Chem. Rev.

1988, 88, 109-124 109

Ion Transport in Solvent-Free Polymers

M. A. RATNER* and D. F. SHRIVER*

Department of Chemistry and Materials Research Center, Northwestern University, Evanston, Illinois 60208

Received July 20, 1987 (Revised Manuscript Received September 15, 1987)

Contents situation, and the conductivity response.

109
Table I categorizes the classes of solid electrolytes
1. Introduction
1. Formation of Metal Salt Complexes 110
that have been extensively investigated. The classifi-
cation is subjective; a number of intermediate situations
3. Structure 111
occur and still other solid electrolytes do not fit into any
4. Ion Transport 112
a. Measurement 112
of these categories. The ceramic framework materials
are characterized by high Debye temperatures and
b. Temperature and Pressure Dependence 113
melting points; in these substances, which are generally
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

c. Interpretation of Ionic Conductivity: 114

Downloaded via UNIV OF WISCONSIN-MADISON on September 13, 2020 at 09:43:15 (UTC).

oxides or chalcogenides, conductivity is due to mobile

Phenomenological Models
ions hopping among energetically favorable sites in the
d. Frequency-Dependent Conductivity: 117
Dynamic Percolation and Relaxation
potential fixed by the framework. The soft framework
materials have far lower Debye temperatures and
5. Specific Polymer Conductors 119
a. Complexes of Linear and Branched 119 melting points; they are often halides, in which the
conduction is due either to ion hopping or to liquid-like
Polymers with Salts
b. Complexes of Comb-like Polymers with 119
diffusion of one ionic sublattice. The class of polymeric
Salts electrolytes is substantially different from the first two;
c. Network Polymers with Salts 119 polymeric materials show fast-ion behavior only above
d. Polyelectrolytes 120
the glass transition temperature and are generally
e. Proton-Conducting Polymers 120 softer, compliant materials. As will be discussed ex-
f. Polymer-Salt Complexes Containing 120 tensively in section 4, in polymer electrolytes the
motions of the polymer host are responsible for ionic
Dipositive Cations
g. Mixed Ionic and Electronic Conductors 120 mobility; the ions move only if polymer segments also
h. Inhomogeneous Systems 121 undergo fairly large-amplitude excursions. Below the
6. Problems and Prospects 121 glass transition, such chain motion does not occur, and
ionic conductivity drops to very small values. This
1. Introduction behavior stands in sharp contrast to framework inor-
ganic electrolytes, where motions of the framework
Solid electrolytes comprise a widely varied set of simply provide activation energy for ionic conduction
materials in which the ionic conductivity a is far higher through structural channels in the framework.
than that of typical ionic solids such as NaCl. The A number of polymer-based systems exhibit ionic
conductivity of typical solid electrolytes lies in the range conductivity. Our emphasis will be on solvent-free
(10-6 a ;S 10"1 S cm-1) characteristic of dilute aqueous polymer-salt complexes, where interest really began
ionic solutions.1 Solid electrolytes include refractory with the pioneering studies of materials based on alkali
covalent solids such as /3-alumina [(Na20)x'llAl203], metal salt complexes with polyethylene oxide), or PEO,
soft ionic crystals such as Agl, glasses such as Ag2GeSe3, reported by Wright7-9 and by Armand10,11 in the 1970s.
and among the most recently discovered and investi- Other types of ionically conductive polymer systems
gated species, polymer-salt complexes. have been prepared and are of great interest. These
Within the past 2 decades, the area of electroactive include solvent-swollen systems such as poly(vinyl al-
polymers has become one of the most challenging and cohol) /H3P0412 in which quite high conductivities are
fruitful realms of polymer science. Both electronically observed, due largely to ionic motion in solvent-rich
conductive polymers and polymeric electrolytes have regions of the swollen polymer host. Ionic motion is also
been prepared and studied in a large number of labo- seen in polyelectrolytes such as Nafion,13 in redox
ratories, and a good deal of both synthetic and mech- polymers,14 and in intercalated electronically conductive
anistic knowledge about these new polymer materials polymers such as lithium-doped polyacetylene. Tra-
has been gained. While these species share some of the ditional solvent-swollen polyelectrolytes such as Nafion
properties of more usual conductive systems such as or sodium poly(styrenesulfonate) are best thought of
metals, semiconductors, and ionic solutions, the poly- as concentrated solutions in which one ionic species is
meric structure provides a new set of conditions, so that very large and multiply charged and the mobile ion is
a number of new features appear in the electrical re- solvated. In recent work, both polar solvating groups
sponse. While reviews of the polymeric ionic conductors and charged groups have been covalently incorporated
recently have appeared,2"6 we are not aware of any into the polymers (see section 5d).15-21 Redox poly-
currently available overview of polymer electrolytes for mers14 such as poly[Os(bpy) (vpy)2(C104)x] (x =
0-3, bpy
a general chemical audience. This article will attempt =
bipyridyl, vpy vinylpyridyl) consist of mixed-valent
=

to survey, very selectively, the polymer electrolyte field; complexes covalently attached to the polymer chains
our emphasis will be on the chemistry, the structural along with counterions (such as halides or nitrate) that
0009-2665/88/0788-0109$06.50/0 © 1988 American Chemical Society
110 Chemical Reviews, 1988, Vol, 88, No. 1 Ratner and Shriver

The remainder of this article deals with solvent-free

polymer electrolytes based on polymer-salt complexes.
Section 2 is devoted to the preparation of these mate-
rials, and section 3 to their structure. Section 4 presents
a discussion of the conductivity in these polymer elec-
trolytes, concentrating on measurements and on the
interpretation of ionic mobility. Section 5 discusses
several particular examples of polymer electrolytes.
Finally, section 6 gives a brief outlook on current
problems.

2. Formation of Metal Salt Complexes

The polymer electrolytes with which we are princi-
Mark Ratner has been Professor of Chemistry at Northwestern
since 1980. His previous positions included graduate study at pally concerned are complexes of alkali metal salts,
Northwestern following undergraduate training at Harvard, post- denoted MX, with polymer hosts. Both the precursor
doctoral work in Aarhus and in Munich, and a faculty position at salt and the neat polymer are solids, so that the com-
NYU. His research interests lie in theoretical chemistry, both in
plex-forming reaction
methodology and, particularly, in the applications of theory to
chemical problems. Recent work from his group includes research mMX + (-RY-)n (MX)m-(-RY-)„
—
(1)
on coupled vibration problems, self-consistent field methods for
dynamics of molecules, van der Waals molecule dissociation, where (-RY-) denotes the polymer repeat unit, is a
polymer fracture, conductivity of molecular metals, nonlinear optical solid/solid reaction. As with most other reactions of
properties of molecules, and ionic conductivity in crystals, glasses, this type, the kinetics of (1) are unfavorable, even when
and polymers. Research in the last four areas involves close
collaboration with experimentalists, as does his substantial interest the complex is stable. Although other schemes for ac-
in canoeing. celerating (1) have been employed, including intimate
grinding/mechanical mixing,23 by far the most common
method has been to dissolve or suspend both the MX
salt and the host polymer in a common solvent and then
to remove the solvent, producing the solvent-free
polymer electrolyte in either bulk or thin-film form.24
Care must be taken to purify the starting materials and
(especially in the case of hygroscopic lithium electro-
lytes) to exclude water. Acetonitrile and methanol have
been the solvents most commonly used. If the poly-
mer-salt complex is partly crystalline, both the mor-
phology and the transport properties of the electrolyte
material produced may vary with choice of solvent.
Clearly, reaction 1 will be thermodynamically favor-
able (AG° negative) only if the Gibbs energy of solva-
tion of the salt by the polymer is large enough to ov-
ercome the lattice energy of the salt. In general, one
then expects a close relationship between the ability to
form homogeneous complexes and the ability of the
Du ward F. Shriver is Morrison Professor of Chemistry at North- monomer to dissolve the salt. Work by the Grenoble
western University. He received his Bachelor’s degree at the and Evanston groups has shown that for a given poly-
University of California, Berkeley, in 1958 and his Ph.D. at the mer host a fairly sharp demarcation line may be es-
University of Michigan in 1961. He joined the chemistry faculty at
Northwestern University in 1961, where he is a member of the tablished between salts that can and cannot form com-
Materials Research Center and the Catalysis Center. His research plexes; the latter simply have too large lattice energies
interests in preparative and physical inorganic chemistry have led (compare Table II). In addition to the very important
his research group to synthesize and characterize new inorganic lattice energy considerations, a number of other criteria
and polymeric electrolytes, as well as mixed ionic-electronic
conductors. In collaboration with Irving Klotz, he studied the that determine the possibility of forming complexes
resonance Raman spectra of biomolecules with special attention have been described. These include the following:11,34"38
to oxygen bonding in hemerythrin. His research group also is active (1) A high concentration of polar (basic) groups on the
in organometallic chemistry where the current emphasis is on the
polymer chain is needed to solvate the salt effectively.
reactivity of simple ligands on metal clusters and the relationships
between organometallic chemistry and heterogeneous catalysis. (2) The cohesive energy of the polymer cannot be too
high, and its flexibility, as indicated by a low glass
transition temperature, should be quite high, so that
move along the solvent-swollen polymer in response to reorientation of the local coordination geometry, to
electronic hopping among the mixed-valent metal sites. achieve effective solvation, may be achieved.
These are mixed ionic-electronic conductors. Interca- Most polymer electrolytes are based on oxygen-con-
lated electronically conductive polymer systems such taining monomers, including ethers in polyethylene
as lithium polyacetylide or (Si(pc)0)„+I3~ (pc = oxide) and poly(propylene oxide) and polysiloxanes,
phthalocyanine)22 are much like the redox polymers. carbonyl in poly(vinylpyrrolidone) or poly(ethylene
However, the electronic motion is of band, rather than succinate), and hydroxyls in poly (vinyl alcohol). Other
hopping, type; once again they are mixed conductors Lewis base groups have been employed, including ni-
(electrodes) rather than ionic conductors (electrolytes). trogen in poly(ethylenimine) and sulfur in poly(alkylene
Ion Transport in Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 111

TABLE I. Classes of Solid Electrolytes

ceramic framework materials soft framework materials polymers0
crystalline species glasses crystalline species glasses partly crystalline amorphous
LiAlSi04 LiAlSi04 (glass) Agl AgCl/Agl/CsCl
^-alumina ((Na20).,'llAl203) Ag^GeSe^* Ag2HgI4 LiSCN-PEO LiSCN-MEEP
Pbl2 NaCF3S03-PEI
Na2S04 NaCF3S03-PP0
“PEI =
poly(ethylenimine), PEO =
poly (ethylene oxide), PPO =
poly (propylene oxide), MEEP =
-(N=P(OC2H4OC2H4OCH3)2)„.

TABLE II. Salts That Form Complex Polymeric

Electrolytes with PEO“
Li+ Na+ K+ Rb+ Cs+
F“ no no no no no
1036 923 821 785 740
cr yes no no no no
853 786 715 689 659
Br yes yes no no no
807 747 682 660 631
I" yes yes yes yes yes
757 704 644 630 604
SCN" yes yes yes yes yes
807 682 619 616 568
CF3S03- yes yes yes yes yes
725 650 605 585 550

“The numbers reported are the lattice energies of the salts (in
kj/mol). “Yes” indicates polymer-salt complex formation and
“no” indicates the lack of complex formation. The stair-step line
indicates the division between complex formation and separate 1. Phase diagram for the PEO-NaSCN system. X de-
Figure
phases. notes mole fraction of NaSCN, L denotes liquid or elastomeric
phase, and cc denotes crystalline complex. SPEO is semicrystalline
sulfides). In general, Lewis base character on the com- PEO with NaSCN in the amorphous phase. From ref 43.
plexing host species is required to coordinate the cation
of the salt and thus provide a favorable Gibbs energy presence of a compound with the composition P-
of polymer-salt interaction. [(EO)3.5NaSCN].
There is considerable diversity in notation used to
3. Structure represent the phases, but the one used here appears to
be the most consistent and unambiguous. In this
Polyethylene oxide), or PEO, which has received the scheme P denotes a polymeric unit of unspecified size,
most attention as a host for electrolyte formation, is a and EO indicates the ethylene oxide repeat unit. Thus
semicrystalline material with about 60% of the bulk in the formula given above there are 3.5 ethylene oxide
being crystalline at room temperature and the remain- (OC2H4) units per NaSCN.
der present as an amorphous elastomeric phase. This As with the properties of free polymers, the properties
phenomenon of partial crystallinity carries over to many of the polymer-salt complex appear to have little if any
of the polymer-salt complexes. Thus the problem of variation with polymer molecular weight once a suitably
structure of the polymer electrolytes can be discussed high molecular weight has been achieved. This dem-
at two levels: the macroscopic identity and arrange- onstrates that the polymer motions relevant to ionic
ment of crystalline and amorphous phases and the conductivity are not the gross backbone diffusion of the
microscopic arrangement of atoms in the polymer and polymer backbone, which scale like M-2 (M = molecular
polymer-salt complex. weight) but rather the side-chain segmental motions,
The morphology of the crystalline phases can be ob- which are independent of molecular weight above a
served with an optical microscope. For both PEO and certain molecular weight.
PEO complexes the form of the polycrystalline phase It was clearly demonstrated by Berthier and co-
is often dendritic or spherulitic.39,40 It is not so easy to workers that ion transport preferentially occurs in the
determine phase diagrams for the polymer electrolytes amorphous phase of PEO-salt complexes.45 Thus the
as it is for simple inorganic substances, because the gross morphological structure of PEO-salt complexes
kinetics of crystallization can be very slow and because may play a role in determining the ion transport, be-
a certain amount of randomness is inevitable in many cause in the process of long-range diffusion an ion will
polymer systems. These complications lead to apparent have to circumnavigate the crystalline regions. This
violations of the Gibbs phase rule and to experimental unfavorable influence on ion transport is mitigated in
problems in determining the liquidus and solidus some applications by the structural integrity that the
boundaries. Despite these problems it has been possible crystalline phase lends to the bulk polymer-salt com-
to obtain useful and revealing phase diagrams for sev- plex. However, cross-linking of the amorphous polymer
eral PEO-salt systems.41-44 These phase diagrams host appears to provide a more favorable solution to the
demonstrate the presence of compounds having discrete structural integrity of the bulk polymer (section 5c).
ratios of PEO polymer repeat units to formula units of At the atomic level, detailed structural information
the salt. For example, the phase diagram for the on the solvent-free electrolytes remains an important
PEO-NaSCN system (Figure 1) demonstrates the but elusive goal. Although interest has shifted from
 112 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

TABLE III. Structures of Some Salt Complexes Inferred addition to the use of vibrational spectroscopy for
from X-ray Fiber Diffraction structural and dynamic information, infrared spec-
PEO material structural features diffraction ref troscopy is particularly sensitive for the detection of
P[(EO)4KSCN] K+ outside PEO helix 46 trace water and it should be more widely used in
P[(EO)sNaI] Na+: 30 at 2.24-2.61 A 47 polymer electrolyte research because water can exert
2P at 2.94-3.16 A a substantial influence on ion transport properties.
P[(EO)4HgCl2] Hg2+: 2 nearest-neighbor Cl a
2 nearest-neighbor 0
Recently, very useful information on the structure
and dynamics of polymer electrolytes has been obtained
"Iwanoto, R.; Saito, Y.; Ishihara. H.; Tadokoro, H. J. Polym. by NMR spectroscopy. For example, both chemical
Sci., Part A-2 1968, 6, 1509. shift and spin-lattice relaxation time data from 13C
NMR have shown that the likely region of coordination
partially crystalline materials to amorphous polymer- of Li+ to the comb polymer MEEP is at the oxygen
salt complexes, oriented fiber diffraction studies on atoms designated /? and y in structure l.56 Aside from
polycrystalline materials have provided useful insight
into the structures of polymer-salt complexes. It is £ r
(0—ch2ch2— o —

CH2CH2—OCH3)2
important to recognize that the oriented fiber technique
does not provide the degree of structural precision or- CP=r:N]„
dinarily associated with structures determined by sin-
gle-crystal X-ray diffraction. The oriented fiber X-ray
1

studies indicate the expected coordination of the cation structural information, NMR has proven to be very
to the oxygen atoms in the polyether and association useful for the determination of the dynamics of the
of the cation and anion.46-48 The most detailed X-ray polymer-salt complexes. Spin-lattice relaxation data
diffraction study to date was conducted on 100 reflec- provided the first conclusive demonstration that ion
tions and indicated that in P[(EO)3NaI], three ether motion preferentially occurs in the amorphous
oxygens and two iodide ions are coordinated to each phase,45,57-58 and pulsed field gradient NMR studies
sodium ion.47 provide detailed information on the diffusion coeffi-
Extended X-ray fine structure analysis (EXAFS) has cients for NMR-active nuclei.58,59
been explored as a means of probing the nearest-
neighbor environment of the cation in some poly- 4. Ion Transport
ethylene oxide)-salt complexes; however, the technique
has proven to be of rather limited utility because radial a. Measurement
distribution patterns contain a broad single feature.49,50
Apparently, inhomogeneity of the coordination envi- In the characterization of the electrical properties of
ronment and/or thermal motion present severe limi- a solid electrolyte the most basic and useful information
tations on the application of this technique. is the total conductivity and the fraction of this con-
Infrared, Raman, and NMR spectroscopies have ductivity that is attributable to each charge carrier.
provided useful insights into the structures and dy- There is a broad class of materials in which both ions
namics of the solvent-free polymer electrolytes. Direct and electrons are mobile; these interesting materials will
evidence for the polymer-cation interaction is available be discussed later in this review. In the electrolyte
from the observation of a cation vibration in the ether materials with which we are now dealing, conductivity
cage. These far-IR studies confirm the general model occurs by the migration of ions. The measurement of
of coordination of the cation by the ether oxygen atoms ionic conductivity, or ion transport as it is often called,
in polyethylene oxide) and related comb polymers is nontrivial, because of resistance to ion flow at the
(polymers with many short side chains).37,51,52 electrode-electrolyte interface. In general, it is much
The degree of ion pairing has also been inferred from more difficult to establish a low-resistance interface for
vibrational spectroscopy. In particular, the perturba- ion flow than for electron flow. This problem is largely
tion of vibrational modes for polyatomic anions has surmounted by the use of ac impedance spectroscopy,
demonstrated ion pairing between BH4-, NCS-, or N03- which is described in-depth elsewhere.60,61 For the
and alkali metal cations.51-53 Strong ion pairing is in- purposes of this review it is adequate to note that the
dicated in the case of sodium borohydride in poly- measurements are often made with the electrolyte
ethylene oxide), whereas ion pairing was not detectable sandwiched between a pair of electrochemically inert
with the BF4- ion. In keeping with these observations electrodes (Figure 2); a sinusoidal potential is applied
the borohydride salt complexes are much poorer ionic and the magnitude (A) and phase shift (0) of the cur-
conductors. The problem remains as to whether one rent (7) are measured. Thus the sinusoidal dependence
can effectively employ infrared spectroscopy to differ- of the current with time (t) is given by
entiate the degree of cation interaction with poorly 7 = A sin (cct + 0) (2)
coordinating anions such as BF4- and S03CF3-. As
discussed in section 4, it is highly likely that strong ion This measurement is repeated at a series of fre-
association occurs in the polymer electrolytes containing quencies which typically may range as low as 10-4 Hz
these ions, but spectroscopic evidence concerning the to as high as 10 MHz. From these data it is possible
nature of the interaction is lacking. to extract the conductivity and dielectric constant of
Vibrational spectroscopy also has provided valuable the bulk electrolyte sample. Information on the re-
information about polymer conformation, coordination sistance to interfacial charge transport also can be de-
of polar groups to cations, and secondary interactions termined. This analysis follows along lines originally
such as hydrogen bonding.16,54,55 There is considerable proposed by Cole and Cole60 and developed in detail by
scope remaining for further research in this area. In Macdonald.61,62 The raw data referred to above can be
Ion Transport in Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 113

A
TABLE IV. Typical Conductivities0
material conductivity
P(E0)12LiC104 5.6 X KT6
P(P0)9LiCF3S03 2.2 X KT6
(MEEP)4LiCF3S03 1.0 X 10"4
RbAg4I5 6 X 10’1
Ge4 5 X 10‘2
Cu4 5.6 X 105

“Results in S/cm. Values at 312 K. 4

Electronic conductivity.

Figure Schematic of an ac impedance experiment. B

2. =

electrode, A
=
polymer electrolyte, E = imposed potential, / =

measured current response.

3. Complex impedance spectrum (Cole-Cole plot) ob-

Figure 4. Arrhenius-type plots for log a vs T"1 for PEO complexes
Figure of Lil and LiSCN. The curved behavior for PEO-LiSCN fits the
tained on (MEEP)4LiI at 20 °C between Pt electrodes. Frequency VTF relation of eq 7. The double Arrhenius behavior from
range 100 Hz to 3 MHz. X is the imaginary impedance (Z" in PEO-Lil corresponds to the conduction of the partially crystalline
eq 4) and R is the real impedance (Z'in eq 4). Rb represents the and elastomeric phases. From J. M. Chabagno, Thesis, Grenoble,
bulk resistance of the electrolyte sample. 1980.

used to express the current vector (/*) in terms of

ac
real (JO and imaginary (J'O parts:
liquid electrolytes. They devised simple but elegant
methods for the determination of transference numbers,
I* = I' + jl" j =
VIl (3a) which are generally designated as t+ or for the frac-
tion of the current carried by the cation and anion,
and a similar expression applies to the ac potential: respectively.63 The measurement of transference num-
E* =E' + jE" (3b) bers, or transport numbers as they are also called, is
experimentally more difficult with the solid electrolytes
The ac impedance, Z* = E*/I*, also can be expressed
than their solution counterparts,5 but measurements of
as a real and imaginary part: fair quality have been made and the general picture for
Z* = Z' + jZ" (4) a variety of polymer electrolytes is that somewhat over

In a Cole-Cole plot, now commonly referred to as an

half of the current is carried by the anion and therefore
less than half by the cation. A major issue that is not
impedance spectrum, the real part of the impedance
(Z 0 is plotted against the imaginary part (Z'O for data yet well resolved is the nature of the mobile species.
collected at a series of frequencies (Figure 3). In fa- With monovalent ions in dilute aqueous solution the
vorable cases various features of the impedance spec- isolated cation and anion are the charge carriers. But
trum can be interpreted in terms of the response of the the much lower dielectric constant of the polymer host
bulk polymer whereas other features arise from the in the solvent-free polymer electrolytes should be con-
ducive to strong Coulombic interaction between ions.
electrode-electrolyte interface. The bulk resistance of In the salt concentration range generally studied in
the electrolyte (I?b) is one of the quantities that can be
derived from such a plot. The value for the resistance polymer electrolytes, the primary charge carriers may
of the sample (J?b) along with thickness of the sample well be ion triplets, quintets, and so on. A recent de-
and electrode area yields the resistivity of the sample tailed study of ion transport in fluid solutions of
or its inverse, the conductivity. Table IV presents some
short-chain polyethylene oxide) provides strong evi-
dence for the importance of ion clusters in the solid
typical values for the total ionic conductivity for a range
of electrolytes, as well as some more familiar electronic electrolytes.24
conductors, which are included to put the electrolytes b. Temperature and Pressure Dependence
into perspective.
The identity of the charge carriers and the fraction The temperature dependence of the conductivity of
of the current carried by each is a more subtle issue that polymer electrolytes indicates an activated process.
is still not well resolved in most studies of solvent-free Thus the conductivity increases with increasing tem-
polymer electrolytes. This issue was addressed by perature, and Arrhenius behavior often provides a good
physical chemists around the turn of the century for representation of the data (Figure 4). On closer in-
114 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

-—
Temp °C <r =
90 30
'LmmQi (5)
100 80 70 60 50 40
where a, nit nh and q, are respectively conductivity, the
-2 -
mobility of the i species, the concentration of carriers
of the i species, and the charge of the i species. The
polymer electrolytes contain no significant conjugation
within the polymer backbone, and the salts on which
they are based have negligible electronic conductivities.
Thus one might suspect, and indeed several experi-
mental measurements have shown,68 that electrons or
electron holes do not contribute to the summation in
eq 5. Both cations and anions do, however, contribute
and as discussed above, important questions concerning
the relative mobilities (transference numbers) of the
cationic and anionic carriers remain. Once again, this
represents a complication in the simple interpretation
2.8 30 33
2.7 2.9 3.1 3 2
of the temperature dependence of conductivity.
10 3/T K'1) —
(
Experimentally, one observes fairly straightforward
Figure 5. Temperature-conductivity plots for P- behavior of the temperature dependence of the con-
[(EOlgNILSOsCFa]. The darkened circles represent data obtained ductivity in homogeneous electrolytes. The straight or
by heating the partially crystalline sample from room temperature. curved lines observed when the conductivity is plotted
The rapid rise in conductivity corresponds to the crystalline to
elastomer transition, (a) Cooling curve in which a metastable
in Arrhenius coordinates (Figure 4) can be fit, respec-
elastomeric phase persists below 45 °C. (O) Heating curve for tively, to the Arrhenius and VTF69 forms
the metastable elastomer. From ref 44 and 91.
aT =
a0e-E^kT (6)
spection, changes in slope and curvature of the log a vs aT =
a0e-B!k(T~T^ (7)
1/T plots are often observed. An abrupt change in
slope can generally be traced to a phase change in the In the Arrhenius form, Ek is the usual activation energy,
polymer-salt complex.11 Of somewhat greater interest whereas in the VTF form, T0 is a parameter to be de-
is the curvature observed in the conductivity plot for termined, and B is a constant, whose dimensions are in
most amorphous polymer electrolytes when data are fact energy, but which is not simply interpreted as an
collected over a wide temperature range (Figure 4). activation term; k is Boltzmann’s constant. The early
Polymeric electrolytes are soft materials, and one investigations of the groups in Grenoble generally
might therefore suspect that they would be easily com- showed curved plots, corresponding to VTF type be-
pressible. Accordingly, several investigations66-67 have havior.10,11,25-27 Cheradame and co-workers discussed
been reported in which the ionic conductivity was these plots in terms of the chain segment mobility of
measured as a function of pressure. The conductivity the polymer host material.25-27,30 They used the rela-
indeed is observed to decrease with increasing pressure, tionship, very common in discussion of polymer dy-
as expected on the basis of the free volume model, namics, of polymer chain viscosity to glass transition
discussed below, or any other vacancy model. Activa- temperature that is summarized in the Williams-Lan-
tion volumes have been discussed, and the results have del-Ferry70 relationship:
been used to suggest the nature of chain relaxations that
aid ion motion. log lv(T)/n(Ts)] =
log aT =
(8)

c. Interpretation of Ionic Conductivity: where Ts is an arbitrary reference temperature, aT is

Phenomenological Models called the mechanical shift factor, and Cx and C2 are
As mentioned in section 3, partially crystalline poly- “universal” constants. Often the WLF equation is
mer electrolytes may have different degrees of crys- coupled with the empirical observation known as
tallinity and a variety of morphologies. The multiphase Walden’s rule63
behavior influences the transport by strongly reducing Dr\ =
const/r,- (9a)
dc conduction and by introducing hysteresis effects as
temperature is varied (Figure 5). It is clear from this
or equivalently with the Stokes-Einstein relationship
figure that the higher conductivity occurs in the D kT/Qirijri
=
(9b)
amorphous phase that is metastable at room tempera- where D is the diffusion coefficient and rl is the radius.
ture. The multiphase behavior presents complications
for fundamental studies and for applications. To gain If in addition the Nernst-Einstein relationship
fundamental insight into the mechanisms of ionic <r = DNq2/kT (9c)
motion in polymer electrolytes, we consider fully ho-
where N is the number of carriers and q is the charge,
mogeneous amorphous polymer-salt complexes. These
materials are in fact now of primary interest in the is employed, then one can rewrite the temperature
study of polymeric electrolytes. Particular examples dependence of the conductivity in the WLF form:70
include the phosphazenes, siloxanes, poly(propylene a(T) Ci(T T.) -

oxide), and cross-linked poly (ethylene oxide), to be °g (10a)

discussed in section 5. a(TJ C2+(T- Ts)
The conductivity of any material can be expressed To a good first approximation, and 10a hold quite
eq 7
in terms of the mobility n by the relationship well for a series of polymer electrolytes. On the basis
Ion Transport In Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 115

of this, Cheradame’s group has argued that WLF be-

havior is the rule in polymer electrolytes and, therefore,
that the fluidity (inverse viscosity) of the polymer chain
segments largely determines the conductivity.
The WLF form of eq 8 was, in fact, originally70 based
on the empirical VTF form69 of eq 7, but written for the
fluidity (jj-1) rather than for oT. The equivalence of eq
10 to the VTF form is easy to show and provides an
identification of the parameters in the form
C2
=
T8
-

T0 (10b)
Cx
=
B/k(Ts -

T0) (10c) -|o9lO(aT>m

Figure 6. Correlation between the conductivity and the shift

where the constants C1 and C2 refer to eq 8 and rep- factor ax (of eq 8) for PEO networks with low concentrations of
resent the mechanical shift factor. salt (less than one ion per oxygen). The direct proportionality,
The WLF and VTF equations themselves are em- with slope of unity, implies that a is inversely proportional to
pirical generalizations, rather than the result of any viscosity, as assumed in the Walden or Stokes-Einstein relations
theoretical approach. As is clear from Figure 4, they (of eq 9a,9b). From ref 32.
describe the thermal dependence of a quite well. In an
When the volume is expanded in terms of the volume
attempt to understand how the conductivity mechanism
at the glass transition temperature plus a linear term,
works, quasi-thermodynamic theories,71-76 originally
the free volume theory yields the form25,30’38
developed to deal with molten salts and neat polymers,
have in fact been applied with some success to consid-
eration of transport properties in polymer electrolytes. 0 =
(uw
These theories are based on considerations involving
the critical role of the glass transition temperature Tg
c°TKip(-t-(r°-~cy))
and of the so-called “equilibrium” glass transition tem- where the constants a and C{ are both inversely pro-
perature T0. Above Tg, the polymeric material becomes portional to the free volume thermal expansion factor.
macroscopically rubbery rather than glassy.71-76 Note that we can rewrite eq lib as
Mechanistically, this is understood by considering that
a local environment on any given polymer chain be-
D -
D°X (HO
comes liquid-like at the glass transition. Therefore, it
is only the thermal energy in excess of the glass tran-
exp( c~+ T )
sition temperature that provides actual mobility of the with C2 =
C2' CA. If one chooses CA
- =
0, the free
local polymer chain segments. In this sense, one is not volume argument thus yields the WLF relationship of
surprised that the Arrhenius behavior, involving inverse eq 8. Alternatively, one can choose any other value of
temperature, is replaced by the VTF involving the in- the arbitrary constant CA; for instance, often one selects
verse of T T0. The concept of equilibrium glass
-

CA o' 50 K, relating DtoT0 ^ Tg- 50 K, the so-called

transition temperature T0 is based on the kinetic feature “equilibrium” glass transition temperature.
of Tg: depending on the rate of cooling, one can observe Figure 6, based on the results of Cheradame and co-
different glass transition temperatures, and T0 is workers,25-33 shows that indeed there is a very close
idealized as the temperature at which all “free volume” relationship between conductivity and polymer relax-
vanishes or at which all polymer segment motion dis- ation behavior. The shift factor aT, whose logarithm
appears or at which the excess configurational entropy is plotted along the abscissa, is the ratio of the me-
of the material vanishes. The theoretical scheme which chanical relaxation time at temperature Tr to that at
treats T0 in terms of volume is called the free volume a reference Ts, which is generally taken to be either the
theory,71,72 and that which treats it in terms of entropies glass transition temperature Tg or the “equilibrium”
is called the excess entropy or configurational entropy glass transition temperature T0. It is easy to show that
model.73,74 if the WLF equation fits data well with a given refer-
The free volume model, although subject to various ence temperature Tg and parameters C: and C2, then
trenchant criticisms75-79 for the discussion of ion it will fit the data equally well if a different reference
transport (to be discussed below), is the simplest way temperature is chosen, provided that Cx and C2 are
to understand the polymer segment mobility. It states allowed to change. This close relationship between the
that as temperature increases, the expansivity of the relaxation property of the polymer host and the con-
material produces local empty space, free volume, into ductivities of the ionic solute very strongly suggests a
which ionic carriers, solvated molecules, or polymer relationship between the chain motion of the polymer
segments themselves can move. The overall mobility and the diffusive or conductive mode of the ions. Such
of the material, then, is determined by the amount of a relationship is also implied by the use of the Walden
volume present in the material. The free volume is relation of eq 9a since the t? refers to the viscosity largely
calculated according to the simplest model, by the usual determined by the polymer chain, whereas the D is the
statistical argument of maximizing the number of ways diffusion of the ionic solute. Several other facts also
in which the volume can be distributed.71 One then imply a close relationship between the segmental
obtains, for the diffusivity D, the form fluidity motions of the polymer and the ionic diffusivity
D = BRT exp(~V*/Vf) (11a)
motion of the salt. For example, when the conductivity
is fit to the WLF form of eq 10a, the constants Cj and
where B and V* are constant and V{ is the free volume. C2 very closely match the “universal” values found for
116 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

1001P A ....
-

-75

* '
03 0j6 09 12 15 18 *2.1
Cations per Polymer Repeat Unit

Figure 8. Dependence of conductivity (left ordinate) (at 70 °C)

and of glass transition temperature (right ordinate, in °C) on
concentration of salt in MEEP-(AgSQ3CF3)I complexes. (MEEP
=
-(N=P(OC2H4OC2H4OCH3)2)-.) Note two effects on the
Figure 7. Conductivity at constant reduced temperatures TL + conductivity: increased carrier number at low salt concentration,
T as a function of the lattice energies of the incorporated salts. and higher glass transition temperature (less flexible chains, slower
The dropoff in conductivity at high lattice energy is indicative segmental motion) at higher salt concentration. From ref 83.
of residual Coulombic attraction among the ions in the complex,
reducing the effective number of current carriers and, therefore, there is a fairly substantial variation of the conductivity
reducing the conductivity. The measurements (ref 82) were made even at constant T + Tg and that the conductivity de-
in PEO-based networks.
creases as the lattice energy of the parent salt increases,
polymer physical properties.19’25,27'28’30’43’81’82 Thus, for indicating that Coulombic attraction between anion and
example, if Ta is chosen as T„, characteristic values for cation acts to reduce the effective number of carriers.
Cj range from 8.3 to 15.2 while those for C2 range from The effects of concentration variation,25,80-84 quali-
30 to 93 °C. These numbers, obtained from fitting the tatively, generally follow the behavior seen in Figure 8:
conductivity to the WLF form, should be compared as the salt concentration increases starting from a dilute
with the “universal” values, C2 = 17.2, C2 = 51.6 °C, complex, the conductivity first increases and then, after
used to describe polymer physical properties (the shift attaining a maximum for a particular concentration,
factor aT). The substantial deviation indicates that the falls off quickly for more concentrated materials.
simple relationship of Figure 6 is correct overall but that Qualitatively, such behavior may be understood on the
differences can occur in any given species. These dif- basis of eq 5: at low concentration, the carrier numbers
ferences reflect, for example, ion interaction effects. increase as salt is added, so that n in eq 5 increases as
In fact, it is the diffusivity or mobility rather than does <r. (The expected dependence is sublinear due to
the conductivity of the ion which might be imagined to effects of ion association.) At higher concentrations,
relate to the polymer chain motions. According to eq the salt acts as a weak sort of cross-linker, raising the
5, the conductivity depends not only upon mobility but value of T0. One then expects a to drop with further
also upon the concentration of carriers. In extremely increases of ion concentration. Angell and Bressel
dilute solutions, it is generally fair to assume that all suggested85 (for liquid solutions) that the conductivity
carriers are in fact available, so that the number n of can be expressed, following eq 11, in the form
eq 5 is simply the stoichiometric number of ions. In <7 ~
X exp[~B'/(X0 -

X)] (12)
most polymer electrolytes, however, the concentrations
are in fact much higher. For example, in P- where X is the mole fraction of salt, 5'is a constant,
[(EO)4LiCNS], assuming a density of 1 g cm-3, the and X0 is the concentration at which the isotherm
concentration of salt is 4 M, and the mean distance temperature matches Tg. Once again, however, the
between cation and anion is roughly 7 A. Under these presence of interactions within the polymer electrolyte
conditions, the ions do not move freely, and one might makes itself clearly felt: the maximum conductivity of
therefore suspect that the concentration of carriers, n, PEO-based solid electrolytes is much smaller (almost
is dependent on the temperature, the stoichiometric a factor of 100) than that of comparable aqueous solu-
concentration, and the physical properties of the poly- tions. Angell86 attributes this to the effects of residual
mer host. Although nearly all the early discussions of ion-ion interaction in the low dielectric constant poly-
ionic conductivity in polymers were based on nonin- ether solvent.
teracting carrier ideas,1011’35 it is clear in retrospect that The observed pressure dependence of ionic conduc-
the polymer electrolytes usually studied are in fact tivity in polymer electrolytes is easily interpreted by
strongly concentrated electrolytes and that Coulombic free volume type concepts. Since the free volume V{
interactions among the carriers will be crucial for de- of eq 11a will decrease with applied pressure, one ex-
termining the conductivity.25 This can be clearly seen pects D to fall off exponentially as pressure is increased
if one studies the conductivity of a number of poly- and free volume decreased. Alternative explanations,38
mer-salt complexes with the same polymer host and however, can be given in terms of the configurational
different solutes. If simple free volume theory were entropy or dynamic percolation models. In the latter
applicable, then the conductivity should be constant for cases, the pressure dependence of the conductivity is
constant T + Tg. Figure 7 shows a plot of such data, considered to arise largely from reduction in the extent
obtained by Watanabe and co-workers.82 Note that of local segmental motion of the polymer chains. In a
Ion Transport In Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 117

free volume picture, the pressure dependence might WLF behavior very well, with identical values for the
arise from motion of ions into voids or of chain seg- Ci and C2 parameters. Once again, this shows the de-
ments into voids.25,87 Since, however, ionic conduction pendence of ion transport on polymer motion. Angell
in these electrolytes is not critically dependent on ion and Torrell have recently used the width of the Bril-
size for the small ions usually studied,83,88 free volume louin peak to obtain the characteristic relaxation times
theory is usually invoked in terms of the volume re- for both glassy and polymeric ionic conductors94 and
quired for motions of the polymer chain segments, to study the dependence on molecular weight of the
rather than of the ions themselves. polymer. They note that the mechanical relaxation
The free volume model is intuitive and helpful. It times appropriate for description of segment motion
has been subjected to thorough analysis in the context and for use in the decoupling ratio R of eq 13 are not
of ionic motion in fused salts, fluid solutions, and the global, macroscopic times, but rather local, micro-
polymer electrolytes. Several shortcomings of the model scopic times corresponding to small (roughly 1-10 A)
have been pointed out in each of these contexts.38,63,77 motions. The proper probes to obtain such relaxation
For example, free volume models give incorrect quan- times are local ones, such as NMR, light scattering, or
titative predictions of the pressure dependence of dielectric relaxation. Poinsignon and Berthier95 have
transport properties.75 Occasionally even the sign is recently completed quasi-elastic neutron scattering
incorrect.79 An improved model, based upon consid- studies of polymer electrolytes; they obtain from the
erations of entropy fluctuations rather than volume quasi-elastic half-width a time of roughly 10-xo s, which
fluctuations, was developed by Gibbs, DiMarzio, and they identify as the time scale for segment motion.
Adam.73,74 The model, called the configurational en- Relaxation processes in amorphous condensed-phase
tropy model, leads to transport properties in agreement systems such as fused salts of polymers often show a
with the empirical VTF form of eq 7 or WLF form of wide spectrum, or distribution, of characteristic relax-
eq 10a. Moreover, it correctly predicts pressure de- ation times. Different experimental measurements,
pendences and provides a satisfying description of T0 depending upon precisely how the experimental probe
as the temperature at which the excess configurational couples to the system under study, can observe different
entropy vanishes. Angell76-78 has pointed out quite relaxation processes. This is clearest with NMR, where
clearly the advantages of configurational entropy, as the resonance frequency selects which nucleus is being
opposed to free volume, interpretations of the VTF monitored, but it is true in general that different ex-
equation, but the appealing simplicity of free volume periments will measure different components of the
ideas has led to their widespread use in the litera- relaxation time distribution. Thus neutron scattering
ture.25,32,80,81,89 Like the free volume model, the con- probes chiefly motions of the protons, Brillouin scat-
figurational entropy theory is quasi-thermodynamic, tering senses relaxation of the phonon modes, etc. The
rather than microscopic. This is both a strength be- measurements of frequency-dependent conductivity
cause of the direct relation to thermodynamic observ- o-(a>), as obtained from microwave measurements, and
ables such as P, V, T, and S and a weakness since no of dielectric relaxation are sensitive to the motions of
microscopic mechanistic or equation-of-motion pictures charged species, including both mobile ions and dipoles
are available. on the polymer. Since the important polymer move-
ments for promoting ionic transport involve motions of
d. Frequency-Dependent Conductivity: Dynamic polymer segments that contain dipolar Lewis base
Percolation and Relaxation groups, the relaxation processes measured by electrical
probes are probably the most relevant ones for inter-
We have dealt thus far with the dc conductivity and preting the conductivity although, as we have already
its dependence on pressure, concentration, and tem- stated in connection with Figure 4, it appears that most
perature. The explanation of the motion mechanism measurements of relaxation processes yield similar
invoked arguments based on the rates of segmental temperature dependence.
motion of the chains of the polymer host. Measure- Polymer solid electrolytes conduct very poorly near
ments of the characteristic time scales for segmental or below their glass transition temperatures; above Tg,
motion are therefore of interest. A number of mea- the local polymer chain motion is in fact liquid-like and
surements, including those of storage modulus and rapid. Angell77,78 has generalized these concepts to
mechanical relaxation,25,27,30,31,43,90 NMR relaxation,33 differentiate two types of amorphous solid electrolytes.
dielectric relaxation and loss,91,92 microwave measure- He defines a decoupling ratio R as
ment of ac conductivity,91,93 Brillouin scattering,94 and
R ra/ra
=
(13)
inelastic neutron scattering,95 have been made to
characterize the dynamic (frequency dependent) re- where ra is a structural relaxation time and t„
sponse of polymer electrolytes. The mechanical loss or {<^±e0e^/erdc, where e„ is the optical dielectric constant)
mechanical relaxation measurements are generally in- is an electrical, or conductivity, relaxation time. For
terpreted in terms of WLF behavior, and, indeed the glassy electrolytes, which are used at temperatures
observed direct proportionality of mechanical relaxation below their glass transition temperature T„, R can be
and ionic conductivity, as illustrated in Figure 6, con- of the order 1013 (since ts can be97 of the order 200 s at
stitutes one of the most telling evidences for the de- Tg), whereas for polymer electrolytes, useful above Tg,
pendence of ionic conductivity on chain motion. Re- R can approach or even drop below unity. If R were
laxation times obtained from NMR studies also fit very close to unity, it would indicate that the ionic motion
well to the WLF relation as a function of temperature; and the structural relaxation occur on the same time
Cheradame’s group has found recently,17,96 for example, scale, thus suggesting that their rate-determining steps
that in polyelectrolytes with phosphorus cross-link sites are the same. Angell78,94 has noted that for polymeric
and mobile Li+ the 7Li, XH, and 3XP NMR signals all fit solid electrolytes R =* 10-3, which implies (since it is
118 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

substantially less than unity) strong residual ion-ion with wbeing some average rate. For polymer electro-
coupling, resulting in reduced conduction. For our lytes above Tg, the segmental motion changes the local
purposes here, however, the most striking aspect of the coordination environment of the ion with a character-
decoupling index is that it is much closer to unity than istic time rren, so that a jump which is unavailable
was true for the glasses, again indicating a very close (because the chains are improperly arranged) at time
relationship of structural relaxation process (due to t can become available, because the chains have re-
chain motion and reflected in rg) and conductivity (in- oriented, at time t + rren. The mathematical statement
versely proportional to t„). of this motion is that the assignment of probabilities
The substantial change in the value of R, from 1013 ~
Wij as zero or w changes on the time scale rren.
in glasses to near unity in soft polymer electrolytes, The dynamic percolation model is characterized by
must in part be due to the low frequencies and large the parameters /, w, and rren. These can be related to
amplitudes associated with polymer segment motion in the phenomenological models,38,87 so that these param-
the elastomeric phase above Tg. In an ionic conducting eters are defined in terms, say, of free volume. The
glass, as in a covalent crystal, only small charges in the dynamic percolation model has several interesting fea-
local geometry are, in general, associated with ionic tures, among the most important of which are the
motion into a vacancy. In these polymer materials, in following:5,38,87,89,98-101
contrast, very large changes in the local geometry are 1. For observation times long compared to the re-
brought about upon complexation of a cation. For the newal time, the motion is always diffusive. That is, the
cation to move, subsequently, the segments complexing mean-squared displacement is always proportional to
it must first exchange the primary coordinating atoms, time. This is observed for polymer electrolytes but is
and such motions require segmental mobility. not true of ordinary, static, nonrenewing percolation
Druger, Nitzan, and Ratner38,87,89’98-101 have developed models.
a dynamic percolation model for description of ion 2. The diffusion coefficient is, in general, propor-
transport in polymer electrolytes. This is a microscopic tional to (X), the average rate of renewal. This corre-
model that characterizes the ionic motion in terms of sponds well both to the wealth of experimental data
jumps between neighboring positions. For anions, indicating that the ionic motion is modulated by the
which are not strongly solvated by the polymer host, segmental motions of the polymer host and to the ex-
such a description is straightforward. For cations, the pectation, based on a value of the decoupling index R
local coordination environment evolves slowly, as a of eq 13 close to or less than unity, that structural re-
single M+—B (M+ = metal ion, B = Lewis base site on organization and conductivity arise from the same un-
polymer) linkage is changed at a time;36,37 the “jump” derlying motion mechanism.
of a cation then corresponds to a completed exchange 3. It is possible to show in great generality that
of one ligand. This process is sketched in structure 2. D0( w + iX) =
D(o>) (16)
That is, the diffusion coefficient at frequency a> in the
renewing (dynamic) percolation problem may be found
from the diffusion coefficient in the static (nonrenew-
ing) percolation problem, analytically continued to
frequency u + iX.
4. The factor /, giving the number of available jumps,
will be substantially different for cationic (strongly
solvated) and anionic (weakly solvated) motions. The
2
transference numbers might reflect this difference. The
study of polyelectrolytes, in which only one ionic species
The rate of “jumping” (that is, of ion motion) between moves, will be a good test of this feature of the theory.
any two “sites” (that is, between two different positions One interesting application of the dynamic percola-
in the material) is then represented in terms of simple tion model has been in understanding the problem of
first-order chemical kinetics, using the so-called master frequency-dependent conduction in the accessed mi-
equation crowave range 1 MHz < u < 37 GHz. Brodwin and
collaborators have measured these conductivities for a
dPi/dt =
ZiPjWji
j
-

PiWij) (14)
number of polymer electrolytes;35,91,93 qualitatively, one
expects that either the hop time rh ur1 or the renewal
=

with Pj the probability to be on site j and wu the rate time xren might lie within the microwave range. The
of ion motion from site j to site i.
general behavior observed for homogeneous polymer
The dynamic percolation model takes into account
electrolytes involves an initially flat response followed
the dependence of ionic motion rates on the fluidity,
by a rise to a peak value. As Figure 9 shows, the dy-
or rate of segmental motion, of the polymer host. A
namic percolation model fits these data quite well. At
characteristic rate of renewal, X rren-1, is defined,
=
low frequencies (long times), only the ions that expe-
which characterizes the rate at which a motion pathway rience the renewal process exhibit (r2) t, so that only
~

from one site to another becomes available for the ion the ions contribute to a. In the partly crystalline sam-
to move. In ordinary (static) percolation theory,102
ple, below 1 MHz a significant dropoff is observed in
~

applied, for example', to electron hopping in amorphous a, again describing limitations to long-range motion of
metals, the rates of eq 14 are taken to be ions due to poorly conducting crystalline regions; to
Wjj
= 0 probability 1 -

f obtain the fit shown in the figure, two renewal times

were involved, one for crystalline, the other for
Wji
= w probability f (15) amorphous regions. At higher frequencies (shorter
Ion Transport in Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 119

Commercial polypropylene oxide) is not stereoregular

and is therefore noncrystalline; however, the commercial
material includes a cross-linking agent, which represents
an unwanted impurity for fundamental studies. Nev-
ertheless, there is an extensive literature on the elec-
trical properties of PPO-salt complexes.11,80,105 A very
promising linear polyether that does not have the
problems of crystallinity consists of medium-chain
polyethylene glycols) linked by methylene groups. By
this device the OC2H4 repeat units are interrupted by
OCH2 units, thus preventing crystallization. The con-
ductivities of the LiS03CF3 complexes are high at room
Figure 9. Room temperature plots of observed microwave con- temperature.104 Many other linear polymers with polar
ductivity as a function of frequency for amorphous P- groups have been investigated as polymer hosts for
[(E0)8NH4S03CF3] ( ) and pure PEO (a). The lines are the fits
to experimental data of the dynamic percolation model. From electrolyte formation,5,6 such as sulfides,106 ketonic ox-
ref 91 and 44. ygens,64,107 and imines.41,103 Poly(ethylenimine) is a close
analogue of polyethylene oxide) and it shares with PEO
times, smaller distances), there is a contribution to the the complications of multiphase behavior.41,103 Closely
observed signal both from the response of the polymer related branched polymers such as branched poly-
and from the ions themselves. Indeed, at high enough (ethylenimine) also have been investigated. Branched
frequency (above ~100 MHz), the conductivity of the poly(ethylenimine) is an amorphous material and
complex becomes comparable to that of PEO itself, therefore generally superior to its crystalline linear
since excursions of dipolar groups on the polymer chain counterpart as a host for polymer electrolyte forma-
contribute to a. At even higher frequencies, into the tion.55
infrared, inertial effects should destroy the response,
so that at very high frequency, the conductivity a b. Complexes of Comb-like Polymers with Salts
should go to zero. The dynamic percolation model, like The realization that high conductivity in polymer
any hopping model, contains no inertial effects and thus electrolytes is dependent on local thermal motion of
predicts, incorrectly, that at very high frequencies the polymer segments led to the exploration of comb
conductivity cr(w) becomes flat (frequency independent polymers as hosts for polymer electrolyte formation.
but finite). The general concept is to utilize a flexible backbone and
The dynamic percolation model has also been applied attach to this short-chain polar oligomers capable of
to understand the frequency-dependent conductivity
in partially crystalline P[(EO)*NaSCN].100 There, a complexing alkali metal salts. These ideas led to the
distribution of renewal times was necessary to fit the synthesis and characterization of polyphosphazene,
observed conductivity; this is quite reasonable, since one
polysiloxane, and polyitaconate comb polymers con-
taining short-chain polyether side chains.83,108-112 Salt
expects far faster renewal, hence better conduction, in
the amorphous regions compared to crystalline ones. complexes of the phosphazene polymer have been fairly
extensively investigated and are found to have much
Applications of the dynamic percolation model to higher conductivity at room temperature than poly-
polymer electrolytes are discussed in detail else- ethylene oxide)-based electrolytes. The somewhat less
where.38,87,101 While the model does have the attractive
flexible polymethacrylate comb polymers also have been
feature of including the effects of segment motion on
ionic conduction, it does not directly include interionic investigated.113
interaction. Thus while it is the best microscopic model c. Network Polymers with Salts
currently available for understanding ionic conduction
in polymer electrolytes, it is inadequate in several ways From practical standpoint amorphous linear poly-
a

(no inertial dynamics, no interionic interaction). A great mers are inconvenient because they tend to flow at
deal remains to be done in understanding, on any rea- somewhat elevated temperatures. This nuisance can
sonable microscopic level, ionic motion in these mate- be remedied in fundamental studies by proper design
rials. of cells for electrical measurements, but it is a serious
drawback for potential commercial applications where
5. Specific Polymer Conductors long-term dimensional stability is required. Cheradame
and co-workers provided a solution to this problem by
a. Complexes of Linear and Branched Polymers the synthesis of network polymers consisting of cross-
with Salts linked poly(ether glycols).24-33 If the degree of cross-
linking is kept low or if flexible cross-links are em-
Polyethylene oxide) complexes of alkali metal salts ployed, segmental chain motion is not significantly im-
have been studied much more extensively than any paired and salt complexes of these network polymers
other solvent-free polymer electrolytes. As discussed have conductivities that are superior to those of the
in sections 3 and 4, ion transport occurs primarily in crystalline linear polymers.24,33 Cross-linked siloxane
the amorphous phase, so the partial crystallinity of most and phosphazene comb polymer electrolytes are found
of these PEO complexes is an unwelcome complication to have good mechanical properties with little sacrifice
when studying the conductivity response of the mate- in conductivity.1080,114 It also has been shown that
rial. As shown in Figure 5, the conductivity is low for amorphous materials with high conductivity can be
the partially crystalline compound and as the crystalline prepared by radiation cross-linking polyethylene ox-
phase melts the conductivity increases dramatically. ide)-salt complexes above their melting point.115
120 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

d. Polyelectrolytes at room temperature. Unlike the polymers discussed

elsewhere in this review, this is a rigid material with a
The comparable conductivities of anions and cations Tg that exceeds room temperature. Therefore the
in the solvent-free polymer electrolytes discussed so far mechanism of conduction is likely to be quite different
have led to the exploration of polyelectrolytes in which from that in the softer low-Tg materials discussed here.
charged groups are bound to the polymer. Most sol- Another interesting difference between this material
vent-free polyelectrolytes, such as sodium poly(styre- and others reviewed here is the relative insensitivity of
nesulfonate), are hard materials with high Tg and low the bulk conductivity to the presence of moisture.
conductivity. Potentially a hard plastic may be ren-
dered flexible (increased in free volume) by the addition f. Polymer-Salt Complexes Containing
of a plasticizer. Attempts to introduce standard plas- Dipositive Cations
ticizers into sodium polystyrene led to materials that
The overwhelming majority of studies of solvent-free
were unstable with respect to phase separation. How-
ever polyethylene glycol) plasticizers were found to give polymer-salt complex electrolytes have focused on 1:1
stable materials.15,16 Judging from infrared data the electrolytes (anion and cation singly charged); recently,
systems with doubly-charged cations have been stud-
stability of these systems arises from hydrogen bonding ied.119'122 Reports largely center on PEO materials,
between the glycol OH group and the sulfonate group.
One of the principal lines of interest in polymer elec- though transport in polyphosphazenes containing Sr2+
has been measured.83 In the PEO-based complexes,
trolytes is their wide potential window of redox stability
but the presence of OH groups is detrimental to this problems have been encountered with residual water
electrochemical stability. However, the glycol-con- content, history-dependent properties, and partial
crystallinity. Nevertheless, some of these complexes do
taining polyelectrolyte suggests that suitable materials show rather high conductivities; with well-defined
might be prepared by covalently linking short-chain
amorphous samples, estimated transport numbers in-
polyethylene oxide) groups to the polyelectrolyte. dicate that these are largely anion conductors.123 An
Recently comb-like polyelectrolytes have been reported
with anionic or cationic groups as well as short-chain interesting report by Moryousseff et al. indicates the
presence of a “mixed-cation effect”, in which a sample
polyethers attached to the backbone. These short of P[(EO)3oCaBr2CaI2] has higher conductivity than
chains provide self-plasticization, so that no added
P[(EO)15CaBr2] or P[(EO)15CaI2].124
plasticizer is needed.96,20,116
g. Mixed Ionic and Electronic Conductors
e. Proton-Conducting Polymers
Polar polymers of the type used in polymer electro-
Polymer electrolytes that transport protons have been lytes also have been shown to be useful for the prepa-
examined for several types of materials. Solvent- ration of mixed conductors in which ionic conductivity
swollen polyelectrolytes have long been recognized as and electronic conductivity coexist. The concept is to
excellent protonic conductors.13 In addition, poly(vinyl incorporate alkali metal salts of an electronically con-
alcohol)/phosphoric acid/water films12 have been ducting anion, such as a polyiodide, into the polymer
studied and show relatively high protonic conductivity. host. In a material of this type the ions will be mobile
Water-containing Nafion in the proton form at room and the polyiodide may serve as an electronic conduc-
temperature reaches a conductivity of 0.01 S cm'1. tor. Starting with an alkali metal iodide in the polar
These materials conduct better with higher water con- polymer, a pure electrolyte results, and as successive
tent, and, as suggested by Slade et al., the proton increments of I2 are added, the conductivity increases
motions are probably more similar to those in systems and electronic conductivity eventually predominates.125
of sorbed water than to ionic conduction in solvent-free The electronic conductivity transition in these materials
polymer electrolytes.13 is analogous to the metal-insulator transitions that can
Recent work by Armand and co-workers has focused be achieved in certain metal oxides as the oxygen partial
on rigorously anhydrous PE0/H3P04 materials, cast in pressure is changed.126
thin-film form from tetrahydrofuran/acetonitrile solu- Thermopower measurements on the poly(ethylene
tion.117 These differ from the solvent-containing ma- oxide)-sodium polyiodide system demonstrate that
terials and are far closer to the alkali polymer electro- holes are the conductors in these materials. Although
lytes. They show VTF rather than Arrhenius behavior further details on the mechanism of conduction must
(eq 7 rather than eq 6) for the temperature dependence be worked out, it is probable that an atom-transfer
of the conductivity and attain conductivity maxima mechanism occurs, similar to that discussed by Gileadi
near 10'1 S cm'1 near 100 °C. Armand’s group has also for polyhalides in fluid solution.127
investigated polyamide electrolytes, including nylon 6 In the range of iodine concentrations over which both
and acrylamide, complexed with H3P04 and cast into the ionic conductivity and electronic conductivity can
films from methanol. These materials are of real in- be measured, it is found that increasing electronic
terest for electrochromic or sensor applications, showing conductivity is accompanied by increasing ionic con-
o- ~
10'4 S cm'1 at ambient temperature and proton ductivity.128 It is not clear as yet whether this phe-
transport numbers t+ roughly equal to 0.9. These nomenon arises from changes in polymer segmental
phosphoric acid/polymer materials degrade if wetted, motion or the influence of mobile electrons (or electron
so that rigorously anhydrous conditions must be holes) on ion motion. The latter phenomenon was
maintained. predicted by Huggins and Huggins129 based on the
A decade ago, Hoel and Grunwald118 investigated earlier work by Wagner.130 They consider the case of
protonic conductivity in polybenzimidazole films. They only two carriers (one ionic, the other electronic) and
found proton conductivity on the order of 10'4 S cm'1 point out that in the case in which the ionic carriers are
Ion Transport in Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 121

more numerous but the electronic carriers are more Coulombic fluid. Important developments have taken
mobile, a substantial enhancement of the chemical place in the areas of polymer hosts, of mechanistic in-
diffusion constant is expected. terpretation of the conduction, of polymer blends, comb
Charge-transfer ions other than polyiodides can be polymers, and copolymers, of microionics preparation,
used to provide the electronic conductivity in these and of battery building and testing. Both the level of
mixed conductors. Interesting materials have recently understanding and the preparation of new materials
been studied by Wright,131 who prepared P[(EO)„LiA], have progressed substantially.
where LiA is an organic acid salt or charge-transfer salt Probably the most complex feature of these materials
(that is, the A" species can be TCNQ-, chloranil' or is the high concentration of charged species, which re-
phenoxy-). He has observed electronic conductivities sults in strong Coulombic correlations of charge carriers.
as high as 10-1 S cm-1. The conductivity is activated, In particular cases, this can lead to carrier trapping in
and when the material was drawn and oriented, the ion pairs or higher multiplets, thus substantially re-
conductivity increased appreciably, as the conducting ducing the conductivity. Even in other situations, the
states of aromatics became taore favorably aligned for residual Coulombic interactions are important. They
electronic overlap. lead to correlation factors in the Nernst-Einstein re-
lation (eq 9c), where the number N of carriers is not the
h. Inhomogeneous Systems stoichiometric concentration N0 of ions, to off-diagonal
Structural polymers are often based on the co- friction coefficients in a Langevin description of the
polymerization of monomers or on polymer blends, to pulsed field gradient NMR measurement of diffusion,139
achieve good mechanical properties. In these materials and to a decoupling index R of eq 13 becoming less than
the two polymers involved in the blend exist as a unity. Curiously, the very high concentration of ions
physical mixture of separate phases held together with may make the problem a bit easier than in the analo-
physical forces and not chemical bonds. Motivated gous case of lower concentration (~0.1-1.0 M) solu-
largely by a desire to improve the structural integrity tions, where the ions can range from tightly ion paired
of polymer electrolytes, investigations of polymer blends to roughly independent and where the theoretical de-
as electrolytes have been carried out. Often these scription is very complicated and seriously lacking.63
polymer blends consist of a polar polymer providing the When the concentrations become as high as they are
conductive paths, and a more rigid structural polymer, in these polymer electrolytes, most carriers are strongly
such as polystyrene132’133 or poly(vinyl acetate).134 An- coupled to counter charges, and the situation is very
other set of popular structural materials is based on similar to that of a solvated fused salt (always remem-
block copolymers. Again, good structural properties bering that the solvent, here, is a polymer, whose own
often result. A very interesting aspect of these materials internal motion time scale acts as a gate for the ionic
is the range of microstructures that can sometimes be motions). In analogous glassy electrolytes, the so-called
achieved by the phase separation of the components. weak-electrolyte model140-143 has been very useful. In
For example, sheetlike and rodlike structures of one this model, emphasis is placed on the carrier number
polymer in a matrix of the other can be achieved. Block term fy, rather than the mobility factor in eq 5; in
copolymers that undergo distinct phase separation have fact, the mobilities of all free ions, once they can act as
been investigated as polymer electrolytes, but as yet carriers, are essentially identical. The scheme is called
detailed studies of morphology and directional con- the weak-electrolyte model because mass action type
ductivity have not been reported. Block copolymers relations are used to obtain the carrier numbers.
with both structural integrity and good conductivity The relative importance of interionic interactions on
have been prepared.135 the structural and transport properties of the polymeric
Solid inorganic phases within the polymer electrolyte electrolyte is probably the most active current research
have been investigated. A moderate amount of an in- topic in these materials. Some reports indicate144 that
sulating phase, alumina, has been found to have little the Nernst-Einstein relation of eq 9c works very well
influence on the electrical properties of PEO electro- over a broad temperature range, thus implying that the
lytes.136 In battery development, extensive use has been correlation factors for diffusion and for conduction are
made of composite electrodes, in which a polymeric very similar. But, as already indicated, several exper-
electrolyte such as P[(E0)„LiCF3S03] is intimately iments, including pulsed field gradient NMR studies
mixed with a solid inorganic cathode material such as of diffusion, show strong ionic correlation effects. From
V6013 or TaS2.137,138 These materials are electronic a modeling viewpoint, it will be important to modify the
conductors via the inorganic phase and ionic conductors currently used pictures (free volume, configurational
in the polymer phase. Although a torturous path for entropy, dynamic percolation) to include correlation
charge migration must exist with resulting percolation effects.
limitations,137 these composites appear to be effective The outstanding experimental problems are con-
cathode materials.138 nected with the precision and specificity of structural
and dynamic information. From the structural stand-
6. Problems and Prospects
point we have evidence for coordination of the polymer
A great deal has been learned about solvent-free polar groups to the cation and some for ion pairing, but
polymeric electrolytes since the first reports by Wright we lack detailed knowledge of coordination numbers
15 years ago on the high conductivity of PEO com- and geometries. Similarly, the dynamical information
plexes. Our picture of conductivity mechanisms has often is based on the response of the bulk material,
evolved from independent cation hopping among fixed which may not be directly related to microscopic models
sites in the crystalline region to strongly correlated for conductivity. For example, dielectric properties can
diffusive motion by both ions in an amorphous solvated be measured and related to relaxation times but the
122 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

detailed connection with motions at an atomic level are on polymer electrolytes. It is a pleasure to thank the
lacking. The problems are similar to those that remain present and past members of The Northwestern Ionics
unsolved for electrolyte solutions. However, experience Groups, including Professors D. H. Whitmore, M.
with fluid electrolytes indicates that further progress Brodwin, A. Nitzan, B. Crist, and W. Halperin and their
can be made with more extensive use of multinuclear students, for close collaboration. We appreciate fruitful
NMR spectroscopy, which has already contributed collaborations with Prof. H. Allcock and P. Austin at
substantially to our knowledge of both structure and The Pennsylvania State University. Finally, we very
dynamics of polymer electrolytes. More detailed in- gratefully acknowledge the members of our research
formation on diffusion coefficients, polymer structure, groups who have contributed to our research in polymer
and relaxation times for the constituent nuclei in electrolytes: A. Barriola, P. Blonsky, S. Clancy, K.
polymer electrolytes can help to test structural and Doan, S. Druger, R. Dupon, C. Hardy, C. Harris, B.
dynamical models. Other techniques that have not Papke, R. Spindler, and J. Tonge.
been fully exploited for the study of polymer electrolyte
dynamics include ultrasound measurements, Brillouin References
scattering, and variable high-frequency conductivity.
(1) (a) Vashishta, P.; Mundy, J. N.; Shenoy, G. K., Eds. Fast-Ion
Recently, kinetic electrochemical measurements were Transport in Solids; North-Holland: Amsterdam, 1979. (b)
made on electroactive ions in polymer electrolytes.145 Bates, J. B.; Farrington, C. G., Eds. Fast-Ion Transport in
The kinetic electrochemical methods can be used to Solids; North-Holland: Amsterdam, 1981. (c) Kleitz, M.;
Sapoval, B.; Ravaine, D., Eds. Solid State Ionics; North-
determine diffusion coefficients for electroactive ions, Holland: Amsterdam, 1983. (d) Boyce, J.; DeJonghe, L.;
thus facilitating the study of how ion charge, size, and Huggins, R. H., Eds. Solid-State Ionics; North-Holland:
concentration influence ionic diffusion. Amsterdam, 1985.
(2) Armand, M. B. Annu. Rev. Mater. Sci. 1986, 16, 245.
There are many synthetic opportunities in the area (3) Gauthier, M.; Armand, M. B.; Muller, D. In Electrorespon-
of polymer electrolytes. The combination of the syn- sive Polymeric Systems; Marcel Dekker, in press.
thesis of new polymer electrolytes with electrical mea- (4) MacCallum, J. R.; Vincent, C. A. Polymer Electrolyte Re-
views; Elsevier: London, 1987.
surements may lead to new electrolytes that are selec- (5) Vincent, C. A. Prog. Solid State Chem., in press.
tive ion binders and transporters. Other synthetic (6) Tonge, J. S.; Shriver, D. F. In Polymers for Electronic Ap-
plications; Lai, J. H., Ed.; CRC Press, in press.
challenges include the preparation of polyelectrolyte (7) Wright, P. V. Br. Polym. J. 1975, 7, 319.
systems that have high ionic conductivity in the absence (8) Wright, P. V. J. Polym. Sci., Polym. Phys. Ed. 1976,14, 955.
of solvents or plasticizers. Materials of this type would (9) Fenton, D. E.; Parker, J. M.; Wright, P. V. Polymer 1973,14,
589.
have a transference number of 1 for the mobile ion, with (10) Armand, M. B.; Chabagno, J. M,; Duclot, M. Extended Ab-
a resulting simplification of the interpretation of the stracts, Second International Conference on Solid Electro-
lytes, St. Andrews, Scotland, 1978.
transport process. Mixed ionic-electronic conductors (11) Armand, M. B.; Chabagno, J. M.; Duclot, M- J- In Fast Ion
based on salts in polar polymers have only recently been Transport in Solids; Duclot, M. J., Vashishta, P., Mundy, J.
reported and considerable scope remains for the syn- N., Shenoy, G. K., Eds.; North-Holland: Amsterdam, 1979.
(12) Petty-Weeks, S.; Polak, A. J. Sensors and Activators, in
thesis of these new materials and studies of their charge press.
transport mechanisms. (13) Slade, R. C. T.; Hardwick, A.; Dickens, P. G. Solid State
Ionics 1983, 9/10, 1053.
Although our review has concentrated on the phe- (14) Jernigan, J. T.; Chidsey, C. E. D.; Murray, R. W. J. Am.
nomenology and theory of polymer electrolytes, a few Chem. Soc. 1985, 107, 2824.
words on applications are in order because substantial (15) Hardy, L. C.; Shriver, D. F. Macromolecules 1984, 17, 175.
(16) Hardy, L. C.; Shriver, D. F. J. Am. Chem. Soc. 1985, 107,
impetus for the study of solvent-free electrolytes has 3823.
come from the prospects for utilizing them in advanced (17) Cheradame, H.; LeNest, J. F. Abstracts. First International
electrochemical devices, such as high energy density Symposium on Polymer Electrolytes, St. Andrews, Scotland,
1987. LeNest, J. F. Doctoral Thesis, L’Universite Scientifi-
batteries. To our knowledge, there are no batteries on que et Medicale, L’lnstitut National Polytechnique, Greno-
the market that incorporate polymer electrolytes, al- ble, France, 1985.
(18) Bannister, D. J.; Davies, G. R.; Ward, I. M.; McIntyre, J. E.
though development projects are under way in several Polymer 1984, 26, 1291.
different countries.146-162 Experience demonstrates a (19) LeNest, J. F.; Cheradame, H.; Dalard, F.; Deroo, D. J. Appl.
low success rate and long development times for new Electrochem. 1986, 16, 75.
(20) Dominguez, L.; Meyer, W. H. Abstracts. First International
high energy density batteries but the potential rewards Symposium on Polymer Electrolytes, St. Andrews, Scotland,
are huge. 1987.
The properties of polymer electrolytes such as their (21) Kobayashi, N.; Uchiyama, M.; Tsuchida, E. Solid State Ion-
ics 1985, 17, 307.
high compliance, good adherence to electrodes, and the (22) Inabe, T.; Lomax, J. F.; Lyding, J. W.; Kannewurf, C. R.;
possibility of fabricating the polymers into thin films Marks, T. J. Synth. Met. 1984, 9, 103.
are attractive not only for batteries but for many other (23) Gray, F. M.; MacCallum, J. R.; Vincent, C. Solid State Ionics
1985, 18/19, 282.
electrochemical devices. A variety of such applications (24) A particularly elegant and extensive series of investigations
have been proposed, including electrochromic dis- has been reported by H. Cheradame and co-workers. '19,25~33
They employ networks based on short-chain PEO and/or
plays,153 electrochromic windows, solid-state photoe- PPO, cross-linked with various agents, especially isocyanates.
lectrochemical cells,154-156 and sensors. In addition, a The materials are prepared by dissolving the salt in the
recent paper from Wrighton’s laboratory describes the polyether sometimes with added solvent, cross-linking the
material with isocyanate in a reaction catalyzed by Sn4+, and
utilization of a polymer electrolyte in an electrochemical drying the resulting membrane.
transistor.157 Finally, polymer electrolytes are conven- (25) Cheradame, H. In IUPAC Macromolecules; Benoit, H.,
ient media for electrochemical experiments under un- Rempp, P., Eds.; Pergamon Press: New York, 1982; p 251.
(26) Killis, A.; LeNest, J. F.; Cheradame, H. Makromol. Chem.,
usual conditions such as high-vacuum photoelectro- Rapid Commun. 1980, 1, 595.
chemistry.168 (27) Killis, A.; LeNest, J. F.; Gandini, A.; Cheradame, H. J. Po-
lym. Sci., Polym. Phys. Ed. 1981, 19, 1073.
(28) Cheradame, H.; LeNest, J. F. Abstracts. First International
Acknowledgments. We are grateful to the ONR, Symposium on Polymer Electrolytes, St. Andrews, Scotland,
DOE, and NSF-MRL program for support of our work 1987.
Ion Transport in Solvent-Free Polymers Chemical Reviews, 1988, Vol. 88, No. 1 123

(29) Levesque, M.; LeNest, J. F.; Cheradame, H. Makromol. (75) Goldstein, M. J. Phys. Chem. 1973, 77, 667.
Chem.., Rapid Commun. 1983, 4, 497. (76) Angell, C. A.; Sichina, W. Ann. N. Y. Acad. Sci. 1976,279, 53.
(30) Killis, A.; LeNest, J. F.; Gandini, A.; Cheradame, H.; Cohen- (77) Angell, C. A. Solid State Ionics 1983, 9/10, 3.
Addad, J. P. Solid State Ionics 1984,14, 231. (78) Angell, C. A. Solid State Ionics 1986,18/19, 72.
(31) Killis, A.; LeNest, J. F.; Gandini, A.; Cheradame, H. Makro- (79) Angell, C. A.; Williams, E. J. Polym. Sci., Polym. Lett. 1973,
mol. Chem. 1982,183, 1037. 11, 383.
(32) Killis, A.; LeNest, J. F.; Cheradame, H.; Gandini, A. Makro- (80) Watanabe, M.; Ogata, N. In Polymer Electrolyte Reviews;
mol. Chem. 1982, 183, 2835. Elsevier: London, 1987.
(33) Killis, A.; LeNest, J. F.; Cheradame, H.; Cohen-Added, J. P. (81) Watanabe, M.; Nagano, S.; Sanui, K.; Ogata, N. Solid State
Ionics 1986, 18/19, 338.
Polym. Bull. 1982, 6, 351.
(82) Watanabe, M.; Itoh, M.; Sanui, K; Ogata, N. Macromolecules
(34) Vincent, C. A.; MacCallum, J. R, In Polymer Electrolyte 1987 20 569
Reviews; MacCallum, J. R., Vincent, C. A., Eds.; Elsevier:
London, 1987. (83) Blonsky, P. M.; Shriver, D. F.; Austin, P.; Allcock, H. R. J.
Am. Chem. Soc. 1984,106, 6854.
(35) Shriver, D. F.; Papke, B. L.; Ratner, M. A.; Dupon, R.; Wong,
(84) Kobayashi, N.; Uchiyama, M.; Shigehara, K.; Tsuchida, E. J.
T.; Brodwin, M. Solid State Ionics 1981, 5, 83.
Phys. Chem. 1985, 89, 987.
(36) Papke, B. L.; Ratner, M. A.; Shriver, D. F. J. Phys. Chem. (85) Angell, C. A.; Bressell, R. D. J. Phys. Chem. 1972, 76, 3244.
Solids 1981, 42, 493.
(86) Angell, C. A. Abstracts. First International Symposium on
(37) Papke, B. L.; Ratner, M. A.; Shriver, D. F. J. Electrochem. Polymer Electrolytes, St. Andrews, Scotland, 1987; Br. Po-
Soc. 1982, 129, 1694.
lym. J., in press.
(38) Ratner, M. A. In Polymer Electrolyte Reviews; MacCallum, (87) Druger, S. D.; Nitzan, A.; Ratner, M. A. Solid State Ionics
J. R., Vincent, C. A., Eds.; Elsevier: London, 1987. 1983, 9/10, 1115.
(39) Wright, P. V.; Lee, C. C. Polymer 1982, 23, 681. (88) Mali, M.; Roos, J.; Brinckmann, D.; Gorecki, W.; Andreani,
(40) Neat, R.; Glasse, M.; Linford, R.; Hooper, A. Solid State R. ; Berthier, C.; Armand, M. Proc. 23 Colloq. Ampere 1986;
Ionics 1986,18/19,1088. Solid State Ionics, submitted.
(41) Chiang, C. K.; Davis, G. T.; Harding, C. A.; Takahashi, T. (89) Druger, S. D.; Nitzan, A.; Ratner, M. A. J. Chem. Phys. 1983,
Macromolecules 1985, 18, 825. 79, 3133.
(42) Robitaille, C. P.; Fauteux, D. J. Electrochem. Soc. 1986,133, (90) Watanabe, M.; Sanui, K.; Ogata, N. Macromolecules 1986,19,
315. 815.
(43) Lee, Y. L.; Crist, B. J. Appl. Phys. 1986, 60, 2683. (91) Ansari, S. M.; Brodwin, M.; Druger, S.; Stainer, M.; Ratner,
(44) Stainer, M.; Hardy, L. C.; Whitmore, D. H.; Shriver, D. F. J. M. A.; Shriver, D. F. Solid State Ionics 1986, 17, 101.
Electrochem. Soc. 1984, 131, 784. (92) Fontanella, J. J.; Wintersgill, M. C.; Calame, J. P.; Andeen,
C. G. Solid State Ionics 1983, 8, 333.
(45) Berthier, C.; Gorecki, W.; Minier, M.; Armand, M. B.; Cha-
bagno, J. M.; Rigaud, P. Solid State Ionics 1983, 11, 91. (93) Wong, T.; Brodwin, M.; Papke, B. L.; Shriver, D. F. Solid
State Ionics 1981, 5/6, 689.
(46) Hibma, T. Solid State Ionics 1983, 9/10, 1101.
(94) Torrell, L. M.; Angell, C. A., unpublished observations.
(47) Okamura, S.; Chatani, Y. Polymer, in press.
Borjesson, L.; Martin, S. W.; Torell, L. M.; Angell, C. A. Solid
(48) Parker, J. M.; Wright, P. V.; Lee, C. C. Polymer 1981, 22, State Ionics 1986, 18/19, 431.
1305.
(95) Poinsignon, C.; Berthier, C. Abstracts. First International
(49) Catlow, C. R. A.; Chadwick, A. V.; Greaves, G. N.; Moroney, Symposium on Polymer Electrolytes, St. Andrews, Scotland,
L. M.; Worboys, M. R. Solid State Ionics 1983, 9/10, 1107. 1987.
(50) Blonsky, P. M., unpublished observations, Northwestern (96) Cheradame, H.; LeNest, J. F. Abstracts. First International
University, 1983. Symposium on Polymer Electrolytes, St. Andrews, Scotland,
(51) Dupon, R.; Papke, B. L.; Ratner, M. A.; Whitmore, D. H.; 1987.
Shriver, D. F. J. Am. Chem. Soc. 1982, 104, 6247. (97) deBolt, M. A.; Easteal, A. J.; Macedo, P. B.; Moynihan, C. T.
(52) Teeters, D.; Freeh, R. Solid State Ionics 1986, 18/19, 271. J. Am. Ceram. Soc. 1976, 59, 16.
(53) Papke, B.; Ratner, M. A.; Shriver, D. F. J. Electrochem. Soc. (98) Harris, C. S.; Nitzan, A.; Ratner, M. A.; Shriver, D. F. Solid
1982, 129, 1434. State Ionics 1986, 18/19, 151.
(54) Spindler, R.; Shriver, D. F. Macromolecules 1986, 19, 347. (99) Druger, S. D.; Ratner, M. A.; Nitzan, A. Phys. Rev. 1985, 331,
(55) Harris, C. H.; Ratner, M. A.; Shriver, D. F. Macromolecules 3939.
1987, 20, 1778. (100) Druger, S. D.; Ratner, M. A.; Nitzan, A. Solid State Ionics
(56) Barriola, A., unpublished observations, Northwestern Univ- 1986, 18/19, 106.
ersity, 1986. (101) Druger, S. D.; Nitzan, A.; Ratner, M. A., to be published.
(102) (a) Landauer, R. J. Appl. Phys. 1952, 23, 779. (b) Kirkpa-
(57) (a) Minier, M.; Berthier, C.; Gorecki, W. J. Phys. (Les Ulis,
Fr.) 1984, 45, 739. (b) Minier, M.; Berthier, C.; Gorecki, W. trick, S. Rev. Mod. Phys. 1973, 45, 574.
Solid State Ionics 1983, 9/10, 1125. (103) Harris, C. S.; Shriver, D. F.; Ratner, M. A. Macromolecules
1986 19 987
(58) Gorecki, W.; Andreani, R.; Berthier, C.; Armand, M. B.; Mali,
M. ; Roos, J.; Brinkmann, D. Solid State Ionics 1986,18/19, (104) Nicholas, C. V.; Wilson, D. J.; Booth, C.; Giles, J. R. M. Ab-
stracts. First International Symposium on Polymer Elec-
295.
trolytes, St. Andrews, Scotland, 1987, p 273.
(59) Bhattacharja, S.; Smoot, S. W.; Whitmore, D. H. Solid State (105) Armand, M. B. Solid State Ionics 1983, 9/10, 745.
Ionics 1986, 18/19, 306.
(106) Clancy, S.; Shriver, D. F.; Ochrymowycz, L. A. Macromole-
(60) Cole, K. S.; Cole, R. H. J. Chem. Phys. 1941, 9, 341. cules 1986, 19, 606.
(61) Macdonald, J. R. J. Chem. Phys. 1973, 58, 4982. (107) Dupon, R.; Papke, B. L.; Ratner, M. A.; Shriver, D. F. J.
(62) Macdonald, J. R. In Superionic Conductors; Mahan, G. D., Electrochem. Soc. 1984, 131, 586.
Roth, W. L., Eds.; Plenum Press: New York, 1976; p 81. (108) (a) Blonsky, P. M.; Shriver, D. F.; Austin, P.; Allcock, H. R.
(63) Smedley, S. The Interpretation of Ionic Conductivity in Solid State Ionics 1986, 18/19, 258. (b) Allcock, H. R.;
Liquids; Plenum: New York, 1980. Austin, P. S.; Neenan, T. X.; Sisko, J. T.; Blonsky, P. M.;
(64) MacCallum, J. R.; Tomlin, A. S.; Vincent, C. A. Eur. Polym. Shriver, D. F. Macromolecules 1986,19,1508. (c) Tonge, J.
J. 1986, 22, 787. S. ; Shriver, D. F. J. Electrochem. Soc. 1987, 134, 269.
(65) Fontanella, J. J.; Wintersgill, M. C.; Smith, M. K.; Samancik, (109) (a) Hall, P. G.; Davies, G. R.; McIntyre, J. E.; Ward, I. M.;
J.; Andeen, C. G. J. Appl. Phys. 1986, 60, 2665. Bannister, D. J.; LeBrocq, K. M. F. Polym. Commun. 1986,
(66) Fontanella, J. J.; Wintersgill, M. C.; Calame, J. P.; Pursel, J. 27,98. (b) Bell, S. E.; Bannister, D. J. J. Polym. Sci., Polym.
P.; Figueroa, D. R.; Andeen, C. G. Solid State Ionics 1983, Lett. 1986, 24, 165.
9/10, 1139. (110) Fish, D.; Kahn, D. M.; Smid, J. Polym. Prepr. 1986, 27, 325.
(67) Chadwick, A. V.; Strange, J. H.; Worboys, M. R. Solid State (111) (a) Cowie, J. M. G.; Ferguson, R. J. Polym. Sci., Polym. Phys.
Ionics 1983, 9/10, 1155. Ed. 1985, 23, 2181. (b) Cowie, J. M. G.; Martin, A. C. S.
(68) Dupon, R.; Whitmore, D. H.; Shriver, D. F. J. Electrochem. Polym. Commun. 1985, 23, 2181.
Soc. 1981, 128, 716. (112) Xia, D. W.; Soltz, D.; Smid, J. Solid State Ionics 1984, 14,
221.
(69) Vogel, H. Phys. Z. 1921, 22, 645. Thammann, G.; Hesse, W.
Z. Anorg. Allg. Chem. 1926,156, 245. Fulcher, G. S. J. Am. (113) (a) Xia, D. W.; Smid, J. J. Polym. Sci., Polym. Lett. 1984,14,
221. (b) Xia, D. W.; Smid, J. Solid State Ionics 1984,14, 221.
Ceram. Soc. 1925, 8, 339.
(114) Spindler, R.: Shriver, D. F. Macromolecules, in press.
(70) Williams, M. L.; Landel, R. F.; Ferry, J. D. J. Am. Chem. Soc.
(115) MacCallum, J. R.; Smith, M. J.; Vincent, C. A. Solid State
1955, 77, 3701. Ionics 1984, 11, 307.
(71) Cohen, M. H.; Turnbull, D. J. Chem. Phys. 1959, 31, 1164. (116) Kobayashi, N.; Uchiyama, M.; Tsuchida, E. Solid State Ion-
(72) Cohen, M. H.; Grest, G. S. Phys. Rev. B 1980, B21, 4113. ics 1985, 17, 307.
(73) Gibbs, J. H.; DiMarzio, E. A. J. Chem. Phys. 1958, 28, 373. (117) Armand, M. B. Abstracts. First International Symposium
(74) Adam, G.; Gibbs, J. H. J. Chem. Phys. 1965, 43, 139. on Polymer Electrolytes, St. Andrews, Scotland, 1987.
124 Chemical Reviews, 1988, Vol. 88, No. 1 Ratner and Shriver

(118) Hoel, D.; Grunwald, E. J. Phys. Chem. 1977, 81, 2135. (138) Steele, B. C. H. In Transport-Structure Relations in Fast
(119) Wetton, R. E.; James, D. B.; Whiting, W. J. Polym. Sci., Ion and Mixed Conductors (Proceedings of the RIS0 Inter-
Polym. Phys. Ed. 1976, 14, 577. national Symposium on Metallurgy and Materials Science);
(120) Yang, L. L.; Huq, R.; Farrington, G. C. Solid State Ionics Paulsen, F. W., Andersen, N. H., Clausen, K., Skaarup, S.,
1986, 18/19, 291. Sorensen, D. T., Eds.; Roskilde: 1985.
(121) Patrick, A.; Glass, M.; Latham, R.; Linford, R. Solid State (139) Whitmore, D. H., unpublished observations, Northwestern
Ionics 1986, 18/19, 1063. University, 1987.
(122) Abrantes, T. M. A.; Alcacer, L. T.; Sequeira, C. A. C. Solid (140) Ingram, M. D.; Moynihan, C. T.; Lesikar, A. V. J. Non-Cryst.
State Ionics 1986,18/19, 315. Solids 1980, 38/39, 371.
(123) Huq, R.; McGhie, A. R.; Farrington, G. C. Abstracts. First (141) Ravaine, D.; Souquet, J. L. Phys. Chem. Glasses 1977,18, 27.
International Symposium on Polymer Electrolytes, St. An- (142) Martin, S. W. Solid State Ionics 1986,18/19, 472.
drews, Scotland, 1987. (143) Kone, A.; Souquet, J. L. Solid State Ionics 1986,18/19, 454.
(124) Moryousseff, A.; Bonnat, M.; Fouletier, M.; Hicter, P. Pro- (144) Ward, I. M.; McIntyre, J. E.; Davies, G. R. Abstracts. First
ceedings of the 6th RISp Symposium on Metallurgy and International Symposium on Polymer Electrolytes, St. An-
Materials Science; Poulsen, F. W., Andersen, N. H.; Clausen, drews, Scotland, 1987.
K., Skaarup, S., Sorensen, D. T., Eds.; Roskilde: 1985. (145) Geng, L.; Reed, R. A.; Longmire, M.; Murray, R. W. J. Phys.
(125) Hardy, L. C.; Shriver, D. F. J. Am. Chem. Soc. 1985, 107, Chem. 1987, 91, 2908.
3823. (146) Owens, B. B. EPRI Rep. 1986, 320.
(126) Mott, N. F. The Metal-Insulator Transition; Taylor and (147) Bonino, F.; Ottaviani, M.; Scrosati, B.; Pistoia, G. J. Elec-
Francis: London, 1972. trochem. Soc., in press.
(127) Kirowa-Eisner, E.; Brestovsky, A.; Reshef, I.; Brand, M.; (148) Hooper, A.; North, J. M. Solid State Ionics 1983, 9, 1161.
Gileadi, E. J. Electrochem. Soc. 1986, 133, 555. (149) Gauthier, M. Abstracts. First International Symposium on
(128) Tonge, J. S., unpublished observations, Northwestern Univ- Polymer Electrolytes, St. Andrews, Scotland, 1987.
ersity, 1987. (150) Gauthier, M., et al. J. Electrochem. Soc. 1985, 132, 1333.
(129) Huggins, R. A.; Huggins, J. M. Macromolecules 1977,10, 889. (151) Gauthier, M.; Fateux, D.; Vassort, G.f Belanger, A.; Duval,
(130) Wagner, C. J. Chem. Phys. 1953, 21, 1819. M.; Ricoux, P.; Chabagno, J. M.; Muller, D.; Rigaud, P.; Ar-
(131) (a) Conheeny, K.; Wright, P. V. Polym. Commun. 1986, 27, mand, M. B.; Deroo, D. J. Electrochem. Soc. 1985,132,1333.
364. (b) Siddiqui, J. A.; Wright, P. V. Polym. Commun. 1987, (152) Armand, M. B. In NATO ASI Series E101 Applied Science,
28, 5. (c) Siddiqui, J. A.; Wright, P. V. Polym. Commun. Solid State Batteries, Sequeira, C. A. C., Hooper, A., Eds.,
1987, 28, 90. 1985, p 363.
(132) Gray, F. M.; MacCallum, J. R.; Vincent, C. A. Solid State (153) Honda, K.; Kuwaro, A. J. Electrochem. Soc. 1986,153, 853.
Ionics 1986,18/19, 256. (154) Skotheim, T. A. Lundstrom, I. J. Electrochem. Soc. 1982,
(133) Xia, D. W.; Smid, J. Solid State Ionics 1984,14, 221. 129 894
(134) Armstrong, A. D.; Clarke, M. D. Electrochim. Acta 1984, 29, (155) (a) Skotheim, T. A. Appl. Phys. Lett. 1981, 38, 712. (b)
1443. Inganas, O.; Skotheim, T. A. Solid State Ionics 1986,18/19,
(135) Watanabe, M.; Nagaoka, K.; Kanba, M.; Shinohara, I. Polym. 332.
J. 1982, 14, 877. (156) Sammells, A. F.; Ang, P. G. J. Electrochem. Soc. 1984,131,
(136) Weston, J. E.; Steele, B. C. H. Solid State Ionics 1982, 7, 75. 617.
(137) Harris, K. D. M.; Rogers, M. D.; Vincent, C. A. Solid State (157) Chao, S.; Wrighton, M. S. J. Am. Chem. Soc. 1987,109, 2197.
Ionics 1986, 18/19, 833. (158) Skotheim, T. A. Synth. Met. 1986, 14, 31.