Unravelling the Role of Oxygen Vacancies in the Mechanism of the Reverse Water–Gas Shift Reaction by Operando DRIFTS and Ultraviolet–Visible Spectroscopy

1 This document is the Accepted Manuscript version of a Published Work that appeared in final form in ACS CATALYSIS copyright © American Chemical Society after peer review and technical editing by the publisher. To access the final edited and published work see https://pubs.acs.org/doi/10.1021/acscatal.8b02121 Unravelling the Role of Oxygen Vacancies in the Mechanism of the Reverse Water-Gas-Shift Reaction by Operando DRIFTS and UV-Vis Spectroscopy Luis F. Bobadilla † , *, José L. Santos †, Svetlana Ivanova †, José A. Odriozola † and Atsushi Urakawa ‡ † Instituto de Ciencia de Materiales de Sevilla, Centro Mixto CSIC-Universidad de Sevilla, Av. Américo Vespucio 49, 41092 Sevilla, Spain ‡ Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology, Av. Països Catalans 16, 43007 Tarragona, Spain ABSTRACT The reaction mechanism of reverse water-gas shift (RWGS) reaction was investigated using two commercial gold-based catalysts supported on Al 2 O 3 and TiO 2 . The surface species formed during the reaction and reaction mechanisms were elucidated by transient and steady-state operando DRIFTS studies. It was revealed that RWGS reaction over Au/Al 2 O 3 proceeds through the formation of formate intermediates that are reduced to CO. In the case of Au/TiO 2 catalyst, the reaction goes through a redox mechanism with the suggested formation of hydroxycarbonyl intermediates, which further decompose to CO and water. The Ti 3+ species, the surface hydroxyls, 2 and oxygen vacancies jointly participate. The absence of carbonyl species adsorbed on gold particles during the reaction for both catalysts indicates that the reaction pathway involving dissociative adsorption of CO 2 on Au particles can be discarded. To complete the study Operando UV-Vis spectroscopy was successfully applied to confirm the presence of Ti 3+ and to understand the role of the oxygen vacancies of TiO 2 support in activating CO 2 and thus the subsequent RWGS reaction. KEYWORDS. RWGS; Au-based catalysts; Operando; DRIFTS; UV-Vis 1. INTRODUCTION During the last decade increasing interests have been witnessed in developing novel catalytic technologies and processes to utilize carbon dioxide as a C1 building block for the synthesis of valuable chemicals and fuels 1 . Among available options, hydrogenation of CO 2 is one of the most intensively investigated reactions for the broad product scope. Various commercially important products, from C1 molecules (CH 4 , CH 3 OH) to higher molecular weight hydrocarbons and alcohols through C–C bond formation, can be produced by CO 2 hydrogenation 2 . Particularly, reverse water-gas shift (RWGS) reaction (CO 2 + H 2 ⇄ CO + H 2 O) is considered as the key step in the catalytic hydrogenation of CO 2 to produce chemical fuels, since the established industrially important syngas-based reactions (e.g. Fischer-Tropsch and methanol synthesis reactions) become within reach 3 . It is therefore obvious that developing more efficient catalysts and understanding the nature of catalysts under RWGS conditions are of fundamental importance. Pioneered by Haruta 4 , Au-based catalysts are well known for their high catalytic activity in the water-gas shift (WGS) and RWGS reactions when Au is highly dispersed on support materials 5- 3 12 . Support material can influence the electronic properties of active metal centers and it can participate directly in the reaction by interacting with the reactants. It has been proposed that activation of CO 2 occurs at the oxide support or the interfacial sites between the active metal and the oxide support 13 . Through a combined experimental and theoretical study, Yang et al. 14 demonstrated that an electronic polarization at the metal−oxide interface of Au nanoparticles anchored and stabilized on a CeO x /TiO 2 substrate generates active centers for CO 2 adsorption and its low pressure hydrogenation, leading to an optimal catalytic performance. They found that reduced oxides have a strong tendency to activate the CO 2 even causing direct C-O bond scission. Thus, a proper tuning of Au-oxide interface can promote the hydrogenation of CO 2 . However, there are still controversies with regard to the effects of support and it roles on the activity of supported gold catalysts. Generally, two major mechanisms are suggested for RWGS reaction; the regenerative redox mechanism and the associative one. In the redox mechanism, hydrogen is proposed to be a reducing agent without direct participation in the formation of intermediates in RWGS reaction. Normally the support is of reducible nature 15 . The validity of the redox mechanism is based on the fundamental assumption that CO 2 can efficiently re-oxidize the partially reduced support at low temperatures leading to the formation of CO. On the other hand, the associative mechanism suggests that CO is formed from formate intermediates decomposition, derived from the association of hydrogen with CO 2 . Since the formate intermediates cannot be decomposed on Au nanoparticles, their transformation is strongly influenced by the support, according to a report 16 . In addition to the main redox and associative mechanisms, other authors proposed the so-called carbonate mechanism for the RWGS reaction 17-18 . For instance, Millar et al. 17 suggested this mechanism over Cu/SiO 2 catalyst based on a combined temperature programmed desorption 4 (TPD) and infrared (IR) study of CO 2 and H 2 interactions. They found that increasing the availability of activated surface symmetric carbonate species induced an increase in the formation of formate species, stable intermediates in the methanol synthesis or the decomposition to CO and water. Monitoring the surface phenomena occurring over catalyst under reaction conditions is essential to elucidate the relevant reaction steps, active sites and intermediates as well as to establish the relationships between catalytic performance (reactivity/selectivity) and structure of the solid. This knowledge is of great importance for the rational catalyst design and to improve their catalytic properties for specific applications 19 . IR spectroscopy is a very useful technique to identify the adsorbed species and elucidate the reaction mechanisms in heterogeneously catalyzed reactions. Nevertheless, IR spectroscopy has serious drawbacks such as the complexity to distinguish reactive intermediates from spectators. IR spectroscopy may be combined with other characterization techniques such as EPR, NMR, Raman, UV-Vis and XAFS, or integrated with an optical microscope to analyze heterogeneities within and between catalyst grains. On the other hand, UV-Vis spectroscopy is a powerful method for electronic structure studies as it can probe, for example, the electronic d–d transitions for transition metals if present in the catalysts 20 . For instance, Gonzalez-Castaño et al. 7 investigated the influence of the support over Au and Pt catalysts in WGS reaction by means of UV-Vis spectroscopy and they found that the catalytic performance is directly correlated with the electronic properties of the metal/support interactions. Herein, we investigated the possible pathways for CO formation in the RWGS reaction focusing our attention on the role played by the support material over highly dispersed Au catalysts. For this purpose, we selected two model supports: a non-reducible (Al 2 O 3 ) and a reducible (TiO 2 ) one. 5 The differences in the reaction mecanism over the different nature oxides were studied by operando DRIFTS and UV-VIS spectroscopy. 2. EXPERIMENTAL SECTION 2.1. Materials and characterization The catalysts used in this investigation were the commercially available materials (AUROlite™) consisting of 1 wt.% of ca. 2.0 nm Au particles supported on TiO 2 and Al 2 O 3 supplied by STREM in collaboration with Project AuTEK 21 . We employed powder X-ray diffraction (XRD) to analyse the structure of the two Au supported catalysts (see Figure S1 in the Supporting Information). The gold particles size and morphology were studied by Transmission Electron Microscopy (HR-TEM) on a FEI Talos electron microscope using an acceleration voltage of 200 kV with a field emission filament and a side mounted Ceta 16M camera. A standard holey carbon-covered copper TEM grid was used for supporting the catalyst. The average gold particle size for both fresh and spent catalysts, was estimated considering around 200 particles as follows:   = ∑     ∑     (Eq. 1) where n i represents the number of particles with diameter d i . 2.2. Catalytic testing The catalytic activity of the gold catalysts in CO 2 hydrogenation was measured using a tubular fixed-bed flow reactor (3.4 mm inner diameter) made of stainless steel. The catalyst (200 mg) was sieved to the range of 200-300 μm and then loaded in the reactor. Before testing, the fresh catalyst 6 was pre-treated at 250 °C for 1 h using a feed that contained 10 vol.% H 2 in He to reduce and remove moisture from the sample. Then, the reaction mixture was introduced into the reactor using a flow rate of 50 mL min -1 (5 mL min -1 CO 2 and 20 mL min -1 H 2 , H 2 /CO 2 molar ratio of 4) and GHSV of 12,000 h -1 for Au/Al 2 O 3 and 12,000 – 40,000 h -1 range for Au/TiO 2 . The temperature was increased to 450 °C in 25 °C increments and was held constant at each setting for 30 min to ensure the stable isothermal activity. The effluent gas stream was passed into a transmission gas cell of an IR spectrometer (Bruker ALPHA) for quantitative product analysis and estimation of CO 2 conversion. CO was the only observed product. 2.3. Operando DRIFTS measurements Operando DRIFTS experiments were carried out using a reaction chamber (HVC-DRP, Harrick) mounted in a Praying Mantis (Harrick) DRIFTS optical system equipped with ZnSe windows. The spectra were collected in the range 650-4000 cm -1 using a Bruker Vertex 70V FTIR spectrometer equipped with an MCT detector. About 50 mg of catalyst was placed in the cell for each measurement. The flow of gases was controlled by means of mass flow controllers (Bronkhorst). The gases pass through the sample (catalyst) packed-bed, mimicking the plug-flow condition of the reaction. Prior to each measurement, catalyst was pre-treated in situ at 250 °C in H 2 diluted in He stream (5 ml min -1 H 2 and 45 ml min -1 He) for 1 h. To gain insights into the evolution of surface species over the Au catalysts, transient experiments were performed by alternatingly passing 20% H 2 in He and 5% CO 2 in He at 10 mL min −1 for sufficiently long times to stabilize the DRIFTS signal. A four-way switching valve (VICI, 2 position and 4 port switching valve) was used to select the CO 2 or H 2 -containing streams. On the other hand, the reaction of CO 2 hydrogenation was performed by passing 20% H 2 and 5% CO 2 in 7 He (H 2 /CO 2 = 4 molar ratio) at 10 mL min -1 . The spectra were recorded every 30 seconds to follow the reaction and stabilization process of the surface species. Both experiments were carried out at the starting temperature of the catalytic test (250 ºC), where the level of product formation was low and all surface species unstable at high temperatures can be more clearly observed. The effluent gas stream was analyzed on line by the IR spectrometer (Bruker ALPHA). 2.4. Operando UV-Vis spectroscopy The catalytic cell used in the operando UV-Vis study is a modified version of the commercial CCR1000 reactor cell from Linkam Scientific Instruments (Figure 1). A high-temperature reflection UV-Vis probe formed by six radiating optical fibers and one reading fiber was located inside the reactor cell above the catalytic bed to analyze the optical properties of the sample. The experimental device allows to heat the sample under desired atmosphere and the gas effluents are analyzed on-line by mass spectrometry (MS). In contrast to the generally used cell reactors with spectroscopic quartz windows 22-23 , our catalytic cell does not use any window and the cell is sealed by O-rings. Figure 1 includes a 3D schematic representation (Figure 1) of the catalytic cell. The sample (about 50 mg) is positioned in a ceramic holder on a fiber filter and the reaction gas mixture is passing through it from the top to the bottom of the catalytic bed. The temperature is controlled by a Linkam TMS94 temperature controller. 8 Figure 1. 3D schematic representation of operando UV-Vis spectroscopic cell for catalyst characterization under reaction conditions. Operando UV-Vis experiments were performed under steady-state conditions in the range of λ = 200 – 1000 nm using an AVASPEC fiber optical spectrometer (Avantes) equipped with a DH- 2000 deuterium-halogen light source and a charge coupled device (CCD) array detector. Barium sulfate was used as a white reference material. The UV-Visible spectra were recorded at room temperature on a fresh sample and during the reduction treatment (total flow of 50 mL min -1 of 10% H 2 in He) at 250 ºC for 1 h and RWGS reaction (total flow of 50 mL min -1 of H 2 /CO 2 /He = 32/8/60) in the 150-450 ºC temperature interval after stabilizing the gas flows and temperature. The spectra are presented either in absorbance (A) mode or in difference spectral mode, with the latter highlighting the spectral changes during catalytic reaction. If A1 and A2 are the spectra before and after the sample treatment a difference (A2-A1) > 0 corresponds to treatment induced absorption and, if negative, disappearance of already existing light absorbing entities. 2.5. Oxygen storage complete capacity (OSCC) measurements 9 Oxygen Storage Complete Capacity (OSCC) of the samples was measured over 50 mg of sample loaded in U-shaped reactor. The samples were pretreated in Ar flow at 350 ºC for 1 h and the temperature decreased to the first studied temperature (150 ºC) at which the oxygen storage capacity was measured. Once the temperature stabilized, ten O 2 pulses of 1 mL each were injected every 2 min followed by ten CO pulses. The OSCC was calculated by the formed CO 2 during the CO pulses. This experiment allows to calculate the stoichiometric coefficient for TiO x , where x falls within the 1.5 - 2 range, with the first value corresponding to the presence of reduced (Ti 3+ ) only and the last value to completely oxidized (Ti 4+ ) titanium ion. From this value, the degree of reduction of Ti 4+ to Ti 3+ could be elucidated. 3. RESULTS AND DISCUSSION 3.1. Steady-state activity measurements Figure 2 shows the RWGS CO 2 conversion as a function of the reaction temperature for Au/Al 2 O 3 and Au/TiO 2 . The CO 2 conversion increased with the temperature for both catalysts. No products other than CO were observed, although the formation of traces of alcohols or other oxygenated compounds formed below our detection limit cannot be ruled out. Compared to Au/Al 2 O 3 , Au/TiO 2 catalyst exhibited higher RWGS activity, reaching values close to the thermodynamic limit. To ensure the kinetic regime of the reaction, the space velocity (GHSV) was increased to 40,000 h -1 for Au/TiO 2 , showing lower CO 2 conversion due to the decrease of the reactants residence time. 16 Figure 6. Transient DRIFT spectra collected at 250 ºC where the feed gas was switched from 5% CO 2 to 20% H 2 (both in He) over Au/TiO 2 . Pretreatment condition: 250 °C in H 2 diluted in He stream (5 ml min -1 H 2 and 45 ml min -1 He) for 1 h In order to gain further insights into the reaction intermediates leading to CO formation, surface species during RWGS (i.e. the mixture of CO 2 and H 2 ) at 250 ºC was studied by DRIFTS (Figure 7A-B). CO 2 was initially adsorbed over Au/Al 2 O 3 , forming surface bicarbonate species via the reaction of CO 2 with surface OH groups. The H 2 dissociation, which takes place on gold, results in the formation of Au-H species that attack surface bicarbonates forming adsorbed formate species. Figure 7C displays the evolution of the band intensities for the bicarbonate and the formate species with the reaction time evidencing a close correlation between both species. These data clearly reveal that formates are directly formed from bicarbonate reduction assisted by hydrogen. The formate species are finally decomposed to CO and the reaction follows clearly the associative mechanism over Au/Al 2 O 3 . 17 Figure 7. (A-B) Evolution of DRIFT spectra during RWGS reaction at 250 ºC and (C) evolution of the band height attributed to bicarbonates (1437 cm -1 ) and formate (1590 cm -1 ) surface species on Au/Al 2 O 3 as a function of the time on stream. Pretreatment condition: 250 °C in H 2 diluted in He stream (5 ml min -1 H 2 and 45 ml min -1 He) for 1 h, and reaction conditions: 20% H 2 and 5% CO 2 in He (H 2 /CO 2 molar ratio = 4) at 10 mL min -1 . For CO 2 hydrogenation over Au/TiO 2 (Figure 8A-B), surface carbonate and formate species were observed as possible intermediates leading to CO and water. A closer look in the region of hydroxyl groups (3500-3700cm -1 ) shows that a band feature at 3675 cm -1 , attributed to terminal hydroxyl groups associated to Ti 4+ 41 , evolved with a distinct kinetics. Figure 8C shows the trends of the intensities for hydroxyls group associated to Ti 4+ , carbonates, and gaseous CO produced with the reaction time. Obviously, the formation of Ti 4+ , carbonates and CO (g) are well correlated. In contrast, formates are accumulated over time. It has been reported that at low temperatures (150- 200 ºC) most formates observed by IR are spectator species although this does not preclude the possibility that there are some formates involved in the reaction. As the temperature is increased, this formate path may become more probable 26 . In a study about the reaction mechanism of dry 18 reforming of methane, Lercher et al. 42 proposed that the main route of CO 2 hydrogenation occurs via initial formation of carbonate close to the metal-support boundary. Carbon on the metal reduces the carbonate to formate by forming CO and the formate decomposes rapidly to CO and a surface hydroxyl group that recombines to form water. Our data suggest that CO 2 is adsorbed over the titania accompanying surface oxidation from Ti 3+ to Ti 4+ and forming carbonates and formates that are decomposed to CO. As discussed above, CO 2 likely reacts with the hydroxyl groups with a neighbor oxygen vacancy producing unstable bicarbonates species that are rapidly reduced to formates. The absence of carbonyl species adsorbed on gold particles during the transient and steady-state experiments for both catalysts indicates that the reaction pathway involving the dissociative adsorption of CO 2 on Au particles followed by the desorption of surface CO to the gas phase can be ruled out. A reaction scheme involving CO 2 adsorption on the support and subsequent reduction to gaseous CO assisted by the active dissociated hydrogen on the metal-support interface more reasonably explain our experimental results. Similar mechanism is reported by Rodríguez et al. 43 for a CeO x /Cu(111) sample. They proposed a mechanism in which CO 2 is hydrogenated yielding a hydroxycarbonyl intermediate (OCOH) with a relatively small calculated energy barrier (40 kJ mol -1 ). This intermediate is easily decomposed into CO and OH, that subsequently produce water from reaction from adsorbed H and OH species. The presence of this route can not be discarded since the hydroxycarbonyl species are much more reactive than formate and carbonates, especially at lower temperatures 43 . 19 Figure 8. (A-B) Evolution of DRIFT spectra during RWGS reaction at 250 ºC, and (C) evolution of the band height attributed to Ti(IV)-OH, carbonates and formate surface species on Au/TiO 2 as well as the band area of gaseous CO as a function of the time on stream. Pretreatment condition: 250 °C in H 2 diluted in He stream (5 ml min -1 H 2 and 45 ml min -1 He) for 1 h, and reaction conditions: 20% H 2 and 5% CO 2 in He (H 2 /CO 2 molar ratio = 4) at 10 mL min -1 and 250 ºC The above differences in RWGS catalytic activity can be attributed to the difference in interaction of CO 2 with the supports. When the interaction of CO 2 with the support is very strong, it is considered to cause low conversion of CO 2 due to the high stability of the adsorbed species. However, the presence of the oxygen vacancies on the support may enhance the formation of surface carbonate intermediates which is further easily transformed into CO and water regenerating the initial oxidation state on the support to initiate a new redox cycle. The adsorption of CO 2 on titania leads to the formation of different types of carbonate-like species. The facile formation and subsequent reduction of these species on titania can be attributed to its electronic and defect structure. It is well known that TiO 2 in its reduced form contains a high concentration of oxygen vacancies and CO 2 could be activated on these sites through an electron transfer from 20 Ti 3+ to CO 2 adsorbed 44 . In order to demonstrate the importance of the oxygen vacancies generated through the reduction of Ti 4+ to Ti 3+ of the support we have performed an operando UV-Vis spectroscopic study. 3.3. Operando UV-Vis spectroscopy Figure 9A presents the UV-Vis spectra of Aeroxide® TiO 2 P25 during activation in 10% H 2 and under RWGS reaction conditions. At room temperature (dashed line) the sample shows only one absorption band at around 320 nm corresponding to the absorption edge of the valence to conduction band transfer of semiconducting TiO 2 . During the activation in H 2 a spectral change appears as a broad band in the 400-800 nm range. Liu et al. 45 reported three types of interactions between H 2 and TiO 2 at different temperatures. Below 300 ºC, H 2 interacts weakly with the lattice whereas when the temperature rises the electrons are transferred from the H atoms to the lattice oxygen of titania and its vacancies are formed with H 2 O release. The formation of Ti 3+ ions are suggested above 250 ºC when electrons are transferred from the oxygen vacancies to Ti 4+ ions. Further increase of temperature leads to Ti 4+ →Ti 3+ conversion rate increase, and higher relative quantity of Ti 3+ and oxygen vacancies. The titania reduction could be also promoted by the UV irradiation 46 . However this possibility has been excluded as the UV-Vis spectra in absence of H 2 were different, allowing to assume that the reduction of titania is provoked solely by H 2 . It was reported that upon reduction the titania crystals, more precisely the rutile ones, displayed very broad absorption signal ranging from 400 to 1000 nm with poor spectral features and originated from a variety of coexisting intrinsic bulk and surface defects, identified as neutral singly and doubly charged oxygen vacancies and trivalent and tetravalent Ti interstitials at different structural positions 47 . All these defects result in three superposed bands in the 400-700 nm range without clear “type defect/band” attribution. The d-d transitions of the Ti 3+ ions could also be present when 21 the number of Ti 3+ sites becomes significant and should appear in the long wavelength region (> 700 nm). In any case, the appearance of the broad band at 400-700 nm unequivocally suggests the presence of oxygen defects on titania. Considering the literature indications and the observed spectral features for the TiO 2 P25 sample under reduction treatments at 250 ºC (Figure 9A, orange spectra), the formation of defects attributed to oxygen deficient sites and Ti 3+ sites are evident. The same material under the RWGS reaction condition (Figure 9A, blue spectra) did not show significant spectral changes till 350 ºC when RWGS starts taking place with CO evolution. The difference spectra (Figure 10A) demonstrates an increasing formation of defect sites under the RWGS gas mixture upon heating up to 300 ºC, supported by the changes registered in the 400-800 nm region and also by the intensity increase of a positive band at around 350 nm suggesting a bandgap modification caused probably by the contribution of the formed Ti 3+ levels. When the temperature rises to 350 ºC, the TiO 2 starts to convert CO 2 and H 2 into CO and H 2 O and the defect bands disappear as confirmed by the negative contribution in the 400-800 nm range suggesting the direct participation of these sites in the reaction. Figure 9. UV-Vis spectra of (A) Aeroxide TiO 2 P25 and (B) Au/TiO 2 catalyst during the activation (50 mL min -1 of 10% H 2 in He at 250 ºC for 1 h) and RWGS reaction at 150-450 ºC on the samples 22 Logically it is expected that Au/TiO 2 presents similar spectral features to those of TiO 2 P25. However, the wide Au plasmon band appearing at around 550 nm renders the identification of spectral features in the 400-800 nm range impossible (Figures 9B). The difference spectra in Figure 10B show contribution of the Au energy levels to titania bandgap changes (the positive band around 350 nm) coupled with the contributions of the titania vacancies formation confirmed by the asymmetric shape of the band toward 450 nm. Higher the temperature, more asymmetrical the shape becomes. This indicates the use of the vacancies created during the reaction (negative contribution appearance tending to increase the asymmetry of the band in the long wavelength range). Figure 10. Evolution of difference UV-Vis spectra in function of the RWGS reaction temperature for the samples (A) Aeroxide TiO 2 P25 and (B) Au/TiO 2 catalyst When the band intensity is compared to the observed CO 2 conversion presented in Figure 2, the increase of the intensity of the band centered at 360 nm is proportional to the RWGS reaction rate as can be observed in Figure 11. This indicates that the catalytic performance of Au/TiO 2 is 23 correlated with the presence of Ti 3+ , and consequently with the formation of oxygen vacancies of the support and their concentration. Figure 11. Relationship between the intensity of the UV-Vis band associated to Ti 3+ and the RWGS reaction rate for Au/TiO 2 catalyst We can conclude that the oxygen vacancies generated over titania surface directly participates in CO 2 activation generating carbonates species, as also suggested by IR, which are easily reducible to CO and water in presence of hydrogen. 3.4. Oxygen storage complete capacity (OSCC) According to the operando UV-Vis results, the superior performance of Au/TiO 2 catalyst in RWGS reaction can be associated with the increase in the oxygen vacancies concentration of titania. The validation of this affirmation in a quantitative manner generally requires a reasonable measurement of the exact number of oxygen vacancies on the surface and on the gold-support interface. Such estimation is provided by measuring the oxygen storage complete capacity 24 (OSCC). The obtained results are included in Table 1. TiO 2 P25 is usually considered as oxide with low oxygen mobility and indeed the measured OSCC value is very low and barely influenced by the temperature. The presence of gold, in contrast, boosts significantly the oxygen mobility of titania; almost doubled when the temperature rises from 150 to 250 ºC. At 250 ºC the calculated coefficient for the TiO x stoichiometry is 1.96 corresponding to maximal titania reduction of around 7 % and suggesting indirectly the presence of Ti 3+ . This is in a good agreement with the results obtained by operando UV-Vis, in which Ti 3+ species are also detected at this temperature. At 350 ºC the OSCC value decreases which can be tentatively explained by a possible changes in the gold/titanium contact. As detected by TEM (Figure 3) gold nanoparticles sinter during the reaction suggesting diminution of the gold/titanium interface which could reflect in a decrease in the oxygen mobility of titania at this temperature. Table 1. OSCC properties of TiO 2 support and Au/TiO 2 catalyst at 150, 250 and 350 ºC Temperature / ºC OSCC for TiO 2 µmol [O] g -1 OSCC for Au/TiO 2 µmol [O] g -1 Stoichiometric coefficient x for Au/TiO x % Surface reducibility on Au/TiO 2 150 10.2 298.4 1.98 4.2 250 11.1 516.6 1.96 7.2 350 12.7 343.6 1.98 4.8 3.5. Discussion of plausible reaction mechanism Data obtained from operando DRIFTS and UV-Vis experiments suggest considerable differences in the prevailing mechanism for the RWGS reaction on Au/Al 2 O 3 and Au/TiO 2 , respectively. 25 According to the proposed mechanistic pathways of RWGS reaction over supported noble metal catalysts reported in the literature, two reaction mechanisms have been widely accepted, namely regenerative redox mechanism and associative mechanism with a further possibility that surface formates could be considered intermediates involved in the rate determining step 5 . RWGS reaction proceeds through a concerted pathway where support and gold particles play key roles in the CO 2 activation and the dissociation of H 2 . Our results support that the activation of CO 2 over the two catalysts does not occur via direct dissociation of CO 2 to CO and O on the Au particles. In the case of Au/Al 2 O 3 , CO 2 first reacts with the hydroxyls on alumina to form bicarbonates species. This step does not require the participation of the gold particles 28 . The following steps involve the dissociation of hydrogen over Au and dissociated H atoms can spillover to the support to react with the bicarbonates leading to the formation of formates, which can be decomposed to CO. It has been suggested that this reaction of formation and decomposition of formates preferentially takes place on the interfacial sites between metal and alumina support 30 . Scheme 1 shows this reaction mechanism via formates intermediates proposed for the RWGS reaction on Au/Al 2 O 3 . Alumina is an irreducible support and thus the absence of oxygen vacancies leads to a decisive role of the hydroxyl groups in the mechanism of the reaction. The participation of hydroxyl groups in the formation and transformation of surface bicarbonates species is fundamental to understand the reaction mechanism on Au/Al 2 O 3 catalyst. If we assume that the formation of surface bicarbonates involves hydroxyls groups on the surface then the rate of transformation into formates species will depend also on the surface OH concentration. The H 2 dissociation takes place on the gold particles and the Au-hydride formed reduces the bicarbonate species to adsorbed formate species at the metalsupport interface. We propose that the surface formates are initially formed at the Au-support 32 REFERENCES (1) Ma, J.; Sun, N.; Zhang, X.; Zhao, N.; Xiao, F.; Wei, W.; Sun, Y., A Short Review of Catalysis for CO 2 Conversion. Catal. Today 2009, 148, 221-231. (2) Centi, G.; Perathoner, S., Opportunities and Prospects in the Chemical Recycling of Carbon Dioxide to Fuels. Catal. Today 2009, 148, 191-205. (3) Daza, Y. A.; Kuhn, J. N., CO 2 Conversion by Reverse Water Gas Shift Catalysis: Comparison of Catalysts, Mechanisms and Their Consequences for CO 2 Conversion to Liquid Fuels. RSC Adv. 2016, 6, 49675-49691. (4) Sakurai, H.; Tsubota, S.; Haruta, M., Hydrogenation of CO 2 over Gold Supported on Metal Oxides. Appl. Catal. A Gen. 1993, 102, 125-136. (5) Burch, R., Gold Catalysts for Pure Hydrogen Production in the Water-Gas Shift Reaction: Activity, Structure and Reaction Mechanism. Phys. Chem. Chem. Phys. 2006, 8, 5483-5500. (6) Tao, F.; Ma, Z., Water-Gas Shift on Gold Catalysts: Catalyst Systems and Fundamental Studies. Phys. Chem. Chem. Phys. 2013, 15, 15260-15270. (7) González-Castaño, M.; Reina, T. R.; Ivanova, S.; Centeno, M. A.; Odriozola, J. A., Pt vs. Au in Water–Gas Shift Reaction. J. Catal. 2014, 314, 1-9. (8) Tabakova, T.; Idakiev, V.; Andreeva, D.; Mitov, I., Influence of the Microscopic Properties of the Support on the Catalytic Activity of Au/ZnO, Au/ZrO 2 , Au/Fe 2 O 3 , Au/Fe 2 O 3 –ZnO, Au/Fe 2 O 3 – ZrO 2 Catalysts for the WGS Reaction. Appl. Catal. A Gen. 2000, 202, 91-97. (9) Tabakova, T.; Ilieva, L.; Ivanov, I.; Zanella, R.; Sobczak, J. W.; Lisowski, W.; Kaszkur, Z.; Andreeva, D., Influence of the Preparation Method and Dopants Nature on the WGS Activity of 33 Gold Catalysts Supported on Doped by Transition Metals Ceria. Appl. Catal. B Environ. 2013, 136–137, 70-80. (10) Jacobs, G.; Ricote, S.; Patterson, P. M.; Graham, U. M.; Dozier, A.; Khalid, S.; Rhodus, E.; Davis, B. H., Low Temperature Water-Gas Shift: Examining the Efficiency of Au as a Promoter for Ceria-Based Catalysts Prepared by CVD of a Au Precursor. Appl. Catal. A Gen. 2005, 292, 229-243. (11) Reina, T. R.; Ivanova, S.; Delgado, J. J.; Ivanov, I.; Idakiev, V.; Tabakova, T.; Centeno, M. A.; Odriozola, J. A., Viability of Au/CeO 2 –ZnO/Al 2 O 3 Catalysts for Pure Hydrogen Production by the Water–Gas Shift Reaction. ChemCatChem 2014, 6, 1401-1409. (12) Shekhar, M.; Wang, J.; Lee, W.-S.; Williams, W. D.; Kim, S. M.; Stach, E. A.; Miller, J. T.; Delgass, W. N.; Ribeiro, F. H., Size and Support Effects for the Water–Gas Shift Catalysis over Gold Nanoparticles Supported on Model Al 2 O 3 and TiO 2 . J. Am. Chem. Soc. 2012, 134, 4700- 4708. (13) Pakhare, D.; Spivey, J., A Review of Dry (CO 2 ) Reforming of Methane over Noble Metal Catalysts. Chem. Soc. Rev. 2014, 43, 7813-7837. (14) Yang, X.; Kattel, S.; Senanayake, S. D.; Boscoboinik, J. A.; Nie, X.; Graciani, J.; Rodríguez, J. A.; Liu, P.; Stacchiola, D. J.; Chen, J. G., Low Pressure CO 2 Hydrogenation to Methanol over Gold Nanoparticles Activated on a CeO x /TiO 2 Interface. J. Am. Chem. Soc. 2015, 137, 10104- 10107. (15) Wang, W.; Wang, S.; Ma, X.; Gong, J., Recent Advances in Catalytic Hydrogenation of Carbon Dioxide. Chem. Soc. Rev. 2011, 40, 3703-3727. 34 (16) Ishito, N.; Hara, K.; Nakajima, K.; Fukuoka, A., Selective Synthesis of Carbon Monoxide via Formates in Reverse Water–Gas Shift Reaction over Alumina-Supported Gold Catalyst. J. Energy Chem. 2016, 25, 306-310. (17) Millar, G. J.; Rochester, C. H.; Howe, C.; Waugh, K. C., A Combined Infrared, Temperature Programmed Desorption and Temperature Programmed Reaction Spectroscopy Study of CO 2 and H 2 Interactions on Reduced and Oxidized Silica-Supported Copper Catalysts. Mol. Phys. 1992, 76, 833-849. (18) Ma, D.; Lund, C. R. F., Assessing High-Temperature Water−Gas Shift Membrane Reactors. Ind. Eng. Chem. Res. 2003, 42, 711-717. (19) Tsakoumis, N. E.; York, A. P. E.; Chen, D.; Rönning, M., Catalyst Characterisation Techniques and Reaction Cells Operating at Realistic Conditions; Towards Acquisition of Kinetically Relevant Information. Catal. Sci. Technol. 2015, 5, 4859-4883. (20) Sattler, J. J. H. B.; González-Jimenez, I. D.; Mens, A. M.; Arias, M.; Visser, T.; Weckhuysen, B. M., Operando UV-Vis Spectroscopy of a Catalytic Solid in a Pilot-Scale Reactor: Deactivation of a CrO x /Al 2 O 3 Propane Dehydrogenation Catalyst. Chem. Commun. 2013, 49, 1518-1520. (21) https://secure.strem.com/catalog/family/Aurolite/ (access date: 2013/06/15) (22) Ovsitser, O.; Cherian, M.; Brückner, A.; Kondratenko, E. V., Dynamics of Redox Behavior of Nano-Sized VO x Species Over Ti–Si-MCM-41 from Time-Resolved In Situ UV/Vis Analysis. J. Catal. 2009, 265, 8-18. (23) Kondratenko, E. V.; Takahashi, N.; Nagata, N.; Ibe, M.; Hirata, H.; Takahashi, H., Operando UV/Vis Analysis of the Synergy Effect between Copper and Gold in Nitric Oxide Reduction over Gold and Copper on Aluminacatalysts. ChemCatChem 2015, 7, 3956-3962. 35 (24) Walther, G.; Mowbray, D. J.; Jiang, T.; Jones, G.; Jensen, S.; Quaade, U. J.; Horch, S., Oxidation of CO and H 2 by O 2 and N 2 O on Au/TiO 2 Catalysts in Microreactors. J. Catal. 2008, 260, 86-92. (25) Jacobs, G.; Patterson, P. M.; Williams, L.; Chenu, E.; Sparks, D.; Thomas, G.; Davis, B. H., Water-Gas Shift: In Situ Spectroscopic Studies of Noble Metal Promoted Ceria Catalysts for CO Removal in Fuel Cell Reformers and Mechanistic Implications. Appl. Catal. A Gen. 2004, 262, 177-187. (26) Meunier, F. C.; Reid, D.; Goguet, A.; Shekhtman, S.; Hardacre, C.; Burch, R.; Deng, W.; Flytzani-Stephanopoulos, M., Quantitative Analysis of the Reactivity of Formate Species seen by DRIFTS over a Au/Ce(La)O 2 Water–Gas Shift Catalyst: First Unambiguous Evidence of the Minority Role of Formates as Reaction Intermediates. J. Catal. 2007, 247 (2), 277-287. (27) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Andreeva, D.; Tabakova, T., FTIR Study of the Low-Temperature Water–Gas Shift Reaction on Au/Fe 2 O 3 and Au/TiO 2 Catalysts. J. Catal. 1999, 188, 176-185. (28) Szanyi, J.; Kwak, J. H., Dissecting the Steps of CO 2 reduction: 1. The Interaction of CO and CO 2 with γ-Al 2 O 3 : an In Situ FTIR Study. Phys. Chem. Chem. Phys. 2014, 16, 15117-15125. (29) Trillo, J. M.; Munuera, G.; Criado, J. M., Catalytic Decomposition of Formic Acid on Metal Oxides. Catal. Rev. 1972, 7, 51-86. (30) Wang, X.; Shi, H.; Kwak, J. H.; Szanyi, J., Mechanism of CO 2 Hydrogenation on Pd/Al 2 O 3 Catalysts: Kinetics and Transient DRIFTS-MS Studies. ACS Catal. 2015, 5, 6337-6349. (31) Wang, X.; Hong, Y.; Shi, H.; Szanyi, J., Kinetic Modeling and Transient DRIFTS–MS Studies of CO 2 Methanation over Ru/Al 2 O 3 Catalysts. J. Catal. 2016, 343, 185-195. 36 (32) Gaur, S.; Wu, H.; Stanley, G. G.; More, K.; Kumar, C. S. S. R.; Spivey, J. J., CO Oxidation Studies over Cluster-derived Au/TiO 2 and AUROlite™ Au/TiO 2 Catalysts using DRIFTS. Catal. Today 2013, 208, 72-81. (33) Denkwitz, Y.; Zhao, Z.; Hörmann, U.; Kaiser, U.; Plzak, V.; Behm, R. J., Stability and Deactivation of Unconditioned Au/TiO 2 Catalysts during CO Oxidation in a Near-Stoichiometric and O 2 -Rich Reaction Atmosphere. J. Catal. 2007, 251, 363-373. (34) Bando, K. K.; Sayama, K.; Kusama, H.; Okabe, K.; Arakawa, H., In-Situ FT-IR Study on CO 2 Hydrogenation over Cu Catalysts Supported on SiO 2 , Al 2 O 3 , and TiO 2 . Appl. Catal. A Gen. 1997, 165, 391-409. (35) Nomura, N.; Tagawa, T.; Goto, S., In Situ FTIR Study on Hydrogenation of Carbon Dioxide over Titania-Supported Copper Catalysts. Appl. Catal. A Gen. 1998, 166, 321-326. (36) Baltrusaitis, J.; Jensen, J. H.; Grassian, V. H., FTIR Spectroscopy Combined with Isotope Labeling and Quantum Chemical Calculations to Investigate Adsorbed Bicarbonate Formation following Reaction of Carbon Dioxide with Surface Sydroxyl Groups on Fe 2 O 3 and Al 2 O 3 . J. Phys. Chem. B 2006, 110, 12005-12016. (37) Fottinger, K.; Schlögl, R.; Rupprechter, G., The Mechanism of Carbonate Formation on Pd- Al 2 O 3 Catalysts. Chem. Commun. 2008, 3, 320-322. (38) Guglielminotti, E.; Giamello, E.; Pinna, F.; Strukul, G.; Martinengo, S.; Zanderighi, L., Elementary Steps in CO Hydrogenation on Rh Catalysts Supported on ZrO 2 and Mo/ZrO 2 . J. Catal. 1994, 146, 422-436. 37 (39) Senanayake, S. D.; Stacchiola, D.; Liu, P.; Mullins, C. B.; Hrbek, J.; Rodriguez, J. A., Interaction of CO with OH on Au(111): HCOO, CO 3 , and HOCO as Key Intermediates in the Water-Gas Shift Reaction. J. Phys. Chem. C 2009, 113, 19536-19544. (40) Gokhale, A. A.; Dumesic, J. A.; Mavrikakis, M., On the Mechanism of Low-Temperature Water Gas Shift Reaction on Copper. J. Am. Chem. Soc. 2008, 130, 1402-1414. (41) Deiana, C.; Fois, E.; Coluccia, S.; Martra, G., Surface Structure of TiO 2 P25 Nanoparticles: Infrared Study of Hydroxyl Groups on Coordinative Defect Sites. J. Phys. Chem. C 2010, 114, 21531-21538. (42) Bitter, J. H.; Seshan, K.; Lercher, J. A., Mono and Bifunctional Pathways of CO 2 /CH 4 Reforming over Pt and Rh based Catalysts. J. Catal. 1998, 176, 93-101. (43) Rodriguez, J. A.; Liu, P.; Stacchiola, D. J.; Senanayake, S. D.; White, M. G.; Chen, J. G., Hydrogenation of CO 2 to Methanol: importance of Metal–Oxide and Metal–Carbide Interfaces in the Activation of CO 2 . ACS Catal. 2015, 5, 6696-6706. (44) Raskó, J., FTIR Study of the Photoinduced Dissociation of CO 2 on Titania-Supported Noble Metals. Catal. Lett. 1998, 56, 11-15. (45) Liu, H.; Ma, H. T.; Li, X. Z.; Li, W. Z.; Wu, M.; Bao, X. H., The Enhancement of TiO 2 Photocatalytic Activity by Hydrogen Thermal Treatment. Chemosphere 2003, 50, 39-46. (46) Xiong, L.-B.; Li, J.-L.; Yang, B.; Yu, Y., Ti 3+ in the Surface of Titanium Dioxide: Generation, Properties and Photocatalytic Application. J. Nanomater. 2012, 2012, 5-8. (47) Kuznetsov, V. N.; Serpone, N., Visible Light Absorption by Various Titanium Dioxide Specimens. J. Phys. Chem. B 2006, 110, 25203-25209. 38 (48) Carrasquillo-Flores, R.; Singh, S.; Li, S.; Alba-Rubio, A. C.; Dumesic, J. A.; Mavrikakis, M. In Formic acid decomposition on Au catalysts: Identifying the active site for improved catalyst design, Catalysis and Reaction Engineering Division 2014 - Core Programming Area at the 2014 AIChE Annual Meeting, 2014; p 450. (49) Singh, S.; Li, S.; Carrasquillo-Flores, R.; Alba-Rubio, A. C.; Dumesic, J. A.; Mavrikakis, M., Formic acid decomposition on Au catalysts: DFT, microkinetic modeling, and reaction kinetics experiments. AIChE J. 2014, 60 (4), 1303-1319. (50) Frost, J. C., Junction Effect Interactions in Methanol Synthesis Catalysts. Nature 1988, 334, 577-580. (51) Hernández, W. Y.; Romero-Sarria, F.; Centeno, M. A.; Odriozola, J. A., In Situ Characterization of the Dynamic Gold−Support Interaction over Ceria Modified Eu 3+ . Influence of the Oxygen Vacancies on the CO Oxidation Reaction. J. Phys. Chem. C 2010, 114, 10857- 10865. (52) Goguet, A.; Meunier, F. C.; Tibiletti, D.; Breen, J. P.; Burch, R., Spectrokinetic Investigation of Reverse Water-Gas-Shift Reaction Intermediates over a Pt/CeO 2 Catalyst. J. Phys. Chem. B 2004, 108, 20240-20246. (53) Kalamaras, C. M.; Dionysiou, D. D.; Efstathiou, A. M., Mechanistic Studies of the Water– Gas Shift Reaction over Pt/Ce x Zr 1–x O 2 Catalysts: The Effect of Pt Particle Size and Zr Dopant. ACS Catal. 2012, 2, 2729-2742. (54) Rodríguez, J. A.; Evans, J.; Graciani, J.; Park, J.-B.; Liu, P.; Hrbek, J.; Sanz, J. F., High Water−Gas Shift Activity in TiO 2 (110) Supported Cu and Au Nanoparticles: Role of the Oxide and Metal Particle Size. J. Phys. Chem. C 2009, 113, 7364-7370. 39 SYNOPSIS (TOC) SUPPORTING INFORMATION Unravelling the Role of Oxygen Vacancies in the Mechanism of the Reverse Water-Gas-Shift Reaction by Operando DRIFTS and UV-Vis Spectroscopy 40 Luis F. Bobadilla † , *, José L. Santos †, Svetlana Ivanova †, José A. Odriozola † and Atsushi Urakawa ‡ † Instituto de Ciencia de Materiales de Sevilla, Centro Mixto CSIC-Universidad de Sevilla, Av. Américo Vespucio 49, 41092 Sevilla, Spain ‡ Institute of Chemical Research of Catalonia (ICIQ), The Barcelona Institute of Science and Technology, Av. Països Catalans 16, 43007 Tarragona, Spain (*) Corresponding author: [email protected] EXPERIMENTAL DETAILS X-ray diffraction (XRD) measurements were performed on an X’Pert Pro PANalytical diffactometer. Diffraction patterns were recorded using Cu Kα radiation (40 mA, 45 kV) over a 10–90° 2ϴ-range and a position-sensitive detector using a step size of 0.05° and a step time of 240 s. 41 Figure S1. XRD patterns of Au/Al 2 O 3 and Au/TiO 2 Due to the observed variations in the gold particle size of both catalysts during the RWGS reaction, the catalytic activity was also evaluated in terms of conversion up/down curves. As example, we have included in Fig. S2 the obtained results for Au/TiO 2 catalyst under kinetic regime conditions at GHSV = 40,000 h -1 . As can be observed the conversion up/down is identical even reaching maximal temperature of 400 ºC (Fig. S2A) and one can guarantee that Ea is not under-estimated (Fig. S2B).