J. MATER. CHEM., 1994,4(8), 1331-1336 Zirconia Formation by Reaction of Zirconium Sulfate in Molten Alkali-metal Nitrates or Nitrites Huda Al Raihani," Bernard Durand,bVc9* F. Chassagneux,b David H. Kerridged and Douglas lnmana a Department of Materials, Imperial College of Science Technology and Medicine, Prince Consort Road, London, UK SW72BP Laboratoire de Chimie Minerale 3, URA CNRS no. 7 76, ISIDT, Universite Claude Bernard Lyon 7, 43 Boulevard du 17 Novembre 7978, 69622 Villeurbanne Cedex, France Laboratoire de Materiaux Mineraux, URA CNRS no. 428, Ecole Nationale Superieure de Chimie de Mulhouse, 3 rue A. Werner, 68093 Mulhouse Cedex, France Department of Chemistry, University of Southampton, High field, Southampton, UK SO9 5NH Thermogravimetry up to 500 "C reveals several multi-stage reactions between Zr(SO,), and molten alkali-metal nitrates which form poorly crystallised mixtures of tetragonal and monoclinic zirconia.The use of NaNO, gives incomplete transformations. The addition of NaCl as well as the use of LiN0,-KNO,, leads the reaction to completion; the increase of basicity by the addition of Na,CO, also has the same effect, if it is assumed that Na,ZrO, is formed together with ZrO,. Pure and finely divided tetragonal zirconia is produced by a single-stage reaction at a temperature lower than 300 "C, when the basicity of the medium is increased by replacing LiN0,-KNO, by NaN0,-KNO,. Finely divided zirconia powder is of interest as a potentially useful or catalyst ~arrier,~ as well as a precursor of ceramics.Historically, a great deal of effort has been spent on the synthesis of small particle size ceramic powder^.^-^ The ideal powder should be chemically homogeneous at the atomic level and of high purity; it should consist of fine particles without hard aggregates. Reactions in molten salts are likely to lead to such oxide powder^.^,^ For zirconia particularly, Jebrouni et al. showed that the reaction of zirconyl chloride, ZrOC12.8H20, with molten alkali-metal nitrates in the range 400-500 "C leads either to pure' or yttrium-stabilised zirconia powders'' characterised by a satisfactory purity, a very homo- geneous distribution of the elements and nano-sized crystal- lites. Because they have specific surface areas > 100 m2 g-', which are stable to annealing at 6OO0C, these powders are very useful in catalysis as carrier^.^." Moreover, high densities are achieved by natural sintering at 1500 OC.I2 The reaction of zirconium sulfate towards a molten LiN03-KN03 eutectic was first investigated by Kerridge and Can~ela-Rey,'~who showed that the formation of insoluble ZrO, proceeded in two stages. The increase in the basicity of the molten medium brought about by the addition of Na202, Na20 or NaOH shifted the transformations towards low temperatures but favoured the formation of alkaline zircon- ates.The characteristics of the zirconia obtained were not determined. The reactions taking place in molten alkali-metal nitrate baths are based on the Lux-Flood principle. NO3-ions are considered as bases as they give rise to oxide ions 0,-;Zr4+ ions, resulting from the dissolution of the zirconium starting salt, are considered as acids because they accept oxide ion^.'^-'^ The mechanism of the whole reaction can be written as follows: dissolution of Zr(SO,),: Zr(S0,),+Zr4+ +2S042-dissociation of nitrates: NO3- +NO,+ +02-recombination of NO,': NO,' +NO3--+[N2O5]+ 2N02++02 precipitation of zirconia: Zr4++20,--+Zr02 Thus it is easily understood that a change in the acid-base properties or of the complexing efficiency of the molten medium may have a strong influence on the development of reactions of transiton-metal salts and on the properties of the powders obtained.Nitrite ions, NO,-, are more basic than nitrate ions, their dissociation constant to oxide anions being some lo1' larger than that of nitrate ions under equivalent conditions.This paper is concerned with a study of the reaction of zirconium sulfate, either hydrated or anhydrous, with various molten nitrate and nitrite media in order to determine the best conditions for obtaining finely divided zirconia powders. Experimental Materials Zirconium sulfate: zirconium sulfate tetrahydrate, Zr(S04),-4H20 (BDH AnalaR), was used dehydrated or as received. The dehydration was performed with sulfuric acid followed by ignition at 350-400°C according to Bear's method.17 lithium nitrate-potassium nitrate eutectic (mp 132 OC): lithium nitrate trihydrate and potassium nitrate (BDH AnalaR) were dehydrated separately in an oven at 180°C for 24 h.After they had been cooled at room temperature in a desiccator, the nitrates were weighed to give a mixture of 43 mol% lithium nitrate which then melted at 180 "C over a few hours with occasional stirring. The melt was then filtered and cooled in a desiccator. Finally the solidified melt was broken into convenient lumps in a dry box and stored in a desiccator. Sodium nitrite-potassium nitrite eutectic (mp 220 ' C): the eutectic (35 mol% potassium nitrite) was prepared in the same way as for nitrates, from sodium nitrite and potassium nitrite (BDH) and the mixture melted at 240°C. Other materials: sodium nitrate, sodium chloride and sodium carbonate (BDH AnalaR) were used as received.Procedure Most of the reactions were performed in the crucible of a Setaram G70 thermobalance to investigate the weight loss produced by gas release. Depending on the final temperature, either a Pyrex or a silica crucible were used. Zirconium sulfate and alkali-metal nitrates or nitrites were quickly hand-mixed in a mortar; a few hundredmg were introduced into the crucible. From the expected weight loss, blank runs indicated that buoyancy corrections were not necessary. No mass losses J. M.4TER. CHEM., 1994, VOL. 4 were observed below 550 "C when the molten salts (nitrates or nitrites) were heated alone. The derivative of the mass us. temperature relationship (thermogram) was used to increase the resolution of overlapping mass losses.Experiments were generally performed with a heating rate of 5 "C min-'. Some experiments were carried out, on a larger scale, inside a vertical cylindrical Pyrex reactor closed at its upper end with a drying tube filled with silica gel. The reactor was heated in a regulated vertical muffle furnace, the temperature being checked by means of a chromel-alumel thermocouple inserted between the reactor and the furnace walls. The reaction products were identified by X-ray diffractome-try (XRD), either on the ground solidified reaction mixture or on the insoluble solids after the solidified melt had been washed with water. A Siemens D500 diffractometer working with Cu-Kcr radiation (A=0.154 nm) and coupled with a Digital PDP 11 computer was employed.Identification was performed by comparison of experimental values with JCPDS files (Table 1).The size of crystallites in the tetragonal zirconia powders was determined from the line-broadening of the ( 111) XRD peak by means of the Warren-Scherrer formula. The extraction of zirconia powders was performed by dissolving the quenched melt in water, filtering of the insoluble oxide and then washing it with 50ml aliquots of water until qualitative tests showed no sulfate or nitrate in the washings. After the powder had been washed, the remaining liquid was removed by vacuum filtration and the zirconia powder dried in a desiccator. The morphology of some powders was characterised by transmission electron microscopy (TEM) on a JEOL 200CX apparatus.The TEM samples were prepared from alcoholic suspensions dispersed in an ultrasound bath. A drop was carefully evaporated on a carbon film predeposited on a copper grid. Results and Discussion Reaction of Anhydrous Zirconium Sulfate with Sodium Nitrate Reactions were performed, using a stoichiometric excess of NaNO, C0.2 mol Zr(S04)2 per mol NaNO,]. Anhydrous ZT(SO~)~seemed soluble in the molten sodium salt (mp 308°C) and began to react visibly just below the melting point at ca. 260"C, evolving a mixture of nitrogen dioxide (easily recognisable by its brown-red colour) and oxygen (confirmed with pyrogallol). Thermogravimetry (TG) (Fig. 1) showed this first reaction together with a second one com-mencing at ca.370 "C and evolving the same gaseous mixture. The maximum rates were observed at 320 and 400 "C, respect-ively. The investigation of a reaction mixture quenched at 350 "C indicated that the first reaction leads to a water-soluble product. The addition of a small quantity of sodium hydroxide, or heating the aqueous solutions, produced a gelatinous precipitate which was amorphous when analysed by XRD. The solidified melt gave two diffraction lines [d=3.89 (70) and 3.23 (100)A] which could not be attributed to zirconium dioxide. Kerridge and Can~ela-Rey'~previously observed Table 1 XRD identification of monoclinic, tetragonal and cubic-zirconia in the range 27.5-32" [L Cu-Kcr, = 1.5405 A] monoclinic tetragonal cubic JCPDS 37-1484 JCPDS 17-923 JCPDS 27-997 d/A 2O/degrees d/A 281degrees d/A 2O/degrees ~~ 3.165 23.17 2.96 30.17 2.93 30.48 2.841 31.46 91 .G 30E .-1 560 6006 0 100 200 300 400 500 TI'C Fig. 1 TG of the reaction of anhydrous zirconium sulfate in molten sodium nitrate. Influence of the addition of sodium chloride and sodium carbonate: 0, NaNO, alone; A, NaNO,+NaCI; a, NaNO, +Na,CO,.identical behaviour and the same peaks for the reaction of zirconium sulfate with a molten LiN03-KN03 eutectic. The product first formed is probably a basic nitrate. The overall mass loss measured at 650°C corresponds to 68.0% of the total Zr(SOJ2 added. This is lower than the calculated value, 76.3%, if the reaction is assumed to occur according to: Zr(S04)2+4N03--+Zr02+2S0,2-+4NO2+o2 (1) The difference indicates incomplete transformation.XRD on the extracted insoluble solid identified a mixture of tetragonal and monoclinic zirconia [Fig. 2(a), Table 11. When sodium chloride was added to sodium nitrate (0.2 mol NaCl per mol NaN0,) the first transformation was not changed, the maximum rate still occurring near 320"C, but the second one was shifted slightly towards low temperatures, with the highest rate near 370°C (Fig. 1). The overall mass loss, measured at 600"C, 73.7% of the total Zr(S04)2added, is in agreement with eqn. (1). It seems that chloride ions are not involved in the reaction. However, remembering that, as explained above, reactions in molten nitrates proceed accord-ing to a dissolution-precipitation mechanism, the fact that the reaction goes to completion in the presence of chloride ions could be interpreted as an effect of the complexing efficiency of Cl-ions promoting the dissolution of Zr(S04)2.As in pure sodium nitrate, XRD indicated the formation of a mixture of tetragonal and monoclinic zirconia [Fig. 2(b), Table 11. The increase in the basicity of the molten medium, by addition of sodium carbonate to sodium nitrate (0.2 mol I' I 28 29 30 31 26ldegrees Fig. 2 X-Ray diffractogram of zirconia obtained from reactions of anhydrous zirconium sulfate in (a)NaNO,; (b)NaNO, +NaCl; and (c) NaNO, +Na,CO, J. MATER. CHEM., 1994, VOL. 4 Na2C03 per mol NaNO,) brought a similar modification to the differential thermal analysis curve as the addition of sodium chloride (Fig.I). XRD of the extracted powder showed the formation of tetragonal zirconia [Fig. 2(c) Table 11,which could be explained by the following reaction: Zr( SO,), +2N0, -+CO,,--+ ZrO, +2S04,-+2N0, +CO, +40, (2) although the calculated mass loss, 50.8% of the total Zr(S04), added, is in total disagreement with the observed loss, 89.3%. Such a significant loss can be understood if it is assumed that the high basicity of carbonate ions involves the simultaneous formation of zirconia and zirconate ions, Zr03,-, according to the following reaction, giving a calculated value, 84.0%, closer to the experimental value: 2Zr(SO,), +8N03-+CO,'--+ ZrO, +z1-0,~-+4So,,-+8N02+CO, +20, (3) Poorly crystallised Na,ZrO, was identified by XRD (Fig. 3).When sodium chloride or sodium carbonate were added to sodium nitrate, a small mass loss was noticed when the temperature rose above 600°C (Fig. 1). Such a loss was previously observed by Kerridge and Cancela-Rey13 in reac- tions of zirconium sulfate with an LiN03-KN03 eutectic and was attributed to the formation of anionic zirconate. These small losses were not taken into account in the values pre- viously given for the reactions (73.7 and 89.3%). In the same way, the small mass losses at ca. 100"C,due to the elimination of water absorbed during the mixing of reactants, were not considered. Reaction of Zirconium Sulfate, Anhydrous or Hydrated, with an LiN03-KN03 Eutectic Anhydrous Sulfate TG of the reaction of anhydrous zirconium sulfate (0.2 mol kg- ') with LiN0,-KNO, eutectic (mp 132 "C)corrobor-ated Kerridge's and Cancela-Rey's previous re~u1ts.l~It showed a reaction taking place in several stages, beginning above the melting point and finishing at ca.500 "C (Fig. 4). The experimental mass loss, 75.2% of the total Zr(SO,),! added, is in good agreement with that calculated for eqn. (l), 76.3%. The XRD pattern of the extracted insoluble solid showed a mixture of monoclinic and tetragonal zirconia [Fig. 5(a)]. Larger-scale experiments (20 g eutectic heated in a furnace) with the same concentration of zirconium sulfate (0.2 rnol kg-I) showed similar reactions.Analysis of the extracted insoluble reaction products indicated the presence of detectable nitrate, though potassium and lithium were absent. This suggested that complete reaction to zirconia had not always occurred and this was supported by the somewhat !-l----r-~-T~ ' ' ' I.. .-1.7 ' 7 1 -' ' TI ' T' ' I ' ' ' J 25 30 35 40 45 50 55 60 2Bldegrees Fig. 3 X-Ray diffraction pattern of zirconia obtained from reaction of anhydrous zirconium sulfate in NaN03 +Na,CO,: *, Na,ZrO, TPC Fig.4 TG of the reactions of the zirconium sulfate in lithium nitrate-potassium nitrate eutectic. Influence of hydration: ( I,anhy-drous zirconium sulfate; A,tetrahydrated zirconium sulfate. I. 1 28 29 30 31 28ldegrees Fig. 5 X-Ray diffractogram of zirconia obtained from reactions of anhydrous zirconium sulfate in: (a) LiN0,-KNO,; and (b) in NaN02-KN0, gelatinous nature of the precipitate, which diminished as the eutectic melt was heated to higher temperatures and/or for longer times.It was also supported by the mass of dry precipitate, which was greater than that calculated for the mass of pure zirconia. The percentage excess also decreased with temperature/time (Fig. 6), while X-ray powder diffraction timelh Fig. 6 Temperature-time profiles for reactions of anhydrcus zir-conium sulfate in LiN03-KN03 and weight percentages of rec overed precipitate. (a) LiN0,-KN03+0.20 mol kg-I Zr(SO,),, -I 10.8%; (b) LiN0,-KN03+ 0.25 rnol kg-' Zr(SO,),, -108.9%; md (c) LiNO,-KNO, +0.20 mol kg-' Zr(S04)2,-98.9%.showed the presence of both monoclinic and tetragonal zir- conia together with amorphous material in all the samples. Tetrahydrated Sulfate When the tetrahydrate was heated in the nitrate eutectic, white fumes containing nitric acid were evolved and a white suspension was produced. TG (Fig. 4) showed the mass loss to begin from the lower temperature of llO"C, as expected for a hydrolysis reaction. The first stage of mass loss, over the temperature range 110-330 "C was due to overlapping reac- tions. The second stage was similar to that of anhydrous zirconium sulfate. The overall loss, 88.4% of the total Zr(S04),.4H,0 added, was rather higher than the calculated value, 81.1%, for the reaction: Zr(S04)2~4H20+4N03--+ ZrO, +2S042-+4H,O +4N02+o2 (4) The difference is undoubtedly due to the loss of some nitrate as nitric acid.Larger-scale experiments led to results similar to those for anhydrous zirconium sulfate. Reaction of Anhydrous Zirconium Sulfate in Molten Nitrates in the Presence of Nitrite and in a Molten Nitrite Eutectic Addition ofKN02 to the Eutectic LiN0,-KNO, Since longer reaction times or higher temperatures are undesir- able industrially, a change was made to more basic and hence more reactive melts, initially by introducing potassium nitrite in the LiN0,-KNO, eutectic. The nitrite has a dissociation constant, to form oxide anions, some 10" larger than that of nitrate under equivalent conditions. TG (Fig. 7) showed a two-stage process with the major mass loss occurring during the first stage in the temperature range 150-270°C.The second stage, which gave a small mass loss, ended at ca. 370 "C.The overall experimental mass loss, 53.3% of the total Zr(S04)2 added, was close to the calculated value, 53.7%0, for the equation: Zr(S04),+4N0,- +ZrO, +2S04,-+2N0, +2N0 (5) The washed precipitates, from furnace reactions performed at 450 or 5OO0C, again contained some nitrate and possibly lithium and also exhibited a dry mass slightly above that for pure zirconia [Fig. 8 (a)-(c)].XRD patterns gave maxima at the positions for monoclinic and tetragonal zirconia. In fur- nace reactions carried out at 200 "C [Fig. 8(d),(e)], longer durations failed to compensate for the low temperature and low masses of precipitate were recovered; this supports the idea that in the two-stage process only the first stage is involved at low temperature.200 300 400 500 TI'C Fig. 7 TG of the reactions of the anhydrous zirconium sulfate in: A, nitrate/nitrite and 0,pure nitrite melts J. MATER. CHEM., 1994, VOL. 4 0 2468 time/h Fig. 8 Temperature-time profiles for reactions of anhydrous Zr( SO,), in LiN0,-KNO, in the presence of KN02 and weight percentages of recovered precipitate. (a) 0.19 mol kg-' Zr(S04),+0.15 mol kg-' KN02-106.1%; (b)0.14 rnol kg-I Zr(S04),+0.98 rnol kg-' KNO,, -103.8%; (c) 0.19 rnol kg-' Zr(S04),+0.99 mol kg-I KNO,, -104.8%; (d) 0.20 rnol kg- ' Zr(SO,), +0.99 mol kg- KNO,, -62.8%; and (e) 0.20 mol kg-' Zr(SO,),$ 1.00 rnol kg-'KNO,, -67.2%.Reaction in Molten NaN02-KN02 eutectic To increase the basicity of the melt further the LiN0,-KNO, eutectic (mp 132 "C) was replaced by NaN02-KN02 eutectic (mp 220°C). TG (Fig. 7) showed a one-stage reaction with a mass loss maximum almost at the melting point and no loss above 270°C. The measured overall loss was 42.2% of the total Zr(S04), added. In order to match this observed value to a possible stoichiometric equation, it was necessary to dissociate nitrite ions into N203 and 02-according to eqn. (6) and to take account of the known reaction" of N203 to produce nitrogen dioxide and nitrogen monoxide according to eqn. (7), with further reaction of nitrogen dioxide with nitrite ions according to eqn.(8). 2N02- +N203 +0'-(6) N203+N02+N0 (7) NO2+NO2-+NO,-+NO (8) Thus the stoichiometric eqn. (9) was obtained, giving a calculated weight loss of 42.4% Zr(SO,), +6N02- +Zr02 +2S04'-+2NO,-+4NO (9) The XRD pattern of the extracted powder identified tetragonal zirconia [Fig. 5 (b)]. Some Characteristics of Zirconia Powders obtained from Reactions of Anhydrous Zirconium Sulfate with Molten Nitrate and Nitrite Eutectics Some of the zirconia powders were more intensely studied, particularly four samples prepared from anhydrous zirconium sulfate in furnace reactions at 450 or 300°C; the reaction temperature was reached at a rate of 150°C h-' and main- tained for 90 min. The nature of the molten salt and the reaction temperature are indicated together with the results of chemical analysis in Table 2.For samples A and B, prepared in the nitrate medium, the presence of a significant amount of remaining sulfate indicates that reaction is not complete. This is corroborated by XRD patterns exhibiting lines of tetragonal and monoclinic zirconia and also the strongest line of zirconium hydroxysulfate, ZI-~.~~(SO~)(OH),at ca. 7.5" 28 (JCPDS file no. 41 0694) (Fig. 9). For sample A prepared in LiN03-KN03 the forma- tion of tetragonal zirconia predominates, whereas for sample J. MATER. CHEM., 1994, VOL. 4 Table 2 Chemical analysis of zirconia obtained from anhydrous zirconium sulfate by reaction in nitrate and nitrate molten media reaction chemical analysis (wt.%)temperature/ sample molten medium "C Zr so4 N03/NO," LilNa" K ~~ A LiNO3-KNO3 450 57.8 4.10 0.29 0.03 0.05 B NaN0,-KNO, 450 60.7 5.50 0.16 0.57 0.16 C NaN0,-KNO, 450 61.9 0.01 0.17 0.82 0.05 D NaN0,-KNO, 300 53.1 0.01 1.67 0.13 0.13 "NO, for samples A and B, NO, for samples C and D."Li for sample A, Na for samples B, C and D. R ' ic, I-II 10 20 30 40 50 60 70 2O/degrees Fig. 9 X-Ray patterns of zirconia samples A@), B(b), C(c) and D(d) (see Table 2) B prepared in NaN0,-KNO, the formation of monoclinic zirconia is mainly observed, For samples C and D prepared in the nitrite media, the transformation of zirconium sulfate into zirconia is complete. The XRD patterns (Fig. 9) reveal the formation of pure tetragonal zirconia at 450°C and the formation of a mixture of amorphous and poorly crystallised zirconia at 300 "C. When annealed at 500 "C, the latter powder crystallises as tetragonal zirconia and a quantity of nitrates greater than for sample C is simultaneously eliminated.An electron micrograph of tetragonal zirconia obtained at 450°C in the nitrite medium [Fig. lO(u)] shows that the powder is made of soft agglomerates containing nearly spheri- cal elementary grains with diameters close to 6 nm. From the size of crystallites determined by X-ray broadening, 5 nm, it is concluded that the elementary grains are monocrystalline. This powder exhibits a specific surface area >150 m2 g-'. An electron micrograph of sample D [Fig. lo@)] reveals the presence of agglomerates formed of smaller elementary grains Fig.10 Transmission electron micrographs of zirconia powders prepared from reactions of zirconium anhydrous sulphate with molten alkali nitrites. (a) Sample C prepared at 450°C (30nm= 1.05cm). (h) Sample D prepared at 300 "C (30 nm =1.45 cm). with diameters close to 3 nm. The amorphous and tetragonal phases cannot be distinguished. The specific surface area is close to 250m2 g-'. Conclusion Differential thermal analysis investigation of the reaction of zirconium sulfate, either hydrated or anhydrous, with molten alkali-metal nitrates showed that the reactions occurred in at least two stages, each indicated by the release of red -brown nitrogen dioxide. The first step leads probably to a soluble zirconium nitrate or oxynitrate, the second to insoluble zir- conia.In pure sodium nitrate (mp 308"C), the reaction remained incomplete even at temperatures approaching 500 "C. The addition of sodium chloride to the molten medium gave a complete reaction at ca. 440"C. The increase of hasicity by the addition of sodium carbonate also favours the reactions, but sodium zirconate is formed simultaneously with zirconia. The decrease of the melting point of the molten salt, by replacing sodium nitrate by the LiN0,-KNO, eutectic (mp 132 "C), allowed complete transformation to occur as low as 500 "C. When zirconium sulfate tetrahydrate is used. white fumes of nitric acid are evolved at the beginning of the reaction. An increase in basicity by addition of potassium nitrite to the nitrate eutectic or by performing the reaction in the eutectic NaN0,-KN02, shifted the reactions towards lower temperatures.In the latter case, thermal analysis shows only one stage occurring under 300 "C. Reactions of zirconium sulfate with the molten medium led either to monoclinic or tetragonal zirconia. When this latter phase is obtained, it irreversibly transforms into monoclinic by annealing. However, it is much more interesting to synthe- size the tetragonal variety directly. Accordingly, among the molten media studied, the eutectic NaN0,-KN02 appears to be the most suitable for obtaining fine zirconia powders of a satisfactory purity and with large specific surface arcas, as shown by the characterisation of powders prepared at 450 "C for 90 min with anhydrous zirconium sulfate.Such pc )wders are likely to find applications as catalyst supports in hcterog- enous catalysis and as ceramic precursors on the condition that the method may be extended to the preparation of stabilised zirconia. These results were obtained in the frame of a twinning contract between Imperial College and the University of Lyon. The authors are indebted to the Commission of European Communities for financial support. References 1 B. Y. Lee, Y. home and I. Yasumori, Bull. Chem. SOC.Jpn., 1981, 54, 13. 2 T. Lizuka, Y. Tanaka and K. Tanabe, J. Card., 1981,76,1. 3 D. Hamon, M. Vrinat, M. Breysse, M. Jebrouni, B. Durand, M. Roubin and P.Magnoux, Catal. Today, 1991,10,613. 4 K. S. Mazdiyasni, Ceram.Int., 1982,8 (I),42. 1336 J. MATER. CHEM., 1994, VOL. 4 5 J. L. Burke, N. L. Reed and V. Weiss, Ultra-jine-Grain Ceramics, 13 D. H. Kerridge and J. Cancela-Rey, J. Inorg. Nucl. Chem., 1977, Syracuse University Press, Syracuse, 1970. 39,405. 6 M. J. Bannister and W. G. Garrett, Ceram Int., 1975, 1 (3), 127. 14 H. Lux, 2.Elektrochem., 1939,45,303. 7 B. Durand, Ceramic Powders: Preparation, Consolidation and 15 G. Charlot and B. Tremillon, Les Riactions Chimiques dans les Sintering, ed. P. Vincenzini, Elsevier, Amsterdam, 1983, p. 413. Solvants et les Sels Fondus, ed. Gauthier-Villars, 1963, ch. XV, 8 B. Durand and M. Roubin, Mater. Sci. Forum, 1991,73-75,663. p. 486. 9 M. Jebrouni, B. Durand and M. Roubin, Ann. Chim.-Sci. Muter., 16 F. R. Duke and S. Yamamoto, J. Am. Ceram. So(., 1959,81,6378.1991, 16, 569. 17 I. J. Bear, Aust. J. Chem., 1966, 19, 357. 10 M. Jebrouni, B. Durand and M. Roubin, Ann. Chim.-Sci. Muter., 18 N. N. Greenwood and A. Earnshaw, in Chemistr,vof the Elements, 1992, 17, 143. Pergamon Press, Oxford, 1984, p. 521. 11 B. Durand, D. de Mareuil, M. Vrinat and T. des Courieres, Brevet Elf Aquitaine no. 900549. 12 M. Descemond, M. Jebrouni, B. Durand, M. Roubin, C. Brodhag Paper 3/07391B; Received 15th December, 1993and F. Thevenot, J. Muter. Sci., 1993,28, 2283.