Elucidating the Role of the Metal Catalyst and Oxide Support in the Ru/CeO2-Catalyzed CO2 Methanation Mechanism

This study addresses the yet unresolved CO2 methanation mechanism on a Ru/CeO2 catalyst by means of near-ambient-pressure X-ray photoelectron spectroscopy (NAP–XPS) and diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) complemented with periodic density functional theory (DFT) calculations. NAP–XPS results show that the switch from H2 to CO2 + H2 mixture oxidizes both the Ru and CeO2 phases at low temperatures, which is explained by the CO2 adsorption modes assessed by means of DFT on each representative surface. CO2 adsorption on Ru is dissociative and moderately endergonic, leading to polybonded Ru-carbonyl groups whose hydrogenation is the rate-determining step in the overall process. Unlike on Ru metal, CO2 can be strongly adsorbed as carbonates on ceria surface oxygen sites or on the reduced ceria at oxygen vacancies as carboxylates (CO2–δ), resulting in the reoxidation of ceria. Carboxylates can then evolve as CO, which is released either via direct splitting at relatively low temperatures or through stable formate species at higher temperatures. DRIFTS confirm the great stability of formates, whose depletion relates with CO2 conversion in the reaction cell, while carbonates remain on the surface up to higher temperatures. CO generation on ceria serves as an additional reservoir of Ru-carbonyls, cooperating to the overall CO2 methanation process. Altogether, this study highlights the noninnocent role of the ceria support in the performance of Ru/CeO2 toward CO2 methanation.


INTRODUCTION
The global warming induced by the rising energy-related CO 2 emissions is a threat that has become a primary focus for environmental research. Moreover, according to official forecasts, the energy demand will grow by 33% in 2050 compared to 2017 together with the CO 2 emissions over the next years. 1 Hence, supplying this demand while meeting the CO 2 emission targets is a major challenge that society has to face in the 21st century via the implementation of sustainable policies and energy practices with a lower carbon footprint. 2 In this scenario, scientists and industrial stakeholders are urged to design alternative technologies to enable a feasible and successful transition from carbon-based to renewable fuels. This involves the deployment of new hydrogen power plants and the reduction of carbon dioxide emissions according to the established pillars of the circular economy. 3,4 In this regard, the CO 2 methanation reaction is an attractive strategy. 5,6 This process consists in the conversion of CO 2 to CH 4 As a remarkable positive incentive, CO 2 methanation technologies based on captured CO 2 and green hydrogen 7 can enable a clean energy distribution without the requirement of new infrastructure or alternative combustion engines. 8 Furthermore, CO 2 methanation can be deemed carbon-neutral if both hydrogen and the thermal activation energy required are supplied from renewable sources and the CO 2 released upon the eventual CH 4 utilization is captured and fed back into the methanation process. In general terms, CO 2 methanation is spontaneous at ambient temperature, but the use of active catalysts is needed to overcome the sluggish reaction kinetics to reach relevant reaction rates and selectivities for industrial applications. 9 Among the studied active phases, nickel is a well-known costeffective CO 2 methanation catalyst that operates between 300 and 450°C providing good CH 4 selectivity. When finely dispersed as Ni nanoparticles, the choice of the support determines the overall performance. For example, literature reports have shown that CO 2 Ni/CeO 2 formulations are more active than their counterparts with TiO 2 , Al 2 O 3 , and ZrO 2 . 10−17 Such improved performance is generally attributed to the well-known oxygen storage capacity and oxygen mobility in ceria. 10,18−21 Yet, Ni catalysts suffer from sintering and the reverse water gas-shift reaction (RWGS) at high operating temperatures, which compromises CH 4 selectivity and CO 2 conversion.
On the other hand, noble-metal-based catalysts such as Rh, Pt, Pd, and Ru 22−27 exhibit a lower CO 2 methanation onset than Ni (from ca. 250°C to ca. 180°C) enabling us to operate at temperatures where the RWGS is rendered thermodynamically unfavorable. Among precious metals, Ru has received great attention 28−36 and the CO 2 methanation mechanism on this metal catalyst has been investigated in detail. 30,36,37 However, many important aspects of the underlying mechanism still remain unclear and no general consensus has been reached yet, mainly due to the strong structure-sensitive nature of this reaction. 24,38,39 On the one hand, many studies have reported plausible reaction pathways featuring formates (*OCHO, where * denotes a surface-active site) as key intermediates formed upon CO 2 adsorption on H-terminated Ru surfaces. According to this approach, further hydrogenation steps from formates using chemisorbed *H would give rise to *C x H y O intermediate species up to methane. 19,22,35,40,41 On the other hand, alternative mechanisms for the Ru-catalyzed methanation have been proposed by different authors. Proanõ et al. 41 postulated that Ru-carbonyls are direct precursors of CH 4 on Ru/TiO 2 catalysts. On the contrary, Falbo et al. 30 proposed a mechanism for Ru/Al 2 O 3 , wherein CO 2 is adsorbed as a bicarbonate on the support and subsequently hydrogenated to CH 4 on Ru, involving formate and carbonyl intermediates. For Ru/CeO 2 , Wang et al. 42,43 reported that CO 2 methanation involves formate intermediates, which dissociate on the vacancy sites in ceria. Despite all of these insights, the detailed roles of ruthenium, the ceria support, and the interfacial sites remain unclear. Similarly to Ni, CeO 2 and CeO 2 -modified supports provide enhanced activity for Ru catalysts, 29,44,45 although the nature of their active sites is not completely understood either.
Herein, we report a thorough mechanistic investigation of the CO 2 methanation reaction on a high-performance Ru/ CeO 2 catalyst using advanced spectroscopy techniques and periodic density functional theory (DFT) calculations. The changes in the electronic structure of Ru and Ce were monitored under reaction conditions using synchrotron radiation by in situ near-ambient-pressure X-ray photoelectron spectroscopy (NAP−XPS). DFT calculations, together with in situ diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy experiments, were used to identify the active sites and the role of the chemisorbed species during reaction. Altogether, this multidisciplinary study demonstrates that CO 2 is activated on ruthenium, leading to strongly bound carbonyl species, while the main role of ceria is to promote CO 2 chemisorption in the form of carbonates and carboxylates. Further activation of CO 2 on ceria is found to proceed via the decomposition of formate species on surface oxygen vacancies in complementary action to ruthenium. DFT calculations also indicate that the hydrogenation of Ru-carbonyls is the ratedetermining step (RDS), while the subsequent hydrogenation of CO x H intermediates is thermodynamically driven.

Catalyst Synthesis and Characterization.
Cerium oxide support was obtained by calcination of Ce(NO 3 ) 3 ·6H 2 O (99.0%, Sigma-Aldrich) for 6 h at 600°C. Ruthenium was loaded by incipient wetness impregnation of ruthenium(III) acetylacetonate (97.0%, Sigma-Aldrich) dissolved in toluene to achieve a nominal 4% w. Finally, to obtain the Ru/CeO 2 catalyst, the impregnated sample was heated under N 2 atmosphere at a 5°C/min pace up to 350°C and was kept at this temperature for 3 h. General characterization results are presented in the Supporting Information, including N 2 adsorption−desorption isotherms (Table S1, Figure S1), Xray diffraction ( Figure S2), and temperature-programmed reduction with H 2 ( Figure S3).
2.2. CO 2 Methanation Catalytic Tests. CO 2 methanation activity tests were performed in a fixed-bed tubular reactor (10 mm inner diameter) containing 200 mg of catalyst mixed with SiC particles (1.00−1.25 mm) to reach the bed volume of 1 cm 3 . The catalyst was pretreated in situ at 550°C for 1 h under 200 mL/min of a 50% H 2 /N 2 mixture. Once this pretreatment was completed, and after cooling down to 100°C, the reaction mixture was introduced into the reactor. The feed consisted of 200 mL/min of 16% CO 2 , 64% H 2 , and 20% N 2 at atmospheric pressure. The GHSV was 9000 h −1 , and the temperature was increased in steps of 100°C up to 325°C. The gas composition was monitored under steady-state conditions at each temperature with specific gas analyzers (AwiteFLEX COOL; NDIR, electrochemical and thermal conductivity detectors) for CO, CO 2 , CH 4 , O 2 , and H 2 . A cold trap at −96°C is placed after the reactor to condense the released water vapor prior to the detectors. The measured outlet concentrations are rescaled to keep the mol balance.
2.3. In Situ NAP−XPS Experiments. X-ray photoemission spectra were recorded in situ under methanation conditions at the near-ambient-pressure photoemission (NAPP) branch from CIRCE beamline at the ALBA Synchrotron Light Source facility, 46 which allows tuning the photon energy within 100− 2000 eV using a PHOIBOS NAP150 energy analyzer (SPECS GmbH). In each experiment, samples were exposed in the analysis chamber to the reaction mixture consisting of 16% CO 2 , 64% H 2 , and 20% N 2 . Two photon energies were used for each region of interest to obtain information at different surface depths, namely, 1372 and 1082 eV for the Ce 3d region, and 972 and 722 eV for the Ru 3d region. The inelastic mean free path for the emitted photoelectrons through the different pure solid phases of Ru and Ce at each energy utilized during the experiments is presented in Table S3. 47 The Ru/CeO 2 sample was pelletized and mounted on a gold mesh to minimize charging during the measurements and to provide an energy reference (Au 4f 7/2 peak at 84.0 eV). An infrared laser system (808 nm) was used to heat the samples while the temperature was monitored using a K-type thermocouple. The pressure in the analysis chamber was kept at 1 mbar during the reaction. The catalyst was pretreated in situ at 550°C for 1 h with 50% H 2 /N 2 and then cooled down to 100°C. After this procedure, the methanation mixture was fed at 30 mL/min and the temperature was increased in steps of 50°C until 450°C. The reaction was monitored with a mass spectrometer (MKS Instruments) installed at the second stage of the differential pumping system of the analyzer, and XPS measurements were recorded at each temperature under steady-state conditions.  51 were used to describe the core electrons of Ce, Ru, O, C, and H ions, while plane waves with a kinetic cutoff energy of 500 eV were employed to represent their valence electrons. For Ce atoms, an additional on-site correction for the electrons localized in the 4f orbital was introduced using an effective Hubbard U term (U eff ) of 4.5 eV following Dudarev's approach. 52 The CeO 2 and metal Ru metal bulk structures were optimized with a Γ-centered Monkhorst−Pack k-point grid of 7 × 7 × 7 and 11 × 11 × 11, respectively, and the equilibrium lattice parameter was calculated using the Birch− Murnaghan equation of state. The most stable facets, i.e., CeO 2 (111) and Ru(0001), 53,54 were modeled as surface slabs with at least a 15 Å vacuum gap in the perpendicular direction to the surface to avoid the interaction between the top and bottom layers. The CeO 2 (111) and Ru(0001) slabs were built with a periodicity of p(2 × 2), consisting of three and four metal layers, respectively. In both cases, the two upper layers were allowed to relax optimizing the structure using a k-point mesh of 3 × 3 × 1 and 7 × 7 × 1, respectively. The CeO 2 (111) and Ru(0001) slabs were constituted by three and four metal layers, respectively. Once optimized, the adsorption energies, ΔE ads , of the CO 2 methanation intermediates were computed as where E slab is the energy of the clean slab, E ads is the energy of the adsorbate species in the gas phase, and E ads+slab is the energy of the slab with the specific adsorbates in the most favorable configuration. These values were employed to calculate the Gibbs adsorption energies, ΔG ads , using the following equation where ΔE ZPE and TΔS ads denote the changes in zero-point energy and entropy, respectively, relative to the clean slab and the adsorbate molecules in the gas phase. The effect of hydrogen and CO 2 partial pressures on the above Gibbs adsorption energies was introduced as follows where Δμ i (T, p) is the change in chemical potential of the n adsorbed species i at a given temperature and pressure, defined as where k B is the Boltzmann constant, α is taken as 1/2 for H 2 and 1 for CO 2 , and Δμ i (T, p 0 ) is the change in chemical potential for H 2 or CO 2 at a given temperature and standard pressure.
After assessing the resting state of the Ru(0001) and CeO x (111) surfaces in the representative thermodynamic reactions conditions, the CO 2 mechanism was investigated by computing the relative energies of the most plausible reaction intermediates on each surface, leading to different reaction pathways. Transition states (TS) for the relevant steps were located by means of the climbing image nudge elastic band method (CI-NEB) using at least five images along the reaction coordinate and the limited-memory Broyden− Fletcher−Goldfarb−Shanno (LBFGS) optimizer. The nature of TSs obtained was verified via vibrational frequency analysis using the finite difference method with a displacement of 0.01 Å.

RESULTS AND DISCUSSION
3.1. Catalytic Activity. CO 2 methanation tests were conducted in a conventional fixed-bed reactor containing the Ru/CeO 2 catalyst, and the evolution of CO 2 and CH 4 gases produced during the reaction was monitored as shown in Figure 1. Prior to the experiment, the catalyst was pretreated with H 2 at 550°C for 1 h to reduce the ruthenium species present, in agreement with H 2 -TPR profiles ( Figure S3). In this experimental setup, the CO 2 methanation activity starts above 170°C showing a sharp increase in the CO 2 conversion (X CO2 ) with temperature until a stationary state is achieved at 225°C. This behavior is similar to that reported in the literature for other Ru/CeO 2 catalysts. 42,43 Notably, only CH 4 and H 2 O were detected as reaction products regardless of the temperature, which demonstrates the high selectivity toward The Journal of Physical Chemistry C pubs.acs.org/JPCC Article CH 4 relative to other byproducts such as CO. The m/z 44 signal monitored by mass spectrometry during the in situ DRIFTS and in situ NAP−XPS experiments under CO 2 methanation reaction conditions is also included in Figure 1 for comparison. CO 2 conversion displays similar onsets in the three experimental setups (ca. 170°C), but the effect of temperature on CO 2 consumption is different for each technique. The comparison of the catalytic behavior in the different experimental setups is essential to obtain a meaningful interpretation of experimental results since the reaction conditions used in the conventional fixed-bed reactor are different from those used in spectroscopic experiments (DRIFTS and NAP−XPS) due to the restrictions imposed by the spectroscopy setups.

3.2.
In Situ DRIFTS Experiments. Figure 2 shows two relevant wavelength ranges selected from the spectra recorded after steady state was achieved at different temperatures. Figure  2a plots the 2100−1200 cm −1 range, which reveals the characteristic bands assigned to different vibration modes of C−O and CO bonds, where formates (*OCHO) and carbonates (*OCOO) can be discerned, besides the formation of molecular H 2 O. The band centered at ca. 2000 cm −1 is attributed to Ru-carbonyls, showing a lower-wavenumber tail characteristic of their multiple configurations. The linear Ru 0 − CO shows between 2045 and 2015 cm −1 , while oxidized Ru n+ −CO appears at lower wavenumbers. 55 This assignment contradicts what one would expect due to the lower backdonation from cationic species, but has been elsewhere reported and attributed to lower adsorbate−adsorbate interactions in the oxidized environment. 56 Regarding this, as precisely detailed by Solís-García et al. in a recent paper, we recognize that there is, unfortunately, a large variety and contradictory band assignments on Ru carbonyls in the literature. 57 For instance, according to some studies, bridged zerovalent Ru carbonyls may also be contributing to the broad carbonyl band in the lower-wavenumber region (i.e., 2000− 1900 cm −1 ), 58 although these species have been more frequently reported at ca. 1800 cm −1 . 59,60 The metallic linear carbonyls display a reversible blueshift to 2045 cm −1 attributed to the reduction of their oxidized surface environment by the effect of the reaction mixture followed by a decrease of CO coverage above the methanation onset, setting the Ru 0 −CO band back to 2015 cm −1 . On the other hand, we do not see a sign of polybonded carbonyls either on metal or cationic Ru sites. Figure 2b, displaying the 3800−2700 cm −1 range, provides information about C−H bonds, including the fingerprint bands of formates, gas-phase CH 4 , and O−H vibration bands assigned to surface hydroxyls and water. Figure 2 proves that CO 2 is chemisorbed on the reduced catalyst at room temperature, as seen by the formation of bidentate carbonates (1583 and 1280 cm −1 bands), formates (2846, 1610, and 1375 cm −1 bands), and Ru-carbonyls. Carbonates are formed upon CO 2 chemisorption on metal oxides and are assumed to sit on the ceria surface as ruthenium is mostly reduced according to H 2 -TPR experiments ( Figure  S3). Notably, the presence of formates at 25°C indicates that partial hydrogenation of the chemisorbed CO 2 takes place at room temperature. The H atom in the formate species (*OCHO) can be sourced from either the dissociated H 2 on the catalyst after reduction at 450°C, presumably on Ru 0 , or the hydroxyl groups formed on ceria. 61,62 Isolated hydroxyls are observed in the 25−300°C range in the spectra as weak bands in the 3750−3600 cm −1 region, which are eventually masked below the broad and intense absorption band of polybonded OH or surface H 2 O. Since the consumption of hydroxyls is not detected, the first hypothesis regarding the generation of formates is more likely. We also note that the formation of Ru-carbonyls involves the dissociation of the CO 2 molecule, which occurs on the reduced catalyst even at room temperature, despite the methanation reaction onset in this experimental setup is only observed at higher temperatures (see Figure 1). From the DRIFT spectra, however, it is not possible to discern whether CO 2 splits directly onto the Ru particles or the CeO 2 surface followed by CO spillover to the Ru phase. As stated above, the carbonyl band shape allows us to infer the coexistence of both on Ru 0 and on Ru n+ species, although Ru is reduced after H 2 pretreatment. The identification of carbonyls on cationic ruthenium (Ru n+ − CO) could be evidence of the partial oxidation of ruthenium upon CO 2 exposure at room temperature. This oxidation can be the result of the direct dissociative CO 2 chemisorption into oxygen atoms and carbonyls. However, this is just a hypothesis since the effect of ceria in keeping Ru oxidized through synergistic interactions cannot be ruled out.
Interestingly, we observed that the increase in temperature has an effect on the intensity of the DRIFTS signals but not on the nature of the surface species observed, as indicated by the presence of bands attributed to carbonyls, formates, and The Journal of Physical Chemistry C pubs.acs.org/JPCC Article carbonates. In addition, a sharp peak at 3015 cm −1 is observed at temperatures above 300°C, which we assign to the C−H stretching mode of gas-phase CH 4 . Similarly, water is detected above 350°C as indicated by the presence of a broad band in the 3700−3450 cm −1 range, corresponding to the symmetric and asymmetric stretching of H-bonded hydroxyl groups. 61 The identification of such broad band can also be related to the presence of large coverages of water on the catalyst surface or the accumulation of water in the reaction chamber as a product of the CO 2 methanation reaction. Several bands from the spectra shown in Figure 2 have been selected to monitor the evolution of carbonyls (2015 and 1920 cm −1 ), formates (2830 cm −1 ), and carbonates (1280 cm −1 ) with temperature. Note that the characteristic formate bands within 2930−2850 cm −1 exhibit a redshift when temperature is increased, which is attributed to the weakening of the C−H bond due to redox changes in the adsorption site. The evolution of band intensities from the selected signals with temperature is presented in Figure 3.
The generation of formates and bidentate carbonates is strongly influenced by the reaction temperature. In particular, their signals peak at 200 and 300°C, respectively, and then drop. The carbonyl signals also increase slightly from 25 to 200°C , and then remain constant until 400°C. This indicates that carbonyls are not only formed on ruthenium at room temperature but also at higher temperatures, and they remain on the catalyst surface even when exposed to CO 2 methanation temperatures (above 170°C). The decrease of the formate, carbonate, and carbonyl bands above ca. 200, 300, and 400°C, respectively, can be attributed to their hydrogenation onset or thermal decomposition. Panagiotopoulou et al. 35 confirmed that Ru-carbonyls are reaction intermediates in the CO 2 methanation on Ru/TiO 2 catalysts, while other authors 30,63 reached the same conclusion for Ru/Al 2 O 3 . However, Wang et al. 43 concluded that formates are the main reaction intermediates. On the contrary, Upham et al. 64 pinpointed carbonates as active intermediates of CO 2 methanation while ruling out the participation of CO and formate-type adsorbates in a Ru-doped ceria catalyst with composition Ru 0.05 Ce 0.95 O x . This controversy highlights that particular features of the ruthenium catalyst and the experimental techniques employed to investigate the CO 2 methanation have a major impact on the nature of the intermediates formed and, therefore, the conclusions drawn about the reaction mechanism.
3.3. In Situ NAP−XPS Experiments. The spectra recorded under steady-state conditions for Ce 3d and Ru 3d XPS regions are shown in the Supporting Information (see Figures S4 and S5, respectively). From these spectra, the percentages of Ru 0 and Ce 3+ were calculated, 65 and the results are plotted as a function of temperature in Figures 4 and 5, respectively. After the reduction pretreatment, NAP−XPS reveals that the surface of the ruthenium catalyst is mainly reduced, with ca. 90% Ru 0 as measured with the 722 eV photon energy (see Figure 4 and Table S2). This percentage is Figure 3. Evolution of formates, bidentate carbonates, and ruthenium carbonyl signals as a function of temperature for the experiments performed in the DRIFTS reactor cell under CO 2 methanation gas mixture (10% CO 2 , 40% H 2 , 50% He) after a reduction pretreatment at 450°C with 50% H 2 /He. Figure 4. Evolution of Ru 0 content relative to the total ruthenium as measured by NAP−XPS under CO 2 methanation gas mixture (10% CO 2 , 40% H 2 , 50% N 2 ) after a reduction pretreatment at 550°C with 50% H 2 /N 2 . The photon energies and probed depths (estimated as 3 times the inelastic mean free path, IMFP) are also indicated. These data can be found in the Supporting Information (Table S1) for selected crystalline phases. Figure 5. Evolution of Ce 3+ percentage relative to the total cerium as measured by NAP−XPS under CO 2 methanation gas mixture (10% CO 2 , 40% H 2 , 50% N 2 ) after a reduction pretreatment at 550°C with 50% H 2 /N 2 . The photon energies and corresponding sample probed depths (estimated as 3 times the IMFP) are also indicated. Data for selected crystalline phases can be found in the Supporting Information (Table S1).
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article slightly lower (86%) when probing deeper regions with an incident energy of 972 eV. The presence of surface oxidized Ru after the reduction treatment can be related to the presence of ceria, which promotes the stabilization of oxidized species at the Ru−CeO 2 interface. The introduction of the methanation reaction gas mixture at 100°C leads to the oxidation of the ruthenium surface as evidenced by the drop of the Ru 0 percentages from 90 and 86 to 79 and 74% for the 722 and 972 eV incident energies, respectively. This observation is consistent with the dissociative chemisorption of CO 2 leading to Ru-carbonyls and oxygen atoms, which could potentially oxidize partially the surface as observed by DRIFTS. As the temperature increases, the percentages of Ru 0 increase alongside CO 2 consumption up to similar values to those observed during H 2 pretreatment (see Figure 4). Hence, we conclude that the oxygen atoms left on ruthenium upon dissociative chemisorption of CO 2 at low temperatures (i.e., below 200°C) are removed at higher temperatures in the form of H 2 O and/or transferred to ceria.
The percentage of surface Ce 3+ cations determined by NAP−XPS experiments, shown in Figure 5, amounts to 37% after the reduction pretreatment at 550°C under H 2 , regardless of the probed depth. This percentage decreases when the methanation gas mixture is fed into the reaction cell at 100°C in accordance with the behavior of ruthenium, evidencing that CO 2 chemisorption oxidizes both ruthenium and ceria. The oxidation of Ce(III) to Ce(IV) under these conditions can occur via the adsorption and dissociation of CO 2 to CO + *O on the reduced surface of ceria, as confirmed by DFT calculations (vide inf ra). According to Figure 5, the change in concentration of Ce 3+ (from 37% after H 2 pretreatment to 25−14% at 100°C in CO 2 + H 2 ) is higher than that for Ru (from 90−86 to 79−74%, respectively, as seen in Figure 4). Hence, considering that there is much more ceria than ruthenium in the Ru/CeO 2 catalyst, we posit that the oxygen generated upon CO 2 dissociation, which goes onto ceria, is much larger than that accumulated on ruthenium, highlighting the key role of ceria as oxygen reservoir.
As observed in Figure 5, increasing the temperature beyond 100°C leads to the reduction of the ceria surface up to the level achieved with the H 2 reduction pretreatment. The monitorization of the oxidation state of bulk cerium in a Ru/ CeO 2 catalyst under CO 2 methanation conditions using synchrotron radiation was reported by Wang et al. 43 Therein, operando XANES experiments showed that the concentration of Ce 3+ increased from 3% at room temperature to 9% at 400°C under reaction conditions. We note that these values are much lower than the ones obtained in this work using NAP− XPS ( Figure 5), which we attribute to XANES being a bulksensitive analysis while XPS is surface-sensitive. 67 Therefore, we conclude that ceria bulk remains mainly oxidized, whereas the surface is highly reduced (37% Ce 3+ ) at CO 2 methanation temperatures.
In summary, during the CO 2 methanation reaction, the Ru/ CeO 2 catalyst is reduced on average with a surface composition of ca. 90−86% Ru 0 and 37% Ce 3+ , which is similar to the catalyst state after the H 2 reduction pretreatment at 550°C. This indicates that, during CO 2 methanation, the dissociative chemisorption of H 2 is faster than that of CO 2 , leading to a larger concentration of H atoms, which most likely remain adsorbed on the ruthenium surface.
3.4. DFT Calculations. 3.4.1. Coverage Analysis and Resting State Assessment. DFT calculations were carried out to shed light on the CO 2 methanation pathway using representative facets of the Ru/CeO 2 catalyst. Given the polycrystalline nature of the catalyst, as depicted by XRD ( Figure S2), the lowest-energy surface slabs were selected and modeled. Namely, Ru(0001) and CeO 2 (111), which are inferred to be the most abundant and exposed terminations in the catalyst. The interaction of the surface slabs with the CO 2 + H 2 mixture at representative conditions was studied to assess the participation of each phase in the multistepped reaction mechanism. Because as-prepared ceria is known to exhibit oxygen vacancies, a nonstoichiometric p(2 × 2)-CeO 2−x (111) surface with one surface oxygen vacancy was modeled, which corresponds to a 25% surface reduction. The lowest-energy configuration for this slab corresponds to two surface Ce 3+ ions located in the nearest neighbor (NN) and next-nearest neighbor (NNN) positions relative to the oxygen vacancy, in agreement with previous theoretical studies. 67,68 The modeled Ru(0001) and CeO 2−x (111) surface slabs are depicted in Figure 6a,b, respectively. To assess the catalyst resting state at the onset of the CO 2 methanation reaction, the surface coverage of the CeO 2−x (111) and Ru(0001) slabs was investigated by computing the Gibbs adsorption energies of all of the reactant species under relevant experimental conditions (see the Materials and Methods section for details). DFT calculations revealed that the dissociative chemisorption of H 2 on Ru(0001) has a negligible activation barrier (see Figure S6), rendering this process extremely fast. Hence, we focused our coverage analysis of this surface on the computation of the Gibbs energy of consecutive H adsorptions as a function of temperature. The results derived from this analysis led to the prediction of a surface with all of the fcc sites occupied by H atoms as the catalyst resting state at room temperature ( Figure  S7). In addition, calculations revealed that, as temperature increases, lower H coverages become more stable until the clean surface prevails at temperatures above ca. 300°C. At 225°C , a temperature slightly above the experimental onset found in methanation catalytic tests (ca. 175°C), the Ru surface is predicted to be partially covered with 2H atoms in fcc positions, as shown in Figure 6a. Hence, this surface termination was selected for the mechanistic studies of the CO 2 methanation reaction on Ru(0001).
Similarly to Ru(0001), H 2 activation on the CeO 2 (111) surface has been shown to demand a relatively low energy barrier (ca. 1 eV), leading to the hydroxylation of the surface in a highly exergonic process. 57 Consequently, we modeled the CeO 2−x (111) surface with different levels of hydroxylation as a function of reaction temperature (Figure S7b), finding a coverage with 3OH groups and 1O vacancy (see Figure 6b) as the most likely surface termination within the experimental temperature range of 25−360°C. Hence, we selected this surface coverage to investigate the CO 2 methanation mechanism on CeO 2−x (111).
3.4.2. CO 2 Methanation on Ru(0001). The binding of a CO 2 molecule was investigated on all possible sites of the Ru(0001) surface. The most favorable adsorption sites are depicted in Figure S8. On the clean Ru(0001) surface, we found that CO 2 is weakly adsorbed in a V-shape on an fcc site with a Gibbs energy of +0.15 eV, while CO 2 adsorption is hampered on the H-covered Ru(0001) surfaces. In particular, CO 2 remains physisorbed at 4.02 Å on the 4H-fcc surface as there are no fcc sites available in this coverage. Hence, a H coverage blocks CO 2 adsorption and only when fcc sites become available at high temperatures, CO 2 can be adsorbed. This remark points out that CO 2 adsorption and activation on Ru can be limited at the initial CO 2 methanation conditions and conditioned by the H coverage.
After determining the 2H-covered Ru(0001) surface as the catalyst resting state under relevant CO 2 methanation conditions, the potential CO 2 evolution pathways were investigated at 225°C. The lowest Gibbs energy diagrams obtained for the formation of Ru-carbonyls observed in experiments are depicted in Figure 7. These pathways involve the hydrogenation of CO 2 by the surface H atoms followed by the restitution of the H coverage as the dissociative chemisorption of H 2 has been shown to be extremely fast and highly exergonic.
As shown in Figure 7, after CO 2 adsorption on the Hcovered surface, CO 2 can either split into *CO and *O or be hydrogenated to other species, namely, formate (*OCHO*, red trace) or hydrocarboxylate (*COOH, green trace). Ru(0001) surfaces displaying high H coverages have been reported to favor CO 2 hydrogenation over CO 2 dissociation since the former frees one active site, whereas the latter requires an additional surface site for the binding of the generated *O. 66 However, because a partial H coverage is predicted under our experimental reaction conditions, we find that CO 2 dissociation is the most thermodynamically favored route by 0.30 eV. This step is also independent of the reaction temperature, as illustrated in Figure S9. DFT calculations indicate that the cleavage of CO 2 is exergonic by −0.70 eV and requires a low relative barrier of 0.44 eV (see Figure S10), rendering this process feasible under experimental conditions. Importantly, we note that this step results in the local reoxidation of Ru due to the adsorption of *O, which is in line with the surface reoxidation observed by NAP−XPS. The presence of carbonyl groups on cationic Ru sites observed in the DRIFT spectra could be explained by means of this mechanism, too. However, although our theoretical findings show that CO 2 split on Ru surfaces is thermodynamically feasible under methanation conditions, it is not accurate to attribute such cationic species to this process since the role of the support in the redox state of the Ru particles cannot be ruled out. After CO 2 dissociation, the *O species is hydrogenated in two consecutive steps to form an *OH group and then H 2 O, which desorbs from the surface reducing Ru back to its original state.
Notably, this mechanism is considerably lower in energy than the alternative paths involving formates and their evolution via hydrocarboxylates (*COOH) or formyl groups (*CHO), which can also lead to very stable *CO groups by releasing water to the gas phase, as shown in Figure 7. This is also in line with the broad and intense Ru n+ −CO band detected in in situ DRIFTS experiments all along the reaction course. Overall, Figure 7 shows that Ru-carbonyls are very stable and that these can be formed via different pathways, although the CO 2 dissociative route is predicted to dominate. This may explain the broad and irregular shape of the carbonyl band in the DRIFT spectra, indicating the accumulation of CO in different local environments on the surface.
Once the stable *CO species has been formed on the Ru(0001) surface, these can be hydrogenated to *COH or *CHO, which have been proposed to be further hydrogenated to CH 4 in a rate-determining process. 66 As can be observed in Figure 8, our calculations indicate that CO hydrogenation to  Figure S11 for the TS2 calculation), which is considerably higher than that for CO 2 dissociation. This observation could explain the carbonyl formation at room temperature and their constant concentration on surface up to 400°C. Subsequently, *COH undergoes hydrogenation to *C and H 2 O, which desorbs into the gas phase in an almost thermoneutral process. From this point, the adsorbed *C evolves as the final product CH 4 through consecutive hydrogenations involving exergonic or moderately endergonic steps. One of the endergonic steps of the mechanism corresponds to the hydrogenation of *C to *CH, although this process exhibits a relative energy barrier of only 0.63 eV (difference between TS3 and IVB). Therefore, we assume low comparable barriers for the successive stepwise hydrogenations.
Overall, because of the marked stability of the Ru-carbonyls, we conclude that *CO to CH 4 hydrogenation requires a global activation energy of ca. 2.0 eV, which corresponds to the formation of *COH via TS2. This high energy barrier is in good agreement with in situ experimental DRIFTS, which shows that the Ru−CO band is already intense at room temperature and remains unaltered throughout the reaction course. In addition, there is no sign of CO 2 conversion at temperatures below the CO 2 methanation onset and no CO is detected in the outlet streams, pointing to *CO hydrogenation as the RDS. Hence, both experiments and theoretical studies indicate that CO is strongly bound on Ru(0001), where it accumulates until the temperature is high enough (CO 2 methanation onset is at ca. 175°C) to hydrogenate *CO to CH 4 . The effect of temperature is observed in Figure S9, where the most favored mechanistic route remains the same at 425°C . In conclusion, *CO adsorbs on the H-covered Ru(0001), the first hydrogenation takes place, and the formation of subsequent intermediates is thermodynamically driven.
3.4.3. CO 2 Methanation on CeO 2−x (111). After modeling the CO 2 methanation mechanism on Ru(0001), we turned our attention to ceria using the resting state predicted above with 3OH groups and 1O vacancy.
In contrast to Ru, CeO 2 has good affinity for CO 2 retention, 69,70 which further grows with the presence of reduced Ce 3+ cations. As shown in Figure 9, the binding of CO 2 onto a surface oxygen to form a carbonate species (*OCOO) is exergonic by −0.62 eV on the reduced surface. However, we note that, in the presence of H 2 , these two processes can compete since both hydroxylation and carbonate formation require an available surface oxygen. Interestingly, unlike with Ru(0001), the presence of surface OH groups favors the formation of carbonates. In fact, the 2*OH + *OCOO state is more stable than the coverage with only 3*OH (−2.47 vs −1.67 eV). Therefore, *OCOO and *OH groups are expected to coexist on ceria, in good agreement with DRIFTS. Alternatively, CO 2 can also bind on the oxygen vacancy site in the 3*OH surface as a V-shaped carboxylate species with an energy of −1.73 eV. This adsorption mode is especially relevant since the oxygen vacancy is partially refilled by one of the oxygen atoms from CO 2 , leading to the reoxidation of ceria as observed by NAP−XPS. These carboxylates may evolve in two different pathways as shown in the Gibbs energy diagram presented in Figure 10, namely, via direct splitting into CO and O, which directly refills the oxygen vacancy (red trace), or through the hydrogenation to formates (blue trace).
We note that the resulting CO released via the carboxylate route is not chemisorbed on ceria according to DFT calculations, neither released to the gas phase since no CO   The Journal of Physical Chemistry C pubs.acs.org/JPCC Article was detected in the outlet gases. Therefore, it is very likely that the weakly physisorbed CO interacts with the Ru phase at the triple-phase boundary of the Ru/CeO 2 catalyst giving rise to Ru-carbonyls. Nonetheless, given the low energy barrier (TS4) found for the splitting of carboxylates on ceria (0.54 eV; see Figure S13), we conclude that carboxylates are a potential additional source of Ru-carbonyls at room temperature. Our calculations also predict that carboxylates can be largely stabilized as formates with a slightly larger activation energy of 0.97 eV via TS5 (see Figure S14) and remain on surface until sufficient energy is provided for their decomposition into CO ( Figure 10, blue trace). Importantly, while formates are thermodynamically much more stable than carboxylates, they are predicted to be rather kinetically inert, in agreement with the observation of the depletion of the formates band in DRIFTs above 200°C. Finally, in contrast to formates, carbonates may be merely released as CO 2 as the temperature increases, whose depletion is triggered beyond 300°C according to DRIFTS.
Overall, theoretical calculations reveal that ceria can source CO to the ruthenium catalyst at low and high temperatures, giving rise to Ru-carbonyls in addition to the intrinsic capacity of Ru to split CO 2 with a moderate activation energy. This highlights the important role of ceria as an oxygen sink by assisting in the Ru-catalyzed reaction mechanism, which is not possible with inert supports like Al 2 O 3 or SiO 2 rendering a poorer catalytic performance. Besides, these results shed light on why Ru/CeO 2 catalysts with Ru and CeO 2 phases in tight interaction exhibit improved catalytic performance for the CO 2 methanation reaction, in good agreement with experiments. Simulations show that Ru on its own can promote both H 2 dissociation and CO 2 methanation, while the ceria support facilitates CO 2 dissociation. Since the binding of CO 2 on Ru metal is weak and even more difficult on the H-covered surface, the assistance of ceria is very beneficial to accelerate the CO 2 methanation rate at temperatures around the reaction onset. Future modeling studies of the Ru−Ce interfacial sites should confirm the key role of synergistic interactions arising in Ru-supported cerium oxide catalysts.

CONCLUSIONS
This work has investigated the catalytic behavior of a highperformance Ru/CeO 2 catalyst on the CO 2 methanation reaction by means of in situ spectroscopic techniques and periodic DFT calculations of the individual Ru and CeO 2 phases under representative reaction conditions. Exposure of the CO 2 + H 2 methanation mixture on the Ru metal and CeO 2 results in the partial reoxidation of these surfaces via CO 2 dissociative chemisorption at room temperature, as confirmed by NAP−XPS. The CO 2 chemisorption on both Ru and CeO 2 modeled by DFT supports the oxidative nature of the process.
H 2 activation and dissociation under CO 2 methanation conditions can occur on both the Ru phase, in an almost barrierless process, and on CeO 2−x (111), with a low activation energy, leading to high H and OH coverages, respectively. For ceria, surface hydroxylation is beneficial for CO 2 retention, while the opposite is predicted for Ru. This complementarity seems to have a positive influence on the overall reactivity, enabling additional CO 2 adsorption sites which are predicted to be accessible to Ru at the Ru/CeO 2 interface. The overall rate-determining step in the CO 2 methanation mechanism on the Ru/CeO 2 catalyst is the hydrogenation of Ru-carbonyls with an energy barrier of 2.16 eV.
On ceria, we have shown that CO 2 is accumulated on the surface in the form of strongly bound carbonates, while carboxylates sitting on oxygen vacancies are thermodynamically driven to the generation of formate species. The evolution of the latter stable species requires a considerable activation energy, and hence, they remain adsorbed on the ceria surface up to 200°C. Beyond this temperature, formates can decompose to CO and yield ruthenium carbonyls, as observed by DRIFTS. On the other hand, at lower temperatures, carboxylates are kinetically very reactive leading to CO and O*, restoring the oxygen vacancy and giving rise to the Rucarbonyls observed at room temperature via a spillover mechanism.
Although Ru shows a very weak CO 2 binding unlike CeO 2 , an incipient Ru reoxidation is also observed in CO 2 methanation reaction conditions even at room temperature. This Ru oxidation must be related to the synergistic Ru−CeO 2 interactions, either by O spillover from ceria at the interfacial sites or by enabling highly active interfacial sites, which split CO 2 and stabilize CO very strongly.
Physicochemical characterization of the catalyst, namely, N 2 adsorption−desorption isotherms (Table S1, Figure  S1), X-ray diffraction ( Figure S2) and temperatureprogrammed reduction with H 2 ( Figure S3, The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. All authors contributed equally.