You are on page 1of 9

Solid State Ionics 147 (2002) 249 – 257

www.elsevier.com/locate/ssi

Polymer electrolytes and polyelectrolytes:


Monte Carlo simulations of thermal effects on conduction
J.F. Snyder *, M.A. Ratner , D.F. Shriver
Department of Chemistry and Materials Research Center, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208-3113, USA

Abstract

Monte Carlo calculations were carried out to simulate ion diffusion through polymer matrices. A dynamic bond percolation
(DBP) model was employed that includes local harmonic motion of covalently bound anions in polyelectrolyte systems. The
temperature dependence of cation diffusion was investigated in polyelectrolytes and polymer – salt complexes for 0 – 100 C.
Systems in which the rate of polymer reorganization is independent of temperature display Arrhenius behavior both above and
below the Tg of 35 C. Systems in which the temperature is coupled to the rate of polymer reorganization display VTF behavior
above the Tg and near Arrhenius behavior below the Tg. In all cases, the temperature is coupled to the rate of successful ion
jumps. Temperature and Tg seem to have no effect on the ion density at which the cation conductivity reaches a maximum.
D 2002 Elsevier Science B.V. All rights reserved.

Keywords: Polymer electrolyte; Polyelectrolyte; Ion conductivity; Temperature dependence; Monte Carlo simulations; Dynamic bond
percolation model

1. Introduction complexes and polyelectrolytes. In polymer – salt


complexes, both ion species are free to diffuse
There has been considerable interest in polymer through the system. In polyelectrolytes, one ion is
electrolytes since Armand et al. [1] recognized their covalently bound to the polymer. For systems in
potential application in solid state batteries [2,3]. which the anion is bound, there is an inherent cation
These systems are not limited by the volatility of transference number of one. This property deters salt
liquid electrolytes or the brittleness of traditional solid buildup in lithium ion batteries. For polyelectrolytes
electrolytes. In addition, they can function as the in the present study, the anions are restricted and the
electronic separator between the electrodes [4]. Sol- lithium cation is the only independently mobile spe-
vent-free single-phase polymer electrolytes may be cies. It has been established that solvent-free poly-
classified into two major categories, polymer – salt electrolyte systems can produce reasonable ambient
temperature conductivities (10  5 S/cm) [5]. The
target conductivity for commercially viable lithium
*
batteries ( > 10  3 S/cm) has not yet been attained with
Corresponding authors. J.F. Snyder is to be contacted at these types of materials when they are free of small
Tel.: +1-847-491-3461; fax: +1-847-491-7713. M.A. Ratner. Tel.:
+1-847-491-5652; fax: +1-847-491-7713.
molecule solvents. It has been demonstrated that ion
E-mail addresses: snyderj@chem.nwu.edu (J.F. Snyder), diffusion occurs principally in the amorphous phase
ratner@chem.northwestern.edu (M.A. Ratner). of polymer electrolytes [6], but the details of ion

0167-2738/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 2 7 3 8 ( 0 2 ) 0 0 0 2 5 - 5
250 J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257

transport are not well understood. We have modeled fusive, the cation transference number is frequently
transport phenomena to understand the mechanistic observed to be less than 0.5. Polyelectrolytes prevent
differences between polymer – salt complexes and these problems by prohibiting counterion diffusion.
polyelectrolytes, and to explore the dependence of However, clusters may still form around the immobile
conductivity on temperature and anion motion. anions. Cations can be trapped in local potential wells
Considerable effort is being devoted to the theo- [20], reducing the overall conductivity. We have pre-
retical study of polymer electrolytes through the use viously shown, through an improvement to the DBP
of molecular dynamics [7– 11]. The studies of ion model [23,24], that local anion motion in polyelec-
dynamics presented here have been performed using trolytes can significantly diminish the negative effects
Monte Carlo calculations, which do not have the of ion correlation and improve cation conductivity,
advantages (full dynamics) or the computational lim- while retaining unit cation transference number. By
itations (short times) of molecular dynamics. Many means of this improvement, a broad range of polymer
previous attempts to describe ion motion have used electrolytes has been simulated and cation conductiv-
free volume type ideas [12 – 14]. Our model is based ity was shown to be optimized as a function of anion
on the dynamic bond percolation (DBP) model [15 – local motion and total ion density. In this current
17], which instead presents a largely microscopic work, we further investigate ion correlation through
description of motion in polymer electrolytes. The the use of this extended DBP model. In particular, the
DBP model incorporates the theory of static percola- effects on cation conductivity are examined as a
tion [18,19], which depicts particle motion as hopping function of temperature, in the range of 0 – 100 C.
between neighboring sites on a lattice. Particle diffu- The effects on cation conductivity for both polyelec-
sion relies on a randomly chosen set of available trolytes and polymer – salt complexes are examined in
pathways, labeled ‘bonds’. Through these pathways, cases when the temperature is either coupled or not
each particle has on average, access to a predeter- coupled to the rate of polymer rearrangement. Cation
mined number of adjacent sites. In the DBP model, conductivity is also studied at a variety of ion/site
the lattice sites represent ion coordinating sites on a densities at several temperatures.
polymer, and the bonds represent the possibility of
hopping to adjacent sites. Elastomeric motion is
simulated by periodically redistributing or ‘repercolat- 2. Simulations
ing’ the bonds at a time sR. The rate of redistribution
is proportional to the rate of segmental rearrangement Our model was recently described [23,24] and it is
and accordingly models the dependence of ion trans- very similar to the one used by Lonergan [20 – 22,25].
port on polymer motion. First-order chemical kinetics The sites of a rigid lattice gas model represent sites on
can describe the rate of hopping using the master the polymer that coordinate ions, such as etheric oxy-
equation: gens. Ions occupy the sites and diffuse through the
matrix by hopping. The lattice currently employed is
dPi X square planar (two-dimensional) with periodic boun-
¼ ðPj Wji  Pi Wij Þ ð1Þ
dt j
daries. The size of the lattice was varied in conjunc-
tion with the number of ions to determine the ion
where Pi is the probability of an ion being found at density. The smallest lattice used contains 2025 sites
site i and Wij is the rate of ion motion from site i to site (density = 0.5 – 0.9 ions/site) and the largest lattice
j. Lonergan [20] extended the DBP theory to include contains >105 sites (density = 0.01 ions/site). The lat-
charged particles. tice length is given as a multiple of the intersite spa-
Ion correlation can have a detrimental effect on cing. Results from the smallest lattices were verified
cation diffusion. When both species are freely mobile, through simulations using larger lattices. No finite
the cations and anions can move in pairs or small size effects were observed. The configurations of the
clusters resulting in reduced conductivity and devia- system can be expressed by
tion from Nernst – Einstein behavior [21,22]. Addi-
tionally, since the anions are often more freely dif- n ¼ fn1 ,n2 , . . . nN g ð2Þ
J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257 251

Using this notation, the potential energy for the


system is given by:
XN X N
ni nj q2
Esys ðnÞ ¼ : ð4Þ
i¼1 j>i
4pe0 erij

In Eq. (4), N is the total number of sites, q is the


ion charge, e is the selected dielectric constant, and rij
is the distance between sites i and j. Charge neutrality
is maintained by fixing qcation =  qanion. Experiments
are averaged over at least five independent simula-
tions to improve statistical accuracy. Except where
noted, the standard deviation of the data points is
within the symbol size. During each simulation, the

Fig. 1. These are two representations of the same data set. Cation
diffusion increases rapidly from the backbone polyelectrolyte value
to the polymer – salt complex value with only a minimal range of
anion motion. (a) Investigates this increase as a function of the
range of anion motion. (b) Explores it as a function of the force
constant H on a logarithmic scale. The data points are fit to a B
spline function. The simulations in this data set were run for 104
Monte Carlo Steps at T * = 0.25 with an ion density of 0.20 ions per
lattice site. The error bars on (a) are shown for both the cation tracer
diffusion coefficient (labeled as Dt) and the anion mean squared
displacement. The error was derived from statistical deviation of the
five simulations over which each data point was averaged.

where
8
>
> 1 anion occupation
>
>
<
ni ¼ 0 vacant : ð3Þ Fig. 2. T * = 0.25, ion density = 0.20. (a) Relates the rate of attemp-
>
> ted jumps that are successful, or blocked by other ions, with the
>
>
: force constant H. Error is within symbol size. (b) Examines the
1 cation occupation relative coulomb energy per particle for these same simulations.
252 J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257

Fig. 3. T * = 0.25. Cation/anion distribution functions for several ion densities. The solid line represents the polyelectrolyte limit and the dotted
line represents the polymer – salt complex limit. The data points are fit to spline functions.

system was evolved through randomly selected single where DEsys(n) follows from Eq. (4), ri is the position
particle hops from site i to nearest neighbor site j with of site i, and H is an effective harmonic force constant
probability per unit time Probij. Following the general that binds each moving anion to its initial position r0
scheme of percolation theory, Wij is set to 1 if the bond [24]. For freely mobile anions, H is 0, while for rigidly
is available and 0 if it is not. Since coulomb inter- bound anions, H is infinity. In all systems studied here,
actions were included, the Metropolis transition prob- the cations are freely mobile (Hcations = 0).
ability [26,27] was used for unblocked jumps to insure Each system was equilibrated over three consec-
the proper equilibrium distribution. Thus Probij is: utive steps: (1) high temperature annealing with
8 Wij = 1 for all i and j; (2) equilibration under exper-
>
> exp½DEij =kB T  for DEij > 0 imental temperature, but retention of Wij = 1 for all i
>
> and j; (3) equilibration with experimental temperature
<
Probij ¼ 1 for DEij V0 and bond allowances. Equilibration is indicated by
>
> fluctuation of the lattice energy around a mean value.
>
>
:
0 for nj a0,Wij ¼ 0 The appropriate values of H were applied during all
ð5Þ stages of equilibration. Once equilibrium was reached,
the systems were allowed to evolve for 104 Monte
where DEij refers to the overall change in energy in Carlo time steps (MCS), for each of which there was
the move, expressed by: Nm jump attempts. Bonds were periodically redistrib-
uted to simulate polymer segmental motion. The time
Ej  Ei ¼ DEij ¼ DEsys ðnÞ þ 2H ½ðrj  r0 Þ2 lengths of experiments allowed convergence of the
properties of the system. Coulomb interactions were
 ðri  r0 Þ2  ð6Þ calculated by Ewald summations [28,29]. Reduced
J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257 253

Fig. 4. Temperature dependence of cation conductivity at ion density = 0.30 ions/site. (a) Includes systems in which the polymer segmental
motion is not coupled to temperature. The time between bond renewals is 10 MCS in all cases. The behavior is Arrhenius. (b) Studies the effect
of coupling temperature to polymer motion. The time between bond renewals was calculated for each temperature by means of Eq. (8). The
parameters are otherwise identical to those in the simulations in (a). There is VTF-like behavior at temperatures above the Tg and Arrhenius
behavior (within the error of the simulations) at temperatures below the Tg.
254 J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257

units were used throughout so distances are given in atures by using s1 = s(T1 = 100 C) = 10 MCS in the
terms of the intersite lattice spacing, a; and thermal equation:
energy in terms of reduced temperature T * = kbT/e(a), A A
T
where e(a) = q2/(4pe0ea). s2 ¼ s1 e T2 T0 1 T0 : ð8Þ

For each set of parameters, the relative changes in


3. Results conductivity were compared for Hanion in both the
backbone polyelectrolyte limit (H ! infinity) and the
Several experimentally realizable parameters were polymer – salt complex limit (H ! 0). At the comple-
varied to explore cation diffusion as a function of tion of each simulation, the time-dependent mean
temperature for different anion mobilities. Simulations squared displacement (MSD), h (Dx(t))2i, was calcu-
were run at eight different total (anion plus cation) lated for each ion species from the migration of the
particle densities ranging from 0.01 to 0.9 for temper- particles over the simulation time. The MSD can be
atures ranging from T* = 0.183 to T * = 0.25. These directly related to the tracer diffusion coefficient:
temperatures correlate to a range of 0 –100 C for
hðDxðtÞÞ2 i
e(a) = 2.060
10  20 J. In a system with screened Dt ¼ ð9Þ
2dt
charges, this could correspond to a = 5 Angstroms and
e = 5.6. In comparison to real polymers, ePEO 5 [30] where d represents the dimensionality. d = 2 in all
and ePPO = 5.78 [31]. The bond density was set to 0.35 cases here. Dt can be used to find the conductivity, r,
on a square planar lattice, which is well below the through the corrected Nernst –Einstein equation:
static percolation limit of 0.50 for square planar
lattices [32]. This guarantees that significant long- 1 nq2 Dt
r¼ ð10Þ
range ion diffusion depends on periodic redistribution HR kB T
of the bonds, as is the case in real polymer electrolytes
[6]. The time between bond repercolations, s, was set where HR = Dt/Dr is the Haven ratio [20], a factor that
to be 10 MCS for one set of data, and adjusted compensates for ion correlations. Dr is the actual
according to temperature for others. A repercolation conductivity diffusion coefficient. HR was set to 1
time that is independent of temperature would be for all calculations.
expected to result in conductivity that exhibits Arrhe- Fig. 1 shows a significant dependence of ion motion
nius behavior, whereas a temperature-dependent on the force constant, H. Fig. 1a suggests that for a
repercolation time should result in conductivity that density of 0.2 ions/site, the most significant growth in
displays VTF type behavior. s is represented by [33]: cation Dt occurs when the anion motion initially
changes from frozen to local motion—a range of three
A
lattice sites, roughly 1.5 nm. This compares favorably
s ¼ ce T T0 ð7Þ
with previous results calculated using s = 100 MCS
where c is a constant, kBA is an activation energy and [24], which have additionally indicated that the cation
T0 is the temperature at which free volume goes to conductivity will increase faster with increasing anion
zero. A was set to 878 K. T0 Tg  50 K. T0 was set motion as the system becomes more dense. Very li-
to  15 C in all simulations here. This corresponds mited anionic motion can still generate a substantial
to the T0 elucidated from the (PEO)5LiSCN system increase in cation conductivity. Fig. 1b reveals a near
[1]. s was set to be 10 MCS for T * = 0.25 (T = 100 logarithmic increase in ion mobility as the restrictions
C). In simulations in which renewal was coupled to defined by H on the anions are relaxed. The slopes for
temperature, s can be calculated for other temper- the two ion species are very similar, but they do not

Fig. 5. Dependence of cation conductivity on ion concentration at three temperatures. The minima at the endpoints arise respectively from a very
low concentration of charge carriers and from severely reduced diffusion at high concentrations owing to blockage. The maxima occur at
different ion concentrations for the polyelectrolyte and polymer – salt complex limits. The simulations in (a) were run at T * = 0.25 (T = 100 C).
The temperature is reduced in (b) to T * = 0.2232 (T = 60 C), and in (c) to T * = 0.1964 (T = 20 C). Tg = f 35 C.
J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257 255
256 J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257

overlap one another. Furthermore, cation diffusion in- As the temperature decreases, polymer motion slows
creases over a range of H in which there is no long- down and the rate of bond repercolation decreases.
range anion motion—that is, while the anions are still Below a certain temperature (Tg), the lattice is essen-
‘‘tied down’’ to their sites of origin. This shows that the tially static. Polymer electrolytes are most effectively
increase in cation diffusion is due to something other modeled by means of this coupled expression.
than increased mobility of ionic pairs and clusters. Fig. 4 explores the behavior of these two types of
Fig. 2 explores this phenomenon further. The solid systems studied. In Fig. 4a, the temperature is coupled
line in Fig. 2a plots the variation in the rate of success to ion jumps alone. Arrhenius behavior for the cation
for attempted jumps for each species. The dotted line conductivity is evident in both the polymer – salt com-
traces the rate of unsuccessful jumps that resulted plex and polyelectrolyte limits. The slopes of the
from physical blockage by another ion. The error is conductivity curves appear to be independent of anion
within symbol size. Although Fig. 1b indicates that mobility. In Fig. 4b, the temperature is also coupled to
there is no long-range anion movement at high values temperature by means of polymer renewal. Above the
of H, Fig. 2a shows that local motion does occur. In Tg (chosen to be 35 C) there is clear VTF behavior.
fact, the cation diffusion curve in Fig. 1b appears to Below the Tg, the lattice is essentially static and the
correlate well with this rate increase in anion success- conductivity curve displays Arrhenius behavior (with-
ful jumps. There also appears to be an inverse in the error of the simulations). This is reasonable since
correlation between the rates of successful jumps long-range cation diffusion is coupled to polymer seg-
and blocked jumps for cations. That is, the fluctuation mental motion.
in successful jumps is caused both by energetic effects Fig. 5 investigates the dependence of cation con-
and by physical blocking. This suggests a strong ductivity on ion density for three temperatures. All of
initial increase in clustering, which is supported by these simulations were of the second type; that is, the
the energy/particle minimum at H = 100 in Fig. 2b. rate of bond renewal was coupled to the temperature.
Fig. 3 compares the degree of clustering at either All of the curves are effectively zero at the endpoints
limit of H over the range of simulated particle and show a maximum in between. At very low ion
densities by means of the cation –anion radial distri- densities, there are very few charge carriers so the con-
bution function g(r). At low and moderate densities, ductivity is effectively zero. At very high densities, the
including d = 0.2 ions/site, there is significant cluster- systems become so crowded the ions have nowhere to
ing. As the concentration of ions is raised, the large move. In our previous studies [24], we have noticed
first peak in g(r) is reduced and the systems begin to that the maximum conductivity in polymer – salt elec-
resemble one large cluster with distributions that are trolytes is at a higher density than that for polyelec-
characteristic of glassy substances. In the case of trolytes. This shift is apparent here as well. Inter-
polymer –salt complexes, the system equilibrates into estingly, temperature and the Tg do not seem to affect
a highly ordered ion – counterion –ion network. In the this shift. Fig. 5a and b are above the Tg and Fig. 5c is
polyelectrolyte limit, this network is hindered by the below the Tg. Temperature does affect the relative
initial, random placement of static anions so a highly increase in conductivity between the two curves. At
ordered system is precluded. lower temperatures, anion mobility seems to make a
To understand better the role of temperature more significant difference in cation conductivity. This
dependence in polymer electrolyte systems, two types can be attributed to a greater influence of anion trap-
of systems were simulated. In one, temperature is ping on cation diffusion as the temperature is lowered.
coupled only to ion jumps. That is, only the temper-
ature and force constant H were varied. The time of
bond repercolation was set to be independent of 4. Conclusions
temperature (10 MCS in all simulations). In the
second type of system, temperature is coupled to We have found that the nature of cation conduc-
polymer segmental motion as well as anion mobility tivity is highly dependent on effective temperature,
[34 – 36]. That is, the time of bond repercolation was the rate of bond repercolation, particle density, and
recalculated for each temperature by means of Eq. (8). local anion mobility. A substantial fraction of the
J.F. Snyder et al. / Solid State Ionics 147 (2002) 249–257 257

conductivity difference between polyelectrolytes and [7] S.W. de Leeuw, A. Van Zon, G.J. Bel, Electrochim. Acta 40
polymer – salt complexes can be regained by allowing (2001) 1419.
[8] E. Hackett, E. Manias, E.P. Giannelis, Chem. Mater. 12 (2000)
for local anion motion in polyelectrolytes. The minor 2161.
range of anion motion required to promote significant [9] O. Borodin, G.D. Smith, Macromolecules 33 (2000) 2273.
changes in cation conductivity suggests this would [10] J.W. Halley, Y. Duan, L.A. Curtiss, A.G. Baboul, J. Chem.
still retain a unit cation transference number. Poly- Phys. 111 (1999) 3302.
[11] S. Neyertz, J.O. Thomas, D. Brown, Comput. Polym. Sci. 5
electrolytes in which the anion is on the terminal end
(1995) 107.
of a flexible side chain [37] may undergo local motion [12] H. Vogel, Phys. Z. 22 (1921) 645.
as described in this model. The flexibility and length [13] G. Tammann, W. Hesse, Z. Anorg. Allg. Chem. 156 (1926)
of the side chain shape the degree of anionic ‘‘bind- 245.
ing’’ described by the force constant H in these [14] G.S. Fulcher, J. Am. Ceram. Soc. 8 (1925) 339.
studies, i.e. the boundaries of the local motion. The [15] S.D. Druger, A. Nitzan, M.A. Ratner, J. Chem. Phys. 79
(1983) 3133.
ratio of anionic side chains to etheric oxygen side [16] R. Granek, A. Nitzan, J. Chem. Phys. 92 (1990) 1329.
chains determines the ion density of the system. [17] S.D. Druger, J. Chem. Phys. 100 (1994) 3979.
Arrhenius behavior is evident in the systems for [18] R. Landauer, J. Appl. Phys. 23 (1952) 779.
which temperature is not coupled to polymer rear- [19] S. Kirkpatrick, Rev. Mod. Phys. 45 (1973) 574.
rangement. In the systems for which the rate of poly- [20] M.C. Lonergan, Ion Transport in Polymer Electrolytes, PhD
Thesis, Northwestern University, Evanston, IL, 1994, p. 319.
mer rearrangement is dependent on temperature, VTF [21] M.C. Lonergan, J.W. Perram, M.A. Ratner, D.F. Shriver, J.
behavior is apparent for temperatures above the Tg, Chem. Phys. 98 (1992) 4937.
and Arrhenius behavior is noticeable below the Tg. The [22] M.C. Lonergan, D.F. Shriver, M.A. Ratner, Electrochim. Acta
nature of anion mobility does not seem to affect the 40 (1995) 2041.
[23] J.F. Snyder, M.A. Ratner, D.F. Shriver, in: R.A. Marsh, Z.
nature or slope of these curves. Anion mobility affects
Ogumi, J. Prakash, S. Surampudi (Eds.), Lithium Batteries,
the ion density dependence of conductivity though, in The Electrochemical Society Proceedings Series, Honolulu,
that the conductivity maximizes at a higher density for HI, 1999, p. 563, PV 99-25.
polymer – salt complexes than for polyelectrolytes. [24] J.F. Snyder, M.A. Ratner, D.F. Shriver, J. Electrochem. Soc., in
These maxima are unaffected by temperature or press.
whether the system is above or below the Tg. Com- [25] Y. Boughaleb, M.A. Ratner, in: A.L. Laskar, S. Chandra
(Eds.), Superionic Solids and Solid Electrolytes, Academic
parable simulations on cubic lattices are underway. Press, Boston, 1989, p. 515.
[26] N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Tell-
er, E. Teller, J. Chem. Phys. 21 (1953) 1087.
Acknowledgements [27] K. Binder, Topics in Current Physics, vol. 7. Springer-Verlag,
New York, 1986, p. 411.
This work was supported by the U.S. Department [28] S.W. de Leeuw, J.W. Perram, E.R. Smith, Proc. R. Soc. Lon-
of Energy, LBL, the ARO, and the Northwestern don, Ser. A 373 (1980) 27.
[29] S.W. de Leeuw, J.W. Perram, E.R. Smith, Annu. Rev. Phys.
Materials Research Center. We would also like to
Chem. 37 (1986) 245.
thank Dr. Abraham Nitzan for his help and insight in [30] M.B. Armand, in: J.R. MacCallum, C.A. Vincent (Eds.), Poly-
this work. mer Electrolyte Reviews, vol. 1. Elsevier, New York, 1987,
p. 3.
[31] A.L. Tipton, Synthesis, Physical, and Electrical Characteri-
References zation of Polymer Electrolytes and Polymer Complexes Con-
taining Polyhalides, PhD Thesis, Northwestern University,
[1] M.B. Armand, J.M. Chabagno, M.J. Duclot, in: P. Vashishta, Evanston, IL, 1992, p. 47.
J.N. Mundy, G.K. Shenoy (Eds.), Fast Ion Transport in Solids, [32] D. Stauffer, Introduction to Percolation Theory, Taylor & Fran-
Elsevier, North-Holland, 1979, p. 131. cis, Philadelphia, PA, 1985, p. 17.
[2] F.M. Gray, Solid Polymer Electrolytes, VCH, New York, 1991. [33] S.D. Druger, M.A. Ratner, A. Nitzan, Solid State Ionics 9 – 10
[3] C.A. Vincent, Electrochim. Acta 40 (1995) 2035. (1983) 1115 – 1120.
[4] W.H. Meyer, Adv. Mater. 10 (1998) 439. [34] C.A. Angell, Solid State Ionics 9 – 10 (1983) 3 – 16.
[5] L.C. Hardy, D.F. Shriver, J. Am. Chem. Soc. 107 (1985) 3823. [35] C.A. Angell, Solid State Ionics 18 – 19 (1986) 72 – 88.
[6] C. Berthier, W. Gorecki, M. Minier, M.B. Armand, J.M. [36] L.M. Torell, C.A. Angell, Br. Polym. J. 20 (1988) 213.
Chabagno, P. Rigaud, Solid State Ionics 11 (1983) 91. [37] J.G. Cowie, G.H. Spence, Solid State Ionics 123 (1999) 233.

You might also like