- Email: [email protected]
SiO2 catalysts
Applied Catalysis A: General 325 (2007) 256–262 www.elsevier.com/locate/apcata
Grafting of titanium alkoxides on high-surface SiO2 support: An advanced technique for the preparation of nanostructured TiO2/SiO2 catalysts M. Cozzolino, M. Di Serio, R. Tesser, E. Santacesaria * University of Naples ‘‘Federico II’’, Department of Chemistry, Via Cintia 4, Compl. Univ. Monte S. Angelo, I-80126 Naples, Italy Received 21 July 2006; accepted 7 February 2007 Available online 3 March 2007
Abstract The present paper reports and discusses results on the preparation, characterization and catalytic performances of silica-supported titania (TiO2/ SiO2) catalysts. Samples were prepared by grafting titanium tetraisopropoxide (Ti-OPri)4, dissolved in toluene, onto a silica surface in an N2 atmosphere, followed by steam hydrolysis and calcination. The samples were characterized by chemical analysis, BET surface area measurements, X-ray diffraction (XRD), DR–UV–vis, FTIR and DRIFT spectroscopy analyses, TEM/EDX and NH3-TPD. The results indicated that the grafting preparation method gives rise to very strong Si–O–Ti bonds, that are responsible for high titanium dispersion. In particular, at low Ti loading, titanium species in tetrahedric coordination resulted prevalent on the catalyst surface until the maximum surface monolayer coating reached (2.2 Ti atoms/nm2). The degree of polymerization of Ti species increases with further TiO2 load increases, giving rise to a large amount of octahedrical Ti-sites grafted on SiO2. The effects on the catalytic activity of TiO2/SiO2 catalysts with increasing quantities of TiO2 were also investigated. The catalytic results obtained in the epoxidation reaction of cyclooctene with cumene hydroperoxide showed the significant effect of titanium loading on the physicochemical and reactivity/selectivity properties of the silica-supported titania catalysts. # 2007 Elsevier B.V. All rights reserved. Keywords: Supported titanium oxide; Liquid-phase grafting preparation technique; TiO2 dispersion; Epoxidation reaction
1. Introduction Over the last decades, applications and uses of supported metal oxides on silica have driven research towards the production of catalysts characterized by the presence of high percentage of the metal oxide in a dispersed state (twodimensional surface metal oxide overlayers) [1–4]. An interesting example is given by silica-supported titania TiO2/ SiO2 catalysts that have been considered as advanced support materials substituting pure TiO2. The higher mechanical strength, thermal stability and specific surface area of the supported titania oxides, compared to pure TiO2, have recently attracted much attention and driven interest towards the use of these materials not only as catalytic supports, but also as catalysts through the generation of new catalytic active sites [5– 9]. It is well known that the silica surface is fairly inert. However, the silica surface hydroxyls generally act as
* Corresponding author. Tel.: +39 081674027; fax: +39 081674026. E-mail address: [email protected] (E. Santacesaria). 0926-860X/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2007.02.032
adsorptive/reactive sites because of their hydrophilic character. Thus, the preparation of highly dispersed metal oxides on silica often involves a highly reactive precursor, such as TiCl4 or titanium alkoxides, which readily react with the surface hydroxyls of the silica support [3,9,10]. It is well known that the preparation method, the surface area and the concentration of the surface hydroxyls on the silica surface could strongly influence the dispersion capacity and the maximum surface coverage of surface titanium oxide species on silica. Depending on titanium loading, often indicated as percentage by weight of TiO2, different types of titanium species could be present, ranging from dispersed surface TiOx species, both in tetrahedral and octahedral coordination, to small TiO2 crystallites. The relative amount depends on the preparation conditions and chemical composition (surface Ti densities) [3,11,12]. The physicochemical and reactivity/selectivity properties of oxide catalysts are often a strong function of their structural characteristics. In the past, extensive studies have been carried out on Ti silicalites and TiO2–SiO2 mixed oxides [13]. In both materials, Ti atoms are shown to substitute Si in the silica framework to form tetrahedral TiO4 units, which work as active
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
sites for the epoxidation reaction. The highly dispersed TiO2/ SiO2 supported oxides have also been shown to be active in liquid phase epoxidation reactions [5,6,9]. However, the structural features of supported TiO2/SiO2 oxides and their relationship with the physicochemical and reactivity/selectivity properties are still not understood well due to the lack of fundamental systematic studies. Therefore, this article focuses on the preparation of highly dispersed TiO2/SiO2 catalysts by grafting different amounts of titanium tetraisopropoxide, dissolved in toluene, onto the silica surface. In addition, a detailed characterization study was carried out to collect information about the structural characteristics of the supported titanium species. In this way, the factors that control the maximum surface coverage of silica by titanium oxide were investigated. Furthermore, the epoxidation reaction of cyclooctene with cumene hydroperoxide was employed to investigate the reactivity of highly dispersed TiO2/SiO2 catalysts since this reaction is very sensitive to the surface structure of the titanium sites. The catalytic results obtained could be considered as the starting point in the development of the molecular structure–reactivity/ selectivity relationships of TiO2/SiO2 catalysts. 2. Experimental 2.1. Preparation of catalysts A commercial silica (Grace S432, specific surface area = 280 m2/g, pore volume = 1.02 cm3/g, hydroxyl groups = 0.92 mmol/g) was used as a support. The solids were prepared by contacting the silica, calcined at 773 K for 8 h, with a solution of tetra-isopropoxide ((Ti-OPri)4, Aldrich) dissolved in anhydrous toluene. The reaction was performed in a jacketed glass-reactor of 200 cm3 for 6 h, constantly stirred, at the boiling temperature of toluene (388 K). The solid was filtered off, washed with toluene, dried at 393 K overnight, hydrolyzed with steam and, finally, calcined at 773 K for 2 h. Steam hydrolysis was performed at 150 8C. The amount of adsorbed titanium was determined by the colorimetric analysis suggested by Snell and Ettre [14], by evaluating the quantity of titanium remaining in solution after the grafting reaction. The operative conditions and the adsorption results are listed in Table 1.
257
2.2. Techniques used in catalyst characterization Different techniques were used for catalyst characterization, such as: X-ray diffraction (XRD), TEM/EDX, diffuse reflectance spectroscopic analyses (DRIFT and DRUV), temperature programmed desorption (TPD). Textural analyses were carried out by using a Thermoquest Sorptomatic 1990 Instrument (Fisons Instrument) and by determining the nitrogen adsorption/desorption isotherms at 77 K. The samples were thermally pretreated under vacuum (105 mbar) overnight up to 473 K (heating rate = 1 K/min). Specific surface area (SBET) and pore size distributions were determined by using the BET and Dollimore–Heal methods [15]. XRD analyses were carried out by using a Philips diffractometer. The scans were collected in the range 5–608 (2u) using Cu Ka radiation with a rate of 0.018 (2u)/s. TEM observations were made by using a Jeol Jem 2010 equipped with an EDX probe and sample preparation was as follows: a drop of the dispersion of the milled catalytic powder in isopropyl alcohol was put on a Lacey carbon grid and the dispersant was removed by evaporation at room temperature. Diffused reflectance spectra were obtained on a UV–vis scanning Jasco spectrometer V-550, equipped with an integrating sphere, using BaSO4 as reference. UV–vis spectra were recorded in the diffused reflectance mode (R) and transformed by a magnitude proportional to the extinction coefficient (K) through a Kubelka–Munk function (F(R)). FTIR and DRIFT were recorded by using a Nicolet AVATAR 360 instrument, equipped with an accessory for the diffused reflectance. Samples were powdered and diluted 1% by weight with KBr. Adsorbed water was completely removed ‘‘in situ’’ by heating at 673 K under vacuum (105 mbar). Thermal programmed desorption (TPD) measurements were carried out by using ammonia as the probe molecule. Samples of about 100 mg were first calcined in situ at 673 K for 1 h under air flow in order to remove adsorbed water. After cooling, the samples were subsequently contacted with a stream of Helium (1000 cm3/min), containing 1000 ppm of NH3, until saturation level is reached. The ammonia desorption was performed under Helium flow from 373 to 773 K (heating rate = 10 K/min).
Table 1 List of the catalysts prepared by grafting titanium alkoxide on SiO2 and related properties—amount of the support (SiO2) = 4 g Catalysts SiO2 TS1 TS2 TS3 TS4 TS-M (1 step) TS-D (2 steps) TS-T (3 steps)
Ti(O-Pri)4 initial amount (g)
Anchored metal (mmolTi =gSiO2 )
molTi molTi
– 0.04 0.2 0.4 0.6 1.04 1.04 1.04
– 0.035 0.176 0.352 0.528 0.915 1.184 1.383
– 0.99 1 1 1 1 0.64 0.50
anchored/ initial
wt.% TiO2
BET surface area (m2/g)
Specific pore volume (cm3/g)
mmolNH3 =g
Yields to epoxide (%)
Selectivity (%)
– 0.3 1.4 2.8 4.2 5.7 9.5 11.1
282 – 280 283 – 280 276 278
1.02 – 0.99 0.81 – 0.23 0.26 0.27
– 30 45 84 120 156 211 234
– – 3.8 11.6 – 31.3 – 21.2
– – 7.3 28.2 – 70.8 – 36.3
258
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
2.3. Catalytic tests Epoxidation of cyclooctene (Fluka 98%) with cumene hydroperoxide (Fluka 80% by weight in cumene) was performed in a 100 cm3 three neck glass vessel equipped with a Liebig condenser and an internal cooling coil. The reaction temperature was kept constant, at 355 K, by putting the vessel in a thermostatted bath. To 22 cm3 of cyclooctene, 0.3 g catalyst was added, then 10 cm3 cumene hydroperoxide was slowly added with a syringe. The reaction mixture was kept at 355 K under a nitrogen atmosphere for 3 h. Unreacted hydroperoxide was measured by iodometric titration [16] at the end of the reaction. The epoxide formed was measured with the method suggested by Dearbon et al. [17]. Selectivities were determined as: selectivity = 100 [formed epoxide (mol)/ consumed peroxide (mol)]. 3. Results and discussion 3.1. Catalyst preparation By taking into account the surface monolayer of titanium tetraisopropoxide on silica **(G1 = 1.02 mmolTi/gSiO2), determined in a previous work [18], the catalysts reported in Table 1 indicated with the acronyms TS1, TS2, TS3 and TS4, can be considered as sub-monolayer catalysts. In order to completely cover the silica surface, catalysts with higher TiO2 loading were prepared by repeating the operations of grafting, hydrolysis and calcination. In particular, TS-D and TS-T catalysts were prepared by submitting the monolayer catalyst (TS-M) to the operations mentioned above, one and two times respectively. The related properties are also reported in Table 1. 3.2. Catalyst characterization 3.2.1. Nitrogen adsorption (BET) measurements Both support and prepared samples were characterized by BET analyses. The specific surface areas are reported in Table 1. It is interesting to observe that the specific surface area does not significantly change with titanium loading for all the prepared samples. This confirms the possibility to prepare highsurface supported titania catalysts by grafting. On the contrary, a considerable decrease in the specific pore volume compared to that of silica, was observed for all the sub-monolayer catalysts until the monolayer coverage is reached. No further variations were observed for higher TiO2 loadings. 3.2.2. X-ray diffraction analysis XRD analyses were performed to estimate surface TiO2 dispersion. The absence of crystallites of TiO2 was detected in all the XRD patterns of TiO2/SiO2 prepared samples, resulting mainly amorphous. These results suggested that TiO2 particles on SiO2 are very small and below the detection ˚ ). Only the TS-T sensitivity of the XRD technique (<40 A catalyst showed the characteristic peak of TiO2 (anatase) at 2u = 258. These peaks resulted broad, meaning that crystal size was very small. The mean dimension of the crystals was
measured along with the crystallographic direction 1 0 1, using the Scherrer formula d¯ h k l ¼
Kl Bd cos u
where K is the Scherrer constant, ( the wavelength, Bd the FWHM (full widht half maximum), and u is the Bragg angle. ˚ , a value quite close to the Crystallites size was about 45 A instrumental sensitivity limit. The amount of TiO2 crystalline phase present in TiO2/SiO2 samples were measured using ZrO2 as reference compound [19]. Physical mixtures of TiO2, SiO2 and ZrO2 containing 5% wt. of ZrO2 and increasing quantities of titania (2, 5, 10, 15, 20% wt.) were prepared and underwent XRD analysis. The area ratio of the characteristic anatase peak at 2u = 258 and the ZrO2 peak at 2u = 288 increased in line with increases in TiO2 concentration. This calibration line was used to determine the crystalline TiO2 contents in a sample containing TS-T catalyst and ZrO2 (5 wt.%). The amount of crystalline TiO2 determined on the TS-T catalyst was 1.8%. Further study was made of the influence of calcination temperature on the amount of TiO2 crystalline phase and crystallite size. For this purpose, some TS-T samples were calcined at different temperatures and then submitted to XRD analysis after each step of calcination. The low amount of crystalline TiO2, observed (from 1.8 to 2.6%) also for the catalysts calcined at higher temperatures (from 500 to 800 8C), and the slight changes of crystallite size confirms the stability of Si–O–Ti bonds and the good titanium dispersion achieved as a consequence. 3.2.3. Morphological analyses by TEM and EDX TEM observations and EDX analyses were conducted on TS3 and TS-T catalysts. Both catalysts resulted nearly completely amorphous to the morphological observations as the presence of well-defined crystalline TiO2 was not detected. These results seem to suggest that the grafted titanium remains homogenously dispersed on the silica support, even after repeating the grafting operation of titanium alkoxide on silica three times, followed by steam hydrolysis and calcination. Moreover, EDX analysis gave a value of supported Ti corresponding approximately to that of the chemical analysis. 3.2.4. FTIR spectroscopic analysis 3.2.4.1. Analysis of 3200–2700 cm1 spectral range. Some of the prepared samples were submitted to FTIR analysis before the calcination treatment. In Fig. 1 we report the IR spectra recorded in the range 3100–2800 cm1 for TS1, TS2 and TS3 catalysts. The spectra showed four bands at 2981, 2929, 2880, 2857 cm1, assigned to symmetric and asymmetric vibrational C–H stretching of CH3 groups [20]. The bands corresponding to the vibrational C–C stretching of isopropilic groups and CH3 (1260–1145 cm1), and those characteristic of vibrational C–O stretching (1150–1050 cm1) did not appear in the spectra since these are completely covered by the strong bands characteristic of vibrational Si–O–Si stretching in the range 1000–1300 cm1. The area of infrared signals of the peaks corresponding to CH bonds give a linear trend as a function of
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
Fig. 1. Infrared spectra (spectral range: 3100–2800 cm1) of TS1, TS2 and TS3 samples.
molar ratio Ti/Si for the sub-monolayer catalysts: TS1, TS2, and TS3. This suggests that the grafting stoichiometry remains almost constant, at least for the catalysts with a titanium loading lower than the monolayer coverage. 3.2.4.2. Analysis of spectral range 4000–3500 cm1. Fig. 2 reports the spectra of silica and sub-monolayer TiO2/SiO2 calcined catalysts, recorded in the range 4000–3500 cm1 after calcination. From the figure, it is interesting to observe the decrease in the intensity of the peak at 3747 cm1, attributed to the isolated surface silanols (single Si–OH groups and geminal Si(OH)2 groups). This is due to the consumption of Si–OH hydroxyls owing to the grafting reaction. Fig. 3 contains (as a function of Ti/Si molar ratio) the ratio of peak area recorded at 3747 cm1 of TiO2/SiO2 catalysts and the peak area at the same wavenumber of silica support. As can be seen, no linear correlation was observed, probably due to the decrease of
Fig. 2. Infrared spectra (spectral range: 4000–3700 cm1) of the support (SiO2) and TS1, TS2, TS3 and TS4.
259
Fig. 3. AðOHÞTS catalysts =AðOHÞSiO2 as function of molðTiÞanchored =mol ðOHÞSiO2 .
titanium dispersion since aggregates of the titanium species form during calcination. We found that only for the submonolayer catalysts the stoichiometric values obtained correspond to that of the grafting reaction and resulted about 2.5. 3.2.5. Determination of TiO2 dispersion by DRIFT analysis in the spectral range 1400–700 cm1 DRIFT analyses were performed on the following catalysts: TS1, TS2, TS4, TS-M, TS-D and TS-T. The spectra obtained showed the following signals: A large band with a maximum at 1100 cm1 attributed to the asymmetric vibrational stretching Si–O–Si, characteristic of SiO4 units with tetrahedrical coordination; a shoulder at 1200 cm1, due to the asymmetric vibrational Si–O stretching. A peak at 800 cm1 due to the corresponding symmetric vibrational stretching. A peak at 950 cm1 characteristic of symmetric vibrational Si–O–Ti stretching. The last peak resulted crucial in determining the dispersion of TiO2 over the silica surface, at least from a semi-quantitative point of view. In fact, it is possible to estimate the catalyst dispersion by comparing the area of the band at 950 cm1 with that of the band at 1200 cm1. The ratio between those two mentioned areas, A(Si–O–Ti) and A(Si–O–Si), calculated by a gaussian deconvolution of the four peaks mentioned above, could be an index of the catalyst’s dispersion. In Fig. 4 the A(Si–O–Ti)/A(Si–O–Si) ratio is reported a as function of molar ratio Ti/Si. As can be seen as the Ti/Si molar ratio increases, the A(Si–O– /A Ti) (Si–O–Si) ratio does not increase linearly. This means that as Ti loading increases the dispersion decreases, and only for a low TiO2 quantities we have a homogeneous dispersion. The data in Fig. 4 shows this quantity to be about 0:2 mmolTi =gSiO2 .
260
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
Fig. 4. A(Si–O–Ti)/A(Si–O–Si) as function of molar composition.
Fig. 6. DR–UV spectra: (a) SiO2; (b) TS-M; (c) TS-D; (d) TS-T; (e) TiO2 (anatase).
3.2.6. Diffused reflectance UV–vis spectroscopic analysis All the TiO2/SiO2 catalysts were submitted to DR–UV analysis in order to provide information about the coordination geometry of the Ti cations (the first coordination sphere) and the ligand environment (the second coordination sphere) under various conditions. The DR–UV–vis spectra of all the prepared samples, characterized by increasing Ti loadings, are compared in Figs. 5 and 6. The spectrum of bulk TiO2 (anatase) is also included in both the figures for a useful comparison. The absorption of silica in the range 200–500 nm can be considered negligible. It is well known that the Ti cations in tetrahedric coordination show a typical band at 212 nm, due to the ligand–metal charge transfer between Ti4+ and oxygen ligands, such as –O–H, –O–Si, –O–Ti, or H2O. On the other hand, a shift of the band towards higher wavelenghts (260 nm) is usually observed for the Ti cations in octahedrical environments. The DR–UV–vis spectra showed adsorption bands in the range 210–360 nm. It has been observed that the maximum
ligand–metal charge transfer (LMCT) transitions shift to higher wavelenghts with increasing TiO2 loading (see Fig. 5), suggesting a gradual increase in the polymerization degree of Ti atoms [21]. In particular, the TS-D and TS-T samples prepared by repeating titanium alkoxide grafting, two and three times, showed a broad absorption band with a shape (see Fig. 6) very similar to that of bulk anatase TiO2. This result suggests that, even though the TiO2 crystallites are very small and beyond the detection sensitivity of XRD measurements, they may still possess the same electronic properties as the pure TiO2 anatase phase. In this way, by adopting the grafting preparation method, it is possible to chemically modify the surface of an oxide, without altering the original mechanical properties. Finally, the decrease in the edge energy of LMCT transitions of Ti atoms with increasing TiO2 loading can also be associated to the increase in the number of the nearest Ti atoms, due to the Ti polymerization at higher TiO2 loadings. The observed shift of the band-gap absorption edge may be explained in terms of quantum effects due to the small titania particles [22]. 3.2.7. Temperature programmed desorption (TPD) analyses The catalysts were submitted to TPD analysis by using ammonia as the probe molecule. The results obtained are also reported in Table 1. From the values reported in Table 1, it can be observed that the grafting of titanium species increases the total number of acid sites on the surface of silica, greatly. On the contrary, by considering the related values of temperature (TM) in correspondence with the maximum of the NH3 desorption, the strength of acid sites remains almost constant. In particular, the results suggest the presence of acid sites of medium strength. 4. Catalytic performances in the epoxidation reaction
Fig. 5. DR–UV spectra: (a) SiO2; (b) TS1; (c) TS3; (d) TS4;(e) TS-M; (f) TiO2 (anatase).
It is well known that the catalytic performance of titanium oxide appears to be completely modified by interaction with the silica support, which is associated with the changes in the
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
molecular structure and coordination environment. The particular catalytic performances of the highly dispersed TiO2/SiO2 catalysts, in comparison with pure TiO2, were previously demonstrated for liquid-phase epoxidation reactions with hydrogen peroxide and alkyl hydroperoxide [6]. Pure TiO2 is not active in the epoxidation reaction, while the highly dispersed TiO2/SiO2 catalysts exhibit high reactivity and high selectivity to epoxide [6]. Two fundamental requirements for good catalytic performance are: high dispersion of the catalytically active component within the matrix (site isolation) and stability to leaching [23–25]. In the present work, the catalytic performances of TiO2/SiO2 prepared catalysts were tested in the epoxidation of cyclooctene with cumene hydroperoxide. Cyclooctene is an interesting test reagent because of the high stability of the corresponding oxide. The results obtained are summarized in Table 1 where yields to epoxide (%) and related selectivities are reported together with the amounts of grafted titanium. As can be seen, both activites and selectivities are affected by Ti loading. In particular, the results showed an increase of both activity and selectivity as TiO2 loading increases until the surface monolayer coating was reached. Then, a decrease occurs for a TiO2 amount greater than a monolayer. This behaviour can be related to the structural environment of the surface-supported Ti species. At low Ti coverage, supported TiO2/SiO2 oxide catalysts mainly show tetrahedrical titanium sites, which probably correspond to the presence of Lewis acid sites responsible for the activity in epoxidation reactions. By further increasing TiO2 loading, the surface dispersion decreases and, consequently, lower activities and selectivities are observed. Fig. 7 reports a comparison of the catalytic performances in the mentioned reactions of the two systems, i.e. TiO2/SiO2 catalysts (SiO2, 282 m2/g) and TiO2/ SiO2 catalysts (SiO2, 450 m2/g) [8], both characterized by different Ti loadings. It is interesting to note that the catalytic systems investigated showed a maximum in terms of both yield and selectivity in
261
correspondence with different values of the molar composition, corresponding for both catalytic systems to the monolayer TiO2 coverage of silica. These catalytic results suggest that the surface area of the starting support significantly determines the maximum surface coverage of titanium oxide species on silica and, consequently, the dispersion capacity. The higher yieldvalues observed in the case of the system with a higher surface area [8] are due to the higher amount of supported Ti active sites. 5. Conclusions As shown in the present work grafting titanium alkoxide onto the silica surface is an innovative method in preparing highly dispersed supported metal oxide catalysts, which possess superior properties relative to catalysts prepared by traditional aqueous methods. A homogenous dispersion of the grafted titanium species was observed until monolayer coverage was reached. The attempt to obtain multilayer coatings of silica by repeating the grafting operation more than once leads to the formation of amorphous agglomerates characterized by a lower surface TiO2 dispersion, as shown by DRIFT analyses on a semi-quantitative basis. However, textural analyses carried out by BET showed that the specific surface area does not significantly change with titanium loading, so confirming in this way the possibility to prepare high-surface supported titania catalysts by grafting. Interesting correlations between the surface properties of TiO2/SiO2 catalysts and their catalytic performances in the epoxidation of cyclooctene with cumene hydroperoxide were found. The catalytic data obtained showed that both activity and selectivity can be influenced by the coordination environment of titanium species grafted on the surface. Thus, homogeneous surface dispersion is a fundamental requirement to develop highly active and selective catalysts. Finally, the surface area and surface density of the hydroxyl groups of SiO2 are considered crucial in obtaining good catalyst dispersion and, consequently, good catalytic performances. Acknowledgements Thanks are due to MIUR for the financial support. We are also grateful to the research group of Prof. G. Capannelli for TEM/EDX analyses. References [1] [2] [3] [4] [5] [6] [7] [8]
Fig. 7. Comparison yield-selectivity vs. molar composition between two systems: TiO2/SiO2 catalysts (SiO2, 282 m2/g) and TiO2/SiO2 catalysts (SiO2, 450 m2/g) [8].
R.D. Roark, S.D. Kohler, J.G. Ekerdt, I. Wachs, Catal. Lett. 16 (1992) 77. R.D. Roark, S.D. Kohler, J.G. Ekerdt, Catal. Lett. 16 (1992) 71. R. Castillo, B. Koch, P. Ruiz, B. Delmon, J. Catal. 161 (1996) 524. G.C. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1. R.A. Sheldon, J.A. Van Doorn, J. Catal. 31 (1973) 427. R.A. Sheldon, J. Mol. Catal. 7 (1980) 107. H.P. Wulf, US Patent 39,233,843 (1975). P. Iengo, G. Aprile, M. Di Serio, D. Gazzoli, E. Santacesaria, Appl. Catal. A: Gen. 178 (1999) 97. [9] G. Deo, A.M. Turek, I.E. Wachs, P.A. Jacobs, Zeolytes 13 (1993) 365. [10] A. Fernandez, J. Leyrer, A.R. Gonzales-Elipe, H. Kno¨zinger, J. Catal. 112 (1998) 489.
262
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
[11] S. Haukka, E. Lakomaa, A. Root, J. Phys. Chem. 97 (1993) 5085. [12] X. Gao, S.R. Bare, J.L.G. Fierro, M.A. Banares, I.E. Wachs, J. Phys. Chem. B 102 (1998) 5653. [13] G.N. Vayssilov, Catal. Rev.-Sci. Eng. 39 (1997) 209. [14] F.R.D. Snell, L.S. Ettre, Enciclopedy of Industrial Chemical Analysis, vol. 19, Interscience, New York, 1974, p. 107. [15] D. Dollimore, G.R. Heal, J. Appl. Chem. 14 (1964) 109. [16] V.R. Kokatnur, M. Jelling, J. Am. Chem. Soc. 63 (1941) 1432. [17] E.C. Dearbon, R.M. Fuoss, A.K. MacKenzie, R.G. Shepherd Jr., Ind. Eng. Chem. 45 (12) (1953) 2715. [18] E. Santacesaria, M. Cozzolino, M. Di Serio, A.M. Venezia, R. Tesser, Appl. Catal. A: Gen. 270 (2004) 177.
[19] J.R. Anderson, K.C. Pratt, Introduction to Characterization and Testing of Catalysts, Academic Press, 1986. [20] K. Witke, A. Lachowicz, W. Bruser, Z. Zeigan, Anorg. All. Chem. 465 (1980) 193. [21] X. Gao, I.E. Wachs, Catal. Today 51 (1999) 233. [22] G. Lassaletta, A. Fernandez, J.P. Espinos, A.R. Gonzalez-Elipe, J. Phys. Chem. 99 (1995) 1484. [23] R. Hutter, T. Mallat, A. Baiker, J. Chem. Soc., Faraday Trans. 91 (1995) 1261. [24] R. Hutter, T. Mallat, A. Baiker, J. Catal. 31 (1973) 427. [25] T. Maschmeyer, F. Rey, G. Sanker, J.M. Thomas, Nature 378 (1995) 159.
Grafting of titanium alkoxides on high-surface SiO2 support: An advanced technique for the preparation of nanostructured TiO2/SiO2 catalysts M. Cozzolino, M. Di Serio, R. Tesser, E. Santacesaria * University of Naples ‘‘Federico II’’, Department of Chemistry, Via Cintia 4, Compl. Univ. Monte S. Angelo, I-80126 Naples, Italy Received 21 July 2006; accepted 7 February 2007 Available online 3 March 2007
Abstract The present paper reports and discusses results on the preparation, characterization and catalytic performances of silica-supported titania (TiO2/ SiO2) catalysts. Samples were prepared by grafting titanium tetraisopropoxide (Ti-OPri)4, dissolved in toluene, onto a silica surface in an N2 atmosphere, followed by steam hydrolysis and calcination. The samples were characterized by chemical analysis, BET surface area measurements, X-ray diffraction (XRD), DR–UV–vis, FTIR and DRIFT spectroscopy analyses, TEM/EDX and NH3-TPD. The results indicated that the grafting preparation method gives rise to very strong Si–O–Ti bonds, that are responsible for high titanium dispersion. In particular, at low Ti loading, titanium species in tetrahedric coordination resulted prevalent on the catalyst surface until the maximum surface monolayer coating reached (2.2 Ti atoms/nm2). The degree of polymerization of Ti species increases with further TiO2 load increases, giving rise to a large amount of octahedrical Ti-sites grafted on SiO2. The effects on the catalytic activity of TiO2/SiO2 catalysts with increasing quantities of TiO2 were also investigated. The catalytic results obtained in the epoxidation reaction of cyclooctene with cumene hydroperoxide showed the significant effect of titanium loading on the physicochemical and reactivity/selectivity properties of the silica-supported titania catalysts. # 2007 Elsevier B.V. All rights reserved. Keywords: Supported titanium oxide; Liquid-phase grafting preparation technique; TiO2 dispersion; Epoxidation reaction
1. Introduction Over the last decades, applications and uses of supported metal oxides on silica have driven research towards the production of catalysts characterized by the presence of high percentage of the metal oxide in a dispersed state (twodimensional surface metal oxide overlayers) [1–4]. An interesting example is given by silica-supported titania TiO2/ SiO2 catalysts that have been considered as advanced support materials substituting pure TiO2. The higher mechanical strength, thermal stability and specific surface area of the supported titania oxides, compared to pure TiO2, have recently attracted much attention and driven interest towards the use of these materials not only as catalytic supports, but also as catalysts through the generation of new catalytic active sites [5– 9]. It is well known that the silica surface is fairly inert. However, the silica surface hydroxyls generally act as
* Corresponding author. Tel.: +39 081674027; fax: +39 081674026. E-mail address: [email protected] (E. Santacesaria). 0926-860X/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2007.02.032
adsorptive/reactive sites because of their hydrophilic character. Thus, the preparation of highly dispersed metal oxides on silica often involves a highly reactive precursor, such as TiCl4 or titanium alkoxides, which readily react with the surface hydroxyls of the silica support [3,9,10]. It is well known that the preparation method, the surface area and the concentration of the surface hydroxyls on the silica surface could strongly influence the dispersion capacity and the maximum surface coverage of surface titanium oxide species on silica. Depending on titanium loading, often indicated as percentage by weight of TiO2, different types of titanium species could be present, ranging from dispersed surface TiOx species, both in tetrahedral and octahedral coordination, to small TiO2 crystallites. The relative amount depends on the preparation conditions and chemical composition (surface Ti densities) [3,11,12]. The physicochemical and reactivity/selectivity properties of oxide catalysts are often a strong function of their structural characteristics. In the past, extensive studies have been carried out on Ti silicalites and TiO2–SiO2 mixed oxides [13]. In both materials, Ti atoms are shown to substitute Si in the silica framework to form tetrahedral TiO4 units, which work as active
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
sites for the epoxidation reaction. The highly dispersed TiO2/ SiO2 supported oxides have also been shown to be active in liquid phase epoxidation reactions [5,6,9]. However, the structural features of supported TiO2/SiO2 oxides and their relationship with the physicochemical and reactivity/selectivity properties are still not understood well due to the lack of fundamental systematic studies. Therefore, this article focuses on the preparation of highly dispersed TiO2/SiO2 catalysts by grafting different amounts of titanium tetraisopropoxide, dissolved in toluene, onto the silica surface. In addition, a detailed characterization study was carried out to collect information about the structural characteristics of the supported titanium species. In this way, the factors that control the maximum surface coverage of silica by titanium oxide were investigated. Furthermore, the epoxidation reaction of cyclooctene with cumene hydroperoxide was employed to investigate the reactivity of highly dispersed TiO2/SiO2 catalysts since this reaction is very sensitive to the surface structure of the titanium sites. The catalytic results obtained could be considered as the starting point in the development of the molecular structure–reactivity/ selectivity relationships of TiO2/SiO2 catalysts. 2. Experimental 2.1. Preparation of catalysts A commercial silica (Grace S432, specific surface area = 280 m2/g, pore volume = 1.02 cm3/g, hydroxyl groups = 0.92 mmol/g) was used as a support. The solids were prepared by contacting the silica, calcined at 773 K for 8 h, with a solution of tetra-isopropoxide ((Ti-OPri)4, Aldrich) dissolved in anhydrous toluene. The reaction was performed in a jacketed glass-reactor of 200 cm3 for 6 h, constantly stirred, at the boiling temperature of toluene (388 K). The solid was filtered off, washed with toluene, dried at 393 K overnight, hydrolyzed with steam and, finally, calcined at 773 K for 2 h. Steam hydrolysis was performed at 150 8C. The amount of adsorbed titanium was determined by the colorimetric analysis suggested by Snell and Ettre [14], by evaluating the quantity of titanium remaining in solution after the grafting reaction. The operative conditions and the adsorption results are listed in Table 1.
257
2.2. Techniques used in catalyst characterization Different techniques were used for catalyst characterization, such as: X-ray diffraction (XRD), TEM/EDX, diffuse reflectance spectroscopic analyses (DRIFT and DRUV), temperature programmed desorption (TPD). Textural analyses were carried out by using a Thermoquest Sorptomatic 1990 Instrument (Fisons Instrument) and by determining the nitrogen adsorption/desorption isotherms at 77 K. The samples were thermally pretreated under vacuum (105 mbar) overnight up to 473 K (heating rate = 1 K/min). Specific surface area (SBET) and pore size distributions were determined by using the BET and Dollimore–Heal methods [15]. XRD analyses were carried out by using a Philips diffractometer. The scans were collected in the range 5–608 (2u) using Cu Ka radiation with a rate of 0.018 (2u)/s. TEM observations were made by using a Jeol Jem 2010 equipped with an EDX probe and sample preparation was as follows: a drop of the dispersion of the milled catalytic powder in isopropyl alcohol was put on a Lacey carbon grid and the dispersant was removed by evaporation at room temperature. Diffused reflectance spectra were obtained on a UV–vis scanning Jasco spectrometer V-550, equipped with an integrating sphere, using BaSO4 as reference. UV–vis spectra were recorded in the diffused reflectance mode (R) and transformed by a magnitude proportional to the extinction coefficient (K) through a Kubelka–Munk function (F(R)). FTIR and DRIFT were recorded by using a Nicolet AVATAR 360 instrument, equipped with an accessory for the diffused reflectance. Samples were powdered and diluted 1% by weight with KBr. Adsorbed water was completely removed ‘‘in situ’’ by heating at 673 K under vacuum (105 mbar). Thermal programmed desorption (TPD) measurements were carried out by using ammonia as the probe molecule. Samples of about 100 mg were first calcined in situ at 673 K for 1 h under air flow in order to remove adsorbed water. After cooling, the samples were subsequently contacted with a stream of Helium (1000 cm3/min), containing 1000 ppm of NH3, until saturation level is reached. The ammonia desorption was performed under Helium flow from 373 to 773 K (heating rate = 10 K/min).
Table 1 List of the catalysts prepared by grafting titanium alkoxide on SiO2 and related properties—amount of the support (SiO2) = 4 g Catalysts SiO2 TS1 TS2 TS3 TS4 TS-M (1 step) TS-D (2 steps) TS-T (3 steps)
Ti(O-Pri)4 initial amount (g)
Anchored metal (mmolTi =gSiO2 )
molTi molTi
– 0.04 0.2 0.4 0.6 1.04 1.04 1.04
– 0.035 0.176 0.352 0.528 0.915 1.184 1.383
– 0.99 1 1 1 1 0.64 0.50
anchored/ initial
wt.% TiO2
BET surface area (m2/g)
Specific pore volume (cm3/g)
mmolNH3 =g
Yields to epoxide (%)
Selectivity (%)
– 0.3 1.4 2.8 4.2 5.7 9.5 11.1
282 – 280 283 – 280 276 278
1.02 – 0.99 0.81 – 0.23 0.26 0.27
– 30 45 84 120 156 211 234
– – 3.8 11.6 – 31.3 – 21.2
– – 7.3 28.2 – 70.8 – 36.3
258
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
2.3. Catalytic tests Epoxidation of cyclooctene (Fluka 98%) with cumene hydroperoxide (Fluka 80% by weight in cumene) was performed in a 100 cm3 three neck glass vessel equipped with a Liebig condenser and an internal cooling coil. The reaction temperature was kept constant, at 355 K, by putting the vessel in a thermostatted bath. To 22 cm3 of cyclooctene, 0.3 g catalyst was added, then 10 cm3 cumene hydroperoxide was slowly added with a syringe. The reaction mixture was kept at 355 K under a nitrogen atmosphere for 3 h. Unreacted hydroperoxide was measured by iodometric titration [16] at the end of the reaction. The epoxide formed was measured with the method suggested by Dearbon et al. [17]. Selectivities were determined as: selectivity = 100 [formed epoxide (mol)/ consumed peroxide (mol)]. 3. Results and discussion 3.1. Catalyst preparation By taking into account the surface monolayer of titanium tetraisopropoxide on silica **(G1 = 1.02 mmolTi/gSiO2), determined in a previous work [18], the catalysts reported in Table 1 indicated with the acronyms TS1, TS2, TS3 and TS4, can be considered as sub-monolayer catalysts. In order to completely cover the silica surface, catalysts with higher TiO2 loading were prepared by repeating the operations of grafting, hydrolysis and calcination. In particular, TS-D and TS-T catalysts were prepared by submitting the monolayer catalyst (TS-M) to the operations mentioned above, one and two times respectively. The related properties are also reported in Table 1. 3.2. Catalyst characterization 3.2.1. Nitrogen adsorption (BET) measurements Both support and prepared samples were characterized by BET analyses. The specific surface areas are reported in Table 1. It is interesting to observe that the specific surface area does not significantly change with titanium loading for all the prepared samples. This confirms the possibility to prepare highsurface supported titania catalysts by grafting. On the contrary, a considerable decrease in the specific pore volume compared to that of silica, was observed for all the sub-monolayer catalysts until the monolayer coverage is reached. No further variations were observed for higher TiO2 loadings. 3.2.2. X-ray diffraction analysis XRD analyses were performed to estimate surface TiO2 dispersion. The absence of crystallites of TiO2 was detected in all the XRD patterns of TiO2/SiO2 prepared samples, resulting mainly amorphous. These results suggested that TiO2 particles on SiO2 are very small and below the detection ˚ ). Only the TS-T sensitivity of the XRD technique (<40 A catalyst showed the characteristic peak of TiO2 (anatase) at 2u = 258. These peaks resulted broad, meaning that crystal size was very small. The mean dimension of the crystals was
measured along with the crystallographic direction 1 0 1, using the Scherrer formula d¯ h k l ¼
Kl Bd cos u
where K is the Scherrer constant, ( the wavelength, Bd the FWHM (full widht half maximum), and u is the Bragg angle. ˚ , a value quite close to the Crystallites size was about 45 A instrumental sensitivity limit. The amount of TiO2 crystalline phase present in TiO2/SiO2 samples were measured using ZrO2 as reference compound [19]. Physical mixtures of TiO2, SiO2 and ZrO2 containing 5% wt. of ZrO2 and increasing quantities of titania (2, 5, 10, 15, 20% wt.) were prepared and underwent XRD analysis. The area ratio of the characteristic anatase peak at 2u = 258 and the ZrO2 peak at 2u = 288 increased in line with increases in TiO2 concentration. This calibration line was used to determine the crystalline TiO2 contents in a sample containing TS-T catalyst and ZrO2 (5 wt.%). The amount of crystalline TiO2 determined on the TS-T catalyst was 1.8%. Further study was made of the influence of calcination temperature on the amount of TiO2 crystalline phase and crystallite size. For this purpose, some TS-T samples were calcined at different temperatures and then submitted to XRD analysis after each step of calcination. The low amount of crystalline TiO2, observed (from 1.8 to 2.6%) also for the catalysts calcined at higher temperatures (from 500 to 800 8C), and the slight changes of crystallite size confirms the stability of Si–O–Ti bonds and the good titanium dispersion achieved as a consequence. 3.2.3. Morphological analyses by TEM and EDX TEM observations and EDX analyses were conducted on TS3 and TS-T catalysts. Both catalysts resulted nearly completely amorphous to the morphological observations as the presence of well-defined crystalline TiO2 was not detected. These results seem to suggest that the grafted titanium remains homogenously dispersed on the silica support, even after repeating the grafting operation of titanium alkoxide on silica three times, followed by steam hydrolysis and calcination. Moreover, EDX analysis gave a value of supported Ti corresponding approximately to that of the chemical analysis. 3.2.4. FTIR spectroscopic analysis 3.2.4.1. Analysis of 3200–2700 cm1 spectral range. Some of the prepared samples were submitted to FTIR analysis before the calcination treatment. In Fig. 1 we report the IR spectra recorded in the range 3100–2800 cm1 for TS1, TS2 and TS3 catalysts. The spectra showed four bands at 2981, 2929, 2880, 2857 cm1, assigned to symmetric and asymmetric vibrational C–H stretching of CH3 groups [20]. The bands corresponding to the vibrational C–C stretching of isopropilic groups and CH3 (1260–1145 cm1), and those characteristic of vibrational C–O stretching (1150–1050 cm1) did not appear in the spectra since these are completely covered by the strong bands characteristic of vibrational Si–O–Si stretching in the range 1000–1300 cm1. The area of infrared signals of the peaks corresponding to CH bonds give a linear trend as a function of
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
Fig. 1. Infrared spectra (spectral range: 3100–2800 cm1) of TS1, TS2 and TS3 samples.
molar ratio Ti/Si for the sub-monolayer catalysts: TS1, TS2, and TS3. This suggests that the grafting stoichiometry remains almost constant, at least for the catalysts with a titanium loading lower than the monolayer coverage. 3.2.4.2. Analysis of spectral range 4000–3500 cm1. Fig. 2 reports the spectra of silica and sub-monolayer TiO2/SiO2 calcined catalysts, recorded in the range 4000–3500 cm1 after calcination. From the figure, it is interesting to observe the decrease in the intensity of the peak at 3747 cm1, attributed to the isolated surface silanols (single Si–OH groups and geminal Si(OH)2 groups). This is due to the consumption of Si–OH hydroxyls owing to the grafting reaction. Fig. 3 contains (as a function of Ti/Si molar ratio) the ratio of peak area recorded at 3747 cm1 of TiO2/SiO2 catalysts and the peak area at the same wavenumber of silica support. As can be seen, no linear correlation was observed, probably due to the decrease of
Fig. 2. Infrared spectra (spectral range: 4000–3700 cm1) of the support (SiO2) and TS1, TS2, TS3 and TS4.
259
Fig. 3. AðOHÞTS catalysts =AðOHÞSiO2 as function of molðTiÞanchored =mol ðOHÞSiO2 .
titanium dispersion since aggregates of the titanium species form during calcination. We found that only for the submonolayer catalysts the stoichiometric values obtained correspond to that of the grafting reaction and resulted about 2.5. 3.2.5. Determination of TiO2 dispersion by DRIFT analysis in the spectral range 1400–700 cm1 DRIFT analyses were performed on the following catalysts: TS1, TS2, TS4, TS-M, TS-D and TS-T. The spectra obtained showed the following signals: A large band with a maximum at 1100 cm1 attributed to the asymmetric vibrational stretching Si–O–Si, characteristic of SiO4 units with tetrahedrical coordination; a shoulder at 1200 cm1, due to the asymmetric vibrational Si–O stretching. A peak at 800 cm1 due to the corresponding symmetric vibrational stretching. A peak at 950 cm1 characteristic of symmetric vibrational Si–O–Ti stretching. The last peak resulted crucial in determining the dispersion of TiO2 over the silica surface, at least from a semi-quantitative point of view. In fact, it is possible to estimate the catalyst dispersion by comparing the area of the band at 950 cm1 with that of the band at 1200 cm1. The ratio between those two mentioned areas, A(Si–O–Ti) and A(Si–O–Si), calculated by a gaussian deconvolution of the four peaks mentioned above, could be an index of the catalyst’s dispersion. In Fig. 4 the A(Si–O–Ti)/A(Si–O–Si) ratio is reported a as function of molar ratio Ti/Si. As can be seen as the Ti/Si molar ratio increases, the A(Si–O– /A Ti) (Si–O–Si) ratio does not increase linearly. This means that as Ti loading increases the dispersion decreases, and only for a low TiO2 quantities we have a homogeneous dispersion. The data in Fig. 4 shows this quantity to be about 0:2 mmolTi =gSiO2 .
260
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
Fig. 4. A(Si–O–Ti)/A(Si–O–Si) as function of molar composition.
Fig. 6. DR–UV spectra: (a) SiO2; (b) TS-M; (c) TS-D; (d) TS-T; (e) TiO2 (anatase).
3.2.6. Diffused reflectance UV–vis spectroscopic analysis All the TiO2/SiO2 catalysts were submitted to DR–UV analysis in order to provide information about the coordination geometry of the Ti cations (the first coordination sphere) and the ligand environment (the second coordination sphere) under various conditions. The DR–UV–vis spectra of all the prepared samples, characterized by increasing Ti loadings, are compared in Figs. 5 and 6. The spectrum of bulk TiO2 (anatase) is also included in both the figures for a useful comparison. The absorption of silica in the range 200–500 nm can be considered negligible. It is well known that the Ti cations in tetrahedric coordination show a typical band at 212 nm, due to the ligand–metal charge transfer between Ti4+ and oxygen ligands, such as –O–H, –O–Si, –O–Ti, or H2O. On the other hand, a shift of the band towards higher wavelenghts (260 nm) is usually observed for the Ti cations in octahedrical environments. The DR–UV–vis spectra showed adsorption bands in the range 210–360 nm. It has been observed that the maximum
ligand–metal charge transfer (LMCT) transitions shift to higher wavelenghts with increasing TiO2 loading (see Fig. 5), suggesting a gradual increase in the polymerization degree of Ti atoms [21]. In particular, the TS-D and TS-T samples prepared by repeating titanium alkoxide grafting, two and three times, showed a broad absorption band with a shape (see Fig. 6) very similar to that of bulk anatase TiO2. This result suggests that, even though the TiO2 crystallites are very small and beyond the detection sensitivity of XRD measurements, they may still possess the same electronic properties as the pure TiO2 anatase phase. In this way, by adopting the grafting preparation method, it is possible to chemically modify the surface of an oxide, without altering the original mechanical properties. Finally, the decrease in the edge energy of LMCT transitions of Ti atoms with increasing TiO2 loading can also be associated to the increase in the number of the nearest Ti atoms, due to the Ti polymerization at higher TiO2 loadings. The observed shift of the band-gap absorption edge may be explained in terms of quantum effects due to the small titania particles [22]. 3.2.7. Temperature programmed desorption (TPD) analyses The catalysts were submitted to TPD analysis by using ammonia as the probe molecule. The results obtained are also reported in Table 1. From the values reported in Table 1, it can be observed that the grafting of titanium species increases the total number of acid sites on the surface of silica, greatly. On the contrary, by considering the related values of temperature (TM) in correspondence with the maximum of the NH3 desorption, the strength of acid sites remains almost constant. In particular, the results suggest the presence of acid sites of medium strength. 4. Catalytic performances in the epoxidation reaction
Fig. 5. DR–UV spectra: (a) SiO2; (b) TS1; (c) TS3; (d) TS4;(e) TS-M; (f) TiO2 (anatase).
It is well known that the catalytic performance of titanium oxide appears to be completely modified by interaction with the silica support, which is associated with the changes in the
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
molecular structure and coordination environment. The particular catalytic performances of the highly dispersed TiO2/SiO2 catalysts, in comparison with pure TiO2, were previously demonstrated for liquid-phase epoxidation reactions with hydrogen peroxide and alkyl hydroperoxide [6]. Pure TiO2 is not active in the epoxidation reaction, while the highly dispersed TiO2/SiO2 catalysts exhibit high reactivity and high selectivity to epoxide [6]. Two fundamental requirements for good catalytic performance are: high dispersion of the catalytically active component within the matrix (site isolation) and stability to leaching [23–25]. In the present work, the catalytic performances of TiO2/SiO2 prepared catalysts were tested in the epoxidation of cyclooctene with cumene hydroperoxide. Cyclooctene is an interesting test reagent because of the high stability of the corresponding oxide. The results obtained are summarized in Table 1 where yields to epoxide (%) and related selectivities are reported together with the amounts of grafted titanium. As can be seen, both activites and selectivities are affected by Ti loading. In particular, the results showed an increase of both activity and selectivity as TiO2 loading increases until the surface monolayer coating was reached. Then, a decrease occurs for a TiO2 amount greater than a monolayer. This behaviour can be related to the structural environment of the surface-supported Ti species. At low Ti coverage, supported TiO2/SiO2 oxide catalysts mainly show tetrahedrical titanium sites, which probably correspond to the presence of Lewis acid sites responsible for the activity in epoxidation reactions. By further increasing TiO2 loading, the surface dispersion decreases and, consequently, lower activities and selectivities are observed. Fig. 7 reports a comparison of the catalytic performances in the mentioned reactions of the two systems, i.e. TiO2/SiO2 catalysts (SiO2, 282 m2/g) and TiO2/ SiO2 catalysts (SiO2, 450 m2/g) [8], both characterized by different Ti loadings. It is interesting to note that the catalytic systems investigated showed a maximum in terms of both yield and selectivity in
261
correspondence with different values of the molar composition, corresponding for both catalytic systems to the monolayer TiO2 coverage of silica. These catalytic results suggest that the surface area of the starting support significantly determines the maximum surface coverage of titanium oxide species on silica and, consequently, the dispersion capacity. The higher yieldvalues observed in the case of the system with a higher surface area [8] are due to the higher amount of supported Ti active sites. 5. Conclusions As shown in the present work grafting titanium alkoxide onto the silica surface is an innovative method in preparing highly dispersed supported metal oxide catalysts, which possess superior properties relative to catalysts prepared by traditional aqueous methods. A homogenous dispersion of the grafted titanium species was observed until monolayer coverage was reached. The attempt to obtain multilayer coatings of silica by repeating the grafting operation more than once leads to the formation of amorphous agglomerates characterized by a lower surface TiO2 dispersion, as shown by DRIFT analyses on a semi-quantitative basis. However, textural analyses carried out by BET showed that the specific surface area does not significantly change with titanium loading, so confirming in this way the possibility to prepare high-surface supported titania catalysts by grafting. Interesting correlations between the surface properties of TiO2/SiO2 catalysts and their catalytic performances in the epoxidation of cyclooctene with cumene hydroperoxide were found. The catalytic data obtained showed that both activity and selectivity can be influenced by the coordination environment of titanium species grafted on the surface. Thus, homogeneous surface dispersion is a fundamental requirement to develop highly active and selective catalysts. Finally, the surface area and surface density of the hydroxyl groups of SiO2 are considered crucial in obtaining good catalyst dispersion and, consequently, good catalytic performances. Acknowledgements Thanks are due to MIUR for the financial support. We are also grateful to the research group of Prof. G. Capannelli for TEM/EDX analyses. References [1] [2] [3] [4] [5] [6] [7] [8]
Fig. 7. Comparison yield-selectivity vs. molar composition between two systems: TiO2/SiO2 catalysts (SiO2, 282 m2/g) and TiO2/SiO2 catalysts (SiO2, 450 m2/g) [8].
R.D. Roark, S.D. Kohler, J.G. Ekerdt, I. Wachs, Catal. Lett. 16 (1992) 77. R.D. Roark, S.D. Kohler, J.G. Ekerdt, Catal. Lett. 16 (1992) 71. R. Castillo, B. Koch, P. Ruiz, B. Delmon, J. Catal. 161 (1996) 524. G.C. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1. R.A. Sheldon, J.A. Van Doorn, J. Catal. 31 (1973) 427. R.A. Sheldon, J. Mol. Catal. 7 (1980) 107. H.P. Wulf, US Patent 39,233,843 (1975). P. Iengo, G. Aprile, M. Di Serio, D. Gazzoli, E. Santacesaria, Appl. Catal. A: Gen. 178 (1999) 97. [9] G. Deo, A.M. Turek, I.E. Wachs, P.A. Jacobs, Zeolytes 13 (1993) 365. [10] A. Fernandez, J. Leyrer, A.R. Gonzales-Elipe, H. Kno¨zinger, J. Catal. 112 (1998) 489.
262
M. Cozzolino et al. / Applied Catalysis A: General 325 (2007) 256–262
[11] S. Haukka, E. Lakomaa, A. Root, J. Phys. Chem. 97 (1993) 5085. [12] X. Gao, S.R. Bare, J.L.G. Fierro, M.A. Banares, I.E. Wachs, J. Phys. Chem. B 102 (1998) 5653. [13] G.N. Vayssilov, Catal. Rev.-Sci. Eng. 39 (1997) 209. [14] F.R.D. Snell, L.S. Ettre, Enciclopedy of Industrial Chemical Analysis, vol. 19, Interscience, New York, 1974, p. 107. [15] D. Dollimore, G.R. Heal, J. Appl. Chem. 14 (1964) 109. [16] V.R. Kokatnur, M. Jelling, J. Am. Chem. Soc. 63 (1941) 1432. [17] E.C. Dearbon, R.M. Fuoss, A.K. MacKenzie, R.G. Shepherd Jr., Ind. Eng. Chem. 45 (12) (1953) 2715. [18] E. Santacesaria, M. Cozzolino, M. Di Serio, A.M. Venezia, R. Tesser, Appl. Catal. A: Gen. 270 (2004) 177.
[19] J.R. Anderson, K.C. Pratt, Introduction to Characterization and Testing of Catalysts, Academic Press, 1986. [20] K. Witke, A. Lachowicz, W. Bruser, Z. Zeigan, Anorg. All. Chem. 465 (1980) 193. [21] X. Gao, I.E. Wachs, Catal. Today 51 (1999) 233. [22] G. Lassaletta, A. Fernandez, J.P. Espinos, A.R. Gonzalez-Elipe, J. Phys. Chem. 99 (1995) 1484. [23] R. Hutter, T. Mallat, A. Baiker, J. Chem. Soc., Faraday Trans. 91 (1995) 1261. [24] R. Hutter, T. Mallat, A. Baiker, J. Catal. 31 (1973) 427. [25] T. Maschmeyer, F. Rey, G. Sanker, J.M. Thomas, Nature 378 (1995) 159.