J O U R N A L O F C H E M I S T R Y Materials Solvent-free synthesis of binary inorganic oxides John N. Hay*† and Hema M. Raval Department of Chemistry, School of Physical Sciences, University of Surrey, Guildford, Surrey, UK GU2 5XH The non-hydrolytic sol–gel route has been used in the solvent-free synthesis of binary inorganic oxides based on silicon, aluminium and titanium. Where necessary, iron(III ) chloride was used as a catalyst.Clear evidence for the formation of true, amorphous binary systems was obtained only in the case of aluminosilicates. A crystalline aluminosilicate was obtained after calcination at 1000 °C. A titanium–silicon binary system gave only crystalline rutile and anatase following calcination, with no evidence for either crystalline silica or a binary oxide.Low temperature routes to inorganic oxides have attracted Mechanistically, the non-hydrolytic route should favour the considerable attention in recent years owing to their reduced formation of homogeneous binary oxides from diVerent ‘metal’ energy demand compared to traditional high temperature glass precursors since the reaction of M(OR)m with M¾Xn should forming processes and the potential opened up for combining form unsymmetrical MMOMM¾ linkages in the absence of the oxides with thermally labile organic compounds.The reversibility and ligand exchange reactions. Another way to oxides and their hybrids have a plethora of uses, including make non-hydrolytic sol–gels is to form the alkoxy groups in catalysis and catalyst supports, ceramics, sensor applications situ, by reacting the metal halide with an alcohol10 or other and active glasses.The most important synthetic route is the organic oxo-compounds, e.g. ethers, aldehydes and ketones.11 hydrolytic sol–gel method which allows inorganic oxides The non-hydrolytic route has been used to make non-metal and/or their immediate precursors to be produced from simple oxides, metal oxides, transition metal oxides and binary oxides, alkoxides or chelates via low temperature hydrolysis and with the outstanding contribution coming from the group of condensation reactions.This route has fairly widespread appli- Corriu.12 Non-metal oxides, e.g. silica gels, have been studied cability and has been the subject of a number of recent books most extensively.Stoichiometric reactions between silicon and reviews.1–7 For certain applications, however, this tetrahalides and various oxygen donors including alcohols, approach has a number of disadvantages, including the need benzaldehyde and tetrabenzyloxysilane, have been examined.13 to add water for the hydrolysis, the formation of condensation Overall, alcohols appear to be more eYcient oxygen donors by-products (e.g. water, alcohols) and the initial formation of than alkoxides, ethers and aldehydes. In the case of metal a solvent-swollen gel which undergoes substantial shrinkage oxides, alumina gels have been studied by Corriu et al., who on drying. In addition, for the production of uniform binary found solvents to be necessary.9,14 The main oxygen donors oxides, the large diVerences in hydrolysis rates of diVerent to aluminium halides were aluminium isopropoxide and diisoprecursors can lead to undesirable inhomogeneities in the propyl ether.The alumina obtained was amorphous up to product. This is true of binary oxides derived from the trans- 750 °C compared to those obtained via the hydrolytic route ition metal alkoxides, especially those of d0 transition metals, which usually crystallise below 500 °C.As the precursors to e.g. titanium and zirconium, which are widely used precursors the transition metal oxides are mostly liquids, these reactions for glasses and ceramics. The lower electronegativities of the often proceed without the need for solvents. Some of the transition metals compared to silicon make their alkoxides oxygen donors used with titanium tetrachloride include tetramore reactive towards nucleophilic reactions such as hydrolysis hydrofuran (THF), diisopropyl ether, dimethoxyethane and condensation. (DME), and titanium isopropoxide.9,15 Gel-times were longer One possible route to avoiding this complication is the non- than under hydrolytic conditions, which may be an advantage hydrolytic sol–gel process, which is based on an observation for titanium since this increases the potential for more control made by Gerrard et al.8 Whilst studying the interactions of the reaction.Other metal halides investigated include between organosilanes and alcohols, Gerrard et al. found silica niobium, molybdenum and tungsten chlorides, but these lead to be formed.Since then, this route has been adapted to to precipitates rather than gels. More recently, Guenther et al. involve the direct condensation of metal halides with metal reacted zirconium chloride with diisopropyl ether.16 Further alkoxides, e.g. silicon, titanium and zirconium (Scheme 1).9 work by our group on single inorganic oxides prepared by the The mechanism for this reaction involves the coordination of non-hydrolytic sol–gel route is reported separately.17 the oxygen donor, e.g.alkoxide, to the metal centre of the The rate of condensation in the non-hydrolytic process is metal halide. This usually evolves via three possible transition highly dependent on the nature of the oxygen donor. The states or intermediates and results in the formation of oxogeneral tendency of the non-hydrolytic process compared to bridges.The non-hydrolytic route involves the cleavage of the the hydrolytic route is to delay crystallisation18 and this carbon–oxygen bond instead of the metal–oxygen bond. The behaviour may find application in the preparation of heteromain limitation of this process is the insolubility of some metal geneous catalysts.DiVerent metals may also be used to form chlorides in non-aqueous solvents. It also has a general tendmonolithic binary mixed metal oxides via a cross-condensation ency to delay crystallisation of the metal oxide. This process reaction (Scheme 2).18–21 appears to be simpler to control than the conventional hydroly- The hydrolytic route to mixed metal oxides often leads to sis–condensation process found in the hydrolytic route.diYculties in controlling stoichiometry and homogeneity. Homogeneity depends on the rate of homocondensation (i.e. formation of MMOMM and M¾MOMM¾) versus the rate of † E-mail: j.hay@surrey.ac.uk J. Mater. Chem., 1998, 8(5), 1233–1239 1233M OR + M X M X M OR M O M R + + X– M O M X – + R+ M X M O R M O R X M + (1) (2) (3) Scheme 1 involved a reflux system, immersed in a temperature controlled mMXn + nM¢(OR)m MmM¢ nOnm + nmRX oil bath, connected to a series of nitrogen drying vessels.A Scheme 2 side arm was included above the reaction tube to allow the safe addition of air-sensitive liquid reactants from a syringe via a septum. heterocondensation (i.e. formation of MMOMM¾). One way to overcome this in the hydrolytic route is to subject the less reactive alkoxide precursor to partial hydrolysis before adding the more reactive one, or to use heterometallic alkoxides.In Instrumentation principle, the non-hydrolytic process potentially provides a A Shimadzu thermogravimetric analyser-50 was used to deterbetter route to mixed metal oxides than the conventional mine sample mass loss over a temperature range of 25–490 °C. hydrolytic sol–gel approach or mixed powder methods because A heating rate of 20 K min-1 was applied under a nitrogen of improved homogeneity during the reaction.22 This is in part atmosphere (flow rate, 50 ml min-1).For mass losses up to a consequence of the reaction mechanism, which is completely higher temperatures, a Perkin-Elmer TGA-7 interfaced with a diVerent from that of the classical sol–gel process.As in the Perkin-Elmer 7700 computer was used to measure sample hydrolytic route, the reactants react at low temperature in a mass loss over a temperature range of 40–900 °C. A heating homogeneous solution or liquid, thus avoiding phase separarate of 10 K min-1 was applied under a nitrogen atmosphere tion, crystallisation and chemical decomposition.DiVerent with a flow rate of 50 ml min-1. combinations of silicon, aluminium and titanium chloride have All calcinations were carried out in a Lenton UAF 16/21 been reacted with the corresponding isopropoxides.9,18,20 In furnace. An SMC 24127 calibrated digital K-type thermometer all cases gels were formed. Compositions can be controlled via was used to control temperatures to ±5 °C and the sample the composition of starting solutions, since the results show was contained in a mullite tube during the heating run.A that calcined gels are close in composition to initial solutions Siemens D500 X-ray diVractometer was used to determine and this agrees with the high yields achieved. Iwasaki et al. sample morphology. Samples were ground into fine powders have recently described the condensation of silicon tetraacetate before being scanned with Cu-Ka radiation at 40 kV, 40 mA with titanium acetylacetonatotriisopropoxide in THF.23 More from 15–70° 2h, with a step size of 0.02° and a count time per recently, it has been demonstrated that homogeneous zirstep of 10 seconds.Surface morphology and evidence of any conium titanate gels can be prepared by the non-hydrolytic microporosity in the samples were evaluated using a sol–gel route without the intermediate formation of zirconia CAMSCAM S4 field emission scanning electron microscope or titania.21 In this paper we describe the synthesis and with an accelerating voltage of 5 kV for the secondary electron characterisation of further binary inorganic oxides via the imaging (SEI) and 20 kV for the energy dispersive X-ray solvent-free non-hydrolytic sol–gel route.To date, there are (EDX) analyses. Prior to mounting, the specimens were carbon relatively few reports of the solvent-free synthesis of binary coated to prevent charging under the electron beam. EDX oxides by the sol–gel route, which is likely to be a prerequisite spectra were acquired from a number of areas to determine for the successful commercial development of products prethe elemental composition of the sample.pared by this method. Infrared (IR) spectra were recorded on a Perkin-Elmer 1750 FT-IR spectrophotometer interfaced with a Perkin-Elmer Experimental computer. Samples were ground and prepared for diVuse reflectance infrared Fourier transform spectroscopy (DRIFTS).Materials Abbreviations used to describe peaks are as follows: vs=very Silicon tetrachloride (Aldrich), tetraethylorthosilicate, TEOS strong; s=strong; m=medium; w=weak; vw=very weak; (Lancaster), aluminium trichloride (Aldrich), aluminium iso- b=broad; sp=sharp and sh=shoulder. propoxide (Aldrich), titanium tetrachloride (Aldrich), titanium Solid-state 29Si nuclear magnetic resonance (NMR) specisopropoxide (Lancaster), ethanol (Hayman Ltd.) and iron(III ) troscopy was undertaken at the University of Durham, at chloride (Aldrich) were all used as received.Propan-2-ol ambient temperature on a Varian UNITYplus spectrometer. (Fisons), diethyl ether (Fisons) and carbon tetrachloride (BDH) The spectrum was recorded against an external TMS standard were dried using 4 A ° molecular sieves before use.with magic angle spinning (MAS) at a spinning rate of 4300 Hz and an angle of 54.7°. The cross polarised (CP) spectrum was Method obtained as a single contact experiment with a contact time of 3 ms and a relaxation delay of 2.0 s (700 repetitions). An Many of the materials used to make non-hydrolytic sol–gels acquisition time of 9.6 ms was used.Silicon sites were labelled are very reactive and have strict handling conditions. As a with the conventional Qn notation where Q refers to tetrafunc- consequence, care should be taken when embarking on new tional SiO4 units and n to the number of bridging oxygen and untried experiments, particularly in the absence of a diluting solvent.The basic set-up used in these experiments atoms surrounding the central silicon atom. 1234 J. Mater. Chem., 1998, 8(5), 1233–1239Fig. 2 FTIR spectrum of aluminosilicate synthesised from aluminium Fig. 1 FTIR spectrum of aluminosilicate synthesised from silicon tetrachloride and aluminium isopropoxide (equimolar) trichloride and TEOS (453 molar ratio) Synthesis of binary oxides Preparation of aluminosilicate from silicon tetrachloride and aluminium isopropoxide (equimolar) with the aid of a solvent.An equimolar amount of silicon tetrachloride (4.00 g, 2.70 cm3, 0.024 mol) was added to aluminium isopropoxide (4.81 g, 0.024 mol) dissolved in a 451 (by mass) mixture of carbon tetrachloride and diethyl ether (4 cm3). A white solid was formed within approximately 40 minutes at 90 °C.This was washed in propan-2-ol and dried for 3 days at 90 °C followed by a further 2 days at 150 °C (2.13 g, 81% yield based on SiAlO3.5. Note: since the non-condensed sites of the network cannot be evaluated, yields are only included as information). The thermogravimetric analysis (TG) showed a mass loss of 14%. The IR spectrum (Fig. 1) showed absorptions at nmax#3400 cm-1 (m, b), 1630 cm-1 (m), 1200 cm-1 (vs, sp), 1150 cm-1 (sh) and 450 cm-1 (vs, sp). Preparation of aluminosilicate from aluminium trichloride and TEOS (equimolar) with the aid of a catalyst. Equimolar amounts of aluminium trichloride (2.06 g, 0.015 mol) and TEOS (3.12 g, 3.35 cm3, 0.015 mol) were reacted with iron(III) chloride (0.070 g, ca. 1% by mass).A solid product was obtained within 3.5 hours at 110 °C. The solid was washed with ethanol and subsequently dried at 90 °C for 3 days and at 150 °C for a further 2 days (1.50 g, 84% yield based on AlSiO4). The resulting monolith was dark brown in colour probably due to trapped catalyst. The TG showed 9% mass loss and the IR spectrum showed absorptions at nmax#3600 cm-1 (b), 1600 cm-1 (vs, sp), 1250 cm-1 (m, sh), 1000 cm-1 (s, b), 950 cm-1 (sh) and 450 cm-1 (m).Fig. 3 XRD spectrum of aluminosilicate prior to calcination (top) and after calcination (below) at 1000 °C for 18 hours accompanied by the Preparation of aluminosilicate from aluminium trichloride peak positions observed for Al6Si2O13 and TEOS (stoichiometric) with the aid of a catalyst. Aluminium trichloride (3.06 g, 0.023 mol) and TEOS (3.52 g, 3.76 cm3, 0.017 mol) were mixed together in the presence of iron(III) be highly porous, with the pores being approximately 200 nm in diameter.EDX analysis of the bulk material (Fig. 4) showed chloride (0.068 g, 1.03% by mass). A clear, dark brown monolith was formed after approximately 10 minutes at 90 °C. The the elements aluminium, silicon and oxygen.Iron was also present in some loose material covering the surface of the product was washed in diethyl ether ( left overnight), filtered and dried at 150 °C for 3 days (2.08 g, 94% yield based on sample. The IR spectrum (Fig. 5) showed absorptions at nmax#3600 cm-1 (vw), 2000 cm-1 (vw), 1880 cm-1 (w), Si3Al4O12). TG showed a 22% mass loss and the IR spectrum (Fig. 2) showed absorptions at nmax#3400 cm-1 (m, b), 1620 cm-1 (w), 1330 cm-1 (vs, sp), 1240 cm-1 (m), 1030 cm-1 (m), 930 cm-1 (m), 850 cm-1 (s), ~640 cm-1 (s), 529 cm-1 (s) 1611 cm-1 (m), 1250 cm-1 (s, sp), 1160 cm-1 (sh), 950 cm-1 (m), 800 cm-1 (m) and 500 cm-1 (m). and 420 cm-1 (m). The dark brown and black coloured product of this reaction was calcined in air at 1000 °C for 18 hours, resulting in Preparation of titanium–silicon binary oxide from silicon tetrachloride and titanium isopropoxide (equimolar).18–21 approximately 30% mass loss.The resulting specimen was heterogeneously coloured black, brown and oV-white. The X- Equimolar amounts of silicon tetrachloride (2.00 g, 1.35 cm3, 0.012 mol) and titanium isopropoxide (3.35 g, 3.5 cm3, ray diVraction (XRD) pattern of the oxide prior to calcination and that after calcination are shown in Fig. 3. The latter shows 0.012 mol), were reacted at 55 °C. A fine white precipitate was seen within 1 minute of stirring at room temperature and this a very good fit with an orthorhombic mullite structure, Al6Si2O13, with no crystalline peaks left unmatched. The disappeared at approximately 45 °C.A clear, pale brown, crazed solid resulted in less than 24 hours and was washed in scanning electron micrograph (SEM) showed this material to J. Mater. Chem., 1998, 8(5), 1233–1239 1235Fig. 7 XRD spectrum of titanium–silicon binary oxide calcined for 18 hours at 1000 °C accompanied by the peak positions observed for rutile and anatase Fig. 4 EDX spectrum of aluminosilicate calcined for 18 hours at 1000 °C (bulk) Fig. 5 FTIR spectrum of aluminosilicate calcined for 18 hours at 1000 °C Fig. 8 SEM of titanium–silicon binary oxide calcined for 18 hours at 1000 °C Fig. 6 29Si NMR spectrum of titanium–silicon binary oxide prior to calcination propan-2-ol and dried at 85 °C for 2 hours producing a creamy, shell-like product (1.70 g, 101% yield based on TiSiO4).TG showed a 20% mass loss. When subjected to further drying at Fig. 9 EDX spectrum of titanium–silicon binary oxide calcined for 18 hours at 1000 °C (bulk) 140 °C for 24 hours, a charcoal black solid was formed (1.36 g, 81% yield). The IR spectrum shows absorptions at nmax#3500 cm-1 (s, b), 1630 cm-1 (m), ~1100 cm-1 (s, b) and 600 cm-1 (s, b). The 29Si NMR spectrum is shown in Fig. 6. tetragonal rutile and anatase. There seemed to be more rutile than anatase present, but no silica phase was detected. A The black material was calcined in air at 1000 °C for 18 hours to leave a white solid (#28% mass loss). A thin black typical SEM is shown in Fig. 8. The EDX spectrum (Fig. 9) showed the bulk of the specimen to have a high titanium coating covered the surface of a few particles.Qualitative XRD results (Fig. 7) show a good fit with two titania phases; content in addition to the silicon and oxygen signals. 1236 J. Mater. Chem., 1998, 8(5), 1233–1239Preparation of titanium–silicon binary oxide from titanium tetrachloride and TEOS (equimolar). An equimolar amount of 4 AlCl3 + 3 Si(OC2H5)4 Al4Si3O12 + 12 C2H5Cl FeCl3 titanium tetrachloride (4.00 g, 2.32 cm3, 0.021 mol) was reacted Scheme 3 with TEOS (4.39 g, 4.70 cm3, 0.021 mol) for 5 minutes at 110 °C.The dark brown gel was washed in ethanol, filtered and dried at 150 °C for 4 days (3.66 g, 124% yield based on formation. At a 453 molar ratio of aluminium trichloride to TiSiO4). TG showed a 28% mass loss and the IR spectrum TEOS, a shorter gel-time was achieved at a lower temperature showed absorptions at nmax#3500 cm-1 (m, b), ~3000 cm-1 and the yield was the highest of the aluminosilicate sol–gel (m), 1850 cm-1 (vw, b), 1630 cm-1 (m, sp), 1430 cm-1 (m, sp), systems; however, the TG showed a higher percentage mass 1370 cm-1 (m, sp), 1250 cm-1 (vs, sp), 1160 cm-1 (s), 960 cm-1 loss, suggesting incomplete reaction.IR analysis (Fig. 2) was (m), 810 cm-1 (m) and 525 cm-1 (m).indicative of formation of a binary oxide, which might be expected to have the empirical formula Al4Si3O12 on the basis Preparation of titanium–aluminium binary oxide from alu- of the stoichiometry (Scheme 3). minium trichloride and titanium isopropoxide (453 molar ratio). The XRD pattern of the oxide prior to calcination (Fig. 3) An appropriate stoichiometric amount of aluminium trichlo- shows the sample to be amorphous with a broad peak at the ride (2.00 g, 0.015 mol) was reacted with titanium isopropoxide low angle end of the spectrum.Following calcination of the (3.20 g, 3.35 cm3, 0.011 mol) for 24 hours at 110 °C. The dark oxide at 1000 °C, XRD analysis reveals that the sample has brown monolith was washed in diethyl ether ( left overnight), crystallised, although an amorphous phase is still present.The filtered and dried for 3 days at 150 °C (1.88 g, 113% yield XRD trace shows a very good fit with an orthorhombic mullite based on Al4Ti3O12). TG showed a 26% mass loss. The IR structure, Al6Si2O13, with no peaks left unmatched. The stoichispectrum produced absorptions at nmax#3400 cm-1 (s, b), ometry of this structure is very diVerent to that expected from 1630 cm-1 (vs, sp), 1100 cm-1 (s), ~900 cm-1 (vs, b), 800 cm-1 the reaction stoichiometry and may result from the presence (s) and 600 cm-1 (m, b).of an amorphous silica phase in the calcined structure. The extent to which molecular or network rearrangement has Preparation of titanium–aluminium binary oxide from occurred during the calcination process is impossible to ascertitanium tetrachloride and aluminium isopropoxide (354 molar tain from these results.It should be noted that at 1000 °C, the ratio). Appropriate stoichiometric amounts of titanium tetra- thermodynamics control the crystallisation process of the chloride (4.00 g, 2.32 cm3, 0.021 mol) and aluminium isoprop- mullite. The possibility of nucleation initiated by the mullite oxide (5.74 g, 0.028 mol) were stirred together for tube used in the calcination can not be totally discounted.approximately 6 hours at 110 °C. The dark brown sample was Optimisation of the reaction stoichiometry might maximise washed with propan-2-ol and then dried for 4 days at 150 °C formation of a crystalline mullite after calcination, but this leaving a black shiny material (4.59 g, 147% yield based on was not carried out as part of this study.Longer heating times Ti3Al4O12). TG showed a 35% mass loss. The IR spectrum might also improve the conversion. showed absorptions at nmax#3500 cm-1 (m, b), 3000 cm-1 (m, The SEMs show the calcined product to be highly porous, sh), 1630 cm-1 (m), 1100 cm-1 (s), 1000 cm-1 (vs, sp), 800 cm-1 with pores of approximately 200 nm diameter.The sample (m, sh) and ~450 cm-1 (w). surface was covered with loose material which, using EDX analysis, was found to contain a high iron content (catalyst residue) in addition to the elements aluminium, silicon and Results and Discussion oxygen which were found in the EDX spectrum of the bulk material (Fig. 4). The aluminium signal is much weaker than Aluminium–silicon oxides the silicon signal, in contrast to the Al5Si ratio of 652 found Three experiments were undertaken on the synthesis of alumifor the crystalline phase using XRD analysis. This apparent nosilicates similar to those reported by Corriu et al.19 A solvent discrepancy may be explained by the fact that XRD analysis was employed for the initial reaction to moderate the potenprovides information only on the crystalline phase, while EDX tially vigorous reaction between silicon tetrachloride and aluanalysis averages the analysis of the bulk sample, including minium isopropoxide.A stoichiometric ratio of 151 was used, amorphous material (such as silica). An additional factor may compared to the halide to alkoxide ratio of 154/3 used by be the influence of the surface roughness of the sample on the Corriu et al.Use of a higher reaction temperature in our case consistency of the quantitative analyses obtained by EDX. led to rapid formation of the initial product, compared to a Following calcination, IR analysis showed that the siloxane literature gel-time of two days at 40 °C. The IR spectrum stretching vibration at #1180 cm-1 had disappeared and been (Fig. 1) of the white product confirmed the formation of replaced by a higher frequency band at 1330 cm-1 (Fig. 5), aluminosilicate.24 The presence of a band at #1200 cm-1 may which could result from crystallisation of the aluminosilicate. be indicative of aluminium substitution in the silica network. The peak at #1150 cm-1 appears as a shoulder and can be Titanium–silicon oxides assigned to the elongation vibration of SiMOMAl.Vibrations between 3500–3400 cm-1 and at 1630 cm-1 are due to the For the titanium–silicon binary oxide system prepared from silicon tetrachloride and titanium isopropoxide, the results stretching of hydroxyl groups from either hydrogen bonded water or propan-2-ol and surface silanol molecules.The latter tend to agree with published findings on related systems.19 IR spectroscopy provides limited evidence for the presence of absorption is due to the deformation mode of these hydroxyl groups. The TG indicated an incomplete reaction, probably TiMOMSi species in the initial amorphous product. Solidstate 29Si NMR spectroscopy was used to obtain data on the due to the presence of unreacted material and trapped solvent or by-product.extent of TiMOMSi bonding in the system (Fig. 6). The spectrum shows the expected SiMOMSi signals at -101 ppm No solvent was used for the reaction of aluminium trichloride with TEOS; however, based on studies by Corriu et al.,25 and -109 ppm. These were assigned to partially condensed, Q3 (ROSiO3) species and fully condensed, Q4 (SiO4) species.a catalyst was employed in the reactions to reduce reaction times and residual unreacted functionalities. Two ratios were No signals (expected at #-20 ppm) could be attributed to the presence of TiMOMSi species,26 suggesting the product studied. At a 151 ratio of aluminium halide to TEOS, a good yield was achieved after 3.5 hours at 110 °C, with a relatively was actually a phase separated mixture of silica and titania.The poor stability of TiMOMSi bonds may help to explain small TG mass loss due to residual volatiles, implying good oxide conversion. The IR analysis24 suggested aluminosilicate this result since any first-formed TiMOMSi species will J. Mater. Chem., 1998, 8(5), 1233–1239 1237rearrange on ageing to form SiMOMSi and TiMOMTi species.The XRD pattern (Fig. 7) verified that after calcination at 3 Ti(OPri)4 + 4 AlCl3 Ti3Al4O12 + 12 PriCl 1000 °C, the majority of the sample had crystallised. A good Scheme 4 fit was obtained with two titania phases, tetragonal rutile and anatase with more of the former present. No crystalline silica was detected, although other studies undertaken by us had shown that pure silica formed by the non-hydrolytic sol–gel 3 TiCl4 + 4 Al(OPri)3 Ti3Al4O12 + 12 PriCl Scheme 5 process could by crystallised under similar calcination conditions to form mixtures of quartz and cristobalite.17 Some possible reasons for this are: (a) the ‘silica’ is actually present Titanium–aluminium oxides as a genuine titanium–silicon hybrid oxide which does not crystallise under these conditions, or (b) the phase size of the Two complementary experiments were carried out to investisilica is too small for crystallisation to occur in this system— gate the formation of titanium–aluminium binary oxides.The a recent study has demonstrated that there is a strong size first is the reaction of aluminium trichloride with titanium dependence of crystallisation kinetics in inorganic nanocrys- isopropoxide (Scheme 4) and the second the reverse reaction tals.27 NMR evidence showing the absence of TiMOMSi of titanium tetrachloride with aluminium isopropoxide species in the original amorphous product suggests ration- (Scheme 5).In the first reaction between aluminium trichloride ale (b) may be the most reasonable explanation for the failure and titanium isopropoxide, the TG result confirms that the of the silica to crystallise.high yield obtained was most likely due to trapped solvent or Corriu et al.18–21 calcined a similar titanium–silicon oxide by-product. The IR spectrum was diYcult to interpret. In the at 500 °C for five hours. They also detected no silica via XRD absence of literature data on the assignment of titanium– analysis, only crystallites of anatase.No rutile was found. aluminium binary oxide peaks, all that can be deduced is Elemental analysis of the oxide showed it to be practically general oxide formation. The second reaction appeared to be carbon-free and have a metal content very close to the composi- even more incomplete. The yield achieved was very high and tion of the starting solution. The microchemical analytical the TG mass loss supported this result. The IR spectrum was study by means of an electron-probe analyser, i.e.energy- also diYcult to interpret. All that can be deduced is oxide dispersive spectroscopy, indicated a constant Ti5Si ratio (close formation, with AlMO absorptions24 at 1100 cm-1 and to unity), which is consistent with homogeneity at the 800 cm-1.A slight shift to a higher frequency in the tentatively micrometer level. Analysis of our sample via EDX (Fig. 9) assigned TiMO stretching band at #1000 cm-1 might indicate agreed qualitatively with this finding. Homogeneity on an AlMO substitution.30 Longer reaction times are undoubtedly atomic scale would lead to a true binary oxide on crystallisation required to force the titanium–aluminium reactions to cominstead of a mixture of individual oxides (TiO2 and SiO2); pletion and much more work would be needed to fully however, the empirical formula predicted for our product is characterise the products.This was not justified in the context TiSiO4 which is not a known compound.22 For pure titania of the present work.systems synthesised via the non-hydrolytic route, Corriu et al.15 concluded that compared to the conventional sol–gel route,28 Conclusions the crystallisation of titania is delayed as is the anatase to rutile transformation.29 In our titanium–silicon system, the The non-hydrolytic sol–gel route to inorganic oxides has been XRD results show that crystallisation of titania phases, in extended to the synthesis of a series of binary oxides based on particular rutile, is the primary crystallisation process, with no silicon, aluminium and titanium.In principle, one advantage evidence for formation of a crystalline binary oxide. of this route over the well established hydrolytic route is that The low magnification SEM shows the general, crazed reactivity diVerences between diVerent ‘metals’ are reduced, morphology of the surface of this oxide.Again, a lot of loose therefore oVering the promise of more homogeneous binary material is present. At a higher magnification, the faceted systems. Since many of the starting inorganic halides and nature of the crystallites is highlighted (Fig. 8). The EDX alkoxides are liquids, this oVers the prospect of carrying out spectrum (Fig. 9) shows the presence of both silicon and the reactions under solvent-free conditions, albeit paying due titanium as well as oxygen. respect to the high reactivity of some of these systems! Binary For the reaction of TEOS with titanium tetrachloride, no systems based on aluminium–silicon, titanium–silicon and catalyst was necessary owing to the high reactivity of the titanium–aluminium combinations were studied.transition metal halide. At 110 °C, this reaction reached gel- As expected, reactions were generally rapid in the absence ation rapidly. The TG result suggested that the high yield of a solvent and the presence, where necessary, of a catalyst obtained was due to the presence of unreacted precursors.The [iron(III ) chloride]. Formation of true aluminosilicates was IR spectrum suggested binary oxide formation, with a demonstrated, although in one case it was shown by XRD SiMOMTi stretching vibration at 960 cm-1 in addition to that the crystalline product resulting from calcination had a TiMOMTi elongation bands30 at 1250 cm-1 and 725 cm-1 structure (mullite) diVerent from that expected solely from the and SiMOMSi stretching absorption bands2,31,32 at reaction stoichiometry. The extent to which this resulted from 1220 cm-1, 810 cm-1 and 525 cm-1.Corriu et al.19 have network rearrangement processes during calcination was not performed a similar reaction but used silicon isopropoxide determined. At the calcination temperature of 1000 °C, thermo- instead of TEOS.They formed a gel in five days at 40 °C, with dynamics control the crystallisation of mullite. Amorphous a 100% oxide yield. A 10% mass loss was seen up to 1200 °C. material such as silica was also present. In the case of titanium– XRD, determined after five hours heat treatment of their silicon systems, there was no clear spectroscopic evidence for product at 500 °C, showed the crystalline phase to be anatase the initial formation of a binary oxide.The results after and the specific surface area, also after calcination at this calcination showed the presence of only pure crystalline titania temperature, was found to be 590 m2 g-1. In the reaction phases. The presence of silica nanophases could explain the between titanium tetrachloride and TEOS, the condensation lack of silica crystallisation under these conditions.In the case rate needs to be reduced in order to achieve greater reaction of the titanium tetrachloride/TEOS reaction, a lower reaction control, i.e. by reducing the reaction temperature a lower rate might improve product homogeneity. The titanium–alu- reaction rate would avoid premature gelation and improve minium reactions led to incomplete oxide formation and precursor mobility.This should improve the homogeneity of unequivocal evidence for binary oxide formation was not the two components and consequently allow the reaction to proceed further towards completion. obtained. Further optimisation of the reaction conditions and 1238 J. Mater. Chem., 1998, 8(5), 1233–123912 (a) R.J. P. Corriu and D. Leclercq, Angew. Chem., Int. Ed. Engl., more in-depth analysis of the products are required to demon- 1996, 35, 1420; (b) D. Leclercq and A. Vioux, Heterog. Chem. Rev., strate the potential for binary oxide formation in these systems. 1996, 3, 65. Overall, this study has shown that in the silicon, aluminium, 13 R. J. P. Corriu, D. Leclercq, P. Lefe`vre, P.H. Mutin and A. Vioux, titanium series, amorphous binary oxides can in some cases J. Non-Cryst. Solids, 1992, 146, 301. be formed directly via the non-hydrolytic sol–gel reaction, but 14 S. Acosta, R. J. P. Corriu, D. Leclercq, P. Lefe`vre, P. H. Mutin and A. Vioux, J. Non-Cryst. Solids, 1994, 170, 234. only the aluminosilicates appear to form true crystalline binary 15 (a) P. Arnal, R.J. P. Corriu, D. Leclercq, P. H. Mutin and A. Vioux, oxides after calcination at 1000 °C. Optimisation of the initial Mater. Res. Soc. Symp. Proc., 1994, 346, 339; (b) P. Arnal, reaction conditions and/or stoichiometry may lead to improved R. J. P. Corriu, D. Leclercq, P. H. Mutin and A. Vioux, J. Mater. homogeneity in the other systems. Chem., 1996, 6, 1925. 16 E. Guenther and M.Jansen, Chem.Mater., 1995, 7, 2110. 17 J. N. Hay and H. M. Raval, J. Sol-Gel Sci. T echnol., in press. The authors are grateful to the Engineering & Physical Sciences 18 (a) M. Andrianainarivelo, R. J. P. Corriu, D. Leclercq, P. H. Mutin Research Council (EPSRC) and the Defence, Evaluation & and A. Vioux, J. Mater. Chem., 1996, 6, 1665; Research Agency (DERA) for the award of a CASE studentship (b)M.Andrianainarivelo, R. J. P. Corriu, D. Leclercq, P. H. Mutin (to H. R.). We also thank DERA for use of their facilities and and A. Vioux, Chem. Mater., 1997, 9, 1098. for undertaking XRD and SEM analyses. Particular thanks 19 R. J. P. Corriu, D. Leclercq, P. Lefe`vre, P. H. Mutin and A. Vioux, Chem. Mater., 1992, 4, 961. are due to Dr M. Clegg and Dr D.Porter (both DERA) for 20 S. Acosta, R. J. P. Corriu, D. Leclercq, P. H. Mutin and A. Vioux, helpful discussions and suggestions. We also thank the Mater. Res. Soc. Symp. Proc., 1994, 345. University of Durham for provision of solid state 29Si NMR 21 M. Andrianainarivelo, R. J. P. Corriu, D. Leclercq, P. H. Mutin services. and A. Vioux, J. Mater. Chem., 1997, 7, 279. 22 H.Dislich, Angew. Chem., Int. Ed. Engl., 1971, 10, 363. 23 I. Iwasaki, S. Yasumroi, S. Shibata and M. J. Yamane, Sol-Gel Sci. T echnol., 1994, 2, 387. References 24 T. Lo�pez, M. Asomosa, L. Razo and R. Go�mez, J. Non-Cryst. 1 C. L. Bird and A. T. Kuhn, Chem. Soc. Rev., 1981, 10, 49. Solids, 1989, 108, 45. 2 C. J. Brinker and G. W. Scherer, T he Physics and Chemistry of 25 L. Bourget, R. J. P. Corriu, D. Leclercq, P. H. Mutin and A. Vioux, Sol–Gel Processing, Academic Press, London, 1990. in First European Workshop on Hybrid Organic-Inorganic 3 D. Avnir, Acc. Chem. Res., 1995, 28, 328. Materials, ed. C. Sanchez and F. Ribot, Paris, 1993, 305, 308. 4 B. M. Novak, Adv. Mater., 1993, 5, 422. 26 (a) F. Babonneau, New J. Chem., 1994, 18, 1065; (b) P. Prasad and 5 J. Livage et al., Prog. Solid State Chem., 1988, 18, 259. B. A. Reinhardt, Chem.Mater., 1990, 2, 660. 6 C. D. Chandler, C. Roger and M. J. Hampden-Smith, Chem. Rev., 27 C.-C. Chen, A. B. Herhold, C. S. Johnson and A. P. Alivisatos, 1993, 93, 1205. Science, 1997, 276, 398. 7 L. Hench and J. K. West, Chem. Rev., 1990, 90, 33. 28 (a) K. Kamiya, K. Tanimoto and T. Yoko, J.Mater. Sci. L ett., 1986, 8 W. Gerrard and P. F. GriVet, Chem. Ind., 1959, 55. 5, 402; (b) M. Aizawa, Y. Nakagawa, Y. Nosaka, N. Fujii and 9 (a) S. Acosta, P. Arnal, R. J. P. Corriu, D. Leclercq, P. H. Mutin H. Miyama, J. Non-Cryst. Solids, 1990, 124, 112. and A. Vioux, Mater. Res. Soc. Symp. Proc., 1994, 346, 43; 29 P. Arnal, R. J. P. Corriu, D. Leclercq, P. H. Mutin and A. Vioux, (b) R. J. P. Corriu, D. Leclercq, P. Lefe`vre, P. H. Mutin and Chem. Mater., 1997, 9, 694. A. Vioux, J. Mater. Chem., 1992, 2, 673; (c) R. J. P. Corriu, 30 M. Schraml-Marth, K. L. Walther and A. Wokaun, J. Non-Cryst. Solids, 1992, 143, 93. D. Leclercq, P. Lefe`vre, P. H. Mutin and A. Vioux, J. Sol-Gel Sci. 31 Analysis of Silicones, ed. A. L. Smith, Wiley-Interscience, New T echnol., 1997, 8, 89. York, 1974. 10 (a) W. Gerrard and J. V. Jones, J. Chem. Soc., 1952, 1690; 32 R. K. Iler, T he Chemistry of Silica, Wiley, New York, 1979. (b) W. Gerrard and K. D. Kilburn, J. Chem. Soc., 1956, 1536. 11 (a) A. Zappel, J. Am. Chem. Soc., 1955, 77, 4228; (b) R. Schwartz and W. Kucher, Chem. Ber., 1956, 89, 169. Paper 7/07549I; Received 20th October, 1997 J. Mater. Chem., 1998, 8(5), 1233–1239