Anion-trapping and polyanion electrolytes based on acid-in-chain borate polymers

Transcription

1 Electrochimica Acta 48 (2003) 2255/ Anion-trapping and polyanion electrolytes based on acid-in-chain borate polymers Wu Xu, Xiao-Guang Sun 1, C. Austen Angell * Department of Chemistry and Biochemistry, Arizona State University, Tempe, AZ , USA Received 19 May 2002; accepted 25 September 2002 Abstract Oligoether branched-and-spaced acid-in-chain polymers, with variable length side chains attached to the acidic borons have been prepared by a simple two-step reaction sequence. These may be used as anion-retarding hosts for salt-in-polymer electrolytes or, alternatively, may be converted to polyanionic electrolytes by reacting with strong Lewis base anions. High ionic conductivities, reaching 7.6/10 5 S cm 1 at 25 8C have been obtained for the former case, using LiTFSI:B proportions of 1:1, and optimized side chain and spacer lengths. The electrochemical windows are wide enough ( /4.5 V) for most applications. # 2003 Elsevier Science Ltd. All rights reserved. Keywords: Anion-trapping; Acid-in-chain; Borate polymers 1. Introduction * Corresponding author. address: caa@asu.edu (C. Austen Angell). 1 Present address: Environmental Energy Technologies Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA. Solvent-free salt-in-polyether electrolytes show relatively high ionic conductivity. However, their conductivities are dominated by anions because the cation, i.e. Li, is strongly chelated by the basic ether oxygen atoms in the chains. The lithium ion transference number (t Li ) is below 0.3 [1,2]. One way to improve the cation transference number is to bind the anions in or onto the polymer chains, when t Li /1 is obtained. Another way is to temporally trap anions by electrondeficient (Lewis base) atoms incorporated in the polymer chains, to form coordinated structures in which t Li is substantially increased. Earlier [3] we reported a class of Lewis acid additives for electrolytes in which borate groups [obtained by forming boric acid esters with glycols (BEGs)] interacted with anions to yield high salt solubilities despite the very low solvent dielectric constants. Related boron-containing additives were reported a little later by McBreen and co-workers [4,5]. BEG-like polymers were briefly investigate in [6], but anion-trapping polymers were not studied in detail until Mehta et al. [7 /9] reported examples based on in-chain boroxine rings. In [7] a t Li of about 0.7 was measured for LiBF 4 and LiSO 3 CF 3 by the combined dc polarization/ac impedance method. The conductivities of these boroxine ring-containing systems were only about 10 5 S cm 1 at 30 8C. Recently [10] we reported a versatile type of chain polymer in which the acidic boron occurs in an open chain rather than a ring structure. Here the Lewis acid groups are separated by oligoethers of variable length. Such acid-in-chain polymers were then turned into polyanions by reacting with a Lewis base anion, leaving the counter-cation free to conduct. Depending on the Lewis base strength of the added anion, this type of system could range from strictly polyanionic to weakly anion trapping in nature. The conductivities of such polymer electrolytes were in the range of 10 6 to 10 4 S cm 1 and t Li was either unity (strong base anion) or above 0.3 for LiTFSI, based on diffusivities measured by the pulsed field gradient NMR spin echo method [11]. However, the boron-containing starting materials for these polymers, phenyl- or methyl-boronic acid [10], are very expensive. Here we describe a simple inexpensive alternative preparation in which short polyether chains replace the pendant phenyl or methyl groups of the /03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved. doi: /s (03)

2 2256 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/2266 original boronic diacid reactant. The polyether chain is added in a preliminary reaction of controlled stoichiometry that produces the selected diacid. This is then reacted as before with the polyether spacer groups. The ionic conductivity of the new polymer electrolytes also ranges from anion-retarding to polyanionic, depending on the salt dissolved. Due to the flexible structure of the pendant oligoether group, the polymer electrolytes from the acid-in-chain borate polymers synthesized in this work have higher ionic conductivities than those with the rigid phenyl structure. The thermal behavior, ionic conductivity and electrochemical properties will be discussed in detail. 2. Experimental section 2.1. Materials The oligoethers used in this work were all from Aldrich Chemical Co. and included 2-methoxyethanol, tri(ethylene glycol) methyl ether, poly(ethylene glycol) methyl ether with molecular weight of 350 and 550 [all can be represented as CH 3 O(CH 2 CH 2 O) m H or MEGmOH, where m/1, 3, 7.2 (:/8) and 11.8 (:/12)], poly(propylene glycol) mono(n-butyl) ether with molecular weight of 430 {n-c 4 H 9 O[CH 2 CH(CH 3 )O] m H or BPG m OH, m/4.6 (:/5)}, di(ethylene glycol), poly(ethylene glycol) with molecular weight of 200, 400 and 600 [all can be represented as HO(CH 2 - CH 2 O) n H or PEG n, n/2, 4.2 (:/5), 8.7 (:/9), 13.2 (:/14)] and poly(propylene glycol) with molecular weight of 725 {HO[CH 2 CH(CH 3 )O] n H or PPG n, n/ 12.2 (:/13)}. The oligoethers were refluxed azeotropically with benzene for 2 days to remove residual water followed by evaporation of the solvent on a rotavapor at reduced pressure, and then dried in a high vacuum oven at about 90 8C for 2 days. Boric acid, lithium hydroxide monohydrate, lithium trifluoromethanesulfonate (Li- SO 3 CF 3 ), lithium thiocyanate (LiSCN), lithium cyanide (LiCN), sodium cyanide (NaCN), lithium sulfide (Li 2 S), lithium methoxide (LiOCH 3 ) were also from Aldrich and used as received except LiSCN which was dried at about 200 8C in high vacuo. Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) was provided gratis by 3M Company and was used as received. Lithium bis(oxalato)borate (LiBOB) was synthesized as described previously [12]. Lithium 2,2,2-trifluoroethoxide (LiOCH 2 CF 3 ) was synthesized by reacting lithium metal with 2,2,2-trifluoroethanol. tube, were placed 0.05 mol boric acid, 0.05 mol poly(ethylene glycol) monomethyl ether or poly(propylene glycol) monobutyl ether and some anhydrous benzene. The reaction mass was stirred and heated in an oil bath of about 100 8C until a clear solution was obtained and no more water was evaporated from the reaction. The benzene solution was cooled to room temperature. Then 0.05 mol poly(ethylene glycol) or poly(propylene glycol) was added quickly into the above reaction solution. The mixture solution was stirred and heated again in the oil bath of 100/110 8C until no more water was released (about 3 days). The solvent in the viscous reaction mass was evaporated on a rotavapor at reduced pressure. The residual polymer was then dried in a vacuum oven at about 90 8C under high vacuum for 2 days, and kept in a dry glove box after drying. The polymers of this study are abbreviated as P(YZB), where Y represents the pendant side chain on the in-chain boron atoms, chosen from MEG m and BPG m, Z is the spacer chain between boron atoms in the main chain, chosen from EG n and PG n, m and n are the numbers of the repeating oxyalkylene units in the side chain and main chain spacer, respectively, and B means borate. The polymers synthesized in this work are depicted as P(MEG m BEG n ), P(BPG m BEG n ) or P(MEG m BPG n ) according to the oligoethers used. The formation of a typical polymer is shown in the following reactions (Scheme 1) Preparation of poly[oligoether-branched borate of oligo(alkylene glycol)] In a dry flask equipped with stirring bar, water separation apparatus, condenser and CaCl 2 drying Scheme 1. The formations of a typical acid-in-chain polymer and its polyanionic electrolyte.

3 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/ Preparation of polymer/salt complexes In this work the molar ratio of boron to anion (B/X ) was fixed to be 1:1 except for the case of Li 2 S where B/ S 2 /2:1. About 2 g of the polymer of Section 2.2 was weighed out in a Vacuum Atmospheres Company (VAC) dry box and dissolved in anhydrous tetrahydrofuran (THF). A quantitative amount of Li or Na salt was added into the solution. Shaking or stirring the mixture was needed to dissolve the salt and to make the solution homogeneous. Filtration was applied for the cyanide and sulfide salts in which a fraction of the base remained undissolved. The fraction undissolved was small and estimated less than 10%. The solvent in the clear solution was evaporated down on a rotavapor at reduced pressure. The residual liquid was dried further in a vacuum oven at about 90 8C at high vacuum for 2 days, and stored in a dry box. The general procedure is summarized in Scheme Measurements The polymers synthesized in this work were all characterized by NMR spectroscopy. 1 H and 13 C NMR spectra of these polymers in DMSO-d 6, using TMS as internal reference, were obtained on a Varian Gemini 300, while 11 B NMR spectra were collected using a Varian Inova 400, with BF 3 /Et 2 O as external reference. The thermal behavior of the polymers and their polymer/salt complexes was determined by a simple differential thermal analysis (DTA). Actual and differential temperatures were recorded by a two-pen HP 7132A recorder during reheating process at approximately 20 K min 1 after initial quenching in liquid nitrogen to low temperature. The values of glass transition temperature (T g ) were obtained from the onset of the glass transition. The devitrification temperature (T c ) where longer oligoether segments became ordered, and the liquidus temperature (T l ) where they remelted, were obtained from the peak temperature of the crystallization exotherm and the melting (redissolution) endotherm, respectively. Electrical conductivities of the complexes were determined by ac impedance measurement as a function of temperature using a HP 4192A LF Impedance Analyzer in the frequency range from 5 Hz to 13 MHz, using either dip-type cells containing two parallel platinum discs (for liquids) or block-type cells containing two stainless steel block electrodes (for solids). The cell constants of the dip-type cells were from 0.7 to 1.3 cm 1, calibrated by a 0.1 m KCl aqueous solution. The cell constant of the block-type cells depended on sample thickness but typically was about 0.3 cm 1. The electrochemical properties were measured using standard cyclic voltammetry with an EG&G Potentiostat/Galvanostat Model 273, with a three-electrode dipcell with stainless steel (SS) or platinum (Pt) disc (surface area for SS is 1.96/10 3 cm 2 and for Pt 4.91/10 4 cm 2 ) as working electrode and lithium as counter and reference electrodes. 3. Results and discussion 3.1. Pure polymers Table 1 shows the physical appearances and thermal properties of the pure polymers, poly[oligoetherbranched borate of oligo(alkylene glycol)]. The physical properties of previous acid-in-chain polymers with pendant groups of phenyl (Ph), 2,4-difluorophenyl (DFPh) and methyl (Me) [10,11] are also included in Table 1 for comparisons. The polymers formed from short pendant oligoethers such as m/1 and 3 are soft gel-like materials while all other polymers are very viscous liquids. This indicates that the molecular weight of the polymers is not very high. The polymers synthesized in this work are actually oligomers due to the dramatic slowdown in polymerization reaction rate with increasing length. The molecular weight can be increased easily by crosslinking the oligomers with LiBH 4 or other crosslinkers. This part of work is in process and will be reported in subsequent papers. Characterization of the acid-in-chain polymers was carried out by NMR spectroscopy. The typical peaks of proton and carbon for CH 2 CH 2 O or CH 2 CH(CH 3 )O groups were obtained from the 1 H and 13 C NMR spectra. The NMR chemical shifts of 11 B in the polymers were all located in the range 23.5 /24.5 ppm, which is typical of three-coordinated boron atoms. The glass transition temperatures (T g ) of the pure polymers with pendant oligoethers are located between /80 and /65 8C, much lower than for the polymers with the same spacer chain but aryl pendants. It is also seen that, for the same pendant oligoether group, e.g. MEG 3 or MEG 8, T g increases slightly with increasing the length of the oligoether spacer. However, for the same oligoether spacer, e.g. EG 9, T g shows a minimum (at MEG 3 ) with increasing length of the pendant oligoether group [/65.0 8C (MEG 1 ), to /74.3 8C (MEG 3 ), /71.6 8C (MEG 8 ) and /71.0 8C (MEG 12 )]. All polymers except for those with short lengths of pendant and spacer oligoethers exhibited exothermic devitrification during warm-up with subsequent endothermic dissolution/melting. These processes involve the long oxyethylene (CH 2 CH 2 O or EO) units. The longer the EO group, the higher the T l and the closer the T c is to T g.

4 2258 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/2266 Table 1 Physical appearances and thermal properties of boron containing acid-in-chain polymers P(YZB) Pendant type (Y) Spacer type (Z) Pendant length (m ) Spacer length (n) Physical appearance DTA results T g (8C) T c (8C) T l (8C) MEG m EG n 1 9 Viscous liquid /65.0 /25.9 / Soft gel /80.0 / / 5 Soft gel /78.5 / / 9 Viscous liquid /74.3 /38.5 / Viscous liquid /72.2 / Viscous liquid /74.4 /43.1 / Viscous liquid /74.3 /33.9 / Viscous liquid /71.6 / Viscous liquid /70.4 / Viscous liquid /71.0 / PG n 8 13 Viscous liquid /72.2 /37.4 /11.2 BPG m EG n 5 9 Viscous liquid /69.2 /28.5 /1.8 Ph EG n / 9 Viscous liquid /53.8 / / / 14 Viscous liquid /59.5 / PG n / 13 Viscous liquid /58.5 / / DFPh EG n / 14 Viscous liquid /56.5 / Me EG n / 14 Viscous liquid /69.8 / Structure and thermal behavior of polymer-salt complexes When the lithium salts of very weakly basic anions such as TFSI, CF 3 SO 3, BOB or even SCN were added into the above poly[oligoether-branched borate of oligo(alkylene glycol)], the solutions cannot be described as polyanionic electrolytes, because the interaction between the anion and the electron-deficient boron center is relatively weak. These electrolytes are very viscous polymer solutions in appearance. The T g of these solutions is much higher than for the original polymers, by differences ranging from 13 to 51 8C (Table 2). The longer the oligoether groups in the polymer, the smaller increase in T g. On the other hand, the shorter the oligoether pendant group, the larger the T g increases. For example, when LiTFSI was added into P(MEG 3 EG n B), T g increases from /80.0 to /28.8 8C for n/2, from /78.5 to /36.5 8C for n :/5, from /74.3 to /45.9 8C for n :/9, and from /72.2 to /52.3 8C for n :/14. This is because the polymer with longer oligoether group has a lower salt concentration, so that the local viscosity of the polymer electrolyte is lower. It is seen in Table 2 that the addition of LiTFSI and LiBOB prevent the crystallization of long EO chains while the other two salts inhibit but do not exclude it, especially for polymers with both long pendant and spacer oligoether chains. The suppression of oligoether ordering reflects the greater concentration of salt in the polyether medium and away from the boron centers due to the weaker acid /base interaction. This is also the reason for the higher T g values relative to these observed for the polyanionic ruptures. T g values are especially high for the LiBOB-containing systems perhaps because of the rigidity of the double ring structure of BOB. By contrast, when lithium salts with strong Lewis base anion such as CH 3 O, CF 3 CH 2 O, S 2 or even CN are added to the pure polymer, the T g of the polyanionic electrolyte does not increase much above that of the original polymer and the polyanionic system shows polyether crystallization and melting processes during warm-up. Evidently the ions are concentrated at the boron sites, leaving the oligoether segments free to order. For the very strong basic alkoxide anions, there are evidently two distinct reactions, as evidenced by the precipitation of a rubbery polymer on mixing of alkoxide and polymer solutions. Part of the alkoxide anions will attack the electron-deficient (under-coordinated) boron atoms to form fully coordinated borate structure with one oligoether side chain and another alkoxide side group. However, a large part must react with the terminal hydroxyl groups of the polymer to generate small alcohol molecules and macromolecular alkoxide anions. The latter attach to the under-coordinated boron atoms of neighboring chains to produce the high molecular weight (crosslinked) precipitate. For the strongly basic sulfide anion (S 2 ) case, there is no reaction between S 2 anion and the terminal hydroxyl group of the polymer chain. Due to its divalent structure, it likes two under-coordinated boron atoms in the same or different polymer chains to form crosslinking polymer electrolytes. There are no such crosslinkings of the polymer chains for the moderate strong Lewis base anion CN. These anions bind to the under-coordinated boron atoms to form fully coordinated borate polyanionic electrolytes

5 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/ Table 2 Thermal properties of polymer/salt complexes Salt Polymer (YZB) DTA results Branch type Spacer type Branch length (m ) Spacer length (n ) T g (8C) T c (8C) T l (8C) LiTFSI MEG m EG n 1 9 /42.6 / / 3 2 /28.8 / / 5 / / 9 /45.9 / / 14 /52.3 / / 8 2 /45.6 / / 5 /48.1 / / 9 /49.7 / / 14 /56.5 / / 12 9 /57.9 / PG n 8 13 /56.2 / / BPG m EG n 5 9 /51.4 / / Ph EG n / 9 /37.8 / / / 14 /47.0 / / EG n / 13 /45.0 / / DFPh EG n / 14 /46.2 / / Me EG n / 14 /52.5 / / LiSO 3 CF 3 MEG m EG n 1 9 /32.5 / / 3 2 /30.9 / / 5 /42.0 / / 9 /46.4 / / 14 /53.4 / / 8 2 /47.6 / / 5 /48.1 / / 9 /53.4 / / 14 /55.4 / /55.7 / LiBOB MEG m EG n 1 9 /24.8 / / 3 9 /33.6 / / 8 2 /31.7 / / 5 /37.4 / / 9 /47.2 / / 14 /53.7 / / 12 9 /51.1 / / LiSCN MEG m EG n 8 2 /40.1 / / 5 /49.5 / / 9 /49.5 / / 14 /49.2 / NaCN MEG m EG n 8 2 /63.7 / / 5 /61.8 / / 9 /64.2 / /64.6 / LiCN MEG m EG n 8 2 /69.5 / /68.4 /42.6 / /76.1 /43.9 / /69.8 / LiOCH 3 MEG m EG n 8 9 /73.7 /53.9 / /64.0 / LiOCH 2 CF 3 MEG m EG n 8 9 /66.0 / /67.8 / Li 2 S MEG m EG n 8 9 /66.0 /44.8 / /67.8 / with two pendant groups, one is oligoether side chain and the other is the cyano group. The polyanionic structure needs to be confirmed by lithium ion transference number measurements, e.g. by electrophoretic NMR studies [13] or other methods [14] Ionic conductivities of polymer /salt complexes Fig. 1(a) and (b) show the temperature dependence of ionic conductivities of 1:1 (mol) LiTFSI/P(XEG 9 B) and LiTFSI/P(XEG 14 B), respectively, where X represents

6 2260 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/2266 Fig. 2. Temperature dependence of ionic conductivities of 1:1 (mol) LiSO 3 CF 3 /P(MEG m EG 9 B). These polymer electrolytes should be considered as anion-trapping rather than polyanionic electrolytes. Fig. 1. Temperature dependence of ionic conductivities of 1:1 (mol) (a) LiTFSI/P(XEG 9 B) and (b) LiTFSI/P(XEG 14 B) (with longer linkers), respectively. These polymer electrolytes should be considered as aniontrapping rather than polyanionic electrolytes. The data symbolized by ', 2, and m are taken from Ref. [10]. MEG m and BPG m in this work and Me, Ph and DFPh are reproduced from the previous report [10,11]. It is clear that the polymer electrolytes synthesized in this work have higher conductivities than the previous examples. The difference is due to the higher T g (Table 2) of the previous cases caused by the rigidity of phenyl side group. Figs. 2 and 3 show the benefit of the new side group by Arrhenius plots for the compositions 1:1 (mol) LiSO 3 CF 3 /P(MEG m EG 9 B) and LiBOB/ P(MEG m EG 9 B), respectively. In these the length of the side group MEG m is varied. It is seen that the longer the pendant oligoether group, the higher the conductivity of the polymer /salt complex. This is correlated to the lower T g values (Table 2). The conductivities for polymers with pendant oligoether length of 8 and 12 are nearly the same, indicating an approach to the optimal effect of the pendant group. The highest ambient conductivity of 7.6/10 5 S cm 1 has been obtained for 1:1 (mol) LiTFSI/P(MEG 8 EG 14 B). The improvement over previous acid-in-chain polymers is also demonstrated in Fig. 4 where the three cases with phenyl side groups have the lowest conductivities. Figs. 4/7 exhibit the Arrhenius plots of conductivity for anion-trapping polymer electrolytes formed by P(MEG 8 YG n B) with LiTFSI, LiSO 3 CF 3, LiBOB and LiSCN, respectively, where YG n represents EG n and PG n. This provides a test of the effect of the spacer structure. For the same salt and the same pendant oligoether group, the anion trapping polymer electrolyte having longer EO spacer has higher conductivity due to its lower T g (Table 2). The propylene glycol spaced polymer electrolyte shows much lower conductivity than Fig. 3. Temperature dependence of ionic conductivities of 1:1 (mol) LiBOB/P(MEG m EG 9 B). These polymer electrolytes should be considered as anion-trapping rather than polyanionic electrolytes.

7 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/ Fig. 4. Ionic conductivity vs. reciprocal temperature plots of 1:1 (mol) LiTFSI/P(MEG 8 YG n B) with comparison of those of LiTFSI/ P(PhYG n B) taken from [10], where YG n is chosen from EG n and PG n. the ethylene glycol spaced analogue because of its weaker ionic dissociating ability (Fig. 4). Fig. 8(a) and (b) exhibit the temperature dependence of ionic conductivities of polyanionic electrolytes formed by reaction of P(MEG 8 YG n B) with LiCN and NaCN, respectively. Except for the equal conductivity for Li and Na in the case of spacer length 14, we find higher conductivities for Li over Na. Again the difference can be correlated with T g differences (Table 2). Although there was some cyanide salt undissolved which resulted the B/X molar ratios for these cases are slightly larger than 1, the fraction undissolved was small, estimated less than 10%. The conductivity is not a strong function of composition near the 1:1 value so that our conclusions are probably not sensitive to Fig. 6. Ionic conductivity vs. reciprocal temperature plots of 1:1 (mol) LiBOB/P(MEG 8 EG n B). incomplete dissolution. The same reason is also applied to the sulfide case. Fig. 9 compares the conductivities of the polymer complexes formed by the same polymer host but different salts. The conductivities show a decreasing order of LiTFSI/LiSO 3 CF 3 /LiSCN/LiBOB/ LiCN/NaCN/Li 2 S/LiOCH 3 /LiOCH 2 CF 3. As described in Section 3.2 for the cases of polyanionforming electrolytes, alkoxide anions may lead to effective crosslinking of polymer chains among undercoordinated boron atoms and terminal oxyalkylene anions of polymer chains. Thus the polymer main chain may not move freely and only the pendant oligoether chains can move. The crosslinking caused by sulfide anions is only between sulfur and boron atoms so polymer main chain and pendant oligoether chain can move. There are no crosslinkings in the case of CN Fig. 5. Ionic conductivity vs. reciprocal temperature plots of 1:1 (mol) LiSO 3 CF 3 /P(MEG 8 EG n B). Fig. 7. Ionic conductivity vs. reciprocal temperature plots of 1:1 (mol) LiSCN/P(MEG 8 EG n B).

8 2262 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/2266 Fig. 9. Comparisons of Arrhenius plots of ionic conductivities of polymer/salt complexes from by P(MEG 8 EG 9 B) with different lithium salts. Note that the polymer electrolytes are anion trapping electrolytes for LiTFSI, LiSO 3 CF 3, LiSCN and LiBOB, and polyanionic electrolytes for Li 2 S, LiOCH 3, LiOCH 2 CF 3, LiCN and NaCN. Fig. 8. Temperature dependence of ionic conductivities of 1:1 (mol) (a) LiCN/P(MEG 8 EG n B) and (b) NaCN/P(MEG 8 EG n B), respectively. Note that the polymer electrolytes formed are polyanion electrolytes. anions. Thus the conductivity of the polyanions follows the order of alkoxideb/sulfide B/cyanide. Combination of conductivity and T g data permits us to learn something about the freedom of Li cations to migrate independent of the polymer segmental relaxation. We obtain this information from the decoupling index R t /t s /t s [15,16], where t s is the structural relaxation time and t s is the conductivity relaxation time. At T g where the structural relaxation time is approximately 100 s [17], so the decoupling index is given by the approximate relation log R t /14.3/ log s Tg [18]. To estimate the conductivity at T g we extrapolate the conductivity data to low temperatures using the well-known Vogel /Fulcher /Tamman (VFT) equation: ss o exp[b=(t T o )] (1) or ss o exp[d s T o =(T T o )] (2) where s o is the pre-exponent conductivity, T o is the vanishing mobility temperature and B is a constant characteristic of the system and the process under study. Provided conductivity is coupled to viscosity (see below), D s is inversely proportional to the fragility of the electrolyte [19]. Values of the VFT parameters such as s o, B, T o and D s are summarized in Table 3. Also included in Table 3 are the differences between T g and T o, the conductivity of the polymer electrolytes at T g (s T g ) and the corresponding decoupling indexes R t. For the polymer/salt complexes, from anion trapping polymer electrolytes to polyanion electrolytes, the values of s o and B are consistent with expectations for dissociated electrolytes, and normally decrease as expected with decreasing Li concentration as increasing the pendant chain (m) or the spacer chain (n). There is some indication of increasing ionic association with increasing m or n value. Although the extrapolations to T g are long, and hence must be regarded with much caution, the values of s Tg show some significant trends. They indicate that as the Li ion concentration increases, the logarithmic decoupling index (log R t ) increases, sometimes from negative to positive values, just as seen in other salt-in-polymer electrolytes [20]. Unfortunately, the extent of decoupling predicted by the VFT equations is unreliable in some cases of the salts with strong Lewis base such as lithium alkoxides. Note that T o parameters for these salts is unrealistically much lower than the T g value, which induces values of s Tg and R t that are also unrealistic. This is because, with

9 Table 3 VFT parameters from best fitting for polymer/salt complexes Salt Polymer (YZB) s o (S cm 1 ) B T o (K) D T g /T o (K) log s Tg log R t Branch type (Y) Spacer type (Z) Branch length (m) Spacer length (n ) LiTFSI MEG m EG n / / / / /23.2 / /15.5 / / / /15.8 / / PG n / BPG m EG n / Ph EG n / /20.4 /6.1 / /18.3 /4.0 PG n / /24.3 /10.0 DFPh EG n / /14.4 /0.1 Me EG n / /16.0 /1.7 LiSO 3 CF 3 MEG m EG n / / / / / / / / /14.9 / / LiBOB MEG m EG n / /16.6 / /15.8 / /15.7 / /15.8 / /17.8 / /15.9 /1.6 LiSCN MEG m EG n / / / / / / / / NaCN MEG m EG n / / / /17.5 /3.2 LiCN MEG m EG n / /15.0 /0.7 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/

10 2264 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/2266 Table 3 (Continued) Salt Polymer (YZB) so (S cm 1 ) B To (K) D Tg/To (K) log st g log Rt Branch type (Y) Spacer type (Z) Branch length (m) Spacer length (n ) 5 4.4/ /29.6 / / /6.9 / / / /0.6 / / LiOCH 3 MEG m EG n / /55.5 / / / / LiOCH2CF3 MEGm EGn / / / Li 2 S MEG m EG n / / / /16.0 /1.7 longer spacer units, Li ions can become chelated and then are not free to move on the segmental relaxation time scale. Consequently they behave like paired ions giving rise to a temperature dependence of conductivity, which is not included in the equations. Hence the equation parameters are distorted and the predictions outside the range of data are unreliable. In the cases of LiCN and LiOCH 3 electrolytes, on the other hand, some values of T g /T o are negative, which is quite unphysical. This is a result of the very small range of conductivity data ( B/1 order of magnitude) being fitted in these cases. No extrapolations can be made in such cases Effect of mixed anions The effect of mixed anions on the ionic conductivity of the polymer electrolytes has also been studied because of the finding that in glassy ionic conductors, mixing the anions leads to an increased decoupling of cations from anions, and consequently to a higher ambient temperature conductivity. Fig. 10 shows the effect of mixed anions on ionic conductivities of anion trapping polymer electrolytes where the molar ratio of anion 1 to anion 2 to boron is 0.5:0.5:1. The thermal properties of these mixed anions polymer electrolytes are summarized in Table 4. Unfortunately this effect does not seem to be manifested in our systems, probably because the anions are too diluted with polyether groups. It is, however, quite possible that in the polymer-in-salt regime (which we will report using our polyanionic materials as one component [21]), we may find the mixed anion effect to be useful. Fig. 10. Effect of mixed anions on temperature dependence of ionic conductivities of 0.5:0.5:1 (molar ratio of anion X1 to anion X2 to boron) LiX1/LiX2/P(MEG 8 EG 14 B) anion trapping polymer electrolytes.

11 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/ Table 4 DTA results, VFT parameters of complexes of P(MEG 8 EG 14 B) and mixed lithium salts with molar ratio of 1:0.5:0.5 Salt 1 Salt 2 DTA results s o (S cm 1 ) B T g (K) D T g /T o (K) log s Tg log R t T g (8C) T c (8C) T l (8C) LiTFSI LiSO 3 CF 3 /55.9 / / / LiSCN LiTFSI /56.2 / / / LiSO 3 CF 3 /53.1 / / / LiClO 4 /52.3 / / / LiI /55.4 / / Electrochemical stability The electrochemical stability of the anion trapping polymer electrolytes was studied on stainless steel and platinum electrodes by cyclic voltammetry. The results of the first cycle for three electrolytes formed by a typical polymer, P(MEG 8 EG n B), with three different lithium salts, are shown in Figs. 11/13. The polymer electrolytes have good electrochemical stability up to 4.7 V versus Li /Li. The preliminary scans of cyclic voltammograms also indicate that the first and second cycle Coulomb efficiencies of the lithium deposition/ stripping process are about 12.7 and 19.4% for LiTFSI, 26.6 and 36.0% for LiSO 3 CF 3, and 48.7 and 54.7% for LiBOB, respectively. The low efficiency is caused by the reactions between lithium and terminal hydroxyl ( /OH) group in the not high molecular weight polymer hosts. The replacement of /OH group with alkyl groups or crosslinking may lead to higher chemical and electrochemical stability against Li and the results will be reported elsewhere. Fig. 11. Cyclic voltammographic results of 1:1 (mol) LiTFSI/ P(MEG 8 EG 14 B) anion trapping polymer electrolyte on stainless steel working electrode at room temperature. SS area: 1.96/10 3 cm 2. Scan rate: 5 mv s 1. Fig. 12. Cyclic voltammographic results of 1:1 (mol) LiSO 3 CF 3 / P(MEG 8 EG 14 B) anion trapping polymer electrolyte on stainless steel working electrode at room temperature. SS area: 1.96/10 3 cm 2. Scan rate: 5 mv s 1.

12 2266 W. Xu et al. / Electrochimica Acta 48 (2003) 2255/2266 Acknowledgements This work was supported by Mitsubishi Chemical Corporation. We acknowledge also Dr. M.D. Williams and Dr. R.A. Nieman in the NMR Laboratory of the Department for analyses on 11 B nuclei under US NSF grant #CHE References Fig. 13. Cyclic voltammographic results of 1:1 (mol) LiBOB/P(ME- G 8 EG 9 B) anion trapping polymer electrolyte on platinum working electrode at room temperature. Pt area: 4.91/10 4 cm 2. Scan rate: 10 mv s Conclusions The cheap new oligoether branched and spaced borate polymers with different length of pendant side chains and spacer chains have been easily prepared through a two-step reaction process. Anion-trapping polymer electrolytes or polyanion electrolytes can be prepared by simply adding different type of lithium salts with weak or strong Lewis base anions. These anion trapping polymer electrolytes have high ambient conductivity up to 7.6/10 5 S cm 1 at 25 8C and wide electrochemical stability window. [1] P.G. Bruce, M.T. Hardgrave, C.A. Vincent, Solid State Ionics 53/56 (1992) [2] W. Gorecki, M. Jeannin, E. Belorizky, C. Roux, M. Armand, J. Phys.: Condens. Matter 7 (1995) [3] S.S. Zhang, C.A. Angell, J. Electrochem. Soc. 143 (1996) [4] H.S. Lee, X.Q. Yang, C.L. Xiang, J. McBreen, J. Electrochem. Soc. 145 (1998) [5] X. Sun, H.S. Lee, X.Q. Yang, J. McBreen, Electrochem. Solid- State Lett. 1 (1998) 239. [6] C.A. Angell, K. Xu, S.S. Zhang, US Patent no. 5,849,432 (1998). [7] M.A. Mehta, T. Fujinami, Solid State Ionics 113/115 (1998) 187. [8] M.A. Mehta, T. Fujinami, T. Inoue, J. Power Sources 81/82 (1999) 724. [9] M.A. Mehta, T. Fujinami, S. Inoue, K. Matsushita, T. Miwa, T. Inoue, Electrochim. Acta 45 (2000) [10] (a) X. Sun, C.A. Angell, Electrochim. Acta 46 (2001) 1467; (b) X. Sun, C.A. Angell, manuscript under preparation. [11] X. Sun, M. Videa, C.A. Angell, manuscript under preparation. [12] W. Xu, C.A. Angell, Electrochem. Solid-State Lett. 4 (2001) E1. [13] H.J. Walls, T.A. Zawodzinski, Jr., Electrochem. Solid State Lett. 3 (2000) 321. [14] M. Videa, W. Xu, B. Geil, R. Marzke, C.A. Angell, J. Electrochem. Soc. 148 (2001) A1352. [15] C.T. Moynihan, N. Balitactac, L. Boone, T.A. Litovitz, J. Chem. Phys. 55 (1971) [16] C.A. Angell, Solid State Ionics 9/10 (1983) 3 and 18/19 (1986) 72. [17] C.T. Moynihan, P.B. Macedo, C.J. Montrose, P.K. Gupta, M.A. DeBolt, 1976 Ann. New York Acad. Sci. 279 (1976) 15. [18] C.A. Angell, Annu. Rev. Phys. Chem. 43 (1992) 693. [19] M. Videa, C.A. Angell, J. Phys. Chem. 103 (1999) [20] M.G. McLin, C.A. Angell, Solid State Ionics 53/56 (1992) [21] W. Xu, L.-M. Wang, C.A. Angell, in: (a) abstracts of the Eighth International Symposium on Polymer Electrolytes, 19/24 May 2002, Santa Fe, NM, USA; (b) Electrochim. Acta. (2003) in press.