MIAMI UNIVERSITY The Graduate School. Certificate for Approving the Dissertation. We hereby approve the dissertation of.

Transcription

1 MIAMI UNIVERSITY The Graduate School Certificate for Approving the Dissertation We hereby approve the dissertation of Diep Vu Ca Candidate for the Degree: Doctor of Philosophy Prof. James A. Cox, Director Prof. Gilbert E. Pacey, Reader Prof. Thomas L. Riechel, Reader Prof. Richard T. Taylor, Reader Prof. Hailiang Dong, Graduate School Representative

2 ABSTRACT NANOSTRUCTURED ASSEMBLIES FOR SOLID PHASE EXTRACTION OF METAL IONS by Diep Vu Ca The main goal of our research was to develop nanostructured materials for i) solid phase extraction of metal ions and ii) electrocatalytic systems. The selective preconcentration of cesium from aqueous solutions containing high concentrations of alkali metals is an important problem in the treatment of radioactive waste. We investigated the use of cobalt hexacyanoferrate (CoHCF) for this purpose. The CoHCF in our work was encapsulated in a silica sol-gel material that is templated to have pores in the nano-domain. A capacity of 0.61 ± 0.01 mmol Cs + g -1 was achieved for the CoHCFdoped sonogel. The CoHCF silica sol-gels are promising both for the solid phase extraction of Cs + and for the capture and storage of this cation. For environmental applications, after sorbing Cs + the glass-like material can be sintered to collapse the pores. We found that silica sol-gel is not only a good hosting material but also a medium for growing crystals and extruding fibers. Here, a sonogel was used to grow CoSO 4 and CuSO 4 crystals and extrude Prussian Blue-containing fibers. Subsequently, we investigated the use of mixed-ligand monolayer-protected gold nanoclusters (MPCs) with crown ether (CE) and carboxylate functionalities for the preconcentration of cesium from aqueous solution. 18-crown-6 ether was used as a functional group because it has a selective affinity to cesium. Here, the MPCs were used to assemble layer-by-layer (LBL) films on a substrate. The uptake of cesium from solution by these films was monitored by a quartz crystal microbalance. Our general interest was to modify electrode surfaces that can selectively interact with substances ranging from metal ions to biological models. We focused on the former as a surrogate. Thus, the study was developed for a metal ion, Pb 2+ which is electrochemically active. In this case, 15-crown-5, which selectively complexes with Pb 2+, was used instead of 18-

3 crown-6. The electrostatic LBL films were assembled on gold and indium tin oxide (ITO) electrodes. Trapping Pb 2+ within these nanostructured films was demonstrated by voltammetry and quartz crystal microbalance measurements. Metal nanoparticles (NPs) can have different electrocatalytic properties from the corresponding bulk metal under given conditions. A hypothesis of our study was that NP catalysts can be optimized by controlling the distribution of metal NPs on an electrode surface. The oxidation of cysteine and arsenite were the test systems for AuNPs and PtNPs, respectively. Generation-4 poly(amidoamine)-encapsulated Au and Pt nanoparticles were synthesized. The metal-pamam NPs were assembled on an ITO electrode. The PAMAM was then decomposed by heating, leaving the NPs on the ITO. The surface excess, Γ, of PAMAM-metal NPs was controlled. The catalytic oxidation of cysteine at the resulting AuNP array was observed. Interestingly, a study of the cyclic voltammetric peak current vs Γ showed that a small amount of metal NPs dispersed on a surface electrode gave high electrocatalytic activities. For example, a mole fraction of Au-PAMAM in the assembled layer of yielded the highest sensitivity. Linear calibration curves were obtained over the range 5 µm µm for cysteine with AuNP catalysts. For As III with PtNPs, linearity was observed over the range 0.2 mm 1.0 mm.

4 NANOSTRUCTURED ASSEMBLIES FOR SOLID PHASE EXTRACTION OF METAL IONS A DISSERTATION Submitted to the Faculty of Miami University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Chemistry and Biochemistry by Diep Vu Ca Miami University Oxford, Ohio 2005 Dissertation Director: Dr. James A. Cox

5 Table of Contents Table of Contents List of Tables List of Figures Acknowledgement Page ii vii viii xi 1 Chapter 1 Sol-gel Chemistry Introduction Overview of sol-gel processing Precursors and methods of synthesis Hydrolysis Condensation Geletion Aging Drying Sintering Factors affecting sol-gel structures Precursors Water ph Template-based nanoporous silica Surfactant templates Dendrimer template Sonogel Physical characterizations 27 ii

6 1.6.1 Basic concepts Theory Surface area and pore size analyzer, Coulter SA Applications Sol-gel based sensors Sol-gel materials as solid matrix for organic dyes, proteins and enzymes Sol-gel derived selective electrodes Solid phase extraction 36 2 Chapter 2 Removal of Cesium and Polynulear Transition Metal Hexacyanometallates Introduction Radioactive cesium Methods of cesium removal Organic ion exchangers Inorganic ion exchangers Polynuclear transition metal haxacyanometallates Synthesis Structures Properties Application Extraction of cesium by transition metal hexacyanoferrates Overview of the method Evaluation of the method Mechanism of cesium sorption by hexacyanoferrates Some characteristics important to practical operations Environmental and medical applications 57 iii

7 3 Chapter 3 Solid Phase Extraction of Cesium from Aqueous Solutions Using Solgel Encapsulated Cobalt Hexacyanoferrate Introduction Experimental Reagents Preparation of the sol-gel composites Analytical procedures Results and discussion Conclusion 4 Chapter 4 Crystal Growth and Fiber Extrusion with Sol-Gel Matrices Introduction Crystal growth in gel materials Crystal growth in inorganic gels Crystal growth in organic gels Experimental Reagents and materials Preparation of salt-doped sol-gel for crystal growth Apparatus Results and discussion Conclusion 89 5 Chapter 5 Gold Nanoparticles Introduction Synthetic method Citrate reduction Alkanethiolate-stabilized AuNPs Place exchange reactions 94 iv

8 5.2.4 Dendrimer-stabilized NPs Other ligands as protecting agents Surfactant-stabilized AuNPs Polymer-stabilized AuNPs Metal-oxide supported AuNPs Controlling size and size distribution of NPs Structure Size-dependent properties Electronic properties The surface plasmon resonance (SPR) Nonlinear optics (NLO) Melting point Catalysis Characterization techniques Applications Catalysts Sensors Electronic devices Chapter 6 Layer-by-Layer Films Containing Metal NPs Introduction Layer-film assembly and electrocatalytic activity of the film Thin film containing thiol monolayer protected NPs Thin film containing dendrimer-encapsulated metal NPs AuNPs as electrode arrays Characterization techniques UV-visible spectroscopy Quartz crystal microbalance Scanning probe microscopy (SPM) 136 v

9 7 Chapter 7 Measurement Platform Fabrication by Layer-by-Layer Assembly of Crown Ether Functionalized Gold Nanoclusters Introduction Experimental Chemicals and materials Apparatus Synthetic methods Procedures Results and discussion Conclusion Chapter Optimization of the Dispersion of Gold and Platinum Nanoparticles on Indium Tin Oxide for the Electrocatalytic Oxidation of Cysteine and Arsenite Introduction Experimental Chemicals and materials Apparatus Synthetic methods Procedures Results and discussion Conclusion 177 vi

10 List of Tables Table Page 1 Characterization of the pore structure of sol-gels processed to contain K 4 Fe(CN) Influence of Na + and Ca 2+ on the uptake of Cs + by CoHCF-doped sonogels 74 vii

11 List of Figures Figure Page 1 Hydrolysis mechanism 5 2 Condensation mechanism 6 3 Model for sol-gel shrinkage 8 4 Schematic illustration of the drying process 10 5 Schematic of the organic template, which is incorporated and removed, for 15 preparation of amorphous silica 6 Schematic phase diagram for C 16 TMABr in water 17 7 Three structure types observed for silica-surfactant mesophases 20 8 Schematic of the liquid-crystal templating mechanism 21 9 Structure of G4-PAMAM-NH IUPAC classification of adsorption isotherms The unit cell of Prussian Blue and analogues M A k [M B (CN) 6 ] l.xh 2 O according to Ludi and Güdel Pore size distribution of sol-gels processed to contain K 4 Fe(CN) Influence of the processing ph on the uptake capacity of CoHCF-doped silica sol-gels Reaction scheme for the formation of CoHCF by sequential immersion of G4-PAMAM-doped sol-gel into Fe(CN) 4-6 and Co Pore size distribution of a sonogel processed to contain K 4 Fe(CN) Fibers on the surface of a monolith FTIR spectra of Prussian Blue and of the blue fiber CoSO 4 crystals formed on the surface of the sol-gel CuSO 4 crystals formed on the surface of the sol-gel CuSO 4 crystals formed inside the sol-gel Synthetic scheme for interdendrimer metal nanoparticles Dendrimer nanocomposite structures 100 viii

12 23 Scheme of intradendrimer metal-displacement Synthetic scheme of dendrimer-encapsulated metal nanoclusters with the use of fatty acid as a phase transfer agent Relationship between particle size and melting point of gold nanoparticles A diagram of CHI400 EQCM Formation of 15CE5-MPC via place-exchange reaction Scheme of stepwise electrostatic multilayer film assembly Complex between Pb 2+ and 15-CE Size distribution of the MPCs Layer-by-layer assembly of ITO 3-APTES (CE-MPC G4-PAMAM) n monitored by spectrophotometry Layer-by-layer assembly of Au 4-ATP (CE-MPC) G4-PAMAM) n monitored by mass measurement with a QCM Cyclic voltammetry of 0.5 mm ferrocene Influence of scan rate on the cyclic voltammetry of ferrocene at an LBL assembly of CE-MPC Cyclic voltammetry at an ITO 3-APTES (CE-MPC) G4-PAMAM) 4 electrode Cyclic voltammetry of cysteine Scheme of modification of electrode with AuNPs Thermogravimetric analysis of generation-4 PAMAM with surface OHgroups AFM image of AuNPs on an ITO substrate Structure of muscovite projected along the sheet-stacking direction AFM of mica PAH (Au-PAMAM, PAMAM) after heating at 350 o C Influence of the AuNP density on the peak current for the oxidation of cysteine Influence of scan rate on the cyclic voltammetry of cysteine Linear scan voltammetry at mv s -1 of the oxidation of AsO - 2 at an ITO APTES (Pt-PAMAM, PAMAM) electrode Linear scan voltammetry of AsO - 2 at scan rates of 1 5 mv s ix

13 46 Influence of the PtNP density on ITO on the peak current for the oxidation - of AsO Calibration curve of cysteine at AuNP-ITO electrode Calibration curve of arsenite at PtNP-ITO electrode 180 x

14 Acknowledgements I would like to thank my mother, my husband and my daughter for all of their love, understanding and supporting me to pursue this degree. I wish my father could see me today. I would like to thank my advisor, Prof. James A. Cox, for his guidance, support and encouragement. Thank you for your lessons; I have learned a lot, about science and life, from you. I would like to thank my friends at Miami, Michelle, Dorota and Laisheng for help, enjoyment, and scientific and non-scientific laughs at Hughes. Special thanks to my dearest friends, Huong, Anna and John, whom I share my happiness and sorrows, and I can always rely on when I m down and need help. xi

15 Chapter 1 Sol-Gel Chemistry 1.1 Introduction Glass and other ceramics have been made classically by mixing fine-grained solid powdered materials, e.g., oxides such as SiO 2, Al 2 O 3, PbO, Fe 2 O 3 (or sometimes other compounds of the metals), and then allowing them to react in solid, but preferably in liquid state, at temperatures between 1000 to C for periods varying from hours to days, often several times in succession with regrinding between melting [1]. The term sol-gel was first introduced in the late 1800s. It generally refers to a low-temperature method using chemical precursors that can produce ceramics and glasses. Historically, Ebelmen, who synthesized ethyl orthosilicate, Si(OEt) 4, for the first time in 1845, observed that on standing at room temperature, Si(OEt) 4 was slowly converted into a glassy gel due to slow hydrolysis by atmospheric moisture. Si(OEt) 4, thus, can be regarded as the first precursor for sol-gel materials [2]. Ebelmen hoped that it could be used in the construction of optical instruments. However, his discovery was not the result of a systematic search for an alternative way of producing glasses. In the 1930 s Geffcken and Dislich of Schott Glass Company in Germany developed an economic way of covering industrial glass with a preparation of oxide films. In 1932, the process of supercritical drying to produce aerogels was first introduced by Kistler. Around this time, the theory that a gel consisted of a solid network with continuous porosity was widely accepted [3]. The interest in the sol-gel route was renewed by the hope of producing optical glass components at low temperatures without the need of subsequent meltings. However, the initial emphasis appears to be on the preparation of single oxide SiO 2 or TiO 2 optical coatings [4]. During the decade preceding 1956, mixed oxide materials were investigated. Roy and co-workers [5, 6] contributed the scientific studies on sol-gel process by introducing preparations of homogeneous ceramic powders and glassy melts of the system including Al, Si, Ti, Mg, Ca and Ba oxides from metal-organic derivatives. The new method was claimed to give final materials which, after just one melting, were more homogeneous 1

16 than the corresponding ones of the same components after several successive melting and crushing operations in the conventional process from individual oxides. The mixed oxides materials were also investigated later by Levene and Thomas [4]. An excellent summary of these achievements was presented in 1971 by Dislich in a publication [7], in which he concluded that the homogeneity of the final product indicated not only intimate mixing at the molecular level in the original solution containing alkoxides of different elements, but also formation of new bonds among the component. The rapidly growing interest by a large number of investigators of sol-gel processing is reflected in the amount of work published since the mid-1970s after it was demonstrated that monoliths could be produced by carefully drying the gel [8]. In general, sol-gel processing consists of three steps: i. Mixing low viscosity solutions of suitable precursors, which could finally yield the oxides, to ensure homogenization at molecular level. ii. Forming a uniform sol and causing it to gel. This is the key step in the process because an interconnected, rigid network consisting of submicron pores and polymeric chains is formed. iii. Removing solvent to form a condensed solid. In this step, the final form is shaped. The sol-gel technique appears attractive because it offers in principle several obvious advantages over the conventional glassy method [2]: 1) Lower processing temperature. 2) High homogeneity and purity of resulting materials 3) Possibility of forming products with various shapes. Sol-gel materials have been used in a wide range of compositions and in various forms, including powders, fibers, coatings and thin films, monoliths, and porous membranes. Organic/inorganic hybrids, where a gel (usually silica) is encapsulated with polymers or organic compounds to provide specific properties, can also be made. This chapter will review the chemistry involved in sol-gel processing, the synthetic conditions that can influence the properties of the resulting materials, and applications of sol-gels to analytical chemistry. 2

17 1.2 Overview of sol-gel processing Precursors and methods of synthesis In the sol-gel process, there are two kinds of precursors used for preparing a colloid: inorganic salts and organic compounds, of which metal alkoxides are most widely employed [3]. The use of precursors, metal salts and organometallics, divides solgel processes into two synthetic routes: aqueous-based and alcohol-based methods, respectively. In the former process, the sol forms when the metal salt is hydrolyzed. The next step is the formation of the gel, which occurs by either removing the solvent or increasing the ph. The network of the gel is produced from association of discrete colloidal particles. Differently, in the latter procedures, there is no distinct sol-formation step. Hydrolysis and condensation processes take place simultaneously until in the end a gel is formed. Here, the gel develops as an interconnected three-dimensional network. For both procedures the remaining steps after gelation, including aging, drying and sintering, are the same [9]. In the rest of this dissertation, only alkoxides of silicon, which involve the alcohol-base method, will be discussed Hydrolysis The initial step in alcohol-based sol-gel processing is hydrolysis, in which the sol is formed by mixing silicon tertrafunctional alkoxide precursors with water as in the following reactions: Si(OR) 4 + H 2 O HO-Si(OR) 3 + ROH (1) R is an alkyl group. The amount of water leads the hydrolysis to completion: Si(OR) 4 + H 2 O Si(OH) 4 + 4ROH (2) or stops while the silicon is only partially hydrolyzed: Si(OR) 4 + H 2 O Si(OR) 4-n (OH) n + nroh (3) The effect of the H 2 O : Si molar ratio (r) on the process and structure of the final material will be discussed in more detail in part It should be noted that water and alkoxysilanes are immiscible, so normally an alcohol is added as a mutual solvent for 3

18 homogenization. However, gels can be prepared without added solvent by applying ultrasound to homogenize water and alkoxides, as presented in part 1.5. The hydrolysis process is most rapid in the presence of catalysts, of which acids and bases are most generally employed. The hydrolysis occurs by bimolecular nucleophilic displacement reactions (S N 2) involving penta-coordinate intermediate or transition states [10, 11]. Under acidic condition, an alkoxide group is protonated in a rapid first step, withdrawing the electron density away from silicon and making it more electrophilic and susceptible to attack by water. In the next step, where a transition state is induced, water attacks silicon, reducing the positive charge on the protonated alkoxide and making alcohol a good leaving group. The transition state decays by displacement of alcohol accompanied by inversion of the silicon tetrahedron as shown in Figure 1 [3, 12]. Under basic condition, nucleophilic hydroxyl anions are formed from dissociation of water in a rapid first step. The hydroxyl anion then attacks the silicon atom and displaces OR - with inversion of the silicon tetrahedron (Figure 1) [3, 12] Condensation Condensation reaction involving the silanol groups produces siloxane bonds (Si- O-Si) and the by-products, alcohol (Eq. 4) or water (Eq. 5). Under most conditions, condensation begins before hydrolysis is complete [3]. Si-OR + HO-Si Si-O-Si + ROH (4) Si-OH + HO-Si Si-O-Si + H 2 O (5) As with hydrolysis, the condensation process can be catalyzed by an acid or a base. In both cases, the reaction proceeds via a rapid formation of a charged intermediate by reaction with a proton or hydroxide ion, followed by slow attack of a second neutral silicon species on this intermediate (Figure 2) [12]. Rate competition between hydrolysis and condensation is one of important factors influencing the structure of the final product; it will be discussed in part

19 Acid catalysed H H RO H + RO OR H O + Si OR δ + O Si O RO H RO OR R H δ + HO OR OR Si OR + ROH H + Base catalysed HO - RO Si RO RO OR OR RO OR OR HO δ Si OR δ HO Si + OR - OR OR Fig. 1 Hydrolysis mechanism (based on ref. 12) 5

20 Acid catalysed H + HO HO H OH Fast R + Si OH Si O+ + HO Si HO HO H R R OH Slow HO Si HO R O + OH R Si OH H 3 O + Base catalysed HO Si HO R Fast HO OH Si HO + OH - R + O - + HO Si HO R OH Slow HO Si HO R O + Si OH OH R H 2 O OH - Fig. 2 Condensation mechanism (based on ref. 12) 6

21 1.2.4 Gelation Hydrolysis and condensation result in formation of silica sol particles, which continue growing or aggregate to clusters. Gelation occurs when links form between these clusters to such an extent that a giant cluster spans across the containing vessel. At this point, although the mixture has a high viscosity, so that it does not pour when the vessel is tipped, many sol particles are still present as such, entrapped and entangled in the spanning cluster. The initial gel has high viscosity but low elasticity. At the gel point, the viscosity increases suddenly. After gelation, further cross-linking and chemical inclusion of isolated sol particles into the spanning cluster continues; the elasticity of the sample continues increasing [3] Aging As long as a gel maintains its pore liquid, its structure and properties continue to change long after the gel point. This process is called aging, which includes polymerization, syneresis, coarsening and phase transformations. These processes can take place singly or simultaneously. Polymerization is characterized by condensation reactions, which continue to occur as long as neighboring silanol groups are close enough to react. This increases the connectivity of the network and its fractal dimension. It can continue for months for samples at room temperature. The rate depends on ph, temperature and gel composition. The net effect of these processes is a stiffness and shrinkage of the sample. Shrinkage occurs because new bonds are formed where there were formerly only weak interactions between surface hydroxyl and alkoxy groups (Figure 3). This shrinkage leads to expulsion of liquid from the pores of the gel, which is known as syneresis. Another process associated with aging is coarsening or ripening. In this process, smaller particles of the gel dissolve and the solute precipitates onto larger particles. This results in an increase in the average pore size of the gel and decrease in the specific surface area, but does not produce shrinkage. In phase transformation, the solid phase can separate from liquid on a local scale or the liquid can separate into two or more phases. An important advantage of aging is an increase in strength of the gel. In general, the network becomes stronger when the gel is aged longer [3, 9, 13]. 7

22 HO OH O O a) OH HO OH HO O b) Fig. 3 Model for sol-gel shrinkage: a) Shrinkage result from condensation between neighboring groups on a surface. b) Movement of flexible chains may permit new bonds to form. This permits extensive shrinkage as long as the network remains flexible (based on ref. 3). 8

23 1.2.6 Drying The drying process involves four different stages. The first stage is the constant rate period, in which the gel shrinks by a volume equal to that of water or other liquid that evaporates. The rate of evaporation per unit area of the drying surface depends on time because the surface of the body is covered with a film of liquid, resulting in an evaporation rate that is similar to that of an open dish of liquid. During this stage a gel will typically shrink in volume by a factor of 5 to 10. Additionally, the concentration of any solute in the pore liquid increases dramatically and may result in precipitation of crystals [3]. This side-result can be exploited to grow crystals by using the sol-gel technique. This issue will be discussed in more detail and our preliminary data will be shown in Chapter 4. The next stage is called the critical point, where the gel is sufficiently stiff to resist further shrinkage as liquid continues to evaporate. At this point, the liquid recedes into the pores of the gel, leaving air-filled pores near the outside of the gel. Due to its surface tension and the small size of the gel pores, very large pressures are generated across the curved interfaces of the liquid menisci in the pore. The gel will crack due to this capillary stress unless it has been prepared to have optimum cross-linking or has been very carefully aged. The following stage is the first falling-rate period. Drying still continues but the rate of evaporation decreases due to: a) the liquid has to flow to the exterior and evaporate and b) the liquid captured in pockets can evaporate but the vapor has to diffuse to the surface. In this stage, there is a rise of capillary stress that may cause cracking. The final stage of drying is the second falling rate period, where the evaporation occurs inside the body because the liquid is isolated in pockets, stopping flow to the surface; it can escape only by diffusion of its vapor to the surface. The schematic of the drying process is shown in figure 4. The gel is subject to cracking during drying due to excessive capillary stress. There are two main conventional routines to avoid or reduce fracturing. First, the network of the gel can be strengthened by aging, so the gel will be stiff enough to overcome the capillary force. The resulting gel, which is dried under ambient conditions, is called a 9

24 a) Initial condition Liquid/vapor meniscus flat Pore liquid Solid phase b) Constant rate period Evaporation Shrinkage c) Falling rate period Empty pore Minimum radius of curvature Fig. 4 Schematic illustration of the drying process (based on ref. 3). 10

25 xerogel and is often reduced in volume by a factor of 5 to 10 compared to the original wet gel. Second, capillary pressure can be eliminated by supercritical drying where there is no liquid-vapor interface. The resulting gel is called an aerogel, which has a volume similar to that of the original sol. Kijak et al. [14] reported that silica prepared from tetramethoxysilane by an acid-catalyzed sol-gel process was strengthened by the inclusion of a generation-zero polyamidoamine dendrimer (G0-PAMAM) in the starting reaction mixture. Here, G0-PAMAM plays a role of crosslinking. The obtained silica monoliths were less subject to fracture when cycled in and out of water. Other methods such as using surfactants and drying-control chemical additives have been reported to be successful reducing fracturing as well [3] Sintering Sintering is a process of densification that is driven by surface energy to eliminate the pores within the gel. In gels, the solid-vapor interfacial area is so large that sintering can take place at relatively low temperature. Generally, there are three temperature regions in the densification process. At temperatures lower than C, weight loss occurs as pore surface liquid is desorbed, but little further shrinkage takes place. At intermediate temperatures from C to C, both weight loss and shrinkage proceed. At above C, the shrinkage rate increases sharply but with no further weight loss. Depending on the temperature condition of sintering, dense glasses or ceramics can be produced. For uses in analytical chemistry, especially applications which involve encapsulating functional organic or biological molecules within gel pores, silica gels are prepared and dried at or near room temperature [3, 9]. 1.3 Factors affecting gel structures Numerous studies have shown that variation in the synthetic conditions, such as the nature of the precursors, the value of r (water/alkoxide ratio), the catalyst type and concentrations, and the temperature cause modifications in the structure and properties of the sol-gel products. Understanding the effects of these factors on the structure and properties of the sol-gel will help to obtain the desired end product. 11

26 1.3.1 Precursors Alkyl chain lengths and branching degrees of alkoxy groups are found to have significant effects on the hydrolysis rate. Larger alkoxide groups have more steric hindrance and cause overcrowding of the transition state, thus leading to slower reactions. For example, the hydrolysis rate constant of tetraethoxysilane (TEOS) is 2.6 times greater than that of tetra-n-butoxysilane. Practically, the bulkier the alkoxy groups, the larger the pore diameter in the resulting gel [15] Water The ratio, water:si (r), influences the reaction rates, which in turn affect structures of the resulting products. From the stoichiometry of equation 2, the value of r for complete hydrolysis process is 4. In fact, less water than this can be used because, as discussed above, the condensation reaction, which usually commences before the hydrolysis completes, leads to production of water. However, if the amount of water is too small, the hydrolysis rate slows down due to the reduced reactant concentration. In contrast, a very large amount of water dilutes the other reactant (alkoxide), resulting increase in gel times. Theoretically, an r value of 2 is sufficient to complete hydrolysis and condensation, yielding anhydrous silica. In term of the product structure, the general trend is that acid-catalyzed hydrolysis with low r values yields weakly branched polymeric sols, while base-catalyzed hydrolysis with large r values produces highly condensed particulate sols. The r value has been experimentally tuned from 1 to over 50, depending on other synthetic conditions, especially ph and catalyst type, to obtain the desired end products. For example, to produce bulk gels, r values are 3.7 and 5.1 for 2nd step-acid and 2nd step-base, respectively. An r value of 1-2 with 0.01 M HCl as catalyst is used in the process for preparing fibers; 10.9, for films; and from 20 to over 50 with 1-7 M NH 3, for monodisperse, spherical particles [3] ph The Influence of ph on the morphology of the gel is explained in a very understandable publication by Curran and Stiegman [16]. The trend observed in this 12

27 report agrees with that in a review [3] from Brinker and Scherer. In the sol-gel process, when all the other factors such as precursor, solvent, r value and temperature are maintained constant, ph significantly controls the hydrolysis and condensation rates, which, in turn, dictate the pore structure of the resulting gel. The ph is divided into three regions: ph < 1; ph 2; and ph 3-8, based on the point-of-zero-charge (PZC) of silica, which is in the range of ph 1-3. In the region of ph from 3-8, a significant amount of silanol groups are deprotonated, pushing the condensation process relatively faster than the hydrolysis. It results in highly branched silica species, which are then gelated to a loosely packed, cluster-like structure. Therefore, mesoporous xerogel is yielded. As ph is lowered, the concentration of deprotonated silanol groups (SiO - ) decreases, making condensation the rate-determining step. The gelation time is increased and less-branched silica species are formed. When the ph approaches PZC of the silica, only neutral SiOH groups are present, thus the gelation times reach a maximum. This results in very linear or randomly branched silica species, which produce highly microporous xerogel (pore sizes less than 5 nm). In the region of ph well below PZC, the gelation process becomes rapid. This is attributed to the protonation of silanol groups, yielding SiO + 2 groups. The protonated silanol groups are readily condensed and thus increase the rate of this process. Similar as in the region of ph from 3-8, the fact that the condensation rate is rapid relatively to hydrolysis leads to more mesoporous material. 1.4 Template-based nanoporous silica As in the above discussion, pore structure of the gel can be controlled by manipulating ph. In this part, another approach to control pore sizes in mesoporous domain will be presented. It is known as the template-base approach. The first research of molecular templating to control pore size and shape of sol-gel was reported by Dickey in 1949 [17]. In this study, silica sol-gel was synthesized in the presence of methyl orange. When gelation occurred, the silica species were tightly organized around methyl orange molecules due to noncovalent interactions, such as van der Waals force, hydrogen bonding, and interionic attraction. The xerogel was then washed with methanol to extract 13

28 methyl orange out of the gel, leaving its footprint inside the gel. Interestingly, the resulting gel showed a selective adsorption of methyl orange over its analog compounds, e.g., n-propyl, and n-butyl orange. After Dickey s publication, molecular and supramolecular templating has been studied broadly because silica with tailor-made pore sizes and shapes is very important in applications of molecular recognition, such as shape-selective catalysis, molecular sieving, chemical sensing, and selective adsorption. A template is defined as a central structure around which a network forms in a way that removal of the template creates a cavity with morphological and/or stereochemical feature related to those templates [17]. A general template scheme is shown in figure 5. There are two types of interaction between template and matrix: a) non-covalent bonds; and b) covalent bonds. Depending on the purpose, the template can be kept after the gel formed or be removed. The ability to remove template mainly depends on three factors: i) The nature of interaction between template and the surrounding matrix. ii) The ability of the matrix to re-shape. iii) The relative sizes of the template and the original units used to build the matrix. In the following part, non-covalent bonded templates will be discussed. This type of templating includes molecules, polymers, and superamolecular arrays. When dissolved in solution, these species can template small organic groups via electrostatic, van de Waals, and hydrogen-bonding interactions to form nanostructured materials with tailorable pore sizes and shapes. Among these templates, surfactants and dendrimers will be the focus Surfactant templates In 1992, scientists at Mobil Oil Research and Development announced the direct synthesis of the first broad family of mesoporous molecular sieves (denoted M41S) using cationic surfactants to assemble silicate anions from solution. These materials exhibit large specific surface area (greater than 1000m 2 /g) and narrow pore size distributions similar as to those found in microporous crystalline zeolites. Unlike the crystalline zeolite 14

29 + Organic template x-linking Monomer Organic template in an inorganic matrix -T +T Pore that mimics the size and shape of the template Fig. 5 Schematic of the organic template, which is incorporated and removed, for preparation of amorphous silica (based on ref. 17). 15

30 materials with maximum pore dimensions of < 2 nm, the pores of M41S can be engineered with diameters from 1.5 to 10 nm. The relatively large pore sizes and ordered structures available make these materials very attractive for various applications such as catalyst supports for large molecules, shape-selective polymerization, miniature reactors and high surface sorbents, especially when the substrate molecules involved are too large to fit into the relatively small channels of conventional zeolites [18]. Besides controlling pore size of the gel, surfactant is also found to reduce cracking because it reduces the shrinkage at the critical point [3]. a) Overview of surfactants A surfactant is an organic compound consisting of two parts: 1) a hydrophobic portion; and 2) a hydrophilic portion. The combined hydrophobic and hydrophilic moieties make the compound surface-active and thus able to concentrate at the interface between a surfactant solution and another phase. Surfactants are divided into: 1. Anionics, where the hydrophilic portion of the molecule carries a negative charge, for example carboxylic acids (C 17 H 35 COOH or C 14 H 29 COOH), phosphates (C 12 H 25 OPO 3 H 2 ). 2. Cationics, where this portion of molecule carries positive charge. For example, alkylammonium salts, such as C n H 2n+1 (CH 3 ) 3 NX, n > 6, X = OH, Cl, Br, H 2 SO Nonionics, where the hydrophilic group is not charged. For example primary amines and poly(oxyethylene oxides) [19]. The amphiphilic nature of surfactants makes them able to associate into supramolacular arrays. The extending of micelle formation, the shape of the micelles, and the aggregation of micelles into liquid crystals depends on the surfactant concentration. Figure 6 is an example of a schematic phase diagram for a cationic surfactant, cetyltrimethylammonium bromide (C 16 TMABr), in water [17]. At very low concentration, surfactant exists as individual molecules in solution. When the concentration increases and reaches the critical micelle concentration (CMC1), the surfactant molecules form small, spherical micelles. If the concentration continues increasing to CMC2, the amount of solvent between the micelles reduces and spherical micelles can coalesce to form elongated cylindrical micelles. At higher concentrations, 16

31 Crystals in water Fig. 6 Schematic phase diagram for C 16 TMABr in water (based on ref.17). 17

32 liquid crystal phases form, including hexagonal close-packed liquid crystals, which consists of rodlike micelles; cubic bicontinuous phases; and lamellar liquid crystals. At very high concentrations, in some system, inverse phases form, where water is solubilized at the interior of the micelle, and the head group points inward. b) Synthesis of surfactant-templated sol-gel For a charged surfactant, electrostatic interactions are claimed to have important roles in the initial organization of the silica-surfactant arrays. For neutral surfactants, hydrogen bonding interactions are important [17]. The size and shape of the surfactant array direct the pore size and geometry of the silica. For example, the pore size can be controlled conveniently in ca. 4.5 Å increments by changing the length of the tail of C n TMABr (n = 8-16) [20]. The geometry of micelle shapes is dictated by the type of surfactant. For single chains and relatively large heads, spherical or ellipsoidal micelles are expected in the aggregate structure. Simple surfactants with relatively small head groups in the aggregation form relatively large cylindrical or rod-shaped micelles. Vesicles or bilayer structures are formed from surfactants having double chains with large heads. Generally, single-chain surfactants are employed in mesophase synthesis, while those that have double chains are avoided because they tend to form lamellar mesophases, which collapse when the template is removed [21]. In the synthesis of surfactant silica mesophases, water, catalyst and surfactant are first combined to form a homogeneous micellar solution. Then alkoxysilane, such as TEOS or TMOS, is added to this micellar solution. A homogenizing agent such as methanol or ethanol can be added to the precursor solution to ensure homogeneity and maximize the product yield. Generally, the mesophase forms in seconds to minutes at temperature as low as -14 o C [22]. The ph is set based on the type of surfactant to facilitate the electrostatic interactions between surfactant and precursors. For example, when a cationic surfactant is chosen, the ph is set at 9-14, so the precursors will be negatively charged. For neutral surfactant, the reaction is done at near-neutral ph [23]. 18

33 Three members of the M41S family of materials were distinguished (figure 7): hexagonal (MCM-41), a 1-d system of hexagonally arrayed cylindrical pores; cubic (MCM-48), a 3-d, bicontinuous system of pores; and lamellar (MCM-50), a 2-d system of metal oxide sheets interleaved by surfactant bilayers [17, 20]. Among them, MCM-41 has been the focus of most study due to its uniform mesopores. c) Mechanism of MCM-41 formation There are two pathways of formation of silica/surfactant mesophases, as shown in figure 8, depending on the concentration (c) of surfactant [17, 18, 20, 24]. The mechanism follows path 1 when c > H 1, which leads to formation of liquid crystal phases. Silica species are deposited between surfactant tubules and then condense to form the inorganic network. In this case, the surfactant arrays serve as templates and the formation of the ordered mesophase is effectively independent of interfacial interaction between the surfactant and silica. Path 2 is operative when c < H 1, even at c < CMC1, where silica is present as free molecules. The reason is that adding silica significantly changes the system thermodynamics, pushing the concentration axis in the phase diagram (figure 6) to the right and expanding the liquid-crystalline phase space in the solvent:surfactant:silica diagram. In this case, the interaction of silica species with the surfactants is very important because it mediates the hexagonal ordering, when there are no liquidcrystalline phases in solution prior to adding silica. It is hypothesized that silica and surfactant cooperatively organize into an organic-inorganic liquid-crystalline phase during the course of the reaction, but the details of the mechanism are still controversial. Therefore, though hexagonal liquid crystals form at surfactant concentration H 1 = 30% (wt), to obtained MCM-41, surfactant can be used at concentrations as low as ca. 0.5% (wt) [22]. It should be emphasized that the possibility of MCM-41 formation also strongly depends on the size of hydrophobic portion of the surfactant molecule. Beck et al. [25] observed that, depending on the chain length, alkyltrimethylammonium surfactants could serve as structure-directing agents, or templates, for the formation of microporous or 19

34 Fig. 7 Three structure types observed for silica-surfactant mesophases: (a) hexagonal (MCM-41); (b) cubic bicontinuous (MCM-48); and (c) lamellar (MCM-50) (based on ref. 17). 20

35 2 Silicate Silicate 1 Surfactant micellar rod Hexagonal micelle array H 1 Calcination MCM-41 Fig. 8 Schematic of the liquid-crystal templating mechanism. Path 1 is liquid crystal initiated and path 2 is silicate anion initiated (based on ref. 17). 21

36 mesoporous silica frameworks, respectively. With short alkyl chain lengths, C n 6,TMABr surfactants produced microporous material because it did not form micelles in aqueous systems and did not form MCM-41 in the presence of silica. Surfactants with n = 8-16 form MCM-41 and, thus, form mesoporous materials. The longer the chain of hydrophobic portion, the greater the favorability of micellization at low surfactant concentration. d) Template removal There are four methods to remove the template: i) Solvent extraction, ii) calcination, iii) oxygen plasma treatment, and iv) supercritical extraction. The first two methods have been well studied. Ethanol has been mostly used as a solvent to extract organic groups from surfactant-templated material. For example, acidic ethanol was used to extract organic groups from electrostatically templated material. Up to 100% of the template can be removed when 1M HCl solution was used at 70 o C [26]. Pure ethanol was used at room temperature to extract up to 85% of organic components from acidsythesized ion-mediated MCM-41 [27]. Hot ethanol was used to extract ca. 100% of organic groups from a neutrally templated silica sample [28]. The MCM-41 samples can be calcinated in a flow of N 2, O 2, or air at temperatures above 500 o C to remove the templates [18, 20, 26] Dendrimer template Ideally, dendrimers are perfect monodisperse macromolecules with a regular and highly branched three-dimensional architecture. Dendrimers are produced in a repeating sequence of reaction steps, in which each additional interaction leads to a higher generation material. The supramolecular properties of dendrimers can be effectively tailored by the introduction of desired functional groups at either the core, the peripheral surface, the branching unit, or at multiple sites within the dendrimer [29]. The first dendritic structures that have been thoroughly investigated and received widespread attention are Tomalia s PAMAM dendrimers [30] and Newkome s arborol system [31]. In figure 9, the structure of the so-called Starburst poly(amidoamine) generation 4.0 (G4- PAMAM) is shown. 22

37 Fig. 9 Structure of G4-PAMAM-NH 2 (based on ref. 30) 23

38 For templating, dendrimers offer the possibility of various chemical interactions. For example, highly charged dendrimers of opposite charge might assemble via electrostatic forces into superlattice structures. Another approach involves the covalent assembly of dendrimer building blocks into network structures using coupling reactions between the functional end groups on the dendrimer surface. Spherically shaped dendrimers may be expected to pack into networks possessing interstitial spaces, whose size is influenced by the radius of the dendrimer building block. With this approach, it is possible to generate new classes of porous materials with tunable pore sizes and inclusion properties. Dendritic macomolecules possess physical properties that, in many cases, greatly differ from their linear analogs. For example, a monodisperse structure of a dendrimer is built by generations (layer by layer) around a core moiety. In organic solvents, they exhibit high solubility and low viscosity compared to their linear analogs. When playing a role of a template, dendrimers are also different from surfactants. They are unique in that: 1) they have the potential, depending upon dendrimer size, to produce mesopores via single molecules instead of micelles; and 2) spheroidal voids are imprinted on inorganic solids. The first report on the synthesis of dendrimer-mediated mesoporous silica is a study of the porosity of carbosilane dendrimer/sol-gels without template removal by Kriesel and Tilley in 1999 [32, 33]. They hydrolyzed second and third generation triethoxysilyl-terminated carbosilane dendrimers via a sol-gel protocol to micro- and mesoporous hybrid dendrimer-silica xerogel, which could be used as catalyst supports. Total surface area and pore volume of the obtained xerogel increased with generation number. Dendrimers have open and vacuous structures characterized by channels and pockets which are especially true for higher generations. Unlike first and second generations, the higher generation dendrimers have greater internal surface area compared with the external surface area. Therefore, the third and higher generations should be well suited for applications where high surface area (both internal and external) is required. Because of their tree-like branched architecture, functionalized dendrons are potential candidates for novel sorbents to be used in analytical sample enrichment and 24

39 separations. Kabir et al. [34] developed a benzyl-terminated, dendron-based sol-gel coating for capillary microextraction and preconcentration. Here, phenyl-terminated dendrimer generation 3 was chemically bonded with triethoxysilane. The resulting materials efficiently extracted both polar and nonpolar analytes from aqueous samples. Different from the studies mentioned above, where dendrimers were used as covalently bonded templates and were not removed, Larsen and co-workers [35] are the first ones who used poly(propylene imine) generation 5 ( G5-PPI) and PAMAM generation 4 (G4-PAMAM) as non-covalent templates. In their first report, G5-PPI was used for silica made from TEOS with a ratio of ~1.5 Si per dendrimer terminal amine group. After the template was removed by calcination, the maximum in the pore size distribution of the mesoporous silica was ~23 Å, whereas the hydrodynamic diameter of the soft-core G5-PPI is 39.6 Å. It was hypothesized that upon heat-induced cross-linking of the gel most dendritic centers adopt a dense-core conformation, which has a molecular diameter of ~25 Å. In the next study, Larsen and co-workers [36] used G4-PAMAM dendrimer as the template. The number of Si atoms per surface amine group in the PAMAM molecule was 6, which indicates that somewhat dilute packing of voids is expected. They reported that template removal by solvent extraction (CH 2 Cl 2, methanol and ethanol) was ineffective compared to the calcination method. The resulting gel after calcination had a pore size with a maximum around Å, while the size of G-4 PAMAM is 40 Å. It was explained that contraction upon heating takes place, which is conceivably due to an inward collapse of the low-density dendrimer structure. This phenomenon leads to templated microcavities. It was suggested that more dense-dendrimer templates (e.g., polyether dendrimers) might be able to resist contraction upon heating. It is noteworthy to know that very low PAMAM generation dendrimers do not act as templates. Generation-zero PAMAM (G0-PAMAM) dendrimer was proposed as a crosslinking agent by interaction of protonated amine sites with silica strands [14]. In this case, the inclusion of G0-PAMAM strengthened silica prepared from TMOS, decreasing the tendency of the gel to fracture or rupture under load and when cycled in and out of water. In contrast, G4-PAMAM was not an effective crosslinking agent because of its 25

40 size. It served as a templating agent similar to a liquid crystal that leads to mesoporous silica when the template is not removed [14]. The chelating capacity of such amine dendrimers can be used to trap metal ions in the composites to ultimately make, after high-temperature and oxidation/reduction treatments, embedded metal oxide and metal clusters of tunable sizes. PPI and PAMAM dendrimers, which have amine groups, are able to chelate many metal ions, such as Cu(II), Au(III), Pt(II), per molecule. Cox and co-workers [37] used G4-PAMAM doped silica to encapsulate Prusian Blue, iron(iii) hexacyanoferrate and cobalt hexacyanoferrate by immersing G4-PAMAM doped silica in a desired metal hexacyanoferrate solution followed by reaction with appropriate reagents. Here, the uptake of hexacyanoferrate was facilitated via ion-exchange of the amine sites of PAMAM. The resulting materials were used in solid state voltammetric experiments. They showed that the G4-PAMAM remarkably increased the lifetime of the silica in terms of its ability to act as a solid electrolyte. Velarde-Ortiz and Larsen reported the use of G5-PPI as a macrochelating agent to produce well-dispersed CuO nanoclusters embedded in a silica sol-gel derived matrix [38]. Ruckenstein and Yin [39] also reported on a similar strategy to incorporate copper in silica/pamam dendrimer composites. 1.5 Sonogels Because the alkoxide precursor and water are immiscible, a mutual solvent such as alcohol is normally used as a homogenizing agent in the conventional method. However, alcohol is not simply a solvent because it is a product of both hydrolysis and condensation reactions. In 1984 Tarasevich [40] proposed an approach to sol-gel processing eliminating the use of additional solvent (alcohol) by exposing the TEOSwater mixture to intense ultrasonic irradiation. Thereafter, the Zarzycki and Esquivias groups conducted extensive work to establish the practical consequences of this approach on the kinetic and textural characteristics of the so-called sonogel [41]. Generally, for high ultrasonic intensity, a high-power ultrasonic horn, by means of a 13 mm titanium tip driven by an electrostrictive device, is used to generate ultrasonic waves [42]. An ultrasonic bath can be used alternatively. However, in these cases, a much lower 26

41 ultrasonic intensity is supplied to the reactants and, consequently, characteristic sonogel features may not be the same as in the former cases [43]. According to Donatti and Vollet [44], the hydrolysis rate of a co-solventless system is practically zero during an initial ultrasonic process due to the immiscible gap of the TEOS-water mixture. Ultrasonic energy acts as an effective starter for the reaction, promoting cavitation and increasing the effective contact area between the alkoxide and water; hence, the reaction commences. Alcohol produced from the hydrolysis helps the mutual dissolution causing further enhancement of the hydrolysis. Jonas et al. [45] used nuclear magnetic resonance spectroscopy of 29 Si to monitor the state of bridging groups (SiO 4 ) during gelation by sonochemical and conventional methods. They observed that after 20 minutes of sonication the sonosol consisted of 42% of network-forming species, whereas in the corresponding classic sol only 10% of the species were in network. In terms of the reaction mechanism, this means that sonication promotes hydrolysis. The concentration of silanol groups is higher and consequently the rate of bridging-o formation is also higher. As results, the gelation time decreases drastically in comparison with the classic technique, where alcohol is present. At constant temperature, the time for gelation decreases with increasing ultrasonic dose [46]. In addition, when high energies are used, cavitation takes place due to collapse of vapor bubbles in the liquid subjected to ultrasonic waves, producing local high pressures and hot spots, which accelerate hydrolysis and condensation reactions [47]. SAXS study by Ramirez-del-Solar et al. [48] during gelation shows that the sonogel structure is coherent with a model of statistical balls. These tangles of gel compact during aging, giving rise (after drying) to a very homogeneously packed aggregate formed from ~1 nm spheroid particles. Consequently, sonogels have a very high bulk density and surface/volume ratio, two or three times higher than gels prepared in alcohol solution. Their very fine and uniform porosity and high apparent density besides the short gelation time, are unique features of sonogels. These parameters are sensitive to the ultrasonic dose applied. The larger the dose, the finer the porosity and the higher the homogeneity. 27

42 1.6 Physical characterizations Porous materials are most frequently characterized in term of specific surface areas and pore sizes. Pores are classified according to pore diameter by IUPAC [49] as follows: micropores have diameters less than 2 nm; mesopores have diameters between 2 and 50 nm; and macropores have diameters greater than about 50 nm. Different methods have been used to determine these two parameters. Examples include gas adsorption porosimetry, mercury porosimetry, thermoporosimetry, small-angle X-ray scattering, fluid flow methods (permeation and diffusion) and NMR relaxation measurements. Optical and electron optical methods are used for qualitative characterization and increasingly for quantitative analysis using image analysis methods. Gas adsorption porosimetry, which was use in our research, will be described in details Basic concepts The adsorption and desorption isotherms The adsorption isotherm is an incremental set of data which describe how much adsorbate gas condenses on to a material at a given pressure and at a constant temperature. The volume of gas adsorbed is reported at standard temperature and pressure (STP) conditions. Desorption is the reverse process and is a decremental set of data reported in the same unit. Relative pressure p/p 0 As gas is being contacted to the sample surface, not all the molecules are adsorbed. Those that are not adsorbed form a residual pressure (p) in the sample chamber. This value of sample pressure divided by the saturation vapor pressure (p 0 ) gives the relative pressure. Saturation vapor pressure The saturation vapor pressure may be considered to be the boiling pressure of the liquid gas. It varies with temperature. In the typical analysis setup, liquid nitrogen is used for the sample coolant bath. The bath temperature will change with the atmospheric 28

43 pressure and the purity of the nitrogen used. In an open container atmospheric gases will continually condense in the liquid nitrogen and so contaminate the liquid. The level of contamination is unknown unless the temperature or saturation pressure is monitored. It is therefore important for the most accurate results, that the saturation vapor pressure is measured throughout the duration of sample analysis [50] Theory Initially, Langmuir described a method to determine the surface area of solids using the physical adsorption of inert molecules onto their surface. The physical adsorption of gas molecules is driven by van der Waals-London forces. If the adsorption of a gas is measured at a temperature well above the condensation temperature of that gas, only a monlayer is formed, i.e. a second layer will not build onto the first one. By measuring the maximum amount of gas adsorbed the molecular area follows from: Total surface area = (number of adsorbed gas molecules) x (area per molecule) In the derivation of the isotherm describing monolayer adsorption, Langmuir used the following assumptions: The surface of the adsorbent is flat All adsorption sites are energetically equivalent The adsorbed gas molecules do not mutually interact The adsorbed gas molecules have a fixed position on the surface. However, in the real world, these assumptions are rarely met: Surfaces are never flat and adsorption sites are not energetically equivalent. In addition, adsorbed gas molecules do have mutual interactions, especially at higher surface coverage, and are highly mobile. Not taking into account surface heterogeneity causes deviations at relative pressures p/p 0 below 0.005, whereas neglecting lateral interactions between the adsorbed gas molecules causes a more serious mismatched at higher pressures (p/p 0 < 0.35). Brunauer, Emmett and Teller developed a more practical model (BET) which still used the above-mentioned suppositions, but allowed for the adsorption of monolayers. In this model the assumption is made that the heat of adsorption of the first layer is higher than that of the following layers. In these other layers the heat of adsorption is assumed to be equal to the latent heat of condensation of the adsorbed gas. The BET equation reads 29

44 p 1 (c 1)(p/p0) = + 0 v(p p) v c v c m m (6) where v is volume of the gas adsorbed (at STP) and v m is the volume of gas (STP) adsorbed in the monolayer; c is equivalent to exp[(q-l)r], in which Q is the heat of adsorption of the first layer and L is the latent heat of condensation of the gas [15]. The adsorption isotherms generally can be grouped in 6 classes as in figure 10 [49]. Isotherm type I is typical for adsorption in microporous materials, where the BET equation is not applicable because of the strong adsorption forces in micropores resulting from the overlap of force fields from opposite pore walls. Consequently, adsorption in micropores occurs at very low pressures (often less than 10-6 atm) by a pore-filling mechanism. Types II, III, and VI are characteristic for non-porous or marcroporous materials. For mesoporous materials isotherms of types IV and V generally hold. Isotherm types IV and V show hysteresis in the adsorption-desorption cycles. This is explained by the Kelvin equation which says that the vapor pressure of curved surfaces differs from that of planar surfaces. As a consequence the rate of desorption in pores can differ from the rate of adsorption, depending on the size and shape of the curve. This means that the shape of the isotherm can be related to the pore size distribution in the material. By applying the Kelvin equation in an appropriate form, a graduated desertion experiment can be used to obtain a pore volume distribution [51] Surface area and pore size analyzer, Coulter SA 3100 In our research, pore size distributions and surface area of sol-gel materials were measured by using a Coulter SA 3100 analyzer. Besides Langmuir and BET methods presented above, t-plot and BJH methods are also employed. This combination allows us to determine pore sizes and surface areas of microporous, mesoporous and macroporous materials. The choice of method depends on the analyzed sample. The BET calculation is the most commonly used for characterization of specific surface area. The Langmuir surface area calculations are applicable to adsorption isotherms data which have type I characteristics, where the volume adsorbed is raised rapidly at very low relative pressure values. 30

45 Fig. 10 IUPAC classification of adsorption isotherms (based on ref. 50) 31

46 In t-plot (thickness-plot) method, the volume of gas adsorbed is plotted against the adsorbate molecular film thickness. The thickness of the film is calculated from either the Halsey calculation (equation 7) or from the Harkins and Jura calculation (equation 8). 1/ t = (7) p0 log p 1/3 5 t = 3.54x (8) p xlog p This t-plot data is then used to calculate the Y-intercept value, which when converted from a gas volume to a liquid volume gives the micropore volume. The slope of the linear section is used to calculate the meso/macro pore surface area by applying the following equations: VOLUME micropore = (t-plot intercept) (9) SURFACE AREA mesopore = 1547(t-plot slope) (10) The pore size distributions are obtained by the analysis of either adsorption or desorption isotherm branches based on the BJH (Barrett, Joyner, and Halenda) method. It is assumed that pores are cylindrical, open-ended and pore networks are absent. The method involves the area of the pore walls. It uses the Kelvin equation to correlate the relative pressure of nitrogen in equilibrium with the porous solid to the size of the pores where capillary condensation takes place [50] 1.7 Applications Sol-gel technology provides an alternative route to the production of ceramic and glasses. Compared to conventional techniques, the sol-gel process offers a number of important advantages that make the method interesting for the production of materials with specific applications. The main potential lies in the fact that sol-gel glass can be produced and processed at low temperatures and under mild chemical conditions. This 32

47 allows the entrapment of molecules with poor thermal stability such as proteins and enzymes. Another advantage is that these materials are obtained from solution which allows the convenient production of films, powder, fibers, and bulk materials of any possible shape. The third major advantage of the sol-gel method is that it produces porous materials whose pore-size distributions can be controlled, both by the chemical composition of the starting material, as well as by the processing conditions. In addition, sol-gel glass is optically transparent (down to 250 nm), which is good for spectrophotometric and spectrofluorimetric measurements. When applying the sol-gel technique, some disadvantages should be taken into account. The sol-gel method is expensive and time consuming. Sol-gel based sensors may have a slow response in aqueous media, particularly when thick films or monoliths are used. The response is usually limited by the diffusion process which is dependent on various parameters such as pore sizes, thickness of the film, and size and concentration of the analyte. Leaching, particularly for small reagent may be a problem. Entrapment in sol-gel glass may slightly change spectroscopy and chemical properties and biological activities of the reagent. Nevertheless, sol-gel materials have various applications in analytical chemistry, which will be summarized below Sol-gel based sensors In 1990, Avnir and co-workers [52] first reported sol-gel glassed doped with ph indicators as ph sensors. It was found that the entrapped indicators behaved similar to those in solution, exhibiting an isosbestic point. With the use of various ph indicators, spectroscopic determination of ph in different ranges was possible. The doped glasses were casted as monoliths and were immersed in the solution for ph measurement. Thereafter, fiber-optic ph sensors were developed [53]. Avnir and co-workers [52, 54, 55] were also the first to report the feasibility of using doped sol-gel materials as chemical sensors for various metal ions, such as Ni 2+, Cu 2+, Fe 2+, Al 3+, Co 2+ and Pb 2+. The determinations of metal ions were based on colored complexes with multiple ligands. For example, a Fe 2+ sensor based on sol-gel doped with 1,10-phenanthroline produced a detection limit less than 0.1 ppm [54]. The low detection limit of the sensor was attributed to its ability to pre-concentrate the analyte into the glass 33

48 phase by chelating with entrapped ligands. The feasibility of these sensors implies that the entrapped ligands have relatively high degree of freedom to move or to reorient inside the pores of sol-gel material in order to form the complexes. Gas sensors have been also developed. Examples include optical detection of NO using cobalt tetrakis(5-sulfothienyl) porphine doped silica sol-gel. This device was demonstrated with the use of optical fiber as a waveguide [56]. Sol-gel doped biological reagents such as myoglobin, cytochrome c, hemoglobin were also employed for sensing of NO, CO, O 2 [57, 58]. Cox et al. [59, 60] developed an amperometric gas-phase sensor for ammonia by coating a V 2 O 5 sol-gel overlayer onto an integrated microelectrode array that was modified with a ruthenium-base catalyst. An H 2 O 2 -selective electrode was developed by entrapping a layer of horseradish peroxidase between the surface of a carbon paste electrode and a thin film of undoped sol-gel glass [61]. Holmstrom and Cox [62] prepared an electrode using nanostructured silica for solid-state voltammetric determination of gaseous hydrogen peroxide. Here, solids, which were prepared by a base-catalyzed sol-gel process, extracted hydrogen peroxide from the gas phase and served as electrolyte for the voltammetric oxidation of HO - 2 in the absence of a contacting liquid phase. Glucose sensors based on enzyme-containing sol-gels have been most investigated. For example, Zink and co-workers [63] studied a system involved two enzymatic reactions. First, glucose oxidase catalyzed the oxidation of glucose to gluconic acid and hydrogen peroxide. Second, peroxidase catalyzed the reactions of dye precursors with hydrogen peroxide to produce color dyes. Thus the glucose could be determined spectroscopically. Tatsu et al. [64] also reported the entrapment of glucose oxidase in a sol-gel material and its application as a glucose sensor. Bright and co-workers [65] investigated three different schemes to immobilize glucose oxidase: i) physisorption onto sol-gel film; ii) entrapment in a sol-gel matrix; and iii) a layer system of a plain sol-gel film: glucose oxidase layer: plain sol-gel film sandwich configuration. There have been also many other glucose selective electrodes developed based on sol-gel derived glasses. Based on the principles similar to the glucose sensors, sol-gel sensors for other organic compounds have been introduced. For example, sol-gel sensors for determination of urea have been developed based on encapsulated urease, in which ammonia is 34

49 produced by enzymatic hydrolysis of urea and then reacts with Nessler s reagent (K 2 HgI 4 ) to form a colored product (NH 2 Hg 2 I 3 ) [66]. In addition to sensor development, many investigations have been conducted on the use sol-gel materials as an inorganic solid matrices for entrapment of laser dyes (to produce dye laser materials or devices) [67] and proteins, enzymes, and microbial cells (for the development of biological catalyst and reactors) [68-70] Sol-gel materials as solid matrix for organic dyes, proteins and enzymes Silica glasses prepared by so-gel techniques have been found to be a very good matrix for some organic dyes such as Rhodamine 6G, which can be used as probes for some chemical compounds and properties [71]. In early 1971, Johnson and Whateley [72] reported their investigation on the use of silica gel for the immobilization of trypsin, an enzyme with esterase activity. They observed that the esterase activity of trapped trypsin was 34% relative to that in solution and that storage at 4 o C for 75 days resulted in less than 10% loss in activity. Avnir and co-workers [73-75] reported on the preparation of bochemically active sol-gel materials made by entrapping alkaline and acid phosphatase, trypsin, and other enzymes. For alkaline phosphatase, the entrapped enzymes has 30% activity compared to that in solution, and the bioactive sol-gel glass was preserved in water at room temperature for two months without losing activity. The entrapped trypsin activity was stable when sol-gel glasses were incubated at ambient temperature and ph 7.5 for several months. The stabilization effect was attributed to the interaction of enzymes with silica matrix (i.e., Si-OH groups). In 1992, Zink and co-workers [76] studied the application of sol-gel material for encapsulation of some proteins by a modified sol-gel procedure. A buffer was added after HCl-catalyzed hydrolysis of TMOS and before the proteins were added. The addition of buffer brought the ph to above 5.5, thus preventing the acid denaturation and aggregation of the proteins. Proteins encapsulated in sol-gel matrices were observed to have similar reactivities and spectroscopic properties to those found in solution. 35

50 1.7.3 Sol-gel derived selective electrodes Many glucose selective electrodes have been developed based on sol-gel doped with glucose oxidase. The glucose electrodes were constructed by different techniques with various base electrodes: by enclosing doped sol-gel on the tip of an oxygen electrode [64]; by coating films or attaching slides of doped sol-gel glass onto a Pt electrode [77]; on indium tin oxide electrode [65] or a glassy carbon electrode [78-80]; and by printing doped sol-gel glass onto alumina ceramic substrate [81]. Glucose concentration was determined by amperometric or voltammetric measurements. In some of the investigations, redox mediators were used to facilitate the electron transfer [78, 80, 82]. Lev and co-workers [79, 82, 83] developed composite electrodes by dispersing graphite powder and enzymes or chemicals in the sol-gel derived carbon-silica materials. Selective electrodes for determinations of ph, halides, and glucose were demonstrated. Such constructed electrodes were shown to have some advantages over carbon paste and glassy carbon electrodes [83]. A novel technique was developed to fabricate strip-type of composite electrodes with the use of standard printing technology to print enzyme-doped sol-gel onto ceramic substrates [81]. The ink was prepared by dispersing the enzymes along with graphite powder and a binder (hydroxylpropyl cellulose) in the sol-gel solution. With this technique, glucose and peroxide selective electrodes were fabricated [81] Solid phase extraction Derivatizing silica gels with complexing agents and organic functionalized compounds also have applications in solid phase extraction (SPE) and chromatography techniques. For example, Garg et al. [84] used 3-hydroxy-2methyl-1,4-naphthoquioneimmobilized silica gel for the adsorption and estimation of zinc, copper and cobalt in milk, steel and vitamin by both batch and column techniques. Cichna and Markl [85] investigated the selective uptake of pyrene in the presence of structurally related polyaromatic hydrocarbons by encapsulating polyclonalanti-pyrene antibodies in a solgel material to provide the solid phase. Seneviratne and Cox [86] synthesized complexing agent-doped silica for SPE of metal ions via two means of immobilizing the agents, namely encapsulation in the pore structure and covalent bonding, in consideration of pore 36

51 size effects. Here, the pore width of sol-gel host was controlled in microporous and mesoporous domains by synthesis at ph of PZC and an inclusion of surfactant, respectively. With mesoporous silica as the host, the exchange capacity was higher than that of analogous material with pore sizes in microporous domain. However, a leaching problem was faced in the mesoporous material because the pore widths exceed the sizeof the molecules of the hosted reagent. The authors solved the leaching problems by employing a mesoporous host to which a complexing agent was covalently bound. Recently, Kabir et al. [34] developed a benzyl-terminated dendron-based sol-gel coating for capillary microextraction (CME). The coating material performed solventless extraction of a wide range of polar and nonpolar analytes. It is claimed that due to the strong chemical bonding with the capillary inner walls the sol-gel dendron coatings showed excellent thermal and solvent stability in CME in hyphenation with chromatographic analysis. 37

52 Chapter 2 Removal of Cesium and Polynuclear Transition Metal Hexacyanometallates 2.1 Introduction As a result of the end of the cold war, the Department of Energy (DOE) began an important program of environmental restoration and decommissioning within the sites that were once used for the production of basic materials for nuclear weapons [87]. Its main objective is to search for alternative solutions for the fast remediation and treatment of all of the radioactive, hazardous, and mixed wastes from basins, burial grounds, landfills, pits, piles, and tanks that resulted as a consequence of almost 40 years of the operation. Many highly radioactive fission products and byproducts are found within these places, among which 90 Sr, 137 Cs, and 60 Co are the most important. 90 Sr and 137 Cs are the main fission products of spent fuels, and because of this, they are found in almost all of the radioactively contaminated places, preferentially in the aqueous phases. 60 Co, in contrast, does not come directly from the fallout of nuclear fuels but as a consequence of the impurities present in the stainless steel that confine nuclear reactors. It is reported that over 90 million gallons of high-level and low-level radioactive waste have been stored in underground tanks at the DOE s Hanford Site, the Savannah River Site (SRS), Idaho National Engineering and Environmental Laboratory (INEEL), and the Oak Ridge National Laboratory (ORNL) awaiting pre-treatment and safe disposal [87]. It is costly and not practical to dispose of all the waste as high-level wastes (HLW). A large portion of the waste is present as a liquid that contains soluble radionuclides, along with high concentrations of sodium and potassium salts. In general, the radionuclides constitute less than 5% of the waste, and cesium is the primary radionuclide [87]. As a result, if the radionuclides are extracted, reducing the radioactivity below the Nuclear Regulatory Commission limits, the majority of the waste could be disposed of as low-level waste (LLW). Then, the concentrated radionuclides would be combined with HLW tank sludge solids and sintered into HLW glass. This chapter will summarize methods which have been used and investigated for removal of cesium. Among these 38

53 methods, the use of polynuclear transition metal hexacyanometallates will be the focus. Synthesis and structure of metal hexacyanoferrates, and the mechanism of cesium removal by using these compounds will be discussed in detail. 2.2 Radioactive cesium Radionuclides emit ionizing radiation that can damage living organisms by initiating chemical reactions in tissues. It takes ten half-lives to lose 99.9% of the radioactivity [88]. If the half life is seconds or minutes, there will be no harm because most of the radionuclides are decayed before they have a chance to enter the food chain or water supply. Similarly, if the radioisotope has a very long half life, for example 99 Tc with half life of 2 million years, there will be no significant amount of radiation generated during its lifetime. Radioactive cesium is 137 Cs, which has a half life of 30.1 years [88, 89]. Thus, it will generate significant gamma-ray radiation and can be harmful to living things (depending on the amount absorbed). 2.3 Methods of cesium removal Many different processes have been investigated for removing cesium from aqueous solutions. They include traditional methods (such as precipitation, where cesium is co-precipitated with copper, cobalt, nickel, and mercury ferrocyanides [90]; complexation, for example the use of crown-calixarenes and crown ethers, which selectively complex with cesium [91-93]; and ion exchange and absorption, which will be discussed in detail later) as well as less conventional or new ones. Of the latter some examples follow. Electrically switched ion exchange is a technique which combines ion exchange and electrochemistry [94]. The ion uptake and the elution can be controlled directly by modulating the potential of an ion exchange film that is electrochemically deposited onto an electrode. Lin et al. [95] developed self-assembled monolayers (SAMs) on mesoporous supports for selective sorption of cesium. The first step is preparing a silica ceramic that contains ordered arrays of cylindrical pores, which have a controlled diameter, for the sorbent. The second step is covalently coating the inside surfaces of the channel-like pores with a monolayer of molecules. Next, copper(ii) ferrocyanide is immobilized on the surface of the monolayer, so that any cesium that binds to it also will 39

54 be immobilized and removed from the solution. Here, the exceptionally high surface area of the mesoporous silica support gives the SAMs a very high binding capacity. Other methods, using materials such as activated carbon, coal fly ash, bauxite wastes, coal ash, and even magnetic particles, also have been reported [88]. Among all methods, the ion exchange and adsorption techniques, using organic and inorganic materials, are most commonly utilized for the extraction of cesium from aqueous solutions. These types of materials will be discussed below Organic ion exchangers An early organic ion exchanger was Dowex-50, a cation-exchange resin, which contains sulphonic acid groups [96, 97]. The Dowex-50 had selectivity for cesium and other alkali metals in the order Li + <Na + <K + <Rb + <Cs +. However, this material was not able to complete separation of small concentrations of cesium from large concentrations of sodium because the separation factor between Na + and Cs + was not large enough. In the same family with Dowex-50 were Duolite C-3 and Diaion SK1 organic resins [98-100]. These materials were bifunctional resins, including phenol groups in addition to sulphonic groups. They have a specific selectivity for cesium from other alkaline solutions, which was attributed to the complex formation between Cs + and phenol groups. Koyama et al. [101] developed a complex anion, [Cr(NH 2 C 6 H 5 ) 2 (NCS) 4 ] - for the quantitative extraction of 137 Cs in nitrobenzene at ph This method, then, was used to concentrate 137 Cs from a national mixture of fission products and cooling water from a nuclear reactor. The radioactive cesium ion-paired with [Cr(NH 2 C 6 H 5 ) 2 (NCS) 4 ] - in the organic phase was back-extracted into an aqueous phase by shaking the organic phase with 6 M HCl for its quantification. Several organic resins have been tested for removal of radioactive cesium at ORNL. Examples include resorcinolformaldehyde from Boulder Scientific, SuperLig 644C from IBC Advance Technologies and WWL Web with SuperLig 664C from 3M Corporation [88] Inorganic ion exchangers In environmental applications, inorganic exchangers have been mostly used to concentrate or extract 137 Cs. The inorganic compounds are popular because they possess 40

55 superior resistance to radiation and usually higher thermal stability as compared to the organic ion-exchange materials [88]. Important examples of inorganic exchangers, which can be used for selective extraction of 137 Cs, are complex cyanides, including insoluble polynuclear transition metal hexacyanoferrate(ii) and hexacyanoferrate(iii); and dodecaheteropolyacid salts of molybdenum and tungsten, such as ammonium 12- molybdophosphate and ammonium 12-tungstophosphate, commonly known as AMP and ATP, respectively. Other ion exchangers having high selectivity for cesium have also been reported. These include crystalline silicotitanate, zeolites, micas, hydroxyl oxides of pentavalent metals, and acid salts, such as silicates, phosphates, molybdates, tungstates of group III and IV metals. The use of hexacyanoferrates in cesium removal will be discussed separately in detail in part 2.5. a) Heteropolyacid salts Among a large number of insoluble salts of dodecaheteropolyacids, AMP, [(NH 4 ) 3 Mo 12 O 40.xH 2 O], is the most widely investigated ion exchange for selective concentration of cesium from aqueous solutions. Smit et al. [102, 103] were the first ones who discovered ion-exchange properties of AMP for cesium. Since then, this exchanger has been broadly used to pre-concentrate and determine cesium in environmental samples. AMP contains NH + 4 ions, which can be exchanged for Cs + and other heavy alkali metal cations such as Rb + and K +. However, AMP has such a high affinity to cesium that cesium ions can be adsorbed quantitatively from the solutions containing large concentrations of potassium and rubidium. For example, the presence of potassium ions in a saturated potassium solution could not stop AMP from the uptake of cesium ions [102]. In addition, AMP has a very large separation factor for Cs + /Na + ion pair. This value reaches 6000 or higher, while that of a conventional organic resin, Dowex-50 in ammonium form, is only 2.4. Therefore, AMP can be used for the preconcentration of Cs + ions where there is an excess of sodium. Morgan and Arkell and Smit et al. [104, 105] reported that the exchange of cesium on AMP was rapid in acidic media (ph 1 or lower) and was not significantly influenced by the high ionic strength of the solution. The distribution coefficient for cesium was still large (1500) even at or close to neutral ph values. The high selectivity 41

56 and the rapid sorption rate of cesium on AMP are the most important advantages of this exchanger over the organic ion-exchange resins for its use in the extraction and determinations of 137 Cs in environmental samples. Other advantages include its stability in the presence of nuclear radiation, ease of preparation, and availability at low cost [106, 107]. Salts of other dedecaheteropolyacids also have similar selectivity for cesium as AMP. For example, the salts of dodecatungstophosphoric acid and dodecatungstosilicic acid are most frequently used for the separation of cesium and other heavy alkali metals, such as rubidium and francium, from acid solutions [ ]. b) Zeolites Wingefors et al. [113] investigated the ability of mordenite to uptake cesium and other fission products from an acidic, radioactive waste liquid. Mordenite columns showed a complete removal of cesium. Other fission products were insignificantly adsorbed. In spite of the acidic nature of the feed solution, no signs of degradation of the zeolite structure were observed. It is shown that the rate of adsorption of cesium increased with the decrease in zeolite particle size. An uptake of close to 100% was seen for cesium and strontium after shaking the solutions with granular zeolite for 5 and 10 h, respectively. c) Silicotitanate The crystalline silicotitanate was first prepared by personnel from Sandia and Texas A&M University [87, 114]. Testing demonstrated that this material had a large affinity for cesium in the presence of high sodium concentrations. In 1996, this material, with a commercial name UOP IONSIV IE-911, was chosen for use in cesium removal in ORNL [87]. The material was in granular form with particle sizes from 30 to 60 mesh and was used in columns to remove cesium. Recently, IONSIV IE-911 was evaluated for the removal of cesium from INEEL acidic radioactive tank waste [115]. Batch contacts and column tests were performed. Overall, IONSIV IE-911 was effective for cesium sorption from highly acidic solutions including sodium and potassium; however, the stability of the material may be reduced in these solutions, resulting in a decrease in 42

57 cesium sorption. Crystalline silicotitanates were also reported to be used for cesium separation from contaminated milk in the Ukraine [116]. c) Hydrous oxides Amorphous hydrous titanium oxide (HTO) materials were first developed at Sandia in the 1960s to prepare electro-active ceramic materials for defense applications [87]. They were then used for high level nuclear waste processing in 1975 at ORNL. Recently, Mishra et al. [117] studied the removal behavior of hydrous titanium oxide and revealed that an increase in cesium concentration from 10-8 M to 10-2 M, temperature from 298 to 328 K, and ph from 2.5 to 10.2 apparently enhanced the uptake capacity of Cs + on hydrous Ti oxide. Another group of hydrous oxides, layered vanadium pentoxide gels, analogous to montmorillonite-layered silica, also has high selectivity for cesium ions in the presence of highly concentrated solutions containing Li +, Na +, K +, Mg 2+, Ca 2+, Ba 2+, and Sr 2+ [118]. d) Mica Miyake et al. [119] developed Na-substituted fluoromicas, which were swellable in water, for the separation of cesium. Na-substituted hectorite (NaH) was reported as Cs + ion sieve with no dependence on the concentration of Cs + ions. The cesium selectivity of Na-substituted taeniolite (NaT) and Na-substituted tetrasilicic mica (NaTS) was also determined and compared with that of NaH. It was suggested that NaH can be utilized in the separation and immobilization of Cs + ions from solutions containing Na +, K +, and Cs + ions. e) Acid salts of group (III) and (IV) metals Acid salts of zirconium showed high selectivity for heavy alkali metals in acid solutions. Zirconium phosphate [120, 121], zirconium molybdate [122], and zirconium tungstate [123] were used to isolate radioactive cesium and to separate cesium from rubidium. The separation can be performed on columns or on filter papers impregnated with these materials. Hydrated titanium dioxide in ammonium form, stannic phosphate and clay minerals also showed similar ion-exchange properties [124]. Betteridge and 43

58 Stradling [125, 126] developed a glassy chromium polyphosphate ion exchanger which exhibited specific selectivity for monovalent cations. The order of selectivity among monovalent ions was Cs + >Rb + >K + >Na + >H +. This ion exchange did not show affinity for multivalent ions. The uptake of potassium, rubidium and cesium ions at ph 2 and 3 was not affected by large molar excesses of Ni(II), Co(II), Mn(II), Cd(II), Zn(II), Ca(II), Sr(II), Mg(II), Fe(III), and Cr(III) ions. Ion-exchange reactions of alkali metals with chromium tripolyphosphate exchanger were done by batch and column methods. Komarneni and Roy [127] reported crystalline γ-zirconium phosphate as a highly selective cesium-ion sieve. The exchanger can be used to separate 137 Cs from circulating water in nuclear reactors. High selectivity of this exchanger for 137 Cs can also be utilized for pre-concentration of this radioisotope present at pg ml -1 in the presence of very large concentrations of sodium. 2.4 Polynuclear transition metal hexacyanometallates Polynuclear transition metal hexacyanometallates form an important class of insoluble mixed valence compounds. They have a general formula M A k [M B (CN) 6 ] l where M A and M B are transition metals with different formal oxidation numbers. These materials can contain ions of other metals, such as potassium and sodium, and varying amounts of water [128] Synthesis Generally, polynuclear transition metal hexacyanometallates can be synthesized by two methods: i) chemically by mixing a salt of metal A with a hexacyanide of metal B; it commonly produces an insoluble product in granular or very fine particles and ii) electrochemically by the consecutive potential cycling of the working electrode over a suitable potential range with a suitable initial potential in a solution of metal A salt and metal B hexacyanide; the product is a film of metal hexacyanometallate formed on the working electrode [ ]. In this dissertation, only the first method will be discussed. The composition, the structure and, hence, the properties of metal hexacyanometallates prepared by a chemical routine depends on the synthetic conditions, mainly on the 44

59 concentration and ratio of the starting reagent. Two important examples of this issue, iron hexacyanoferrates and cobalt hexacyanoferrates, will be presented in this section. a) Iron hexacyanoferrates Ferric ferrocyanide, or iron(iii) hexacyanoferrate(ii), well known as insoluble Prussian Blue (PB), was first prepared by Diesbach in Berlin in 1704 by mixing an excess of an iron(iii) salt with a potassium hexacyanoferrate(ii) solution [132]. The action of the excess of an iron(ii) salt on potassium hexacyanoferrate(iii) solution also produces an insoluble deep blue substance like PB; this material, which is formally known as Turnbull s Blue, puzzled chemists working in this area for a long time [128, 132]. Tracer studies showed that the two materials contained two types of iron and only small amounts of potassium; both gave the same X-ray powder pattern. It was proven that both were iron(iii) hexacyanoferrate(ii), Fe 4 [Fe(CN) 6 ] 3, in various hydrated forms. It was explained that an electron-transfer reaction between the Fe(II) and Fe(III) occurred to give rise to the same mixed-valence species wherein the Fe(II) occupied the M B position of the M A [M B (CN) 6 ].xh 2 O structure [132]. PB is also obtained by thermal decomposition of H 4 [Fe(CN) 6 ] or H 3 [Fe(CN) 6 ] at 160 o C [133]. Depending on the oxidizing conditions, hexacyanoferrate(ii) can be oxidized to PB or iron(iii) hexacyanoferrate(iii) in the presence of acid, but in neutral or alkaline solutions the reaction does not take place at an appreciable rate. Interestingly, PB produces iron(ii) hexacyanoferrate(iii), Fe 3 [Fe(CN) 6 ] 2, under the vacuum pyrolysis; it implies that the water present generates an internal pressure which favors the conversion. Re-conversion into iron(iii) hexacyanoferrate(ii) takes place slowly in moist air and rapidly in contact with diluted HCl [ ]. When solutions of FeCl 3 and K 4 [Fe(CN) 6 ], or of FeCl 2 and K 3 [Fe(CN) 6 ], are mixed at 1:1 molar ratio, soluble PB or the so-called soluble Turnbull s Blue, respectively, are obtained [132]. Both have the approximate composition KFe[Fe(CN) 6 ].H 2 O, though the water content is variable. The soluble forms of PB are in fact colloidal suspensions of KFe[Fe(CN) 6 ]. Heating the colloidal solution obtained from 3 moles of FeCl 3 and a mole of K 3 [Fe(CN) 6 ] at 90 o C in the dark, produces iron(iii) hexacyanoferrate(iii), which was named Berlin Green [132, 133]. Berlin Green also can 45

60 be obtained by oxidation of PB with nitric acid [133]. It should be noted that Berlin green has unit cell dimensions the same as those of PB, and it was, for many years, formulated Fe[Fe(CN) 6 ] with all the iron in the tripositive state. However, this formulation is incorrect. The intense color is present only after at least a low concentration of Fe(II) species is produced in preparations starting from Fe(III) species or as long as oxidation of Fe(II) species is incomplete [133]. When completely oxidized, the material is formally Prussian Yellow. b) Cobalt hexacyanoferrates Cobalt hexacyanoferrates are formed by the reaction of a soluble cobalt salt and a soluble ferrocyanide (K, Na, or H compound). All the precipitation products can be represented by a general formula: A 2n Co 2-n [Fe(CN) 6 ].xh 2 O; where A is an alkaline metal (commonly Na + and K + ), H +, or NH + 4 ; n = 0, 1, or 2 for simple compositions of definite formulae [136]. Values of n commonly vary from 0 to 1 and less commonly, from 1 to 2. Many of the precipitation products show an average composition with non-integral values of n. For example, when 0.43 M K 3 [Fe III (CN) 6 ] 3- is co-precipitated with an aqueous solution of CoCl 2, cobalt(ii) hexacyanoferrate(iii) is formed with a formula K 0.2 Co 1.4 [Fe(CN) 6 ].7H 2 O [137]. When the cobalt nitrate solution was slowly added to a K 4 [Fe(CN) 6 ] solution, pure K 2 Co[Fe(CN) 6 ] was obtained. The reverse order of mixing, however, produced a mixture of Co 2 [Fe(CN) 6 ] and K 2 Co[Fe(CN) 6 ] [138] Structures Attempts to establish chemical formulae, oxidation states of the metals, and solid state structures of metal hexacyanoferrates began in the 1930s by Keggin and Miles [139]. They proposed a face-centered cubic structure for PB type compounds with a ~ 10.2 Å; the positions of the iron atoms are the same in all of them, in each case being linked by cyano bridging. In K 2 Fe II [Fe II (CN) 6 ] every small cube, and in KFe III [Fe II (CN) 6 ] every other small cube, contains a K + ion at its center. However, the structures for these two compounds were not supported by comparison of calculated and observed densities or X-ray intensities [132]; thus, they are not regarded as firmly established. But Cs 2 Mg[Fe II (CN) 6 ] and Cs 2 Li[Fe III (CN) 6 ] certainly have the structure attributed to 46

61 K 2 Fe II [Fe II (CN) 6 ] [132]. According to Lehto [140], K 2 Co[Fe(CN) 6 ] has a cubic structure. Iron and cobalt ions are located at the corners of the elementary cubes, cyano groups on the edges, and the exchangeable potassium ions in the body center. In 1962, Robin [133] used ligand field theory based on an electronic study to explain the charge-transfer transition that induces the intense blue color of PB. In the early 1970s, Ludi and Güdel [ ] proposed a structure for Prussian Blue and its analogues M A k [M B (CN) 6 ] l.xh 2 O, as shown in figure 11. In this face-centered cubic structure, there are some vacancies which occur randomly, thus leaving a highly symmetrical and empty structure in which zeolitic water occupies the centers of the octants of the unit cell. For example, in the case of Mn 3 [Co(CN) 6 ] 2.12H 2 O, one-third of the Co, C and N positions are unoccupied. Insoluble Prussian Blue also has a low density structure with vacant lattice sites, and zeolitic water again occupies the centers of the octants of the unit cell [142, 143]. This proposed structure was supported by singlecrystal X-ray and neutron diffraction studies. The number of sites occupied by M B, C and N and the number of coordinated water molecules in the unit cells vary with values of k and l. It should be borne in mind that there are some exceptions. For example, Mn 2 [Ru(CN) 6 ].8H 2 O contains two water molecules, which act as bridging ligands; the unit cell is monoclinic, with a = 9.48 Å [144]. Many compounds of formula K 2 M II [Fe(CN) 6 ] have a face-center cubic lattice with a ~ 10.2 Å and variable water content. Among these compounds are those for which M II = Co, Ni, Zn or Cd. However, for M II = Cu, the Jahn-Teller Effect results in a distortion and the unit cell is tetragonal. The cobalt, nickel, and copper salts of formula M II 2 [Fe(CN) 6 ] also have face-centered cubic unit cells with a = 10.1 Å, but the zinc salt has an orthorhombic unit cell with a = 11.52, b = 13.17, c = 9.90 Å [133] Properties PB and many PB analogues exhibit magnetic behavior. The first investigation of the magnetic behavior was reported in 1928 by Davison and Welo, who described the magnetic susceptibility of the compound at three temperatures between 200 and 300 o C [133]. There are two basic environments for the metals centers, M A and M B, in the 47

62 N C N N C C C C N N C C C N N N N C N C M A M B H 2 O Fig. 11 The unit cell of Prussian Blue and analogues M A k [M B (CN) 6 ] l.xh 2 O (base on ref. 132) 48

63 framework molecule with the formula M A k [M B (CN) 6 ] l. The M B metal ions are coordinated in a strong cyanide ligand field, and thus are low spin, whereas the M A metal is coordinated in a weak nitrile ligand field [128, 133]. The weak coupling observed in PB resulted from the interaction of high-spin ferric centers through intervening diamagnetic ferrous centers. The ordering of the Fe(III) ions, in spite of the long distances that separate them in the structure, is an indication that the intervening Fe(II) sites participate in the magnetic interaction. As an extension of the PB magnetic studies, a series of rare earth ferricyanides, chromicyanides, and cobalticyanides were crystallized and subjected to X-ray and magnetic susceptibility measurement in the range K. To pursue high temperature magnetic materials, one or both of the Fe sites are replaced with a different metal ion with unpaired electrons [133]. The ability of hexacyanometallates to uptake metal ions and neutral molecules is an interesting area because of their potential applications in separation science. Many studies show that ferrocyanides and ferricyanides have a strong affinity to heavy alkali metals and have a different selectivity to each one. These features can be applied to the pre-concentration or separation of heavy metals in solutions. Compared to ferrocyanides ([Fe(CN) 6 ] 4- ), ferricyanides ([Fe(CN) 6 ] 3- ) are usually more soluble [136]. This should be taken into account when using the hexacyanoferrates to remove heavy metals from solutions. In addition, ferricyanides tend to have lower capacities since there is only one easily exchangeable cation instead of the two that occur in ferrocyanides [145]. In the early 1940s, it was noted that PB reversibly absorbed atmospheric water or ammonia in a manner similar to zeolites [133]. In addition, PB can be dehydrated without destruction of the 3D architecture by heating or storage over a dehydrating agent [133]. In the 1950s and 1960s, Seifer [146] investigated the sorption of neutral molecules, such as ethanol, acetone, toluene, and water, by a series of divalent ferrocyanides, and found that the uptake was the most rapid for acetone followed by ethanol, toluene, and water. Seifer [147] also concluded that the cyanoferrate solids contained 67% surface sites for adsorption and 18.5% internal cavities that are related to zeolitic sorption and ionexchange properties. This explained why some organic molecules, with a size larger than the internal cavity diameters of the cyanoferrates, are still able to be incorporated in the materials. In addition, the solids were observed to change color during the process of 49

64 adsorption in the presence of certain vapors such as ethanol, which suggested a chemical interaction and not zeolitic behavior. Later, Cartraud et al. [148] studied the zeolitic properties of K 2 Zn 3 [Fe(CN) 6 ] 2.xH 2 O in the presence of the gases CO 2, CO, and N 2, as well as water vapor and small hydrocarbons. They concluded that this ferrocyanide compound, with a structure consisting of ellipsoidal cavities, exhibits true zeolite properties. In the unactivated state, these cavities are occupied by water that is lost upon dehydration without destruction of the host cyanide structure. The activated material behaves as a zeolite with molecules that are smaller than the pore diameters and performs a molecular sieving effect toward molecules too large to diffuse into the cavities. When studying the neutral molecules taken up on cyanide compounds, Boxhoorn et al. [149] discovered that those compounds, with the typical structure of PB, for example, Zn 3 [Co(CN) 6 ] 2, have unoccupied sites in unit cells, producing cavities with the size of 5.6 x 8.6 Å after dehydration. These cavities dictate the uptake selectivity of the compound. In the case of a compound which does not possess vacancies in the structure, e.g. Zn 3 [Co(CN) 5 NO] 2, only very small molecules are able to enter the openings in the cubic structure. A more extensive study revealed that unlike the classical aluminosilicates, hexacyanometallates are able to be tailored with respect to the geometries and size of the vacancies. For example, simple substitution, such as replacement of CN - with NO or OH -, leads to new crystal symmetry, and, therefore, different pore shapes and sizes [133] Applications The earliest application of transition metal hexacyanometallates should be given to PB which is among the oldest recognized members of a class of transition metal compounds referred to as mixed-valence complexes. The extraordinary intensity and longevity of PB color led to heavy industrial use of PB as a pigment for paints, printing inks, laundry dye, and other color-related uses [128]. The presence of the different oxidation-state ions gives rise to a number of phenomena that can be studied by various electrochemical and spectral techniques. The first report in 1978 by Neff [150], of the electrochemical behavior of PB thin films, involved the deposition of the compound on Pt foil. Since then, researchers have carried out numerous studies on modified electrodes 50

65 comprised of PB and analogous films on conducting and semi-conducting substrates for various applications, such as electrochromism, electroanalysis, ion selectivity and the ability to store cations, solid state batteries, and molecular magnetism [128, 132]. Many insoluble hexacyanoferrates(ii) have been reported as effective ionexchange materials. The fixation of alkali metal ions follows the sequence Cs + >Rb + >K + >NH + 4 >Na + >Li + [133]. One of the main applications of the ion exchange properties of PB and related materials is for the environmental clean-up of 137 Cs formed in nuclear fission reactions, and for medical uses, which will be presented in detail in part Extraction of cesium by transition metal hexacyanoferrates Overview of the method There are three methods commonly used in applications of insoluble metal hexacyanoferrates to remove cesium from waste solutions [136]. The first one is to mix the ferrocyanide solution into the waste solution for an in-situ preparation of the solids. The Co, Ni, Zn, or other transition metal components may be components of the waste, or they may be added as salt solutions. This procedure is not very suitable for caustic wastes because most transition metals are not soluble at high phs and would precipitate as hydroxides. A second procedure is to premix the ferrocyanide and metal salt solutions to precipitate a slurry and then add the slurry to the waste solution. The third procedure is to precipitate the slurry and then separate, wash, dry, size classify, or otherwise pre-treat the solids, such as impregnation on support materials (silica gel, resins, membranes, filters, etc.), before supplying them to the waste. There is a compromise among these three methods. The first two methods are simple and low cost. However, they have some disadvantages. The precipitate of hexametallates is usually in a form of very fine particles [151], which may be a problem for the separation of these solids from the solution and be difficult for column performance. In the case of using hexacynoferrates, if the solids accumulate in the waste storage, they will produce periodic discharges of H 2 and nitrogen oxide gases [136]. These discharges are a severe hazard. The last procedure involves solid treatments, which may reduce the uptake rate and capacity of hexacyanometallates, 51

66 compared to the freshly prepared slurries [136]. However, it can be modify to improve the physical properties of the solids, such as, increasing the particle size and encapsulation of the solids in porous materials. In both cases, the final products are suitable for columns and batches. Particularly, after using the hexacyanoferrateencapsulated porous materials for cesium removal, the solids can be sintered into glasses, thereby reducing the waste volume. In the next paragraph, some examples of preparing hexacyanoferrate-doped porous materials are given. Several chemically un-reactive solids have been used as a support for ferrocyanide precipitates. Konecny and Caletka [152] encapsulated a metal hexacyanoferrate in silica gel. First, the silica gel and K 4 Fe(CN) 6 were mixed followed by drying. This process was repeated. Then, the resulting solid was reacted with the appropriate metal salt solution. The loaded metal ferrocyanide was reported as 0.10 to 0.31 g/g of silica gel for single precipitation and 0.28 to 0.45 g/g silica gel for repeated precipitations. Another method was introduced by Valenity et al. [153]. Here, silica gel was added to a CuSO 4 solution followed by adding Na 4 Fe(CN) 6. The resulting gel contained copper hexacyanoferrate(ii), Cu 2 Fe(CN) 6. The copper hexacyanoferrate(iii) gel was obtained by oxidizing the samples with 1 M HNO 3. Metal hexacyanoferrates are also encapsulated in ion-exchange resins [154], which have good properties for operations of the exchange columns. For example, an ion resin can be converted to the ferrocyanide form by a reaction with a K 4 Fe(CN) 6 solution. This ferrocyanide resin then can be reacted with metal salt solutions to precipitate the metal hexacyanoferrate in the resin pores. Up to ten repetitions of this procedure are applied to prepare the product Evaluations of the method The performance of a solid for cesium removal is usually evaluated by three measurements [136]: 1) distribution coefficient, K, to measure the equilibrium cesium concentrations with an excess of solids; 2) capacity to measure of how much cesium is removed by a unit amount of solids in an excess of cesium in the solution; and 3) rate to measure how rapidly the cesium reacts with the solid. All of these depend on other variables such as the concentration of cesium in solution, the concentration of other 52

67 cations, the ph, and the temperature. More details are given about the capacity parameter in the next paragraph, because it is commonly used to report results. The capacity of a cesium absorbent is the amount of cesium retained after the extraction is completed. In hexacyanoferrates with a general formula A 2n M 2-n Fe(CN) 6, the estimated capacity is 2n for n 1 [136]. It assumes that the alkali metals, H + or NH + 4, are exchangeable and the multivalent cations are not. This capacity is the upper limit for the ion exchange and is rarely observed in practice. For n = 0 or all multivalent cations, the uptake of cesium is usually governed by a sorption rather than of an ion exchange; these capacities are unpredictable. The capacity is also reported as grams of cesium uptake per gram of the hexacyanoferrate or of the composite. The capacity results have a wide variability with compositions and preparation conditions. For example, zinc ferrocyanide, with very fine particles, had a capacity for cesium of 0.91 to 2.1 mol Cs/mol Fe [155]; with granular and with the separation performed on columns, the maximum values were 0.9 to 1.0 mol Cs/mol ferrocyanide [156]. The hexacyanoferrates(ii) supported on silica gel prepared by Konecny [151] showed a cesium uptake capacity of 0.25 to 1.3 mol Cs/mol hexacyanoferrate Mechanism of cesium sorption by hexacyanoferrates The sorption of cesium from solutions by hexacyanoferrates goes through two main steps [138]. First, cesium is transferred from the bulk solution to the solid surface; this transfer is commonly controlled by diffusion through a liquid film on the solid surface. Second, cesium is adsorbed within the hexacyanoferrate solids. The mechanism of this step is more complicated and dependent on both the composition and the physical properties of the solids. It varies from a fast ion exchange to other chemical reactions, which are much slower and cause changes in the structure. Intermediate rates and mixed mechanisms are often observed. A true ion exchange mechanism is usually observed in hexacyanoferrates containing K +, Na +, H +, or NH + 4. The sorption is rapid by the ion exchange of Cs + for K +, Na +, H +, or NH + 4 ; these exchanged cations appear in stoichiometric amounts in the solution. In contrast, hexacyanoferrates, with the simple formula M l [Fe(CN) 6 ] k, show a complex mechanism with release of metals that are smaller than the amount of Cs + 53

68 sorbed. A study of the mechanism of granular solids with the average formula K 2 Cu 3 [Fe(CN) 6 ] 2.4H 2 O by Lee and Streat [157] shows that K + was released with a molar amount a little larger than the Cs + sorbed and a very small amount of Cu 2+ was released to the solution. In another study, Lee and Streat [158] observed that Cu 2 Fe(CN) 6 sorbed Cs + without releasing Cu 2+ to the solution. Lehto et al. [140, 159] proposed a mechanism of the cesium uptake of K 2 [CoFe(CN) 6 ]. The ionic radius of non-hydrated cesium is 1.69 Å, which is considerably larger than the effective window radius of the K 2 [CoFe(CN) 6 ], 1.47 Å. Therefore, cesium ions cannot penetrate through the crystal. In addition, it was observed that the excess of potassium released from the hexacyanoferrate was related to the amount of cesium ions sorbed. Thus, it is assumed that the uptake of cesium ions by K 2 [CoFe(CN) 6 ] is most probably a real ion exchange process. However, not all the potassium ions are exchanged for cesium but rather only those in the surface layer of the crystal. Vlasselaer et al. [160] and Loos-Neskovic [155] found a 1-to-1 Cs + /K + exchange for K 2 Zn 3 [Fe(CN) 6 ] 2 without any release of zinc. The equilibrium was approached in 10 minutes. For Zn 2 Fe(CN) 6, the zinc released was much smaller than the Cs + sorbed, and the crystal lattice changed from a trigonal structure to a cubic structure. It took several hours to reach the equilibrium. It should be noted that the rate of reaction also depends on the physical properties of the hexacyanoferrate solid. For example, for hexacyanoferrates having larger particles, it will take a longer time to approach equilibrium than for the smaller sized particles, because of the longer diffusional distance. Loos-Neskovic and co-workers [161] investigated sorption mechanisms of cesium on Cu II 2 Fe II (CN) 6 and Cu III 3 [Fe III (CN) 6 ] 2. Both compounds have cubic structures with iron vacancies that can be occupied by water or salt molecules. The former had a higher cesium uptake than the latter. It was believed that this was due to the different type of sites occupied by copper in each compound. In Cu II 2 Fe II (CN) 6, copper occupied two different sites: Cu1 in the position linked to Fe II through CN groups, and Cu2 not linked to the CN groups and partially occupying the interstitial positions. The second type of site was not present in Cu III 3 [Fe III (CN) 6 ] 2. The presence of the Cu2 sites seemed to play a favorable role in the sorption. It demonstrated that the sorption of cesium was neither an ion exchange nor a dissolution-precipitation. The sorption was the incorporation of ion 54

69 pair, Cs + and NO - 3, which are hosted in the vacancies of the structure, replacing the water molecules. This process was followed by a reorganization of the solid, resulting in one or more new solid phases Some characteristics important to practical operations In addition to the preparation of suitable particles for cesium removal, some other characteristics of the hexacyanoferrates also should be taken into account when applied to practical operations. The cesium-removal effectiveness of the material may be lost during use. Such losses that are caused by the chemical environment are reviewed in the following section. Losses by simple dissolution are determined by the solubility (see section c). Some other important causes of lack of physical stability are radiation damage, high temperatures, and mechanical erosion from flow or mixing effects (section b). a) Chemical stability The efficient removal of cesium from solutions requires a limited chemical decomposition of the hexacyanoferrate solids during contact with the solution. The waste solution may have low ph or high salt concentrations or both. For these reason, the chemical stability of hexacyanoferrates in different solutions has been studied. However, the results are difficult to evaluate in general quantitative terms because they vary depending on the composition and preparation of the hexacyanoferrate. Loos-Neskovic [162] investigated the chemical stability of zinc and nickel hexacyanoferrates. Zinc hexacyanoferrates could be used from ph 1 to 10 and nickel hexacyanoferrates, from ph 0 to 12 in both aqueous solutions and organic solvents such as acetone. The ferrocyanide solids are chemically decomposed at either high caustic or high acid concentrations and are oxidized to the ferricyanides by nitric or other oxidizing agents. The followings are some example reactions [136]: Cs 2 CoFe(CN) 6 + 3H 2 SO 4 Cs 2 SO 4 + CoSO 4 + FeSO 4 + 6HCN (11) Cs 2 CoFe(CN) 6 + 6NaOH 2CsOH + Co(OH) 2 + Fe(OH) 2 + 6NaCN (12) 4-3Fe(CN) 6 + 4H NO 3 3Fe(CN) 6 + NO + 2H 2 O (13) 55

70 All of the hexacyanoferrate solids are chemically stable over a range from moderately acid solution to about ph 11. The caustic decomposition (Eq. 12) occurs above ph 11; the rate depends on the solid preparation procedure. While fine powders lose effectiveness very rapidly, the granular potassium hexacyanocobalt(ii) ferrate(ii) synthesized by Prout s method [163, 164] is stable for periods ranging from hours to several days, up to ph 13. Campbell et al. [165] found that potassium copper hexacyanoferrate lost its effectiveness more rapidly than potassium cobalt hexacyanoferrate as the ph increased. Campbell also found that the hexacyanoferrates retained their effectiveness down to ph 2, even in solutions of high salt concentrations; but a nickel compound was better than the cobalt compound. Valentini et al. [166] observed much different adsorption behavior for cesium on Zn 2 Fe(CN) 6 from NH 4 NO 3 solutions as compared to HNO 3 solutions. The behavior in NH 4 NO 3 can be explained by an ion-exchange mechanism; while that in the HNO 3 solution was affected by oxidization to ferricyanide, releasing zinc, and by decomposition of the ferro- and/or ferricyanide compounds. In addition, the sorbed cesium may be desorbed in nitric acid solutions as a consequence of the oxidization of ferrocyanides [89]. b) Physical stability Generally, nuclear fuels are allowed to decay before reprocessing, so the wastes have commonly decayed for long periods of time before treatment to remove cesium. Therefore, the radiation damage to hexacyanoferrate solids by 137 Cs is small for normal processing times. The radiation damage by other radioisotopes is generally insignificant [136]. Phillips et al. [167] reported that copper hexacyanoferrate showed no effects from radiation exposures up to 10 7 MGy. Watari et al. [168] showed no change in the cesium removal capacity after a R (from 60 Co) dose to iron hexacyanoferrate precipitated in an anion-exchange resin. Barton et al. [169] reported that cesium zinc hexacyanoferrate slurries of high radioactive cesium content did not show radiation decomposition for a week. During heating, weight losses of the metal hexacyanoferrate were observed. It is attributed to dehydration followed by losses of CN or oxygen depending on the atmosphere [136]. Valentini et al. [170] reported that Cu 2 Fe(CN) 6.12N 2 O in nitrogen lost 56

71 11H 2 O by 145 o C and decomposed to ferrous and copper cyanides at 425 o C. In the air, at 185 o C, copper oxide, (CN) 2, and ferrihexacyanoferrate was produced from the oxidization. Flow or mixing processes may break the hexacyanoferrate into very fine particles, which is a problem for the treatment process. Encapsulating hexacyanoferrates in porous materials may solve this problem. c) Solubility Most of the applications of the hexacyanoferrate solids are used to treat diluted solutions of cesium, using large solution/solid volume ratios, commonly 10 3 to 10 4 L/kg [136]. Therefore, it requires hexacyanoferrate solids to have a very low solubility to minimize losses. Valentini [166] reported that zinc hexacyanoferrate had a lower solubility than that of the cobalt and nickel compounds. Narbutt et al. [171] demonstrated that titanium ferrocyanide with high Ti/Fc ratios had a very low solubility. Additionally, as mentioned before, metal hexacyanoferrates(ii) are less soluble than metal hexacyanoferrates(iii) Environmental and medical applications a) Medical applications The use of PB compounds to treat animals and humans that have been exposed to 137 Cs began with studies in rats and dogs and was first tested on humans in 1966 [132]. The nuclear disaster at Chernobyl in 1986 led to the widespread contamination of many regions of Europe with 137 Cs and 134 Cs, affecting both humans and livestock. PB feed additives, to prevent absorption of radioactive cesium in contaminated animals, have been administered to pigs, sheep, cows, goats, and deer in many countries in Europe, for example, Russia, Norway, Austria, France, Ukraine, and Belarus [172]. A major concern of treatments involving PB is the issue of toxicity. Studies of the metabolism of ammonium ferric cyanoferrate, NH 4 Fe[Fe(CN) 6 ], in dairy cows revealed that between 90-95% of the compound was excreted. The compound thus, is a safe food additive to prevent the transfer of dietary radiocesium to milk. Daily administration of hexacyanoferrate powder, which is a mixture of 5% KFe[Fe(CN) 6 ] and 95% of 57

72 Fe 4 [Fe(CN)] 6, at a rate of 3-5 g per cow reduced 137 Cs transfer by up to 90% in milk. It was explained that when hexacyanoferrates were supplied, the bound radiocesium was eliminated from the body in feces and only a limited and dose-dependent portion of ingested radiocesium was absorbed. In addition, hexacyanoferrates can bind endogenously excreted radiocesium in the gut, thereby accelerating the excretion of the radiocesium from the body. Similar results were reported for sheep studies. The effect of a natural clinoptilotite modified with iron hexacyanoferrate, named RADEKONT, on 137 Cs uptake into meat was tested in experiments with broiler chickens in the Czech Republic [173]. It showed that RADEKONT as a supplement during the decontamination period decreased the biological half-life of 137 Cs to less than 1 day and the uptake of 137 Cs was faster in leg meat than in breast meat. Human experiments in 1966 involved the ingestion of 3 grams of PB per day for 20 days by volunteers. The studies showed that the compound enhanced excretion of radioactive cesium and humans suffered no side-effects after large doses of PB. After this study, PB was used to treat those who were contaminated with 137 Cs. b) Environmental applications This section will give some examples of utilizing metal hexacyanoferrates to remove radioactive cesium from liquid waste. Pioneering work in the 1960s established the utility of PB compounds for binding radioactive cesium in large-scale waste treatment in Hanford [133]. In 1991, at the Loviia Nuclear Power Station (NPS) in Finland, hexacyanoferrates, packed in stainless steel columns, were tested for the removal of cesium from the waste liquid. The test run was conducted by Lehto and co-workers [174]. The ph of the waste solution was 13.7, so it was adjusted to 11.5 before the treatment with 60% nitric acid. In the test run, 253,000 liters of concentrate was purified. A decontamination factor of ~2000 was achieved. The original solution consisted of 3x10 5 Bq L -1 ; in the purified solution there was on average < 150 Bq L -1. The volume reduction factor was reported as The used columns would be solidified in a concrete container, which then would be stored in a rock vault. The savings that were calculated for the test run were totally over 2 million USD. 58

73 Oji et al. [175] reported the use of a 3M Empore membrane filter, loaded with potassium cobalt hexacyanoferrate, for the uptake of radioactive cesium from the R- Disassembly Basin at the Savannah River Site. The membrane was configured as a cartridge and inserted into a commercial filter housing. An effective pre-filtration system was used upstream of the Empore cartridges to remove the particulates and to prevent the clogging of the Empore cartridges during lengthy unattended operation periods. R- Disassembly Basin water was flowed at a linear processing rate of 1200 liters/minute/m 2. During the seven-day demonstration, over 210,000 liters of the basin water were processed through the system without cesium breakthrough. The estimated minimum cesium-137 decontamination factor was 1604, which is 58 times better than the DOE requirement. 59

74 Chapter 3 Solid Phase Extraction of Cesium from Aqueous Solution Using Sol-Gel Encapsulated Cobalt Hexacyanoferrate 3.1 Introduction As discussed in Chapter 2, the selective preconcentration of cesium from aqueous solutions containing high concentrations of alkali metals is an important problem in the treatment of radioactive wastes. Among the materials being studied are precipitates of hexacyanoferrate, HCF, with divalent cations of metals such as copper, cobalt, nickel and zinc. These precipitates are formed as bulk solids or as films on a variety of supports [136]. Hexacyanoferrates such as that of cobalt (CoHCF), which in some cases are charge-balance with other cations such as K + and H +, have perhaps been the compounds most used. Lehto et al. [140] investigated the distribution coefficients for alkali and alkali earth cations on K 2 Co II Fe II (CN) 6.4H 2 O; a high selectivity toward Cs + was reported. Ramaswamy [176] compared the distribution of Cs + on HCFs of Cu II, Co II, Ni II, and Zn II. The metal HCFs were in the form of composites on anion exchange resins. The efficacy of these metal HCFs for the extraction of Cs + was in the sequence Co>Ni>Cu>>Zn. The composite materials were reported as advantageous relative to the unsupported solids in terms of total capacity [176]. In acid solution, a low concentration of hydrazine was added to obtain high distribution coefficients. An analogous study by Ramaswamy but with silica as the support reported that the exchange capacity for Cs + from neutral solution was greatest with CuHCF [177]. The above [136, 140, 176, 177] and other studies as presented in Chapter 2 support the hypothesis that metal hexacyanoferrates are potentially useful as reagents for the solid phase extraction of Cs + from aqueous solution. For applications it is necessary to consider effective means of supporting these compounds. As stated above, Ramanswamy has investigated making composites on anion exchange resins [176] and on conventional silica [177]. Applications using metal HCFs in the form of a powder have also been studied. Applications of the powders include supporting them on a membrane filter and using them as a packing material on a column [175]. Simple 60

75 precipitation of hexacyanoferrate with a salt of the selected transition metal commonly gives very fine particles that are very difficult to separate from solution and are not suitable for making columns. Special preparation procedures have been developed to prepare granular solids suitable for column operation. Silica was used as the supporting material, on which K 4 Fe(CN) 6 was deposited and then reacted with Co(NO 3 ) 2 to form CoHCF [178]. Another granular form was a composite prepared by CoHCF precipitation in a commercial silica sol matrix [151]; the resulting hydrogel was dried at 115 o C to give angular composite particles. The mechanism of the uptake of Cs + by metal hexacyanoferrates has been reported as either ion-exchange or sorption into the cage structure of these compounds. Lehto et al. [141] and Ramaswamy [176] concluded that the mechanism of cesium sorption by copper, cobalt, nickel and zinc hexacyanoferrates was true ion-exchange. In contrast, Ayrault et al. [161] found that true ion-exchange played, at most, a very minor role. The mechanism of cesium sorption by this compound was attributed to the incorporation of ion pairs of Cs + and a coordinating anion such as NO - 3 into vacant sites of the lattice structure, replacing the water molecules. One goal of our research was to further elucidate the uptake mechanism. In the present study, we developed a method for preparing a composite of CoHCF and a silica sol-gel using room temperature processing. The mechanism of uptake of Cs + by CoHCF and the influence quantity of CoHCF doped in the sol-gel on the uptake capacity were investigated. Moreover, the pore size of the support, namely silica prepared by a sol-gel processing at room temperature, for the CoHCF was a variable. Specifically, the pore size distribution was varied by control of ph during the hydrolysis and condensation of the sol [16] and by templating, a method that is described in a cited review [17]. Regarding the former method, near the point-of-zero-charge of silica, ph ~2, the rate of condensation is slow relative to that of hydrolysis, which results in a microporous sol-gel. At ph<1, these relative rates are reversed; subsequently, mesoporous sol-gels are produced. When mesoporosity was imparted by templating, generation-4 polyamidoamine dendrimer (G4-PAMAM) was included in the sol under 61

76 acidic conditions. The protonated amine sites cause the negatively charged silica to form around this macromolecule, thereby resulting in mesopores [14]. One goal of this study was to investigate methods for the synthesis of high concentrations of CoHCF in silica, which, in turn, was expected to result in a solid with a high capacity for the uptake of Cs +. The role of pore structure on this capacity was studied. Of concern was whether micropores allow facile diffusion of Co 2+ during the CoHCF synthesis and of Cs + during the extraction experiments. To meet these goals, we investigated dispersion of the sol by sonication as an alternative to dissolution in alcohol. Although the use of G4-PAMAM was in part related to the pore size study, our interests in this dendrimer also were based on its potential ability to increase the CoHCF level in the silica by sequentially sorbing anionic HCF and the Co Experimental Reagents Tetramethylorthosilicate (TMOS, 99% purity), generation-4 poly(amidoamine) dendrimer (G4-PAMAM), 10% (wt) in methyl alcohol, and cesium nitrate (99% purity) were purchased from Aldrich Chemical Company, Inc. (Milwaukee, WI). The Co(NO 3 ) 2 6H 2 O and K 4 Fe(CN) 6 3H 2 O were purchased from Fisher Scientific (Fair Lawn, NJ). Unless otherwise noted, all other chemicals were ACS Reagent Grade and used without further purification. Solutions were prepared with house-distilled water that was further purified with a Barnstead NANOPure II system (Boston, MA) Preparation of the sol-gel composites For experiments investigating the influence of ph on the pore size, silica sol-gels were prepared from precursor solutions comprising 2 ml each of tetramethyl orthosilicate (TMOS), methanol, 0.04 M K 4 Fe(CN) 6 and HCl. The HCl concentrations were as follows: 1.2, 0.12, 0.04, and M. When the objective was to use this general processing condition but with a higher concentration of K 4 Fe(CN) 6, the concentration of the added HCl was increased to1.2 M, which permitted addition of 2 ml of 0.3 M K 4 Fe(CN) 6 without precipitate formation. 62

77 Two other processing conditions were used in experiments designed to achieve higher loadings of K 4 Fe(CN) 6. One eliminated the methanol from the precursor solution; instead, the sol was sonicated to obtain a homogeneous mixture of TMOS and water. In this case the precursor was 2 ml each of TMOS, 0.5 M K 4 Fe(CN) 6, and 1.0 M HCl. The second approach was to include G4-PAMAM in the sol. Specifically, this composite was prepared from a sol containing 2 ml each of TMOS, 1.0 mm G4-PAMAM in methanol, 0.2 M HCl, and distilled water. In all of the above cases, the processing was initiated by either stirring or sonicating the sol for 30 minutes, during which the precursor solution was covered with Parafilm. After the mixing step, 2-mL aliquots of sol were pipetted into hexagonal polystyrene weighing dishes (1.5" x 1"), which were then sealed with Parafilm to control the vapor pressure in the container, thereby allowing the sol to age before gelation. After one day, several pinholes were made in the film so that the solvent and the alcohol that is released during gelation can evaporate gradually. The Parafilm was removed after 2 days. The gels were dried under the ambient conditions for another 7 days. After the 10-day period, the K 4 Fe(CN) 6 -doped silica monoliths were ground into powders. For sol-gels made with 0.04 M K 4 Fe(CN) 6 as the dopant, 1.0 g of the powder was stirred in 10 ml of 0.1 M Co(NO 3 ) 2 in acetone for 27 hours. With silica formed using the higher K 4 Fe(CN) 6 concentrations, the concentration of Co(NO 3 ) 2 was adjusted to yield a mole ratio of Co:HCF of 3:1, thereby assuring an excess of Co 2+ in the 10-mL bathing solution, which is in excess of the maximum Co:HCF ratio in the product (2:1). The CoHCF-doped gel powder was then filtered and washed with acetone five times (15 ml acetone/each time). Monoliths prepared with G4-PAMAM as the dopant did not contain K 4 Fe(CN) 6 initially. Here, 1.0 g of sol-gel was ground and stirred in 150 ml of 0.1 M K 4 Fe(CN) 6 for 24 hours, during which ion-exchange between the protonated amines and the HCF anions occurred. The powder was filtered and allowed to be dried; then, it was transferred to 150 ml of 0.1 M Co(NO 3 ) 2 in acetone. Acetone was used rather than water to prevent leaching of HCF during this step. After 27 hours, the CoCHF-doped silica was recovered by filtration, rinsed with acetone, and dried. In all cases, the stirring process of the ground sol-gel in a desired solution was in a sealed propylene bottle. 63

78 3.2.3 Analytical procedures Unless otherwise stated, the uptake capacity measurements were performed by stirring a 10-mg sample of powdered, CoHCF-doped sol-gel in 20 ml of 0.5 mm CsNO 3 in a propylene bottle for 24 h at 25 o C. The sample was filtered, and the Cs + in the residual solution was determined by ion chromatography with a DX-120 system (Dionex, Sunnyvale, CA). Selectivity of the CoHCF-doped composites toward Cs + was determined by the same experiment except that the samples were 0.5 mm CsNO 3, including either NaNO 3 at concentrations of 10-4, 10-3, 10-2 and 10-1 M or Ca(NO 3 ) 2 at concentrations of 10-2 and 10-1 M. All experiments were performed five times. The pore size distribution and the specific surface area of certain composites were determined by the nitrogen adsorption (BET) method. An SA 3100 Surface Area Analyzer was used (Beckman Coulter, Miami, FL). 3.3 Results and discussion The goal of this work was to investigate the solid phase extraction (SPE) of Cs + from water using CoHCF-doped sol-gels. The study involved elucidation of, first, the role of pore size on this SPE and, second, the development of methods to increase the level of CoHCF doping in the sol-gel. The initial experiments on the influence of pore size on the SPE of Cs + with CoHCF-doped silica were performed on sol-gels processed from HCFcontaining precursors of various ph; subsequently, the CoHCF dopant was synthesized by immersing powders of this sol-gel in solutions of Co(NO 3 ) 2 in acetone. The use of acetone rather than water was used to minimize leaching of HCF from the silica. When acetone was used, there was no visual evidence of CoHCF formation in the solution, and the color of the doped sol-gel suggested that Co 2+ permeated the solid matrix. With the analogous experiment with Co(NO 3 ) 2 in water, formation of CoHCF primarily on the surface of the silica and in the surrounding solution was observed. In these experiments, ph variation was used to obtain silica with a range of pore size. The data in Fig. 12 demonstrate that the HCF-doped sol-gels formed at low ph had a greater distribution of pore sizes in the larger of the two domains that are plotted, which is consistent with the reported formation of mesoporous (albeit undoped) silica under 64

79 Gel pore vol. % Sol ph Pore dia. range 3-6 nm 6-20 nm Fig. 12 Pore size distribution of sol-gels processed to contain K 4 Fe(CN) 6. The measurements were made by the nitrogen adsorption method. 65

80 analogous conditions [3, 16, 179]. The general dependence of pore diameter on ph is that the minimum occurs approximately at the point-of-zero-charge, pzc (ca. ph 2); that is, at either higher or lower ph values, the pore size is greater. The reason is that the hydrolysis rate near the pzc is much faster than the condensation rate. As a result, silica units build at the ends of the developing chains, thereby yielding linear, closely packed strands with microporous interstitial spaces. At low ph values, such as the entry for ph 0.5 in Fig. 12, the hydrolysis rate is slower than that of the condensation rate, so the addition of partially hydrolyzed monomers to the growing strands occurs. Continued hydrolysis leads to highly branched polymers with interstitial spaces in the mesoporous domain. The above-described set of CoHCF-doped sol-gels was evaluated for the SPE of Cs +. Of importance is that the level of CoHCF doping was the same for all of these solgels (see part for the details on the HCF doping). The results are summarized in Fig. 13. The data suggest that pore size is not a significant factor in the SPE of Cs + by CoHCF-doped sol-gels. For example, the data set obtained with silica processed from a ph 0.5 sol (mesoporous product) was compared to that for silica from the ph 3.0 precursor (microporous product) using the t-test (8 degrees of freedom). The results were statistically identical at the 95% confidence level. The 25 trials that comprised the data set in Fig. 13 yielded an uptake capacity of ± mmol Cs g -1. The uptake capacities achieved by the sol-gels used to obtain the data in Fig. 13 are too low for practical application. Because these data along with the results in Fig. 12 also showed that increasing pore size is not a viable route to increasing uptake capacity, the remainder of this study was aimed at finding methods for increasing the doping level of CoHCF. The limiting factor is the quantity of HCF that can be doped in the silica. When it is added as a solute in the sol, the solubility of K 4 Fe(CN) 6 determines this value. The solubility is dependent on ph. When the sol contains 0.03 M HCl, the maximum quantity of K 4 Fe(CN) 6 that can be dissolved therein is M. Increasing the acid content to 0.30 M HCl results in a solubility of M K 4 Fe(CN) 6 in the sol. Based on this solubility data, the experiment in Fig. 13 with a silica prepared from a ph 0.5 sol was repeated except that the concentration of K 4 Fe(CN) 6 in the sol was increased from M to M. The resulting solid had an uptake capacity of 0.36 ± 0.01 mmol Cs g -1 (5 trials). 66

81 Capacity, mmol Cs + g Sol ph Fig. 13 Influence of the processing ph on the uptake capacity of CoHCF-doped silica solgels. This ph has a known influence on pore size. All sols contained 0.01 M K 4 Fe(CN) 6. The error bars represent the standard deviation of five trials. 67

82 The second approach to increasing the quantity of HCF in the silica was to prepare the sol-gel in the absence of K 4 Fe(CN) 6 and then sorb the HCF by immersion in an aqueous K 4 Fe(CN) 6 solution. Because silica is a cation-exchange material, this approach requires modification of the sol-gel processing to impart an anion-exchange characteristic. Therefore, generation-4 polyamidoamine (G4-PAMAM) was included in the sol as described in the Experimental section. Consistent with a prior study [14], the pore size distribution suggested mesoporosity. Each G4-PAMAM center contains 64 amine sites that are protonated at the ph of the sol-gel processing methods used in this study. Those protonated amine sites facilitate the uptake of Fe(CN) 4-6 into this matrix as shown in Fig.14. The mesoporous composite of G4-PAMAM and silica was doped with CoHCF as described in the Experimental section. An uptake capacity of 0.43 ± 0.01 mmol Cs g -1 (5 trials) was determined. This value was statistically greater (95% confidence level) than that obtained with the previously described sol-gels. Clearly, the quantity of HCF that can be doped into the silica is an important factor in determining the level of CoHCF in the solid matrix and, in turn, the SPE of Cs +. In an attempt to further improve the uptake capacity of CoHCF-doped silica, solgel processing in the absence of added alcohol was investigated. The purpose of the methanol is to make a homogeneous solution of otherwise immiscible TMOS and water. However, methanol as a bulk component of the precursor is problematic in that it limits the solubility of K 4 Fe(CN) 6. An alternative is to use ultrasound to homogenize the mixture of the tetraalkoxysilane (TMOS in the present study) and water [44]. Before using the CoHCF-doped sonogels for SPE, the pore structure was characterized by the BET method. Data were obtained both before and after conversion of the HCF to CoHCF, and the results were compared to those obtained from conventional sol-gels and from composites of silica sol-gels and G4-PAMAM. A comparison of the pore size distribution for the sonogel (Fig. 15) to the analogous data for the conventional sol-gel (the ph 0.5 entry of Fig. 12) suggested that sonicating the gelling mixture causes the resulting solid to have a microporous structure rather than the mesoporosity that is observed under conventional processing. This result is consistent with that of a study that was similar to the present one except that processing leading to 68

83 HCl H 3 + N NH 3 + H 3 + N NH 3 + H 3 + N NH 3 + H 3 + N H 3 + N NH 3 + NH 3 + H 3 + N NH 3 + H 3 + N NH 3 + H 3 + N NH 3 + H 3 + N H 3 + N NH 3 + NH 3 + H 3 + N NH 3 + H 3 + N NH 3 + H 3 + N NH 3 + H 3 + N H 3 + N NH 3 + NH 3 + Fig. 14 Reaction scheme for the formation of CoHCF by sequential immersion of G4- PAMAM-doped sol-gel into Fe(CN) 4-6 and Co

84 an aerogel was used [47]. In terms of apparently changing the pore size distribution, the efficacy of the procedure employed herein was somewhat surprising in that sonication was performed in a bath rather than with immersion of a probe directly into the sol. The data in Fig. 15 do not conclusively prove the assignment of microporosity of the sonogel because pore sizes below 6 nm are not resolved. However, further analysis of BET data supports this suggestion (Table 1). The entry for the sonogel prepared from a ph 0.5 sol relative to that for a conventional sol-gel prepared at that ph shows that the former has a surface area about 20% larger than the latter even though the pore volumes are similar. This observation is consistent with a decrease in mean pore diameter when the sol is sonicated during gelation. Moreover, the effect of ph on the pore volume in Table 1 is consistent with the known behavior for conventional processing, namely that at ph 1.5 a microporous sol-gel is formed whereas at ph 0.5 the material is mesoporous. The purpose of investigating gelation during sonication was to increase the quantity of K 4 Fe(CN) 6 that can be encapsulated, as discussed above. An outcome of this strategy is that the resulting increase in the quantity of CoHCF in the sol-gel has a profound influence on the pore structure. For example, the surface area for the sonogel in Table 1 decreases from 457 to 4.2 m 2 g -1 upon conversion of the encapsulated HCF to CoHCF. Apparently, the solid product, CoHCF, blocks or partially fills the micropores of the material. With the conventional sol-gel processed at ph 0.5, the analogous result was a change from 387 m 2 g -1 to 86 m 2 g -1. Nevertheless, the sonogel yielded a greater uptake capacity, 0.61 ± 0.01 mmol Cs + g -1 (5 trials) than the corresponding sol-gel processed with conventional stirring, 0.36 ± 0.01 mmol Cs + g -1 (5 trials). These relative values show that the increased amount of HCF that is doped into the sonogel (0.17 M vs M with the conventional sol-gel) is a more important factor than the possibly lower diffusion coefficient of Cs + in the microporous sonogel than in the mesoporous sol-gel processed at ph 0.5. The statistical data on the sonogels, indeed on all of the sol-gels in this report, are independent of batch; that is, the standard deviations of the uptake capacities are in the 3% range when the preparations of the solid media are repeated. Moreover, the results on uptake capacity do not change when the sol-gels are older than 10 days (up to at least 2 months) even though the pore size decreases with time. 70

85 % total volume pore under over 20 Pore dia. range (nm) Fig. 15 Pore size distribution of a sonogel processed to contain K 4 Fe(CN) 6. 71

86 Table 1. Characterization of the pore structure of sol-gels processed to contain K 4 Fe(CN) 6 Sol ph Convection K 4 Fe(CN) 6 Pore volume Surface area M cm 3 g -1 m 2 g stirred stirred sonicated Measurements made by the nitrogen adsorption method 72

87 The selectivity of the uptake of cesium in relation to that of sodium and calcium was investigated using the CoHCF-doped sonogel. Typical results are summarized in Table 2. When the Na + concentration was varied over the range, 0.5 mm-100 mm, the uptake Cs + from a 0.5 M sample was not changed at the 95% confidence level. The same observation was made with 10 mm Ca 2+ as the test interferent. However, increasing the concentration of Ca 2+ to 100 mm caused a statistical decrease in the uptake capacity of the CoHCF-doped sonogel. The data not only demonstrated selectivity of this solid phase toward Cs + but also showed that the mechanism of cesium sorption on CoHCF system was not ion exchange. If it were ion exchange, the propensity of a dication relative to a monocation to interact with an ionic site would have resulted in preferential sorption of Ca 2+ over that of Cs +. Instead, the Cs + is probably incorporated into the CoHCF cage structure, which is shown in Fig. 11 (here, M A is Co 2+ and M B is Fe 2+ ), as an ion-pair with NO - 3. The selectivity of CoHCF would then relate to the relative sizes of the cage and the Cs + ion. Thus, this study is in agreement with the uptake mechanism that was suggested by Ayrault et al. [161]. Finally, the question of whether cesium sorbed by the CoHCF-doped sonogel will leach from this material was addressed. After sorption of Cs +, the sonogel was dried at 25 o C, and 10 mg of the powder was stirred in 20 ml of 0.01 M HCl for 24 hours. The filtrate was analyzed for cesium by ion chromatography. No Cs + was detected in the filtrate. The detection limit of the ion chromatograph used was 100 µg Cs + L Conclusion The capacity of CoHCF-doped sol-gel materials for solid phase extraction of Cs + is dependent on the amount of CoHCF doped in the solid and not on the pore size of the material. The use of alcohol in conventional sol-gel processing limits the doping level because it limits the amount of K 4 Fe(CN) 6 that can be dissolved in the sol. In turn, this limits the quantity of CoHCF that can be formed by immersion in a Co 2+ -containing solution. Two methods were found to increase the doping level. First, a silica sol-gel composite with G4-PAMAM was prepared; sequential immersion in an HCF and a Co 2+ solution doped the silica with CoHCF. Second, sonication of a sol that contained 73

88 Table 2. Influence of Na + and Ca 2+ on the uptake of Cs + by CoHCF-doped sonogels Na +, mm Ca 2+, mm capacity, mmol Cs + g ± a ± ± ± ± ± 0.05 a Determined by loss of Cs + from 20 ml of 0.5 mm sample when 10 mg of powdered sonogel was added and stirred for 24 h 74

89 K 4 Fe(CN) 6 was used to avoid the use of alcohol as a solvent. The resulting CoHCFdoped silica sonogel provided the highest uptake capacity (0.6 mmol Cs + g -1 ) of the systems we investigated. Concentrations of Na + up to 100 mm and of Ca 2+ up to 10 mm did not influence the sorption of Cs + from a 0.5 mm solution. The interference study supported the view that the sorption of Cs + by CoHCF is by insertion into the cage structure of this material rather than by ion-exchange. The amount of Cs + leached from the doped sonogel into a ph 2 solution over a 24-h period was too low to be detected by ion chromatography. Hence, CoHCF silica sol-gels are promising both for the solid phase extraction of Cs + and for capture and storage of this cation. The latter application would be aided by sintering the glass-like material after sorption of Cs + to collapse the pores. 75

90 Chapter 4 Crystal Growth and Fiber Extrusion with Sol-Gel Matrices 4.1 Introduction Methods used to produce crystals have been developed over the years as a result of increased understanding and advancing technology, but crystallization strategies continue to be rooted in trial-and-error approaches refined over the past decades [180]. A very large number of materials have already been grown as single crystals, some with relative ease and others only after long research. Some substances have never been obtained in the required size or degree of perfection [181]. The aim is to create a solution supersaturated with a substance that will produce single, well-ordered crystal. However, sometimes, supersaturated solutions, especially in the case of proteins, produce precipitates or phase separation instead. There is no priori theory for discerning which solution will produce crystal and which will produce a precipitate. Hydrogels are known to be convenient, efficient, and inexpensive media for growing high-quality crystals of small molecules [181]. Although they were introduced for protein crystallization before [182], they are still seldom used despite their many advantages. A reason might be that the potential advantages of gels have not been accompanies by real applications in structural biology. Further, the expected enhancements on crystal quality have not been supported by strong experimental evidence in the macromolecular field [183]. Nevertheless, among their properties, they reduce convection in the mother solution and favor diffusive transport of the molecules towards crystal nuclei and growing crystals, mimicking in some ways conditions occurring in a microgravity environment [183, 184]. The framework of the gel also traps crystal nuclei, preventing crystal sedimentation, and promotes their three-dimensional growth. Gels also reduce incorporation of impurities, thereby suppressing twinning formation. In addition, gels may be used to control the number of nuclei [182]. This chapter will briefly review methods to grow crystal in gels and present our new method to grow inorganic crystals and extrude fibers using sonogels (with TMOS as 76

91 precursors) at room temperature. Our work is a first step toward growing protein crystals and fibers, which are usually difficult to obtain. 4.2 Crystal growth in gel materials In general, a crystallization process is separated into two stages: screening and optimizing [180, 185]. Each stage is conducted with a largely empirical approach. The screening process discovers first crystallization conditions, which typically produce microcrystals, thin rods or thin plates. There are three phases of crystal growth: nucleation, growth and cessation of growth. Screening is largely concerned with finding condition that will support nucleation and some crystal growth. The number of variables affecting crystallization, such as concentration, temperature, ph, ionic strength, specific additives and precipitants, is large [185]. As a result, the total number of possible solution conditions to be tested is very large. One approach to overcome this problem is initially to use the incomplete factorial method of Carter & Carter [185], in which a very coarse matrix of crystallization conditions is explored and the results analyzed to build a grid around deduced and projected conditions of the initial incomplete factorial. The crystallization conditions are then optimized to produce a crystal suitable for x-ray diffraction, i.e. large and well-ordered. Here, large may mean just tens of microns on edge [180] Crystal growth in inorganic gels The hydrogel method to grow crystals is comprised of two steps [181]. The first step involves preparation of a hydrogel from commercial waterglass or, more preferably, from reagent grade Na 2 SiO 3.9H 2 O (to eliminate undesired impurities). The second step is to place some other solution, once a firm gel is formed, on top of the gel without damaging its surface. This solution supplies one of the components of the reaction and also prevents the gel from drying out. If the reagent in the gel is tartaric acid and the supernatant reagent is an approximately 1M solution of calcium chloride, then, in due course, crystals of calcium tartrate tetrahydrate are form in the gel as the result of an exchange reaction. The crystals are only sparingly soluble in water. Another example is growing calcite crystal in a hydrogel from Na 2 SiO 3.9H 2 O [186]. Acetic acid or HCl was 77

92 added as acidifying agent. The gel s ph was ranging between 7 and The solution containing the Ca 2+ ion was prepared from CaCl 2 aqueous solution, whereas the CO 2-3 ion was obtained from NaHCO 3 or Na 2 CO 3. Gel growth experiments were conducted in two types of tubes: a three-arm device, where the central arm received the counterdiffusing solutions and held the seed crystals, and a classic U tube. The speed of crystal formation depends on the concentrations involved [181]. The ph of the surrounding environment influences the shape of the crystal. For example, when high acid concentrations are used, sodium tartrate crystals tend to grow in the form of long clear needles. As a general rule, very dense gels produce poor crystals. On the other hand, gels of insufficient density take a long time to form and are mechanically unstable. A specific gravity of 1.02 g cm -1 appears to be the lower practical limit [181]. The second reagent is not necessary solution. Gas reagents under varying pressure can be used, and these offer the additional possibility of extending the temperature range within which the experiments are carried out [181]. The silica itself need not be necessarily acidic nor need it be based on sodium metasilicate; various proprietary silicas as well as agar gel can also be used. The proprietary silicas are free of sodium, which should be an advantage in principle, since sodium is usually a contaminant and not part of any essential reaction. However, the presence of sodium was found beneficial for large crystal growth. The gel acts as a three-dimension crucible, which support the crystal and, at the same time, yields to its growth without exerting major forces upon it [187]. This relative freedom from constraint is believed to be an important factor in the achievement of high structural perfection. The other important function of the gel is to suppress nucleation [181], thereby reducing the competitive nature of the growth. In fact, nucleation control is one of the keys to the ultimate success of the gel method for single crystal growth, and just because nucleation is suppressed, a very high degree of supersaturation can be obtained without leading to immediate and total precipitation in an amorphous form. The growth mechanisms operative in the gel after nucleation should not be very different from those in stagnant solution, though complication may arise in cases in which the gel serves both as a reaction medium and a diffusion medium [181]. 78

93 It should be noted that very little has been done on the gel-growth of watersoluble crystals, though this subject is of special interest because for such crystals a comparison is possible between the degree of perfection achievable in a gel and that obtained in solution. There are two approaches to growing water-soluble crystals in a gel: a) lowering the temperature, and b) allowing a solvent in which the crystal is less soluble to diffuse into the gel medium [187], as presented in the next paragraph. It was found that the growth rates in the gel and in solution are about the same. Gel-growth crystals were fount to be superior to those grown in solution. Their dislocation density was especially low [181]. Nasimova at el. [187] studied crystal growth of copper(ii) sulfate pentahydrate crystals in covalently bonded, slightly crosslinked gels of poly(n-vinylcaprolactam) (PVCa). PVCa is a nonionic synthetic polymer soluble in both water and organic compounds. This property can be used for the growth of soluble crystals that can make use of the variation of solubility by the addition of another solvent. The types of interaction between the macromolecule and the low molecular weight substance can be electrostatic, van de Waals interaction, and hydrogen bonding. PVCa gel was first prepared; then the gel sample was immersed into CuSO 4 solutions at room temperature for one week to concentrate CuSO 4 inside the gel. The contraction of gel was observed and was explained by i) creation of additional bonds as a result of complex formation of Cu 2+ ions and amide groups of PVCa chains and ii) increasing hydrophobic interactions as a result of water structure changes in the presence of salt ions. In the next step, the CuSO 4 -bearing hydrogel sample was placed in a solution of 97% ethanol. The high saturation necessary for crystal growth in the gel was related to the fact that i) ethanol diffused into the gel but water diffused from the gel into the external solution and ii) CuSO 4 is readily soluble in water but poorly soluble in alcohol. Following are some other examples of crystals grown in hydrogels. Kumar and Wang [188] grew brushite (CaHPO 4.H 2 O) single crystals by using Ca(NO 3 ) 2.4H 2 O and (NH 4 ) 2 HPO 4 as the inner and outer reactants, respectively. Sodium silicate was the gel medium. The growth of large and high quality protein crystal has been a bottleneck for the determination of three dimensional structures. Microgravity experiments [189] showed that protein crystals which nucleated and grew at the center of a large container 79

94 were much bigger than those that nucleated at the wall. Theses experiments demonstrated that the concentration gradient layer around a growing crystal was a key factor in crystal growth, influencing protein transport and therefore, crystal growth rate, size and quality in the diffusion controlled region. Hou et al. [190] exploited the use of a gel to mimic microgravity conditions to investigate mass transport properties during protein crystallization by visualizing the concentration gradient layer around a protein crystal in a gel and comparing concentration gradient properties in gel and solution. It was found that concentration gradients were wider and transport rate were slower for gel-grown crystals than for the solution counterparts. Sanz et al. [191] grew nanocrystals of 1-cyano-1-(4-nitrophenyl)-2-(4- methoxyphenyl)ethane, CMONS, in silica gel, using TMOS and methyltrimethoxysilane (MTMOS) as gel precursors. The bulk gel matrix was prepared under acidic catalyzed conditions. CMONS was included in the initial solution of alkoxide precursors. The solution was first annealed at 80 o C. After few hours, a transparent gel was obtained. It was aged at 80 o C for 2 days in order to adjust the gel porosity. The nucleation of CMONS in the pores of the gels was induced by a high supersaturation created by dropping the temperature from 80 to 20 o C. The samples were then dried slowly for 1 month to avoid cracking. CMONS nanocrystals had better thermal stability in comparison to pure CMONS. It was attributed to hydrogen bonding interactions between the surface of the CMONS nanocrystals and the sol-gel matrix Crystal growth in organic gels Crystal growth in organic gels is applied to not only small molecules, but also to proteins. Agarose gel has been most widely used, probably because it seems more reliable for biological products due to being neutral. Its advantage as a gel growth medium is to promote nucleation [192, 193]. Agarose is polysaccharide extracted from red seaweed. The element unit of the polymer is made of two cycles carrying different substituents, methyl and sulfate. It is a powder soluble in water above a temperature of about 100 o C. By lowering the temperature, the solution gels at a temperature, T g, that depends on the exact nature of the gel. The gelation shows an important hysteresis effect: 80

95 the gel melts again only about 40 o C above T g. The temperature of gelation depends on the concentration in agarose, salt, ph, and the cooling rate. Sauter et al. [183] conducted a microgravity experiment in which an agarose gel was added to the protein solution. The gel was believed to play a role during and after crystallogenesis. It was designed to protect the crystal against vibrations and shocks upon landing and during subsequent transport prior to their analysis. Saute et al. [183] reported a growth of thaumatin crystals in a sodium tartrate solution gelified with 0.15% (m/v) agarose. Tartrate played a role as a crystallizing agent for thaumatin. It was also considered as an additive that contributed to crystal cohesion. Thaumatin crystallized as individual and well-shaped tetragonal dipyramids. The quality of the crystals was sufficiently high for solving the structure at atomic resolution. However, the interaction between thaumatin and the agarose gel could not be detected. Boue and co-workers [194] demonstrated that hen egg white lysozyme crystals nucleate more easily in agarose gel than in gel-free solutions. It was attributed to lysozyme clusters of several tens to several hundreds of nanometers that increased with increasing supersaturation; it was noticeably stronger in gel than in solution. This effect was not observed in the agarose gel. A 1% (wt/vol) agarose sol was prepared at high temperature. An aliquot of this sol was added to a warm premixed solution containing NaCl and acetate buffer; then a solution of protein in acetate buffer was added. The temperature of gelation of agarose was chosen in order to avoid thermal degradation of the protein. The mixture was allowed to cool to room temperature. The salt content was used to increase supersaturation. There was no influence of the agarose gel on the arrangement of the protein gel; this reinforced the view that agarose is neutral to proteinprotein interactions. 4.3 Experimental Reagents and materials The CuSO 4.7H 2 O and CuSO 4.5H 2 O were purchased from Fisher Scientific (Fair Lawn, NJ). Triton X-114 (octylphenoxy polyethoxyethanol) was from Sigma (St. Louis, 81

96 MO). The other reagents used in this experiment were the same as those described in part Preparation of salt-doped sol-gels for crystal growth K 4 Fe(CN) 6 -doped sonogels were prepared the same as described in part Crystals of CoSO 4 and CuSO 4 were grown from sol-gel doped with CoSO 4 and CuSO 4 salt, respectively. The doping silica sol-gels were prepared from precursor solutions comprising 4 ml each of TMOS, CoSO 4 or CuSO 4 solution (comprising of 2.7 ml of the salt-saturated solution and 1.3 ml distilled water), and 1 M HCl. The processing was initiated by sonicating the sol for 30 minutes, during which the precursor solution was covered with Parafilm. After the mixing step, 2-mL aliquots of sol were pipetted into hexagonal polystyrene weighing dishes (1.5" x 1"), which were then sealed with Parafilm to control the vapor pressure in the container, thereby allowing the sol to age before gelation. After five days, several pinholes were made in the film so that the solvent and the alcohol that is released during gelation can evaporate gradually. The Parafilm was removed after 10 days. Doped sol-gels templated by surfactant were prepared by the same procedure as described above. The only difference was that 0.2 ml of methanol and 2 drops of surfactant, TX-114 were added to the initial sol mixture, and the sol was stirred instead sonicated Apparatus Elemental analyses were done by ion chromatography with a DX-120 system (Dionex, Sunnyvale, CA) and inductively couple plasma-atomic emission spectroscopy (ICP-AES) with a Liberty 150 system (Varian, Chicago, IL). 4.4 Results and discussion When doing the experiment, as described in the Experimental section, Chapter 3, we observed that after two weeks of aging and drying, very fine fibers like hair grew on the surface of the sonogel monolith when it was doped with the maximum amount of K 4 Fe(CN) 6. They are shown in Fig. 16. In some batches the fibers had a white color, while in the others, the fibers were deep blue. The fibers were about 100 µm in diameters, 82

97 Fig. 16 Fibers on the surface of a monolith. The approximate widths are 100 µm 83

98 and they had a maximum length of 2 cm (after about a two-month growth). The fibers were harvested for elemental analysis. The IC and ICP-AES were used to analyze the white fiber in solution since they dissolved completely in water. The data showed the presence of iron and cyanide in the white fibers, though in very small amounts (less than 1% wt); the main component of the fibers was KCl, which must have been produced from metathesis of K 4 FeCN 6 and HCl contained in the sol-gel. Since the blue fibers were not dissolved completely in water, some blue precipitate still remained in the solution. The fibers were analyzed in their solid state by a FTIR. The FTIR spectra showed a presence of Prussian Blue in the blue fibers (Fig. 17). It raised a question about the source of Fe(III), since the iron doped in sol-gel was in the form of Fe(II)(CN) 4-6. A possibility was Fe(III) came from HCl as an impurity; also possible was slow decomposition of hexacyanofferate. This explained why in some batches the fibers had blue color and in the other batches they had white color. The amount of PB produced was dependent on the amount of Fe(III). Another question was about how the fibers grew. We conducted the following experiment for the answer. When the white fiber was grown up to few mm, we used Co(NO 3 ) 2 to mark the head of the fiber by gently contacting the head of the fiber with Co(NO 3 ) 2 crystals until a deep blue spot appeared on the fiber. The blue spot was the product, CoHCF, of reaction between Co(NO 3 ) 2 and K 4 Fe(CN) 6 present in the fiber. The fiber was then allowed to continue to grow. It was shown that the fiber grew from its bottom, like hair growth. That is, it was extruded from the monolith rather than the alternative, growth by precipitation of salt at the end of the fiber (which has surface water present). Crystals of CoSO4 and CuSO4 on the surface of the sol-gel monoliths (Fig. 18 and Fig. 19) and inside the gel (Fig. 20) were also grown. The mechanism of the crystal growth can be explained as follows. The salt solution was confined in pore pockets of the doped sol-gel. During drying process, the solvents (alcohol and water) were evaporated, resulting in supersaturation for crystal growth. Because the liquid in the pore receded into the sol-gel due to the capillary force (see discussion in Chapter 1), the only way that crystals formed outside the sol-gel was via extrusion due to the collapse of the gel during drying. Therefore, the strength of the sol-gel was critical for the inside-gel or outside-gel 84

99 Fig. 17 FTIR spectra of Prussian Blue (A) and of the blue fiber (B) 85

100 Fig. 18 CoSO 4 crystals formed on the surface of the sol-gel 86

101 Fig. 19 CuSO 4 crystals formed on the surface of the sol-gel 87

102 Fig. 20 CuSO 4 crystals formed inside the sol-gel 88

103 formation of the crystals. The crystals formed outside of the gel were easy to harvest but had imperfections. The size of the crystals varied, depending on the rate of aging and drying sol-gel. Crystals with a size of few mm were obtained when the sol-gels were aged for 5 days; rod-like crystals with a length of 2 mm were observed when the sol-gel was aged for 3 days. The differences probably reflected the structural evolution of the gel. With shorttime aging, the gel network was weak and was subject to collapse during the drying process, resulting in smaller pore sizes, which in turn shaped the extruded crystals as rodlike. The following are factors that influenced the crystal growth: 1) concentration of the doped salt; it must be sufficiently high to reach supersaturation for crystal growth. For example, when the concentrations of the salt doped in sol-gel were below 0.04 M, no crystals were observed even when the sol-gel was allowed to dried for months; 2) humidity; it was observed that a humidity of 40% was necessary for the fiber extrusion, but it is unclear why; 3) the rate of aging and drying of the sol-gel; bigger crystals were observed when the gel was aged and dried for long period of time (10 days) while small, rod-like crystals were obtained from those gels aged for a shorter period (3 days); 4) morphology of the gel; for the surfactant-templated sol-gels, where mesoporous domains exist, no crystal growth was observed (by eyes). Perhaps, the crystals were still formed due to supersaturation when the solvent in the gel evaporated; however, relatively large pores and rigid sol-gel network may have held the crystals inside the pores. 4.5 Conclusion PB-containing fibers have been fabricated by using sol-gel media, but the content of PB in the fiber was small. The importance of this material lies in its interestingly electrochemical properties, as mention in Chapter 2. There are two goals for future studies: 1) increase the amount of PB in the fibers by introducing trace Fe(III) into the initial sol mixture instead of relying on an uncontrolled source; and 2) immobilize the blue fiber onto a gold interdigitated array electrode for electrochemical study. We successfully demonstrated the crystal growth of water-soluble substances by using sonogels as the media. Our approach was different from the two approaches in the 89

104 literature, namely cooling the gel or introducing into the gel a solvent, in which the substance is poorly soluble, to enhance supersaturation. In addition, the present study has two other significant differences from previous reports. First, the sol-gel was synthesized via sonication. Second, we were able to crystallize a substance outside the gel. The importance of this study was that silica gel-grown crystals of CoSO 4 and CuSO 4 are surrogates for gel-growth of protein crystals, which are usually difficult to obtain by conventional methods. It highlights sol-gels as not only a good matrix for biological substances, as discussed in Chapter 1, but also as crystallization media. 90

105 Chapter 5 Gold Nanoparticles 5.1 Introduction It is probably that soluble gold appeared around the 4 th or 5 th century B.C. in Egypt and China [195]. In antiquity, colloidal gold was most used to make ruby glass and to color ceramics. The most famous example is the Lycurgus Cup that was made in this time. It is ruby red in transmitted light, and green in reflected light due to the presence of gold colloids. During the Middle Age, colloidal gold was also used as a medicine for various diseases. In 1857, Faraday reported the formation of a deep red solution of colloidal gold by reduction of an aqueous solution of tetrachloroaurate using phosphorus in CS 2 [195]. The 20 th century can be regarded as a boom of investigation of gold, some other noble metal, and semiconductive colloids. Various methods to prepare the colloids have been introduced. Fundamental studies of material at nanoscale have been conducted more intensively. The terms nanoparticles and nanotechnology were coined. The term nanocluster or nanoparticle is used to name particles of any kind of matter whose size is between that of single molecules and about 100 nm [196]. Novel properties of such nanoclusters, whether consisting of atoms or composed of building blocks, occur on this scale compared with those of the corresponding bulk material. In the nano-scale regime, besides composition, control of size and shape means control of new structures and macroscopic properties of the material. Achievements in this area can change significantly the fields of information technology, communication hardware, electronics, optics, and others. This issue will be illustrated in this chapter; gold nanoparticles (AuNPs) will be the focus. The rest of the chapter is constructed of six parts: synthetic methods, controlling size and size distribution, structure, size-related properties, characterization techniques, and applications of AuNPs. 91

106 5.2 Synthetic methods Metal nanoparticles can be produced by chemical, electrochemical and physical methods. The first method has been used most widely; only this method will be discussed in this dissertation. The requirement to study the behavior of nanoparticles is that the particles must be separated from each other to avoid coalescence and to keep the individual nature of the particles. Numerous materials have been employed to protect metal NPs from coalescence; among them ligand protection is most commonly used. The ligands do not only to protect nanoparticles from aggregation but also impart the nanoparticle characteristics, such as solubility, electrochemical properties, and charges. For example [196], owing to their ligand shell, even 30 nm particles become soluble in appropriate solvents. From non-polar solvents such as pentane, to water as an extremely polar medium, all kinds of solvents can, in principle, be used if the ligands bear appropriate substituents. This is a big advantage for the treatment of clusters, also with respect to the generation of cluster arrangement. This section will review the utilized ligands and, briefly, some other materials often used in the synthesis of AuNPs. Thiol compounds and dendrimers, which were used in our research for protecting NPs will be the focus Citrate reduction One of the conventional methods to synthesize AuNPs is reducing HAuCl 4 to Au 0 by citrate under heating and stirring the solution [197]. The method was introduced by Turkevitch et al. in Here, citrate plays two roles: the reducing and the protecting agent. The method leads to AuNPs of ca. 20-nm diameter. Later, Frens [198] reported a method to control the size of AuNPs by manipulating the ratio of citrate and gold salt. With this method it is able to obtain AuNPs in a range of 16 to 147 nm. The Kunitake group [199] modified the method to obtain AuNPs with smaller sizes. A mixture of trisodium citrate and sodium 3-mercaptopropionate, which was used as a stabilizer, was added to a HAuCl 4 solution, which was refluxed. The particle size was controlled in a range of nm by varying the stabilizer/gold ratios. 92

107 5.2.2 Alkanethiolate-stabilized AuNPs a) Two-phase synthesis In 1993, Mulvaney and Giersig [200] were the first ones who reported the stabilization of AuNPs with alkanethiols and showed the possibility of using thiols of different chain lengths. However, citrate was still employed as the reducing agent. As a result, AuNPs were protected by a mixture of thiol and citrate. In 1994, Brust et al. [201] introduced a method, using two-phase (water-toluene) reduction of AuCl - 4 by sodium borohydride in a presence of an alkanethiol to synthesize AuNPs. The AuNPs had diameters in a range 1-3 nm, and a maximum in the particle size distribution at nm. This is the first time the AuNPs sizes were controlled in that nanoscale range. Importantly, these AuNPs could be repeatedly isolated and re-dissolved in common organic solvents without irreversible aggregation or decomposition. They were air-stable and easy to handle and functionalize as stable organic and molecular compounds. These AuNPs can be purified and kept in the solid state under ambient conditions for months without showing significant aging effects. In more detailed application of the method, - AuCl 4 was transferred from an aqueous solution to toluene using a phase-transfer reagent, tetraoctylammonium bromide, and reduced by NaBH 4 in the presence of dodecanethiol. The color of the organic phase changed from orange to deep brown within a second upon addition of NaBH 4. Larger thiol/gold ratios produced smaller average core sizes, and fast reductant addition and low-temperature reaction also resulted in smaller, more monodisperse particles. It should be noted that not all types of thiol ligands lead to the same extraordinarily high degree of stability that can be achieved by the use of alkane thiols with a hydrocarbon chain length of C5 to C18. For example, hydroxofunctionalized NPs stabilized by 4-mercaptophenol are less stable and degrade slowly over several weeks under ambient conditions [202]. The same trend is observed if shorter (C2 to C4) thiols are employed as stabilizers. Recently, Schiffrin [203] reported a method to purify dodecanethiol-stabilized AuNPs from tetraoctylammonium bromide impurities by Soxhlet extraction. It should be noted that the Brust-Schiffrin method is well suited for the functionalization of the AuNPs with simple n-alkanethiols. However, in the case of ω-substituted thiols, forming 93

108 polar surfaces, a second layer of the ammonium surfactant covers the nanoparticles and the purification process is long [204]. A solution to this problem was a one-phase synthesis, as described below. b) One-phase synthesis Brust et al. [202] developed the synthesis of AuNPs with a bifunctional stabilizing thiol ligand; here, it was p-mercaptophenol. The preparation was carried out in singlephase system, methanol. Hydrogen tetrachloroaurate and p-mercaptophenol were dissolved in methanol, followed by an addition of a freshly prepared NaBH 4 solution. The product in a diameter range of 2.4 to 7.6 nm (a mean of 5 nm) and was insoluble in water and nonpolar solvents but dissolved readily in alcohols, ethyl acetate, and in alkaline aqueous solutions (ph 13). After the Brust et al. publications, numerous papers appeared describing the use and modifications of Brust-Schiffrin methods for synthesis of other stable AuNPs, which are also commonly called monolayer-protected clusters (MPCs). It may be troublesome to apply either of Brust s methods (as described above) to synthesize AuNPs stabilized by some thiol ligands, such as some 4 -substituted-4- mercaptobiphenyls, having low solubility in both toluene and alcohols. Yee et al. [204] developed a method allowing for the formation of alphatic and aromatic thiolfunctionalized AuNPs that cannot be done by the one-phase or two-phase synthesis. An example was preparation of octadecanethiol-functionalized gold, iridium and palladium NPs, in surfactant-free condition, using tetrahydrofuran (THF) as a solvent. The alkanethiol and HAuCl 4 were mixed in anhydrous THF, followed by an addition of lithium triethylborohydride, the reducing agent, in THF. The octadecanethiolfunctionalized AuNPs, synthesized by this method, had diameters of ca. 4-nm average size Place-exchange reactions Murray and co-workers [ ] studied and reported the place-exchange process of a controlled proportion of thiol ligands on pre-synthesized alkanethiolstabilized AuNPs. This method can generate AuNPs soluble in both water and organic 94

109 solvents. The most importance of the method is the ability to combine appropriate functionalities, including redox centers, complexing groups, negative or positive charge, mixed together on the same cluster surface. This broadens the applications of the AuNPs. When one ω-functionalized alkanthiolate is used as a replacing ligand, the resulting AuNPs are called poly-homofunctionalized monolayer-protected cluster (MPCs). When more than one replacing ligands are used, the resulting Au MPCs are called poly-heteroω-functionalized MPCs [206]. In the place-exchange reaction, a new thiolate ligand (R S) is incorporated into a cluster monolayer by mixing its thiol (R SH) and the alkanethiolate (RS) AuPMCs in solution [205]. x(r SH) + (RS) m MPC x(rsh) + (R S) m (RS) m-x MPC (14) Where x is the number of ligands placed-exchanged (1 to 108) and m is the original number (about 108) of alkanethiolate ligands per Au 314 cluster. The R SH ligand enters the monolayer and protonates a bound thiolate ligand in an associative rate-determining step; the displaced thiolate becomes a thiol product [209]. Disulfides and oxidized sulfur species were not involved in the reaction. The rates of the ligand exchange on Au MPCs depend on the mole ratio of the entering and existing ligands; it decreases with an increase in the size of the entering ligand and the chain length of the initial protecting ligand. The difference in susceptibility of the surface sites on the cluster to placeexchange was attributed to differences in binding sites [209]. Au core edge and vertice sites were presumed more reactive than terrace sites. The ligand-exchange reaction on Au MPCs was proposed similar to that on 2D-SAM (self-assembled monolayer). Exchange at defect sites lead to the incorporation of a new surface-bound thiolate. After exchange at these minority sites, the rate-determining step was either the slow migration of ligands on terrace sites to more reactive vertice/edge sites or the very slow exchange of thiol directly with terrace sites. There are two synthetic routes to synthesize poly-hetero-ω-functionalized MPCs, simultaneous and stepwise exchanges [206]. In the former route, place-exchange reactions were accomplished by co-dissolving the alkanethiolate-mpcs and the incoming ω-functionalized alkanethiols in methylene chloride, followed by stirring for at least 48 h at room temperature. Up to five different ligands were simultaneously exchanged onto the alkanethiolate-mpcs. In the latter route, progressive exchange of different thiols, 95

110 isolating and characterizing the cluster product after each step was conducted. The feed mole ratio (the entering thiolate/mpc-bound thiolate before the reaction) was not necessary the same as the product mole ratio (the introduced thiolate/un-exchanged thiolate after the place exchange reaction). Longer chain length ω-functionalized ligands tended to displace those with short, bulky linker chains; this suggests that short chain length, bulky, ω-functionalized alkanethiolates are the least thermodynamically stable ligands. Importantly, this strategy allowed for functionalization of clusters while retaining the dimension of the Au core [207], which potentially could be changed when employing functionalized alkanethiols in a de novo synthetic fashion [203]. In later study on steric effects, Murray and co-workers [208] synthesized ω- bromoalkanethiolate-functionalized MPCs by place exchange reaction on alkanethiolate MPCs and then conducted the reaction of ω-bromoalkanethiolate-functionalized MPCs with primary amines. They showed that the S N 2 reactivity of ω-bromoalkanethiolatefunctionalized MPCs responded to the steric bulk of the incoming nucleophile (rate of n- propylamine > isopropylamine > tert-butylemine) and to the relative chain lengths of ω- bromoalkanethiolate and surrounding alkanethiolate chains (rates of C12:C12Br > C12:C8Br > C12:C3Br). In another study, Murray and co-workers [207] reported amide and ester coupling reactions on ω-functionalized MPCs. The coupling reactions were either of alcohols or amines with MPC ω-carboxylic acid groups or of carboxylic acids with MPC ω-alcohol groups. This approach offers several advantages over the place-exchange route. The latter route consists of two steps. First, the amide and ester reactions are performed to attach interesting structural groups to the free alkanethiol. Second, by the place-exchange process, those amide or ester thiols are bound to the gold core. However, this route involves a number of steps: protection of the thiol via a disulfide or other appropriate protecting group prior to the coupling reaction, the removal of excess reactant and side product, and finally deprotection to again obtain the thiol. In contrast, the coupling reaction conducted on ω-functionalized MPCs alleviates the needs for protecting/deprotecting the thiol functionality. In addition, unreacted reagents are simply removed by filtration. 96

111 5.2.4 Dendrimer-stabilized NPs In 1998, Crooks and co-workers [210], and Tomalia and Balogh [211] introduced methods to synthesize dendrimer-encapsulated metal nanoclusters. The two methods were very similar. Both groups investigated two types of G4-PAMAM dendrimers, the amine-terminated and the hydroxyl-terminated dendrimers. The main difference was that Crooks and co-workers used NaBH 4 as the reducing agent, while Tomalia and Balogh used aqueous hydrazine. In both methods, copper(ii) ions were first loaded inside G4- PAMAM via complexing between Cu(II) and the interior tertiary amine groups of the PAMAM. Second, chemical reduction of Cu(II)-loaded G4-dendrimer, resulted in intradendrimer Cu 0 clusters. The Crooks group s synthetic scheme is shown in Fig 21. Interestingly, the Cu 0 clusters obtained by Tomalia and Balogh reportedly had diameters in the rage nm; while those from the Crooks group had reported diameters less than 1.8 nm, much smaller than the 4.5 nm diameter of G4-PAMAM. Crooks believed that his group s data indicated that the copper atoms coalesced into a single nanocluster inside the dendrimer [212]. According to Balogh and Tomalia, their data suggested that the interior of the dendrimer was loaded with many single-copper atoms or very small copper clusters [212]. It is seen in all of the Crooks group s published schemes that only one particle appears inside, but offset from the center of the dendrimers, while in the Tomalia group s schemes, several particles are shown as located in the voids of the dendrimers. In term of solubility and some other chemical behavior, these nanocomposites behave like the dendrimer, but at the same time, they also exhibit the unique optical, magnetic, or other properties of the encapsulated metal nanocluster [212]. The dendrimer s outer surface can be functionalized with primary amines, hydroxyls, or other groups. These surface groups can be modified to make the dendrimer soluble in any solvent. Using dendrimers is a new and good way to controllably produce and stabilize metal nanoclusters. Such nanoclusters are of great interest not only for fundamental studies but for numerous potential applications in catalyst, optoelectronic, nanodevices, environmental clean up and medication. After these publications, a variety of assemblies between different PAMAM dendrimers with different generations, and different metal 97

112 M n+ BH 4 Metal nanoparticle Fig. 21 Synthetic scheme for interdendrimer metal nanoparticles (based on ref. 216) 98

113 NPs (mostly gold, silver, platinum and palladium) or inorganic compounds (CuS, CdS) has been reported. In term of the number of publications on dendrimer metal NPs, groups of Crooks, Tomalia, and Esumi dominate. Other groups, including those of Amis and Murphy also have strong influence. According to Tomalia and co-workers [213], there are three basic dendrimer nanocomposite structures described by the location of the inorganic domains: internal (I), mixed (M) and external (E) structures as shown in Fig. 22. The Crooks group endeavors to synthesize metal ions dominantly within the interiors of dendrimers (I type). There are two approaches reported. The first approach is using dendrimers with non-complexing terminal groups. For example, nth-generation hydroxyl-terminated PAMAM dendrimers (Gn-OH) are used in the synthesis process to avoid complexation of the metal ions with the terminal groups [210, ]. The studies showed that many metal ions, including Cu 2+, Pd 2+, Pt 2+, Ni 2+, Au 2+, and Ru 3+, sorb into Gn-OH interior over a wide range of ph via complexation with the interior tertiary amines. Therefore, upon chemical reduction, the loaded metal ions inside the Gn-OH form intradendrimer nanoclusters. In the second approach, complexation of the metal-ion with the amine-terminated dendrimers is prevented by adjusting ph so that the surface primary amines (pk a = 9.5, which are more basic than the interior tertiary amines, pk a = 5.5) are selectively protonated [210, ]. Interdendrimer nanoparticles synthesized by these methods had diameters less than 1.8 nm. In contrast, in a synthetic process, where surface primary amines were not protected, the size of clusters was larger than 4 nm [210, 223]. It was attributed to the agglomeration of metal particles formed on the dendrimer exterior. The Esumi group has reported a number of preparations of dendrimer-protected metal NPs of types E and M. The NPs of type E obtained when dendrimers with unprotected surface amine groups [224, 225], sugar-persubstituted PAMAM (sugar ball) [226, 227], and dendrimers having carboxyl groups [225] were used. Here, the metal ions complexed with the surface functional groups of dendrimer before the reduction. As a result, the NPs were formed on the peripheries of the dendrimers after reduction. Type M NPs were produced when the PAMAM dendrimer with methyl ester-terminal groups in ethyl acetate was used [228, 229]. In this case, there were three possible types of - interactions between AuCl 4 and the dendrimer: a) ion pairs formed between 99

114 "I" "M" "E" Fig. 22 Dendrimer nanocomposite structures: internal I, mixed M and external E type of dendrimer-nanocluster structures. The spheres represent individual atoms or molecules (based on ref. 213). 100

115 - AuCl 4 and the protonated tertiary amines of PAMAM; b) chelations of Au(III) by one (or two) tertiary amine(s) and two terminal methyl ester groups; and c) chelations of Au(III) by two tertiary amines and two adjacent amine groups of the dendrimer. Zhao and Crooks [230] developed an intradendrimer metal-displacement method to synthesize dendrimer-encapsulated metal clusters, whose precursor ions, for example Ag +, are not strongly complexed with the interior amines of the dendrimer. In this approach, dendrimer-encapsulated Cu nanoclusters are first prepared; then, upon an exposure to Ag +, the Cu particles are oxidized to Cu 2+ ions, and Ag + is reduced to yield dendrimer-encapsulated zerovalent AgNPs (see Fig. 23). Other noble metal nanoclusters, such as Au, Pt, and Pd, can be synthesized by this method as well. In 2001, Crooks and co-workers [231] proposed three approaches to preparation of dendrimer-encapsulated bimetallic NPs. First, mixed-metal intradendrimer NPs could be prepared by a partial displacement reaction. For example, when less than a stoichiometric amount of Ag +, Au 3+, Pd 2+, or Pt 2+ is added to a G6-OH(Cu n ) solution, it is possible to form dendrimerencapsulated Ag/Cu, Au/Cu, Pd/Cu, or Pt/Cu bimetallic NPs, respectively. Second, bimetallic NPs could be synthesized by the sequential loading methods. For example, dendrimer-encapsulated PtNPs is prepared; then, PdCl 2-4 is added to form a mixed metalion intradendrimer composite, followed by the second reduction. The third approach involves a simultaneous co-complexation of two different metal ions, followed by a single reduction step. In 1999, the Tomalia and Balogh group [232] presented a method, like the second approach from the Crooks group to prepare interadendrimer Au-Ag alloy NPs. The product formation was confirmed by the UV-visible spectrum. In 2003, Chung and Rhee [233] were the first to report a method, which was similar to the third approach proposed by the Crooks group in 2001, to prepare Pt-Pd bimetallic NPs encapsulated in G4-OH and G4-NH 2 PAMAM dendrimers. It is claimed that the resulting NPs had a diameter about 2.3 nm and were nearly spherical and uniform. The formation of bimetallic NPs rather than a mixture of Pt and Pd were confirmed by comparing the EDS (X-ray energy dispersive spectroscopy) analysis and the UV-visible spectra of the resulting NPs with: i) that of the monometallic Pt and Pd NPs and ii) their physical mixture. They also demonstrated a cooperative catalytic effect, 101

116 Cu 2+ G6-OH G6-OH(Cu 2+ ) n Reducing agent G6-OH(Cu n ) Ag + Au 3+, Pt 2+, or Pd 2+ G6-OH(Ag 2n ) G6-OH(Au 0.67n, Pt n, or Pd n ) Fig. 23 Scheme of intradendrimer metal-displacement (based on ref. 231) 102

117 which was attributed to the presence of bimetallic metal NPs. Later, Crooks and coworkers [218] published their synthesis of bimetallic Pd-Pt dendrimer-encapsulated NPs. The method used to prepare the NPs was generally the same as their 2001 procedure. The resulting NPs had an average size of about 2 nm. The formation of bimetallic NPs was also confirmed by EDS analysis. Their conclusion agreed with that from Chung and Rhee [233] that higher Pd loadings resulted in higher activities of the bimetallic NPs. The main 2- differences between these methods is that the Crooks group pointed out that PtCl 4 2- formed a complex with the dendrimer very slowly, while PdCl 4 was very quickly complexed with the dendrimer. Therefore, they let the mixture of PdCl 2-4, PtCl 2-4 and the dendrimer react for 4 days, instead of 1 h, as Min and Rhee did, to ensure complete complexation. In addition, the Crooks group showed that the atomic % s of Pd and Pt in NPs corresponded to the mol % s of PdCl 4 and PdCl 4 used in the original synthesis mixture. As mentioned above, the solubility of dendrimers in organic solvents can be controlled via functionalization of the surface groups. One method is adding functional groups to the dendrimer by covalent grafting; metal ions are extracted from an aqueous phase into non-aqueous phase and then reduced to zerovalent metal NPs in the nonaqueous phase [234, 235]. This method is slow and requires several washing and extraction iterations. Crooks et al. [220] demonstrated another method which is simpler and more versatile. This method is based on an acid-base interaction between fatty acids and dendrimer terminal-amines. First, PdNPs encapsulated in G4-PAMAM with terminal amines were prepared following the method presented previously. Then, the PdNPs were quantitatively transported from an aqueous phase to toluene in a presence of dodecanoic acid to the organic phase. The authors showed that this was a consequence of the formation of monodisperse, inverted micelles templated by the dendrimer (see Fig. 24). In another study, Niu and Crooks [236] described a preparation of dendrimerencapsulated NPs in organic solvent based on the differences in metal-ion solubility between the solvent and the dendrimer interior. These differences were used to drive metal ions into the dendrimer. This approach is basically different from the previous strategy, where interactions between metal ions and functional groups of dendrimers rely 103

118 Toluene Toluene Water Phase transfer Water Fig. 24 Synthetic scheme of dendrimer-encapsulated metal nanoclusters with the use of fatty acid as a phase transfer agent (based on ref. 220) 104

119 on complexation. Therefore, this new approach facilitates the preparation of dendrimerencapsulated NPs where the metal ions do not complex with the dendrimer-interior functional groups; for example, Cu 2+ does not complex with functional groups inside a poly(propylene) dendrimer. Importantly, this method eliminates the size limitation imposed on dendrimer-encapsulated metal NPs by the finite number of ligand sites within dendrimers. An important publication by Amis and co-workers [237] described the use of different generations (G2-G10) of the PAMAM dendrimer in a synthesis of dendrimerencapsulated gold NPs. First, they found that the gold/dendrimer ratio needed to be adjusted because relatively high concentrations of dendrimer resulted in precipitation; no explanation was given. Perhaps, a relatively large amount of hydroxyl-termination dendrimers causes a catalytic reduction of gold ions by hydroxyl before they have a chance to get inside the dendrimer and complex with the interior tertiary amines. In addition, PAMAM dendrimers are commonly stored in methanol solution, which also can reduce gold ions in the solution. Here, gold NPs aggregate because they are not protected. Second, they observed that the size distribution was influenced by the dendrimer concentration used. For example, when the mass fraction of the dendrimer was 1%, a stable colloidal solution was formed but the resulting gold colloids were relatively polydisperse. Whereas, if the dendrimer mass fraction was 0.12% or lower, the reduction resulted in uniform gold colloids. This observation supports the hypothesis about the pre-reduction of gold ions. Next, with evidence from TEM, SAXS, and SANS scattering curves, the Amis group concluded that: i) the gold ions from the load of one dendrimer form one particle, ii) the gold particle is offset from the center of the dendrimer, and iii) with the same dendrimer, the lower the ratio between gold ions to dendrimer end groups, the smaller the gold NPs. The first two conclusions incidentally supported the depicted scheme by the Crooks group, as described above. According to Amis and co-workers [237], the influence of the dendrimer generation on the AuNPs size is not simple. They observed that under the same reduction condition, for G2, the resulting NPs were larger than those obtained with G4-G5 PAMAM (4 nm compared to 2 nm diameter). For G6-G9 PAMAM, the AuNP diameter increased with increasing generation number. These results were contrary to those 105

120 obtained by the Crooks group for the case of dendrimer-encapsulated PdNPs [238]. The Crooks reported that a Gn-OH PAMAM dendrimer/pdcl - 4 ratio of 1:40 yielded PdNPs having a diameter of 1.7 ± 0.2 nm regardless of the dendrimer generation (n = 4-8); and the authors suggested the reason was that the encapsulated NPs may have complex shapes. The Amis group s study [237] found that for G10 PAMAM a multiple number of smaller AuNPs (about four 3 nm-au-particles) per dendrimer were formed. These results were attributed to the structural nature of different generations. The G2 PAMAM has a starlike structure with 16 end groups; G3-G5 PAMAMs have both starlike and spherical structures; and higher generation dendrimers possess a spherical structure. A schematic of multiple dendrimers attached to one AuNP was proposed to explain the unsystematically large size obtained from the G2 dendrimer [237, 239, 240]. For low-generation dendrimers, a single molecule is too small to stabilize the surface of one AuNP. Additionally, very small metal NPs are less stable, so they tend to aggregate, forming a bigger NP with more than one dendrimers attached. This stabilization can be regarded as an analogue to the classical stabilization mechanism with low molar mass molecules such as citric acid. A partial aggregation occurred to NPs synthesized with G4, G5 and even G6 PAMAM dendrimers. The Amis group s data [237] showed that for G7-G9 PAMAM, individual dendrimer molecules contained single gold particles located offset from the center of the dendrimer. The authors explained that the structure of these dendrimers provided sufficient polymer to stabilize the surface of the forming gold particle, along with enough flexibility for the growth of only one particle from all of the gold ions loaded in the dendrimer. The higher the generation, the more interior amine sites, which results in more gold ions loaded, thereby forming larger AuNPs. For G10 PAMAM, the number of the voids increases but the volume of each void decreases compared to that of the lower generations. In addition, the flexibility of the dendrimer chains decreases so that the growth of one NP is prevented. However, the internal surface was sufficient to stabilize multiple small particles inside the voids [237]. 106

121 5.2.5 Other ligands as protecting agents Sulfur-containing ligands Besides thiols, some other sulfur-containg ligands have been also used as stabilizers though not broadly. Examples include disulfides [ ], which do not form as strong a bond with AuNPs as do thiols; di- and trithiols, and resorcinarene tetrathiols [ ]; xanthates [248]; thioethers [249]; and polythioethers [250]. Phosphines, amines, acetone and iodine ligands In 1981, Schmid [251] reported a synthetic method for AuNPs, Au 55 [P(C 6 H 5 ) 3 ] 12 Cl 6, by using tetrachloroaurate and tetraoctylammonium bromide in a mixture of water and toluene, to which P(C 6 H 5 ) 3 was added, followed by the addition of NaBH 4. Indeed, Brust et al. [201] based on this process, developing the two-phase method to prepare alkanethiol-stabilized AuNPs as presented in part (a). There have been several publications [ ] on using complexes of amines and gold ions to synthesize amine-stabilized AuNPs. It should be noted that amines form week covalent bonds with bulk gold surfaces; they do not form stable, close-packed, ordered monolayers on the Au 0 (111) surface [255]. In addition, although alcohols exhibit no appreciable interaction with the Au 0 surface, amine-terminated n-alkanes do not spontaneously adsorb onto Au 0 surfaces from ethanol and other high-polarity solvents, because the Au 0 /amine interaction is insufficient to compete with the solvent interaction [256]. However, alkylamine-stabilized AuNPs [252] were nearly as stable as their thiolstabilized counterparts. The physical characterization indicated that the stability of these NPs is predominately kinetic, rather than thermodynamic. Gold NPs also could be obtained by using acetone [257] or iodine [258] as reducing and stabilizing agents. DNA Thiol- and phosphine-modified oligonucleotides have been used to stabilize AuNPs [195]. The groups of Mirkin [ ] and of Alivisatos [262] have pioneered strategies in this area. 107

122 5.2.6 Surfactant-stabilized AuNPs As described in Chapter 2, surfactants provide various assemblies at the nanoscale such as micelles in aqueous solutions, reverse micelles (water-in-oil and oil-in-water microemulsions), vesicles, and liquid crystals. Therefore, NPs can be synthesized, sized, and shaped by using surfactants as templates in the presence or in the absence of thiol ligands [ ]. Water-in-oil microemulsions are particularly attractive reaction media for preparing metal NPs. These microemulsions consist of nanosized water droplets that are dispersed in a continuous oil medium and stabilized by surfactant molecules accumulated at the oil/water surface. The highly dispersed water pools are ideal nanostructured reaction media for generating ultrafine, monodisperse NPs. The syntheses involve a two-phase system. AuNPs were produced by the reduction of the salts, in the microemulsions, by NaBH 4 or hydrazine. AuNPs with 4-nm diameters can be made by using this method Polymer-stabilized AuNPs There are a number of methods for preparing polymer-stabilized AuNPs that have been reported. The following are two main approaches. The first one is in-situ synthesis of the nanoparticles in the polymer matrix either by reduction (NaBH 4 is most often used) of the metal salts dissolved in that matrix [268] or by evaporation of the metals on a heated polymer surface [269]. Second method involves polymerization of the matrix around the NPs [270]. Blending of pre-made AuNPs with pre-synthesized polystyrene polymer bound to a thiol group was also reported [271]. Some common polymers used for stabilization of AuNPs include (but are not limited to) poly(vinyl propylene), poly(ethylene glycol), poly(vinyl alcohol), poly(acrylic acid), and block copolymer [195] Metal-oxide supported AuNPs The preparation of AuNPs on a metal oxide can be achieved by one of the following approaches [ ], namely, deposition-precipitation, coprecipitation, cosputtering, gas-phase grafting, liquid-phase grafting, and the use of organic-ligand- or polymer-stabilized AuNPs. The first approach is the easiest to handle and is commonly used for producing commercial Au catalysts. Traditionally [275], a metal-oxide support is 108

123 immerged in an aqueous solution of HAuCl 4 followed by the evaporation of water to disperse HAuCl 4 crystallites over the support surfaces. Then, the dried precursor is calcinated in the air, usually above 473 o K. This method produces large AuNPs (> 30 nm in diameter) because interactions of the HAuCl 4 crystallites and the metal oxide are weak and the chloride remaining in the support surfaces markedly promotes the coagulation. Latter, the method was modified [278]. The HAuCl 4 solution was converted to Au(OH) 3 with NaOH so that the excess reagent (sodium and chloride) can be washed from the surface before drying. The metal oxides that are usually used as supports for AuNP, are TiO 2, MgO, ZrO 2, Co 3 O 4, α-fe 2 O 3 and NiO. 5.3 Controlling size and size distribution of NPs When the size of materials lies between that of a molecule and of a bulk substance, their electronic, optical, catalytic, magnetic and thermal properties are sensitively dependent on the particle shape and size (will be discussed in part 5.5). Therefore, the ability to synthesize NPs with well-controlled shape and size provides significant opportunities to generate unique-property materials. There have been a numerous reports of methods for controlling the size and size distribution of NPs. Presenting all of these methods is beyond the scope of this dissertation. Therefore, only two approaches, using thiols and using dendrimers, which are involved in our work, will be described. In the case of using thiols, monodisperse AuNPs were obtained because NPs were stabilized by long-chain thiols in the same manner as the case of water-in-oil microemulsions [281]. Controlling the AuNP size via the thiol-to-gold ratio was first reported by Leff et al. [281]. The synthesis was based on the Brust method [201], which was in two phases, water and toluene. NaBH 4 was used as reducing agent, and dodecanethiol was used as a protecting agent. The particle size over a range of diameters - from 1.5 to 20.0 nm was controlled at room temperature by varying the RSH:AuCl 4 mole ratio from 1:1 to 1:6 while the other conditions, including the ratios of N(C 8 H 17 ) 4 Br/AuCl - 4 and NaBH 4 /AuCl 4, remained constant. The trend was that the size of the resulting AuNPs decreased with increasing thiol:gold ratio. For example, the particle sizes of AuNPs from the thiol:gold ratios, 1:3.0 and 1:3.5, were 2.95 and 6.81 nm, 109

124 respectively. Later, Murray et al. [282] extended the thiol:gold ratios to 1:12-4:1, and they also manipulated the temperature and the rate of reductant addition (deliver rate). The temperatures were varied from -70 to 90 o C, and times of the addition of NaBH 4 were 10 s, 2 min, and 5 min. The diameter of AuNPs was controlled in a range from 1.5 to 5.2 nm at room temperature. Generally, the trend of the thiol:gold ratio effect was similar to that seen in the Leff et al. study [281]. Reaction at lower temperatures resulted in smaller NP sizes. Fast delivery tends to generate bigger NPs compared to slow delivery; this effect is more pronounced at low temperatures (< 0 o C) than at high temperature (> 20 o C). When the ratio was high ( 2), at room temperature the delivery rate did not influence the size of NPs. Briefly, one can manipulate the three condition keys to obtain the desired AuNP size. For example, a very small AuNP size ( 0.76 nm) would be produced when the ratio of thiol:gold is 2:1, the deliver rate is 10 s, and the reaction is carried out at -78 o C. A large size of the AuNPs ( 2.6 nm) will be obtained if the conditions are as follow: the ratio of thiol:gold, 1:12; the deliver rate, 10 s; and the temperature, 90 o C. The conditions can be tuned between these two extreme conditions to get intermediate sizes. Dendrimers are monodisperse macromolecules built with a high level of synthetic control over structure [234]. Each dendrimer contains a controlled number of interior binding sites (the number of which is generation-dependent), which are homogeneously dispersed within the dendrimer. As a result, a controlled number of metal ions will reside in each dendrimer, and, therefore, the dendrimer-nps generated from reduction should be monodisperse. In addition, metal ions are organized according to the position of the binding sites of the dendrimer. Subsequently synthesized metal particles reflect the shape of the dendrimer. Size and shape control of NPs can be achieved by using dendrimers of different sizes, shapes, and architectures. These may be different dendrimers of different compositions (core and repeat unit), different generations of dendrimers, or combinations of the two. The influence of using different dendrimer generations on the NP size was discussed in part The size of NPs can be also controlled via the mole ratio of dendrimer surface groups to the metal salt. With the same dendrimer, the size of nanoparticles decreased with an increase in the mole ratio of the surface groups and the metal ions [224]. 110

125 5.4 Structure There are two very distinct regions of size of NPs [283]. The first one is when the particles are smaller than ~2 nm, where quantum size effects are present. The large fraction of atoms (about 50%) that are on the surface becomes a dominant factor. In this regime the dominant shapes deviate from the regular face-center-cubic (fcc) structure. For the case of gold, all the forms of the truncated decahedron and the icosahedron become the most favorable shape. The truncation is the most common mechanism chosen by nature to reduce the total energy of the particles. The formation of extra facets induces the reduction in the contributions to the energy coming from the surface area and from the radius of curvature of the particles. The second regime starts when the particle size grows larger, where they start to produce internal stress. This can be considered as a slow transition to the bulk state. This regime is dominated by a mechanism to relieve the internal stress. Dislocations, disclinations, and linear defects dominate the structure of the particles and their properties. Yacaman et al. [283] found that dodecanethiol-stabilized AuNPs rotate and change in external shape, but the structure remains fcc, which is the same as that of bare AuNPs. In this regime, most of the shapes correspond to the cubotahedron. The lattice spacing in the NPs is slightly smaller than that in the bulks, and the density of fcc structures depend on the growth conditions. 5.5 Size-dependent properties Physical properties of NPs, which are intermediate between the size of small molecules and that of the bulk metals, are neither those of the bulk metal nor those of the molecular compounds [284, 285]; they strongly depend on the particle size. Gold is an outstanding example. When the size of the particles goes down to the wavelength of the electrons, they behave electronically as zero-dimensional quantum dots [195]. It means that the laws of classical physics, valuable for bulk materials have to be substituted by quantum mechanical rules. Specific heat, susceptibility, conductivity and other fundamental characteristics of a metal are going to be lost, at least at low temperatures. This section will review some physical properties, which are size-dependent, of AuNPs. 111

126 5.5.1 Electronic properties Generally, there are three typical situations for electronic density of states [284]: 1) the bulk state with its electronic band structure of d and s electrons; 2) the characteristic situation of discrete energy levels in a molecule; and 3) the transition state between these two, which is typical for nano-sized cluster. It is suggested that the transition from bulk to molecule and vice versa is not a well defined situation, but develops continuously. The difference of the electronic structure of NPs from the corresponding bulk is enhanced with decreasing size of particles. The method applied and the temperature of investigation affects the definition of the situation. The most sensitive and valuable technique used to study the electronic situation in a single, nanosized particle is tunneling spectroscopy (STS), resulting in current (I)-voltage (V) characteristics. For example, the I-V curve of 17 nm Pd particles, at 293 o K, is linear, as is the case of the bulk. However, at 4.2 o K there is so-called Coulomb gap observed, indicating a situation where the thermal energy, k B T, of the electrons is small compared to the electrostatistic energy, e 2 /2C (C is capacitance of the combination cluster/tip). This situation should be also realized at higher temperature when the particles become small enough. This is the case for ligand-protected Au 55 clusters with a metal core diameter of only 1.4 nm. When the temperature is lowered to 90 o K, a series of Coulomb steps (staircase) was observed in the I-V curve. Single electron transitions (SET) between neighboring clusters can be initiated if they are excited. When the de Broglie wavelength is the same as or larger than the diameter of the cluster nucleus, in the excited state the electrons become able to tunnel through non-conducting medium between them [284] The surface plasmon resonance (SPR) One interesting and important characteristic of AuNPs is that their color in a solution changes from light red via purple-red to bluish-red, depending on their particle size [285]. This effect is induced by the surface plasmon resonance (SPR), which is initiated by the interaction of the electric field of visible light with the confined electron gas at the surface of NPs. The SPR of AuNPs has three main characteristics [ ]: The wavelength is in the region of nm, depending on the size and shape of the particle and the dielectric properties of the surrounding medium. 112

127 SPR sharply decreases with decreasing core size diameters in the rage nm, due to the quantum size effects. SPR also has a slight blue shift. The decrease of intensity of the SPR as particle size decreases is accompanied by broadening of the plasmon bandwidth. Steplike spectral structures indicate transitions to the discrete unoccupied levels of the conduction band with monodispersed AuNPs having core diameters between 1.1 and 1.9 nm. Therefore, the SPR disappears for AuNPs with core diameters less than 2 nm, as well as for bulk gold. For AuNPs in aqueous media having an average diameter of 9, 15, 22, 49, and 99 nm, the SPR maximum (λ max ) was observed at 517, 520, 521, 533, and 575 nm, respectively. It should be noted that the SPR maxima and the bandwidth are also influenced by the particle shape, media dielectric constant, and temperature [ ]. The refractive index, which is directly related to its dielectric constant, of the solvent induced a shift of the SPR from the value predicted by Mie theory. For example, a solution of dodecanethiol-stabilized AuNPs with 5.2 nm average diameter has an 8-nm shift in SPR as the solvent refractive index is varied from n 20 d = 1.33 to 1.55 [289]. In addition, the ligand shell alters the refractive index, subsequently causing either red or blue shift. This shift is especially significant with thiolate ligands because of the strong ligand field interacting with the surface electron cloud [289]. Murray and co-workers [282] showed that the optical dielectric of the alkanethiol ligand protecting the AuNPs influenced the SPR more strongly than did the bulk solvent. The SPR of those AuNPs remained almost unchanged in different solvents. Therefore, the spectroscopic data obtained often deviate from the prediction of Mie theory that deals with naked NPs. The agreement with Mie theory is obtained only when the shift induced by the ligand shell is taken into account. Very recently, Xu et al. [292] used Ru(bpy) 2+ 3 as a probe to study the dependence of SPR of AuNPs on the size of the NPs, the temperature and the surface environment. Ru(bpy) 2+ 3 was directly adsorbed on the Au surface by the hydrophobic interaction of tris-bipyridine with surface Au atoms. They found that, the rate of color change of AuNPs was extremely sensitive to the amount of Ru(bpy) 2+ 3 ; the minimum concentration 113

128 of Ru(bpy) 2+ 3 needed to change the color of AuNPs was strongly dependent on the NP size and the temperature Nonlinear optics (NLO) An interesting emerging application of NLO is photonics, the photon-based analog of electronics, which utilizes nonlinear phenomena (e.g., sum- and differencefrequency generation and frequency doubling) for predictable modification, translation, and switching of optical signals [293]. AuNPs, with their intense SPR, might prove particularly effective as NLO chromophores. In fact, AuNPs have a large third-order nonlinear susceptibility and a near-resonance nonlinear response that is fast, for example on the 50-ps time scale [294]. Therefore, AuNP-embedded materials, such as glass, polymer and mesoporous silica, are becoming potential alternatives to the commercially available (but expensive) inorganic crystals, such as LiNbO 3, KH 2 PO 4 and BaB 2 O 4. For practical application, it is desirable to have a low absorption coefficient and a large thirdorder optical susceptibility. With silica glass, in which AuNPs were synthesized by ion implantation, the third-order nonlinear optical susceptibility of this material was found to be proportional to the fourth power of the radius of the NPs (or the fourth power of the absorbance coefficient at the peak of the SPR) when the total volume of the AuNPs was constant; and inversely proportional to the third power of the total volume of AuNPs when the absorption coefficient of the SPR was constant [295] Melting point When the particle size of AuNPs decrease to a certain point the melting point starts dropping below that of bulk gold [296]. This trend is also true for other metal NPs. As can be seen from Fig. 25, the melting point of AuNPs, with a radius about 2 nm, is around 500 o C; while that value for bulk material is 1064 o C. It is attributed to the increasing number of the surface atoms with decreasing particle size. The surface atoms have lower coordination numbers than the inner atoms and therefore become mobile more easily [285]. 114

129 Melting point ( o C) Particle radius (nm) Fig. 25 Relationship between particle size and melting point of gold nanoparticles (data based on ref. 285) 115

130 5.5.5 Catalysis For a long time, gold has been known as an essentially useless material for catalysis. The first document about catalysis by gold was in 1906, in which Bone and Wheeler showed that gold foil catalyzed the combustion of dihydrogen and dioxygen to give water [196]. In 1989, Haruta [297] discovered that some composite oxides of AuNPs exhibited surprisingly high catalytic activity for the oxidation of CO even at 200 o K. Since then, the catalytic activity of AuNPs has been studied extensively. In general, there are three groups of catalytic AuNPs: 1) AuNPs supported on oxides, 2) alkanethiolatestabilized AuNPs, and 3) dendrimer-encapsulated AuNPs. a) AuNPs supported on oxides The catalytic activity of oxide-supported AuNPs, especially for CO oxidation and propylene epoxydation, has been investigated extensively by Haruta and co-workers. Haruta and co-workers and many other authors showed that when AuNPs are dispersed across the surface of certain oxides, they become very active, even under ambient conditions, for various reactions [ ]. There are three main factors that influence the catalytic activity and selectivity of AuNPs: i) type of the support material; ii) interactions between AuNPs and the support; and iii) size of the AuNPs. The influence of these factors is exhibited very clearly in CO oxidation and propylene epoxidation reactions. The selection of suitable metal oxides as supports depends on the reaction that AuNPs catalyze. For CO oxidation, except for strong acidic materials such as SiO 2 - Al 2 O 3, many oxides can be used as supports. However, semiconductive metal oxides, such as TiO 2, Fe 2 O 3, Co 3 O 4 and NiO, provide more stable Au catalysts than insulating metal oxides, like Al 2 O 3 and SiO 2 [275, 279]. With alkaline earth metal hydroxides, such as Ba(OH) 2 and Mg(OH) 2, AuNPs exhibit a very high activity at even low temperature (as low as 196 o K). However, the Au catalysts are stable for only few months [298]. For the combustion of hydrocarbons, Co 3 O 4 is the most active metal-oxide support for complete oxidation [299]. Owing to the high affinities to nitrogen, ferric oxide and nickel ferrites lead to the highest activity of AuNPs for reactions of nitrogen-containing compounds [275]. For the selective oxidation of hydrocarbons in the presence of O 2 and 116

131 H 2, only TiO 2 and Ti-silicates exhibit as effectiveness supports, and only anatase makes Au selective [ ] for the CO oxidation reaction. For oxidations of alcohols, ZnO works the best as a supports [303]. Catalytic activity of AuNPs is influenced by the interactions between the particles and the supporting material. Haruta and co-workers [304] showed that hemispherical AuNPs with their flat planes, which were strongly attached to the TiO 2, had a much higher catalytic activity for the CO oxidation than spherical particles, which are simply loaded on the TiO 2. A similar trend was observed in the epoxidation of propylene in the presence of O 2 and H 2 [305]. Hemispherical AuNPs attached to the TiO 2 support produced propylene oxide with 100% selectivity at a low temperature, 323 o K; whereas, spherical AuNP needed a higher temperature, 355 o K, for the reaction to occur. Tsubota et al. [306] also observed that the strong contact between Au and TiO 2, generated by calcination at high temperature, led to a higher catalytic activity, though the size of AuNPs were larger. However, Mohr et al. [274] showed that, the desired activation of the C=O group on the selective hydrogenation of acrolein was not related to the gold-support interface. Additionally, in the case of Pt/TiO 2, the interaction between PtNPs and the support did not affect the Pt catalysis on the oxidation of CO [304]. The contrast between the two catalysts and the two types of reactions suggests that the active sites responsible for the catalytic activity of gold are located at either the accessible periphery of the metal particles, where the supporting oxide may also contribute directly, or on the surface of the gold particles, whose properties not are altered by the interaction with the support. Depending on the reaction type, one of these situations will be superior. While bulk Au is an extremely poor catalyst for CO oxidation, the phenomenal oxidation activity of supported Au particles with average size below 5 nm clearly illustrates that the particle size plays a crucial role in determining the catalytic activity. Choudhary and Goodman [279] showed that TiO 2 -supported AuNPs with particle size ca. 3.2 nm were much more active for the CO oxidation compared to the particles with a size of about 6 nm or of 2 nm. Mavrikakis et al. [307] observed an increase in the normalized rate with a decrease in the mean diameter of AuNPs by a factor of 2/3. This increase can be explained if the active sites are edge, corner or step sites, whose fractions increase with a decrease in the size of AuNPs. Investigating the active sites of ZnO-supported 117

132 AuNP on the catalyzed C=O hydrogenation, Mohr et al. [274] concluded that the hydrogenation of C=O group, producing allyl alcohol, worked best on the edges of the cuboctahedral AuNPs. Haruta et al. [305] noticed that the particle size of the AuNPs supported on TiO 2 influenced the product yield of the reaction of propylene with O 2 and H 2. When the AuNP size was in the diameter range nm, the main product was propylene oxide, and propylene was formed only in a presence of O 2. AuNPs with diameters from nm produced almost 100% propylene regardless the presence or absence of O 2. Additionally, in hydrogen oxidation and the hydrogenation of alkenes and alkylnes, the product selectivity and the rate of almost independent of the type of support and the size of AuNPs as long as they are in the range of 2-10 nm. This fact was attributed to the changes in electronic structure of the small-size particles. For spherical particles of noble metals, 2 nm in diameter is quantum size. For such sizes, the fraction of atoms exposed to the surface exceeds 50%. The electronic structure of such particles deviates from that of the bulk metal, even when they are unsupported [275]. Häkkinen et al. [276] found that the smallest MgO-supported AuNPs that catalyzed CO oxidation was Au 8 ; Au 4 was catalytically inert, and Au 9 was less catalytic than Au 8. It was explained by a combination of two key factors: 1) dynamic structural fluxionality, which is the ability of a given isomer to adapt its structure so that the reactant (oxygen in this case) can be adsorbed on the AuNP with the most favorable freeenergy path and 2) size-dependent activation of the reactant (oxygen). In addition, the metal-oxide support, and in particular surface oxygen vacancy sites like a MgO(100) surface, were found to play a dominant role in anchoring the metal NPs and activating the NPs by (partial) charge-transfer. Before closing this section, it should be noted that the melting point of Au is much lower than that of Pd and Pt; it can be as low as 500 o K when the particle diameter is 2 nm or less due to the quantum-size effect (Fig. 25). Therefore, the upper limit of reaction temperatures for AuNP catalysis is important to consider. 118

133 b) Alkanethiolate-stabilized AuNPs According to Zhong and Maye [277], in comparison with the catalytic activity of the oxide-supported AuNPs and the dendrimer-encapsulated AuNPs, alkanethiolatestabilized AuNPs have many intriguing parallels. A major distinction lies in the fact that the protecting monolayer imparts the NPs with a shell reactivity and processibility dictated by the functional groups in a three-dimensional framework. The discussion about using AuNPs as catalysts will be limited to electrocatalysis. The use of AuNP thiolates as electrocatalysts mostly involves two main steps. The first step is imparting to the protecting monolayer of the AuNPs the desired functional group(s) by a ligand exchange process as described in part The second step involves the immobilization of the modified AuNPs on an electrode surface by a single-layer or layer-by-layer assembly. The process of the assembly, the nature of electronic conductivity, the role of interfacial reactivity, and the influence of the nanoporous properties of the nanostructure films will be presented in detail in Chapter 6. c) Dendrimer-encapsulated NPs The catalytic activity of dendrimer-encapsulated NPs has been studied extensively by the Crooks group. A variety of catalytic reaction involving dendrimer-nps has been reported; these include, but are not limited to, hydrogenations, hydroformylation, olefin metathesis, Heck reactions, Suzuki coupling alkylation, and oxidation [217]. The main advantage of using dendrimer-nps in catalysis is the ease of controlling the structure, size, and location of catalytically active sites. The NPs can be located within the dendrimer or on the periphery of the dendrimer, as presented in part The discussion here will be limited to the former case and extended to PtNPs and PdNPs. The advantages of using NPs encapsulated inside the dendrimer over those that are stabilized on the dendrimer periphery are [217]: 1) the NPs are stabilized by encapsulation within the dendrimer; therefore, they do not agglomerate during catalytic reactions; 2) the NPs are retained within the dendrimer primarily by steric effects; therefore, a substantial fraction of their surface is unpassivated and available to participate in catalytic reactions; 3) the dendrimer branches can be used as selective gates to control the access of small molecules to the encapsulated NPs; 4) the dendrimer 119

134 periphery can be tailored to control the solubility of the composite and used to facilitate linking to a surface or other polymers. Following are some examples of catalytic dendrimer-encapsulated NPs. It was shown that G4-OH(Pd 40 ) efficiently catalyzed hydrogenation of both the linear and branches alkenes in aqueous solutions [214]. The result implies that the substrates (the alkenes and hydrogen in this case) can penetrate the dendrimer, encounter the NPs, and undergo chemical reactions inside the denrimer. Crooks and co-workers [238] developed an approach to using dendrimers as size-selective nanofilters to control the access of small molecules to the dendrimer interior. This introduces another method to solve the limitation of intrinsically nonselective catalysts. Here, different generations were used to size-select reagents, because higher generation dendrimers have more crowded peripheries and are thus less porous, which limits access of the reagents to the interior metal NPs. On the other hand, this feature of dendrimers should be taken into account when deciding the generation to be used. Metal NPs encapsulated in lower - generation dendrimers have higher catalytic efficiency due to the open architectures of the dendrimer compared to those in lower-generations, but they may be less stable. For example, in Suzuki cross-coupling reactions, G3-OH(Pd 10 ) was shown to be more efficient for coupling phenylboronic acid and iodobenzene than G4-OH(Pd 10 ), but G3- OH(Pd 10 ) was found to be less stable [308]. It was also observed that G4-NH 2 PdNPs had better hydrogenation activity for allyl alcohol in organic solvent (toluene) than in water [309]. It was attributed to the relatively hydrophylic interior of the PAMAM molecule, which may enhance partitioning of the allyl alcohol into the vicinity of the encapsulated NPs. Dendrimers are soluble in not only aqueous and non-aqueous solvent, but also in supercritical CO 2 (scco 2 ), which is an important solvent because it is easily tunable and is nontoxic. For example, PdNPs encapsulated in perfluorinated dendrimers were reported to be soluble in scco 2, and they catalyzed Heck coupling [310]. Dendrimer-encapsulated metal NPs also can be immobilized onto substrates, such as mica and highly oriented pyrolytic graphite, for heterogenerous catalysis [217]. The study on electrocatalysis of dendrimer-encapsulated NPs that were assembled on an electrode surface will be presented in Chapter

135 5.6 Characterization techniques Size and size distribution are the two crucial parameters of NPs. The most common techniques used to determine these two parameters and provide images of NPs are high-resolution transmission electron microscopy (HRTEM), scanning tunneling microscopy (STM), atomic force microscopy (AFM), small-angle X-ray scattering (SAXS), laser desorption-ionization mass spectroscopy (LDI-MS) and X-ray diffraction [195]. The AFM technique, which was used in our research to determine the size of the AuNPs assembled on substrates, will be presented in Chapter 6 accompanied by a discussion about assemblies of AuNP-containing monolayer and multilayer films. Though the light scattering technique does not give images of NPs, it is a convenient method to obtain size and size distribution of NPs. The advantages of the method lie in the fact that it is fast, readily works with solution, and does not require lengthy data interpretation. This technique was used to determine the size and size distribution of hexanethiol-stabilized AuNPs in our study. 5.7 Applications Owing to the unique properties induced from the nano-scale size, metal NPs, particularly AuNPs, provide a variety of potential applications in different fields. Some important examples are given below Catalysts The finding of catalytic activity of AuNPs supported on oxide was followed by the first commercial application in Gold deposited on Fe 2 O 3, which is supported on a zeolite wash-coated honeycomb, has been used as an odor eater in modern Japanese toilets [278]. It is reported that most pollution from U.S. automobiles is emitted in the first 5 minutes after startup. This is because Pt- or Pd- based catalysts, currently used in the exhaust cleanup, are inactive below 200 o C [272]. Small-size AuNPs supported on a suitable oxide (such as TiO 2 ), which are catalytically active to CO oxidation at room temperature, will be a potential solution to this cold-startup problem. In addition, the high activity of AuNPs in selectively oxidizing CO in the presence of excess H 2 makes them 121

136 particularly suitable for trace CO clean-up of hydrogen streams for application in PEM (proton exchange membrane) fuel cells, which are extremely sensitive to CO poisoning [279]. AuNPs supported on Fe 2 O 3 or La 2 O 3 are the most active among the noble metal catalysts for the oxidative decomposition of dioxin at temperature below 473 o K. By integrating Ir, Pt and Au catalysts supported on La 2 O 3, SnO 2 and TiO 2, respectively, Haruta and co-workers [278] achieved 95% decomposition of dioxin, even at 423 o K. Dendrimer-encapsulated metal NPs have catalytic activities for a broad range of reactions, including hydrogenations, Heck coupling, and Suzuki reactions [217]. The important advantages of these materials are: i) selectivity toward reactants, ii) soluble in a wide range of solvents, and iii) easily recovered and recycled. Dendrimer-encapsulated NPs can be also immobilized onto an electrode surface and serve as electrocatalysts, which will be discussed in the next chapter Sensors The sensitivity of the SPR of AuNPs to core size and core environment provides a good source of sensors. For example, AuNPs functionalized with 15-crown-5 ether was used to selectively recognize K + ions in a solution containing Na + and Li + [311]. The recognition was based on the color change of the AuNPs in solution. This color change was induced by the reduction in NP proximity, resulted from the selective complexation between 15-crown-5 ether and K +. The well-known-red-to-blue color change of AuNPs upon aggregation also has been applied to a sensitive colorimetric DNA sensing method. The method uses oligonucleotide-coated NPs designed to aggregate in the presence of a specific DNA base sequence [312]. Accurate sensing of specific DNA sequences has a technological potential for biosensors, rapid disease diagnosis, and gene expression. Other approaches, such as using hydrogen-bonding, π-π complexing, host-guest complexing, van der Waal assembly, electrostatic attraction, and charge-transfer complexation, to fabricate AuNPs-based sensors also have been reported [195]. Tomalia [212] suggested dendrimer nanocomposites coupled to magnetic fields for potentially directing drug delivery. For example, first, render a dendrimer magnetic by trapping iron nanoclusters in its interior. Then hook an anticancer drug like cisplatin to the surface of the dendrimer. Inject the magnetic dendrimer near a tumor and hold a 122

137 magnet over the tumor so that the dendrimer s cisplatin warhead will home in on the tumor cell instead of being circulated throughout the body, as happens with most drugs Electronic devices Metal and semiconducting NPs promise to substitute for traditional materials in the micrometer size regime in the future. New computer and laser generations will be developed. Data storage capacities of orders of magnitude better than at present are to be expected. Even three-dimensional neuronal networks, impossible to realize with traditional materials, become possible. Meanwhile, there are numerous possible applications to be realized in shorter periods of time. Examples include in the field of electroluminescence, non-linear optic, surface-enhanced spectroscopy, sensors and catalysis. For some of these applications, organized NPs are necessary; this topic will be discussed in Chapter

138 Chapter 6 Layer-by-Layer Films Containing Metal NPs 6.1 Introduction Alternate layer-by-layer (LBL) adsorption which was introduced by Detcher and coworkers [313, 314] has been developed as a powerful and convenient technique for preparation of organized molecular films. In the process of LBL assembly, a charged solid substrate is immersed in a solution containing a polyelectrolyte having an opposite charge to that of the surface so that a layer of this polyelectrolyte is adsorbed. The concentration of polyelectrolyte is selected to be sufficiently high (several mg per milliliter) so a number of ionic group remain exposed to the solution, effectively reversing the charge of the original surface [315]. After washing and drying, the substrate is immersed in the second solution, which contains another polyelectrolyte or macromolecule with a charge opposite to that of the first polyelectrolyte. A new layer is formed, but now the original surface charge is restored. Subsequent depositions result in a multilayer film that is stabilized by strong electrostatic forces. Since electrostatic interactions are a very general principle, the process is very versatile with respect to the incorporation of different charged compounds or nano-objects [316]. In addition, because the process only involves adsorption from solution, there are, in principle, no restrictions with respect to substrate size and topology; multilayers have been prepared with colloids and on objects with dimensions of several tens of centimeter [317]. Polyelectrolytes rather than small molecules are used because good adhesion of a layer to the underlying substrate or film requires a certain number of ionic bonds. Therefore, overcompensation of the surface charge by the incoming layer can occur. This is because polymers can simply bridge over underlying defects. Their conformation at the surface is mostly dependent on the selected polyelectrolytes and the adsorption conditions and much less dependent on the substrate or the substrate charge density [318]. The linear increase of film thickness with the number of deposited layers is often similar even if different substrate are used, which makes the film properties rather independent of the substrate [319]. 124

139 The field of colloid and cluster science offers a wide range of NPs with various sizes and properties. There is a need to control the organization of these nanoparticles at the smallest-possible scale. The LBL assembly has been employed to address this challenge. Here, one of the polyelectrolytes is replaced by charged NPs [320, 321], or NPs encapsulated in charged macromolecules [215]. This chapter will describe LBL assemblies of thiol-stabilized AuNPs and dendrimer-encapsulated AuNPs. Characterization of these assemblies and electrocatalysis by these LBL films will be addressed. Specific assemblies of thin films for nanoarrays will be discussed in a separate part. 6.2 Layer-film assembly and electrocatalytic activity of the film Generally, the assembling procedure of LBL films involves alternating immersion of a well-cleaned substrate into desired solutions. Each immersion is followed by rinsing and drying. In order to produce these films, initial surface modification of the substrate is usually required, and the quality of this functionalization is thus of importance in determining the quality of the resulting films. Substrates most often used include (but not are limited to) gold, silver, platinum, indium tin oxide (ITO), glassy carbon, and highly ordered pyrolytic graphite (HOPG). For surfaces not used as electrode, glass, quartz, and mica are used. Surface modification of two types of substrates, glass slides and gold surface, which were used in our research, will be described. In the case of glass slides, the use of organosilanes as surface functionalizing materials has been investigated as a precursor to the adsorption of ligand stabilized AuNPs and the build-up of NP/polymer multilayer films [322]. The purpose of surface functionalization is to produce a uniform surface with a dense positive charge to enable the efficient adsorption of negatively charged AuNPs. Evans and co-workers [322] investigated different silanizing materials, 3-aminopropyltriethoxysilane (APTES), 3-aminopropyltrimethoxysilane (APTMS), 3- aminopropyldimethylmethoxysilane (APDMS), poly(diallydimethylammonium) (PDDA) and n-octadecyltriethoxysilane (OTS) together with various methods of their deposition. The degree of NPs adsorption in the resulting films was characterized using atomic force microscopy and X-ray photoelectron spectroscopy. The surface potential of the aminosilane films was also measured to provide information regarding the surface charge 125

140 density. The results showed that films of APTES (with one-hour deposition) yielded the highest level of NPs adsorption. Surfaces of glass substrates can also be modified with 3-mercaptopropopyl trimethoxysilane, which will anchor the MPCs by binding a layer of the clusters by placeexchange of the surface-attached thiol into the MPC monolayer [323]. In the case of Au substrates, the surfaces are modified with mercaptoalkanoic acid or 4-aminothiophenol (ATH) to impart negative or positive charges to the surface via a self-assembled monolayer (SAM) of the thiol molecules [323]. The Au surface can also be modified with alkanedithiols, which will undergo place-exchange reaction with thiolate protecting monolayer on metal NPs [324, 325] Thin film containing thiol monolayer protected NPs Generally, there are three routines to assemble LBL films containing thiol MPCs: - Using dithiol to link MPCs, accompanied with place-exchange process. As a result, a network of MPCs is produced. - Using metal ions as crosslinking agents via coordination between the metal ions and the functional groups (e.g. carboxylate) on MPCs - Using a spacer between layers of MPCs; spacers can be charged species or functional polymers and dendrimers Although the first two techniques are not self assembly, they are described to compare with the third technique, which is related to our research. a) Dithiols as crosslinking agents Zhong and co-workers [277, 326, 327] constructed a 3D-network thin film of AuNPs and Au-Pt alloy NPs on electrode surfaces for CO and methanol electrooxidation. The nanostructure assemblies were prepared from alkanethiol-stabilized NPs via a onestep, exchange-crosslinking-precipitation route. One example involved the exchange of ω-functionalized alkyl thiols (e.g., 1,9-nonanedithiol), with alkanethiolates (e.g., decanethiol) capped on AuNPs. The procedure was carried out by simply immersing a substrate, such as a gold or a glassy carbon electrode, in a solution containing 1,9- nonanedithiol and decanethiol-stabilized AuNPs. The exchange reaction was followed by 126

141 crosslinking, leading to nucleation and growth of a 3D network thin film of the AuNPs. The thickness of the film was controlled by immersion time. The modified electrode with the AuNP film was found to electrochemically catalyze the oxidation of CO and MeOH in solutions. The oxidation current was found to increase with concentration of CO or MeOH. The thiolate encapsulation did not block the catalytic site in a significant way. Depending on the activation and reaction conditions, partial or complete removal of the shell thiolates and dithiolates occurs [277]. It results in morphological changes associated with the structural evolution. The shell encapsulation may become partially open as a result of the thiolate reorganization or partial desorption. This evolution involved re-encapsulation by surface gold oxides [277, 327]. The shell structural evolution from initial organic encapsulation to partial or complete oxide encapsulation of the AuNP catalyst was responsible for the catalytic activity. The formation of the surface oxide film in the catalytic activation could also lead to the formation of porous morphology for the nanostructured catalysts [328]. Such a shell nanoporosity was operative for admitting reactants or releasing products, thereby allowing catalysis to occur at the peripheral (interfacial) between core and shell involving Au or Au δ+ species [277]. Electrochemical quartz-crystal nanobalance monitoring of the electrocatalytic process detected a mass increase during both the catalytic activation and methanol oxidation process. The quantitative correlation of the oxidation charge and the detected mass changes supported the formation of oxide species (e.g., AuO x ) in the assembled catalysts and mass fluxes of reactant, product and solvent [277, 326]. b) Metal ions as crosslinking agents In this technique, the AuMPCs have a mixed monolayer of alkanethiolate (e.g. hexanethiolate) and mercaptoalkanoic acid (e.g. mercaptoundecanoic acid, MUA) [ ]. A multilayer-assembled film was produced by linking the AuMPCs to each other and to the Au surface with carboxylate-metal ion-carboxylate bridges. Metal ions can be Zn 2+ [329, 330] or Cu 2+ [329, 331]. The assembly begins with a thiolated surface such as a glass surface that is reacted with (3-mercaptopropyl)trimethylsiloxane. In step 2, the thiolated surface is metalated with metal ions (e.g. Cu 2+ ). In step 3 this surface is exposed to a solution of the mixed-monolayer MPCs, leading to accumulation of a multilayer film 127

142 of MPCs. Steps 2 and 3 can be repeated to build the MPC film to a desired thickness. The multilayer MPC formation can be monitored by UV-visible spectroscopy and cyclic voltammetry. Monitoring by UV-visible spectroscopy accompanied with calculation based on the absorbance change showed that ca. 30 layers were formed in a single exposure to the MPC solution (step 3), depending on the length of MPC exposure [331]. Formation of a multilayer of MPCs in a single exposure implied that the initial monolayer of metal ions in the thiolated surface must become redistributed. However, it is unclear whether the metal ions are redistributed by some solvent mechanism or if the ions can move in the film [331]. The formation of the MPC multilayer film was found chemically reversible [329, 331]; the MPC multilayer film was completely dissolved by acid (acetic) and by strong Cu 2+ ligands such as thiols, both steps serving to destabilize the Cu 2+ - carboxylate coordination. The electron transport in multilayer films follows diffusion principles [330]. A change in electrode potential moves the Fermi level of the NPs to more positive values, corresponding to a positive charge state change on the MPC cores. The charge state change is initiated at the electrode/np interface and propagates outward into the bulk of the MPC film, until the entire film comes into equilibrium with the applied potential. The rate of the diffusion-like electron hopping is measured using a potential step method, chronoamperometry. The longer the linker bridge, the slower the electron-hopping rate throughout the MPC film. c) Layer-by-layer assembly The electrostatic layer-by-layer (LBL) assembly method has been applied not only for the multilayer formation of oppositely charged polyelectrolytes [ ], for which this method was originally developed, but also to construct NPs multilayers [323, ] by alternatively immersing the charge substrate surface into oppositely charged AuNPs and ionic polymeric solutions with rinsing and drying after each deposition. Combining the versatility of polymeric materials with AuNPs by LBL sequential formation of ordered nanostructures offers significant merits such as the use of environmental-friendly aqueous solutions, low cost, and high throughput. 128

143 The formation of multilayer films of organothiol-stabilized AuNPs, using LBL techniques, has been investigated. Ionic AuNPs are produced by employing placeexchange reactions as described in chapter 4 or by using the Brust method; for example, anionic AuNPs were synthesized by the Brust method, using 4-mercaptobenzoic acid (MBA) as the stabilizing agent [340]. In films of polymer/aunps multilayers, the polymers that have been most often used are poly(styrene sulfonic acid) (PSS) [341, 342], and poly(allylamine hydrochloride) (PAH), and poly(diallyldimethylammonium chloride) (PDDA) [ ]; in addition, PAMAM dendrimers have been used [345]. Murray and co-workers [323] demonstrated assemblies of multilayer films, which were composed of alternative polymers and charged AuMPCs. The AuMPCs were given a positive or negative charge via place-exchange reaction. For example, by a placeexchange reaction, MUA ligands were incorporated into the monolayer shell of hexanethiol-mpc of Au, resulting in anionic MPCs having a mixed monolayer of hexanethiolate (C 6 SH) and MUA. Cationic MPCs having a mixed monolayer of C 6 SH and ATH were obtained in a similar way. A multilayer film was generated by alternately exposing a charged substrate surface to a solution of the mixed MPCs and a solution of encounter polymer; C 6 SH-MUA MPCs/PAH and C 6 SH ATH MPCs/PSS films were formed by that approach. The factors that promoted layer-by-layer growth in these systems were primarily electrostatic interactions. Hydrogen boding was seen to play a less dominant role. The polymer solution ph played an important role in the growth of the film because it influenced the solubility of the MPCs (e.g. C 6 SH-ATH MPCs) and the protonation of the amine-containing compounds such as PAH and ATH on the MPCs. The growth of MPC layers was followed by UV-visible spectroscopy and quartz microbalance. It was found that ca. 3.7 ± 1.3 monolayers of MPC were added per dipping cycle. Stolarczyk et al. [324, 325] immobilized 4-hydroxythiophenol (4-HTP)-stabilized AuNPs on a gold electrode modified by 1,9-nonanedithiol. Here, ligand exchange between the thiolate of the protecting monolayer on the AuNPs and the dithiolate on the Au electrode surface occurred. The AuNPs immobilized on an Au electrode via dithiol bridges transferred charge to ascorbic acid (AA) [324] and to 3,4- dihydroxyphenylalanine (DOPA) [325] molecules in the solution more efficiently than 129

144 when the same 4-HTP monolayer was formed directly on the Au electrode. Presumably, the NPs attached to the monolayer on the electrode surface played the role of charge transporting unit, while 4-HTP was the catalyst for charge transfer to AA. Hydrogen bonding has been employed to assemble multilayer films of polymers and AuNPs. For example, Hao and Lian [346] alternatively deposited poly(4- vinylpyridine) (PVP) and MBA-stabilized AuNPs on a modified substrate. The multilayer buildup was monitored by UV-visible spectroscopy, which showed a linear increase of the film absorbance with the number of adsorbed Au layers. FTIR spectroscopic data verified hydrogen bonding between the pyrindine and carboxyl groups, which was believed to be the driving force for the formation of polymer/au multilayer thin films. Depending on the polyelectrolyte structure and nanoparticle morphology as well as conditions of self-assembling, the final properties of charge transport and permeability within the assembly can be varied from a film with bulk metal conductivity [347] to a film exhibiting electronic charge transport from the electrode through the film via electron hopping from nanoparticle to nanoparticle [323]. A tunable mobility of electrolyte ions moving through the film (necessary for electroneutrality) has also been described [323]. It is suggested that the film s structure is not static but somewhat fluid. When an electrode potential is applied and electronic charge is passed between the electrode and the film, electrolyte ingress or egress occurs so that local electroneurality is maintained within the film. The double layer charging of each nanoparticle, at equilibrium, is independent of the other nanoparticles. Electron transport through the film to charge the nanoparticle cores is a bimolecular process in which electrons hop between nearby nanoparticles. If the films were static and the nanoparticles lacked microscopic mobility, the intervening polyelectrolyte structure would act as a tunneling barrier that retards the rate of electron hopping and transport. There is a dual mobility of electrolyte ions and nanoparticles within the film, whose layered structure cannot be regarded as being static. However, it should be noted that different polymer/mpc film has different rate of electron transport. For example, although the MUA-MPC/PAH and ATH-MPC/PSS films were grown to contain approximately the same amounts of MPC, the rate of electron transport through 130

145 the ATH-MPC/PSS film was much more higher than that through the MUA-MPC/PAH film [323]. Evans and co-workers [340] found that multilayer films, containing MBAstabilized AuNPs and PDDA fabricated on APTES-functionalized silicon substrates, had poor conductivity. It was attributed to the insufficient thickness of each bilayer for complete monolayer growth. It has been also demonstrated that electronic charge transport within LBL selfassembled multilayers of polyelectrolytes can occur by electron hopping between adjacent molecular redox centers that are covalently grafted on the polyelectrolyte backbone. Such systems involved redox-polyelectrolyte assemblies such as poly(butanylviologen)/pss [348], poly(allylamine)ferrocene or osmium complexderivatized poly(allylamine)/glucose oxidase [349, 350], and viologen-functionalized poly(vinylpyridinium)/nitrate reductase [351] Thin films containing dendrimer-encapsulated metal NPs Dendrimers, within which NPs are encapsulated, provide reactive surface groups for linking the nanoparticles to surfaces and other polymers [222]. The stability of the resulting monolayer film depends on the nature of the interactions between the dendrimer surface groups with the substrate surface or with the intermediate layer. For example, monolayers of Pt dendrimer encapsulated nanoparticles (DENPs) terminated with hydroxyl functional groups could be physisorbed to Au surfaces and used for the electrocatalytic reduction of O 2 [215]. However, the peripheral hydroxyl groups have low reactivity, so the stability of the dendrimer monolayer is poor under electrochemical conditions. Most of the DENP monolayer desorbed after only one voltammetric scan between -0.2 and +0.6 V. In contrast, DENPs, having primary amine groups [222] or having both quaternary ammonium groups and primary amines on their periphery [352], can be covalently linked to a monolithic Au surface using an intermediate MUA SAM adhension layer. This type of film was stable under electrochemical scans (up to 20 scans between ±0.6 V). The DENPs can also be attached to the surface via electrostatic interactions between the acid terminal groups of MUA and the peripheral primary amines of the dendrimer; physisorption (van der Waals interactions) may also play a role. Crooks et al. 131

146 [222] showed that the monolayer of Pd 30 NPs encapsulated within G4-NH 2 PAMAM dendrimer provided a significantly more robust film and higher coverage than those encapsulated within G4-OH PAMAM dendrimer; both two types of monolayer films were prepared on either a naked Au substrate or MUA SAM. Especially, the PdDENP monolayers with the covalent bonds with the MUA SAM remained robust even under aggressive conditions intended to desorb the dendrimer. G6-OH(Pt 55 ) and G6-OH(Pt 55 ) PAMAM which were immobilized on Au electrodes as single monolayers, showed electrocatalysis for O 2 reduction in acid electrolytes [215, 216]. The results demonstrate that the surface of the DENPs are accessible to reactants in the solution and can exchange electrons with the underlying electrode surface without the need for mediators. Sun and Crooks [353] developed another method for immobilization of DENPs in a flat substrate. Dendrimers are first chemisorbed onto the substrate. Then DENPs are prepared directly on the surface by loading metal ions into the dendrimer, followed by chemical reduction. This method possesses two potential advantages: i) the metal particles are inherently small and monodispersed because the size of the particles is controlled by the number of metal precursor ions coordinated to one dendrimer molecular, and ii) the location of immobilization can be controlled through specific chemical interactions between dendrimer and functional groups on the solid support. Tomalia and co-workers [354] demonstrated electrostatic LBL assembly of the DEAuNPs using PSS as the oppositely charged polyelectrolyte. UV-visible absorption spectra from the consecutive multilayers indicated that each bilayer growth was regular, even though a 20 nm absorption bathochromic shift took place in the film. TEM of PSS/DEAuNP film showed that AuNPs appeared as aggregated within a DEAuNP monolayer. 6.3 AuNPs as electrode arrays Highly ordered nanostructures of metal NPs are expected to exhibit high catalytic activity. However, it is challenging to produce such systems with highly dispersed metal nanoparticles with controllable interparticle spaces and distribution. This part reviews methods applied to produce highly ordered metal NPs systems and to test their 132

147 electrocatalytic activity. Among the potential candidates, a method using block copolymer micelle is one. Amphiphilic block copolymers can self-assemble in selected solvents to produce core-shell micelles in which the cations or anions preferentially segregate in the core or shell and react to create inorganic NPs directly, depending on the specific affinity between the polymer block and the ions [355]. These inorganic/micelle nanohybrids can self-assemble to form highly ordered arrays on a substrate and, after removing the polymers, ordered NPs are produced. The following is an example. McFarland and co-workers [356, 357] used the diblock copolymer, poly(styrene)-blockpoly(2-vinylpyridine), which formed spherical micelles in toluene (the polar poly-2- vinylpyidine heads constitute the center and the nonpolar polystyrene tails extend - outward), to encapsulate the gold precursor via complexing AuCl 4 with the pyridine groups. The micelles containing the gold precursors were dip-coated onto ITO-coated glass substrates, followed by reduction of the gold precursors and removal of the polymer by oxygen plasma. AuNPs consisting of an Au 0 core, and an Au 2 O 3 shell was formed. The AuNPs showed highly efficient electrocatalytic oxidation of CO. For particles with diameters between approximately 1 and 6 nm, the smallest NPs studied (1.5 nm) were the most active for electrooxidation of CO and had the largest fraction of oxygen associated with gold at the surface as measured by the Au 3+ /Au 0 ratio. The oxide component of the Au 0 / Au 2 O 3 was believed to be of crucial importance for the catalytic activity. The method possessed some advantages: i) quasi-hexagonal ordering of the Au clusters giving rise to high dispersions, ii) control of the intercluster spacing by changing the length of the block copolymer tail [358], iii) control of the cluster size distribution by the concentration of gold solution and/or by selection of the appropriate length block copolymer head [ ]. However, the method is limited to solvents in which the copolymer can produce micelles suitable for an application as nanoreactors. Hepel [361] developed a method to form composite polypyrrole films containing a highly dispersed three-dimensional array of PtNPs. PtCl 2-4 anions were trapped inside the polypyrrole matrix during the electropolymerization of pyrrole. In the next step, PtCl 2-4 anions were reduced to Pt 0 NPs with an average size of 10 nm. The presence of PtNPs in composite polypyrrole films and their uniform distribution were confirmed by energy-dispersive X-ray spectroscopy and X-ray diffraction. Electrocatalytic activity 133

148 toward methanol oxidation was observed. Higher catalytic activity was found for electrodes with PtNPs dispersed in the polypyrrole matrix than electrodes with Pt 0 electrodeposited on the surface of the conductive polymer. Ohsaka and co-workers [362] assembled AuNPs arrays by immobilizing citratestabilized AuNPs on a cystamine-modified Au electrode. The system selectively catalyzed electrooxidations of dopamine and ascobate. However, no method of controlling the distribution and the interparticle space of the AuNPs array was presented. Template methods have been employed to fabricate long-range 1D chains of metal NPs. For example, Hornyak et al. [363] prepared 1D chains of AuNPs by using the pore channels of alumina membranes. The pore channels were filled by one of three methods: vacuum induction, electrophoresis, or immersion. Oku and Suganuma [364] exploited the adhesive force at the step edge of amorphous carbon thin films to form 1D self-organization of AuNPs. Berven et al. [365] used a biopolymer template, poly-llysine, between electrode pairs of a gold interdigitated array to produce an extended nanoparticle arrays. Miyake and co-workers [366] fabricated 1D chains of size-controlled AuNPs using nanoscale ridge-and-valley structured carbon. The AuNPs were predominatly immobilized in valleys and partly on ridges. Joo et al. [367] synthesized highly dispersed PtNPs with a narrow particle-size distribution around 2.5 nm using highly ordered, rigid arrays of nanoporous carbon as supports. The supporting material was produced by using ordered mesoporous silica as templates. The high dispersion of the PtNPs gave rise to promising electrocatalytic activity for oxygen reduction. Kulesza et al [368] fabricated dispersed PtNPs on a carbon working electrode by an electrochemical method. When a sufficiently negative potential was applied to a carbon working electrode for a long time (up to 6 hours), a platinum counter underwent corrosion and served as a source of platinum for subsequent electrodeposition. As a result, ultrafine, dispersed PtNPs were produce on the working electrode. These NPs were approximately spherical, had diameters of nm, and were separated by distances of less than nm. Even very small amounts of interfacial Pt coverage (on the order of a few percent) actively promoted electrocatalytic reactions, including oxygen reduction, hydrogen evolution and oxidation of As(III). 134

149 6.4 Characterization techniques Techniques used to characterize the layer films containing metal NPs in our research are described as below UV-visible spectroscopy Layer-by-layer growth on glass slide can be monitored by spectrophotometry exploiting the feature of the surface plasmon resonance band of AuNPs [323]. Importantly, when the molar absorptivity (ε) of MPCs in dilute solution and the molar of MPCs per monolayer (Γ) are determined, the number of monolayers deposited in each dipping cycle can be calculated from changes absorbance by using equation A = 10 3 εγ Quartz crystal microbalance Historically, in 1959 Sauerbrey [369] published a paper that showed that the frequency shift of a quartz crystal resonator is directly proportional to the added mass. Sauerbrey s work is generally taken as the breakthrough and the first step towards a new quantitative tool to measure very small masses i.e., the quartz crystal microbalance (QCM). The heart of the QCM is the piezoelectric AT-cut quartz crystal sandwiched between a pair of electrodes [370]. When the electrodes are connected to an oscillator and an AC voltage is applied over the electrodes the quartz crystal starts to oscillate at its resonance frequency due to the piezoelectric effect. This oscillation is generally very stable due to the high quality of the oscillation. If a rigid layer is evenly deposited on one or both of the electrodes the resonant frequency will decrease proportionally to the mass of the adsorbed layer according to the Sauerbrey equation: 2 2f 0 f = m A µρ (15) where: f = measured frequency shift, f 0 = resonant frequency of the fundamental mode of the crystal, m = mass change per unit area (g cm -2 ), A = piezo-electrically active area, 135

150 ρ = density of quartz, g cm -3, µ = shear modulus of quartz, 2.947x10 11 g cm -1 s -2. There are situations where the Sauerbrey equation does not hold, for example, when the added mass is a) not rigidly deposited on the electrode surface(s), b) slips on the surface or c) is not deposited evenly on the electrode(s). Therefore, the Sauerbrey equation is only strictly applicable to uniform, rigid, thin-film deposits. The growth of multilayer films containing AuNPs in our research was monitored by a Model 400 Electrochemical QCM (CHI400 EQCM) from CH Instruments (see Fig. 26). Instead of measuring the frequency directly, the CHI400 series uses a time-resolved mode [370]. The frequency signal of the QCM is subtracted from a standard reference frequency. The difference is then measured by the reciprocal technique. The reference crystal has an oscillation frequency of 8.000M Hz. The working crystal oscillation frequency should be 7.995M Hz M Hz. The area (A) of gold disk coated onto the crystal is cm 2. A mass change of 0.14 ng for a 0.1 Hz frequency change describes the sensitivity Scanning probe microscopy (SPM) SPM consists of a family of microscopy forms where a sharp probe is scanned across a surface, and the probe:sample interaction or interactions are monitored [371]. There are two primary forms of SPM: scanning tunneling microscopy (STM), which was developed in 1982 by Binning, Rhrer, and Weibel at IBM in Switzerland, and atomic force spectroscopy (AFM), which was developed in 1986 by Binning, Quate, and Gerber in a collaboration between IBM and Stanford University. AFM was chosen for our research because STM is typically limited to conductive and semiconducting surfaces; the AuNPs in our study were assembled on mica substrate. The AFM probes the surface of a sample with a sharp tip, a few microns long and often less than 100 Å in diameter [371]. The tip is located at the end of a cantilever that is 100 to 200 µm long. Forces between the tip and the sample surface cause the cantilever to deflect. A detector measures the cantilever deflection as the tip is scanned over the sample or the sample is scanned under the tip. The measured cantilever deflections allow 136

151 Fig. 26 A diagram of CHI400 EQCM (based on ref. 370) 137

152 a computer to generate a map of surface topography. AFM can be used to study insulators and semiconductors as well as electrical conductors. There are three modes of AFM: Contact mode, non-contact mode, and tapping mode [372]. Contact mode AFM operates by scanning a tip attached to the end of a cantilever across the sample surface while monitoring the change in cantilever deflection with a split photodiode detector. The tip contacts the surface through the adsorbed fluid layer on the sample surface. In the non-contact mode, the cantilever is oscillated at a frequency which is slightly above the cantilever s resonance frequency, typically with an amplitude of a few nanometers (<10 nm), in order to obtain an AC signal from the cantilever. The tip does not contact the sample surface but rather oscillates above the adsorbed fluid layer on the surface during scanning. Tapping mode AFM operates by scanning a tip attached to the end of an oscillating cantilever across the sample surface. The cantilever is oscillated at or near its resonance frequency with an amplitude ranging typically from 20 nm to 100 nm. The frequency of oscillation can be at or on either side of the resonance frequency. The tip lightly taps on the sample surface during scanning, contacting the surface at the bottom of its swing. Although tapping mode AFM has slightly slower scan speed than contact mode, it was employed to our research for the following advantages over contact mode and non-contact mode: Higher lateral resolution on most samples (1 nm to 5 nm) Lower forces and less damage to soft samples imaged in air Lateral forces are virtually eliminated, so there is no scraping. 138

153 Chapter 7 Measurement Platforms Fabricated by Layer-by-Layer Assembly of Crown Ether Functionalized Gold Nanoclusters 7.1 Introduction The fabrication of surface layers that have organization at the atomic or molecular level and have controlled functionalities are of interest for a wide range of applications including analytical platforms for separation science and sensing, electronics, photonics, catalysis, and corrosion inhibition. One approach to forming films for these applications is layer-by-layer electrostatic assembly of supramolecules [332, ]. In a typical procedure an electrode surface is modified with a monolayer of a charged substance. For example, we fabricated an electrocatalytically active layer-by-layer assembly on gold by first modifying the electrode with a positively charged self-assembled monolayer of 4- aminothiophenol and then alternating its immersion in an anionic supramolecule, phosphomolybdate (PMA), and a cationic supramolecular spacer, generation-4 poly(amidoamine) dendrimer [378]. In an analogous manner, quartz was modified with LBL assemblies containing either phosphotungstate or PMA; here, the initial monolayer was formed by reaction with poly(diallyldimethylammonium chloride) [379]. Feldheim et al. [380] used LBL assembly to prepare multilayers that contained citrate-protected colloidal gold particles with diameters in the 1 4 nm range. The intermediate layers comprised insulators that gave an insignificant amount of short circuits between gold layers in a 1 cm 2 array. Indeed, the general topic of electron and ion transport, including quantized charging, with assemblies that contain monolayerprotected gold nanoclusters (MPC) has been a focus of several studies, typified by citations 323, 329, A complication relative to the LBL fabrication of multilayers of conventional supramolecules is that exposure of a surface of a charged polyelectrolyte to an oppositely charged MPC solution does not necessarily result in a single monolayer. For example, Hicks et al. [323] found that these nanoclusters, when converted to mixed-ligand MPCs 139

154 using either mercaptoundecanoic acid (MUA) or mercaptophenylamine (ATH), electrostatically assembled with poly(allylamine hydrochloride), PAH, and poly(styrene sulfonate), PSS, respectively, in a manner that resulted in the deposition of about 3.7 ± 1.3 monolayers of MPC for each exposure. They compared the ability of these two assemblies to transport charge by investigating the voltammetry at electrodes thereby modified to contain similar quantities of MPC. The combination of cationic PAH and MUA/MPC provided facile charge transport, whereas the LBL assembly containing PSS and ATH/MPC was sufficiently less conductive and was of more dense structure so that the voltammetry of ferrocene was blocked. We are interested in the development of modified electrode surfaces that can selectively interact with substances ranging from metal ions to biological compounds. The present study is focused on the former. Crown ethers are functionalities that are well known to have the ability for selective complexation of metal ions. A challenge is to fabricate electrode surfaces of defined structure that contain controlled quantities of these reagents. One approach is to tag a crown ether with an alkanethiol and subsequently form a self-assembled monolayer, SAM, of the product on an substrate. Using a method reported by Flink et al. [384], Kijak and Cox we synthesized 2-[(6-mercaptohexyl)oxy]methyl-15-crown-5 and formed a SAM of this compound on Au [385]. The functionalized assembly was used to trap Pb 2+ from solution. An important finding was that the SAM was formed with a surface coverage of 0.97 ± 0.01 in s. Because of the rapid re-formation of the SAM, it was practical to release the trapped lead ion (and the entire functionalized SAM) into a flow stream by oxidizing the gold thiolate bond. A detection limit of 6 pmol Pb was thereby achieved by amperometry at a down-stream electrode. Flink et al. [384] used a similar SAM on Au to sense electrochemically inactive cations such as Na + ; here, either impedance spectroscopy or attenuation of the cyclic voltammetric current of an electroactive complex in the contacting liquid phase was used to generate the signal. Moore et al. [386] used a somewhat different approach in that electroactive sites, tetrathiafulvalene, were incorporated into the macrocyclic ligand; complexation by cations such as Li + was signaled by a shift in the TTF voltammetry. Zhang and Echegoyen assembled calixarene derivatives of a crown ether and sensed Cs + 140

155 by impedance spectroscopy in a study that demonstrated the influence of conformational change on the recognition chemistry [387]. A limitation of these approaches is that a single monolayer has a limited uptake capacity. A premise of this study is that the same chemistry that yields a functionalized SAM on Au is applicable to forming functionalized MPC, which in turn can be assembled into multilayers, thereby increasing the uptake capacity. Others have incorporated chemical trapping agents into multilayer assemblies of gold nanoparticles. Namely, Shipway et al. [388] prepared multilayers of citratestabilized gold colloid and incorporated a tetra-cationic cyclophane as an electrostatic crosslinking agent. The ability of the cyclophane to form pi-donor complexes was employed to subsequently trap p-hydroquinone. Cyclic voltammetry demonstrated that the quantity of p-hydroquinone that was trapped increased with the number of Au particle layers. The present study is not only aimed at testing the premise of trapping by functionalized multilayers of MPC but also is intended to characterize the assembly process, to establish whether the resulting films have a sufficiently open structure to permit facile ion transport therein, and to test whether the conformation and density of the crown ether groups on the CE-MPC will permit complexation of a metal ion via a sandwich mechanism. Data from our first attempt to conduct a selective solid phase extraction of Cs + ions from an aqueous solution by LBL film, containing 18-crown-6 MPCs is also included. 7.2 Experimental Chemicals and materials Unless otherwise stated, the chemicals were ACS Reagent Grade from Aldrich Chemical Company (Milwaukee, WI).The 2-(hydroxymethyl)-15-crown-5 (95%); the 2- (hydroxymethyl)-18-crown-6 (95%); 1,12-dibromododecane (98%); tetrabutylammonium fluoride, TBAF, 1.0 M solution in tetrahydrofuran, THF; hydrogen tetrachloroaurate (III) trihydrate (99.9%), 1-hexanethiol (95%), tetraoctylammonium bromide (98 %); sodium borohydride, 12% wt solution in aqueous conc. sodium hydroxide; toluene (99.5%); 141

156 poly(amidoamine), generation 4 (G4-PAMAM), 10 wt % solution in methyl alcohol; 3- aminopropyl triethoxysilane (3-APTES); 4-aminothiophenol (4-ATP) (97%); tetrabutylammonium hexafluorophosphate, Bu 4 NPF 6 (98%); ferrocene, Fc (98%); and silica gel, mesh, 60 Å, also were purchased from Aldrich. The hexamethyldisilathiane (98%) was from Sigma-Aldrich Inc. (St Louis, MO). Poly(4- styrene sulfonate), PSS, (MW ca ) was from Johnson Matthey (Ward Hill, MA). Glass slides with indium tin oxide (ITO) coated one surface were from Delta Technologies Limited (Stillwater, MN). Solutions were prepared with house-distilled water that was further purified with Barnstead NANOpure II system (Boston, MA) Apparatus The voltammetry experiments were performed with a CHI Model 750 electrochemical workstation or a CHI Model 800 electrochemical detector (CH Instruments, Austin, TX). The mass change experiments were carried out on a CH Instruments Model 400 electrochemical quartz crystal microbalance, EQCM, with an MHz reference crystal. UV-Visible spectrophotometry experiments were done on a Hewlett Packard 8453 system (Agilent Technologies, Palo Alto, CA). Size distribution of AuNPs was determined by dynamic light scattering detector (Precision Detectors, Inc., Franklin, MA) Synthetic methods 2-(12-bromododecyloxy)methyl-15-crown-5 ether (Compound A) was prepared by a modification of a procedure reported by Fox and Wooten [389]. Briefly, 2- (hydroxymethyl)-15-crown-5 (1 g, 4 mmol), 3.75 eq. 1,12-dibromododecane (5.08 g, 15 mmol), 1.1 eq. potassium hydroxide (0.246 g, 4.4 mmol), t-butyl alcohol (32.2 ml) and water (4 ml) were refluxed for 24 h. Water (50 ml) was added and the crude product was extracted with CH 2 Cl 2 (50 ml). The organic layer was washed with water 5 times (50 ml of water each time), and the solvent was removed in vacuo at temperature below 40 o C to avoid decomposition of the compound. Excess 1,12-dibromododecane was separated from the product by column chromatography on silica with hexane as the eluent, and Compound A was eluted off the column with methanol/chloroform (1:9). The 142

157 column made of silica had a length of 25 cm and a diameter of 2.5 cm. The flow rate was maintained at 2 ml/min. After the solvent was removed in vacuo, Compound A was obtained as light-yellow oil (0.45 g). The 1 H NMR and IR confirmed the presence of the compound. 2-(12-mercaptododecyloxy)methyl-15-crown-5 ether (15CE5, Compound B) was synthesized by preparing a 0.5 M solution of A in dry THF (1.654 ml) under argon [390]. The solution was cooled to -10 o C by a bath of mixture of ice and NaCl, and 1.1 equivalents of hexamethyldisilathiane and 1.2 equivalents 1.0 M TBAF in THF (10% H 2 O) were added and stirred for 30 min. The solution was brought to room temperature and stirred for an additional 24 h. Saturated ammonium chloride (10 ml) was added, and the crude product was extracted with CH 2 Cl 2 (50 ml). The organic layer was dried over Na 2 SO 4, and the solvent was removed in vacuo at temperature below 40 o C. The crude product was purified by column chromatography (SiO 2, 25 cm in length and 2.5 cm in diameter, hexane/ethyl acetate 1:1), followed by solvent removal in vacuo to yield Compound B as a yellow oil (0.209 g, 58%). The 1 H NMR (200 MHz, CDCl 3 ) was identical to that reported by Flink et al. [384]. 2-(12-mercaptododecyloxy)methyl-18-crown-6 ether (18CE6, Compound C) was synthesized from 2-(12-bromododecyloxy)methyl-18-crown-6 ether in the same routine as that for Compound B. The final products, B and C, were stored under an inert gas, argon, in sealed vials to eliminate oxidation of the thiol; the vials were purged with argon every 10 days. Those vials were sealed with a septum to facilitate a use of a syringe to transfer the compound for the next experiment. The synthesis of the crown-ether functionalized monolayer-protected gold clusters was accomplished in two steps. In this synthesis, all of the glassware was cleaned with aqua regia solution (3:1, v/v, HCl:HNO 3 ), followed by rinsed thoroughly with distilled water. Caution: Aqua regia solution reacts violently with organic material and metals, thus it must be handled with great care. First, preparation of hexanethiolate MPC with an average core radius of 1.7 nm was done by the method of Hostetler et al. [282]. Specifically, to a vigorously stirred solution of 1.5 g of tetraoctylammonium bromide in 80 ml of toluene (room temperature) was added 0.8 mmol of HAuCl 4 in 25 ml of water (it should be noted that HAuCl. 4 3H 2 O was transferred with a non-metal spatula). The 143

158 organic layer, having an orange color, was separated, and 0.8 mmol of 1-hexanethiol was added. The resulting solution was stirred for 1 h at room temperature. A freshly prepared solution of 0.38 g of NaBH 4 in 25 ml of water was added over 10 s. The color of the solution immediately changed to brown color. This solution was further stirred for 4 h at room temperature. The organic phase was collected and the solvent was removed in vacuo (the temperature was kept below 50 o C to minimize product decomposition). Methanol (100 ml) was added and the black solid was recovered by filtration and washed with methanol 5 times (30 ml of methanol each time). The solid was dried in air and then stored in a sealed vial for the next use. Subsequently, the crown-ether functionalized monolayer-protected gold clusters (18CE6-MPCs and 15CE5-MPCs) were prepared from the above MPC using a ligand place-exchange reaction in a modification of a procedure reported by Murray and coworkers [206] as shown in Fig. 27 (presented for 15CE5-MPC). A solution of 160 mg of the hexanethiolate MPC, mg of B, and 6.9 mg of MUA was prepared in 80 ml CH 2 Cl 2. The mixture was stirred at room temperature under argon for 120 h. The solvent was removed in vacuo below 40 o C, and the resulting precipitate was collected by vacuum filtration. 18CE6-MPCs were obtained in the same way Procedures Layer-by-layer (LBL) electrostatic deposition of multilayers of CE-MPCs and G4-PAMAM on ITO glass slides was done using solutions of 10 mg CE-MPC in 10 ml of dry ethanol and 1 wt % G-4 PAMAM in water. Sufficient HCl was added to each to give nominal ph 1 solutions. The ITO glass slide was first sonicated in dry ethanol for 10 min, followed by rinse with dry ethanol and drying under a stream of helium. Then the slide was functionalized with 0.06 M 3-APTES in methanol for 1 h. The use of 3-APTES and of a 1-h immersion was chosen because, as presented in Chapter 5, the film of 3- APTES yielded the highest level of NP adsorption [322]. The slide was washed with methanol and dried with a helium gas stream. During the experiment the slide must be handled with care to avoid finger prints and impurities left on the slide surface. The first layer of CE-MPCs was assembled on the ammonium-functionalized surface by contacting the ITO to the above solution for 30 min. By repeating the rinsing (dry 144

159 145 Fig. 27 Formation of 15CE5-MPC via place-exchange reaction S S S O S O O O O O O S O O O O O O S O O O O O O S O O O O O O S O O O O O S S S S O S O O O O O S COO- S COO- S COO- S COO- S -OOC S -OOC S -OOC S S S S S Au O O O O O O(CH 2 ) 12 SH HOOC(CH 2 ) 10 S S S S S S S S S S S S S S S S S S S S Au CE-MPC

160 ethanol), drying (helium stream), and immersions in the CE-MPC and the G4-PAMAM solutions, multilayers were deposited. A scheme of the stepwise electrostatic multilayer assembly is shown in Fig. 28. The LBL assemblies of PSS G-4 PAMAM were fabricated in the same manner except ca. 14 µm PSS was used instead of the CE-MPC solution. It should be pointed out that quality of the multilayer film (linear and homogenous growth) strongly depends on the consistency of the immersion time and the cleanness. The trapping of metal ions, Cs I and Pb II, on the CE-MPC assemblies was based on the projected formation of sandwich complexes with the crown ether (Fig. 29). For minimization of additional complexation with the carboxylates, the outer layer of the film was G-4 PAMAM. The coated ITO was immersed in a 1.0 mm Pb(NO 3 ) 2, 0.1 M HNO 3 solution for 10 min, washed with water, and dried with helium gas. Characterizations by QCM measurements were performed by first modifying the gold surface of the crystal with 1 mm 4-ATP in ethanol and then fabricating CE-MPC G-4 PAMAM multilayers as described above. The only difference was instead of dipping the crystal into the solutions, 1 drop of the selected solution was put onto the surface, and the crystal was kept in the chamber of saturated vapor of the selected solvent to prevent evaporation. The trapping of Cs I and Pb II was the same as described above except the sample was now a drop of solution on the surface and the vapor chamber was used to block evaporation. Electrochemical measurements were in a conventional three-electrode cell. With Pb II as the test species, the supporting electrolyte was 0.1 M tetraethylammonium chloride. With 0.5 mm ferrocene as the test species, the solution was 0.1 M Bu 4 NPF 6 in CH 2 Cl 2, and the working electrode was ITO. All potentials were recorded and reported vs. a Ag AgCl reference electrode. 7.3 Results and discussion The size distribution of the MPCs, which was measured by dynamic light scattering, is shown in Fig. 30. The average diameter was 1.87 nm, which is agreement with the literature [282]. The previous study that used a functionalized MPC in a LBL assembly employed the functional group as a crosslinking agent in the formation of the 146

161 OH OH OH OH OH OH OH OH 3-APTES O O Si NH 2 O O O Si NH 2 O O O Si NH 2 O O O Si NH 2 O (a) O O Si NH 3 + O O O Si NH 3 + O O O Si NH 3 + O O O Si NH 3 + O (b) O O Si NH 3 + O O O Si NH 3 + O O O Si NH 3 + O O O Si NH 3 + O Glass ITO [(a), (b)] n O O Si NH 3 + O O O Si NH 3 + O O O Si NH 3 + O O O Si NH 3 + O = CE-MPC = G4-PAMAM Fig. 28 Scheme of stepwise electrostatic multilayer film assembly 147

162 O O O O Pb 2+ O Pb 2+ OO O O O O O O O HOH 2 C HOH 2 C HOH 2 C Fig. 29 Complex between Pb 2+ and 15-CE-5 148

163 15 % Dia. (nm) Fig. 30 Size distribution of the MPCs (measured by a dynamic light scattering detector) 149

164 film [388]. In this case, the functional group on the MPC, namely a crown ether (CE), was neutral, so the LBL assembly required the attendant citrate in the mixed-ligand MPC to provide the necessary electrostatic charge without interfering from the tethered CE. Spectrophotometric measurements, as shown in Fig. 31, demonstrated the formation of ITO 3-APTES (15CE5-MPC) G4-PAMAM) n with n = 16, at which point the experiment was terminated. A linear least squares curve fit of the absorbance to n (at 520 nm) gave an r-value of Among the ligands on the MPCs, the CE ligand was defined to have the longest carbon-chain length for two reasons. First, to facilitate place-exchange reaction, the entering ligands should have longer carbon-chain length than that of the leaving ligand on the MPC, especially in this case where the CE ligand was a bulky group [206]. Second, we hypothesized that in LBL film, the flexibility of the ligands might be limited and might hinder the formation of a sandwich complex with metal ions; a long carbon-chain length would make the CE ligand sufficiently flexible to form the sandwich complex. Previously, Hicks et al. [323] used spectrophotometry to determine the number of monolayers of a MPC deposited per dipping cycle. Here, their approach was adapted to the case of the 15CE5-MPC. That is, the expression A= kεг was used where k is a constant (obtained as described below), ε is the molar absorptivity of the MPC, and Г is the surface excess. To correct for differences in operational parameters and the molar absorptivities of the MPCs employed, their experiment was first replicated by preparing the hexanethiolate and MUA mixed-ligand MPC they used. An n = 6 assembly was prepared on an ITO 3-APTES surface. Relative to the report of 3.7 ± 1.3 monolayers of the hexanethiolate and MUA mixed-ligand MPC per dipping, 1.1 ± 0.3 monolayers of CE-MPC per dipping were deposited. The deposition of 15CE5-MPC G4-PAMAM on a Au 4-ATP electrode of a QCM was monitored (Fig. 32). A linear least squares fit of frequency change with bilayer number (n = 4) yielded an r-value of The importance of the QCM result was that it established the veracity of the experimental approach for LBL assembly on the QCM crystal. This result eliminated our concern that solvent evaporation may have caused deposition by other than electrostatic assembly on the QCM electrode. 150

165 y = 0.007x R 2 = Absorbance at 520 nm Absorbance Num ber of layers Wavelength (nm) Fig. 31 Layer-by-layer assembly of ITO 3-APTES (CE-MPC G4-PAMAM) n monitored by spectrophotometry. Insert is a plot of the maximum absorbance at 520 nm vs the number of AuMPC layers (R 2 = 0.993) 151

166 - delta f (Hz) n Fig. 32 Layer-by-layer assembly of Au 4-ATP (CE-MPC) G4-PAMAM) n monitored by mass measurement with a QCM. Each entry is a single point. 152

167 Proposed applications require accessibility of Cs II and Pb II to the CE functionalities in the inner layers of the assembly. A previous study [323] showed that with a somewhat related MPC, an LBL provided facile charge transport, presumably because of a rather open structure imparted by the presence of these supramolecular assemblies. Hence, the first step in studying transport in the present assemblies was to investigate the relative diffusion of a non-complexing species, ferrocene, through the LBL assemblies by cyclic voltammetry. Initially, the cyclic voltammetry of 0.5 mm Fc was studied at ITO 3-APTES (15CE5-MPC) G4-PAMAM) 4 and ITO 3-APTES (PSS) G4- PAMAM) 4. As shown in Fig. 33, under identical voltammetric conditions the peak currents were comparable. From the ratio in their values, the diffusion coefficients, where the former is denoted D MPC and the latter is denoted D PSS, calculated from the Randles- Sevcik Equation showed that D MPC = 1.6D PSS. The result is in contrast to one system studied by Hicks et al. [323] in which a bilayer comprised of poly(allylamine hydrochloride) and an MPC with mercaptoundecanoic acid and hexanethiol ligands essentially blocked the voltammetry of Fc. This comparison points to an important role of the G4-PAMAM in facilitating mass and charge transport through these LBL assemblies. That the voltammetry of Fc at ITO 3-APTES (15CE5-MPC) G4-PAMAM) n was diffusion limited was shown by a study of peak current, i p, vs. the v 1/2, where v is the scan rate (Fig. 34). Here, the solution was 1.0 mm Fc in CH 2 Cl 2 with 0.1 M Bu 4 NPF 6 as the supporting electrolyte. The scan rate at the 4.2 cm 2 electrode, upon which 6 bilayers were assembled, was varied over the range mv s -1. The data were fitted by a log i p vs. log v plot. The r 2 value was and the slope was Although the theoretical slope for a diffusion limited process is 0.5, the experimental value supports the assignment of a diffusion limited process. The difference between experimental and theoretical values is probably due to uncompensated resistance that caused the peak potential difference to increase with scan rate with a concomitant decrease in peak current. The results with an identical study on a ITO 3-APTES (PSS) (G4-PAMAM) 6 electrode yielded a slope of The ITO 3-APTES (CE-MPC) G4-PAMAM) n electrodes were used to extract Pb II as described in the Experimental section. A typical result is shown in Fig. 35. Integration of the current voltage curve yielded the result that 8.4 x mol cm -2 were trapped. In 153

168 Current/mA Potential/V Fig. 33 Cyclic voltammetry of 0.5 mm ferrocene at ITO 3-APTES (CE-MPC) G4- PAMAM) 4 (solid line) and ITO 3-APTES (PSS) G4-PAMAM) n (dashed line) electrodes. Supporting electrolyte, 0.1 M Bu 4 NPF 6 in CH 2 Cl 2 ; scan rate, 100 mv s -1 ; initial potential, -0.2 V. 154

169 4.0 y = 0.409x R 2 = Log Ip Log V 1.0 Current/mA Potential/V Fig. 34 Influence of scan rate on the cyclic voltammetry of ferrocene at an LBL assembly of CE-MPC. Except for scan rate, the conditions are those in Fig. 34. Insert is a plot of logi p vs logv 155

170 0.0 Current/mA Potential/V Fig. 35 Cyclic voltammetry at an ITO 3-APTES (CE-MPC) G4-PAMAM) 4 electrode in 0.1 M tetraethylammonium chloride before (dashed line) and after (solid line) equilibration with Pb II from a 1.0 mm Pb(NO 3 ) 2, 0.1 M HNO 3 solution. Scan rate, 100 mv s

171 that three bilayers were used, an average of 2.8 x mol cm -2 per monolayer was obtained, which is comparable to the trapping of (6 ± 3) x mol Pb II cm -2 for a selfassembled monolayer (SAM) with a tethered crown ether [385]. Given that the complex between 15-crown-5 and Pb II is a sandwich, the comparison suggested that the CE-MPC has a population density of the crown ethers that can participate in the trapping of the metal comparably to that of the SAM. It was also important that the quantity of the Pb II trapped was proportional to the number of bilayers. Because the study only involved n in the range 1-3 the statistical data are of limited value; the standard deviation of the fit was 30%. After the trapping of Pb II, the LBL assembly was studied by UV-visible spectrophotometry. The spectrum was the same as that reported in Fig. 31, which suggests that the general organization of the network was not altered. Apparently, the Pb II is complexed by the crown ethers in a given monolayer rather than crosslinked between either crown ethers or G4-PAMAM of adjacent bilayers. To test this postulate, the extraction experiment was repeated with an ITO 3-APTES (MUA- MPC G4-PAMAM) 3 electrode that was immersed in a 0.1 M HNO 3, 1.0 mm Pb(NO 3 ) 2 solution for 30 min. Subsequently, cyclic voltammetry was performed at 100 mv s -1. No evidence of the reduction of Pb II was observed, which supports the premise that the Pb II resided on the crown ether in the experiment reported in Fig. 29. The voltammetric study was supported by QCM results. Using an Au 4-ATP (CE- MPC) G4-PAMAM) 3 surface, the average amount of Pb II trapped per monolayer was 2.1 x mol cm -2. The solid phase extraction of Cs + from solution by the 18C6-MPC film conducted on the EQCM showed that 1.7 x mol cm -2 of Cs were trapped per monolayer. Moreover, increasing the time of exposure of the surface to the Pb II solution did not increase the quantity of the metal that was trapped, which suggests that the system reached equilibrium in 10 min. 7.4 Conclusions The electrostatic layer-by-layer assembly of cationic G4-PAMAM and anionic mixed-ligand monolayer protected gold nanoclusters with 15-crown-5 functionality was successful on both indium tin oxide and on gold. In the former case, the surface was 157

172 treated with 3-aminopropyl triethoxysilane, and in the latter, 4-aminothiophenol. Up to 16 bilayers were formed with a constant quantity of the MPC in each. The assembly yielded 1.1 monolayer of the MPC per cycle, the trapping capacity of which was 2.8 x mol Pb II cm -2 per monolayer. Facile diffusion of a neutral compound, ferrocene, through an assembly comprising 4 bilayers was demonstrated. These LBL assemblies are not only useful for fundamental studies but also have promise for electrochemical stripping analysis. In this regard, functionalities other than crown ethers can be used as trapping agents, thereby extending the scope of possible analytes. In addition, in the manner reported by Wagner et al. [391], the complexing agent may enhance the analytical utility of the LBL assembly by blocking the transport of an interfering species to the electrode surface. 158

173 Chapter 8 Optimization of the Dispersion of Gold and Platinum Nanoparticles on Indium Tin Oxide for the Electrocatalytic Oxidation of Cysteine and Arsenite 8.1 Introduction The catalysis of electrode reactions by metal nanoparticles has recently become accepted. A summary of the theory and examples of systems are presented in the treatise edited by Wieckowski et al. [392]. The present study focused on fabrication of arrays of nanoparticles on electrode surfaces with the intent of identifying the minimum density needed to achieve a mass-transport limited current for the electrocatalytic oxidation of selected test systems. To meet this objective required an investigation of methods to fabricate nanoparticle-modified electrodes with controlled packing densities. Described herein is the preparation of such electrodes on the basis of synthesizing nanoparticles (NPs) in the void volume of generation-4 poly(amidoamine) dendrimer (PAMAM), electrostatically depositing a monolayer on an electrode from mixtures of PAMAM and NP-containing PAMAM, and thermally converting the surface to an electrode modified with the desired surface coverage of NPs. As mentioned in Chapter 5, the use of PAMAM in the template synthesis of NPs was first reported in 1998 [210, 211]. In these studies, Cu II was complexed with the amine sites of the dendrimer and reduced to form NPs. Crooks et al. [231] have reviewed the early studies of template synthesis of NPs and their application to catalysis. Of particular importance to the present study is the use of this method to prepare PtNPs [215]; here, the PAMAM was OH-terminated (external sites) so that the complexation of the PtCl - 4 was limited to the interior. The size the resulting NPs was correlated to the intradendrimer loading by PtCl 2-4 prior to the reduction with BH - 4. An extension of that approach led to the synthesis of near-monodisperse gold NPs [393]. With 55 and 140 equivalents of AuCl - 4 per dendrimer, AuNPs of 1.3 ± 0.3 and 1.6 ± 0.3 nm diameter, respectively, were obtained after chemical reduction. 159

174 The present study required a method of preparing a controlled array of NPs on electrode surfaces. Previously, the use of nanowells as templates had been employed. For example, deposition of a metal into the void volume of a nanoporous membrane yields an array with features analogous to a distribution of NPs on an electrode [394, 395]. However, this and related approaches are not readily amenable to variation of the NP density. Methods involving impregnation of polymers with NPs are promising. For example, the formation of AuNPs in micelles of block copolymers [356] yields a distribution that is potentially variably. The approach taken in this study is related to layer-by-layer (LBL) electrostatic assembly of supramolecules [313, 317, 374, 375, 377, 396]. Generally, the electrode surface is first modified by deposition of a charged self-assembled monolayer. Subsequently, it is dipped sequentially in solutions of supramolecules of opposite charge. An example that yields a modified electrode with electrocatalytic properties is to first coat a gold surface with a positively charged self-assembled monolayer of 4- aminothiophenol and then immersing it sequentially in an anionic supramolecule, phosphomolybdate, and a cationic spacer, generation-4 PAMAM [378]. Multilayers that contained citrate-protected AuNPs with diameters in the 1 4 nm range were prepared by Feldheim et al. [380] using the LBL method. The generation-4 PAMAM spacer hindered electron transfer between the layers. A nanocomposite of generation-5 PAMAM and poly(styrenesulfonate), in which the dendrimer contained AuNPs, was prepared by LBL assembly [354]. Of importance to the present study was the demonstration by atomic force microscopy that a uniform distribution of AuNPs was obtained. As stated above [392], NPs have been extensively studied as electrochemical catalysts. Most of the applications are to redox of species of interest in fuel cell technology. Among the studies related to the present report is oxidation of dopamine at a gold electrode that was modified by a self-assembled monolayer (SAM) with terminal amines; subsequently, AuCl - 4 was bound by ion-exchange and reduced with BH - 4 in the presence of citrate to form the AuNP catalysts [362]. The same general type of electrode, namely Au modified by a SAM (1,9-nonanedithiol) on which AuNPs were deposited, was employed, along with a related set of modified Au electrodes, for the electrocatalytic oxidation of ascorbic acid [324]; here the AuNPs were monolayer-protected clusters 160

175 (MPCs) with 4-hydroxythiophenol as the sheath. The Au SAM blocked the oxidation of ascorbic acid, whereas the Au SAM MPC promoted it. The possibility that the Au centers act as microelectrodes was discussed. The present study has two significant differences from previous reports. First, the organic components of the assemblies, which were a SAM-modified electrode on which generation-4 PAMAM that contained NPs was electrostatically bound, were thermally decomposed to form a NP array on the surface. Second, the electrode material and test systems were selected on the basis of not yielding a faradaic process in the absence of NPs in the potential window studied. This approach permitted the evaluation of the role of the density of the NPs on the surface in the generation of catalytic current. 8.2 Experimental Chemicals and materials Unless otherwise stated, the chemicals were ACS Reagent Grade from Aldrich Chemical Company (Milwaukee, WI). Also from Aldrich were the following: HAuCl 4 (99.9%), K 2 PtCl 4 (99.99%), generation-4 poly(amidoamine) dendrimer with hydroxyl surface groups as a 10% (weight) solution in methanol, sodium borohydride (98%); 3- aminopropyltriethoxysilane (99%) and poly(allylamine hydrochloride) with a molecular mass of 70,000. The L-cysteine was from Sigma Chemical Company (Louis, MO), and the sodium arsenite (98%) was from Fisher Scientific Company (Fair Lawn, NJ). Glass slides with indium tin oxide (ITO) coated in one surface were from Delta Technologies Limited (Stillwater, MN). Mica (muscovite) was from Ward s Natural Science (Rochester, NY). Solutions were prepared with house-distilled water that was further purified with Barnstead NANOpure II system (Boston, MA). Double-sided tape was from Scapa (Windsor, CT) Apparatus The voltammetry experiments were performed with a CHI Model 750 electrochemical workstation or a CHI Model 800 electrochemical detector (CH Instruments, Austin, TX). The atomic force microscopy (AFM) images were made on a 161

176 Digital Instruments NanoScope MultiMode scanning probe microscope that was used in the Tapping Mode (Veeco Instruments Inc., Woodbury, NY). Thermogravimetric analysis (TGA) data were collected with a Perkin Elmer Thermogravimetric Analyzer TGA7 (Perkin Elmer, Chicago, IL) Synthetic methods All of the glassware was cleaned with aqua regia solution (3:1, v/v, HCl:HNO 3 ), followed by rinsed thoroughly with distilled water before used. Caution: Aqua regia solution reacts violently with organic material and metals, thus it must be handled with great care. Dendrimer-encapsulated AuNPs were prepared by a modification of a procedures reported by Amis et al. [237]. Briefly, 5.0 ml of a 0.12 % (wt.) generation-4 poly(amidoamine) dendrimer with surface OH-groups, which is hereafter symbolized as PAMAM unless otherwise stated, was prepared by dilution of the purchased reagent with water. It was mixed with AuCl - 4 (12.5 ml of 2 mm solution) for 20 minutes to obtain a solution with 64 equivalents of 2 mm HAuCl 4. A 5-fold molar excess of freshly prepared 0.5 M NaBH 4 in 0.3 M NaOH (0.25 ml) was added to this stirring solution. The color of the solution immediately changed from light-yellow to brown. The solution was stirred for an additional 20 minutes. Dendrimer-encapsulated PtNPs were synthesized by a modification of method reported by Crooks and co-workers [238]. An aliquot of the 10% (wt.) PAMAM in methanol was evaporated to dryness, from which a 5 mm aqueous solution (0.25 ml) was prepared, and 25 ml of a 2 mm K 2 PtCl 4 aqueous solution was slowly added with stirring. The Pt II : PAMAM mole ratio was 40 : 1. The solution had pale yellow color. 2- This mixture solution was stirred for 4 days to ensure complete complexation of PtCl 4 with the interior amine groups of PAMAM [218]. A 5-fold molar excess of freshly prepared 0.5 M NaBH 4 (0.5 ml) was quickly added to this solution, followed by stirring for an additional 20 minutes. Both the gold and platinum nanoparticles encapsulated in the generation-4 poly(amidoamine) dendrimers with surface OH-groups, herein designated as Au-PAMAM and Pt-PAMAM, respectively, were used without further purification. 162

177 Procedures ITO was cleaned and handle with care as described in Chapter 7. Assemblies of Au-PAMAM and Pt-PAMAM on ITO were fabricated as follows. The ITO was modified by formation of a monolayer of 3-aminopropyltriethoxysilane (APTES) by immersion in a methanol solution of 0.06 M APTES for 1 h. The ITO APTES was rinsed with dry methanol and dried with He gas. A layer of Au-PAMAM or Pt-PAMAM was electrostatically assembled on the modified ITO by immersion of this material for 1 h in either the Au-PAMAM or the Pt-PAMAM solution that was prepared as described above. Before immersion, the dendrimer solutions were adjusted to ph 4 with HCl. In most cases, the Au-PAMAM and the Pt-PAMAM solutions were diluted with 0.12% (wt.) and 0.08% (wt.) PAMAM (with surface OH-groups), respectively, to obtain the desired mole fraction, moles NP-PAMAM / (moles NP-PAMAM + moles PAMAM), of the dendrimer that contains NPs in the monolayer that was attached to the ITO APTES surface. The mixture of NP-PAMAM and PAMAM was always prepared from a freshly diluted PAMAM solution. The resulting assemblies were rinsed with water and dried with He gas. Mica was cut into ~1 cm x 1 cm pieces and cleaved before each new experiment by peeling with double-sided tape. A piece of double sided tape was pressed to the mica surface then was gently peeled off the surface. The resulting mica surface should appear flat to the naked eye. The above surface-peeling procedure was repeated until the mica surface appears to be flat. Assembly of Au-PAMAM and Pt-PAMAM on mica was done by the same procedure as with ITO. The only difference is that freshly cleaved mica was modified in a poly(allylamine hydrochloride), PAH, aqueous solution (10 mg PAH in 10 ml of water) for at least 1 h. Finally, unless otherwise stated, the PAMAM and APTES (or PAH) were thermally decomposed by heating the modified ITO or mica at 350 o C in the air for 2 h. Electrochemical measurements were in a conventional three-electrode cell, in which the working electrode was ITO and the reference electrode was Ag AgCl. With 1 mm cysteine as the test species, the supporting electrolyte was a mixture containing 0.1 M KNO 3 and 1 mm HNO 3. With 1 mm NaAsO 2 as the test species, the electrolyte was a 0.1 M acetate buffer at ph

178 8.3 Results and discussion Initial experiments were performed with ITO as the base electrode, AuNPs as the catalyst, and cysteine as the test substance. The selection of this system was based on preliminary trials which showed that cysteine was not oxidized at ITO in the absence of AuNPs (Fig. 36). An electrode was prepared by immersion of ITO APTES in a mixture with a mole fraction of Au-PAMAM of (see the Experimental section). Subsequently, it was heated at 350 o C to remove the organic compounds (see Fig. 37). This temperature was selected on the basis of TGA of an ITO, APTES, PAMAM sample (Fig. 38). Addressing the objectives of this study required a means of preparing electrodes with a variable density of NPs on the surface. The method that was tested was to vary the mole fraction of NP-PAMAM in the solution used to bind this material to the surface. There were two general concerns with this approach. First, if the free energies of binding of NP-PAMAM and of PAMAM with the surface modified with a positively charged monolayer differ significantly, varying this mole fraction will not be reflected in an analogous change in the density of the NPs. Second, at the decomposition temperature of the PAMAM, 350 o C, the possibility of annealing the NPs was not discounted. To test for the presence of either, or both, of these potential problems, ITO APTES (Au-PAMAM, PAMAM) was heated at 350 o C and imaged by AFM. Although ITO is too rough for unambiguous conclusions about the AuNP distribution, the results supported the model that it varied systematically with the mole fraction of Au-PAMAM in the solution used in the fabrication of the assembly. Moreover, individual AuNPs were observed after the heat treatment (Fig. 39). The above result was verified by repeating the experiment except with a mica PAH (Au-PAMAM, PAMAM) sample. Here, mica, muscovite, serves as a smooth substrate for the AFM imaging. Before presenting the data obtained, some characteristics of muscovite will be briefly described. Muscovite is potassium aluminum silicate, KAl 2 (OH) 2 Si 3 AlO 10 [397]. It consists of infinite sheets of corner-shared SiO 4 tetrahedra, with the apical oxygen atoms located at the corners of a hexagon. In this structure, onefourth of the Si is replaced by Al, with the remaining K + ions and Al 3+ ions lying between the aluminosilicate sheets [Fig.40]. Because the ionic bonding between the 164

179 Fig. 36 Cyclic voltammetry of cysteine at a) ITO and at b) AuNP-ITO electrodes. Scan rate, 100 mv s -1 ; electrolyte, 1 mm cysteine in 0.1 M KNO 3,1 mm HNO 3 ; geometric area of the ITO, 3.3 cm 2. The electrode was prepared from a mixture of Au-PAMAM and PAMAM with a mole fraction of Au-PAMAM of

180 OH OH OH OH OH 3-APTES OH OH OH OH OH OH OH ITO glass O O Si NH 2 O O O Si NH 2 O O O Si NH 2 O O O Si NH 2 O Au-PAMAM + PAMAM O O Si + NH 3 O O O Si + NH 3 O O O Si + NH 3 O O O Si + NH 3 O 350 o C 2h Fig. 37 Scheme of modification of electrode with AuNPs 166

181 Fig. 38 Thermogravimetric analysis of generation-4 PAMAM with surface OH-groups. 167

182 Fig. 39 AFM image of AuNPs on an ITO substrate 168

183 = K = SiO 4 = AlO 6 Fig. 40 Structure of muscovite projected along the sheet-stacking direction. A single octahedral AlO 6 sheet is sandwiched between two tetrahedral SiO 4 sheets, with a layer of K + ions located between the trilayer aluminosilicate sheets (based on ref. 397). 169

184 K + layers and the trilayer aluminosilicate sheets is rather weak, mica cleaves rather easily at the positions of the K + layers, leaving oxygen atoms on the surface. After cleavage, no further surface preparation, nor cleaning is required. Clean mica surfaces are useful for AFM studies because large, single-crystal terraces of over 1 µm can be easily observed. There are two likely sources of strong interaction energies between the PAH and the mica surface. First, there is electrostatic attraction between positively charged PAH and the negatively charged mica surface. Second, PAH may also interact with the mica surface through hydrogen bonding. Typical results from a set of experiments with mica surfaces in which the mole fraction of Au-PAMAM was varied from 1.0 to 0.0 are shown in Fig.41. The image quality does not permit measurement of the size of the AuNPs, but from the literature, the procedure that is used yields particles with diameters of ca. 1-2 nm [393]. Having validated that the NPs density was variable, the voltammetric experiment in Fig. 38 was repeated using electrodes prepared with a varying mole fraction of AuNP. As shown in Fig. 42, the maximum current for the oxidation of cysteine was obtained under conditions with a highly dispersed AuNP array that was obtained with a mole fraction of Au-PAMAM of With this mole fraction, the mica AuNP electrode had an estimated spacing of 200 nm between nanoparticles. This value cannot be used directly for the ITO AuNP system because the surface roughness led to a more heterogeneous distribution of nanoparticles than that shown for mica. Nevertheless, the combination of the size of the NPs and the distribution on ITO suggested that these catalytic centers will behave as an array at which current is potentially limited by hemispherical diffusion. To test the hemispherical model suggested by the AFM results, the current for the oxidation of cysteine was measured as a function of scan rate (Fig. 43). The peak current, i p, increased systematically over the scan rate, v, range, mv s -1. A plot of log i p vs. log v had a slope of 0.6 and r 2 of when analyzed by a linear least squares fit. Hence, over this v-range, the system behaved much like the semi-infinite linear diffusion case, where the theoretical value of the slope is 0.5. This range of scan rate was too great for hemispherical diffusion to limit the current. However, even at values below 5 mv s -1, 170

185 A B C D E F Fig. 41 AFM of mica PAH (Au-PAMAM, PAMAM) after heating at 350 o C. Mole fraction of Au-PAMAM in the mixture = (A), (B), (C), (D), (E), and (F). 171

186 Fig. 42 Influence of the AuNP density on the peak current for the oxidation of cysteine. Scan rate, 100 mv s -1 ; electrolyte, 1 mm cysteine in 0.1 M KNO 3, 1 mm HNO 3 ; electrode, ITO APTES (Au-PAMAM, PAMAM) that was treated at 350 o C; geometric area of the ITO, 0.39 cm 2. The mole fraction of Au-PAMAM was varied. Each entry is a single point. 172

187 Fig. 43 Influence of scan rate on the cyclic voltammetry of cysteine. The conditions are those in the caption of Fig. 40 except that the mole fraction of Au-PAMAM was 0.07, the geometric area was 3.3 cm 2 and the scan rate range was mv s -1. Traces a i correspond to scan rates of 1, 5, 10, 20, 50, 80, 100, 150, and 200 mv s

188 the current was not independent of scan rate, which is a characteristic of hemispherical diffusion. There are at least two potential problems with the experiment in Fig. 43. First, because of the roughness of the ITO, the NPs can be envisioned as residing preferentially in valleys and along edges of the material. In that event, achieving hemispherical diffusion patterns will be problematic. Second, cysteine is a thiol that may chemisorb to gold, thereby imparting an element of oxidation of a surface-confined species to the voltammetry. A current that is limited to a surface process will be characterized by a slope of 1.0 in the log i p vs. log v plot described above. In support of a contribution of a surface reaction, note that in Fig. 43 there is a suggestion of two anodic processes involving cysteine, possibly that of the solution species and that of an adsorbed species. In an attempt to clarify the voltammetry at the ITO NP electrode, the test system was changed to the oxidation of As III at PtNPs. The cyclic voltammetry over the mv s -1 scan rate range (Fig. 44) was similar to that for cysteine except that only a single anodic process was evident. A linear least squares fit of the log i p vs. log v plot yielded a slope of 0.5 and an r 2 of 0.997, which suggested semi-infinite linear diffusion control of the current without any complication by adsorption over this scan rate range. When the v- range was restricted to 1 5 mv s -1, in which hemispherical diffusion control is possible, the results shown in Fig. 45 were obtained. Two important points are evident. First, the shape of the voltammogram is sigmoidal in which a steady-state current plateau is developed. This shape is indicative of hemispherical control of the current. The other common factors that give this shape, catalytic regeneration of the electroactive species and chemical kinetic control of the current, are inconsistent with the oxidation of As III. However, independence of current of scan rate is not observed, so the current limitation is not simply hemispherical diffusion. To determine whether the topography of ITO is preventing the system from having hemispherical diffusion control of the current at low scan rates, further work is in progress using mica as the substrate for vapor-deposited conductors as electrodes and supports for NPs. 174

189 Fig. 44 Linear scan voltammetry at mv s -1 of the oxidation of AsO - 2 at an ITO APTES (Pt-PAMAM, PAMAM) electrode. Mole fraction of Pt-PAMAM in the Pt- PAMAM, PAMAM mixture, 0.1; the electrode that was treated at 350 o C prior to voltammetry; the electrolyte was 1 mm As III in 0.1 M acetate buffer at ph 4.8; geometric area of the ITO, 3.3 cm 2. Scans a h correspond to scan rates of 10, 20, 40, 60, 80, 100, 150, and 200 mv s

190 Fig. 45 Linear scan voltammetry of AsO - 2 at scan rates of 1 5 mv s -1. The conditions are those in the caption to Fig. 42 except geometric area of the ITO was 0.39 cm 2. Scans a, b, and c are at 1, 2, and 5 mv s -1, respectively. 176

191 Analogous to the case of the oxidation at cysteine at ITO that was partially covered with AuNPs (Fig. 42), a low surface coverage of PtNPs on ITO was sufficient to develop the maximum current for the oxidation of As III (Fig. 46). Here, an electrode that was prepared from a monolayer with a mole fraction of 0.2 of Pt-PAMAM relative to the total of Pt-PAMAM and PAMAM was sufficient to develop the maximum current for the oxidation of As III. The effective surface coverage was estimated for this mole fraction by measuring the peak current for the oxidation of 1 mm As III at a 100 mv s -1 scan rate. The data were fitted to the Randles-Sevcik Equation assuming linear diffusion control, which is suggested by the above scan rate study using a diffusion coefficient of 5 x 10-6 cm 2 s -1 and a two-electron oxidation. The effective area of the electrode thereby calculated was 3 x 10-5 cm 2, whereas the geometric area of the ITO was 0.39 cm 2. Also assumed was an approximately homogeneous distribution of PtNPs on the ITO, but based on AFM images (not shown), the NPs existed in limited domains, which may account for a slope of 0.5 in the i p vs. v 1/2 plot, as discussed in the interpretation of Fig. 42. As in the case of AuNPs on electrodes, presently electrodes prepared on a mica base are being prepared and tested. A difficulty is that the simplest electrode to prepare on a mica base is vapor-deposited Au. The chemical system used in the study must not be oxidized on such a surface in the absence of NPs. The oxidation of As III on PtNPs on an Au surface is the system under investigation in this regard. In term of analytical experiments, calibration curves of cysteine and asernite were obtained at AuNP- and PtNP-electrodes (Fig. 47 and Fig. 48), respectively. A linear response over the range 5 µm µm was obtained for cysteine with AuNP catalysts. For As III with PtNPs, linearity was observed over the range mm. 8.4 Conclusions The oxidations of cysteine and of AsO - 2 at ITO that was partially coated with AuNPs and PtNPs yielded two surprising results. First, even though AFM images of NPs on ITO and on mica suggested that the particles were physically separated, analysis of the peak current as a function of scan rate gave a behavior that approached the linear diffusion limit. This result may be indicative of mass transport vectors that are, first, perpendicular to the ITO (with the diffusion coefficient being that of the electroactive 177

192 Fig. 46 Influence of the PtNP density on ITO on the peak current for the oxidation of AsO - 2. The conditions are those in the caption of Fig. 40 except that the mole fraction of Pt-PAMAM in the Pt-PAMAM, PAMAM solution was varied. Each entry is a single point. 178