“Soggy Sand” Electrolytes - American Chemical Society

Apr 29, 2010 - Solid State and Structural Chemistry Unit, Indian Institute of Science, ... Static and dynamic rheology measurements show formation of ...
0 downloads 0 Views 743KB Size
6830

J. Phys. Chem. B 2010, 114, 6830–6835

Influence of Oxide Particle Network Morphology on Ion Solvation and Transport in “Soggy Sand” Electrolytes Shyamal K. Das and Aninda J. Bhattacharyya* Solid State and Structural Chemistry Unit, Indian Institute of Science, Bangalore 560012, India ReceiVed: March 21, 2010

The role of oxide surface chemical composition and solvent on ion solvation and ion transport of “soggy sand” electrolytes are discussed here. A “soggy sand” electrolyte system comprising dispersions of hydrophilic/ hydrophobic functionalized aerosil silica in lithium perchlorate-methoxy polyethylene glycol solution was employed for the study. Static and dynamic rheology measurements show formation of an attractive particle network in the case of the composite with unmodified aerosil silica (i.e., with surface silanol groups) as well as composites with hydrophobic alkane groups. While particle network in the composite with hydrophilic aerosil silica (unmodified) were due to hydrogen bonding, hydrophobic aerosil silica particles were held together via van der Waals forces. The network strength in the latter case (i.e., for hydrophobic composites) were weaker compared with the composite with unmodified aerosil silica. Both unmodified silica as well as hydrophobic silica composites displayed solid-like mechanical strength. No enhancement in ionic conductivity compared to the liquid electrolyte was observed in the case of the unmodified silica. This was attributed to the existence of a very strong particle network, which led to the “expulsion” of all conducting entities from the interfacial region between adjacent particles. The ionic conductivity for composites with hydrophobic aerosil particles displayed ionic conductivity dependent on the size of the hydrophobic chemical moiety. No spanning attractive particle network was observed for aerosil particles with surfaces modified with stronger hydrophilic groups (than silanol). The composite resembled a sol, and no percolation in ionic conductivity was observed. 1. Introduction 1,2

“Soggy sand” electrolytes are composite of dispersions of fine oxide particles (size: 7-300 nm) such as silica (SiO2) in nonaqueous liquid salt solutions (e.g., LiPF6-EC:DMC). The composites have higher ionic conductivity and solid-like mechanical strength compared to the neat liquid electrolytes. These beneficial features combined with an electrochemical voltage window of 3-4 V3 make “soggy sand” electrolytes promising for application in electrochemical devices such as rechargeable lithium batteries.4 Apart from the novelty in terms applications, what is fundamentally interesting and intriguing is the phenomenon of ion transport in “soggy sand” electrolytes. The ionic conductivity in “soggy sand” electrolytes does not arise out of trivial additive effects but is highly synergistic in nature. In the published reports prior to 2009,1,2,4 the ionic conductivity was qualitatively accounted for by the concept of heterogeneous doping.5 Maier (MPI, Stuttgart) initially introduced this concept for prediction and quantitative description of the beneficial influence of insulator (such as Al2O3, SiO2) dispersions in weak electrolytes (such as LiI, PbF2). A distinguishing feature compared to normal or “homogeneous” doping is that, in addition to the varying dopant concentration, heterogeneous doping allows one to tune electrolyte ion transport via variation in dopant morphology (size, shape)4 and surface chemical composition.6 The composite ionic conductivity displays a percolation-type behavior owing to the formation of a highly conductive interface space charge layer at the oxide conductor interface.5 Briefly, the concept states that, in presence of the * Corresponding author. E-mail: [email protected]. Tel: +91 80 22932616. Fax: +91 80 23601310.

insulator, the undissociated ground state is broken up, and the counter carrier, i.e., the vacancy is set free. The sign of the carrier depends significantly on the dispersant acidity/basicity. While dispersion of Al2O3 in LiI increases the lithium ion vacancy concentration, SiO2 dispersion in PbF2 leads to fluorine ion vacancies. These arguments to a great extent account well for similar enhancements in ionic conductivity observed in the case of polymer composite electrolytes.7 Modeling of ion transport in solid composite electrolytes has also been successfully done using computer simulation studies.8,9 The composites were modeled assuming a random distribution of the oxide phase. Coming to the case of liquid electrolytes such as Libattery electrolytes, the “ground state” is the undissociated ionpair that does not contribute to the Li+ conductivity. Hence, adsorption of the anion (say) will lead to dissociation of the LiX ion pair and enhancement in Li+ ions in the space charge layer in the vicinity of the oxide layer. It was proposed that at silica volume fraction φ ) φonset, a spanning particle network is formed, resulting in overlap of space charge and percolation in ionic conductivity. We feel that mere formation of a connected particle network and space charge are not sufficient criteria for percolation in ionic conductivity. Complete accountability needs to take into account other nontrivialities of “soggy sand” electrolytes. We previously demonstrated6 the surface chemical composition of the oxide particles and that their configuration in the liquid played a key role in determining the network structure and percolation in ionic conductivity. The importance of the particle network on ion transport has also been highlighted by the group of Maier. Jarosik et al. recently demonstrated via Monte Carlo computer simulation10 that the temporal stability of the network also plays an important role

10.1021/jp102548e  2010 American Chemical Society Published on Web 04/29/2010

Role of Oxide Composition on “Soggy Sand” Electrolytes in the percolation of ionic conductivity in “soggy sand” electrolytes. In this work we highlight the importance of the nature of the solvent on the network formation and influence on ionic conductivity. Correlation of particle network structure with ion transport has been systematically studied in a “soggy sand” electrolyte system comprising of various surface functionalized aerosil silica dispersed in lithium perchlorate (LiClO4)-methoxy polyethylene glycol (m-PEG) solutions. The study is implemented via static and dynamic rheology and ionic conductivity. Zeta (ζ)-potential of the oxide surfaces was also done to supplement observations from rheology and ionic conductivity. 2. Experimental: Materials and Methods The starting material Aerosil 300 silica (Degussa; A300) (Brunauer-Emmett-Teller (BET) surface area ) 300 m2g-1, primary particle size ) 7 nm), has a hydrophilic surface containing OH groups 2-3 [-OH] nm-2.11 Chemical modification of the A300 surface is performed via treatment of A300 with various hydrophobic/hydrophilic methoxysilane (MS, all Sigma-Aldrich) moieties, viz., methyltrimethoxysilane (MTMS), octyltrimethoxysilane (OTMS), (3-aminopropyl)trimethoxysilane (APTMS). Appropriate quantities of MTMS, OTMS, and APTMS were added to a stirred suspension of A300 in toluene at room temperature (25 °C). The mixture was stirred for 24 h. The solid phase recovered by filtration was washed several times with isopropanol and toluene and then dried under vacuum at 100 °C for 24 h. The dried particles are designated as MA300, OA300, and NA300 corresponding to MTMS-, OTMS-, and APTMS-functionalized A300 silica, respectively. Functionalization and quantification of chemical moieties of the oxide surface was done using CHN microanalysis (Perkin-Elmer, 2400 CHNSO system), Fourier transform infrared (FTIR) spectroscopy (Perkin-Elmer FT-IR Spectrometer Spectrum 1000), transmission electron microscopy (TEM) (FEI Tecnai F30, images are taken at 200 kV) and thermogravimetry analysis (TGA) (Mettler Toledo). LiClO4 (lithium battery grade, Chemetall) was preheated to 120 °C under vacuum for several days prior to use. m-PEG (Mw ) 350 mol g-1, Sigma Aldrich) were dried over 4 Å molecular sieves (Qualigens) and stored in a glovebox under pure argon atmosphere (MBraun, water < 0.1 ppm). All oxides were subjected to a preheat treatment prior to electrolyte preparation. A300 and functionalized A300 (i.e., MA300, OA300, NA300) were preheated to 300 and 180 °C, respectively, for 24 h to rule out effects due to physisorped water. The preheat treatment temperatures were selected on the basis of the TGA (Supporting Information, Figure 1) performed under N2 atmosphere (heating rate ) 5 °C min-1). The weight loss observed for MA300, OA300, and NA300 until approximately 110 °C was due to physisorped water. For temperatures greater than 200 °C, weight loss was significant, and this is attributed to decomposition of various surface functional groups. In the case of A300, no significant weight loss was observed in the measuring temperature range (room temperature to 700 °C). Additionally, 0.1 M LiClO4-m-PEG solution was continuously stirred at 60 °C for several hours and cooled to room temperature prior to electrolyte preparation. The “soggy sand” composite electrolytes were prepared by mixing the 0.1 M lithium perchlorate (LiClO4) in m-PEG solution with preheated A300/ MA300/OA300/NA300 particles using a vortex mixture (∼2000 vibrations min-1) inside a glovebag (H2O ≈ 1 ppm). To check the effect of water on the electrolyte conductivity, the requisite amount of deionized water in parts per million was added to

J. Phys. Chem. B, Vol. 114, No. 20, 2010 6831 the m-PEG/0.1 M LiClO4 solution inside the argon-filled glovebag (Supporting Information, Figure 4). Ionic conductivity was obtained from ac-impedance spectroscopy (Alpha, Novocontrol) by scanning the sample in the frequency range: 1-106 Hz (signal amplitude: 0.05 V). The electrolyte was sandwiched between stainless steel electrodes in a home-built quartz glass cell at room temperature (25 °C), and the sample was loaded inside a glovebox. The roomtemperature conductivity was evaluated from the intercept of the low frequency spike with the real axis obtained from impedance measurements. The variation in ionic conductivity estimates were approximately (15%. Intrinsic viscosity and storage/loss modulus at room temperature (25 °C) of various composite electrolytes were obtained from rheology (Advanced Rheometer AR2000, TA Instruments) measurements involving steady and oscillatory shear. Various cell configurations (cone and plate, parallel plate) were employed as per sample and test requirements. A steady preshear was applied at a shear rate of 0.5 s-1 for 60 s followed by a rest time of 120 s before each dynamic experiment to avoid any previous shear histories and maintain the equilibrium state of the materials.12 Additionally, the linear viscoelastic (LVE) region for each composite material was determined by performing a dynamic measurement as a function of strain amplitude at a constant oscillation frequency of 1 rad s-1. Surface potential was obtained from electrophoretic ζ-potential measurements (Zetasizer Nano ZS, Malvern Instruments). For ζ-potential, a solution of 1 mM LiClO4 in m-PEG was prepared, and the silica volume fraction (φ) was adjusted to φ ) 0.005.1,2,13 For prevention of particle agglomeration, the electrolyte solution was stirred and sonicated prior to the measurement. The basis for selection of salt and oxide concentrations is significantly lower than that employed for ionic conductivity and rheology measurements as a result of the technical constraints and accuracy involved with electrophoretic measurements. 3. Results and Discussion Elemental estimates (from CHN analysis) and FTIR (Supporting Information, Figure 2) of MA300, OA300, and NA300 show successful chemical functionalization of A300 surface. The surface concentrations of various groups depend on the hydrophilicity/hydrophobicity and their size. Estimation of the grafting density of the functional group shows partial replacement of silanol groups (Si-OH) by various functionalized moieties. The grafting density of the functional groups were obtained using results of the elemental analysis normalized to the surface area (300 m2 g-1) of as-received A300. Monolayer coverage of the functional moieties with no alkoxy groups retained on the silica surface (A300) was assumed for the calculation of density of various functional groups. In general, the hydrophilic group (as in NA300) showed a higher surface coverage (78%; grafting density ) 2.4 nm-2) compared to the hydrophobic groups of methyl (MA300: 67%; grafting density ) 2.0 nm-2) and octyl (OA300: 56%; grafting density ) 1.7 nm-2). Functionalization resulted in only growth in particle size. The intrinsic morphology for the functionalized particles remained unchanged, displaying a branched particle aggregation similar to that of unfunctionalized A300 (Supporting Information, Figure 3). Static (Figure 1) and dynamic (Figure 2) rheology tests were performed to investigate particle arrangement in the composite electrolyte. Figure 1 shows the variation of viscosity (η) as a function of shear rate (γ˙ ) for all composites. The silica volume

6832

J. Phys. Chem. B, Vol. 114, No. 20, 2010

Figure 1. (A) Steady-shear viscosity as a function of shear rate for various composites (A300: as received, MA300: methyl-capped A300, NA300: amine-capped A300, and OA3000: octyl-capped A300). For composites with silica and functionalized silica, the volume fractions (φ) are 0.06 and 0.07, respectively. Photographs show the visual appearance of various composite electrolytes. (B) Variation of viscosity as a function of various silica surface functional groups in m-PEG and EG. In all composites, the silica volume fraction is 0.07, except for the composite indicated by the down arrow, which was 0.06.

Figure 2. Variation of elastic (G′) and viscous (G′′) moduli as a function of frequency for various composites, as obtained from dynamic frequency sweep measurement (A300: as received, MA300: methylcapped A300, NA300: amine-capped A300, and OA300: octyl-capped A300). For composites with silica and functionalized silica, the volume fractions (φ) are 0.06 and 0.07, respectively.

Das and Bhattacharyya fractions (φ) were 0.06 and 0.07 for A300 and MA300/OA300/ NA300, respectively. A300 show shear thinning, i.e., macroscopic viscosity decreases with an increase of shear rate. This observation is in accordance with the physical appearance, which resembled a gel in the oxide concentration regime 0.1 e φ e 0.06 (onset for gel formation, φonset ≈ 0.01). Contrary to physical observation, MA300 and OA300 samples also demonstrated shear thinning. Shear thinning behavior is possible if and only if the silica particles form an attractive network in the LiClO4-mPEG medium. Fitting η versus γ˙ plots with the power law equation of η ) mγ˙ (n-1) reveals that n < 1 and implies existence of a connected particle network.14 The zero shear (γ˙ ∼ 10-3 s-1) viscosities of A300, MA300, and OA300 are the same, nearly equal to 1 MPa s. However, it should be mentioned that the degree of shear thinning is very rapid in the cases of MA300 (φ ) 0.07) and OA300 (φ ) 0.07) compared to A300 (φ ) 0.06). The differences can be accounted for in terms of the nature and strength of interactions between adjacent silica particles discussed at a later stage. The reason for exhibiting shear thinning behavior in A300, OA300, and MA300 is primarily due to rupture of the attractive network into smaller and smaller connective flow units (smaller portions of the attractive particle network that contain a number of connected particles). Since the flow units also contain a nonaqueous solution volume that is proportional to the size of the flow unit, breaking of the flow unit into smaller units releases this liquid in which the flow units can move. With increasing shear rates, the size of the flow units decreases, thus releasing more and more liquid. Ultimately, at very high shear rates, the flow unit is the particle itself. At such shear regimes, the viscosity becomes identical to that of a slurry where the particles are highly repulsive. For NA300, η is almost independent of γ˙ , resembling a Newtonian fluid. No attractive particle network is formed for the NA300 sample. The slight increase in viscosity compared to m-PEG () 0.027 Pa s at 25 °C15) is attributed to the combined effect of added salt and silica. Dynamic rheology measurements were performed to further probe the composite microstructure. The experiments were performed under constant strain, the magnitude of which was within the LVE regime. Figure 2 shows the frequency dependence of elastic (G′) and viscous (G′′) moduli for all composite electrolytes. Over the measured frequency range, G′ is greater than G′′ for A300 (φ ) 0.06), OA300 (φ ) 0.07), and MA300 (φ ) 0.07) samples. Barring small deviations as observed in the case of G′′, both moduli are fairly independent of frequency. G′ > G′′ suggests that dispersion of A300, OA300, and MA300 in m-PEG-LiClO4 results in a composite electrolyte with intrinsic elasticity. The degree of elasticity is found to be dependent on the surface functionalities. At 1 rads-1, G′ approximately equals 120, 60, and 50 kPa for A300, OA300, and MA300 composites, respectively. It is interesting to note that the composite with A300 (φ ) 0.06) particles had a higher elastic strength than the composite containing MA300/OA300 (φ ) 0.07). This favorable mechanical property of liquids resulting from dispersion of A300 was already utilized in the assembly of lithium cells without any conventional separators in ref 4. The frequency-independent response of both moduli also demonstrates the stability of the composites over a wide range of time scales. This strongly suggests that A300, OA300, and MA300 composites are physical gels, where silica particles are interconnected via physical bonds to form a space-filling network structure.12,14 In contrast to A300, OA300, and MA300, for NA300, composite G′′ is greater than G′, and both moduli display strong frequency dependence in the measured frequency

Role of Oxide Composition on “Soggy Sand” Electrolytes

J. Phys. Chem. B, Vol. 114, No. 20, 2010 6833

SCHEME 1: Schematic Representation of Various Interactions Prevailing in the Composite Systems: (a) A300, (b) MA300, (c) OA300, and (d) NA300

range. The viscous liquid-like behavior of NA300 is also reflected by the dynamic rheology experiment. The NA300 composite resembles a sol consisting of nonflocculated silica units.16 Thus the findings from dynamic rheology experiments are consistent with those from steady-state rheology experiments for all composites. We now discuss in greater detail the particle network in various composites. The chemical composition of the silica particle surface solely determines the macroscopic mechanical consistency of the composites. A gel composite electrolyte as in the case of A300, OA300, and MA300 results essentially from the formation of an attractive particle network spanning the electrolyte dimensions. The absence of such a spanning particle network leads to a sol, as observed in the case of NA300 composite. We interpret the particle configuration network in A300, OA300, MA300, and NA300 composites in terms of van der Waals forces or chemical (hydrogen/covalent) bonding existing between adjacent oxide surface chemical moieties and solvent molecules. We suppose that an attractive network solely due to overlap of space charge layer is not probable in the symmetric case of an AB (i.e., LiClO4) salt.17 Network formation via space charge overlap is more probable in the case of salts with higher valency of the type AB2 and so forth, where charge ordering may lead to an attraction between adjacent particles. The different kinds of colloidal interactions prevailing in the composite systems are depicted in Scheme 1. In the case of A300, which shows gel behavior even at low silica concentrations (φ g 0.01) there are two possibilities for the formation of a spanning attractive particle network. The silanol (Si-OH)

group present on A300 interacts via hydrogen bonding either with solvent molecules or with the silanol group of adjacent silica particles. In this context, the chemical nature of the solvent is highly significant. In the case of a highly polar solvent (ε f εwater ) 80, say), the solvent molecules organize in the vicinity of silica particles possibly via hydrogen bonding with the silanol groups on the silica surface. The layer of solvent molecules (or solvation layer) provides short-range repulsive forces between silica particles resulting in sol morphology. On the other hand, due to negligible interaction between solvent molecules and surface silanol groups, no solvation layer exists in the case of nonpolar solvent. At an optimum silica concentration, i.e., when φ ≈ 0.01 (onset for gel formation), a three-dimensional particle network is formed via hydrogen bonding between surface silanol groups of adjacent silica particles.18-22 The existence of such a particle network results in a colloidal gel, which agrees well with the conclusions drawn from static and dynamic rheology of the composite comprising hydrophilic A300 in m-PEG (ε ) 10.9)/LiClO4. As demonstrated in our previous work, the dominance of the interaction between surface silanol groups of adjacent silica particles over the solvent-surface interaction exists also in slightly more polar solvents such as ethylene glycol (EG; ε ) 37.7), leading again to gel morphology.6 However, due to a lesser degree of screening, the particle network in the case of the less polar m-PEG is stronger compared to the more polar EG (Figure 1B).6 It is interesting to note that, in spite of the presence of a strong inter particle network, no enhancement of conductivity was observed for the A300 composite electrolyte (Figure 3). This is contrary to the observation in the case of

6834

J. Phys. Chem. B, Vol. 114, No. 20, 2010

Figure 3. Variation of effective composite (0.1 M LiClO4/m-PEG: aerosil-SiO2) ionic conductivity as a function of silica volume fraction (φ) at 25 °C, as obtained from AC impedance spectroscopy (A300: as received, MA300: methyl-capped A300, NA300: amine-capped A300, and OA3000: octyl-capped A300).

EG-LiClO4-A300 composite, where a typical percolation-like ionic conductivity was observed1,6 (percolation thresholds in ionic conductivity: φonset ) 0.01, φmax ) 0.02, φcutoff > 0.07). There are possibly two reasons for the absence of nonpercolative conductivity. One reason may be that the hydrogen bonding between adjacent particles may consume all silanol groups, resulting in no dissociation and trapping/adsorption of anions (X-) on the silica surface. Another reason could be that, due to close proximity, the particles may physically be in contact with each other, resulting in blocked pathways. Both these states of aggregation result in the absence of any conducting species in the region between particles. This is also supported via ζpotential measurements, which estimated a potential nearly equal to 0 mV. This strongly suggests the absence of residual free silanols on the silica surface. Static and dynamic rheology measurements as discussed earlier clearly demonstrated the formation of an attractive particle network in MA300, and OA300 shows the formation of network structure. Both composites comprise silica particles with surfaces being predominantly hydrophobic in character. Therefore, the microstructure of MA300 and OA300 can be explained primarily on the basis of van der Waals (vdw) interaction as shown in Scheme 1b,c. To understand the existence and magnitude of vdw forces between hydrophobic silica particles, we estimated the Hamaker constant (Aeff)12,15,23 using the following equation: 2 3hνe (nm2 - nδ2)2 3 (∈m - ∈δ) Aeff ) kT + 2 2 3/2 4 (∈ + ∈ )2 16√2 (nm + nδ ) m δ

Here, ∈ is the dielectric constant, and n is the refractive index. The subscript δ refers to the functionalized group layer, and m refers to continuous medium, k is the Boltzmann constant, T is the absolute temperature, h is Planck’s constant, and νe ) 3 × 1015 s-1. m-PEG has ∈ ) 10.9 and n ) 1.455 at 20 °C.24 For OA300, considering the value of n-octane, ∈ )1.95 and n ) 1.387 at 20 °C,23 and for MA300, considering the value of methane, ∈ ) 1.00079 and n = 1.00029 at 27 °C,25 the estimated value of Aeff is found to be ∼10-21 J for both OA300 and MA300. This is typical for vdw interaction in a nonpolar medium.12,15,23 Although the zero shear viscosities of MA300 and OA300 are the same, Figure 1A clearly shows steep shear thinning in the case of MA300 in the shear rate range (10°-102)

Das and Bhattacharyya s-1 compared to OA300, suggesting a weaker particle network in the former with respect to the latter. This is also supported via dynamic rheology (Figure 2), which shows a higher elastic modulus for OA300 compared to MA300 over the measured frequency range. The differences in the strength of the particle network arise out of the interaction between the solvent molecules and the remaining surface silanols. The estimated percentage coverage of tethered methyl and octyl groups on A300 are 67% and 56%, respectively. The smaller methyl groups are incapable of providing an effective steric barrier to the solvent molecules for accessing the remaining 33% of silanol groups, and the solvent molecules possibly form hydrogen bonds via the functional group [OH] of the polyether chain [m-PEG: HO-(CH2-CH2-O)n-CH3] with the surface silanols. The formation of hydrogen bonding between the glycol and silica is detrimental for stronger interparticle networks, and, as a result, the MA300 composite can not sustain high shear rates, as observed in Figure 1. On the other hand, although the percentage of residual silanols in OA300 is high, the tethered large octyl groups are expected to remain close to the silica surface, forming a dense hydrophobic layer and thus preventing the solvent molecules from interacting with the surface silanols. As a consequence, the vdw interaction is stronger for OA300 compared to MA300 and results in a stronger network formation. Similar observations were reported in ref 12 with regard to octylfunctionalized fumed silica (R805, Degussa) in polyethylene glycol [HO-(CH2-CH2-O)n-H] (ε ) 15.6), polythylene glycol monomethyl ether [HO-(CH2-CH2-O)n-CH3] (ε ) 10.9), and polyethylene glycol dimethyl ether [CH3-O(CH2-CH2-O)n-CH3] (ε ) 7.9). As OA300 and MA300 form attractive particle networks, both are expected to exhibit enhancement in conductivity. However, similar to our previously published results,6 we observed that percolation-like ionic conductivity in the composite electrolytes is dependent on the tethered alkane chain length. In spite of a higher number of residual silanols in OA300 ([OH] ) 44%) in comparison to MA300 ([OH] ) 33%), no conductivity enhancement is observed in the case of OA300. The residual free silanols on the silica surface breaks the ion-pairs and adsorbs anions (X-) on the surface. This results in an increase in the concentration of free cations (M+) in the vicinity of silica surface. At silica concentrations equal to the onset percolation threshold, a percolating particle network is formed, and excess cations in the space charge layer can percolate freely through this network, resulting in enhancement of ionic conductivity compared to the nonaqueous salt solution. At the offset percolation threshold, blockage of percolation pathways occurs due to particles being in contact. Although the free silanols present on the OA300 surface adsorb anions (ζ-potential ) -36.80 mV), the dense hydrophobic layer provides a steric barrier to the movement of free cations (M+) resulting in nonpercolative ion transport in spite of the existence of a spanning particle network. As the small methyl groups are less effective as steric barriers to ion movement compared to the octyl groups, enhancement in ionic conductivity is observed in the case of MA300 (ζ-potential ) -27.55 mV). The ζ-potential values are quite consistent with density of silanol groups. As depicted in Figure 1d, the solvation layer plays a key role in sol formation in the case of NA300. Due to the strong polar nature of the tethered amino propyl group, the solvent molecules [m-PEG: HO-(CH2-CH2O)n-CH3] (via interaction either between the OH end group) instantaneously organize themselves in the vicinity of the silica surface. Therefore a solvation layer is expected to cover up the silica particles. Nonflocculation of silica particles occurs due

Role of Oxide Composition on “Soggy Sand” Electrolytes to a short-range repulsive force, which arises from the solvation layer. This stable nonflocculated silica units form the sol in NA300, similar to what has been already reported in the literature.26-29 There is no enhancement in ionic conductivity in NA300 due to adsorption of cations (M+) by the amine groups (Figure 3). The measured ζ-potential for NA300 was found to be +3.32 mV, suggesting the adsorption of positive charge on the silica surface. 4. Conclusions We have demonstrated here the importance of particle configuration and the nature of solvent in determining both mechanical strength and ion transport of “soggy sand” electrolytes. The nature and strength of interaction between particles or a particle with solvent is influenced by the particle surface chemical moiety and properties of the solvent. Although spanning network of electrolyte dimensions is a prerequisite, our studies reveal that size and concentration of the surface chemical moiety is also highly important in determining ionic conductivity. We believe that the present study will further help in fundamental understanding of ion transport in “soggy sand” electrolytes and lead to better optimization of such electrolytes for various electrochemical applications, especially lithium batteries. Acknowledgment. The authors thank I. S. Jarali for TGA and FTIR, Amit Mondal (INI-IISc., Bangalore) for TEM and CEN, and IISc, Bangalore, for glovebox facilities. We also thank S. Balakrishnan and N. Gopalakrishnan of TA Instruments, Bangalore, for rheology measurements, Vikas Rane, Evonik, India, for aerosil samples, and AIMIL Ltd., Bangalore, for the ζ-potential measurements. Supporting Information Available: TGA, FTIR spectroscopy, and transmission electron micrographs of various aerosil particles were employed for preparing the composites. Water effect on the ionic conductivity of the composite is shown. This material is available free of charge via the Internet at http:// pubs.acs.org.

J. Phys. Chem. B, Vol. 114, No. 20, 2010 6835 References and Notes (1) Bhattacharyya, A. J.; Maier, J. AdV. Mater. 2004, 16, 811. (2) Bhattacharyya, A. J.; Maier, J.; Bock, R.; Lange, F. F. Solid State Ionics 2006, 177, 2565. (3) Das, S. K.; Bhattacharyya, A. J. under preparation, 2010. (4) Bhattacharyya, A. J.; Dolle´, M. Maier. Electrochem. Solid-State Lett. 2004, 7, A432. (5) Maier, J. Prog. Solid State Chem. 1995, 23, 171. (6) Das, S. K.; Bhattacharyya, A. J. J. Phys. Chem. C 2009, 113, 6609. (7) Croce, F.; Settimi, L.; Scrosati, B. Electrochem. Commun. 2006, 8, 364. (8) Bunde, A.; Dieterich, W. J. Electrochem. 2000, 5, 81. (9) Bunde, A.; Dieterich, W.; Roman, H. E. Phys. ReV. Lett. 1985, 55, 5. (10) Jarosik, A.; Kaskhedikar, N.; Traub, U.; Bunde, A.; Maier, J. Conductivity and stability of particle networks in liquids. International Workshop on Fundamentals of Lithium-Based Batteries, Schloss Ringberg, Tegernsee, Germany, 23rd-28th Nov, 2008. (11) Technical Bulletin Fine Particles No. 11, Degussa Corporation. (12) Raghavan, S. R.; Riley, M. W.; Fedkiw, P. S.; Khan, S. A. Chem. Mater. 1998, 10, 244. (13) Reed, J. S. In Principles of Ceramics Processing, 2nd ed.; John Wiley & Sons, Inc.: New York, 1995; p 152. (14) Macosko, C. W. Rheology: Principles, Measurements and Applications; VCH Publishers: New York, 1994. (15) Raghavan, S. R.; Walls, H. J.; Khan, S. A. Langmuir 2000, 16, 7920. (16) Atkins, D. T.; Ninham, B. W. Colloids Surf. 1997, 129, 23. (17) Moreira, A. G.; Netz, R. R. Phys. ReV. Lett. 2001, 87, 078301. (18) Khan, S. A.; Zoeller, N. J. J. Rheol. 1993, 37, 1225. (19) Korn, M.; Killmann, E.; Eisenlauer, J. J. Colloid Interface Sci. 1980, 76, 7. (20) Kiraly, Z.; Turi, L.; Dekany, I.; Bean, K.; Vincent, B. Colloid Polym. Sci. 1996, 274, 779. (21) Ketelson, H. A.; Pelton, R.; Brook, M. A. Langmuir 1996, 12, 1134. (22) Depasse, J. J. Colloid Interface Sci. 1997, 194, 260. (23) Israelachvili, J. Intermolecular and Surface Forces, 2nd ed.; Academic Press: San Diego, 1991. (24) Web site of Sigma Aldrich: http://www.sigmaaldrich.com/catalog/; methoxy polyethylene glycol 350, product number M6768 (Sigma) or 202479 (Aldrich). (25) Younglove, B. A.; Ely, J. F. J. Phys. Chem. Ref. Data 1987, 16, no 4. (26) Besseling, N. A. M. Langmuir 1997, 13, 2113. (27) Forsman, J.; Woodward, C. E.; Jo¨nsson, B. J. Colloid Interface Sci. 1997, 195, 262. (28) Kiraly, Z.; Turi, L.; Dekany, I.; Bean, K.; Vincent, B. Colloid Polym. Sci. 1996, 274, 779. (29) Ketelson, H. A.; Pelton, R.; Brook, M. A. Langmuir 1996, 12, 1134.

JP102548E