[13] J. Cruickshank, H.V.St.A. Hubbard, N. Boden and I.M. Ward,. Polymer, in press. [14] I.M. ... 151 J.R. MacCallum, A.S. Tomlin and C.A.Vincent, Eur. Polym.
SOLID STATE ELSEWIER
Iolws
Solid State Ionics 90 (1996) 39-48
studies of triflate ion association in polymer gel electrolytes and their constituents
Spectroscopic
SF. Johnston*,
I.M. Ward, J. Cruickshank,
G.R. Davies
IRC in Polymer Science and Technology, Universiry of Leeds, Leeds, LS2 9JT, UK Received
25 September
1995; revised 21 March
1996; accepted
I April 1996
Abstract Raman and NMR spectroscopy have been used to investigate the state of ion association in systems comprising salt, polymer and/or solvent. The dissolved salt in each case was lithium triflate (lithium trifluoromethane sulphonate). Five systems were studied, comprising polymer gel electrolytes or constituents of such electrolytes. These were: salted (a) NJ’-dimethyl formamide (DMF), (b) tetraethylene glycol dimethylether (tetraglyme), (c) polyethylene glycol (PEG), (d) end-esterified PEG and (e) poly(vinylidene fluoride) (PVDF) gels containing either DMF or tetraglyme. Raman results give a broad indication of the significance of end-group, solvent and polymer choice in triflate-salted systems. In all cases, anion association rises with temperature, often with a significant increase in ion aggregation. The reliability of these results is supported by an analysis of systematic errors incident in this technique for ion association measurements. A limited comparison performed on the system having the highest concentration of ion aggregates (end-esterified PEG) suggests that Raman and NMR spectroscopy provide different information concerning ion association. Reasons for the difference are discussed, concluding that complementary information is obtained owing to the different time constants relevant to the two techniques and to the dependence of the Raman results on molecular proximity. Keywords: Ion association;
Polymer
gel electrolytes;
Lithium triflate; Tetraglyme;
1. Introduction Previous work has indicated that Raman and infrared spectroscopy are useful methods for determining the state of ions in polymer electrolytes [l-7]. The Raman and infrared spectra of triflatesalted polymer electrolytes display one or more peaks in the 1030-1055 cm-’ region. Such studies have ascribed the various observed spectral positions
*Corresponding
author.
0167-2738/96/$15.00 Copyright PII SO167-2738(96)00381-5
01996
PEG; DMF; Raman; NMR
of such multiple peaks to the presence of free ions, ion pairs and to larger aggregations of ions. By determining the relative areas of these peaks the fraction of ions in each state can be determined. A complementary source of information regarding ion state is pulse field gradient spin echo NMR. This technique can measure the self diffusion coefficients of both the cation and anion in polymer electrolyte systems as a function of salt concentration and temperature. The mobility, charge state and relative distribution of the charge carriers can thus be inferred _by combining this information with the measured conductivity.
Elsevier Science B.V All rights reserved
40
S.F. Johnston
et al. I Solid State Ionics
90 (1996)
39-48
2. Experimental
2.2. Raman measurements
2.1. Materials
A Coderg double-pass dispersive spectrometer and cooled photomultiplier detector were employed. The 488 nm argon-ion laser power incident on the sample was -0.5 to 1 W. The detailed spectra employed for quantitative work covered the region 975-1200 cm-’ with 1 resolution and 0.25 cm-’ steps, with cm-’ wavenumber accuracy better than 1 cm-‘. Two to five scans were co-added to detect any long term instrumental changes and to improve noise. The net averaging time per measured point was some 6- 10 s. Sample temperature, controlled by a Neslab watercirculation system, spanned a temperature range of about l-97°C (274-370 K), with stability typically better than +l”C. The resulting spectra were curve-fitted using a commercial software analysis package (Spectra-Calc, Galactic Industries) over the spectral range 9751175 cm-’ using up to nine Lorentzian peaks plus a linear background. Three of the Lorentzians were attributed to ‘free’, contact-paired and aggregates of CF,SO, ions (near 1032, 1041 and 1051 cm-‘, respectively). The remaining Lorentzians were used to model the background spectrum attributable to the solvent and slow temporal fluctuations in the Raman equipment. The random error in the determinations of free ion fraction was estimated by comparing the results from repeated spectra. The resulting uncertainties are indicated by the error bars for each data point, and are typically 20.02 to kO.07. The error in the determinations was reduced by repeating the curvefitting procedure using various initial fitting parameters. The fitting error is also a function of spectral signal-to-noise ratio and of perturbations (principally temperature fluctuations) during data acquisition, and is therefore difficult to avoid entirely. The total error in free ion fraction random and fitting, is estimated to be better than +-0.1. Systematic errors were also investigated carefully. In general, different peaks in the Raman spectrum are unrelated in intensity, i.e. the efficiency of scattering is likely to be different for each. As the method of determining the free ion fraction employs the ratio of peak areas, it is important that these areas
Lithium triflate dissolved in DMF, tetraglyme or PEG yields transparent liquids of low to moderate viscosity. The samples had various salt concentrations, typically ranging from near the solubility limit for high salt concentrations down to concentrations at the limit of instrumental detectivity. Following convention, the polymer salt concentration is quoted as the ratio of monomer to alkali metal cation concentration. The tetraglyme samples covered an oxygen to lithium concentration range from 12:l to 100: 1 (molality m = 0.11 to 0.88 mol kg-‘). DMF samples were prepared covering the range EO:Li = 3: 1 to 100: 1 (m = 0.14 to 4.56 mol kg-‘), and PEG samples had EO:Li = 1O:l to 768:l (m = 0.02 to 1.7 mol kg-‘). DMF, tetraglyme, PEG and lithium triflate were obtained from Aldrich and dried prior to use. The PEG samples fell into three classes:
1. PEG, mol. wt. 300 (denoted PEG300). 2. Esterified PEG, mol. wt. 200 (PEG2OOE). 3. Esterified PEG, mol. wt. 300 (PEG300E).
The ‘esterified’ samples (CH,CO(OCH,CH,),OCOCH,) had the OH end-groups of PEG end-estermed with acetic acid. The completeness of the esterification reaction (- 96%) was verified by ‘H NMR analysis. Lithium triflate was dissolved in the solvent in a glove box under a dry nitrogen atmosphere. Errors in salt concentration are estimated to be l-2%. Besides Raman and NMR investigations, such samples were also studied using a.c. conductivity and viscosity measurements, detailed elsewhere [8]. The NMR and initial Raman samples were sealed in glass tubes back-filled with dry nitrogen gas to minimise exposure to oxygen and moisture (water contamination could alter conductivity measurements by about 1% for samples left for 24 h in a poorly sealed container). Subsequent Raman samples were contained in capillary tubes sealed with paraffin film.
S.F. Johnston et al. I Solid State Ionics 90 (1996) 39-48
be proportional to the fraction of ions in the various states. To verify this the area of the peaks in the region 1000-1075 cm-’ (i.e. the region including all three peaks, if present) was measured as a function of triflate concentration for lithium triflate in DMF. The peak areas for DMF vary dramatically with salt concentration, and so provide a good test. The integrated intensities vs. salt concentration were measured for four temperatures and found to be linear within the experimental error. An additional concern is the dependence of Raman scattering upon the polarisation of the input laser beam, which normally allows quantitative work only if spectra are acquired using more than one input and output polarisation geometry. All experimental work described here employed vertical polarisation of the input beam and detection of the vertically polarised component of the output radiation scattered at 90”, denoted Vv polarisation geometry. To investigate whether this was adequate, spectra were acquired with three other polarisation states, all using 90 scattering: Vh, Hv and Hh, where the letter H signifies horizontal and the upper/lower case letter signifies input/output. The results for a sample of tetraglyme + lithium triflate indicate that the scattered radiation is completely polarised, and thus that the Vv polarisation provides reliable quantification. A final consideration in analysing the Raman data is the attribution of the ‘free’ ion peak. Bernson and Lindgren [9] have recently suggested that the 1032 cm-’ peak in PPG + Li triflate should be attributed to solvent-shared ion pairs as well as to free ions. The terms ‘solvent-separated’, ‘solvent-shared’ and ‘contact’ ion pairing refer to the relative proximity of the anion and cation as discussed by Marcus [lo]. Bernson and Lindgren cite infrared evidence to support their argument that systems having hydroxyl end-groups contain solvent-shared ions. From the present Raman (and unpublished infrared) data, however, no qualitative difference between PEG (hydroxyl end-group) and the other systems was noted. If, nevertheless, the Bernson and Lindgren interpretation is correct, then the measured free ion fractions for at least the PEG system are somewhat high, because they would have overestimated the free ion fraction with respect to ion pairs. The determinations of the fraction of contact-ion paired
or aggregate changed.
ions,
41
in either
case,
would
be un-
2.3. NMR measurements The NMR pulse field gradient spin echo (PFG) technique was used to determine ‘Li and 19F self diffusion coefficients for some of the samples at various triflate concentrations and sample temperatures [ll]. The range of concentrations and the temperature range were constrained by the applicability of the NMR method. An extensively modified Bruker SXP 100 MHz spectrometer was used, with a Varian V-7400 fifteen inch magnet providing the static field. In the PFG experiment, a 90” resonant rf pulse is applied at time zero, a 180” pulse at time r, and a spin echo is found at time 27. A square shaped field gradient pulse with amplitude and direction characterised by G and duration S is applied between the two rf pulses. A second identical gradient pulse follows the 180” pulse at a time A after the first one. The echo amplitude will be attenuated by an amount dependent upon how much the spins have changed their position by self diffusion in the interval A. It can be shown that the attenuation of the echo amplitude is given by: R = exp{ - y2S2G2(A
- S/3)D}
(1)
where D is the self diffusion coefficient and y is the gyromagnetic ratio of the spin. Typically, S =7 ms, A= ~=20 ms and G would be varied from 0.2-2 Tm-‘, so that D can be determined from a plot of log R vs. G2. By applying the Nernst-Einstein equation, u ca,c = Nq2(D+ + D_)lkT
(2)
where N is the number of anions per cm3 of solution, q is the charge per ion, and D, and D_ are the diffusion coefficients of the cation and anion, respectively, and then defining a parameter Sdlff to represent the difference between this calculated conductivity and the measured conductivity,
S.F. Johnston et al. I Solid State tonics 90 (1996) 39-48
42
(T meas =
gccalc(l -
(3)
‘diff)
and finally assuming that the difference is due principally to the coupled diffusion of anions and cations, then a,,,, is a measure of the extent of ion association. Thus (1 - adi,,) is identified as the free ion fraction.
3. Results and discussion 3.1. DMF (Raman) Raman spectra of 9 samples of salted DMF all show the dominant ‘free’ ion peak near 1032 cm-‘. A secondary ‘hump’ centred near 1041 cm-’ for EO:Li concentrations below 15: 1, and the suggestion of a third peak centred on 105 1 cm-’ for concentrations below 5: 1, are attributed to contact-ion pairs and ion aggregates, respectively. Fig. 1 plots the free, paired and larger aggregate ion fractions at room temperature. Except at the highest salt concentrations, where aggregate fractions rise, changes in the free ion fraction are complemented by inverse changes in the fraction of contact-ion pairs. Similar observations have been made by Tore11 et al. [5]. DMF proves to be completely dissociated for salt concentrations below about 20: 1, a result in accord with a chemical shift data on the similar sodium triflate/DMF system [12].
3.2. Tetraglyme
(Raman)
For all salt concentrations, the free ion (1032 cm-‘) and contact-ion pair (1041 cm-‘) Raman peaks were prominent. The ion pair peak predominated, indicating that the salt was generally less than half dissociated. For the two highest salt concentrations (EO:Li ratios of 12:l and 15:l) a weak ‘aggregate’ shoulder was detectable at 1051 cm- I. At higher sample temperatures, the ion pair and aggregate peaks had larger amplitudes, indicating increased ion association. Free, contact-ion paired and aggregated ion fractions are plotted as a function of temperature in Fig. 2 for an EO:Li ratio of 12. Unlike DMF, it is notable that, as temperature rises, the fraction of contact-ion pairs appears to remain static while the drop in free ions is matched by a rise in the fraction of ion aggregates. All salt concentrations show a similar temperature dependence, i.e. a moderately reduced free ion fraction with increasing temperature. As with DMF and PEG, the free ion fraction in tetraglyme falls as the concentration of lithium triflate salt increases. The fraction of aggregated ions increases with salt concentration. 3.3. PVDF gels (Raman) 3.3.1. DMF +Li trijate system The gel samples containing PVDF, unlike the liquid tetraglyme + Li triflate samples, fluoresced when initially illuminated by the argon-ion laser.
~_*..,._.._.._.____-..* ._.._____-*_______-__*
2 0 p
25
50
o
0 O@o
0000
0
00
0
0
0
75
EO to LI concentration
0
20
40
60
80
Temperature (deg C) Fig. 1. Fraction of free ions (0). contact-ion pairs (0) and ion aggregates (0) for lithium triflate in DMF, from room temperature Raman measurements.
Fig. 2. Free (0). paired (0) and aggregate ion (A) fractions tetraglyme + Li triflate vs. temperature (EO:Li = 12: 1).
for
43
S.F. Johnston et al. I Solid State Ionics 90 (1996) 39-48
association with temperature the salted solvent.
Fig. 3. Raman spectra of DMF+ Li triflate without (top) 40% PVDF (EO:Li= 151).
with (bottom)
and
Spectra were therefore acquired following an exposure of 2 h, by which time fluorescence was negligible. Fig. 3 shows subsequent room temperature spectra for DMF+ Li triflate with and without PVDF. The spectra differ in two respects. Firstly, the contact-ion pair peak is relatively more prominent in the PVDF-containing sample. Secondly, the relative intensities of the two Lorentzian components in the compound 1080-1100 cm-’ peak differ in the two cases. Similar observations have been made previously for DMF samples having different lithium triflate concentrations. Analysis of the ‘ion association’ peaks shows that the addition of PVDF increases ion association at all measured temperatures. The increase amounts to some 1530%. Fig. 4 illustrates the rise in ion
for the gel as well as
3.3.2. Tetraglyme + Li triflate system In all cases, the 1032 cm-’ peak becomes relatively less prominent with increasing temperature. The very minor effect of adding PVDF is illustrated by the two room temperature spectra shown in Fig. 5. The free ion fraction for one concentration is plotted in Fig. 6, again illustrating the rise in ion association with temperature. As suggested by direct examination of the spectra, the PVDF-containing samples and PVDF-free samples yield similar values
ST,
“V’
Fig. 5. Raman spectra of salted tetraglyme with (bottom) and without (top) PVDF. EO:Li= 12:1, with 40% PVDF in lower curve.
5 f
20
40
Temperature Fig.
60
(deg C)
4. Free ion fraction vs. temperature for salted DMF with (0) and without (0) 40% PVDF (EO:Li= 15:l).
Temperature
(deg C)
Fig. 6. Free ion fraction vs. temperature for salted tetraglyme (0) and without (0) 40% PVDF (EO:Li= 12:l).
with
SF.
44
Johnston et al. I Solid State Ionics 90 (1996)
for free ion fraction, but with the PVDF-containing samples being consistently more associated. For these samples, and over the temperature range of 0-80°C the presence of 30-40% PVDF appears to reduce systematically the free ion fraction by some 5--20%.
6
39-48
,.....,....,,,..-
Q. 2
b,
4
'\
8 G= 5 p ti
3.4. Fits to temperature
dependence
,/' ,'
For all materials studied, the Raman results indicate that ion association increases with temperature. Although the quality of the data precludes precise fitting, the temperature-dependence conforms to the empirical expression Free ion fraction -f
= u{ 1 - exp( -b/T)}
__.~...................--------4
a., 2.
,/ f,' ,'
“8
H 0
0
',"',"',,"',"
10
20
30
EO to LI concentration Fig. 7. Comparison of Raman (0) and NMR fractions for salted PEG3OOE at 348 K.
(0)
free ion
(4)
where a and b are fitting constants and T is expressed in kelvin. An alternate representation, plotting log (f) vs. 1 /T, indicates a linear relationship within the experimental uncertainty. Three-parameter expressions such as the reduce-temperature Arrhenius dependence f=A exp(-Bl(TTO)) are not adequately constrained by the data. These results are qualitatively similar to those found by Schantz et al. [3] for PPG complexed with NaCF,SO, and LiClO,. They found a decrease in free ions with an increase in temperature and attributed this to entropic effects, arguing that the formation of ion pairs provides greater freedom to the polymer chains by removing free ions, which act as transient cross-links. Schantz et al. found their temperaturedependence data to be well described by an equation of the form Eq. (4). They also found that the free ion fraction fell with increasing salt concentration, particularly below a molality of about m = 1.
results is limited by the few materials studied using both techniques. Fig. 7 plots the free ion fraction inferred from NMR diffusion measurements and from Raman spectroscopy for PEG3OOE as a function of concentration and at one temperature. Fig. 8 shows the temperature dependence of ion association inferred by the two techniques for similar EO:Li ratios. Conductivity, viscosity and further NMR data related to this system have been published elsewhere [ 13,141. Despite the limited data, some broad points of similarity can be cited. The NMR results indicate that the end-esterified PEG3OOE system does not conduct as well as the hydroxyl-terminated PEG300, presumably due to its lower solvating ability. It also exhibits a distinct
3.5. PEG (NMR and Raman) Raman spectra of samples of salted PEG300 display, as with DMF, a contact-paired ion peak at higher salt concentrations. 3.5.1. Trends in ion association Analysis of the NMR measurements made on some of these systems demonstrates both similarities and differences with respect to the Raman data. The amount of comparison between the Raman and NMR
"0
20
40
M)
80
Temperature (deg C) Fig. 8. Comparison of free ion fraction vs. temperature, determined by Raman and NMR, for PEG3OOE. Raman: EO:Li= 3O:l (V), 15:l (Cl); NMR 20~1 (*).
S.F. Johnston et al. I Solid State Ionics 90 (19%)
curvature, this deviation being most likely caused by changing ion association below about 1 mol kg-‘. These two observations - the dependence of association on end-group and on salt concentration - are borne out by the Raman interpretations. A plot of conductivity against salt concentration for OH-capped polymers in Fig. 9 shows a lack of significant deviation from log-linearity (the uncertainty in the conductivity data was less than 1% at the lowest values, and improved as the temperature and conductivity increased). A change in the degree of association over this large concentration range would reveal itself as a deviation from this behaviour most commonly as an horizontal S-shaped curve, as observed by MacCallum et al. [15], Cameron et al. [16] and Hall [17]. Such deviation was seen for the PEG2OOE/lithium triflate system (Fig. 10). The initial drop at low concentration is ascribed to ion pair formation, as is known to occur for solvents of low dielectric constant. The subsequent increase beyond the minimum, which occurs at about 0.1 mol kg-‘, could be a result of either triple ion formation or a redissociation effect, as put forward by Cameron et al. [16]. The minimum can be seen to shift to lower concentrations with increasing temperature on close examination of Fig. 11. This shift is consistent with further ionic aggregation, from ion pairs to higher charged aggregates which can contribute to the conductivity. Alternatively, we can picture the situation at these higher concentrations as a clustering of ions, with single ions being able to move to
-’ I i
2
-a-
B
E
i
E
_g-
4 2 pf
-10
-
E -117 0
1
Concentration,
2
3
molal
Fig. 9. Molal conductivity vs. concentration for various polymers with lithium triflate. at 303 K: PEG300 (k), PEG3OOE (A) and PEG600 (0).
1
-9.5
0
39-48
45
1 0.2
0.4
0.6
0.8
Concentration,
Fig. 10. Molal conductivity lithium triflate at 298 K.
1.0
1.2
1.4
molal
vs. concentration
for PEG2OOE with
-7.0 -_
1
L I
z0
-9.0 -9.5
I 1
0.1
0.2
0.3
0.4
(Concentration)“‘.
0.5
0.6
0.7
0.8
molal
Fig. 11. Molal conductivity against concentration for PEG2OOE with lithium Uiflate at 298 K (A), 308 K (0). 318 K (V), 328 K (A,, 338 K (0). 348 K (V).
different positions within a cluster, a view suggested by the dynamic modelling of Forsyth et al. [18], and the work of Boden et al. [193. The NMR self diffusion coefficient measurements throw further light on this issue of ion association. It can be seen from Table 1 that the values of free ion fraction, which should be unity for a fully dissociated system, are typically around 0.3 but increase significantly as the salt concentration is increased. This result is consistent with the previous NMR results of Boden et al. [19] on PEG6OO/lithium triflate systems. The present and published Raman results show the opposite trend, however. The results imply that ion association increases significantly at concentrations below about 1.4 mol kg-’ (12: 1 EO:Li) i.e. below the commonly seen conductivity
S.F. Johnston et al. I Solid State Ionics 90 (1996) 39-48
46 Table 1 Free ion fractions measurements EO:Li ratio
for PEG3OOEILi
Temperature
triflate solutions
(K)
from NMR
Free ion fraction
6:l
348
0.46kO.08
8:l
328 338 348
0.37~0.05 0.43kO.05 0.40~0.04
12:l
318 328 338 348
0.25?0.04 0.29kO.03 0.29t0.03 0.21 kO.03
2o:l
308 318 328 338 348
0.25 +0.03 0.27+0.03 0.24+0.02 0.21+0.02 0.20+0.01
The diffusivities of ‘Li and 19F are roughly equal, which is not surprising for a system in which perhaps 60-70% of the ions are associated. maximum.
3.5.2. Effect Table 2 concentration inferred from
of end-group
lists the free ion fraction vs. salt for four of the systems studied, as Raman measurements. The tetraglyme
Table 2 Free ion fraction vs. salt concentration on Raman measurements
at room temperature,
based
Material
EO:Li ratio
Free ion fraction
Tetraglyme
12:l 15:l 18:l 24:l 30: 1 40: 1 6O:l 8&l 1OO:l
0.26+0.03 0.27+0.02 0.21+0.04 0.31 to.05 0.63-cO.04 0.75?0.04 0.5420.04 0.79?0.12 0.53kO.11
PEG300
1O:l 24:l 36:l 50: 1
0.63-tO.03 0.79?0.04 0.83kO.04 0.76+0.03
PEG3OOE
8:l 15:l 30: 1
0.093?0.01 0.285t0.03 0.37kO.02
12:l 24: 1
0.065 t0.006 0.14kO.02
PEG2OOE
and PEG data suggest that ion association is reduced below a salt concentration of 0.6tO.l mol kg-‘. As tetraglyme is essentially a particular weight variant of PEG, and all the systems have molecular weights in the range 200 to 300, the effects of other influences - in this case end-group - can be noted. The highest free ion fraction is found in those polymers having hydroxyl end-groups (PEG300), methyl end-groups (tetraglyme) and ester end-groups (PEG3OOE), respectively. This behaviour is qualitatively similar to that reported recently by Forsyth et al. [20]. For the NMR measurements, values of the diffusion coefficients in hydroxyl-capped polymers, such as PEG300, were not systematically studied. However, one value of D( 19F) was obtained for a solution of PEG300 with lithium triflate at 13: 1 concentration at 45°C. This allowed a value of d to be calculated for this concentration assuming that D( ‘Li)= D(19F)), which was found to be zero, within experimental error. This result implies that the salt is fully dissociated, in contrast to the end-esterified systems looked at so far. The Raman results suggest a slightly lower value of about 80% dissociated. In either case, the two techniques suggest that the hydroxyl groups make a crucial contribution to the solvating power of these polymers. We can further investigate the effect of the endgroup substitution by comparing the PEG300 and 300E systems. Fig. 12 shows the conductivities of these two systems, as well as the expected con-
-51
-a
1
1
I 3
2
Concentration,
molal
Fig. 12. A comparison of measured and calculated for PEG300 and PEG3OOE: PEG3OOE measured measured (A), PEG3OOE calculated (0).
conductivities (4), PEG300
S.F. Johnston et al. / Solid State Ionics 90 (1996) 39-48
ductivity of the PEG3OOE system calculated from the Nemst-Einstein equation. It can be seen that the conductivity of the hydroxyl-ended PEG300 is very similar to that expected for the PEG3OOE system from its measured diffusivities. This is further evidence for nearly complete dissociation in the PEG300 polymer systems, which explains the lack of deviation from log-linearity in their molal conductivity against temperature plots. A value of D( 19F) for PEG600 with lithium triflate was also measured, at 13: 1 concentration and 45°C. In this case the estimated value of d was 0.25kO.08. This higher value may be a result of a lower concentration of hydroxyl groups, although the assumption that D(‘Li)=D( 19F) for this system may also be erroneous. However, this value is in the range expected from the work of Leng [21], who calculated A values for the same system at higher temperature and found that they increased with temperature and decreased with salt concentration. 3.6. Raman vs. NMR data In a few particulars, the Raman and NMR data sets appear to differ. The free ion fraction according to Raman spectroscopy is somewhat lower than that inferred from NMR. Moreover, the data indicate that the free ion fraction drops as salt concentration falls, while Raman results suggest the opposite trend. Furthermore, some of the NMR data suggest that ion association falls as temperature rises, a result at odds with the present Raman work. It has to be recognised that the NMR measurements may be somewhat inaccurate, so that these differences between the NMR and Raman measurements may be in several instances not insignificant, particularly at low ion concentration and for the ‘Li measurements. However, it does appear that, even taking a generous view of the inaccuracies of both techniques, there are significant differences both numerically and in terms of the trends. It is evident that the Raman and NMR techniques are probing different time scales of the ionic species. PFG NMR is observing ionic motions of the orders of milliseconds, while Raman spectroscopy detects configurational environments that are temporally orders of magnitude shorter. Finally, Raman spectroscopy is sensitive to the relative proximity of anions and
47
cations, while the NMR technique independently senses the diffusion of the various ion types.
4. Conclusions The Raman technique appears to be applicable to DMF, tetraglyme and PEG samples, and to PVDF gels. For these various salt/solvent/polymer systems, several similarities are apparent: Ion association is strongly affected by end-group. Hydroxyl, methyl and ester end-groups, respectively, exhibit decreasing free ion fractions. Both contact-ion pairs and ion aggregates are found. Ions in DMF+ Li triflate appear to have little or no contact-ion pairing or aggregation for salt concentrations below an EO:Li fraction of 20: 1. The fraction of ions associated as contact-ion pairs rises dramatically as salt concentration rises from 20: 1 to 3:l. The addition of PVDF to either salted DMF or salted tetraglyme causes a slight reduction in free ion fraction. The effect appears to be more marked for DMF than for tetraglyme. For all samples measured, anion association, in the form of contact-ion pairs and larger aggregates, rises with temperature. The quality of the data precludes precise fitting, but the results are consistent with either an Arrhenius or VTF temperature dependence. As to the applicability of the Raman/infrared surement technique itself:
mea-
The measurement of ion association proved feasible for all the sample systems for salt concentrations above about 1OO:l. The measurement precision of the free ion fraction was typically in the +5% to 220% range. Other workers have suggested that the Raman peaks assigned here solely to free ions may be due partly to solvent-shared ion pairs. These conclusions do not appear to be confirmed for the systems studied here. In either case, the data imply that association becomes stronger with increasing temperature.
48
S.F. Johnston et al. I Solid State lonics 90 (1996) 39-48
There are clear differences between the estimated free ion fraction measured by Raman spectroscopy and by a combination of NMR diffusion and conductivity measurements for lithium triflate/PEG systems. These are attributed to the different temporal scales probed by the two techniques.
References [II D.W. James and R.E. Mayes, Amt. J. Chem. 35 (1982) 1775. VI S. Schantz, L.M. Tore11 and J.R. Stevens, J. Appl. Phys. 64 (1988) 2038. [31 S. Schantz, J. Chem. Phys. 94 (1991) 6296. [41 S. Schantz, L.M. Tore11 and J.R. Stevens, J. Chem. Phys. 94 (1991) 6862. [51 L.M. Torell, P. Jacobsson and G. Petersen, Polym. Adv. Techn. 4 (1992) 152. [61 A. Brodin, B. Mattsson, K. Nilsson, L.M. Torrell and J. Hamara, Solid State Ionics 85 (1996) 111. [71 A. Bemson, J. Lindgren, W. Huang and R. Frech, Polymer, in press. F31 P. Southall, H.VSt.A. Hubbard, S.F. Johnston, V Rogers, G.R. Davies, J.E. McIntyre and I.M. Ward, Solid State Ionics 85 (1996) 51.
[9] A. Bemson and J. Lindgren, Polymer 35 (1994) 4842. [IO] Y. Marcus, in: Ion Solvation (Wiley, New York, 1985), Ch. 8. [ 1I] E.O. Stejskal and J.E. Tanner, J. Chem. Phys. 42 (1965) 288. [12] A.G. Bishop, D.R. MacFarlane, D. McNaughton and M. Forsyth, Solid State Ionics 85 (1996) 129. [13] J. Cruickshank, H.V.St.A. Hubbard, N. Boden and I.M. Ward, Polymer, in press. [14] I.M. Ward, N. Boden, J. Cruickshank and S.E. Leng, Electrochim. Acta, in press. [ 151 J.R. MacCallum, A.S. Tomlin and C.A.Vincent, Eur. Polym. J. 22 (1986) 787. [16] G.G. Cameron, M.D. Ingram, G.A. Sorrie, J. Chem. Sot., Faraday Trans. 83 (1987) 3345. [17] P.G. Hall, G.R. Davies, I.M. Ward and J.E. McIntyre, Polym. Commun. 27 (1986) 100. [ 181 M. Forsyth, VA. Payne, M.A. Ratner, SW. de Leeuw and D.F. Shriver, Solid State Ionics 53-56 (1992) 1011. [19] N. Boden, S.A. Leng and I.M. Ward, Solid State Ionics 45 (1991) 261. [20] M. Forsyth, M. Garcia, D.R. MacFarlane, P. Meakin, S. Ng and M.E. Smith, Solid State Ionics 85 (1996) 209. [21] S.A. Leng, PhD thesis, University of Leeds, 1988.
