Faraday Discuss. Chem. SOC., 1989, 88, 65-76 Charge Transfer at Polymer Electrolytes Michel Armand Laboratoire d' Ionique et d' Electrochimie du Solide CNRS UA 1213, ENSEE-Grenoble B. P. 75 Domaine Universitaire, 38402 Saint-Martin-d' Hkres, France Polymer electrolytes ( PEs) are ion-conducting complexes formed between a metal salt and a solvating polymer, usually a polyether like poly(ethy1ene oxide). Electrochemical reactions at PEs/electrode interfaces are reviewed and compared with those of liquid electrolytes. One characteristic of these materials is their exceptional inertness which spans the 0-4.6 V/Li+ : Li" stability window. Nevertheless, the conduction mechanism, a solvation- desolvation of the cation along the chain, governs the kinetics: only cations with fast ligand exchange are electroactive, irrespective of thermodynamics; anions, which are not solvated are always mobile.Transfer of ions into intercalation compounds is usually well defined, with no solvent co-intercala- tion. The possibility is discussed that some level of redox conduction appears in PEs at potentials either cathodic (metal solubility) or anodic beyond the stability window (oxonium radical). Polymer electrolytes (PEs) cover a wide variety of materials in which ions exist and are mobile in the absence of any solvent or low molecular weight pla~ticizer.'-~ In this respect, they are termed 'immobile solvents' as the ionic motion does not correspond to a net displacement (centre of gravity) of the macromolecule. This is especially evident with cross-linked materials in which the chains are bound together to form a single three-dimensional network.The most frequently studied compounds are those obtained from the dissolution of an alkali-metal salt in a poly(ethy1ene oxide) [CH2-CH2-O-], (PE0)- or poly(propy- lene oxide) [CH,-CH(CH,)-0-1, (PP0)-based polymer. poly( ethylene oxide)( PEO) \ 0 / CH2-c~C H 2 - b C q A p o I poly( propylene oxide) (PPO) \ 1 0 The corresponding units can be arranged with a linear chain, comb or network structure, keeping the basic solvation properties of the pol yether. For highly symmetrical monomer units, like PEO, both the pristine polymer and the corresponding stoichiometric com- plexes (usually 3 : 1 and 6 : 1 repeat units per salt) tend to form regular non-conductive structures (crystallites). A conductive eutectic (elastomeric) is formed at 40-60 "C between these two phases.Modifications of the linear chain structure favour the existence of the conductive amorphous phase; in addition, cross-links improve the mechanical properties. Typical conductivities for PEs are within 10-6-10-3 0 cm-' for temperatures ranging from 25 to 120 "C. The best electrolytes reach conductivities of LR cm-' at room temperature. 6566 H C l o t , CF3SOT +LI+ Be .> + + "Ne' 'Mg+ Al - + t + + - K+ + C i +SC r T 1 + V +I9Cr+*M;' Fe+ goQ:Ni +Cu ' 2 : Ga Go - + + ++ + 3 + 2 + + 2 + + + 2 + + + + + 2 + + + + I- + ' + + + + 4 + 5 + 5 + 2 - + + + + + 2 + . +-). + + + Rb Sr Y Zr Nb Mo Tc Ru R h P d Ag'+Cd+ In +Sn+ Sb Cs Be + L a Hf ;a W Re 0 s Ir Pt A u TI P b +Bi P o * + + + + + + + ' + + + + + + 5 + 3 + 4 +% 1 + + + 3 Fr Ra Ac Charge Transfer at Polymer Electrolytes - + +4 + + Y + + t + + + Ce Pr Nd Pm Sm Eu Gd + + + + + + + + t 2/3t+ ++ Th Pa LO,' + + 2 + , + +4 ++ +I+ + + +I+ +r+ + Tb +Dy+ Ho+ Er Tm Yb Lu + + + + + + + + + + In addition to the alkali metals, by far the most studied, a wide variety of other derivatives including alkaline, rare-earth and transition metals are now known, though they are not yet fully characterized, neither by phase diagram nor by electrochemistry (fig.1). Polyethylene imine with the repeat unit [CH2-CH2-NH-] poly(ethy1ene imine)( PEI) \ / C H 2 - C \ H 2 n n NH NH NH NH NH forms similar complexes, which are especially stable in the case of the 'soft' transition- metal cations (cu*+, Zn2+, co2+ etc.).Charge Transfer at Electrodes Charge Transfer within the Electrolytic Domain A distinguishing feature of the polymer electrolytes is, among other technological advantages, their expected lack of reactivity. This is especially the case for PEO, which shows the well known chemical inertness of ethers. As a result, electron transfer is limited in most cases to the solute and the redox reaction limited at the electrode interface. We shall discuss here the reduction and oxidation processes accessible with polymer electrolytes. Metal Deposition and Alloy Formation At cathodic potentials, a variety of metal cations have been shown to be reduced to the metal. M''+(e/ectro/yte) 4- ne-(e/ecrrode) ---* (M"). The alkali metals have received most attention, especially lithium deposition onto a metallic substrate for battery applications.6-8 Research in this field has been discussedM.Armand r 67 -0.8 voltage/V Fig. 2. Cyclic voltammetry for P( E0)9LiCF3S03 electrolyte at 80 "C. 20 mV s-' on non-alloying nickel substrate. Cycle number indicated on figure. From ref. (8). t I I I 1 1 1 1 I I 1 1 - 4 -3 -2 -1 0 1 2 3 4 voltage/V Fig. 3. Cyclic voltammetry trace for P(EO),,,NaCF3S03 ; platinum electrode, 85 "C; sweep rate 60 mV s-'; reference Ag"/Ag,SI. recently by Bilanger;' the metal can be reversibly plated and stripped at low overpotential when using a salt derived from a thermodynamically (I-) or kinetically (C104-, CF,SO,-) stable anion as shown in fig. 2. The well known difficulties associated with an electrode of the first kind (Li+/ Li") appear much less dramatic with PEs.Dendrite formation, an especially preoccupying problem with liquid electrolytes, can be elminated for > 1000 deep cycles in batteries with surface capacities of ca. 2 C cm'. Sodium deposition as shown in fig. 3 is also observed as a reversible process and even a potassium reduction wave can be obtained despite the higher reactivity of the metal. Nevertheless, the electrochemical activity is dependent upon the transport number of the species, i.e. how68 Charge Transfer at Polymer Electrolytes voltage/ V Fig. 4. Cyclic voltammogram for P(EO),-(0.9)LiC10,-(0.1)Mg(C104)2 ; platinum electrode, 85 "C; 40 mV s-'; reference Ag"/Ag,SI. fast the electroactive species are supplied to the electrode.As pointed out by Vincent, the transference number of the cations can be very small, yet they can diffuse as neutral ion pairs, MX, or triplets, XMX-. Such species are expected to exist in appreciable concentrations in these iow dielectric constant media ( E = 5-10). Neutral species do not appear to be mobile in PEs, as shown by the coincidence of the conductivity values directly measured using various electrochemical techniques and of those deduced from NMR pulsed field gradient measurements through the Nernst-Einstein equation."?' This implies that ionization is the required step for motion and diffusion in PEs, and thus these materials differ markedly from liquids. Strong polymer-cation interactions tend to limit the mobility of the positive charges ( t + = 0), despite an appreciable ionization.This is the case for Mg2+; Mg(CIO,), in PEO has no apparent redox activity, as shown in fig. 4. Conversely, the larger Ba2+, which is less strongly solvated, is able to move towards the electrode and can be reversibly plated and stripped at a potential of ca. 500mV, positive of the Li couple, as shown in fig. 5. An interesting correlation can be made between the motion of cations in PEs and the measured rates at which the same ions exchange water molecules within their solvation shell, drawn in fig. 6 and covering 18 orders of magnitude. Only cations with exchange rates exceeding lo8 sC' are expected to have an appreci- able mobility in polyethers whose solvation characteristics are quite similar to those of water. Tieline at Zn2+ ( t f = 0.05) corroborates the observation of electroactivity of the heavier alkaline- and rare-earth metals with the addition of the 's' metals and the larger transition-metal cations, or those sensitive to the Jahn-Teller effect with high ligand liability (see fig.7 for an example on lead deposition and stripping): Sn2+, Pb2+, Hg", Mn2+, Cu2' etc. Alloy formation is also possible with many host metals and lithium; in fact, this element forms alloys with almost all metals except those with high lattice energy (Fe, Ni, Mo, W, Cr) and the diffusion is fast enough to be observed around room temperature.M. Armand 69 - L - 3 -2 -1 1 2 3 4 voltage/ V Fig. 5. Cyclic voltammogram for P(EO),- (0.9)LiCI04-(0. l)Ba(ClO,), ; platinum electrode, 85 "C; 40 mV s-'.Arrow shows the Ba" deposition peak; reference Ago/Ag,SI. I : t i P6 Co' Fc" V" TI'' Mn'j Lanthanides 1 1 I I ' dl - 8 -6 - 4 - 2 0 2 4 b 8 10 Fig. 6. Rate constants (C1) at 25°C for exchange of water ligand in aqueous solutions. From ref. (23). In fig. 8, the cyclic voltammetry scan of a mixture of LiC10, and LiI (10%) shows in the cathodic domain a reduction wave superposing the formation of a Li/R alloy ( 1 : 2) and Li" deposition, while the two processes of dissolution and de-alloying are well separated on the following anodic sweep. Lithium dissolution in aluminium has been well studied in various electrolytic media, again for battery applications. With the PEO electrolyte, the coulometric titration curve as shown in fig. 9 clearly indicates the transformation of the cy aluminium structure ( < 7 at.% Li) into the p phase which possesses a wide range of non-stoichiometry, 0.85 < y < 1.15, in Li,,Al, in agreement with the phase diagram.Similarly, zinc and magnesium form alloys with lithium which are either stoichiometric compounds (Zn) or true solid solution (Mg). The existence of intermetallic compounds with sodium and potassium is limited to 'softer', more electronegative elements like Pb or Bi.70 Charge Transfer at Polymer Electrolytes + 2 +1 i -1 -2 +3 +4 voltage/V Fig. 7. Cyclic voltammetry trace for P(E0)30Pb(CF3S03),; 90 “C platinum electrode; 10 mV s-’. From ref. (12). Proton-conducting PEs, exemplified by the PEO--H3PO3l3 complexes or the L ~ r m ~ l y t e ~ ’ 1 4 \ NH2 NH2 NH2 NH2 which are intermediates between glasses and polymers both show hydrogen evolution at cathodic potentials, a reversible process on platinum electrodes.Intercalation The most general solid-state redox reaction is intercalation, where an ion (+ne or -ne) diffuses into a solid host structure (H) together with a compensating electronic chargeM. Armand 71 4 - 2 -1 0 - 2 - 4 - -4 -3 -2 -1 0 1 2 3 4 voltage/V Fig. 8. Cyclic voltammogram for P( EO),-(0.9)LiC104-(0. 1)LiI platinum electrode, 85 "C; 40 mV s-'. A", Li dissolution peak from LiPt, alloy; reference Ag"/Ag,SI. 4 0 0 300 > iu < 200 0 - , , I . . . . , . . . . 0 1 0.5 1.5 y(Al-Li, ) Fig. 9. EMF us. composition (coulometric titration curve) for the lithium dissolution in aluminium; reference Li" 65 "C electrolyte P( E0)8LiC104.(e- or h+), keeping the basic structure-building bonds mainly intact (topochemical reaction). Further refinements of this simple description are in terms of a higher order structure (stages, sublattice ordering, site exchange) which allows minimization of the strains (mechanical, electrostatic, Jahn-Teller) due to charge injection. The insertion reaction, written in its electrochemical form as xM+(e/ecrro/yre) + Xe-(e/ecrrode) + (H) ( M ~ H ) o<x<x,,,72 Charge Transfer at Polymer Electrolytes > 0 02 01, 0 6 08 1 X Fig. 10. Ultra-slow scan voltammetry trace for lithium insertion in ZrSe,,,,, ; electrolyte P(EO),LiClO,; 1.2 mA h electrode capacity; reference Li"; 100 "C; 10 mV h-'; from ref. (16). ( a ) CV curve; ( b ) integral EMF=f(x).consists of a simple charge transfer across the electrode interface. Again, such a process has been studied in detail with various host (H) structures like TiS2, W03 or V205 for battery applications, the most immediate driving force in the field of PEs. The reversibil- ity, stability and absence of co-intercalation of the polymer are among the main advan- tages. This latter phenomenon is often seen with liquid electrolytes and two-dimensional structures which are not sterically selective, and occurs when the cations entering the structure are too strongly solvated to shed their coordination shell. As a result, the structure exfoliates and loses its reversibility. For instance, it is not possible to prepare electrochemically the graphite-alkali-metal intercalate with any liquid electrolytes, while they are readily obtained from PEO-MX electr01ytes.l~ Recently, Chabre et have taken advantage of the absence of side-reactions in PE electrochemical cells to set up digital ultra-slow scan voltammetry ( V s-'). This technique allows direct differen- tiation between the sites according to their energies in the intercalation compound.A good example is given with the cyclic voltammetry trace for sub-stoichoimetric zirconium selenide [fig. 10( a ) ] interpreted as the preferential occupancy of the sites created in the vicinity of the Zr-Zr pairs within the slabs, followed by the van der Waals sites. Integration of the cyclic voltammetry trace gives directly the thermodynamic coulometric titration curve [fig. 10( b ) ] .Another aspect of intercalation chemistry, misnamed as 'doping', is observed with conjugated polymers (CPs). Either reduction or oxidation reactions are possible, the latter corresponds to an ingress of anions to compensate for delocalized carbocationsM. Armand 73 40 2 0 - 2 0 I - 4 0 V( Li/ Li') Fig. 11. Cyclic voltammetry trace for solid-state cell Li/ P( E0)8LiC104/polyviologen; sweep rate = 0.5 mV s-', temperature = 90 "C. formed on the polymer backbone: The superposition of both conductivities, ionic in PEs and electronic in CPs, is very appealing. Simple mechanical contact between the two polymers gives poor overall kinetics due to the sluggish diffusion of X- in CPs, yet the elegance of an all polymeric system: (CH),/PEO-NaI/(CH), is certain. l 7 Combination at the molecular level, as for instance oligoPEO-grafted polypyrrole, shows much improved kinetics of charge transfer in the bulk of the composite.'* A non-conjugated redox polymer, like the polyviologen.has been studied in an all solid-state cell, for various MX and n:19 - Li/ P( EO) MX/ polyviologen+. The reversible electroactivity of the viologen is clearly demonstrated on the cyclic voltammogram of fig. 11, showing the formation of the radical cation and of the neutral species: v2+ + e-(elecrrode) a V'+ e V". The pol viologen electrode was thus used as an anion-specific electrode at the equivalent point Vr+2X-/V*+X- to determine the activity of MX in PEO.I974 Redox Processes and Conduction in the Electrolyte Metal Solubility An interesting possibility is that alkali metals are soluble in the polymer, in a similar way to in ether solvents (THF, glymes) where the metals form ion pairs in concentrations up to ca.10-3moldm-3: Charge Transfer at Polymer Electrolytes The formation of solvated electrons, as in ammonia MTOruoted eLolUated is less likely due to the lower dielectric constant of ethers. The high solvation ability of polyethers should favour the metal solubility, and PEO has been shown to increase markedly the solubility of K in THF mol dm-3).20 Such ion pairs in polymers could lead to 'n'-type electronic conductivity, similar in principle to that of 'electrides' formed from cryptates and alkali metals. 7nnn-r :o: :o: :o+ ro: :o: In any case, the solubility is expected to become negligible in the presence of salts, a situation normally encountered when operating PEs ('salting-out' effect).The possibility remains that, under prolonged cathodic polarization with Mo deposition, the concentra- tion gradient due to the anion transference number results in local salt depletion, allowing for metal solubility. Such a phenomenon may have been observed with Lio as a coloured layer moving away from the cathode surface and may explain the 'soft dendrite' healing mode of solid-state lithium batteries: a cell short-circuited by a dendritic growth recovers after a few cycles with little effect on life span. Anodic reactions are also limited to the solute within the 0-4.6 V/Li: Li+ stability window of the polyethers. A good example is given again in fig. 8 where the oxidation of the minority carrier I- into 1,- appears as a reversible reaction.In this case considering the low concentration on the polyiodide species, the reaction is controlled by diffusion of the ionic species, the anions always being mobile. The polysulphides M2S, behave similarly and both systems have been used in photoelectrochemical A different situation is encountered with higher iodine concentrations, as shown in fig. 12 for CsI, which displays an almost linear current-voltage relationship, indicative of predominant electronic conductivity. Similarly, Wright and Siddiqui have found high redox conductivity in the system PEO TCNQO-NaTCNQO-. Again, the polymers differ from simple solutions since the attainable concentrations are higher than those of liquids, corresponding to shorter average distances betwen species, a situation favourable to electron hopping.There are no clear examples yet of a redox conduction mechanism obtained with the same metal at two different oxidation states. Good candidates are the Fe""", Cu"", , UOZ1'" systems which are known or expected to give complexes either in PEO or PEI. However, relatively low levels of redox conduction are expected as the elements often require a different coordination environment when changing valency; this would lead to polaron-type conductivity. Eu I I / 111M. Armand 75 - 4 -3 -2 -1 1 2 3 4 voltage/V Fig. 12. Cyclic voltammetry trace for the complex P( EO)&sI, ; 65 "C, platinum electrode; reference Ago/ Ag, S I. Beyond the anodic stability limit, in the presence of oxidation-resistant anions, the transient formation of oxonium radical cations is possible: 7Af7AAf- :o: :o: :o+ :o: :o: X- = C104-, CF3S03- and could lead to 'p'-type conductivity before chain degradation takes place.The latter process is in fact markedly accelerated by impurities, like CI-. For instance DDQ (dichlorodicyanoquinone) gives DDQ'- in the presence of purified PEO with a minimal loss in molecular weight; addition of chloride results in almost immediate chain scission. Liquid electrolytes like THF or dioxolane also form a cation radical which initiates an irreversible polymerization with the formation of a passivating layer, a problem which was long underestimated with battery electrolytes. Conclusions A decade after their discovery, the electrochemistry of polymer electrolytes has begun to be well documented. It is now clear that the behaviour of these materials cannot be transposed directly from that of liquids: the mechanism for conductivity is unique and more selective, restricting the number of potentially electroactive species from the larger choice based on the thermodynamics alone.On the other hand, this is a definite advantage in terms of stability and absence of side-reaction at the electrode interface. We should also retain the possibility of working under high vacuum, as for in situ SEM76 Charge Transfer at Polymer Electrolytes observation of electrode reactions.23 PEs appear, in fact, to be the best compromise between liquids and true solid electrolytes (AgI, p alumina, glasses), which are even more selective but very limited in choice, and for which the realization of complete electrochemical systems is a challenge, with arduous material processing and interfacial problems.The almost infinite choices offered by macromolecular chemistry present many interesting possibilities for a variety of applications, including alloys of conjugated polymers and polymer electrolytes, the attachment of redox centres to the backbone to mediate or catalyse redox processes, or the confinement of reagent or species to a predesigned spatial location, based on specific ion-polymer interactions ( e.g. Cu2+ in PEI); shuttle redox couples may also be used to reveal the electroactivity of species immobilized in the macromolecule. References 1 M. B. Armand, Ann.Rev. Muter. Sci., 1986, 16, 245. 2 Polymer Electrolytes Reviews 1, ed. J. R. MacCallum and C. A. Vincent (Elsevier, London, 1987). 3 C. A. Vincent, Prog. Solid State Chem., 1987, 17-3. 4 M. Gauthier, M. Armand and D. Muller, in Electroresponsive Molecular and Polymeric Systems, ed. T. Skotheim (Marcel Dekker, New York, 1988), vol. 1, p. 41. 5 M. Armand and M. Gauthier, in Solids with High Conductivity, ed. T. Takahashi (World Publishers, Singapore, 1989), in press. 6 M. Gauthier, A. BClanger, B. Kapfer, G. Vassort and M. Armand, in Polymer Electrolytes Review IZ ed. J. R. MacCallum and C. A. Vincent (Elsevier, London, 1989), in press. 7 C. A. C. Sequeira and A. Hooper, Solid State Ionics, 1983, 9/10, 1131. 8 F. Bonino, B. Scrosati and A. Selvaggi, Solid State Ionics, 1986, 18/19, 1050. 9 A. BClanger, Proc. 2nd Int. Meeting Polymer Electrolytes, Sienna, Italy, June 14-17, 1989. 10 W. Gorecki, R. AndrCani, C. Berthier, M. B. Armand, M. Mali, J. Roos and D. Brinkman, Solid State Zonics, 1986, 18/19, 295. 11 S. Bhattacharja, S. W. Smoot and D. H. Whitmore, Solid Sfate Ionics, 1986, 18/19, 306. 12 A. Bouridah, F. Dalard, D. Deroo and M. Armand, Solid State Zonics, 1986, 18/19, 287. 13 F. Defendini, M. Armand, W. Gorecki and C. Berthier, Electrochemical Society Sun Diego Meeting, 1986, 86-2 extended abstract No. 3. 14 Y. Charbouillot, D. Ravaine, M. Armand and C. Poinsignon, J. Non Cryst. Solids, 1988, 103, 325. 15 R. Yazami, P. Touzain, J. Power Sources, 1983, 9, 365. 16 Y. Chabre, P. Deniard and R. Yazami, Solid State Ionics, 1988, 28/30, 1153. 17 C. K. Chiang, Polymer, 1981, 22, 1651. 18 M. G. Minett and J. R. Owen, Solid Sfate Zonics, 1988, 28/30, 1193. 19 A. Bouridah, F. Dalard and M. Armand, Solid Stare Ionics, 1988, 28/30, 950. 20 I . M. Panayotov, C. B. Tsvettanov, I. V. Berlinova and R. S. Velchkova, Makromol. Chem., 1970, 134, 21 T. A. Skotheim, S. W. Feldberg and M. B. Armand, J. Phys. (Paris), 1983, C344, 615. 22 B. Marsan, Ph.D. Thesis, UniversitC du QuCbec INRS Energie, 1988. 23 P. Baudry, M. Armand, M. Gauthier and J. Massounave, Solid State Zonics, 1988, 28/30, 1567. 24 D. W. Margerum et al., in Coordination Chemistry, ed. A. E. Martell (ACS Monograph No. 174, 313. Washington, D. C., 1978) vol. 2. Paper 9/02136A; Received 19th May, 1989