Oxidation of alkaline-earth-metal sulfide powders and thin films Janos Madarasz,?Tuula Leskela, Janne Rautanen and Lauri Niinisto" Laboratory of Inorganic and Analytical Chemistry, Helsinki University of Technology, FIN-021 50 Espoo, Finland The oxidation of CaS and SrS powders and thin films was studied in situ up to 1300 "C by thermoanalytical techniques (TG, DTA and high-temperature XRD) while FTIR and powder XRD were used to analyse ex situ the solid reaction intermediates and products. CaS powder starts to oxidize to CaSO, around 500 "C but the oxidation is not complete because of a competing reaction, i.e. the decomposition of CaSO, to CaO, which is significant above 1000"C. SrS0, is more stable and therefore SrS completely oxidizes to SrSO, before the decomposition starts.FTIR and XRD failed to detect CaSO, or SrS0, at any stage of the heating process. Particle size appears to have a marked effect on the decomposition of powders, while moisture plays only a minor role. Thin films appear to be more stable towards oxidation but they react around 1000 "C with the silicon substrate. Alkaline-earth-metal sulfides, especially the isostructural CaS and SrS, are important matrix materials for electroluminescent (EL) thin films used in flat-panel full-colour displays. When these sulfide layers are doped with rare-earth-metal ions, e.g. trivalent Eu, Tb or Ce, they emit in red, green and green-blue, re~pectively.'-~ Although the emissive layer in an electrolumi- nescent structure is well protected from the atmosphere, there is a risk that at least its surface may become oxidized or degraded during deposition or annealing.Decreasing the extent of oxidation and degradation would enhance the luminescence efficiency and stability of these devices. While systematic studies on oxidation and degradation of alkaline-earth-metal sulfide thin films are lacking, there are several reports on the effects of oxygen contamination. Okamoto and Hanaoka6 reported that the luminance of elec- tron-beam (EB)-deposited SrS :Ce3+ thin films was reduced owing to the presence of a significant amount of oxygen in form of SrC0, in the precursor pellet. Abe et aL7 have systematically examined the effects of oxygen partial pressure in the EB evaporation atmosphere and found out that an increase in the oxygen pressure led to a higher oxygen content in the CaS :Eu2+ films and simultaneously decreased the electroluminescent efficiency owing to the oxidation of some Eu2+ ions to Eu3+.A reductive atmosphere appears generally advantageous for the Eu2+ and Ce3+ dopants. Thus, for instance, H, gas applied during EB evaporation could improve up to 2.3 times the luminance of a SrS:Ce,Eu,K thin film.' Heat treatment of an SrS :Pr,Ce EB precursor pellet in a H,S atmosphere resulted in films showing luminances five times greater than was found for films without the heat treatment.g On the other hand, there are reports that oxygen in EL layers appears in some cases not to be detrimental. Park et a!." observed (by XPS) oxygen-related luminescence centres in a CaS :Bi3+ phosphor.Furthermore, they found that CaS :Eu and SrS :Eu phosphors can be stabilized against atmospheric moisture and carbon dioxide by annealing above 700 "C in air for several hours." According to their XPS study such a treatment results in the formation of CaSO, and a spurious 'strontium oxysulfide (SrSO)'. Based on the XPS spectra of the SrS :Pr,Ce powder, several sulfur species including elemen- tal S and SO3,-are suggested to appear on the surface owing to the oxidizing effect of moist air.g These suggestions do not seem to be well founded because, for instance, the chemical shifts of the S species show SO4,-formation.', Poelman et al.I3 correlated the additional peaks in the emission spectra of an t Permanent address: Institute of General and Analytical Chemistry, Technical University of Budapest, H-1521 Budapest, Hungary.SrS, -$ex :Ce layer to its SrSO, contamination. When oxygen is present the rare-earth-metal dopant ions in alkaline-earth- metal or zinc sulfide matrices may form (LnO),,"+ polycation clusters as shown by extended X-ray absorption fine structure (EXAFS) studies.14 The thermal oxidation processes of CaS are frequently studied above 500°C in order to investigate the complex desulfurization processes of gases obtained from coal gasifi- cation or from combustion using lime or limestone as absorb- ent~.'~,'~Some of the model components in such a system were recently studied by thermogravimetry (TG), differential thermal analysis (DTA) and X-ray powder diffraction (XRD).17 Chemical species formed in reactions of SO, with CaO and CaCO, have been monitored by Fourier-transform IR pho- toacoustic spectroscopy (FTIR-PAS)." The primary product found in the desulfurization process of combustion gases was CaSO, which, however, is thermodynamically unstable in comparison to CaSO,.As part of our ongoing research into the deposition and characterization of SrS and CaS thin films for EL appli- ~ations,'~-~'the present investigation was undertaken using a combination of thermal and diffraction methods as well as by employing FTIR spectroscopy. The main emphasis was on SrS, but CaS was studied also and the results compared with literature data.Experimenta1 Preparation of the samples SrS and CaS powders. SrS and CaS powders were prepared by two slightly different methods. In the first22 (Method A) we used two different strontium carbonates [from Schering A.-G. and Baker, containing 130 and 60 pg (g SrC03)-' barium, respectively] as starting materials for SrS while the starting material for CaS was calcium carbonate (from Merck). The carbonates (3-4 g) were weighed in A1,0, boats which were placed inside a ceramic tube in the tube furnace (Carbolite Furnaces CTF 1200). The temperature of the furnace was raised to 1000 "C and a stream of H2S (purity 99.999%) diluted with Ar-H, (5%) gas (flow rate 3-5ml s-') was passed through the furnace for 3 h. During cooling only the Ar-H, mixture was passed through the furnace.,, The other SrS and CaS samples (B) were prepared in CNRS, Talence, France.These samples were synthesized from the corresponding sulfates by reduction in an Ar-H, (10%) stream at 1000 "Cfor 15 h. After grinding the products were annealed at 1100°C under H2S for 1 h. J. Mater. Chem., 1996, 6(5), 781-787 781 SrS and CaS thin films. Sr(thd),, Ca(thd), (thd= 2,2,6,6,- tetramethylheptane-3,5-dione)and H,S were used as precursors for the deposition of SrS and CaS thin films by atomic layer epitaxy (ALE) ''23 The hot-wall, flow-type F-120 reactor, manufactured by Microchemistry Ltd (Espoo, Finland), was employed in the depositions Phosphorus-doped Si( 100) wafers were used as substrates The substrate and source temperatures were 360, 225 "C for SrS and 350, 197 "C for CaS, respectively The thicknesses of the SrS and CaS films were 620 and 630 nm, respectively, as evaluated from reflectance spectra measured using a Hitachi U 2000 spectrophotometer, and the program of Ylilammi and Ranta-aho 24 Thermal analysis TG measurements were carried out in a Seiko Instruments TG-DTA 320 analyser, equipped with a SSC/5200 disk station The heating rates were 2 and 10°C min-' A120, and Pt crucibles were employed and Al,O, was used as a reference material The experiments were carried out in flowing air using flow rates of 80 or 220 ml min-' The sample size was typically 20-30 mg Some experiments were also performed in moist air produced by passing air at a rate of 80ml min-' through a gas washing bottle filled with water IR, XRD and SEM measurements SrS and CaS powders and the products after heating in air were characterized by IR spectroscopy and X-ray diffraction IR spectra of the samples were recorded in the region 4000-400 cm -'with a Nicolet Magna-FTIR 750 spectrometer using the KBr disk technique X-Ray diffractograms were recorded with a PhilipsoMPD 1880 diffractometer using Cu-Ka radiation (A= 154060 A) An Anton Paar HTK 10 high-temperature goniometer attachment was used for zyt sztu X-ray diffraction (XRD) measurements during heating of CaS and SrS powders and thin films in air The powder samples were glued onto Pt wire and measured using a linear position-sensitive detector (PSD) while the thin films were measured with a normal proportional counter equipped with a graphite monochromator Cu-Ka was used as the X-ray source in both cases For powder samples the Pt(ll1) reflection was used as an internal standard Scanning electron microscopy (SEM) was used to check the particle size and its distribution in the CaS samples prepared by the two methods Results and Discussion Thermal analysis of powder samples Oxidation of SrS powders in air.The thermal behaviour of SrS powders in air is basically straightforward and proceeds as expected the oxidation to SrS0, begins slowly around 700"C, the reaction rate is increased at 800°C and the mass at 1300°C corresponds to SrSO, (Fig 1) Oxidation to SrSO, does not appear to take place under these conditions According to FTIR measurements the formation of SrSO, was clearly observed in the samples heated to 600 "C but the first indications of oxidation to sulfate were observed at 450 "C because in this case IR spectroscopy is a much more sensitive method than TG SrS synthesized from SrSO, (method B) was oxidized almost completely to sulfate by 1300°C when a heating rate of 10°C min-' was used The flow rate of air influenced the oxidation slightly (97 8% with 80 ml min-' and 99 7% with 220 ml min-') In the case of method A samples only 70% of the theoretical yield of SrS0, was achieved even at 1300°C The remaining SrS was unchanged according to XRD analysis (Fig 2) The conversion could be increased to 90% using the slower heating rate (2°C min-l) and smaller sample size (5 mg), but 100% conversion to sulfate could not 782 J Mater Chem, 1996, 6(5),781-787 160 I G 110 100 ex0 80 70 -tendo Fig.1 Simultaneously recorded TG and DTA curves for SrS (sample B) in flowing air (80 ml min-I), recorded with a heating rate of 10 "C min Initial mass 22 55 mg be reached The main reason for the different behaviour between the two SrS samples dunng heating in air is most probably the different particle size, which will be discussed in more detail below After SrSO, has been formed there is an additional change as the DTA curve reveals an endothermic transition at 1150°C without any change in the TG curve This reversible transition can be assigned to the orthorhombic- cubic phase transition of SrSO, 25 The oxidation in moist air does not deviate much from the oxidation in dry air, except that at temperatures above 1000"C the oxidation takes place somewhat quicker in moist air Therefore, in contrast to concern expressed in the literat~re,~ SrS phosphors appear not to be significantly moisture-sensitive Oxidation of CaS powders in air.The oxidation of CaS samples begins around 500 "C (Fig 3, 4) The oxidation rate in sample B is initially very slow, but it increases above 750 "C and is very fast up to 1050°C (Fig 3) According to the IR measurements the formation of CaSO, was clearly observed in the samples heated to 600°C but, also in this case, the first indications of oxidation to sulfate were observed at 450°C Again, no formation of sulfite could be seen (Fig 5) The theoretical yield of CaS0, (188 7% of the original mass) was not achieved by either sample, however The factors affecting the amount of CaSO, formed include the heating rate, sample amount and particle size When the heating rate was slow (2°C min-') the mass gain was greater (77%) compared to the faster heating rate (10 "C min-') which resulted in a mass gain of 62% According to thermodynamic data26 27 SrSO, is more stable at higher temperatures than CaSO, Theoretically it is not possible to completely oxidize CaS to CaSO, because the decomposition of CaSO, to CaO begins while the oxidation of CaS is still occuring The decomposition reaction cannot be seen from the TG curve of sample B (Fig 3) because the two competing reactions balance each other In the A sample a sudden mass loss between 1060 and 1115 "C followed by a rapid mass increase can be seen (Fig 4) According to XRD analysis the amount of CaO is small until 950"C, but when the sample is heated to 1085°C the product contains a considerable amount of CaO as well as CaSO, and the unreacted CaS (Fig 6) At 1200°C CaS could no longer be observed The main reason for the different thermal behaviour of the two CaS samples is the particle size, which in the A sample is >5 pm while in the B sample it is t2 pm as measured by SEM When CaS with a larger particle size (sample A) was ground in an agate mortar the sudden mass loss at 1085°C almost disappeared A higher conversion to sulfate was also 20.00 30.00 40.00 50.00 60.00 2e/degrees Fig.2 XRD patterns of SrS (sample A) at room temperature (a) and after heating up to 600 (b),950 (c), 1300 "C (d)at 10 "C min- rate in flowing air (80 ml min -') :e% i r140 130s!;120 g 110 100 90 80 endo J70' * ' ' ' 1 ' ' ' ' ' 0 200 LOO 600 800 1000 1200 1400 TI"C Fig.3 Simultaneous TG and DTA curves for CaS (sample B) in air (80ml min-'), recorded with a heating rate of 10°C min-'. Initial mass 22.04mg. 11 0 -105k? v) Y 1002 95 90 85 80 0 200 400 600 800 1000 1200 1400 TPC Fig.4 Simultaneous TG and DTA curves for CaS (sample A) in air (80ml min-'), recorded with a heating rate of 10°C min-'.Initial mass 18.56 mg. achieved by grinding the sample. An obvious explanation is that if the particle size is larger the diffusion of oxygen into the particles is slower and the CaSO, formed at the outer surface of CaS particles begins to decompose before the inner part has oxidized. It may also be possible that the CaSO, acts 21 0 220200 k677 I 592190 610 170 120 -110 -100 7 (b) -90 80 -70 -60 -50 -(a) 1800 1600 lL00 1200 1000 800 600 Fig.5 FTIR spectra of CaS (sample A) recorded in the range 1800-400 cm-' after heating the sample up to (a) 350, (b)600, (c) 950 and (d) 1300"C as a barrier for the oxygen diffusion and slows the oxidation.Grinding the powder provides a larger surface for oxidation and shortens the diffusion length needed to complete oxidation. However, even with small particle sizes a 100% conversion to sulfate cannot be achieved because of the instability of CaSO, at high temperatures. The DTA curve of CaS also reveals an endothermic trans- ition at 1220°C. This reversible transition is due to the orthorhombic-cubic phase transition of CaSO, .28 Our results can be compared with those presented recentlyI7 where it was found that the oxidation of CaS began, under the experimental conditions used, at 500"C and the maximum J. Mater. Chem., 1996, 6(5),781-787 783 C8S (220 o',,,,,,,,, .,,,,,,,, ,,",",, ,,.,,,,,,10 d0 40 $0 I 0 2@/degrees Fig.6 XRD pattern of CaS (sample A) cooled from 1085 "C to room temperature The unassigned reflections belong to CaSO, mass was reached around 950°C The total mass increase was only 7%, which can be explained by a small gas flow rate (15 ml min-l) and a large sample size (42 mg) The oxidation was a two-step process and a possible explanation, according to the author, was that CaS oxidizes first to CaSO, which then oxidizes to CaSO, The existence of CaSO, was not proven by any analysis, however According to the ex sztu XRD analysis the different phases present at 1100°C were CaSO,, CaO and CaS The oxidation in moist air is very similar to the oxidation in dry air initially, but at temperatures above 850°C the oxidation rate is higher in moist air The total mass gain is also ca 10% higher in moist air Although the differences in the observed oxidation behaviour for CaS in dry and moist air are small, they are clearly observable below 1000 "C whereas a similar effect was not seen for SrS This may be connected to the higher oxidation tendency of CaS compared with SrS High-temperature XRD of powders and thin films Mainly because of the small amount of sample, thin films cannot normally be studied using the conventional thermo- analytical techniques such as TG and DTA 29 High-tempera-ture X-ray diffraction (HTXRD) is one of the few methods available for zn sztu monitoring of the thermally induced changes occurring in thin films3' One of the problems in HTXRD studies is sample alignment, which may lead to erroneous peak positions Also, the linear PSD detector causes errors to the peak positions near the edges of the measuring windows A further source of error is the sample temperature, which may significantly deviate from the programmed one if the conduction of heat from the sample through the substrate to the thermocouple is not good32 In this work the errors in the peak positions were minimized by using the Pt(ll1) reflection and the silicon substrate reflections as internal standards for the powder and thin film samples, respectively The temperature gradients between the sample and the Pt heating element were minimized by waiting 10min at each measuring temperature before performing the XRD measure- ment Also the heating rate between each temperature was relatively slow A fourth, more serious problem in the HTXRD study of thin films is caused by the reaction of the substrates with the 784 J Muter Chem, 1996, 6(5),781-787 thin film For silicon substrate this takes place at higher temperatures and may lead to a number of silicon-containing crystalline phases which are sometimes hard to identify Fig 7 displays the XRD pattern of an SrS powder recorded zn sztu at temperatures from 30 to lOOO"C, and Fig 8 displays the XRD pattern of CaS powder recorded zn sztu at tempera- tures from 30 to 1100 "C According to the HTXRD measure- ment, the SrS powder began to oxidize at 750 "C The oxidation product was SrSO, but the SrS phase was still present after heating the powder to 1000 "C 33 The CaS powder also started to oxidize at 750°C and the product was initially CaSO, but at 1000°C the formation of CaO started because of the decomposition of CaSO, 34 The CaS peaks disappeared between 1000 and 1100°C The intensity of the CaSO, peaks decreased after 1000 "C while the intensities of the reflections due to CaO increased Fig 9 shows the zn sztu HTXRD results for an SrS thin film sample from room temperature to 1000°C It appears that the (111) oriented SrS thin films are thermally very stable because the intensity of the SrS reflections begins to decrease only above 800 "C At 1000 "C a considerable amount of SrS is still left together with SrSO, and the silicon-containing phases SrSiO, and Sr,SiO, (Fig 9) 35 Prolonged heating (15 h) at 1000°C causes the SrS phase to disappear, however Fig 10 shows the zn sztu HTXRD results for a CaS thin film sample heated from room temperature up to 1000°C The pattern at room temperature displays a typical mixed orien- tation with (lll), (200), (220) and (222) as the most dominant reflections l9 The first signs of thermal reactions are visible at 600°C where, for instance, the (111) and (200) peaks of CaS have a diminished intensity as compared to a sample recorded at room temperature At 800°C CaS is still present but part of it has been converted to CaSO, At the same time CaSO, has partly decomposed to CaO The most prominent peaks at 1000"C are due to silicon-containing phases, however, showing that oxygen diffuses into the substrate and the SiO, formed reacts with CaO Indeed, the presence of CaSiO, can be concluded from the XRD pattern (Fig 10) when comparing it to the JCPDS data 36 A difference to the SrS thin film pattern at the same temperature is the absence of the 2 1 phase (Ca2Si0,), instead the (111) and (222) reflections of SiO, (cri~tobahte)~~are relatively strong Although the SiO, detected 50 62 00 22.00 26.00 30.00 34.00 2Wdegrees Fig.7 SrS powder (sample B) measured in situ by high-temperature XRD at (a) 30, (b)500, (c) 750 and (d) 1OOO"C in air using linear PSD. The measuring time for each spectrum was 500 s. Intensities are in root scale. -00N Y 1 .60 3 L.60 I I I 40 (f 1 0.ool I I I I I I I e00 23.00 27.00 31 .OO 35.00 26,/degrees Fig.8 CaS powder (sample B) measured in situ by HTXRD at (a) 30, (b)500, (c) 750, (d) 900, (e) 1000 and (f) 1100"C in air using linear PSD. The measuring time for each spectrum was 500 s. Intensities are in root scale. is a high-temperature phase (>1200"C), its presence at 1000"C be seen in the TG curves. The sample characteristics, uiz. its can be explained by the effect of calcium ions which lower the synthesis and especially the particle size, have a significant transition temperature. 37 The peak at 28= 17.0" could not be effect on the reaction temperatures and mechanism. This is indexed with certainty; it probably belongs to a transient phase especially obvious for CaS where the formation of CaSO, and because it almost disappeared when the sample was kept at its decomposition to CaO are competing reactions.1000°C for 1.5 h. The alkaline-earth-metal sulfide thin films, in particular the (111) oriented SrS, appear to be remarkably resistant towards Conclusions oxidation which is a promising sign in view of their processing and for possible application in EL devices. The reaction with The oxidation of both CaS and SrS powders in air leads to the substrate observed by high-temperature XRD at 1000"C the formation of the sulfate phase without the sulfite intermedi- is not a problem in real applications where the deposition and ate. The first signs of oxidation are observed in the IR spectra annealing temperatures used are much lower, for instance in at 450 "C, but only at higher temperatures can a mass increase the atomic layer epitaxy (ALE)technique around 400-500 0C.38 J.Muter. Chem., 1996, 6(5),781-787 785 40 10 00 20 00 24 00 28 00 32 00 36 00 40 00 2Bldegrees Fig. 9 SrS thin film sample measured zn situ by HTXRD at (a) 28, (b) 600, (c) 800 and (d) 1000 "C using normal detector The data collection time per spectrum was 1 25 h and the spectrum was collected in the range 20 =20-60" a =SrSO,, b =Sr2Si04,c =SrSiO, and d =unidentified peak Si(200) is a forbidden reflection of Si Intensities are in root scale '1 nn zoo e75 cc C ccc00I e 00I I I I I I I I 15 00 25 .OO 3s 00 45 00 55 00 2Bldegrees Fig. 10 CaS thin film sample measured zn situ by HTXRD at (a) 28, (b) 600, (c) 800 and (d) 1000°C using normal detector The data collection time per spectrum was 1 25 h and the spectrum was collected in the range 20 =20-60" a =CaO, b=CaSO,, c =CaSiO, and d =unidentified peak (see text) Intensities are in root scale Professor C Foaussier (CNRS, Talence) is gratefully acknowl- 2 M Leskela and L Niinisto, Mater Chem Phys 1992,31,7 edged for providing the CaS and SrS powder samples The 4 M Leskela, M Makela, E Nykanen and M Tammenmaa, Chemtronics 1988,3, 113 authors are also indebted to Ms E Nykanen for growing the 4 M Leskela, L Niinisto, E Nykanen, P Soininen and M Tiitta, SrS thin film sample and to Dr M Ritala for the SEM J Less-Common Met 1989,153,219 measurements An Eotvos scholarship from the State of 5 M Leppanen, M Leskela, L Niinisto, E Nykanen, P Soininen Hungary and partial support from the National Scientific and M Tiitta, SID91 Digest 1991,282 Research Foundation (OTKA, Budapest, Grant No F014518) 6 K Okamoto and K Hanaoka, Jpn J Appl Phys 1988,27, L1923 to Janos Madarasz are gratefully acknowledged 7 Y Abe, K Onisawa, Y A Ono and M Hanazono, Jpn J Appl Phys 1990,29,1495 8 Q Z Gao, J Mita, T Tsuruoka, M Kobayashi and K Kawamura, J Crystal Growth 1992,117,983 References 9 K Onisawa, Y Abe, K Tamura, T Nakayama, M Hanazono and Y A Ono, J Electrochem SOC 1991,138,599 1 S Tanaka, H Deguchi, Y Mikami, M Shiiki and H Kobayashi, 10 H L Park, H K Kim and C H Chung, Solid State Commun SID86 Digest 1986,28 1988,66,867 786 J Muter Chern, 1996, 6(5), 781-787 M Han, S-J Oh, J H Park and H L Park, J Appl Phys, 1993, 73,4546 24 25 M Ylilammi and T Ranta-aho, Thin Solid Films, 1993,232, 56 F D Rossini, D D Wagman, W H Evans, S Levine and I Jaffe, T A Carlson, Photoelectron and Auger Spectroscopy, Plenum Circ Bur Stand, 1952, No 500,785 Press, New York and London, 1975, pp 200,205,356 D Poelman, R Vercaemst, R L Van Meirhaeghe, W H Lafrere and F Cardon, Jpn J Appl Phys Part 1, 1993,32,3477 26 27 I Barin, Thermochemical Data of Pure Substances, VCH Verlag, Weinheim, 1989 M Fredriksson and E Rosen, Chem Scr , 1980,16,34 15 16 17 18 19 20 21 22 Y Charreire, 0 Tolonen-Kivimaki, E Nykanen, M Leskela, L Niinisto, D Bonnin and 0 Heckmann, Thin Solid Films, 1994, 247, 151, and references therein K Schwerdtfeger and I Barin, Erdol Kohle, Erdgas, Petrochem, 1993,46,103 P F B Hansen, K Dam-Johansen and K Oestergaard, Chem Eng Sci, 1993,48,1325 K Wieczorek-Ciurowa, J Therm Anal, 1992,38,523 M A Martin, J W Childers and R A Palmer, Appl Spectrosc, 1987,41,120 J Rautanen, M Leskela, L Niinisto, E Nykanen, P Soininen and M Utriainen, Appl Surf Sci, 1994,82/83, 553 P Soininen, L Niinisto, E Nykanen and M Leskela, Appl Surf Sci , 1994,75,99 S Lehto, P Soininen, L Niinisto, J Likonen and R Lappalainen, Analyst, 1994,119, 1725 J W Brightwell, B Ray and C N Buckley, J Crystal Growth, 1982,59,210 28 29 30 31 32 33 34 35 36 37 38 0 W Florke, Naturwissenschaften, 1952,39,478 M Leskela, T Leskela and L Niinisto, J Therm Anal, 1993, 40,1077 D D L Chung, P W DeHaven, H Arnold and D Ghosh, X-Ray Diflractzon at Elevated Temperatures-A Method for In Situ Process Analysis, VCH, New York, 1993 Shinichi Ohya and Yasuo Yoshioka, Adv X-Ray Anal 1990, 33,397 N E Brown, S M Swapp, C L Bennett and A Navrotsky, J Appl Crystallogr ,1993,26,77 JCPDS cards 8-489,5-593 JCPDS cards 8-464,37-1496,4-777 JCPDS cards 34-99,39-1256 JCPDS cards 3 1-300,27-605 F Liebau, Structural Chemistry of Silicates, Structure, Bonding and Classrjication, Springer-Verlag, Berlin, 1985 L Niinisto and M Leskela, Appl Surf Scz , 1994,82/83,454 23 M Tammenmaa, H Antson, M Asplund, L Hiltunen, M Leskela, L Niinisto and E Ristolainen, J Crystal Growth, 1987,84, 151 Paper 5/06493G, Recieved 2nd October, 1995 J.Muter Chem, 1996, 6(5),781-787 787