Ionization Condition of Lithium Ionic Liquid Electrolytes under the

Feb 28, 2008 - In case of the silica dispersion, rLi decreased to 0.7 nm at 10 wt % silica. Although the silica surface with silanol groups fundamenta...
12 downloads 9 Views 462KB Size
Download PDF
J. Phys. Chem. B 2008, 112, 3357-3364

3357

Ionization Condition of Lithium Ionic Liquid Electrolytes under the Solvation Effect of Liquid and Solid Solvents Tatsuya Umecky,† Yuria Saito,*,† Yasue Okumura,† Seiji Maeda,‡ and Testuo Sakai† National Institute of AdVanced Industrial Science and Technology, 1-8-31 Midorigaoka, Ikeda, Osaka 563-8577, Japan, and The Nippon Synthetic Chemical Industry Company Limited, 2-13-1 Muroyama, Ibaraki, Osaka 567-0052, Japan ReceiVed: December 10, 2007; In Final Form: December 20, 2007

Ionization condition and ionic structures of the lithium ionic liquid electrolytes, LiTFSI/EMI-TFSI/(PEG or silica), were investigated through the measurements of ionic conductivity and diffusion coefficient. The size of the hydrodynamic lithium species (rLi) evaluated from the Stokes-Einstein equation was 0.90 nm before gelation with the PEG or silica. This reveals that the TFSI- anions from the solvent are coordinated on Li+ for solvation, forming, for example, Li(TFSI)43- and Li(TFSI)2- in the electrolyte solution. By the dispersion of PEG for gelation, rLi increased up to 1.8 nm with the 10 wt % of PEG. This indicates that the lithium species is directly interacted with the oxygen sites on the polymer chains and the lithium species migrate, reflecting the polymer by hopping from site to site. In case of the silica dispersion, rLi decreased to 0.7 nm at 10 wt % silica. Although the silica surface with silanol groups fundamentally attracts Li+, the lithium does not migrate from site to site on the silica surface as in the gel of the polymer and follows random walk behavior in the network of the liquid-phase pathways in the two-phase gel. In the process, that solvated TFSI- anions are partially removed may be due to the attractive effect of H+, which was dissociated from the silanol group. It is concluded that the dispersed silica is effective to modify the hydrodynamic lithium species to be appropriate for charge transport as reducing the size and anionic charge of Li(TFSI)43- by removing one or two TFSI- anions.

Introduction Ionic liquids have been actively applied to electrochemical devices as electrolyte solvents and capacitors recently due to their high safety performance characterized by the low vapor pressure and non-flammability.1 Taking advantage of the merit, substitution of the ionic liquid solvent for the combustible organic solvent as propylene carbonate (PC) or the mixture of ethylene carbonate and diethyl carbonate (EC/DEC) has been actively studied in the field of lithium secondary batteries.2,3 To design the conductive materials controlled in ionic mobility and chemical stability in the charge transport system, it is significant to learn not only the dynamic property of the carrier species but also the static features responsible for the carrier transport, the ionization condition of the dissolved salt, and the solvation condition of the dissociated species. The peculiar feature of the lithium ionic liquid electrolytes different from the conventional electrolytes with the organic solvent is that the solvent itself is ionized in the solution. The dissociation degrees of the solute and solvent salts are associated with each other depending on their mixing concentration ratio. This complicated situation makes it difficult to imagine the existing state of the individual species and the factors responsible for ion mobility. We had previously proposed a new approach to evaluate the dynamic and static features of electrolyte materials using the originally developed electric field by applying NMR technique.4 By the measurement of the ionic mobility (µobs) and * Corresponding author. Telephone: +81-72-751-4527. Fax: +81-72751-9623. E-mail: [email protected]. † National Institute of Advanced Industrial Science and Technology. ‡ The Nippon Synthetic Chemical Industry Co. Ltd.

diffusion coefficient (Dobs), inherent ionic mobilities, µcation and µanion, dissociation degree of the lithium salt, x, and interactive force between the lithium and the effective site of the solvent, γ, are determined simultaneously through the technique. If we apply this to the ionic liquid electrolyte, however, it would be impossible to evaluate the individual static and dynamic value because the interactive conditions among the species in the electrolyte are too complicated to be modeled for analyses. In addition, application of the electric field at the measurement has a risk to change the equilibrium state of the electrolyte because not only the ions from the solute lithium salt but those from the solvent drift in response to the applied electric field. In this situation, it is acceptable that the Dobs is the most reliable for the guide of the dynamic feature of the probed nuclear species because it can be measured without any perturbation, which affects the state of the electrolyte. To be exact, Dobs has contributions from the dissociated ion and associated ion pair coexisting in the electrolyte.4,5 We then have to suggest new analytical ideas for the evaluation of Dobs to obtain more detailed information concerning the ionization condition and structure of the individual ionic species. One of the effective analytical ways was to take the ratio of Dobs of the cation and anion species, DLiobs/DFobs of the solute lithium salt and DHobs/DFobs of the ionic liquid solvent.6 According to the study, we revealed that the Li+ cation is solvated by the anions of the ionic liquid solvent, consequently forming an anionic species of lithium such as Li(TFSI)32- and Li(TFSI)43- (TFSI- : N(CF3SO2)2-, bis(trifluoromethanesulfonyl)imide). From the aspect of the lithium cell operation, the larger anionic lithium solvated by the anions is not suitable as the carrier species because the lithium cations move from the

10.1021/jp711625r CCC: $40.75 © 2008 American Chemical Society Published on Web 02/28/2008

3358 J. Phys. Chem. B, Vol. 112, No. 11, 2008

Umecky et al.

TABLE 1: List of Samples and Their Macro Viscosity, Lithium Concentration in the A Type Gels, and Molar Concentration Ratio of Li+ to the Oxygen Sites on the Solid Dispersant

mark

content

η/mPa‚s at 50 °C

A0 A-P1 A-P3 A-P5 A-S5 A-S10 B0 B-P3 B-P5 B-P10 B-S5 B-S10

LiTFSI/EC+DEC A0 + PMMA 1 wt % A0 + PMMA 3 wt % A0 + PMMA 5 wt % A0 + silica 5 wt % A0 + silica 10 wt % LiTFSI/EMI-TFSI B0 + PEG 3 wt % B0 + PEG 5 wt % B0 + PEG 10 wt % B0 + silica 5 wt % B0 + silica 10 wt %

4.3 17 120 1550 3.1 × 107 1.5 × 108 36 43 47 64 2.6 × 104 4.1 × 105

a

F/g‚cm-3 at 25 °C

A0 solution content in the gel [g/(100 cm3)]

LiTFSI content in the gel [mol/(100 cm3)]

[Li+]/[-Obulk]

[Li+]/[-Osurface]a

1.49 1.49 1.49 1.38 1.41

147.5 144.5 141.6 131.1 126.9

0.514 0.503 0.493 0.460 0.442

3.24 1.06 0.62 0.37 0.18

0.42-0.62 0.20-0.30

0.50 0.29 0.14 0.40 0.19

0.45-0.67 0.21-0.32

The values are estimated assuming the spherical-cubic particles.

anode to the cathode for discharging and vice versa for charging. This result urges us to remove the clothing on the lithium to be more mobile and cationic in the ionic liquid solvent. It is generally accepted that the gelation of liquid electrolytes using the solid solvent is one of the significant and practical approaches for the preparation of self-supporting electrolytes eliminating the liquid leakage. Composite electrolytes based on fumed silica, poly(ethylene glycol), and lithium salt have been prepared by several groups to improve the electrochemical stability, compatibility with lithium metal, low volatility, and high processability for the fabrication of lithium secondary batteries.7 In the case of lithium ionic liquid electrolytes, however, we found that dispersion of the solid solvent was an inevitable approach to prepare lithium species suitable for the charge transporting species.6 The polar sites on the dispersant interact with the lithium competitive to the solvated TFSIanions. That leads to the change of the structure, size, and charge of the lithium species, and the consequent migration mechanism. Especially, we suggested that the silica was effective to release the solvated TFSI- anions from the lithium, making lithium species cationic and more mobile in the ionic liquid solvent. It is expected that we will design the lithium ionic electrolytes with the suitable form of lithium for charge transport by the selection and modification of the solid solvent. In this research, we propose a new and more quantitative approach to evaluate the ionic condition in the electrolytes. We estimate the size and represent the structure of the individual ionic species in the lithium ionic liquid electrolytes using the observed diffusion coefficients of the species. We discuss the change of the size of the lithium species by gelation with the polymer or silica associated with their interactive effect. We also show that the ionization condition of the lithium salt is sensitively affected by the solid solvent through the evaluation of the conductivity ratio. Experimental Section To investigate the solvent effects on the ionization condition of the dissolved lithium salt, we started from two electrolyte solutions: A0, 1 M LiTFSI/(EC+DEC) (EC:DEC ) 1:1 in vol. ratio); B0, 20 wt % LiTFSI/EMI-TFSI (Z1 M LiTFSI/EMITFSI). The A0 solution with the conventional organic solvent was selected as a standard to compare the solvation effect with that of the B0 solution. Gel electrolytes of the solutions were prepared by adding a polymer or silica to elucidate the ionization condition change by the solid solvent. To prepare homogeneous gels with polymer dispersion, we tried some kinds of polymer

and finally selected polymethyl methacrylate (PMMA, molecular weight, M ) 6000) from Kishida Chemicals as the dispersant of the A0 solution and polyethylene oxide dimethyl ether (PEG, M ) 2000) from Aldrich for the B0 solution. The gels were prepared by mechanical mixing at 80 °C with several mixing concentrations, A0 + 1,3,5 wt % of PMMA (A-P type electrolytes) and B0 + 3,5,10 wt % of PEG (B-P type electrolytes). For the gels with silica, the hydrophilic nanosize silica powder (AERO-SIL300, NIPPON AEROSIL, primary particle size; ∼7 nm, specific surface area 300 ( 30 m2/g) was mixed with the A0 in the concentrations 5, 10 wt % to form homogeneous gels (A-S type electrolytes). For gelation of the B0 solution, the silica powder was first mixed with EMI-TFSI solvent to be a homogeneous gel, and then LiTFSI was dissolved in it to complete the gel electrolyte (B-S type electrolytes). The electrolytes prepared are summarized in Table 1. Ionic conductivity was measured by the complex impedance measurements using the Solartron frequency analyzer, 1250B combined with the potentio/galvano stat 1287 in the frequency range of 0.1 Hz to 65 kHz and in the temperature range of 2550 °C. Viscosity was measured using an AR-1000 rotational type rheometer (TA Instruments) in the temperature range from 25 to 60 °C for the A0, A-P, B0, and B-P electrolytes. Cone and plate geometry for sandwiching the sample was 2° and 4 cm in diameter. Applied shear stress was kept constant, 20 Pa for A-P, B0, B-P samples and 5 Pa for A0 sample by controlling the shear rate. Viscosity was estimated from the product of shear stress and shear rate. It was difficult to reliably measure the temperature dependence of viscosity of the gels with silica (A-S and B-S types of electrolytes) due to the high viscosity values over the ability of the instrument. For reference, however, we measured the viscosity of the gels with silica at 50 °C. Diffusion coefficients of the ionic species were measured using the pulsed gradient spin-echo NMR (PGSE-NMR) technique with a JNM-ECP300W spectrometer and wide-bore probe units adjusting the probed nuclear species 1H (300.5 MHz), 7Li (116.8 MHz), and 19F (282.7 MHz). The spin-echo pulse sequence was used for measurements of diffusion coefficients, DLiobs contributed from lithium species, DFobs from TFSI- species, and DHobs from the solvent species of A type electrolytes and EMI+ species of B type electrolytes.8 The half sine-shaped gradient pulse was applied twice in sequence after the first and second 180° pulses to detect attenuation of echo intensity according to the diffusive migration of the probed species. The typical values of the field gradient pulse parameters

Ionization of Lithium Ionic Liquid Electrolytes

Figure 1. Temperature dependence of ionic conductivity of (a) A type electrolytes and (b) B type electrolytes. Eσ is the activation energy for conduction.

were g ) 1-10 T/m for the pulse strength, δ ) 0 - 7 ms for the pulse width, and ∆ ) 50 ms for the interval between the two gradient pulses. The diffusion coefficient was measured in the temperature range 25-70 °C. For the A0 solution, observed diffusion coefficients anomalously increased deviated from the Arrhenius relation above 35 °C. This is due to the thermal convection due to the low solution viscosity following the composition change of the volatile solvent. Therefore, we exceptionally measured the diffusion values of the A0 solution in the temperature range of 0-30 °C. The 1H signal of the A type electrolytes has two components from EC and DEC. To evaluate diffusivity of the solvent species independent of other ionic species, we used the peak of DEC because EC has a higher dielectric constant and tends to coordinate to Li+ for solvation to form, for example, Li(EC)4+.9 The 1H signal of the B type electrolytes comprises several peaks attributed from the proton configuration on the imidazolium cation species.10 We then averaged the diffusion coefficient values estimated from each peak attenuation. To illustrate the ionic and molecular structures of the carrier species, we first arranged the TFSI- and EC around an Li+ to form a temporary structure according to the diffraction and MD results,9,11-15 and then optimized the structure by the calculation of the potential energies of the species through the MM2 route provided by the ChemBats3D Ultra 10.0 (ChemOffice CambrigeSoft). Results and Discussion Dynamic Features, Viscosity, Ionic Conductivity, and Diffusion Coefficient. Figures 1 and 2 represent the Arrhenius plots of ionic conductivity (σ) and macro viscosity (η), respectively, of the A and B types electrolytes. Ionic conductivity of the A type electrolytes decreased with the increase in the polymer or silica content in the gel. Activation energies for conduction (Eσ) were almost constant, ∼13 kJ/mol independent of the type and concentration of the dispersant. Macro viscosity

J. Phys. Chem. B, Vol. 112, No. 11, 2008 3359

Figure 2. Temperature dependence of macro viscosity of (a) A type electrolytes and (b) B type electrolytes. Eη is the activation energy for viscous flow.

of the A-P type electrolytes increased with PMMA content correlated with the change of conductivity. Activation energies for viscous flow (Eη) of them were in the range of 14-17 kJ/ mol, which were slightly larger than Eσ. This indicates that the rate-determining step and mechanism between the ionic species responsible for conductivity and the neutral species contributing to the viscosity are different in the A-P type electrolytes.16 Viscosity of the gel with the silica (A-S type electrolytes) was higher in more than 4 orders of magnitude as compared to the gel with PMMA as shown in Table 1 despite the same order of conductivity between the A-P and A-S types electrolytes. This is attributed to the difference of the fundamental ion migration mechanism associated with the gel morphology, which dominates the pathways for the ion transport. Ionic conductivity of the B type electrolytes did not show the apparent change with increasing PEG for gelation. Ionization degrees of both the solute and the solvent of the B type electrolytes are influenced by the solid dispersant.6 Therefore, the conductivity change independent of the dispersant reveals that the total concentration of the ionic species increased contrary to the macro viscosity increase, which leads to the mobility decrease by the PEG dispersion. Ionic conductivities of the B-S type electrolytes were lower than that of the B0 solution. This is mainly due to the high viscosity of the gel by the silica dispersion. To elucidate the dynamic properties of the individual species in the electrolyte, we measured diffusion coefficients of the Li+ species (DLiobs), TFSI- species (DFobs), and the solvent species of the DEC of the A type electrolytes and EMI+ of the B type electrolytes (DHobs). It should be noted that the observed diffusion coefficient has contributions from not only the dissociated ions but also the associated ion pairs in the electrolyte.4,5 Figure 3 represents the Arrhenius plots of DLiobs, DFobs, and DHobs of the A type electrolytes. Diffusion coefficient values of all species gradually decreased with increasing the concentration of the dispersant. This mainly reflects the reduction of mobility of the ionic species due to the macro viscosity

3360 J. Phys. Chem. B, Vol. 112, No. 11, 2008

Umecky et al.

Figure 4. Temperature dependence of conductivity ratio, Λimp/ΛNMR, of the A type electrolytes. Λimp, molar conductivity obtained from the impedance measurement; ΛNMR, conductivity evaluated from the DLiobs and DFobs using the Nernst-Einstein equation. In case of A0 + silica 10 wt %, the ratio exceeds 1, which is in conflict with the definitions of Λimp and ΛNMR. This is probably due to the fact that DLiobs and DFobs are reduced by restriction of the physical barrier effect of a large amount of silica, although the restricted condition is not so strong to make the species deviate from the random walk behavior.17

Figure 3. Temperature dependence of diffusion coefficients: (a) DLiobs, (b) DFobs, and (c) DHobs of the A type electrolytes. ED is the activation energy for diffusion.

increase. Comparing the samples, A-P5 (A0 + 5 wt % PMMA) and A-S10 (A0 + 10 wt % silica), it is noted that DLi(A-S10) < DLi(A-P5), DH(A-S10) < DH(A-P5), and DF(A-S10) < DF(A-P5), although σ(A-S10) > σ(A-P5). When we evaluate the lithium content in the unit volume of the A type gels (Table 1), we found that the lithium molar concentration of the A-P5 was higher than that of the A-S10. Therefore, the result of σ(AS10) > σ(A-P5) despite the higher concentration of LiTFSI and higher diffusivity of all species in the A-P5 indicates that the net carrier content contributing to the conduction is larger in the A-S10 as compared to that of the A-P5. To demonstrate this situation, we evaluated the dissociation condition of LiTFSI in the electrolytes using the conductivity ratio, Λimp/ΛNMR, estimated from the impedance and NMR measurements in Figure 4. We have previously revealed that the increase of Λimp/ ΛNMR represents the progress of the lithium salt dissociation.18 According to the interpretation, we can find from Figure 4 that the gelations of the A0 solution with the silica and PMMA enhance the lithium salt dissociation. Especially, the dissociation effect of the silica is stronger than that of the PMMA according to the Λimp/ΛNMR of the A-S type electrolytes. It is generally accepted that the surface of silica forms silanol groups such as Si-OH and Si-(OH)2.19 The acidic surface forms negative sites accompanying adsorption of Li+ in the gel electrolytes. Simple estimation of the Li+ concentration per an oxygen site was listed in Table 1. Assuming that the surface sites of the silica particles are effective for the interaction with Li+, we can see that [Li+]/[-O] of the A-S5 was comparable to that of the A-P5. Next, we can roughly conclude that the larger values of Λimp/

Figure 5. Temperature dependence of diffusion coefficients: (a) DLiobs, (b) DFobs, and (c) DHobs of the B type electrolytes. ED is the activation energy for diffusion.

ΛNMR of the A-S type electrolytes are associated with the stronger attraction of the silica site for Li+ than of the PMMA site. Figure 5 represents the temperature dependence of observed diffusion coefficients of the B type electrolytes. We can find characteristic features of the ionic liquid electrolyte from this

Ionization of Lithium Ionic Liquid Electrolytes

J. Phys. Chem. B, Vol. 112, No. 11, 2008 3361

TABLE 2: van der Waals Volume and Radius of DEC and EMI+ Species 1H

species

DEC EMI+ a

VvdW/nm3

rvdW/nm

0.111a 0.116b

0.298 0.303

Reference 23. b Reference 24.

result. Concerning the B-P type electrolytes, DHobs and DFobs were independent of the PEG dispersion. On the other hand, DLiobs apparently decreased with increasing PEG content. The independent features of DFobs and DHobs on PEG dispersion are similar to that of the conductivity represented in Figure 1. The peculiar changing feature of DLiobs with PEG dispersion indicates that the ether oxygen sites on PEG selectively attract Li+ species to restrict the free migration that is performed in the solution.18,20 It is noted that the positively charged EMI+ species were not influenced by the active sites on PEG, which may be due to the steric hindrance of the larger size of EMI+ as compared to Li+. In case of the B-S type electrolytes, DLiobs became higher than that of the B0 solution. This may be due to the selective interaction of the silica sites on the Li+, which is discussed later. We can also find that the decreases of DHobs and DFobs are fairly small against the presumption from the viscosity change of several orders of increase by the silica dispersion. This makes us imagine the phase separated gel morphology in which the carriers migrate through the liquid network of pathways among the silica particles. It is not easy to evaluate the ionization rate of the lithium salt in the ionic liquid solvent as that of the A type electrolytes because the ionization conditions of the solute and solvent are correlated with each other depending on their mixing concentration. We will promote discussion on this in future work. Ionic Structure and the Size Evaluation. It is confirmed that the Li+ ions are stabilized by forming the solvated species such as Li(EC)n+ and Li(TFSI)n+1n- in the A0 and B0 solutions, respectively.6 On the analogy of it, it is acceptable that the lithium species in the gel is solvated by the solid solvent through the interactive effect of the oxygen sites on the polymer or silica. It is expected that the anomalous decrease of DLiobs as compared to the constant DHobs and DFobs of the B0 solution by the PEG dispersion comes from the selective enlargement of the lithium species by the solvation of the PEG. Further, enhanced dissociation of the A0 solution by the solid dispersion would reflect the solvation effect of the PMMA and silica. On the basis of the considerations until now, we decided to evaluate the structure and size of the hydrodynamic ionic species in the electrolyte quantitatively. Diffusion process of the hydrodynamic species in a medium is correlated with the viscous flow according to the Stokes-Einstein equation,

D)

kBT 6πrη

(1)

where D is the diffusion coefficient, kB is the Boltzmann constant, T is the absolute temperature, r is the radius of the hydrodynamic species in the continuous medium, and η is the viscosity of the medium. It should be noted that η is the local viscosity in the neighborhood of the species that should be taken rather than the macro viscosity.21 When this relation can be applied to the migration of the ionic or molecular species of the electrolytes, solvation effect is reflected on r. The local viscosity, η, can be estimated from the diffusion coefficient value of the isolated species independent of the solvation and the Coulombic effects in the electrolytes. In case of the A type

Figure 6. DLiobs versus T/η and DFobs versus T/η of the A type and B type electrolytes.

electrolytes, DHobs is contributed from DEC species, which is independent of the salt dissociation and solvation and is applicable for the η estimation. In case of the B type electrolytes, all species in the electrolyte are originated from the salts, and their ionization rates change depending on the concentrations

3362 J. Phys. Chem. B, Vol. 112, No. 11, 2008

Umecky et al.

Figure 8. (a) rLi and (b) rF as a function of solid dispersant content in the A type and B type electrolytes.

for viscosity standard due to the following reason. In the B0 solution, the molar ratio of Li+/TFSI- is 1/3.9. This means that most of TFSI- anions of EMI-TFSI coordinate to the Li+ for solvation because Li(TFSI)43- species are dominant in the LiTFSI/EMI-TFSI system.6,22 Consequently, isolated EMI+ cations are dominantly present in the electrolyte solution. Further, we had no evidence of the interactive effect of the polymer or silica sites on EMI+ contrary to their attractive effect on Li+ species from the observations of diffusion coefficient change by gelation. Therefore, we propose that the DHobs is contributed from the isolated EMI+ species and can be applied for the estimation of the viscosity of the B type electrolytes according to

η)

Figure 7. rLi versus T and rF versus T of the A type and B type electrolytes.

of LiTFSI, EMI-TFSI, and the solid dispersant. That means the contributing species on DHobs differ from gel to gel. Therefore, strictly speaking, there is no independent species available for the standard of viscosity in the B type electrolytes. In this situation, however, we suppose that EMI+ is an acting species

kBT 6πrHDHobs

(2)

in which rH values of DEC and EMI+ species were estimated from their van der Waals volumes assuming the spherical species, listed in Table 2. It is significant to first confirm the applicability of Dobs of each species of the electrolytes to the Stokes-Einstein relation. Figure 6 shows the plots of Dobs versus T/η. We can find linear changes of DLiobs and DFobs of each electrolyte. This proves that the migration of the hydrodynamic species approximated to be a spherical particle with the radius r is affected by the viscous resistance η from the surrounding medium. Distribution of the slope of the plots for DLiobs is wider than that of DFobs. This reveals the fact that the size of the Li+ species, which

Ionization of Lithium Ionic Liquid Electrolytes

J. Phys. Chem. B, Vol. 112, No. 11, 2008 3363

Figure 9. Ionic and molecular structures of lithium in the A type electrolytes (Li(EC)4+, TFSI-, LiTFSI) and B type electrolytes (LiTFSI, Li(TFSI)43-, Li(TFSI)2-).

corresponds to the magnitude of the slope, sensitively reflects the solvation effect of the liquid and solid solvents. On the basis of these preliminary confirmation, we estimated the apparent hydrodynamic radii of the hydrodynamic species, rLi and rF, using DLiobs and DFobs, respectively, represented in Figure 7. Figure 8 shows the changes of rLi and rF with the solid dispersant content. As DLiobs and DFobs have contributions from the ion and ion pair including the probed nuclei, 7Li and 19F, respectively, r and r represent the averaged size of the Li F contributing species.4 Therefore, the difference between rF ≈ 0.46 nm of the A0 solution and rF ≈ 0.65 nm of the B0 solution is attributed to the different contributing species, TFSI- and LiTFSI for the A0 solution and TFSI- and Li(TFSI)n+1n- for the B0 solution. Figure 9 illustrates the ionic and molecular structures of the typical species in the electrolytes.8,10-14 Estimated radii of the species from the optimized structures were r(TFSI-) ≈ 0.4 nm, r(LiTFSI) ≈ 0.4 nm, r(Li(TFSI)43-) ≈ 0.8 nm, and r(Li(TFSI)2-) ≈ 0.7 nm. Therefore, these values are well consistent with the rF values derived from the StokesEinstein relation in Figure 7. It is worth noting that the rF values of the B0 solution clearly prove the formation of the solvated species, Li(TFSI)43-, Li(TFSI)2-, which were initially presumed by the evaluation of DLiobs/DFobs.6 After the dispersion of the polymer or silica for gelation, rF did not change apparently as

shown in Figure 8. This means that the anionic species are not affected directly by the attractive sites on the dispersant. In case of lithium species, estimated rLi was 0.56 nm for the A0 solution and 0.90 nm for the B0 solution at 300 K. These values indicate that the hydrodynamic lithium species are coordinated by the EC molecules and TFSI- anions in the A0 and B0 solutions, respectively. Spectral and MD calculation results suggest that four EC molecules or TFSI- anions coordinate tetrahedrally or two TFSI- anions coordinate at the two oxygen sites of each TFSI- anion as in Figure 9.9,11-13,25 The sizes of the solvated species presumed from the optimized structures in Figure 9 are well consistent with the estimated values from DLiobs of the A0 and B0 solutions. By the gelation of the A0 solution with PMMA, rLi may have the tendency of increase with PMMA from Figure 8. Dispersion of silica in the A0 solution, on the other hand, slightly decreased rLi. This opposite effect between the PMMA and silica was more apparent in the B type electrolytes. The rLi of the B type electrolytes anomalously increased from 1.0 nm of the B0 solution to 1.8 nm with 10 wt % PEG and decreased to 0.7 nm by 10 wt % silica dispersion. Enlargement of the size of the lithium species with the PMMA and PEG would be attributed to the polymer interactive effect that the oxygen site attracts Li+ to form a coordinate bond. This is reasonable according to

3364 J. Phys. Chem. B, Vol. 112, No. 11, 2008 the fact that the lithium species migrate from site to site supported by the segmental motion of the polymer chains in the polymer electrolytes and the selective interaction between the Li+ and PEO type polymer are present even in the gel electrolyte, which includes the liquid solvent.18,20 That is, it is acceptable that the Li+ species transport in the gel, reflecting the interacting site of the polymer on the hydrodynamic size. On the other hand, reduction of rLi with the silica dispersion can be recognized as that the solvating EC or TFSI- species are partially released from the Li+. Further, we have seen that the lithium salt dissociation was promoted by the silica dispersion, which is due to the attracting effect of the silica surface O- on the Li+.19 Simultaneously, H+ species released from the silanol group of silica would interact with TFSI-, which contributed to the solvation on the Li+. This results in the reduction of rLi. However, the attracting effect of the silica on the Li+ species did not lead to the increase in rLi as the increasing behavior by the polymer dispersion discussed above. This is due to the difference in the Li migration mechanism between the gels with the polymer and silica. In the gel with the silica, lithium species do not transport by hopping from site to site on the silica but diffuse through the liquid-phase network of pathway, although the Li+ is actually influenced by the Osites on the silica particle surface. Larger changes of rLi of the B type electrolytes as compared to that of the A type electrolyte by the solid dispersion would reflect the strength of the coordinate bond of the EC and TFSI-. This is supported by the Raman study of lithium coordination in EMI-TFSI, in which it is confirmed that the lithium ion no longer interacts with the TFSI- when the EC species are added.25 It is expected that that the solvating TFSI- anions are easily released by the approach of the basic sites to the Li+ as compared to the EC on the Li+. As a result, Li+ on the B0 solution is effectively attracted on the PEG and silica, leading to the larger change of rLi as compared to that in the A type electrolytes. In conclusion, we found that the Li+ species are present with the solvated form as Li(TFSI)43- in the ionic liquid solvent. The solvation effect is weaker than that of the neutral solvent EC. As a result, the lithium is sensitively affected by the polymer sites in the gel with the PEG, behaving as mobile species solvated by the polymer. In case of silica dispersion, the oxygen sites on the silica surface attract the lithium to promote the lithium salt dissociation and affect Li+ migration. Consequently, it is probable that the removed H+ from the silanol groups approaches TFSI-, dissociating the solvating TFSI- species on the Li+. The different changing feature of rLi by the dispersion of the PEG (rLi increase) and silica (rLi decrease) is attributed to the difference in the ion migration mechanism. The lithium species in the gel with the polymer migrates from site to site of the oxygen along or across the polymer chains supported by

Umecky et al. the segmental motion of the polymer chains. Therefore, the way of the Li transport depends on the structure and morphology of the network of polymer chains. This is the reason why rLi strongly reflects the coordination of polymer. The lithium species in the gel with silica is also affected by the silica surface site. As a result, the dissociation of LiTFSI is induced and the solvated TFSI- anion on Li+ was removed by the H+ species on the silanol group. The lithium species diffuse in the liquid phase at random, although they are, to be exact, restricted in the diffusion rate due to the collision with the silica barrier. This is the reason why rLi is reduced by the silica dispersion. References and Notes (1) Ohno, H. In Electrochemical Aspect of Ionic Liquid; Ohno, H., Ed.; John Wiley & Sons, Inc.: New Jersey, 2005; p 1. (2) Webber, A.; Blomgren, G. E. In AdVanced in Lithium Ion Batteries; van Scalkwijk, W. A., Scrosati, B., Eds.; Kluwer Academic/Plenum Publishers: New York, 2002; p 185. (3) Hayashi, K.; Nemoto, Y.; Akuto, K.; Sakurai, Y. J. Power Sources 2005, 146, 689. (4) Kataoka, H.; Saito, Y.; Sakai, T.; Deki, S.; Ikeda, T. J. Phys. Chem. B 2001, 105, 2546. (5) Kataoka, H.; Saito, Y. J. Phys. Chem. B 2002, 106, 13064. (6) Saito, Y.; Umekey, T.; Niwa, J.; Sakai, T.; Maeda, S. J. Phys. Chem. B 2007, 111, 11794. (7) Fan, J.; Pedkiw, P. S.; Raghavan, S. R.; Kahn, S. A. In Proceedings of the Symposium on Lithium Polymer Batteries; Broadhead, J., Scrosati, B., Eds.; The Electrochemical Society, 1997; pp 54, 74. (8) Tanner, J. E. J. Chem. Phys. 1970, 2, 2523. (9) Borodin. O.; Smith, G. D. J. Phys. Chem. B 206, 110, 4971. (10) Saito, Y.; Hirai, K.; Matsumoto, K.; Hagiwara, R.; Minamizaki, Y. J. Phys. Chem. B 2005, 109, 2942. (11) Nowinski, J. L.; Lightfoot, P.; Bruce, P. G. J. Mater. Chem. 1994, 4, 1579. (12) Kameda, Y.; Umebayashi, Y.; Takeuchi, M.; Wahab, M. A.; Fukuda, S.; Ishiguro, S.; Sasaki, M.; Amo, Y.; Usuki, T. J. Phys. Chem. B 2007, 111, 6104. (13) Borodin, O.; Smith, G. D.; Henderson, W. J. Phys. Chem. B 2006, 110, 16879. (14) Lasse`gues, J.-C.; Grondin, J.; Talaga, D. Phys. Chem. Chem. Phys. 2006, 8, 5629. (15) Umebayashi, Y.; Mitsugi, T.; Fukuda, S.; Fujimori, T.; Fujii, K.; Kanzaki, R.; Takeuchi, M.; Ishiguro, S. J. Phys. Chem. B 2007, 111, 11794. (16) Bockris, J. O’M.; Reddy, A. K. N. Modern Electrochemistry: 2nd ed. Ionics; Plenum Press: New York and London, 1998; p 656. (17) Saito, Y.; Hirai, K.; Murata, S.; Kishii, Y.; Kii, K.; Yoshio, M.; Kato, T. J. Phys. Chem. B 2007, 109, 11563. (18) Kataoka, H.; Saito, Y.; Uetani, Y.; Murata, S.; Kii, K. J. Phys. Chem. B 2002, 106, 12084. (19) Foissy, A.; Persello, J. In The Surface Properties of Silica; Legrand, A. P., Ed.; John Wiley & Sons Ltd.: England, 1998; p 365. (20) Watanabe, M.; Ogata, N. Br. Polym. J. 1988, 20, 181. (21) Bockris, J. O’M.; Reddy, A. K. N. Modern Electrochemistry: 2nd ed. Ionics; Plenum Press: New York and London, 1998; p 453. (22) Borodin, O.; Smith, G. D.; Henderson, W. J. Phys. Chem. B 2006, 110, 16879. (23) Bondi, A. J. Phys. Chem. 1964, 68, 441. (24) Ue, M.; Murakami, A.; Nakamura, S. J. Electrochem. Soc. 2002, 149, A1385. (25) Hardwick, L. J.; Holzapfel, M.; Wokaun, A.; Nova´k, P. J. Raman Spectrosc. 2007, 38, 110.