Quaternary Ammonium Room-Temperature Ionic Liquid Including an

Jan 8, 2008 - Undoing Lithium Ion Association in Ionic Liquids through the Complexation by Oligoethers. Paul M. Bayley , G. H. Lane , L. J. Lyons , D...
2 downloads 0 Views 404KB Size
Download PDF
J. Phys. Chem. B 2008, 112, 1189-1197

1189

Quaternary Ammonium Room-Temperature Ionic Liquid Including an Oxygen Atom in Side Chain/Lithium Salt Binary Electrolytes: Ionic Conductivity and 1H, 7Li, and 19F NMR Studies on Diffusion Coefficients and Local Motions Kikuko Hayamizu,*,† Seiji Tsuzuki,† Shiro Seki,‡ Yasutaka Ohno,‡ Hajime Miyashiro,‡ and Yo Kobayashi‡ National Institute of AdVanced Industrial Science and Technology (AIST), AIST Tsukuba Center 5, Tsukuba, Ibaraki 305-8565, Japan, and Materials Science Research Laboratory, Central Research Institute of Electric Power Industry (CRIEPI), 2-11-1, Iwado-kita, Komae, Tokyo 201-8511, Japan ReceiVed: September 26, 2007; In Final Form: NoVember 6, 2007

A room-temperature ionic liquid (RTIL) of a quaternary ammonium cation having an ether chain, N,Ndiethyl-N-methyl-N-(2-methoxyethyl)ammonium bis(trifluoromethylsulfonyl)amide (DEME-TFSA), is a candidate for use as an electrolyte of lithium secondary batteries. In this study, the electrochemical ionic conductivity, σ, of the neat DEME-TFSA and DEME-TFSA-Li doped with five different concentrations of lithium salt (LiTFSA) was measured and correlated with NMR measurements of the diffusion coefficients D and the spin-lattice relaxation times T1 of the individual components DEME (1H), TFSA (19F), and lithium ion (7Li). The ion conduction of charged ions can be activated with less thermal energy than ion diffusion which contains a contribution from paired ions in DEME-TFSA. In the doped DEME-TFSA-Li samples, the σ and D values decreased with increasing salt concentration, and within the same sample generally DLi < DTFSA < DDEME except for the sample having the lowest salt concentration at low temperatures. Since plots of the temperature dependence of T1 of the 1H and 7Li resonances showed T1 minima, the correlation times τc(H) and τc(Li) were calculated for reorientational motions of DEME and the lithium jump, respectively. At the same temperature, τc(Li) is longer than τc(H), suggesting that the molecular motion of DEME occurs more rapidly than the lithium jump. Combining the DLi and τc(Li), averaged distances for the lithium jump were estimated.

Introduction Room-temperature ionic liquids (RTILs) have attracted much attention given their widespread potential for practical applications. Many basic studies have been performed to determine the specific ion structures and mutual ionic interactions.1 Quaternary ammonium cations form RTILs with various anions. Because of the variety of side chains that can be connected to central N+, many quaternary ammonium RTILs have been synthesized, and the details have been compiled in a database of physical properties.2 Among them, we focus on an RTIL with an ether structure in a side chain.3,4 In the present paper, the target RTIL is N,N-diethyl-N-methyl-N-(2-methoxyethyl)ammonium bis(trifluoromethylsulfonyl)amide (DEME-TFSA).

The cation includes a chain -CH2CH2OCH3. Attractive interactions between oxygen atoms and lithium ions are well* To whom correspondence should be addressed. Phone and fax: + 8129-861-6295. E-mail: [email protected]. † National Institute of Advanced Industrial Science and Technology (AIST). ‡ Central Research Institute of Electric Power Industry (CRIEPI).

known. To use DEME-TFSA as an electrolyte of lithium batteries, the lithium ion (Li+) dopant is expected to interact with oxygen atoms in the side chain of the cation DEME. We have demonstrated the excellent performance of lithium secondary batteries using lithium salt doped DEME-TFSA.5-7 Also, RTILs composed of DEME cations are used as electrolytes in electric double layer capacitors.8 Electrochemical approaches for studying the doped RTIL are important to improve the battery performance, and we have reported systematic studies on binary DEME-TFSA and LiTFSA systems in which the degree of Li doping was increased.9 Also, we will report ab initio molecular orbital calculations for DEME-Li-TFSA configurations and interactions of a lithium ion with an oxygen atom in the side chain of DEME.10 In this paper, the results of the ionic conductivity measurements are presented for neat DEME-TFSA and DEME-TFSA-Li doped with five different concentrations of LiTFSA. The diffusion coefficients of the individual ions were measured for the cation DEME, the anion TFSA, and the Li ion by 1H, 19F, and 7Li NMR, respectively, and are discussed in connection to the ionic conductivity. We have reported the ionic conductivity and ion diffusion coefficients of neat RTILs without including salts or solvents.11-15 Also, we reported the ionic conductivity and the ion diffusion coefficients including lithium diffusion of the binary system composed of 1-ethyl-3-methyl imidazolium tetrafluoroborate EMI-BF4 and LiBF4.16 The ion transport properties are closely related to the bulk physical properties such as viscosity and density. The ion self-diffusion coefficients are 10-10 to 10-13

10.1021/jp077714h CCC: $40.75 © 2008 American Chemical Society Published on Web 01/08/2008

1190 J. Phys. Chem. B, Vol. 112, No. 4, 2008 SCHEME 1: The Stable Structures of DEME and TFSA Estimated by ab Initio Calculation

m2 s-1 in the temperature range between 373 and 253 K, and averaged distances for the ion migration with the time interval of 0.1 s are about 10-6 m which corresponds to the ion migration measured by the ionic conductivity. The spin-lattice relaxation time T1 is related to local motions like a jump of the lithium ion or molecular reorientational motion. Usually, in the solution state, it is difficult to derive physical parameters like rates of molecular motions from 1H T1 values. However, Arrhenius plots of 1H T1 having T1 minimum were observed for the viscous liquid electrolytes of etheneglycol dimethyl ethers (CH3O(CH2CH2O)nCH3 (n ) 3-5)) with a lithium salt and neat liquids polyetheneglycol dimethyl ethers of the molecular weight 400 and larger.17 Also, the protons in the hydrated and anhydrous phosphoric acids showed a T1 minimum, and the correlation times were evaluated for a proton one-flip motion at various temperatures for the various phosphoric acid concentrations.18 13C T1 measurements of imidazolium RTILs have been reported in which the minima were observed in the Arrhenius plots.19-23 Recently, the dynamic properties of trimethylsilylmethyl substituted imidazolium ionic liquids were reported where the minima were observed for the 1H and 19F T NMR.24 In the present work, minima in the liquid 1 state were observed for both 1H and 7Li T1’s for DEME-TFSA without and with doping LiTFSA. Analysis of 1H and 7Li relaxation data using the Bloembergen-Purcell-Pound (BPP) equation gave estimates of the molecular motions of DEME (1H NMR) and the lithium jump (7Li NMR) without assuming the relaxation mechanisms. The correlation times of the local motions of the DEME are fast as they are about 10-9 to 10-11 s. In this paper, we estimated average distances for a lithium jump depending on the salt concentration and temperature above 303 K. In Scheme 1, HF/6-311G** level optimized geometries for isolated DEME and TFSA are shown. Experimental Section Sample Preparation. N,N-Diethyl-N-methyl-N-(2-methoxyethyl)ammonium bis(trifluoromethylsulfonyl)amide (DEMETFSA) was purchased from Kanto Kagaku (synthesized by Nissinbo Industries, Inc.). Lithium bis(trifluoromethanesulfonyl)amide (LiTFSA, Kishida Chemical Co., Ltd) was used to prepare the doped electrolytes. First, DEME-TFSA and LiTFSA were

Hayamizu et al. dried in a vacuum chamber at 323 K for more than 48 h and were stored in a dry argon-filled glovebox ([O2] < 0.4 ppm, [H2O] < 0.1 ppm, Miwa Mfg. Co., Ltd.). Electrolyte mixtures of DEME-TFSA and LiTFSA were prepared by dissolving given amounts of LiTFSA (the concentrations were 0.16, 0.32, 0.48, 0.64, and 0.80 mol kg-1) and DEME-TFSA and by allowing each mixture to stand by stirring at room temperature for over 48 h. The preparation was carried out in a dry-argon-filled glove box. For NMR diffusion measurements, the samples were placed into 5 mm NMR microtubes (BMS-005J, Shigemi, Tokyo) to a height of 5 mm and were sealed with Araldite in a glovebox to prevent moisture. Ionic Conductivity Measurements. The temperature dependencies of the ionic conductivity of the neat DEME-TFSA and the doped DEME-TFSA-Li were measured in stainless steel (SUS)/electrolyte/SUS (SUS ) Stainless Used Steel) symmetric blocking cells and were determined by the complex impedance method using an ac impedance analyzer Princeton Applied Research, PARSTAT-2263, 200 kHz-50 mHz, and an applied voltage of 10 mV at temperatures over the temperature range 353-233 K with cooling. Measurements of the Self-Diffusion Coefficients. The selfdiffusion coefficients (D’s) were measured by the pulsedgradient spin-echo (PGSE) NMR method using a Techmag Apollo-NTNMR and a wide-bore 6.4 T SCM equipped by a JEOL pulsed-field gradient (PFG) multiprobe. The temperature was controlled by a JEOL console. The spectra of the DEME, TFSA, and Li were measured by the 1H, 19F, and 7Li NMR at the frequencies of 270.17, 254.19, and 105.00 MHz, respectively. The attenuation of the echo-signal E was obtained by varing the duration time δ of the PFG at the fixed amplitude g. The D values were determined by regressing the Stejskal equation25

(

(

E ) exp γ2g2δ2D ∆ -

δ 3

))

(1)

onto the attenuation data where γ is the gyromagnetic ratio of the observing nuclei and ∆ is the interval period during the diffusion measurements. D is independent of ∆ for the homogeneous samples, and as implied in eq 1, singleexponential echo attenuation indicates a free-diffusive mode. At low temperatures, since the D values of all the samples were slightly dependent on ∆ at short values, the measurements were made by setting ∆ between 20 and 70 ms from 353 to 273 K with cooling. The shorter ∆ was necessary to prevent convection artifacts at high temperatures.26 At low temperatures, long ∆ was desirable for obtaining equilibrium values.27 The maximum g used was 16 T/m and the longest δ was 4 ms for the 7Li resonance, and the PFG measuring conditions were milder for 1H and 19F PGSE-NMR. The gradient parameters were chosen to give 20 data points that spanned at least 1 order of magnitude of signal attenuation. When the lithium salt concentration became larger and the temperature decreased, the diffusion plots were no longer single-exponential especially for the lithium diffusion. We adopted the D values as long as the diffusion plots were single exponentials. Measurements of the Spin-Lattice Relaxation Time T1. Temperature-dependent 1H, 19F, and 7Li T1 values were measured by the inversion recovery pulse sequence (180°-τ-90°acquisition) together with the diffusion measurements and, when necessary, the measurements were made over the wider temperature range. The 1H NMR spectrum is shown in Figure 1,

NMR Studies on Lithium Salt Doped DEME-TFSA

J. Phys. Chem. B, Vol. 112, No. 4, 2008 1191

Figure 1. The 1H NMR spectrum of the doped DEME-TFSA-Li with a concentration of 0.16 mol kg-1 at 313 K.

Figure 2. Plots of the ionic conductivity versus temperature for the neat DEME-TFSA and the doped DEME-TFSA-Li.

Figure 3. The VFT fitting parameters (see eq 2) of the ionic conductivity by the equation derived from Figure 2 are plotted versus temperature for (a) A, (b) B, and (c) T0.

and the 1H T1 values were measured for the separated peaks of CH2O in NCH2CH2O and two methyl signals of OCH3 and NCH2CH3. Results Ionic Conductivity. The ionic conductivities versus temperature are plotted for the neat and four doped samples in Figure 2 for the wide temperature range from 353 to 233 K. The Arrhenius plots were all curved, and the curvatures became significant as the salt concentration increased. The plots were fitted by the Vogel-Fulcher-Tammann (VFT)-type equation as

σ ) AT1/2 exp(-B/(T0 - T))

(2)

where A, B, and T0 are fitting parameters. The fitted lines are shown in Figure 2, and the parameters obtained are plotted versus the salt concentration in Figure 3. A and B are related to several carrier ions and activation energies, respectively, and T0 is believed to have a close relationship with glass-transition temperature. The three fitting parameters change smoothly with the salt concentration, and A and B increase with the salt concentration, while T0 becomes lower which is reasonable referring to the glass-transition temperature. Ion Self-Diffusion Coefficients. The temperature-dependent diffusion coefficients of the cation (DEME) and the anion (TFSA) are shown in Figure 4 for the neat DEME-TFSA. Convection effects were observed above 343 K at long ∆, where the apparent D values became larger similar to nonviscous liquids even if the sample height was only 5 mm,26 and thus, short ∆ values were used for actual measurements. Below 333

Figure 4. The self-diffusion coefficients of DEME (solid down triangle) and TFSA (solid up triangle) of the neat DEME-TFSA at the equilibrium ∆. The open marks are the data measured at the ∆ being 20 ms. The VFT parameters D0, B, and T0 of DEME and TFSA were 5.6 ((1.8) and 6.6 ((1.9) × 10-8 m2 s-1, 1380 ((130) and 1449 ((120) K-1, and 124 ((9) and 122 ((8) K, respectively, for the fitting of the D(T) ) D0 exp(-B/(T0 - T)).

K, the D values were independent of ∆ (i.e., free from the convection artifacts). At low temperatures below 303 K, the apparent D became faster as ∆ became shorter because of anomalous diffusion in heterogeneous space as reviewed by Metzler and Klafter.27 The apparent D approached an equilibrium value as ∆ became longer, and the equilibrium values are plotted in Figure 4. In this figure, the apparent D values measured with ∆ ) 20 ms are included for the low temperatures.

1192 J. Phys. Chem. B, Vol. 112, No. 4, 2008

Hayamizu et al.

Figure 5. The temperature dependence of the self-diffusion coefficients of DEME (circle), TFSA (down triangle), and Li (up triangle) of the doped DEME-TFSA-Li at a concentration of 0.32 mol kg-1. The D’s used in this figure were derived from single exponential attenuation plots. The 7Li diffusion plot at 263 K was curved.

Interestingly, the D obtained at ∆ ) 20 ms gave almost linear Arrhenius plots as shown in Figure 4. The D values in the higher temperature region suggest that the anion and cation diffuse in a free space, while at low temperatures some specific interactions exist to prevent free diffusion for both the anion and the cation. The DEME diffusion is always only a little faster than the TFSA diffusion, although the calculated van der Waals radius of DEME (0.351 nm) is larger than that of TFSA (0.325 nm).28,29 The temperature-dependent D’s of DEME, TFSA, and Li in the doped DEME-TFSA-Li with the concentration of 0.32 mol kg-1 are shown in Figure 5. At low temperatures, the diffusion plots of the components, especially the Li, were curved and the curvatures became more pronounced as ∆ became shorter. Similar phenomena were observed for the lithium diffusion in cross-linked poly(ethylene oxide) electrolytes.30 Then, data obtained from straight lines with long ∆ were plotted in Figure 5. The temperature-dependent D’s for other samples with the different salt concentrations (i.e., 0.16, 0.48, 0.64, and 0.80 mol kg-1) are shown in Figures S1, S2, S3, and S4, respectively, in the Supporting Information. Within the same sample, the DEME diffusion was the fastest, the next fastest was the TFSA, and the Li diffusion was the slowest except for the sample with the lowest salt concentration at low temperature (see Figure S1). All the diffusion plots of the 0.80 mol/kg sample below 303 K showed diffraction effects representative of undergoing restricted diffusion at low temperatures. The diffusions of all the species are restricted in the concentrated samples at low temperatures. The D’s of all of the individual components for the samples of neat DEME-TFSA and doped DEME-TFSA-Li are shown versus temperature in Figure 6 (a) DEME, (b) TFSA, and (c) Li. Clearly, as the salt concentration increased, the D values decreased. Such phenomena were also observed in the ionic conductivity in Figure 2, and the curvatures became more pronounced with salt concentration, which is similar to the ionic conductivity. The reduction of the ion diffusion induced by the salt concentration can be interpreted mainly by the increase of the viscosity. As reported in our previous papers,11-15 the plots of the temperature dependence of the D’s for a cation and an anion of neat RTILs are also described by a VFT type (i.e., D(T) ) D0 exp(-B/(T - T0)), where D0, B, and T0 are fitting parameters). The fitting was possible for each ion in the doped samples. As

Figure 6. The temperature dependence of the self-diffusion coefficients of (a) DEME, (b) TFSA, and (c) Li ions for the neat DEME-TFSA (open square) and the doped DEME-TFSA-Li at concentrations of 0.16 (up triangle), 0.32 (down triangle), 0.48 (diamond), 0.64 (left triangle), and 0.80 (right triangle) mol kg-1.

TABLE 1: The Activation Energies of the Diffusion Coefficients of the Neat DEME-TFSA and Doped DEME-TFSA-Li above 303 K (kJ/mol) concentration mol kg-1

DEME

TFSA

Li

0 0.16 0.32 0.48 0.64 0.80

30.1 ( 0.6 31.8 ( 0.7 32.9 ( 0.2 34.5 ( 0.4 35.7 ( 0.5 40.0 ( 0.6

30.8 ( 0.8 33.2 ( 0.5 34.2 ( 0.4 35.2 ( 0.8 37.2 ( 0.3 41.5 ( 0.9

33.0 ( 0.6 35.5 ( 0.9 36.9 ( 0.9 39.0 ( 0.6 44.0 ( 0.7

described above, the ambiguity exists in the D values measured at low temperatures. Since the fitting parameters are sensitive to these values, we did not describe the values in this paper except for the undoped sample in Figure 4, where the fitting results are given in the caption. The Arrhenius plots in Figure 6 are linear in the high-temperature region, and the calculated activation energies for each ion above 303 K are summarized in Table 1. Clearly, following the increase in salt concentration, the activation energies of the D’s increased for all the ion species. Within the same sample, the activation energy of DEME was always the smallest. The Li activation energy was larger than that of TFSA, and the difference became larger as the salt concentration increased. The D’s of the individual ions at 303 K are shown in Figure 7 for all the samples. It is remarkable that in the neat and 0.16 mol kg-1 samples, where the lithium ion was not concentrated, the individual ion diffusion is not much different, and as the salt concentration increased, the mutual differences became larger, and this trend was also obtained in the activation energies in Table 1.

NMR Studies on Lithium Salt Doped DEME-TFSA

J. Phys. Chem. B, Vol. 112, No. 4, 2008 1193 temperature range. Thus, the temperature-dependent 1H T1 data was analyzed using the Bloembergen, Purcell, and Pound (BPP) equation32

(

4τc τc 1 )C + 2 2 T1 1 + ω 0 τc 1 + 4ω02τc2

)

(3)

where ωo is the observed frequency (rad s-1), τc is the correlation time of the dipole-dipole interaction, and the summation index j is over all interacting dipoles. The 1H T1 can be calculated by

C)

Figure 7. The D’s of DEME (square), TFSA (up triangle), and Li (down triangle) are shown versus the lithium salt concentration at 303 K.

3 10

Spin-Lattice Relaxation Time T1 and Local Motions. The spin-lattice relaxation times of DEME, TFSA, and Li were measured for all the samples over a wider temperature range compared with the diffusion measurements, since the T1’s are reflected by local motions which are much faster than the mobility of the center of gravity in the translational diffusion and are little affected by the anomalous or restricted diffusion. Actually, the tentative calculation of longer correlation times derived from the D values indicated that the translational diffusion is not effective for the relaxation mechanism.31 As an example, the temperature dependencies of the T1’s are shown for the DEME-TFSA-Li with the concentration of 0.32 mol kg-1 in Figure 8a. Except for CF3 in TFSA of the 19F resonance, the T1 of the 7Li and 1H resonances showed minima in the present

∑j

1

(4)

rj6

where γH is the gyromagnetic ratio of 1H, p is the reduced Planck’s constant, and r is the separation between dipoles (the H-H distance). In the dipole-dipole relaxation model, because of the term in brackets in eq 3, the spin-lattice relaxation time is a minimum when ωoτc ) 2πν0τc ) 0.616. In the current study, the 1H observed frequency, ν0, is 270.17 MHz and the τc at the T1 minimum can be calculated to be 3.63 × 10-10 s (363 ps). C can be calculated from the T1 value at the minimum point. Thus, eq 3 can be used to calculate τc from the observed T1 values at every temperature. When the T1 minimum is obtained, it is possible to calculate the τc without determination of the precise H-H distances. The τc’s calculated are given in Figure 8b for the methylene protons adjacent to the oxygen atom of NCH2CH2O and the two methyl groups in NCH2CH3 and OCH3 for the separated peaks in the 1H spectrum in Figure 1. In the Supporting Information, the temperature dependencies of the T1 data of all the species and the τc’s are given for the neat DEME-TFSA in Figure S5 and the doped DEME-TFSALi of the concentrations of 0.16, 0.48, 0.64, and 0.80 mol kg-1 in Figures S6, S7, S8, and S9 in the Supporting Information, respectively. For the 19F T1 data which reflects the internal rotation around the three symmetrical axis of CF3 in the TFSA, eq 4 becomes

C)

Figure 8. (a) The T1’s and (b) the correlation times calculated for species having the T1 minima are shown versus temperature for the 0.32 mol kg-1 doped sample.

γH4p2

9 4 21 γ p 20 F r6

(5)

where r is the F-F atomic distance in CF3 and γF is the 19F gyromagnetic ratio. Since the 19F T1 became longer as the temperature increased without minimum, the extreme narrowing condition may be assumed: ωoτc , 1. Hence, for TFSA, eq 3 becomes

1 9 4 21 ) γ p τ T1 4 F r 6 c

(6)

The temperature dependencies of the 7Li T1 values are shown in Figure 9a for all the doped DEME-TFSA-Li samples. Clearly, 7Li T minima were obtained, and the T minimum temperature 1 1 became higher with longer T1 as the salt concentration increased. The temperature-dependent behaviors of the T1’s for the 0.32 and 0.48 mol kg-1 samples resemble each other, and also the 0.64 and 0.80 mol kg-1 samples behaved similarly. The T1 of the 7Li (I ) 3/2) undergoing isotropic reorientational diffusion is given by33

(

2 4τc τc 1 ωq ) + T1 50 1 + ω 2τ 2 1 + 4ω 2τ 2 0 c 0 c

)

(7)

1194 J. Phys. Chem. B, Vol. 112, No. 4, 2008

Figure 9. The temperature dependencies of (a) the 7Li T1 and (b) the correlation time (τc(Li)) calculated from the T1 minima are shown for the doped DEME-TFSA-Li at concentrations of 0.16 (square), 0.32 (circle), 0.48 (up triangle), 0.64 (down triangle), and 0.80 (diamond) mol kg-1.

where ωq ) 2πνq is the quadrupolar coupling constant. Similar to the analysis for the 1H dipole-dipole relaxation above, the relaxation minimum should now occur when ω0τc ) 2πν0τc ) 0.616 (for 7Li, ν0 ) 105.0 MHz). The 7Li correlation time τc(Li) at the minimum temperature is 9.34 × 10-10 s (934 ps) and the temperature-dependent values are shown in Figure 9b for all of the doped samples. It is well-known that the 7Li T1 minimum can be observed in polymer electrolyte doped lithium salts and even in liquids like polyetheneglycol dimethyl ethers (CH3O(CH2CH2O)nCH3 (n ) 3-5)) doped with lithium salt.16 The quadrupolar relaxation mechanism of 7Li resonance for a lithium ion in a spherical environment gives short T1’s, and the mode of the lithium motions is a jump from one position to another. From ab initio molecular orbital calculations, the oxygen atom in the DEME side chain has negative charge and a close contact with lithium ion.10 A lithium ion has sites to jump at an oxygen atom in DEME and an anion TFSA. In the present doped DEME-TFSA-Li samples, the rate of the lithium jump increased as the temperature increased, that is, τc(Li) became shorter with increased temperature. As shown in Figure 9b, the τc(Li) became slower as the increase of the salt concentration and the Arrhenius plots are not always linear over the entire temperature range. Since the slopes changed around the ambient temperature, a switch between different modes of the lithium jump can be assumed. The activation energies calculated from the bent Arrhenius plots are listed in Table 2, where the values are smaller at lower temperatures; in other words, the lithium jumps are less activated thermally. At the same time, the lithium motion is significantly affected by the lithium concentration of the samples as shown in Figure 10, and the τc(Li) at 303 K changed from 3.9 × 10-10 to 2.1 ×

Hayamizu et al. 10-9 s (390 ps to 2.1 ns) for the DEME-TFSA-Li of the concentrations 0.16-0.80 mol kg-1, respectively. At higher lithium salt concentrations, the relative number of the lithium ions is increased and the lithium jump is reduced. Generally, in solution-state 1H NMR, molecular motions are assumed to be under the extreme narrowing condition and, consequently, 1H T1 minima can only be observed in the solid state. Clearly, as shown in Figures 6 and S5-S9, the 1H T1’s of the cation DEME in the RTIL show minima in the present study. It is well-known that the methyl rotations are averaged out in the solid state and that the relaxation process which gives the minimum in the present temperature range in the liquid state is not the methyl rotation around the three symmetric axes. Also, the methylene CH2O in the NCH2CH2O structure gives similar τ values to the methyl groups. Thus, we can assume that the process related to the T1 minimum is the molecular reorientational motions. The activation energies in Table 2 were obtained from the τc(H) of the CH2O in NCH2CH2O which give almost linear Arrhenius plots over the whole temperature range indicating a single motional mode. The motions measured at the NCH2CH2O were the slowest, the next slowest was CH3 in NCH2CH3, and the OCH3 motion of the NCH2CH2OCH3 was the fastest in the longest chain of the DEME. The mutual differences of the τc(H) were very small in the same sample at the same temperature. Local proton positions and the environment only slightly affect the τc(H) of the molecular reorientational motions. To see the salt concentration effects on the local motions of DEME, the τc(H)’s at 303 K are plotted in Figure 10, and the molecular motion had a slight concentration dependence. Since the methylene protons of NCH2CH2O are located near the center of DEME molecule, the correlation time must be most reflected by the local motions of the whole molecule. Depending on the concentration at 303 K, the τc(H) changed from 1.8 × 10-10 s (180 ps) in neat DEME-TFSA to 3 × 10-10 s (300 ps) in DEMETFSA-Li with the concentration of 0.80 mol kg-1. The τc(H)’s of the methyl groups are only a little shorter (motion is faster) probably because of the greater freedom at the periphery. Within the same sample, the motion of DEME is faster than the lithium jump. The temperature dependence of the 19F T1 for CF3 of TFSA is shown in Figure 11a for each of the samples. Since the T1 became longer as the temperature increased similar to usual liquid-state samples, we assumed extreme narrowing conditions, ωoτc , 1 for ωo ) 2πν0 where ν0 is 254.19 MHz for the 19F resonance. Following eq 6, τc(F) was calculated with rF-F ) 0.216 nm as shown in Figure 11b. The activation energies above and below room temperature were calculated from the Arrhenius plots of the T1 and are given in Table 2. The thermal activation of the TFSA motion was larger at the high-temperature range than at the lower temperatures, and the motional modes may be different. The τc(F) calculated under uncertain assumption of the extreme narrowing condition may be important for comparing the rates of the motions of DEME, Li, and TFSA. The τc(F) values of all the samples at 303 K are compared in Figure 10. The TFSA motion is quicker than the DEME motion and the concentration dependency was similar to DEME. The slow down of local motions of DEME and TFSA with the salt concentration may be related to the solvation effects around the lithium ions.34 Discussion Ionic conductivity is a measure of the migration of charged ions, while the diffusion coefficients are averaged values of

NMR Studies on Lithium Salt Doped DEME-TFSA

J. Phys. Chem. B, Vol. 112, No. 4, 2008 1195

TABLE 2: The Activation Energies (kJ/mol) for the DEME, Lithium, and TFSI Motions Li jump (from τc)

TFSI rotation (from T1)

salt concentration (mol/k g)

DEME motion (from τc of NCH2CH2O)

higher temperature

lower temperature

higher temperature

lower temperature

0 0.16 0.32 0.48 0.64 0.80

21.6 ( 0.4 21.1 ( 0.3 20.1 ( 0.5 22.1 ( 0.5 23.3 ( 0.9 22.3 ( 0.6

19.9 ( 0.5 19.1 ( 0.3 19.1 ( 0.2 19.0 ( 0.5 18.2 ( 0.5

15.8 ( 0.7 20.8 ( 0.8 16.2 ( 0.5 17.5 ( 1.0 19.2 ( 0.8

14.3 ( 0.3 13.1 ( 0.3 11.4 ( 0.3 11.0 ( 0.2 10.0 ( 0.2 9.3 ( 0.1

7.9 ( 0.2 7.6 ( 0.2 5.9 ( 0.2 7.4 ( 0.3 5.2 ( 0.2 5.1 ( 0.1

charged and paired ions. Although every RTIL is composed of ions, each ion does not always migrate having charge in a distance where the diffusion or ionic conductivity is measured. To compare the VFT parameters derived from the ionic conductivity to those obtained from the diffusion coefficients of the neat DEME-TFSA, the ionic conductivity data was reanalyzed using the VFT equation σ ) σ0 exp(-B/(T - T0)) but in the same temperature range as the diffusion coefficients. The analysis gave B ) 511 ( 33 K-1 and T0 ) 189 ( 4 K. For the DDEME and DTFSA shown in Figure 4, the B and T0 values were about 1410 K-1 and 123 K, respectively. The smaller B value and the lower T0 of the ionic conductivity suggest that the charged ions in DEME-TFSA can be thermally activated compared with the diffusion including paired ions. For RTILs, the larger B values and the lower T0 values were more generally obtained for ionic conductivity than for ion diffusion.15 Up to now, we have observed that the cation diffusion coefficient of an RTIL is always larger than that of the anion in the same sample at the same temperature.11-15 As described above, the calculated van der Waals radius of DEME is a little larger than TFSA. In a recent paper, bulky imidazolium anions exhibited similar or slower diffusion than counter cations.24 Thus, ion sizes are related to relative D values. In the lithiumdoped DEME-TFSA-Li samples, the DEME diffuses the fastest, the next fastest is the TFSA, and the lithium diffusion is the slowest as shown in Figures 3 and S1-S4 except for the least concentrated sample at low temperatures as shown in Figure S1. The same trends were observed in the binary EMI-BF4-Li systems, where the reduction of the Li diffusion was much larger.16 The lithium ion size is the smallest. The result that the lithium ion diffuses the slowest implies that it diffuses accompanied by DEME or TFSA. Using Raman spectroscopy,

the existence of Li-TFSA interactions was shown in RTILs.34 The activation energies for the D’s in Table 1 indicate that thermal activations are in the order DEME < TFSA < Li and that they became larger as the concentration increased. Especially, in the most concentrated sample (i.e., 0.80 mol kg-1), increased activation energies were observed for all components. The concentration dependencies of the D’s at 303 K in Figure 7 show clearly that the salt concentration brings a continuous decrease of the transport mobility of center of gravity for all components. The root-mean-squared displacement for a diffusing species is given by 〈R〉 ) x6D∆. Thus, for a period of 0.1 s at 303 K, the DEME and the TFSA in the neat DEME-TFSA move a distance of about 3.8 µm which decreases to 1.9 and 1.6 µm for the DEME and TFSA in the DEME-TFSA-Li of the 0.80 mol kg-1 sample. The average displacement distance for the lithium ion was 3.2 µm for the 0.16 mol kg-1 sample and decreased to 1.3 µm for the 0.80 mol kg-1 sample at 303 K. The ions in the neat DEME-TFSA and doped DEME-TFSA-Li can diffuse average distances of about 1-10 µm during 0.1 s at 303 K and can change in rate and distance depending on temperature and salt concentration.

Figure 10. Plots of the correlation times at 303 K versus the salt concentration for the lithium ion (asterisk) determined using 7Li NMR; the NCH2CH2O (circle), CH3 signals in NCH2CH3 (down triangle), and OCH3 (up triangle) determined using 1H NMR and the CF3 of TFSA (diamond) derived from 19F NMR measurements.

Figure 11. The temperature dependencies of (a) the 19F T1 and (b) the correlation time (τc(F)) calculated under the assumption of the extreme narrowing condition for the neat DEME-TFSA (square) and the doped DEME-TFSA-Li at concentrations of 0.16 (circle), 0.32 (up triangle), 0.48 (down triangle), 0.64 (diamond), and 0.80 (left triangle) mol kg-1.

1196 J. Phys. Chem. B, Vol. 112, No. 4, 2008

Figure 12. Plots of the one-jump distances for the lithium ion above room temperature versus temperature for the doped DEME-TFSI-Li at 0.16 (circle), 0.32 (square), and 0.64 (up triangle) mol kg-1.

The T1 process corresponds to a motion. For example, if the mechanism is the dipole-dipole relaxation, a dipole-dipole axis changes the position or direction once for an effective T1 process. In other words, the τc calculated from the T1 is the time of a jump or a molecular motion. The τc(H) of DEME, attributed to the reorientational motion, ranges between 5 × 10-9 and 5 × 10-10 s from the low to high temperatures, and the activation energies in Table 2 are about 20-23 kJ/mol for all of the samples. The DEME reorientational motions have larger activation energies compared with the lithium jump for τc(Li) and CF3 motions. The local motions of DEME can be activated with smaller energy than the diffusion. With reference to the Stokes-Einstein-Debye (SED) relationship,35 an attempt was made to calculate the DEME radius aT from the equation aT ) xτcD where D should be the rotational diffusion coefficient and τc is a rotational correlation time for a whole molecular rotation. The approximate calculation was made by using the translational diffusion coefficient and the τc obtained from the 1H T1. The calculated aT was about 0.09 nm for neat DEME-TFSA at 353 K and was much smaller than the DEME size estimated by ab initio calculations in Scheme 110 or by the van der Waals radius (0.351 nm). Similarly, small radii were reported from the τc obtained by the analysis of 13C T1.23 The correlation time derived from the T1 does not always correspond to an overall molecular rotation. Hence, the τc(H) in the present study may correspond to local motions like small-angle reorientation of DEME.36 The rotational diffusion coefficient may be faster than the translational diffusion coefficient in DEME-TFSA. The values of correlation times τc(F) in Figure 10, which were calculated under the assumptions that the relaxation results from the CF3 rotation and that the motion is in the extreme narrowing condition (τcωo , 1), indicate that the CF3 motions are faster than the reorientation motion of DEME. Also, the activation energies calculated from the 19F T1 (more reliable than the calculated τc) in the higher temperature range are smaller than those for the DEME. The much smaller activation energies of the CF3 in the lower temperature range (5-8 kJ/mol) may correspond to free CF3 rotation. The physical interpretation of τc(Li) is different, since the lithium ions may jump from one place to another. The lithium jump in the doped polyethyleneoxide (PEO) is a well-established concept where the lithium jumps from an oxygen to another oxygen atoms in the -(CH2CH2O)n- structures. In

Hayamizu et al. the electrolyte solution of the LiTFSA doped pentaglyme CH3(CH2CH2O)5CH3, the τc(Li)’s were about 8 × 10-11 and 3 × 10-10 s at 353 and 303 K, respectively.16 In the present study, as shown in Figure 9, the τc(Li)’s of the 0.32 mol kg-1 sample were longer at 3.2 and 7.0 × 10-10 s at 353 and 303 K, respectively. At the same time, a significant slow down of the lithium jump was observed with increasing salt concentration and the lithium ions. The average distances of a single lithium jump was calculated by assuming that the DLi of the jump is the same as the measured value in micrometer space, that is, 〈Rone-jump〉 ) x6Dτc(Li). The one-jump distances calculated above room temperature are shown in Figure 12. The lithium jump distance became longer with increasing lithium concentration and temperature. When the lithium concentration was small, the lithium jump distance was about 0.2 nm at 303 K and was very short, which gives insight into the lithium jump. The jump distance in the lithium-concentrated samples became longer with the longer jump time τc(Li). Conclusion DEME-TFSA is an RTIL with a quaternary ammonium cation having an ether side chain. The ionic conductivity data shows that the charged ion migration is thermally activated easily compared to the migration of whole ions including paired ions. With LiTFSA doping, the ionic conductivity and the ion diffusion decrease with increasing salt concentration. The diffusion of the lithium is generally slowest in spite of the smallest size. The correlation time of the DEME molecular reorientation obtained from the 1H T1 data is shorter than that of the lithium jump calculated from the 7Li T1 values. The estimated lithium one-jump distance varies with the temperature and salt concentration and affords important information on lithium transfer. Acknowledgment. The authors express their sincere thanks to Professor W. S. Price for reading the manuscript critically and for fruitful discussion. Supporting Information Available: Temperature-dependent diffusion coefficients of DEME, TFSA, and Li for the doped samples (0.16, 0.48, 0.64, and 0.80 mol kg-1) and the Arrhenius plots of 1H, 19F, and 7Li T1’s and the 1H and 7Li τc values at each temperature for the neat (without 7Li) and the doped samples (0.16, 0.48, 0.64, and 0.80 mol kg-1). This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Electrochemical Aspects of Ionic Liquids; Ohno, H., Ed.; Wiley: Hoboken, NJ, 2005. (2) Zhang, S.; Sun, N.; He, X.; Lu, X.; Zhang, X. J. Phys. Chem. Ref. Data 2006, 35, 1475. (3) Matsumoto, H.; Yanagida, M.; Tanimoto, K.; Nomura, M.; Kitagawa, Y.; Miyazaki, Y. Chem. Lett. 2000, 8, 922. (4) Zhou, Z.-B.; Hajime Matsumoto, H.; Tatsumi, K. Chem. Lett. 2004, 33, 886. (5) Seki, S.; Kobayashi, Y.; Miyashiro, H.; Ohno, Y.; Mita, Y.; Usami, A.; Terada, N.; Watanabe, M. Electrochem. Solid-State Lett. 2005, 8, A577. (6) Seki, S.; Kobayashi, Y.; Miyashiro, H.; Ohno, Y.; Usami, A.; Mita, Y.; Watanabe, M.; Terada, N. Chem. Commun. 2006, 544. (7) Kobayashi, Y.; Mita, Y.; Seki, S.; Ohno, Y.; Miyashiro, H.; Terada, N. J. Electrochem. Soc. 2007, 154, A677. (8) Sato, T.; Masuda, G.; Takagi, K. Electrochim. Acta 2004, 49, 3603. (9) Seki, S.; Ohno, Y.; Miyashiro, H.; Kobayashi, Y.; Usami, A.; Mita, Y.; Terada, N.; Hayamizu, K.; Tsuzuki, S.; Watanabe, M. Submitted for publication. (10) Tsuzuki, S.; Hayamizu, K.; Seki, S. to be submitted. (11) Noda, A.; Hayamizu, K.; Watanabe, M. J. Phys. Chem. B 2001, 105, 4603.

NMR Studies on Lithium Salt Doped DEME-TFSA (12) Tokuda, H.; Hayamizu, K.; Ishii, K.; Susan, M. A. B. H.; Watanabe, M. J. Phys. Chem. B 2004, 108, 16593. (13) Tokuda, H.; Hayamizu, K.; Ishii, K.; Susan, M. A. B. H.; Watanabe, M. J. Phys. Chem. B 2005, 109, 6103. (14) Tokuda, H.; Ishii, K.; Susan, M. A. B. H.; Tsuzuki, S.; Hayamizu, K.; Watanabe, M. J. Phys. Chem. B 2006, 110, 2833. (15) Tokuda, H.; Tsuzuki, S.; K.; Susan, M. A. B. H.; Hayamizu, K.; Watanabe, M. J. Phys. Chem. B 2006, 110, 19593. (16) Hayamizu, K.; Aihara, Y.; Nakagawa, H.; Nukuda, T.; Price, W. S. J. Phys. Chem. B 2004, 108, 19527. (17) Hayamizu, K.; Akiba, E.; Bando, T.; Aihara, Y. J. Chem. Phys. 2002, 117, 5929. (18) Aihara, Y.; Sonai, A.; Hattori, M.; Hayamizu, K. J. Phys. Chem. B 2006, 111, 24999. (19) Antony, J. H.; Mertens, D.; Dolle, A; Wasserscheid, P; Carper, W. R. Chem. Phys. Chem. 2003, 588. (20) Carper, W. R.; Wahlbeck, P. G.; Antony, J. H.; Mertens, D.; Dolle, A; Wasserscheid, P. Anal. Bioanal. Chem. 2004, 378, 1548. (21) Carper, W. R.; Wahlbeck, P. G.; Dolle, A. J. Phys. Chem. A 2004, 108, 6096. (22) Antony, J. H.; Dolle, A.; Mertens, D.; Wasserscheid, P.; Carper, W. R.; Wahlbeck, P. G. J. Phys. Chem. A 2005, 109, 6676. (23) Heimer, N. E; Wilkes, J. S.; Wahlbeck, P. G.; Carper, W. R. J. Phys. Chem. A 2006, 110, 868.

J. Phys. Chem. B, Vol. 112, No. 4, 2008 1197 (24) Chung, S. H.; Lopato, R.; Greenbaum, S. G.; Shirota, H.; Castner, E. W., Jr.; Wishart, J. F. J. Phys. Chem. B 2007, 111, 4885. (25) Stejskal, E. O. J. Phys. Chem. 1965, 43, 3597. (26) Hayamizu, K.; Price, W. S. J. Magn. Reson. 2004, 167, 328. (27) Metzler, R.; Klafter J. Phys. Rep. 2000, 339, 1. (28) Ue, M. J. Electrochem. Soc. 1994, 141, 3336. (29) Ue, M.; Murakami, A.; Nakamura, S. J. Electrochem. Soc. 2002, 149, A1385. (30) Hayamizu, K.; Akiba, E.; Bando, T.; Aihara, Y.; Price, W. S. Macromolecules 2003, 36, 2785. (31) Abragam, A. The Principles of Nuclear Magnetism; Oxford: London, 1961. (32) Traficante, D. D. In Relaxation: An Introduction, in Encyclopedia of Nuclear Magnetic Resonance; Grant, D. M., Harris, R. K., Eds.; Wiley: New York, 1996; Vol. 6, p 3988. (33) Werbelow, L. G. In Relaxation Theory for Quadrupolar Nuclei; in Encyclopedia of Nuclear Magnetic Resonance; Grant, D. M., Harris, R. K., Eds.; Wiley: New York, 1996; Vol 6, p 4092. (34) Umebayashi, Y.; Mitsugi, T.; Fukuda, S.; Fujimori, T.; Fujii, K.; Kanzaki, R.; Takeuchi, M.; Ishiguro, S. J. Phys. Chem. B 2007, 111, 13028. (35) Koenderink, G. H.; Zhang, H.; Aarts, D. G. A. L; Lettinga, M. P.; Philipse, A. P.; Nagele, G. Faraday Discuss. 2003, 123, 335. (36) Hayamizu, K.; Tsuzuki, S.; Seki, S. J. Chem. Phys., submitted.