wei et al. (2013)

References Backlund et al. (2007)

n.s.

2–14

1.4

–

Micro (nm) –

20–217

n.s. 15–137

350

2–20

Meso (nm) 3–4

~0.8–8 5

n.s.

130–180 1.6–5

150

61–88

SSA Macro (m2g 1) (μm) 10–180 0.4–4

Table 1 Overview over several non-siliceous, hierarchically structured monoliths prepared via sol–gel processing

20 A. Feinle et al.

ZrO2

Fe3O4

ZrOCl2*8H2O/H2O/EtOH/ PO

CuCl2*2H2O/H2O/EtOH/ glyc./PO/2-propanol CrCl3*6H2O/PO/EtOH/urea/ H2O FeCl3*6H2O/H2O/glyc./PO/ TMO/2-propanol Zr(OnPr)4/H2O/HNO3/NFA

Cu(OH)2

Cr2O3

F-127

Cu(NO3)2*5H2O/H2O

CuO

PEO

PEO

PAAm

PAAm

PAAm

–

PEO

PEO

TiO2/FA/HNO3 (freezedrying) AlCl3*6H2O/H2O/EtOH/PO

Al(NO3)3*9H2O/H2O/ boehmite

Al2O3

–

Ti(OiPr)4/HAc/HCl/MeOH

Meso, macro Amorphous Micro/ meso, macro Crystalline Meso, macro Crystalline Meso, macro

Crystalline

Meso, macro Amorphous Meso, macro Crystalline Macro

Crystalline

Meso, macro Anatase Meso, macro Amorphous Meso, macro γ-Al2O3 Meso, macro Boehmite Meso, macro γ-Al2O3 Meso, macro

Anatase

–

117

n.s.

n.s.

200 584

0.3–2

n.s.

n.s.

n.s.

108

224

~0.8

127

0.8–8

10–40

77–85

20–230

n.s.

76–89

n.s.

Kido et al. (2012)

–

59

–

2.5–4.8 –

n.s.

3–4

Guo et al. (2015) (continued)

Konishi et al. (2008)

Kido et al. (2014)

–

–

1.8

Fukumoto et al. (2015)

–

7

39

Naikoo et al. (2014)

– –

3–20

n.s.

4–12

Gaweł et al. (2012)

Tokudome et al. (2007b), Hartmann et al. (2012)

–

182–396 0.4–1.8 2.6

–

Fujita et al. (2006)

~0.7–5 n.s.

–

n.s.

13–20

Zhao et al. (2011)

n.s.

–

77

Hierarchical Organization in Monolithic Sol–Gel Materials 21

AlCl3*6H2O/H3PO4/H2O/ MeOH/PO

Precursors/solvent/additives YCl3*6H2O/AlCl3*6H2O/ H2O/EtOH/PO FeCl3*6H2O/LiCO3/H3PO4/ PO/PVP CaCl2*2H2O/H3PO4/H2O/ MeOH/PO/TMO/EB ZrOCl2*8H2O/H3PO4/H2O/ HCl/glyc. PEO

PAAm/ PEO

PAAm

PEO

Pore dimensions Macro

Micro, macro Crystalline Meso, macro Low Meso, crystallinity macro Crystalline Macro Amorphous Meso, macro Crystalline Macro

Crystalline

Polymer Cry. PEO Crystalline

– – –

– 30 –

n.s. n.s. n.s.

15

Li et al. (2013b)

Zhu et al. (2015)

n.s. 120

Tokudome et al. (2011)

Hasegawa et al. (2011)

References Tokudome et al. (2007a)

n.s.

n.s.

–

n.s.

n.s.

Micro (nm) –

4–5

n.s.

~1–5

Meso (nm) –

110–600 n.s.

23–58

5–68

SSA Macro (m2g 1) (μm) n.s. n.s.

Abbreviations: AcAc acetylacetone, EB 1,2-epoxybutane, EDA ethylenediamine, EG ethylene glycol, EtAcAc ethyl acetoacetate, EtOH ethanol, FA formamide, glyc glycerol, HAc acetic acid, MeOH methanol, NFA N-methyl formamide, n.s. not specified, PEG poly(ethylene glycol), PO propylene oxide, PrOH 1-propanol, TMO trimethylene oxide, PAAm poly(acrylamide), PEO poly(ethylene oxide), PVP poly(vinylpyrrolidone), SSA specific surface area

AlPO4

Zr(HPO4)2

CaHPO4

LiFePO4

Material Y3Al5O12

Table 1 (continued)

22 A. Feinle et al.

CrN

Cr3C2

Ni–C composite

Fe3O4

Cr2O3

CrN

Ni(OH)2

Products Cu, Cu2O

CaTiO3, SrTiO3, BaTiO3 Fe, Fe3C

TiO2

Starting material Cu(OH)2

Heat treatment under argon

Addition of urea and heat treatment under nitrogen Carbothermal reduction

Carbothermal reduction

Impregnation of TiO2

Conditions Solvothermal treatment

Table 2 Post-synthetic reactions of monolithic non-siliceous materials

Micro/meso, macro Micro, macro Mirco, macro Micro/meso, macro

Meso, macro

Pore dimension Meso, macro

14–191

454

56

22–226

2–49

SSA (m2g 1) 21–149

0.4

~0.8

~0.8

n.s.

n.s.

Macro (μm) n.s.

n.s.

–

n.s.

n.s.

–

n.s.

n.s.

–

Micro (nm) –

n.s.

4–50

Meso (nm) 28–33

Kido et al. (2013)

Kido et al. (2014)

Kido et al. (2014)

References Fukumoto et al. (2015) Ruzimuradov et al. (2011) Kido et al. (2012)

Hierarchical Organization in Monolithic Sol–Gel Materials 23

24

A. Feinle et al.

These sol–gel techniques are often combined with a polymerization-induced phase separation process that has already been described in the previous section for siliceous materials. In these cases, typical bimodal porosities are obtained with micro- and mesopores that often result from interstices between metal oxide crystallites. The size and shape of the macropores, however, can mostly be adjusted by the concentration of the polymer, which is used as phase separator and/or the gelation time of the sol. Table 1 gives an overview over the syntheses known to date toward non-siliceous monolithic materials with hierarchically organized pore sizes prepared via sol–gel processing. In Table 2 selected follow-up reactions, e.g., carbothermal reduction or solvothermal treatment, of the monolithic compounds of Table 1 are summarized. A selection of some recent key examples is presented below in more detail. Particular focus is given to meso-/macroporous titania monoliths since several of the underlying principles have been developed for this class of non-siliceous monoliths. Commonly, metal alkoxides, metal salts, or even preformed metal oxide particles are used as precursor molecules, and the wet gels are typically dried at 30–60  C in an oven and calcined at higher temperatures. This and further details for each synthesis are described in the individual examples.

Phase Separation by Adjusting the Sol–Gel Processing Conditions The interest in hierarchically organized titania monoliths in diverse application areas, e.g., for separation science is significant. Despite the difficulty to decrease and control the reaction rates of the precursor molecules, a variety of approaches toward monolithic materials with bicontinuous porosity has been published. Lindén et al. reported a template-free synthetic approach toward hierarchically macro-/ mesoporous anatase monoliths based on the sol–gel reaction of titanium isopropoxide in the presence of two different acids, namely, hydrochloric and acetic acids (Backlund et al. 2007). Hydrochloric acid as a strong acid enhances hydrolysis, but decreases the reaction rate of condensation. Acetic acid, however, slows down both hydrolysis and condensation rates, by acting as a chelating agent. Thus, the prepared wet gels were first dried at 60  C for 24 h prior to calcination at 300  C. After thermal treatment the material consists of fully crystalline anatase particles with crystallite sizes between 10 and 15 nm. Aggregation of these near to spherical particles leads to the formation of an interconnected, three-dimensional macroporosity. The macropore size can be controlled by the careful adjustment of the sol–gel processing conditions. Above a molar ratio of HAc/Ti(OiPr)4 2:1, small amounts of water are released by an esterification reaction between acetic acid and isopropanol. This in turn causes an increase of the condensation rate of the titanium alkoxide and further as well as of the solvent volume. Both contribute to an increase in the macropore diameter. A slightly modified form of this approach was used by Wei et al. for the synthesis of anatase monoliths for chromatographic applications (Wei et al. 2013). They successfully synthesized a material that comprises several properties which are prerequisites for the use in chromatographic applications, e.g., high specific surface areas and large-pore diameters.

Hierarchical Organization in Monolithic Sol–Gel Materials

25

Fig. 12 SEM images of dried TiO2 gels with increasing ratios of water/TiO2 (a–e). Digital picture of monolithic TiO2 gels prepared in Teflon tubes and a coin (f) (Figure adapted with permission from ref Konishi et al. (2006a). Copyright (2006) American Chemical Society)

A further possibility to generate multiscale porous titania monoliths in templatefree conditions was reported by Konishi et al. and is based on sol–gel processing of Ti(OnPr)4, hydrochloric acid, formamide, and water (Konishi et al. 2006a). Formamide is known to react with strong acids, thereby producing ammonia and thus increases the pH value gradually, e.g., from below 0 to 5 after 24 h aging time. This increase in pH can promote condensation reactions and induce sol–gel transitions. At the same time, the number of OH groups and thus the polarity of the gel phase are lowered with progressive degree of condensation. Since the polarity of the solvent remains high, phase separation between the condensed and the liquid phase occurs, and macropores are formed after drying of the sample at 60  C. In addition, the material crystallizes during drying and small anatase crystallites are formed. The authors further show that the macropore size of the crystalline network can be controlled via the composition of the starting solution and/or the temporal relationship between phase separation and gelation time (Fig. 12). In this context, high amounts of water delay the onset of phase separation, resulting in a finer bicontinuous structure of the monoliths, whereas, with small amounts, mainly spherical particles are formed.

Polymers as Phase Separation Agents TiO2 Konishi et al. extended their studies to the application of the abovementioned monoliths as chromatographic separation media (Konishi et al. 2009). The need for high mechanical strength was addressed by increasing the titanium precursor content in the starting sol. On one hand this indeed strengthens the network, whereas on the other the reactivity of the precursor solution in such concentrated systems is

26

A. Feinle et al.

Fig. 13 Ethyl acetoacetate converts to acetoacetic acid by hydrolysis. The generated acetoacetic acid immediately decomposes into acetone and carbon dioxide

dramatically increased resulting in a loss of control over phase separation. To regain better control, the ammonium source formamide was replaced by N-methyl formamide (NFA), which acts as acid scavenger and hydrolyzes much slower than formamide. Therefore, the pH is gently increased enabling a better control of the sol–gel transition even at high precursor concentrations. As a consequence of the high precursor concentration, the amount of propanol in the system is relatively high and almost no phase separation was observed. To improve the phase separation tendencies, the authors added the water-soluble polymer poly(ethylene oxide) PEO to the system. The polymer adsorbs to TiO2 oligomers via hydrogen bonding and is therefore able to reduce the solubility of the oligomers in the solvent. The careful choice of the Ti(OnPr)4–NFA–PEO composition finally enables the production of TiO2 monoliths with controllable porous morphology. Several similar approaches in which polymers are used as phase separation agents are listed in Table 1. The influence of mineral salts on the sol–gel processing of TiO2 oligomers and the role of strong acid anions as blocking agent to prevent titanium atoms from nucleophilic reactions were reported by Hasegawa et al. (2010b, c). The authors reported the synthesis of monolithic titania with multiscale porosity by utilizing PEO as phase separator, ethyl acetoacetate as chelating agent, and ammonium nitrate as mineral salt; the latter one is reported to further stabilize the chelated species and decrease the hydrolysis rate. In order to retain the monolithic shape of the wet gels during calcination, the authors removed the employed chelating agent by hydrolysis in EtOH/H2O, followed by decarbonation into acetone and carbon dioxide (Fig. 13). After crystallization of the amorphous gel skeleton in warm water, macroporous TiO2 monoliths with well-defined mesopores attributed to interstices between anatase crystallites were obtained. A biocompatible approach toward meso-/macroporous titania monoliths was described by Brook and his co-workers (Chen et al. 2006). They used glycerol to slow down the reaction rate of hydrolysis by transesterification of Ti(OiPr)4 (Fig. 14) and to accomplish the sol–gel reaction at neutral pH without the need for the addition of any catalyst. With the addition of PEO, they obtained bimodal meso- and

Hierarchical Organization in Monolithic Sol–Gel Materials

27

Fig. 14 Glycerol-modified titanium precursors (Figure adapted with permission from reference Chen et al. 2006. Copyright (2006) American Chemical Society)

macroporous amorphous structures that could be crystallized to anatase monoliths at temperatures above 600  C. In addition to the fully oxidized phases, reduced materials with tailored pore structures are known. Hasegawa and co-workers reported a novel synthesis toward reduced titanium oxide monoliths with well-defined hierarchical pore structure by the use of an ethylenediamine-modified titanium precursor (Hasegawa et al. 2013). The alkoxy groups of the titanium precursor Ti(OiPr)4 were substituted by ethane1,2-diamine, and the authors obtained inorganic–organic gels with Ti–N linkages (Fig. 15). Heating of the samples in an argon atmosphere initially resulted in the formation of anatase or rutile that converted toTi4O7 and Ti3O5 at 800–900  C by carbothermal reduction. The temperatures required for these reactions are exceptionally low, since the reduction reactions of anatase and rutile phases to Ti4O7 by H2 gas (Kolbrecka and Przyluski 1994), metals (Hauf et al. 1999; Kitada et al. 2012), or carbons (White et al. 1992) known to date require temperatures of more than 1000  C. The authors explain this low-temperature reduction with: (i) the small size of the anatase and rutile crystallites, (ii) the carbon coating, and (iii) the N-doping which distorts the Ti-O lattice and decrease the stability against reduction. With increasing temperature, the amount of micro- and mesopores increases, and specific surface areas of up to 200 m2g 1 have been obtained. Further reduction of Ti3O5 to Ti2O3 initially results in the loss of micropores and the formation of mesopores, whereas the formation of TiOxNy at 1400  C is accompanied with the loss of micro- and mesoporosity. Further studies on the effect of calcination conditions on the micro- and mesoporosity of these samples as well as on the electric conductivity have been reported recently (Hasegawa et al. 2015).

28

A. Feinle et al.

Fig. 15 Scheme of sol–gel processing of titanium alkoxides in the presence of ethane-1,2-diamine and the development of the crystalline character of the resulting TiO2 monoliths upon treatment at different temperatures

Fig. 16 Scheme of sol–gel processing of titanium alkoxides in the presence of ethane-1,2-diamine and the development of the crystalline character of the resulting TiO2 monoliths upon treatment at different temperatures

Hierarchical Organization in Monolithic Sol–Gel Materials

29

Fe3O4, Fe Similar approaches were reported for various iron- (Fe3O4, iron, and Fe2O4) and chromium-based (CrN and Cr3C2) crystalline monoliths (Kido et al. 2012, 2014). The main difficulty in preparing iron(III) oxide monoliths lies in the tendency of the precursor sol to form precipitates of iron(III) hydroxide. Nakanishi and his group reported a synthetic route toward iron-based monoliths from an aqueous solution of iron(III) chloride hexahydrate (Kido et al. 2012). Upon adjusting the solvent composition, polymer, and epoxide content, the authors were able to control the morphology and the gel formation of iron(III) hydroxide. As a simultaneous phase separation inducing and network-forming agent poly(acrylamide) was added. The conversion of the amorphous iron(III) hydroxide structure into crystalline hematite (α-Fe2O3) by calcination of the material in air was accompanied with collapse of the monolithic form, whereas it can be preserved by calcination in an inert argon gas flow (Fig. 16). In nonoxidizing atmospheres, the organic species are converted to carbon, which in turn acts as a reducing agent to yield crystalline Fe3O4, iron, and Fe3C from iron(III) species. With this process, the monolithic form as well as the macrostructure can be retained; however, the pore size and pore volume decrease with increasing temperature. The mesopore size decreases from 5 to 6 nm for the as-dried gels to 3–4 nm for the heat-treated samples. For the samples heated above 400  C, micropores appeared due to the combustion of carbon in the skeleton. Since the specific surface area of the heat-treated samples mainly depends on the proportion of micropores, the BET values increased from 5 m2g 1 at 300  C, through 224 m2g 1 at 700  C to 262 m2g 1 at 1000  C. This approach can be transferred to the synthesis of nickel–carbon composite monoliths from rigid nickel hydroxidebased xerogels (Kido et al. 2013). For the preparation of chromium-based monoliths, the authors combined the “urea glass route” in which urea is employed as the nitrogen and/or carbon source with the epoxide-mediated sol–gel route (Kido et al. 2014). Subsequent heat treatment under an inert atmosphere led to the formation of crystalline chromium nitride and chromium carbide. Al2O3 Porous alumina (Al2O3) is another class of oxidic material that attracts considerable attention due to its high thermal stability and moderate Lewis acidity. The first report on macroporous Al2O3 monoliths prepared by combining the phase separation with sol–gel processing was reported by Tokudome et al. in 2007 (Tokudome et al. 2007b). Their recipe of success was the use of: (i) aluminum salts instead of aluminum alkoxides to decrease the hydrolysis rate of the precursor, (ii) propylene oxide to start gelation by gradually increasing the pH value of the sol, and (iii) PEO to induce the formation of phase-separating structures. With variation of the PEO concentration, the authors were able to change the gel morphology from nonporous through bicontinuous to particle aggregates while the macropore size can

30

A. Feinle et al.

Fig. 17 Formation of a three-dimensional network by condensation of boehmite particles with hydrolyzed aluminum molecules

simultaneously be controlled in the range of 400 nm to 1.8 μm. Additionally, the authors were able to influence the degree of agglomeration grade and size of primary particles via the drying process (Tokudome et al. 2009). Hartmann et al. extended this work and mixed the aluminum precursor with the PEO/EtOH/H2O solution under ice-cooled conditions prior to the addition of propylene oxide (PO) at 25  C (Hartmann et al. 2012). With this method, they were able to perfectly control condensation and phase separation. Gawel et al. reported a method in which an aluminum nitrate precursor was hydrolyzed in the presence of boehmite (Gaweł et al. 2012). The interactions between the hydrolyzed aluminum molecules and the wetted surface of boehmite particles are formed leading to a three-dimensional network by condensation reactions (Fig. 17). Crystalline alumina with boehmite crystal structure is formed at temperatures up to 300  C, whereas at temperatures higher than 400  C, the crystalline phase is transformed into γ-alumina. The authors confirmed the hierarchical structure of the alumina samples with nitrogen sorption and mercury porosimetry measurements revealing the existence of meso- and macropores. Both originate from the aggregation of plate crystallites or interaggregate voids formed during the drying and calcination steps. CuO Recently, Fukumoto et al. reported the possibility to convert hierarchically organized copper hydroxide-based monolithic xerogels into copper oxides (CuO and Cu2O) under preservation of the monolithic form and macrostructure (Fukumoto

Hierarchical Organization in Monolithic Sol–Gel Materials

31

et al. 2015). A mixture of copper(II) chloride dihydrate, water, ethanol, glycerol, and PAAm was used as starting solution. In their approach, the presence of glycerol had a significant influence on the crystal growth with lower amounts leading to crystalline precipitates and higher amounts suppressing crystallization and allowing for preservation of the monolithic shape. In contrast, the added PAAm in the starting solution had no influence on the crystallization process, but the morphology was strongly influenced due to the incorporation of PAAm in the gel skeletons. The obtained morphologies varied from co-continuous structures consisting of similar-sized globular units for low PAAm concentrations to isolated macropores for excessive amounts. Therefore, it not only controls phase separation but also physically supports the network. The latter is particularly evident in the calcination process from copper(II) hydroxide to copper(II) oxide. Calcination in air resulted in the formation of crystalline copper(II) oxide, but also in collapse of the monolithic shape. This was improved by a pre-calcination step in argon at 800  C, followed by calcination in air at 400  C. Although, the monolith was retained, calcination in argon at such high temperatures is accompanied by the collapse of the mesopores. Simply a solvothermal treatment resulted in meso-/macroporous monoliths. However, the crystalline phase consisted of a mixture of copper(0)/PAAm and copper(I)/PAAm. The positive effect of solvothermal treatment in tailoring the meso- and crystal structure without destroying the macroporous structure was also reported by Hirao et al. and Guo et al. for the synthesis of meso-/macroporous zirconia (Konishi et al. 2008; Guo et al. 2015). With the solvothermal treatment, the authors obtained a high density of mesopores in the crystallized material via Ostwald ripening and high specific surface areas of up to 584 m2g 1.

Carbon Monoliths With Hierarchical Pore Structure Porous carbon materials have remarkable physicochemical properties, such as hydrophobicity, high corrosion resistance, good thermal stability, easy handling, and in many cases rather low costs in manufacturing, resulting in materials suitable for many areas of applications, such as energy storage or conversion, e.g., as battery electrodes or supercapacitors, in capacitive desalination, chemical catalysis, and electrocatalysis, to mention only a few examples (Zhai et al. 2011; Roberts et al. 2014a). As in the previous sections discussed, pore structure control is of major importance for carbon materials as well, not only to increase the surface area but also to adjust the accessibility of the active sites and to deliberately tailor the material for the specific application. Hierarchically organized, porous carbons are of special interest, for the reasons mentioned above. Numerous reports on the preparation of porous carbon monoliths via hard-templating approaches (sometimes even in combination with soft templating) can be found in the literature, and many excellent articles have been published (Yang and Zhao 2005; Ungureanu et al. 2015). As mentioned above, we will limit ourselves to “soft-templating” routes in combination with solution-based processes relying on polycondensation reactions as the networkforming mechanism due to the nature of this chapter. In addition, only hierarchical

Carbon/silica composites via phase separation Dual templating/ colloidal

Phase separation Dual phase separation

Strategy of pore design Acid-catalyzed sol–gel polymerization Living radical polymerization

Thermal curing, carbonization

Phenol, FA

BTEB

Mesophase pitch Phloroglucinol, FA

PMMA, F127

F127

F127, glycolic solvents

PS, PMMA

PDMS

DVB

Thermal activation with CO2

Annealing, carbonization Prepolymerization, carbonization, high-temperature treatment Calcination, silica etching

Structure directing agent –

Carbon precursor Resorcinol, FA

Other treatment steps Thermal activation with CO2

Meso, macro

Micro, macro

Micro, meso, macro Meso, macro Meso, macro

Pore structure Micro, macro

0.34

1

>1500

464–505

3

100

~2

Macro (μm) ~2

350

20–170

2360

SSA (m2 g 1) >3000

Table 3 Different approaches to monolithic carbon materials with hierarchical pore structures

3.0

–

8.0

10–100

~20

Meso (nm) –

Adelhelm et al. (2007) Liang and Dai (2009)

–

Hasegawa et al. (2012)

Wang et al. (2008)

0.5

–

–

Hasegawa et al. (2010a)

References Baumann et al. (2008) n.s.

Micro (nm) n.s.

32 A. Feinle et al.

Styrene

DVB, VBC, span 80

Polyglycerol polyricinoleate

Colloidal silica

Stöber silica, F127

PMMA, TEOS, F127

Micro, meso, macro Micro, meso, macro Meso, macro

Meso, macro

Meso, macro

~25

10

4, fierce ignition of the gel occurred during calcination, thereby increasing the temperature and creating hard agglomerates. It is also desirable to keep the amount of added organics low from an economic perspective. However, if the ratio was too low, < 0.5, the complexation was not sufficient to avoid secondary phases. A total of 150 mL water was necessary to aid the dissolution of CA when 1 mol of CA and 1 mol of EG were used. Excess water beyond this had two drawbacks. First, the evaporation of water also removed some quantities of the polymeric substance out from the boiling solution. And second, the formation of energetic water vapor during the evaporation could break the polymeric network that was already forming in the solution.

High-Temperature Superconductors Superconductors are materials which exhibit zero electrical resistance and expulsion of magnetic fields below a critical temperature, Tc. The discovery of high-temperature superconductors in copper oxide-based material systems with a Tc above 77 K (can be cooled by liquid nitrogen) in the late 1980s created enormous excitement and these materials were intensively investigated (Bednorz and M€uller 1986). Particularly oxides in the Y-Ba-Cu system (YBCO), like YBa2Cu3O7-x, received a significant amount of interest (Wu et al. 1987). There are numerous reports on different synthesis routes to prepare YBCO powder, as reviewed by Kakihana and Pathak and Misra (Kakihana 1996; Pathak and Mishra 2005). The powder characteristics, like homogeneity, phase purity, and particle size, will have a large influence on the properties of the final device, and both solid-state (Cava et al. 1987) and wet chemical synthesis routes (Kamat et al. 1991) have been employed for the synthesis of YBCO. Different sol–gel syntheses to prepare YBCO have been applied, and they are argued to give better homogeneity and be less complicated than other wet chemical methods, such as coprecipitation (Kakihana et al. 1989). The cation precursors used in modified Pechini syntheses for YBCO can either be nitrates (Kakihana et al. 1991) or Y2O3 and CuO dissolved in HNO3 together with BaCO3 (Lee et al. 1989). A possible precipitation of Ba(NO3)2 could be avoided by adding ammonia to adjust the pH to about 6 (Chu and Dunn 1987). In all of the reports, CA and EG were employed as complexing and polymerization agents. The ratio between metal cations and CA varied from 1:0.33 (Kakihana et al. 1991) to 1:6.5 (Shiomi et al. 1993). After evaporation of the solvent, the gels were calcined and phase-pure YBCO was obtained at around 900  C (Mazaki et al. 1991). Sometimes BaCO3 could be detected after calcination at lower temperatures (Lee et al. 1989). The resistivity of YBCO prepared with CA and EG is shown in Fig. 7, demonstrating the superconducting transition. The narrow transition

resistivity (arbitrary unit)

Modified Pechini Synthesis of Oxide Powders and Thin Films

17

92 K

90 K R=0 70

80

90

100 temperature

110

120

Fig. 7 Resistivity as a function of temperature for a polycrystalline sample of YBa2Cu3O7-δ prepared by a modified Pechini method with CA and EG. (From Kakihana et al. 1991)

temperature interval is an indication that YBCO with high purity and structural homogeneity could be prepared (Kakihana et al. 1991) by this method. In addition to the modified Pechini routes, there are also several reports on amorphous citrate-syntheses to prepare YBCO (Yang et al. 1989) without the use of EG. Also oxalic acid has been used as the complexing agent (Sanjines et al. 1988). In some of these reports, Y2O3, together with barium and copper carbonate, was dissolved directly in citric acid, without the use of HNO3 (Karen and Kjekshus 1994). The molar ratio between cations and CA was typically 1:2 (Sanjines et al. 1988). Also in these syntheses the pH was often increased to prevent precipitation of Ba(NO3)2. Liu et al. argued that it was advantageous to use ethylenediamine to adjust the pH instead of ammonia, as it could react with nitric acid and thereby further prevent precipitation of Ba(NO3)2 (Liu et al. 1989). Another reported way to mitigate the issue with Ba(NO3)2 precipitation is to employ a combination of CA and EDTA as complexing agents (Van der Biest et al. 1991). Here, one solution is prepared where Y3+ and Cu2+ are complexed by CA and another with Ba2+ complexed by EDTA. After mixing the two solutions and adjusting pH to 7, the precipitation could be avoided (Niou et al. 1992).

Cathode Materials for Li Ion Batteries Lithium ion batteries have become increasingly important and have been at the focus of intense research the last decades. They are today routinely used in portable electronic devices and also show great promise for storage of energy from renewable sources and their use in electronic vehicles (Bruce et al. 2008; Goodenough and Kim 2010). The anode is typically graphite, whereas the cathode traditionally was LiCoO2 (Mizushima et al. 1980). Later a variety of potential cathode materials has received interest, among them several materials based on manganese oxide (LiMn2O4, Li(Ni1/2Mn1/2)O2, Li(Mn1/3Co1/3Ni1/3)O2), phosphates (LiFePO4), and

18

T.O.L. Sunde et al.

silicates (Li2FeSiO4) (Tarascon et al. 1991; Padhi et al. 1997; Nytén et al. 2005). It has been shown for many of these materials that their performance is highly dependent on the synthesis method (Li et al. 2004), and indeed, the modified Pechini process has proven very useful (Liu et al. 1996). The choice of cation precursors for Li ion battery cathodes are typically nitrates and sometimes acetates (Predoana et al. 2015). Some exceptions exist, for instance, was SiO2 particles used in the synthesis of Li2FeSiO4 (giving a colloidal suspension) (Dominko et al. 2008) and tetra-n-butyl titanate was used to dope LiMn2O4 with titanium (Xiong et al. 2012). In the latter case, lactic acid was used to stabilize the titanium solution. The first report of Pechini synthesis of LiMn2O4 used EG as a solvent (Liu et al. 1996), but it has since become increasingly common to use aqueous solutions (Kunduraci and Amatucci 2006; Predoana et al. 2015). The most commonly used complexing agents are, as in Pechini’s original patent, CA and EG (Zhao et al. 2013). There are a few reports where the amount of EG has been optimized (Han and Kim 2000). Kunduraci and Amattucci prepared LiMn1.5Ni0.5O4 with a 1:1 ratio between cations and CA and an accompanying ratio of EG varying between 0, 2, 4, and 7 (Kunduraci and Amatucci 2008). Interestingly, it was found that the presence of EG had a large influence on the obtained microstructure after calcination. The materials prepared with EG had smaller particles and also a more mesoporous structure, as can be seen in Fig. 8. Both of these factors had a large impact on the electrochemical performance of the cathode material, and the optimal ratio between CA and EG was found to be 1:4. Other authors have applied only CA (Predoana et al. 2015) or a combination of CA and EDTA (Liu et al. 2014). Duncan et al. used EG without a carboxylic acid for the preparation of LiMn1.5Ni0.5O4 (Duncan et al. 2010). Here, an ester reaction occurred between EG and the acetate precursors; however, a polymer network will not form as the acetate has only one functional group. The calcination temperature has been shown to be an important parameter to optimize the electrochemical properties of the cathode materials. First of all, the volatility of Li at high temperatures is well known (Rossen et al. 1993), and sometimes an excess of Li is used in the solution to compensate for this (Predoana et al. 2015). Furthermore, a low calcination temperature will give smaller particle sizes, giving a high discharge capacity; however, the crystallinity of the particles can often be impaired, leading to fast capacity fading (Zhao et al. 2013). Typically a compromise must be found between these two aspects. A similar contradiction was encountered in the preparation of the layered material Li(Mn1/3Co1/3Ni1/3)O2 (Xia et al. 2009). Again, a low calcination temperature favorably produced smaller particles. However, in this case it also resulted in cation disorder, where Li goes into the transition metal layers, which in turn significantly reduced the lithium mobility. Finally, most of these cathode materials suffer from inherently poor electron conductivity. A carbon coating on the particle surfaces is typically necessary to mitigate this issue (Dominko et al. 2005). Sometimes this is obtained by adding a polymer, like polyethylene glycol, to the solution, followed by a heat treatment in reducing atmosphere (Mei et al. 2012). However, interestingly, when the modified

Modified Pechini Synthesis of Oxide Powders and Thin Films

19

Fig. 8 Scanning electron microscopy images of LiMn1.5Ni0.5O4 prepared by a modified Pechini method with a 1:4 ratio between CA and EG (a) and without EG (b). (From Kunduraci and Amatucci 2008)

Pechini method is used, this layer can be formed without the use of additives, simply from carbonization of the complexing agents CA and EG (Dominko et al. 2008). This carbon layer provides the necessary conductivity, but also serves to prevent particle growth and agglomeration (Moskon et al. 2007). Zhang et al. used a modified Pechini route to prepare a composite cathode material of Li3V2(PO4)3 and graphene (Zhang et al. 2013). Here, the complexing agents, CA and EG, were not only used to chelate the precursors, but also to attach the phosphate to functional groups on graphene oxide. In the following reducing heat treatment, CA and EG decomposed to coat the surface of the phosphate particles with a carbon layer and the graphene oxide was reduced to graphene. A schematic depiction of this process is given in Fig. 9.

Thin Films Superconducting Thin Films Superconducting thin films have been an important part of the science of superconductivity for more than six decades (Lin et al. 2015). Potential applications for hightemperature superconducting films include high-frequency electronics, microwave communications, and magnetic field detectors (Norton 2003). Different substrates are used for different application, and lattice match to give epitaxy and chemical compatibility are among the important considerations. Special attention has also been directed towards the preparation of superconducting wires, typically consisting of a metallic tape with a superconducting coating, which can enable large-scale applications for electric-power and magnetic applications (Larbalestier et al. 2001; Rupich et al. 2004; Kang et al. 2006). Here, a buffer layer between the metal substrate and the superconducting layer is needed to protect the metal substrate from oxidation and to give epitaxial growth of the superconducting film (Obradors

20

T.O.L. Sunde et al.

COOH -LVP

COOH -LVP

O

OH

+ CA -LVP + EG

80∞C stirring

O

COOH

Carbon shell Graphene sheet

750∞C OH - CA -LVP

N2/H2

LVP LVP

LVP

COOH-EG

Fig. 9 A schematic depiction of how a composite cathode of graphene and Li3V2(PO4)3 (LVP) was prepared from a modified Pechini process. (From Zhang et al. 2013)

et al. 2004). Chemical solution deposition has emerged as a highly competitive method for the preparation of YBCO coatings, both in the shape of wires and films, where high-quality films can be prepared cost-effectively with high speed without the need for vacuum (Obradors et al. 2006). A highly successful route for the preparation of superconducting films is the trifluoroacetate (TFA) route (Iguchi et al. 2002). The advantage of this synthesis method is that it avoids the formation of BaCO3, which can precipitate at the grain boundaries and limit the performance. However, there are some drawbacks from using fluorine, most notably the evolution of highly corrosive and dangerous hydrofluoric gas, making it less suitable for industrial upscaling (Cui et al. 2009). As a consequence, several modified Pechini routes have been developed as alternatives (Bubendorfer et al. 2003), where the formation of barium carbonate can also be avoided (Cui et al. 2009). The typical cation precursors are nitrates or acetates, but there is a larger variety in the choice of complexing agents, with CA and EG (Cui et al. 2009), EDTA (Brylewski and Przybylski 1993), triethanolamine and acetic acid (Thuy et al. 2009), and trimethylacetate and propionic acid (Shi et al. 2004) all being applied. Bubendorfer et al. investigated several organic acid, such as lactic, tartaric, glycolic, malonic, and diglycolic acid, but found that malic acid was the best choice, due to its superior ability to chelate Y3+ (Bubendorfer et al. 2003). Wang et al. used the addition of polymers, such as PVB, PEG, and PVP to improve the wettability and viscosity of the solution (Wang et al. 2008a). Solutions with a very long shelf life can be prepared (Thuy et al. 2009). By spin or dip coating on substrates such as strontium titanate or lanthanum aluminate, films with a preferential orientation can be prepared, as can be seen in the diffractogram in Fig. 10.

Transparent Conducting Oxides Powder synthesis and preparation of bulk materials, as described for ITO above, are important for the preparation of high-quality sputtering targets, but the materials are used as transparent thin films in their final application. Several physical deposition techniques for the preparation of TCOs exist, but chemical solution-based techniques offer several advantages. TCO thin films of In2O3 (Legnani et al. 2007), SnO2 (Sladkevich et al. 2011), and ZnO (Lima et al. 2007) have all been prepared by the

Modified Pechini Synthesis of Oxide Powders and Thin Films

21

Fig. 10 XRD of an YBCO thin film with preferential orientation deposited by spin coating on a strontium titanate substrate. The solution was prepared with malic acid and glycerol as complexing agents. (From Bubendorfer et al. 2003)

modified Pechini method, both by spin coating (Sunde et al. 2012) and dip coating (Kundu and Biswas 2008). The most common synthesis route is to dissolve nitrate or chloride precursors in water with the addition of CA and EG (Legnani et al. 2007). For the deposition of ZnO thin films, He et al. used a mixture of water and ethanol as solvent. In some cases, it was shown that an additive, like PVA, was necessary in order to increase the viscosity of the solution and improve the wettability on the substrate (Kundu and Biswas 2008; Sunde et al. 2014). The thickness of the films was tailored by the number of depositions, but also by varying the cation concentration and viscosity of the solution and by changing the spinning or dipping speed (Sunde et al. 2014). Layers ranging from 10 to several hundreds of nanometre could be prepared. Bernardi et al. prepared films of antimony-doped tin oxide (ATO) with varying thickness by changing the viscosity by carefully adding or evaporating water before dip coating (Bernardi et al. 2002). Two films with the same total thickness, consisting of seven and one layers, respectively, are shown in Fig. 11. They found that the highest density and best optical and electrical properties were obtained in the film made by several depositions. The TCO films are typically deposited on glass slides, but more refractory substrates, like sapphire or YSZ, are necessary if the calcination temperature is higher than about 500  C. Sladkevich et al. also prepared ATO films on sheet-like clay particles, thereby demonstrating the flexibility of the modified Pechini process (Sladkevich et al. 2011). Also patterning, which is important for TCOs used in photovoltaic devices, has been demonstrated by the modified Pechini process.

22

T.O.L. Sunde et al.

Fig. 11 SEM micrographs obtained at 45 inclination of ATO films prepared by dip coating of a modified Pechini solution. The top film (a) is prepared by seven layers from a solution with a viscosity of 4 cP. The bottom film (b) is a single layer from a solution with 20 cP. (From Bernardi et al. 2002)

Sladkevich et al. prepared films of Cd2SnO4 by spin coating of an aqueous solution with CA and EG. By using a lithographic nanoimprint technique, a pattern was made in the as-spun polymeric precursor film, which remained in the film after calcination (Schumm et al. 2011). Here, it was also demonstrated that they could obtain the metastable cubic phase of Cd2SnO4 in the thin film, which cannot be obtained in bulk materials, by optimizing the calcination procedure. Finally, the calcination temperature and atmosphere have been shown to be very important for the electrical properties of the TCOs (Choppali and Gorman 2008). After heat treatment at high temperatures and in reducing atmospheres, ITO thin films with excellent properties have been obtained. With a specific resistance down in the 10 4 Ω
cm range, the films prepared by the modified Pechini method are comparable to the best values from physical deposition techniques (Sunde et al. 2014).

Thin Film Phosphors Thin films of luminescent materials are attractive for many technological applications, especially related to display technology (Yu et al. 2005), but also for lightconversion layers for photovoltaics (Huang et al. 2013) and optical waveguides (Chae et al. 2013). Already in 1980 Robertson and van Tol demonstrated that epitaxial luminescent films of rare earth-doped garnets could withstand much higher power densities in cathode ray tubes without degradation than their powder counterparts (Robertson and Van Tol 1980). Since then, thin film phosphor materials have received significant attention (Choe et al. 2001; Garskaite et al. 2010). The uniform thickness and smoother surface of the thin films makes it possible to define smaller pixels, thereby giving a higher resolution. Thin films prepared by the modified

Modified Pechini Synthesis of Oxide Powders and Thin Films

23

Pechini method can also be patterned, which can be achieved by relatively simple and inexpensive soft-lithography techniques (Pang et al. 2003). A demonstration of such patterning is given in Fig. 12. These patterns were prepared by the micromolding in capillaries technique. Here, a droplet of the modified Pechini solution, containing the cation precursors and complexing agents, is deposited next to a micromold, upon which capillary forces will pull the solution into the mold. After drying the mold can be removed, and the patterned oxide thin film remains after calcination (Lin et al. 2007). Phosphors made up from monodisperse, small, and spherical particles are attractive due to the possibility of a low light-scattering and a high packing density, leading to good resolution (Martinez-Rubio et al. 2001). In this regard, the flexibility of the modified Pechini method can be utilized to produce core-shell structures. Here, monodisperse and spherical silica particles are typically produced by the Stöber method (Stöber et al. 1968). These particles are added to a modified Pechini solution, containing the chosen cations in the desired stoichiometry together with complexing agents, typically CA and PEG (Wang et al. 2005). The silica particles contain OH groups, which the chelated cations can bond to, thereby producing a thin coating after calcination. By using a silica core, the total cost of the phosphor particle is significantly reduced. The optical properties of the core-shell particles can be tuned by the number of coatings, the calcination temperature, and the size of the silica particle. The crystallinity of the coating improves with increasing annealing temperature; however, if the temperature is too high, a reaction between the core and the shell can occur (Lin et al. 2007).

Fig. 12 Optical photographs of patterned thin films of LaPO4 doped with Ce3+ and Tb3+ prepared by the modified Pechini process and the micromolding in capillaries technique. (Modified from (Lin et al. 2007). Reprinted with permission from Lin et al. Copyright 2007 American Chemical Society)

24

T.O.L. Sunde et al.

References Abbas HA, Youssef AM, Hammad FF, Hassan AMA, Hanafi ZM. Electrical properties of nanosized indium tin oxide (ITO) doped with CuO, Cr2O3 and ZrO2. J Nanopart Res. 2014;16:2518. Arima M, Kakihana M, Nakamura Y, Yashima M, Yoshimura M. Polymerized complex route to barium titanate powders using barium-titanium mixed-metal citric acid complex. J Am Ceram Soc. 1996;79:2847–56. Baumann FS, Fleig J, Habermeier HU, Maier J. Impedance spectroscopic study on well-defined (La, Sr)(Co, Fe)O3-δ model electrodes. Solid State Ion. 2006;177:1071–81. Bednorz JG, M€uller KA. Possible high Tc superconductivity in the Ba-La-Cu-O system. Z Phys B: Condens Matter. 1986;64:189–93. Bernardi MIB, et al. Influence of the concentration of Sb2O3 and the viscosity of the precursor solution on the electrical and optical properties of SnO2 thin films produced by the Pechini method. Thin Solid Films. 2002;405:228–33. Bruce PG, Scrosati B, Tarascon JM. Nanomaterials for rechargeable lithium batteries. Angew Chem Int Ed. 2008;47:2930–46. Brylewski T, Przybylski K. Physicochemical properties of high-TC (Bi, Pb)-Sr-Ca-Cu-O and Y-Ba-Cu-O superconductors prepared by sol–gel technique. Appl Supercond. 1993;1:737–44. Bubendorfer AJ, Kemmitt T, Campbell LJ, Long NJ. Formation of epitaxial YBCO thin films by ex-situ processing of a polymerized complex. IEEE Trans Appl Supercond. 2003;13:2739–42. Cava RJ, et al. Bulk superconductivity at 91 K in single-phase oxygen-deficient perovskite Ba2YCu3O9-δ. Phys Rev Lett. 1987;58:1676–9. Chae KW, Park TR, Cheon CI, Cho NI, Kim JS. Transparent and highly luminescent Eu-oxide thin film phosphors on sapphire substrates. Electron Mater Lett. 2013;9:59–63. Chiang C, Shei CY, Wu SF, Huang YT. Preparation of high-purity Tl-based “1223” superconductor phase by modified Pechini process in water solution. Appl Phys Lett. 1991;58:2435–7. Chick LA, et al. Glycine-nitrate combustion synthesis of oxide ceramic powders. Mater Lett. 1990;10:6–12. Cho SG, Johnson PF, Condrate Sr RA. Thermal decomposition of (Sr, Ti) organic precursors during the Pechini process. J Mater Sci. 1990;25:4738–44. Choe JY, et al. Alkoxy sol–gel derived Y3-xAl5O12:Tbx thin films as efficient cathodoluminescent phosphors. Appl Phys Lett. 2001;78:3800–2. Choppali U, Gorman BP. Preferentially oriented ZnO thin films on basal plane sapphire substrates derived from polymeric precursors. Mater Chem Phys. 2008;112:916–22. Chu C-T, Dunn B. Preparation of high-Tc superconducting oxides by the amorphous citrate process. J Am Ceram Soc. 1987;70:c375–7. Courty P, Ajot H, Marcilly C, Delmon B. Oxydes mixtes ou en solution solide sous forme très divisée obtenus par décomposition thermique de précurseurs amorphes. Powder Technol. 1973;7:21–38. Cui W, et al. YBa2Cu3O7-x thin films by citrate-based non-fluorine precursor. J Supercond Nov Magn. 2009;22:811–5. Da Conceição L, Silva CRB, Ribeiro NFP, Souza MMVM. Influence of the synthesis method on the porosity, microstructure and electrical properties of La0.7Sr0.3MnO3 cathode materials. Mater Charact. 2009;60:1417–23. Da Conceição L, Ribeiro NFP, Souza MMVM. Synthesis of La1-xSrxMnO3 powders by polymerizable complex method: evaluation of structural, morphological and electrical properties. Ceram Int. 2011;37:2229–36. Ding X, Liu Y, Gao L, Guo L. Synthesis and characterization of doped LaCrO3 perovskite prepared by EDTA-citrate complexing method. J Alloys Compd. 2008;458:346–50. Dominko R, et al. Impact of the carbon coating thickness on the electrochemical performance of LiFePO4/C composites. J Electrochem Soc. 2005;152:A607–10. Dominko R, Conte DE, Hanzel D, Gaberscek M, Jamnik J. Impact of synthesis conditions on the structure and performance of Li2FeSiO4. J Power Sources. 2008;178:842–7.

Modified Pechini Synthesis of Oxide Powders and Thin Films

25

Duncan H, Abu-Lebdeh Y, Davidson IJ. Study of the cathode-electrolyte interface of LiMn1.5Ni0.5O4 synthesized by a sol–gel method for Li-ion batteries. J Electrochem Soc. 2010;157:A528–35. Fan B, Liu X. A-deficit LSCF for intermediate temperature solid oxide fuel cells. Solid State Ion. 2009;180:973–7. Farbun IA, Romanova IV, Kirillov SA. Optimal design of powdered nanosized oxides of high surface area and porosity using a citric acid aided route, with special reference to ZnO. J Sol–Gel Sci Technol. 2013;68:411–22. Fontaine ML, Laberty-Robert C, Barnabé A, Ansart F, Tailhades P. Synthesis of La2-xNiO4+δ oxides by polymeric route: non-stoichoimetry control. Ceram Int. 2004;30:2087–98. Gai S, Li C, Yang P, Lin J. Recent progress in rare earth micro/nanocrystals: soft chemical synthesis, luminescent properties, and biomedical applications. Chem Rev. 2014;114:2343–89. Garskaite E, Lindgren M, Einarsrud MA, Grande T. Luminescent properties of rare earth (Er, Yb) doped yttrium aluminium garnet thin films and bulk samples synthesised by an aqueous sol–gel technique. J Eur Ceram Soc. 2010;30:1707–15. Gaudon M, Laberty-Robert C, Ansart F, Stevens P, Rousset A. Preparation and characterization of La1-xSrxMnO3+δ (0  x  0.6) powder by sol–gel processing. Solid State Sci. 2002;4:125–33. Gehman BL, et al. Influence of manufacturing process of indium tin oxide sputtering targets on sputtering behavior. Thin Solid Films. 1992;220:333–6. Ghosh S, Dasgupta S, Sen A, Maiti HS. Low-temperature synthesis of nanosized bismuth ferrite by soft chemical route. J Am Ceram Soc. 2005a;88:1349–52. Ghosh S, Dasgupta S, Sen A, Maiti HS. Low temperature synthesis of bismuth ferrite nanoparticles by a ferrioxalate precursor method. Mater Res Bull. 2005b;40:2073–9. Ginley DS, Perkins JD. Transparent conductors. In: Ginley DS, Hosono H, Paine DC, editors. Handbook of transparent conductors. New York: Springer US; 2011. p. 1–25. Goodenough JB, Kim Y. Challenges for rechargeable Li batteries. Chem Mater. 2010;22:587–603. G€ulg€un MA, Nguyen MH, Kriven WM. Polymerized organic–inorganic synthesis of mixed oxides. J Am Ceram Soc. 1999;82:556–60. Gupta RK, Whang CM. Effects of anion and synthesis route on the structure of (La0.9Sr0.1) (Cr0.85Fe0.05Co0.05Ni0.05)O3-δ perovskite and removal of impurity phases. Solid State Ion. 2007;178:1617–26. Han YS, Kim HG. Synthesis of LiMn2O4 by modified Pechini method and characterization as a cathode for rechargeable Li/LiMn2O4 cells. J Power Sources. 2000;88:161–8. Hardy A, et al. Gel structure, gel decomposition and phase formation mechanisms in the aqueous solution-gel route to lanthanum substituted bismuth titanate. J Sol–Gel Sci Technol. 2005;33:283–98. Hardy A, et al. Effects of precursor chemistry and thermal treatment conditions on obtaining phase pure bismuth ferrite from aqueous gel precursors. J Eur Ceram Soc. 2009;29:3007–13. Huang X, Han S, Huang W, Liu X. Enhancing solar cell efficiency: the search for luminescent materials as spectral converters. Chem Soc Rev. 2013;42:173–201. Iguchi T, Araki T, Yamada Y, Hirabayashi I, Ikuta H. Fabrication of Gd-Ba-Cu-O films by the metalorganic deposition method using trifluoroacetates. Supercond Sci Technol. 2002;15:1415–20. Jiang SP. Development of lanthanum strontium manganite perovskite cathode materials of solid oxide fuel cells: a review. J Mater Sci. 2008;43:6799–833. Jiang QH, Nan CW, Shen ZJ. Synthesis and properties of multiferroic La-modified BiFeO3 ceramics. J Am Ceram Soc. 2006;89:2123–7. Kakihana M. Invited review “sol–gel” preparation of high temperature superconducting oxides. J Sol–Gel Sci Technol. 1996;6:7–55. Kakihana M, Yoshimura M. Synthesis and characteristics of complex multicomponent oxides prepared by polymer complex method. Bull Chem Soc Jpn. 1999;72:1427–43. Kakihana M, Börjesson L, Eriksson S, Svedlindh P, Norling P. Synthesis of highly pure YBa2Cu3O7-δ superconductors using a colloidal processing technique. Phys C. 1989;162–164:931–2.

26

T.O.L. Sunde et al.

Kakihana M, Börjesson L, Eriksson S, Svedlindh P. Fabrication and characterization of highly pure and homogeneous YBa2Cu3O7 superconductors from sol–gel derived powders. J Appl Phys. 1991;69:867–73. Kakihana M, et al. Polymerized complex route to synthesis of pure Y2Ti2O7 at 750 C using yttriumtitanium mixed-metal citric acid complex. J Am Ceram Soc. 1996;79:1673–6. Kakihana M, Arima M, Yoshimura M, Ikeda N, Sugitani Y. Synthesis of high surface area LaMnO3 +d by a polymerizable complex method. J Alloys Compd. 1999;283:102–5. Kamat RV, Vittal Rao TV, Pillai KT, Vaidya VN, Sood DD. Preparation of high grade YBCO powders and pellets through the glycerol route. Phys C. 1991;181:245–51. Kang S, et al. High-performance high-Tc superconducting wires. Science. 2006;311:1911–4. Karen P, Kjekshus A. Citrate-gel syntheses in the Y(O)-Ba(O)-Cu(O) system. J Am Ceram Soc. 1994;77:547–52. Kim BC, Kim SM, Lee JH, Kim JJ. Effect of phase transformation on the densification of coprecipitated nanocrystalline indium tin oxide powders. J Am Ceram Soc. 2002;85:2083–8. Kim JK, Kim SS, Kim WJ. Sol–gel synthesis and properties of multiferroic BiFeO3. Mater Lett. 2005;59:4006–9. Kim SM, et al. Preparation and sintering of nanocrystalline ITO powders with different SnO2 content. J Eur Ceram Soc. 2006;26:73–80. Kodaira CA, Brito HF, Felinto MCFC. Luminescence investigation of Eu3+ ion in the RE2(WO4)3 matrix (RE = La and Gd) produced using the Pechini method. J Solid State Chem. 2003;171:401–7. Kundu S, Biswas PK. Synthesis of nanostructured sol–gel ITO films at different temperatures and study of their absorption and photoluminescence properties. Opt Mater. 2008;31:429–33. Kunduraci M, Amatucci GG. Synthesis and characterization of nanostructured 4.7 V LixMn1.5Ni0.5O4 spinels for high-power lithium-ion batteries. J Electrochem Soc. 2006;153: A1345–52. Kunduraci M, Amatucci GG. The effect of particle size and morphology on the rate capability of 4.7 V LiMn1.5+δNi0.5-δO4 spinel lithium-ion battery cathodes. Electrochim Acta. 2008;53:4193–9. Larbalestier D, Gurevich A, Feldmann DM, Polyanskii A. High-Tc superconducting materials for electric power applications. Nature. 2001;414:368–77. Lee HK, Kim D, Suck SI. Superconducting transition and Raman spectrum of Y1Ba2Cu3O7-x prepared by polymeric precursor synthesis. J Appl Phys. 1989;65:2563–5. Legnani C, et al. Indium tin oxide films prepared via wet chemical route. Thin Solid Films. 2007;516:193–7. Leonard NM, Wieland LC, Mohan RS. Applications of bismuth(III) compounds in organic synthesis. Tetrahedron. 2002;58:8373–97. Li DC, Muta T, Zhang LQ, Yoshio M, Noguchi H. Effect of synthesis method on the electrochemical performance of LiNi1/3Mn1/3Co1/3O2. J Power Sources. 2004;132:150–5. Li L, et al. Synthesis and luminescent properties of high brightness MLa(WO4)2:Eu3+ (M = Li, Na, K) and NaRE(WO4)2:Eu3+ (RE = Gd, Y, Lu) red phosphors. J Lumin. 2013;143:14–20. Lima SAM, Cremona M, Davolos MR, Legnani C, Quirino WG. Electroluminescence of zinc oxide thin-films prepared via polymeric precursor and via sol–gel methods. Thin Solid Films. 2007;516:165–9. Lin J, Yu M, Lin C, Liu X. Multiform oxide optical materials via the versatile pechini-type sol–gel process: synthesis and characteristics. J Phys Chem C. 2007;111:5835–45. Lin YH, Nelson J, Goldman AM. Superconductivity of very thin films: the superconductorinsulator transition. Phys C. 2015;514:130–41. Liu RS, Wang WN, Chang CT, Wu PT. Synthesis and characterization of high-Tc superconducting oxides by the modified citrate gel process. Jpn J Appl Phys. 1989;28:L2155–7. Liu W, Farrington GC, Chaput F, Dunn B. Synthesis and electrochemical studies of spinel phase LiMn2O4 cathode materials prepared by the Pechini process. J Electrochem Soc. 1996;143:879–84.

Modified Pechini Synthesis of Oxide Powders and Thin Films

27

Liu X, Lin C, Lin J. White light emission from Eu3+ in CaIn2O4 host lattices. Appl Phys Lett. 2007;90:81904. Liu T, Xu Y, Zhao J. Low-temperature synthesis of BiFeO3 via PVA sol–gel route. J Am Ceram Soc. 2010;93:3637–41. Liu X, Huang T, Yu A. Fe doped Li1.2Mn0.6-x/2Ni0.2-x/2FexO2 (x  0.1) as cathode materials for lithium-ion batteries. Electrochim Acta. 2014;133:555–63. L€u Y, et al. Color-tunable luminescence of YNbO4:Ln3+ (Ln3+ = Dy3+ and/or Eu3+) nanocrystalline phosphors prepared by a sol–gel process. Eur J Inorg Chem. 2015;2015:5262–71. Magnone E, Traversa E, Miyayama M. Synthesis and thermal analysis of the strontium and irondoped lanthanum cobaltite nano-powder precursors. J Ceram Soc Jpn. 2007;115:402–8. Marcilly C, Delmon B, Sugier A. New process for obtaining finely divided homogeneous oxides of many elements. In: Editor^Editors. French Patent, P.V. 110,438 applied June 14, 1967. Marcilly C, Courty P, Delmon B. Preparation of highly dispersed mixed oxides and oxide solid solutions by pyrolysis of amorphous organic precursors. J Am Ceram Soc. 1970;53:56–7. Martinez-Rubio MI, Ireland TG, Fern GR, Silver J, Snowden MJ. A new application microgels: novel method for the synthesis of spherical particles of the Y2O3:Eu phosphor using a copolymer microgel of NIPAM and acrylic acid. Langmuir. 2001;17:7145–9. Mazaki H, Kakihana M, Yasuoka H. Synthesis of YBa2Cu4Oy from citrate sol–gel precursors. J Jpn Soc Powder Metall. 1991;38:211–4. Mei T, Zhu Y, Tang K, Qian Y. Synchronously synthesized core-shell LiNi1/3Co1/3Mn1/3O2/carbon nanocomposites as cathode materials for high performance lithium ion batteries. RSC Adv. 2012;2:12886–91. Mizushima K, Jones PC, Wiseman PJ, Goodenough JB. LixCoO2 (0 < x < 1): a new cathode material for batteries of high energy density. Mater Res Bull. 1980;15:783–9. Moskon J, Dominko R, Cerc-Korosec R, Gaberscek M, Jamnik J. Morphology and electrical properties of conductive carbon coatings for cathode materials. J Power Sources. 2007;174:683–8. Nadaud N, Nanot M, Boch P. Sintering and electrical properties of titania- and zirconia-containing In2O3-SnO2 (ITO) ceramics. J Am Ceram Soc. 1994;77:843–6. Niou CS, Ma YT, Li WP, Javadpour J, Murr LE. Preparation of superconducting YBa2Cu3O7-x powders by a solution technique. J Mater Sci-Mater Electron. 1992;3:181–6. Nityanand C, Nalin WB, Rajkumar BS, Chandra CM. Synthesis and physicochemical characterization of nanocrystalline cobalt doped lanthanum strontium ferrite. Solid State Sci. 2011;13:1022–30. Norton DP. Epitaxial growth of superconducting cuprate thin films. In: N. Khare, editors. Handbook of high-temperature superconductor. CRC Press, Boca Raton, Florida 2003. Nytén A, Abouimrane A, Armand M, Gustafsson T, Thomas JO. Electrochemical performance of Li2FeSiO4 as a new Li-battery cathode material. Electrochem Commun. 2005;7:156–60. Obradors X, et al. Chemical solution deposition: a path towards low cost coated conductors. Supercond Sci Technol. 2004;17:1055–64. Obradors X, et al. Progress towards all-chemical superconducting YBa2Cu3O7-coated conductors. Supercond Sci Technol. 2006;19:S13–26. Padhi AK, Nanjundaswamy KS, Goodenough JB. Phospho-olivines as positive-electrode materials for rechargeable lithium batteries. J Electrochem Soc. 1997;144:1188–94. Pang ML, et al. Patterning and luminescent properties of nanocrystalline Y2O3:Eu3+ phosphor films by sol–gel soft lithography. J Mater Sci-Mater Electron. 2003;100:124–31. Park TJ, Papaefthymiou GC, Viescas AJ, Moodenbaugh AR, Wong SS. Size-dependent magnetic properties of single-crystalline multiferroic BiFeO3 nanoparticles. Nano Lett. 2007;7:766–72. Pathak LC, Mishra SK. A review on the synthesis of Y-Ba-Cu-oxide powder. Supercond Sci Technol. 2005;18:R67–89. Patra H, Rout SK, Pratihar SK, Bhattacharya S. Effect of process parameters on combined EDTAcitrate synthesis of Ba0.5Sr0.5Co0.8Fe0.2O3-δ perovskite. Powder Technol. 2011;209:98–104.

28

T.O.L. Sunde et al.

Pechini MP. Method of preparing lead and alkaline earth titanates and niobates and coating method using the same to form a capacitor. In: Editor^Editors. US Patent 3:330,697 1967. Popa M, Crespo D, Calderon-Moreno JM, Preda S, Fruth V. Synthesis and structural characterization of single-phase BiFeO3 powders from a polymeric precursor. J Am Ceram Soc. 2007;90:2723–7. Predoana L, Jitianu A, Voicescu M, Apostol NG, Zaharescu M. Study of formation of LiCoO2 using a modified Pechini aqueous sol–gel process. J Sol–Gel Sci Technol. 2015;74:406–18. Psuja P, Hreniak D, Strek W. Fabrication, properties and possible applications of pure and Eu3+ doped SnO2 and In2O3/SnO2 (ITO) nanocrystallites. In: Proceedings of 2007 International Students and Young Scientists Workshop “Photonics and Microsystems”, STYSW 2007; 2007. Robertson JM, Van Tol MW. Epitaxially grown monocrystalline garnet cathode-ray tube phosphor screens. Appl Phys Lett. 1980;37:471–2. Rocha RA, Muccillo ENS. Synthesis and thermal decomposition of a polymeric precursor of the La2Mo2O9 compound. Chem Mater. 2003;15:4268–72. Rørvik PM, Tadanaga K, Tatsumisago M, Grande T, Einarsrud MA. Template-assisted synthesis of PbTiO3 nanotubes. J Eur Ceram Soc. 2009;29:2575–9. Rossen E, Reimers JN, Dahn JR. Synthesis and electrochemistry of spinel LT-LiCoO2. Solid State Ion. 1993;62:53–60. Rupich MW, Verebelyi DT, Zhang W, Kodenkandath T, Li X. Metalorganic deposition of YBCO films for second-generation high-temperature superconductor wires. MRS Bull. 2004;29:572–8. +539-541. Sanjines R, Ravindranathan Thampi K, Kiwi J. Preparation of monodispersed Y-Ba-Cu-O superconductor particles via sol–gel methods. J Am Ceram Soc. 1988;71:512–4. Schumm B, Wollmann P, Fritsch J, Grothe J, Kaskel S. Nanoimprint patterning of thin cadmium stannate films using a polymeric precursor route. J Mater Chem. 2011;21:10697–704. Schwartz RW, Schneller T, Waser R. Chemical solution deposition of electronic oxide films. C R Chim. 2004;7:433–61. Selbach SM, Einarsrud MA, Tybell T, Grande T. Synthesis of BiFeO3 by wet chemical methods. J Am Ceram Soc. 2007;90:3430–4. Selbach SM, Einarsrud MA, Grande T. On the thermodynamic stability of BiFeO3. Chem Mater. 2009;21:169–73. Serra OA, Cicillini SA, Ishiki RR. A new procedure to obtain Eu3+ doped oxide and oxosalt phosphors. J Alloys Compd. 2000;303–304:316–9. Shang M, et al. Blue emitting Ca8La2(PO4)6O2:Ce3+/Eu2+ phosphors with high color purity and brightness for white LED: soft-chemical synthesis, luminescence, and energy transfer properties. J Phys Chem C. 2012;116:10222–31. Shao J, Tao Y, Wang J, Xu C, Wang WG. Investigation of precursors in the preparation of nanostructured La0.6Sr0.4Co0.2Fe0.8O3-δ via a modified combined complexing method. J Alloys Compd. 2009;484:263–7. Shetty S, Palkar VR, Pinto R. Size effect study in magnetoelectric BiFeO3 system. Pranama J Phys. 2002;58:1027–30. Shi D, et al. The development of YBa2Cu3Ox thin films using a fluorine-free sol–gel approach for coated conductors. Supercond Sci Technol. 2004;17:1420–5. Shiomi Y, Asaka T, Tachikawa K. Superconducting properties and structures of high-Tc oxides prepared by a citric acid salt process. IEEE Trans Appl Supercond. 1993;3:1170–3. Shur MS, Žukauskas A. Solid-state lighting: toward superior illumination. Proc IEEE. 2005;93:1691–703. Sladkevich S, et al. Antimony doped tin oxide coating of muscovite clays by the Pechini route. Thin Solid Films. 2011;520:152–8. Sletnes M, Skjærvø SL, Lindgren M, Grande T, Einarsrud MA. Luminescent Eu3+-doped NaLa (WO4)(MoO4) and Ba2CaMoO6 prepared by the modified Pechini method. J Sol–Gel Sci Technol. 2016;77:136–44.

Modified Pechini Synthesis of Oxide Powders and Thin Films

29

Smet PF, Parmentier AB, Poelman D. Selecting conversion phosphors for white light-emitting diodes. J Electrochem Soc. 2011;158:R37–54. Stöber W, Fink A, Bohn E. Controlled growth of monodisperse silica spheres in the micron size range. J Colloid Interface Sci. 1968;26:62–9. Sunde TOL, et al. Transparent and conducting ITO thin films by spin coating of an aqueous precursor solution. J Mater Chem. 2012;22:15740–9. Sunde TOL, Einarsrud MA, Grande T. Solid state sintering of nano-crystalline indium tin oxide. J Eur Ceram Soc. 2013;33:565–74. Sunde TOL, Einarsrud MA, Grande T. Optimisation of chemical solution deposition of indium tin oxide thin films. Thin Solid Films. 2014;573:48–55. Tai LW, Lessing PA. Modified resin – intermediate processing of perovskite powders: Part I. Optimization of polymeric precursors. J Mater Res. 1992a;7:502–10. Tai LW, Lessing PA. Modified resin – intermediate processing of perovskite powders: Part II. Processing for fine, nonagglomerated Sr-doped lanthanum chromite powders. J Mater Res. 1992b;7:511–9. Tao S, Irvine JTS. Synthesis and characterization of (La0.75Sr0.25)Cr0.5Mn0.5O3-δ, a redox-stable, efficient perovskite anode for SOFCs. J Electrochem Soc. 2004;151:A252–9. Tarascon JM, Wang E, Shokoohi FK, McKinnon WR, Colson S. Spinel phase of LiMn2O4 as a cathode in secondary lithium cells. J Electrochem Soc. 1991;138:2859–64. The IUPAC Stability Constants Database, Academic Software. http://www.acadsoft.co.uk/ Thuy TT, et al. Sol–gel chemistry of an aqueous precursor solution for YBCO thin films. J Sol–Gel Sci Technol. 2009;52:124–33. Van der Biest O, et al. Ceramic superconductors synthesized by sol–gel methods. Phys C. 1991;190:119–21. Vojnovich T, Bratton RJ. Impurity effects on sintering and electrical resistivity of indium oxide. Am Ceram Soc Bull. 1975;54:216–7. Wang J, et al. Epitaxial BiFeO3 multiferroic thin film heterostructures. Science. 2003;299:1719–22. Wang H, Lin CK, Liu XM, Lin J, Yu M. Monodisperse spherical core-shell-structured phosphors obtained by functionalization of silica spheres with Y2O3:Eu3+ layers for field emission displays. Appl Phys Lett. 2005;87:1–3. Wang WT, et al. Chemical solution deposition of YBCO thin film by different polymer additives. Phys C. 2008a;468:1563–6. Wang Z, et al. NaEu0.96Sm0.04(MoO4)2 as a promising red-emitting phosphor for LED solid-state lighting prepared by the Pechini process. J Lumin. 2008b;128:147–54. Wang H, Xu X, Li X, Zhang J, Li C. Synthesis and sintering of indium tin oxide nanoparticles by citrate-nitrate combustion method. Rare Met. 2010;29:355–60. Wu MK, et al. Superconductivity at 93 K in a new mixed-phase Y-Ba-Cu-O compound system at ambient pressure. Phys Rev Lett. 1987;58:908–10. Xia H, Wang H, Xiao W, Lu L, Lai MO. Properties of LiNi1/3Co1/3Mn1/3O2 cathode material synthesized by a modified Pechini method for high-power lithium-ion batteries. J Alloys Compd. 2009;480:696–701. Xiong L, Xu Y, Tao T, Goodenough JB. Synthesis and electrochemical characterization of multications doped spinel LiMn2O4 used for lithium ion batteries. J Power Sources. 2012;199:214–9. Yang YM, et al. Characterization of YBa2Cu3O7-x bulk samples prepared by citrate synthesis and solid-state reaction. J Appl Phys. 1989;66:312–5. Ye S, Li Y, Yu D, Yang Z, Zhang Q. Structural effects on Stokes and anti-Stokes luminescence of double-perovskite (Ba, Sr)2CaMoO6: Yb3+, Eu3+. J Appl Phys. 2011;110:013517. Yu M, Lin J, Wang SB. Effects of x and R3+ on the luminescent properties of Eu3+ in nanocrystalline YVxP1-xO4:Eu3+ and RVO4:Eu3+ thin-film phosphors. Appl Phys Mater Sci Process. 2005;80:353–60. Zalga A, Moravec Z, Pinkas J, Kareiva A. On the sol–gel preparation of different tungstates and molybdates. J Therm Anal Calorim. 2011;105:3–11.

30

T.O.L. Sunde et al.

Zhang H, Fu X, Niu S, Xin Q. Synthesis and photoluminescence properties of Eu-doped AZrO3 (A = Ca, Sr, Ba) perovskite. J Alloys Compd. 2008;459:103–6. Zhang L, et al. Li3V2(PO4)3@C/graphene composite with improved cycling performance as cathode material for lithium-ion batteries. Electrochim Acta. 2013;91:108–13. Zhao Y, Sun G, Wu R. Synthesis of nanosized Fe-Mn based Li-rich cathode materials for lithiumion battery via a simple method. Electrochim Acta. 2013;96:291–7. Zhu WZ, Deevi SC. Development of interconnect materials for solid oxide fuel cells. Mater Sci Eng A. 2003;348:227–43.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications Douglas A. Loy

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sol-Gel Polymerizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surfactant and Polymer Templating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hierarchical Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pendant Polysilsesquioxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Copolymers of Silsesquioxanes and Silica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bridged Polysilsesquioxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Properties of Mesoporous Polysilsesquioxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polysilsesquioxane Xerogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polysilsesquioxane Aerogels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surfactant and Polymer Templated Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of Mesoporous Polysilsesquioxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adsorbents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anticorrosion Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chromatographic Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Low k Dielectric Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Membranes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sunscreens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 4 4 5 5 6 7 7 11 12 16 19 22 22 23 24 25 25 26 26 26 27 29 29 30 30 30

D.A. Loy (*) Department of Materials Science and Engineering, Department of Chemistry and Biochemistry, The University of Arizona, Tucson, AZ, USA e-mail: [email protected] # Springer International Publishing AG 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_131-1

1

2

D.A. Loy

Abstract

Mesoporous polysilsesquioxanes are an important class of hybrid organicinorganic materials with numerous applications in microelectronics, membranes, chromatographic materials, catalysis, and more. Prepared by sol-gel polymerization of monomers with trialkoxysilyl groups, these materials are based on siloxane networks modified by organic groups. Mesoporosity in these materials is formed through sol-gel polymerizations by themselves or can be directed into ordered arrays using surfactant templating. Surfactant templating allows mesopores in geometries ranging from cylindrical pores to gyroid porosities. Surfactant templating also enables the facile development of hierarchical structures where multiple modes of pores can independently coexist in the same material. Formulations for preparation of all of these mesoporous materials as monoliths, particles, and coatings will be discussed and examples will be provided.

Introduction Polysilsesquioxanes are a class of hybrid organic-inorganic materials that has developed from silane coupling agent chemistry into a class of thermally stable materials with tailored porosity and organic chemical functionalities. Based on a monomer repeat unit with up to three siloxane linkages and one organic group for every silicon atom, polysilsesquioxanes can form cross-linked networks with mesoporosity (2 < width < 50 nm) when solidified around solvent, self-assembled surfactants or phaseseparated polymers. The fundamental macromolecular architecture of polysilsesquioxanes is that of interconnected branched and cyclic structures. These networks are formed by the sol-gel polymerization of organotrialkoxysilane precursors in which the alkoxide groups hydrolyze and the resulting silanols condense to afford Si–O–Si linkages (Figs. 1 and 2). In this review, polysilsesquioxanes are divided into pendant polysilsesquioxanes, silica-silsesquioxane copolymers, and bridged polysilsesquioxanes. Pendant polysilsesquioxanes are those with organic groups attached by a single Si–C bond to the silicon atom (Fig. 1). This makes the organic group, if it is larger than a methyl group, act to block further development of the siloxane network. As a result, polyhedral oligomeric silsesquioxanes (POSS) can be formed where the organic groups block network development in all directions. POSS will not be further discussed in this review but have been extensively reviewed elsewhere (Cordes et al. 2010). More extended networks are possible, but attaining high enough molecular weight to form solid particles necessary for gelation or building a periodic mesoporous structure is not common for pendant polysilsesquioxanes. An approach that minimizes the blocking effect of the pendant groups is to copolymerize the organotrialkoxysilane with silica or other metal oxides monomers that more readily form high molecular weight polymers and gels (Lim and Stein 1999). These materials are often called organosilicates or Ormosils.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

Fig. 1 Hydrolysis polysilsesquioxanes

and

condensation

of

organotrialkoxysilanes

to

3

afford

pendant

Bridged polysilsesquioxanes are materials where two or more silsesquioxane groups are connected through an organic bridging group (Fig. 2). These are also formed by a sol-gel polymerization, most commonly of monomers with two or more trialkoxysilyl groups attached to an organic bridging group. The result of the bridged configuration is that the organic group facilitates network growth rather than acting as a blocking agent. Consequently, bridged polysilsesquioxanes readily form gels without the need of a silica comonomer, at monomer concentrations considerably lower than the pendant polysilsesquioxanes systems. Surfactant and polymer templating are techniques that replace solvent porogens with supermolecular templates created by phase separated solvent-surfactant or solvent-polymers systems. Organotrialkoxysilane or bridged monomers polymerize in the solvent phase and, once the templates are extracted, materials with periodic mesopores (periodic mesoporous organosilica or PMO) are obtained. The geometries of the PMO reflect the diverse geometries of the surfactant systems. If the polysilsesquioxanes scaffolding in the PMO is itself porous, the result is a hierarchical material. This chapter will cover the methods for preparing porous polysilsesquioxanes by sol-gel polymerizations with and without surfactant templating. It will also describe the properties of sol-gel processed xerogels and aerogels, surfactant-templated materials, and hierarchical structures, and some of this collection of materials applications.

4

D.A. Loy

Fig. 2 Hydrolysis and condensation of monomers with two trialkoxysilyl groups to afford bridged polysilsesquioxanes

General Procedures Porosity Porosity of materials provides surface area on which adsorption and catalysis can occur and pores whose diameter selects what species are admitted to the surface (Barton et al. 1999). Pores are divided by convention into micropores (pore width < 2 nm), mesopores (2 < pore width < 50 nm) and macropores (pore width > 50 nm). They are most frequently characterized by nitrogen porosimetry that affords sorption isotherms from which surface area, pore volume, pore size, and pore size distributions can be calculated using various models. Sol-gel processed silica often exhibits broad distributions of pores ranging from micropores to macropores. The pores in silica gels arising from fractal surfaces developed in the formation and aggregation of particles making up the materials. In contrast, micropores in crystalline zeolites arise from the crystal structure and are narrow enough to allow molecular sieving. Sol-gel processed materials typically exhibit broader pore size distributions than crystalline materials, but until surfactant templating was developed, the pore sizes of zeolites were limited to less than 2 nm. With surfactant and polymer templating, mesoporous materials with narrow pore size distributions were possible up to the macropore domain.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

5

Sol-Gel Polymerizations Sol-gel polymerization refers to the chemical creation of solid particles from monomeric precursors and the percolation of the particles in a continuous solvent phase to afford a gel. Sol-gel polymerizations most commonly start with pendant or bridged trialkoxysilyl monomers, occasionally with tetraalkoxysilane comonomers, reacting with water in the presence of a base, acid, or fluoride catalyst. Since most monomers are sparingly soluble in water, alcohols or other solvents capable of forming a homogeneous polymerization solution are used. The solvents provide a continuous liquid phase for particle nucleation and gelation. They also serve as porogens or pore templates that can be preserved into the dried gels as air-filled pores. Surfactant and polymer templating are the logical extensions of using solvent porogens. Sol-gel polymerizations proceed by hydrolysis of the alkoxide groups to yield silanols, followed by condensation of silanols or, in the case of methoxysilane monomers, silanols and methoxysilanes to afford Si-O-Si linkages. Cyclization to tetrameric rings occurs quickly once linear tetramers form. Sol-gel polymerization of monomers with trichlorosilyl groups is also possible but, due to their greater reactivity and the generation of large quantities of HCl, are rarely used. The most common alkoxide groups in these monomers are either methoxide or ethoxide, with the latter showing significantly slower rates of hydrolysis and a greater propensity to form cyclic siloxane structures due to sterics. Alkaline catalysts include sodium or potassium hydroxide or aqueous ammonia, while hydrochloride acid is the most commonly used acid catalyst. Fluoride, despite its greater efficacy at catalyzing polymerizations, is used relatively rarely in preparing polysilsesquioxanes. Like silica systems, the extent of reaction in polysilsesquioxanes is higher than would be expected from the Flory–Stockmeyer equation, indicating formation of cyclic structures is successfully competing with network formation. Furthermore, the extents of reaction are greater with ethoxide monomers than methoxide monomers, pendant than bridged polysilsesquioxanes, and basic catalysts than acidic catalysts. Depending on the sol-gel polymerization, the result can be particles, coatings, or gels. Particles can be precipitated or used in suspension (in the sol). Coatings are deposited by dip-coating, spin-coating, or spraying. Gels can be air-dried to afford monolithic xerogels or supercritically dried to produce monolithic aerogels. Gels can also be broken up into granules or particles then processed by air- or supercriticaldrying or by washing with water, solvent, and drying by filtration. This allows larger quantities of particles to be prepared with an added element of control over porosity.

Surfactant and Polymer Templating Surfactant templating developed out of micropore templating of zeolites, save that the mesoscale order of the self-assembled surfactant or phase separated polymer (s) allows the structural material to be organized without being crystalline (Beck et al. 1992). Thus, many templated silicas will have small angle X-ray diffraction patterns from the mesoporous structure, but no wide angle scattering from the silica.

6

D.A. Loy

Over the last few decades, a variety of mesoporous architectures have been prepared from an even vaster field of surfactants and polymer templates. The self-assembled surfactants can be lamellar, hexagonal close-packed cylinders (p6mm) (Cho and Char 2004; Park et al. 2008), simple cubic (Pm3n) (Sayari et al. 2000; Rebbin et al. 2012), face-centered cubic (Fm3m) (Liang et al. 2005; Manchanda and Kruk 2016), 3D body-centered cubic (Im3m) (Grudzien et al. 2006; Cho et al. 2012), and bicontinuous cubic, gyroidal networks (Ia3d) (Wang et al. 2003; Xia et al. 2014) (Fig. 3). Sol-gel polymerization around the template generates the material architecture, after which the template is removed to liberate pores where the surfactant or polymer resided (Fig. 4). In silica gels and other fully inorganic materials, the templates are calcined with air and high temperature, a process that lends greater stability to the resulting mesoporous material. In organically modified materials, such as the polysilsesquioxanes, the templates are typically extracted with solvent since calcining may also destroy the organic constituents. Polymer templating also involves phase-segregated templates during the sol-gel polymerization with formation of the organically modified silica in the solvent phase. These polymers are block copolymer surfactants that allow the same structures as with small molecule surfactants to be formed, but in greater scale (Goto and Inagaki 2002). The size of the resulting pores can be further increased by swelling the template with added solvents. Alternatively, the solutions of polysilsesquioxane can undergo physical phase separation due the changes in the thermodynamics of mixing as the sol-gel polymerization proceeds. Nucleated phase separation provides large spherical pores while spinodal phase separation generates bicontinuous structures (Fig. 5) (Kanamori et al. 2011; Kanamori and Nakanishi 2011). Once the self-assembled templates have been removed, the periodic materials are characterized by small angle (2θ < 10
) X-ray diffraction, transmission and scanning electron microscopies, nitrogen sorption porosimetry, solid state Si-29 & C-13 NMR, and elemental analysis (if the materials are new compositions). Small angle XRD and transmission electron microscopy will allow periodicity to be identified and assigned to a space group (Fig. 3). Nitrogen sorption porosimetry will allow the surface area and pore size to be determined so that the latter, in particular, can be compared to the d-spacings observed in the TEM or XRD. Most periodic organosilicas exhibit high surface areas and nitrogen sorption Type IV isotherms with a significant hysteresis, often associated with noncylindrical pores, but so do many silsesquioxane xerogels made without templating (Loy et al. 2000); (Hu and Shea 2011). Characterization of the molecular composition using NMR and elemental analyses is also important, particularly for copolymer-based systems.

Hierarchical Materials When a single material has several different architectures defined at different length scales, the material is said to have a hierarchical structure. This can be accomplished by incorporating microporosity through crystallinity or solvent templating of the

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

7

Fig. 3 Some of the mesophase structures (space groups) possible through templating (Reprinted with permission (Innocenzi et al. 2009). Copyright (2009) American Chemical Society)

sol-gel structure, mesoporosity through surfactant or polymer templating, spinodal phase separation (Mushiake et al. 2012; Guo et al. 2015), and macroporosity through colloidal templates (Jun et al. 2005). For example, sol-gel polymerization of 1,2bistriethoxysilylethane in the presence of P-123 or F127 block copolymers as surfactant templates and urea to induce spinodal phase separation, mesoporosity and macroporosity can be created in the same material (Mushiake et al. 2012). In some of these systems, there appears to be additional microporosity within the walls of the mesostructures (Brandhuber et al. 2006).

The Materials Pendant Polysilsesquioxanes Polymerization Without Templating There are hundreds of commercially available trialkoxysilanes with pendant organic groups (Fig. 6) and thousands more synthesized due to their utility as silane coupling agents for surface modification, composite interface engineering, and, more recently, the preparation of polyhedral oligosilsesquioxanes and mesoporous organosilicas. As a result, the hydrolysis and condensation chemistry of organotrialkoxysilanes

8

Fig. 4 Surfactant templating of silica with surfactant in cylindrical mesophases

Fig. 5 Mesoporous polysilsesquioxanes prepared by spinodal phase separated during the sol-gel polymerization in methanol with formamide (Reprinted with permission (Kanamori et al. 2009). Copyright (2009) Royal Society of Chemistry)

D.A. Loy

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

9

was extensively investigated as it pertained to surface modification. Formation of gels was an indication of a formulation with poor shelf life and was not vigorously investigated. Due to the electron donating behavior of most organic groups, organotrialkoxysilanes hydrolyze and undergo condensation more quickly than the analogous tetraalkoxysilanes under acidic conditions, but slower under basic conditions or with fluoride. Compared with the relatively facile gelation of tetrafunctional silica monomers, organotrialkoxysilanes only gel with small organic substituents or functional organic groups that can react under sol-gel conditions to generate additional connectivity needed to form networks (Fig. 1). Instead, sol-gel polymerization frequently results in phase separation of droplets of liquid oligosilsesquioxanes, or crystallization of POSS (Fig. 7), even when polymerizations are performed without solvent. The pendant polysilsesquioxanes most commonly formed as gels are hydrido-, chloromethyl-, and methyl-substituted polysilsesquioxanes (Loy et al. 2000). These are also the only pendant polysilsesquioxanes that regularly form transparent or translucent gels (Fig. 8). Polyhydridosilsesquioxane has Si-H groups that react completely with base catalysts to afford silica gels. Only under acidic conditions, do the gels retain the hydride groups. Vinyl-, ethyl-, cyanoethyl-, and chloromethylphenyl-substituted polysilsesquioxane gels are opaque white in appearance. Dodecyl-, hexadecyl-, and octadecyl-substituted polysilsesquioxane gels are also white and opaque, but unlike the gels mentioned above, these will turn back into a liquid with heating or dissolve in hydrocarbon solvents. Thermoreversible gels are also obtained with polysilsesquioxanes modified with amine groups, like aminopropylsilsesquioxane and ethylenediaminopropylsilsesquioxane and imidazolepropylsilsesquioxane, but these are rubber and soluble (Sanchez and Loy 2001). Some organotrialkoxysilanes (R = phenyl, mercaptopropyl) that do not form gels or even solid polymers upon removal of solvent under sol-gel conditions will form solid particles (Arkhireeva et al. 2004). However, these particles are soluble and melt at relatively low temperatures. Lastly, a number of pendant silsesquioxane monomers will form gels through chemical reactions of the organic group. These systems form in situ bridging groups which increase the functionality of the monomer and make it easier to build polymers with the high molecular weights needed for phase separation of colloids. Systems that most readily react during sol-gel polymerizations include monomers with isocyanate (Fig. 9a), epoxy (Fig. 9b), alkyl chloride, bromide, or iodides (Fig. 9c) that react with comonomers with a nucleophilic functionality on its organic group or additives with two or more nucleophilic groups, alkene groups in combination with a second group bearing thiols and with exposure to light (Fig. 9d), and ligand groups (amines, thiols, etc.) in the presence of metal ions (Fig. 9e). This reactivity makes it relatively easy to prepare new monomers or incorporate metals into organosilicas, but also can be a problem in retaining water reactive groups, like isocyanates or epoxides, through the hydrolysis and condensation conditions used in sol-gel polymerizations or templating.

10

D.A. Loy

Fig. 6 Some examples of some of the many commercially available organotrialkoxysilanes

With Surfactant and Polymer Templating Because most pendant polysilsesquioxanes are not able to afford gels, surfactant and polymer templating (Fig. 10) is not as common with this class as with the silicasilsesquioxane and bridged polysilsesquioxanes discussed later in the chapter. The list of monomers capable of forming surfactant templated materials is very short because very few pendant polysilsesquioxanes give rise to intractable solids without having silica or bridged polysilsesquioxane comonomers or through the formation of bridging groups with the orthogonal reaction of functionalities on the organic groups (Fig. 9). The only examples that were found for this review were prepared from hydridotrialkoxysilanes (Xie et al. 2011), methyltrialkoxysilanes (Nakanishi and Kanamori 2005; Tan and Rankin 2006), and vinyltrialkoxysilanes (Nakanishi and Kanamori 2005). We have not been able to find periodic mesoporous

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

11

Fig. 7 Crystals of ethyl substituted T8 POSS

organosilicas made from chloromethyl-, chloromethylphenyltrialkoxysilanes.

ethyl-,

cyanoethyl-,

or

Copolymers of Silsesquioxanes and Silica Without Templating The inability of many pendant silsesquioxane monomers to make gels can be alleviated by copolymerizing the monomers with tetramethoxysilane or tetraethoxysilane. The addition of 25 mol% tetraalkoxysilanes to most organotrialkoxysilane monomer will generally afford solid colloids in a sol that can form a gel (Schwertfeger et al. 1994). However, unless the concentration of tetraalkoxysilane is greater than 75%, these gels often collapse during drying to afford nonporous solids. Many cases where “polysilsesquixoane gels” have been reported are actually copolymers of organotrialkoxysilanes with tetraalkoxysilanes, silsesquioxane-silica copolymers. Surface modification with organic groups can be directed by reacting less reactive organotriethoxysilanes with more reactive tetramethoxysilanes. With Surfactant and Polymer Templating The most common approach using templating with silica and silsesquioxane systems is to surfactant template the silica, then surface modify the silica surface with organotrialkoxysilane once the template has been removed (Fig. 11) (Lim and Stein 1999; Moller et al. 1999). This grafting approach will modify the surface tension and chemistry of the silica. Surface modification also stabilizes the mesoporous silica and reduces the diameter of pores (Stein et al. 2000). Alternatively, copolymerization of tetraalkoxysilane with organotrialkoxysilanes can be used as a one-pot formulation of the organically modified silica (Fig. 12), as long as the silica:organotrialkoxysilane ratio is high enough to ensure phase

12

D.A. Loy

Fig. 8 Formulation of polymethylsilsesquioxane gels

separation of solid particles needed to build the mesostructured material (Lim et al. 1997). Table 1 has examples of periodic mesoporous copolymers of organotrialkoxysilanes with tetraalkoxysilanes or bridged monomers as the crosslinking agents.

Bridged Polysilsesquioxanes Monomers with two or more trialkoxysilyl groups on a bridging organic group have six or more potential siloxane linkages and are known to polymerize readily to afford gels (Loy and Shea 1995) (Fig. 2). Before 1989, only a handful of reports described the ethylene-, phenylene-, and acetylene-bridged polysilsesquioxanes as coatings

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

13

Fig. 9 Pendant silsesquioxanes reacting to generate bridging groups in situ during sol-gel polymerizations

and dipropylamine- and dipropyltetrasulfide-bridged polysilsesquioxanes as coupling agents. Today, there are thousands of papers and patents based on bridged polysilsesquioxanes or their monomers. A large number of bridged polysilsesquioxane monomers are now commercially available (Fig. 13) and more can be readily synthesized from commercial coupling agents (Fig. 9).

Sol-Gel Polymerizations Without Templating Since bridged monomers have more than one trialkoxysilyl group, more water is needed for their sol-gel polymerization (Fig. 2) (Shea and Loy 2001). Most bridged polysilsesquioxanes will form as gels relatively quickly at concentrations well below 1Mole/Liter. Only those with flexible alkylene-bridging groups (ethylene, propylene, butylene) take longer to gel at these lower concentrations, due to their intramolecular cyclization early in the sol-gel process. Even so, at concentrations typically used to prepare silica gels from tetraethoxysilane, 1,3-bis(triethoxysilyl)propane reacts under acidic conditions to form gels in seconds. Bridged polysilsesquioxanes gels form with acid, base, or fluoride catalysts. Alkylene-bridged gels are more frequently opaque, with arylene-bridged gels are more often transparent. However, the optical properties of the gel will vary with monomer, solvent, and catalyst, so transparency should be determined experimentally. In Fig. 14, methods for preparing phenylene-bridged gels in ethanol with HCl, NaOH, or ammonium fluoride catalyst are shown along with the photographs of the gels. Whether the methoxy or ethoxy monomer and the nature of the bridging group and catalyst used all influence the ultimate appearance of the gels. The gels form irreversibly and generally undergo syneresis during aging (minutes to hours after

14

D.A. Loy

Fig. 10 Surfactant templating of pendant silsesquioxanes, R = H, Me, Et, vinyl

gelation). Gels should typically be aged until after syneresis to obtain the highest surface area.

With Surfactant and Polymer Templating Surfactant templating of bridged polysilsesquioxanes (Fig. 15) has been extensively reported and reviewed (Mizoshita et al. 2011; Van Der Voort et al. 2013). As with sol-gel polymerizations, bridged monomers in templated systems react quickly and establish robust scaffolding that prevents collapse of porosity. These bridged systems have replaced pendant polysilsesquioxanes-modified silica and cocondensed silsesquioxane-silica in the current literature on periodic mesoporous materials. Table 2 lists a variety of templated, bridged polysilsesquioxanes. An example of a formulation leading to a periodic mesoporous, phenylene-bridged polysilsesquioxanes with cubic morphology is shown in Fig. 16.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

15

Fig. 11 Mesoporous, organically modified silica made by surface modification after templating

A number of researchers had speculated about ordered structure at molecular length scales in bridged polysilsesquioxanes, particularly in confined environments where intramolecular cyclization is more favorable (Ben et al. 2000). Inagaki discovered that the bicontinuous mesophase provided just such an environment (Inagaki et al. 2002). Instead of amorphous particles forming in the water phase of the mesostructured surfactant water system then aggregating to form the structures around the templates, the walls of the surfactant templated, phenylene-bridged polysilsesquioxanes were found to be ordered polymeric structures (Fig. 17). It appears that having rigid bridging groups capable of pi stacking or bridges with strong interbridged nonbonding interactions is necessary to achieve these ordered structures. Most mesoporous, bridged polysilsesquioxanes with nonrigid bridging groups are made of amorphous aggregates of colloidal particles.

Hierarchical Materials Hierarchical materials using bridged polysilsesquioxanes most frequently use particle, fiber or ribbon morphology as the longest length scale feature. Surfactanttemplated mesoscale features make up the mid-length scale structures and microporosity with the walls of the mesopores makes up the smallest morphological phase, though some have counted the macromolecular and monomeric structures as addition, smaller levels in the hierarchy (Brandhuber et al. 2006). Monolithic foams based on spinodal phase separation of macrostructures, surfactant templating of

16

D.A. Loy

Fig. 12 Surfactant templating of copolymers of silica and organotrialkoxysilanes

mesopores, and microporosity in mesopore walls have also been reported with bridged polysilsesquioxanes (Kanamori and Nakanishi 2011).

Coatings Significant advances in coating techniques have resulted from the development of polysilsesquioxane low κ dielectric materials. Coatings of sol-gel processed polysilsesquioxanes, silica-polysilsesquioxanes, or bridged polysilsesquioxanes are applied to surfaces as the sol before gelation is reached. Coatings are commonly be applied by spin-coating, dip-coating, or spraying the sols at the appropriate viscosities (30–40 centipoise) (Abe et al. 2000). One concern with sol-gel coatings is the shelf-life of the sol if the polymerization chemistry continues. Bridged polysilsesquioxane systems need lower concentration sols or sub-stoichiometric water to prevent coating solutions from gelling.

Surfactant

Monomer

1208/3.6

963/4.0

p6mm

p6mm

Cubic Pm3n P6mm

NaOH

HCl

Surfactant CO2H NaF

740/2.8

1424/2.2

Space group p6mm

Reagent NaOH

Surface area (m2/g-1) /pore diam. (nm) Not reported/3.9

(Brown et al. 2000)

(Che et al. 2003)

(Nie et al. 2004)

(Lim et al. 1997)

Ref (Fowler et al. 1997)

Table 1 Pendant silsesquioxanes cocondensed with silica or bridged polysilsesquioxanes to afford organically modified periodic mesoporous materials

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications 17

18

D.A. Loy

Fig. 13 Commonly, commercially available bridged polysilsesquioxane monomers

Fig. 14 Sol-gel polymerization and gelation of phenylene-bridged polysilsesquioxanes made from 1,4-bis(triethoxysilyl)benzene

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

19

Fig. 15 Surfactant templating of bridged polysilsesquioxanes

Coating with surfactant template silica-silsesquioxanes and bridged polysilsesquioxanes can be readily accomplished by applying the solutions by evaporation induced self-assembly (Castricum et al. 2014), electrochemically assisted selfassembly (Herzog et al. 2013), or spin coating (Zhang et al. 2016). Development of periodic mesoporous structures can be detected using microscopy and XRD (de Theije et al. 2003) and characterized by scraping off the coating for surface area, TEM, and solid-state NMR characterization (Wahab and He 2009).

Particles Particles are made when the colloids generated by sol-gel polymerization do not achieve percolation and gel. The prerequisite of phase separation of a solid particle is common to both gelation and particle synthesis. This would lead one to expect that if you can make particles of a polymer, it should be possible to form a gel. However, this has not been the case for many pendant polysilsesquioxanes. These pendant

Surfactant

xylene

Monomer

Ia3d

Fm3m

HCl/ NaCl HCl/KCl

Im3m

P6mm

HCl

HCl

Pm3n

NaOH

P6mm

Not reported

HCl

HCl

Space group P6mm

Reagent NaOH

Table 2 Surfactant templating of bridged polysilsesquioxanes

1080/5.6

1150/9.9

514/21

833/5.1

1182/4.6

877/3.8

1222/2.9

Surface area (m2/g-1) /pore diam. (nm) 818/3.8

(Hao et al. 2012)

(Hao et al. 2010)

(Mandal and Kruk 2010)

(Xia et al. 2014)

(Kapoor et al. 2005)

(Lin et al. 2014)

(Ren et al. 2002)

Ref (Inagaki et al. 2002)

20 D.A. Loy

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

21

Fig. 16 Formulation leading to high surface area, mesoporous phenylene-bridged polysilsesquioxane with Im3m structure (Cho et al. 2009)

Fig. 17 Amorphous aggregated particulate structure in template walls (left) versus highly ordered, polymeric walls (right)

polysilsesquioxanes have been prepared as solid particles, but not gels. These include mercaptopropyl-, phenyl-, styryl-, and methacryloxypropyl-substituted silsesquioxanes (Arkhireeva et al. 2004). The particles have been successfully prepared using Stober conditions or emulsion polymerizations. In many cases the particles are not intractable and will melt with heating or dissolve in more nonpolar solvents than the solvents used in their preparation. A typical procedure for their preparation is given in Fig. 18. The challenge with making particles from bridged monomers is that concentrations lower than 0.1 M are necessary to avoid gelation. Exact application of Stober conditions to bridged monomers for particle formation is mostly unsuccessful. Changes in solvent are often necessary with each monomer and particle size distributions are often broader than would be desired (Khiterer and Shea 2007). Emulsion polymerizations have been successfully used to prepared monodispersed

22

D.A. Loy

particles for chromatographic applications (O’Gara and Wyndham 2006). Surfactant templating uses low concentrations of monomer that get concentrated in the selfassembled template and so particle formation is easier than monolithic gels or thin films.

Properties of Mesoporous Polysilsesquioxanes Polysilsesquioxane Xerogels Porosity Xerogels refer to “dried” gels or gels that were dried by allowing solvent to evaporate. Air-drying or work-up with water and solvent provide solvent drying fronts in the gels that lead to shrinkage and collapse of a significant fraction of porosity that existed in the gel. As a result xerogels have smaller pores than expected. Polysilsesquioxane xerogels prepared from trimethoxysilyl monomers with pendant hydrido- and methyl-groups have surface areas between 400 and 620 m2/g with pore diameters between 4.0 and 6.7 nm (Loy et al. 2000). The analogous xerogels prepared from the triethoxysilyl monomers have slightly higher surface areas for the hydrido materials with comparable pore diameters and lower surface areas and larger pores for the methyl materials. Most gels prepared from vinyl, chloromethyl, and chloromethylphenyl monomers are nonporous or have surface areas under 100 m2/g. In contrast, bridged polysilsesquioxane xerogels generally have much higher surface areas than pendant polysilsesquioxanes or silica xerogels. Bridged polysilsesquioxane xerogels often shrink 80–95% during drying. The porosity of bridged polysilsesquioxane xerogels can be controlled through the nature of the organic bridging group, especially when prepared under acidic conditions. Bridged polysilsesquioxanes with rigid arylene, alkenylene, and alkynylene groups are porous with surface areas as high as 1200 m2/g and significant micropores and mesopores whether or not they are prepared under acidic or basic conditions. Bridged polysilsesquioxanes prepared with flexible bridging groups longer than six carbons in length under acidic conditions are nonporous. The same xerogels prepared under basic conditions are mesoporous with surface areas as high as 600 m2/g hundreds of meters squared. One of the key advantages of surfactant templated bridged polysilsesquioxanes is that they can be made under acidic conditions with long, flexible bridging groups and still be porous, whereas the sol-gel processed ones are nonporous (Lin et al. 2015). Pore size distributions for some alkylene- and arylenebridged xerogels are as narrow as many surfactant template systems (Fig. 19). Mechanical Properties Outside of film hardness, very little has been reported regarding the mechanical properties of bridged polysilsesquioxanes. A recent study of the flexural strength and moduli of phenylene- and hexylene-bridged polysilsesquioxane xerogels was conducted on monolithic cylindrical xerogels prepared under basic conditions

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

23

Fig. 18 Formation of polyphenylsilsesquioxane particles (Li et al. 2013)

(Boday et al. 2012). The xerogels had surface areas of 667  33 m2/g and 472  23, respectively, and average pore diameters of 4 nm. Phenylene-bridged xerogels, with densities near 0.79 g/cc, had flexural strengths of 28.0  2.0 MPa and moduli of 124  13 MPa. The hexylene-bridged xerogels of similar density had flexural strengths of 26.0  1.0 MPa and moduli of 119  23 MPa.

Thermal Properties Polysilsesquioxanes are often touted for their thermal stability. For pendant polysilsesquioxanes, those that do not form crystalline oligomers (POSS) or intractable, cross-linked solids, have relatively low glass transition and “melting” temperatures. In those that are cross-linked and intractable, thermal stability will depend on the organic group and the thermal stabilities described below concern the degradation of the organic group, not the resulting ceramic products of the degradation that often are stable to 700
C or higher. Polysilsesquioxanes with aryl substituents or bridging groups are typically stable to over 400–500
C, even in air. In contrast, polysilsesquioxanes with alkyl substituents and alkylene-bridging groups are only stable to 300–400
C and allylic, acetylenic, and alkyl groups with leaving groups (halides, alkoxides, acetates) beta to silicon will degrade at even lower temperatures. This thermal instability has been used to modify the porosity of bridged polysilsesquioxanes through the thermal cleavage of bridging groups, such as carbonates or carbamates (Loy et al. 1999). As with many organosilicon-based materials, thermolysis can lead to the formation of silicon-oxycarbide products that are stable (Shea et al. 1992).

Polysilsesquioxane Aerogels Porosity Silica-silsesquioxane (Schwertfeger et al. 1994) and bridged polysilsesquioxanes aerogels (Loy et al. 1992) appeared in the literature in the early 1990s, while pendant

24

D.A. Loy

Fig. 19 The pore size distribution plot for octylenebridged polysilsesquioxanes xerogel prepared by sol-gel polymerization in tetrahydrofuran with sodium hydroxide as catalyst

polysilsesquioxanes aerogels, presumably due to the greater difficulties associated with making gels, appeared in 2006 (Rao et al. 2006). Surface areas of aerogels range from 200 to 1900 m2/g with arylene-bridged polysilsesquioxanes having the highest surface areas. Densities have been reported as low as 0.05 g/cm3. Pore size distributions are broad with meso- and macropores dominating. Silicasilsesquioxane and pendant polysilsesquioxanes aerogels are frequently more hydrophobic than most silica aerogels, whereas bridged polysilsesquioxanes are in between silica and pendant polysilsesquioxanes in hydrophilicity, due to their retention of more surface silanols.

Mechanical Properties Polymethylsilsesquioxane aerogels prepared by hydrolysis and condensation of methyltriethoxysilane and gelation by spinodal phase separation were shown by Kanamori to be elastic with reversible compressive strengths as high as 9 MPa (Kanamori 2011). Similarly, hexylene-bridged polysilsesquioxanes aerogels were discovered to be elastic (Loy et al. 1995), but were not further investigated due to their opacity, until recently rediscovered (Meador et al. 2009). Despite the flexibility of some examples, organosilica aerogels, like silica aerogels, are relatively weak materials due to their low density and delicate network architecture. Phenylene- and hexylene-bridged polysilsesquioxane aerogels’ flexural strengths (0.048  0.002 and 0.031  0.003 MPa, respectively) were 20% higher and comparable, respectively, to that of silica aerogels of the same density (0.092 g/mL) (Boday et al. 2012).

Surfactant and Polymer Templated Materials Porosity Surfactant templating in polysilsesquioxanes generates periodic mesoporosity once the templates have been removed. This mesoporosity gives rise to high surface areas and narrow pore size distributions, the size of which can be controlled from 2 to 60 nm diameters through selection of surfactant and swelling of the template

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

25

mesophase with solvents. Porosity of surfactant templated polysilsesquioxanes is often touted as one of the primary advantages of these mesoporous materials. Interestingly enough, a few papers provide comparisons to control polysilsesquioxanes made without templates. One report that does provide a direct comparison looked at hexylene-bridged polysilsesquioxanes prepared as xerogels and with surfactant templating under both acidic and basic conditions. The result was that comparable surface areas and pore sizes, including narrow pore size distributions, could be accomplished by both approaches. The surfactant templating did allow the formation of mesoporous material under acidic conditions where the sol-gel processed hexylene-bridged polysilsesquioxanes dries to form nonporous xerogels. Two more significant advantages of sol-gel processing are faster processing and larger amounts of materials formed. Additional significant advantages for surfactant templating include better retention of porosity with film formation and the potential for controlling the orientation of pores by using the appropriate template mesostructure.

Mechanical Properties Since few monolithic samples have been prepared with periodic mesoporous polysilsesquioxanes, the hardness of surfactant templated thin films will be used to examine the mechanical properties of this class of materials. Early reports of silicaethylene-bridged cocondensed films template with Brij-56 showed that sample hardness increased with increasing ethylene-bridged polysilsesquioxanes content from 0.35 GPa with 75:25 silica to ethylene bridge to 0.48 GPa for 25:75 silica: ethylene bridge (Lu et al. 2000). A mesoporous biphenylene-bridged polysilsesquioxanes film was shown to have a hardness of 0.23 GPa. Finally, a P-123 templated film of ethylene-bridged polysilsesquioxanes was shown to have a hardness of 0.8 GPa (Jiang et al. 2014). As a rule, the bridged polysilsesquioxanes films are harder than comparable silica films.

Applications of Mesoporous Polysilsesquioxanes Adsorbents Polysilsesquioxane adsorbents have been studied since the 1980s in Russia, where thiourea bridged polysilsesquioxanes prepared by sol-gel polymerizations were used to scavenge a variety of precious and toxic metals from aqueous solutions (Vlasova et al. 1989). With the advent of surfactant templating, a large number of groups have investigated using bridged polysilsesquioxanes as adsorbents for metal ions (Wu et al. 2010) and organic compounds from air (Borghard et al. 2009) or water. While many compare the efficacy of their surface-modified silica, silicasilsesquioxane copolymer or bridged polysilsesquioxanes adsorbents against other commercially available adsorbents, no one compares these materials to the materials made without templating. What are frequently characterized as extraordinary

26

D.A. Loy

adsorption capacities are often the same as could have been achieved with the sol-gel polymerized materials.

Anticorrosion Coatings Sol-gel processed coatings of polysilsesquioxanes have been successfully used as replacements for chromium-based anticorrosion coatings (van Ooij and Child 1998). These have included copolymers of pendant polysilsesquioxanes with silica and bridged polysilsesquioxanes (Gu et al. 2015). In all cases, porosity would be detrimental to the protection of the substrate, so surfactant templating is not useful. Sol-gel coatings of pendant-silica copolymers where the pendant monomer is present in greater than 25 mol% are often non-porous, but strong enough to make thin layers. Bridged polysilsesquioxanes with flexible bridging groups prepared under acidic conditions will form as nonporous coatings as well (Pan et al. 2007).

Catalysts Most catalysts based on mesoporous polysilsesquioxanes fall into one of three types: Bronsted acid catalysts (Shylesh et al. 2008), ligand-supported metal catalysts (Dufaud and Davis 2003), and organic catalysts (Borah et al. 2015). Some bridged polysilsesquioxanes bearing sulfonic acid groups have been prepared in porous xerogels, aerogels, and periodic mesoporous forms (Nakajima et al. 2005). The sulfonic acid groups have been built into the structure or added after its formation (Jones et al. 1998). Metal centers for catalyst can be supported by ligands on pendant silsesquioxanes from copolymers with silica (Fig. 20a) (Sharma and Singh 2014) or grafted from the surface (Fig. 20b) (Naeimi et al. 2015) ligands that are part of the bridging group itself (Fig. 20c) (Gruening et al. 2014), or as part of the bridge with the ligands (Fig. 20d) (Borah et al. 2012). A variety of chiral catalysts for hydrogenations has been developed using chiral bridging groups in surfactant template materials (Wahab and Beltramini 2015).

Chromatographic Materials Modification of chromatographic packing materials with silane coupling agents has long been used to modify the column adsorption properties. More recently, the packing materials have been prepared by emulsion copolymerization of organotrialkoxysilanes and tetraalkoxysilanes or from bridged polysilsesquioxane monomers (O’Gara and Wyndham 2006). These materials are required for the high pressures used in modern liquid chromatographic systems. Sol-gel polymerization of monomers in emulsions are typically the best method for controlling the size of the particles. Surfactant templating can also be used to impart ordered porosity to packing materials, but these materials have not been commercialized. Monolithic,

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

27

Fig. 20 Catalysts made from periodic mesoporous organosilicas

porous polysilsesquioxanes have also been prepared as chromatographic adsorbents by sol-gel, spinodal phase separation, and surfactant templating techniques (Guo et al. 2013). Progress has also been made in using molecular imprinting or imprinting by using removable bridging groups (Lofgreen et al. 2011) to modify chromatographic materials’ selectivities.

Low k Dielectric Films Due to decreasing scale of microelectronics, materials with low κ dielectrics are needed to prevent electronic cross-talk between circuits and to reduce power consumption. Silsesquioxanes have lower κ values than silica and have excellent mechanical properties in film form. Their first application started with pore templating in polysilsesquioxanes (Remenar et al. 1998) and their use has expanded with the development of polyhydridosilsesquioxane, methylsilsesquioxane, and methylene-bridged polysilsesquioxanes coatings (Hatton et al. 2006).

Fig. 21 Mesoporous, surfactant templated porphyrin bridged polysilsesquioxanes coating for photovoltaic applications

28 D.A. Loy

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

29

Fig. 22 Salicylate bridged silsesquioxane-silica particle prepared by sol-gel polymerization for use as a sunscreen

Membranes Most polysilsesquioxanes membranes have been developed for nonporous gas separation membranes or proton conducting, gas barriers for fuel cells. Nonporous polysilsesquioxane membranes were first evaluated by Stern for gas separations (Stern et al. 1987). Pendant and bridged monomers were used as pore templates in silica gas separation membranes that were calcined to remove the organic functionality after the membranes were cast (Raman et al. 1996). Microporous silicamethylsilsesquioxane membranes, prepared by sol-gel copolymerizations, are hydrophobic, yet highly permeable to oxygen and nitrogen (de Vos et al. 1999). More recently a number of microporous bridged polysilsesquioxane membranes were evaluated for gas and pervaporation separations (Kreiter et al. 2011).

Optical Coatings Polymethylsilsesquioxane coatings for optical fibers are among the earliest commercial applications of the class of materials (Dislich and Jacobsen 1973). The most durable coatings generally have organic groups that can be cured using chemistries orthogonal to the sol-gel chemistry used to prepare the silsesquioxane (Baney et al. 1995). An added benefit of bridged polysilsesquioxanes is that complicated and bulky chromophores can be incorporated into networks to prevent the organic from interfering with polymerization or from causing phase segregation. For example,

30

D.A. Loy

porphyrin-bridged monomers have been surfactant templated into photovoltaic thin films (Fig. 21) (Li et al. 2013).

Sunscreens Polysilsesquioxanes covalently modified with organic sunscreens provides another method of preparing polymers or particles that can absorb ultraviolet light without passing into tissues or through cell membranes (Tolbert et al. 2016). The monomers used to make these sunscreen polysilsesquioxanes can readily be prepared by modifying reactive coupling agents with commercial sunscreens with nucleophilic substituents (Fig. 22). For the sunscreen to pass into the body the sunscreen would have to be chemically cleaved from the polysilsesquioxane. Particles of polysilsesquioxane sunscreen were made by copolymerizing the organically modified trialkoxysilane with a tetraalkoxysilane or using bridged monomers where the sunscreen is the organic bridging group. The resulting sunscreen particles, when dispersed in glycerol or water retained their UV absorbing characteristics far longer than commercial organic sunscreens.

Concluding Remarks In this chapter, mesoporous polysilsesquioxanes, silica-polysilsesquioxanes, and bridged polysilsesquioxanes are described. In this group of materials, the presence and size of pores is dependent on the monomer(s) used, the polymerization chemistry, the macromolecular structure of the resulting polymers and how they interplay with the solvent, self-assembled surfactants or polymers as porogens and subsequent processing. A particular emphasis was placed on comparing the polysilsesquioxane materials prepared without self-assembled templates to those prepared with them. The beauty of the periodic structures formed by the latter is without question, but, in some cases, the same ultimate properties can be attained with disordered mesopores found in sol-gel processing without templates.

References Abe Y, Kagayama K, Takamura N, Gunji T, Yoshihara T, Takahashi N. Preparation and properties of polysilsesquioxanes. Function and characterization of coating agents and films. J Non-Cryst Solids. 2000;261(1–3):39–51. Arkhireeva A, Hay JN, Lane JM, Manzano M, Masters H, Oware W, Shaw SJ. Synthesis of organicinorganic hybrid particles by sol-gel chemistry. J Sol-Gel Sci Technol. 2004;31(1–3):31–6. Baney RH, Itoh M, Sakakibara A, Suzuki T. Silsesquioxanes. Chem Rev. 1995;95(5):1409–30. Barton TJ, Bull LM, Klemperer WG, Loy DA, McEnaney B, Misono M, Monson PA, Pez G, Scherer GW, Vartuli JC, Yaghi OM. Tailored porous materials. Chem Mater. 1999;11 (10):2633–56.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

31

Beck JS, Vartuli JC, Roth WJ, Leonowicz ME, Kresge CT, Schmitt KD, Chu CTW, Olson DH, Sheppard EW, et al. A new family of mesoporous molecular sieves prepared with liquid crystal templates. J Am Chem Soc. 1992;114(27):10834–43. Ben F, Boury B, Corriu RJP, Le Strat V. Evidence for an organization of nanostructured silica-based hybrid materials prepared by sol-gel polymerization. Chem Mater. 2000;12(11):3249–52. Boday DJ, Stover RJ, Muriithi B, Loy DA. Mechanical properties of hexylene- and phenylenebridged polysilsesquioxane aerogels and xerogels. J Sol-Gel Sci Technol. 2012;61(1):144–50. Borah P, Ma X, Nguyen KT, Zhao Y. A vanadyl complex grafted to periodic mesoporous organosilica: a green catalyst for selective hydroxylation of benzene to phenol. Angew Chem Int Ed. 2012;51(31):7756–61 .S7756/7751-S7756/7757 Borah P, Mondal J, Zhao Y. Urea-pyridine bridged periodic mesoporous organosilica: an efficient hydrogen-bond donating heterogeneous organocatalyst for Henry reaction. J Catal. 2015;330:129–34. Borghard WG, Calabro DC, DiSanzo FP, Disko MM, Diehl JW, Fried JC, Markowitz MA, Zeinali M, Melde BJ, Riley AE. Characterization and testing of periodic mesoporous organosilicas as potential selective benzene adsorbents. Langmuir. 2009;25(21):12661–9. Brandhuber D, Peterlik H, Huesing N. Facile self-assembly processes to phenylene-bridged silica monoliths with four levels of hierarchy. Small. 2006;2(4):503–6. Brown J, Richer R, Mercier L. One-step synthesis of high capacity mesoporous Hg2+ adsorbents by non-ionic surfactant assembly. Microporous Mesoporous Mater. 2000;37(1–2):41–8. Castricum HL, Paradis GG, Mittelmeijer-Hazeleger MC, Bras W, Eeckhaut G, Vente JF, Rothenberg G, ten Elshof JE. Tuning the nanopore structure and separation behavior of hybrid organosilica membranes. Microporous Mesoporous Mater. 2014;185:224–34. Che S, Garcia-Bennett AE, Yokoi T, Sakamoto K, Kunieda H, Terasaki O, Tatsumi T. A novel anionic surfactant templating route for synthesizing mesoporous silica with unique structure. Nat Mater. 2003;2(12):801–5. Cho E-B, Char K. Macromolecular templating approach for the synthesis of hydrothermally stable mesoporous organosilicas with high periodicity and thick framework walls. Chem Mater. 2004;16(2):270–5. Cho E-B, Kim D, Gorka J, Jaroniec M. Three-dimensional cubic (Im3m) periodic mesoporous organosilicas with benzene- and thiophene-bridging groups. J Mater Chem. 2009;19 (14):2076–81. Cho E-B, Kim D, Mandal M, Gunathilake CA, Jaroniec M. Benzene-silica with hexagonal and cubic ordered mesostructures synthesized in the presence of block copolymers and weak acid catalysts. J Phys Chem C. 2012;116(30):16023–9. Cordes DB, Lickiss PD, Rataboul F. Recent developments in the chemistry of cubic polyhedral oligosilsesquioxanes. Chem Rev. 2010;110(4):2081–173. de Theije FK, Balkenende AR, Verheijen MA, Baklanov MR, Mogilnikov KP, Furukawa Y. Structural characterization of mesoporous organosilica films for ultralow-k dielectrics. J Phys Chem B. 2003;107(18):4280–9. de Vos RM, Maier WF, Verweij H. Hydrophobic silica membranes for gas separation. J Membr Sci. 1999;158(1–2):277–88. Dislich H, Jacobsen A. Fiber optics for the ultraviolet spectral region. Angew Chem. 1973;85 (11):468–73. Dufaud V, Davis ME. Design of heterogeneous catalysts via multiple active site positioning in organic-inorganic hybrid materials. J Am Chem Soc. 2003;125(31):9403–13. Fowler CE, Burkett SL, Mann S. Synthesis and characterization of ordered organosilica-surfactant mesophases with functionalized MCM-41-type architecture. Chem Commun (Camb). 1997;2 (18):1769–70. Goto Y, Inagaki S. Synthesis of large-pore phenylene-bridged mesoporous organosilica using triblock copolymer surfactant. Chem Commun (Cambridge, U. K.). 2002;2(20):2410–1. Grudzien RM, Pikus S, Jaroniec M. Periodic mesoporous organosilicas with Im3m symmetry and large isocyanurate bridging groups. J Phys Chem B. 2006;110(7):2972–5.

32

D.A. Loy

Gruening WR, Siddiqi G, Safonova OV, Coperet C. Bipyridine periodic mesoporous organosilica: a solid ligand for the iridium-catalyzed borylation of C–H bonds. Adv Synth Catal. 2014;356 (4):673–9. Gu L, Liu S, Zhao H, Yu H. Anticorrosive oligoaniline-containing electroactive siliceous hybrid materials. RSC Adv. 2015;5(69):56011–9. Guo XZ, Yu H, Yang H, Kanamori K, Zhu Y, Nakanishi K. Pore structure control of macroporous methylsilsesquioxane monoliths prepared by in situ two-step processing. J Porous Mater. 2013;20(6):1477–83. Guo XZ, Wang R, Yu H, Zhu Y, Nakanishi K, Kanamori K, Yang H. Spontaneous preparation of hierarchically porous silica monoliths with uniform spherical mesopores confined in a welldefined macroporous framework. Dalton Trans. 2015;44(30):13592–601. Hao N, Yang Y, Wang H, Webley PA, Zhao D. Synthesis of large-pore phenyl-bridged mesoporous organosilica with thick walls by evaporation-induced self-assembly for efficient benzene adsorption. J Colloid Interface Sci. 2010;346(2):429–35. Hao TT, Shi JY, Zhuang TY, Wang WD, Li FC, Wang W. Mesostructure-controlled synthesis of chiral norbornane-bridged periodic mesoporous organosilicas. RSC Adv. 2012;2(5):2010–4. Hatton BD, Ozin GA, Perovic DD, Landskron KMM. Transformation of bridging organic groups in organosilica low dielec. const. (k) materials in microelectronics, Can. 2006. 79 p. Herzog G, Sibottier E, Etienne M, Walcarius A. Electrochemically assisted self-assembly of ordered and functionalized mesoporous silica films: impact of the electrode geometry and size on film formation and properties. Faraday Discuss. 2013;164:259–73. Hu L-C, Shea KJ. Organo-silica hybrid functional nanomaterials: how do organic bridging groups and silsesquioxane moieties work hand-in-hand? Chem Soc Rev. 2011;40(2):688–95. Inagaki S, Guan S, Ohsuna T, Terasaki O. An ordered mesoporous organosilica hybrid material with a crystal-like wall structure. Nature (London). 2002;416(6878):304–7. Innocenzi P, Malfatti L, Kldchob T, Falcaro P. Order-disorder in self-assembled mesostructured silica films: a concepts review. Chem Mater. 2009;21(13):2555–64. Jiang T, Zhu B, Ding S-J, Fan Z, Zhang DW. High-performance ultralow dielectric constant carbonbridged mesoporous organosilica films for advanced interconnects. J Mater Chem C. 2014;2 (32):6502–10. Jones CW, Tsuji K, Davis ME. Organic-functionalized molecular sieves as shape-selective catalysts. Nature (London). 1998;393(6680):52–4. Jun Y, Leatherdale CA, Norris DJ. Tailoring air defects in self-assembled photonic bandgap crystals. Adv Mater (Weinheim, Ger). 2005;17(15):1908–11. Kanamori K. Organic inorganic hybrid aerogels with high mechanical properties via organotrialkoxysilane-derived sol-gel process. J Ceram Soc Jpn. 2011;119(1385):16–22. Kanamori K, Nakanishi K. Controlled pore formation in organotrialkoxysilane-derived hybrids: from aerogels to hierarchically porous monoliths. Chem Soc Rev. 2011;40(2):754–70. Kanamori K, Nakanishi K, Hanada T. Spinodal decomposition in siloxane sol-gel systems in macroporous media. Soft Matter. 2009;5(16):3106–13. Kanamori K, Kodera Y, Hayase G, Nakanishi K, Hanada T. Transition from transparent aerogels to hierarchically porous monoliths in polymethylsilsesquioxane sol-gel system. J Colloid Interface Sci. 2011;357(2):336–44. Kapoor MP, Setoyama N, Yang Q, Ohashi M, Inagaki S. Oligomeric polymer surfactant driven selfassembly of phenylene-bridged mesoporous materials and their physicochemical properties. Langmuir. 2005;21(1):443–9. Khiterer M, Shea KJ. Spherical, monodisperse, functional bridged polysilsesquioxane nanoparticles. Nano Lett. 2007;7(9):2684–7. Kreiter R, Rietkerk MDA, Castricum HL, van Veen HM, ten Elshof JE, Vente JF. Evaluation of hybrid silica sols for stable microporous membranes using high-throughput screening. J Sol-Gel Sci Technol. 2011;57(3):245–52.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

33

Li Y, Auras F, Loebermann F, Doeblinger M, Schuster J, Peter L, Trauner D, Bein T. A photoactive porphyrin-based periodic mesoporous organosilica thin film. J Am Chem Soc. 2013a;135 (49):18513–9. Li Z, Tolbert S, Loy DA. Hybrid organic-inorganic membranes on porous supports by size exclusion and thermal sintering of fluorescent polyphenylsilsesquioxane nanoparticles. Macromol Mater Eng. 2013b;298(7):715–21. Liang Y, Hanzlik M, Anwander R. Ethylene-bridged periodic mesoporous organosilicas with Fm3m symmetry. J Mater Chem. 2005;15(35–36):3919–28. Lim MH, Stein A. Comparative studies of grafting and direct syntheses of inorganic-organic hybrid mesoporous materials. Chem Mater. 1999;11(11):3285–95. Lim MH, Blanford CF, Stein A. Synthesis and characterization of a reactive vinyl-functionalized MCM-41: probing the internal pore structure by a bromination reaction. J Am Chem Soc. 1997;119(17):4090–1. Lin D, Hu L, You H, Tolbert SH, Loy DA. Comparison of new periodic, mesoporous, hexylenebridged polysilsesquioxanes with Pm3n symmetry versus sol-gel polymerized, hexylenebridged gels. J Non-Cryst Solids. 2014;406:139–43. Lin DR, Hu LJ, Tolbert SH, Li Z, Loy DA. Controlling nanostructure in periodic mesoporous hexylene-bridged polysilsesquioxanes. J Non-Cryst Solids. 2015;419:6–11. Lofgreen JE, Moudrakovski IL, Ozin GA. Molecularly imprinted mesoporous organosilica. ACS Nano. 2011;5(3):2277–87. Loy DA, Shea KJ. Bridged polysilsesquioxanes. Highly porous hybrid organic-inorganic materials. Chem Rev. 1995;95(5):1431–42. Loy DA, Shea KJ, Russick EM. Preparation of aryl-bridged polysilsesquioxane aerogels. Mater Res Soc Symp Proc. 1992;271(Better Ceramics through Chemistry V):699–704. Loy DA, Jamison GM, Baugher BM, Russick EM, Assink RA, Prabakar S, Shea KJ. Alkylenebridged polysilsesquioxane aerogels: highly porous hybrid organic-inorganic materials. J Non-Cryst Solids. 1995;186:44–53. Loy DA, Beach JV, Baugher BM, Assink RA, Shea KJ, Tran J, Small JH. Dialkylenecarbonatebridged polysilsesquioxanes. Hybrid organic-inorganic sol-gels with a thermally labile bridging group. Mater Res Soc Symp Proc. 1999;576(Organic/Inorganic Hybrid Materials II):99–104. Loy DA, Baugher BM, Baugher CR, Schneider DA, Rahimian K. Substituent effects on the sol-gel chemistry of organotrialkoxysilanes. Chem Mater. 2000;12(12):3624–32. Lu Y, Fan H, Doke N, Loy DA, Assink RA, LaVan DA, Brinker CJ. Evaporation-induced selfassembly of hybrid bridged silsesquioxane film and particulate mesophases with integral organic functionality. J Am Chem Soc. 2000;122(22):5258–61. Manchanda AS, Kruk M. Synthesis of large-pore face-centered-cubic periodic mesoporous organosilicas with unsaturated bridging groups. Microporous Mesoporous Mater. 2016;222:153–9. Mandal M, Kruk M. Large-pore ethylene-bridged periodic mesoporous organosilicas with facecentered cubic structure. J Phys Chem C. 2010;114(47):20091–9. Meador MAB, Weber AS, Hindi A, Naumenko M, McCorkle L, Quade D, Vivod SL, Gould GL, White S, Deshpande K. Structure-property relationships in porous 3D nanostructures: epoxycross-linked silica aerogels produced using ethanol as the solvent. ACS Appl Mater Interfaces. 2009;1(4):894–906. Mizoshita N, Tani T, Inagaki S. Synthesis, properties and applications of periodic mesoporous organosilicates prepared from bridged organosilane precursors. Chem Soc Rev. 2011;40 (2):789–800. Moller K, Bein T, Fischer RX. Synthesis of ordered mesoporous methacrylate hybrid systems: hosts for molecular polymer composites. Chem Mater. 1999;11(3):665–73. Mushiake A, Kanamori K, Nakanishi K. Evolution of mesopores in monolithic macroporous ethylene-bridged polysilsesquioxane gels incorporated with nonionic surfactant. Int J Polym Sci. 2012;2012:1.

34

D.A. Loy

Naeimi H, Nejadshafiee V, Masoum S. Copper@PMO nanocomposites as a novel reusable heterogeneous catalyst for microwave-assisted green synthesis of β-hydroxy-1,2,3-triazoles through experimental design protocol. Appl Organomet Chem. 2015;29(5):314–21. Nakajima K, Tomita I, Hara M, Hayashi S, Domen K, Kondo JN. A stable and highly active hybrid mesoporous solid acid catalyst. Adv Mater (Weinheim, Ger.). 2005;17(15):1839–42. Nakanishi K, Kanamori K. Organic-inorganic hybrid poly(silsesquioxane) monolith with controlled macro- and mesopores. J Mater Chem. 2005;15(35–36):3776–86. Nie C-F, Zhao R, Suo J-S. Synthesis and characterization of bifunctional periodic silica with surface and framework benzene functionality. J Porous Mater. 2004;11(3):141–6. O’Gara J, Wyndham K. Porous hybrid organic-inorganic particles in reversed-phase liquid chromatography. J Liq Chromatogr Relat Technol. 2006;29(7&8):1025–45. Pan G, Watkins E, Majewski J, Schaefer DW. Effect of thickness on the water-barrier properties of silane films. J Phys Chem C. 2007;111(42):15325–30. Park SS, An B, Ha C-S. High-quality, oriented and mesostructured organosilica monolith as a potential UV sensor. Microporous Mesoporous Mater. 2008;111(1–3):367–78. Raman NK, Anderson MT, Brinker CJ. Template-based approaches to the preparation of amorphous, nanoporous silicas. Chem Mater. 1996;8(8):1682–701. Rao AV, Bhagat SD, Hirashima H, Pajonk GM. Synthesis of flexible silica aerogels using methyltrimethoxysilane (MTMS) precursor. J Colloid Interface Sci. 2006;300(1):279–85. Rebbin V, Rothkirch A, Vainio U, Funari SS. Non-ionic block copolymers leading to a highly ordered organosilica material. J Phys Conf Ser. 2012;351:012015/012011-012015/012017. Remenar JF, Hawker CJ, Hedrick JL, Kim SM, Miller RD, Nguyen C, Trollsas M, Yoon DY. Templating nanopores into poly(methylsilsesquioxane): new low-dielectric coatings suitable for microelectronic applications. Mater Res Soc Symp Proc. 1998;511(Low-Dielectric Constant Materials III):69–74. Ren T, Zhang X, Suo J. Synthesis of periodic mesoporous organosilicas via the “S+X-I+” route. Microporous Mesoporous Mater. 2002;54(1–2):139–44. Sanchez A, Loy DA. Sol-gel chemistry of aminopropyltrialkoxy silanes. Polym Prepr (Am Chem Soc Div Polym Chem). 2001;42(1):182–3. Sayari A, Hamoudi S, Yang Y, Moudrakovski IL, Ripmeester JR. New insights into the synthesis, morphology, and growth of periodic mesoporous organosilicas. Chem Mater. 2000;12 (12):3857–63. Schwertfeger F, Huesing N, Schubert U. Influence of the nature of organic groups on the properties of organically modified silica aerogels. J Sol-Gel Sci Technol. 1994;2(1/2/3):103–8. Sharma P, Singh AP. Synthesis of a recyclable and efficient Pd(II) 4-(2-pyridyl)-1,2,3-triazole complex over a solid periodic mesoporous organosilica support by “click reactions” for the Stille coupling reaction. RSC Adv. 2014;4(81):43070–9. Shea KJ, Loy DA. Bridged polysilsesquioxanes. Molecular-engineered hybrid organic-inorganic materials. Chem Mater. 2001;13(10):3306–19. Shea KJ, Loy DA, Webster O. Arylsilsesquioxane gels and related materials. New hybrids of organic and inorganic networks. J Am Chem Soc. 1992;114(17):6700–10. Shylesh S, Samuel PP, Sisodiya S, Singh AP. Periodic mesoporous silicas and organosilicas: an overview towards catalysis. Catal Surv Jpn. 2008;12(4):266–82. Stein A, Melde BJ, Schroden RC. Hybrid inorganic-organic meso-porous silicates-nanoscopic reactors coming of age. Adv Mater (Weinheim, Ger). 2000;12(19):1403–19. Stern SA, Shah VM, Hardy BJ. Structure-permeability relationships in silicone polymers. J Polym Sci B Polym Phys. 1987;25(6):1263–98. Tan B, Rankin SE. Effects of progressive changes in organoalkoxysilane structure on the gelation and pore structure of templated and non-templated sol-gel materials. J Non-Cryst Solids. 2006;352(52–54):5453–62. Tolbert SH, McFadden PD, Loy DA. New hybrid organic/inorganic polysilsesquioxane-silica particles as sunscreens. ACS Appl Mater Interfaces. 2016;8(5):3160–74.

Mesoporous Polysilsesquioxanes: Preparation, Properties, and Applications

35

Van Der Voort P, Esquivel D, De Canck E, Goethals F, Van Driessche I, Romero-Salguero FJ. Periodic mesoporous organosilicas: from simple to complex bridges; a comprehensive overview of functions, morphologies and applications. Chem Soc Rev. 2013;42(9):3913–55. van Ooij WJ, Child T. Protecting metals with silane coupling agents. Chemtech. 1998;28(2):26–35. Vlasova NV, Pestunovich AE, Pozhidaev YN, Zemyanushnova OV, Kirillov AI, Voronkov MG. N, N0 -bis(triethoxysilylpropyl)thiourea-modified silica gels and their sorption activity. Izv Akad Nauk Belarusi Ser Khim Nauk. 1989;5:74–7. Wahab MA, Beltramini JN. Recent advances in hybrid periodic mesostructured organosilica materials: opportunities from fundamental to biomedical applications. RSC Adv. 2015;5 (96):79129–51. Wahab MA, He C. Self-assembly, optical, and mechanical properties of surfactant-directed biphenyl-bridged periodic mesostructured organosilica films with molecular-scale periodicity in the pore walls. Langmuir. 2009;25(2):832–8. Wang YQ, Yang CM, Zibrowius B, Spliethoff B, Linden M, Schueth F. Directing the formation of vinyl-functionalized silica to the hexagonal SBA-15 or large-pore Ia3d structure. Chem Mater. 2003;15(26):5029–35. Wu H-Y, Chen C-T, Hung IM, Liao C-H, Vetrivel S, Kao H-M. Direct synthesis of cubic benzenebridged mesoporous organosilica functionalized with mercaptopropyl groups as an effective adsorbent for mercury and silver ions. J Phys Chem C. 2010;114(15):7021–9. Xia LY, Hu YC, Rong MZ, Liang YJ. Preparation of bifunctionalized phenylene-bridged periodic mesoporous organosilica for solid-phase microextraction. RSC Adv. 2014;4(1):168–74. Xie Z, Henderson EJ, Dag O, Wang W, Lofgreen JE, Kubel C, Scherer T, Brodersen PM, Gu Z-Z, Ozin GA. Periodic mesoporous hydridosilica – synthesis of an “impossible” material and its thermal transformation into brightly photoluminescent periodic mesoporous nanocrystal siliconsilica composite. J Am Chem Soc. 2011;133(13):5094–102. Zhang J, Zhang G, Gao Y, Sun R, Wong CP. Ultra-low-κ HFPDB-based periodic mesoporous organosilica film with high mechanical strength for interlayer dielectric. J Mater Sci. 2016;51 (17):7966–76.

Anti-soiling Effect of Porous SiO2 Coatings Peer Löbmann

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 General Approaches for Anti-soiling Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Hydrophobic Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Porous Corrugated Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Superhydrophilic Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Photocatalytic Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Validation of Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Porous Anti-soiling SiO2 Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Abstract

Various surface topographies and chemical compositions may result in antisoiling properties. In this chapter, different strategies are compiled and respective solutions based on sol-gel processing are reviewed. Methods for the validation of material performance are discussed with focus on particulate contaminations. Then porous SiO2 films combining antireflective (AR) and anti-soiling are presented.

Introduction Contamination by liquids and particles is a widespread problem for architectural and technical surfaces. Graffiti on building facades and trains basically are only annoying to the aesthetic appeal, but fingerprints, for example, may also go along with P. Löbmann (*) Optik und Elektronik, Fraunhofer-Institut f€ ur Silicatforschung ISC, W€ urzburg, Germany e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_132-1

1

2

P. Löbmann

bacterial contagion. As soon as technical surfaces are concerned, pollution may seriously obstruct their functionality: The output of lighting elements as well as the efficiency of solar cells is obstructed by soiling; dirt on windshields and ophthalmic devices impairs vision and therefore may lead to accidents. At the best case, adhesion of contaminations is generally suppressed, but also their easy removal under environmental conditions such as wind, rainfall, or solar radiation is an option. Alternatively, the easier elimination of dirt under standard cleaning procedures may be facilitated. It is clear that anti-soiling properties result from the complex interaction between surfaces, pollutants, and the respective environmental conditions. To make things even more complicated, different types of contaminations deposited may interact and create properties being different from the initial virgin surfaces. Hence there may be severe disparities between testing under controlled/simplified laboratory conditions and real-life outdoor exposure. Nevertheless different strategies for anti-soiling surfaces may be distinguished according to their surface functionality and exterior topography, an overview is given in Fig. 1. First of all surfaces may be differentiated according to their structures. Dense and even materials are the simplest case, whereas they may also be corrugated. Porous surfaces may be considered having a roughness on a very fine scale and thus should be distinguished from waviness on a larger range. Despite from these structural features, the respective base materials can be classified according to their interaction with polar liquids. Hydrophobicity denotes

Fig. 1 Overview on anti-soiling properties related to different surface functionalities and surface topography

Anti-soiling Effect of Porous SiO2 Coatings

3

a weak interaction with water, and the related contact angles are above 90 . Therefore, aqueous contaminations are more likely to flow off as droplets from hydrophobic than off hydrophilic surfaces. If hydrophobic surfaces are corrugated, their wetting behavior further decreases. The self-cleaning properties of so-called Lotus surfaces results from their superhydrophobicity and a reduced interaction with solid contaminations (Barthlott and Neinhuis 1997). Hydrophilic surfaces show contact angles with water smaller than 90 . Titania exposed to UV light may exhibit superhydrophilicity with contact angles close to 0 . On such surfaces, any pollutants may be undercut by water and thus easily be washed away. On TiO2, this effect goes along with photocatalysis, the oxidative decomposition of organic materials: In combination, the adhesion of organic contaminations to photocatalysts is weakened and their removal by superhydrophilicity is facilitated. In that case, solid or liquid pollutants are removed from dense titania surfaces, and gaseous organics may be decomposed by porous photocatalysts. The above considerations are generally true for surfaces irrespective of their individual origin: Dense photocatalytic TiO2 films may be prepared by chemical vapor deposition (CVD) or physical vapor deposition (PVD) techniques such as sputtering. The Lotus effect primarily observed in nature may be mimicked by the embossing of hydrophobic substrates. In the last decades, however, sol-gel processing (Brinker and Scherer 1990; Schneller et al. 2013) has evolved into a mature technology for the preparation of inorganic and hybrid thin films. Various film compositions and microstructures as depicted in Fig. 1 may be realized by the wet chemical deposition method. This chemical and microstructural flexibility is a strong advantage of sol-gel processing. In the next paragraph, different strategies leading to anti-soiling surfaces are shortly reviewed, and special focus is put on the respective realization by sol-gel processing. Subsequently some methods for the validation of anti-soiling properties are highlighted. Finally sol-gel-derived porous SiO2 coatings are introduced that combine both dust-repellant and antireflective properties.

General Approaches for Anti-soiling Surfaces As already stated above, real-life contaminations originate from the complex interaction of solid and liquid pollutions with surfaces. As the final focus of this chapter is porous SiO2 coatings with anti-soiling properties for solar applications in arid regions, the interaction with particulate pollutions will be emphasized.

Hydrophobic Surfaces The contact of surfaces with liquids basically is ruled by the interfacial energy between the phases. Surface energies of liquids and solids have both a polar and dispersive proportion (Yan et al. 2011). In short, water is repelled from low-energy surfaces such as organic polymers, and wetting angles are high. Regarding

4

P. Löbmann

particulate contaminations, droplets that contain dust particles are more likely to be rinsed off hydrophobic than off standard hydrophilic materials before drying. In general, surfaces are considered as being hydrophobic if the contact angle with water is higher than 90 . Besides the wetting behavior, the roll-off angle of the water droplets plays an important role (Marmur 2004).

Smooth Hydrophobic Surfaces Hydrophobicity can be provided either by the substrate material itself or by specific surface functionalization. Glasses and ceramics, for example, may react with alkoxysilanes, incorporating nonpolar organic or fluorinated sidechains (Dey and Naughton 2016). Also hydrophobic hybrid polymer films are feasible (Schottner 2001). In this context, it is important that droplets containing solid dust are preferably removed by rolling off before the liquid is evaporated and the dry surface is contaminated by particles. Porous Hydrophobic Surfaces On transparent substrates, antireflective properties can be facilitated by porous λ/4 layers (Löbmann 2013). Therefore porous hydrophobic surfaces are of general interest for architectural glazing and solar applications. As already shown in Fig. 1, it has to be noted that there is a smooth transition between roughness on a nanoscopic scale induced by porosity and longer-ranged waviness. Porous hydrophobic SiO2 films have been prepared by sol-gel processing and subsequent hydrophobization (Gurav et al. 2011; Dey and Naughton 2016). Alternatively functionalized silica nanoparticles with radii of 20 nm and 100 nm were used for the preparation of transparent films (Karunakaran et al. 2011). When the smaller nanoparticles were deposited after a first coating with the larger ones, water contact angles exceeded 150 and roll-off angles below 10 were realized. In this case, superhydrophobicity originates from structures far smaller than those originally observed on Lotus leafs (Barthlott and Neinhuis 1997). Such topographies are discussed in the next paragraph. Corrugated Hydrophobic Surfaces In 1997, Barthlott and Neinhuis discovered that contaminations from certain plant leaves are very effectively removed by rainfall. This observation was attributed to the cooperative effect of hydrophobic surface corrugations both in the μm and nm scale. Since this report, considerable efforts were being made to utilize artificial Lotus surfaces (Jiang et al. 2015). Since μm-sized protrusions will scatter light, transparent application based on “classic” Lotus surfaces cannot be realized. Corrugated surface may be prepared by roughening hydrophobic materials or by hydrophobization of rough substrates. Sol-gel films have been imprinted both by natural (Saison et al. 2008) and artificial (Back et al. 2014) structures. It has been debated, though, that microstructures on two orders of magnitude are no prerequisite for self-cleaning properties (Ma and Hill 2006). Transparent superhydrophobic coatings and their preparation recently have been reviewed (Yu et al. 2015).

Anti-soiling Effect of Porous SiO2 Coatings

5

Porous Corrugated Surfaces The anti-soiling properties of hydrophobic (section “Hydrophobic Surfaces”) and superhydrophilic (section “Superhydrophilic Surfaces”) surfaces depend on the presence of water due to rainfall or cleaning procedures. Rainfall is a scarce event in arid regions, and technical cleaning procedures, e.g., solar energy plants, are costly. Therefore, a sol-gel-based coating procedure was developed by combining both anti-soiling and antireflective properties. In paragraph 4 preparation, microstructure and performance of these systems will be described in detail.

Superhydrophilic Surfaces Since the first report of light-induced amphiphilic TiO2 surfaces (Wang et al. 1997), superwettability has gained substantial academic and industrial attention (Su et al. 2016). Superhydrophilic surfaces with contact angles 92 %

Table 18 Examples for preparation of water-repellent coatings using flowerlike alumina in combination with water-repellent agents

34 M. Nofz

Alumina Thin Films

35

Summary and Conclusions Alumina thin films enable a fascinating diversity of possible applications. This is based on (i) the large variety of formulations developed for coating sols, (ii) the variability of posttreatments of initially deposited films which allow preparing aluminas with specific physical and chemical properties, and (iii) and the applicability to metallic and ceramic materials as well as polymers having different shapes. For the preparation of alumina thin films, more than half of the analyzed papers are based on the use of boehmite or pseudoboehmite sols. Their overwhelming majority was produced by the Yoldas process, whereas commercially available boehmite/pseudoboehmite powders or sols are seldom used. Spin and dip coating are the most often used deposition methods. To fully exploit the potential of sol-gel alumina coatings, the application of spray deposition would be favorable. Spray coating using an air-pressure-driven nozzle was mentioned for deposition of alumina-sol-based slurries. Development of alumina sol formulations usable for aerosol deposition seems to be necessary. Research on membrane applications and barrier coatings was continued. Some new ideas opening further fields of possible applications were developed: (i) laser processing of coatings on materials which do not survive the high processing temperatures usually required to convert sol-gel materials; (ii) producing a flowerlike structure, i.e., a “tailored” surface roughness enabling realization of superhydrophobic coatings; and (iii) modification of textile fibers and fabrics.

References Adraider Y, Pang YX, Nabhani F, Hodgson SN, Zhang ZY. Laser-induced deposition of sol-gel alumina coating on stainless steel under wet condition. Surf Coat Technol. 2011;205:5345–9. Adraider Y, Hodgson SNB, Sharp MC, Zhang ZY, Nabhani F, Al-Waidh A, Pang YX. Structure characterization and mechanical properties of crystalline alumina coatings on stainless steel fabricated via sol-gel technology and fibre laser processing. J Eur Ceram Soc. 2012;32:4229–40. Adraider Y, Pang YX, Nabhani F, Hodgson SN, Sharp MC, Al-Waidh A. Laser-induced deposition of alumina ceramic coating on stainless steel from dry thin films for surface modification. Ceram Int. 2014;40:6151–6. Agrafiotis C, Tsetsekou A. Deposition of meso-porous γ-alumina coatings on ceramic honeycombs by sol-gel methods. J Eur Ceram Soc. 2002;22:423–34. Ahmad AL, Leo CP, Shukor SRA. Preparation of γ-alumina thin layer with bimodal pore size distribution for diminution of transport resistance in bi-layered membrane. Thin Solid Films. 2008;516:4319–24. Ahmad NA, Leo CP, Ahmad AL. Synthesis of superhydrophobic alumina membrane: effects of sol-gel coating, steam impingement and water treatment. Appl Surf Sci. 2013;284:556–64. Al-Atia MH. Developing the alumina sol-gel coating for steel oxidation protection at high temperatures. Eng Tech J. 2014;32:744–58. Al-Nakoua MA, El-Naas MH, Abu-Jdayil B. Characterization and testing of sol-gel catalysts prepared as thin layers in a plate reactor. Fuel Process Technol. 2011;92:1836–41. Alongi J, Ciobanu M, Malucelli G. Thermal stability, flame retardancy and mechanical properties of cotton fabrics treated with inorganic coatings synthesized through sol-gel process. Carbohydr Polym. 2012;87:2093–9.

36

M. Nofz

Alongi J, Carosio F, Malucelli G. Current emerging techniques to impart flame retardancy to fabrics: an overview. Polym Degrad Stab. 2014;106:138–49. Assih T, Ayral A, Abenoza M, Phalippou J. Raman study of alumina gels. J Mater Sci. 1988;23:3326–31. ASTM D3359, Standard test methods for measuring adhesion by tape test. ASTM International, West Conshohocken, Pennsylvania; 2009. ASTM D3363, Standard test method for film hardness by pencil test. ASTM International, West Conshohocken, Pennsylvania; 2011. ASTM D4145, Standard test method for coating flexibility of prepainted sheet. ASTM International, West Conshohocken, Pennsylvania; 2010. ASTM D4541, Standard test method for pull-off strength of coatings using portable adhesion testers. ASTM International, West Conshohocken, Pennsylvania; 2009. ASTM D6677, Standard test method for evaluating adhesion by knife. ASTM International, West Conshohocken, Pennsylvania; 2007. Atakan B, Eckert C, Pflitsch C. Light emitting diode excitation of Cr3+:Al2O3 as thermographic phosphor: experiments and measurement strategy. Meas Sci Technol. 2009;20:075304 (9 pp). Avci N, Smet PF, Lauwaert J, Vrielinck H, Poelman D. Optical and structural properties of aluminium oxide thin films prepared by a non-aqueous sol-gel technique. J Sol-Gel Sci Technol. 2011a;59:327–33. Avci N, Cimieri I, Smet PF, Poelman D. Stability improvement of moisture sensitive CaS:Eu2+ micro-particles by coating with sol-gel alumina. Opt Mater. 2011b;33:1032–5. Bahlawane N. Novel sol-gel process depositing α-Al2O3 for the improvement of graphite oxidation resistance. Thin Solid Films. 2001;396:126–30. Barbé CJ, Cassidy DJ, Triani G, Latella BA, Mitchell DRG, Bartlett JR. Low temperature bonding of ceramics by sol-gel processing. J Sol-Gel Sci Technol. 2003;26:1145–50. Belleville PF, Floch HG. A new room-temperature deposition technique for optical coatings. J Sol-Gel Sci Technol. 1994;3:23–9. Benkreira H. The effect of substrate roughness on air entrainment in dip coating. Chem Eng Sci. 2004;59:2745–51. Bi S, Wang C, Cao Q, Zhang C. Studies on the mechanism of hydrolysis and polymerization of aluminium salts in aqueous solution: correlations between the “Core-links” model and “Cagelike” Keggin Al13 model. Coord Chem Rev. 2004;248:441–55. Birnie III DP. Surface skin development and rupture during sol-gel spin-coating. J Sol-Gel Sci Technol. 2004;31:225–8. Biswas RG, Woodhead JL, Bhattacharaya AK. Corrosion studies of inorganic sol-gel coatings on 316 stainless steel. J Mater Sci Lett. 1997;16:1628–33. Brinker CJ, Hurd AJ. Fundamentals of sol-gel dip coating. J Phys III France. 1994;4:1231–42. Brinker CJ, Frye GC, Hurd AJ, Ashley CS. Fundamentals of sol-gel dip coating. Thin Solid Films. 1991;201:97–108. Caiut JMA, Nakagaki S, De Lima OJ, Mello C, Leite CAP, Nassar EJ, Ciuffi KJ, Sacco HC. Encapsulation of tetraazaannulenato compounds in matrix by sol-gel process. J Sol-Gel Sci Technol. 2003;28:57–64. Caldarelli A, Raimondo M, Veronesi F, Boveri G, Guarini G. Sol-gel route for the building up of superhydrophobic nanostructured hybrid-coatings on copper surfaces. Surf Coat Technol. 2015;276:408–15. Cebollada PAR, Garcia-Bordejé E. Optimisation of physical properties of γ-alumina coating microreactors used for the growth of a carbon nanofiber layer. Chem Eng J. 2009;149:447–54. Chen Y, Jin L, Xie Y. Sol-gel processing of organic-inorganic nanocomposite protective coatings. J Sol-Gel Sci Technol. 1998;13:735–8. Chen Y, Ai X, Huang C. Preparation of alpha alumina coating on carbide tools. Mater Sci Eng. 2000a;B77:221–8. Chen Y-C, Ai X, Huang C-Z, Wang B-Y. Preparation of α-alumina coated carbide tools by the sol-gel process. Mater Sci Eng A. 2000b;288:19–25.

Alumina Thin Films

37

Chen X, Hu Y, Wilson GS. Glucose microbiosensor based on alumina sol-gel matrix/electropolymerized composite membrane. Biosens Bioelectron. 2002;17:1005–13. Chui-Sabourin M, Garofalini SH. Effect of substrate on crystallization of sol-gel coating. Thin Solid Films. 1991;198:177–90. Cooper CA, Lin YS. Microstructural and gas separation properties of CVD modified mesoporous γ-alumina membranes. J Membr Sci. 2002;195:35–50. Cregan V, O’Brien SBG. A note on spin-coating with small evaporation. J Colloid Interface Sci. 2007;314:324–8. Cristiani C, Grossale A, Forzatti P. Study of the physic-chemical characteristics and rheological behavior of boehmite dispersions for dip-coating applications. Top Catal. 2007;42– 43:455–9. Das N, Maiti HS. Ceramic membrane by tape casting and sol-gel coating for microfiltration and ultrafiltration application. J Phys Chem Solids. 2009;70:1395–400. Das S, Pal P, Roy S, Chakraboarty S, Biswas PK. 3-D mapping with ellipsometrically determined physical thickness/refractive index of spin coated sol-gel silica layer. Bull Mater Sci. 2002;25:557–60. De Visschere P, Vanbesien K. Capacitance spectroscopy of alumina sol-gel capacitors with Al top contacts. J Sol-Gel Sci Technol. 2008;45:225–35. De G, Bhattacharyya S. Au nanoparticles in alumina sols and coatings. J Mater Chem. 2008;18:2816–24. Ding DY, Rao JC, Wang DZ, Ma ZY, Geng L, Yao CK. Sol-gel alumina coatings for whisker reinforced metal matrix composites. Mater Sci Eng. 2000a;A279:138–41. Ding DY, Wang DZ, Zhang WL, Yao CK, Rao JC, Li DX. Interfacial reactions and mechanical properties of 6061Al matrix composites reinforced with alumina-coated Al18B4O33 whiskers. Mater Lett. 2000b;45:6–11. Djošić MS, Mišković-Stanković VB, Janaćković DT, Kačarević ZM, Petrović RD. Electrophoretic deposition of boehmite coatings on titanium substrate. Colloids Surf A. 2006;274:185–91. Dou P, Kasada R. Preliminary study on nano- and micro-composite sol-gel based alumina coatings on structural components of lead-bismuth eutectic cooled fast breeder reactors. J Nucl Mater. 2011;409:177–82. Douy A, Odier P. The polyacrylamide gel: a novel route to ceramic and glassy oxide powders. Mater Res Bull. 1989;24:1119–26. Dressler M, Nofz M, Gemeinert M. Rheology, UV-vis transparency and particle size of modified Yoldas sols. J Sol-Gel Sci Technol. 2006;38:261–9. Dressler M, Nofz M, Malz F, Pauli J, J€ager C, Reinsch S, Scholz G. Aluminium speciation and thermal evolution of aluminas resulting from modified Yoldas sols. J Solid State Chem. 2007;180:2409–19. Dressler M, Nofz M, Dörfel I, Saliwan-Neumann R. Influence of sol-gel derived alumina coatings on oxide scale growth of nickel-base superalloy Inconel-718. Surf Coat Technol. 2008a;202:6095–102. Dressler M, Nofz M, Saliwan-Neumann R, Dörfel I, Griepentrog M. Sol-gel derived alumina layers on nickel-base superalloy Inconel-718 (IN-718). Thin Solid Films. 2008b;517:786–92. Dressler M, Nofz M, Klobes P, Dörfel I, Reinsch S. Differences between films and monoliths of sol-gel derived aluminas. Thin Solid Films. 2010;519:42–51. Dressler M, Nofz M, Dörfel I, Saliwan-Neumann R. Diffusion of Cr, Fe, and Ti ions from Ni-base alloy Inconel-718 into a transition alumina coating. Thin Solid Films. 2012;520:4344–9. Du YJ, Damron M, Tang G, Zheng H, Chu C-J, Osborne JH. Inorganic/organic hybrid coatings for aircraft aluminium alloy substrates. Prog Org Coat. 2001;41:226–32. Durán A, Castro Y, Conde A, de Damborenea JJ. Sol-gel protective coatings for metals. In: Sakka S, editor. Handbook of sol-gel science and technology, vol. III. Kluver Academic Publishers, Boston-Dordrecht-London; 2005, Chapter 19. Eckert C, Pflitsch C, Atakan B. Sol-gel deposition of multiply doped thermographic phosphor coatings Al2O3:Cr3+, M3+) (M = Dy, Tm) for wide range surface temperature measurement application. Prog Org Coat. 2010;67:116–9.

38

M. Nofz

EN 1071-3, Advanced technical ceramics – methods of test of ceramic coatings – part 3: determination of adhesion and other mechanical failure modes by a scratch test. Beuth, Berlin-WienZ€urich; 2005. EN ISO 4287, Geometrical Product Specifications (GPS) – surface texture: profile method – terms, definitions and surface texture parameters. Beuth, Berlin-Wien-Z€ urich; 1998. Feigl M, Nofz M, Sojref R, Kohl A. Improved wetting of bare and pre-coated steels by aqueous alumina sols for optimum coating success. J Sol-Gel Sci Technol. 2010;55:191–8. Felde B, Mehner A, Kohlscheen J, Gl€abe R, Hoffmann F, Mayr P. Deposition of alumina coatings on monocrystalline diamonds by sol-gel techniques. Diam Relat Mater. 2001;10:515–8. Feng L, Li H, Song Y, Wang Y. Formation process of a strong water-repellent alumina surface by the sol-gel method. Appl Surf Sci. 2010;256:3191–6. Feng L, Liu Y, Zhang H, Wang Y, Qiang X. Superhydrophobic alumina surface with high adhesive force and long-term stability. Colloids Surf A. 2012;410:66–71. Floch HG, Priotton J-J, Thomas IM. Optical coatings prepared from colloidal media. Thin Solid Films. 1989;175:173–8. Giani L, Cristiani C, Groppi G, Tronconi E. Washcoating method for Pd/γ-Al2O3 deposition on metallic foams. Appl Catal B Environ. 2006;62:121–31. Guido E, Alongi J, Colleoni C, Di Blasio A, Carosio F, Verelst M, Malucelli G, Rosace G. Thermal stability and flame retardancy of polyester fabrics sol-gel treated in the presence of boehmite nanoparticles. Polym Degrad Stab. 2013;98:1609–16. G€und€uz B, Cavaş M, Yakuphanoğlu F. Production of Al2O3 thin films for FET and MOSFET transistor gate applications. 6th international advanced technologies symposium (IATS’11); 2011 May 16–18, Elaziğ. http://web.firat.edu.tr/iats/cd/subjects/Metallurgy%26Material/ MSM-119.pdf Guo W, Lu H, Hu N. Comparative bioelectrochemical study of two types of myoglobin layer-bylayer films with alumina: vapor-surface sol-gel deposited Al2O3 films versus Al2O3 nanoparticle films. Electrochim Acta. 2006;52:123–32. Han L, Karthikeyan KG, Anderson MA, Wouters JJ, Gregory KB. Mechanistic insights into the use of oxide nanoparticles coated asymmetric electrodes for capacitive deionization. Electrochim Acta. 2013;90:573–81. Hawthorne HM, Neville A, Troczynski T, Hu X, Thammachart M, Xie Y, Fu J, Yang Q. Characterization of chemically bonded composite sol-gel based alumina coatings on steel substrates. Surf Coat Technol. 2004;176:243–52. Heinonen S, Huttunen-Saarivirta E, Nikkanen J-H, Raulio M, Priha O, Laakso J, Storgårds E, Lev€anen E. Antibacterial properties and chemical stability of superhydrophobic silvercontaining surface produced sol-gel route. Colloids Surf A. 2014;453:149–61. Horowitz F, Michels AF, Yeatman EM. Optical viscometry of spinning sol coatings. J Sol-Gel Sci Technol. 1998;13:707–12. Hu H, Xiao W, Yuan J, Shi J, He D, Shangguan W. High photocatalytic activity and stability for decomposition of gaseous acetaldehyde on TiO2/Al2O3 composite films coated on foam nickel substrates by sol-gel processes. J Sol-Gel Sci Technol. 2008;45:1–8. Hu B, Yao M, Yang P, Shan W, Yao X. Preparation and dielectric properties of dense and amorphous alumina film by sol-gel technology. Ceram Int. 2013;309:7613–8. Hu B, Yao M, Xiao R, Chen J, Yao X. Optical properties of amorphous Al2O3 thin films prepared by a sol-gel process. Ceram Int. 2014;40:14133–9. H€ubert T, Svoboda S, Oertel B. Wear resistant alumina coatings produced by a sol-gel process. Surf Coat Technol. 2006a;201:487–91. H€ubert T, Schwarz J, Oertel B. Sol-gel alumina coatings on stainless steel for wear protection. J Sol-Gel Sci Technol. 2006b;38:179–84. Hwang S, Kwon OJ, Kim JJ. Method of catalyst coating in micro-reactors for methanol steam reforming. Appl Catal A Gen. 2007;316:83–9. Idrissi-Kandri N, Ayral A, Guizard C, El Ghadraoui EH, Cot L. Synthesis, characterization and first application of new alumina hydrosols. Mater Lett. 1999;40:52–8.

Alumina Thin Films

39

Ienei E, Milea AC, Duta A. Influence of spray pyrolysis deposition parameters on the optical properties of porous alumina films. Energy Procedia. 2014;48:97–104. Islam T, Kumar L, Khan SA. A novel sol-gel thin film porous alumina based capacitive sensor for measuring trace moisture in the range of 2.5–25 ppm. Sensors Actuators B Chem. 2012;173:377–84. ISO 4624, Paints and varnishes – pull-off test for adhesion. Beuth, Berlin-Wien-Z€ urich; 2002. Jia J, Zhou J, Zhang J, Yuan Z, Wang S. The influence of preparative parameters on the adhesion of alumina washcoats deposited on metallic supports. Appl Surf Sci. 2007;253:9099–104. Jing C, Zhao X, Tao H. An approach to predict the solid film thickness possibly yielded from an alumina sol-gel liquid film. Surf Coat Technol. 2006;201:2655–61. Jing C, Xu X, Hou J. Preparation of compact Al2O3 film on metal for oxidation resistance by polyvinylpyrrolidone. J Sol-Gel Sci Technol. 2007a;43:321–7. Jing C, Zhao X, Zhang Y. Sol-gel fabrication of compact, crack-free alumina film. Mater Res Bull. 2007b;42:600–8. Jing C, Zhao X, Zhang Y. Elimination of a zero-growth in thickness of Al2O3 protective film deposited by cycles of dip-coating method. J Sol-Gel Sci Technol. 2007c;42:151–6. JIS K 5400 Testing method for paints, Japanese Standards Association; 1990. Jung Y-I, Kim H-G, Park J-Y, Park D-J, Park J-H. Effect of heat treatment on phase formation and crystallization of sol-gel derived Al2O3, ZrO2-Y2O3, and Ta2O5. J Asian Ceram Soc. 2015;3:217–20. Kákoš J, Bača L, Veis P, Pach L. Photoluminescence spectra and crystallization of θ-Al2O3 and α-Al2O3 from AlOOH-Fe(NO3)3 gels. J Sol-Gel Sci Technol. 2001;21:167–72. Kheirollahi I, Abdellahi M, Emamalizadeh M, Sharifi H. Preparation and characterization of multilayer mesoporous alumina nao membrane via sol-gel method using new precursors. Ceram Int. 2015;41:15083–8. Kim H, Troczynski T. Thick alumina dielectric films on aluminium through chemically bonded composite sol-gel. Ceram Int. 2007;33:333–6. Kim S-H, Kim C-E, Oh Y-J. Influence of Al2O3 diffusion barrier and PBTiO3 seed layer on microstructural and ferroelectric characteristics of PZT thin films by sol-gel spin coating method. Thin Solid Films. 1997;305:321–6. Kim H, Chen M, Yang Q, Troczynski T. Sol-gel alumina environmental barrier coatings for SiC grit. Mater Sci Eng A. 2006;420:150–4. Kozuka H, Ishikawa Y, Ashibe N. Radiative striations of spin-coating films: surface roughness measurement and in-situ observation. J Sol-Gel Sci Technol. 2004;31:245–8. Krivoshapkin PV, Mikhaylov VI, Krivoshapkina EF, Zaikovskii VI, Melgunov MS, Stalugin VV. Mesoporous Fe-alumina films prepared via sol-gel route. Micropor Macropor Mater. 2015;204:276–81. Krivoshapkina EF, Petrakov AP, Krivoshapkin PV, Zubavichos YV, Melgunov MS. Small-angle scattering of synchrotron radiation investigations of nanostructured alumina membranes synthesized by sol-gel method. J Sol-Gel Sci Technol. 2013;68:488–94. Krozel JW, Palazoglu AN, Powell RL. Experimental observation of dip-coating phenomena and the prospect of using motion control to minimize fluid retention. Chem Eng Sci. 2000;55:3639–50. Langlet M. Ultrasonic pulverization of an aerosol: a versatile tool for the deposition of sol-gel thin films. In: Sakka S, editor. Handbook of sol-gel science and technology, vol. I. Kluver Academic Publishers, Boston-Dordrecht-London; 2005, Chapter 13. Liu Z, Liu B, Zhang M, Kong J, Deng J. Al2O3 sol-gel derived amperometric biosensor for glucose. Anal Chim Acta. 1999;392:135–41. Liu Z, Deng J, Ding L. A new tyrosinase biosensor based on tailoring the porosity of Al2O3 sol-gel to co-immobilize tyrosinase and the mediator. Anal Chim Acta. 2000;407:87–96. Liu L, Wang X, Li H, Chen L, Huang X. Al2O3-coated LiCoO2 as cathode material for lithium ion batteries. Solid State Ion. 2002;152–153:341–6. Luo L, Yao J, Li J, Yu J. Preparation and characterization of sol-gel Al2O3/Ni-P composite coatings on carbon steel. Ceram Int. 2009;35:2741–5.

40

M. Nofz

MacKenzie KJD, Temuujin J, Smith ME, Angerer P, Kameshima Y. Effect of mechanochemical activation on the thermal reactions of boehmite (γ-AlOOH) and γ-Al2O3. Thermochim Acta. 2000;359:87–94. Malenovska M, Neouze M-A, Schubert U, Monnier V, Scolan E, Pugin R. Defect-free nanostructured alumina coating doped with noble metal nanoparticles. Appl Surf Sci. 2009;256S:S9–11. Mange F, Fauchadour D, Barré L, Normand L, Rouleau L. A microstructural investigation of nanostructured boehmite films prepared by the sol-gel route. Colloids Surf A. 1999;155:199–210. Marsal A, Ansart F, Turq V, Bonino JP, Sobrino JM, Chen YM, Garcia J. Mechanical properties and tribological behavior of a silica or/and alumina coating prepared by sol-gel route on stainless steel. Surf Coat Technol. 2013;237:234–40. Matsuda A, Tatsumisago M. Electrophoretic sol-gel deposition. In: Sakka S, editor. Handbook of sol-gel science and technology, vol. I. Kluver Academic Publishers, Boston-Dordrecht-London; 2005, Chapter 14. Miki T, Nishizawa K, Suzuki K, Kato K. Microstructure control of porous alumina film using aqueous sol containing poly(ethylene glycol). J Electroceram. 2007;21:524–7. Mitra A, Jana D, De G. Facile synthesis of hexagonally ordered mesoporous aluminium oxide thin films with high catalytic activity. Microporous Mesoporous Mater. 2012;158:187–94. Morgado Jr E, Lam YL, Menezes SMC, Nazar LF. Characterization of peptized boehmite systems: an 27Al nuclear magnetic resonance study. J Colloid Interface Sci. 1995;176:432–41. Morgado Jr E, Lam YL, Nazar LF. Formation of peptizable boehmites by hydrolysis of aluminium nitrate in aqueous solution. J Colloid Interface Sci. 1997;188:257–69. Murali KR, Thirumoorthy P. Characteristics of sol-gel deposited alumina films. J Alloys Compd. 2010;500:93–5. Musić S, Dragčević D, Popović S, Vdović N. Chemical and microstructural properties of Al-oxide phases obtained from AlCl3 solutions in alkaline medium. Mater Chem Phys. 1999;59:12–9. Nitta SV, Jain A, Wayner Jr PC, Gill WN, Plawsky JL. Effect of sol-gel rheology on the uniformity of spin-on xerogel films. J Appl Phys. 1999;86:5870–8. Nofz M, Stösser R, Scholz G, Dörfel I, Schultze D. The thermally induced transformation of pseudoboehmite gels – a comparison of the effects of corundum seeding and iron doping. J Eur Ceram Soc. 2005;25:1095–107. Nofz M, Pauli J, Dressler M, J€ager C, Altenburg W. 27Al NMR study of Al-speciation in aqueous alumina-sols. J Sol-Gel Sci Technol. 2006;38:25–35. Nofz M, Dörfel I, Sojref R. Microstructure of sol-gel derived corundum containing coatings. Thin Solid Films. 2007;515:7145–54. Nofz M, Zietelmann C, Feigl M, Dörfel I, Saliwan NR. Microstructural origin of time-dependent changes in alumina sol-gel coated Inconel 718 exposed to NaCl solution. J Sol-Gel Sci Technol. 2015;75:6–16. Ohya Y. Aqueous precursors. In: Sakka S, editor. Handbook of sol-gel science and technology, vol. I. Kluver Academic Publishers, Boston-Dordrecht-London; 2005, Chapter 5. Olding T, Sayer M, Barrow D. Ceramic sol-gel composite coatings for electrical insulation. Thin Solid Films. 2001;398–399:581–6. Pan C, Shen L, Shang S, Xing Y. Preparation of superhydrophobic and UV blocking cotton fabric via sol-gel method and self-assembly. Appl Surf Sci. 2012;259:110–7. Pandey M, Tyagi K, Mishra P, Saha D, Sengupta K, Islam SS. Nanoporous morphology of alumina films prepared by sol-gel dip coating method on alumina substrate. J Sol-Gel Sci Technol. 2012;64:282–8. Pandey M, Mishra P, Saha D, Sengupta K, Jain K, Islam SS. Nanoporous alumina (γ- and α-phase) gel cast thick film for the development of trace moisture sensor. J Sol-Gel Sci Technol. 2013;68:317–23. Parola S, Verdenelli M, Sigala C, Scharff J-P, Velez K, Veytizou C, Quinson J-F. Sol-gel coatings on non-oxide planar substrates and fibers: a protection barrier against oxidation and corrosion. J Sol-Gel Sci Technol. 2003;26:803–6.

Alumina Thin Films

41

Partlow DP, Yoldas BE. Colloidal versus polymer gels and monolithic transformation in glassforming systems. J Non-Cryst Solids. 1981;46:153–61. Peng P, Li XD, Yuan GF, She WQ, Cao F, Yang DM, Zhuo Y, Liao J, Yang SL, Yue MJ. Aluminium oxide/amorphous carbon coatings on carbon fibers, prepared by pyrolysis of an organicinorganic hybrid precursor. Mater Lett. 2001;47:171–7. Pflitsch C, Siddiqui RA, Atakan B. Phosphorescence properties of sol-gel derived ruby measurement as functions of temperature and Cr3+ content. Appl Phys A. 2008a;90:527–32. Pflitsch C, Siddiqui RA, Eckert C, Atakan B. Sol-gel deposition of chromium doped aluminium oxide films (ruby) for surface temperature sensor application. Chem Mater. 2008b;20:2773–8. Phani AR, Santucci S. Evaluation of structural and mechanical properties of aluminium oxide thin films deposited by a sol-gel process: comparison of microwave to conventional anneal. J Non-Cryst Solids. 2006;352:4093–100. Pietrzyk B, Miszczak S, Szymanowski H, Sobczyk-Guzenda A, Gawronski S. Plasma enhanced aerosol-gel method: a new way of preparing ceramic coatings. J Sol-Gel Sci Technol. 2013a;68:455–63. Pietrzyk B, Miszczak S, Szymanowski H, Kucharsky D. Plasma enhanced aerosol-gel deposition of Al2O3 coatings. J Eur Ceram Soc. 2013b;33:2341–6. Pillonet-Minardi A, Marty O, Bovier C, Garapon C, Mugnier J. Optical and structural analysis of Eu3+-doped alumina planar waveguides elaborated by the sol-gel process. Opt Mater. 2001;16:9–13. Popa AF, Rossignol S, Kappenstein C. Ordered structure and preferred orientation of boehmite films prepared by the sol-gel method. J Non-Cryst Solids. 2002;306:169–74. Qin W, Guan K, Lei B, Liu Y, Peng C, Wu J. One-step coating and characterization of α-Al2O3 microfiltration membrane. J Membr Sci. 2015;490:160–8. Quéré D, de Ryck A, Ramdane O. Liquid coating from a surfactant solution. Europhys Lett. 1997;37:305–10. Radić N, Grbić B, Rožić L, Novaković T, Petrović S, Stoychev D, Stefanov P. Effects of organic additives on alumina coatings on stainless steel obtained by spray pyrolysis. J Non-Cryst Solids. 2011;357:3592–7. Ruhi G, Modi OP, Singh IB, Jha AK, Yegneswaran AH. Wear and electrochemical characterization of sol-gel alumina coating on chemically pre-treated mild steel substrate. Surf Coat Technol. 2006;201:1866–72. Ruhi G, Modi OP, Sinha ASK, Singh IB. Effect of sintering temperatures on corrosion and wear properties of sol-gel alumina coatings on surface pre-treated mild steel. Corros Sci. 2008;50:639–49. Ruhi G, Modi OP, Singh IB. Pitting of AISI 304 stainless steel coated with nano structured sol-gel alumina coatings in chloride containing acidic environments. Corros Sci. 2009a;51:3057–63. Ruhi G, Modi OP, Singh IB. Corrosion behavior of nano structured sol-gel alumina coated 9Cr-1Mo ferritic steel in chloride bearing environment. Surf Coat Technol. 2009b;204:359–65. Saha D, Das S, Sengupta K. Development of commercial nanoporous trace moisture sensor following sol-gel thin film technique. Sensors Actuators B Chem. 2008;128:383–7. Saha D, Ghara DK, Pal M. Nanoporous γ-alumina based novel sensor to detect trace moisture in high temperature and high pressure environment. Sensors Actuators B Chem. 2016;222:1043–9. Schulz W, Nofz M, Feigl M, Dörfel I, Saliwan Neumann R, Kranzmann A. Corrosion of uncoated and alumina coated steel X20CrMoV12-1 in H2O-CO2-O2 and air at 600  C. Corros Sci. 2013a;68:44–50. Schulz W, Feigl M, Dörfel I, Nofz M, Kranzmann A. Influence of a sol-gel alumina coating on oxidation of X20CrMoV12-1 in air up to 650  C. Thin Solid Films. 2013b;539:29–34. Sehrawat P, Mishra P, Islam SS. Spectroscopic analysis of multi-walled carbon nanotube-alumina composite films: optimization of temperature coefficient of resistance and thermal hysteresis for thermal sensor applications. Mater Sci Semicond Process. 2015;31:116–23.

42

M. Nofz

Seok SI, Kim JH, Choi KH, Hwang YY. Preparation of corrosion protective coatings on galvanized iron from aqueous inorganic-organic hybrid sols by sol-gel method. Surf Coat Technol. 2006;200:3468–72. Shafiei A, Oprea C, Troczynski T. Effect of porous alumina coatings on decreasing hydrogen damage to Pb(Zr, Ti)O3. Surf Coat Technol. 2013;224:29–33. Singh IB, Singh M, Das S, Yegneswaran AH. Corrosion behavior of sol-gel Al2O3 coated Al-Si alloy in 3.5% NaCl solution. Indian J Chem Technol. 2010;17:419–24. Singh IB, Modi OP, Ruhi G. Development of sol-gel alumina coating on 9Cr-1Mo ferritic steel and their oxidation behavior at high temperature. J Sol-Gel Sci Technol. 2015a;74:685–91. Singh IB, Gupta P, Maheshwari A, Agrawal N. Corrosion resistance of sol-gel alumina coated Mg metal in 3.5 % NaCl solution. J Sol-Gel Sci Technol. 2015b;73:127–32. Song K-H, Kim H-Y, Kim B-M, Lee K-R. Boehmite sol-coated membrane preparation for the separation of aqueous alcohol. Desalination. 2008;234:244–51. Tadanaga K. Preparation and application of alumina- and titania-nanocrystals dispersed thin films via sol-gel process with hot water treatment. J Sol-Gel Sci Technol. 2006;40:281–5. Tadanaga K, Katata N, Minami T. Formation process of super-water-repellent Al2O3 coating films with high transparency by the sol-gel method. J Am Ceram Soc. 1997;80:3213–6. Tadanaga K, Morinaga J, Minami T. Formation of superhydrophobic-superhydrophilic pattern on flowerlike alumina thin film by the sol-gel method. J Sol-Gel Sci Technol. 2000;19:211–4. Tadanaga K, Kitamuro K, Matsuda A, Minami T. Formation of superhydrophobic alumina coating films with high transparency on polymer substrates by the sol-gel method. J Sol-Gel Sci Technol. 2003;26:705–8. Tadanaga K, Yamaguchi N, Uraoka Y, Matsuda A, Minami T, Tatsumisago M. Anti-reflective properties of nano-structured alumina thin films on poly(methyl methacrylate) substrates by the sol-gel process with hot water treatment. Thin Solid Films. 2008;516:4526–9. Tiwari SK, Sahu RK, Pramanick AK, Singh R. Development of conversion coating on mild steel prior to sol gel nanostructured Al2O3 coating for enhancement in corrosion resistance. Surf Coat Technol. 2011;205:4960–7. Topuz B, Çiftçioğlu M. Sol-gel derived mesoporous and microporous alumina membranes. J Sol-Gel Sci Technol. 2010;56:287–99. Truyen D, Courty M, Alphonse P, Ansart F. Catalytic coatings on stainless steel prepared by sol-gel route. Thin Solid Films. 2006;495:257–61. Van Gestel T, Sebold D, Meulenberg WA, Bram M, Buchkremer H-P. Manufacturing of new nanostructured ceramic-metallic composite microporous membranes consisting of ZrO2, Al2O3, TiO2 and stainless steel. Solid State Ion. 2008;179:1360–6. Vasconcelos DCL, Oréfice RL, Vasconcelos WL. Processing, adhesion and electrical properties of silicon steel having non-oriented grains coated with silica and alumina sol-gel. Mater Sci Eng A. 2007;447:77–82. Vasconcelos DCL, Nunes EHM, Vasconcelos WL. AES and FTIR characterization of sol-gel alumina films. J Non-Cryst Solids. 2012;358:1374–9. Velez K, Quinson JF. Structural characterization by SIMS of alumina coatings on stainless steel. J Sol-Gel Sci Technol. 2000;19:469–72. Vieira Coelho AC, Rocha GA, Souza Santos P, Souza Santos H, Kiyohara PK. Specific surface area and structures of aluminas from fibrillar pseudoboehmite. Revista Materia 2008;13:329–41 Villegas L, Masset F, Guilhaume N. Wet impregnation of alumina-washcoated monoliths: effect of the drying procedure on Ni distribution and on autothermal reforming activity. Appl Catal A Gen. 2007;320:43–55. Vinogradov VV, Agafonov AV, Vinogradov AV. Superhydrophobic effect of hybrid organoinorganic materials. J Sol-Gel Sci Technol. 2010;53:312–5. Vitanov P, Harizanova A, Ivanova T, Dimitrova T. Chemical deposition of Al2O3 thin films on Si substrates. Thin Solid Films. 2009;517:6327–30. Wang Z, Zeng R. Comparison in characterization of composite and sol-gel coating on AZ31 magnesium alloy. Trans Nonferrous Met Soc China. 2010;20:s665–9.

Alumina Thin Films

43

Wang J, Yang S, Chen M, Xue Q. Preparation and characterization of arachidic acid self-assembled monolayers on glass substrate coated with sol-gel Al2O3 thin film. Surf Coat Technol. 2004;176:229–35. Wang T, Pu J, Bo C, Jian L. Sol-gel prepared Al2O3 coatings for the application as tritium permeation barrier. Fusion Eng Des. 2010;85:1068–72. Wang Z, Xu X, Gong J, Ma J, Xu J. Structure and fabry-pérot fringes of polymer-boehmite hybrid thin film by sol-gel dip-coating. J Sol-Gel Sci Technol. 2015;73:154–60. Wilson S, Hawthorne HM, Yang Q, Troczynski T. Sliding and abrasive wear of composite sol-gel alumina coated Al alloys. Surf Coat Technol. 2000;133–134:389–96. Wilson S, Hawthorne HM, Yang Q, Troczynski T. Scale effects in abrasive wear of composite sol-gel alumina coated light alloys. Wear. 2001;251:1042–50. Wouters JJ, Lado JJ, Tejedor-Tejedor MI, Perez-Roa R, Anderson MA. Carbon fiber sheets coated with thin-films of SiO2 and γ-Al2O3 as electrodes in capacitive deionization: relationship between properties of the oxide films and electrode performance. Electrochim Acta. 2013;112:763–73. Wu Z, Liu L, Shen B, Zhong C, Hu W. Effect of α-Al2O3 coatings on the mechanical properties of Ni/SiC composites prepared by electrodeposition. Mater Sci Eng A. 2012;556:767–74. Xie Y, Hawthorne HM. Measuring the adhesion of sol-gel derived coatings to a ductile substrate by an indentation-based method. Surf Coat Technol. 2003;172:42–50. Xu M, Chen Z, Li L, Zhu H, Zhao Q, Xu L, Peng N, Gong L. Highly crystalline alumina surface coating from hydrolysis of aluminium isopropoxide on lithium-rich layered oxide. J Power Sources. 2015;281:444–54. Yamaguchi N, Tadanaga K, Matsuda A, Minami T, Tatsumisago M. Anti-reflective coatings of flowerlike alumina on various glass substrates by the sol-gel process with the hot water treatment. J Sol-Gel Sci Technol. 2005;33:117–20. Yamaguchi N, Tadanaga K, Matsuda A, Minami T, Tatsumisago M. Formation of anti-reflective alumina films on polymer substrates by the sol-gel process with hot water treatment. Surf Coat Technol. 2006;201:3653–7. Yamaguchi N, Tadanaga K, Matsuda A, Minami T, Tatsumisago M. Antireflective properties of flowerlike alumina thin film on soda-lime silica glass substrates prepared by the sol-gel method with hot water treatment. Thin Solid Films. 2007;515:3914–7. Yang P, Yao M, Xiao R, Chen J, Yao X. Preparation of Al2O3 film by sol-gel method on thermally evaporated Al film. Vacuum. 2014;107:112–5. Yao M, Shan W. Dielectric behavior of alumina thin film under high DC electric field prepared by sol-gel method. J Adv Dielectr. 2013;3:1350017 (5 pages). Yao M, Xiao R, Peng Y, Chen J, Hu B, Yao X. The influence of titanium doping on the electric properties of amorphous alumina films prepared by sol-gel technology. J Sol-Gel Sci Technol. 2015;74:39–44. Yarbrough WA, Roy R. Microstructural evolution in sintering AlOOH gels. J Mater Res. 1987;2:494–515. Yoldas BE. A transparent porous alumina. Ceram Bull. 1975a;54:286–8. Yoldas BE. Alumina sol preparation from alkoxides. Ceram Soc. 1975b;54:289–90. Yoldas BE. Alumina gels that form porous transparent Al2O3. J Mater Sci. 1975c;10:1856–60. Yoldas BE. Design of sol-gel coating media for ink-jet printing. J Sol-Gel Sci Technol. 1998;13:147–52. Yuan Q, Zhang X, Wang Z, Shibraen MHMA, Yang S, Xu J. Dimerization of rhodamine B in alumina sol and corresponding dip-coated film. Colloids Surf A. 2015;486:139–44. Zemanová M, Fellner P, Chovancová M. The preparation of combined aluminium oxide coatings on aluminium substrates. Chem Pap. 1996;50:55–9. Zhang XJ, Li Q, Zhao SY, Gao CX, Wang L, Zhang J. Improvement in the oxidation resistance of γ-TiAl-based alloy by sol-gel derived Al2O3 film. Appl Surf Sci. 2008a;255:1860–4.

44

M. Nofz

Zhang XJ, Li Q, Zhao SY, Gao CX, Zhang J. Sol-gel Al2O3/ZrO2 duplex films for protection of a γ-TiAl-based alloy against high-temperature oxidation. J Sol-Gel Sci Technol. 2008b;47:107–14. Zhang X, Honkanen M, Pore V, Lev€anen E, M€antyl€a T. Effect of heat treating gel films on the formation of superhydrophobic boehmite flaky structures on austenitic stainless steel. Ceram Int. 2009;35:1559–64. Zhang X, Mao L, Du J, Wei HJ. Atomic oxygen erosion resistance of sol-gel oxide films on Kapton. J Sol-Gel Sci Technol. 2014;69:498–503. Zhao J, Ma N, Liu B, Zhou Y, Xu C, Kong J. Photoelectrochemistry of photosynthetic reaction centers embedded in Al2O3 gel. J Photochem Photobiol A. 2002;152:53–60. Zhong X, Quing L, Chen B, Wang J, Hu J, Hu W. Effect of sintering temperature on corrosion properties of sol-gel based Al2O3 coatings on pre-treated AZ91D magnesium alloy. Corros Sci. 2009;51:2950–8. Zhou L, Wang YF, Ren FM, Cui XR, Wang XY, Zeng JH. Alumina coatings on fluorine-doped tin oxide@titanium dioxide as photoanode for dye-sensitized solar cells. Electrochim Acta. 2015;173:534–9. Zhu M, Li M, Li Y, Zhou Y. Influence of sol-gel derived Al2O3 film on the oxidation behavior of a Ti3Al based alloy. Mater Sci Eng A. 2006;415:177–83.

Polyhedral Silsesquioxanes Abby R. Jennings, Scott T. Iacono, and Joseph M. Mabry

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthetic Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fully Condensed POSS (Tn) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Incompletely Condensed POSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ladderlike Condensed Silsesquioxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polymeric Nanocomposites Containing POSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Blended POSS Nanocomposites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pendant Copolymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bead Copolymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cross-Linkable Copolymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metallasilsesquioxanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2 3 4 5 6 7 10 14 15 17 20 21

Abstract

This chapter aims to highlight the selected applications of polyhedral oligomeric silsesquioxanes (POSS). A brief introduction will be given and used to highlight the use of solution–gelation chemistry for the synthesis of POSS. Furthermore, the different three-dimensional structures, including fully and incompletely condensed, random, and ladder structures, will be discussed. Subsequently, the use of the various POSS structures as fillers in polymeric nanocomposites will be examined, with a focus on polymer blends, pendant polymers, bead polymers,

A.R. Jennings (*) • S.T. Iacono (*) Department of Chemistry & Chemistry Research Center, United States Air Force Academy, Colorado Springs, CO, USA e-mail: [email protected]; [email protected] J.M. Mabry Aerospace Systems Directorate, Air Force Research Laboratory, Edwards AFB, CA, USA # Springer International Publishing Switzerland (outside the USA) 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_134-1

1

2

A.R. Jennings et al.

and POSS used as cross-linkers. Applications of polyhedral oligometallasilsesquioxanes will then be reviewed.

Introduction Polyhedral oligomeric silsesquioxanes (POSS) are molecular silicas having several rings connected in a three-dimensional confined structure and are represented by the empirical formula (RSiO1.5)n. In this formula, R can be hydrogen, alkyl, aryl, or a number of different organically modified functional groups. POSS molecules are classified as trifunctional siloxanes (T-siloxanes) or containing a silicon atom bound to three oxygen atoms. This designation was coined by General Electric (GE) and further classifies the remaining siloxanes as mono (M)-, di (D)-, and quaternary (Q)-siloxanes (Fig. 1). POSS compounds can be fully or incompletely condensed and take on a number of different three-dimensional structures including cages, ladders, and random conformations. Caged silsesquioxanes were initially developed in the 1930s by GE and Corning Glass Works (now Dow Corning), with the first fully condensed cages being isolated and reported by Scott (1946). Nearly 20 years later, Brown and Vogt (1965) published on the first partially condensed POSS compounds and proposed the first ladderlike POSS materials (Brown 1965). Since then, the field has continued to mature and these hybrid organic–inorganic molecular frameworks have found applications in catalysis (Abbenhuis 2000), biomaterials (Ghanbari et al. 2011), optics (Hartmann-Thompson 2011), and coatings (Ramirez and Mabry 2014). There are a number of noteworthy reviews that have been published on the preparation and applications of POSS (Lickiss and Rataboul 2008; Cordes et al. 2010). This chapter aims to overview the adaptation of solution–gelation (sol–gel) chemistry for obtaining various types of POSS materials and highlight the implementation of these hybrid organic frameworks in polymeric materials. Furthermore, expanded topics including the synthesis and utilization of metallasilsesquioxanes for various applications will be discussed.

Synthetic Strategies The synthesis of a number of POSS materials has relied on the implementation of acid- or base-catalyzed hydrolysis of reactive silanes (Scheme 1). By tailoring reaction conditions, such as reagent concentrations, solvent, pH, temperature, Fig. 1 Basic structures of siloxanes and their corresponding designations

Polyhedral Silsesquioxanes

3

Scheme 1 General hydrolytic condensation reaction for the synthesis of condensed POSS compounds

Fig. 2 Examples of fully condensed POSS cage structures

reaction time, and catalyst, the structures, e.g., cage, ladder, and random, of the resulting POSS materials can be altered.

Fully Condensed POSS (Tn) Fully condensed POSS compounds are cage structures that are generally represented by the notation Tn or RnTn, and cage structures of n = 4, 6, 8, 10, 12, 14, and 16 are known and have been isolated. Figure 2 shows some examples of fully condensed POSS compounds. Hydrolytic condensations of reactive silanes typically form the most stable POSS cage, usually the T8 cage, and as a result, these compounds have dominated the field. Many of the simple T8 compounds prepared by the hydrolysis of reactive silanes can be further modified to form new T8 POSS compounds. Modification techniques that have been employed for this purpose include simple nucleophilic substitutions, oxidations, addition reactions to alkenes, alkene metathesis, thiol–ene chemistry, hydrosilylation reactions, and many more (Cordes et al. 2010). Furthermore, a number of R10T10 and R12T12 cages have been prepared by the cage rearrangements of the corresponding R8T8 cages (Rikowski and Marsmann 1997). Again, once these cage structures are isolated, they can be altered using known chemical transformations into new T10 and T12 cages. The inability to form larger cages (n = 10, 12, 14, and 16) by the simple hydrolysis of reactive silanes has been attributed to the T8 precipitating out of solution prior to the other cages being able to form (Feher and Buzichowski 1989) and the stability of the Si4O4 rings formed in the T8 cage structures (Cordes et al. 2010). In fact, many of the larger POSS cages are isolated as by-products of

4

A.R. Jennings et al.

the T8 cages, in very low yield and after multiple purifications steps. For example, the T10, T12, T14, and T16 cages, where R = H, can be obtained during the synthesis of the H8T8 cage (Agaskar 1991; Agaskar and Klemperer 1995). Noteworthy examples where larger cages were successfully formed in moderate to high yields by the simple hydrolysis of reactive silanes include the Cp10T10 (Bent and Gun’ko 2005) and the R12T12 cages where R = Ph and 2-C4H3S (Brown et al. 1964; Olsson and Axen 1964). Smaller cage structures, both T4 and T6, prepared by acid-catalyzed hydrolysis are synthetically known, although much less common. Only two T4 ((i-pr)4T4 and (t-bu)4T4) cage structures have been successfully isolated, and the driving force for their formation is the bulky R groups employed in the starting chlorosilanes and the stability of the resulting silanol intermediates (Wiberg and Simmler 1955). Moreover, if the acid catalyst is the only source of water in the hydrolysis, T6 derivatives, where R = n-C7H15, n-C8H17, and n-C9H19, can be obtained and isolated after distilling under vacuum (Andrianov and Izmailov 1966, 1976). Other T6 cages have been prepared from reactive silanes by nonaqueous hydrolysis (Bassindale et al. 2003), as well as dehydrating silanol intermediates isolated when bulky R groups are employed in the reactive silanes (Unno et al. 1996, 2003).

Incompletely Condensed POSS Rather than forming discrete Tn clusters, random resinous siloxane oligomers or T-gels can also be formed through acid- or base-catalyzed hydrolysis of reactive silanes (Fig. 3). The implementation of these materials for high-temperature lubricants is currently being examined. Also of significance to the POSS community are the incompletely condensed POSS compounds containing terminal silanol functionalities. Examples of incompletely condensed POSS materials are shown in Fig. 4. Once an incompletely condensed cage has been formed, the reactive silanol groups can be partially or completely reduced to form a wide variety of new POSS materials. Furthermore, incompletely condensed POSS compounds have shown to be ideal for modeling various silica surfaces, surface silanol sites, and silica-supported catalysts (Feher et al. 1989; Herrmann et al. 1994). Incompletely condensed cages can be prepared via incomplete hydrolysis of reactive silanes (Brown 1965), or as shown in Scheme 2, by ring-opening fully condensed cages (Feher et al. 1998, 1999).

Fig. 3 Random siloxane network (T-gel)

Polyhedral Silsesquioxanes

5

Fig. 4 Examples of partially condensed POSS structures

Scheme 2 Examples of partially condensed POSS cages obtained from ringopening fully condensed cages

Fig. 5 Ladderlike condensed silsesquioxane structures

Ladderlike Condensed Silsesquioxanes In addition to the other condensed POSS compounds, the ladderlike polycyclic materials have also gained noteworthy interest. Ladderlike siloxanes encompass a linear cyclic construction with bi- (Shklover et al. 1980), tri- (Brown 1965), penta(Unno et al. 2002), and polycyclic forms being known (Fig. 5).

6

A.R. Jennings et al.

Scheme 3 Synthesis of a syn-tricyclic ladderlike POSS compound by Brown (1965)

Furthermore, they can take on a number of different geometric conformations, including cis-, trans-, syn-, and anti-, with most ladder structures being obtained as a mixture of isomers (Unno et al. 2013). These ladderlike silsesquioxanes possess high thermal stability and exhibit excellent oxidative resistance and have been employed for dielectrics, optical fiber coating, gas separation membranes, and ceramic binders (Xie and Zhang 1997; Zhou et al. 2008). The first ladderlike siloxane compound was proposed in 1965 (Brown 1965) and was prepared by reacting a partially condensed POSS compound with 1,3-dichloro-1,1,3,3-tetramethyldisiloxane (Scheme 3). Many subsequent ladderlike silsesquioxane materials have been obtained following similar synthetic methods.

Polymeric Nanocomposites Containing POSS Polymers and polymeric materials play a vital role in a variety of fields such as aeronautical, aerospace, pharmaceutical, automotive, and petrochemical industries. Generally, polymers are lightweight and easy to process, but exhibit lower modulus, reduced mechanical strength, and thermal resistance when compared to other bulk materials like ceramics and metals. In order to obtain polymers that can withstand more demanding conditions, various inorganic fillers can be employed as additives, yielding hybrid organic–inorganic nanocomposite materials (Fig. 6). Types of inorganic fillers that have been utilized for polymeric nanocomposites include nanoparticles, nanofibers, single- and multi-walled carbon nanotubes, layered silicates, and POSS materials (Wu and Mather 2009; Ayandele et al. 2012). Compared to other nanoreinforcement components, POSS solely exists as a threedimensional or spherical core structure within a polymer matrix and has the ability to build up to higher orders of dimensionality through mechanisms such as aggregation or matrix functionalization. POSS segments can be either blended or covalently linked to the polymer matrix. POSS blends can form various degrees of aggregation inducing the formation of crystalline domains to build up dimensionality, whereas covalently linked POSS structures allow for the degree of aggregation to be minimized. As shown in Fig. 7, when the POSS component is covalently incorporated into the polymer matrix, it can be done so through pendant polymers, bead polymers, or as cross-linkers.

Polyhedral Silsesquioxanes

7

Fig. 6 Interrelationship of organic–inorganic nanocomposites and properties

Fig. 7 Various POSS copolymer architectures

As demonstrated, numerous possibilities of POSS-incorporated scaffolds can be envisioned to produce inorganic–organic hybrid materials that can be integrated for high-strength properties and oxidative resistance of ceramics with ease of processability of polymers. Therefore, by altering molecular-level structure, one can tailor bulk properties to suit specific applications, with each system possessing unique intermolecular bonding motifs.

Blended POSS Nanocomposites An example where a POSS material was physically blended with a polymer to form a nanocomposite was presented by Tuteja et al. (2007). In this instance, the authors synthesized 1H, 1H, 2H, 2H-heptadecafluorodecyl POSS or fluorodecyl POSS

8

A.R. Jennings et al.

Scheme 4 Synthesis of fluorodecyl POSS using sol–gel chemistry

shown in Scheme 4 utilizing sol–gel chemistry. As a result of the fluorine content, the fluorodecyl POSS is a low-surface energy material and is still the most hydrophobic single molecule ever reported (Mabry et al. 2008)). The authors blended the fluorodecyl POSS with poly(methyl methacrylate) (PMMA) and electrospun various surfaces. By incorporating the fluorodecyl POSS molecule at different wt% of PMMA, the authors were able to tailor the hydrophobicity of the blends. Furthermore, it was determined that the POSS molecule underwent surface migration, which induced surface roughness and contributed to the hydrophobicity. The formulations containing low wt% of POSS were found to be both superoleophilic and superhydrophobic. More recently, Ragesh et al. (2014) prepared photocatalytic, hydrophobic fluorinated nanofibers by electrospinning the fluorinated polyhedral oligomeric silsesquioxane shown in Fig. 8 with poly(vinylidene fluoride) (PVDF) and TiO2. Comparing the POSS/PVDF/TiO2 nanocomposite to a control consisting of PVDF/ TiO2 by atomic force microscopy (AFM), it was determined that the incorporation of the POSS compound increased the roughness of the electrospun fibers. Further examination of the nanocomposite and the control by water contact angle measurements determined that the incorporation of the fluorinated POSS also increased the hydrophobicity of the nanocomposites. Finally, scratch tests were performed on the POSS/PVDF/TiO2 nanocomposite and the control. It was found that the POSS/ PVDF/TiO2 nanocomposite exhibited improved mechanical strength. As a result of the incorporation of the POSS compound via blending techniques, the physical properties of the nanocomposite were tailored. Illescas and Arostegui (2014) investigated the influence of the organic functional group incorporated into the POSS materials on the resulting properties of poly (methyl methacrylate) (PMMA)/POSS nanocomposite blends. The POSS materials utilized in this study are shown in Fig. 9. SEM analysis of melt-blended nanocomposites showed that the PMMA formulation made with the fully condensed cages contained aggregates, whereas aggregation could be prevented or induced using the incompletely condensed cage depending on the wt% POSS incorporation. The optical transparency of the blends could also be tailored with the type of POSS

Polyhedral Silsesquioxanes

9

Fig. 8 Fluorinated POSS compound used in blended photocatalytic–hydrophobic nanocomposites

Fig. 9 POSS frameworks studied in PMMA/POSS nanocomposites

material that was incorporated. It was determined that the completely condensed PMMA/POSS blends were transparent at 2.5 wt.% and 5 wt.% POSS incorporation. The 5 wt.% PMMA/i-octyl-POSS formulation had an oil-like coating on the surface due to POSS migration. The PMMA/POSS blend made using the incompletely condensed cage was only transparent at 2.5 wt.% of POSS. Furthermore, an improvement in the oxidative degradation was observed in the incompletely condensed PMMA/POSS and in the octa-functional methyl methacrylate PMMA/POSS melts. More recently, in an effort to make polylactide (PLA) more flexible and robust, Zubrowska et al. (2015) melt blended PLA with two different POSS cages containing different lengths of ethylene glycol derivatives. One of the POSS cages was prepared by employing a hydrosilylation reaction and the synthesis is shown in Scheme 5. The other cage was a mixture [Rn(SiO1.5)n, where n = 8, 10, and 12 and R = CH2CH2(OCH2OCH2)13.3OCH3] and was purchased from commercial sources. It was determined that the commercially available POSS acted as a plasticizer, reducing the Tg values of the melt blends with respect to PLA alone. Analysis by scanning electron microscopy (SEM) showed that these melt blends contained phase separation of the PLA/POSS; however, due to the inability of the commercially available POSS to crystallize, these inclusions were liquid. Furthermore, formulations containing 15 and 20 wt.% were ductile and transparent. Analysis of the melt blends that utilized the POSS shown in Scheme 5 showed that the inclusion of this ethylene glycol-terminated POSS also reduced the glass transition temperature (Tg)

10

A.R. Jennings et al.

Scheme 5 Ethylene glycol-terminated POSS cage used in PLA melt blends

of the resulting materials. Moreover, these melt blends were transparent and elastomeric, and SEM analysis showed no evidence of phase separation. The melt blends prepared from the POSS shown in Scheme 5 were also aged over a 6-month period and were able to retain their appearance, flexibility, and thermomechanical properties.

Pendant Copolymers In order to prepare nanocomposites containing POSS molecules as pendant groups, the POSS material must be terminated by a single reactive functional group including but not limited to an olefin, an amine, or an epoxide. The POSS molecule can then be incorporated into a polymer network by copolymerization, grafting to/from, or end-capping techniques. For instance, copolymers with POSS as pendant groups can be prepared using heptaethyldec-9-enyl POSS monomer as shown in Scheme 6. In the presence of either titanocene- or zirconocene-based catalyst systems activated by methylalumoxane (MAO), high molecular weight polymer can be obtained as a plasticizer enhancer (Tsuchida et al. 1997). More interestingly, they can also be copolymerized with a feedstock of propylene or ethylene that shows a depression in melting point by nearly 20  C compared with the homopolymer. Furthermore, thermostability enhancement was observed with a 50 % increase in decomposition temperature. POSS-incorporated epoxy resins illustrate the ability of POSS to control polymer chain motion within a matrix. Epoxy-functionalized POSS, when reacted with difunctional epoxides in the presence of various diamines, produce epoxy resins that show an increase in glass transition temperatures of the bulk material (Scheme 7) (Lee and Lichtenhan 1998; Lee et al. 2000). Another example of a property enhancing pendant polymer system employing a POSS compound with a single terminal reactive group involves the ring-opening metathesis polymerization (ROMP) of a norbornyl-functionalized POSS cage with norbornene (Scheme 8) (Mather et al. 1999). In the presence of molybdenum catalysts, random copolymers were obtained that were observed to significantly increase glass transition to 69  C compared with polynorbornene at 52.3  C. Not

Polyhedral Silsesquioxanes

11

Scheme 6 Vinyl POSS monomers for polymerization or copolymerization in the presence of metallocene catalysts

Scheme 7 Epoxy-functionalized POSS incorporated into epoxy nanocomposites

Scheme 8 Norbornyl POSS as monomer for ring-opening metathesis polymerization

only was the glass transition increased, X-ray diffraction (XRD) analysis showed no change in the amorphous nature compared to pure polynorbornene. This is another remarkable feature of POSS incorporation where desired properties are enhanced without the expense of significantly altering the molecular composition inherent of the bulk material. As shown in Scheme 9, a further illustration of the diverse utility of POSS copolymer systems is the ability to use styryl-functionalized monomers in chaingrowth polymerizations (Haddad and Lichtenhan 1996). The thermoplastic materials produce improved mechanical properties. At POSS minimal loadings of up to 8 wt.%, glass transition and thermal decomposition temperatures were observed to increase

12

A.R. Jennings et al.

Scheme 9 AIBN initiated chain-growth polymerizations of styryl-functionalized POSS monomers

nearly by 16 and 14  C, respectively. Further rheological studies show the melt and solid phase features are similar to those of poly(4-methylstyrene) homopolymer (Romo-Uribe et al. 1998). POSS molecules containing methyl methacrylate functional groups have also been utilized to make pendant polymers. For instance, Ramirez et al. (2013) reacted an incompletely condensed fluorinated POSS cage with 3-methacryloxypropylmethyldichlorosilane to prepare the methyl methacrylate POSS monomer shown in Scheme 10. The authors then utilized reversible addition-fragmentation chain transfer (RAFT) copolymerization to prepare POSS pendant nanocomposites. By incorporating the fluorinated POSS as a pendant group, the authors were able to tailor the surface energy and thus the wetting properties of the resulting polymers. It was determined that the POSS pendant polymers were non-wetting when compared to poly(methyl methacrylate) alone. Furthermore, when the POSS pendant

Polyhedral Silsesquioxanes

13

Scheme 10 Synthesis of a methyl methacrylate POSS monomer used to prepare pendant POSS polymers

Scheme 11 Synthesis of POSS end-capped polystyrene-block-poly(ethylene oxide) polymers

nanocomposite was used to coat surfaces and fabrics, the materials were superhydrophobic and oleophobic. More recently, Ni et al. (2014) corner capped a fluorinated POSS compound with an azide group using an esterification reaction (Scheme 11). The authors then utilized the copper-catalyzed Huisgen azide/alkyne cycloaddition (CuAAC) to end cap polystyrene-block-poly(ethylene oxide) polymers and studied the self-assembly of the resulting nanocomposites (Dong et al. 2015). By preparing a nanocomposite with three immiscible components, giant surfactants were generated and utilized to study the self-assembly behavior of shape amphiphiles, chain conformation, and phase behaviors of tethered block copolymers. The authors determined that both nanocomposites displayed primary lamellar

14

A.R. Jennings et al.

frames as a result of the phase separation of the block copolymer tails and fluorinated POSS compound. Separation of the fluoroalkyl side chains of the POSS molecules and the silicon-oxygen core, in conjunction with the segregation of the polyethylene oxide and polystyrene blocks, yields secondary lamellar structures.

Bead Copolymers Bead polymers, such as siloxane-based copolymers, are an example of processable precursors for ceramics (Lichtenhan et al. 1993). Similarly devised copolymers with hydrosiloxane-based polymers afford a 14.5 % char yield at a processing temperature of 1,000  C producing SiC, SiO2, and SiOxCy species (Scheme 12) (Mantz et al. 1996). Thermal degradation analysis shows POSS can promote distinct decomposition mechanisms based on the amount of weight percent incorporation. A more recent example where the material properties were enhaced by the incorporation of a POSS molecule was presented by Wang et al. (2015). The authors initially prepared a diacylchloride/POSS-containing monomer from an incompletely condensed POSS compound and then copolymerized the POSS monomer with terephthaloyl chloride (TPC) and a tert-butyldimethylsilyl-functionalized 4,6-diaminoresorcinol (DAR) monomer to make bead polymers (Scheme 13). The authors prepared solution cast films from the bead polymers and investigated film resistance to atomic oxygen (AO) erosion. In comparison to films cast from a DAR/TPC polymer which showed a mass loss 14.06 % after 15 h, the POSS bead polymer exhibited only a 0.41 % mass loss due to AO erosion. Further analysis by AFM and field emission scanning electron microscopy showed that the POSS bead polymer films contained some defects due to oxygen exposure, but were much less severe when compared to the DAR/TPC films. The POSS bead polymer films also demonstrated improved tensile strength and ductility before and after exposure to AO with respect to the DAR/TPC films.

Scheme 12 Siloxane POSS bead copolymers from dimethyl-functionalized dihalo- or aminosilanes

Polyhedral Silsesquioxanes

15

Scheme 13 Synthesis of a diacylchloride–POSS monomer for incorporation into bead polymers

Cross-Linkable Copolymers Unlike the pendant copolymers where the POSS molecule is incorporated through a single binding site, cross-linkable POSS monomers constitute multifunctional bonding with a monomer on the Si O Si core shell. In order to utilize POSS materials as cross-linkers, they must be functionalized with multiple reactive organic groups and similar groups as to those described in the previous section can be employed. For example, a commercially available POSS compound, vinyl POSS, was used as a cross-linker for the preparation of fluorescent thiol–ene films (Ma et al. 2015). In this example, the thiol–ene click reaction was used to crosslink the POSS, 1,6-hexanedithiol, and a vinyl-functionalized porphyrin (Scheme 14). The films were investigated as candidates for the detection of trace explosives through fluorescence signal quenching. It was determined that the incorporation of the POSS molecules in the thiol–ene thin films provided a stable chemical environment for the porphyrin. Uniform films were also achieved, which prevented aggregation-induced self-quenching. Furthermore, it was identified that the inclusion of the POSS molecules yielded films that contained a well-defined porous network with stable dye dispersion, permitting faster diffusion and detection of the explosives.

16

A.R. Jennings et al.

Scheme 14 Synthesis of POSS thiol–ene thin films

Fig. 10 Monomers and POSS cross-linkers utilized by Lungu et al.

More recently, Lungu et al. (2016) prepared binary and ternary methacrylate and epoxy polymer systems that were reinforced with octa-functional POSS compounds containing either methacrylate or epoxide reactive functional groups. The monomers utilized in this system are shown in Fig. 10. For comparison, the authors also conducted studies on the binary and ternary systems with the incorporation of monofunctional methacrylate or epoxide POSS compounds. It was determined that there was a direct relationship between the curing behaviors and the thermomechanical properties of the resulting nanocomposites depending on the type of POSS material utilized. Within the ternary nanocomposites, the polymerization of the methacrylate groups was relatively unaffected by the incorporation of the POSS cross-linkers; however, the polymerization of the epoxy groups was hindered by the incorporation of the POSS compounds. Thermomechanical analysis of the binary nanocomposites showed that the incorporation of the POSS molecules resulted in an increase in the Tg values as a result of the restricted motion of the macromolecular chains. The opposite trend was seen in the ternary nanocomposites with the methacrylate- and epoxideterminated POSS cross-linkers. This was assigned to the nanocomposites having a higher degree of flexibility as a result of the POSS materials and the reduced reactivity of the epoxy groups. The inclusion of the POSS materials within the nanocomposites resulted in an increase in the thermostability. Analysis by SEM showed that there was an increased tendency for agglomeration when octa-functional POSS was used with respect to monofunctional POSS compounds. Additionally, amine-terminated POSS materials have been used as cross-linkers in polyimides. Raaijmakers et al. (2014) synthesized POSS-polyimide thin films by

Polyhedral Silsesquioxanes

17

Scheme 15 Synthesis of POSS cross-linked polyimides for gas separation membranes

Scheme 16 Incompletely condensed POSS utilized as a cross-linker for vulcanized silicone rubber

cross-linking the commercially available OctaAmmonium POSS with 4,4-(hexafluoroisopropylidene) diphthalic anhydride (Scheme 15). The thin films were then investigated as gas separation membranes. In comparison to traditional polymeric membranes, the POSS-polyimide nanocomposites exhibited an elevated thermal stability, and this was attributed to inclusion of the POSS cages. Additionally, the films were found to function over a much wider temperature range and maintain gas selectivity up to 300  C, which is unparalleled to other gas separation membranes. POSS materials utilized as cross-linkers are not unique to fully condensed structures. For example, Shi et al. (2014) prepared a partially condensed crosslinking POSS compound utilizing sol–gel chemistry. As demonstrated in Scheme 16, the authors reacted the partially condensed cage with hydroxyl-terminated polydimethylsiloxane and prepared vulcanized silicone rubber. Morphology analysis of the POSS–silicon rubber nanocomposites showed that the POSS was successfully incorporated into the rubber as a cross-linker and is uniformly dispersed without the presence of large aggregates. The resulting nanocomposites were found to exhibit increased thermal stability and this was partially attributed to the POSS cross-linker. Moreover, the incorporation of POSS resulted in the materials having discrete two-step degradation processes that could be delayed by increasing the wt% of the POSS cross-linker.

Metallasilsesquioxanes Polyhedral oligometallasilsesquioxanes (POMS) were first introduced to the POSS community in the mid-1980s by Feher (1986) and were initially comprised of a POSS framework with a transition metal atom incorporated directly into the Si–O

18

A.R. Jennings et al.

Fig. 11 Examples of POMS structures

Fig. 12 Titanium POMS compounds utilized as epoxidation catalysts

framework. Since then, this unique class of POSS molecules has evolved to include other complex structures, such as having the metallic component attached to the periphery of the POSS structure and containing bridged/sandwich metal–silsesquioxane complexes. Additionally, the more recently developed POMS materials have grown to include other metals and metalloids (Fig. 11). Applications of POMS materials include ligands for metal complexes, homo-/ heterogeneous catalysis, catalyst supports, and tools for modeling the chemical transformations occurring on silica surfaces. The use of POMS compounds for the aforementioned applications can be attributed to organosilicon chemistry being well understood, the POMS compounds having discrete three-dimensional structures where bond lengths and angles are known, the ability to study the POMS compounds utilizing simple techniques such as multinuclear NMR, and the enhanced stability of the POMS materials (Kung et al. 2014). For example, Crocker et al. (1997) prepared a number of different titaniumcontaining POMS compounds by reacting incompletely condensed POSS cages with different homoleptic titanium(IV) complexes to be utilized as catalysts for epoxidation of alkenes. Figure 12 shows some of the POMS structures that were studied. Analysis of the epoxidation of 1-octene with tert-butyl hydroperoxide showed that the POMS compounds were active and selective as epoxidation catalysts. 134-1As an extension of this initial work, Wada et al. (2012) used similar bridged titanium POMS materials for the epoxidation of cyclooctene (Fig. 13). In this

Polyhedral Silsesquioxanes

19

Fig. 13 POMS materials utilized in conjunction with silica supports for the epoxidation of cyclooctene

Fig. 14 Materials utilized to prepare molybdenum catalysts

instance, although the POMS materials were catalytic on their own, the authors combined the POMS materials with solid silica supports. It was determined that the titanium-bridged POMS, when combined with the silica supports, showed an increase in catalytic activity and was assigned to the titanium species being bound to the silica. POMS compounds have also been employed as catalysts for metathesis reactions. For example, Cho et al. (2006) prepared a number of different POMS molybdenum catalysts by mixing the molybdenum complex with the various POSS structures shown in Fig. 14. It was determined that when the fully condensed POSS cage is mixed with the molybdenum complex, a pentacoordinated POMS complex forms and is only active toward alkyne metathesis (Fig. 15). In contrast, the POMS materials prepared using the incompletely condensed cages were reactive toward alkyne metathesis and alkyne polymerizations. In a study presented by Frackowiak et al. (2015), the authors prepared different germanium-containing POMS where the metalloid was incorporated into the cage structure or on the periphery of the POMS. These POMS compounds were prepared by reacting incompletely condensed POSS materials with reactive chlorogermane compounds. The POMS materials were used to prepare new ruthenium POMS complexes, and these materials were investigated as catalysts in the coupling reactions with styrene (Scheme 17). It was determined that all POMS materials were stable toward moisture and air. Furthermore, the catalytic studies performed with the POMS compounds containing the germanium–ruthenium complex and

20

A.R. Jennings et al.

Fig. 15 Pentacoordinate POMS complex used for alkyne metathesis

Scheme 17 Synthesis of POMS-containing germanium and germanium–ruthenium complexes

styrene showed that the structure incorporating the metalloid on the periphery was active toward the coupling of olefins, while the POMS materials that had the germanium incorporated within the framework were inactive.

Conclusions This chapter aims to summarize the implementation of sol–gel chemistry for the synthesis of POSS compounds of various three-dimensional structures. Although there are few examples in the present literature of new POSS materials prepared by sol–gel methods, many of the previously established POSS compounds can be utilized, in conjunction with other chemical transformations, to yield new POSS structures. While polymer nanocomposites and metallasilsesquioxanes were the focus of this chapter, as nanometer-sized materials, POSS compounds exhibit a unique set of structure–property relationships that make these materials useful in a multitude of applications in all facets of materials science. As demonstrated in this chapter, POSS compounds can be included into polymers by a variety of different

Polyhedral Silsesquioxanes

21

900 800

Patents Publications

700 600 500 400 300 200 100 0 95 996 997 998 999 000 001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 1 1 1

19

Fig. 16 Growth of POSS-related publications and patents over the past 20 years

methods, and this in turn can have significant effects of the resulting nanocomposite properties. Given the growth in POSS-related publications and patents over the past 20 years (Fig. 16), fundamental and applied developments in this area continue to grow in pace with the demand of high-performance materials. Analysis of the publications and patents over the last 3 years indicates that research within the POSS community has begun to level off. As the field advances, researchers are faced with the rising cost of synthesizing POSS compounds as well as their limited commercial availability and this could be leveraging the slight decline. Furthermore, research endeavors have evolved into very specialized and focused applications, and the field is transitioning from fundamental to more applied research in areas including biological, medicinal, advanced surfaces and coatings, catalysis, and membrane materials. Regardless, there is still a need to fully understand the structure–property relationships in the advanced POSS polymers, and the anticipation is that the field of POSS-based nanocomposites will only continue to grow.

References Abbenhuis HCL. Advances in homogeneous and heterogeneous catalysis with metal-containing silsesquioxanes. Chem Eur J. 2000;6:25–32. Agaskar PA. New synthetic route to the hydridospherosiloxanes Oh-H8Si8O12 and D5h-H10Si10O15. Inorg Chem. 1991;30:2707–8.

22

A.R. Jennings et al.

Agaskar PA, Klemperer WG. The higher hydridospherosiloxanes: synthesis and structures of HnSinO1.5n (n = 12, 14, 16, 18). Inorg Chim Acta. 1995;229:355–64. Andrianov KA, Izmailov BA. Hydrolytic polycondensation of higher alkyltrichlorosilanes. Zh Obshch Khim. 1966;36:341–5. Andrianov KA, Izmailov BA. Hydrolytic polycondensation of higher alkyltrichlorosilanes in concentrated hydrochloric acid. Zh Obshch Khim. 1976;46:329–35. Ayandele E, Sarkar B, Alexandridis P. Polyhedral oligomeric silsesquioxane (POSS)-containing polymer nanocomposites. Nanomaterials. 2012;2:445–75. Bassindale AR, MacKinnon IA, Maesano MG, Taylor PG. The preparation of hexasilsesquioxane (T6) cages by “non aqueous” hydrolysis of trichlorosilanes. Chem Commun. 2003;1382–3. Bent M, Gun’ko Y. Cyclopentadienyl-functionalised polyhedral silsesquioxanes as building blocks for new nanostructured materials. J Organomet Chem. 2005;690:463–8. Brown JF. The polycondensation of phenylsilanetriol. J Am Chem Soc. 1965;87:4317–24. Brown JF, Vogt LH. The polycondensation of cyclohexylsilanetriol. J Am Chem Soc. 1965;87:4313–7. Brown JF, Vogt LH, Prescott PI. Preparation and characterization of the lower equilibrated phenylsilsesquioxanes. J Am Chem Soc. 1964;86:1120–5. Cho HM, Weissman H, Wilson SR, Moore JS. A Mo(VI) alkylidyne complex with polyhedral oligomeric silsesquioxane ligands: homogeneous analogue of a silica-supported alkyne metathesis catalyst. J Am Chem Soc. 2006;128:14742–3. Cordes DB, Lickiss PD, Rataboul F. Recent developments in the chemistry of cubic polyhedral oligosilsesquioxanes. Chem Rev. 2010;110:2081–173. Crocker M, H. M. Herold R, Crocker M, Guy Orpen A. Synthesis and structural characterisation of tripodal titanium silsesquioxane complexes: a new class of highly active catalysts for liquid phase alkene epoxidation. Chem Commun. 1997;2411–2. Dong X-H, Ni B, Huang M, Hsu C-H, Chen Z, Lin Z, et al. Chain overcrowding induced phase separation and hierarchical structure formation in fluorinated polyhedral oligomeric silsesquioxane (FPOSS)-based giant surfactants. Macromolecules. 2015;48:7172–9. Feher FJ. Polyhedral oligometallasilsesquioxanes (POMSS) as models for silica-supported transition-metal catalysts. Synthesis and characterization of (C5Me5)Zr[(Si7O12)(c-C6H11)7]. J Am Chem Soc. 1986;108:3850–2. Feher FJ, Budzichowski TA. New polyhedral oligosilsesquioxanes via the catalytic hydrogenation of aryl-containing silsesquioxanes. J Organomet Chem. 1989;373:153–63. Feher FJ, Newman DA, Walzer JF. Silsesquioxanes as models for silica surfaces. J Am Chem Soc. 1989;111:1741–8. Feher FJ, Soulivong D, Nguyen F. Practical methods for synthesizing four incompletely condensed silsesquioxanes from a single R8Si8O12 framework. Chem Commun (Cambridge). 1998;1279–80. Feher FJ, Terroba R, Ziller JW. Base-catalyzed cleavage and homologation of polyhedral oligosilsesquioxanes. Chem Commun (Cambridge). 1999;2153–4. Frąckowiak D, Żak P, Spólnik G, Pyziak M, Marciniec B. New vinylgermanium derivatives of silsesquioxanes and their ruthenium complexes – synthesis, structure, and reactivity. Organometallics. 2015;34:3950–8. Ghanbari H, Cousins BG, Seifalian AM. A nanocage for nanomedicine: polyhedral oligomeric silsesquioxane (POSS). Macromol Rapid Commun. 2011;32:1032–46. Haddad TS, Lichtenhan JD. Macromolecules. 1996;29:7302. Hartmann-Thompson C. Polyhedral oligomeric silsesquioxanes in electronics and energy applications. In: Hartmann-Thompson C, editor. Applications of polyhedral oligomeric silsesquioxanes. New York: Springer; 2011. p. 247–325. Herrmann WA, Anwander R, Dufaud V, Scherer W. Molecular siloxane complexes of rare earth metals – model systems for silicate-supported catalysts? Angew Chem Int Ed. 1994;33:1285–6. Illescas S, Arostegui A. Influence of polyhedral oligomeric silsesquioxanes on thermal and mechanical properties of melt-mixed poly(methyl methacrylate)/polyhedral oligomeric silsesquioxanes composites. High Perform Polym. 2014;26:307–18.

Polyhedral Silsesquioxanes

23

Kung MC, Riofski MV, Missaghi MN, Kung HH. Organosilicon platforms: bridging homogeneous, heterogeneous, and bioinspired catalysis. Chem Commun. 2014;50:3262–76. Lee A, Lichtenhan JD. Macromolecules. 1998;31:4970. Lee A, Lichtenhan JD, Reinerth WA. Polym Mater Sci Eng. 2000;82:235. Lichtenhan JD, Vu NQ, Carter JA, Gilman JW, Feher FJ. Silsesquioxane-siloxane copolymers from polyhedral silsesquioxanes. Macromolecules. 1993;26:2141–2. Lickiss PD, Rataboul F. Fully condensed polyhedral oligosilsesquioxanes (POSS): from synthesis to application. In: Anthony FH, Mark JF, editors. Advances in organometallic chemistry. Oxford: Academic; 2008. p. 1–116. Lungu A, Florea NM, Manea M, Vasile E, Iovu H. Polyhedral oligomeric silsesquioxanes nanoreinforced methacrylate/epoxy hybrids. J Appl Polym Sci. 2016;133. Ma J, Lv L, Zou G, Zhang Q. Fluorescent porous film modified polymer optical fiber via “click” chemistry: stable dye dispersion and trace explosive detection. ACS Appl Mater Interfaces. 2015;7:241–9. Mabry JM, Vij A, Iacono ST, Viers BD. Fluorinated polyhedral oligomeric silsesquioxanes (F-POSS). Angew Chem Int Ed. 2008;47:4137–40. Mantz RA, Jones PF, Chaffee KP, Lichtenhan JD, Gilman JW, Ismail IMK, et al. Chem Mater. 1996;8:1250. Mather PT, Jeon HG, Romo-Uribe A, Haddad TS, Lichtenhan JD. Mechanical relaxation and microstructure of poly(norbornyl-POSS) copolymers. Macromolecules. 1999;32:1194–203. Ni B, Dong X-H, Chen Z, Lin Z, Li Y, Huang M, et al. “Clicking” fluorinated polyhedral oligomeric silsesquioxane onto polymers: a modular approach toward shape amphiphiles with fluorous molecular clusters. Polym Chem. 2014;5:3588–97. Olsson K, Axen C. Octakis(arylsilsesquioxanes), (ArSi)8O12. II. Octa- and dodecakis (2-thienylsilsesquioxane) and their perbromo derivatives. Ark Kemi. 1964;22:237–44. Raaijmakers MJT, Hempenius MA, Schön PM, Vancso GJ, Nijmeijer A, Wessling M, et al. Sieving of hot gases by hyper-cross-linked nanoscale-hybrid membranes. J Am Chem Soc. 2014;136:330–5. Ragesh P, Nair SV, Nair AS. An attempt to fabricate a photocatalytic and hydrophobic self-cleaning coating via electrospinning. RSC Adv. 2014;4:38498–504. Ramirez S, Mabry J. Fluorinated silsesquioxanes. In: Smith DW, Iacono ST, Iyer SS, editors. Handbook of fluoropolymer science and technology. Hoboken: Wiley; 2014. p. 545–63. Ramirez SM, Diaz YJ, Sahagun CM, Duff MW, Lawal OB, Iacono ST, et al. Reversible additionfragmentation chain transfer (RAFT) copolymerization of fluoroalkyl polyhedral oligomeric silsesquioxane (F-POSS) macromers. Polym Chem. 2013;4:2230–4. Rikowski E, Marsmann HC. Cage-rearrangement of silsesquioxanes. Polyhedron. 1997;16:3357–61. Romo-Uribe A, Mather PT, Haddad TS, Lichtenhan JD. J Plym Sci B Polym Phys. 1998;36:1857. Scott DW. Thermal rearrangement of branched-chain methylpolysiloxanes. J Am Chem Soc. 1946;68:356–8. Shi Y, Gao X, Zhang D, Liu Y, Huang G. Synthesis and thermal properties of modified room temperature vulcanized (RTV) silicone rubber using polyhedral oligomeric silsesquioxane (POSS) as a cross linking agent. RSC Adv. 2014;4:41453–60. Shklover VE, Struchkov YT, Makarova NN, Zhdanov AA. cis-1,7-diphenyl-3,3,5,5,9,9,11,11octamethylbicyclohexasiloxane, C20H34O7Si6. Cryst Struct Commun. 1980;9:1–5. Tsuchida A, Bolln C, Sernetz FG, Frey H, Mulhaupt R. Ethene and Propene Copolymers Containing Silsesquioxane Side Groups. Macromolecules. 1997;30:2818. Tuteja A, Choi W, Ma M, Mabry JM, Mazzella SA, Rutledge GC, et al. Designing superoleophobic surfaces. Science. 2007;318:1618–22. Unno M, Alias SB, Saito H, Matsumoto H. Synthesis of hexasilsesquioxanes bearing bulky substituents: hexakis((1,1,2-trimethylpropyl)silsesquioxane) and hexakis(tert-butylsilsesquioxane). Organometallics. 1996;15:2413–4. Unno M, Suto A, Matsumoto H. Pentacyclic laddersiloxane. J Am Chem Soc. 2002;124:1574–5. Unno M, Imai Y, Matsumoto H. Hexakis(2,4,6-triisopropylphenylsilsesquioxane). Silicon Chem. 2003;2:175–8.

24

A.R. Jennings et al.

Unno M, Suto A, Matsumoto T. Laddersiloxanes – silsesquioxanes with defined ladder structure. Russ Chem Rev. 2013;82:289. Wada K, Sakugawa S, Inoue M. Facile preparation of silica-supported Ti catalysts effective for the epoxidation of cyclooctene using Ti-bridged silsesquioxanes. Chem Commun. 2012;48:7991–3. Wang P, Tang Y, Yu Z, Gu J, Kong J. Advanced aromatic polymers with excellent antiatomic oxygen performance derived from molecular precursor strategy and copolymerization of polyhedral oligomeric silsesquioxane. ACS Appl Mater Interfaces. 2015;7:20144–55. Wiberg E, Simmler W. Über eine neue Klasse von Siliconen mit Urotropin- Struktur. Z Anorg Allg Chem. 1955;282:330–44. Wu J, Mather PT. POSS polymers: physical properties and biomaterials applications. Polym Rev. 2009;49:25–63. Xie P, Zhang R. Functionalization and application of ladder-like polysilsesquioxanes. Polym Adv Technol. 1997;8:649–56. Zhou Q, Yan S, Han CC, Xie P, Zhang R. Promising functional materials based on ladder polysiloxanes. Adv Mater (Weinheim, Ger). 2008;20:2970–6. Zubrowska A, Piorkowska E, Kowalewska A, Cichorek M. Novel blends of polylactide with ethylene glycol derivatives of POSS. Colloid Polym Sci. 2015;293:23–33.

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in Sol–Gel Synthesis of Silicate–and Phosphate–Based Materials Makoto Kobayashi, Hideki Kato, and Masato Kakihana

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water-Dispersed Silicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conventional Silicon Raw Materials for Aqueous Solution-Based Sol-Gel Methods . . . . . . Preparation of Glycol-Modified Silane (GMS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chemistry of GMS Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stability of GMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of GMS for Synthesis of Silicates by an Amorphous Metal Complex Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrothermal Gelation Synthesis of Silicates Using GMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Synthesis of Silicates by Homogeneous Precipitation Using GMS . . . . . . . . . . . . . . . . . . . . . . . . . Preparation of Monodispersed Silica Using GMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water-Soluble Phosphates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Problems of Conventional Phosphorous Raw Materials for Aqueous Solution-Based Sol-Gel Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preparation of an Ethylene Glycol-Conjugated Phosphorous Ester . . . . . . . . . . . . . . . . . . . . . . . . . Preparation of a Polyethylene Glycol-Conjugated Phosphorous Ester . . . . . . . . . . . . . . . . . . . . . . Stability and Reactivity of EG-P and PEG-P in Aqueous Conditions . . . . . . . . . . . . . . . . . . . . . . . Synthesis of Phosphates by a Polymerizable Complex Method Employing EG-P and PEG-P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Development of New Phosphates Using EG-P and PEG-P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 3 3 4 5 6 9 10 14 16 16 16 18 20 22 24 25 25

Abstract

Silicon and phosphorous are important elements in materials science. There are less proper raw materials for synthesis of their compounds by aqueous solutionbased sol-gel methods, because the conventional raw materials have low stability M. Kobayashi (*) • H. Kato • M. Kakihana Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai, Japan e-mail: [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_135-1

1

2

M. Kobayashi et al.

in aqueous conditions and/or high reactivity with other elements such as metal ions. In this chapter, preparation, chemistry, property, and stability of glycolmodified silanes (GMSs) and glycol-conjugated phosphate esters are summarized. The former is dispersed in an aqueous solution as silicon clusters, and the latter is dissolved in water. In addition, their applications for synthesis of silicates and phosphates by aqueous solution-based sol-gel methods are also documented using examples. The materials obtained showed high performance and phase purity caused by homogeneity of constituents at the atomic level. It is also clearly showed that introduction of such unconventional raw materials enables us to develop new compounds and new synthetic routes for synthesizing highperformance materials.

Introduction As discussed in chapter 5, water can be regarded as the best solvent for the sol-gel synthesis of materials. However, elements such as titanium, tantalum, and niobium show strong hydrolysis behavior in aqueous conditions. In the case of titanium, although TiCl4 and Ti(SO4)2 are available as aqueous solutions, they are stable only under highly acidic conditions. An aqueous solution of TiOCl2 can be prepared when TiCl4 is added into iced water, and precipitates are not formed in the solution even under neutral conditions. Nevertheless, long-term storage of such solutions is still a problem (Kakihana et al. 2010). The hydrolysis behavior of compounds is sometimes discussed in terms of the average electronegativity of the central cation and the ligands or counter ions. According to the partial charge model (Livage et al. 1988; Henry et al. 1992), small partial charges of the cation (central ion) and anion (ligand) could reduce the tendency for hydrolysis. Therefore, exchange of ligands is an effective way to prepare water-soluble and stable source materials. This chapter focuses on silicon and phosphorous. Both elements are important in the field of materials science, due to their abundance and the high functional abilities of their compounds. Silicon-containing materials with high functions and controlled structures could be obtained using alkoxides, and the related chemistry is established (see other chapters). However, for aqueous solution-based sol-gel methods, fumed silica is often used because there are no suitable water-soluble precursors. In the field of organic chemistry, ligand exchange reactions are well-documented. However, those techniques have hardly been applied in materials science due to a gap in knowledge sharing. Using knowledge from organic chemistry, water-dispersed silanes and water-soluble and stable phosphate esters can be prepared. This chapter summarizes their preparation and stability under aqueous conditions. In addition, their applications for the synthesis of ceramics by sol-gel methods are documented.

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

3

Water-Dispersed Silicates Conventional Silicon Raw Materials for Aqueous Solution-Based Sol-Gel Methods There are very few suitable raw materials for the synthesis of multicomponent silicon-containing materials, such as Ca2SiO4 and glass, by aqueous solutionbased sol-gel methods. Usually, alkoxides such as tetraethyl orthosilicate (TEOS) are chosen for the synthesis of silicon-containing ceramics by sol-gel methods. However, TEOS is not soluble in water and it hydrolyzes to form an insoluble gel. Therefore, alcohols and other polar solvents are used with alkoxides to prepare silicon-containing materials by sol-gel methods, as discussed in other chapters. In addition, silicon alkoxides are highly volatile, which means that precise control of its constituent ratio with alkoxides often requires numerous trial and error experiments. Therefore, more reproducible approach is desired. Hexafluorosilicate (H2SiF6) and sodium/potassium silicates (known as water glass, liquid glass, or soluble glass) are water-soluble silicon compounds. The former can be obtained through a reaction between silica and hydrogen fluoride, but it has strong acidity and high toxicity. The latter exist in highly concentrated alkaline solutions and also contain alkaline metals. Therefore, these well-known watersoluble silicates are not suitable for the synthesis of multicomponent silicates using sol-gel methods in aqueous solutions. The silicate anions are another candidate water-soluble silicon raw material. Although they form random polymers in general, distinct silicate anions with unique structures can be formed using tetraalkylammonium under appropriate conditions. This approach has enabled us to obtain water-soluble silicon materials such as Si8O208
with a double four-membered cage (Hasegawa et al. 1986).

Preparation of Glycol-Modified Silane (GMS) According to the partial charge model mentioned earlier in this chapter (Livage et al. 1988; Henry et al. 1992), the stability of a compound against hydrolysis can be improved by changing its ligands. In 1967, Mehrotra and Narain reported the reaction between silicon alkoxides and glycols and succeeded in preparing glycolsubstituted (modified) silane (GMS) (Mehrotra and Narain 1967). The reaction was carried out as follows: A glycol such as ethylene glycol was added to a solution containing TEOS as the silicon source and benzene as the solvent, with molar ratio of silicon/glycol = 1 or 2. Glycol cannot be mixed into the solution homogeneously; therefore, two phases were observed. The mixture was refluxed and alcohols formed by the reaction between the alkoxide and glycol were removed by azeotropic distillation, then a white precipitate was gradually formed. After drying, silane partially substituted with glycol was obtained as a white powder insoluble in water. Mehrotra and Narain also reported the preparation of silanes modified by α-hydroxy acids such as glycolic acid, using alkoxide in a similar way (Mehrotra and Narain 1968).

4

M. Kobayashi et al.

A solution of GMS conjugated with ethylene glycol had been available from Sigma-Aldrich as an ethylene glycol solution of tetrakis(2-hydroxyethyl) orthosilicate (CAS No. 17622–94–5). The chemical structure is illustrated in Fig. 1. Using an acid catalyst, such GMS solutions can be prepared without additional solvent such as benzene (Suzuki and Kakihana 2009; Takahashi et al. 2009). Glycol and tetraethoxysilane are put into a conical flask. An inorganic acid such as HCl and HNO3 is added into the mixture as catalyst. Then the flask is capped and the mixture stirred. First, the mixture becomes an emulsion. After a few hours (depending on the temperature), a homogeneous GMS solution is obtained. The reaction proceeds faster if the mixture is heated (up to 373 K). For example, at 353 K, the reaction is completed within 10 min. Note that if the conical flask is too small, the cap can be blown out due to the vapor pressure of alcohol formed by ligand substitution. Using a large flask (e.g., a 200 mL flask for 0.1 mol of silicon) is therefore highly recommended. The as-prepared solution contains partially substituted GMS, that is, Si(OR)4-n(OR0 OH)n (n  4), because the reaction mixture is an equilibrium between alkoxides and silanes. The formed alcohol can be easily removed by vacuum, resulting in the formation of fully substituted GMS (Fig. 1), which has a high viscosity. The concentration of silicon is not changed during this process, meaning that the solution has a low vapor pressure (according to safety data sheet (SDS) supplied by Sigma-Aldrich, the boiling point of ethylene glycolmodified silane is 473 K at 1 atm), which is one of the advantages for this synthesis of silicon-containing ceramics over that using the volatile alkoxides. Polyalcohols (e.g., glycerine C3H5(OH)3) and polyester alcohols (e.g., polyethyleneglycol, PEG; HO-(CH2-CH2-O)n-H) can also be used as ligands. On the other hand, the introduction of α-hydroxy acids leads to the formation of insoluble precipitates using the same procedure. Silicon tetrachloride (SiCl4), which is a raw material of silicon alkoxides, is not applicable for such preparation of GMS solutions because a volatile solid is formed. This phenomenon can be explained as follows: Cl
is easily detached from SiCl4, then the bridge formation between Si ions by glycol proceeds rapidly, resulting in the insoluble solid (Fleming 1987).

Chemistry of GMS Formation GMS is formed through the alkoxy substitution reaction between alkoxide and glycol. The reaction is well-known to be a bimolecular nucleophilic substitution reaction as shown in Fig. 2 (Klein 1985). Silicon in the alkoxide is positively charged with a low electron density due to the difference in electron negativity between silicon and the alkoxy group. Therefore, silicon attracts negatively charged molecules (nucleophiles) readily. On the other hand, the hydroxy groups in polyalcohols have high electron density and therefore readily react with silicon. In the present reaction, silicon in the alkoxide will be attacked by polyalcohol from behind as the initial step. Then a transition state of alkoxide with a trigonal bipyramidal molecular geometry is formed, in which one alkoxy and one polyalcohol are positioned along

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

5

Fig. 1 Chemical structure of tetrakis(2-hydroxyethyl) orthosilicate, an ethylene glycol-modified silane

Fig. 2 Reaction scheme of alkoxy substitution reaction between alkoxide and glycol

the same axis. Alcohol is released from the transition state and the reaction is terminated.

Stability of GMS Figure 3 shows the 29Si NMR spectrum of an aqueous solution of propylene glycolmodified silane (PGMS). Peaks at
73,
82, and
91 ppm are assigned as Q0, Q1, and Q2, which may correspond to Si(OH)4, (HO)3Si-O-Si(OH)3, and (HO)3Si-O-Si (OH)2-O-Si(OH)3, respectively. Therefore, GMS cannot remain dissolved in water as it is: the aqueous solution is gradually hydrolyzed to form an insoluble, transparent, colorless gel-like substance. The gelation time is dependent on the temperature and pH. Figure 4 shows the time required for a 0.7 M PGMS aqueous solution adjusted to various pH values to hydrolyze and solidify at room temperature and at 363 K (Yoshizawa et al. 2010). At pH less than 4 and room temperature, the solution remained a colorless transparent liquid even after 1 month. As the pH was increased, the hydrolysis progressed faster. The solution became a gel and solidified in several hours at pH 5.5 and within 1 h at pH 6–8.4. However, at pH 8.8–10, the time to gel became longer. At pH 11 the hydrolysis progressed rapidly after adding PGMS to the aqueous solution, and a white precipitate was formed. This hydrolysis behavior of PGMS at room temperature is similar to that of TEOS (Iler 1979). At 363 K the hydrolysis progressed more quickly than at room temperature, and the solution gelled in 1–6 h even when the pH was less than 4. This trend in hydrolysis behavior by pH is similar to that of GMS modified by other glycols, although the time required

6

M. Kobayashi et al.

Fig. 3 29Si NMR spectrum of an aqueous solution prepared by adding propylene glycol-modified silane in water Over 1 month

RT

GElation time / h

102

101

363 K

10−0

10−1 2

3

4

5

6

7

8

9

pH

Fig. 4 Dependence of the gelation time on the pH of the aqueous PGMS solution at room temperature and 363 K (Yoshizawa et al. 2012)

for gelation is slightly different (Table 1). As a whole, GMSs conjugated with longer chains (e.g., PEG) have higher stability.

Application of GMS for Synthesis of Silicates by an Amorphous Metal Complex Method Using GMSs, a homogenous aqueous solution containing silicon and other cations can be prepared for aqueous solution-based sol-gel synthesis of silicates. As a model,

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . . Table 1 Time required for the gelation of 0.25 M aqueous solutions of different GMSs at 393 K

Ligand Propylene glycol (PG) 1,3-butanediol (1,3-BGMS) 1,2-butanediol (1,2-BGMS) Polyethylene glycol (PEG) Polypropylene glycol (PPG) a

7 Gelation time/h 3.5 4.0 5.0a 12.0 2.5b

Turbid gel Phase was separated

b

this section introduces the synthesis of a Eu2+-doped BaZrSi3O9 phosphor by an amorphous complex method, which is recognized as one of the sol-gel methods, using a chelate and various GMSs. Figure 5 is a flowchart of the synthesis of BaZrSi3O9 powder doped with Eu2+ by an amorphous metal complex method using GMS. BaCO3, ZrOCl2, and Eu(NO3)3 with a stoichiometric ratio of Ba:Zr:Eu = 1
X:1:X are dissolved together in an aqueous solution of citric acid. The molar ratio of citric acid to the total metal amount is usually 4. To enhance complexation between the metal ions and citric acid, the solution is stirred at 353 K for 2 h. Then an aqueous solution of GMS is added, and the mixture is heated at 393 K. The treatment produces a transparent gel. When the reaction occurs successfully, the gel is transparent. When using GMS prepared with polyester alcohols, gelation may not occur at 393 K, and the temperature needs to be increased to 423 K or above. Note that there is no need to consider excess silicon, as the GMSs are not volatile. The gel is heated at 823 K for 6 h to eliminate organic compounds and calcined at 1,073 K for 12 h to obtain the precursor. If a large sample is prepared at once, repeated calcination is required. The ideal precursor is amorphous or with the target material BaZrSi3O9. If ZnO, BaCO3, or other impurity phases are present, the following reaction is the same as a conventional solid-state reaction, that is, ion distribution occurs only between interfaces of each bulk and homogeneity is not achieved at the atomic level. The precursor is put in a double crucible with graphite (Komukai et al. 2015) and calcined at 1,673 K for 2 h. During heating, carbon creates a reducing atmosphere by producing CO and CO2. Figure 6 shows the photoluminescence spectra of 3 mol% Eu-doped BaZrSi3O9 (Ba0.97Eu0.03ZrSi3O9) synthesized by the amorphous metal complex method employing various GMSs. Samples using different GMSs showed the same peak shapes in excitation and emission spectra, because all the samples were of almost single-phase target material. Nevertheless, the samples had slightly different emission intensities. The powder synthesized using 1,3-butanediol-modified silane (1,3-BGMS) showed the highest emission intensity, with the internal (IQE) and external quantum efficiencies (EQE) being 66.1 % and 52.9 % at the excitation of 300 nm, respectively. On the other hand, samples synthesized using GMS conjugated with polyethylene glycol (PEG) or polypropylene glycol (PPG) showed lower efficiencies (IQE = 60.4 % and EQE = 50.9 % for PEG-modified silane, IQE = 53.0 % and EQE = 40.5 % for PPG-modified silane). As shown in Table 1, 1,3-BGMS has the higher stability against gelation than that of PGMS. It implies that

8

M. Kobayashi et al.

Fig. 5 A flowchart of the synthesis of Eu2+-doped BaZrSi3O9 powder by an amorphous metal complex method using GMS

Fig. 6 Photoluminescence spectra of 3 mol% Eu2+-doped BaZrSi3O9 samples synthesized by the amorphous metal complex method using various GMSs (λex = 300 nm, λem = 490 nm)

the gelation of citric acid complexes occurs prior to the hydrolysis of GMS, resulting in a gel whose constituents are homogeneously dispersed at the atomic level. Using 1,2-BGMS, PEGMS, or PPGMS, gels are not transparent, implying agglomeration

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

9

Fig. 7 A schematic illustration of the ideal gelation of GMS in an aqueous solution containing other cations, for synthesizing multicomponent ceramics

of ions. In general, a more homogeneous distribution of emission centers leads to higher emission intensity for phosphors activated by doping. Therefore, the sample obtained using 1,3-BGMS exhibited the strongest emission. Other silicates, not only phosphors (Sato et al. 2015) but also photocatalysts such as Ba3Ta4Si4O24 (Yanagisawa et al. 2010) with better property than those synthesized by other methods, can also be obtained using similar approaches.

Hydrothermal Gelation Synthesis of Silicates Using GMS As discussed in the previous section, the gelation of GMSs can be accelerated by heating. Using this property, other elements can be incorporated into the gel as shown in Fig. 7. This section discusses a hydrothermal gelation (HG) method utilizing gelation of GMSs by heating and describes the synthesis of a Ce3+-doped Ca3Sc2Si3O12 (CSS) green-emitting phosphor as an example. Single phase of CSS can also be obtained by a conventional solid-state reaction method. However, it requires a much higher calcination temperature (usually over 1,673 K) than that needed following the HG method (1,073 K). Figure 8 displays the flowchart of HG synthesis of the Ce3+-doped CSS. An aqueous solution of CaCl2, Ce2(NO3)3 H2O, and Sc(NO3)3 5H2O and an aqueous solution of GMS with a given stoichiometric ratio are mixed in water. The solution is hydrothermally treated in an autoclave at 473 K for 3 h. GMS is hydrolyzed during heating and becomes a gel, as shown in Fig. 9. Note that extended hydrothermal treatment produces a sol instead, and single phase of the target material could not be obtained. That is, the formation of the wet gel shown in Fig. 9 is important for synthesizing materials with high homogeneity. Solvent is incorporated in the gel. Experiment has revealed that the gel retained five times its volume of water when 2.0 M PGMS was used. The gel is dried at 393 K and then heated at 823 K for 3 h to remove organic compounds, followed by calcination at 1,073 K for 12 h to obtain the

10

M. Kobayashi et al.

Fig. 8 The synthesis procedure of Ce3+-doped CSS by the HG method using GMS

Fig. 9 A photograph of the gel obtained by the HG method

precursor. The precursor consists of almost single phase of CSS. In contrast, crystalline CaCO3 and Sc2O3 are present in CSS synthesized using the conventional solid-state reaction method or other sol-gel methods (such as a polymerizable complex method) after the same heat treatment (Yamaguchi et al. 2010). Therefore, the present approach produces a precursor with homogeneously dispersed constituents. As a result, single-phase Ce3+-doped CSS with high luminescence upon irradiation is obtained after the precursor is reduced at 1,273 K for 2 h under CO-CO2 atmosphere using carbon and a double crucible.

Synthesis of Silicates by Homogeneous Precipitation Using GMS The hydrolysis behavior of GMSs is strongly pH dependent, as shown in Fig. 4. Homogeneous precipitation (HP) is a ceramic synthesis method using an aqueous

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

11

Fig. 10 A flowchart of the synthesis of Zn2+-doped Zn2SiO4 by the HP method using GMS

solution (Matijevic 1994; Velikov and Blaaderen 2001). The HP method involves changing the pH of the solution uniformly to promote the slow hydrolysis of a precipitant (such as urea) by heating, in order to obtain a homogeneous precipitation. The GMSs can be applied for the synthesis of silicates using the HP method. According to the solubility curves of a zinc ammine complex and a manganese carbonate-ammine complex (Kragten 1978), in both systems, the solubility of metal ions rapidly increases at approximately pH = 6 irrespective of the ammonia concentration. GMSs are also rapidly hydrolyzed in the same pH range, as shown in Fig. 4. In this section, the synthesis of a Mn2+-doped Zn2SiO4 phosphor by the HP method using GMS is introduced (Yoshizawa et al. 2010). The synthesis of a Ce3+-doped Y2SiO5 phosphor based on the same strategy is also described to clarify the mechanism and usability of the HP method using GMSs. Figure 10 is the flowchart of the synthesis of Zn(2
X)MnXSiO4 powder by the HP method using GMSs. Stoichiometric amounts of Zn(NO3)2 and Mn(NO3)2 are dissolved in an aqueous solution and placed in a conical flask or a beaker. An aqueous solution of GMS is added, and then HCl is added to the mixed solution to set the initial pH to 2.5 (if needed). The faster hydrolysis of GMS at a higher pH makes it suitable for the HP method. However, the hydrolysis of GMS also becomes faster at higher temperatures. This suggests that higher operating temperature and the rate of pH increase are crucial factors for obtaining a homogeneous precipitation, in which zinc, manganese, and silicon are intimately mixed at the atomic level. In order to obtain Mn2+-doped Zn2SiO4 by the HP method, a mixed aqueous solution of GMS, Zn(NO3)2, and Mn(NO3)2 is prepared at room temperature and adjusted to an initial pH of 2–3 to reduce the hydrolysis of GMS. Then, urea or hexamethylenetetramine (HMT) is added as a precipitant, and the mixture is stirred. Distilled water is added to the mixed solution to adjust the total concentration of metal and silicon to

12

M. Kobayashi et al.

Fig. 11 Time course of the change in pH of solutions containing metal ions with (a) 4.5 M urea, (b) 1.0 M urea, (c) and 1.0 M HMT at 363 K (Yoshizawa et al. 2010). Photographs of a solution with metal ions and 4.5 M urea after 1, 2, 3, and 4 h heat treatment at 353 K are also displayed

0.1 M and that of urea to 4.5 M. The flask or beaker was capped or covered and then heated in an oil or water bath to bring the solution temperature to 363 K. The solution is held for 4 h and stirred at 400 rpm by a magnetic stirrer to obtain a precipitate. The pH of the solution should be raised quickly to the critical pH value at which cation precipitation starts forming by heating. Figure 11 shows the time dependence of solution pH using urea and PGMS and photographs of the solution (transferred to a test tube for clearer images). Within 1 h of heating the pH of the solution reached 6, and at 2 h a white precipitate was observed. The precipitate was separated by filtration, washed with distilled water, and dried to obtain a precursor. This precursor was annealed at various temperatures for 2 h to obtain the final sample. The precursor obtained with the HP method employing urea was amorphous. It remained amorphous after annealing at 973 K but became single-phase Zn2SiO4 when annealed at 1,073 K. That is, the Zn2SiO4 phase without ZnO or other phases was directly formed by low-temperature annealing of the amorphous precursor synthesized using the HP method. The direct crystallization of Zn2SiO4 phase at low temperatures from the amorphous precursor is likely due to the fact that forming a uniformly distributed Zn-O-Si-O network in the amorphous precursor does not require thermal diffusion of the constituent cations. Elemental analysis revealed that the composition of the samples was almost the same as that of the starting mixture. However, Cl was not detected in the precursor, indicating that the by-product NH4Cl during synthesis was thoroughly removed by washing. On the other hand, in the

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

13

sample synthesized by a conventional solid-state reaction method, a large amount of ZnO starting material remained after heating at 1,073 K. Although the phase purity gradually increased with calcination temperature, calcination at 1,373 K or higher was required to obtain single-phase Zn2SiO4. Using a polymerizable complex method, crystalline ZnO was confirmed to form in the precursor, and a singlephase sample was formed at 1,473 K (Takahashi et al. 2009). When using TEOS instead of the GMSs, the precursor consisted of Zn5(OH)6(CO3)2 and the sample annealed at 973 K contained crystalline ZnO. Even after annealing at 1,373 K, ZnO remained resulting in the formation of the mixed phase of Zn2SiO4 and ZnO. The insoluble TEOS caused emulsions to form at an early stage of the reaction. Since the hydrolysis of TEOS progresses only at the TEOS-water interface, it was difficult to obtain a precipitate in which silicon and zinc were complexed homogeneously at the atomic level. Consequently, ZnO segregated and became a mixed phase with Zn2SiO4. These facts indicate that the synthesis using TEOS is essentially an inhomogeneous process, in which the constituent elements do not mix well at the atomic level. Thus, GMS is applicable for synthesizing silicate materials using the HP method. A quite different result was obtained with the HP method when HMT instead of urea was used as a precipitant. In this synthesis, the recovery yield of the precursor after reaction for 4 h was low, ca. 50 % while it achieved 90–95 % in the synthesis using urea. Even after annealing at 1,073 K, this precursor remained amorphous, and no crystalline Zn2SiO4 phase was formed. The pH of the mixed solution after reaction was below 6. At this pH, zinc may not precipitate because the zinc ions are not significantly hydrolyzed. When the precipitation period was extended to 24 h, the pH increased to 6.6, and the resulting precursor showed a weak diffraction peak of ZnO. When this sample was annealed at 1,073 K, a mixed phase of Zn2SiO4 and ZnO was obtained. In other words, single-phase Zn2SiO4 could not be obtained by the HP method using HMT. Therefore, urea is a more suitable precipitant in the HP method using GMS. As shown in Fig. 11, when using 4.5 M urea, the pH of the solution increased within 15 min from 2.5 to a critical pH of ca. 6. At higher pH values, zinc was precipitated due to its reduced solubility (Fig. 10), and the hydrolysis of GMS progressed quickly (Fig. 4). In contrast, when 1.0 M HMT was used although the pH increased to 5 within 15 min, the subsequent rate of pH increase was slow compared with the rate when urea was used, and the pH was only 5.8 after 4 h. The slow increase in pH and the lower attained pH with HMT were caused by the characteristics of HMT rather than the precipitant concentration, because when 1.0 M urea was used the pH rose to 6 within 15 min upon heating and reached 7.8 after 4 h. When HMT was used, although the pH reached 5 at the early stage of heating and the hydrolysis of GMS took place, the homogeneity of the precipitate was poor since the hydrolysis of zinc occurred later in the reaction. As a result, ZnO was formed in the HP method using HMT. In summary, in the HP method using GMS, urea is a suitable precipitant because it can quickly raise the pH and allows the simultaneous hydrolysis of GMS and zinc while the reaction solution is heated. According to the dissolution curves of metal ions (Kragten 1978), the solubilities of yttrium and cerium ions in aqueous solutions change dramatically at pH 6–7.

14

M. Kobayashi et al.

Therefore, Ce3+-doped Y2SiO5 with high-phase purity can be obtained at low temperature using the HP method employing GMSs. Following the synthesis flowchart described in Fig. 10, Y2-XCeXSiO5 was synthesized using Y(NO3)3 and Ce (NO3)3 in stoichiometric ratio, instead of Zn(NO3)2 and Mn(NO3)2. When the total cation concentration was set to 0.1 M, the precipitate was composed of Y2(CO3)3(H2O)2 (PDF No. 01–081–1538). After annealing at 973 K for 2 h, crystalline Y2O3 and an amorphous phase were formed. Even after annealing the precursor at above 1,673 K, Y2O3 and Y4.67(SiO4)3O (PDF No. 00–030–1457) were formed as secondary and tertiary phases, respectively. The fact that yttria was formed in the precursor might indicate that segregation of yttrium occurred during the precipitation stage. That is, the hydrolysis of each cation proceeded separately. According to the dissolution curves (Kragten 1978), yttrium precipitated at approximately pH 6.9 in an aqueous solution when the total metal concentration was 0.1 M ([Y3+] = 0.067 M) as in the present case. The precipitation caused by hydrolysis proceeds at a lower pH when using a higher [Y3+]. Thus, a precursor with the constituents homogeneously dispersed at the atomic level could be obtained, using a solution with high metal concentration. For example, the powder sample prepared using a solution with 0.2 M total metal was almost amorphous after annealing at 973 K, even though the precursor was assigned to Y2(CO3)3(H2O)2, which was observed in the sample obtained with lower metal concentration. As a result, singlephase Y2SiO5-X1 (the low-temperature phase of Y2SiO5) could be obtained at 1,273 K, and the high-temperature phase Y2SiO5-X2 was obtained at 1,873 K, while impurity phases such as Y2O3 were formed using a solution with low metal concentration even after the annealing above 1,673 K. Therefore, simultaneous hydrolysis of the constituent ions is necessary for the present method, and under the proper conditions high-quality silicates can be obtained using GMSs.

Preparation of Monodispersed Silica Using GMS The method developed by Yokoi et al. (2009) can be used to prepare monodispersed silica using GMS. The procedure is shown in Fig. 12. An aqueous solution of GMS is added to a cooled ammonium solution and stirred at 333 K for 48 h to cause Fig. 12 A flowchart of the synthesis of monodispersed silica nanoparticles using GMSs

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

15

nucleation of silica. L(+)-lysine is added, then the solution is stirred at 353–423 K for 24 h to obtain a silica sol. Figure 13a, b shows the transmission electron microscopy (TEM) image and size distribution histogram of the silica particles obtained after aging at 353 K. Monodispersed silica particles approximately 10 nm in diameter were observed. The zeta potential of the dispersion was
39 mV, which was similar to the value reported by Yokoi et al. (2009). With a higher concentration of ammonia, the size of obtained silica particles became larger. The solubility of silica increased with increasing basicity (Iler 1979). Therefore, the growth of silica particles proceeds by the dissolution and reprecipitation processes. Interestingly, beaded silica nanoparticles could be obtained using aqueous solutions of PGMS with high concentration of silicon (Fig. 13c). Such strings of silica particles usually are formed with templates such as polymers (Zhou et al. 2015); however, here, the one-dimensional architecture of silica was formed using GMS and without any template. With higher silicon concentration in the aqueous solution of GMS, the

Fig. 13 (a) Size distribution histogram and (b, c) TEM images of colloidal silica synthesized using PGMS. (a, b) [NH3] = 1.6 M, [PGMS] = 10 mM, 353 K, 24 h, (c) [NH3] = 0.21 M, [PGMS] = 25 mM, 393 K, 24 h

16

M. Kobayashi et al.

zeta potential increased, leading to increased collision frequency among silica particles, which promoted the observed agglomeration of silica.

Water-Soluble Phosphates Problems of Conventional Phosphorous Raw Materials for Aqueous Solution-Based Sol-Gel Methods The lack of proper phosphorus sources is a critical limitation in the synthesis of multicomponent phosphates using aqueous solution-based methods. The conventional phosphorous sources H3PO4 and NH4H2PO4 are highly reactive with many cations; therefore they easily form undesirable precipitates in an aqueous solution containing other cations, as shown in Table 2 (Weng et al. 1999). Consequently, the final products show low performance owing to the low homogeneity and the presence of impurity phases. Previous reports have demonstrated that such undesirable precipitates could be avoided by using condensed chain-structured phosphates, resulting in highly stable water-soluble complexes between the metal and phosphate ions (Wazer and Callis 1958; Rashchi and Finch 2000). Pretula et al. described a direct synthesis process of oligo-poly(ethylene phosphate) from ethylene glycol (EG) and H3PO4 (Pretula et al. 2008). The phosphate oligomer was found to be soluble and stable in aqueous solution, owing to its condensed chain structure.

Preparation of an Ethylene Glycol-Conjugated Phosphorous Ester A stable water-soluble phosphate precursor can be synthesized using a reflux process by following the procedure shown in Fig. 14 (Pretula et al. 2008; Kim 2013a). For example, 0.05 mol H3PO4 and 0.08 mol EG are mixed in heptane under azeotropic Table 2 Reactivity between phosphorous sources and metals in water (P precipitation, N negligible precipitation, and  no precipitation)

Element Li Na K Rb Mg Ca Sr Ba Y La Ce Eu Fe

NH4H2PO4 P    P P P P P P P P P

H3PO4 P    P P P P P P P P 

EG-P P    P P N N   P  

PEG-P     P      P  

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

17

Fig. 14 A flowchart of the reflux process for the synthesis of ethylene glycolconjugated phosphate esters

Fig. 15 31P{1H} NMR spectra of EG-Ps prepared by reflux method (P H3PO4, M monoester, D diester, T triester)

distillation condition with vigorous stirring under a N2 atmosphere. A Dean-Stark water trap is used to remove the formed H2O through condensation. The reaction is carried out at various temperatures (383–403 K) for different periods of reflux (12–30 h). After reflux, the obtained gel-like matter is dried in vacuum at 353 K to remove residual heptane and H2O. Figure 15 shows the 31P{1H} NMR spectra of phosphate precursors prepared by reflux at 393 K for 24 h and at 393 K for various times. Prior to the NMR measurement, the formed gels were dissolved in D2O and the pH was adjusted to 13 using 10 M NaOH to determine the chemical shift standard. The signals observed at 5.25, 4.20, 0.81, and 0.56 ppm were attributable to H3PO4, monoester, diester, and triester, respectively (Fig. 16). The numbers of n and m were found to be 2–4 (Pretula et al. 2008; Kim et al. 2013a). Table 3 summarizes the proportion of H3PO4 and each

18

M. Kobayashi et al.

Fig. 16 The chemical structures of the synthesized EG-P: (a) monoester, (b) diester, (c) and triester

Table 3 The proportion of each EG-P ester and its solubility, depending on reflux conditions Reaction conditions Temp Time 383 K 24 h 393 K 18 h 393 K 24 h 393 K 30 h 403 K 24 h

Proportion in product (%) P M 84.2 10.0 53.9 28.6 40.3 26.8 36.0 21.5 20.5 33.9

D, T 5.8 17.5 32.9 42.5 45.6

Solubility in water Good Good Good Poor Poor

P H3PO4, M monoester, D diester, T triester

ethylene glycol-conjugated phosphorous (EG-P) ester in the samples obtained under different conditions. At 383 K, H3PO4 was found to be the main product with small amounts of monoester and diester, indicating that this temperature is not sufficient to drive the esterification reaction between H3PO4 and EG. When the reflux temperature was raised to 393 K, the amount of ester products significantly increased. Mainly, monoester formation was observed in addition to small amounts of di- and tri-esters. The proportion of the di- and tri-esters increased when the temperature was further increased to 403 K. Thus, it is clear that a temperature above 393 K is required for this esterification reaction. At the fixed temperature of 393 K, the proportions of the different EG-Ps were found to be dependent on the reaction time, with the formation of di- and tri-esters increasing at longer reflux times. The samples prepared at 393 K for 30 h and at 403 K for 24 h contained matters insoluble in water. These insoluble by-products were formed via the decomposition or selfcondensation of EG under the severe reflux conditions of high temperature or long reaction time (Jensen and Neese 1972).

Preparation of a Polyethylene Glycol-Conjugated Phosphorous Ester EG-P has good solubility and stability in aqueous condition, although approximately 50 % H3PO4 remains unreacted. The preparation of a phosphorus source

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

19

Fig. 17 A flowchart of the reflux process for the synthesis of polyethylene glycol-conjugated phosphate esters

with less residual H3PO4 would greatly expand the range of available phosphates with better properties. The condensed chain-structured phosphates are commercially available and can be synthesized in a rather simple reaction using alcohol, phosphorus pentoxide, and polyphosphoric acid (Tracy and Reierson 2002). PEG in particular is considered a good candidate for condensation with the phosphate moiety, because it is widely used as a cross-linking agent to promote the formation of gel in a variety of solution-based sol-gel methods. Actually, polyethylene glycol-conjugated phosphate (PEG-P) ester is soluble and stable in an aqueous solution as discussed later. Figure 17 shows a flowchart of the typical synthesis of PEG-P. Pyrophosphoric acid and polyethylene glycol 300 (PEG300) with a molar ratio of 1:1 is mixed at 323 K and kept for 2 h. Equimolar phosphorus pentoxide P4O10 to pyrophosphoric acid is weighed in an inert atmosphere to prevent its hydrolysis and then slowly added into the mixture. If the addition of P4O10 is not performed slowly, a black precipitate, which is not soluble for the following procedure, is formed. After P4O10 was dispersed homogeneously, the mixture is heated to 358 K and kept for 5 h. Finally, additional equimolar PEG300 to PEG300 added at first is added, and the reaction was continued at 358 K for 12 h. The entire reaction was conducted in N2 atmosphere. Gel with high viscosity was formed and it was dissolved in distilled water. The scheme of the reaction is illustrated in Fig. 18. Figures 19 and 20 show the 31P{1H} NMR spectra of the obtained PEG-P and the estimated chemical structures of the synthesized PEG-conjugated monoester and diester, respectively. The 1H NMR spectrum indicates that only one side OH group of PEG was conjugated with the phosphate. It is estimated that more than 90 % PEG300 was bonded with phosphate (Kim et al. 2013b). The 31P{1H} NMR spectrum also confirms that the product contained H3PO4, monoester, and diester with a ratio of 17.4:72.5:10.1. Considering that the yield of esters using EG was about 55.4 % (44.6 % unreacted H3PO4), the method using PEG can produce the condensed chain-structured phosphate with less residual H3PO4 compared with that mentioned in the previous section. This phosphorous source with a large proportion of esters and a small amount of H3PO4 is considered suitable for solution-based synthesis of phosphate-based phosphors, because there is less H3PO4 to interact with metal ions in the solution to form precipitates.

20

M. Kobayashi et al.

Fig. 18 The overall reaction sequence for the preparation of polyethylene glycolconjugated phosphate ester

Fig. 19

31

P{1H} NMR spectra of PEG-P in D2O at pH 13

Stability and Reactivity of EG-P and PEG-P in Aqueous Conditions Table 4 summarizes the ratios of H3PO4 and esters (EG-P and PEG-P) in aqueous solutions calculated from the 31P{1H} NMR spectra, after keeping the solutions for 0, 2, and 8 weeks at room temperature. The stability and hydrolysis behaviors of each solution were evaluated by analyzing the content ratios at different holding times, as the esters would be hydrolyzed into H3PO4 with time. In the case of EG-P, the di- and triesters were converted into monoesters in the aqueous solution by hydrolysis, as noted by the increase in the monoester proportion from 26.8 % to

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

21

Fig. 20 Estimated chemical structures of the synthesized PEG-P: (a) monoester, (b) diester

Table 4 Proportion of H3PO4 and esters in the as-prepared EG-P and PEG-P and in aqueous solutions of EG-P and PEG-P kept for 2 and 8 weeks at room temperature Time/week 0 2 8

EG-P P 40.6 45.2 52.2

M 26.8 29.3 37.1

D, T 32.6 25.5 10.7

PEG-P P 17.4 16.9 15.4

M 72.5 71.3 71.8

D 10.1 11.8 12.8

P H3PO4, M monoester, D diester, T triester

29.3 % between 0 and 2 weeks. The monoester was hydrolyzed to H3PO4 at the same time. Over 2 weeks, the proportion of H3PO4 to esters increased from 40.6 % to 45.2 %. Hydrolysis was observed to continue with time, further increasing the proportion of H3PO4 to esters (52.2 %) after 8 weeks. On the other hand, the amounts of H3PO4, monoester, and diester in the aqueous solution of PEG-P did not change significantly even after 8 weeks. Therefore, PEG-P is much more stable in aqueous conditions than EG-P. It has been reported that the degradation of such esters is strongly dependent on molecular weight (Roberts et al. 2002). For example, an oligomer with low molecular weight is easily degraded, while a polymer with high molecular weight shows good stability over longer periods. Therefore, EG-P shows more pronounced hydrolysis behavior due to its oligomeric structure, while PEG-P with a polymeric structure has greater stability. The construction of an “element library” by compiling the reactivity between phosphorous (P) sources and cations is important for identifying potential P sources for functional materials. Table 3 shows the results when P sources such as NH4H2PO4, H3PO4, EG-P, and PEG-P are combined with various cations in aqueous conditions at 323 K for 1 week. There are no precipitates with Na, K, and Rb salts irrespective of the P source, while white precipitates are observed in solutions containing Mg or Ce. NH4H2PO4 and H3PO4 show rapid precipitation with almost all the metals here. A precipitate between Fe and NH4H2PO4 is observed, while other P sources are stable and do not react with Fe. When EG-P is reacted with Li, Ca, Sr, and Ba, the precipitates formed with Sr or Ba are negligible. Therefore, EG-P could be used as a P source in the synthesis of phosphate compounds without Li, Mg, Ca, Y, La, or Ce. On the other hand, PEG-P forms no precipitate with most of the cations tested. It is speculated that the proportions of H3PO4 and the hydrolysis behaviors of the P sources affect their stabilities in solutions containing various

22

M. Kobayashi et al.

Fig. 21 A flowchart of the synthesis of Eu2+-doped KSrPO4 by a polymerizable complex method

cations. Although PEG-P precipitates with Mg and Ce, it is considered a desirable P source for aqueous solution-based sol-gel-based methods.

Synthesis of Phosphates by a Polymerizable Complex Method Employing EG-P and PEG-P Considering Table 3, the performance of prepared phosphates including Li, Ca, Sr, Ba, Y, La, Eu, and Fe may be improved by using EG-P or PEG-P. Figure 21 shows a typical preparation procedure of Eu2+-doped KSrPO4 as a model compound by the polymerizable complex method. EG-P or PEG-P, KNO3, Sr(NO3)2, and Eu(NO3)3 at a ratio of P:K:Sr:Eu = 1:1:0.97:0.03 were dissolved in an aqueous solution of citric acid. The solution was first heated at 353 K to allow chelation, then propylene glycol was added to achieve a molar ratio of (K, Sr, Eu):citric acid:glycol = 1:4:4. The temperature was subsequently increased to 423 K to facilitate gel formation. When using conventional phosphorous sources such as H3PO4 and NH4H2PO4 instead of EG-P and PEG-P, the solution became turbid. The resulting gel was heated at 1,073 K in air to remove the organic content, and then reduction was conducted by heating the samples at 1,173 K under a flow of Ar containing 4 % H2 for 3 h. Figure 22 shows the X-ray diffraction (XRD) patterns of the 3 mol% Eu2+-doped KSrPO4 phosphors synthesized separately using EG-P, PEG-P, NH4H2PO4, and H3PO4. The samples prepared via the PC method using EG-P or PEG-P had no impurity phases, whereas those synthesized using H3PO4 or NH4H2PO4 contained

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

23

Fig. 22 XRD patterns of Eu2+-doped KSrPO4 synthesized by the polymerizable complex method using (a) NH4H2PO4, (b) H3PO4, (c) EG-P, (d) and PEG-P

K2SrP2O7 or Sr3(PO4)2 impurities. These results are in agreement with the earlier experiments that show a precipitate would form when H3PO4 or NH4H2PO4 is added to an aqueous solution containing K, Sr, and Eu salts, whereas the use of EG-P or PEG-P maintains a transparent homogeneous solution (Table 2). Note that single phase of the sample requires 1,373 K calcination and cannot be obtained at 1,173 K by a solid-state reaction method (Kim et al. 2014). The photoluminescence properties of the Eu-doped KSrPO4 samples prepared using EG-P and PEG-P are almost the same, while those synthesized using NH4H2PO4 or H3PO4 by the solid-state reaction method showed lower intensity. The reason is that EG-P or PEG-P enabled the synthesis of phosphors with a stoichiometric composition throughout the material (Kim et al. 2013a, 2014). These results demonstrate that the polymerizable complex method in conjunction with EG-P and PEG-P is a suitable process for multicomponent phosphate synthesis, owing to the superior phase purity and the significantly lower temperature required. Figure 23 shows the XRD patterns of 3 mol% Eu2+-doped LiCaPO4 synthesized by the polymerizable complex method employing PEG-P, EG-P, NH4H2PO4, or H3PO4. LiCaPO4 with precise stoichiometry is difficult to prepare by solution-based methods, because of the high reactivity of Li and Ca with P sources in aqueous conditions. Nevertheless, the sample synthesized by the PC method employing PEG-P consisted of almost single-phase LiCaPO4. In contrast, impurities such as Li3PO4, Ca3(PO4)2, and Ca4P2O9 were formed when NH4H2PO4, H3PO4, or EG-P was used. As indicated by Table 2, Li and Ca formed precipitates with NH4H2PO4, H3PO4, and EG-P in aqueous solutions. Indeed, during the synthesis, white

24

M. Kobayashi et al.

Fig. 23 XRD patterns of Eu2+-doped LiCaPO4 synthesized by the polymerizable complex method using (a) NH4H2PO4, (b) H3PO4, (c) EG-P, (d) and PEG-P

precipitates were observed when using NH4H2PO4, H3PO4, and EG-P, while no precipitates were observed using PEG-P. The undesirable precipitates consist of agglomerations of specific elements, and they prevent the homogeneous distribution of the constituents at the atomic level in the final product. The distribution of a tiny amount of rare-earth ions is directly related to the performance of the phosphor material and therefore very important. Here only PEG-P allowed us to synthesize single-phase LiCaPO4, and the prepared sample exhibited the highest emission intensity (Kim et al. 2013b, 2014). Thus, the compiled “element library” is quite useful for choosing a proper P source for the synthesis of phosphates.

Development of New Phosphates Using EG-P and PEG-P The exploration of novel compounds is necessary for developing high-functional materials to meet the demands of modern technology. The precise characterization of these new compounds then leads to additional knowledge of material chemistry and improved material design. Over the years, a variety of approaches, such as the combinatorial method, has been applied to finding new compounds (Potyrailo et al. 2011). In spite of the large number of compounds found and characterized, new useful compounds continue to be reported in recent years (Hirosaki et al. 2014; Pust et al. 2014a, b; Fujii et al. 2014; Yamada et al. 2015; Kanno et al. 2016). This

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

25

suggests that many potentially revolutionary materials with high and unique functions remain to be uncovered. Reliable and compositionally controllable synthesis methods allow us to discover new compounds. Solution-based synthesis has been regarded as a desirable approach to obtain materials with compositions controlled at the atomic level (Hasegawa et al. 2013). As discussed above, the application of water-soluble and stable phosphorus sources EG-P and PEG-P has led to the synthesis of phosphors with high constitutional homogeneity. As a result, two novel compounds, Na3Ba2Ca(PO4)3 and NaBa0.5Ca0.5PO4 were prepared using a polymerizable complex method employing PEG-P (Kim et al. 2015). Although these compounds appear to be solid solutions between NaBaPO4 and NaCaPO4 in formula, they have different coordination environments and PO4 orderings. The reason why these compounds had not been identified earlier is the difference between the crystal systems of NaBaPO4 (trigonal) and NaCaPO4 (orthorhombic) caused by the large difference between the ionic radii of Ba2+ and Ca2+. The mismatching crystal systems do not allow us to obtain the solid solutions and make it difficult to prepare these compounds without optimal raw material. Thus, new raw materials directly contribute to the discovery of novel compounds.

Conclusions Even though conventional sol-gel methods using nonaqueous solutions and/or alkoxides have produced high-performance materials with controlled structures, alternative methods based on aqueous solutions are highly desirable due to the growing interest in green chemistry. Up to several years ago, it was believed that “aqueous solution processes are environmentally friendly, but cannot produce materials with excellent properties.” Now, we know that aqueous solution processes are not only environmentally friendly but also produce materials with excellent properties. A large number of materials fabricated by biomimetic and bioinspired processes have been reported (Hellmich and Katti 2015). Some of these processes can produce materials with amazing properties under ambient conditions. This chapter has summarized the preparation methods, characteristics, and applications of silicon and phosphorus raw materials for aqueous solution-based material synthesis. The development of new raw materials leads to new synthesis methods, resulting in new materials with higher performance.

References Fleming JL. Sol
gel processing of glass powder. In: Messing GL, Mazdiyasni KS, McCauley JW, Haber RA, editors. Advanced in ceramics, vol. 21: ceramic powder science ceramic powder science. Westerville: The American Ceramic Society; 1987. p. 155–61. Fujii K, Esaki Y, Omoto K, Yashima M, Hoshikawa A, Ishigaki T, Hester JR. New Perovskiterelated structure family of oxide-ion conducting materials NdBaInO4. Chem Mater. 2014;26:2488–91.

26

M. Kobayashi et al.

Hasegawa I, Kuroda K, Kato C. The effect of tetraalkylammonium ions on the distribution of the silicate anions in aqueous solutions. Bull Chem Soc Jpn. 1986;59:2279–83. Hasegawa T, Kato H, Yakatsuka Y, Kobayashi M, Yamane H, Kakihana M. Orange emission from new (Ba1-XSrX)4Al2S7:Eu2+ thioaluminate phosphors with visible light excitation. ECS J Solid State Sci Technol. 2013;2:R3107–11. Hellmich C, Katti D. Multiscale mechanics of biological, bioinspired, and biomedical materials. MRS Bull. 2015;40:309–13. Henry M, Jolivet JP, Livage J. Aqueous chemistry of metal cations: hydrolysis, condensation and complexation. In: Reisfeld R, JJørgensen CK, editors. Chemistry, spectroscopy and applications of sol–gel glasses. Berlin: Springer; 1992. p. 153–206. Horosaki N, Takeda T, Funahashi S, Xie RJ. Discovery of new nitridosilicate phosphors for solid state lighting by the single-particle-diagnosis approach. Chem Mater. 2014;26:4280–8. Iler RK. The chemistry of silica. New York: Wiley; 1979. Jensen FR, Neese RA. Preparation of trans–2,3–trans–5,6–dioxane–d4. Use of a hindered base to prevent acid–catalyzed side reactions. J Org Chem. 1972;37:3037–8. Kakihana M, Kobayashi M, Tomita K, Petrykin V. Application of water-soluble titanium complexes as precursors for synthesis of titanium-containing oxides via aqueous solution processes. Bull Chem Soc Jpn. 2010;83:1285–308. Kanno M, Yamada T, Yamane H, Nagai H. Synthesis, crystal structure, and thermoelectric properties of Na2+xAl2+xSn4–x (x = 0.38–0.24). Chem Mater. 2016;28:601–7. Kim M, Kobayashi M, Kato H, Kakihana M. Enhancement of luminescence properties of a KSrPO4:Eu2+ phosphor prepared using a solution method with a water–soluble phosphate oligomer. J Mater Chem C. 2013a;1:5741–6. Kim M, Kobayashi M, Kato H, Kakihana M. A highly luminous LiCaPO4:Eu2+ phosphor synthesized by a solution method employing a water-soluble phosphate ester. Opt Photonics J. 2013b;3:13–8. Kim M, Kobayashi M, Kato H, Kakihana M. The significance of phosphate source in the preparation of functional luminescent phosphate materials. J Ceram Soc Jpn. 2014;122:626–9. Kim M, Kobayashi M, Kato H, Yamane H, Sato Y, Kakihana M. Crystal structures and luminescence properties of Eu2+-activated new NaBa0.5Ca0.5PO4 and Na3Ba2Ca(PO4)3. Dalton Trans. 2015;44:1900–4. Klein LC. Sol–gel processing of silicates. Annu Rev Mater Sci. 1985;15:227–48. Komukai T, Takatsuka Y, Tezuka S, Kato H, Kakihana M. Luminescence properties of BaZrSi3O9: Eu synthesized by an aqueous solution method. J Lumin. 2015;158:328–32. Kragten J. Atlas of metal–ligand equilibria in aqueous solution. Sussex: Ellis Horwood Limited; 1978. Livage J, Henry M, Sanchez C. Sol–gel chemistry of transition metal oxides. Prog Solid State Chem. 1988;18:259–341. Matijevic E. Uniform inorganic colloid dispersions. Achievements and challenges. Langmuir. 1994;10:8–16. Mehrotra RC, Narain RP. Reactions of tetramethoxy– & triethoxysilanes with glycols. Indian J Chem. 1967;5:444–7. Mehrotra RC, Narain RP. Reactions of tetramethoxy– & triethoxysilanes with α–hydroxyl carboxylic & salicylic acids. Indian J Chem. 1968;6:110–1. Potyrailo R, Rajan K, Stoewe K, TakeuchiI, Chisholm B, Lam H. Combinatorial and highthroughput screening of materials libraries: review of state of the art. ACS Comb Sci. 2011;13:579–633. Pretula J, Kaluzynski K, Wisniewski B, Szymanski R, Loontjens T, Penczek S. Formation of poly (ethylene phosphates) in polycondensation of H3PO4 with ethylene glycol. Kinetic and mechanistic study. J Polym Sci Part A Polym Chem. 2008;46:830–43. Pust P, Weiler V, Hecht C, T€ ucks A, Wochnik AS, Henß AK, Wiechert D, Scheu C, Schmidtz PJ, Schnick W. Narrow-band red-emitting Sr[LiAl3N4]:Eu2+ as a next-generation LED–phosphor material. Nat Mater. 2014a;13:891–6.

Water–Dispersed Silicates and Water–Soluble Phosphates, and Their Use in. . .

27

Pust P, Wochnik AS, Baumann E, Schmidt PJ, Wiechert D, Scheu C, Schnick W. Ca[LiAl3N4]: Eu2+—a narrow-band red-emitting nitridolithoaluminate. Chem Mater. 2014b;26:3544–9. Rashchi F, Finch JA. Polyphosphates: a review their chemistry and application with particular reference to mineral processing. Miner Eng. 2000;13:1019–35. Roberts MJ, Bentley MD, Harris JM. Chemistry for peptide and protein PEGylation. Adv Drug Deliv Rev. 2002;54:459–76. Sato Y, Kato H, Kobayashi M, Kim JH, Kakihana M. Syntheses of silicate phosphors by aqueous solution techniques using water–dispersible inorganic Si cluster. J Jpn Soc Powder Metall. 2015;62:127–33. Suzuki Y, Kakihana M. Preparation of water soluble silicon compound and its application for synthesis of (Y, Ce, Gd)2SiO5 blue emission phosphor. J Ceram Soc Jpn. 2009;117:330–4. Takahashi N, Suzuki Y, Kakihana M. Synthesis of Zn2SiO4:Mn2+ green emission phosphor by hydrothermal gelation method using a novel water soluble silicon compound. J Ceram Soc Jpn. 2009;117:313–5. Tracy DJ, Reierson RL. Commercial synthesis of monoalkyl phosphates. J Surfactant Deterg. 2002;5:169–72. Velikov KP, Blaaderen AV. Synthesis and characterization of monodisperse core
shell colloidal spheres of zinc sulfide and silica. Langmuir. 2001;17:4779–86. Wazer JRV, Callis CF. Metal complexing by phosphates. Chem Rev. 1958;58:1011–46. Weng W, Huang J, Han G. The alkoxide sol–gel process in the calcium phosphate system and its applications. Appl Organomet Chem. 1999;13:555–64. Yamada T, Yamane H, Nagai H. A thermoelectric zintl phase Na2+xGa2+xSn4–x with disordered Na atoms in helical tunnels. Adv Mater. 2015;27:4708–13. Yamaguchi T, Suzuki Y, Kakihana M. Low temperature synthesis of (Ca, Ce)3Sc2Si3O12 phosphor by hydrothermal gelation method using novel water soluble silicon compound. J Jpn Soc Powder Metall. 2010;57:706–10. Yanagisawa R, Petrykin V, Kakihana M. Synthesis of Ba3Ta6Si4O26 using aqueous solution processes and its photocatalytic activity. J Jpn Soc Powder Metall. 2010;57:701–5. Yokoi T, Wakabayashi J, Otsuka Y, Fan W, Iwama M, Watanabe R, Aramaki K, Shimojima A, Tatsumi T, Okubo T. Mechanism of formation of uniform–sized silica nanospheres catalyzed by basic amino acids. Chem Mater. 2009;21:3719–29. Yoshizawa K, Kato H, Kakihana M. Synthesis of Zn2SiO4:Mn2+ by homogeneous precipitation using propylene glycol–modified silane. J Mater Chem. 2010;22:17272–7. Zhou S, Oda Y, Shimojima A, Okubo T, Aoshima S, Sugawara–Narutaki A. Ring assembly of silica nanospheres mediated by amphiphilic block copolymers with oxyethylene moieties. Polym J. 2015;47:128–35.

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and Intrusion René Pirard, Christelle Alié, and Jean-Paul Pirard

Abstract

This chapter is dedicated to the characterization of the porosity and mechanical behavior of the aerogels and xerogels. The main method of study involved is mercury porosimetry. Mercury porosimetry is a method used to characterize the texture of porous materials. It enables determining pore volume, specific surface area, and also distributions of pore volume and surface area versus pore size. It can be applied to powders, as well as to monolithic porous materials. The basic hypothesis usually accepted is that mercury penetrates into narrower and narrower cavities or pores as pressure increases.

Contents Texture of Porous Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Shrinkage of Porous Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Global Mechanical Behavior of Sol–Gel Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Textural Representation of Aerogel and Aerogel-Like Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Reversibility of Shrinkage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Collapse of Largest Pores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Relation Between Pore Size and Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Two Successive Mechanisms: Buckling and Intrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Relation Between the Buckling Constant Kf and the Aggregate Size . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

R. Pirard (*) Laboratoire de Génie Chimique, Institutde Chimie B6a, University of Liège, Liège, Belgium e-mail: [email protected] C. Alié Department of Chemical Engineering, School of Engineering, University of Liège, Liège, Belgium e-mail: [email protected] J.-P. Pirard Department of Applied Chemistry, School of Engineering, University of Liège, Liège, Belgium e-mail: [email protected] # Springer International Publishing AG 2017 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_136-1

1

2

R. Pirard et al.

Membrane Test to Avoid Intrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Determination of the Buckling Constant Kf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pore Size Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relation Between Global Mechanical Behavior and Textural Representation . . . . . . . . . . . . . . . . . . Accuracy of the Buckling Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples for Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15 16 17 18 20 21 24 26

Texture of Porous Materials The characterization of a complex porous texture of materials is a difficult task, because no method exists that gives a direct and complete measurement. In most characterization methods, it is not the pore size that is directly measured, but rather a physical variable influenced by the porous texture. It is necessary to analyze the results using a model based on hypotheses which are generally not fully satisfied by the material behavior. So, it is preferable to use a set of methods to obtain a good picture of the texture. There are few experimental techniques leading to pore volume and specific surface distribution as a function of pore size. The most commonly used technique is based on nitrogen adsorption–desorption isotherms at 77 K. A coherent set of methods has been proposed to analyze the data and to obtain the pore size distribution (Lecloux 1981). Nitrogen adsorption gives information on the material’s microporosity and mesoporosity. Mesoporosity is the pore size range in which capillary condensation takes place; it is analyzed using the Kelvin equation, which depends on a pore shape factor. The results differ slightly as a function of the hypothesis on the pore shape that is chosen, and, in all cases, the distribution is only obtained with good precision for pore sizes under 25 nm. Thermoporometry is a technique based on the thermodynamic study of the freezing of liquids enclosed in the material porosity (Quinson and Brun 1988). The precise calorimetric measurement of the freezing or melting enthalpy as a function of temperature also leads to volume distribution as a function of pore size, with a satisfactory precision for pore sizes under 25 nm. X-ray scattering (SAXS) and neutron scattering (SANS) at small angles is increasingly used to obtain information on particle and aggregate sizes and their hierarchical arrangement in the size range between 0.1 and 100 nm (Renouprez 1970). It is, however, difficult to infer without ambiguity a pore size distribution from SAXS data. Mercury porosimetry thus appears as necessary, to complete the data in the range of large mesopores and macropores. Mercury porosimetry is a method currently used to characterize the texture of porous materials. It enables determining pore volume, specific surface area, and also distributions of pore volume and surface area versus pore size. It can be applied to powders, as well as to monolithic porous materials. The basic hypothesis usually accepted is that mercury penetrates into narrower and narrower cavities or pores as pressure increases. Data analysis is performed using the intrusion equation proposed by Washburn (1921):

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

r¼

2γ cos θ P

3

(1)

in which r is the cylindrical pore radius, P, the pressure at which mercury enters into the pore, γ, the surface tension of mercury (0.485 N/m) and θ, the contact angle between the mercury meniscus and a flat surface of the material under investigation (θ  140 ). In early applications of the method, it appeared that the materials, porous or not, were elastically compressed at high pressures. To obtain accurate results, the volume variation due to compression must be subtracted from the rough data, but a method to do so has not yet been well established. Fortunately, in the case of porous materials, the volume variation due to elastic compression is often negligible with regard to the volume variation due to mercury intrusion in the empty pores. Therefore, mercury porosimetry data are generally handled without any correction. In the case of less porous materials like carbons, Friesen and Ogunsola (1995) point out that the volume variation by elastic compression at high pressure is an important part of the total pore volume, but the authors do not suggest any method to take into account this contribution.

Shrinkage of Porous Materials The compression phenomenon of some porous materials under mercury isostatic pressure was already shown by Brown and Lard (1974). By examining samples of porous silica, the authors experimentally noted that the sample compression could be so important that the intrusion becomes negligible in some pressure ranges. This observation and its correct analysis have long been ignored, even when the compression phenomenon is significant; as a consequence the total volume variation during mercury porosimetry is attributed to intrusion. Some publications and patents (Micoulin and Chevallier 2001) continue to use the Washburn Eq. 1 for data analysis, while this equation is not applicable in the case of material compression. The pore size distribution obtained in this way is obviously erroneous. However, in most research on porous silica (Fadeev et al. 1996) or carbon blacks (Milburn et al. 1988), mercury porosimetry data are not analyzed to obtain a pore size distribution, if a volume variation due to compression is suspected. During the first attempts of texture characterization of silica aerogels by mercury porosimetry, the phenomenon of compression by mercury isostatic pressure was systematically observed and its importance relative to intrusion has been recognized. Broecker et al. (1986) assert that it is necessary to consider the compression of the aerogel pellets by the mercury pressure. At low mercury pressures, only compression and no penetration occur. At medium pressures, with part of the porous pellet being penetrated, the degree of compression decreases. At the highest pressures, the penetration is completed (except for the pores in the low-nm range) and compression ceases. The quantitative evaluation of this process is very complicated, as Young’s or compressive modulus change with the fraction of penetrated porous volume over about four orders of

4

R. Pirard et al.

magnitude. Because the authors cannot clearly distinguish between the compression and intrusion contributions, they present the data as an “apparent relative intrusion volume” and an “apparent relative surface” as a function of “apparent pore radius.” This means that the data are nevertheless analyzed using Washburn’s equation. Broecker et al. thus consider that the main phenomenon is mercury intrusion and that compression only alters the phenomenon in its purity. They conclude as follows: “It will be an important task for the future to separate the two effects, compression and penetration, and so derive the true pore size distribution.” It seems that the distinction can be made because the phenomena of mercury intrusion in the pore network and material compression are mutually exclusive (Minihan et al. 1994) and some materials among the most porous are only compressed (Vittoratos and Auburn 1995). Materials synthesized by particle aggregation in vapor phase, like carbon blacks and fumed silica, have a low bulk density and high pore volume, like aerogels. They also show the compression phenomenon under mercury isostatic pressure (Smith et al. 1990). For these materials, it has been shown that, if there are intrusion and compression in the same material, compression occurs at low pressure and intrusion at high pressure. The two phenomena do not occur simultaneously in the same pressure range and it is possible to prevent mercury intrusion in the pore network by wrapping the sample in a membrane. The whole volume variation is then due to sample densification under the effect of isostatic pressure. In the absence of a clear relation between pressure and pore size, Friesen et al. (1988), considering the reproducibility of the variation of volume V as a function of pressure P, suggested to measure the variations of the apparent bulk modulus K0 , as a function of pressure, using the following equation: d P ¼ K 0 ðdV=V Þ

(2)

It is, however, fundamental to point out that the total volume variation recorded by mercury porosimetry is a true measure of the material pore volume, whether the phenomenon is due to intrusion or densification. Except for rare materials with a large micropore volume, the specific volume VHg measured by mercury porosimetry is equal to the difference between the apparent specific volume, 1/ρ and the specific volume of the solid part of the material, 1/ρs. V Hg ¼ 1=ρ  1=ρs

(3)

with ρ, the bulk density and ρs, the skeletal density obtained by helium pycnometry.

Global Mechanical Behavior of Sol–Gel Materials Aerogels and xerogels are highly porous and compliant materials. They can be made with very low densities and a number of studies have shown that their elastic Young’s modulus is generally very small. Moreover, the Young’s modulus measured on aerogels of various densities can be written as a power law

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

E ¼ E0 ðρ=ρ0 Þα

5

(4)

where E is the Young’s modulus, ρ is the density, and E0 and ρ0 are their corresponding values in a reference state (Gronauer et al. 1986). The value of the α exponent varies from 2.6 (Gross and Fricke 1992) to 3.7 (Woignier et al. 1989), according to the nature of the gel and the method of measurement. The relation: K ¼ E=3ð1  2vÞ

(5)

expresses the proportionality between the elastic bulk modulus and Young’s modulus, as long as Poisson’s ratio v stays constant in the considered domain. The elastic bulk modulus K, thus, can also be written as a power law of the density with the same exponent α. Because aerogels are densified under mercury pressure without intrusion of mercury into the pore volume, the volume of the aerogel sample is progressively reduced by mercury porosimetry from a reference volume V0, until a volume V is obtained when the pressure P is reached in the porosimeter. Scherer et al. (1995) suggest determining the bulk modulus by Eq. 2, from mercury porosimetry measurements. However, it should be noticed that the parameter calculated in this way is a true elastic bulk modulus only at very low pressures, i.e., when the material compression is completely reversible. At higher pressures, the gel exhibits yield, followed by plastic hardening, and the parameter calculated is not constant any more and cannot be assimilated to the elastic bulk modulus K. For this reason, we distinguish between the actual elastic bulk modulus K and the apparent bulk modulus K0 . In the plastic regime, Scherer et al. show that the modulus K0 changes with the volume, according to a power law: K 0 ðV Þ ¼ K 00 ðV 0 =V Þm

(6)

where K 00 and V0 are the modulus and volume of the sample at the beginning of the power-law range. It is noticeable that the parametric adjustment, which enables the determination of the exponent m, is very sensitive to the value of the material initial density or initial volume V0, which thus must be determined with high accuracy. The exponents α and m generally obtained are astonishingly close to each other, despite the fact that the deformation process, elastic or plastic, is clearly different. Woignier et al. (1997) prepared an aerogel series with different densities by densification in a mercury porosimeter and showed that the bulk modulus K0 of these samples obtained from Eq. 2 from the mercury porosimetry data is very close to the elastic bulk modulus K obtained from measurements of ultrasound propagation rates. Materials that undergo densification in the whole pressure range generally show a porosimetry curve of volume variation versus pressure that starts by a concave curvature, but then the curvature reverses at high pressure (Fig. 1). As a consequence, the bulk modulus K0 goes through a minimum (Duffours et al. 1996) and the power law is only valid in the part of the porosimetry curve with convex curvature. Aerogels with very low density exhibit volume variation from the lowest pressures

6

R. Pirard et al.

Fig. 1 Mercury porosimetry curves (volume variation as a function of mercury pressure) of two aerogels samples. (○): aerogel of silica–zirconia; bulk density ρ0 = 0.1 g/cm3 and exponent m = 3.52 computed on the whole curve. (Δ): aerogel of pure silica; bulk density ρ0 = 0.165 g/ cm3 and exponent m = 3.33 computed for the convex part of the curve

applied, i.e., below 0.1 MPa. In this case, the whole curve is convex and the power law (Eq. 6) is observed in the entire pressure range. The analysis method of the mercury porosimetry curves developed by Scherer et al. (1995) has the great advantage to enable the complete characterization of the mechanical properties of a series of aerogels with different densities, by one single mercury porosimetry experiment. The porosimeter is at the same time the experimental device for measuring the bulk modulus and the apparatus that leads to increasing aerogel density. In this approach, the global mechanical characteristics of the material are obtained as if the material had been a homogeneous entity. The method does not give information on the material porous texture.

Textural Representation of Aerogel and Aerogel-Like Materials The size of the particles constituting aggregates in aerogel and xerogel samples can be measured by TEM. The general aggregate shape can only rarely be observed by this technique and the pore size between particles and aggregates cannot be estimated. The observation of aerogels by SEM reveals the texture proper to materials obtained by the sol–gel process, but this examination can only be done on very low-density aerogels, due to the low resolution of SEM. It is observed (Fig. 2) that aerogels are formed by particles aggregated into filament-like clusters, which in turn form a three-dimensional cross-linked network. This filamentary structure has been reported in many porous materials, including sol–gel materials (Brinker and Scherer 1990), vapor phase aggregated materials

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

7

Fig. 2 SEM micrograph of an aerogel composed of 50% silica and 50% zirconia. The supercritical drying took place without replacement before drying of the pore liquid, consisting of ethanol with 5% methoxyethanol and water traces. The aerogel was calcined at 400  C in air for 8 h. The sample bulk density is ρ0 = 0.02 g/cm3 (Reprinted from (Pirard et al. 1995) with permission from MRS)

(Smith et al. 1990), or precipitated silica (Iler 1979). The pores are interconnected, and it does not seem possible to define a precise individual pore volume and certainly not a pore surface. This structure suggests that the aerogels can be modeled by a three-dimensional network in which the pores appear as an assembling of empty polyhedra and where only the edges of the polyhedra are materialized by filaments or strings of beads. Gel synthesis has often been described as a process of particle and cluster–cluster aggregation (Brinker and Scherer 1990). Indeed this model leads to structures similar to those observed. Beurroies et al. describe the gel densification phenomenon under isostatic pressure as a compaction and interpenetration of clusters, which can be considered as fractal objects (Beurroies et al. 1998). This model explains the volume variation as a function of applied pressure and the decrease of the cluster center to center distance observed by SAXS. Open cell polymeric foams show a texture similar to that of an aerogel. Gibson and Ashby (1988) proposed to model this texture as a cubic array of rods of length L and square cross section of side d. The idea was picked up and applied to aerogels by Hrubesh and Pekala (1994) and by Scherer et al. (1995), who used as a model a set of cubes, all of the same size, equal to the average pore size. The calculation shows that the bulk modulus of such a structure, elastically compressed, should be proportional to the density squared. The higher exponents observed experimentally can be explained by the presence of dead ends, which are branches connected at only one end, which do not bear a tensile load (Scherer et al. 1995). Pirard et al. (1995) consider a large distribution of cube sizes to represent the material texture. When the material is submitted to external pressure, the solid filaments, which constitute the pore edges, are subjected to axial compressive stress, as a response to pressure. The stresses in the filaments are proportional to the square of the sizes of the pores that they subtend and the most loaded filaments are deforming. After reviewing the different deformation and break modes for a filament submitted to axial compressive stress, Pirard et al. conclude that the most probable deformation mode is buckling.

8

R. Pirard et al.

Reversibility of Shrinkage The densification of aerogels under isostatic mercury pressure can be totally irreversible after pressurization–depressurization, as is the case for silica–zirconia aerogels (Pirard et al. 1995). But most of the time there is a partial reversibility, which can be relatively important if the sample is maintained at moderate pressure for a short time (Alaoui et al. 1998). Anyway, the sample volume is smaller after densification, even after a possible relaxation. Therefore, it is possible to study the effect of isostatic pressure on aerogels by characterizing the texture of materials first densified to a known value of isostatic pressure. Some aerogels show a large reversibility, while others do not. This variability has been attributed to the presence or absence of functional groups able to react with each other on the filament surface. During compression, the filament-shaped aggregates get close to each other and the reactive functional groups, mostly hydroxyls, react, creating links which fix the filaments in their distorted position. It has been observed that calcined silica aerogels, rich in hydroxyl groups Si–OH, able to react to give Si–O–Si  links, show less reversibility than the same aerogels uncalcined. The latter are rich in alkoxide functions Si–O–CH3, less likely to react with each other (Duffours et al. 1995). Smith et al. have shown that silica aerogels treated with trimethylchlorosilane exhibit complete reversibility in the whole pressure range accessible to mercury porosimetry. The authors attribute this behavior to the replacement of hydroxyl functions by trimethylsilane functions, which do not react with each other. Therefore, the filaments distorted by the pressure are not fixed in their buckled position (Smith et al. 1992).

Collapse of Largest Pores Brown et al. have used nitrogen sorption to characterize porous silica partially densified by isostatic pressure. From this experiment, they concluded that, during densification, the largest pores disappear first at the lowest pressures (Brown and Lard 1974). Using this same characterization technique, the same conclusion was derived from studies on silica–zirconia aerogels (Pirard et al. 1995) and pure silica aerogels (Dieudonne and Phalippou 1999). The isotherms of aerogels with the same composition, but with different densities, are practically identical for all the samples, for a relative pressure smaller than 0.8. For higher pressures, the isotherms show an adsorbed volume plateau which indicates that the compacted aerogels do not adsorb nitrogen anymore and thus have no macropores. The plateau starts at a relative pressure that depends on the level of densification of the compacted sample; the higher the densification pressure, the sooner the plateau is reached (Fig. 3). From the analysis of these isotherms, it can be inferred not only that the largest pores disappear first but also that the smaller pores remain unaffected by pressure and that the specific surface area measured does not vary (Pirard et al. 1995). Partially densified silica aerogels have also been characterized by SAXS and, despite the fact that this method does not allow to directly determine pore size, the results obtained are compatible with the disappearance of pores of decreasing size as the pressure increases (Dieudonne 1998).

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

9

Fig. 3 Nitrogen adsorption isotherms at 77 K, for a series of pure silica aerogels first densified under different isostatic pressures. Desorption isotherms were left out for clarity (Pirard 2000)

In a particularly nice theoretical study, Ma et al. (2000) use the mathematical algorithm DLCCA – diffusion limited cluster–cluster aggregation – (Kolb et al. 1983) to simulate the gel construction in 3D and obtain a structure similar to that observed on very low density aerogel micrographs (Fig. 2). The loading of this structure by an isostatic pressure is then calculated by the finite element method. This simulation shows that the load is not evenly distributed in the filaments. On the contrary, the results show that the fraction of bonds bearing the strain in Ma’s aerogel model decreases with decreasing density, and this is why the network is so compliant. The longest filaments are submitted to far higher stresses than the others and are consequently distorted. When these longest filaments are highly deformed or broken, they cannot bear the load anymore. The stresses are then redistributed on a population of shorter and more numerous filaments, which are distorted in their turn. This confirms the disappearance of the largest pores at low pressure and explains why the exponent m of the power law that links the bulk modulus to the density, Eq. 6, can be higher than 2. The explanation previously given by several authors to explain m values higher than 2 was based on the presence of dead ends; this explanation was invalidated by Ma et al. in a complementary study (Ma et al. 2001).

Relation Between Pore Size and Pressure Quantitatively establishing the relation between the size of the pores that are compressed and the pressure of densification is obviously essential in order to determine pore volume distribution as a function of pore size. This relation has been established by Pirard et al. (1995) from the analysis of nitrogen adsorption–desorption isotherms

10

R. Pirard et al.

of aerogels that were partially densified by mercury porosimetry at increasing pressures. Two types of aerogels were analyzed: silica–zirconia aerogels, whose densification is completely irreversible, and pure silica aerogels, whose densification is partially reversible. The pore size distributions have been calculated from Broekhoff and de Boer’s method (Lecloux 1981), in order to determine the size of the largest pores that resisted the pressure without being destroyed. There exists a linear relation between the logarithm of the size of the pores that disappear and the logarithm of the pressure that is responsible for pore compression (Fig. 4). The linear regression established on experimental data leads to a straight line with slope  0.27. Figure 4 compares the slopes of straight lines representative of different mechanisms which can be considered: mercury intrusion in the pores is characterized by slope  1, purely elastic compression of the material by slope  0.5, and compression in the plastic domain by slope  0.33. Slope  0.25 is characteristic of a deformation of the material filaments by buckling. The fact that the straight line slope obtained is close to 0.25 allows writing the following relation: L ¼ kf =P0:25

(7)

where L is the pore size (typically the diameter, if the pores are spherical), P the pressure, and kf is a constant, depending on the material structure. This relation has been identified with the equation describing the buckling of long rods and the

Fig. 4 Logarithm of size of the largest pores that remain after densification versus logarithm of densification pressure. The error bars and squares are represented on this figure. Lines are drawn as guides to the eye. Two series of measurements are presented: silica–zirconia aerogels which are irreversibly densified and for which the densification pressure is precisely determined (error bar) and pure silica aerogels that recover some volume during depressurization and whose densification pressure is determined with some uncertainty (error squares) (Pirard 2000)

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

11

constant kf has been identified with the buckling constant of Euler’s equation and takes the form: kf ¼ nπ 2 E=16


0:25

d

(8)

where d is the diameter of filamentous aggregates, E is the Young’s modulus of the aggregates and n is the critical strength coefficient, that is equal to 1 or 4, depending on whether the filaments are considered as articulated or embedded. The constant kf generally cannot be calculated with sufficient accuracy, and it is necessary to determine it experimentally by measuring the size of the largest pores that remain in a material densified at a known pressure.

Two Successive Mechanisms: Buckling and Intrusion Numerous materials show a behavior more complex than aerogels, when submitted to isostatic mercury pressure. During texture analysis of various porous silica, Minihan et al. (1994) noted that these materials show successively two mechanisms of volume variation during mercury porosimetry. The authors come to this conclusion by collecting samples after mercury porosimetry experiments and by submitting them to nitrogen adsorption–desorption isotherm measurements, subsequently to elimination of the mercury traces trapped in the sample after intrusion. Vittoratos and Auburn (1995) observe a similar behavior during characterization of silica-based catalysts for ethylene polymerization. The authors interrupt mercury porosimetry experiments at different pressures, analyze the curve obtained during depressurization, measure the mass variations, and observe aspect variations of the sample collected after the experiment. They conclude that, at low pressure, the material is densified, whereas above a given pressure, Pt, mercury penetrates into the pore network. In these two papers, the texture data obtained from nitrogen desorption isotherms by BJH method are compared to mercury porosimetry data analyzed by Washburn’s equation. They note very important differences and conclude that Washburn’s equation is not appropriate for mercury porosimetry data analysis in the pressure range where the material undergoes densification. Vittoratos rightly points out that the inadequate use of Washburn’s analysis leads to an interpretation of the densification curve in terms of the presence of large pores. The succession of two mechanisms, compression followed by intrusion, has also been pointed out by Smith et al. on samples of fumed silica (Smith et al. 1990) and by Pirard et al. on carbon black samples (Pirard et al. 1999), silica precipitated from alkaline silicates (Pirard and Pirard 2000), and polyurethane xerogels (Pirard et al. 2003). The presence of two successive mechanisms in mercury porosimetry curves is systematically observed for low-density xerogels synthesized from mixtures of tetraethoxysilane (TEOS) with alkoxides derived from tetramethoxysilane (TMOS), in particular 3-(2-aminoethylamino)propyltrimethoxysilane (EDAS) (Alié et al. 1999). Considering that the texture of xerogels and aerogels is related, the relation between

12

R. Pirard et al.

Fig. 5 Mercury porosimetry curve, volume versus pressure, of a silica xerogel, showing two successive mechanisms. The point of mechanism change, Pt, is located at 26 MPa (Reprinted from Pirard et al. (1997a) with permission from the Royal Society of Chemistry)

the size of the pores that are compressed and the pressure should be the same. Therefore, Pirard proposes the use of Eq. 7 to determine the pore size distribution from mercury porosimetry data, in the pressure range where the samples are densified (Pirard et al. 1997a). In the pressure range above the pressure of mechanism change, Pt, the intrusion of mercury in the pore network takes place and Washburn’s equation should be used. The pressure of mechanism transition, Pt, is generally easy to locate on mercury porosimetry curves, because there is an abrupt slope change at Pt (Fig. 5). In any case, there is no mass variation of the sample during pressurization till Pt, which means no mercury entrapment. When pressure is increased above Pt, the large mass variation shows that mercury is entrapped in the material. The photos of a monolithic sample (Fig. 6) before experiment (A), after pressurization just below Pt (B), and after pressurization to 200 MPa (C) show that the sample pressurized till Pt does not contain entrapped mercury and its volume is reduced in a homothetic way. After compression at 200 MPa, the sample contains entrapped mercury and the surface is scattered with craters and cracks, due to intrusion and extrusion of mercury in the remaining pore volume. By considering that, at Pt, both the equation of buckling Eq. 7 and that of intrusion Eq. 1 are simultaneously valid, the buckling constant kf can be determined by: kf ¼

4γ cos θ P0:75

(9)

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

13

Fig. 6 Monolithic silica xerogel before mercury porosimetry (a); the same sample after densification till a pressure just below Pt (b); and the same sample after pressurization till 200 MPa (c) (Reprinted from Pirard et al. (2002) with permission from Elsevier)

Fig. 7 Cumulative volume distribution (♦) and volume distribution versus pore size (□) of a silica aerogel, obtained by analysis of the curve on Fig. 5 by the appropriate equations (Reprinted from Pirard et al. (1997a) with permission from the Royal Society of Chemistry)

It is thus possible to determine the pore size distribution in the whole pressure domain (Fig. 7). When Eq. 7 is used in the pressure range below Pt, where the sample is densified, and Eq. 1 is used in the pressure range above Pt, where intrusion occurs, the two distribution parts link up without threshold, with a common tangent at the mechanism change.

14

R. Pirard et al.

Relation Between the Buckling Constant Kf and the Aggregate Size According to Eq. 8, the buckling constant kf is proportional to the diameter of the filaments which constitute the aerogel and xerogel skeletons. In order to assess the existence of a relation between kf and a characteristic dimension of the gel structure, Pirard et al. (1998) observed various xerogel samples by transmission electron microscopy (TEM). The mean diameters of the gel particles were determined by direct measurements on the micrographs. However, the particles themselves may not provide an adequate comparison, as they are not a good indicator of the structure of the network and they cannot be subjected to buckling. Buckling occurs at the aggregate level and the aggregates could be considered as the real building blocks of the three-dimensional network of the gel. So, the mean diameter of the filament-shaped aggregates was assessed visually from the micrographs. Despite the inaccuracy of particle and aggregate size measurement by TEM, the linear relation between the constant kf and the aggregate diameter has been evidenced on samples of Pd-Ag catalysts supported on silica xerogels (Pirard et al. 1998) and on low-density silica xerogels synthesized from TEOS and EDAS (Alié et al. 2000). The mixing, in controlled proportion, of an alkoxide with ethoxy groups (TEOS) and an alkoxide with methoxy groups (EDAS), enables tailoring the particle and aggregate size. Figure 8 shows that the linear relation between the buckling constant kf and aggregate diameter is relatively well established for aggregate sizes lower than 100 nm and constitutes a confirmation of the buckling theory (Alié 2001).

Fig. 8 Correlation between the particles diameter (Δ), or the filament-shaped aggregate diameter (○) and the buckling constant kf (Modified from Alié et al. (2000) with permission from Elsevier)

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

15

Membrane Test to Avoid Intrusion The idea of wrapping a sample in a tight membrane to prevent any intrusion during pressure increase inside the porosimeter has been worked out by several researchers. By submitting raw samples and wrapped ones to mercury porosimetry, it is possible to compare their curves of volume versus applied pressure and to determine in which domain there is densification and in which there is intrusion for the raw sample. Scherer et al. did this on a silica aerogel and confirmed that the whole volume variation observed on this sample is due to material densification (Scherer et al. 1995). The experiment is even more interesting when the raw material is densified at low pressure and penetrated at higher pressure. Smith et al. showed, on two fumed silica samples, that the intrusion part of the curve at high pressure is replaced by the extension of the densification part of the curve when the sample is wrapped in a membrane (Smith et al. 1990). Mosquera et al. did the same experiment on silica xerogels and they drew the same conclusions (Mosquera et al. 2002). The wrapping method was used by Pirard et al. on precipitated silica (Pirard and Pirard 2000) and by Alié et al. on low-density silica xerogels (Alié et al. 2001), not only to correctly identify the two successive mechanisms of volume variation but also to demonstrate the validity of Eq. 7 in the densification part. The mercury porosimetry curves of a sample wrapped in a tight membrane and the same material without membrane are identical between 1 MPa and Pt (Fig. 9). This confirms that the mechanism of volume variation in this pressure domain is truly crushing, without intrusion. At pressures above Pt, the two curves are very different, as is expected

Fig. 9 Mercury porosimetry curve of a monolithic sample of precipitated silica and of the same sample wrapped in a tight membrane (Modified from Pirard and Pirard (2000) with permission from Elsevier)

16

R. Pirard et al.

Fig. 10 Volume distribution versus pore size of a monolithic sample of precipitated silica, obtained by analysis of porosimetry curves (Fig. 9), of samples wrapped in a membrane, or not (Reprinted from Pirard and Pirard (2000) with permission from Elsevier)

because the two mechanism are different; for the sample wrapped in a membrane, the only possible mechanism is the densification, whereas the sample without membrane is invaded by mercury. Below Pt, the sample with or without membrane collapses without mercury intrusion. Consequently, Eq. 7 is used to analyze the data. The kf constant is obtained from the Pt value, using Eq. 9. At pressures above Pt, the sample without membrane is invaded by mercury and Eq. 1 is used to analyze the data, but the sample wrapped in the membrane continues to be densified and Eq. 7 is still valid. The pore size distributions obtained from the two different curves with the adequate equation are identical, as shown in Fig. 10. It is noticeable that the range of pore sizes concerned by intrusion and densification is not the same in the same pressure range. Due to the exponent that acts on the pressure in Eqs. 1 and 7, a larger pore size range is accessible if the mechanism is intrusion rather than densification.

Determination of the Buckling Constant Kf It is essential to determine the value of the buckling constant kf, characteristic of the material, in order to calculate the size of the pores that collapse through Eq. 7. This constant, whose theoretical expression is given by Eq. 8, depends on different parameters, proper to the aggregates, which can neither be measured nor theoretically predicted at the present time. It is therefore necessary to determine the value of the constant experimentally, by measuring the length of the filaments that buckle at a

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

17

given pressure. In a way, it is a calibration of Eq. 7 by another method of characterization of the porous texture. When the material shows two successive mechanisms at increasing pressures, densification followed by intrusion, Eq. 9 is used to determine the constant kf. In this case, Eq. 7 is calibrated by mercury intrusion porosimetry. On very different xerogels, pressures Pt have been found ranging from 0.25 (Alié et al. 2002) to 200 MPa, which is the maximum pressure of the porosimeter employed. The constants kf calculated from these values of Pt range between 4250 and 28 nm MPa0.25. When the porosimetry curves show only densification, it is necessary to measure the size of the largest pores remaining after a partial densification at a known pressure. This size can be obtained from nitrogen sorption isotherms, but the calibration work is more difficult and less precise because the determination of the mesopore size slightly depends on the isotherm analysis method and on the pore shape adopted in Kelvin’s equation. Pirard et al. (1995) obtained a constant kf = 48 nm MPa0.25 on silica–zirconia aerogels, by analyzing the nitrogen adsorption isotherms by Broekhoff and de Boer’s method, with a shape factor characteristic of spherical pores. When this value was determined and used, the authors were aware of the importance of the pore size concept developed hereafter. Lower constant values, ranging between 26 and 39 nm MPa0.25, have been determined afterwards on pure silica aerogels, by using a shape factor characteristic of cylindrical pores. It is clear that, compared to the calibration method using mercury intrusion porosimetry, constants higher than 28 nm MPa0.25 are overestimated. Indeed, it can be assumed that all materials presenting a filamentous structure should successively show a densification mechanism followed by an intrusion mechanism. If a material shows only densification during mercury porosimetry, it is likely that the mechanism change is taking place at a pressure higher than the maximum pressure accessible to the mercury porosimeter (usually 200 MPa). The constant kf of such a material cannot be higher than 28 nm MPa0.25, if determined by mercury intrusion calibration.

Pore Size Concept The determination of the buckling constant kf by calibration from mercury intrusion porosimetry, or from nitrogen adsorption–desorption, can lead in some cases to different results. It is likely that, beyond the imprecision due to the method, the differences in pore sizes observed arise from a fundamental difference in the pore size concept. IUPAC proposed to define pore size as the distance between two opposite pore walls (Rouquerol et al. 1994). In the case of materials from the sol–gel process, this definition is not applicable, because pores are not included between walls, but are only delimited by interconnected filaments. In practice, it is considered that the sizes obtained from analysis methods of the texture of porous materials are characteristic pore sizes. Because the different analysis methods are based on different physical phenomena, it is not astonishing that they lead to slightlyy different characteristic pore sizes. Discrepancies resulting from using different characterization methods appear in several publications, often when the same material is analyzed by

18

R. Pirard et al.

nitrogen adsorption–desorption and mercury intrusion porosimetry (Smith et al. 1990; Brown and Lard 1974; Milburn et al. 1988; Minihan et al. 1994). McEnaney et al. noted that the distribution profiles obtained by different characterization techniques are often similar, but that differences, sometimes important, between the absolute values of characteristic pore sizes are almost unavoidable (McEnaney and Mays 1995). The analysis of the nitrogen adsorption isotherms in the mesopore range is based on the capillary condensation phenomenon put into equation by Kelvin. The geometric dimension computed directly from this equation is the curvature radius of the condensed nitrogen meniscus. The pore radius is obtained by adding the thickness of the layer of nitrogen adsorbed onto the solid surface to the curvature radius of the nitrogen meniscus. The reference pore shape is a cylindrical pore in the analysis made by BJH method. It can be cylindrical, spherical, or slit-shaped, by using the adequate shape factor in the analysis by Broekhoff and de Boer’s method (Lecloux 1981). In mercury intrusion porosimetry, the pressure at which intrusion takes place in a pore is related to the curvature of the mercury meniscus at the pore entrance. The knowledge of the contact angle between mercury and the solid enables calculating the pore radius. The characteristic size defined here is the diameter of a reference cylindrical pore, in which the curvature of the mercury meniscus is the same as it is at the entrance of the true pore in which mercury penetrates. The disagreements between pore sizes are particularly expected when the true pore shape of the material studied is very different from a cylinder with circular cross section, which has been chosen as a reference shape for the development of the theories of the various analysis methods. This is certainly the case for sol–gel materials and more generally for materials with a structure of interconnected filaments and whose pores are under no circumstance cylinders with circular sections. The measurement technique of texture of porous materials that collapse under isostatic mercury pressure is based on the mechanical behavior of the network of interconnected filaments. This method defines, as characteristic size, the length of the edge of the reference cubic pore that has the same resistance toward buckling that the true pore that collapses under mercury pressure. The calibration of this technique by other characterization methods is necessary, but results can differ depending on whether the adjustment has been made from nitrogen adsorption-desorption isotherms, or from mercury intrusion porosimetry. In the case of discrepancy, preference is given to this last adjustment method.

Relation Between Global Mechanical Behavior and Textural Representation As explained previously, the densification of a material under isostatic mercury pressure can be used to analyze the material in two different manners. Scherer et al. (1995) study the relation between pressure and volume, by considering the material as a whole without trying to describe its microscopic texture. The authors show that the evolution of bulk modulus K0 versus density is a power law (Eq. 6),

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

19

valid in a certain pressure range. The material is characterized by the exponent m of this power law. Pirard et al. (1995) reasonably assume that the texture shrinkage is due to the buckling of the filaments that constitute the material skeleton. They derive a relation between pressure and pore size (Eq. 7) that is valid at the scale of the individual pores. The material is then characterized by a pore volume distribution versus pore size. As it is possible to characterize the same material in two ways from the same set of experimental data, it is logical to consider that there exists a link between the two characterization methods. To establish this link, it is necessary to express the pore size distribution under an analytical form. This can generally not be done for the entire distribution, but locally by giving a hierarchical representation to the distribution. For this purpose, the pore size is represented as a geometrical progression of ratio η > 1. The ratio between two successive size classes is thus Ln/Ln–1 = η. The volume distribution can be locally defined by a parameter q, so that the ratio of pore volume of two successive classes is expressed by Vn/Vn–1 = ηq. Depending on the value of q, there exist domains where the distribution is uniform (q = 0), increasing (q < 0) or decreasing (q > 0), as a function of pore size. The pore collapse at a pressure corresponding to Eq. 7 leads to establish mathematically a relation m = 4/q between the parameter q, describing the volume distribution versus pore size and the parameter m, that describes the evolution of material bulk modulus versus density increase (Pirard and Pirard 1997). According to Eq. 7, the profile of pore volume versus pressure curves can be theoretically calculated for some values of the parameter q (Fig. 11) and the corresponding values of m can be determined.

Fig. 11 Calculated lost pore volume, as a function of mercury pressure, for various pore volume distributions, identified by their q value (Modified from Pirard and Pirard (1997) with permission of Elsevier)

20

R. Pirard et al.

In a narrow pressure range, the porosimetry curves classically obtained on aerogels start by a concave part, represented by a parameter q < 0. In this pressure range, which exists only for aerogels with relatively high initial densities, K0 decreases slightly when the density increases. This is due to a locally negative value of the exponent m. This has been evidenced by Woignier et al. on silica aerogels (Woignier et al. 1997). The major part of the porosimetry curve is convex and is described by a constant value of q, generally comprised between 1 and 1.3. This value leads to an exponent m comprised between 4 and 3.1, a value that is often observed for silica aerogels (Scherer et al. 1995).

Accuracy of the Buckling Method The study of the behavior of materials that are constituted of interconnected filaments has established that the filament-shaped aggregates go out of shape by buckling, when submitted to mercury porosimetry. Equation 7 resulting from this study enables one to unambiguously relate the pore size to the pressure that induces pore destruction and thus to determine a pore size distribution from mercury porosimetry data. The detailed study of the buckling theory (Timoshenko 1947) shows that the complete collapse of a rod delimiting a pore does not occur completely by application of Euler’s critical stress. Consequently, the complete collapse of a cubic arrangement of filaments representing a pore is not entirely isobaric and it is necessary to increase the pressure above a critical value given by Eq. 7, in order to reduce the cube volume down to zero. Taking into account that the matter filaments are not rectilinear rods, but they present some initial curvature, the beginning of deformation of this filament by bending occurs at a pressure below the critical pressure given by Eq. 7. The collapse of a given pore does not take place at a given pressure, but in a more or less large pressure range around the critical pressure. However, the extension of the range of complete buckling toward higher pressure is limited by the fact that a pore submitted to isostatic pressure does not deform in an isotropic way, but in a uniaxial way, even if its form is a regular polyhedron. The stress that is applied on filaments stays constant during buckling at constant pressure, despite the cube size decrease. These considerations allow calculating the exact volume variation of a cube of size L as a function of applied pressure. Figure 12 shows that 90% of the volume loss takes place in a pressure interval between half and twice Euler’s critical pressure, for filaments with an initial curvature such that the relative deflection (Y ) represents 10% of the filament length (L). From relation (7) between pore size and pressure, it can be inferred that the pore size can be determined with an accuracy of about 20%. It should be noted that the accuracy obtained for mercury intrusion porosimetry is not better, due to the possible, but generally ignored variations on the contact angle between mercury and the solid surface, as well as the hypothesis of cylindrical pores. It is clear that, if a pore of large size L1 is isolated in an environment of pores of small size L2, the resistance toward buckling of the large pore is considerably increased by the presence of the small pores. The whole formed by the large pore surrounded by

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

21

Fig. 12 Volume variation of a cubic pore versus applied pressure, for different initial deflections of the edges. The curves presented as examples are calculated for a pore of size L = 100 nm. The thickness of filaments (rods) is d = 5 nm and the elastic modulus is E = 2500 MPa (Pirard 2000)

the small pores can possibly buckle at a pressure that corresponds to the buckling of pores of small size L2, according to Eq. 7. The distribution obtained is erroneous due to the fact that the volume of the pore of large size L1 is assigned to pores of size L2. This situation is identical to that encountered in mercury intrusion porosimetry, when mercury has access to a large pore of size L1 after penetrating a network of small pores of size L2. In this case, the distribution obtained is also erroneous, because the volume corresponding to the large pore of size L1 is assigned to pores of size L2.

Examples for Application The porous texture characterization by analysis of material compression under isostatic mercury pressure can be used for numerous materials, presenting a structure constituted by particles that are aggregated as interconnected filaments. This structure is systematically observed in materials from the sol–gel process. Silica aerogels with different bulk densities have been prepared from precursor solutions with different concentrations (Rigacci et al. 1998). The texture of these aerogels, whose density ranges from 0.077 to 0.227 g/cm3, has been measured by mercury porosimetry (Fig. 13) and the data have been analyzed using the buckling Eq. 7, in order to obtain the volume distribution versus pore size (Fig. 14). The mercury porosimetry curves show that the total pore volume increases regularly as an inverse function of the density. The results obtained from an analysis with the buckling model show that, for the six samples, the volume distribution is nearly the same for small pores (10–22 nm). The decreasingly dense aerogels show a

22

R. Pirard et al.

Fig. 13 Mercury porosimetry curves (volume vs. pressure) of a series of silica aerogels of increasing density, obtained by increasing concentrations of alkoxide precursors (Pirard 2000)

Fig. 14 Volume distribution versus pore size for a series of aerogels of increasing density, obtained by analysis of the porosimetry curves presented on Fig. 13 (Pirard 2000)

distribution that extends more and more toward larger pores. The texture evolution, as a function of density, of aerogels with the same composition is identical, whether the samples are densified by isostatic pressure application, or the density is increased during synthesis by increasing the alkoxide precursor concentration. Indeed, in both cases the larger pores disappear when the density increases, whereas the distribution of small pores remains unchanged.

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

23

Fig. 15 Mercury porosimetry curve of a silica–zirconia aerogel, calcined at three different temperatures (Pirard 2000)

Aerogels composed of 60% silica and 40% zirconia were calcined in air, at 400  C, 800  C, and 1000  C. After thermal treatment, the porous texture of the samples have been analyzed by mercury porosimetry (Fig. 15; Pirard et al. 1997b). The three samples are irreversibly densified by isostatic pressure in the whole pressure domain, from 0.01 to 200 MPa. The data analysis (Fig. 16) has been done using Eq. 7, with a constant kf estimated at 48 nm MPa0.25, by nitrogen adsorption–desorption isotherm analysis. The volume distributions versus pore size obtained show that the pore volume decreases for all pore sizes during aerogel calcination at increasing temperatures. This is an important difference in the texture evolution of aerogels densified either by isostatic pressure or sintering at high temperatures. This has also been pointed out by Calas et al. (1998), who come to the same conclusions by using other analysis methods. Mercury porosimetry is particularly effective for the determination of volume distribution versus pore size of materials that show two successive mechanisms: material densification, followed by mercury intrusion in the pore network. In this case, the determination of the kf constant proper to the material is easy and does not require any additional experimental work. The two mechanisms are generally easy to distinguish by an abrupt change of slope in the curve of volume versus pressure and the pressure of change of mechanism, Pt enables to directly determine the value of kf, using Eq. 9. The distribution is obtained by analyzing each part of the porosimetry curve by the equation describing the mechanism involved: Eq. 7, for buckling mechanism and Eq. 1 for intrusion. The method has been applied to precisely determine the texture of Pd–Ag/SiO2 catalysts synthesized in a single step by sol–gel process (Heinrichs et al. 1997). It has also been applied to characterize the texture of numerous silica low-density xerogels, synthesized from TEOS and additives derived from TMOS (Alié et al. 2000). Other

24

R. Pirard et al.

Fig. 16 Volume distribution as a function of pore size, of a silica–zirconia aerogel calcined at three different temperatures, obtained by analysis of the porosimetry curves presented in Fig. 15 (Pirard 2000)

materials, like high dispersive silica HDS, strictly speaking not obtained by the sol–gel process, have also been characterized by mercury porosimetry (Pirard et al. 2002). An example of porosimetry curve is given in Fig. 5 and the distribution is presented in Fig. 7. Figure 17 shows the entire pore volume distribution of a low-density xerogel obtained from a succession of characterization methods. Below 2 nm, the distribution is obtained from analysis of the nitrogen adsorption isotherm by Brunauer’s method, and between 2 and 7.5 nm, it is given by the same isotherm analyzed by Broekhoff de Boer’s method (Lecloux 1981). From 7.5 to 53 nm the distribution is obtained from the part of the mercury porosimetry curve that exhibits mercury intrusion, analyzed by Eq. 1 and, between 53 and 350 nm, it is derived from the part of the same mercury porosimetry curve that shows the buckling phenomenon, analyzed by Eq. 7.

Conclusions Several materials, among the most porous, are constituted by particles aggregated in filaments interconnected at their end. In these materials, pores are the spaces between adjacent filaments. They are not limited by a continuous surface, but are largely interconnected, so that the volume of one individual pore can only be defined as the volume comprised between interconnection points of the filaments that delimit the pore. Due to their high porosity, these materials have a low bulk density and also a

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

25

Fig. 17 Entire cumulative distribution of the pore volume versus pore size, for a silica xerogel synthesized from TEOS and additives derived from TMOS

low mechanical resistance toward crushing. Nearly all very porous materials from the sol–gel process show this structure. When submitted to mercury porosimetry, the structure of these materials is compressed under isostatic pressure, before mercury can penetrate by intrusion in the largest pores. The curves obtained show a progressive volume reduction as a function of pressure, and this volume reduction is due to material densification and not to mercury intrusion inside the pore volume. Consequently, Washburn’s Eq. 1 should not be used to analyze the porosimetry data of these materials. However, the data obtained during densification are of great interest, because they enable characterizing both the global mechanical behavior and the porous texture of the material. The curve of volume versus pressure enables one to calculate the bulk modulus in the plastic domain and to determine its evolution as a function of increasing density. The material densification takes place by buckling of the filaments constituting the solid skeleton, starting with the longest filaments that limit the largest pores. There exists a simple relation between pressure and size of the pores that collapse. The relation (7) enables analyzing the densification curve of volume versus pressure in terms of volume distribution versus pore size, as Washburn’s equation does for intrusion curves. The material can thus be characterized by a pore size distribution, both in the case of compression porosimetry and in the case of intrusion porosimetry. It can also be demonstrated that the characterizations by volume distribution versus pore size, or by the evolution of bulk modulus with density, are equivalent concerning the information given. At increasing pressures, numerous materials show mercury porosimetry curves that are successively constituted by a volume variation due to filament buckling and a volume variation due to mercury intrusion in the pores that have not been destroyed

26

R. Pirard et al.

at low pressure. In this case, the pressure at which the mechanism change occurs is easy to identify, because the curve of volume versus pressure shows an abrupt slope change. The data analysis must be done by the buckling equation, for the part of the curve related to compression and by Washburn’s equation, for the part of the curve related to intrusion. The constant of the buckling law is particularly easy to determine in this case, by using the pressure of mechanism change. The detailed developments of the elastic buckling theory based on plausible hypotheses concerning the properties and behavior of the filaments constituting the material show that an accuracy of the same order of magnitude can be obtained for distributions from compression curves, as from intrusion curves.

References Alaoui AH, Woignier T, Phalippou J, Scherer GW. Room temperature densification of aerogels by isostatic compression. J Sol–Gel Sci Technol. 1998;13:365. Alié C, Pirard R, Lecloux AJ, Pirard J-P. Preparation of low density xerogels through additives to TEOS-based alcogels. J Non Cryst Solids. 1999;246:216. Alié C, Pirard R, Pirard J-P. Study of the morphology of porous silica materials. Stud Surf Sci Catal. 2000;128:177. Allié C, Pirard R, Pirard JP. Mercury porosimetry: applicability of the buckling-intrusion mechanism to low-density xerogels. J Non Cryst Solids. 2001;292:138. Alié C, Pirard R, Pirard J-P. The role of the main silica precursor and the additive in the preparation of low density xerogels. J Non Cryst Solids. 2002;311:304. Beurroies I, Duffours L, Delord P, Woignier T, Phalippou J. Fractal geometry change induced by compression densification. J Non Cryst Solids. 1998;241:38. Brinker CJ, Scherer GW. Sol–Gel science: the physics and chemistry of Sol–Gel processing. San Diego: Academic Press; 1990. Broecker FJ, Heckmans W, Fischer F, Mielke M, Schroeder J, Stange A. Structural analysis of granular silica aerogels. In: Fricke J, editor. Aerogels: 1st international symposium proceedings; 1985 Sept 23–25; Würzburg. Proceedings in physics. vol 6. Berlin: Springer; 1986. p. 160. Brown SM, Lard EW. A comparison of nitrogen and mercury pore size distribution of silica of varying pore volume. Powder Technol. 1974;9:187. Calas S, Sempéré R. Textural properties of densified aerogels. J Non Cryst Solids. 1998;225:215. Dieudonne Ph, Delord P, Phalippou J. Small angle X-ray scattering of aerogel densification. J Non Cryst Solids. 1998;225:220. Dieudonne P, Phalippou J. Textural properties of densified aerogels. J Sol–Gel Sci Technol. 1999;14:1. Duffours L, Woignier T, Phalippou J. Plastic behaviour of aerogels under isostatic pressure. J Non Cryst Solids. 1995;186:321. Duffours L, Woignier T, Phalippou J. Irreversible volume shrinkage of silica aerogels under isostatic pressure. J Non Cryst Solids. 1996;194:283. Fadeev AY, Borisova OR, Lisichkin GV. Fractality of porous silicas: a comparison of adsorption and porosimetry data. J Colloid Interface Sci. 1996;185:1. Friesen WI, Mikula RJ. Mercury porosimetry of coals. Pore volume distribution and compressibility. Fuel 1988;67:1516. Friesen WI, Ogunsola OI. Mercury porosimetry of upgraded western Canadian coals. Fuel. 1995;74:604. Gibson LJ, Ashby MF. Cellular solids. New York: Pergamon; 1988.

Specific Behavior of Sol–Gel Materials in Mercury Porosimetry: Collapse and. . .

27

Gronauer M., Kadur A., Fricke J. Mechanical and acoustic properties of silica aerogel. In: Fricke J, editor. Aerogels: 1st international symposium proceedings; 1985 Sept 23–25; Würzburg. Proceedings in physics. vol 6. Berlin: Springer; 1986. p. 167. Gross J, Fricke J. Ultrasonic velocity measurements in silica, carbon and organic aerogels. J Non Cryst Solids. 1992;145:217. Heinrichs B, Delhez P, Schoebrechts J-P, Pirard J-P. Palladium–silver sol–gel catalysts for selective hydrodechlorination of 1,2-dichloroethane into ethylene. I synthesis and characterization. J Catal. 1997;172:322. Hrubesh LW, Pekala RW. Thermal properties of organic and inorganic aerogels. J Mater Res. 1994;9:737. Iler RK. The chemistry of silica. New York: Wiley; 1979. Kolb M, Botet R, Jullien R. Scaling of kinetically growing clusters. Phys Rev Lett. 1983;51:1123. Lecloux AJ. Texture of catalysts. In: Anderson JR, Boudart M, editors. Catalysis: science and technology. 2 Berlin: Springer; 1981. p. 171. Ma HS, Roberts AP, Prévost JH, Jullien R, Scherer GW. Mechanical structure-property relationship of aerogels. J Non Cryst Solids. 2000;277:127. Ma HS, Prévost JH, Jullien R, Scherer GW. Computer simulation of mechanical structure property relationship of aerogels. J Non Cryst Solids. 2001;285:216. McEnaney B, Mays TJ. Characterization of macropores in carbons. In: Patrick JW, editor. Porosity in Carbons. London: Edward Arnold; 1995. p. 94. Micoulin J-M, Chevallier Y. Silica-base rubber composition intended for the manufacture of highway tires improved resistance to rolling. Patent US 6,191 205 B1. 2001 Feb 20. Milburn DR, Adkins BD, Davis BH. Comparison of results from nitrogen adsorption and mercury penetration for spherical-particle carbon blacks. Stud Surf Sci Catal. 1988;39:501. Minihan AR, Ward DR, Whitby W. Interpretation of the differences between the pore size distributions of silica measured by mercury intrusion and nitrogen adsorption. In: Bergna HE, editor. The colloid chemistry of silica – proceedings of the 200th national meeting of the American Chemical Society. 17 Washington DC: American Chemical Society; 1994. p. 341. Mosquera MJ, Pozo J, Esquivias L, Rivas T, Silva B. Application of mercury porosimetry to the study of xerogels used as stone consolidants. J Non Cryst Solids. 2002;311:185. Pirard R. Etude de la texture des matériaux hyperporeux par porosimétrie au mercure. Ph.D. thesis. Liège: Université de Liège; 2000. Pirard R, Pirard J-P. Aerogel compression theoretical analysis. J Non Cryst Solids. 1997;212:262. Pirard R, Pirard J-P. Mercury porosimetry applied to precipitated silica. Stud Surf Sci Catal. 2000;128:603. Pirard R, Blacher S, Brouers F, Pirard J-P. Interpretation of mercury porosimetry applied to aerogels. J Mater Res. 1995;10:2114. Pirard R, Heinrichs B, Pirard J-P. Mercury porosimetry applied to low density xerogels. In: McEnaney B, Mays TJ, Rouquerol J, Rodriguez-Reinoso F, Sing KSW, Unger KK, editors. Characterisation of porous solids IV. Cambridge: The Royal Society of Chemistry; 1997a. p. 460. Pirard R, Bonhomme D, Kolibos S, Pirard J-P. Textural properties and thermal stability of silica–zirconia aerogels. J Sol–Gel Sci Technol. 1997b;8:831. Pirard R, Heinrichs B, Van Cantfort O, Pirard J-P. Mercury porosimetry applied to low density xerogels. Relation between structure and mechanical properties. J Sol–Gel Sci Technol. 1998;13:335. Pirard R, Sahouli B, Blacher S, Brouers F. Sequentially compressive and intrusive mechanisms in mercury porosimetry of carbon blacks. J Colloid Interface Sci. 1999;217:216. Pirard R, Alié C, Pirard J-P. Characterization of porous texture of hyperporous materials by mercury porosimetry using densification equation. Powder Technol. 2002;128:242. Pirard R, Rigacci A, Marechal JC, Quenard D, Chevalier B, Achard P, Pirard JP. Characterization of hyperporous polyurethane-based gels by non-intrusive mercury porosimetry. Polymer. 2003; 44(17):4881. Quinson JF, Brun M. Progress in thermoporometry. Stud Surf Sci Catal. 1988;39:307.

28

R. Pirard et al.

Renouprez A. Small angle scattering of X-rays. In: Everett DH, Otterwill RH, editors. Proceedings of the international symposium on surface area determination. London: Butterworths; 1970. p. 361. Rigacci A, Achard P, Ehrburger-Dolle F, Pirard R. Structural investigations in monolithic silica aerogels and thermal properties. J Non Cryst Solids. 1998;225:260. Rouquerol J, Avnir D, Fairbridge CW, Everett DH, Haynes JH, Pernicone N, Ramsay JDF, Sing KSW, Unger KK. Recommendations for the characterization of porous solids. Pure Appl Chem. 1994;66:1739. Scherer GW, Smith DM, Qiu X, Anderson M. Compression of aerogels. J Non Cryst Solids. 1995; 186:316. Smith DM, Johnston GP, Hurd AJ. Structural studies of vapor-phase aggregates via mercury porosimetry. J Colloid Interface Sci. 1990;135:227. Smith DM, Deshpande R, Brinker CJ. In: Hampden-Smith MJ, Klemperer WG, Brinker CJ, editors. Better ceramics through chemistry V, Material Research Society symposium proceedings. Pittsburgh: Materials Research Society; 1992. p. 567. Timoshenko S. Théorie de la Stabilité Elastique. Paris-Liège: Librairie Polytechnique Béranger; 1947. Vittoratos ES, Auburn PR. Mercury porosimetry compacts SiO2 polymerization catalysts. J Catal. 1995;152:415. Washburn EW. Note on a method of determining the distribution of pore sizes in a porous material. Proc Natl Acad Sci. 1921;7:115. Woignier T, Phalippou J, Vacher R. Parameters affecting elastic properties of silica aerogels. J Mater Res. 1989;4:688. Woignier T, Duffours L, Beurroies I, Phalippou J, Delord P, Gibiat V. Plasticity in aerogels. J Sol–Gel Sci Technol. 1997;8:789.

Application of SVET/SIET Techniques to Study Healing Processes in Coated Metal Substrates Alexandre Bastos

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metals and Corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Corrosion as a Natural Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Corrosion Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Types of Corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Corrosion Prevention and Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inorganic Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metallic Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Organic Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manufacture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paint Application and Film Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Corrosion Protection by Organic Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples of Paint Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Degradation of Coated Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Self-Healing Coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Scanning Vibrating Electrode Technique (SVET) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Typical Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Selected Examples of SVET Measurements on Painted Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scanning Ion-Selective Electrode Technique (SIET) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Potentiometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ISE Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ion-Selective Microelectrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Constitution of an ISME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cocktail Composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 3 4 5 6 6 7 7 8 8 9 9 10 10 12 14 16 17 20 21 22 24 24 27 29 30 30

A. Bastos (*) DEMaC – Department of Materials and Ceramic Engineering, CICECO – Aveiro Institute of Materials, University of Aveiro, Aveiro, Portugal e-mail: [email protected] # Springer International Publishing AG 2017 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_138-2

1

2

A. Bastos

Fabrication of ISME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SIET Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of SIET to Studies with Coated Metal Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Other Ion-Selective Microelectrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Limitations of Using ISME in Corrosion Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Other Microelectrochemical Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analysis of Corrosion and Healing of Coatings by SVET and SIET . . . . . . . . . . . . . . . . . . . . . . . . . . . Review of Works Using SVET and SIET to Probe Healing in Coated Metals, Emphasizing Coatings Produced by the Sol–Gel Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32 33 35 36 38 40 41 45 52 52

Abstract

Self-healing has become a hot topic in the field of protective coatings where an intense effort is being carried out to find ways to prolong the service life of both coating and metal substrate. This chapter reviews the use of two localized electrochemical techniques, the Scanning Vibrating Electrode Technique (SVET) and the Scanning Ion-Selective Electrode Technique (SIET) for studying the performance, degradation, and healing processes of coated metals. First, a brief outline of corrosion and corrosion protection, with emphasis on organic coatings, is given to provide the context of the work. This is followed by the concept of self-healing coatings. The principles of SVET and SIET are then presented, together with selected examples of their use. The chapter closes with a discussion on the strategies to probe healing processes and a review of published work.

Introduction Polymeric coatings are the most universal method of corrosion prevention and control. The search for coatings with self-healing ability is a recent trend and many spectroscopic, surface analysis, and electrochemical techniques are being used to assess the performance and study the mechanisms of action. When defects are produced, either intentionally or unintentionally, localized techniques are needed for a better characterization of the processes. SVET and SIET are well suited to analyze healing of defects in coatings but have limited use for intact systems. The two techniques were originally applied in biology and later adopted in corrosion research. SVET measures the ionic currents in solution associated with the corrosion process. The localization of anodes and cathodes, their magnitudes, and the evolution in space and time can be easily obtained. SIET, in turn, is able to identify the ions related with the ionic fluxes in solution. It can provide information about the local pH and the concentration of selected ions of interest, like metal cations from the anodic dissolution and inhibitive ions leached out from the coating. This chapter was written to be an introduction to the SVET and SIET techniques, a discussion on how they can be used to probe self-healing processes, and a review of published work using them to characterize healing processes in coated metals.

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

3

Metallurgy + Energy Corrosion Ore

Metal

Fig. 1 Corrosion as the reverse of metallurgy

Metals and Corrosion The evolution of mankind across history took place side by side with the advances in metallurgy. Progress in agriculture, first, and in transportation and industry, later, owe much to the developments in metal processing. Polymers and composites are increasingly important, but the relevance of metals and alloys will persist due to their abundance in the earth’s crust and outstanding mechanical properties. The main problem of technical metals is their tendency for corrosion, defined as the spontaneous (electro)chemical interaction of a metal with its environment, resulting in changes in the properties of the metal, often leading to impairment of the function of the metal, the environment, or the technical system of which they form a part (definition based on ISO 8044-2015). The impact of metallic corrosion is enormous, and figures around 3–5 % of gross domestic product (GDP) in each country have been found in different studies, since the first ones (Uhlig 1949; Hoar 1971) to the more recent surveys (Koch et al. 2002; Kruger 2011). Corrosion costs include: (i) material and labor expenditures associated with the protective measures; (ii) expenses arising from the extra material and labor for prevention; (iii) replacement costs and lost production; and (iv) the expenses incurred by using information, technology transfer, research, development, and insurance (Kruger 2011). To these costs it must be added all injuries and deaths resulting from material failure due to corrosion. Given the importance of metals and being corrosion omnipresent, it is important to understand why metals corrode.

Corrosion as a Natural Process Most metals exist in the Earth’s crust in the oxidized form (silicates, carbonates, oxides, sulfides, etc.). The ores (economically viable minerals) are extracted and converted to the metallic (reduced) form by metallurgical processes at the expense of energy. The amount of this energy is the exact measure of the metal tendency to return to its native state. The natural process by which this occurs is corrosion, and it can be regarded as the reverse of metallurgy – Fig. 1. As soon as metals are in the

4

A. Bastos

Acidic media

H2

Neutral and alkaline

Mn+

H+

OH-

H2 O O2

M e-

eMetal

Fig. 2 The corrosion cell

reduced form, they become in contact with the natural environment that over the eons kept them in the oxidized form. The only metals that do not corrode in normal conditions are those already metallic in nature.

Corrosion Reactions Metallic corrosion is an electrochemical phenomenon involving the oxidation of the metal and reduction of some species from the environment. The overall redox process can be divided in half-reactions, the most common being: Metal oxidation MðsÞ ! Mnþ ðaqÞ þ ne

(1)

Main reduction in acidic media 2Hþ ðaqÞ þ 2e ! H2 ðgÞ

(2)

Main reduction in neutral or basic media O2 ðgÞ þ 2H2 OðlÞ þ 4e ! 4OH ðaqÞ

(3)

Figure 2 shows the half-reactions in a schematic of the corrosion cell, with electrons flowing from anode to cathodes through the metal phase and the electric circuit being closed in the environment side by the flow of anions towards the anode and cations towards the cathode. Metal oxidation occurs spontaneously when the equilibrium electrode potential of the metal is more negative than the equilibrium electrode potential of half-reactions (2) and (3). This is true for many metals and is the driving force for their oxidation. The metals that do not naturally corrode are the ones not forced to oxidize by

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . . Crevice Uniform

IG

Pitting

stress

5

Galvanic mG

Selective

stress

SCC

Fig. 3 Some types of corrosion

reactions (2) and (3). Water and oxygen are ubiquitous in the earth’s atmosphere; therefore, as long as the metal is in contact with them, the process will take place. The metal oxidation written as Eq. 1 hides more complex mechanisms. For iron see, for example, Lorenz and Heusler (1987) and for zinc see Zhang (1996). The reduction reactions also have different pathways, depending on the metal and on the environment. For more details about the hydrogen evolution reaction see, for example, Vetter (1967) and (Newman and Thomas-Alyea 2004, pp. 217–225) and for the oxygen reduction reaction see Hoare (1968), Damjanovic (1969), and Tarasevich et al. (1983). The products of the half-reactions react chemically between them or with species in the environment and precipitate as corrosion products. Metal oxides and hydroxides, often containing chloride, sulfate, or carbonate, are common. The initial corrosion products may evolve to more stable corrosion products in a process called aging. In some cases the metal cation may further oxidize, and other cathodic reactions are also possible, as long as oxidizing species are available. The rate at which metals corrode depends on the concentration of oxidizing species in the environment and on how easy oxidation and reduction reactions occur at the metal surface.

Types of Corrosion The anodic and cathodic half-reactions take place separated in space, in countless small anodes and cathodes, changing position very rapidly and leading to homogeneous metal dissolution. This is called general or uniform corrosion. Though, this is not always the case. Depending on the metal, environment, and geometry of the system, corrosion appears in different forms, some of them represented in Fig. 3. Localized types like pitting, crevice, intergranular (IG) and stress corrosion cracking (SCC) are typical of passive metals (stainless steels, nickel, and aluminum alloys), i. e., metals that under certain conditions are covered by a thin interfacial solid film, usually an oxide, that decreases the corrosion rate by orders of magnitude. Corrosion appears in points where the passive layer is disrupted, with high local corrosion rates, because the cathodic reactions are able to proceed in the rest of the surface. Galvanic corrosion is a very common corrosion type which results from the contact of

6

A. Bastos

different metals in the same ionic medium and causes the corrosion of the metal with more negative potential. Sometimes the dissimilar metals are part of the same alloy, one being the matrix and the other being present in intermetallic phases, impurities, or precipitates, leading to microgalvanic couples (μG). In other cases the phases of the same alloy may work as a galvanic couple, causing the selective leaching of the more active phase, leaving voids in its place, decreasing the mechanical resistance, and creating nuclei for fracture initiation. More information about corrosion types can be found in any corrosion textbook (Fontana 1986; Jones 1996) or handbook (Cramer and Covino 2003; Revie 2011; Richardson et al. 2010).

Corrosion Prevention and Control Corrosion is a thermodynamically favored process. While it is not possible to fight thermodynamics, the principles of kinetics can be used to slow down the process, sometimes quite radically. Corrosion rate depends on the concentration of oxidizing species and on how easy is the charge transfer for both oxidation and reduction. This gives the clue for all methods of prevention and control, which can be divided in three groups: those that act on the environment, those that act on the metal, and those that provide a barrier between metal and environment (Wranglen 1985). In the first group is the control of humidity (e.g., indoors), removal of oxygen content (e.g., in closed circuit water systems), removal of aggressive species (e.g., chloride ions), and addition of inhibitors (liquid or vapor phase). These strategies are used to decrease the environment corrosivity. Inhibitors, as the name implies, inhibit the interfacial reactions by “poisoning” the surface. The methods that act on the metal change its potential either in the anodic direction (anodic protection) to form a protective passive layer (possible only when such a layer can be formed) or in the cathodic direction (cathodic protection), bringing the potential well inside the immunity region, where the metal oxidation becomes thermodynamically prohibited. A barrier isolating the metal from the environment, if truly effective, prevents both halfreactions at the surface and blocks the ionic path between them.

Coatings Coatings represent a barrier between the substrate and the environment. The field of surface coatings is enormous, embracing systems of completely different nature, with many forms of application and film formation. The functionalities are also very wide and diverse: decoration, protection (not necessarily against corrosion), wear resistance, hardness, wettability, light reflection, thermal barrier, electrical conductivity, etc. The coatings for corrosion protection can be divided in three main groups: organic coatings, inorganic coatings, and metallic coatings.

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

7

Inorganic Coatings The most important inorganic coatings for corrosion protection are the chemical conversion coatings, which is a generic name to designate those produced by chemical means in which part of the metal substrate is converted into the coating: chromate conversion coatings, anodized layers, and phosphate coatings. Chromate conversion coatings are used to passivate the surface of many metals (steel, zinc, aluminum, cadmium, copper, silver, and others). This type of surface treatment was initially based on acidic baths of Cr(VI), but due to its high toxicity it has been replaced by Cr(III) which is far less toxic but also more difficult to achieve equivalent performance. Chromate treatments can be applied as a finish, a base for paint, or over phosphate coatings to enhance their protection. Phosphate coatings are applied primarily on steel and zinc surfaces as pretreatments for paint. The coatings are produced by spraying or immersing the metal for a few minutes in a solution containing phosphoric acid, metal salts (Fe2+, Zn2+, or Mn2+), and accelerators (NO3, NO2). Reaction with the metallic surface forms a layer of insoluble phosphate crystals. The porosity of this layer facilitates the anchoring of the paint film. The layer is also an isolation between anodes and cathodes, minimizing underfilm corrosion. Alone, phosphate coatings offer weak corrosion resistance. This increases in combination with sealers like oils or waxes. Anodizing is an electrolytic process where an oxide layer (1–100 μm) is produced on the metal that works as anode. The layer is porous and has to be sealed to increase the corrosion resistance, which is done in boiling water, sometimes with dyes or corrosion inhibitors. This is the most common method of corrosion protection for aluminum and its alloys and, to a lesser extent, magnesium alloys. Other inorganic coatings are vitreous enamels used for decoration and protection and thermal barrier coatings (ceramics like yttria-stabilized zirconia, YSZ) used in exhaust systems of gas turbines and airplane engines.

Metallic Coatings Metallic coatings are inorganic in strict chemical terms, but due to their importance, different forms of application, and mechanisms of protection, they are classified separately. Many processes exist for applying metallic coatings, the most common being electrodeposition, hot-dipping, metal spraying, cladding, electroless plating, high temperature diffusion, chemical vapor deposition (CVD), physical vapor deposition (PVD), and ion implantation. Some produce thin films to modify particular surface properties and are rarely used for protection. Coatings for corrosion protection can be as thin as 0.4 μm in tin cans or reach 500 μm as in zinc or aluminum coats applied by thermal spraying to large-scale steel structures such as bridges and girders. They protect the metallic substrate by two mechanisms: acting as a barrier and, some of them, also through cathodic protection. The coatings have lower corrosion rate than the substrate. Therefore, as long as the coat remains intact the underlying metal is freed from corrosion. When pores or

8

A. Bastos

defects appear on the coating, a case of galvanic corrosion emerges and two things can happen. Either the corrosion potential of the substrate is more negative than that of the coating (like steel covered by a chromium coat) and the substrate corrodes or the exact opposite occurs (like a zinc layer on steel) and the coating will corrode preferentially, protecting the exposed substrate by turning it into a cathode (cathodic protection).

Organic Coatings Organic (or polymeric) coatings comprise related products like paints, varnishes, lacquers, enamels, and linings. In this chapter, organic coating and paint are treated as synonyms and defined as a pigmented, liquid or solid, composition applied to a surface in thin layer which is converted after a certain time into a continuous, solid, colored, and opaque film. Painting is the most simple, widespread, versatile, and often cheapest form of corrosion prevention. Apart from protection, paints are applied with many purposes such as decoration, cleanliness, illumination efficiency, visibility, and safety. Paints are complex composite materials with many aspects to be considered, each one topic of entire books (Wicks et al. 2007; Patton 1979; Munger 1986; Schmid 1981). A brief description of each aspect is now presented to acquaint readers with their importance. It may suffice a simple change in one of these aspects to turn an outstanding coating into a poor performing one.

Composition A paint is typically composed by five types of constituents: binder, which provides the adhesion to the substrate and is the medium where all other components are dispersed; pigments for color and hiding power; fillers (also called inert pigments or extenders) to maintain coating thickness at low cost; solvents to adjust the paint viscosity during manufacture and application; additives for special properties and aid pigment dispersion. The binder is the most important component of a paint, determining the major properties such as adhesion, hardness, chemical resistance, and exterior durability. The first binders used were natural oils, but nowadays most organic coatings are based on synthetic resins like acrylics, epoxies, polyurethanes, polyesters, vinyls, and others. Additives, in spite of the small amounts, are essential for production, application, and performance of paints. Dispersants help the pigment dispersion and prevent reagglomeration and sedimentation during storage. Wetting agents also help pigment dispersion and assist the paint penetration into scratches and pores of the substrate. Defoamers prevent the formation of foam in the manufacturing stage and during paint application. Thickeners and rheology modifiers optimize the paint viscosity for

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

9

storage in containers and for application by brush, hand-roller, or spray. Polymerization catalysts, driers, flash-rust inhibitors, and antiskin agents are also common additives. A paint formulation contains several of the above ingredients in a composition balanced to comply with easy manufacture, simple application, good performance, and long shelf and service lives.

Manufacture The production of a paint is divided in two main stages: the pigments dispersion and the let-down. The dispersion of pigments is necessary for the homogeneous distribution of pigments in the matrix. Pigments are delivered with the particles in agglomerated form to prevent dust in the manufacturing plants. The agglomerates must be broken in small aggregates or individual particles. This requires mechanical energy (high-speed disk dispersers, ball mills, or roll mills) with the help of dispersants and defoamers. A bad dispersion leaves large agglomerates with air entrapped. In the dry film these will be preferential places for water ingress and early corrosion. Many academic laboratories that nowadays develop new particles and micro- or nanocontainers overlook this step, and bad results in corrosion testing cause the rejection of systems that if well dispersed would show good performance and maybe even commercial potential. The let-down stage involves mixing the pigment paste with resin and finishing the paint to the desired viscosity. Once the paint is finished, due to the final low viscosity, no further pigment dispersion can be achieved, regardless the disperser speed and the addition of dispersants. The production ends with the quality control of the paint and the transfer to containers such as buckets and drums for delivery or stock.

Paint Application and Film Formation Paints can be applied in many ways like brushing, rolling, dipping, or spraying. In industry, powder coatings or electrophoretic deposition lines permit to coat areas of difficult access, reduce paint waste, and avoid volatile organic compounds. The film formation corresponds to the chemical and physical processes involved in the conversion of the liquid layer into a dry film. Depending on the type of binder, the film is formed by solvent evaporation, reaction with atmospheric O2, reaction between two components, particle coalescence, and polymerization activated by heat or radiation. It is never enough to stress the importance of the surface preparation before paint application, because it seriously affects the adhesion to the substrate and the probability of osmotic blistering, ultimately playing a decisive role in the service life of the paint scheme. Further, foam during application and a dusty atmosphere on drying create weak points in the film and preferential paths for water ingress.

10

A. Bastos

Corrosion Protection by Organic Coatings Paints protect the metal substrate from corrosion by several mechanisms: (i) Barrier effect. The most evident action of a protective coating is to provide a barrier to oxygen, water, and ions. However, there are many examples of polymers and paints where the water and oxygen permeabilities are too high to be the corrosion rate-determining step (Thomas 1991). The permeability of ions, on the other hand, is usually low. The ionic resistance has been considered to be the single most important corrosion protection mechanism of organic coatings (Forsgren 2006). The difficult ionic flow through the paint means a path of high electrical resistance between anodes and cathodes retarding corrosion. This has been first noted by J.E.O. Mayne (1949). In some cases, even if corrosion is able to proceed, the paint film may block the corrosion products which remain in the metal-paint interface counteracting the advancement of the process. The barrier properties are enhanced by higher film thickness and lamellar pigments (micaceous iron oxide, mica, glass, aluminum, and stainless steel flakes) which increase the diffusion path for water and oxygen. (ii) Reservoir of corrosion inhibitors. Primers for corrosion protection are usually formulated with anticorrosive pigments, that is, inorganic pigments of sparingly solubility containing ions with inhibiting properties. The most effective pigments are red lead (Pb3O4 plus some PbO) and chromates (basic zinc chromate, strontium chromate and zinc tetroxychromate), but they have been banned due to their toxicity and replaced by zinc phosphate and its modifications. Zinc phosphate shows good performance in the field, but it does not perform well in accelerated tests, probably due to its low solubility. For this reason modifications like polyphosphate silicates with zinc, calcium, strontium, or aluminum have been developed (Forsgren 2006). All above pigments are thought to work by creating passivating films on the metal surface. A different type of pigment with success for steel is zinc powder which works by cathodic protection. (iii) Adhesion to the substrate. Strong adhesion means less points available at the metal surface for electrochemical reactions. The build-up of a water layer under the coating also becomes difficult. A coating with high ionic resistance is an electrical insulator and together with strong adhesion creates a path of extremely high electrical resistance at the metal surface to close the corrosion circuit. This can explain the excellent performance of some coatings in spite of being formulated without any anticorrosive pigments.

Examples of Paint Schemes To better perceive the variety of applications and the multiplicity of types of paints, number of layers, thicknesses, and pretreatments, a few examples of paint schemes are now presented. It must be emphasized that for each example many alternatives exist and they have varied in time, as a result of technical progress and

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

11

environmental regulation. Likewise, different regions of the globe and different companies in the same sector opt for different schemes for the same application. When studying coating performance and the mechanisms of degradation, it is important to know in which position of this multidimensional plot is the coating under study located. (i) First protective system for steel structures. In the nineteenth century, to protect the steel bridges and other large structures in railways, a coating system was developed consisting of a red lead linseed oil primer and a graphite flake linseed oil topcoat applied in two or more coats (Munger 1986). The red lead in an oil binder provided very good protection for steel and the graphite flake structure increased the diffusion path for water. This system conferred long life to steel structures in many environments except in extremely corrosive marine or industrial environments and remained in service until the recent prohibition of red lead. (ii) Underwater hull in naval ships. Coating systems for the immersed parts of a ship should be corrosion inhibiting, antifouling, abrasion resistant, smooth, and compatible with cathodic protection. They may consist of an anticorrosive primer and an antifouling topcoat with total dry film thickness from 250 to 400 μm, sometimes more. The corrosion protection is obtained by a strongly adherent barrier paint combined with cathodic protection. The antifouling action is achieved with paints containing biocides and self-polishing capability. The surface of the paint is washed away by the movement of the ship, thus preventing fouling growth while maintaining the underwater hull smooth. Each part of the ship has its own paint scheme: superstructure above the waterline, deck, inside areas, tanks for fuel, water, ballast, cargo, etc. (iii) Aircraft. Apart from excellent adhesion to the substrate (which might be aluminum, aluminum alloys, titanium, or composite plastics) and provision of corrosion protection, the coatings must also resist swelling by hydraulic fluids, lubricating oils, and fuel; have good flexibility to endure rapid temperature changes from the hot ground to the freezing temperatures at service altitude; and also bear excellent abrasion resistance to dust, sand, rain, and sleet at high speed. Common paint schemes comprise a pretreatment by chromate (or nonchromate) conversion coatings or anodizing, plus a strontium chromate polyamide epoxy primer (30 μm), a polyurethane basecoat (27–75 μm), and a polyurethane clearcoat (25–40 μm) on the top to improve color and gloss retention, cleanliness, and durability. (iv) Automotive industry. The 12 years warranty against perforation offered by many manufacturers is possible thanks to a series of protective coatings applied in automated sequence. A typical protection scheme of the steel car body includes 7 μm electrodeposited layer of zinc (or zinc alloy), phosphate pretreatment, 20 μm electrophoretic epoxy primer for corrosion resistance, 30 μm polyester primer for surface leveling and stone chip resistance, 10–30 μm polyester basecoat for color, and a final 30–60 μm acrylic transparent clearcoat for weather and wear resistance and for gloss and color retention.

12

A. Bastos

(v) Casings for household appliances. Paints must withstand chemicals, food acids, pollutants, and water; protect from corrosion; be tear, wear, and scratch resistant; and maintain a like-new appearance for an extended period of time. For the sake of time and cost savings, many companies opt for powder or electrophoretic lines with a single 20–50 μm polyester-epoxy coat applied on steel or galvanized steel. (vi) Do it yourself (DIY) all-purpose coating. Water- or solvent-based alkyd paints are cost-effective and provide excellent flow, ideal for brush application in many DIY jobs. One or two coats by brush or spraying with final dry film thickness of 50–100 μm are common.

Degradation of Coated Metals The corrosion of a painted metal starts immediately in pores and defects. In intact systems the degradation begins with the attack of the paint. Abrasion, impact, cracking, or crazing at low or high temperatures, polymer bonds breakage by ultraviolet light, freeze-thaw cycles, hydrolysis and oxidation reactions all contribute to the degradation of the polymer film (Schmid 1981). Such degradation increases the access of water, oxygen, and ions to the metal through the polymeric matrix. The corrosion of painted metals can then be divided in (Leidheiser 1987): (i) (ii) (iii) (iv)

Transport of water, oxygen, and ions through the coating Development of an aqueous phase at the metal/coating interface Activation of the metal surface for the anodic and cathodic reactions Deterioration of the coating/metal interfacial bonding

Corrosion of the metal beneath the polymeric coating is an electrochemical process that follows the same principles as corrosion of the bare metal. It may appear in different forms: blistering, cathodic delamination, anodic undermining, filiform corrosion, and flash-rusting. Loss of adhesion when wet is another type of deterioration that may or may not be related to corrosion. Blistering. Blisters are local regions where the coating loses adhesion to the substrate, water accumulates, and corrosion begins. Several mechanisms have been proposed (Funke 1981). The most common is osmotic blistering which occurs when soluble salts are present in the metal surface beneath the coating. Water crossing the film dissolves the salt and forms a blister which swells in an attempt to equalize the higher concentration of the internal solution with the concentration of the outer solution. Other mechanisms for blister formation are electroosmotic blistering (water moves through a membrane or a capillary system under the influence of an electrical potential gradient that may be caused by local corrosion cells), blistering by volume expansion due to swelling, blistering due to gas inclusion or gas formation, and blistering due to phase separation during film formation. Cathodic delamination. Beneath the blister a separation of anodic and cathodic reactions takes place with the cathodic process occurring at the edge of the blister,

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

Paint

Dust

Agglomerates Pores

13

Defects Voids, air, solvent

Thin

Salts, dirt, fats, oxides

Metal substrate

Aqueous environment Swelling Water absorption

Metal substrate

Blistering

Anodic undermining

Cathodic delamination

Fig. 4 Weak points of paints applied to metals and forms of degradation

where O2 has a faster access. The resulting local high pH leads to the attack of the polymer, or breakage of the bonding to the substrate or even the dissolution of the oxide layer that always exists at the interface between the organic coating and the metal. In any case this results in the delamination of the coating. The same phenomenon can occur when cathodic protection is applied on coated steel. Overprotection increases the cathodic reactions resulting in the same chain of events described for the delamination at the edge of the blister but this time spread all over the steel surface under cathodic protection. Anodic undermining. Corresponds to the separation of the coating from the substrate by the anodic dissolution of the metal underneath. Filiform corrosion. This is a type of attack that manifests in the form of threadlike filaments, occurring generally in a combination of humid environments and coatings applied on not well-cleaned (salt-contaminated) substrates. Flash-rusting is sometimes observed when water-based primers are applied on steel. Before the paint dries, the aqueous film permits the corrosion of the substrate if not enough amount of soluble corrosion inhibitors was added to the formulation. Loss of adhesion when wet. This is observed with some coatings that lose adhesion when immersed but regained it after drying. The bonding between paints and metals was initially assumed to be an interaction between polar groups, later treated as a result of dispersion forces, and now it is considered to be an acid–base interaction. Water may interfere with this chemical interaction and when it does, loss of adhesion when wet occurs. Figure 4 is an illustration of the paint film with several weak points and some forms of degradation by exposure to an aqueous environment.

14

A. Bastos

Self-Healing Coatings A recent trend in the field of polymeric coatings is the development of “smart” and self-healing systems. “Smart” is used to define coatings that are able to sense changes in the environment and respond accordingly in a predetermined manner (Feng et al. 2007). “Self-healing” refers to the automatic and autonomous recovery of the original properties of the coating lost after an external aggression. Three driving forces marked the impetus for the investigation of self-healing in coatings for corrosion protection: the first was the need for new corrosion pigments in paints due to the prohibition of red lead and chromates; the second was the success of selfhealing in polymers (van der Zwaag 2007; Wu et al. 2008; García et al. 2011), particularly after the publication of White’s paper in Nature (White et al. 2001); and the third was the advances of micro- and nanoreservoirs for controlled release of drugs in the medical sector. It is worth noticing that in 1960 Evans was already calling self-healing films to those whose corrosion resistance was improved by the blocking of pores by corrosion products (Evans 1960a). The present approaches to achieve self-healing in corrosion-protection films are based on responsive prepaint layers, self-sealing materials, and micro- or nanoreservoirs with “smart” release of corrosion inhibitors. Comprehensive reviews exist, most of them published in very recent years, demonstrating the interest on the topic (Zheludkevich 2009; Fedrizzi et al. 2011; Saji and Cook 2012; Makhlouf 2014; Montemor 2014; Wei et al. 2015; Tiwari et al. 2015; Hughes et al. 2016). The ideal course of action is to combine different approaches in a multilevel response as illustrated in Fig. 5. The first level is water repellency obtained by functionalizing the topcoat surface with hydrophobic groups. The second level of protection comes from nanotraps to capture water and aggressive ions like chloride, delaying their transport towards the surface. Examples are nanosized clays which will either absorb water or retain the aggressive ions by ionic exchange. This may be just a palliative in conditions of intense and continuous flux of those species. The third level, which in many systems is in fact the first level of action, takes place when mechanical aggressions open cracks in the film. These can be sealed by polymerization, precipitation, or swelling of reactive agents. The most common are polymeric capsules with oils (linseed oil or tung oil), monomers or uncured polymers (ureaformaldehyde, melamine-formaldehyde, etc.), and capsules that break upon the mechanical impact, releasing the content, which polymerizes and heals the defect (Kumar et al. 2006; Cho et al. 2009; Choi et al. 2016). The last level of protection concerns the corrosion of the substrate. The most investigated approach is the introduction of corrosion inhibitors into micro- or nanoreservoirs and their incorporation into the paint formulation. In the dry film the inhibitor remains inside the reservoirs to be released only when necessary, triggered by an external stimulus produced by the corrosion reactions. The reservoirs allow the use of soluble inhibitors of recognized efficiency but inadequate for direct incorporation in the formulation because of the negative effects on water sensitivity, curing, rheology, surface wetting, and adhesion. The inhibitor in nanosized reservoirs can be in very small amounts and very close to the surface. Its release will occur only at the critical points

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

15

+

+

+ Polymeric capsules

Inorganic shells

+

Zeolites

Water repellent

++++++++++++++++++++ +

Halloysites

LDH

+

+

+

+

+

+

+

+

Bentonite

LbL

“smart” release of corrosion inhibitors

Sealing agents H2O and Clnanotraps Inhibitor reservoirs

Topcoat

pH < 3 Corrosion inhibitor

pH > 8

Primer anode

Metal substrate

e-

cathode

Fig. 5 Multilevel approach of self-healing coatings for corrosion protection

while the rest of the containers remain closed. With some reservoirs it is expected that the release ceases as soon as the corrosion is stopped. The most common trigger is pH, taking advantage of reactions (2) and (3) which increase the local pH or reaction (1) that produces metal cations whose hydrolysis might decrease the pH. Many particles can be produced to be stable at near neutral pH and open or be destroyed at high or low pH, releasing their content. Other triggers are concentration of chloride ions, metal cations from the anodic dissolution, ionic strength, temperature change, light, mechanical impact, etc. Examples of some reservoirs are presented in Fig. 5. The inhibitor can be deposited at the surface of inorganic beads (Snihirova et al. 2012), or inside inorganic capsules (Balaskas et al. 2012), zeolites (Dias et al. 2012) or natural nanotubes like halloysites (Shchukin et al. 2008). The particles can be coated by layers of polyelectrolytes, using the layer by layer deposition method (LbL) which prevent the inhibitor liberation until a moment of need. When corrosion starts, the change in local pH opens the polyelectrolyte cover and releases the inhibitor. This one can also be introduced between the polyelectrolyte layers (Shchukin et al. 2006). Different types of clays have been used to trap aggressive agents such as chloride or water and in the process release corrosion inhibitors (Mishra et al. 2013). Examples are bentonites (Bohm et al. 2001) and layered double hydroxides (LDH) (Buchheit et al. 2003). Synergistic effects can be obtained by combining different containers each one with its own inhibitor, acting over different reactions (Serdechnova et al. 2014) or at different times of immersion (Montemor et al. 2012). In some cases the technology used to produce particles was employed to treat the surface with good results. Examples are the use of the LbL method to store inhibitors at the substrate-paint interface (Andreeva et al. 2008) and

16

A. Bastos

Table 1 Properties of “ideal” and “minimal” self-healing materials (van der Zwaag 2007) An “Ideal” self-healing material Can heal the damage many times Can heal the damage completely Can heal defects of any size Performs the healing autonomously Has equal or superior properties to current materials Is cheaper that current materials

A “Minimal” self-healing material Can heal the damage only once Can heal the damage partially Can heal small defects only Needs external assistance to heal Has inferior properties to current materials Is terribly expensive

the production of a LDH layer at the metal surface before painting (Tedim et al. 2011). Presently all above systems are still confined to the research laboratory and a long way seems to lie ahead before efficient, cost-effective, and robust solutions are available commercially. Until then the paint industry will keep using inorganic anticorrosive pigments, namely modified zinc phosphates, as legitimate replacements for red lead and chromates. The sophistication of the self-healing approach embodies many different strategies, as described above, with different levels of efficiency. Table 1 presents the properties of “ideal” and “minimal” self-healing materials as discussed by van der Zwaag (2007). Most current systems are close to the minimal condition. Hopefully, future will bring systems approaching the ideal situation. The improvement of existing systems and the development of new ones require experimental techniques to compare their performance and to study their mechanisms of action. This must be done probing the system from various perspectives and at different scales. Many techniques are used including surface analysis and spectroscopic and electrochemical techniques. For corrosion protection, coatings with the capability of healing defects is of prime importance. For that, techniques capable of providing localized information, namely detailing the processes at the defects, are needed. Two of those techniques are now described.

The Scanning Vibrating Electrode Technique (SVET) SVET provides the distribution of anodic, cathodic, and inactive areas on a corroding sample. This is done by sensing the electrical field in solution close to the surface. In each point of interest SVET measures the potential, ΔV, between two points distanced by Δr. This potential is then converted into the local current density between the two points by calculation or a previous calibration. To understand the functioning of SVET, it is important to know that in electrolyte solutions the electrical current is transported by ions which move following the electrical field in solution. In most of the practical cases, the current density, i, and the electrical field, E, are directly proportional, with the solution conductivity, κ, being the proportionality factor, in a relation expressed by Ohm’s law Newman and Thomas-Alyea (2004, p. 275),

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

e-

e-

Cathode

+

Corrosion

Anode

Electrolytic cell

17

Biology H+

-

_

e-

+

Cathode

-

e-

Anode

+

Cathode

Current lines Equipotential surfaces

Fig. 6 Current and potential in solution

i ¼ κE ¼ κ

ΔV Δr

(4)

Figure 6 shows three examples of current and potential in solution. The first represents a generic electrolytic cell with the current flowing between two electrodes driven by a power source. The second example is the corrosion cell, a type of galvanic (driving) cell and, as in the previous example, electrons flow from anodes to cathodes with the ions in solution closing the circuit. A final example shows ionic currents in biology, specifically in this case, the transhyphal proton flux of the growth of Achlya (a water mold). The current lines represent H+(aq) entering in the apical zone by symport and being expelled distally by a proton-translocation ATPase (Kropf 1986). The current lines in Fig. 6 follow the conventional direction of current (direction of positive charge). Each line transports the same amount of current which means that the closer the lines the higher the current density. The current lines cross the equipotential surfaces in angles of 90 (not correctly drawn in Fig. 6).

Instrumentation The first measurements of current in solution were made with two reference electrodes, one in a fixed position and the other moving and scanning the solution (Thornhill and Evans 1938; Copson 1943; Evans 1960b) – Fig. 7a). The potential difference in each point of measurement permitted to make a map of the potential in solution and, knowing the solution conductivity, determine the current flowing therein using Eq. 4. Later, the two electrodes were made to scan the surface together with constant distance (Δr) between them, and ΔV was obtained directly – Fig. 7b, c. This was done either with Luggin-Haber microcappilaries (Rosenfeld and Danilov 1967) or with metal microelectrodes (Trethewey et al. 1993). These techniques have been termed SRET – scanning reference electrode techniques (Isaacs et al. 1981). A further improvement was achieved by making the electrode vibrate (Fig. 7d). The

18

A. Bastos

a

b

c

d

r r r r

Fig. 7 Measurement of potential and current in solution

vibration modulates the signal to be filtered and amplified in a lock-in amplifier, increasing the signal to noise ratio and sensitivity (Bluh and Scott 1950; Jaffe and Nucitelli 1974). SVET is the name given to systems based on the vibration of a microelectrode to sense the electrical field and convert it into current density. Several systems existed over the years, but today only two designs persist. One consists roughly on a platinum disk about 125 μm in diameter embedded in a glass sheath with a diameter of 250 μm and vibrating with a peak-to-peak amplitude of 30 μm (Williams and McMurray 2008). Commercial versions of a similar design exist (www.princetonappliedresearch.com; www.biologic.com; last view on June 2016). A different design was developed around 1988 (Scheffey 1988) and is available commercially (www.applicableelectronics.com, last view on June 2016). This system is described in more detail because it is the one used by the author, and all results shown in this chapter were obtained with it. Notwithstanding, the type of measurements and the principles are the same, independently of the system used. The typical electrochemical cell is shown in Fig. 8, with the sample isolated from the solution except for the area to be mapped, glued in an epoxy holder of 3 cm in diameter and adhesive tape around it to make the solution reservoir. The vibrating electrode is a PtIr (80/20 %) metal electrode, 1.5 cm long, coated with a polymer (parylene C) except at the tip and produced by Microprobes Inc. (https://micro probes.com, last view on June 2016). The electrode tip is a ~10–40 μm platinum black deposit produced by passing of negative current in the order of 200–1000 nA for a few minutes. This is necessary to create a large surface area to decrease the tip impedance (Scheffey 1986). The electrode vibrates in two directions, x and z with respect to the surface. For that the probe is connected to a plastic arm fixed to a linkage with two piezoelectric oscillators which are responsible for the x and z vibrations. The amplitude is usually between 10 and 20 μm (peak-to-peak), and the frequency is in the range between 40 and 1000 Hz. Usually they are selected once and are not changed during the rest of the system lifetime. The measurement is performed with the probe vibrating at the reference frequencies (sent by the lock-in

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

19

x and z frequency signals from 2 lock-in amplifiers

Ground and pseudo-reference z

x vib. x

z vib.

to 3D motor positioning system

Piezo vibrators Vibrating 10 µm electrode

Sample Epoxy holder Fig. 8 Schematics of one SVET system

amplifiers to the piezoelectric oscillators), and the potential difference between the probe and a pseudo-reference electrode (Pt black wire) is sent to a preamplifier and then to the lock-in amplifiers (up to a 50000 amplification can be achieved). Another Pt black electrode in solution is connected to the ground of the measurement system. The probe moves by means of 3 stepper motors (resolution of 1 μm) under the control of a camera located above the cell which allows monitoring the sample surface. The image of the surface shows the accumulated degradation until the moment of acquisition while maps show the activity during the period of measurement. The measurements are controlled by the ASET software developed by Sciencewares, Inc. (USA). An antivibration table, a Faraday cage, and a UPS (uninterruptible power source) are recommended for optimum performance. A calibration is needed to relate the measured potential with the current density that originates it. This is done by a routine in which the SVET probe is placed at a certain distance (usually 150 μm) from a point current source (metal microelectrode with 3 μm electroactive area or a glass microelectrode with 2 μm opening) that drives a given current, I (normally 60 nA). The current density i at the distance r from the source is given by (Scheffey 1986; Reid et al. 2007), i¼

I 4 π r2

(5)

The typical SVET output is current density converted from potential measurement. The calibration remains valid for different solutions as long as the software is updated with the new solution conductivity. It is important to note that no redox processes take place at the tip of the electrode. No faradaic current passes in the circuit due to the 1015 Ω impedance of the preamplifier. In addition, the influence of any redox equilibria that can be sensed by a stationary platinum electrode is insignificant, because the vibration mixes the solution around the probe nulling any concentration gradients close to the SVET probe (Ferrier and Lucas 1986; Dolgikh et al. 2016).

20

A. Bastos xz map

a

z line x line single point vs time

xy map

path

b 1 mm

Zn

Fe

c

1 cm

40 µA cm-2 40

Zn

Fe

1 mm

1 mm

d

e

Z component of current density

X component of current density

µA cm-2 µA cm-2

Zn

Fe

1 mm

Zn

Fe

1 mm

Fig. 9 (a) Examples of measurements possible to be performed by SVET; (b) map of 2D vectors of the current density above a zinc-iron galvanic couple (shown in insets); (c) line of 2D vectors of the current density measured parallel to the surface and through the center of the Zn and Fe electrodes; (d) map of the z component of vectors shown in (b); (e) map of the x component of vectors shown in (b)

Typical Measurements SVET measurements start by a reference point away from the area of interest, where no current flows. The value should be close to zero and will be subtracted from all subsequent points. Then a variety of possibilities exist up to the imagination of the operator and the objective of the study. Figure 9a shows the most common types of measurement that can be performed by SVET. Maps parallel to the surface (xy plane) are clearly the preferred option because they provide the distribution of the current emanating from the surface, the polarity – positive (anodic) or negative (cathodic) –

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

21

the magnitude, and the localization. With a sequence of maps it is possible to follow the corrosion evolution in time. Alternatively, a measurement in a plane normal (xz map) to the surface shows the current flow from the source towards the bulk of the solution. Lines can be acquired instead of maps especially if faster measurements are desired. When a point of particular interest has to be studied (a corrosion pit, a defect in a coating, or a feature in the surface of the sample) the probe can be placed in a fixed position above it and measured over time, sampling every 0.1 or 1 s, depending on the kinetics of the process. All SVET systems give the z component of the current density (normal to the surface). Some systems also give the x component, but it is rarely used in corrosion. The signal from both vibrations gives a better picture about the actual current direction. Figure 9b–e are forms of presenting the results of one map acquired parallel to the surface, 100 μm above a zinc-iron galvanic couple (1 mm diameter wires embedded in epoxy resin and electrically connected at the back – insets of Fig. 9b) corroding in 0.005 M NaCl. The mapped area was 6  8 mm2, comprised 50  50 points, and the measurement time was about ~1000 s. Figure 9b is a map of 2D vectors of the current density measured by the two vibrations of the probe. Figure 9c shows the vectors in a line parallel to the surface and passing through the center of the zinc and iron electrodes. Figure 9d depicts a map of the z component of the current in b). This is the typical form of presented SVET results. The positive current (in red) corresponds to anodic activity (zinc oxidation) and the negative currents (in blue) to cathodic activity (O2 reduction at the iron electrode with production of OH, Eq. 3). Regions where the z component of the current is null or too small to be detected appear in green and are usually associated to absence of electrochemical activity at the surface. Figure 9e shows the x component of the current. Positive and negative currents keep the signal when flowing left to right but change signal when moving in the opposite direction.

Limitations SVET is a noninvasive technique, often producing self-explanatory results. Though, some limitations exist, and it is important to have them in mind when analyzing the data. SVET detects the currents crossing the plane of measurement but omits the current flowing beneath the plane as well as the current outside the map (Fig. 10a). This is the reason for isolating all samples except the area of interest and for the need to completely map the exposed area. The current density is a 3D vector and SVET measures only one (z) or two (x and z) components. As a consequence, SVET gives an underestimation of the true current. Coatings on metals represent a barrier to the ionic current. If pores and defects are in the micrometer range or smaller, the corrosion cells are too small and the current path does not reach the probe level and is not detected (Fig. 10b). Blistering also hinders the ionic current, and usually the current density flowing outside is too small to be sensed by the SVET probe (Fig. 10c). In addition, systems too fast or too slow may be not adequately characterized by SVET.

22

Scan height

A. Bastos

a

c z

x

b

Fig. 10 Limitations of the use of SVET with coatings

Selected Examples of SVET Measurements on Painted Metals Figure 11 presents a collection of results specially selected to show the possibilities and difficulties of the application of SVET to study painted metals. Figure 11a shows aluminum alloy 2024-T3 coated with a 2 μm film produced by the sol–gel method with (3-glycidoxypropyl)trimethoxysilane and tetra-n-propyl zirconate as precursors, after 1 month of immersion in aqueous 0.05 M NaCl. The sample is still intact, without corrosion, and SVET does not detect any currents. The same system but with a lower curing time (80 min compared to the 17 h of the preceding sample) shows fast corrosion (map obtained after 2 days of immersion) owing to an incomplete film formation and consequent deficient barrier to the testing solution – Fig. 11b. The dark areas in the optical picture correspond to cathodic regions in the SVET map. Only one anodic point is detected, with higher current density. The system depicted in Fig. 2c is another sol–gel film, based on methacryloxypropyltrimethoxysilane + tetraethylorthosilicate + phenyltrimethoxysilane (5:1:4) with 3 % of benzotriazole as corrosion inhibitor, with an artificial scribe exposing the AA2024-T3 substrate. In the first days of immersion, corrosion was restricted to the scribe indicating that the film was adherent enough to prevent film delamination from the scribe. After 1 month, corrosion was able to progress beneath the film. White corrosion products appeared on the scribe and many dark areas on the remaining surface. SVET measured positive currents above the defect and negative currents in points where the film was ruptured allowing the passage of ionic currents in an amount enough to be detected by SVET. The fact that these films are transparent is fortunate because it is then possible to confront the corrosion at the metal surface with the ionic currents in solution above the paint. Corrosion may exist underneath the film and not be detected by SVET. This is still useful for it gives information about the low porosity and lack of defects on the film. Another way of producing artificial defects is by making round-shaped defects with a needle. Often only two defects are produced and it has been observed the tendency for one becoming anodic and the other cathodic – Fig. 11e. The separation of anodic and cathodic activities in the defects has been recently investigated (Bastos et al. 2016). Sometimes no currents are detected. In Fig. 11f two defects were produced on a system consisting of AA2024-T3 alloy with an anodized layer, a sol–gel film as pretreatment and an opaque TiO2 pigmented topcoat. The map was measured after 118 days of

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . . 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2

a

1 mm

8

40

1 mm 8 7 6 5 4 3 2 1 0 -1 -2 -3 -4 -5

500 μm

40

1 mm

1 mm

h

38 33 28 23 18 13 8 3 -2

500 μm 1 mm

0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8

f

g

-4

27 23 19 15 11 7 3 -1 -5

1 mm

1 mm

e

500 μm

1 mm

d

10 8 6 4 2 0 -2 -4

c

1 mm

10 8 6 4 2 0 -2 -4 -6 -8

b

1 mm

1 mm

23

20 30

3

4

10

10

30 30

1 mm

1 mm

Fig. 11 Examples of SVET results on coated systems: (a) intact sol–gel film on AA2024-T3 after 1 month of immersion in 0.05 M NaCl; (b) same sol–gel with incomplete curing after 2 days of immersion; (c) sol–gel film with inhibitor (benzotriazole) and a scribe after 7 days of immersion; (d) same sample after 1 month of immersion; (e) sol–gel film of example (a) with two round-shaped defects after 20 h in 0.05 M NaCl; (f) epoxy coat on anodized AA2024 with two artificial defects after 118 days in 0.5 M NaCl; (g) epoxy clearcoat applied on pure zinc with two artificial defects after 12 h in 0.05 M NaCl; (h) silane film applied on tinplate after 24 h in 0.05 M NaCl. The unit of the current density in all maps is μA cm2

immersion in 0.5 M NaCl. In spite of the corrosion products at the defects and the strong blistering, no currents were detected in the SVET map. The activity was occurring mainly inside the blisters, resembling case (c) in Fig. 10. In addition, due to the height of the blisters (~0.5 mm), the probe was too far from the source (metal surface) to detect meaningful current. There are, however, cases in which SVET is able to detect current above blisters, and Fig. 11g shows one example. It depicts an epoxy clear coat applied on zinc. The system was not optimized for zinc, and extensive delamination with many blisters appeared in less than 12 h, demonstrating

24

A. Bastos

the incompatibility between paint and substrate. SVET detected anodic and cathodic activity above the defects and also many small points of cathodic activity (blue points in the map), coincident with the blisters. Addition of a pH indicator (phenolphthalein, normally colorless but red at pH > 8) confirmed the alkaline environment inside the blisters, due to the cathodic process. A final example is shown in Fig. 11h and concerns a 400 μm thick steel sheet coated by 0.4 μm thick tin coat and 1–2 μm hybrid silane film, immersed in 0.05 M NaCl. Despite the two layers of protection, steel corrosion appeared in many small spots on the surface as early as 24 h. The SVET is a noninvasive technique that provides a whole image of the corrosion process, identifying the anodic and cathodic regions, their current magnitudes, and evolution in time. This is unique as no other technique is able to provide such an information. It says nothing about the chemical species involved in the interfacial processes or in the measured currents. This extra information can be given by SIET or SECM (scanning electrochemical microscopy). Used together these techniques complement each other giving a pretty good description of the corrosion process from the solution side.

Scanning Ion-Selective Electrode Technique (SIET) SIET is the acronym of Scanning Ion-Selective Electrode Technique, which is the trade name of one experimental set-up to perform measurements with potentiometric microelectrodes. It can be used alone or as a complement to SVET identifying some of the ions involved in the corrosion process. The most common case is H+(aq). In corrosion, the cathodic reactions (2) and (3) increase the pH and the hydrolysis of metal cations produced in the anodic reaction (1) decreases the pH according to: Mnþ ðaqÞ þ mH2 OðlÞ ! MðOHÞm ðnmÞþ ðaqÞ þ mHþ ðaqÞ

(6)

Local acidification appears when the hydrolysis occurs in confined places with restriction to the flow of solution. Other important ions that can be measured with potentiometric microelectrodes are the cations from the metal dissolution, the electrolyte ions (Na+, K+, Ca2+, Cl, SO42, NO3), and some corrosion inhibitors added to solution or leached out from the coating.

Potentiometry The theory of ion-selective microelectrodes (ISME) is the theory of potentiometry, probably the most simple and used instrumental technique in chemical analysis. A potentiometric measurement relates one potential difference with the activity of a particular ion of interest, called target ion or primary ion. The measurement is done with two electrodes, one reference electrode, typically a silver-silver chloride electrode, and a sensing electrode (the ion-selective electrode, ISE), which is also a silver-silver chloride electrode with a sensitive and selective membrane at the end

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

a

b E/V

Volt Meter

Reference Electrode

25

Linear range

Sensing Electrode Slope (E vs -log ai)

DL

7

5

d X-

M+

1 mV

4 3 2 1 -log ai or -log ci

ΔE

E/V

c

6

Δt

M2+ M3+

59 mV

-log ai

t (ΔE/Δt)

time

Fig. 12 Potentiometric cell and potentiometric responses

(Morf 1981). That membrane can be a special glass or made with polyvinyl chloride (PVC). The potential difference between the two electrodes is measured with a high impedance voltmeter in zero current condition (virtually no current passing in the cell). This ensures the cell is at equilibrium and no chemical reactions occur. However, a minute current is always necessary for a reading to take place. Figure 12a is a representation of such electrochemical cell. The two electrodes can be combined in a single body, as is typical for commercial pH electrodes. A representation of the cell is

Ag | AgCl | KCl || test solution || Membrane | internal solution | AgCl | Ag Reference electrode (RE)

Ion-selective electrode (ISE)

The cell potential, Ecell, is the difference between the potential of the ISE internal reference, Eref(int), and the potential of the reference electrode, Eref, plus the membrane potential, Emem, and the liquid junction potential, Elj, established at the interface between the reference electrode and the test solution:

26

A. Bastos

Ecell ¼ Eref ðintÞ  Eref þ Emem þ Elj

(7)

In general all terms are constant except Emem, which is sensitive to the differences in activities of the target ion in the internal and external solutions. The cell potential is then described by the Nernst equation Ecell ¼ Econst: þ

R T aext ln z F aint

(8)

where aext is the activity of the target ion in the external or testing solution, aint is the activity of the target ion in the internal solution, R is the gas constant (8314 J K1 mol1), T is the temperature (K), z is the charge of the target ion, F is the Faraday constant (96485 C mol1), and Econst is the sum of all terms of Eq. 7 except Emem. Equation 8 shows that the cell potential changes linearly with the logarithm of the activity of the target ion in the external solution – Fig. 11b. This type of graph is called calibration plot. Using decimal logarithms, which is more intuitive and used in practice, the theoretical (nernstian) slope is 2.303 RT/zF (or 59.16 mV/z) per decade of activity of the target ion at 25  C. Figure 11c shows responses for ions with charges 1, +1, +2, and +3. Potentiometric measurements are preceded by a calibration step where the cell potential is measured for a set of solutions of known activity. The obtained Ecell versus log aext function is used to determine the activity of the target ion in a new solution from the potential recorded in the cell. It is possible, and many times preferred, to calibrate using solutions of known concentration rather than activity. It is then important to use calibration solutions with ionic strength similar to that of the testing solution. The ionic strength of an electrolyte solution reflects the amount of charges and interionic interactions in solution. It is defined as I¼

1X Ci z2i 2

(9)

where Ci is the concentration of ion i in solution and zi is its charge. Activity can be defined as the effective concentration of a chemical species. At very dilute solutions, C < 104 M, the ionic strength is low, the interionic interactions are minimum, and activity approximates concentration. As the ionic strength increases, so do the interionic interactions and activity deviates from concentration. The two are related by ai ¼ γ i

Ci Ci 

(10)

with γ i being the activity coefficient of ion i and Ci the concentration of a standard state. Activity is thus a nondimensional quantity. The activity coefficient can be calculated by the extended Debye-Huckel equation with the modifications by Robinson and by Guggenheim and Bates (Dean 1999):

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

logγ i ¼

z2i

! pffiffi A I pffiffi  bI I 1 þ B r eff i

27

(11)

where A and B are constants dependent of temperature and dielectric constant of the solvent, reffi is the approximate effective ionic radius of ion i and b = 0.2 at 25  C. The values of A, B, and reffi can be found in tables (Dean 1999). Other versions of Eq. 11 exist (Robinson and Stokes 1959; Wright 2007), and a comparison of different methods to determine ionic activities in natural waters can be found in Appelo and Postma (2005) and Bethke (2008). It is important to note that ISEs offer one of the few ways to access directly the activity of free ions in solution, in contrast with the majority of analytical methods which measure the total concentration.

ISE Characteristics Important ISE characteristics are (i) linear response range of calibration curve, (ii) detection limit, (iii) slope, (iv) selectivity to foreign ions, (v) response time, (vi) potential stability, (vii) working pH range, (viii) lifetime, and (ix) membrane resistance. They are now briefly introduced based on IUPAC recommendations (Buck and Lindner 1994). (i) Linear response range. The response of any ISE exhibits linear response in a certain concentration range, usually between 101 and 106 M. The practical use of ISE coincides with this region. (ii) Detection limit (DL). This is the lowest ion activity that can be detected with confidence. In practice it is determined by the intercept of the two asymptotes of the calibration response curve (Fig. 11b). If measurements below the detection limit are problematic (too long response time, bad reproducibility, etc.), a line parallel to the x-axis should be drawn through the mean potential values of the solution of lowest concentration in the linear range. (iii) Slope. This is an indication of the sensitivity of the technique, a measure of the ability to discriminate between small differences in the activity of the target ion. Ions with higher charges have lower slopes (Eq. 8). (iv) Selectivity. Refers to the electrode ability of responding only to the target ion (primary ion). This is an ideal situation, and under most conditions interfering ions (ions giving the same response as the target ion) are present in solution contributing to the electrode response and adding to the true activity of the primary ion. In the presence of interferents the electrode potential can be described by a modified Nernst equation, the Nickolsky-Eisenman equation: x X RT zi =zj E ¼ Econst: þ 2:303 log ai þ K pot i , j ð ai Þ zF j¼1

! (12)

28

A. Bastos

where z is the charge of the ion, the subscripts i and j are for the primary and interfering ions, respectively, and Kpot i,j is the potentiometric selectivity coefficient, a quantitative measure of the ability of the ISE to distinguish the target pot ion from interfering ions. The smaller is Ki,j the higher the selectivity of the electrode, i.e., the preference for the primary ion. The selectivity coefficients can be determined by the following methods: (a) Fixed Interference Method (FIM). The cell potential is measured in solutions with fixed activity of the interfering ion and varying activity of the primary ion. The potential values obtained are plotted versus the logarithm of the activity of the primary ion. The intersection of the extrapolation of the linear portions of this plot indicates the value of ai to be used to calculate the potentiometric selectivity coefficient with K pot i, j ¼

ai z =zj

aj i

(13)

(b) Separate Solution Method (SSM). This is a simple and fast alternative to FIM. The potential of the cell is measured in separate solutions of i and j with the same activity. The potentiometric selectivity coefficient is calculated from: 

log K pot i, j

   E j  Ei z i F zi þ 1 ¼ logai 2:303 R T zj

(14)

The FIM is preferable since its conditions are closer to real situations, when both ions coexist in the same solution. On the other hand, the SSM is simple and fast, especially suitable for comparing the selectivity of a novel ISE based on different membrane compositions. It must be stressed that other methods exist (Umezawa et al. 1995; Bakker et al. 2000; Egorov et al. 2014), and the determination of meaningful and reproducible selective coefficients is one of the most difficult and complex tasks in potentiometry. (v) Response time. The response time, τlim, is the time elapsed between the instant an ISE and a RE are brought into contact with a new sample solution and the instant at which the measured potential reaches a limiting value or a selected ΔE/Δt slope (e.g., 1 mV/min) (Fig. 12d). Other values are τ95 and τ90, the time needed to obtain 95 % or 90 % of the total potential jump from one solution to the other. A critical analysis can be found in Maccá (2004). (vi) Potential stability. The potential stability of ISEs is defined by the potential drift, potential reproducibility, and hysteresis. Drift is the slow nonrandom change with time in the potential of an ion-selective electrode cell assembly in a solution of constant composition and temperature. Potential reproducibility is the standard deviation of the cell potential in a series of measurements in solutions of different concentrations. Hysteresis, or electrode memory, occurs

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

29

when there is a difference between the first observed value of cell potential in a solution of the target ion and a second recorded value of the cell potential in the same solution after exposing the electrode to a different concentration of the target ion. (vii) Working pH range. ISE may respond differently at different pH. This must be characterized. Also the target ion may react and form complexes or precipitates depending on the sample pH. (viii) Lifetime. This is defined as the time elapsed from the ISE first use until the moment it loses functionality. (ix) Membrane resistance. Most ISEs are based on PVC membranes with resistances of 107–109 Ω. Ion-selective microelectrodes usually show resistances on the order of 1010–1012 Ω. These values are very high and require measuring devices with higher internal resistance for a reading with a small loading error. This error (difference between the measured and the true value) can never be eliminated but can be reduced to insignificant proportions. Equation 15 shows that the error is a relation between the cell resistance, Ecell, (dominated by the ISE membrane resistance) and the meter resistance, Emeter (Skoog et al. 2014), Errorð%Þ ¼ 

Rcell  100 Rmeter þ Rcell

(15)

A meter having an internal resistance of the same magnitude as the potentiometric cell gives a reading that is 50 % of the true value. The reading becomes 91 % for a meter resistance 10 times higher than the cell resistance and 99.9 % for 100 times the cell resistance. The importance of a high-resistance meter or a low-resistance membrane for reliable readings is evident.

Ion-Selective Microelectrodes The theory just described is valid for ISMEs as it is for macroelectrodes, the main differences being imposed by the miniaturization (Ammann 1986). The electrical resistance is inherently high due to the small opening of the tip. Potentiometric microelectrodes can be found in the literature under different names: potentiometric microscopy (Horrocks et al. 1993; Klusmann and Schultze 1997), Local Ion Concentration Technique (LICT) (Maile et al. 2000), Scanning Ion-Selective Technique (SIET) (Ding and Hihara 2005; Kunkel et al. 2006; Lamaka et al. 2010), Microelectrode Ion Flux Estimation (MIFE) (Newman 2001), and SelfReferencing Ion-Selective probes (SERIS) (Doughty and Langton 2001). The name adopted in this chapter (SIET) is simply the tradename used by the manufacturing company of the system employed by the author. What will be described in the next pages is common to all systems.

30

A. Bastos To amplifier

a Glass micropipette

b Ag wire with AgCl coat

Reference electrode 1 cm

Preamplifier

for 3D motor positioning system

Electrical grounding

Internal solution Sensitive cocktail

20-100 µm

1 cm

Fig. 13 (a) Liquid membrane ion-selective microelectrode and (b) typical measuring cell

Constitution of an ISME Figure 13a shows the most common type of ISME. It consists of a glass micropipette with an ion-selective viscous membrane at the tip to provide the potentiometric response, filled with an internal solution containing a fixed activity of the target ion and sometimes reagents to adjust pH and ionic strength. A Ag|AgCl wire in a Teflon support with rubber O-ring for capillary clamping is introduced from the back side of the micropipette and serves as internal reference. The electrode is inserted in a preamplifier (1015 Ω) together with a reference electrode to close the circuit – Fig. 13b. This electrode set-up resembles the one in Fig. 12a. The preamplifier moves with the ISME during measurements. The proximity of the microelectrode to the measuring device is important because the extremely low currents involved in the measurement are easily affected by stray currents. The great tendency for noise is minimized by short distance cabling. Sometimes reference electrode and ISME are constructed in the same body using double tip microelectrodes produced from two barrel capillaries.

Cocktail Composition The key component of a liquid membrane ISME is a sensitive cocktail at the tip of the electrode separating the outer and inner solutions. It contains three main components: ionophore (5–10 wt%), lipophylic ionic additive (1–20 wt%), and solvent (60–90 wt%). The ionophore, also called mobile carrier, is the ion-selective substance, responsible for the membrane potential response. It is a chemical compound capable of selectively and reversibly transporting the target ion across the membrane. It should be lipophilic to prevent leaching into solutions. Chemically it may be an ion

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

31

exchanger, a charged complexing agent, or a neutral carrier (with uncharged complex ligands). The lipophylic ionic additive or ionic site is necessary for the ionic conductivity of the membrane. It is an ion-exchanger and reacts with lipophilic ions that could compete with the target ion in the interaction with the ionophore. Being an ion exchanger it may act as ionophore in the absence of other, more selective, ion-carriers. The Solvent is the medium where all other ingredients are mixed. It must be viscous and hydrophobic to remain at the tip of the glass microcapillary. The most common are plasticizers used in the fabrication of conventional PVC macro ISEs. The solvent influences the dielectric constant of the membrane, the mobility of the ionophore, and the reactivity of the ligands. It can substantially affect selectivity, stability, and lifetime of the sensors. Table 2 presents the composition of some ISME, and Table 3 shows some of the characteristics of their response. The chemical structures of the compounds in the cocktails are depicted in Fig. 14. Table 2 Cocktails composition and construction of ISMEs Cocktail composition

ISME construction LIX column length/μm Internal solution 25–30 0.01 M KH2PO4 + 0.1 M KCl 25–30 0.01 M KH2PO4 + 0.1 M KCl 20–30 0.1 M NaCl + 0.01 M NaH2PO4

Ion H+ (a)

Ionophore 10 wt.% Tridodecylamine

Plasticizer 89.3 wt.% o-NPOE

Ionexchanger 0.7 wt.% KTClPB

H+ (b)

6 wt.% 4nonadecylpyridine 5 wt.% 5,10,15,20tetraphenyl21H,23H-porphyrin manganese(III) chloride 10 wt% N,N0 -Dibenzyl-N, N0 -diphenyl-1,2phenylenedioxydiacetamide 5 wt% Valinomycin

93 wt.% o-NPOE

1 wt.% KTClPB

5 wt.% 1-decanol 89 wt% o-NPOE

1 wt.% TDDATClPB

89.5 wt% o-NPOE

0.5 wt.% NaTPB

50

0.1 M NaCl + 0.01 M NaH2PO4

93 wt% 1,2 dimethyl3-nitrobenzene 62 % o-NPOE 25 % chloroparaffin

2 wt.% KTClPB

200–300

0.01 M KCl

3% KTClPB

35–40

0.1 M MgCl2

Cl (c)

Na+ (d)

K+ (e)

Mg2+ (f)

10 wt% N,N00 -octa methylenbis (N0 -heptylN0 -methylmethylmalonamide)

(continued)

32

A. Bastos

Table 2 (continued) Cocktail composition

Ion Ca2+ (g) Zn2+ (h)

Ionophore 10 wt% Calimycin 7 wt.% tetra-n-butyl thiuram disulfide

Plasticizer 89 wt.% o-NPOE 68.8 % o-NPOE

Ionexchanger 1% NaTPB 22.8 wt.% NaBARF 1.4 wt.% TDDATCPB

ISME construction LIX column length/μm Internal solution 20–30 0.1 M CaCl2 60–70

0.1 M KCl + 0.01 M KH2PO4 + 105 M ZnCl2

Other ISME H+ (i) Solid contact electrode: membrane produced with 3 wt.% 4-nonadecylpyridine + 0.5 wt.% p-(MMA-DMA) + 50.75 wt.% KTClPhB + solvent; internal solution replaced by conductive polymer (POT) Boron-doped diamond electrode: Diamond film grown on tungsten wire by Chemical Vapor H+ (j) Deposition in a CH4-H2 atmosphere with B2O3 as doping source H+ (k) Ir|IrOx: pure iridium wire submitted to several thousand cycles in 0.5 M H2SO4, from 0.25 to +1.25 V vs. SCE at 3 V/s Cu|CuSe: layer of CuSe formed on the copper surface by cathodic deposition of selenium Cu2+ (l) from 0.1 M Na2SeO3 with pH adjusted to about 6 with sulfuric acid p-(MMA-DMA) poly(methylmethacrylate-decylmethacrylate) copolymer; POT poly(3-octhylthiophene2,5-dyil) (a) Selectophore Hydrogen ionophore I – Cocktail B (95293), (b)Selectophore Hydrogen ionophore II – Cocktail A (95297), (c)Selectophore Chloride ionophore I – Cocktail A (24902), (d)Sodium ionophore II – cocktail A (71178), (e)Potassium ionophore I – cocktail B (60398), (f)(Lamaka et al. 2008), (g)Calcium ionophore I – Cocktail A (Selectophore), (h)(Bastos et al. 2010a), (i)(Taryba and Lamaka 2014), (j)(Silva and Lamaka 2014), (k)(El-Giar and Wipf 2007), (l)(Papeschi et al. 2000)

Fabrication of ISME The cocktails can be produced in the laboratory, but many compositions are available commercially (Selectophore, Ion sensor materials, Sigma-Aldrich, 2011). The fabrication of liquid membrane ISME follows a sequence of steps: (i) pulling the glass microcapillaries, (ii) silanization, (iii) filling with internal solution and sensitive cocktail, (iv) calibration, and (v) testing. Each step may vary from laboratory to laboratory but the basic procedure is as follows. The first step is the production of micropipettes from glass capillaries. Usually single-barreled borosilicate glass capillaries 10 cm long and outer diameter of 1.5 mm are stretched in a pulling machine, resulting in micropipettes about 5.5 cm long with the last 0.5 cm of one end thinned to a 2 μm tip. The glass surface is highly hydrophilic – up to 4.6 free hydroxyl groups per 10 nm2 of glass (Ammann 1986) – and water rapidly displaces the membrane cocktail from the tip. To avoid it, the glass must be hydrophobic which is accomplished by reacting the hydroxyl groups with an organosilicon compound (Ammann 1986). A common procedure is to put a number of micropipettes (~100) in a grid holder inside an oven at 200  C for a few hours (to remove water and fats from the glass surface) and then inject 200 μL of N,

Application of SVET/SIET Techniques to Study Healing Processes in Coated. . .

33

Table 3 Response of the ISME in Table 2 Ion H+(a) H+(b) Cl(c)

K+(e)

Linear range 5.5–12.0 2.0–9.0 103–3  101 M 2  103–101 M 104–101 M

Mg2+(f) Ca2+(g)

4.8–1.4 pMg 102–107 M

Na+(d)

Zn2+(h) 3.5–1.5 pZn Other ISME H+(i) 2–12 H+(j) 2–12 H+(k) 2–12 Cu2+ 103–5  105 M

Slope (mV/dec) 58.0  0.4 57.1  0.8 57.5  0.5

Time of response (s) τ90  5 τ90  0.6

54

τ90  3

57.8  1.2

1 μm, and the patterns had feature sizes on the order of 100 μm. A sheet resistance down to 0.4 kΩ/□ and transmittance over 85 % in the visible region were achieved. Heusing et al. prepared ITO thin films on PET and PEN foils by gravure printing to fabricate organic photodiodes (Heusing et al. 2009). The films had sheet resistance down to 1 kΩ/□ and transmittance over 80 % in the visible region.

Liquid Phase Deposition Liquid phase deposition (LPD) is a technique that is based on the precipitation of oxides from supersaturated solutions (Nagayama et al. 1988; Deki et al. 1996a, b). Shimizu et al. utilized the pH dependence of the solubility of titania and deposited anatase thin films on PMMA, ABS, and PC substrates as well as cotton, paper, and wool by immersing the substrates in aqueous solutions of TiF4 at 25–70  C (Shimizu

6

H. Kozuka

et al. 1999). The films dried at 60  C had thickness up to 170 nm and were porous, consisting of anatase particles several tens nm in size. The anatase films had many cracks and were easily peeled off when the substrates were relatively hydrophobic PMMA, ABS, and PC. On the other hand, good adhesion was achieved on hydrophilic cotton, paper, and wool. Ou et al. attempted to prepare titania thin films on hydrophobic polytetrafluoroethylene (PTFE), PE, and PET substrates (Ou et al. 2010). Polydopamine thin films were first prepared on the substrates by dipping them into a hydrochloric acid solution of 3-hydroxytyramine hydrochloride and tris(hydroxymethyl) aminomethane. Then the substrates were immersed in an aqueous solution of (NH4)2TiF6 and H3BO3, which allowed titania nanoparticles several nanometers in size to be precipitated on the surface. The titania films thus obtained had rough surface and some cracks. Goutailler et al. proposed a different reaction scheme for depositing uniform titania films on cellulose fibers (Goutailler et al. 2003). Cellulose fibers were dipped in a hexane solution of Ti(OC3H7i)4, and a boiling aqueous solution of N(C4H9n)4Br was added. Anatase and brookite nanocrystals were precipitated on the cellulose fibers to form crack-free titania thin films, which showed good adhesion due to the hydrophilic properties of the cellulose fibers.

Sol-Gel Process with Posttreatments Various posttreatments are proposed for crystallizing the precursor gel films on plastic substrates. The posttreatments include humid vapor treatment, hot water or hydrothermal treatment, and irradiation with UV or IR lasers as shown in Fig. 2.

Humid Vapor Treatment Imai et al. crystallized alkoxide-derived titania gel films on silicon wafers by exposing them to water vapor at 180  C (Imai and Hirashima 1999; Imai et al. 1997). The films were brittle and were easily delaminated from the substrates. They thought that Ti-OH groups are formed by hydrolysis, which leads to the molecular rearrangement and the formation of anatase nanocrystallites. Langlet et al. prepared anatase thin films on PC and PMMA substrates, stimulated by Imai et al.’s work (Langlet et al. 2002). Alkoxide-derived titania gel films on plastic substrates were heated at 140  C in an autoclave containing a small amount of an ethanol-water mixture, resulting in 160 nm thick anatase thin films with crystallite size of 15–20 nm. The crystallization was accompanied by a slight increase in thickness and by a noticeable increase in porosity. The photocatalytic activity was demonstrated for anatase films on PC substrates.

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

7

Crystalline oxide film Plastics Hot water or hydrothermal treatment Crystalline oxide film

Humid vapor treatment Gel film

Plastics

Plastics

UV-laser irradiation Crystalline oxide film Plastics

IR-laser irradiation Crystalline oxide film Plastics

Fig. 2 Sol-gel method with posttreatments for fabricating crystalline oxide thin films on plastics

Hot Water or Hydrothermal Treatment Matsuda et al. and Kotani et al. proposed “hot water treatment,” where alkoxidederived titania-silica gel film are immersed in hot water, resulting in precipitation of anatase nanocrystals (Matsuda et al. 2000; Kotani et al. 2000). They prepared anatase thin films on various plastic substrates including PET, PC, and PMMA by immersing titania-silica gel films in hot water of 90  C (Matsuda et al. 2003). The films obtained were 50–200 nm in thickness, consisting of 30–50 nm anatase nanocrystals, and were rather porous, which are beneficial for their photocatalytic activity. The hydrolysis of Si-O-Ti bonds, the dissolution of silica, and the migration of hydrolyzed titania species were thought to induce the precipitation of anatase nanocrystals. The same group also prepared pseudoboehmite thin films on plastic substrates by hot water treatment (Tadanaga et al. 2003; Yamaguchi et al. 2006). Alkoxide-derived gel films were immersed in hot water of 60  C, resulting in flowerlike pseudoboehmite thin films, which were thought to be formed by dissolution and reprecipitation process. The films showed good adhesion to PET and PC substrates while poor adhesion to PMMA substrates. The adhesion to PMMA substrates could be improved, however, by plasma treatment. The flowerlike pseudoboehmite films on the plastic substrates had antireflective properties. Superhydrophobic nature was given by depositing partially hydrolyzed heptadecafluoroalkyltrimethoxysilane on the flowerlike films. Hashizume and Hirashima subjected alkoxide-derived titania gel films on PI films to hydrothermal treatment at 150  C (Hashizume and Hirashima 2012). The films

8

H. Kozuka

obtained were composed of nanometer-sized granules, and the surface was smooth but porous. The film/PI adhesion was much improved by soaking the PI films in aqueous solution of NaOH, which cleaved the imide rings, generating COOH groups on the surface.

Irradiation with UV and IR Lasers Imai et al. and Asakuma et al. found that the irradiation with UV lasers induces the crystallization of gel films (Imai et al. 1998, 1999b, c, Asakuma et al. 2000). They utilized this technique to prepare ITO thin films on PET, PI, PC, and PEEK sheets (Asakuma et al. 2003). Alkoxide-derived gel films were irradiated with an ArF excimer laser beam after being dried at 100  C, where the laser was operated at 1 Hz with 19 nm pulse width, and the fluence was adjusted at 20 mJ cm 2 per shot. The 95 nm thick ITO thin films on PET and PI sheets had resistivity of 6.3  10 2 and 6.2  10 2 Ω cm, respectively. The ITO films on PC and PEEK sheets, on the other hand, did not exhibit electrical conductivity because of the damage of the plastic substrates by the laser irradiation. Kim et al. utilized a deep-UV (DUV) irradiation technique for preparing amorphous and semiconducting In2O3, indium gallium zinc oxide (IGZO), and indium zinc oxide (IZO) thin films on plastic substrates (Kim et al. 2012). They prepared the precursor film from nitrate, acetate, and chloride solutions, and DUV irradiation was conducted using a mercury lamp. The DUV irradiation promoted the formation of metalloxane bonds as well as the film densification, resulting in thin-film transistors on 200 μm thick polyarylate (PAR) films. Königer et al. irradiated ITO thin films on PET films with CO2 laser (Königer et al. 2010). The PET films were treated with plasma, and the films were prepared by a doctor blade using a dispersion of ITO nanoparticles ca. 100 nm in size. The CO2laser irradiation slightly induced sintering and improved the electrical conductivity. Three micrometer thick ITO films with a sheet resistance of 0.4 kΩ/□ and an 80 % transmission in the visible range were obtained. Salar Amoli et al. irradiated ITO thin films on PC substrates with pulse Nd-YAG laser (Salar Amoli et al. 2012). They prepared the precursor films using a suspension of ITO nanoparticles that were modified by 3-methacryloxypropyltremethoxysilane. A 350 nm thick ITO thin film on a PC substrate had X-ray diffraction (XRD) peaks of crystalline ITO, a sheet resistance of 0.6 kΩ/□, and optical transmittance of 50–70 % in the visible region.

Self-Combustion of Solution-Derived Films The following techniques based on self-combustion have potential for the preparation of oxide thin films on plastics. Kim et al. proposed a technique based on selfcombustion of solution-derived thin films to prepare electrically conductive thin films at temperatures as low as 200  C (Kim et al. 2011). In this technique, metal

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

9

nitrates act as “oxidizers” and AcAc or urea as “fuels,” generating significant heat around 200  C. In2O3 and ITO thin films were prepared from solutions containing nitrates and AcAc, crystallized and significantly increased in electron mobility at temperatures as low as 200  C by self-combustion. They also prepared amorphous ZTO (Zn0.3Sn0.7O1.7) and IZO (In0.7Zn0.3O1.35) thin films, which exhibited a significant increase in mobility when heated up to around 200–225  C. Amorphous indium yttrium oxide (a-IYO) thin films were also prepared from AcAc-containing nitrate solutions, exhibiting a mobility as high as 5.0 cm2V 1s 1 when heated at 250  C (Hennek et al. 2012). Kim et al. prepared 60 nm thick crystalline In2O3 thin films on silicon wafers at 150  C using the precursor solutions containing “oxidizers” and “fuels” as the binders for oxide nanoparticles (Kim et al. 2012b). Although not being applied to the preparation of oxide thin films on plastics, these techniques have potential.

Sol-Gel Transfer Technique The sol-gel transfer technique that we have proposed is different from the above solution process in that film crystallization is guaranteed by a firing process (Kozuka et al. 2012a, b, Kozuka 2013, Kozuka et al. 2013, 2015, 2016). Figure 3 illustrates the procedure of the technique. First a silicon substrate (mother substrate) is coated with polyvinylpyrrolidone (PVP), PI, or their mixture, which acts as a release layer. The release layer 1–2 μm in thickness is prepared by spin-coating, followed by heat treatment under prescribed conditions, which makes PVP and PI-PVP release layers partially decomposed while no decomposition for PI release layers. Then the precursor gel film is deposited on the release layer by spin- or dip-coating, followed by firing for crystallization (Fig. 3a). Finally the crystallized film is transferred to the plastic substrate surface (Fig. 3b, c). The release layer allows the liftoff of the film from the mother substrate. The film transfer is performed with the use of adhesives (Fig. 3b) or without it (Fig. 3c) as will be described later.

Transfer Process Using Adhesives Anatase thin films were fabricated on PC substrates in the following manner (Kozuka 2013; Kozuka et al. 2012a, 2015). An alkoxide-derived titania gel film was deposited on the PVP layer on a silicon substrate, followed by heating up to 500  C to obtain an anatase film. An epoxy resin adhesive was pasted on the anatase film, and a PC plate was mounted on it. After 2 days, the PC plate was detached from the silicon substrate, where the anatase film was transferred to the PC plate (Fig. 4a, b). No cracks were formed during the transfer process (Fig. 4c). The transferred anatase film had a brown-colored, partially decomposed release layer on its surface (Fig. 4c), which could be easily removed by a cellophane adhesive tape, leaving a colorless, transparent anatase film on the PC substrate (Fig. 4d). The anatase film on the PC substrate was optically transparent as is evidenced in the

Precursor film

a

Firing (> 500oC) Si(100)

Ceramic (crystallized) film

Si(100)

Plastics

Si(100)

Heating & softening

c

Si(100)

Plastics

Pressing

b

Si(100)

Plastics

Pressing Adhesive

Si(100)

Ceramic film

Plastics

Product

Si(100)

Ceramic film

Plastics

Product

Fig. 3 Sol-gel transfer technique that is proposed by the author’s group. (a) Process for converting the precursor gel film into a crystalline oxide film on the release layer on a silicon substrate. (b) Process for transferring the fired film onto a plastic substrate using an adhesive. (c) Process for transferring the fired film onto a plastic substrate without using adhesives, where the softened or molten plastic substrate surface acts as an adhesive

Si(100)

Release layer

Adhesive

Plastics

10 H. Kozuka

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

a

Si(100) substrate

b

11

PC substrate

Araldite® TiO2 (fired) PVP (fired)

Cellophane adhesive tape

PC substrate

c

d

0.5 mm Fig. 4 Transfer of a fired titania thin film from a PVP-derived release layer on a silicon substrate to a 0.5 mm thick PC substrate. (a) Photograph showing the mother silicon substrate, and the PC substrate on which the fired titania thin film was transferred. (b) Schematic illustration showing the removal of the release layer from the titania film surface by a cellophane adhesive tape. (c) Optical micrograph of the titania film transferred to the PC substrate. (d) Titania film on the PC substrate after the removal of the release layer

optical absorption spectra (Fig. 5a) and exhibited high optical reflectance due to its high refractive index (Fig. 5b). The anatase film on the PC substrate could be bent to some extent without cracking as is seen in Fig. 5c. ITO thin films were prepared on PMMA substrates. The precursor film was prepared by spin-coating on the PI release layer on a silicon substrate using an indium nitrate and tin(II) chloride solution, followed by firing at 500  C. The spincoating and the firing were repeated eight times, resulting in a 660 nm thick ITO film. An epoxy resin adhesive was used for the transfer, and a transparent ITO film on the PMMA substrate was obtained (Fig. 6a). The ITO film on the PMMA substrate was crack-free and transparent (Fig. 6b). The film on the PMMA substrate exhibited XRD peaks (Fig. 6c) thanks to the repeated firing at 500  C. The film exhibited electrical conductivity on the PMMA substrate as is evidenced in Fig. 6d, its resistivity being 8.69  10 3 Ω cm. Solution-derived ZnO thin films are known to be crystallographically oriented easily even on glass substrates (Ohyama et al. 1997, 1998). Utilizing such nature, a crystallographically oriented oxide ZnO thin film was fabricated on a PMMA substrate. The precursor film was prepared by dip-coating on a PI release layer on a silicon substrate, using a zinc acetate solution. Dip-coating and calcination at

12

H. Kozuka

b 100

80

Bare PC substrate

60

PC substrate coated with TiO2

80 Reflectance (%)

Transmittance (%)

a 100

40

PC substrate coated with TiO2

40 20

20 0 300

60

400

500

600

700

800

Wavelength / nm

0 300

Bare PC substrate 400

500 600 700 Wavelength / nm

800

c

PC substrate coated with TiO2

500 mm

Fig. 5 (a) Optical absorption spectra and (b) reflection spectra of the TiO2 film transferred on a 0.5 mm thick PC substrate using an adhesive. The absorption spectra were measured with a bare PC substrate as the reference. The spectra are also given for the bare PC substrate in (a, b). (c) Bending test and optical micrograph taken after the bending test for the TiO2 film on the PC substrate, where the TiO2 film is not cracked even at a radius of curvature of 20.3 mm

300  C were repeated three times, and the film was fired at 600  C, resulting in a 85 nm thick, (002) oriented ZnO film. The oriented ZnO film was transferred to a PMMA substrate using an epoxy resin adhesive. The transferred film was crack-free and optically transparent as is revealed in Fig. 7a. The (002) orientation of the ZnO film on the PMMA substrate is evidenced in the XRD pattern (Fig. 7b).

Transfer Process Without Adhesives We proposed a technique that realizes direct bonding between crystalline oxide thin films and plastic substrates without adhesives (Kozuka 2013, 2016; Kozuka et al. 2012b, 2013, 2016). As schematically illustrated in Fig. 3c, the transfer is achieved by softening or melting the plastic substrate surface. The oxide film on the release layer is transferred either in a near-IR image furnace or on a hot plate. The transfer in

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics 100

a

Acrylic resin substrate

b

80 60 40 20 0 300

222

c

400 500 600 700 Wavelength / nm

440 611 622

332 431

400

Acrylic resin substrate

ITO film

10

800

d

211

Intensity (a.u.)

Transmittance (%)

ITO film

13

20

30 40 50 2θ / deg. (Cu Kα)

60

Fig. 6 (a) Appearance, (b) optical absorption spectra, (c) XRD pattern, and (d) demonstration of electrical conduction for the ITO film transferred on a 3 mm thick PMMA substrate using an adhesive. The absorption spectra were measured with a bare PMMA substrate as the reference. Miller indices noted in c are those of ITO with bixbyite structure, and a halo observed at 2θ around 14o is originated from the PMMA substrate

b

ZnO film

ZnO (002)

Intensity (a.u.)

a

Acrylic resin substrate

20

30 40 50 2q/ deg. (Cu Kα)

60

Fig. 7 (a) Appearance and (b) XRD pattern of the ZnO film transferred on a 3 mm thick PMMA substrate using an adhesive

14

H. Kozuka

a near-IR image furnace is achieved by irradiating the plastic-substrate/fired-oxidefilm/release-layer/silicon-substrate stack in the image furnace, where the silicon substrate absorbs the near-IR light, radiating the heat (Yamano and Kozuka 2007, 2008), which is conducted to the plastic substrate (Fig. 8a). The transfer can also be conducted by heating the fired oxide film on a hot place and placing and pressing a plastic substrate on the film (Fig. 8b). The plastic substrate can also be pressed on the film using a roller as schematically shown in Fig. 8c. An alkoxide-derived titania gel film was deposited on a PI-PVP mixture release layer on a silicon substrate, followed by heating up to 600  C to obtain an anatase film 60 nm in thickness. The anatase film was transferred to a PMMA substrate by heating in a near-IR image furnace where the stack was heated at 50  C min 1 up to 170  C. The anatase film transferred onto the PMMA substrate was crack-free and looked white because of the white light reflection (Fig. 9a). The high smoothness of the transferred film surface is evidenced in the scanning probe microscopic (SPM) image (Fig. 9b), where an Ra value as small as 0.62 nm was obtained. The optical transparency and the high reflectivity of the film are demonstrated in the absorption and reflection spectra, respectively (Fig. 9c, d). Using a roller and a hot plate, anatase thin films can also be transferred on bent plastic substrates as shown in Fig. 10. ITO thin films were transferred to PC substrate surface. The ITO thin film was prepared by spin-coating on a PI-PVP release layer under the same conditions as those described in section “Transfer Process Using Adhesives.” The film transfer was conducted either on a hot plate of 200  C or in a near-IR image furnace where the stack was heated at 75  C min 1 up to 200  C. The electrical conductivity and optical transparency of the ITO film on the PC substrate are demonstrated in Fig. 11a, b, respectively. The smooth surface of the transferred film is seen in the SPM image (Fig. 11c), where an Ra value as small as 15 nm was obtained. The SEM image of the cross section fabricated by focused ion-beam machining revealed the high smoothness for the film/substrate interface as well as for the film surface (Fig. 11d). Patterned crystalline oxide thin films can also be prepared on plastic substrates. Figure 12 illustrates the process. The release layer and the precursor gel film are deposited on a silicon substrate that has periodic grooves. After being fired for crystallization, the oxide film is heated on a hot plate, and the plastic substrate is placed under a load. The plastic substrate surface that is in contact with the oxide film is molten or softened, leading to the formation of patterned oxide thin films (oxide thin ribbons) on the plastic substrate. The optical micrograph of the patterned ITO thin films thus prepared on a PC substrate is shown in Fig. 13a. ITO thin films 50 μm in width are aligned parallel on the PC substrate. In the transfer process, the softened PC substrate slightly intrudes into the grooves of the silicon substrate. As a result, periodic PC ridges ca. 3 μm in height were formed (Fig. 13b), leaving the ITO ribbons in the valleys of the PC substrate. The SEM pictures of the ITO ribbons on the PC substrate reveal that the cross section is slightly wavy, and the ribbons are buried in the PC substrate (Fig. 13c, d). Once oxide ribbons are formed on a plastic substrate, periodic ridges are generated between the ribbons. Another oxide thin film can be transferred to the periodic

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

15

a Pressing

Melting or softening

Plastics Crystalline oxide film

Release layer Si(100) Heat

Near IR light

b Plastics

Pressing Crystalline oxide film

Plastics Plastics

Melting

Release layer Si(100)

Heat Hot plate

Si(100)

Heat Hot plate

Crystalline oxide film

Si(100)

Hot plate

c 6.4 kg Plastic substrate

Fired oxide film Silicon substrate

Hot plate (160 - 200oC)

Fig. 8 Schematic illustration showing the transfer process without using adhesives. (a) Softening the plastic substrate surface in a near-infrared image furnace, where the silicon substrate absorbs the near-IR light and radiates the heat, which is conducted to the plastic substrate. (b) Softening the plastic substrate surface using a hot plate, where the heat is conducted from the silicon substrate side to the plastic substrate. (c) Transfer using a roller and a hot plate

16

H. Kozuka

Fig. 9 (a) Appearance of the anatase film on a 5 mm thick PMMA substrate. (b) SPM image of the surface of the anatase film transferred on a 5 mm thick PC substrate. (c) Absorption and (d) reflection spectra of the anatase film on a 5 mm thick PMMA substrate. The anatase films were 60 nm in thickness. The transfer was conducted in the near-IR image furnace with heating up to 170  C at 50 and 75  C min 1 for the PMMA and PC substrates, respectively Fig. 10 Anatase thin film transferred to a bent PC substrate

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

17

Fig. 11 (a) Demonstration of electric conduction, (b) optical absorption spectra, (c) SPM image of the surface, and (d) SEM image of the cross section fabricated by FIB for the ITO thin films transferred to PC substrates. The absorption spectra were measured with a bare PC substrate as the reference. The thickness of the PC substrates were 0.5 mm for (a and c) and 5 mm for (b and d). The transfer was conducted on a hot plate for (a–c) and in a near-IR furnace for (d)

ridges of the plastic substrate as shown in Fig. 14, resulting in an array of alternating two kinds of oxide ribbons on the plastic substrate. An array of ITO-ZnO ribbons was fabricated on a PC substrate by this dual transfer technique (Kozuka 2013). An array of micron-sized crystalline oxide dots can also be prepared on plastic substrates. Cronin et al. combined the dip-pen nanolithography patterning of precursor solutions on a polymer release layer and their subsequent transfer, after heat treatment, to a flexible substrate (Cronin et al. 2014). They prepared precursor dots on a PVP-coated glass substrate using a zinc acetate solution, followed by firing at 500  C, and the resulting ZnO dots were transferred to a polyethylene naphthalate substrate on a hot plate of 190  C under pressure. We believed that the release layer should not be completely decomposed in the firing process so that it aids the fired oxide film to be detached from the silicon substrate and transferred to a plastic substrate. However, later we noticed that fired oxide films can be transferred to plastic substrates even when the release layers are completely decomposed by firing at higher temperatures (Kozuka et al. 2016). Rutile

18

H. Kozuka Pressing

Deposition and firing Si(100)

Ceramic film Release layer

Si(100)

Plastics

Si(100)

Heat

Hot plate

Product Plastics Plastics

Pressing

Ceramic film Plastics

Si(100)

Si(100)

Heat

Hot plate

Fig. 12 Process for fabricating an array of crystalline oxide ribbons on plastic substrates

thin films could be fabricated on PC substrates via a firing process at 900 or 1,000  C. Such a high-temperature process should be welcome, opening a door to the realization of more highly crystalline oxide thin films on plastics, while the factors that enable the film delamination and transfer should be clarified.

Comparison with the Transfer Techniques Reported on the Arrays of Sputtered and Fired Oxide Ribbons Transfer techniques combined with firing process are also reported by other groups, where PZT, BaTiO3, and ZnO ribbons are prepared not by solution process but by radio frequency (rf) sputtering (Qi et al. 2010; Park et al. 2010a, b). They have common key features in their techniques. Oxide ribbons are prepared on thermally resistant substrates (mother substrates), followed by firing for their crystallization. (Oxide ribbons can also be fabricated from oxide thin films by dry etching after the firing process.) Then the mother substrates or the sacrifice layers on them are etched through the intervals between the ribbons by wet process, which undercuts and loosens the ribbons. Finally the oxide ribbons are transferred to elastically stretched, sticky PDMS plates. It should be noted that the wet etching for the undercut is available only when the films have the shapes of ribbons with intervals between them. Then, compared with

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

a

b

19

7

Vertical deflection / µm

6

ITO

5 4 3 2 1

200 200 µm

0

0

100

200

300

400

500

600

700

Horizontal distance / µm

c

d A cut made with a knife

ITO

ITO A cut made with a knife 100 µm

50 µm

Fig. 13 (a) Optical micrograph, (b) surface roughness profile, (c) SEM image, and (d) magnified SEM image of patterned ITO thin films transferred on a 0.5 mm thick PC substrate using a hot plate

these techniques, our sol-gel transfer technique needs no etching but employs a release layer to enable the film/mother substrate separation. Therefore, the thin films need not be shaped into ribbons, and even large-area films can be transferred to plastics. In addition, the films can be transferred basically to any kinds of plastic substrates by our sol-gel transfer technique because the film/plastic adhesion is achieved by utilizing adhesives or by melting or softening the plastic substrate surface. Because of such reasons, the sol-gel transfer technique described here may have higher versatility.

Summary Solution process for fabricating crystalline oxide thin films on plastics has been reviewed. The techniques include crystalline nanoparticle deposition, LPD, and sol-gel method or CSD. Most of the techniques focus on how to crystallize the films on plastic substrates without firing. Sol-gel transfer technique that our group has proposed was also described, which guarantees the film crystallization by

Heat Hot plate

Si(100)

Release layer

Oxide film (B)

Oxide film (A)

Oxide film (B)

Oxide film (A)

Plastics Plastics

Heat Hot plate

Si(100)

Plastics Plastics

Pressing

Release layer

Fig. 14 Process for the fabrication of alternating two kinds of oxide ribbons on a plastic substrate and alternating ITO-ZnO ribbons fabricated on a PC substrate

Oixde film (B)

Oxide film (A)

Plastics Plastics

20 H. Kozuka

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

21

employing a firing process. It was demonstrated that large-area as well as patterned thin films can be fabricated on plastic substrates by the technique.

References Aegerter MA, Al-Dahoudi N. Wet-chemical processing of transparent and antiglare conducting ITO coating on plastic substrates. J Sol-Gel Sci Technol. 2003;27:81–9. Al-Dahoudi N, Bisht H, Göbbert C, Krajewski T, Aegerter MA. Transparent conducting, anti-static and anti-static–anti-glare coatings on plastic substrates. Thin Solid Films. 2001;392:299–304. Asakuma N, Fukui T, Aizawa M, Toki M, Imai H, Hirashima H. Ultraviolet-laser-induced crystallization of sol-gel derived inorganic oxide films. J Sol-Gel Sci Technol. 2000;19:333–6. Asakuma N, Fukui T, Toki M, Imai H. Low-temperature synthesis of ITO thin films using an ultraviolet laser for conductive coating on organic polymer substrates. J Sol-Gel Sci Technol. 2003;27:91–5. Chen YC, Yang CF, Hsueh EY. The application of AZOY transparent conductive oxide film in multifilm-coated polycarbonate optical glasses. J Electrochem Soc. 2010;157:H987–90. Cronin SD, Sabolsky K, Sabolsky EM, Sierros KA. Dip pen nanolithography and transfer of ZnO patterns on plastics for large-area flexible optoelectronic applications. Thin Solid Films. 2014;552:50–5. de Carvalho CN, Lavareda G, Fortunato E, Vilarinho R, Amaral A. ITO films deposited by rf-PERTE on unheated polymer substrates-properties dependence on In-Sn alloy composition. Mater Sci Eng B Solid State Mater Adv Technol. 2004;109:245–8. Deki S, Aoi Y, Hiroi O, Kajinami A. Titanium(IV) oxide thin films prepared from aqueous solution. Chem Lett. 1996a;6:433–4. Deki S, Aoi Y, Yanagimoto H, Ishii K, Akamatsu K, Mizuhata M, Kajinami A. Preparation and characterization of Au-dispersed TiO2 thin films by a liquid-phase deposition method. J Mater Chem. 1996b;6:1879–82. Goutailler G, Guillard C, Daniele S, Hubert-Pfalzgraf LG. Low temperature and aqueous sol-gel deposit of photocatalytic active nanoparticulate TiO2. J Mater Chem. 2003;13:342–6. Hashizume M, Hirashima M. Sol-gel titania coating on unmodified and surface-modified polyimde films. J Sol-Gel Sci Technol. 2012;62:234–9. Hennek JW, Kim M-G, Kanatzidis MG, Facchetti A, Marks TJ. Exploratory combustion synthesis: amorphous indium yttrium oxide for thin-film transistors. J Am Chem Soc. 2012;134:9593–6. Heusing S, de Oliveira PW, Kraker E, Haase A, Palfinger C, Veith M. Wet chemical deposited ITO coatings on flexible substrates for organic photodiodes. Thin Solid Films. 2009;518:1164–9. Howson RP, Avaratsiotis JN, Ridge MI, Bishop CA. Properties of conducting transparent oxide films produced by ion plating onto room-temperature substrates. Appl Phys Lett. 1979;35:161–2. Hu Y, Yuan C. Low-temperature preparation of photocatalytic TiO2 thin films on polymer substrates by direct deposition from anatase sol. J Mater Sci Technol. 2006;22:239–44. Imai H, Hirashima H. Preparation of porous anatase coating from sol–gel-derived titanium dioxide and titanium dioxide-silica by water-vapor exposure. J Am Ceram Soc. 1999;82:2301–4. Imai H, Morimoto H, Tominaga A, Hirashima H. Structural changes in sol-gel derived SiO2 and TiO2 films by exposure to water vapor. J Sol-Gel Sci Technol. 1997;10:45–54. Imai H, Tominaga A, Hirashima H, Toki M, Aizawa M. Ultraviolet-laser-induced crystallization of sol-gel derived indium oxide films. J Sol-Gel Sci Technol. 1998;13:991–4. Imai H, Tominaga A, Hirashima H, Toki M, Asakumua N. Ultraviolet-reduced reduction and crystallization of indium oxide films. J Appl Phys. 1999a;85:203–7. Imai H, Hirashima H, Awazu K. Alternative modification methods for sol–gel coatings of silica, titania and silica–titania using ultraviolet irradiation and water vapor. Thin Solid Films. 1999b;351:91–4.

22

H. Kozuka

Itoyama K. Properties of Sn-doped indium oxide coatings deposited on polyester film by high rate reactive sputtering. J Electrochem Soc. 1979;126:691–4. Kim D. Deposition of indium tin oxide films on polycarbonate substrates by direct metal ion beam deposition. Appl Surf Sci. 2003;218:71–8. Kim H, Horwitz JS, Kushto GP, Kafafi ZH, Chrisey DB. Indium tin oxide thin films grown on flexible plastic substrates by pulsed-laser deposition for organic light-emitting diodes. Appl Phys Lett. 2001;79:284–6. Kim M-G, Kanatzidis MG, Facchetti A, Marks TJ. Low-temperature fabrication of highperformance metal oxide thin-film electronics via combustion processing. Nat Mater. 2011;10:382–8. Kim M-G, Hennek JW, Kim HS, Kanatzidis MG, Facchetti A, Marks TJ. Delayed ignition of autocatalytic combustion precursors: low temperature nanomaterial binder approach to electronically functional oxide films. J Am Chem Soc. 2012a;134:11583–93. Kim Y-H, Heo J-S, Kim T-H, Park S, Yoon M-H, Kim K, Oh MS, Yi G-R, Noh Y-Y, Park SK. Flexible metal-oxide devices made by room-temperature photochemical activation of sol-gel films. Nature. 2012b;489:128–32. Königer T, M€unstedt H. Coatings of indium tin oxide nanoparticles on various flexible polymer substrates: influence of surface topography and oscillatory bending on electrical properties. J Soc Inform Display. 2008;16:559–68. Königer T, Rechtenwald T, Al-Naimi I, Frick T, Schmidt M, M€ unstedt H. CO2-laser treatment of indium tin oxide nanoparticle coatings on flexible polyethyleneterephthalate substrates. J Coat Technol Res. 2010;7:261–9. Kotani Y, Matsuda A, Tasumisago M, Minami T, Umezawa T, Kogure T. Formation of anatase nanocrystals in sol-gel derived TiO2-SiO2 thin films with hot water treatment. J Sol-Gel Sci Technol. 2000;19:585–8. Kozuka H. Wet processing for the fabrication of ceramic thin films on plastics. J Mater Res. 2013;28:673–88. Kozuka H, Takahashi M, Niinuma K, Uchiyama H. Fabrication of highly crystalline oxide thin films on plastics: sol-gel transfer technique involving high temperature process. J Asian Ceram Soc. 2016;4:329–36. Kozuka H, Yamano A, Fukui T, Uchiyama H, Takahashi M, Yoki M, Akase T. Large area ceramic thin films on plastics: a versatile route via solution processing. J Appl Phys. 2012a;111:016106. Kozuka H, Fuku T, Takahashi M, Uchiyama H. Ceramic thin films on plastics: a versatile transfer process for large area as well as patterned coating. ACS Appl Mater Interfaces. 2012b;4:6415–20. Kozuka H, Uchiyama H, Fukui T, Takahashi M. Technique for fabricating ceramic films on plastic substrates. Japanese Patent 2016;No. 5924615. Kozuka H, Fukui T, Uchiyama H. Sol-gel and transfer technique for fabricating dual ceramic thin film patterns on plastics. J Sol-Gel Sci Technol. 2013;67:414–9. Kozuka H, Uchiyama H, Fukui T, Takahashi M. Technique for fabricating ceramic films on plastic substrates. Japanese Patent 2015;No. 5717181. Krebs FC. All solution roll-to-roll processed polymer solar cells free from indium-tin-oxide and vacuum coating steps. Org Electron. 2009a;10:761–8. Krebs FC. Polymer solar cell modules prepared using roll-to-roll methods: knife-over-edge coating, slot-die coating and screen printing. Solar Energy Mater Solar Celles. 2009b;93:465–75. Lam SW, Soetanto A, Amal R. Self-cleaning performance of polycarbonate surfaces coated with titania nanoparticles. J Nanopart Res. 2009;11:1971–9. Langlet M, Kim A, Audier M, Herrmann JM. Sol-gel preparation of photocatalytic TiO2 films on polymer substrates. J Sol-Gel Sci Technol. 2002;25:223–34. Langlet M, Kim A, Audier A, Guillard C, Herrmann JM. Transparent photocatalytic films deposited on polymer substrates from sol-gel processed titania sols. Thin Solid Films. 2003;429:13–21. Latella BA, Triani G, Zhang Z, Short KT, Bartlett JR, Ignat M. Enhanced adhesion of atomic layer deposited titania on polycarbonate substrates. Thin Solid Films. 2007;515:3138–45.

Sol-Gel Preparation of Crystalline Oxide Thin Films on Plastics

23

Matsuda A, Kotani Y, Kogure K, Tatsumisago M, Minami T. Transparent anatase nanocomposite films by the sol–gel process at low temperatures. J Am Ceram Soc. 2000;83:229–31. Matsuda A, Matoda T, Kogure T, Tadanaga K, Minami T, Tatsumisago M. Formation of anatase nanocrystals-precipitated silica coatings on plastic substrates by the sol-gel process with hot water treatment. J Sol-Gel Sci Technol. 2003;27:61–9. Miyake A, Yamada T, Makino H, Yamamoto N, Yamamoto T. Properties of highly transparent conductive Ga-doped ZnO films prepared on polymer substrates by reactive plasma deposition with DC arc discharge. J Photopolym Sci Technol. 2009;22:497–502. Nagayama H, Honda H, Kawahara H. A new process for silica coating. J Electrochem Soc. 1988;135:2013–6. Ohyama M, Kozuka H, Yoko T. Sol-gel preparation of ZnO films with preferential orientation along (002) plane from zinc acetate solution. Thin Solid Films. 1997;306:78–85. Ohyama M, Kozuka H, Yoko T. Sol-gel preparation of transparent and conductive Al-doped ZnO films with highly preferential crystal orientation. J Am Ceram Soc. 1998;81:1622–32. Ott AW, Chang RPH. Atomic layer-controlled growth of transparent conducting ZnO on plastic substrates. Mater Chem Phys. 1999;58:132–8. Ou J, Wang J, Zhang D, Zhang P, Liu S, Yan P, Liu B, Yang S. Fabrication and biocompatibility investigation of TiO2 films on the polymer substrates obtained via a novel and versatile route. Colloids Surf B Biointerfaces. 2010;76:123–7. Park KI, Xu S, Liu Y, Hwang GT, Kang SJ, Wang ZL, Lee KJ. Piezoelectric BaTiO3 thin film nanogenerator on plastic substrates. Nano Lett. 2010a;10:4939–43. Park K, Lee D-K, Kim B-S, Jeon H, Lee N-E, Whang D, Lee H-J, Kim YJ, Ahn J-H. Stretchable, transparent zinc oxide thin film transistors. Adv Funct Mater. 2010b;20:3577–82. Puetz J, Aegerter MA. Direct gravure printing of indium tin oxide nanoparticle patterns on polymer foils. Thin Solid Films. 2008;516:4495–501. Qi Y, Jafferis NT, Lyons Jr K, Lee CM, McAlpine MC. Piezoelectric ribbons printed onto rubber for flexible energy conversion. Nano Lett. 2010;10:524–8. Reuss RH, Chalamala BR, Moussessian A, Kane MG, Kumar A, Zhang DC, Rogers JA, Hatalis M, Temple D, Moddel G, Eliasson BJ, Estes MJ, Kunze J, Handy ES, Harmon ES, Salzman DB, Woodall JM, Alam MA, Murthy JY, Jacobsen SC, Olivier M, Markus D, Campbell PM, Snow E. Macroelectronics: perspectives on technology and applications. Proc IEEE. 2005;93:1239–56. Salar Amoli H, Shokatian S, Abdous M. Thermal annealing combination with pulse Nd-YAG laser treatment on ITO on polycarbonate using spin coating process. J Sol-Gel Sci Technol. 2012;62:319–23. Shimizu K, Imai H, Hirashima H, Tsukuma K. Low-temperature synthesis of anatase thin films on glass and organic substrates by direct deposition from aqueous solutions. Thin Solid Films. 1999;351:220–4. Sobajima S, Okaniwa H, Takagi N, Sugiyama I, Chiba K. Production and properties of transparent electroconductive coating on polyester film. Jpn J Appl Phys. 1974;2(Suppl 2–1):475–8. Su W, Wang S, Wang X, Fu X, Weng J. Plasma pre-treatment and TiO2 coating of PMMA for the improvement of antibacterial properties. Surf Coat Technol. 2010;205:465–9. Sun Y, Rogers JA. Inorganic semiconductors for flexible electronics. Adv Mater. 2007;19:1897–916. Tadanaga K, Kitamuro K, Matsuda A, Minami T. Formation of superhydrophobic alumina coating films with high transparency on polymer substrates by the sol-gel process. J Sol-Gel Sci Technol. 2003;26:705–8. Yamaguchi N, Tadanaga K, Matsuda A, Minami T. Formation of anti-reflective alumina films on polymer substrates by the sol-gel process with hot water treatment. Surf Coat Technol. 2006;201:3653–7. Yamamoto T, Miyake A, Yamada T, Morizane T, Arimitsu T, Makino H, Yamamoto N. Properties of transparent conductive Ga-doped ZnO films on glass, PMMA and COP substrates. IEICE Trans Electron. 2008;E91C:1547–53.

24

H. Kozuka

Yamano A, Kozuka H. Effects of the heat treatment conditions on the crystallographic orientation of Pb(Zr, Ti)O3 thin films prepared by polyvinylpyrrolidone-assisted sol-gel method. J Am Ceram Soc. 2007;90:3882–9. Yamano A, Kozuka H. Single-step sol-gel deposition and dielectric properties of 0.4 μm thick, (001) oriented Pb(Zr,Ti)O3 thin films. J Sol-Gel Sci Technol. 2008;47:316–25. Yang TL, Zhang DH, Ma J, Ma HL, Chen Y. Transparent conducting ZnO:Al films deposited on organic substrates deposited by r.f. magnetron-sputtering. Thin Solid Films. 1998;326:60–2. Yang JH, Han YS, Choy JH. TiO2 thin-films on polymer substrates and their photocatalytic activity. Thin Solid Films. 2006;495:266–71. Zhang ZM, Triani G, Fan LJ. Amorphous to anatase transformation in atomic layer deposited titania thin films induced by hydrothermal treatment at 120 oC. J Mater Res. 2008;23:2472–9.

Radiative Striations in Spin-Coating Films Hiromiutsu Kozuka

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantitative Evaluation of Striations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effect of the Amount of Sol Dispensed, Sol Viscosity, and Spinning Rate . . . . . . . . . . . . . . . . . . . . . Striations in Gel Films Deposited on Stationary Substrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . In Situ Observation of Striations in Gel Layers Deposited on Stationary Substrates . . . . . . . . . . . Possible Mechanism of the Formation of Striations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How Evolution of Striations Can Be Suppressed: Effect of Solvent Volatility . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 2 5 9 11 13 15 18 18

Abstract

Radiative striations that are formed on the surface of spin-coating films were quantitatively evaluated where the height and spacing of striations were measured by a contact probe surface profilometer. How the height and spacing of striations are affected by sol viscosity, spinning rate, and solvent volatility is described based on our experimental data. Striations are demonstrated to be formed even on stationary substrates, which indicates that the substrate rotation is not necessary for the formation of striations. On the basis of the idea that striations are formed by the same mechanism irrespective of the presence or absence of the substrate rotation, in situ observation was made on sols placed on substrates for understanding of the mechanism. When less volatile alcohols are used, the striations once formed are demonstrated to be diminished during gelation.

H. Kozuka (*) Department of Chemistry and Materials Engineering, Faculty of Chemistry, Materials and Bioengineering, Kansai University, Osaka, Japan e-mail: [email protected] # Springer International Publishing AG 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_148-1

1

2

H. Kozuka

Introduction “Radiative striations” are circumferential unevenness in thickness or radially extended ridges and are often observed on the surface of spin-coating films, detectable even with the naked eyes. Striations scatter the light, and hence are generally undesired in optical coatings, and are also regarded as defects when precision in surface flatness is required. Table 1 summarizes the literature on radiative striations reported so far. Daniles et al. studied the patterns formed on the surface of spin-on photoresist materials (Daniels et al. 1986). They related Bénard cells and striations and demonstrated that surface tension plays an important role in the formation of striations. Birnie’s group published several papers (Haas and Birnie 2002; Taylor and Birnie 2002; Birnie 2001; Haas and Birnie 2001a, b, Haas et al. 2001), who detected the onset of striation formation real-time during spinning by laser diffraction method, finding that striations are formed on solvent evaporation. Birnie also suggested that a careful crossreferencing of solvent volatility with surface tension effects could suppress the evolution of striations, based on the idea that Bénard or Marangoni convections are the origin of striations (Birnie 2001). Decrease in the spacing of striations with increasing spinning rate is also reported (Haas and Birnie 2001a; Haas et al. 2001). Reduction of striation is also demonstrated to occur when spin coating is conducted in a closed chamber or in an atmosphere filled with solvent vapor by Daniels et al. and Du et al. (Daniels et al. 1986; Du et al. 1995). For practical use of spin-coating technique, systematic knowledge on the relationship between striations and processing parameters is quite important, which, however, is still lacking. From that view point, the author started studies with quantitative evaluation of striations by surface roughness measurement (Kozuka and Hirano 2000; Kozuka et al. 2001, 2002). Second, we found that striations are also formed even in the absence of the substrate rotation, and then in situ observation was made on sol layers placed on stationary substrates in order to obtain information on the mechanism of the formation of striations (Kozuka et al. 2002). Third, how the volatility of the solvent affects the formation of striations was experimentally studied (Kozuka et al. 2004). These results obtained by the author’s group are described in the following sections.

Quantitative Evaluation of Striations In order to quantitatively evaluate the height and spacing of striations, surface roughness was measured on films using a contact probe surface profilometer (Kozuka and Hirano 2000; Kozuka et al. 2001, 2002, 2004). The measurement was conducted on straight lines that are vertical to the spinning diameter and apart from the spinning center at a prescribed distance (Fig. 1). Wavy transverse profiles are obtained because the test line crosses almost vertical to striations. Surface roughness parameters, Rz (ten-point height of irregularities) and S (mean spacing

Radiative Striations in Spin-Coating Films

3

Table 1 Literature on striations of spin-coating films (not including ours) References Haas and Birnie (2002)

Taylor and Birnie (2002) Birnie (2001) Haas and Birnie (2001a) Haas et al. (2001)

Haas and Birnie (2001b) Mellbring et al. (2001) Du et al. (1995)

Rehg and Higgins (1992)

Daniels et al. (1986)

Remarks Evolution of a temperature gradient was examined by a one-dimensional finite difference model Marrangoni numbers (Mn) were calculated under numerous conditions Thermocapillary instability may be responsible for striation features Trends related to spin speed, solvent volatility, and initial solution thickness are discussed Addition of ethyl acetate in sols suppressed the evolution of striations on Al2TiO5 coatings A plot of solvent vapor pressure versus solvent surface tension was introduced to help discuss the impact of solvent choice Laser diffraction was employed to measure the spacing of striations Spacing decreased with increasing spinning rate Optical profilometry was employed for quantitative evaluation of striations Spacing decreased but amplitude remained constant when spinning rate increased Spacing was independent of the distance from the spinning center Detecting the onset of striation formation in real-time during spinning by laser diffraction method Striations were found to be formed on solvent evaporation Polyethylene spin-coating films Low process temperatures suppressed the formation of striations owing to lower solvent evaporation rate Striations were observed on Nd-doped SiO2-TiO2 films codoped with phosphorus A saturated ethanol atmosphere over the spinning substrate suppressed the evolution of striations Coupled, unsteady Navier-Stokes, convective diffusion, and thermal energy equations were solved numerically for spin-coating of colloidal suspensions Rapid substrate acceleration, high rotation rates, partial saturation of the overlying gas phase, and high initial solids concentration were identified as spin-coating protocols that suppress a convective instability Mellbring (Araki et al. 2001) Metal organic deposition of YBa2Cu3O7
x films resulted in striations at higher spinning rates Striations were observed by an optical microscope precisely on spin-on photoresist materials A relationship between Bénard cells and striations was proposed

of local peaks) (Fig. 2), were automatically calculated by the profilometer. Rz and S represent the height and spacing of striations. The thickness was measured by the contact probe surface profilometer. For this measurement a part of the gel film was scraped off with a surgical knife immediately after deposition, and the level difference between the coated part and the scraped part was measured 1 day later or after firing.

4

H. Kozuka Striation Spinning center Test line Prescribed distance Film Substrate

8 mm

Fig. 1 Surface roughness measurement for quantitative evaluation of radiative striations

Ten point height of irregularities, R z

Vertical deflection

Phase correct filtered mean line Yp

Yp

1

Yv

1

Yp

2

Yv

Yp

3

Yp

4

5

2

Yv

Yv3

4

Yv

5

Sampling length, l

Mean spacing of local peaks, S 1 n S= Xi n i =1 X1

Xi

Xn

Transverse profile

l Fig. 2 Definition of surface roughness parameters, Rz and S

Horizontal distance

Radiative Striations in Spin-Coating Films

5

Effect of the Amount of Sol Dispensed, Sol Viscosity, and Spinning Rate First, the effect of the amount of the sol dispensed on radiative striations of silica gel films was studied (Kozuka et al. 2002). A solution of molar compositions, Si(OC2H5)4:H2O:HNO3:C2H5OH = 1:4:0.01:2 was prepared as the starting solution. After being aged, a sol of a viscosity of 4.2 mPa s at 25  C was obtained and served as the coating solution. 0.2–2.0 mL sol was dispensed without intermittence using a syringe on a sodalime silicate glass substrate (52  76  1.3 mm3) rotated at 2,000 rpm. Surface roughness and thickness of the gel films were measured 10 mm apart from the spinning center. Thickness, Rz, and S are found to be almost constant irrespective of the amount of the sol dispensed (Fig. 3). Because Rz and S represent the height and spacing of the striations, these results indicate that the evolution of striations is not affected by the amount of the sol dispensed (Fig. 3a–c).

Fig. 3 Dependence of (a) thickness, (b) Rz, and (c) S on the amount of sol dispensed measured on spin-on silica gel films (Kozuka et al. 2002)

6

H. Kozuka

Fig. 4 Thickness of spin-on silica gel films prepared from sols aged for various periods of time (Kozuka et al. 2002)

The viscosity of alkoxide-derived sols kept standing in sealed containers at a constant temperature increases with time due to the progress of hydrolysis and polycondensation reaction. Spin coating was made using a silica sol aged for different periods of time in order to study the effect of the sol viscosity on formation of radiative striations (Kozuka et al. 2001, 2002; Kozuka 2003). A solution of molar compositions, Si(OC2H5)4:H2O:HNO3:C2H5OH = 1:4:0.01:2, was prepared and kept at 30  C in a sealed glass container for 237 h. The viscosity increased from 3 to 20 mPa s due to the growth of the siloxane polymers. Spin coating was performed at 3,440 rpm using the sols kept for various periods of time, namely, of various viscosities. Surface roughness of the gel films was measured on lines 10 mm apart from the spinning center. The thickness of the gel films increased with sol-aging time due to the increase in viscosity as seen in Fig. 4. Figure 5 shows the transverse profiles obtained on lines crossing vertically to the striations. (You have to be careful on the difference in vertical and horizontal scales. The profile is much more expanded in the vertical axis. The actual cross-sectional profile of striations is much flatter than those shown in the figure.) Apparently, the height of striations increases with increasing sol viscosity, which is more clearly seen in Rzviscosity relation shown in Fig. 6a; Rz increased with sol viscosity. In other words, the height of the striations increased with sol viscosity and/or film thickness. The dependence of S on sol viscosity is rather vague. S, representing the space of striations, seems to be slightly increased or almost constant when the sol viscosity increased (Fig. 6b). The above results indicate that thicker gel films could result in striations of larger height. Thickness is also dependent on spinning rate. In order to study the effect of

Radiative Striations in Spin-Coating Films

7

Fig. 5 Transverse profiles obtained on spin-on silica gel films prepared from sols aged for various periods of time (Kozuka et al. 2002)

Fig. 6 Dependence of (a) Rz and (b) S of spin-on silica gel films on viscosity of sols aged for various periods of time (Kozuka et al. 2001, 2002; Kozuka 2003)

8

H. Kozuka

spinning rate on formation of striations, silica gel films were deposited at different spinning rates, and surface roughness measurement was conducted (Kozuka et al. 2001, 2002). Silica gel films were also prepared at various spinning rates using a sol of a viscosity of 4.2 mPa s prepared from the starting solution of mole ratios, Si(OC2H5)4:H2O:HNO3:C2H5OH = 1:4:0.01:2. The sol was dispensed on a glass substrate rotated at 500 rpm, and after 5 s the spinning rate was increased up to 1,000–6,000 rpm, kept there for 1 min, resulting in formation of silica gel films ca. 0.6–1.5 μm in thickness (Fig. 7a). Surface roughness measurement conducted on straight lines 25 mm apart from the spinning center indicated that Rz slightly decreased and S decreased with increasing the second stage spinning rate as seen in Fig. 7b, c. In other words, striations increased again in height and spacing as the film thickness increased. Haas et al. made quantitative evaluation of striations on

Fig. 7 Dependence of (a) thickness, (b) Rz, and (c) S on spinning rate for spin-on silica gel films (Kozuka et al. 2001, 2002; Kozuka 2003)

Radiative Striations in Spin-Coating Films

9

phosphosilicate-derived spin-on glass and reported that spacing of striation decreases with increasing spinning rate (Haas et al. 2001), which agrees with our result. However, they reported that the amplitude, corresponding to the height of striations, remains constant at different spinning rates. Possible understanding on the effect of sol viscosity and spinning rate is described in “Possible Mechanism of the Formation of Striations.”

Striations in Gel Films Deposited on Stationary Substrates One could have an impression that the centrifugal force and/or viscous drag against it would be the key for the evolution of radiative striations. And, the substrate rotation is believed to be needed for formation of radiative striations. However, we have found that striations are formed even when the substrate is stationary, not being rotated (Kozuka et al. 2002; Kozuka 2003). A solution of mole ratio, Ti(OC3H7i)4:H2O:NH(C2H4OH)2:HNO3:C2H5OH = 1:1:1:0.2:30, was prepared, and gel films were deposited by spin coating at 900 rpm and fired at 700  C for 10 min. As seen in the optical micrographs of the surface of the film (Fig. 8), cell-like patterns are observed near the spinning center (a), and chain-like patterns about 3 mm apart from the spinning center (b). The chains had

Fig. 8 Optical micrographs of a spin-on titania film taken (a) near and (b) apart from the spinning center (Kozuka 2003)

10

H. Kozuka

Sol 200 µm

b

Stationary substrate a 200 µm

c

Fig. 9 (a) Schematic illustration showing the placement of a sol on a stationary substrate and the optical micrographs of a titania gel film thus prepared taken (b) near and (c) apart from the center (Kozuka et al. 2002)

more continuous shape about 6 mm apart from the spinning center (c), and striations are observed around 15 mm apart from the center (d). Similar dependence of the patterns on location is already reported by Daniels et al. on spin-on photoresist materials (Daniels et al. 1986). An interesting thing found there was that such cell-like patterns and striations are also formed in a gel layer deposited on a stationary substrate (Fig. 9a) as shown in Fig. 9b, c. Titania gel films deposited at various spinning rates showed again, 0, lc, decrease in Rz and S with increasing spinning rate (Fig. 10) as in the case of silica-coating films. Rz and S at a spinning rate of 0 rpm correspond to that of the layer deposited on a stationary substrate, and they appear to lie on the extension lines drawn from the higher to lower spinning rates as seen in Fig. 10. As described above, striations can also be formed in gel films deposited on a stationary substrate, and the surface patterns observed are so similar to those of spincoating films. These facts indicate that the substrate rotation is not a necessary condition for the evolution of striations; in other words, neither the centrifugal force nor viscous drag against it is the cause of striations. Also, the height (Rz) and spacing (S) of the film deposited on a stationary substrate lie on the extension lines drawn from higher to lower spinning rates plotted for the spin-on films. This suggests that the mechanism of the evolution of striations is the same in principle whether the substrate is rotated or not.

Radiative Striations in Spin-Coating Films

11

Fig. 10 Dependence of Rz and S of titania gel films on spinning rate (Kozuka et al. 2002; Kozuka 2003)

In Situ Observation of Striations in Gel Layers Deposited on Stationary Substrates According to Bornside et al. spin-coating process can be divided into four stages: “deposition,” where the sol is dispensed; “spin-up,” where the sol flows radially outward, driven by centrifugal force; “spin-off,” where the excess sol leaves as droplets; and “evaporation,” where the solvent evaporates (Bornside et al. 1987). Then, where in the four stages the striations are formed could be a fundamental question. Haas and Birnie performed sophisticated experiments, detecting the onset of striation formation in real time during spinning by laser diffraction method (Haas and Birnie 2001), and concluded that striations are formed during the drying stage. On the basis of the experimental facts described in “Striations in Gel Films Deposited on Stationary Substrates,” one can assume that the origin of striations is the same whether the substrate rotation is present or not. Therefore, in situ observation on gel layers deposited on stationary substrates could provide some information. A drop of the titania sol was dispensed on a stationary substrate, and in situ observation was made with an optical microscope (Kozuka et al. 2002). The photographs taken at various periods of time after dispensing the sol are shown in Fig. 11. It was observed that cell-like patterns near the center and the striation-like patterns away from the center are formed almost simultaneously on solvent evaporation after the sol was spread, not on proceeding of the sol front toward the outer. This suggests that striations are formed during the drying stage as was demonstrated by Haas and Birnie (2001b). In an attempt to visualize the convection or flow in a sol dropped on a stationary substrate, fine Silicon powders were added in silica and titania sols, and in situ

12

H. Kozuka

Fig. 11 In situ observation of a titania sol placed on a stationary glass substrate. The optical micrographs were taken about every 7 s after the sol was spread over the substrate. The time passed from (a) to (f) (Kozuka et al. 2002; Kozuka 2003)

observation was made with an optical microscope (Kozuka et al. 2004; Ishikawa 2002; Ashibe 2003). Near the location where the sol was dropped, convection was observed occurring in the thickness direction, and cell-like patterns were seen as

Radiative Striations in Spin-Coating Films

13

Fig. 12 Photograph of silica sol containing CH3CH(OH) CH2OH placed on a stationary substrate. Fine silicon powders were added to the sol, and the photograph was taken near the center (Kozuka et al. 2004)

shown in Fig. 12. Apart from that location, radial flow was added to the convection, finally forming striations.

Possible Mechanism of the Formation of Striations Haas et al. made quantitative evaluation of striations on phosphosilicate-derived spin-on glass using optical profilometry, and studied the dependence of striation spacing on the distance from the spinning center (Haas et al. 2001). They found that the spacing is constant at any distance from the center, making important suggestion as follows. If the striations are formed via a radial stretching out of the cellular features that are initially formed in the spinning center, the spacing should increase linearly with the distance from the center. However, the spacing remains independent of the distance, disproving this mechanism. The striations cannot arise from the centrifugal elongation of the cellular pattern starting near the spinning center (Haas et al. 2001). In situ observation made on a sol placed on a stationary substrate also supports it. Formation of cell-like patterns near the center did not precede the evolution of striations apart from the center; they developed almost simultaneously (Fig. 11). We obtained similar data on TEOS-derived spin-on silica gel films as shown in Fig. 13, where S is plotted against the distance from the spinning center. Although slightly increasing with the distance from the center, S does not increase linearly, not passing through the origin of the coordinates, suggesting that striations are not straight lines extending without intermittence. The slightly smaller values of S at smaller distances may result from cell-like or chain-like patterns formed near the center as were seen in Fig. 8. The research on the morphological patterns that are formed in a liquid layer placed on a solid plate have long history. The patterns are believed to be formed by

14

H. Kozuka

Fig. 13 Dependence of S on the distance from the spinning center for a TEOS-derived spin-on silica gel film

70 60

S / µm

50 40 30 20 10 0

0

10

20

30

40

Distance from the spinning center / mm

- Lower spinning rates - Higher viscosity

- Higher spinning rates - Lower viscosity

Fig. 14 Schematic illustration of rolling cells in thick and thick sol layers

Marangoni and/or Bénard convection and are known to be variable depending on the materials, the temperature gradient across the layer, and the layer thickness. Berg et al. experimentally demonstrated that the patterns formed in liquid layers depend on the layer thickness; cell- and worm-like patterns were formed depending on the layer thickness (Berg et al. 1966). Considering that the striations are formed on solvent evaporation in gel layers, rolling convections could be one of the possible origins for the evolution of striations (Fig. 14). Even when the rolling convections are assumed to be the origins, a question still remains: why the striations be radiative. One possible factor is the gradient in thickness across the substrate radius occurring in the intermediate stages of the film formation, namely, in the spin-off and evaporation stages. Sol layers could have radial distribution in thickness; thicker near the spinning center and thinner outward. Another possible factor is the flow of the sols that is added to the convections as suggested by Birnie (2001). In the case of spin coating, the sol flow radially, driven by centrifugal force. Even when the substrate is not rotated,

Radiative Striations in Spin-Coating Films

15

the sol dropped on a stationary substrate is spread out radially. Therefore, in both cases, radial flow is added to the convections. Very recently the author’s group has found that even in dip-coating films, striations are formed as long as the gel layer has high viscosity and the substrate is withdrawn at a high rate (Morita 2004). TEOS was hydrolyzed in the presence of PVP, and the substrate was withdrawn at a rate as high as 70 cm min
1. The gel films obtained had striations aligning parallel to the withdrawing direction, i.e., to the gravity, which were observed near the lower part of the films. Considering that the sol drains down along the gravity, the experimental fact strongly suggests that striations are formed via rolling convections, where the rolling convections are realized by the down-flow that is added to Marangoni and/or Bénard convections. It should be noted that striations were formed in the part with larger thickness in the case of dip coating, which is opposite to what is observed in spin-coating films. This indicates that the gradient in thickness may not be the important factor that causes rolling convections and formation of striations. Striations larger in height and spacing were found at higher sol viscosities and lower spinning rates, i.e., in thicker films, as described in “Effect of the Amount of Sol Dispensed, Sol Viscosity, and Spinning Rate.” If the rolling convections occurring on solvent evaporation are the origins of striations, a simple explanation could be given. When the sol layer is thick in the evaporation stage, convections or rolling cells can occur large in size, leading to striations large in height and spacing as is schematically illustrated in Fig. 14.

How Evolution of Striations Can Be Suppressed: Effect of Solvent Volatility Daniels et al. and Du et al. found that striation is reduced when spin coating is conducted in a closed chamber or in an atmosphere that is filled with solvent vapor (Daniels et al. 1986; Du et al. 1995). Therefore, solvents of low volatility are expected to suppress the formation of striations. In order to study the effect of the volatility of solvents on the formation of striations, spin-coating films were prepared from TEOS solutions containing various alcohols (Kozuka et al. 2004). Si(OC2H4)5-HNO3-H2O-ROH solutions were prepared where ROH = CH3OH, C2H5OH, CH3OC2H4OH, and CH3CH(OH)CH2OH. The boiling point, vapor pressure, and surface tension of the alcohols are listed in Table 2, together with the sol viscosity and the film thickness. For ROH = C2H5OH, the mole ratio to Si(OC2H4)5 was 0.01, 4, and 4 for HNO3, H2O, and C2H5OH, respectively. For the other alcohols, the same volume as that of C2H5OH was used. The starting solutions were kept in a sealed glass container at 30  C for 72 h, and then served as the coating solutions. The coating solution was placed on a sodalime silicate glass substrate (76  52  1.3 mm), and then the substrate rotation speed was increased at a rate of 1,000 rpm s
1 up to 2,000 rpm, kept there for 60 s. The gel films were transferred into an electric furnace of 200  C, kept there for 10 min. The roughness parameters were obtained on lines 15 mm apart from the spinning center.

16

H. Kozuka

Table 2 Solvent properties, the sol viscosity and the thickness of the films heated at 200  C

Solvent CH3OH C2H5OH CH3OC2H4OH CH3CH(OH) CH2OH

Boiling point/oC 64.7 78.3 124.5 187.6

Vapor pressure (at 25  C)/kPa 16 8 1.6 ffi0

Surface tension/mN m
1 22.06 21.97 30.84 35.80

Sol viscosity/ mPa s 1.5 2.1 2.7 19

Film thickness/ μm 0.480.02 0.530.02 0.300.02 0.400.02

Fig. 15 Optical micrographs of the films prepared from the sols containing (a) CH3OH, (b) C2H5OH, (c) CH3OC2H4OH, and (d) CH3CH(OH)CH2OH. Gel films were heated at 200  C, 1 min after the spin coating (Kozuka et al. 2004)

Great difference was not seen in the thickness of the heated films (Table 2). In the optical micrographs of the film surface (Fig. 15), striations are detected for the films prepared from sols containing CH3OH, C2H5OH, and CH3OC2H4OH, while they are not for that from the sol containing CH3CH(OH)CH2OH, the alcohol of the lowest volatility. Rz (the height of the striations) and S (the spacing of the striations) measured on the film surface were found to decrease and increase as the boiling point of the alcohols increased as shown in Fig. 16. Decrease in Rz represents that

Radiative Striations in Spin-Coating Films

17

Fig. 16 Dependence of surface roughness parameters of spin-on silica gel films on boiling point of alcohols in sols. Gel films were heated at 200  C, 1 min after the spin coating (Kozuka et al. 2004)

alcohols of low volatility apparently suppress the evolution of striations, which qualitatively agrees with what Daniels et al. and Du et al. observed (Daniels et al. 1986; Du et al. 1995). The increase in S with decreasing solvent volatility possibly results from the coalescence of the neighboring striations occurring before gelation. A question arises; is it the lack of convections due to low volatility that reduces the formation of striations? In order to examine whether convections occur or not in sols containing less volatile alcohols, a drop of the sol containing fine Silicon powders was placed on a glass substrate, and in situ observation was made with an optical microscope. As a result, convections were observed occurring in the thickness direction within cells even in the sol containing CH3CH(OH)CH2OH (Fig. 12). Then, the as-deposited gel film containing CH3CH(OH)CH2OH was observed in situ by the optical microscope during being kept in the ambient atmosphere. The pictures taken 5 s and 240 s after spin coating are shown in Fig. 17. Striations were present immediately after spin coating and then gradually disappeared in the ambient atmosphere. Such change in appearance was not observed for the gel film prepared from the sols containing alcohols of high volatility. In addition, when the gel film containing CH3CH(OH)CH2OH was heated at 200  C immediately after spin coating, the striations remained, and Rz was 0.086 μm, which is much larger than that of the film heated 1 min after spin coating (0.005 μm). As described above, convections do occur and striations do form on spin coating even when less volatile alcohols are used as solvents. In other words, less volatile alcohols do not prevent the evolution of striations, just allowing the striations to disappear during drying. Less volatile solvents evaporate slowly, and it takes long time before gelation (solidification) occurs. The sol layer retains fluidity as long as gelation is not achieved, and hence undergoes any changes that reduce the surface area, including disappearance of striations. When the as-deposited gel films are heated immediately after deposition, the solvent evaporates rapidly, allowing

18

H. Kozuka

Fig. 17 Optical micrographs of the gel film prepared from the sol containing CH3CH(OH)CH2OH, taken (a) 5 s and (b) 240 s after spin coating

gelation to occur before striations disappearing. This is what was observed in the increased Rz for the film heated immediately after spin coating.

Summary Radiative striations were quantitatively evaluated by surface roughness measurement, and the dependence of the height and spacing of striations on sol viscosity, spinning rate, and solvent volatility was described. Striations were found to be formed even on stationary substrates, suggesting that the rotation of substrate is not necessary for the formation of striations. On the basis of the idea that striations are formed by the same mechanism irrespective of the presence or absence of the substrate rotation, in situ observation was made on sols placed on substrates for understanding of the mechanism. The use of less volatile alcohols was demonstrated to be effective in avoiding striations, where striations once formed were found to diminish in the course of gelation.

References Araki T, Kurosaki H, Yamada Y, Hirabayashi I, Shibata J, Hirayama T. Coating processes for YBa2Cu3O7-x superconductor by metalorganic deposition method using trifluoroacetates. Supercond Sci Technol. 2001;14:783–6. Ashibe, N. Effects of the solvents on the formation of striations of spin-coating films. Bachelor Thesis, Department of Materials Science and Engineering, Kansai University 2003 [in Japanese]. Berg JC, Boudart M, Acrivos A. Natural convection in pools of evaporating liquids. J Fluid Mech. 1966;24:721–35. Birnie DP. Rational solvent selection strategies to combat striation formation during spin coating of thin films. J Mater Res. 2001;16:1145–54.

Radiative Striations in Spin-Coating Films

19

Bornside DE, Macosko CW, Scriven LE. On the modeling of spin coating. J Imaging Technol. 1987;13:122–30. Daniels BK, Szmanda CR, Templeton MK, Trefonas III PK. Surface tension effects in microlithography – striations. Proc SPIE. 1986;631:192–201. Du XM, Orignac X, Almeida RM. Striation-free, spin-coated sol-gel optical films. J Am Ceram Soc. 1995;78:2254–6. Haas DE, Birnie DP. Nondestructive measurement of striation defect spacing using laser diffraction. J Mater Res. 2001a;16:3355–60. Haas DE, Birnie III DP. Real-time monitoring of striation development during spin-in-glass deposition. In: Feng X, Klein LC, Pope EJA, Komarneni S, editors. Sol-gel commercialization and applications. Westerville: American Ceramic Society; 2001b. p. 133–8. Haas DE, Birnie DP. Evaluation of thermocapillary driving forces in the development of striations during the spin coating process. J Mater Sci. 2002;37:2109–16. Haas DE, Birnie III DP, Zecchino MJ, Figueroa JT. The effect of radial position and spin speed on striation spacing in spin on glass coatings. J Mater Sci Lett. 2001;20:1763–6. Hartner W, Bosk P, Schindler G, Bachhofer H, Mort M, Wendt H, Mikolajick T, Dehm C, Schroeder H, Waser R. SrBi2Ta2O9 ferroelectric thin film capacitors: degradation in a hydrogen ambient. Appl Phys A Mater Sci Process. 2003;77:571–9. Ishikawa, Y. Fundamental studies on the mechanism of formation of spin-coating films. Bachelor Thesis, Department of Materials Science and Engineering, Kansai University 2002 [in Japanese]. Kozuka H. On ceramic thin film formation from gels: evolution of stress, cracks and radiative striations. J Ceram Soc Jpn. 2003;111:624–32. Kozuka H, Hirano M. Radiative striation and surface roughness of alkoxide-derived spin coating films. J Sol-Gel Sci Technol. 2000;19:501–4. Kozuka H, Takenaka S, Kimura S. Nanoscale radiative striations of sol-gel-derived spin-coating films. Scr Mater. 2001;44:1807–11. Kozuka H, Takenaka S, Kimura S, Haruki T, Ishikawa Y. Effects of processing parameters on radiative striations of alkoxide-derived spin-coating films. Glass Technol. 2002;43C:265–71. Kozuka H, Ishikawa Y, Ashibe N. Radiative striations of spin-coating films: surface roughness measurement and in-situ observation. J Sol-Gel Sci Technol. 2004;31:245–8. Liu J, Lam YL, Chan YC, Zhou Y, Ooi BS, Yun ZS. Experimental and theoretical study of the cracking behavior of sol-gel-derived SiO2 film on InP substrate. Appl Phys A Mater Sci Process. 2000;70:341–3. Mellbring O, Oiseth SK, Krozer A, Lausmaa J, Hjertberg T. Spin coating and characterization of thin high-density polyethylene films. Macromolecules. 2001;34:7496–503. Mendiola J, Calzada ML, Ramos P, Martin MJ, Agullo-Rueda F. On the effects of stresses in ferroelectric (Pb,Ca)TiO3 thin films. Thin Solid Films. 1998;315:195–201. Morita T. Fundamental studies on the formation of striations on silica/polyvinylpyrrolidone hybrid coating films. Bachelor Thesis, Kansai University 2004 [in Japanese]. Rehg TJ, Higgins BG. Spin coating of colloidal suspensions. AICHE J. 1992;38:489–501. Taylor DJ, Birnie DP. A case study in striation prevention by targeted formulation adjustment: aluminum titanate sol-gel coatings. Chem Mater. 2002;14:1488–92.

Graphene and Carbon Dots in Mesoporous Materials Luca Malfatti, Davide Carboni, and Plinio Innocenzi

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graphene in Mesoporous Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graphene and Its Derivatives (GO and rGO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graphene in Sandwich Mesoporous Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graphene in Mesoporous Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Dots in Mesoporous Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mesoporous Particles with Carbon Nanodots: Bio-related Applications . . . . . . . . . . . . . . . . . . . . Mesoporous Particles with Carbon Nanodots for Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mesoporous Particles and Carbon Nanodots for Sorption and Catalysis . . . . . . . . . . . . . . . . . . . . Carbon Nanodots in Mesoporous Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . From Carbon Quantum Dots to Graphene Quantum Dots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 3 4 7 11 16 18 21 22 25 26 26

Abstract

This chapter wishes to provide the reader with some examples of composite materials obtained through integration of carbon-based nanostructures, graphene and carbon dots, into mesoporous materials and some of the issues thereof. After a brief introduction to the chemistry of graphene and its derivatives, relevant for understanding the problems associated with the preparation of graphene-based nanocomposites, the reader is guided through some examples illustrating the problems related to the control of pore organization into these mesoporous materials. Some significant applications of graphene-based sandwich-like structures and films, such as energy storage or optical sensors, are also described before introducing the reader into the chemistry of carbon dots and its integration

L. Malfatti (*) • D. Carboni (*) • P. Innocenzi (*) Laboratory of Materials Science and Nanotechnology, LMNT – D.A.D.U., University of Sassari and CR-INSTM, Alghero, Sassari, Italy e-mail: [email protected]; [email protected]; [email protected]; [email protected] # Springer International Publishing Switzerland 2016 L. Klein et al. (eds.), Handbook of Sol-Gel Science and Technology, DOI 10.1007/978-3-319-19454-7_150-1

1

2

L. Malfatti et al.

into mesoporous structures. Some important examples of the applications of carbon dots, such as bio-related applications or mesoporous particles embedding carbon dots for sorption and catalysis, are then illustrated. Finally, the integration and application of carbon dots into mesoporous films and some hints onto carbon quantum dots and graphene quantum dots are explored.

Introduction The controlled organization of porosity has always been one of the most important features in sol–gel-derived materials, due to the great advantages and versatility that this characteristic is able to provide. It is worth mentioning, in particular, the high surface area attainable by a mono-dispersed porosity or the degree of functionalization achievable through hosting nanostructures, such as nanoparticles, or via grafting of specific groups. Mesoporous materials, featuring an average pore size ranging from 2 to 50 nm, are easier to obtain by a template-assisted self-assembly making use of micelles, though the achievement of an organized porosity is not always straightforward (Innocenzi and Malfatti 2013). The micellar templates are often prepared starting from surfactants characterized by different chemical natures: cationic (i.e., alkyl-ammonium halides), anionic (i.e., sodium alkyl-sulfate), or neutral (i.e., block copolymers). The latter is a family of surfactants featuring a polymeric chain made of monomers with different, and sometimes opposite, solvating behaviors. One of the most used is Pluronic, formed by polyethylene oxide and polypropylene oxide building blocks, which can induce an organized mesoporosity characterized by an average pore size of tens of nanometers depending on the type and length of the chain. Alternatively, the use of porogenic solvents (such as acetonitrile) can induce a porosity whose organization is more difficult to control. The extensive research performed in the sol–gel chemistry of self-assembled materials has produced a considerable amount of experimental procedures describing in detail the best conditions to obtain mesoporous materials with highly organized porosities. However, new challenges are nowadays arising from the need of adapting well-established procedures to the insertion of new carbon-based nanostructures, such as graphene and its derivatives or carbon dots. In fact, the control of the pore organization inside sol–gel materials can be greatly affected by the chemical nature of this nanostructures. Nonetheless, the great advantages deriving from the synergistic effect of carbon-based nanostructures embedded into mesoporous materials have triggered the need of developing new strategies toward the synthesis of these novel nanocomposites. Most of the carbon-based nanostructures, in fact, show peculiar properties (such as high electron mobility for graphene or strong photoluminescence properties for the carbon dots), which can be tuned by a strong interaction with a host porous matrix (i.e., crystalline titania or zinc oxide) enabling enhanced or totally innovative applications. For instance, given the high mechanical and thermal stability of graphene layers, a graphene-based nanocomposite could show a considerable improvement in those areas. Likewise,

Graphene and Carbon Dots in Mesoporous Materials

3

the integration of a carbon-based mesoporous matrix with metal nanoparticle, semiconductor nanoparticles, or organic molecule with specific functionalities could lead to the design of smart nanodevices. Notwithstanding the prospected potential of carbon-based mesoporous nanocomposites, the development of general approaches capable of fully integrating these nanostructures remains still an open challenge. For example, attaining a homogeneous distribution of nanostructures into the mesoporous matrices is not a trivial task, and it depends upon the chemical interactions between the surface chemistry of the carbon-based nanostructures and both the sol–gel precursors and the structure-directing agents. A number of smart approaches have been designed so far for specific tasks, and some of them are reported as examples in this chapter. Nonetheless, further studies on the self-assembly chemistry of mesoporous materials embedding graphene-based nanostructures are needed. On the other hand, though carbon dots can be integrated easier than graphene into mesoporous matrices, also thanks to their good solubility in sol–gel media, the number of studies on this topic is still relatively small. This chapter is far from being comprehensive, and it would rather like to provide the reader with an overview of the current state of the art inherent in some mesoporous materials embedding carbon-based nanostructures and some issues thereof.

Graphene in Mesoporous Materials The technological use of graphene in composite materials bloomed a great deal of interest in the scientific community since 2004, when Geim and Novoselov et al. published a paper reporting the preparation of few-layer graphene (G