Mixed oxides Si02-Zr02 and Si02-Ti02 by a non-hydrolytic sol-gel route Mahandrimanana Andrianainarivelo, Robert Corriu, Dominique Leclercq, P. Hubert Mutin" and AndrC Vioux Laboratoire des Prkcurseurs Molkculaires de Matkriaux, UMR CNRS 44, Universitk Montpellier 2, case 007, Place E.Bataillon, 34095 Montpellier Ckdex 5, France SO,-ZrO, and SiOz-TiOz mixed oxides with various metal contents have been prepared by a non-hydrolytic sol-gel route involving the condensation between chloride and isopropoxide functions at 110"C. Well condensed, monolithic gels were obtained in one step, without the use of additives. The Si/M ratio of the oxide may be controlled easily by the composition of the starting mixture. The Si/Zr oxides remained amorphous after calcination for 5 h at 600 "C; IR and 29Si NMR spectroscopy showed a large amount of Si-0 -Zr bonds, indicating a homogeneous distribution of the components on the atomic scale.The crystallization of tetragonal zirconia took place at higher temperature; the transformation of tetragonal to monoclinic zirconia was strongly retarded and did not take place after 2 h at 1300"C. The crystallization of zircon (for a sample containing 50 mol% Zr) started at 1500"C and was completed after 20 h at 1500"C. IR spectroscopy indicated the presence of a limited number of Si-0-Ti bonds in all the Si/Ti oxides after calcination for 5 h at 500 "C. The sample within the stable glass region (5 mol% Ti) appeared perfectly homogeneous: it crystallized at 900 "C as single-phase cristobalite oxide, with Ti4+ ions substituting Si4+ ions at random.On the other hand, the precipitation of anatase was observed for the %/Ti oxides with a high Ti content (20-50 mol% Ti), which are outside the stable glass region. The transformation of anatase to rutile was not observed even after 2 h at 1300"C. The technological interest in silica-titania and silica-zirconia mixed oxides arises from their chemical resistance and their thermomechanical or optical properties: Si0,-TiO, glasses and zircon, SiZrO,, are characterized by very low thermal expansion, which confer them a high thermal shock resistance; Si0,-TiO, and SiO,-ZrO, glasses have high refractive indices, and they are also of great importance as catalysts or as cata!yst supports.Owing to their refractoriness, these oxides are difficult to produce by conventional melting techniques. Accordingly, the sol-gel processing of alkoxides,' which allows the prep- aration of glasses at low temperature, has been used widely for the preparation of Si02-Ti022-7 and Si0,-Zr028-10 glasses or zircon."-'4 The homogeneity of binary oxide gels has a great influence on the structural evolution of the gels during the heat treat- ment. However, obtaining homogeneous Si0,-Ti02 or Si0,-ZrO, gels by hydrolysis/condensation is not straightfor- ward, owing to the difference in reactivity between silicon alkoxides and transition-metal alkoxides, leading to the fast formation of M-0-M bonds and to the precipitation of the metal oxide. To overcome this limitation several procedures have been elaborated, such as the pre-hydrolysis of the less reactive silicon alk~xide~.~.' or the stabilization of the metal alkoxide by comple~ation.~~'~ The aim of these procedures is to promote the formation of mixed Si-0-M bonds, i.e.homogeneity on the atomic scale. In recent years, we have been developing a new sol-gel route, based on non-hydrolytic reaction^.'^-'^ In this paper we wished to investigate the application of these reactions to the preparation of SiO,-ZrO, and SiOz-TiOz mixed oxides with various metal contents. We used IR and 29Si solid-state NMR spectroscopy to evidence mixed Si-0-M bonds. The crys- tallization behaviour of the mixed oxides as a function of the metal content and the calcination temperature was studied using powder X-ray diffraction.Background In conventional sol-gel processes, the formation of sols or gels results from the formation of M-0-M bridges through hydrolysis and polycondensation reactions. In the non-hydrolytic sol-gel process reported here, M -0-M bridges are obtained by condensation between halide and alkoxide groups, with elimination of alkyl halide: M-X+M-OR-M-0-M+R-X In most cases these reactions are thermally activated and, depending on the reactives involved, temperatures ranging from room temperature to about 100"C are required. Actually, the condensation competes with the redistribution of the ligands, which usually takes place rapidly at room temperature and leads to a complicated mixture of halogenoalkoxides:20-22 MX,+M(OR),+MX,(OR),-, with Odxdn To avoid the use of alkoxides, which are often quite expensive, the alkoxide groups may be formed in situ by the etherolysis of the metal halide by an organic ether (such as diethyl or diisopropyl ether): l6,l9 M-X+R-OR-M-OR+R-X The stoichiometry of these reactions requires an equal number of alkyl groups (borne by the alkoxide or by the ether) and halide groups.Thus, the idealized equations of reaction corre- sponding to the preparation of mixed oxide gels with different M/Si ratios are: halide-alkoxide route: x MX4 + y M(OR), + z M'X, + t M'(OR), -, M(x+y)M)1=+t~02(x+y+z+t)+4(X+Z) R-X with x+z=y+t, M=Si, M'=Ti or Zr halide-ther route: x MX4+y M'X4+2(x+y) ROR+M,M'yO~(,+y)+ 4(x +y)R-X with M = Si, M' =Ti or Zr Experiment a1 Si/Zr and Si/Ti gels with varied compositions were prepared from chlorides and isopropoxides (halide-alkoxide route) or chloride and diisopropyl ether (halide-ther route) (Table 1). J.Mater.Chern.,1996, 6(lo), 1665-1671 1665 Table 1 Preparation of the gels sample moles of reactantsa moles of solvent 1 Si/ZrA 1 ZrCl,+ 1 Si(OPr'), 10 CH,Cl, 1Si/ZrB 1 ZrCl, +1 SiCl, +4Pr,'0 10 CH,Cl, 5Si/Zr 1 ZrC1, +2 SiC1, +3 Si(OPr'), 4 CHZC1, 10Si/Zr 1 ZrC1, +4.5 SiCI, + 5.5 Si(OPr'), 30 CH,C1, 99Si/Zr 49 SiC1, +50 Si(OPr'), +1 ZrC1, 110 CH,Cl, 1 Si/TiA 1 Ti(OPf),+l SiC1, 2 CHCI, 1 Si/TiB 1 TiC1, +.l SiC1, +4Pr,'0 10 CHC1, -4Si/Ti 1 Ti(OPf),+ 2.5 SiCl, + 1.5 Si(OPf), 4Si/Ti 1.5 Si(OPr'),+ 2.5 SiCl, + 1 Ti(OPr'), 6 CHCI, -20Si/Ti 19 Si(OPr'),+20 SiC14+ 1 Ti(OPr'), "In the order of addition.Starting materials Silicon tetrachloride (Aldrich), titanium tetrachloride (Aldrich), zirconium tetrachloride (Jansen) and titanium tetraisopropox- ide (Jansen) were used as received. Silicon tetraisopropoxide was prepared from silicon tetrachloride and isopropyl alcohol according to the method described by Bradley and Hill.23 Diisopropyl ether, dichloromethane and chloroform (Aldrich) were distilled before use. Preparation of the samples The starting solutions were obtained by adding the silicon compound(s) to the metallic compound in a Schlenk tube (Table 1). Solvent was added when needed to obtain a clear solution.After mixing at room temperature, the starting solu- tions were transferred into sealed tubes and heated for 4 days at 110°C. In all cases monolithic gels formed in less than 1 day. The tubes were opened in a glove box under argon, the gels washed with CHCI, and dried for 4 h at 110°C under vacuum, leading to fine, brown powders. The oxides were obtained by the calcination in air of the dried gels, leading to white powders. The overall oxide yield was better than 90% in all cases. Characterization techniques X-Ray powder diffraction was used with Cu-Ka radiation to identify the oxide phases after thermal treatment (Siefert MZ IV diffractometer). Thermal analysis was performed in dry air, at a heating rate of 10 K min-l, on a Netzsch STA 409 thermobalance.Specific surface areas were determined by N, adsorption-desorption experiments using the BET method, using a Micromeritics ASAP 2400. Elemental analyses were performed by the 'Service Central d'Analyses du CNRS' at Vernaison, France: Si, Ti and Zr contents were determined by inductively coupled plasma (ICP) from aqueous solutions; C and H contents by high-temperature combustion and IR spectroscopy; chlorine content by potentiometric titration. The Si/M atomic ratios of the oxides were also determined by EDXA using an energy dispersive X-ray analyser Link AN 1000 fitted to an SEM Cambridge Stereoscan 360. The IR spectra of the solids in Nujol were recorded on a Perkin Elmer 1600 series FTIR spectrophotometer.The 29Si NMR spectra were collected on a Bruker ASX 200 spectrometer at 39.73 MHz, with a 7 mm MAS NMR probe (spinning fre- quency: 3.5 kHz), using proton decoupling, 46 pulses, 60 s recycle delay and adding 500- 1000 scans. Exponential multipli- cation using a line broadening of 100 Hz was performed prior to Fourier transformation. Results and Discussion Formation of the gels As mentioned in the introduction, the easy redistribution of the C1 and OPr' ligands around the silicon and the titanium 1666 J. Muter. Chem., 1996,6( lo), 1665-1671 or zirconium centres has to be kept in mind. This reaction takes place at room temperature during the mixing of the reactants.22 For example, after stirring a mixture of SiCl, and Ti(OPr'), for 3 h at room temperature, 29Si NMR shows the formation of SiCl,(OPr') at 6 -18.9, SiCl,(OPr'), at 6 -41.4, SiCl(OPr'), at 6 -74.3 and a small amount of Si(OPr'), at 6 -87.2.Owing to the occurrence of these redistribution reac- tions, Si-Cl and Si-OPr' as well as M-Cl and M-OPr' (M=Ti or Zr) functions are present. Thus, Si-0-Si, M-0-M and Si-0-M bridges (M=Ti or Zr) may result from the condensation between these functions; in this case, homogeneous samples (on the atomic scale) will be formed only if the rate of the heterocondensation reactions (formation of Si-0-M bridges) is comparable to the rate of the homo- condensations (formation of Si -0-Si or M -0-M bridges). The condensations take place during the heat treatment, leading to the formation of light brown, monolithic gels.The liquid expelled by the spontaneous shrinkage of the gels (syneresis liquid) was analysed using gas chromatography (GC) and 'H NMR spectroscopy. No isopropyl alcohol arising from hydrolysis was detected. The only byproduct detected was isopropyl chloride, indicating that the gelation arose only from non-hydrolytic condensations between M -C1 and M -OPr' groups. Elemental analysis of the Si/Zr and %/Ti xerogels indicated the presence of residual isopropoxide and chloride groups. The empirical formulae of the gels are given in Table 2. The oxygen content was calculated assuming that our samples contained no hydroxy groups or adsorbed water. The mass loss obtained by thermogravimetry in air agreed well with the theoretical mass loss calculated from the empirical formula, assuming formation of the oxide: This result indicates that the empirical formulae and the condensation degrees are reliable.The condensation degrees of the %/Ti xerogels were quite high, >85%; the condensation degrees appeared significantly lower for the Si/Zr xerogels, except for the sample with the lowest Zr content (1 mol%). The formation of well condensed, monolithic Si/Zr and Si/Ti gels suggests that the condensations around the silicon atoms and the zirconium or titanium atoms have comparable rates. From other studies, we know that Ti-Cl/TiO-Pr' and Zr -Cl/ZrO-Pr' condensations effectively takes place at 110 0C,15924 whereas Si-Cl/SiO-Pr' condensations are very slow in the absence of titanium or zirconium.Accordingly, fast precipitation of titania or zirconia should occur. However, Si -Cl/SiO -Pi condensations are efficiently catalysed by Lewis acids (FeCl,, AlCI,, TiCl,, et~.).,~,,~ZrC1, also catalyses these condensations, as shown by the rapid gelation of the sample 99Si/Zr. Thus, the fact that no macroscopic precipi- tation of titania or zirconia occurred and the formation of monolithic gels may be ascribed to a catalysis of the conden- sations around the Si atoms by the transition-metal species, leading to a levelling of the condensation rates around the silicon and the transition-metal atoms. Characterizationof the oxides obtained after calcination The mixed oxides obtained after calcination in air at 500 or 600°C of the dried gels were characterized by elemental analysis, energy dispersive X-ray analysis (EDXA) and X-ray diffraction (XRD). Considering that one of the potential appli- cations of mixed Si/Ti and Si/Zr oxides is heterogeneous catalysis, we have also measured the specific surface area of some samples.The presence of mixed Si-0-M bonds, which prove that mixing at the molecular level occurred during the gel formation, was sought using Fourier-transform IR (FTIR) spectroscopy and 29Si NMR spectroscopy. The crystallization behaviour of the mixed oxides upon heat-treatment in air was studied using XRD. Table 2 Empirical formulae of the gels after drying and mass loss during calcination condensation YO mass loss degree (YO) expl.' (calc.') 57 38 (42) 70 36 (36) 75 33 (37) 80 27 (34) 89 15 (24) 87 21 (23) 86 23 (27) 88 25 (24) 94 20 (17) sa mple empirical formula" 1 lS5Si/Zr 191 Si/ZrA i/ZrB 0Si/Zr 9Si/Zr SiITiA ~iZr0.94(sizr0.79~~izr0.2SiZrO. SiZr0.01(siTi0.92~ 0Pri)1.51C11.8102.22 0p~~1.26c10.8602.52 0~0P~~0.70c10.S1~1.80 )0.56c10.3 lo 1.77 0P~)0.29C10.1601.80 0P2~0.S2c10.4503.36 lS i/TiB ~iTi0.86( 0P2)0.76C10.2903.20 4Si/Ti 2~~~ ~ 0Si/Ti ~ SiTi0.25(0SiTiO.06 Pf)0.40C10. (Opt 10.24c10. 18O2.21 03O 1.99 "From elemental analysis. 'Experimental mass loss, from thermogravimetry, 20-1000 "C in air at + 10 deg. min ~ '. 'Theoretical mass loss derived from the empirical formula, assuming complete oxidation.Composition of the oxides and adsorption-desorption data. According to elemental analysis, the oxides obtained after calcination at 500 or 600°C contained less than 0.3 mass% carbon and chlorine. The metal contents of the glasses (deter- mined either by EDXA or by elemental analysis) were close to the expected values (Table 3). EDXA was also used to verify that no macroscopic phase separation took place during the synthesis of the oxides. For this purpose, 10-point measure- ments were performed, indicating practically constant Si/M ratios in the samples, which is consistent with homogeneity at the micrometre level. The surface areas of our oxides were relatively high, varying between 300 and 1000 m2 g-' (Table 3). All the samples were mesoporous, except the 20Si/Ti sample, which exhibited a type I1 isotherm without desorption hysteresis, indicative of a dominating contribution of microporosity.Microporosity for %/Ti oxides with a low Ti content has been reported by several a~thors*~,~~ Powder X-ray diffraction. All the Si/Zr oxides obtained by calcination at 600°C for 5 h were amorphous. On the other hand, anatase was detected in all the Si/Ti samples calcined for 5 h at 500"C, except the sample with the lowest concen- tration of titanium (20Si/Ti) which was amorphous. FTIR spectroscopy. The FTIR spectra of the Si/Zr oxides are reported in Fig. 1. The spectrum obtained for the 99Si/Zr sample showed the typical absorptions of silica at ca. 1080, 1220 cm-' (Si-0-Si asymmetric stretching vibrations) and 800 cm-' (network Si-0-Si symmetric bond stretching vibration).The presence of hydroxy groups led to a shoulder at 970cm-' (Si-OH bond stretching) and broad bands between 3000 and 3700 cm-' arising from isolated and hydro- gen-bonded SiO-H stretching vibrations and hydrogen-bonded water. When the zirconium content increased, the absorption at 800 cm-' decreased and the shoulder at Table 3 Characterization of the Si/Zr and %/Ti oxides: metal contents of the glasses [from elemental analysis (EA) and EDXA results] and specific surface areas (from BET analysis of N2 adsorptionaesorp- tion data) sample mol% M (nominal) mol% M (EA) mol% M (EDXA) surface area/ m2 g-l 1 Si/ZrA" 50 50 53 334 1 SilZrB' 50 44 48 359 5 Si/Z ra 10Si/Zr" 16.7 8.3 17 10 17 10 651 - 99Si/Zr" 1 0.9 1 - 1 Si/TiA' 1 Si/TiB' 50 50 46 48 47 47 640 528 4Si/Tic 20Si/Ti' 20 4.8 19 6 20 - 790 1000 "Calcined at 600°C for 5 'Calcined at 500°C for 5 h.h; 'Calcined at 600"C, n o holding time. J 8c (de 8n (d 5siIzrs"" A"\,A I I I I I I 4000 3000 2000 1500 1000 500 wavenumberkm -l Fig. 1 IR spectra of the Si/Zr samples calcined for 5 h at 500°C. Samples diluted in a Nujol emulsion, bands due to the Nujol are indicated by asterisks. 970cm-' increased. The decrease of the absorption at 800 cm-' indicates that the silicate network is broken down by the introduction of Zr4+ ions.' In the case of Si0,-ZrO, gels and oxides, the band at 970cm-' has been related to vibrational modes involving Si-0-Zr linkage^.^^,^' This band might also arise from the stretching of Si-OH groups, as in silica.However, the intensity of the broad absorption at ca. 3400 cm-I (SiO-H stretching vibrations) does not signifi- cantly increase with the zirconium content. Therefore, the increase in the intensity of the band at 970 cm-' should arise mainly from the formation of Si-0-Zr linkages. The FTIR spectra of the Si/Ti oxides are reported in Fig. 2. Besides the typical absorptions of silica, a strong band was found at 950 cm-' for all the samples calcined at 500 "C for 5 h. According to IR31-33 and Raman34 spectroscopic studies on mixed Si0,-TiO, gels and oxides, this absorption at about 950 cm-' is associated with a vibrational mode involving Si04 tetrahedra bonded to a titanium in four-fold coordination.In J. Mater. Chem., 1996,6(lo), 1665-1671 1667 I I I I 4000 3000 2000 1500 1000 500 wavenumberkm-l Fig. 2 Comparison between the IR spectra of the 99Si/Zr sample and the &/Ti samples calcined for 5 h at 500°C Samples diluted in a NUJO~ are indicated by asterisks emulsion bands due to the NUJO~ our samples, the intensity ratio of the band at 950cm ' to that at 1080 cm-' did not significantly vary with the Ti content of the oxide (5-50 mol% Ti), which suggests that the number of Si-0-Ti bonds is limited in &/Ti mixed oxides A similar behaviour has already been observed for mixed oxides prepared by hydrolysis of alkoxides thus, Best and C~ndrate~~ reported that the number of Si-0-Ti bonds reached a maximum between 6 3 and 11 2 mol% T10, These observations are in good agreement with the fact that stable S102-T102 glasses in which tetrahedrally coordinated Ti atoms replace Si atoms at random exist up to ca 8 5 mol% TiO, only 35 29Si MAS NMR spectroscopy.The Si/Zr and %/Ti oxides calcined for 5 h at 500 or 600 "C in air have been charactenzed by 29S1 MAS NMR spectroscopy In all cases, the spectra displayed single, broad peaks, as expected for these highly disordered materials The corresponding chemical shifts and the widths at half-height (FWHH) are reported in Table 4 The 29S1 spectra of the 99Si/Zr, lSi/ZrA and lSi/TiA samples, Table4 NMR data for the mixed oxide samples obtained by calcination in air at 500 "C for 5 h (heating rate 5 "C min ') mol% M starting gel (nominal) 6 FWHH (PPd 1Si/ZrA 50 -98 5 23 4 1Si/ZrB 50 -101 5 23 1 5Si/Zra 17 -106 1 16 3 10Si/Zr" 10 -106 1 16 2 99Si/Zr 1 -107 4 17 6 1Si/TiA 50 -107 6 17 0 1 Si/TiC 50 -107 4 17 3 4Si/Tl 20 -107 9 16 7 "Calcined for 5 h at 600°C 1668 J Muter Chem , 1996,6(10), 1665-1671 which are representative of all the spectra obtained, are com- pared in Fig 3 The 99Si/Zr sample contains only about 1 mol% Zr, and its spectrum is typical of the spectra obtained for silica gels calcined under the same conditions, with a peak (FWHH= 17 5 ppm) centred at 6 ca -107 5 The incorpor- ation of large amounts of zirconium in the silica network leads to a significant low-field shift of the peak maximum and to an increase of the linewidth Thus, the spectra of the lSi/ZrA and lSi/ZrB oxides, which contain about 50 mol% Zr, display much broader peaks (FWHH =23 ppm) that are centred at 6 -98 5 and -101 5, respectively On the other hand, after heat treatment at 800 "C, leading to the crystallization of tetragonal zirconia, the peak maximum shifted to 6 -107 and the FWHH decreased to 17 ppm In the Si/Ti samples, the chemical shifts and the linewidth appear independent of the Ti content, and they are close to the values observed for the 99Si/Zr sample and for calcined silica The sensitivity of 29S1 NMR spectroscopy to the nature of the second-nearest neighbours is well documented Thus, the 29S1 chemical shifts of natural aluminosilicates and zeolites depend mainly on the number of A1 atoms in the second coordination sphere of silicon atoms The five possible Si(OSi),(OAI), ,tetrahedra (Q4 ,,At)have characteristic 29S1 shift ranges, each additional aluminium substitution leading to a low-field shift of ca 5-7 ppm 36 Unfortunately, there are much less unambiguous data concerning the effect of the substitution of silicon atoms by titanium or zirconium atoms In the case of zirconium, the 29S1 chemical shift reported for Q4Z= units in zircon is very close to those reported for Q4 units in crystalline aluminosilicates, andalusite and kyanite (Table 5) In the case of titanium, no crystalline binary oxide with silicon exists, however, a low-field shift of 6-8 ppm per titanium substituent may be derived from the chemical shifts of the Q2Ba units in fresnoite3' and of the QIBa units in benitoite (assuming a 10 ppm shift per barium substituent) In addition, the 29S1 NMR spectrum of the ETS-10 titanosilicalite shows a 6 2-9 2 low-field shift between Q4and QlTI3tetrahedral sites 38 Thus, the effect of zirconium or titanium second-nearest neighbours on the 29S1 chemical shift is most probably similar to the effect of aluminium However, it is not straightforward to identify Si-0-Ti or I.I.I.I.1 -60 -80 -100 -120 -140 s Fig 3 29S1MAS NMR spectra of the 99Si/Zr (full line) 1Si/Zr (dashed line) and 1Si/Ti (dotted line) samples calcined for 5 h at 500 "C Table 5 29Si NMR chemical shifts in some silicates and silica gels (from ref.35) tetrahedral site 6 compound -78.8 andalusite (A12Si05) -82.9 kyanite (A12Si05) -81.6 zircon (ZrSiO,) -94.2 benitoite (BaTiSi,O,) -82.0 fresnoite (Ba,TiSi,O,)" -70.3 barium silicate (Ba,SiO,) -109.3 -99.8 silica gels -90.6 'From ref. 36. Si-0-Zr bonds in mixed gels and glasses using ,'Si NMR spectroscopy. Indeed, these samples usually contain numerous surface hydroxy groups bonded to silicon atoms. The low-field shift caused by these hydroxy groups on the ,'Si chemical shift, about 9-10 ppm per OH group, is significantly larger than the effect of aluminium (Table 5). The same silicon atom may be bonded simultaneously to OM and OH groups, in QxOH,(4-n-x)Mnsites.The total number of such sites is 15. In addition, variations in the Si-0-Si angles also lead to a variation in the chemical shift. In the amorphous samples we are dealing with, the different Q signals are not resolved, leading to a single, broad peak. As the chemical shifts and linewidths of the different Q species are not known precisely, the deconvolution of such peaks is not reliable, and the number of Si-0-M bonds cannot be determined. Nevertheless, in SO,-MO, oxides, an increasing number of Si-0-M bonds should lead to an increase of the intensity of QlM3, QlM2, Q2M2,etc. resonances at the expense of the Q" resonance, which will lead to a low-field shift of the peak maximum and to a significant increase of the linewidth.This is what we observe in the lSi/Zr oxides. There is no doubt that these samples are highly homogeneous and contain a large number of Si-0-Zr bonds. Similar spectra were recently reported for SO,-ZrO, calcined gel particles obtained by hydrolysis of aerosols of tetraethoxysilane and zirconium n-propoxide.'" In the same way, a peak centred at 6 -102 with a half-width of 13ppm was reported recently for an Si0,-ZrO, oxide prepared by another non-hydrolytic sol-gel process.39 In the Si/Zr oxides with lower Zr contents (10 and 17 mol% ZrO,), the presence of Si-0-Zr bonds cannot be ascertained, showing the limitations of 29Si NMR spectroscopy in the detection of Si-0-M bonds in mixed oxides.In these cases, "0 NMR spectroscopy would be certainly more sensi- tive4', but it would imply the use of 170-enriched samples, and thus the synthesis of I70-enriched alkoxides or diisopropyl ether, which is not straightforward. In the case of our Si/Ti oxides, Si-0-Ti bonds are present according to the FTIR spectra and the XRD results. However, the 29Si spectra showed no noticeable influence of the Ti content on the chemical shift and linewidth, whatever the Ti content of the glass. This shows that the amount of Si-0-Ti Crystallization behaviour of the oxides The powder XRD results concerning the mixed oxides calcined at different temperatures are summarized in Tables 6 and 7. As mentioned above, all the Si/Zr oxides obtained by calcination at 600 "C for 5 h were amorphous. Crystallization of tetragonal zirconia took place between 600 and 13OO"C, depending on the Zr02 content of the samples; however, the 99Si/Zr sample remained amorphous after heating for 2 h at 1300"C.The distinction between tetragonal and cubic zirconia polymorphs from the XRD spectra only is not straightforward owing to the broadness of the lines.On the other hand, the Raman spectra are readily di~tinguishable.~' In our case, the tetragonal phase was identified clearly by Raman spectroscopy. The temperatures of crystallization of t-ZrO, reported for Si0,-ZrO, gels prepared by hydrolytic sol-gel processes vary between about 450 and 900 0C,10714942p45whereas pure ZrO, crystallizes to t-ZrO, at 330 0C.46 A strong retarding effect has often been ascribed to a good chemical homogeneity of the starting gels, i.e.to a high degree of Si-0-Zr bonding.14 If this criterion was used, the homogeneity of non-hydrolytic Si/Zr gels would be the average of that of hydrolytic gels. The transformation of tetragonal to monoclinic zirconia was retarded strongly in our samples and did not take place after 2 h at 1300 "C,instead of 600-700 "C in pure zirc~nia~~?"~ or 1000-1300 "C in Si0,-ZrO, gels prepared by hydrolytic sol-gel proces~es.~~,~~*~~-"~This retarding effect has been ascribed to blocking by the low-expansivity silica matrix of the tetragonal to monoclinic transformation which involves a positive volume change,45 and the hindering of particle growth by silica, which maintains the size of zirconia particles below the critical size (ca.30 nrn),,, where they transform to the stable monoclinic f~rm.'~,~~ The 1Si/Zr samples were heated at higher temperature to study the formation of zircon, ZrSiO, (Fig. 4). The first traces of ZrSiO, are detected after 2 h at 1500"C; the samples were completely crystallized to zircon after 20 h at 1500 "C. According to the literature, the conversion of Si/Zr gels (without additives) to zircon requires rather high temperatures, between about 1200 and 1600 0C.11,14,43,44*47348 It is noteworthy that in all cases the crystallization of zircon takes place after the crystallization of zirconia. Thus, a high chemical homogeneity of the starting gel is not necessarily favourable for the crystallization of zircon at low temperature.Indeed, Vilmin et al. reported that the formation of zircon in nanoheterogeneous gels (obtained from silica and zirconia sols) I m bonds is too low to be detected by ,'Si NMR spectro~copy.~~ According to Evans, stable %/Ti glasses exist up to ca. 8.5 mol% TiO, only.35 In these glasses, tetrahedrally coordinated Ti atoms replace Si atoms at random. In such cases, the percentage of Si-0-Ti bonds would not exceed 8.5% what- ever the Ti content; their detection by ,'Si NMR spectroscopy would be hindered by the presence of hydroxy groups. This interpretation is consistent with the FTIR spectra, which show a nearly constant intensity of the 950 cm- band whatever the TiO, content of the samples.Here too, 170NMR spectroscopy would be more appropriate for the study of these glasses.37 5 10 15 20 25 30 35 28ldegrees Fig.4 X-Ray diffraction patterns of the lSi/ZrA sample calcined at different temperatures. The most intense reflection of each phase is indicated as follows: m, monoclinic zirconia; t, tetragonal zirconia; z, zircon; c, cristobalite. J. Muter. Chem., 1996, 6(lo), 1665-1671 1669 Table 6 Crystalline phases formed in Si0,-Zr02 samples heat treated at several temperatures (from X-ray powder diffraction) sample Zr mol% 6OO"C, 5 h 1Si/ZrA 50 Am 1Si/ZrB 50 Am 5Si/Zr 17 Am 10Si/Zr 8 Am 99Si/Zr 1 Am "Am, amorphous, t-ZrO,, tetragonal zirconia Table 7 Crystalline phases formed in S1O2-TiO2 samples heat-treated at several temperatures (from X-ray powder diffraction)" heat treatment ~~~~~~~~ ~ sample mol%Ti 500 "C, 5 h 900 "C,2 h 1300"C, 2 h 1Si/TiA 50 Am +An An An 1Si/TiB 50 Am+An An An 4Si/Ti 20 Am+An AnfCr An+Cr 20Si/Ti 5 Am Cr Cr aAm, amorphous, An, anatase, Cr, cnstobalite occurred at a lower temperature than in homogeneous, mono-phasic gels 48 In the case of the Si/Ti oxides, only the 20Si/Ti sample remained amorphous after calcination for 5 h at 500°C (Table 7) It is noteworthy that this sample, containing ca 5 mol% TiO,, is within the 'stable' glass region, which exists up to about 8 5 mol% TiO, according to Evans,35 whereas the other samples are outside this stable glass region Interestingly, the 20Si/Ti sample crystallized as single phase cristobalite at 900°C (note that the 99Si/Zr sample was still amorphous at 1300"C), neither anatase nor rutile was detected, even after annealing for 1h at 1540"C (Fig 5) In addition, the positions of the diffraction lines of cristobalite were modified slightly for instance, the peaks corresponding to the [1131 and [212] reflections were observed at 28 =46 70 and 48 28", instead of 47 05 and 48 63" in pure SiO, cnstobalite The same behaviour was observed for ultra-low expansion (ULE) glasses prepared by flame hydrolysis of chlorides,35 it was ascribed to the formation of a solid solution of T10, in SiO, the substitution of Sl4+ ions by larger Ti4+ ions leads to a linear increase of the tetragonal unit-cell constants The above 28 yalues corre-spond to cell parameters a= 5 00 A and c= 6 98 A, which are in excellent agreement with the values reported by Evans for h l Ih v)c.c V' .A ..M'Ls 1300 "C, 2 h V' > c v) C al + E 900"C,2h 600°C, 5 h 5 10 15 20 25 30 35 40 2Wdegrees Fig.5 X-Ray diffraction patterns of the 20Si/Ti sample calcined at different temperatures 1670 J Muter Chem, 1996, 6(10), 1665-1671 heat treatment 800"C, 5 h lOOO"C, 2 h 13OO0C,2 h t-ZrO, t-ZrO, t-ZrO, - - - t-ZrO, t-ZrO, t ZrO, Am t-ZrO, t-ZrO, Am Am Am an oxide containing 5 mol% Ti35 Accordingly, all the Ti atoms in the 20Si/Ti sample are incorporated in the silica network, suggesting that the starting gel was perfectly homogeneous In the 4Si/Ti sample, which is outside the stable glass region, both anatase and cristobalite crystallized on heating This behaviour shows that part of the T10, formed a solid solution with SiO,, and the remainder crystallized as anatase A similar behaviour was reported for S1O,-T1O2 glasses containing 10 and 30 mol% TiO,, prepared by hydrolysis of alkoxides 49 In the other samples, anatase remained the only crystalline phase detected up to 1300 "C The crystallization of anatase in Si/Ti oxides prepared by hydrolytic sol-gel processes occurs between about 500 and lOOO"C, depending on the conditions of synthesis and the titania content 24 Accordingly, the crystallization temperature of anatase in our samples other than 20Si/Ti is rather low This suggests that the starting gels are inhomogeneous, which is consistent with the small number of Si-0-Ti bonds detected by FTIR spectroscopy, and the 29S1NMR spectra Although anatase crystallized at a rather low temperature in our &/Ti samples, the transformation of anatase to rutile was not observed, even after heating for 2 h at 1300°C The transformation of anatase to rutile is usually detrimental to the applications of T10, oxides as catalysts, owing to the related decrease of surface area In pure titania gels, this transformation takes place between 600 and 1000"C 46 The same suppressive effect on the transformation of anatase to rutile has been reported for mixed SiO,-TiO, oxides prepared by a hydrolytic sol-gel process,so provided that the gels are relatively homogeneous Conversely, the formation of rutile at temperatures as low as 600 "C is reported for samples prepared by physical mixing of silica and titania gels5' or for alkoxide gels prepared in excess water 51 Conclusions The non-hydrolytic sol-gel route described here, involving chlorides and isopropoxides (or diisopropyl ether), applies well to the preparation of Si/Zr and Si/Ti mixed oxides Monolithic Si/Zr and &/Ti gels are readily obtained in one step, without the use of additives The Si/M ratio of the gel and of the oxide may be controlled easily by the composition of the starting mixture The surface areas of the oxides obtained by calcination of the gels are quite high In both Si/Zr and &/Ti oxides, IR and/or 29S1 NMR spectroscopies showed the presence of Si-0-M bonds, indi-cating some degree of homogeneity on the atomic scale However, the number of such bonds remained relatively low in Si/Ti mixed oxides, compared to Si/Zr oxides, which agrees well with the behaviour reported recently for mixed oxides prepared by a hydrolytic sol-gel process 40 The Si/Zr oxides (up to 50 mol% Zr) remained amorphous after 5 h at 600°C The transformation of tetragonal to mono-clinic zirconia was strongly retarded in our samples and did not take place after 2 h at 1300"C The crystallization of zircon (for a sample containing 50 mol% Zr) started at 1500°C and was completed after 20 h at 1500°C The precipitation of anatase was observed after 5 h at 500°C for the Si/Ti oxides with high Ti contents (20-50 mol% Ti), which are outside the stable glass region On the other hand, the transformation of anatase to rutile was not observed even after 2 h at 1300°C 23 24 25 26 D C Bradley and D A W Hill, J Chem SOC, 1963,2101 P Arnal, These, Universite de Montpellier 2, 1995 K A Andnanov,T N GaninaandN N Sokolov J Gen Chem USSR, 1956,26,1897 W Noll, Chemistry and Technology of Szlzcones, 2nd edn ,Wiley, New York, 1951 The sample within the stable glass region (<8 5 mol% Ti) 27 D C M Dutoit, M Schneider and A Balker, J Catal 1995, crystallized at 900 "C as single phase cristobalite oxide, with Ti4+ ions substituting Si4+ ions at random, which suggests that a highly homogeneous gel was formed in this case 28 29 153,165 B E Hardy, M Maciejewski, A Balker and A Wokaun, J Muter Chem ,1992,6,673 S W Lee and R A Condrate, Sr ,J Muter Sci ,1988,23,2951 30 J A Navio, M Macias, G Colon, P J Sanchez-Soto, References 31 V Augugliaro and L Palmisano, Appl Surf Sci , 1994,81,325 C F Smith, Jr, RA Condrate, Sr and W E Votava, Appl 1 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 C J Brinker and G W Scherer, SoI-Gel Science The Physics and Chemistry of Sol-Gel Processing, Academic Press, San Diego, 1990 H Hosaka and K Meguro, Bull Chem SOC Jpn, 1971,44,1252 S Sakka and K Kamiya, J Non-Cryst Solids, 1980,42,403 B E Yoldas, J Non-Cryst Solids, 1980,38&39,81 Y Abe, N Sugimoto, Y Nagao and T Misono, J Non-Cryst Solids, 1988, 104, 164 M Emih, L Incoccia, S Mobilio, G Fagherazzi and M Guglielmi, J Non-Cryst Solids, 1985,74, 129 I M Miranda Salvado and J M Fernandez Navarro, J Non-Cryst Solids, 1992,1471148,256 M Nogami, J Non-Cryst Solids, 1985,69,415 M Nogami, J Non-Cryst Solzds, 1994,178,320 I M Miranda Salvado, C J Serna and J M Fernandez Navarro, J Non-Cryst Solids, 1988,100, 330 Y Kanno, J Muter Sci ,1989,24,2415 T Mory, H Yamamura, H Kobayashi and T Mitamura, J Am Ceram SOC,1992,75,2420 R Oheim, H Paulus and C Russel, J Muter Scz Lett, 1991, 10,1171 P Tartaj, J Sanz, C J Serna and M Ocana, J Muter Sci, 1994, 29,6533 R J P Corriu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Muter Chem ,1992,6,673 R J P Corriu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Non-Cryst Solids, 1992,146,301 R J P Cornu, D Leclercq, P Lefevre, P H Mutin and A Vioux, Chem Muter, 1992,4,961 S Acosta, R J P Corriu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Non-Cryst Solids, 1994,170,234 S Acosta, R J P Cornu, D Leclercq, P Lefevre, P H Mutin and A Vioux, J Sol-Gel Sci Technol ,1994,2,25 K Moedntzer, in Organometallzc Reactzons, ed EI Becker and M Tsutsui, Wiley Interscience, 1971, vol 2, pp 1-116 H Weingarten and J R Van Wazer, J Am Chem SOC, 1965, 87,724 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 Spectrosc ,1975,29,79 Z Dend, E Breval and C G Pantano, J Non-Cryst Solids, 1988, 100,364 A Kasgoz, K Yoshimura, T Misono and Y Abe, J Sol-Gel Sci Technol, 1994,1, 185 M F Best and R A Condrate, Sr ,J Muter Scz Lett, 1985,4,994 D L Evans, J Non-Cryst Solids, 1982,52, 11 5 G Engelhardt and D Michel, High-Resolution Solzd-State NMR of Silicates and Zeolites, Wiley, Chichester, 1987 P J Dirken, M E Smith and H J Whitfield, J Phys Chem, 1995, 99,395 M W Anderson, 0 Terasaki, T Oshuna, A Philippou, S P MacKay, A Ferreira, J Rocha and S Lidin, Nature (London), 1994,367,347 M Jansen and E Guenther, Chem Muter, 1995,7,2110 P J Dirken, R Dupree and M E Smith, J Muter Chem, 1995, 5,1261 C M Phillippi and K S Mazdiyasni, J Am Ceram SOC, 1971, 54,254 S K Saha and P Pramanik, J Non-Cryst Solids, 1993,159,37 J Campaniello, E M Rabinovich, P Berthet, A Revcolevschi and N A Kopylov, in Better Ceramics through Chemistry IV Muter Res SOC Symp Proc 1990, vol 180, p 541 G Monros, M C Marti, J Carda, M A Tena, P Escribano and M Anglada, J Muter Sci ,1993,28, 5852 V S Nagarajan and K J Rao, J Muter Sci ,1989,24,2140 M Ocaiia, V Fornes and C J Serna, Ceram Inter, 1992,18,99 A B Hardy and W E Rhine, Chemical Processing of Advanced Materials, ed L L Hench and J K West, Wiley, 1992, p 577 G Vilmin, S Komarneni and R Roy, J Muter Sci , 1987,22,3556 I M Miranda Salvado and J M Fernandez Navarro, J Non-Cryst Solids, 1992,1471148,256 H Nakabayashi, K Nishiwaki and A Ueno, Muter Res Bull, 1988,23,555 B E Handy, M Maciejewski, A Balker and A Wokaun, J Muter Chem , 1992,2,833 22 D C Bradley, D C Hancock and H Wardlaw, J Chem SOC, 1952,2773 Paper 6/02404A, Received 9th April, 1996 J Muter Chem, 1996, 6(10), 1665-1671 1671