J. MATER. CHEM., 1991, 1(2), 157-162 FEATURE ARTICLE Solid Electrolytes and Mixed Ionic-Electronic Conductors: An Applications Overview Anthony R. West Department of Chemistry, University of Aberdeen, Meston Walk, Aberdeen AB9 2UE, UK A review of the various applications of solid electrolytes and mixed conductors is presented. Discussion focuses on their uses in: sensors; high-density batteries; solid-oxide fuel cells; ion-exchange materials; optical waveguides; solid-state batteries; thin-film electrochromic devices; and high-T, ceramic superconductors. The potential for future development is also discussed. Keywords: Feature article; Solid electrolyte; Ionic-electronic conductor Solid electrolytes are an unusual group of materials which have high ionic conductivity with negligible electronic conduc- tivity.Examples are now known involving high conductivity of most monovalent and some divalent ions, e.g. Ag+ in RbAg415, Na' in /?-alumina, Li' in H-doped Li,N, 02-in 9Zr0, * 1Y 203(yttria-stabilised zirconia), F-in PbSnF4 and H+ in HU02P04*4H20 (hydrogen uranyl phosphate, HUP) (Fig. 1). There is another group of materials, mixed ionic- electronic conductors, that have high conductivities of both ions and electrons. Solid electrolytes and mixed conductors have been surveyed recently in ref. 1 and more comprehen- sively in ref. 2. The purpose of this review is to summarise the current status of the applications of these materials (Table 1) and to Table I Applications of solid electrolytes and mixed ionic/electronic conductors batteries, primary or secondary sensors, gas pumps fuel cells, especially solid-oxide fuel cells electrochemical reactors supercapacitors synthesis of new materials by ion exchange waveguide fabrication by ion exchange optimisation of superconductivity by oxygen (de)intercalation lithium (de)intercalation, new materials, solid solution electrodes electrochromics, smart windows and displays indicate some likely future developments.Specifically excluded from consideration here are polymeric electrolytes and poly- meric mixed conductors; these have been reviewed recently in ref. 3 and 4. Batteries The great upsurge of research into solid electrolytes (also called superionic conductors or fast-ion conductors) in the 1960s, which commenced with the discovery of the ion- conducting properties of fl-alumina' and various Ag salts,6 was motivated to a great extent by the possibility of building new high-density secondary battery systems.The concept of the Na/S cell, using molten electrodes separated by a solid p-alumina electrolyte, originated from the Ford Motor Com- pany and has since been developed vigorously in several major laboratories worldwide. The main Na/S cell designs that have been tested contain a p-alumina tube, closed at one end, with one molten electrode inside and one outside (Fig. 2). The cell operating temperature is 300-350 "C, in order to maintain both electrodes and the sodium polysulphide dis- charge products in the liquid state (Fig.3). The cell voltage Ak03 insulator Na Al can j3-aluminaelectrolyte t \ sulphur + C felt 1 103 KIT Fig. 2 Schematic drawing of a sodium-core, Na/S cell: Fig. 1 Some conductivity Arrhenius plots 2Na +XS eNa,S, I I I I I I I I i cell400 -operatingregion -i= I Ill I 1 1+Na Na2S2 Na2S4Na2S5 S composition Fig. 3 Phase diagram for the Na-S system and cell open-circuit voltage at 350 "C is 2.08 V during initial discharge, while the discharge products are in the region of liquid immiscibility (Fig. 3), and then falls gradually to 1.78 V, at which point Na2S2 starts to precipitate. For practical reasons, the onset of such precipitation is taken as the fully discharged state.Na/S cells have a theoretical power density of 760 W h kg-'; in practice, power densities of 150-200 W h kg-' are routinely achieved in individual cells. The principal research and development groups that have been assembling and testing multicell batteries, including Brown Boveri (West Germany), Chloride Silent Power and Beta R & D (UK), Ford (USA) and Yuasa/NGK (Japan), report performance levels that approach the 1000 cycles of reliable operation that are required for electric vehicle applications. For instance, the most recent data available on the 360 cell battery from Brown Boveri reveal densities of 85 W h kg-' and 120 W kg-'.7 This compares favourably with the power density of lead acid batteries, typically 20-40 W h kg-'.There are many problems that can lead to a decrease in battery performance, e.g. short circuiting through the electro- lyte walls, resistance rise associated with the precipitation of impurities (especially Ca leached from the ceramic electrolyte), corrosion of the container and consequent loss of sulphur. Individually these problems appear not to be insurmountable; however, there is still the necessity to improve cell reliability. The early fears that the p-alumina electrolyte may be thermo- dynamically unstable when in contact with molten Na (owing to the leaching of oxygen from the conduction planes, followed by partial collapse of the crystal structure and loss of ionic conductivity) appear to be groundless. Given the twin factors of diminishing fossil-fuel supplies and increasing environmental awareness surrounding atmos- pheric pollution, it seems reasonable to expect some degree of commercialisation of Na/S batteries within the next decade, either for electric-vehicle or power-station load-levelling appli- cations.An interesting alternative to the Na/S cell that has recently been announced*-" is the Zebra cell, Na/MCl,: M =Fe, Ni. It uses much the same design and technology as the Na/S cell but the cathode compartment has, instead, a mixture of liquid NaAlC1, and solid Fe (or Ni) Cl,. The cell discharge J. MATER. CHEM., 1991, VOL. 1 reaction is 2Na +(Ni, Fe)Cl, +2NaC1+ (Ni, Fe) The purpose of the NaAlCl, is to act as an ionically con- ducting liquid contact between the solid electrolyte and the solid electrode, i.e.the NaC1-FeC1,-Fe, mixture. The cell may be readily assembled in the discharge state, from a mixture of NaCl and Fe/Ni, impregnated with NaAlCl,. The cell has a higher voltage, 2.35 V for Fe and 2.57 V for Ni, and lower operating temperature, 250 "C,than the Na/S cell; prototype batteries giving several hundred cycles of operation, have been tested. It remains to be seen whether the perform- ance (densities of 88 W h kg-' and 65 W kg-' have been quoted for a 66 cell Na/NiCl, battery) can be optimised to exceed that of Na/S batteries. A variety of alternative battery systems using solid electro- lytes have been proposed over the years but none has received the same amount of attention as the 'a batteries' described above. The only example that has achieved commercialisation is the Li/12 miniature heart pacemaker primary battery.It contains as the solid electrolyte a thin film of LiI electrolyte which forms in situ on assembly when the two electrodes are placed in contact. This and other potential batteries, many based on polymeric materials, have been reviewed.' ' Gas Sensors and Pumps Oxygen sensors have been important in various applications for determining oxygen contents of gases and liquids. Most are fabricated from a tube of an oxide ion conductor such as yttria-stabilised zirconia (YSZ), bismuth oxide (in oxidising environments) or thoria (in reducing environments), Fig. 4. The tube is coated with inner and outer electrodes of porous Pt and the potential difference that develops between the electrodes may be related to the difference in oxygen partial pressure in the two compartments.One of the compartments contains a reference gas, e.g. air, 0, or a metal-metal oxide mixture such as Ni-NiO. Oxygen sensors are used commercially for monitoring gas compositions in combustion-plant and metallurgical processes and for determining the amount of oxygen dissolved in molten metals.12 They are also used in car exhaust systems (2 probe) to help optimise the fuel :air ratio. The same cell principle that is illustrated in Fig. 4 is also used in oxygen pumps.13 The two electrodes are short-circuited and oxygen gas may then be pumped from one electrode compartment to the other.Commercial devices are available and are used to purify oxygen or to provide con- trolled oxygen atmospheres in studies of, for example, the corrosion of metals and the cultivation of micro-organisms. Yl environment /c to be measured-1 Fig. 4 Design of an oxygen sensor based on yttria-stabilised zirconia solid electrolyte. I/= (RT/nF)In [p"(O,)/p'(O,)] J. MATER. CHEM., 1991, VOL. 1 Solid Oxide Fuel Cells (SOFC) Dramatic advances have been made in the development of solid oxide fuel cells in the past few years, as signalled by the first international SOFC conference in 1989.14 Much of this impetus has come from Westinghouse, USA but recently, major EEC and Japanese programmes have also commenced.The basic principle of the SOFC is shown in Fig. 5." It uses the oxide ion conducting ceramic, yttria-stabilised zir- conia, to act as separator and solid electrolyte between the fuel (CO, H,, CH, etc.) and air/oxygen. The air electrode is based on the mixed conductor perovskite, LaMnO,, and the fuel electrode is an electronically conducting Ni/ZrO, cermet. Cell operating temperature is 1000 "C, at which the electrolyte, the most resistive component in the cell, has a conductivity of 0.1 $2-'cm -Typical cell voltage is 0.7 V. The choice of cell components is such that they should be compatible with each other over long periods of time at high temperatures and it may be that the compositions of the components are not yet optimised. The original Westinghouse design, which has operated at the 3 kW level for 2500 h, is a tubular design, but there is now increasing interest in a flat- plate configuration (Fig.6). This is made feasible by advances in ceramic fabrication, using tape-casting methods to fabricate the planar components, although there are still doubts about the long-term mechanical stability of such structures. The individual three-layer cell 'sandwiches' are separated by a bipolar plate of electronically conducting LaCrO, which has a corrugated structure on either side to permit the easy flow of air and fuel over the respective electrode surfaces. There are several intrinsic features of SOFCs that make them attractive as power sources. (1) At the temperature of cell operation, 1000°C, methane (natural gas) may be used directly as the fuel.In the competing molten carbonate fuel cell (MCFC) fuel reforming is necessary thereby leading to a reduction in its efficiency. (2) Fuel conversion efficiencies of 50-60% should be possible, which is much higher than is obtainable with, e.g. MCFCs. (3)There should be few problems 10 Ni.Zr02 cermet anode 1 I I -40 LaMn03 cathode EXCESS e-02' 4e -20~-* /AIR AIR Fig. 5 Schematic of a solid oxide fuel cell. Adapted from ref. 15 CURRENT FLOW y YSZ PLATECELLf REPEAT 1 ,, FUEL Fig. 6 Schematic of a parallel plate SOFC design. Adapted from ref. 15 with electrolyte management, corrosion and maintenance. (4) Air pollution should be small.(5) High-grade waste heat is produced, leading to possible combined heat and power (CHP) applications for SOFCs. Recent targets for both Japan and the USA include 25 kW cells for 1990; it therefore seems likely that, within a few years, fuel cells based on the SOFC principle, may finally make a major contribution to energy management programmes. Electrochemical Reactors Yet another application of YSZ and similar oxide ion conduc- tors, in a configuration similar to that of Fig.4, is for the electrochemical partial oxidation of hydrocarbons, e.g. natural gas, to give industrially useful products such as CH30H and C2H4. Development work is still at an early stage and yields are low, but this is, nevertheless, seen as an important growth a~ea.'~,'~Choice of electrode is critical so as to catalyse the oxidation.Vayenas has shown, in the NEMCA (non-Faradaic electrochemically modified catalytic activity) technique, that application of a voltage to mixed conducting electrodes such as Bi2O3-Pr60,, leads to greatly enhanced rates for the oxidation of methane.I7 The mechanism of oxidation is unclear but may involve the active 0-species as an inter- mediate. Supercapacitors The amount of charge that can be stored in a capacitor is limited by the area of the electrodes; double-layer capacitances formed between an ionically conducting electrolyte and a metal electrode are typically optimised at a value between 1 and 10 pF cmV2, since capacitance, C, is proportional to A/d, where A is the electrode area and dis the double-layer thickness.Greatly enhanced apparent capacitances have been achieved by using as the electrode a fine mixture of electrode and solid electrolyte. Thus, in cells of the type: graphite, RbAg41 ,/R bAg,I ,/grap hi te, R bAg,I , a finely ground mixture of electronically conducting graphite and ionically conducting RbAg,I, is used as the electrode material. With this, interfacial contact areas of many square metres per gram are obtained and capacitances as high as 1-10 F are achieved in small, gram-size devices.18 The time constant, z,of a capacitor is given by the magni- tude of the RC product, which for example may have a value of 100 for a supercapacitor containing an electrolyte of resistance 1OR.Such a capacitance can be effective only at low frequencies, since from the relation or=1, co corresponds to the frequency at which the charge stored on a capacitor reaches lje of its limiting value. In the above case, the full magnitude of the capacitance would be observed only at angular frequencies, o,considerably less than 10-'Hz. Synthesis of New Materials by Ion Exchange Solid electrolytes are ideal materials for carrying out ion- exchange reactions since they have mobile ions of one type within a rigid host framework. Using ion-exchange methods, new materials can be synthesised that, thermodynamically, are metastable and could not be synthesised by other means, such as direct reaction of the components. Sometimes, the new materials have propertiesjstructures that are of techno- logical importance.Following on from the discovery of the high Na+ ion conductivity in /?-alumina, there was considerable interest in studying the structures and properties of ion-exchanged /?-aluminas.l9 This remained a topic of essentially academic interest until the discovery, by Farrington and Dunn, that Na+ ions in Nap"-alumina could be ion exchanged for a range of divalent and trivalent cation^.^'-^^ These materials provided the first examples of mobile divalent cations in solid electrolytes; the most remarkable is Pb2 +p"-alumina whose conductivity is comparable to that of Nab"-alumina over a very wide temperature range (Fig. 7). The synthesis of trivalent p"-aluminas has led to a new family of solid-state laser materials.In particular, Nd3 +pff-alumina has luminescence properties that compare very favourably with those of Nd-YAG. For instance, its absorption spectrum contains an anomalously large absorption coefficient at 573 nm, with an oscillator strength nearly 10 times that of Nd-YAG at the same wavelength. Other transitions in the two materials have comparable oscillator strengths.24 The high oscillator strength, coupled with long fluorescence life- times at high Nd concentrations, leads to potential appli- cations as lasers. One of the current problems that prevents full commercial exploitation is associated with the difficulty in growing large single crystals of p"-alumina: crystals tend to be small plates of cross-sectional area several mm2 but of thickness <1 mm.The wide range of optical properties exhib- ited by the lanthanide, transition metal and Cu' /I"-aluminas have been reviewed in ref. 23. Fabrication of Waveguide Materials by Ion Exchange The ion-exchange process discussed above may be used, under closely controlled conditions, to give inhomogeneous mater- ials. The associated compositional variations may lead to variations in refractive index and therefore, potential appli- cations in waveguides. LiNb03 single crystals may be converted into high-index optical waveguides by proton exchange of the surface lay- er~.~~.~~The surface layers form a solid solution (Li, -xHx)Nb03 whose structure depends on both the crystal- lographic orientation of the surface and the composition x.~' LiNb03 is not usually regarded as a solid electrolyte, but the Li+ ions have sufficient mobility under the conditions used, 200 "C in benzoic acid for several days, for partial ion exchange to occur.Waveguides have been fabricated from p-alumina single I I I I 1 2 3 4 5 6 lo3KIT Fig. 7Conductivity Arrhenius plots of p"-aluminas, from ref. 23 J. MATER. CHEM., 1991, VOL. 1 crystals by partial Na Ag exchange,28 but the interiors only of the crystals are allowed to ion exchange. This is done by first coating the crystals with a diffusion mask of Ni-Cr, with a polyamide overcoat, and then etching the crystal end faces uia laser lithography to expose a selected set of conduc- tion planes.The masked crystal is immersed in a bath of molten AgN03 and these inner conduction planes undergo ion exchange. The resulting material has a high refractive index inner region, associated with the Ag p-alumina and this buried waveguide structure is found to be very efficient for coupling and guiding light. Optimisation of Superconductivity by Oxygen (De)intercalation The high T, ceramic superconductors such as YBa2Cu30, and Bi2Sr2CaCu20d are mixed oxide ionic/electronic conduc- tors. Their composition 6 is variable and the different values are achieved by processing the materials at different tem- peratures and oxygen partial pressures. The critical tempera- tures, T,, generally vary greatly with oxygen content.In Bi2Sr2CaCu,0B,T, is optimised at 87 K for 6 =8.185 (Fig. 8). Such a value of 6 can be obtained, for example, in air at 820 "C or in N2 at 400 0C.29 At temperatures above 400-500 "C, especially in fine-grained powders, oxide-ion diffusion rates are sufficiently rapid that samples can respond fairly rapidly to changing T and p(02), but in dense ceramics full equilibration and hom- ogenisation may be difficult. In the well studied YBa2CuJOd materials, there is still confusion as to how T, varies with 6 (Fig. 9). In samples that have been quenched after high- temperature equilibration, T, varies approximately linearly with 8.30But in samples that have been prepared by oxygen deintercalation at relatively low temperatures (400 "C) there is evidence for plateaux at 90 and 60 K in the plot of T, us.8.31Very recent results suggest that the plateaux are caused by ordering of oxygen ions at low temperatures, together with associated changes in the electronic structure. Thus, it is possible to take a quenched material, anneal it for several days at 100-200 "C and generate increased T, values, with plateaux in plots of T, us. x similar to those in Fig. 9. It is clear that oxide ion conduction is fundamental to the processing and optimisation of the properties of the ceramic superconductors, and is important not only in controlling the overall oxygen content, 8, and therefore the hole concen- 85 80 s L" 75 70 I I I 1 I 8.15 8.16 8.17 8.18 8.19 8.20 6 in Bi,Sr,CaCu,O, Fig.8 Critical temperature us. composition for Bi,Sr,CaCu,O, (ref. 29) J. MATER. CHEM., 1991, VOL. 1 7.0 6.8 6.6 6.4 6 in YBa,Cu,O, Fig. 9 Critical temperature us. composition for Y Ba,Cu,O, for samples prepared by (0)quenching from high temperatures3' and by (0)low-temperature deintercalation of oxygen3' tration, or average oxidation state of copper, but also in achieving structures with ordered defect arrangements and modified T, values. Lithium (De)intercalation, New Materials, Solid Solution Electrodes The ability to introduce lithium ions into or remove lithium ions from certain transition metal compounds gives rise to a variety of new phases and solid solutions, some of which find application as reversible electrodes in prototype high-density battery systems.32 The principle is illustrated in Fig.10 in which TiS, behaves as an intercalation host, accepting lithium ions from the electrolyte/anode and electrons from the external circuit to form solid solutions, LixTiS2. The requirements for the host structure (TiS,) are to (a) accept electrons and be electronically conducting and (b)accept Li' ions and for them to exhibit high mobility. Many host (oxide, sulphide, etc.) structures have been intercalated successfully with lithium. Currently there is much interest in V,OI3 as a possible battery cathode since its Li+ diffusion rate is higher than that of TiS2. A convenient means of carrying out 'lithiation' is to immerse samples in n-butyl lithium dissolved in hexane.The n-butyl lithium acts as a source of lithium and the residual n-butyl useful power electrolyte TIS~ LI LI+ +Li+ + Fig. 10 Intercalation of Li into TiS2, adapted from ref. 32 groups dimerise to forin octane. This is essentially an internal redox/intercalation reaction. The reverse process of deintercalation or delithiation may be carried out using a variety of methods, e.g. electrochemical deintercalation or treatment of a sample with a solution of I2 in acetone (LiI is insoluble in acetone and gradually precipitates as Li is removed). As well as acting to reverse intercalation reactions, this method may be used to synthesise entirely new materials. For example, a new form of COO, has been synthesised by deintercalation of Li from LiCoO,.Electrochromics, Smart Windows and Displays Commercial electrochromic devices are now available based on the reversible intercalation of protons into thin films of W03, yielding a coloured tungsten bronze.33 The device structure is shown schematically in Fig. 11 and is made by evaporation onto a glass substrate of successive layers of indium tin oxide electrodes and W03. The proton source is hydrated Ta205. In the OFF state, the device is colourless and transparent. In the ON state, H+ ions intercalate the W03 and the accompanying electrons enter the 5d band of W, giving a bronze of nominal stoichiometry HXW~!-,W~O3.Absorption of light by the 5d electrons is responsible for darkening, which occurs in a matter of seconds.Such structures have been proposed as large-scale coatings on 'smart windows' for better heating/lighting management of buildings, but none are yet at the commercial stage owing to problems in fabricating large-area thin-film structures of sufficient quality and uniformity. Two small-area devices are commercially available, antidazzle car mirrors from Schotts and electrochromic spectacles from Nikon. Future Prospects All of the possible applications listed in Table 1 have been demonstrated convincingly in laboratory-scale experiments. Several are undergoing development and testing on a larger scale, and others, such as sensors and electrochromic thin- film devices, are actually on the market place.Since these various feasibility studies have already been successfully car- ried out, future developments are likely to depend on economic and environmental factors and on whether sufficient resources are made available to turn laboratory-scale devices into commercial products. It is therefore difficult to make predic- tions as to which applications are likely to achieve full commercialisation, other than for those which are already undergoing large-scale development. Of more interest is to enquire whether any further develop- ments in new materials or improved properties are possible or desirable. Many research groups are looking for new solid electrolytes with the particular objectives of finding (a) high oxide ion conductivity at intermediate temperatures (200-500 "C), (b) high protonic conductivity at similar tem- peratures, since most materials with high proton conductivity at room temperature have a high water content and are not I TO I 1 H+SOURCE Fig.11 Schematic construction of a thin-film electrochromic device 162 J. MATER. CHEM., 1991, VOL. 1 stable at high temperatures, (c) high lithium ion conductivity in atmosphere stable materials at ambient temperature. Such materials could find applications in improved fuel cells (a),(b) or batteries (c). There is also much interest in finding new mixed conductors especially those with (a) high oxide ion 10 11 12 13 R. J. Bones, J. Coetzer, R. C. Galloway and D. A. Teagle, J. Electrochem. SOC., 1987, 134, 2379. R.G. Linford, p. 564 in ref. 2. P. Jagannathan et al., in Solid Electrolytes and Their Applications, ed. E. C. Subbarao, Plenum Press, New York, 1980, p. 201. H. Iwahara, in Solid State Ionic Devices, ed. B. V. R. Chaudari conductivity for catalyst and reversible electrode applications and (b)high Li', Na' ion conductivity for reversible elec- trodes in solid-state batteries. New sensor materials and devices are required that are selective to, for instance, oxides of sulphur, oxides of nitrogen or C02. There is also much activity in developing multilayer, thin- film solid-state devices particularly for miniature power 14 15 16 17 18 and S. Radhakrishna, World Scientific, Singapore, 1989, p. 309. Proc. Int. Symp. Solid Oxide Fuel Cells, Nagoya, Japan, 1989, Science House, Tokyo.J. T. Brown, p. 630 in ref. 2. B. C. H. Steele, I. Kelly, H. Middleton and R. Rudkin, Solid State Zonics, 1988, 28-30, 1547. C. G. Vayenas, Solid State Zonics, 1988, 28-30, 1521. R. A. Huggins, p. 664 in ref. 2. sources. An advantage of miniaturisation is that the electrolyte resistance decreases linearly as thickness decreases and there- fore, the normally stringent requirements of having low specific resistivity for ionic conduction can be relaxed somewhat. In summary, there are probably a considerable number of 19 20 21 22 J. T. Kummer, Prog. Solid State Chem., 1972, 7, 141. B. Dunn and G. C. Farrington, Mater. Res. Bull., 1980, 15, 1773. G. C. Farrington and B. Dunn, Solid State Zonics, 1982, 7, 267. G. C. Farrington, B. Dunn and J. 0.Thomas, Appl. Phys. A, 1983, 32, 159. potential ionic conductors/mixed conductors waiting to be discovered and a sufficient number of perceived applications to ensure continued vigorous activity in this area. 23 24 25 G. C. Farrington, B. Dunn and J. 0.Thomas, p. 327 in ref. 2. M. Jansen, A. J. Alfrey, 0.M. Stafsudd, D. L. Yang, B. Dunn and G. C. Farrington, Opt. Lett., 1984, 9, 119. J. L. Jackel, A. M. Glass, G. E. Peterson, C. E. Rice, D. H. Olson and J. J. Veeselka, J. Appl. Phys., 1984, 55, 269. References 26 J. L. Jackel, C. E. Rice and J. J. Veeselka, Electron. Lett., 1983, 19, 387. A. R. West, Ber. Bunsenges. Phys. Chem., 1989,93, 1235. 27 M. Ito and H. Takei, Jpn. J. Appl. Phys., 1989, 28, 144. High Conductivity Solid Ionic Conductors, Recent Trends and Applications, ed. T. Takahashi, World Scientific, Singapore, 1989. 28 B. Dunn, G. C. Farrington and J. 0. Thomas, ZSSZ Lett., 1990, 1, 1. J. R. MacCallum and C. A. Vincent, Polymer Electrolyte Reviews, 29 C. Namgung, J.T. S. Irvine, J.H. Binks, E. E. Lachowski and Elsevier, Barking, 1987, 1989, vol. 1 and 2. S. Etemad, A. J. Heeger and A.G. MacDiarmid, Rev. Phys. 30 A. R. West, Supercond. Sci. Tech., 1989, 2, 181. C. Namgung, J.T. S. Irvine and A. R. West, Physica C, 1990, Chem., 1982, 33, 443. 168, 346. Y. F. T. Yao and J. T. Kummer, J. Znorg. Nucl. Chem., 1967, 29 31 R. J. Cava, B. Batlogg, C. H. Chen, E. A. Rietman, S. M. Zahurak 2453. T. Takahashi, p. 1 in ref. 2. W. Fischer, p. 595 in ref. 2. J. Coetzer, J. Power Sources, 1986, 18, 377. 32 33 and D. Werder, Phys. Rev. B, 1987, 36, 5719. K. West, p. 447 in ref. 2. F. G. K. Baucke and J. A. Duffy, Chem. Br., 1985, 643. R. C. Galloway, J. Electrochem. SOC., 1987, 134, 256. Paper 0/02205E;Received 17th May, 1990