pubs.acs.org/Langmuir © 2009 American Chemical Society
Grafting Strategy to Develop Single Site Titanium on an Amorphous Silica Surface M. C. Capel-Sanchez,† G. Blanco-Brieva,† J. M. Campos-Martin,† M. P. de Frutos,‡ W. Wen,§ J. A. Rodriguez,§ and J. L. G. Fierro*,† †
Instituto de Cat alisis y Petroleoquı´mica, CSIC, Marie Curie 2, Cantoblanco, 28049 Madrid, Spain, ‡Centro de Tecnologı´a Repsol YPF, A-5, Km. 18, 28931 M ostoles, Madrid, Spain, and §Chemistry Department, Brookhaven National Laboratory, Upton, New York, 11973 Received July 16, 2008. Revised Manuscript Received April 7, 2009
Titanium/silica systems were prepared by grafting a titanium alkoxide (titanium isopropoxide and titanium (triethanolaminate) isopropoxide) precursor onto amorphous silica. The grafting process, which consisted of the hydrolysis of the Ti precursor by the hydroxyl groups on the silica surface, yielded samples containing Ti-loadings of 1-1.6 wt %. The as synthesized and calcined TiO2-SiO2 samples were characterized by UV-vis, FTIR, XPS, and XANES spectroscopic techniques. These systems were tested in the liquid-phase epoxidation of oct-1-ene with hydrogen peroxide reaction. Spectroscopic data indicated that titanium anchoring takes place by reaction between the alkoxide precursor and surface OH groups of the silica substrate. The nature of surface titanium species generated by chemical grafting depends largely on the titanium precursor employed. Thus, the titanium isopropoxide precursor yields tetrahedrally coordinated polymeric titanium species, which give rise to a low-efficiency catalyst. However, if an atrane precursor (titanium (triethanolaminate) isopropoxide) is employed, isolated titanium species are obtained. The fact that these species remain isolated even after calcination is due to the protective effect of the triethanolaminate ligand that avoids titanium polymerization. These differences in the titanium environment have a pivotal role in the performance of these systems in the epoxidation of alkenes with hydrogen peroxide.
Introduction Titania-silica mixed oxides are considered as advanced support materials substituting pure TiO2. The higher mechanical strength, thermal stability, surface acidity, and specific area of the Ti-loaded silica substrates, compared to pure TiO2, have recently attracted much attention and driven interest toward the use of these materials not only as supports but also as catalysts through the generation of new catalytic active sites. These latter properties have frequently been exploited in catalysis either as catalysts1-4 or as carriers of an active ingredient.5-10 Moreover silica-supported titania is used on the industrial scale for propylene epoxidation, using alkylhydroperoxides as oxidants.11 Among the commodity chemicals, epoxides, and especially propylene oxide, have enormous importance in the chemical industry. In particular, the epoxidation of alkenes with diluted hydrogen peroxide is one of the main goals in this field. *Corresponding author. Fax: +34 915 854 760. E-mail: jlgfierro@ icp.csic.es. Website: http://www.icp.csic.es/eac/index.htm. (1) Niwa, M.; Sago, M.; Ando, H.; Murakami, Y. J. Catal. 1981, 6969. (2) Imamura, S.; Tarumoto, H.; Ishida, S. Ind. Eng. Chem. Res. 1989, 28, 1449. (3) Neumann, R.; Chava, M.; Levin, M. J. Chem. Soc., Chem. Commun. 1993, 1685. (4) (a) Dutoit, D. C. M.; Schneider, M.; Baiker, A. J. Catal. 1995, 153, 165. (b) Hutter, R.; Mallat, T.; Baiker, A. J. Catal. 1995, 153, 177. (5) Baiker, A.; Dollenmeier, P.; Glinski, M.; Reller, A. Appl. Catal. 1987, 35, 365. (6) Reichmann, M. G.; Bell, A. T. Appl. Catal. 1987, 32, 315. (7) Vogt, E. T. C.; De Boer, M.; van Dillen, A. J.; Geus, J. W. Appl. Catal. 1988, 40, 255. :: (8) Rajadhyakssha, R. A.; Knozinger, H. Appl. Catal. 1989, 51, 81. (9) Galan-Fereres, M.; Mariscal, R.; Alemany, L. J.; Fierro, J. L. G.; Anderson, J. A. J. Chem. Soc., Faraday Trans. 1994, 90, 3711. (10) Galan-Fereres, M.; Alemany, L. J.; Mariscal, R.; Ba~nares, M. A.; Anderson, J. A.; Fierro, J. L. G. Chem. Mater. 1995, 7, 1342. (11) Ragers, R. C. Br. Patent 1,249,079, 1971. (12) Capel-Sanchez, M. C.; Campos-Martin, J. M.; Fierro, J. L. G. Appl. Catal., A 2003, 246, 69.
7148 DOI: 10.1021/la900578u
Several Ti-containing catalysts, including TS-1,12-14 Ti-β,15,16 Ti-containing ordered mesoporous silica,16-18 amorphous xerogels,19,20 and Ti-supported silica21,22 record high efficiency and molecular selectivity in oxidation reactions of olefinic substrates with hydrogen peroxide under mild conditions. Moreover, the performance of these catalysts is highly dependent upon the titanium environment. As a result of their relevance in industrial applications, the local environment of Ti in these catalysts has been studied in depth using several techniques, such as diffuse reflectance-ultraviolet-visible (DRS-UV-vis),23,24 Fourier transform-infrared (FTIR), and Raman25-29 spectroscopic (13) Notari, B. Adv. Catal. 1996, 41, 253. :: (14) Ratnasamy, P.; Srinivas, D.; Knozinger, H. Adv. Catal. 2004, 48, 1. (15) Saxton, P. J.; Chester, W.; Zajacek, J. G.; Crocco, G. L.; Wijesekera, K. S. U.S. Patent 5621122, 1997. (16) Lane, B. S.; Burgess, K. Chem. Rev. 2003, 103, 2457. (17) Thomas, J. M.; Raja, R.; Lewis, D. W. Angew. Chem., Int. Ed. 2005, 44, 6456. (18) Brutchey, R. L.; Ruddy, D. A.; Andersen, L. K.; Tilley, T. D. Langmuir 2005, 21, 9576. (19) Van Grieken, R.; Sotelo, J. L.; Martos, C.; Fierro, J. L. G.; Lopez-Granados, M.; Mariscal, R. Catal. Today 2000, 61, 49. (20) Mariscal, R.; Lopez-Granados, M.; Fierro, J. L. G.; Sotelo, J. L.; Martos, C.; Van Grieken, R. Langmuir 2000, 16, 9460. (21) Capel-Sanchez, M. C.; Campos-Martin, J. M.; Fierro, J. L. G.; de Frutos, M. P.; Polo, A. P. Chem. Commun. 2000, 855. (22) Tang, H.; Yu, C. H.; Oduoro, W.; He, H; Tsang, S. C. Langmuir 2008, 24, 1587. (23) Geobaldo, F.; Bordiga, S.; Zecchina, A.; Giamelo, E.; Leofanti, G.; Petrini, G. Catal. Lett. 1992, 16, 109. (24) O’Shea, V. A. D.; Capel-Sanchez, M.; Blanco-Brieva, G.; Campos-Martin, J. M.; Fierro, J. L. G. Angew. Chem., Int. Ed. 2003, 42, 5851. (25) Trukhan, N. N.; Panchenko, A. A.; Roduner, E.; Mezgunov, M. S.; :: Kholdeeva, O. A.; Mrowiec-Biazon, J.; Jarzeubski, A. B. Langmuir 2005, 21, 10545. (26) Armaroli, T.; Milella, F.; Notari, B.; Willey, R. J.; Busca, G. Top. Catal. 2001, 15, 63. (27) Notari, B.; Willey, R. J.; Panizza, M.; Busca, G. Catal. Today 2006, 116, 99. (28) Astorino, E.; Peri, J. B.; Willey, R. J.; Busca, G. J. Catal. 1995, 157, 482. (29) Deo, G.; Turek, A. M.; Wach, I. E.; Huybrechts, D. R. C.; Jacobs, P. A. Zeolites 1993, 13, 365.
Published on Web 05/08/2009
Langmuir 2009, 25(12), 7148–7155
Capel-Sanchez et al.
Article Table 1. Sample Labeling, Titanium Content, and Treatment
sample TiP TiPc TY TYc
titanium precursor
% wt Ti
treatment
titanium isopropoxide titanium isopropoxide titanium (triethanolaminate) isopropoxide titanium (triethanolaminate) isopropoxide
1.6 1.6 1.0 1.0
as synthesized calcination at 773 K as synthesized calcination at 773 K
techniques, powder neutron diffraction,30 X-ray absorption spectroscopy,31-36 and computer modeling studies.24 All these studies support the general consensus that the most active and selective sites are isolated, mononuclear, 4-coordinate Ti(IV) centers.14,16 However, since traditional spectroscopic techniques such as UV-vis, X-ray photoelectron spectroscopy (XPS), FTIR, and Raman provide only limited information about the nature of the active site, the use of X-ray absorption near-edge spectroscopy (XANES) is an extremely useful tool for studying the coordination of titanium with oxide anions.31-37 An important challenge in the synthesis of the Ti/SiO2 catalyst system is the development of a chemical route that allows the chemical modification of the silica surface with the introduction of a Ti-containing surface species with a precise structure and composition. Unfortunately, there is a current lack of reliable, general synthetic methods that allow such fine control over solid-state structures. To circumvent this problem, other approaches have been undertaken. The preparation of silicasupported titanium materials has been described first by Wulf38 and more recently by other authors employing different recipes for preparing this system, usually by treating the silica substrate with Ti(OiPr)4,21 TiF4,39 tetraneopentyltitanium,40 or ionbeam implantation.41 Our previous studies21,42-45 emphasized the importance of the preparation method of amorphous silicasupported titanium catalysts to obtain very active and selective catalysts for the alkene epoxidation reaction with hydrogen peroxide in which the titanium ions have a tetrahedral chemical environment. Titanium K-edge XANES spectroscopy is widely used to gather information on the coordination environment of titanium ions in structurally complex oxide materials whose matrix can be either amorphous or crystalline.46,47 In particular, the position
(30) Henry, P. F.; Weller, M. T.; Wilson, C. C. J. Phys. Chem. B 2001, 105, 7452. (31) Thomas, J. M.; Catlow, C. R. A.; Sankar, G. Chem. Commun. 2002, 2921 and references cited therein. (32) Liu, Z.; Davis, R. J. J. Phys. Chem. 1994, 98(4), 1253. (33) Davis, R. J.; Liu, Z. Chem. Mater. 1997, 9(11), 2311. (34) Liu, Z.; Crumbaugh, G. M.; Davis, R. J. J. Catal. 1996, 159(1), 83. (35) Blanco-Brieva, G.; Capel-Sanchez, M. C.; Campos-Martin, J. M.; de Frutos, M. P.; Fierro, J. L. G.; Lede, E. J.; Adrini, L.; Requejo, F. G. Adv. Synth. Catal. 2003, 345, 1314. (36) Notestein, J. M.; Andrini, L. R.; Kalchenko, V. I.; Requejo, F. G.; Katz, A.; Iglesia, E. J. Am. Chem. Soc. 2007, 129, 1122–1131. (37) Notestein, J. M.; Solovyou, A.; Andrini, L. R.; Requejo, F. G.; Katz, A.; Iglesia, E. J. Am. Chem. Soc. 2007, 129, 15585–15595. (38) Wulf, H. P. U.S. Patent 3923843, 1973. (39) Jorda, E.; Tuel, A.; Teissier, R.; Kerneval, J. J. Catal. 1998, 175, 93. (40) Holmes, S. A.; Quignard, F.; Choplin, A.; Teissier, R.; Kerneval, J. J. Catal. 1998, 176, 182. (41) Yang, Q.; Li, C.; Wang, S.; Lu, J.; Ying, P.; Xin, Q.; Shi, W. Stud. Surf. Sci. Catal. 2000, 130, 221. (42) Capel-Sanchez, M. C.; Campos-Martin, J. M.; Fierro, J. L. G. J. Catal. 2003, 217, 195. (43) Capel-Sanchez, M. C.; Campos-Martin, J. M.; Fierro, J. L. G. J. Catal. 2005, 234(2), 488. (44) Capel-Sanchez, M. C.; Campos-Martin, J. M.; Fierro, J. L. G. In Focus on Catalysis Research; Bevy, L. P., Ed.; Nova Science Publishers, Inc.: New York, 2006; Chapter 7. (45) Capel-Sanchez, M. C.; de la Pe~na-O’Shea, V. A.; Campos-Martin, J. M.; Fierro, J. L. G. Top. Catal. 2006, 41, 27. (46) Grunes, L. A. Phys. Rev. B 1983, 27, 2111. (47) Fraile, J. M.; Garcia, J.; Mayoral, J. A.; Proietti, M. G.; Sanchez, M. C. J. Phys. Chem. 1996, 100, 19484.
Langmuir 2009, 25(12), 7148–7155
and intensity of the pre-edge peaks in Ti K-edge XANES give information on the Ti coordination sphere because the intensity of the pre-edge transitions is sensitive to the symmetry of the surrounding atoms, being dipole-forbidden. They are weak in symmetrical environments (i.e., octahedral coordination) and increase in intensity as the environment is less centrosymmetric (i.e., tetrahedral coordination). Determination of pre-edge position and intensity can thus be used for the analysis of Ti centers in catalysts.31,36,37,47-53 The performance of titanium grafted onto a silica matrix depends largely on the nature of the titanium derivative used.41,54,55 The synthetic approach followed in this work consisted of the use of two oxygen-rich molecular Ti precursors, titanium isopropoxide and titanium (triethanolaminate) isopropoxide solution, for the controlled synthesis of well-dispersed, mixedelement oxides with specific microstructures. The generation of homogeneously dispersed structures of this type appears to be promoted by the facile, low-temperature elimination of the ligands from the initially grafted, molecular species to give a stable inorganic “single site”. In this endeavor, we have targeted a lowcost, commercial SiO2 substrate displaying a high surface area to provide the highest concentration of active sites per unit volume, and with the highest loading of site-isolated titanium. To understand the origin of these different behaviors, several characterization techniques (UV-vis, XPS, and X-ray absorption spectroscopy) techniques were employed to shed light on the mechanism of titanium incorporation on the amorphous silica substrate, and to relate surface structures with their performance in the epoxidation of alkenes with hydrogen peroxide.
Experimental Methods Samples were prepared following this method: titanium (IV) precursor (Aldrich) (2.0 mmol) was dispersed in isopropyl alcohol (25 mL). The titanium precursors employed were titanium isopropoxide (TYZOR TIP) or titanium (triethanolaminate) isopropoxide solution (TYZOR TE). The solution was heated to 353 K under stirring, and then 5.0 g of silica (Grace Davison XPO 2407, specific area = 213 m2 g-1, pore volume = 1.4 mL g-1) was added to the solution, with the mixture maintained under vigorous stirring at 353 K for 2 h. The solid thus obtained was filtered off and washed twice with 25 mL of solvent. The solid was dried at 383 K and finally calcined in air at 773 K for 5 h. The labeling of samples is included in Table 1. The titanium content of the TiO2/SiO2 systems was determined using inductively coupled plasma absorption spectrometry, with a Perkin-Elmer Optima 3300 DV instrument. Ultravioletvisible spectra were measured on a Varian Cary 5000 UV-vis (48) Grunwaldt, J.-D.; Beck, C.; Stark, W.; Hagenc, A.; Baiker, A. Phys. Chem. Chem. Phys. 2002, 4, 3514. (49) Waychunas, G. A. Am. Mineral. 1987, 72, 89. (50) Luca, V.; Djajanti, S.; Howe, R. F. J. Phys. Chem. B 1998, 102, 10650. (51) Mountjoy, G.; Pickup, D. M.; Anderson, G. W.; R.; Cole, J. M.; Newport, R. J.; Smith, M. E. Chem. Mater. 1999, 11, 1253. (52) Fraile, J. M.; Garcia, J. I.; Mayoral, J. A.; Salvatella, L.; Vispe, E; Brown, D. R.; Fuller, G. J. Phys. Chem. 1996, 100, 19484. (53) Thomas, J. M.; Raja, R.; Lewis, D. W. Angew. Chem., Int. Ed. 2005, 44, 6456 and references cited therein. (54) Sinclair, P. E.; Sankar, G.; Catlow, C. R. A.; Thomas, J. M.; Maschmeyer, T. J. Phys. Chem. B 1997, 101, 4232. (55) Jarupatrakorn, J.; Tilley, T. D. J. Am. Chem. Soc. 2002, 124, 8380.
DOI: 10.1021/la900578u
7149
Article
Capel-Sanchez et al.
spectrophotometer equipped with an integrating sphere. A BaSO4 disk was used as reference. All spectra were acquired under ambient conditions. Infrared spectra were recorded at room temperature on a Nicolet 7600 FTIR spectrophotometer. The FTIR spectra were obtained using self-supporting wafers of the samples (ca. 10-12 mg/cm2) previously degassed under high vacuum at different temperatures for 1 h in a glass cell fitted with greaseless stopcocks and KBr windows. For each spectrum, 100 scans were accumulated at a spectral resolution of 4 cm-1. X-ray photoelectron spectra were recorded on a VG Escalab 200R spectrometer equipped with a hemispherical electron analyzer and a magnesium X-ray source (12 kV and 10 mA) (Mg KR = 1253.6 eV). The powder samples were packed into small aluminum cylinders and mounted on a sample rod in the pretreatment chamber and degassed at 573 K for 1 h. The base pressure of the ion-pumped analysis chamber was maintained below 3 10-9 mbar during data acquisition. Peak intensities were estimated by calculating the integral of each peak after smoothing and subtraction of the “S-shaped” background. All binding energies (BEs) were referenced to the adventitious C 1s line at 284.9 eV. This reference gave BE values with an accuracy of (0.1 eV. Ti K-edge XANES spectra were recorded at the X-19A beamline facilities at the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory (BNL). The X-ray absorption spectra were taken repeatedly in the “fluorescence yield mode” using a special cell with a PIPS (passivated-implanted planar silicon) detector.56 All the XANES spectra were taken at a constant scan step of 0.5 eV through the edge region, and a Ti foil was employed to calibrate the energy. During the measurements, the photon flux was detuned by 40% in order to avoid the disturbance of the higher order harmonics. Samples were placed in an in situ cell and dried in air at 393 K to remove adsorbed water. All thorough analysis of the pre-edge adsorption energy region was performed by fitting the experimental spectrum to four Gaussian lines, which correspond to the transitions A1, A2, A3, and B defined by Grunes.46 Previously, a baseline was subtracted with a Gaussian curve to correct the edge rising. Epoxidation reactions were carried out batchwise in a mechanically stirred 250 mL thermostatted glass reactor equipped with a thermometer, a reflux condenser, and a septum for withdrawing samples. In a typical run with hydrogen peroxide, alkene (0.2 mol), tert-butanol (11 g), and 1 g of catalyst were mixed in the reactor and the suspension was heated at 333 K, whereupon 4 g of a 5 wt % organic solution of H2O2 (in tert-butanol) was added dropwise to the reactor suspension while maintaining vigorous stirring over 0.5 h. The 5 wt % organic solution of H2O2 was prepared by mixing a 70 wt % H2O2 solution in water (kindly provided by Solvay Quı´ mica S.A.) and tert-butanol. The final content of water of this organic solution is very low (2.5 wt %). In a typical run of catalytic epoxidation of 1-octene with ethylbenzene hydroperoxide (EBHP), 45 g of alkene (0.4 mol) and 2 g of catalyst were mixed in the reactor, and the suspension was heated at 393 K. Subsequently, 33 g of a solution of EBHP (33 wt %) in ethylbenzene (0.08 mol of EBHP), kindly provided by Repsol-YPF, was added dropwise. The organic compounds were analyzed on a GC-FID HewlettPackard 6890-plus device equipped with a HP-WAX capillary column. Hydrogen peroxide and ethylbenzene hydroperoxide consumptions were evaluated by standard iodometric titration, using a Radiometer VIT-90 titrator. Epoxide selectivity was related to the hydrogen peroxide or hydroperoxide converted according to the equations:
S ð%Þ ¼ 100 ½epoxide=ð½EBHP0 -½EBHPÞ
ð2Þ
where epoxide represents the 1,2-epoxyoctane, the subscript 0 stands for initial values, and all concentrations are expressed on a molar basis. No other product derived from the olefin was detected.
ð1Þ
Results and Discussion All samples were prepared with the same molar amount of precursor. Assuming that the precursor was quantitatively incorporated to the silica surface, 1.88 wt % of this is the expected loading in each sample. However, the amount of titanium incorporated into the samples (Table 1) depends on the titanium precursor used in the synthesis step. In both cases, only a fraction of the titanium precursor was incorporated into the catalyst. While for the TY system only 1% Ti was loaded, the titanium isopropoxide precursor yielded a higher Ti-loading (1.6%). This observation is related with to relative reactivity of each precursor against hydrolysis. Titanium (triethanolaminate) isopropoxide is less reactive than titanium isopropoxide, and hence, the grafting reaction by hydrolysis of the precursors with surface OH groups takes place to a lesser extent. The chemical state, coordination, and relative dispersion of titanium on the silica surface were evaluated by photoelectron spectroscopy. High-resolution photoelectron spectra of the Ti 2p core levels of the samples degassed in situ at room temperature (RT) and 573 K were recorded. The spectra display the characteristic spin-orbit splitting of Ti 2p levels (Ti 2p3/2 and Ti 2p1/2) in Ti compounds. As chemical information can be derived from an analysis of the binding energies of each level, we focus solely on the most intense Ti 2p3/2 component of the doublet. Peak fitting of the experimental Ti 2p3/2 component revealed two contributions: one at high binding energy (460.0 eV), typical of TiIV ions tetrahedrally coordinated by oxide anions,57 and another at lower binding energies (458.5 eV), usually assigned to titanium in octahedral coordination and/or Ti(IV) interacting with adsorbed water or pentacoordinated species.57 The XP spectra of the samples degassed at RT are shown in Figure 1. All the calcined and uncalcined samples recorded varying proportions of tetrahedral and octahedral titanium species. The differences in the samples are shown in Table 2, and the percentages of these peaks are shown in brackets. For the uncalcined samples, the proportion of titanium in octahedral or pentacoordinated sites is higher than the titanium in tetrahedral sites (Table 2). After calcining, the proportion between tetrahedral and octahedral titanium in both samples changes and the major component appears at 460.0 eV (Figure 1), typical of titanium in a tetrahedral environment.21,42,57 Table 2 also lists the Ti/Si surface atomic ratios derived from the XPS measurements. As expected, the concentration of titanium ions on the surface increases with increasing titanium loading. Figure 2 depicts the XPS spectra of the samples degassed at 573 K. All the calcined and uncalcined samples recorded different proportions of tetrahedral and octahedral titanium species. As expected, after degassing at 573 K, the proportion of the Ti2p3/2 peak component at high binding energy (460.0 eV) increased and the Ti2p3/2 peak component at lower binding energies (458.5 eV) decreased as compared with the spectra recorded upon degassing at RT (Table 2). This is due to the presence of water molecules adsorbed at the titanium center at room temperature that distort the information about the coordination of titanium. The calcined samples display a higher ratio of titanium species in
(56) Rodriguez, J. A.; Hanson, J. C.; Kim, J.-Y.; Liu, G.; Iglesias-Juez, A.; Fernandez-Garcı´ a, M. J. Phys. Chem. B 2003, 107, 3535.
(57) Blasco, T.; Camblor, M. A.; Fierro, J. L. G.; Perez-Pariente, J. Microporous Mater. 1994, 3, 259.
S ð%Þ ¼ 100 ½epoxide=ð½H2 O2 0 -½H2 O2 Þ
7150 DOI: 10.1021/la900578u
Langmuir 2009, 25(12), 7148–7155
Capel-Sanchez et al.
Article
Figure 1. XP spectra in the Ti 2p core level region of samples outgassed at RT.
Figure 2. XP spectra in the Ti 2p core level region of samples outgassed at 573 K.
Table 2. Internal Core Binding Energy of Titanium Degassed at Different Temperatures and Surface Atomic Ratio Measured by XPS
of testing titanium coordination.24,45 As a general rule, LMCT in titanium compounds containing octahedrally coordinated Ti(IV) takes place at wavelengths (λ > 300 nm) higher those than in compounds in which the titanium ions have only a tetrahedral coordination (200-250 nm). The DRS UV-vis spectra of the samples degassed at 573 K (Figure 3) are clearly different among the catalysts. The spectrum of TYc has a fairly narrow band centered at 215 nm; this peak is a clear indication of the presence of titanium in tetrahedral coordination.24,45 The electronic spectrum for the TiPc sample exhibits a broad band centered at 225 nm, which is associated to a mixture of isolated tetrahedral titanium species and polymeric titanium surface species.45,63 The fact that uncalcined counterparts (TY and TiP) display a broader adsorption band shifted to high wavelengths can be taken as indicative of a change in the coordination of titanium species.24,45 The absence of a band at 370-410 nm rules out the presence of free TiO2 moieties in all the samples. The FTIR spectra of the OH stretching vibration mode, in the 4000-3000 cm-1 region, of self-supported wafers degassed at 573 K of bare silica, used to prepare the samples, and the calcined samples are shown in Figure 4. The spectrum of silica shows a broad band extending between 3500 and 3700 cm-1. This band comes from H-bridging hydroxyl (-Si-OH 3 3 3 O-Si-) groups and is accompanied by a sharp band at approximately 3750 cm-1 arising from isolated silanol (-Si-OH) groups. In the FTIR spectra of calcined samples, the band from H-bridging hydroxyl groups decreases strongly. This is a clear indication that the titanium precursor is preferentially anchored over the silica surface by the bridging hydroxyl groups.
BE Ti 2p3/2 (eV) sample TiP TiPc TY TYc
atomic ratio
RT
573 K
Ti/Si
O/Si
460.0 (27) 458.5 (73) 460.0 (65) 485.5 (35) 460.0 (44) 458.5 (56) 460.0 (72) 458.5 (28)
460.0 (41) 458.5 (59) 460.0 (78) 485.5 (22) 460.0 (63) 458.5 (37) 460.0 (100)
0.03
1.88
0.03
1.95
0.02
2.02
0.02
1.94
C/N
N/Ti
6.02
1.98
tetrahedral coordination. This fact is more evident in the TYc sample, in which the component associated to titanium in a tetrahedral environment is observed. See Table 1 for the labeling of samples. Likewise, the electronic spectra recorded from samples exposed to ambient environment are severely distorted by adsorbed water molecules onto the titanium center. As the true coordination of titanium species is complicated by the coordination of water molecules to the titanium sites,44 samples have been degassed at 573 K prior to recording their electronic spectra. UV-vis spectroscopy is one of the most useful tools for studying the titanium environment. The electronic spectra record an absorption associated with the ligand-to-metal charge transfer (LMCT) from the oxygen to the Ti(IV) ion: Ti4+O2- f Ti3+O-. The wavelength at which this transition occurs is highly sensitive to the coordination of titanium sites, and this peculiarity has been proposed as a way Langmuir 2009, 25(12), 7148–7155
DOI: 10.1021/la900578u
7151
Article
Capel-Sanchez et al.
Figure 3. DRS UV-VIS spectra of solids degassed at 573 K.
Figure 5. FTIR spectra of self-supported wafers of the TiP samples.
Figure 4. FTIR spectra of OH vibration regions of silica and calcined samples outgassed at 573 K.
The FTIR spectra of self-supported wafers of both uncalcined and calcined samples degassed under high vacuum at 373 K for 1 h are shown in Figure 5 (TiP and TiPc) and Figure 6 (TY and TYc). Uncalcined TiP (Figure 5) and TY (Figure 6) samples were also degassed at different temperatures from 373 to 823 K to follow the precursor modification. The energy region of stretching vibrations of hydroxyl groups for calcined samples has a sharp peak at 3700 cm-1 typical of isolated silanol (-Si-OH) groups.42 The FTIR spectra of uncalcined samples (Figures 5 and 6) show the characteristic bands of the C-H stretching vibration (in the region 2800-3000 cm-1), and the other between 1430 and 1550 cm, corresponding to C-H bond deformation vibrations of 7152 DOI: 10.1021/la900578u
ligands anchored to the silica surface during the preparation process.58 These bands are absent in calcined samples. The FTIR spectra of uncalcined samples degassed at different temperatures, from 373 to 873 K for TiP (Figure 5) and from 373 to 723 for TY (Figure 6), revealed the temperature at which the hydrocarbon fragments anchored to the silica surface are removed. The C-H vibration suppression signal occurs at different temperatures depending on the precursor employed. Thus, for TiP, this temperature is 723-823 K, while it decreases to between 623 and 723 K for TY. The isopropoxide precursor is somewhat more difficult to remove than the atrane. The complete disappearance of the IR bands due to the vibration modes of C-H bonds at 823 K for TiP and 723 K for TY indicates that nearly all the (58) Bouh, A. O.; Rice, G. L.; Scott, S. L. J. Am. Chem. Soc. 1999, 121, 7201.
Langmuir 2009, 25(12), 7148–7155
Capel-Sanchez et al.
Article
Figure 6. FTIR spectra of self-supported wafers of the TY samples.
Figure 8. Fits of the pre-edge region of XANES spectra measured at ambient temperature in air.
Figure 7. Ti K-edge XANES spectra of samples as synthesized and calcined.
organic groups have been removed from the silica surface. In addition, a clear increase in the signal at 3700 cm-1 in calcined samples is attributed to isolated silanol groups. This phenomenon can be related to changes in the proportion between bonded and isolated silanol groups due to reaction between adsorbed titanium moieties and surface OH, producing Ti-O-Si bonds.58 Figure 7 shows Ti K-edge XANES spectra of samples in both uncalcined and calcined states. Near-edge spectra for all samples show a dominant pre-edge feature at a photon energy of around 4970 eV. It has been shown that the shape, position, and intensity of XANES pre-edge absorption are related to the coordination symmetry of Ti-atoms in Ti oxides59 and silicates.60 For this reason, the spectra presented here are compared with those (59) Waychunas, G. A. Am. Mineral. 1987, 72, 89. (60) Farges, F.; Brown, G. E.; Rehr, J. J. Phys. Rev. B. 1997, 56, 1809.
Langmuir 2009, 25(12), 7148–7155
previously reported in the literature, whose coordination is wellknown.36 Spectra of uncalcined samples differ markedly from those of their calcined counterparts. For the calcined samples, spectra are similar to each other and have a sharp pre-edge peak, usually attributed to titanium in tetrahedral coordination. The uncalcined TiP and TY samples show a broad complex pre-edge peak. The near-edge spectrum of TiP clearly corresponds with titanium species in octahedral coordination with an environment similar to that of nanosized TiO2 anatase.36 However, the spectrum of the TY sample is better resolved and appears a little more intense. These characteristics can be taken as conclusive that Ti(IV) remains in a pentacoordinated environment. As indicated in previous works,36,61 these qualitative observations on titanium coordination can be analyzed in more detail, especially for highly dispersed phases in which pentacoordinated Ti(IV) structures are dominant. The detailed analysis includes convolution of the preedge peak in several components corresponding to individual electronic transitions defined by Grunes.46 Figure 8 shows the fits of the pre-edge region of XANES spectra of the samples measured at ambient temperature in air. For all spectra, the pre-edge region is resolved into four Gaussian peaks corresponding to different electronic transitions and labeled in order of ascending energy A1, A2, A3, and B. The baseline was subtracted with a Gaussian curve to correct the edge rising. Specifically, the energy positions of transitions in A2 and A3 and their combined area relative to that of all pre-edge peaks can be used to measure Ti 3d level occupancy and consequently its coordination.36,61 As in the case of the overall pre-edge feature intensity and location, these relative peak areas can be related empirically to structures of known Ti coordination number.36 As shown in Figure 8, the B transition is difficult to distinguish in samples TYc, TiPc, and TY, suggesting (61) Stewart, S. J.; Fernandez-Garcia, M.; Belver, C.; Mun, S. B.; Requejo, F. G. J. Phys. Chem. B 2006, 110, 16482.
DOI: 10.1021/la900578u
7153
Article
Capel-Sanchez et al. Table 3. Oct-1-ene Epoxidation with Hydrogen Peroxide after 1 h of Reaction (T = 353 K) sample
H2O2 conversion (%)
selectivity to epoxide (%)
96 97 87 99
15 80 3 95
TiP TiPc TY TYc
Table 4. Oct-1-ene Epoxidation with Ethylbenzene Hydroperoxide after 2 h of Reaction (T = 393 K) sample EBHP conversion (%) selectivity to epoxide (%) TON (h-1) TiPc TYc
Figure 9. Position and relative area of A2 + A3 components for Ti-containing reference material taken from the literature and samples analyzed.
that titanium is in different coordination than octahedral. However, as this transition is clear for sample TiP, it is indicative that a significant part of titanium is in octahedral coordination. Sound conclusions can be drawn from Figure 9. Both calcined samples fit perfectly with the literature36 reference of tetrahedral coordination (Ba2TiO4). The opposite happens with uncalcined samples. The energy position of the absorption peak of the TY sample indicates pentacoordinated Ti(IV) sites, and it is very close to fresnoite (pentacoordinated reference).36 By combining this information with other characterization techniques (XPS, IR, elemental analysis, UV-vis), data are conclusive that the atrane precursor is present on the surface of the uncalcined sample. However, a small difference in energy position is observed with respect to the reference material: this effect could be attributed to the presence of a nitrogen atom coordinated to titanium, in contrast with the sole presence of oxygen in fresnoite. Indeed, changes in the pre-edge peak energy position have been associated to changes in the nature of the ligands surrounding titanium.62 Finally, the absorption peak of the uncalcined TiP sample belongs to a photon energy corresponding to distorted 6-coordinated Ti (IV) sites, but its relative proportion of A2 + A3 components area is too high. This effect could be attributed to the small size of the particles. This behavior can be related to the nature of the precursor, titanium(IV) isopropoxide, which is very reactive and has a tendency to form oligomeric structures, that is, two or even more Ti(IV) moieties in the structure. This situation contrasts with the atrane precursor that has a chelant ligand blocking the ability to polymerize. To minimize this oligomerization, use is recommended of very diluted solutions of titanium(IV) isopropoxide, as reported previously in the literature.21,42-45,63 After calcination, this highly dispersed polymeric titanium with 6-coordination yields titanium species in tetrahedral coordination (TiPc) but forms nonisolated Ti-O-Ti bonds (see UV-vis spectrum). Finally, there is currently widespread agreement that these species are noneffective in the epoxidation of alkenes with hydrogen peroxide.12 Hydrogen peroxide conversion and selectivity to epoxide of the catalysts are shown in Table 3. All the Ti-containing silica samples recorded very high hydrogen peroxide conversions. The uncalcined samples recorded very low selectivity values toward epoxide. This trend in catalytic behavior is in complete (62) George, S. D.; Brant, P.; Solomon, E. I. J. Am. Chem. Soc. 2005, 127, 667. (63) Gao, X.; Bare, S. R.; Fierro, J. L. G.; Banares, M. A.; Wachs, I. E. J. Phys. Chem. B 1998, 102, 5653.
7154 DOI: 10.1021/la900578u
93 94
77 93
806 1048
accordance with the characterization data, where uncalcined catalysts have a high proportion of octahedrally coordinated Ti (IV) species (UV-vis, XPS, and XANES). The selectivity for the uncalcined TY sample is nearly zero, which is due to the presence of nitrogen in the structure and to the leaching of Ti in the reaction. The same occurs in the TiP sample, albeit to a lesser extent due to the lower coordination of alkoxides in comparison with triethanolamine precursor. However, the calcined samples show a high selectivity to epoxide. The catalyst TYc, calcined prepared with titanium (triethanolaminate) isopropoxide, has the highest selectivity to epoxide. Similar results were obtained in the epoxidation of octene with ethylbenzene hydroperoxide (EBHP). This reaction is of high industrial importance, as it is the core of the PO/SM process. EBHP conversion and selectivity to epoxide for both calcined catalysts are collected in Table 4. Both catalysts show a very high EBHP conversion. The selectivity to epoxide for calcined samples is very high for both catalysts but substantially higher for the TYc catalyst, which is in agreement with the higher proportion of titanium in tetrahedral coordination in this catalyst. The turnover number (TON) value for conversion = 0 (TON0) was calculated for each catalyst showing a high value (Table 4). However, this value is difficult to compare with previously published results because there are some differences in the reactivity of alkenes or hydroperoxides employed. In short, titanium incorporation by grafting onto silica is produced by reaction with surface OH groups present on the support. This behavior is consistent with the IR studies of the uncalcined samples treated at different temperatures that record a removal of organic species with a decrease of surface OH groups due to the reaction with titanium centers. However, the species present on the uncalcined samples and the final catalysts depend clearly on the titanium precursor. When titanium isopropoxide precursor is employed, a polymeric surface species is formed (Figure 10); this observation is consistent with previous work.58 This structure gives titanium moieties mainly in polymeric tetrahedral coordination (Figure 10), as observed by characterization techniques. Due to the presence of polymeric tetrahedral titanium species, the catalyst obtained yields only a moderate efficiency of hydrogen peroxide (80%) in the epoxidation of alkenes with H2O2, as this species decomposes hydrogen peroxide. In contrast, surface species are clearly different when a strongly coordinating precursor is employed (titanium (triethanolaminate) isopropoxide). Uncalcined samples present pentacoordinated titanium similar to the starting material (Figure 11), which is anchored to the solid by reaction with one or two OH groups.36 These structures give rise to the formation of isolated titanium species in tetrahedral coordination. Due to the presence of these titanium species, this catalyst is very active and efficient in the epoxidation of alkenes with hydrogen peroxide (>95%). Langmuir 2009, 25(12), 7148–7155
Capel-Sanchez et al.
Article
Figure 10. Scheme of surface species present in the sample prepared with titanium isopropoxide.
Figure 11. Scheme of surface species present in the sample prepared with titanium (triethanolaminate) isopropoxide.
Conclusions On the basis of data presented in this work, the following conclusions can be drawn: (i) Characterization techniques indicate that the grafting of titanium is produced by reaction between alkoxide precursor and surface OH groups, as the FTIR and XPS spectra have shown. Nevertheless, there is a strong influence between the nature of the precursor and the surface species generated in the catalysts by chemical grafting of titanium onto the silica surface. (ii) Samples prepared with titanium
Langmuir 2009, 25(12), 7148–7155
isopropoxide yield polymeric titanium species in tetrahedral coordination, which give rise to a low-efficiency catalyst. However, when an atrane is used (titanium (triethanolaminate) isopropoxide), the isolated titanium species is obtained before and after calcination due to the protective effect of the triethanolaminate ligand that makes titanium species much less reactive toward polymerization. (iii) As confirmed by Ti K-edge XANES, UV-vis, and XPS spectra, the Ti atrane precursor yields pentacoordinated titanium species, which anchor to the surface by reaction with one or two OH groups. As the resulting catalyst after calcination is very active and efficient in the epoxidation of alkenes with hydrogen peroxide (>95%) and ethylbenzene hydroperoxide, it is inferred that isolated titanium species in tetrahedral coordination are related to active sites for this reaction. Analysis of TON has shown a quite high value for the TYc sample while it decreased for TiPc, which can be related to the differences in species present in the catalyst. (iv) Due to the close parallelism between catalyst performance and nature of the active sites, spectroscopic information can be taken in a first instance for catalyst screening. Acknowledgment. The authors acknowledge financial support from Repsol-YPF (Spain) and the Spanish Ministry of Science and Education in Projects PSE-310200-2006-2 and FIT-3201002006-88. G.B.-B. and M.C.C.-S. gratefully acknowledge fellowships granted by Repsol-YPF. The research carried out at Brookhaven National Laboratory was supported by the U.S. Department of Energy (Chemical Sciences Division, DE-AC0298CH10886).
DOI: 10.1021/la900578u
7155
