The use of SBA-15-supported Ru catalysts in reactions such as hydrogenation [1, 2] and hydrogenolysis [3, 4] is becoming increasingly important because of the good performance of such catalysts in H₂ activation. The surface area of SBA-15 is higher than those of previously reported supports for Ru catalysts [5, 6], and it has an ordered mesoporous structure, which favors high dispersion of Ru. SBA-15 has few acidic or basic sites and did not affect the studied reactions. The high reactivity of Ru/SBA-15 is therefore normally attributed to high dispersion of small Ru particles [7]. However, strong binding of spillover hydrogen to SiO₂ has been detected on SiO₂-supported Ru catalysts [8]. Moreover, the addition of Ru as a promoter in SiO₂-supported catalysts improves the reducibilities and activities of the catalysts; this is also ascribed to hydrogen spillover [9]. These findings emphasize the importance of hydrogen spillover in reactions involving hydrogen, and raise the following issues in understanding of catalysts under working conditions. (1) Are the active sites for hydrogenation on Ru/SBA-15 catalysts Ru metal particles or the support? (2) How does hydrogen spill over from Ru to SBA-15? However, to the best of our knowledge, such a fundamental understanding of reaction mechanisms and the nature of active sites is still lacking.

In situ or in operando experimental methods need to be used to address these issues of real working catalysts [10]. In situ infrared (IR) spectroscopy is a powerful non-destructive technique, and is particularly used to probe surface sites and reaction mechanisms at the molecular level during catalytic reactions [11]. For example, IR spectroscopy has shown that benzene adsorption increases back-bonding of Pt d electrons to CO antibonding orbitals, based on the shift in the ν(CO) vibration to lower frequencies. This indicates that benzene is adsorbed on a single Pt atom (as in π arene complexes) with global transfer of electrons to Pt [12]. In situ IR spectroscopy has also shown that during the catalytic cycle in Baeyer-Villiger oxidation on Sn-zeolite beta, a ketone is coordinated to the Lewis acid center, resulting in carbonyl group activation [13].

Here, we report that the active sites for hydrogenation reactions over Ru/SAB-15 can be identified using in situ Fourier-transform IR (FTIR) spectroscopy. The formation of these active sites is probably the intermediate step in hydrogen spillover.

SBA-15 was synthesized using a previously reported method [14]. P123 (2 g) was completely dissolved in 1.6 mol/L HCl (75 mL), and tetramethyl orthosilicate (3.3 g) was added under vigorous stirring; stirring was continued for about 20 h. The temperature was kept at 40 ℃. The resultant white mixture was transferred to a Teflon-lined autoclave and aged at 100 ℃ for 24 h without stirring. The white powder was recovered, washed, and dried at 110 ℃ for 6 h. Calcination at 550 ℃ for 10 h in air gave mesoporous SBA-15.

Supported Ru catalysts with specific Ru loadings were prepared using a two-solvents method [15, 16]. Typically, an aqueous solution of RuCl₃ (3.0 mL) was added gradually to cyclohexane (25.0 mL) containing a support material (SBA-15 or Al₂O₃, 1.0 g) at 25 ℃. After vigorous stirring for 15 min, the dark-brown solution became colorless and the originally white support became black. The black powder was separated by centrifugation and dried at 100 ℃ for 6 h; the product is denoted by Ru/support (wt%). Pd/SBA-15 (4%) was prepared by impregnation with PdCl₂.

In situ FTIR spectroscopy was performed using a Thermo Nicolet NEXUS 6700 spectrometer equipped with a mercury cadmium telluride (MCT) detector. A custom-made IR cell enabled spectra to be recorded at ambient temperature and at catalyst activation temperatures (up to 500 ℃). Self-supporting disks (9.0 mg) were prepared for IR studies and physisorbed molecules were removed using a He flow. FTIR spectra were collected with 32 scans at a resolution of 4 cm^-1 at different time intervals until a steady state was reached. The catalyst was reduced in situ with H₂ at 400 ℃ for 30 min at a ramping rate of 20 ℃/min, swept with He for 30 min, and cooled to 110 ℃ under He. These spectra were used as background for obtaining differential FTIR spectra.

The catalyst was subsequently treated using the following procedures; the gas flow rate was kept at 30 mL/min, unless otherwise stated. (a) H₂ adsorption. The sample was treated in a H₂ flow at 110 ℃ until a stable FTIR spectrum was obtained. Blank SBA-15 was treated in the same way for comparison. (b) H-D exchange. After procedure (a), the gas was switched to D₂ to achieve H-D exchange. Spectra were collected until no change in the OH vibration peak was detected. The sample was then swept with He to expel D₂ until a clean spectrum was obtained. The sample was then retreated with D₂ and IR spectra of the sample without any interference from H₂ were obtained. (c) CO adsorption. After the catalyst had been activated and cooled to 110 ℃ under He, CO was pulsed into the He stream three times using a six-way valve. After each pulse, the sample was swept with He for more than 10 min and spectra were recorded. After three pulses, the sample was swept with He for more than 8 h.

Transmission electron microscopy (TEM) was performed using a Tecnai G2 F30 S-Twin instrument (Philips-FEI Company) operated at 300 kV. After H₂ adsorption, the samples were dispersed in ethanol and then supported on a Cu grid for TEM measurements.

The in situ FTIR spectra in the range 2250-1850 cm^-1 of the reduced samples after adsorption of H₂ at 110 ℃ are shown in Fig. 1. No vibration bands from Pd/SBA-15 (4%), Ru/Al₂O₃ (4%), and SBA-15 were detected in this range, but a peak at 1996 cm^-1 was observed for Ru/SBA-15 (4%). These results imply that formation of this band is associated with the presence of both Ru and SBA-15, but is not related to the Ru-H bonds in Ru metallic clusters, which would be detected for Ru/Al₂O₃. It has been reported that Ru-H species cannot be observed using IR, although adsorbed hydrogen species on various supported metals, e.g., Pt/Al₂O₃, Ir/Al₂O₃, Ni/Al₂O₃, Rh/Al₂O₃, and Ni/SiO₂ have been observed [17, 18]. In addition, the intensity of the band arising from the combination tone vibration [19] of Si-O on Ru/SBA-15 at 1866 cm^-1 did not change after exposure to H₂, thus the intensity of the Si-O overtone band at 1985 cm^-1 was constant. The possibility that the band at 1996 cm^-1 can be ascribed to the Si-O overtone can therefore be ruled out.

CO was adsorbed on Ru/SBA-15 (4%) to enable the band at 1996 cm^-1 to be distinguished from those for adsorbed CO species. Fig. 2(a) shows that at low coverage of the Ru surface with CO, a band at 2068 cm^-1 ascribed to CO linearly adsorbed on Ru^δ+ appeared [20]. This band developed together with a band at 2132 cm^-1, both related to Ruⁿ⁺(CO)_x multi-carbonyls [21], as coverage with CO increased until a band at 2177 cm^-1 indicating physisorbed CO appeared. Few of these chemisorbed CO molecules were removed when the system was flushed with He at 110 ℃, even after 8 h (Fig. 2(b)). In addition, the intensity of the carbonyl band at 2068 cm^-1 created by the first CO pulse (Fig. 2(a)) was lower than that of the band at 1996 cm^-1 shown in Fig. 1. The wavenumber difference is therefore not caused by a surface coverage effect because carbonyl bands with wavenumbers higher than 2068 cm^-1 are expected to appear with increasing CO coverage. The CO pulse adsorption results prove that the band at 1996 cm^-1 does not arise from any carbonyl species formed on Ru.

Formation of the band at 1996 cm^-1 was reversible in the case of Ru/SBA-15 (4%). The band developed gradually when the catalyst was exposed to H₂ and reached its maximum intensity after 60 min (Fig. 3(a)). Subsequent flushing with He led to the disappearance of this band, as shown in Fig. 3(b). This band therefore cannot be assigned to spillover hydrogen on SBA-15 because the migration of hydrogen from Ru to SiO₂ is irreversible [8]. The slow growth of this band during exposure for 1 h also indicates that it cannot be attributed to hydrogen chemisorbed on Ru because hydrogen saturates the Ru particle surfaces in a shorter time (less than a few minutes) [8].

Based on the above discussion, we conclude that the band at 1996 cm^-1 cannot be ascribed to adsorbed H species on either Ru metal particles or the SBA-15 support. We postulate that the adsorbed H species are located at the interface between the Ru metal particles and the support. To confirm this hypothesis, we performed experiments using Ru with various interfacial circumferences on SBA-15 by tuning the Ru metal particle size. If the amount of H species is proportional to the interfacial circumference of Ru, these species reside along the Ru-SiO₂ interface on the periphery of the exposed Ru particles. Fig. 4 shows TEM images of Ru/SBA-15 catalysts with various Ru contents. The mean Ru particle sizes calculated from 100 to 150 Ru particles in these TEM images are listed in Table 1. Based on a spherical model (equations (1) and (2)), the Ru particle circumferences were calculated using equation (3); they are presented in Table 1.

(1)

(2)

(3)

where ρ is the density of Ru (12.37 g/cm³), N is the number of Ru particles, r is the Ru particle radius, ω is the Ru loading, and C is the total circumference of the Ru particles.

Fig. 5 shows the evolution of the in situ FTIR spectra of reduced Ru/SBA-15 with various Ru loadings at 110 ℃. The IR spectra of Ru/SBA-15 (4%) are shown in Fig. 3(a). The band areas at surface saturation with H₂ were calculated to be 0.578, 3.01, and 3.83 for Ru/SBA-15 (1%), Ru/SBA-15 (4%), and Ru/SBA-15 (10%), respectively (Table 1). It can be seen from Table 1 that the band area at saturation increased linearly with increasing Ru particle circumference, indicating that the species corresponding to the bands in the range 2100-1800 cm^-1 reside at the interfacial sites between Ru particles and the support. For the sake of convenience, we denote these sites by S_x. A precondition of the formation of special species is the presence of both Ru metal and SiO₂ in the hydrogen atmosphere, therefore the process can be represented by the proposed model shown in Scheme 1. Molecular H₂ dissociates on active Ru metal sites and saturates the Ru surface within a few minutes. The dissociated hydrogen atoms then migrate from the Ru metal sites to the S_x sites between Ru and SiO₂; this process takes a longer time. The migration is also dependent on the Ru particle size; the smaller the Ru particle size is, the faster the migration of H atoms from Ru metal sites to S_x is, as shown in Figs. 3 and 5.

The reactivities of these H species were evaluated using H-D exchange experiments. When D₂ was introduced, the intensities of the hydroxyl bands at 3740 and 3660 cm^-1 decreased, whereas the intensities of the deuteroxyl bands at 2760 and 2630 cm^-1 increased, as shown in Fig. 6(a). Some of the hydroxyl groups corresponding to the band at 3660 cm^-1 remained even after a few hours in D₂. No changes in the bands at 2100-1800 cm^-1 were observed. After flushing with a He flow for more than 1 h, the adsorbed deuterium had been completely swept away, as in the case of adsorbed hydrogen in He. We then introduced D₂ again to investigate the nature of the bands in the range 2100-1800 cm^-1. As shown in Fig. 6(b), a band centered at 1987 cm^-1 was formed in D₂ but it was red shifted relative to the band at 1994 cm^-1 in Fig. 5(b), and of much lower intensity than that in H₂; the former change can be ascribed to the deuterium isotope effect [22]. No replacement of H species by D species occurred during H-D exchange (Fig. 6(a)), therefore the much weaker band for D species cannot be attributed to the deuterium isotope effect alone. It mainly arises because D species are less stable than H species.

Based on the above analysis, we propose a possible structure in which an H atom is coordinated to both Ru and Si, as shown in Scheme 2. The formation of this structure is probably the intermediate step in the hydrogen spillover process. The red shift in the Si-H stretching band [23] from 2100 to 1996 cm^-1 is probably caused by the presence of Ru-H bonds from which electrons can be transferred to Si-H species. This electron transfer is responsible for the higher reactivity of Ru-H-Si species compared with those of isolated Si-H species reported in the literature [24-26]. It should be noted that hydrogenation may also take place on the Ru metal particle surfaces, but these active sites are more important when high selectivity toward targeted products is desired, especially when multi-bond hydrogenation is required [27].

In situ FTIR spectroscopy showed the presence of special adsorbed H species on reduced Ru/SBA-15 catalysts but not on Pd/SBA-15, Ru/Al₂O₃, or SBA-15. These H species were differentiated from adsorbed CO species and were reproducible in H₂. The formation rate of these H species depended on the Ru metal particle size, with small Ru particles giving faster formation of these H species. The amount of these H species was proportional to the Ru interfacial circumference, indicating that these species reside along the Ru-SiO₂ interface on the periphery of exposed Ru particles. H-D exchange experiments suggested that these H species were active in hydrogenation. A possible structure for these active sites was proposed.