Identification of Active Sites for Reverse Water–Gas Shift Reactions on Pt/TiO₂ Cluster Catalysts

Abstract

The reverse water–gas shift (RWGS) reaction is a key process for CO₂ conversion and sustainable fuel production, yet the nature of the active sites on Pt/TiO₂ cluster catalysts remains elusive. Using first-principles microkinetic simulations, we systematically investigated the catalytic behavior of Pt clusters on TiO₂ under operational reaction conditions. We studied three distinct catalytic sitesPt cluster surfaces, oxygen vacancies (O_V) on TiO₂, and Pt–O_V–Ti interfacesand revealed that the Pt–O_V–Ti interface exhibited the highest RWGS activity via a redox mechanism. This synergy enhances CO₂ activation and facilitates oxygen reduction more effectively than the isolated O_V on TiO₂, which show 4-fold lower activity. In contrast, CO-covered Pt clusters show minimal CO₂ activation but serve as H₂ dissociation sites, enabling hydrogen spillover to adjacent O_V on TiO₂, thereby sustaining the RWGS process. Kinetic analysis revealed OH reduction to H₂O as the rate-determining step on both interfacial Pt–O_V–Ti and at the O_V on the TiO_2–X support. These findings highlight the pivotal role of the Pt–O_V–Ti interface in driving the RWGS and offer a design strategy for optimizing high-temperature CO₂ hydrogenation catalysts by maximizing the number of interfacial active sites.

Introduction

The urgent need to mitigate climate change and reduce reliance on fossil fuels has driven intense research into CO₂ conversion technologies. , The reverse water–gas shift (RWGS) reaction, which converts CO₂ into CO via hydrogenation, is a critical step in carbon recycling, enabling the production of fuels and chemicals − through processes such as Fischer–Tropsch synthesis and the CAMERE (CO₂ hydrogenation to methanol via the RWGS) process. However, the RWGS reaction is endothermic and requires high temperatures , to overcome thermodynamic limitations, making the development of efficient catalysts essential.

Metal catalysts supported on reducible oxides have been widely explored for RWGS because of their ability to activate CO₂ through oxygen vacancies (O_V) and metal–oxide interfacial interactions. − Among them, Pt/TiO₂ catalysts demonstrate outstanding RWGS activity and selectivity, particularly in the 300–500 °C range, − owing to strong metal–support interactions (SMSIs). Highly dispersed Pt clusters (1–2 nm) are particularly promising, offering high atomic utilization, tunable electronic structures, and diverse coordination environments that enhance catalytic performance. While three potential active sitesPt cluster surfaces, O_V on the TiO₂ support, and Pt–O_V–TiO₂ interfacial siteshave been proposed, the precise nature of the active site and its role in the RWGS process remain debated. ,−

Experimental techniques such as XPS, IR, SEM, and XAS provide valuable insights into catalyst structure but face limitations in capturing dynamic surface phenomena and reaction intermediates under operando conditions. − Prior studies have proposed multiple RWGS mechanisms on Pt/TiO₂, including a redox pathway involving both Pt sites and O_V, − a carbonate-mediated mechanism at support vacancies, and a formate/bicarbonate pathway at Pt–O_V–Ti³⁺ interfaces. However, the lack of consensus on these mechanistic pathways has hindered the establishment of the structure–activity relationship necessary for rational catalyst design.

In this study, we employ a multiscale theoretical approach to resolve the active sites and mechanism of RWGS on Pt/TiO_2–X clusters. Ab initio thermodynamic analysis determines the stability of hydroxyl groups and O_V on the TiO_2–X support, whereas grand canonical Monte Carlo (GCMC) simulations identify the most stable Pt cluster structures under reaction conditions, with Pd₈(CO)₆ emerging as the most stable configuration. Microkinetic simulations, which are based on first-principles calculations, reveal that RWGS occurs predominantly via a redox mechanism at both the TiO₂–O_V and Pt–O_V–Ti interfaces. Notably, the Pt–O_V–Ti interfacial sites exhibit the highest intrinsic activity compared with the Pd₈(CO)₆ cluster surface and O_V sites on TiO_2–X , enhancing CO₂ activation by lowering the dissociation barrier. This superior performance is attributed to its enhanced CO₂ activation and efficient O_V regeneration via OH reduction. The Pt cluster has minimal RWGS activity but serves as a H₂ dissociation site, facilitating hydrogen spillover to O_V sites on TiO_2–X , where the reaction proceeds with 4-fold lower activity than the Pt–O_V–Ti interface. Kinetic analysis further identified OH + H → H₂O as the rate-determining step. These findings provide fundamental insights into RWGS catalysis and offer a design strategy for optimizing high-temperature CO₂ hydrogenation catalysts by maximizing the number of interfacial active sites.

Results and Discussion

Oxygen Vacancy-Mediated RWGS on a TiO_2–X Support

O_V on TiO₂ surfaces are well-known active sites for CO₂ activation, facilitating the formation of key intermediates such as carbonates, bicarbonates, and formates. ,,, To assess their role in the RWGS reaction, we systematically investigated the surface structure and vacancy formation dynamics of TiO₂ under reaction conditions. Using rutile TiO₂(110), the most stable and frequently exposed surface, , we simulated its behavior in RWGS-relevant environments. At temperatures above 912 K and a partial H₂ pressure of 0.67 bar, a 1/6 monolayer (ML) of surface oxygen atoms undergoes hydroxylation. As the temperature decreases, the entropy-driven effect of H₂ diminishes, allowing a higher hydroxyl coverage of up to 1/3 ML (Figure a and Figure S1).

1.

Surface structures and PES for RWGS on TiO₂(110). (a) Phase diagram of the OH coverage on the TiO₂ (110) surface with a partial H₂ pressure of 0.67 bar. (b) Gibbs formation energy of oxygen vacancies on TiO₂ (110) with a hydroxylation coverage of 1/3 ML at 803 K and partial pressures of H₂ and H₂O of 0.67 bar and 0.1 mbar, respectively. (c) Configurations (above) and potential energy surface (below) for oxygen vacancy formation on the hydroxylated TiO₂ (110) surface. (d) Reaction mechanism (above) and potential energy surface (below) for CO₂ dissociation activation at the O_V site on the hydroxylated TiO_2–X (110) surface.

At 803 K and with H₂ and H₂O partial pressures of 0.67 bar and 0.10 mbar, respectively, a single O_V per periodic p(2 × 4) unit cell forms on TiO₂(110) with a formation energy of −0.01 eV. However, additional vacancy formation is thermodynamically endothermic, requiring a minimum energy input of 0.07 eV (Figure b and Figure S2). These vacancies originate from the recombination of adjacent hydroxyl groups, generating H₂O and an oxygen defect with an activation barrier of 1.53 eV and an enthalpy of 1.44 eV. The desorption of H₂O, with an energy requirement of 0.95 eV, stabilizes a significant low O_V concentration under high-temperature RWGS conditions (Figure c).

Our calculations show that CO₂ strongly chemisorbs at the O_V on the TiO₂ support, forming CO₃ ^## intermediates with an adsorption energy of −1.13 eV (Figure S3). # indicates the oxygen vacancy site on the TiO₂ support. When hydroxylation occurs at the oxygen atoms adjacent to the O_V, forming an OH^# site, CO₂ either weakly adsorbs or physically binds to the surface three-coordinated lattice oxygen, forming CO₂ ^# and CO₃’^## intermediates with weaker adsorption energies of −0.29 eV and −0.22 eV, respectively. In contrast, in the absence of oxygen vacancies, the CO₂ adsorption energy is much weaker at −0.14 eV. These findings emphasize the critical role of oxygen vacancies in the activation and adsorption of CO₂, which is key to enhance the catalytic activity of the RWGS.

Furthermore, we systematically evaluated three CO₂ hydrogenation pathways involving CO₃ ^##, CO₃H^##, and HCO₂ ^# (Figure d). Our results indicate that CO formation is most favorable via the direct dissociation of CO₃ ^##, which has the lowest activation barrier of 1.70 eV (Figure d, Figure S4 and Table S1). The hydrogenation of CO₃ ^## to CO₃H^## followed by dissociation presents a slightly higher barrier of 1.75 eV, whereas the conversion of CO₃H^## to HCO₂ ^# and its subsequent dissociation is the least favorable, requiring 2.81 eV. Therefore, these findings highlight that oxygen vacancies on TiO₂ primarily facilitate RWGS via the CO₃ ^## and CO₃H^## intermediates, with the CO₃ ^## pathway being the most energetically efficient route for CO production.

Structures of Pt₈/TiO_2–X under RWGS Conditions

To comprehensively explore the nature of the active sites in Pt/TiO₂ cluster-catalyzed RWGS in situ, we employed genetic algorithm (GA)-driven GCMC simulations to determine the Pt₈/TiO_2–X configuration under reaction conditions. We studied the configurations of Pt₈ clusters (∼1 nm) on TiO_2–X with an oxygen vacancy concentration of 1/24 ML. The ‘magic-numbered’ Pt₈ cluster is chosen as a representative model owing to its exceptional stability , and size, which closely aligns with those of experimentally synthesized cluster catalysts. ,,,, In the most stable Pt₈/TiO_2–X structure, the O_V is fully occupied by the Pt₈ cluster and positioned centrally, as depicted in Figure a.

2.

In situ configuration of Pt ₈ /TiO_2–X under RWGS conditions. (a) Adsorption energy of CO and dissociative adsorption energy of H₂ on the Pt₈/TiO_2–X cluster and the structures of Pt₈/TiO_2–X and Pt₈(CO)₆/TiO_2–X identified by GA-driven GCMC simulations. (b) Adsorption free energy for different CO coverages on the Pt₈/TiO_2–X cluster at 803 K and a CO partial pressure of 0.13 bar. (c) Phase diagram of hydroxylation of the interfacial lattice oxygen of Pt₈(CO)₆/TiO_2–X with a partial H₂ pressure of 0.67 bar. (d) Adsorption free energy for varying coverage of H on the Pt₈(CO)₆/TiO_2–X cluster (left) and free energy of oxygen vacancy formation at the Pt₈(CO)₆/TiO_2–X interface (right) at 803 K and partial pressures of H₂ and H₂O of 0.67 bar and 0.1 mbar, respectively. The blue circle indicates the position of interfacial O_V.

Among the species present in the reaction environment, H₂ and CO adsorb strongly onto the Pt cluster surfaces, forming the predominant surface coverage species (Figure S5). Our calculations show that the dissociative adsorption energy of H₂ on Pt₈/TiO_2–X clusters ranges from −0.96 to −0.81 eV, primarily occurring at Pt–Pt bridge sites. In contrast, CO adsorption on Pt is significantly stronger by 1.00–1.20 eV, predominantly occupying the top sites of the Pt atoms (Figure a and Figure S6). These findings indicate that, under reaction conditions, the Pt cluster surfaces are preferentially covered by CO, which could have implications for the reaction pathway and active site identification in the RWGS reaction.

GCMC simulations were conducted at 803 K and a CO partial pressure of 0.13 bar (resulting in 40% CO₂ conversion at a total pressure of 1 bar and a H₂:CO₂ feed ratio of 2:1). As shown in Figure b, the Pt₈/TiO_2–X cluster stabilized with six CO molecules exhibited the lowest formation energy of −2.49 eV. The stability of the Pt₈(CO)_n clusters decreases with increasing CO coverage, suggesting coverage of approximately 3/4 ML on the Pt₈ clusters under reaction conditions (Figures a, b, and Figure S7). In the stable Pt₈(CO)₆/TiO_2–X structure, five CO molecules occupy the top sites of the Pt atoms, one CO molecule occupies a bridge site, and the Pt₈ cluster relocates to its edge to cover the O_V on the support.

We examined the hydroxyl species at the Pt₈(CO)₆/TiO_2–X interface under reaction conditions (Figure c). Across the temperature range of 600–950 K, nearly all the two-coordinated oxygen atoms around the interface were hydroxylated, except for one lattice oxygen atom covered by the Pt clusters. This resulted in a slight decrease in surface hydroxyl coverage compared with that of the clean TiO₂(110) surface, which decreased to 6/23 ML (Figure S8). Additionally, we explored hydrogen species adsorption on the Pt₈ cluster surfaces and the distribution of O_V at the interface. Our analysis revealed that although free sites exist on the Pt₈(CO)₆ cluster surfaces, hydrogen adsorption was thermodynamically unfavorable, with the lowest free energy for adsorbing three hydrogen atoms reaching 0.29 eV (Figure d and Figure S9). This aligns with experimental observations that hydrogen adsorption on Pt surfaces covered by carbonyl groups is challenging. Finally, our calculations demonstrate that the Pt₈(CO)₆/TiO_2–X interface promotes oxygen vacancy formation by −0.14 eV, increasing the O_V concentration to 1/8 ML by generating two adjacent bare oxygen vacancies (Figures d, S10–S11).

Mechanistic Insights into the RWGS over the Pt₈(CO)₆/TiO_2–X System

We calculated the potential energy surface (PES) of RWGS at the interface of the most stable Pt₈(CO)₆/TiO_2–X surface identified by the GCMC simulations under reaction conditions. First, we artificially desorbed a CO molecule from the surface of Pt₈(CO)₆ to form Pt₈(CO)₅, creating vacancies for the subsequent adsorption and dissociation of CO₂ and H₂. CO₂ adsorption on the Pt cluster surface was weak, with an energy of only −0.31 eV (Figure S12). In contrast, at the Pt–O_V–Ti interface, CO₂ adsorption was significantly enhanced, reaching −1.81 eV (Figure S13 and Table S2), far surpassing the adsorption strength at oxygen vacancies on the TiO_2–X support (Figure S3 and Table S1). This highlights the superior CO₂ activation ability of the Pt–O_V–Ti interfacial sites.

Once adsorbed at the Pt–O_V–Ti interface, CO₂ ^& undergoes either direct dissociation or hydrogenation to form intermediates such as COOH^& or HCOO^& (Figure a). & and * are the sites at Pt–O_V–Ti and on the Pt cluster. Direct CO₂ ^& dissociation is the most favorable process, with the lowest activation barrier (E _a) of 0.43 eV (Figure b and Table S2). The oxygen fragment from CO₂ ^& binds to the O_V site, whereas the CO fragment is stabilized on Pt, facilitating CO₂ ^& decomposition (Figure S13). Alternatively, the second pathway involves the hydrogenation of CO₂ ^& to form the HCOO^& intermediate, which then undergoes C–O bond cleavage to form HCO* under the synergistic action of Pt–O_V–Ti interface, eventually dissociating to produce CO, with an apparent activation barrier of 1.06 eV (Figure c). This pathway is hindered primarily by the initial hydrogenation step, whereas the subsequent dissociation steps of HCOO^& and HCO* are relatively facile, with lower activation barriers of 0.60 and 0.33 eV, respectively (Figure b, Figure S13, and Table S2). In contrast, the COOH^& pathway, which involves the formation and dissociation of COOH^& intermediates, presents the highest activation barrier of 1.78 eV, making it the least favorable (Figure c).

3.

RWGS mechanism study over the Pt₈(CO)₆/TiO_2–X system. (a) Scheme of the mechanism of the RWGS reaction involving three distinct CO₂ activations at the interface. (b) Potential energy surface for the RWGS at the Pt–O_V–Ti interfaces. The black, blue and pink lines represent direct CO₂ dissociation, the COOH pathway and the HCOO pathway, respectively. The elementary activation barriers are indicated in eV. (c) Overall activation barriers for CO₂ activation at the Pt–O_V–Ti interfaces. The black, blue and pink columns represent direct CO₂ dissociation, the COOH pathway and the HCOO pathway, respectively. All the energies indicated are in eV. (d) TOF for RWGS and coverage for the intermediates CO₂, H and OH at the Pt–O_V–Ti interface and hydroxylated TiO_2–X (110) surface at 803 K, 1 bar, and a feed ratio of H₂:CO₂ of 2:1. (e) Potential energy surface for H₂ adsorption and migration on a Pt cluster (blue line) and spillover to the TiO_2–X support (green line). All the energies indicated are in eV.

Our microkinetic simulations, which are based on the calculated PES, provide a detailed assessment of the RWGS activity at the Pt–O_V–Ti interface and TiO_2–X support. At 803 K, a total pressure of 1 bar, and a H₂:CO₂ ratio of 2:1, the turnover frequency (TOF) at the Pt–O_V–Ti interface is 2351 s^–1, which is four times greater than the value of 570 s^–1 observed for the TiO_2–X support (Figure d). Consistent with our PES analysis, the microkinetic simulations revealed that the optimal CO₂ activation pathway at these two sites involves direct CO₂ dissociation and CO₃ ^## dissociation, respectively. The degree of rate control (DRC) analysis indicates that the rate-determining step (RDS) at both active sites is the reduction of OH^# to generate H₂O^#. The activation barriers for this elementary step are 0.35 eV at the Pt–O_V–Ti interface and 1.53 eV at the O_V sites on the TiO_2–X support (Figures b and d), respectively. Consequently, the Pt–O_V–Ti interface dominated the RWGS activity, whereas TiO_2–X support has a less contribution to the overall activity.

The coverage of key intermediates, such as CO₂ ^&, CO₃ ^##, H*, H^# and OH^#, provides crucial insight into the differing reactivities between the Pt–O_V–Ti interface and the TiO_2–X support. Specifically, the coverage of CO₂ ^& at the Pt–O_V–Ti interface is 10^–7, which is significantly greater than the CO₃ ^## coverage of 10^–9 on the TiO_2–X support, confirming that the interfacial Pt–O_V–Ti sites greatly enhance CO₂ adsorption and activation. However, the lower equilibrium coverage of H* on the Pt clusters and OH^# at the interface, compared with the higher coverage of H^# and OH^# on TiO_2–X (where H^# refers to H adsorbed on the two-coordinated lattice oxygen adjacent to the O_V, Figure d), mitigates the activity advantage of the Pt–O_V–Ti interface, narrowing the TOF difference between the two structures.

While metallic Pt readily facilitates H₂ dissociation, our calculations confirm that this process is barrierless and strongly exothermic (−1.05 eV) (Figure S14). However, owing to strong competitive adsorption by CO and the optimal site for subsequent CO₂ adsorption at the interface, H* must migrate to the remaining free Pt sites. This migration increases the H₂ dissociative adsorption energy on the Pt₈ cluster to −0.29 eV (Figure e and Figure S15), leading to low H* coverage on the Pt surface. Consequently, the reduction of O^# or OH^# at the interface is hindered. As a result, although OH^# hydrogenation to H₂O^# has a lower activation barrier (0.35 eV) than CO₂ dissociation (0.43 eV) and O^# reduction (0.87 eV), the OH^# reduction step becomes kinetically limiting due to its low coverage, thus governing the overall rate.

On the TiO_2–X support, hydrogen for the RWGS reaction originates from both direct H₂ dissociation and spillover from Pt clusters. The apparent activation barriers for these processes are 1.57 and 1.46 eV, respectively, with a net energy release of −1.30 eV (Figure e and Figures S16 and S17). This strong exothermicity led to high H^# coverage on TiO_2–X , facilitating OH^# reduction. Consequently, despite weaker CO₂ adsorption, a greater CO₂ dissociation barrier, and a greater activation barrier for the RDS on TiO_2–X , its intrinsic RWGS activity is just 4-fold lower than that at the Pt–O_V–Ti interface.

Our theoretical simulations are consistent with experimental studies on Pt/r-TiO₂ clusters in RWGS, combining DFT, KMC simulations, experimental testing, and in situ FTIR characterization, , which confirm that the Pt–O_V–Ti³⁺ interface is the active site. The reported TOF of 6.3 s^–1 for Pt/r-TiO₂ (Pt size ∼ 1 nm) at 623 K and a H₂:CO₂ ratio of 2:1 closely matches our simulation result of 10.4 s^–1 under the same conditions. Further Arrhenius analysis revealed an activation energy of 0.80 eV for CO formation on 0.1%Pt/r-TiO₂ at 250–400 °C, which deviates slightly from our calculated 1.11 eV barrier from microkinetic simulations. These experimental results validate our theoretical model. While the rate-determining step for Pt/TiO₂-catalyzed RWGS remains debated, our integrated structural and microkinetic analysis identifies the OH+H→H₂O step as the rate-determining step, offering new insights into RWGS catalysis.

Origins of Superior Catalytic Performance over Pt–O_V–Ti Interfacial Sites

In first-principles microkinetic simulations, we evaluated three distinct catalytic sites for the RWGS reaction under reaction conditions: the TiO_2–X support, the Pt/TiO_2–X interface, and the Pt cluster surfaces. Our results highlight that the Pt–O_V–Ti interfacial site has the highest intrinsic catalytic activity (Figure d). This enhanced performance stems from the significant promotion of CO₂ activation and the efficient reduction of OH^# species at the interface.

The catalyst surface facilitates electron transfer to CO₂, inducing a rearrangement of its molecular orbitals and disrupting the linear CO₂ configuration. This distortion weakens the CO bond, thereby enhancing adsorption activation. ,, On the TiO_2–X support and Pt clusters, CO₂ adsorption is weaker, with adsorption energies of −1.13 eV and −0.31 eV, respectively. Bader charge analysis revealed minimal charge transfer to CO₂, with values of −0.24 |e| and −0.39 |e|, respectively (Figure a). Specifically, CO₂ adsorption on Pt clusters leads to slight bending of the molecule, resulting in an O–C–O bond angle of 143.7°, which further decreases to 125.6° at the O_V on the TiO_2–X support. In contrast, at the Pt–O_V–Ti interface, charge transfer to CO₂ is significantly greater (−0.79 |e|), with a corresponding adsorption energy of −1.81 eV, and the O–C–O bond angle decreases to 121.4°, indicating more activated CO₂ molecules.

4.

Electronic structure analysis of adsorbed intermediates and active sites. (a) Structures of CO₂ adsorbed on the TiO_2–X (110) support, Pt–O_V–Ti interface and Pt cluster surface sites and their corresponding isosurfaces of differential charge density with isovalues of 0.02 |e|/ų. Yellow and cyan represent charge accumulation and depletion, respectively. (b) Density of state (DOS) plot of CO₂ molecules and bonded Ti, O and Pt atoms at oxygen vacancies on the TiO_2–X (110) support, interfacial Pt–O_V–Ti and Pt cluster surfaces, respectively. (c) Projected crystal orbital Hamilton population (pCOHP) plot of O–Ti bonds for the initial and transition states of OH reduced to H₂O by H at the oxygen vacancy on the TiO_2–X (110) support and interfacial Pt–OV–Ti sites, respectively. (d) Electron density map of the TiO_2–X (110) surface and Pt–O_V–Ti interface.

From a molecular orbital perspective, on the TiO_2–X support, the s and p orbitals of CO₂ strongly hybridize with the d orbitals of Ti atoms at oxygen vacancies below the Fermi level at −9.00, −7.50, and −3.60 eV (Figure b). Additionally, CO₂ forms strong chemical bonds with the lattice oxygen atoms adjacent to O_V. This hybridization, rather than tiny charge transfer, explains the relatively strong adsorption energy of −1.13 eV. In contrast, the interaction of CO₂ with Pt clusters is weak, with limited molecular orbital hybridization below −6.00 eV, which underscores the poor adsorption and activation of CO₂ by Pt clusters, rendering them less effective as active sites for the RWGS reaction. At the Pt–O_V–Ti interface, however, CO₂ undergoes both hybridization with the d orbitals of Ti atoms at the O_V site and strong overlapping interactions with the d orbitals of metallic Pt atoms. This synergy at the Pt–O_V interface significantly enhances CO₂ adsorption and activation, facilitating more efficient CO₂ dissociation.

Moreover, the reduction of OH^# species left behind by CO₂ dissociationidentified as the RDS by microkinetic analysisexhibits significant differences between the O_V sites on the TiO_2–X support and the interfacial Pt–O_V–Ti active sites. Projected crystal orbital Hamilton population (pCOHP) analysis demonstrated that the Ti–O bond formed by the OH^# fragment at O_V sites on the TiO_2–X support was weakened in the transition state compared with the initial state, with a -pCOHP integral of 1.88. In contrast, at the Pt–O_V–Ti interface, the O–Ti chemical bond is significantly stronger, with an -IpCOHP of 2.60 (Figure c). Furthermore, the extent of bond weakening relative to the initial Ti–O bonding strength is significantly enhanced at the Pt–O_V–Ti sites, resulting in a lower reaction barrier at the interface. In summary, the interfacial synergistic sites of Pt–O_V–Ti exhibit stronger binding abilities for both CO₂ and OH^#, facilitating the RWGS reaction. This enhancement is due to the SMSI between the Pt clusters and the TiO_2–X support, which promotes the generation of oxygen vacancies and results in greater charge accumulation at the interface than does the bare TiO_2–X support (Figure d). Consequently, the adsorption of key species and intermediates is stronger, the reaction PES is lower, and the intrinsic catalytic activity is increased.

Conclusions

In this study, we investigated the active sites of Pt/TiO₂ cluster catalysts for the RWGS reaction, focusing on the TiO_2–X support, Pt/TiO_2–X interface, and Pt clusters. GA-driven GCMC simulations revealed the stable Pt₈(CO)₆/TiO_2–X configuration under reaction conditions, followed by first-principles calculations and microkinetic analysis. Our findings reveal that Pt–O_V–Ti interfacial sites exhibit the highest intrinsic activity (TOF = 2351 s^–1), which is attributed to efficient CO₂ activation and oxygen reduction. Although oxygen vacancies on TiO_2–X also contribute (TOF = 570 s^–1), their activity is 4-fold lower. In contrast, Pt clusters show negligible CO₂ activation capacity, but serve as H₂ dissociation sites, facilitating hydrogen spillover to TiO_2–X for RWGS. The strong metal–support interaction (SMSI) between the Pt clusters and TiO_2–X promotes the formation of Pt–O_V–Ti sites, further enhancing CO₂ activation and oxygen reduction. These insights underscore the critical role of Pt–O_V–Ti interfaces in driving efficient CO₂ conversion and provide a strategy for optimizing metal–oxide interfaces to enhance RWGS performance.

Computational Methods

Spin-polarized density functional theory (DFT) calculations were conducted via the Vienna Ab initio Simulation Package (VASP) , with the projector augmented wave (PAW) potential. The generalized gradient approximation (GGA) with the Perdew–Burke–Ernzerhof exchange–correlation functional was adopted to describe the electron exchange–correlation interaction. A kinetic energy cutoff of 400 eV was applied. Convergence criteria for electronic optimization were set at 10^–4 eV/atom, and geometry optimizations were deemed converged when the forces on each ion were less than 0.05 eV/Å. A p(4 × 2) supercell slab with a two-layer oxide thickness was used to simulate the (110) surface of rutile TiO₂. The supported Pt₈ clusters were identified via a GA, and their configurations under reaction conditions were determined via GCMC simulations. A vacuum of 15 Å was used to separate neighboring slabs to avoid dipole effects between them. A Monkhorst–Pack k-point mesh of size 2 × 2 × 1 was used to sample the Brillouin zones. GGA+U correction was used when the strongly correlated d electrons of Ti with U _eff were 4.2.

During structure optimization, the adsorbates, the metal cluster and the topmost oxide are relaxed. The adsorption energies (E _ads) were calculated via the following equation:

$${\mathit{E}}_{\mathrm{ads}}{=\mathit{E}}_{\mathrm{total}}{-\mathit{E}}_{\mathrm{slab}}{-\mathit{E}}_{\mathrm{molecule}(\mathrm{g})}$$ 1

where E _total, E _slab, and E _molecule(g) are the total energies of the adsorbed system, clean slab and adsorbate in the gas phase, respectively. The activation barrier (E _a) and reaction energy (E _r) were calculated as

$$\begin{array}{l}{E}_{\mathrm{a}}=E_{\mathrm{TS}}-E_{\mathrm{IS}}\\ E_{\mathrm{r}}=E_{\mathrm{FS}}-E_{\mathrm{IS}}\end{array}$$ 2

where E _IS, E _FS and E _TS are the total energy of the initial state, final state, and corresponding transition state, respectively. The transition states (TSs) were identified via the improved force reversal method until all the forces were less than 0.05 eV/Å. Some of the TSs were confirmed by climbing-image nudged elastic band (CI-NEB) and dimer methods.

The turnover frequency (TOF), DRC, coverage of intermediates, and reaction mechanism were analyzed via CATKINAS software. The simulations were conducted at a reaction temperature of 803 K, with an initial reactant feed ratio of CO₂ to H₂ of 1:2 and a total pressure of 1 bar. These conditions are consistent with those reported in the literature. The detailed methodologies used for the microkinetic analysis, GA, and GCMC can be found in the Supporting Information.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/prechem.5c00010.

• Detailed methodology, DFT calculation structures, potential energy surfaces and calculated energetics (PDF)

Jin-Xun Liu led the conceptualization and design of the DFT calculations. Feng Li contributed to the DFT calculations and data analysis. All the authors participated in writing the manuscript and in the overall scientific interpretation.

The authors declare no competing financial interest.