J. MATER. CHEM., 1994, 4( ll), 1755-1761 Characterization of Silicated Anatase Powders Li Yi,+ Gianguido Rarnis, Guido Busca and Vincenzo Lorenzelli lstituto di Chimica, Facolta di Ingegneria, Universita, P. le Kennedy, I- 16129 Genova, Italy Silicated titanias have been prepared by reaction of commercial anatase-based Ti0, powders with tetraethylorthosilicate, followed by calcination. At low loadings orthosilicate species are found, that interact mainly with surface defects of anatase where surface hydroxy groups are located. At higher loadings polysilicate species are found, while amorphous silica is observed if the nominal loading is comparable to that allowing the coverage of the entire TiO, surface with a silicate layer. Surface silicate species progressively mask the TiO, surface, and do not generate Bransted acidity that is sufficiently strong to protonate pyridine.Silication strongly hinders anatase crystal growth and loss of surface area upon heating, as well as the anatase-to-rutile phase transformation. Thus, supported silica in small amounts can act as an efficient morphology and structure promoter of anatase Ti0,-based catalysts, although it also perturbs the TiO, surface chemistry significantly. Anatase is largely used in heterogeneous catalysis as the support of vanadia catalysts for the selective catalytic reduction of NO, by ammonia (SCR process') and for the selective oxidation of o-xylene to phthalic anhydride.2 Moreover, it is used as the support of molybdena-based catalysts for hydrotreatments and for the Claus proce~s.~ Anatase is also a good photocatalyst, and is better than r~tile.~However, anatase is a metastable form of TiO,, the stable polymorph being rutile at all temperatures at 1 atm.5 Anatase tends to transform into rutile and this phase trans- formation is accelerated by some catalytically active oxides like vanadium o~ide.~,~ The anatase-to-rutile phase transform- ation is much faster for high-area anatase than for low-area anatase',' and is accompanied by a dramatic decrease of surface area.Furthermore, rutile-based catalysts are generally less active than anatase-based catalyst~.~,~* The anatase-to- rutile phase transformation is therefore the main cause of deactivation of vanadia-titania catalysts." This is probably why the supports of SCR catalysts (characterized by high surface areas, 50-loom2 g-') also contain W03, which strongly inhibits both anatase sintering and the anatase-to- rutile phase transf~rmation,~"~~'~ besides having other favour- able effects.Most industrial SCR catalysts also contain sil- iceous materials. Much has been reported on SCR catalysts constituted by vanadia supported on silica-titania mixed o~ides.~~'~~Bulk silica-titanias prepared by coprecipitation techniques have also been investigated in detail in relation to their own catalytic activity in acid-type reaction^'^,'^ and as catalysts for the oxidation of halogenated organic com-pounds." Interest has also been devoted to titania-on-while little data has been reported, to our knowledge, on silica-on-titania or silicated titania, i.e.anatase with small amounts of silica deposited on its surface.22 This paper reports our data on a study of the solid-state chemistry of anatase-based TiO, commercial preparations doped with small amounts of silica, and on the effect of silica on the surface properties of commercial anatase. Experimental Commercial Ti02 powders, P25 from Degussa (Hanau, Germany) and Eurotitania from Tioxide (Billingham, UK) were used. A home-made anatase sample and amorphous silica (Aerosil 130 from Degussa) were also used for comparison. t Permanent address: MMEI Tianjin Institute of Reprographic Technology, Tianjin 300131, Peoples' Republic of China.Silicated samples were prepared by a method similar to that proposed for the preparation of silicated al~minas.~~ The support was immersed in solutions of tetraethylorthosilicate [TEOS, Si(OC,H,),] in butan-2-01. After impregnation of the support the solvent was evaporated and the solid was dried. The dried samples were then calcined in air at 773 or 1023 K for 3 h. Samples were also characterized after differential thermal analysis-thermogravimetry (DTA-TG) runs up to 1240K. The samples are hereinafter denoted by E (Eur otitania support) or D (Degussa support) followed by the nominal percentage of SiO, (e.g. E2 means 2% SiO,-TiO, with Eurotitania support). Table 1 summarizes some characteristics of the Ti02 supports and of the silicated samples.Blank experiments with the supports treated with butan-2-01 have been performed. No other detectable differences we1 e found in the behaviour of the supports. FTIR spectra were recorded using a Nicolet SZDX instru- ment equipped with conventional gas manipulation 'outgas- sing lines. For IR measurements, samples precalcined at 773 K were pressed into self-supporting discs of appropriate thick- nesses and heated in air at 620K in the IR cell for 30min, and finally evacuated at the same temperature for 1 h. I'yridine (1 Torr, 5 min) and CO (50 Torr, 5 min) were adsorbed from the gas phase. The outgassing time at the given temperatures was 5 min. FT Raman spectra were recorded using a Brucker RFSlOO (Nd:YAG laser). Simultaneous TG-DTA experiments were performed with a Setaram TGA92, while XRD patterns were recorded using a Philips PW1130-1049/10, with a Co-Kcr source.Crystal sizes were measured by the Scherrer and the anatase-to-rutile ratios were calculated u\ing the Spurr and Myers formula.25 Adsorbates and reagents were hyperpure commercial prod- ucts from SIO and Carlo Erba (Milano, Italy). Results Solid-state Characterization In Table 1 the crystal phase composition of our samples, deduced by XRD analyses, is reported. The Eurotitaniii sample as obtained is rutile-free, but its XRD pattern shows, together with the peaks typical of anatase (JCPDS Table no. 2 1-1272), also weak diffractions from brookite (JCPDS Table no. 29-1360). Instead, the Degussa sample as received is a mixture of anatase and rutile (JCPDS Table no.21-1276). As shown in Table 1, EO is still stable in the anatase form (with brookite) after calcination at 773 K but is completely converted to rutile by heating at 1023 K. The anatase-to- J. MATER. CHEM., 1994, VOL. 4 Table 1 XRD phase composition, surface areas (S/m2 g-') and crystal sizes (r,/A)of titania and silicated titania samples" sample Yo SiO, T, XRD S r €0 0 300 Ab 125 86 770 Ab 78 236 1020 R 2 452 1240' R >500 E2 2 770 Ab 119 86 1020 Ab 67 157 1240' A 91% 248 R 9% 21 1 El0 10 770 Ab 99 82 1020 Ab 102 82 1240' Ab 128 DO 0 300 A 68% 55 230 R 32% 317 770' A 64% 40 225 R 36% 328 1020 A 12% 15 315 R 88% 413 1240' R 365 D0.5 0.5 7 70 A 66% 49 215 R 34% 297 1020 A 68% 45 248 R 32% 339 1240' A 32% 295 R 68% 328 D1.3 1.3 770 A 68% 51 201 R 32% 288 1020 A 68% 48 210 R 32% 297 1240' A 61% 255 R 39% 306 D2.5 2.5 770 A 68% 51 205 R 32% 288 1020 A 68% 47 210 R 32% 279 1240' A 66% 220 R 34% 297 D5 5 770 A 69% 50 205 R 31% 288 1020' A 68% 48 205 R 32% 288 a T,=calcination temperature; A =anatase; R =rutile.Traces of brookite. Samples after DTA-TG runs. rutile phase transition during DTA runs is evident at 1030 K from the detection of a distinct exothermic peak (Fig. 1). This peak is not observed at all in the case of the silicated Eurotitania samples. Accordingly, XRD analyses show that the rutile phase is not present in sample El0 either after DTA runs or after calcination at 1023 K for 3 h.Sample E2 is still rutile-free (with brookite impurities) after calcination at 1020K, but after DTA runs it contains 9% rutile and no brookite. It is known that brookite is converted thermally into rutile, not into anatase26 and faster than anata~e,~~ so rutile should mostly be formed by brookite transformation, while anatase is stable. This allows us to conclude that the brookite content in Eurotitania TiOz is 69%. Analysis of the behaviour of the E samples suggests that silica stabilizes both anatase and brookite with respect to their phase transformation to rutile.However, in the case of E2, brookite transforms to rutile rather than anatase. The pure Degussa sample DO shows almost no conversion upon calcination at 770 K, while it is only partly converted to rutile after calcination at 1020 K. As shown previ~usly,~ the Degussa sample is more resistant to phase transformation with respect to other materials, in spite of the presence of rutile crystals, due to its relatively low surface area and small porosity. This is also evident from Fig. 1, which shows the r 673 873 1073 1273 TIK Fig. 1 DTA traces for samples EO (a),E2 (b),DO (c)and D2.5 (d). Heating rate 10 K min-' in air. DTA peak associated with the anatase-to-rutile phase trans- ition near 1173 K with respect to 1030 K for EO.D0.5 is partly converted at 1240 K (DTA runs) while none of the other silicated Degussa samples are converted signifi- cantly even at 1240 K. Thus, the data on D samples confirm that silica also strongly inhibits the anatase-to-rutile phase transformation in the presence of rutile. To obtain more information on the state of supported silica species we also studied the skeletal IR spectra with KBr pressed disks. All the spectra present the typical absorption of TiOz in the region 1000-400 cm-', as discussed previ~usly.~~ However, silica-containing samples show additional absorption in the region 1300-400 cm- '. These absorptions, after subtraction of the spectrum of the pure TiO, support, are compared in Fig. 2 with the spectrum of amorphous silica for the D samples.The spectra observed are analogous with those of the E samples. D5 shows a strong band centred at 1065 cm-' with a definite shoulder near 1200 cm-'. Additional shoulders are very evident at 940 and 895cm-' together with a very broad band centred near 740cm-'. The spectra of D2.5 and D1.3 no longer show the shoulder near 1200 cm-', while the components in the region 1000-900 cm-I are more intense 1400 1000 600 wavenurnberIcm-' Fig.2 FTIR spectra of silicate species in D0.5 (a), D1.3 (b),D2.5 (c) and D5 (d)samples (KBr pressed disks, anatase spectrum subtracted) (e)skeletal spectrum of amorphous silica J. MATER. CHEM., 1994, VOL. 4 with respect to the highest-frequency maximum that also shifts to 1030 cm-l.Moreover, the broad band near 740 cm-' decreases for D2.5 and is absent for D1.3 and D0.5. The features in the 1300-1000cm-' region observed for D5 correspond rather well to those of the envelope due to the Si -0-Si asymmetric stretching modes of amorphous silica.28 However, amorphous silica does not present shoulders between 1000 and 850cm-' and shows a sharp band at 800 cm-' due to the Si-0-Si symmetric stretching/bending modes. Thus, while amorphous silica is probably present as a separate phase in D5 (as well as in DlO), it is absent in the samples of lower silica content. D2.5 the persistence of the broad band at 740 cm-', associated with the symmetric stretching of Si- 0-Si bridges, indicates that polysilicate anions are present.The position and broadness of this band allow us to exclude pyrosilicates as predominant species29 while suggesting pyroxene-like chains3' or layer silicate-type sheets31 as predominant in D2.5 but present also in D5. For D1.3 and D0.5 the absence of the band in the 800-700 cm-' region strongly supports the idea that only orthosilicate species are present,32 the band with components at 1030,920 and 880 cm-' being associated with the stretching of terminal Si-0 bonds of isolated sio44-. In conclusion, FTIR spectra suggest that on D0.5 and D1.3 isolated orthosilicate anions exist, on D2.5 polysilicates pre- dominate, while on D5 and D10 together with polysilicates, bulk silica particles are also present. These data can be discussed taking into account the amount of silica added to titania in the samples, in relation to the surface areas.The area available for each Si atom (assuming that they ade homogeneou!ly distributed a! the surface) is aboyt 100 A2 for D0.5, 38 A2 for D1.3, 20 A2 for D2.5 and 10A2 for D5. These areas can be compared with the areas occupied by isolated and polymeric silicate anions. While the arca occupied by an isolated orthosilicate anion is less than 3 A2, in the sheets of layer silicateso like kaolinite there is no more than one silicon atom per 11 A2.33Previously, ImFmura et a[.'' evaluated the area of a silicate unit as 25.4 A2, by comparison with that occupied in the cristobalite structure. However, the density of the most stable silica polymorphs is far lower than that of the silicate sheets of layer silicates.The area available for silicate species in D5 and D10 necessarily implies the formation of polymeric species. On the contrary, the area available for each silicate species in D0.5 and D1.3 allows their dispersion as isolated ions. Obviously, their dispersion implies that the interaction of silicate species with the titania surface is stronger than the mutual interaction between silicate species. For D2.5, the area theoretically available is only twice that of layer silicates, in accord with the presence of both isolated and polymeric silicate anions. Fig. 3 shows the Raman spectra of the Si0,-Ti02 samples and of the corresponding Ti02 supports.As discussed else- where,27 the Raman spectrum of anatase shows strong peaks at 639, 516,397 and 144 cm-' and a weak peak at 197 cm-'. Additional peaks are observed at 590, 550 (broad shoulder), 452, 366, 322, 290, 246 and 210cm-' in the pattern of the Eurotitania support, due to brookite. For the Degussa sample the peak at 447 cm-' and a shoulder at 610 cm-' are indica- tive of the presence of rutile. The spectra of the Si02-Ti02 samples do not differ qualitatively from those of the corre- sponding support. No new bands associated with silicate species are found. Correspondingly, amorphous silica gives an extremely weak and broad Raman pattern.21 However, the intensity of the Ti02 patterns is significantly reduced by the addition of silica.The intensity of the peak at 144cm-' for El0 is 75% of that of EO, while for D10 it is about 50% of that for DO. This weakening appears to be roughly pro- portional to the amount of loaded silica. According to previous 6 3 0 800 600 400 200 wavenumber/cm-' Fig. 3 FT Raman spectra of EO (a),El0 (b),DO (c), D1.3 (d I, D2.5 (e) and D5 (f)[spectra (c) and cf) are repeated using an expantled scale] data2' this could be associated with increased disorder at the surface. Morphology Characterization Table 1 shows the surface areas and the crystal sizes of the SO2-Ti02 samples and of the corresponding Ti02 hupports, as prepared or after calcination. The surface area of the Eurotitania sample is decreased to less than 30"/0 upon calcination at 773 K, when the anatase and brookile phases are still stable, but the anatase crystal size has gropn signifi- cantly.The surface area drops to 2 m2 g-' after calcination at 1020 K, when the sample is completely converted to rutile. The addition of silica strongly hinders the surface area loss, together with the anatase-to-rutile phase transition. For El0 the surface area remains near 100 m2 g-' even after calcination at 1020 K. Correspondingly, the sintering of anatahe is also inhibited. El0 retains the same crystal size of the starting support even after calcination at 1020 K. The area available for each silicate anion on the El0 s?mple, which appears to be morphologically stable, is just 10 A2, which is very similar to the area per silicate ion in the sheets of layer silicates. The IR spectra show that silica particles can be present together with polysilicates in this sample, whose loading nearly corre- sponds to that of D5.As discussed elsewhere, the Degussa sample, although it already contains rutile, is converted to rutile more slowly than Eurotitania owing to its much lower surface area arid higher crystal size.7 Correspondingly, it also retains a highcr surface area after calcination at 1020K (15 m2 g-'). Also, in this case, the addition of silica strongly hinders the surface-area loss and anatase crystal growth. D1.3 and D2.5 appear to be almost morphologically stable upon calcination at 1020K, although the silica loa4ing relative to the surface area is nearly half fp-D2.5 (20 A2 per Si atom) than that of the El0 sample (10 A2 per Si atom).J. MATER. CHEM., 1994, VOL. 4 Surface Characterization Surface Hydroxy Groups Fig. 4 shows the spectra of the surface hydroxy groups of DO and of the corresponding SO,-TiO, samples, after activation at 623 K in vacuum. The spectrum of DO exhibits several components: 3734, shoulder, 3713,3690,3658 and 3641 cm-'. These components are due to the different coordination states of different OH groups on anatase (and on rutile), as discussed previou~ly.~~We previously showed that anatase samples with smaller surface areas give much simpler OH ~pectra,~,~~ showing that most OH groups of high-area anatase are located on the corners, edges or surface defects.The spectrum of D0.5 is very different from that of DO, showing one band only, centred at 3738 cm-', with a shoulder near 3745 cm-' and a tail at lower frequency. This band does not correspond to that of low-area anatase samples, observed at 3680 cm-'.7*34 By increasing the silica loading, the intensity of this band increases progressively, while the maximum also progressively shifts towards 3744 cm-' as measured for D10. The same frequency was observed by us under the same conditions for the silanol groups of pure amorphous silica. This allows us to assign the main band in all silica-containing samples to OH-stretching of surface silanol groups. The shift towards lower frequency by decreasing the silica loading agrees with the previous data reported for coprecipitated ~ilica-titanias.~~This shift is likely to be associated with the higher electron densities at silicon atoms in orthosilicate anions with respect to polysilicates and framework silica (because of the more ionic nature of SiO-Ti bonds with respect to SiO-Si bonds), although the dependence of the OH stretching frequency on the electron densities on atoms and on its own acidity is a complex matter, as already remarked upon.36 Note that in D0.5 the hydroxy groups of titania appear to be completely removed, so a very complex TiOH pattern is substituted by a single SiOH band.This apparently points to the complete dispersion of the orthosilicate anions, probably on surface defects, and to the role of the surface hydroxy groups of TiO, as reaction sites with respect to TEOS, giving rise to silicate species.Adsorption of Carbon Monoxide In Fig. 5 the spectra of carbon monoxide adsorbed at room temperature on the supports DO and EO, and on a third TiO, anatase sample that, according to XRD and Raman analyses, appears to be both rutile-free and brookite-free are reported. The samples were outgassed at 623 K before adsorption. The 3800 3750 3700 3650 3600 wavenurnbedcm-' Fig.4 FTIR spectra of the surface hydroxy groups of DO (a), D0.5 (b),D1.3 (c), D2.5 (d)and D5 (e)samples, all outgassed at 623 K 2300 2200 2100 2000 wavenurnber/cm-' Fig. 5 FTIR spectra of the surface carbonyl species arising from CO adsorption on DO (a),a home-made pure anatase sample (b)and EO (c) spectra show in all cases a band split between a main component at 2188 cm-' and a minor one at 2206 cm-'.This split band is typical of CO adsorbed on anata~e,~~.~~,~~in contrast to rutile that shows only one band near 2185 cm-'. The comparison of the three samples, one of which is pure anatase, with the others containing rutile (DO) or brookite (EO), seem to suggest that brookite does not have a significant effect on the nature of the active sites. On the contrary, the stronger intensity of the lower-frequency component with respect to the higher-frequency one on DO related to the other samples could be associated with the additional effect of rutile surface in this sample. A third band is observed in the three cases near 2105 cm-'.As discussed elsewhere, this band is associated with CO adsorbed on reduced Ti3 centre^?^,^' This band is not reported by several authors reporting the CO adsorption on titania~~',~~ because they pretreated the sample in oxygen and, consequently, worked with nearly stoichiometric TiO,. Its intensity cannot be compared with that of the higher-frequency band: in fact it is well known that the C-0 stretching of coordinated C(.) on reduced centres gains much intensity when n-type back-donation occurs from d electrons of the metal centre, also causing a lowering of the C-0 stretching frequency, with respect to the gas-phase value (2143cm-I). So, this band is probably associated with a very small number of reduced centres, in spite of its remarkable intensity.In Fig. 6 the spectra of CO adsorbed on DO, D0.5, D1.3 and D2.5 previously outgassed at 623 K are compared. On all three silica-containing samples the higher-frequency com- ponent is present as a poorly resolved shoulder, probably because its maximum is slightly shifted downwards, to 2202 cm-', while the band at 2188 cm-' decreases sharply in intensity, almost disappearing in D2.5. The band at 2105 cm-' has a different behaviour: it grows in intensity and shifts significantly upwards in DO, D0.5 and D1.3, but it almost disappears in D2.5. On D5 and D10 no bands of adsorbed CO are observed. In Fig. 7 the spectra of CO adsorbed on silicated Degussa samples outgassed at 773 K are reported.After this pretreat- ment a band at 2205 cm-' is observed together with the main one, always observed at 2188 cm-', but it is much weaker compared with the relative intensities observed on the pure support. Moreover, the intensity of the band at lower frequen- cies, associated with CO adsorbed on Ti3+ centres, is stronger, as expected for a stronger outgassing treatment that can cause partial oxygen depletion with increasing non-stoichiometry of the semiconducting TiO,-, phase. However, in this case the J. MATER. CHEM., 1994, VOL. 4 1.05 wavenum ber/cm-' Fig. 6 FTIR spectra of the surface carbonyl species arising from CO adsorption on DO (a), D0.5 (b), D1.3 (c) and D2.5 (d); samples previously outgassed at 623 K a, t 95: a I I I I 1 I 2300 2200 2100 2000 wavenum ber/crn-' Fig.7 FTIR spectra of the surface carbonyl species arising from CO adsorption on D0.5 (a), D1.3 (b) and D2.5 (c); samples previously outgassed at 773 K intensity of this band is not as dependent on the silica loading, but its position is dependent on the loading (it progressively shifts from 2105 to 2125 going from DO to D2.5, see Fig. 6 and 7). These reduced species were also observed on outgassed mixed silica-titanias, and were characterized by a split band at 2118 and 2130 cm-'.35 The upwards shift of this frequency with increasing silica loading may be due to a higher cationic charge on these reduced sites when they are surrounded by silicate ions than when they are surrounded by the more basic oxide ions.These data can be interpreted as follows: (i) silication causes the progressive disappearance of Ti4+ sites active in CO adsorption; (ii) the stronger sites, responsible for the band above 2200 cm-', are less affected quantitatively by silication than the weaker sites, responsible for the band at 2188 cm-l, but the corresponding band is slightly shifted downwards; and (iii) silication causes a decreased electron densit!. on the Ti3+ reduced sites produced by outgassing, with a consequent shift of the corresponding band from 2105 to 2125 crr-'. Adsorption of Pyridine Fig. 8 shows the spectra of pyridine adsorbed on the Degussa TiO, support, on the corresponding Si0,-TiO, samples and on pure amorphous SiOz after activation at 623 K in vacuum.In all cases the absence of bands in the regions near 1630 and 1550cm-' point to the lack of sufficient Brsnsted ataidity to protonate pyridine. On the contrary, the positions of the 8a (shifting from 1603 to near 1610 cm-' during outgassmg) and 19b (1443-1447 crn-') components provide evidence (or pyri- dine species coordinated on Lewis-acid sites. However, the intensities of these bands strongly decrease with increasing silica loading without evident shifts, being very weak already for the sample D2.5. However, another 8a component grows, close to that near 1610 cm-', with increasing silica loading, at 1595 cm-l. This species is also associated with the forma- tion of strong absorption in the region 2500-4000~m-~ at the expense of the sharp OH band near 3740 cm-', .issigned above to silanol groups (Fig.9). It is evident that the silanol groups associated with the surface silicate species interact with pyridine via hydrogen bonding. Their Brsnsted strength is sufficient to interact rather strongly with pyridine but is not sufficient to cause proton transfer to the adsorbed base. The OH stretching band of the silanol-pyridine H-bonded complexes is observed near 3100 cm-' on silicated tit;mia and near 2900 cm-' on silica (Fig. 9). This should indicate that the interaction is stronger on silica than on silicated titania, thus providing evidence of a lower Brsnsted acid strength of the surface silanols of silicated titania with respect to ,imorph- ous silica.It is interesting to compare these results with those reported on coprecipitated silica-titania with silica contents higher than 50%17 that show that H-bonding oi surface silanols with pyridine result in the formation of very strong H-bonds, with the OH stretching band being split due to its Fermi resonance with the first overtone of the in-plane Si- OH deformation, at 2700 and 2300 cm-l. This indicates that coprecipitated silica-titanias are stronger Brsnsted solid acids than silicated titania and silica. Upon interaction of the sample with pyridine, the spectrum of the adsorbant in the region 1300-1000 cm-' is pcrturbed, as shown by the negative bands appearing in the suhtraction spectra in the region 1300-1000 cm-' (Fig.10).These negative bands correspond to absorptions in the spectrum of the activated sample that disappear (or are strongly displaced) r wavenurnbe r/cm-' Fig.8 FTIR spectra of adsorbed pyridine on DO (a) (-1, D0.5 (b) (---), D2.5 (c) (---) and amorphous silica (d) (-.-), all previously outgassed at 623 K 4000 3200 2400 wavenum berkm-' Fig. 9 FTIR spectra of DO (a), D0.5 (b),D1.3 (c), D2.5 (d) and silica (e), outgassed at 623 K (-) and after pyridine adsorption (---). The transmittance scale for the spectra in (a) is twice that of the other spectra. upon pyridine adsorption. For D0.5 this negative band is broad, centred near 1100 cm-', while for D1 it is centred near 1150 cm- '.For D2.5 this band is split, with components at 1173 and 1034 cm-'.These features are still present when the band at 1595cm-' has disappeared. This indicates that they are due to species that are perturbed upon Lewis-type coordinative interaction on Ti4+ sites. Accordingly, these bands can be assigned neither to the Si-0 stretching nor to the Si--OH in-plane deformation of surface silanol groups. In fact, it seems that these modes fall at 980cm-' and near 800 cm-' for the silanol groups of silica.39 It is unlikely that these modes can be strongly shifted in the case of silicated titania with respect to bulk amorphous silica. Thus, these bands are associated with the stretchings of terminal Si- O-Ti4+ bonds in orthosilicate or polysilicate species. Pyridine coordination on Ti cations increases their electron densities and allows the Si-O(Ti) bonds to relax, so their stretching frequency shifts down.Conclusions (i) Silication strongly stabilizes anatase phase with respect to its conversion to rutile. (ii) Silication inhibits anatase sintering and loss of surface area. (iii) At small loadings silication of anatase results in the formation of isolated orthosilicate species. At higher loadings J. MATER. CHEM., 1994, VOL. 4 1300 1100 wavenum berkm-' Fig. 10 Perturbations, arising from pyridine coordination, of the FTIR spectra of D0.5 (a), D1.3 (b)and D2.5 (c), prebiously outgassed at 623 K. (-) Outgassing at room temperature, after pyridine adsorption; (---) outgassing at 423 K; (-*-) outgassing at 623 K.The upwards sharp bands near 1235,1219,1150,1070 and 1040 cm-' are due to coordinated pyridine, while the downwards broad bands are due to perturbed Si-0 stretching modes. polysilicate species, possibly pyroxene-like chains or layer silicate-type sheets are formed. (iv) When the amount of loaded silica is increased to a value similar to that corresponding to coverag: of the whole titania surface with a layer silicate (about 10 A2 per silicon atom) amorphous silica also appears. (v) The surface silicate species cause a decrease in the intensity of the anatase Raman pattern, possibly associated with increased disorder at the surface. (vi) Isolated orthosilicate species contain surface silanol groups whose OH stretching is observed at lower frequency than those of polysilicate species and of silica (3738 us.3744 cm-') (vii) Silicate species are coordinated to Lewis-acidic Ti4+ sites and are perturbed by pyridine coordination on them. (viii) Silication causes the progressive destruction of the active sites for CO and pyridine adsorption on Ti02. (ix) Silicated titanias do not have sufficiently strong Brsnsted acidity to protonate pyridine. Surface silanol groups of silicated titanias are more active in hydrogen bonding than TiOH groups of titania, but have less Brarnsted acidity than silanols of amorphous silica and of coprecipitated silica- titania. (x) The characteristics of TiOz surface are almost complete!y destroyed when the surface silicon density approaches 20 A2 per Si atom, corresponding to half the density of layer silicate sheets.These data indicate that silica deposited on the anatase surface acts as a strong morphology and structure stabilizing agent. However, silica also modifies significantly the surface chemical properties of anatase. This work was supported by MURST, Rome, Italy. L.Y. acknowledges the governments of Italy and of the Peoples' Republic of China for an exchange grant. The authors are indebted to Dr. G. Oliveri for the surface area measurements. J. MATER. CHEM., 1994, VOL. 4 1761 References 20 A. Fernandez, J. Leyrer, A. R. Gonzalez-Elipe, G. Munuera and H. Knozinger, J. Catal., 1988,112,489. 1 H. Bosch and F. Janssen, Catal. Today, 1988,2, 369.21 S. Srinivasan, A. K. Datye, M. Hampden-Smith, I. E. Wachs, 2 M. S. Wainwright and N. R. Foster, Catal. Rev., 1979,19, 211. G. Deo, J. M. Jehng, A. M. Turek and C. H. F. Peden, J. Catal., 3 S. Matsuda and A. Kato, Appl. Catal., 1983,8, 149. 1991,131,260. 4 J. Augustynski, Electrochim. Acta, 1993,38,43. 22 S. Sato, Langmuir, 1988,4, 1156. 5 I. Barin, Thermochemical Data of Pure Substances, VCH, Berlin, 23 F. Buonomo, V. Fattore and B. Notari, US Pat. 401.;589 and 1990, part 11, pp.1546, 1547. 4013590, 1977; G. Manara, V. Fattore and B. Notari, US Pat. 6 D. J. Cole, C. F. Cullis and D. J. Hucknall, J. Chem. Soc., Faraday 4038337,1977. Trans. I, 1976,72,2185. 24 H. P. Klug and L. E. Alexander, X-Ray DiBraction Piocedures, 7 G. Oliveri, G.Busca, G. Ramis and V. S. Escribano, J. Muter. Wiley, New York, 1970. Chern., 1993, 3, 1239. 25 R. A. Spurr and H. Myers, Anal. Chern., 1957,29,760. 8 S. R. Yoganarasimhan and C. N. R. Rao, Trans. Faraday Soc., 26 C. N. R. Rao, S. R Yoganarasimhan and P. A. Faetti, Trans. 1962,58,1579. Faraday Soc., 1961,57,504. 9 1. Gasior, M. Gasior, B. Grzybowska, R. Kozlowski and 27 G. Busca, J. M. Gallardo Amores, P. Piaggio, G. R'rmis and J. Sloczynski,Bull. Acad. Pol. Sci., 1970,27, 829. V. Sanchez Escribano, J. Chem. Soc., Faraday Trans., in the press. 10 I. M. Pearson, H. Ryu, W. C. Wong and K. Nobe, Ind. Eng. Chem. 28 H. H. W. Moenke, in The Infrared Spectra of Minrmls, ed. Prod. Res. Det.., 1983,22,381. V. C. Farmer, The Mineralogical Society, London, 1974.p. 365. 11 V. A. Nikolov and A. I. Anastasov, Ind. Eng. Chem. Rex, 1992, 29 M. Gabelica-Robert and P. Tarte, Spectrochim. Acta Part A, 1979,35A, 649. 31, 80. 30 M. S. Bilton, T. R. Gilson and M. Webster, Spectrochim. Actu, 12 G. Ramis, G. Busca, G. Cristiani, L. Lietti, P. Forzatti and Part A, 1972,28A, 21 13. F. Bregani, Langrnuir, 1992,8, 1744. 31 V. C. Farmer and J. D. Russel, Spectrochim. Acta 1964,20, 1149. 13 L. Lietti, P. Forzatti, G. Busca and E. Giamello, submitted for 32 P. Tarte, Spectrochirn. Acta 1963,19, 25. publication. 33 D. T. Griffen, Silicate Crystal Chemistry, Oxford I rniversity14 C. U. I. Odenbrand, S. T. Lundin and L. A. H. Andersson, Appl. Press, 1992.Catal., 1985,18, 335. 34 G. Busca, H. Saussey, 0.Saur, J. C. Lavalley and V. I orenzelli,15 A. Baiker, P. Dollenmeier, M. Glinski and A. Reller, Appl. Catal., Appl. Catal., 1985,14, 245. 1987,35, 365. 35 C. U. I. Odenbrand, S. L. T. Andersson, L. A. H. Andersson,16 K. Shibata, T. Kiyourd, T. Kitagawa, T. Sumiyoshi and J. G. M. Brandin and G. Busca, J. Catal., 1990,125,541L. Tanabe, Bull. Chem. Soc. Jpn., 1973,46,2985. 36 K. Hadjiivanov, D. Klissurski, G. Busca and V. I orenzelli, 17 C. U. I. Odenbrand, J. G. M. Brandin and G. Busca, J. Catal., J. Chem. Soc., Furaday Trans., 1991,87,175. 1992,135,505. 31 D. J. T. Yates, J. Phys. Chem., 1961,65,746. 18 S. Imamura, S. Ishida, H. Tarumoto and Y. Saito, J. Chem. Soc., 38 E. Garrone, V. Bolis, B. Fubini and C. Morterra, Langniuir, 1989, Faraday Truns., 1993,89,757; S. Imamura and H. Tarumoto, Ind. 5,892. Eng. Chem. Rex, 1989,28, 1949. 39 B. A. Morrow and A. J. McFarlan, J. Phys. Chem., 1992.96,1395. 19 M. G. Reichman and A. T. Bell, Langmuir, 1987, 3, 111; M. G. Reichmann and A. T. Bell, Appl. Catal., 1987,32, 315. Paper 4/02606C; Received 3rd May, 1994