Gold nanoparticles supported on titanium oxides are highly active catalysts for CO oxidation even at low temperature [1, 2]. Haruta [3] reported that the catalytic performance of gold nanoparticles was controlled by three major factors: the structure of the contact between the gold and the oxide support, the nature of the support, and the particle size, with the contact structure being the most important factor. Despite extensive research on the reaction mechanism and the nature of the active sites on a molecular scale, no consensus has been reached about (ⅰ) the nature of the active sites and the mechanism of activation of O2 molecules and (ⅱ) the role of H2O and its influence on the activities of the catalysts. The origins of the catalytic activity of gold are considered to stem from some combination of contributions from quantum size effects, the presence of low-coordination gold sites, charge transfer between the support and gold, and the formation of a perimeter of the interface between gold and the oxide support.
Although most investigators agree that small metallic gold nanoparticles are the predominant catalytic species [4-6], others have proposed that cationic gold species [7-9], undercoordinated sites on the gold nanoparticles [10-12], or electron-rich gold nanoparticles [13] play an essential role in the reaction. Furthermore, O2 dissociation reportedly occurs when undercoordinated gold atoms are present on extended gold surfaces under model conditions [14, 15]. On the basis of high-intensity in situ X-ray absorption near-edge structure analysis, van Bokhoven et al. [16] indicated that O2 molecules could dissociate on gold nanoparticles supported on Al2O3 and TiO2 substrates. Goodman and coworkers [17, 18] claimed that the TiOx support acted mainly as a dispersant and a promoter and that a bilayer of gold that blocks direct access of the reactants to the support was responsible for CO oxidation. Furthermore, Liu et al. [19] calculated that the barrier to dissociation of O2 from unsupported gold was > 2 eV and that even at the Au/TiO2 interface, the dissociation barrier was still 0.52 eV. These values indicate that O2 interacts weakly with gold, and thus spontaneous dissociation of O2 molecules from the gold surface is not energetically favorable. The perimeter of the interface sites between gold nanoparticles and the oxide support have been recently considered as reaction sites for CO oxidation [20-24]. Behm et al. [21] found that the amount of active oxygen species on the Au/TiO2 surface was linearly related to the number of perimeter sites at the interface between the oxide support and the gold nanoparticles, indicating that the gold-support interface plays a dominant role in O2 activation. Yates et al. [22] reported that CO oxidation over a Au/TiO2 catalyst at low temperature (120 K) occured at a perimeter zone or on the support surfaces. Theoretical studies [25-27] have also highlighted the importance of the periphery of gold particles.
Furthermore, CO oxidation is greatly influenced by H2O in the reactant gas [28-31]. Dáte et al. [30], on the basis of their research and that of others, proposed that H2O had two effects: it activates O2 molecules, and it decomposes a carbonate species. However, the mechanism of the promotional effect of H2O is not fully understood. Here, we describe the reaction mechanism and active sites of CO oxidation over Au/TiO2, along with the role of H2O, on a molecular scale.
The experiments were performed in an ultra-high-vacuum apparatus composed of three chambers: a preparation chamber ( < 3 × 10-10 Torr) equipped with an ion gun for Ar+ sputtering, a cathodic arc plasma source, and an evaporator for Ti deposition. An analysis chamber ( < 1 × 10-10 Torr) held equipment for X-ray photoelectron spectroscopy (XPS) and quadrupole mass spectrometry (QMS). A reaction chamber ( < 1 × 10-9 Torr) was connected to the analysis chamber via a leak valve, to measure the reaction gases by QMS. XPS were measured with Mg Kα radiation.
Single crystals of TiO2(110) (8 × 8 × 0.5 mm, 99.999% purity, SPL) were used as supports for the Au/TiO2 model catalyst, and the crystals were cleaned by three cycles of Ar+ sputtering and annealing at 900 K under a vacuum after oxidation at 900 K for 90 min under 200 Torr of O2. Single-crystal discs of Au(111) and Au(100) (8 mm diameter, 1 mm thickness, 99.999% purity, SPL) were polished on only one side. The polished surface was cleaned by Ar+ sputtering and annealing at 900 K under vacuum.
Gold was deposited on the TiO2(110) surface with a cathodic arc plasma gun (ULVAC, ARL-300) at 300 K, an arc voltage of 70 V, a condenser capacity of 360-2200 μF, and a pressure of 10-9 Torr. The size of the gold particles could be tuned by adjusting the condenser capacity of the arc plasma gun [20, 32]. For example, when a condenser capacity of 360 μF was used, gold nanoparticles with a mean particle diameter of 1.3 nm (composed of approximately 55 atoms) were observed on the TiO2(110) surface. At a condenser capacity of 2200 μF, the size distribution of gold particles was bimodal (4.3 and 5.8 nm with a calculated mean particle diameter of 5.6 nm) [32]. For all condenser capacities used, the deposited gold particles were almost hemispherical, and the shape was confirmed by atomic force microscopy. Thus, the size of gold particles deposited on a single crystal of TiO2 could be controlled in the range of 1-10 nm.
We estimated the number of deposited gold atoms using a quartz microbalance. Then, the XPS of the surface-deposited gold on TiO2(110) were measured under the same deposition conditions to obtain the Au 4f7/2 peak areas. A coverage of 1 monolayer equivalent corresponded a Au(111) surface atom density of 1.39 × 1015 atoms cm-2. We fixed the gold coverage at 1 monolayer equivalent by controlling the generation frequency of the arc.
We also prepared an inverse catalyst by depositing titanium on a Au(111) surface by evaporation from a Ti rod (1.5 mm diameter) with an electron beam evaporator (AVC AEV-1). The Ti coverage was estimated by XPS measurements on the basis of the saturation coverage of atomic oxygen produced by exposure of the Au(111) surface to ozone at 323 K; a coverage of 1 corresponded to a Au(111) surface atom density of 1.39 × 1015 atoms cm-2. We determined the Ti and O coverages from the Ti 2p3/2/Au 4f7/2 and O 1s/Au 4f7/2 peak area ratios, respectively, by using the O 1s/Au 4f7/2 peak area ratio obtained at saturation oxygen coverage (θO=1.1) produced by ozone exposure and the sensitivity factors for O 1s and Ti 2p3/2. The Ti deposition rate was 0.05 min-1 at a constant flux of 10 nA. The TiO2/Au(111) surface was produced by oxidizing the Ti-deposited Au(111) surface at 700 K for 10 min in 3 × 10-7 Torr of O2.
CO oxidation reaction was carried out under 1-25 Torr of CO, 1-625 Torr of O2, and 0-0.5 Torr of H2O at a sample temperature of 270-400 K in a batch reactor. The sample was mounted with two 0.25 mm diameter tantalum wires for resistive heating. The temperature of the sample was measured with an alumel-chromel thermocouple spot-welded to the back of the crystal. The stainless steel walls of the reactor, the sample holder, and tantalum wires showed no activity for CO oxidation in a blank test, that is, reaction with a TiO2(110) sample at 270-400 K. The reaction gases were introduced into the analysis chamber at a constant pressure of 4 × 10-9 Torr through a leak valve, and the pressures of CO, O2, and CO2 were measured in real time during the reaction by monitoring of mass numbers 28, 32, and 44, respectively. The concentrations of CO, O2, and CO2 were calculated from the pressure of each gas and sensitivity factors of 1.06, 1.0, and 1.4, respectively.
First, we investigated the effect of H2O on CO oxidation over Au/TiO2(110) at 300 K, a reaction temperature at which the typical properties of the Au/TiO2 catalyst with respect to CO oxidation can be observed [20]. In the absence of H2O, no CO2 formed, indicating that CO was not oxidized under these reaction conditions. However, in the presence of H2O, the amount of CO2 formed increased linearly with reaction time. The rate of CO2 formation increased with increasing H2O pressure up to 0.1 Torr of H2O, but the amount of CO2 formed decreased at 0.5 Torr of H2O. CO oxidation over gold catalysts is known to be strongly influenced by H2O in the reactant gas [29, 33], indeed our results strongly indicate that CO oxidation did not proceed in the absence of H2O.
To clarify the role of H2O in the CO oxidation, we examined the relationship between the CO2 formation rate and the H2O partial pressure at two reaction temperatures, 300 and 400 K (Fig. 1). At 300 K, the rate initially increased substantially with increasing H2O partial pressure, peaked at 0.1 Torr of H2O, and then gradually decreased with increasing H2O pressure. In contrast, at 400 K, CO oxidation proceeded even in the absence of H2O, and the CO2 formation rate did not depend on the H2O partial pressure. The observation that H2O promoted CO oxidation over the Au/TiO2(110) surface only at low temperature suggests that the mechanisms of CO oxidation at 300 and 400 K differed. We propose that the process by which O2 molecules were activated strongly depended on the reaction temperature. That is, at high temperature, O2 molecules were activated directly over the Au/TiO2(110) surface, whereas at low temperature, H2O took part in the activation.
In early studies, Haruta [34] suggested that the mechanism of CO oxidation over a gold catalyst depended on reaction temperature. Haruta found that the apparent activation energy of CO oxidation over powdered Au/TiO2 changed drastically with reaction temperature (Fig. 2(A)): the estimated activation energy is 2.0±0.7 kJ mol-1 above 320 K, whereas the estimated activation energy below 320 K is 34.0±1.8 kJ mol-1. In this study, we examined the temperature dependence of the rate of CO2 formation over Au/TiO2(110) in the presence of H2O (Fig. 2(B)). An Arrhenius plot of the data clearly showed a sudden change in slope at around 320 K, and we estimated the apparent activation energies above and below 320 K to be 2.9±0.9 and 28.9±2.5 kJ mol-1, respectively. These values and the overall dependence on reaction temperature agreed well with Haruta’s results for a powdered Au/TiO2 catalyst, indicating that Au/TiO2(110) was a good model surface for powdered Au/TiO2 and that the active site and the reaction mechanism over the Au/TiO2(110) surface might change at 320 K.
We next examined the dependence of CO2 formation rate on CO and O2 partial pressures at 300 and 400 K. Increasing the CO partial pressure increased the CO2 formation rate at CO pressures up to 3 Torr, and the reaction order at both temperatures was estimated to be 0.96 at CO pressures below 3 Torr (Fig. 3). Previously, we investigated CO adsorption on Au(111) and Au(100) single-crystal surfaces using polarization modulation infrared reflection absorption spectroscopy and found that CO adsorption on both surfaces showed similar CO-pressure dependences: the peak intensities increase with increasing pressure up to 3 Torr and then remain constant at pressures above 3 Torr [35]. In the same study, we also found that CO adsorption over the gold surfaces showed first-order dependence on CO pressure. Thus, the dependence of the CO2 formation rate on CO pressure observed for Au/TiO2(110) was in good agreement with our previously reported results for CO surface coverage, indicating that the CO2 formation rate was strongly governed by the CO surface coverage. However, the dependence of the CO2 formation rate on CO partial pressure became zero order at CO pressures above 3 Torr, because CO adsorption became saturated at 3 Torr. These results indicate that CO adsorbed on the gold surface was converted to CO2 at both reaction temperatures.
In contrast, the dependence of the CO2 formation rate on O2 partial pressure changed with reaction temperature (Fig. 4). The rate of CO2 formation increased with increasing O2 partial pressure up to 12 Torr at both temperatures. At 300 K, the reaction order was estimated to be 0.97 from the slope of the straight line, whereas at 400 K, the reaction order was estimated to be 0.51, which was approximately half of that at 300 K. This result supports the idea that the mechanism of O2 activation strongly depended on reaction temperature.
We also examined the rates of CO2 formation at 300 and 400 K over Au/TiO2(110) surfaces deposited gold particles of different mean particle diameters (Fig. 5). At 300 K, the CO2 formation rate sharply increased when the particle diameter dropped below 2 nm, whereas at 400 K, the rate increase with decreasing particle size was more gradual. The difference in the particle diameter dependences at the two reaction temperatures suggests that the active sites for CO oxidation at 300 and 400 K differed. Furthermore, we estimated the apparent activation energies of CO2 formation over Au/TiO2(110) surfaces at various mean particle diameters and found that the slopes of the Arrhenius plots were nearly the same at all the particle diameters (Fig. 6). The apparent activation energies for CO oxidation above and below 320 K were estimated to be about 4 and 30 kJ mol-1, respectively, and were independent of the gold particle diameter. This result strongly suggests that the nature of the active sites for CO oxidation on Au/TiO2(110) did not vary with the particle diameter.
To investigate the active sites for CO2 formation, we plotted the turnover frequencies (TOFs) for CO2 formation at the two reaction temperatures as a function of mean gold particle diameter (Fig. 7). To determine whether the active sites for CO oxidation were exposed gold atoms on the gold particles or perimeter sites at the interface between the gold particles and the TiO2 support, we calculated the TOFs in two ways: (ⅰ) by normalizing the number of CO2 molecules formed per second to the total number of exposed gold atoms on the gold particles (TOF-S) and (ⅱ) by normalizing the number of CO2 molecules formed per second to the total number of gold atoms at the perimeter of the interfaces (TOF-P). The number of gold atoms at the interface perimeter was estimated from the length of the perimeter and the gold interatomic distance (0.288 nm). The results clearly show that the relationship between TOF and mean gold particle diameter depended strongly on reaction temperature. At 300 K, TOF-S decreased with increasing particle diameter, whereas TOF-P remained nearly constant regardless of particle diameter (Fig. 7(A)). These results suggest that the active sites for CO oxidation were gold atoms located at the perimeter of the gold particles attached to TiO2 and that the catalytic activity for CO oxidation was correlated neither to a change in the fraction of edge or corner sites nor to a change in the electronic nature of gold particles induced by the quantum size effect. In contrast, TOF-S at 400 K remained nearly constant regardless of the mean gold particle diameter (Fig. 7(B)), suggesting that the active sites for CO oxidation were newly created on the gold metal surface at this temperature. Thus, we concluded that both the reaction mechanisms and the active sites differed between the low-temperature region ( < 320 K) and the high-temperature region ( > 320 K).
To obtain more information about the active sites, we investigated the kinetics of CO oxidation over an inverse catalyst, that is, a Ti-deposited gold single-crystal surface. First, we carried out the CO oxidation reaction over Au(111), Au(100), and TiO2(110) single-crystal surfaces (Table 1). No CO2 formed over the Au(111) and Au(100) surfaces (surfaces not modified with Ti) at 300 K, even when H2O was present in the reaction gas. In contrast, at 400 K, CO oxidation proceeded on both gold surfaces at an estimated rate of 6.4 × 1016 to 6.7 × 1016 molecules s-1. The TOFs, which were calculated by normalizing the number of CO2 molecules formed per second to the total number of surface gold atoms, over Au(111) and Au(100), were estimated to be 48.2 and 52.5 molecules site-1 s-1, respectively; and these values were in good agreement with the value for the Au/TiO2(110) surface (47.8 molecules site-1 s-1). Furthermore, we compared the activation energies for CO oxidation over Au(111) and Au(100) with the activation energy over a Au/TiO2(110) surface at high temperature ( > 320 K). From the slopes of Arrhenius plots of the CO2 formation rates (Fig. 8), we determined the activation energies for CO2 formation on Au(111) and Au(100) to be 3.5±0.3 and 4.5±0.9 kJ mol-1, respectively. These activation energies were almost the same as the activation energy for the reaction over the Au/TiO2(110) surface (2.9±0.9 kJ mol-1). These kinetic data for the Au single-crystal surfaces and the Au/TiO2(110) surface were consistent, indicating that at high temperature, CO was oxidized on the metallic gold surface. Some investigators have suggested that the presence of low-coordinated gold atoms on the surface of nanoparticles may contribute to their high activity for CO oxidation [10, 11]. We believed that during CO oxidation at high temperature, the Au(111) and Au(100) surfaces were reconstructed and that low-coordinated gold atoms appeared. The low-coordinated gold atoms adsorbed O2, which then dissociated and oxidized CO on the metallic gold surface. That is, CO oxidation over the Au/TiO2 catalyst proceeded by the Langmuir-Hinshelwood mechanism, which explains the observed reaction order with respect to O2 partial pressure for CO oxidation at 400 K (CO + 1/2O2 → CO2).
In contrast, we observed CO2 formation over the TiO2-deposited Au(111) surface at low temperature with H2O in the reaction gas. At 400 K, the CO2 formation rate decreased with increasing Ti coverage, and no CO2 formation was observed at coverages of > 1.2 (Fig. 9). The relationship between the CO2 formation rate and Ti coverage agreed well with the relationship between the number of exposed gold atoms and the Ti coverage over TiO2/Au(111). Furthermore, the TOFs for TiO2/Au(111) at 400 K were estimated to be 40.2-44.4 molecules site-1 s-1 (Table 2), and these values are in agreement with those for the Au/TiO2(110) and gold single-crystal surfaces. The agreement demonstrates that the active sites for CO oxidation over the Au/TiO2(110) surface were identical to the metallic gold sites. In contrast, at 300 K, the CO2 formation rate initially increased with increasing Ti coverage, and the rate peaked at a coverage of 0.65. This result clearly indicates that the TiO2 deposited on the gold surface directly promoted CO oxidation at 300 K at a coverage range from 0 to 0.65. However, the deposited TiO2 inhibited the CO2 oxidation reaction at high coverage.
If the perimeter interface between TiO2 islands and the Au substrate acted as the active site for CO oxidation, the TOF (the CO2 formation rate per Au atom at the perimeter of the interface) over the TiO2/Au(111) surface must agree with the TOF for Au/TiO2(110). Thus, we estimated the TOFs for CO2 formation over the TiO2/Au(111) surface at various Ti coverages (Table 2). In our previous study, we found that monolayer TiO2 islands with diameters of about 20 nm formed on Au(111); the number of TiO2 islands increased with increasing Ti coverage below 0.65, and as the coverage was increased further, multilayer islands of TiO2 began to form [36]. Then, we determined the number of gold atoms from the length of the perimeter interface between TiO2 islands and the gold substrate; the length of the perimeter was calculated from the mean diameter and the number of TiO2 islands observed by noncontact atomic force microscopy. The TOFs for TiO2/Au(111) at 300 K were nearly independent of Ti coverage and were estimated to be 73.4-78.6 molecules site-1 s-1, which were consistent with that for Au/TiO2(110). Furthermore, the activation energies for CO2 formation over the TiO2-deposited Au(111) surface were examined at low temperature (Fig. 10). The activation energies for CO2 formation over TiO2/Au(111) at Ti coverages of 0.39 and 0.65 were estimated to be 25.1±2.3 and 26.1±2.7 kJ mol-1, respectively. These values are in good agreement with the Au/TiO2(110) value. Thus, the kinetic data for the reaction over TiO2/Au(111) agreed well with those for the reaction over Au/TiO2(110), strongly supporting the idea that the active site for CO oxidation at low temperature was the perimeter interface between the gold nanoparticles and the TiO2 support and that H2O played an essential role in low-temperature CO oxidation on Au/TiO2.
Both theoretical calculations and experimental data have demonstrated that over gold clusters, OOH can be produced directly from O2 and H2O [37-39]. We thus proposed the reaction mechanism as follows:
First, H2O dissociated to OH and H over the perimeter of the interface between the gold nanoparticles and the TiO2 support (Auδ+-Oδ--Ti). The H atom reacted with O2 to form OOH. The O-O bond was activated by the formation of OOH, and that the reaction with CO to form CO2 occurred subsequently. Recently, Lyubinetsky et al. [40] reported two pathways for H2O interaction with oxygen adatoms on TiO2(110); these investigators observed the recombination reaction of two OH molecules to form H2O and oxygen adatoms over rows of Ti4+. Thus, we believe that in our system, two OH molecules were converted to H2O and O2 by a recombination reaction on Ti4+ sites over the Au/TiO2 surface. However, the O2 molecules formed on the TiO2 surface were not involved in the CO oxidation reaction, because CO cannot be adsorbed on TiO2. This reaction mechanism explains the dependence of the CO2 formation rate on O2 pressure at 300 K (CO + O2 → CO2 + 1/2O2).
The reaction mechanism and active sites for CO oxidation over a Au/TiO2 model surface, gold single-crystal surfaces, and a Ti-deposited gold single-crystal surface, along with the role of H2O, were investigated on a molecular scale. At low temperature ( < 320 K), H2O played an essential role in promoting CO oxidation. The active site for CO oxidation at low temperature was the perimeter of the interface between the gold nanoparticles and the TiO2 support (Auδ+-Oδ--Ti). We believe that the O-O bond was activated by the formation of OOH, which was produced directly from O2 and H2O at the perimeter of the interface between the gold nanoparticles and the TiO2 support, and consequently OOH reacted with CO to form CO2. This reaction mechanism explains the dependence of the CO2 formation rate on O2 pressure at 300 K. In contrast, at high temperature, low-coordinated gold atoms built up on the surface as a result of surface reconstruction due to exposure to CO. The low-coordinated gold atoms adsorbed O2, which then dissociated and oxidized CO on the metallic gold surface.
We are grateful to Professor M. Haruta at Tokyo Metropolitan University for constructive discussions.
2016,
Vol. 37 Issue (10): 1676-1683 
PDF
2016,
Vol. 37 Issue (10): 1676-1683 PDF