J. MATER. CHEM., 1994. 4( 1 ), 113-1 18 Vanadium Phosphate Glasses. Effect of Composition on their Structure and Performance as Cathodes in High-temperature Lithium Polymer-electrolyte Cells Andrew Davies: Richard J. Hobson: Michael J. Hudson: William J. Macklinb and Robin J. Neat!' a Department of Chemistry, University of Reading, Whiteknights, Reading, UK RG6 2AD Applied Electrochemistry Department, A€A Industrial Technology, Harwell Laboratory, Oxfordshire, UK OX77 ORA Five vanadium phosphate (v@,-P@,) glasses, containing between 58 and 88 mol% V205, have been prepared using the melt-quenching technique. IR spectroscopy, in conjunction with powder X-ray diffraction (XRD) of the devitrified materials, has been used to show that the glasses containing d70molo/o V205 have a microstructure which is similar to that of B-VPO, whereas the remainder show both orthorhombic-V,O, and p-VPO, structural features. Molar-volume data suggest that there is a fairly abrupt change in microstructure at ca.75 mol% V,05. The glasses have been employed as the active component of the cathode in lithium polymer-electrolyte cells operating at 120 "C and the cycling performance has been investigated as a function of glass composition. After an initial capacity decline the cells showed good reversibility, although this was not achieved at capacities as high as that of V,O,, in a similar cell. Glasses containing less than 78 mol% V205 cycled at considerably lower capacities than those with higher vanadium content and this has been related to the above change in microstructure from a layered V,O,-type to a network p-VPO, type.The electrochemical insertion of lithium into vanadium (v) oxide containing glasses has been studied by a number of researchers. Many of these studies involved ternary materials with a glass-forming oxide also present, such as the systems V205-P205,1p4 V205-B203,5 and V205-Te0,.6 The V205-P205 system, which may be prepared using a moder- ately fast quenching technique ( 102-103 "C s-'), has shown some interesting features. Lithium-ion insertion into a glass of this type containing 60 mol% V205 was first demonstrated by Pagnier et al. in 1983.' Unfortunately, when tested in a lithium polymer-electrolyte cell operating at 85 "C the glass exhibited poor reversibility which was attributed to the elec- trolyte becoming oxidized near the upper voltage limit.However, a cathode consisting of a single solid piece of glass rather than an intimately mixed composite cathode was employed, which would in any case have led to a poor cell performance regardless of the cathode material because inti- mate contact between the cathode active material and the electrolyte is critical in determining cell perf~rmance.~ Reversible insertion of Li' ions in V205-P@5 glasses was first reported in ambient-temperature liquid-electrolyte cells by Sakurai and Yamaki? who later demonstrated that the glassy materials were superior to crystalline V205 in terms of long-term re~ersibility.~ Reversibility over several hundred cycles was reported for all glass compositions.A relationship between microstructure and composition has been shown for V205-P205 gla~ses.~.~Bhargava and Condrate; on the basis of IR spectral evidence, suggested that the short-range order of the glasses, particularly those with a high P205content (up to 50mol0/0), resembles that of c(-VP05. Sakurai and Yamaki,4 however, proposed that glasses containing <75 mol% V205have a structure which resembles that of the p-VP05 network, whereas above this boundary a change to a 'V205-like' structure occurs. These authors found a significant difference in electrochemical properties either side of this boundary. The good long-term reversibility reported for vanadium phosphate glasses in room- temperature liquid-electrol yte cells24 has led us to reinvestigate the rechargeability of these materials in lithium polymer-electrolyte cells operating at 120"C.The percentage theoretical capacities and reversibility of the glasses are compared with those of V,OI3 for the first time; to warrant further study a material would have to show a significant improvement in reversible capacity and/or energy density over V6OI3. In addition, we have re-examined the IR spectra of the glasses and compared these with the spectra of their crystalline devitrification products, for which we were able to obtain definitive structural information. Experimental Synthesis The vanadium phosphate glasses were prepared from in timate mixtures of powdered V205 and P,O,.Mixtures containing initially 55, 65, 75, 80, and 85 mol% V,O, were employed. These were prepared in a nitrogen-filled dry box and then heated in air to 750°C in a platinum crucible for 1 h. The melt was quenched by pouring it onto a pre-cooled (5°C) stainless-steel plate. Devitrified samples were prepared by heating the glasses at 100"C above their crystallization tem- peratures for 2 h. Techniques Scanning electron micrographs were recorded using a JEOL JXA 840 scanning electron microscope. The vanadium :phos-phorus molar ratios of the glasses were established using electron microprobe X-ray analysis on the same instrument. Reduced vanadium [V"] and total vanadium were deter- mined by potassium permanganate titration and atomic absorption spectroscopy (AAS), respectively.Densities were obtained by pycnometry using toluene or cyclohexane. Electronic conductivities were measured on solid pieces of glass, cut using a diamond stylus, of area =1 cm2 and thickness 2.5 1.0mm (measured using a micrometer). Conducting adhesive copper tape was attached to cover both faces and the resistance through the thickness of the sample was meas- ured at 25°C. Powder XRD patterns were recorded using graphite monochromated Cu-Kcc radiation on a Philips PW1710 diffractometer controlled by a Citrons Cray 112 system run- ning Sietronics 112 software. Data were collected for the 28 range 4-64" at a rate of 2" min-' in steps of 0.04". Differential thermal analyses were carried out using a Stanton Redcroft STAlOOO instrument equipped with data manipulation software using a heating rate of 10"C min-'.IR spectra were measured between 1400 and 400 cm-' as KBr discs using a Perkin-Elmer 1720-X FT-IR spectrometer. Cell Fabrication and Cycling Composite cathodes containing ground vanadium phosphate glass (45 vol.%, particle size <50 pm), ketjenblack carbon (5 vol.%), poly(ethy1ene oxide) (PEO) (Union Carbide MW 4 000 000) and LiC104 (Aldrich) (50 vol.% PEO-LiC104, CEO units]," Li] = 12) were prepared via doctor blade casting from the appropriate solvent slurry onto a nickel current collector. Sheets of the electrolyte PEO-LiC104 (CEO units]/[Li] =12) were cast from acetonitrile solution onto silicone release paper.Solid-state cells, Fig. 1, incorporating a lithium foil anode (Lithco 150 pm) with an active area of 40 cm2 were constructed in a dry room (T=20 "C, dew-point temperature -30 "C) using a combination of heat and pressure. The cells had a capacity, C, of ca. 30mA h. This was calculated using the theoretical capacity of V6013, 417 mA h 8-l for Li,V,O,,, to enable a direct comparison between the performance of the glasses and V601,. Cell cycling of the packaged cells was performed galvanostatically under computer control between limits of 3.5 and 1.7 V at a rate of C/10 and a temperature of 120°C. Prior to cycling the cells were allowed to equilibrate at 120 "C for 2 h when their open- circuit voltages were measured.For each cathode material two cells were cycled, typically 30 times, in order to establish the reproducibility of the results. Results and Discussion The glasses prepared were hard and smooth with no visible crystallites. Their amorphous nature was confirmed using powder XRD, which showed no Bragg peaks, and scanning electron microscopy (SEM). The electron micrographs of fractured samples showed no crystallites (Fig. 2) at any magni- fication up to 2800, at which crystallites as small as 5 x m should be clearly visible. Micrographs of the powdered samples employed in the fabrication of cells showed particles of random size (<50 pm) and shape with no regular faces. The amount of V205 found in the glasses was in all cases slightly larger than the initial content of the mixtures, Table 1, owing to the sublimation of a small amount of P,O, prior to 45 vol% V2O5-P2O5glass Li / PEOl2:LiC1O4/ 5 vol% carbon / Ni 50 vol% PEOl2:LiCIO4 Fig. 1 Lithium polymer-electrolyte cell configuration J.MATER. CHEM., 1994, VOL. 4 Fig. micrograph Of a powdered Of 82 molo/o V,O, glass Table 1 Analytical data of glasses V,O, (mol%) initial found V'"/V,,, density/g cmP3 TJC lJT 55 58 0.19 2.78 390 572 65 70 0.17 2.79 295 410 75 78 0.12 2.84 266 403 80 82 0.09 2.90 241 326 85 88 0.07 3.OO 235 275 melting. The figures given are based on the vanadium :phos-phorus molar ratios determined using electron microprobe analysis and were confirmed by the AAS analyses for total vanadium.All of the glasses were dark blue and contained some vanadium(Iv), see Table 1, produced during the prep- arations with concomitant loss of oxygen. (No correction for this has been applied to the mol% of V205.) The proportion and amount of V" decreases with increasing vanadium con- tent. The measured densities increase with vanadium content as would be expected. Differential thermal analyses of all of the glasses showed an endothermic discontinuity in the baseline indicating the glass-transition temperature,' q,followed by an exotherm due to crystallization of the glass, the onset of which' gives the crystallization temperature, T,. The data are summarized in Table 1. As expected, both Tp and T, increase with the amount of the better glass-forming component, i.e.P,Os. The measured electronic conductivities, which increased with increasing V205 content, ranged from 7.9 x to 3.2 x lop5S cm-l, of the same order as those determined previously" for similar materials at ambient temperature. The conductivities of the glasses would be expected" to be higher at the cell operating temperature of 120 "C. Microstructures of Glasses The microstructures of the glasses would be expected to have a profound influence on their performance as cathode active materials in lithium batteries. Amorphous materials contain- ing solely vanadium(v) oxide have microstructures which are related" to that of orthorhombic V2OS and therefore have sites which are suitable for lithium insertion and V-0-V linkages which, together with the presence of some VIV, are necessary for electronic conduction.The vanadium phosphate glasses, however, are likely to have microstructures based on VPO,, which is dimorphic. Previous investigator^'^^^^ have demonstrated the presence of VO, and PO4 moieties in the glasses and on the basis of IR spectral studies, the glasses J. MATER. CHEM., 1994, VOL. 4 have been assigned microstructures related to r-VPO, by one group' and p-VPO, by a second,, (see earlier). VPO, crystallizes in the tetrag~nal'~ (a) or orthorh~mbic'~ (p)systems. Both modifications contain highly distorted VO, octahedra which are linked into chains via corner sharing along (001)(a)or (100) (p).However, the V-0, bonds linking the oc!ahedra are unequal (a, 1.580 and 2.857 A; p, 1.566 and 2.591 A) and a better description of the structures can be obtained in terms of chains of weakly linked VO, square-based pyramids, the four basal oxygen atoms of which are corner-shared with four PO, tetrahedra. The PO4 tetrahedra in a-VPO, link four chains of VO, units giving rise to layers lying parallel to (001) with only weak links between them.Thus there are sites which may be suitable for lithium-ion insertion, similar to those16 in orthorhombic V205. However, the PO4 tetrahedra in p-VPO, link three chains of VO, units and bridge two adjacent VO, moieties in the same chain, giving rise to a network. This bridging is likely to hinder the free movement of lithium ions.Both structures lack symmetric V-0-V bridges so a glass with a microstructure based on either form would be expected to have a poorer electronic conductivity than one with a V,O,-like microstructure. Molar Volume Drake17 et al. studied a series of vanadium phosphate glasses containing between 47 and 74 mol% V205.They showed that the molar volume per gram atom of oxygen, V2;, showed a 'monotonic and quasi-linear' change with composition, indi- cating that there was no phase separation or structural change in their composition range. Sakurai and Yamaki4 examined glasses containing between 58 and 95 mol% V205 and sug- gested that an inflection in the Vg us. composition plot at ca. 75 mol% V205 indicated a change in microstructure in this region.We calculated Vg using the expression where x =molar fraction, M =relative molar or atomic mass, r =molar ratio VIV:Vtot,and p =density. Similar values to those of Sakurai and Yamaki4 were obtained. However, rather than a point of inflection in the V2; us. composition plot, a maximum was observed between 70 and 78 mol% V205, see Fig. 3, confirming that a change in microstructure or amorph- ous phase separation occurs in this composition range. The latter interpretation is more consistent with our IR evidence given later. The differences between our results and those of Sakurai and Yamaki can be attributed largely to significant 7' 13.4 lP4t A T W W 2.8 b10.8. Fig. 3 Dependence of (0)V;S (see text) and (H)density on composi-tion.The 100mol% data are for crystalline V,O,. The smoothed lines were drawn by fitting a third-degree polynomial piece-wise to the data. differences in the measured densities, see Fig. 3, to which VT, is particularly sensitive. The inclusion of the data for crystalline V205 shows that it fits well our observed trends. Powder X-Ray DifSraction of the Deuitrifed Glasses The diffraction patterns of the devitrified glasses, examples of which are shown in Fig. 4, demonstrate clearly that the products of devitrification are p-VPO, and orthorhombic- V205. The devitrified 58 mol% V,O, glass gave a diffraction pattern which is an excellent match with the calc~lated'~~'' diffraction pattern of p-VPO,, with the exception of two weak peaks at 20=23.22 and 28.34' which do not correspond with either a-VPO, or orthorhombic V205.This match is in contrast with the result of Sakurai and Yamaki4 who obtained an unidentified crystalline material upon devitrification of a similar glass. The devitrified 70 mol% glass gave a very similar diffraction pattern to the 58 mol% material. In both of these some vanadium atoms must presumably occupy some of the tetrahedral phosphorus sites. Such a suggestion' with respect to the glasses has been made previously on the basis of a low angle X-ray scattering study. The other devitrified glasses all gave diffraction patterns which showed a mixture of V20, and p-VPO,, Fig. 4. We found no evidence for the existence of solid solutions in the crystalline V205-P,O, system :is has been reported previ~usly.~ IR Spectra of Glasses and Devitrifed Glasses The IR spectra of the glasses and their devitrification products, for which definitive structural information had been obtained, were studied in order to resolve the differences in the interpret- ation4q8 of their microstructure.The spectra of the glasses showed bands which were substantially broader than those of the crystalline materials, as shown in Fig. 5. This is consist- ent with structures lacking long-range order that have a range of bond lengths and strengths. However, the general features 10 20 30 40 50 60 28ldegrees Fig. 4 Powder XRD patterns of (a)orthorhombic V20,, (b)delritrified 82 mol% V20, glass, (c)devitrified 58 mol% V,O, glass and (d) the pattern calculated for p-VPO, using Lazy Pulverix.18 1200 800 4 wavenumberkm-' Fig.5 IR spectra, (a) orthorhombic V20,; (b)88 mol% V205 devitri-fied material; (c) 88 mol% V205 glass; (d) 58 mol% V20, devitrified material; (e)58 mol% V20, glass of the IR spectra of the vitreous and corresponding non-vitreous materials are similar. Between 1250 and 1050 cm-' the spectra of the glasses show a strong, broad feature and those of the devitrified materials two strong bands which are characteristic' of p-VPO, (1052 and 1149 cm-I). a-VPO, has no bands in this region but instead shows a weak band' at 1211 cm-I absent in all our spectra. The V=O stretching vibration in crystalline V205, p-VPO, and a-VPO, is at 1029, 1000, and 990 cm-', respectively,' and is also evident in the vitreous (1008-1021 cm-') and non-vitreous (1000-1010 cm-') materials.In the region 950-800 cm-' both V205 and p-VPO, show a single, strong absorption (828 and 942 cm- ', respectively) whereas a-VPO, has only very weak features.' The three devitrified materials with higher vanadium content display both bands, whereas the other two show only the 940cm-' band. The maxima of the broad absorption envelopes are at 840 cm-' for the three glasses containing 378 mol% V205 and 920 and 928 cm-' for the 70 and 58 mol% glasses, respectively. This is consistent' with the loss of V-0-V linkages as the proportion of P205increases. Between 800 and 400cm-' the three devitrified materials with higher vanadium content show the two intense bands of orthorhombic V205 (602 and 478 cm-') whereas the 58 and 70mol% materials display the five bands' of p-VPO,.The glasses with >,78 mol% V205show a strong band at 640 cm- and a broad indistinct band around 400 cm-' whereas the 70 and 58 mol% glasses both show a medium intensity shoulder at 780cm-' and a medium or low intensity band around 630 cm-'. The broad absorption envelopes of these glasses, with the exception of the shoulder at 780cm-', encompass the bands of the devitrified materials, see Fig. 5. The superficial similarity between the spectra of the low vanadium glasses and a-VPO, in this region (a-VPO, has two bands: 780 crn-', very weak; 602 cm-', sharp, medium) does not, we believe, show that they have an a-VP0,-like microstructure.' These bands are difficult to assign specifically to an isolated group vibration' and because of the lack of long-range periodicity and differences in local symmetry the vibrational modes of J.MATER. CHEM., 1994, VOL. 4 the glasses are likely to differ from those of the crystalline materials. The considerable similarities between the TR spectra of the glasses and the equivalent non-vitreous materials strongly suggest that the components present in the latter reflect the microstructural domains present in the glasses. Thus the glasses containing <70 mol% V205 have microstructures which are predominantly p-VPO,-like, whereas the glasses containing 378 mol% V205 have domains which are V20,- like and domains which are p-VPO,-like.Without further evidence it is not possible to state unequivocally that this indicates amorphous phase separation although this has been reported previously by several groups of u'orkers.8*'s21 The consequence of this structural change is a profound depen- dence on composition in the performance of the glasses as cathode-active materials. Performance in High-temperature Lithium Polymer-electrolyte Cells Cells containing four of the glasses (70, 78, 82, and 88 mol%) were constructed and cycled as described. The open-circuit voltages of the cells, ca. 3.5 V, were similar to the values forrep~rted~.~ vanadium phosphate glasses in room-temperature liquid-electrolyte cells and or thorhombic V20, in an identical polymer-electrolyte cell but higher than that of cells incorporating V6OI3 (2.8 V). The first discharge curves of the cells are shown in Fig.6 in comparison with those of crystalline V6OI3 and V205 in identical cells. The smooth curves given by the glasses show that, unlike crystalline vanadium oxides, there are no specific sites for lithium inser- tion, which is characteristic of the amorphous state. The 70 mol% glass discharges to a much lower capacity on the first cycle (73% theoretical) than the glasses with higher V205 content, which all gave similar first discharge capacities (> 90% theoretical). This large difference cannot be attributed solely to the vanadium content of the different glasses but rather seems to be associated with the change in microstruc- ture which occurs at ca.75mol% V205 (see earlier). A network structure similar to that of 0-VPO, would restrict the diffusion of Li+ ions and give rise to a lower electronic conductivity (as observed). Our molar volume data, which showed a maximum around 75 mol% V20,, are not consistent with a previous suggestion4 that lithium diffusion is being restricted by a more densely packed structure in the 70 mol% glass. The later discharges for all of the glass-containing cells were also monotonous, as shown in Fig. 6, demonstrating that cycling at elevated temperatures does not induce crys- tallization of the vanadium phosphate glasses.The percentage theoretical capacity obtained is plotted against cycle number in Fig. 7. All of the cells showed a large capacity decline over the first four or five cycles which was primarily associated with a steepening of the voltage-capacity curve between 3.5 and 2.7 V, see Fig. 6,indicating that lithium was being retained in the lower energy sites. This was followed by a region of slow recovery so that by the 25th cycle the cells containing the three glasses with 378 mol% V205 cycled reversibly at around 60% theoretical capacity, with the highest utilization occurring for the highest vanadium content, as expected. This recovery, we believe, is due to the formation of a more open structure with continued cycling. Cathodes containing crystal- line V6013 show very similar behaviour to those containing the glasses, see Fig.7. This oxide has been shown" to become amorphous in high-temperature polymer-electrolyte cells after a few cycles, so the similarities seen here are perhaps not surprising. None of glasses we examined showed a utilization approaching that of V6013 (80% theoretical capacity). J. MATER. CHEM., 1994, VOL. 4 (b) 1 l.I.I I. 0 20 40 60 80 100 (c 1 3.0-1.51 I 1.51 II I I I I 1 1 I0 20 40 60 80 100 0 20 40 60 80 100 theoretical capacity (%) Fig.6 First and later discharge curves (as numbered) for cells containing V205-P205glasses as the cathode-active component and the first discharge curves for similar cells containing V6OI3 or Vz05.(a) 82 mol% V205 (b) 70 molo/o V205; (c) V6OI3;(d)crystalline V205 1001 = I 01 II I I 1 0 6 12 18 24 30 cycle number Fig. 7 Cycling performance of cells containing V205-P205 glasses as the cathode-active component in comparison to a cell containing V6OI3.(a) VOOl3;(h) 82 mol% V205;(c) 70 mol% Vz05 Cells containing the 70 mol% glass recovered after 30 cycles to a much lower capacity than those containing the other glasses (cci. 30% theoretical) which cannot be attributed solely to the lower vanadium content but, like the lower capacity on the first discharge, must be associated with the change in microstructure discussed above. However, our results show a significant improvement in reversibility compared with those of Pagnier et a/.' who experienced difficulty in recharging their cells.The disappointing reversibility they observed may have been due to the low electronic conductivity of their cathodes, which were prepared without the addition of either ketjenblack carbon, for increased electronic conductivity, or polymer electrolyte, for improved interfacial contact between cathode material and electrolyte. Our results, we believe, demonstrate the significant contribution of the composite cathode to cell performance. Fig. 8 compares the observed energy densities of two of the glasses with that of V6013in a similar cell. Although cells incorporating the glasses gave higher open-circuit voltages than V6OI3, and exhibited comparable average cell voltages on cycling (2.1-2.4 V), their specific capacities were slightly lower than that of V6OI3(ca.330 mA h g-') at the capacity plateau. This results in lower observed energy densities 1OOOr -800 iY0) .c 6oo. c).-v)c % 400-P i im -*-------'\A-+---+--------0 5 10 15 20 25 cycle number Fig.8 Variation of observed energy density with cycle number for (*) 70 and (A)82 mol% V205glasses compared with that of V6OI3(m)in a similar cell (500-600 W h kg-' for glasses containing 278 mol% V205) after 25 cycles than those observed for V6013 (ca. 770 W h kg-'). Conclusions The microstructures of vanadium phosphate glasses contain- ing < ca. 75 mol% V205 are predominantly fl-VPO,-like whereas those containing 3 ca. 75 mol% V205 have micro- structures with both V,O,-like and P-VP0,-like domains. All of the glasses we examined can be cycled reversibly when employed as cathodes in secondary polymer-electrolyte cells operating at 120"C, with those containing greater amounts of V205 giving the largest specific energy densities.This difference in performance is not solely related to the vanadium content but is also profoundly influenced by the microstruc- ture of the glasses. The better materials gave lower specific 118 J. MATER. CHEM., 1994, VOL. 4 energy densities over the first 30 cycles than obtained for V6OI3in a similar cell. 9 10 11 M. E. Brown, Introduction to Thermal Analysis, Chapman and Hall, London, 1988. M. Sayer and A. Masingh, Phys. Rev. B, 1972.6,4629.P. Aldebert, H. W. Haesslin, N. Baffier and J. Livage, J. Coll. We thank SERC and AEA Industrial Technology, Harwell Laboratory, for a CASE award to A.D. 12 Interface Sci., 1984,94,484, and references therein. Bh. V. Janakirama-Rao, J. Am. Ceram. Soc., 1966, 49, 605; Bh. V. Janakirama-Rao, J.Am. Ceram. Soc., 1965,48, 31 1. 13 F. R. Landsberger and P. J. Bray, J. Chem. Pkys., 1970,53,2757. References 14 15 B. Jordan and C. Calvo, Can. J, Chem., 1973,51,2621. R. Gopal and C. Calvo, J. Solid State Chem., I972,5,432. 1 T. Pagnier, M. Fouletier and J. L. Souquet, Solid State Ionics, 1983.9 & 10,649. 2 Y. Sakurai and J. Yamaki, J. Electrochem. Soc., 1985, 132,512. 3 Y. Sakurai, S. Okada, J. Yamaki and T. Okada, J. Power Sources, 1987,20, 173. 4 Y. Sakurai and J. Yamaki, J. Electrochem. SOC.,1988,135,791. 5 A. C. Leech, J. R. Owen and B. C. H. Steele, Solid State Ionics, 16 17 18 19 20 D. W. Murphy, P. A. Christian. F. J. DiSalvo and J. V. Waszczak, Inorg. Chem., 1979,18,2800. C. F. Drake, J. A. Stephan and B. Yates, J. Non-Cryst. Solids, 1978,28,61. K. Yvon, W. Jeitschko and E. Parthe, J. Appl. Cryst., 1977, 10, 73. G. W. Anderson and F. U. Luehrs, J. Appl. Phys., 1968,39, 1634. G. W. Anderson and D. W. Compton, J. Chem. Phys., 1970, 52, 6166. 1983,9 & 10,645. 21 E. J. Freibele, L. K. Wilson and D. L. Kinser. J.Am. Ceram. Soc., 6 M. Levy, P. Rousseau and M. J. Duclot, Solid State lonics, 1988, 1972,55, 164. 28-30,736. 22 W. J. Macklin, R. J. Neat and S. S. Sandhu. Electrochim. Acta, 7 A. Hooper and J. M. North, Solid State lonics, 1983,9 & 10, 1161. 1992,37,1715. 8 R. N. Bhargava and R. A. Condrate Sr., Appl. Spectrosc., 1977, 31,230. Paper 3/04197B; Receired 19th July, 1993