Unraveling active sites regulation and temperature-dependent thermodynamic mechanism in photothermocatalytic CO₂ conversion with H₂O

Abstract

In the photothermal synergistic catalytic conversion of CO₂ and H₂O, the catalyst harnesses solar energy to accumulate heat, thereby elevating the reaction system’s temperature. The influence of this temperature effect on surface chemical reactions remains an underexplored area. Here the impact of temperature on the surface-level thermodynamic reactions and conversion of CO₂ with H₂O on oxide semiconductors at the atomic scale was investigated using first-principle calculations. 13 different metal oxides and 5 transition metal clusters were used to introduce surface functional sites on the TiO₂ supporting catalyst. The potential metal oxide cocatalysts that could be most beneficial to the following conversion of CO₂ by H₂O were initially screened by calculating the degrees of promotion of CO₂ adsorption and activation of surface H to provide protons. The proton donation and hydrogen evolution difficulty from H₂O were further analyzed, identifying transition metal cocatalysts that promote direct CO₂ hydrogenation. Upon introducing bifunctional sites to facilitate adsorption and reduction, the production of CH₃OH and CH₄ could be further enhanced through the facilitation of the proton donation process of H₂O. The results of Gibbs free-energy calculations revealed that increasing temperature enhances the reaction thermodynamics for each C1 product formation at different surface sites to varying degrees. These findings offer valuable theoretical insights for designing and regulating active sites on oxide semiconductor surfaces for efficient photothermal catalytic CO₂ reduction by H₂O.

Introduction

Increasing apprehensions concerning global energy demand and greenhouse impacts have fueled the advancement of clean and renewable energy systems. The utilization of solar energy for instigating chemical reactions to convert CO₂ into solar fuels is considered a potential solution, which can simultaneously mitigate the greenhouse effect and ameliorate the energy crisis^1,2. Photothermal synergistic catalytic conversion of CO₂ with H₂O, when viewed from a catalytic pathway perspective, amalgamates the benefits of photogenerated carrier-driven photocatalysis and temperature-driven traditional thermal catalysis, and enables the coupling of non-thermal and thermal effects of solar radiation³. In terms of the feedstock, the abundance and environmental friendliness of hydrogen as a source make this approach exceptionally promising, and the development in this field is rapidly advancing^4,5. In the process of advancing studies on CO₂ photothermal conversion, it is imperative to understand the separate roles of light and heat and, subsequently, grasp the synergistic mechanism. Directly elucidating photothermal synergy and systematically assessing the combined effects of light and heat excitation on catalytic conversion has always been a challenging and complex issue⁶.

Several recent reviews have presented varying definitions for the terminology concerning the impacts of light and heat^{7,8,9,10,11,12}, signifying a growing trend in research that seeks to elucidate these effects from a fundamental mechanistic standpoint. The photothermal process is jointly contributed by photochemistry and thermochemical mechanistic pathways. Accordingly, Mateo et al.⁷ categorized it as “hot-carrier” and “thermal” effects, while Song et al.⁸ referred to it as “light” and “thermal” effects. The more general term, “light” and “heat”, are commonly encountered in current research literature^9,10. Ozin introduced the concepts of “photochemical” and “photophysical” effects⁶, and highlighted the ongoing challenge of determining the relative contributions of these two effects to reaction rate and selectivity. Becker et al.¹¹ adopted a simplified model for evaluating the “photochemical” and “thermochemical” effects in the context of exothermic reactions involving solid catalysts exposed to irradiation. Fang et al.’s review focused on the synergistic effects of “thermal and photo energies” in catalysis¹². Chen et al.¹³ have unveiled the synergistic mechanism involving the coupling of “photo” and “photothermal” effects with specific catalysts like Ru@Ni₂V₂O₇. It should be noted that the temperature effects mentioned here collectively refer to the increase in system temperature after the accumulation of thermal energy caused by thermal effects, which may be attributed to the non-radiative relaxation of semiconductors, localized heating induced by plasma, and thermal vibration of molecules¹⁴. This is relative to the non-thermal effects involving the generation and utilization of high-energy carriers¹⁵. The elevation in reaction temperature will significantly influence the entire reaction system, encompassing reaction thermodynamics and kinetics, product selectivity and yield, thermal active sites, and catalyst structure^9,12. Some recent studies have identified the temperature effects including enhancements in diffusion rates, adsorption-desorption processes, charge carrier migration, and chemical reactions^16,17,18. Temperature also facilitates the surmounting of activation energy barriers¹⁹, activation of thermally active sites²⁰, and adjustments to the redox potential of half-reactions^21,22, along with alterations in catalyst band gaps²³. Nevertheless, there remains a lack of atomic-level understanding of the intricate surface chemistry and the thermodynamic reaction path mechanism affected by temperature in the CO₂ catalytic reduction process^24,25.

While trying to break through research on the mechanisms underlying both non-thermal and thermal effects, scholars turn their eyes toward the target reaction of CO₂ conversion with H₂O, a pivotal and fundamental reaction in the field of solar energy conversion. It is well-known that the chemical properties of CO₂ and H₂O are notably stable, necessitating materials characterized by favorable non-thermal effects to surmount the stringent reaction conditions required for the activation of reactant molecules. Numerous conventional metal oxide semiconductors, including TiO₂, have been utilized for photothermal CO₂ reduction in the presence of pure H₂O. However, the majority of the reported systems thus far have demonstrated relatively low CO₂ conversion rates, typically in the magnitude of μmol·g^-1·h^{-1 26,27}. Based on the basic principles and processes of photothermal catalytic CO₂ reduction by H₂O, ongoing research efforts on improving CO₂ conversion performance mainly focus on strengthening the following three key steps: 1. Photon absorption; 2. Photogenerated carrier separation; 3. Surface catalytic reactions involving CO₂ and H₂O (including adsorption of reactants, and desorption and reduction of intermediates, etc.)²⁸. For the third step, multiple active sites need to be designed on the catalyst surface to simultaneously achieve CO₂ reduction and H₂O dissociation, which still remains tremendously challenging²⁹. In the photothermal catalytic reaction system where CO₂ and H₂O coexist, either H₂O or H₂ might become proton sources. There are three situations in which CO₂ reduction occurs. One is that H₂O first undergoes photooxidation (assisted by photogenerated holes) to produce protons, the second is that H₂O directly provides protons, and the third is that photogenerated holes also assist H₂ dissociation producing protons. Then protons reduce CO₂ with photogenerated electrons. While the process of H₂O splitting for hydrogen evolution can supply H⁺ to reduce CO₂, it simultaneously competes with conduction band electrons for CO₂ reduction, diminishing the effective reduction of CO₂ by photoexcited electrons³⁰. Additionally, H₂O tends to have a higher propensity for surface adsorption on photocatalysts due to its high polarity. In contrast, initial adsorption and activation of CO₂ usually require a huge potential barrier. These result in low CO₂ reduction efficiency and poor hydrocarbon product selectivity^31,32. Surface defect engineering³³ and co-catalyst engineering³⁴ have important applications in photothermal catalytic CO₂ reduction reactions³⁵. In addition to bringing about alterations in the band gap and electronic structure, thereby changing the radiation absorption and conversion performance of the material, the introduction of oxygen vacancies can also function as reaction active sites for the adsorption and activation of surface species during the catalytic process^36,37. Similarly, the incorporation of metal oxides and the loading of nanometal particles can further create tailored functional sites on the material surface, enhancing the selective adsorption and activation of reactants and facilitating the subsequent reduction of key intermediates^38,39. Both approaches have been proven to be powerful means to improve the performance of CO₂ reduction by pure H₂O on the surface of oxide semiconductors⁴⁰. Alkaline-earth metal oxides including MgO, CaO, SrO, BaO^41,42, as well as rare-earth metal oxides such as La₂O₃, CeO₂, Pr₆O₁₁^43,44,45 have been widely used as adsorption cocatalysts to introduce basic active sites on the surface of support oxides. Other metal oxides like Al₂O₃⁴⁶, Ga₂O₃⁴⁷, ZrO₂⁴⁸, Ta₂O₅⁴⁹, HfO₂⁵⁰, and WO₃⁵¹ have also been experimentally explored for their potential in photocatalytic reduction of CO₂ and H₂O, which show promise as promoters to enhance catalytic activity by enriching surface-active sites or modifying the electronic structure. Additionally, Mn, Fe, Co, Ni, and Cu are relatively low-cost representative transition metals capable of activating CO₂ and H₂^52,53,54.

This work aims to reveal the effect of temperature on the surface reaction path mechanism coupling CO₂ reduction and H₂O decomposition involving multiple surface intermediates and hydrodeoxygenation steps. Anatase TiO₂ (aTiO₂) served as the supporting catalyst, and successive modifications were made on its surface using 13 different metal oxides (La₂O₃, Al₂O₃, Ga₂O₃, CaO, MgO, SrO, CeO₂, HfO₂, ZrO₂, BaO, Ta₂O₅, Pr₆O₁₁, WO₃) and 5 transition metal (Mn, Fe, Co, Ni, and Cu) clusters. Subsequently, the adsorption and conversion processes of CO₂ and H₂O at the distinct surface adsorption and reduction sites correspondingly introduced through these modifications were studied. It was found that pure Ti-O-M sites can promote CO₂ adsorption, but do not necessarily facilitate subsequent reduction reactions. The metal oxide cocatalysts that may be most beneficial to the following conversion of CO₂ were initially screened as La₂O₃, Al₂O₃, and Ga₂O₃, which can not only promote CO₂ adsorption but also activate surface H to provide protons. Subsequent loading of Co₄ and Ni₄ transition metal clusters further promoted the proton-donating process of H₂O splitting while inhibiting hydrogen evolution, and are expected to improve the CH₃OH and CH₄ selectivity by further promoting the direct hydrogenation of CO₂. In addition, the results of Gibbs free-energy calculations indicate that increasing temperature enhances the thermodynamics of C1 product formation on the surface at different sites to varying degrees.

Results and discussion

The CO₂ photoreduction with H₂O on the surface of metal oxide semiconductor involves sophisticated multi-reaction paths, accompanied by multiple proton-coupled electrons transfer⁵⁵, and protonation, deoxygenation, and hydroxylation steps. It results in a variety of intermediate species and a complex reaction process. The elaborated thermodynamic reaction network of CO₂ conversion with H₂O on aTiO₂(101)-based surface has been reported in our previous work⁵⁶. To observe its temperature-dependent characteristics, here the reaction paths of CO₂ with H₂O to C₁ products (mainly CO, CH₃OH and CH₄) on pure aTiO₂(101) surface at 100~400 °C were calculated, and the energy barriers comparison of all intermediate steps were shown in Supplementary Fig. 3a. During the process of photothermal catalytic CO₂ conversion with H₂O, both H₂O and H₂ might provide proton sources and reduce CO₂ together with photogenerated electrons. Here we focused on the forward protonation steps of CO₂ conversion (as shown in Fig. 1). It was found that, on pure aTiO₂(101) surface, the energy barriers of certain intermediate steps exhibited an increase as temperature ascends, while others displayed a decrease. However, the fundamental trajectory of the reaction remained unaffected by changes in temperature. The rate-determining step (RDS) of producing CO, CH₃OH, and CH₄ was always the CO₂ adsorption step but with its energy barrier increased from 0.41 eV to 1.04 eV (as shown in Supplementary Fig. 3b), which might be attributed to the fact that the adsorption is exothermic commonly while the increasing temperature would inhibit the reactants adsorption. Therefore, it means that reactant adsorption under high-temperature conditions plays a critical role in the whole reaction. Optimizing surface adsorption sites will be considered first. Furthermore, the subsequent conversion of reactants on the catalyst also exhibited a strong correlation with the presence of active sites on the surface. Basically, introducing adsorption sites and reduction sites on the support semiconductors has been regarded as an effective approach to enhance the performance of photothermal catalytic CO₂ reduction⁵⁷. In this context, the discussion revolves around elucidating the mechanism that governed the influence of temperature on the thermodynamic reaction process path (including the adsorption and conversion of reactants and intermediates) and product selectivity during the photothermal catalysis of CO₂ conversion with H₂O on semiconductor surfaces featuring functional active sites.

Fig. 1: The protonation steps and reaction network of CO₂ conversion with H₂O to C₁ products.

During the photothermal catalytic CO₂ conversion with H₂O, light irradiates H₂O molecules to split and provide protons and electrons to reduce CO₂. It takes two protons and two electrons to generate CO, six protons and two electrons to generate CH₃OH, and eight protons and two electrons to generate CH₄.

Adsorption and activation of CO₂, H₂O, and H atom

The pure aTiO₂(101) surface owns relatively simple reaction sites for reactant adsorption and activation. For the purpose of designing and providing selective sites for the adsorption of CO₂, H₂O, and H, 13 types of M_xO_y (La₂O₃, Al₂O₃, Ga₂O₃, CaO, MgO, SrO, CeO₂, HfO₂, ZrO₂, BaO, Ta₂O₅, Pr₆O₁₁, WO₃) were decorated on the pure aTiO₂(101) surface. As demonstrated in Fig. 2a, the energy barriers for adsorption and activation of CO₂ and H₂O molecule, H atom on surface Ti-O (V_O)-M sites were calculated. It can be concluded that both photoinduced V_O and doped M_xO_y promoted CO₂ adsorption to varying extents. Compared with this, H₂O adsorption was slightly affected, which emphasized the favorable effects of surface oxygen defects and basic sites on promoting the initial activation and selective adsorption of CO₂. Regarding the adsorption of H atoms, La₂O₃, Al₂O₃, and Ga₂O₃ doping exerted the most significant influence among all modified metal oxides. On the Ti-O-M (M=La, Al, Ga) site, the adsorption energy of H attained a notably diminished value. The charge density difference analysis was conducted for H adsorption on M_xO_y-decorated surfaces. As depicted in Supplementary Fig. 4, the electron gains and losses between the H atom and the surface were notably pronounced on surfaces doped with La₂O₃, Al₂O₃, and Ga₂O₃, with an increased charge transfer observed. A higher level of activation of H on the surface could potentially enhance the favorability of CO₂ hydrogenation. It should be noted that the difficulty in creating oxygen vacancies on the surface post-doping with distinct oxides exhibited differing degrees, and the calculated oxygen vacancy formation energy was presented in Fig. 2b for reference. Broadly speaking, when the valence state of the doped oxide’s metal cations M^2y/x+ increased, detaching surface oxygen became more difficult, which might be attributed to the more pronounced charge compensation effect stemming from low-valent metal cations⁵⁸. On surfaces doped with CaO, MgO, SrO, and BaO, the formation energy of V_O experienced a significant reduction, leading to the ease of forming Ti-Vo-M sites. This made reactant molecules and atoms exhibit a propensity for adsorption and activation directly on the defective surface. Comparatively, the presence of photoinduced V_O and the introduction of La₂O₃, Al₂O₃, and Ga₂O₃ on the aTiO₂ surface demonstrated a relatively potent promotional effect on CO₂ activation. Additionally, during photothermal catalysis reactions, light irradiation triggered the generation of oxygen vacancies on the material’s surface. Under these circumstances, the adsorption energy of CO₂ on the surface of La₂O₃, Al₂O₃, and Ga₂O₃-doped TiO₂ was indicated in Supplementary Table 1. Notably, the adsorption energy of CO₂ evolved from −1.20 eV, −0.77 eV, and −0.76 eV (in the absence of surface defects) to −0.61 eV, −0.52 eV, and −0.52 eV respectively, signifying an increase yet maintaining a promotional effect. The enhancement of the basic sites introduced via doping for subsequent CO₂ conversion reactions relied on the availability of protons and electrons. The collaborative presence of synergistic surface oxygen defects and basic sites worked in tandem to amplify the reaction’s thermodynamics, facilitating the multi-electron conversion of CO₂ to yield C1 products⁵⁶. Based on the efficacy of La₂O₃, Al₂O₃, and Ga₂O₃ in augmenting the adsorption and activation of CO₂ molecules and H atoms, comprehensive calculations of the entire thermodynamic reaction paths for photothermal catalytic CO₂ conversion to C₁ products (CO, CH₃OH, and CH₄) at varying temperatures were performed on the respective defective surfaces doped with these compounds.

Fig. 2: Comparison of the difficulty of adsorbing and activating CO₂, H₂O molecules, H atom, and forming V_O on TiO₂ surfaces before and after M_xO_y decorating.

a Energy barriers (E_ads) for activating CO₂, H₂O molecules, and H atoms to CO₂^*, H₂O^*, and H^*, respectively, on different surface sites. b Formation energy of V_O on corresponding surfaces.

Temperature influence on the thermodynamic path of surface CO₂ conversion via enhanced adsorption sites

Thermodynamically, adsorption is exothermic, with a typically negative temperature impact. In other words, as the temperature increases, the strength of adsorption decreases. Yet, when examining surface CO₂ conversion reactions that encompass various intermediate steps, the interplay between temperature and these reactions becomes notably intricate compared to the simpler phenomenon of adsorption. On the surface-active sites that strengthen the adsorption of reactants, the reaction pathways of CO₂ conversion to CO, CH₃OH, and CH₄ at varying temperatures were illustrated in Figs. 3 and 4. As shown in Figs. 3a and 4a, on the defective aTiO₂ surface, the RDS of each product was different from that on the pure aTiO₂ surface. In the process of CO generation through ‘CO₂→^*CO₂→^*COOH→^*CO→CO’, CO₂ adsorption was no longer the RDS on the defective surface. Instead, the RDS could involve either CO desorption or protonation of ^*CO₂ to form ^*COOH. With a temperature increased from 100 °C to 150 °C, CO desorption became the dominant RDS for CO production, resulting in a reduction of the associated energy barrier (from 0.49 eV to 0.31 eV). However, upon further elevation to 200 °C, the RDS transitioned to ‘^*CO₂ protonation to ^*COOH’, accompanied by an energy barrier increase (from 0.27 eV to 0.38 eV) as temperature rose. Similarly, the RDS for CH₃OH production underwent alterations (from ‘^*CH₃OH→CH₃OH’ to ‘^*CHO→^*HCOH’), and the energy barrier first decreased and then increased. This suggested the existence of an optimal thermodynamic temperature range for CO and CH₃OH generation, specifically between 150 °C and 200 °C. Meanwhile, the energy barrier associated with the RDS (^*CHO→^*HCOH) for CH₄ production showed an upward trend with increasing temperature, eventually being consistent with the energy barrier observed for CH₃OH production.

Fig. 3: Reaction free-energy profiles of thermodynamic reaction pathways for multi-electron conversion of CO₂ to C₁ (CO, CH₃OH, and CH₄) products at varying temperatures (100~400 °C).

Different reaction pathways and key intermediate steps are experienced on the surfaces of a aTiO₂ with V_O, b Al₂O₃-doped aTiO₂ with V_O, c La₂O₃-doped aTiO₂ with V_O, and d Ga₂O₃-doped aTiO₂ with V_O that strengthen the adsorption of reactants.

Fig. 4: Comparison of temperature-dependent alterations in the RDS energy barriers for CO, CH₃OH, and CH₄ production.

The Gibbs free-energy change of RDS under the temperature range of 100~400 °C on the surface of a aTiO₂ with V_O, b Al₂O₃-doped aTiO₂ with V_O, c La₂O₃-doped aTiO₂ with V_O, and d Ga₂O₃-doped aTiO₂ with V_O, respectively.

Across defective surfaces featuring various introduced basic sites, the reaction paths of C1 products exhibited discrepancies. Different surfaces tended to favor distinct optimal reaction paths, and engaged diverse reaction intermediates, as presented in Supplementary Fig. 5. When the temperature increased, notable on the Al₂O₃ doped surface as shown in Figs. 3b and 4b, the RDS for CO production experienced a sequential transformation: transitioning from ‘^*COOH→^*CO’ to ‘^*CO₂→^*COOH’, and finally culminating in CO₂ adsorption. This sequential evolution was accompanied by an initial reduction followed by an eventual increase in the associated energy barrier. Notably, a thermodynamically favorable temperature range for CO production emerged, peaking around 150 °C. Likewise, for CH₃OH production, the RDS shifted from ‘^*CH₃OH→CH₃OH’ to ‘^*CH₃O→^*CH₃OH’ with increasing temperature, leading to an initial energy barrier decrease followed by an increase. The most thermodynamically favorable temperature for CH₃OH production was approximately 250 °C. Conversely, the energy barrier for the RDS of CH₄ production (‘^*CH₃O→^*CH₃OH’) increased consistently with temperature. On the La₂O₃ doped surface as shown in Figs. 3c and 4c, the RDS of CO production was always ‘^*CO₂ protonation to ^*COOH’ as the temperature rose, but the energy barrier continued to increase (from 0.98 eV at 100 °C to 1.31 eV at 400 °C), which was not conducive to the formation of CO. The RDS for CH₃OH and CH₄ production decreased first and then increased. In the temperature range from 100 °C to 200 °C, the pivotal step was the conversion of ^*H₂COOH to ^*HCHO, and the energy barrier declined with rising temperature, reaching its minimum value of 0.76 eV at 200 °C. Subsequently, as the temperature continued to rise, the RDS transitioned to ‘^*HCOO→^*HCOOH’, suggesting that the most thermodynamically favorable temperature for CH₃OH and CH₄ production might be approximately 200 °C. In addition, the thermodynamic tendency of the C₁ products evolved with increasing temperature, distinct from both undoped and Al₂O₃-doped surfaces. As shown in Fig. 4c, at lower temperatures (approximately lower than 50 °C), the energy barrier for CO production was the lowest, suggesting a thermodynamic preference for CO generation at this time. While as the temperature rose, the RDS energy barrier for CO production increased until it surpassed that for CH₃OH and CH₄ production, and remained high thereafter, leading to increased selectivity for CH₃OH and CH₄. With Ga₂O₃ doping, as shown in Figs. 3d and 4d, an increase in temperature results in a rising energy barrier for the RDS (‘^*CH₃O→^*CH₃OH’) of CH₄ production. The RDS for CH₃OH production transitioned from CH₃OH desorption to ‘^*CH₃O conversion to ^*CH₃OH’, with its energy barrier displaying an initial decrease followed by an increase. A similar trend was observed for the RDS governing CO production. This transition involved a shift from ‘^*COOH→^*CO’ to CO₂ adsorption as the temperature escalated. Furthermore, at temperatures below about 175 °C, the RDS energy barrier for CH₄ production was lower than that for CO and CH₃OH. However, as the temperature exceeded 175 °C, the energy barrier for CO production decreased below that for CH₃OH and CH₄, signifying that this increasing temperature enhanced the thermodynamic inclination toward CO production.

It can be seen from the above that surface doping with basic metal oxides notably influenced the reaction path for CO₂ reduction to C₁ products by altering the reaction intermediates and key steps. The energy barriers of the rate-determining steps for the formation of C1 products corresponding to the different adsorption-assisting active sites introduced changed and were affected by temperature. The higher temperature accentuated the thermodynamic favorability of multi-electron conversion of CO₂ into C₁ products. The temperature variation resulted in shifts in the relative ease of producing CO, CH₃OH, and CH₄, as shown in Supplementary Fig. 6. To elaborate, at lower temperatures, the RDS energy barrier for CO production was the lowest on Al-V_O-Ti site, while the RDS energy barriers for CH₃OH and CH₄ production were the lowest on pure Ti-O-Ti site, which are lower than those of all defective and doped surfaces. Before 250 °C, it was difficult to produce C1 on La-V_O-Ti, with the energy barrier for CO exceeding that of CH₃OH and CH₄ and higher than that on Ti-O-Ti site. However, as the temperature rose, Ti-V_O-Ti site exhibited the lowest RDS energy barrier for CO production, while Ga-Vo-Ti site exhibited the lowest RDS energy barrier for CH₃OH and CH₄ production, maintaining this status at higher temperatures. The RDS energy barrier for CO production on La-V_O-Ti remained high, but CH₃OH and CH₄ production was enhanced. Although the temperature did not alter the overall reaction path, it did impact the selectivity of each surface C₁ product by modifying the rate-determining steps and energy barriers associated with each C₁ product. Elevated temperatures might promote CO selectivity (with Ga₂O₃ doping) and CH₃OH/CH₄ selectivity (with La₂O₃ doping).

Temperature influence on CO₂ conversion with H₂O at bifunctional sites via further loading transition metal clusters

During the photothermal catalytic reaction, the photoinduced oxygen vacancies on the surface of the oxide semiconductor were likely to be occupied in the intermediate steps of CO₂ conversion (especially in the deoxygenation step), which encompass the following steps: ‘V_O-^*CO₂→^*CO’, ‘V_O-^*COOH→^*COH’, ‘V_O-^*HCOO→^*CHO’, ‘V_O-^*HCOOH→^*HCOH’, ‘V_O-^*HCOO→^*CHO’, ‘V_O-^*CO→^*C’, ‘V_O-^*CHO→^*CH’, ‘V_O-^*HCHO→^*CH₂’, and ‘V_O-^*CH₃O→^*CH₃’. Once the oxygen vacancies were consumed, the doped metal oxide was not conducive to the reduction reaction. As shown in Supplementary Fig. 7, it was calculated that at the M-V_O-Ti (M=Al, La, Ga) site, oxygen vacancies were more prone to being occupied by intermediate species ^*HCOOH, ^*CO₂, and ^*CO₂ respectively to yield ^*HCOH, ^*CO, and ^*CO. Subsequently, ^*HCOH and ^*CO underwent protonation. However, the energy barrier for the subsequent steps, specifically from ^*CH₂OH to form ^*CH₃OH or from ^*CH₃ to form ^*CH₄, was exceedingly high. Consequently, the production of CH₃OH and CH₄ becomes exceptionally challenging at this stage, underscoring the need for further enhancement of the reduction of key intermediates. This also showed that the introduction of adsorption-assisted sites primarily facilitates reactant adsorption, without necessarily guaranteeing the promotion of subsequent reduction reactions.

It was generally believed that transition metal clusters can activate H₂ to ^*H⁵², or promote the H₂O dissociation⁵⁹, thereby providing ^*H to facilitate the direct hydrogenation of CO₂. But at the same time, they might also catalyze the hydrogen evolution of H₂O^60,61, which engendered competition with hydrocarbon products during the reduction of CO₂ by H₂O. Various transition metal clusters were further loaded onto Ga₂O₃-doped TiO₂ surface, to delve deeper into the conversion of CO₂ with H₂O into hydrocarbon products by introducing dual sites that assist in adsorption and reduction. In metal oxide semiconductor systems, the hydrogen evolution step was often identified as the rate-determining stage in the decomposition of H₂O on the surface^62,63. It was considered that a higher energy barrier for the hydrogen evolution step signifies effective suppression of competitive hydrogen evolution. The proton-donating process of water splitting was calculated on the surfaces supported by a series of transition metal clusters (Mn, Fe, Co, Ni, and Cu), including key steps such as water-splitting adsorption, H activation, and hydrogen evolution, as shown in Fig. 5. The results revealed that Mn, Fe, and Cu clusters, when loaded, exhibited lower energy barriers for the hydrogen evolution step compared to the pure aTiO₂ (101) surface, thereby promoting H₂ evolution. Furthermore, Mn and Fe impeded water splitting to yield ^*H. In contrast, the inclusion of Ni- or Co-facilitated water splitting to produce ^*H and activates H while discouraging the formation of H₂. Therefore, we proceeded to calculate the path for CO₂ reduction to produce C₁ on the surface of Ga₂O₃-doped TiO₂ loaded with Ni or Co metal clusters. As shown in Fig. 6, it was found that, compared with the unloaded Ga₂O₃-doped TiO₂ surface, the energy barriers for the RDS in CH₃OH and CH₄ production were significantly lowered when Ni or Co metal clusters were introduced. The impact of Ni was more significant than that of Co. In Comparison with the pure TiO₂ surface, Co could promote the formation of CH₃OH and CH₄ primarily at higher temperatures. In addition, Co loading did not enhance the production of CO, and Ni loading only made the CO’s RDS energy barrier lower when the temperature gradually increased (after 200 °C), with both Ni and Co improving the selectivity of CH₃OH and CH₄. This substantiated the positive role played by the supported Ni₄ or Co₄ cluster in facilitating the direct hydrogenation and multi-electron conversion of CO₂. The electron transfer by Bader charge analysis has been investigated to further explore the possible mechanism and synergistic effect of Ga₂O₃ doping combined with Co or Ni loading on the CO₂ conversion with the H₂O reaction process. The results in Table 1 demonstrated that on Ti-O-Ga··Co₄ and Ti-O-Ga··Ni₄ sites, there was a decrease in positive charge on H and C atoms, while the negative charge on O atoms of CO₂ and those connected to H atoms increased. This alteration suggests an augmented electron transfer between the surface and H or CO₂, leading to an accumulation of electrons on H and CO₂ species. Consequently, the adsorption of CO₂ and H was enhanced, as shown in Supplementary Table 2. Furthermore, the longer CO₂ bond lengths signify a reduction in C=O bond energy and the weakening of the C=O bond. Based on the collective computational findings, it can be inferred that Ga₂O₃ doping combined with Co or Ni loading facilitates CO₂ adsorption and the proton-donating process involved in H₂O splitting by providing additional electrons, thereby facilitating the CO₂ conversion reaction.

Fig. 5: Analysis of energy barriers for key steps (^*H₂O→^*OH+^*H, H→^*H, ^*OH+^*H→^*O+H₂, and ^*H+^*H→^*+H₂) in the proton-donating process of H₂O splitting on different surface-active sites.

The upper dotted line of the figure is aligned with the energy barrier of the hydrogen evolution step (^*OH+^*H→^*O+H₂) at the pure Ti-O-Ti site, and the dotted line below is aligned with the energy barrier of the water-splitting adsorption step (^*H₂O→^*OH+^*H) at the pure Ti-O-Ti site. The surface site where the pink rectangular area is located represents which has the potential to promote water splitting to generate ^*H and activate H while inhibiting the formation of H₂.

Fig. 6: Comparison of RDS energy barriers for CO, CH₃OH, and CH₄ production on the surfaces before and after featuring bifunctional sites under different temperatures (100~400 °C).

Each graph shows the RDS energy barriers of surface sites (Ti-O-Ti, Ti-O-Ga, Ti-O-Ga··Co₄, and Ti-O-Ga··Ni₄) at one specific reaction temperature: a 100 °C, b 150 °C, c 200 °C, d 250 °C, e 300 °C, f 350 °C, and g 400 °C.

Table 1 The Bader charge analysis on Ti-O-Ti, Ti-O-Ga, Ti-O-Ga··Co₄, and Ti-O-Ga··Ni₄ sites after the adsorption of CO₂, H, and corresponding calculated bond lengths

As the temperature rose from 100 °C to 400 °C, combined with Supplementary Fig. 8, it was observed that the reaction path on Ti-O-Ga-M_T4 site for yielding C₁ product remained constant, while the rate-determining steps underwent modifications. On the Co-loaded surface, the RDS for CO production consistently involved CO desorption, with the energy barrier decreasing as temperature increased, but was still higher compared to other surfaces. The RDS for CH₃OH production transitioned from CH₃OH desorption to ‘^*HCHO→^*CH₂OH’, and the energy barrier decreased with increasing temperature. For CH₄ production, the RDS shifted from ‘^*CH₂OH→^*CH₂’ to ‘^*HCHO→^*CH₂OH’. Initially, the energy barrier decreased with rising temperature, but it subsequently began to increase at significantly high temperatures (350 °C). On the Ni-loaded surface, the RDS for CO production changed from ‘^*COOH→^*CO’ to CO₂ adsorption, and the energy barrier diminished as temperature increased. In contrast, the RDS for CH₃OH and CH₄ production shifted from CH₃OH desorption to ‘^*HCOO→^*HCOOH’ and then to CO₂ adsorption, resulting in a larger energy barrier with increasing temperature. Irrespective of temperature, the RDS energy barrier for CH₃OH and CH₄ production on the Ni-loaded surface was always lower than any other surface. The coexistence of dual sites on the surface resulted in CH₃OH and CH₄ dominating the product selectivity even at very high temperatures. In contrast to surfaces solely doped with M₂O₃, wherein only adsorption-assisted sites were introduced, on these surfaces, as temperature increased, the energy barrier for the RDS in CO production remained persistently high, like the production of CH₃OH and CH₄. Alternatively, it gradually decreased until it governed C₁ product selectivity. The dual-site configuration, combining adsorption and reduction sites, not only accelerated the multi-electron conversion of CO₂ to generate CH₃OH and CH₄ but also reduced temperature sensitivity, that is, CH₃OH and CH₄ selectivity was still very high even at very high temperatures.

In this study, how temperature affects the complex intermediate process of photothermal catalytic CO₂/H₂O reaction and C₁ product selectivity was investigated by employing periodic density functional theory methods. The surface reaction path and mechanism before and after regulating the active site were analyzed from the atomic-level perspective. It was revealed that on the pure TiO₂ semiconductor surface, the adsorption of reactants consistently emerged as the rate-determining step for C₁ production at high temperatures and that high temperatures did not favor the exothermic adsorption process. After adsorption activation, the subsequent conversion of CO₂ and H₂O on the semiconductor surface involved multiple surface intermediates and multiple hydrodeoxygenation steps, resulting in the formation of a complex reaction network and potential energy surface. To establish a more universally applicable conclusion regarding the impact of temperature on the thermodynamics of surface reactions, we introduced diverse adsorption sites by incorporating a range of metal oxides onto the TiO₂ surface. Our investigations unveiled that during the photothermal conversion of CO₂, Ti-O-M sites could always enhance CO₂ adsorption, but did not invariably promote subsequent reduction reactions. Temperature variations did not alter the overall direction of the reaction path but brought about alterations in the rate-determining step and the corresponding energy barrier for each C₁ product, thereby changing the selectivity of each C₁ product on the surface which followed the adjustment of active sites. Elevating the temperature might lead to an increased CO selectivity at the Ti-O-Ga site and a higher propensity for CH₄ and CH₃OH selectivity at the Ti-O-La site. Upon introducing bifunctional sites (Ga-O-Ti··M_T4) to facilitate adsorption and reduction, the production of CH₃OH and CH₄ could be further enhanced through the facilitation of the proton donation process of H₂O. However, it was noteworthy that as the temperature increases, it might become less favorable at the Ga-O-Ti··Ni₄ site and might favor an increased selectivity for CO. While higher temperatures could directly augment hydrocarbon product generation at the Ga-O-Ti··Co₄ site, there might be an inflection point where the rate-determining step’s energy barrier started to rise. From a thermodynamic perspective, an increase in temperature did promote the reaction within a certain range, yet there might exist an optimal reaction temperature. The temperature elevation enhanced the reaction thermodynamics of C₁ product formation to varying degrees at different surface sites.

Understanding the surface thermodynamic reaction mechanism affected by temperature and the roles it plays in the adsorption and desorption of reactants, intermediates, and products should be addressed. Here, the present first-principles computational work is complementary to experimentally elucidate the processes promoted by solar-induced temperature, and might provide valuable theoretical insights to guide active site regulation and further experimental design of photothermal catalysts dedicated to the conversion of CO₂ and H₂O into hydrocarbons.

Methods

Computational methods

All calculations have been conducted in the framework of DFT^64,65 using the Perdew–Burke–Ernzerhof (PBE) functional of the generalized gradient approximation (GGA)⁶⁶ as implemented in the Vienna Ab-initio Simulation Package (VASP)^67,68. The core-valence electron interaction was described by the project-augmented wave (PAW) pseudopotential. The empirical correction in the DFT-D3 method with Becke–Jonson damping⁶⁹ was used for long-range dispersion corrections. To balance computational accuracy and cost, the energy cutoff of 400 eV was set in the calculations, along with the convergence threshold for the self-consistent-field iteration set to 10^-6 eV. Geometry optimizations converged when the force components were less than 0.01 eV/Å. The integration in the Brillouin zone was done on a 3 × 3 × 1 k-point grid sampled by the Gamma-centered scheme⁷⁰.

The adsorption energy of CO₂, H₂O, and H, labeled as E_ads-${\text{CO}}_{2}$, E_ads-${\text{H}}_{2}\text{O}$, and E_ads-H were calculated as E_ads = E(total) − E(slab) − E(adsorbed substance). To study how temperature influences CO₂ conversion with H₂O from a thermodynamic point of view, the Gibbs free energy (ΔG) of all intermediates during the reaction were calculated based on statistical thermodynamics^71,72,73 by adding the importance of entropic contribution (S) and zero-point energy (ZPE) corrections as:

$$\Delta {\rm{G}}=\Delta {\rm{H}}-{\rm{T}}\Delta {\rm{S}}$$

(1)

$$\Delta {\rm{H}}\,\approx\, \Delta {\rm{U}}=\Delta {\rm{E}}+\Delta {\rm{ZPE}}+\Delta {{\rm{U}}}_{{\rm{t}}{\rm{ran}}}+\Delta {{\rm{U}}}_{{\rm{rot}}}+\Delta {{\rm{U}}}_{{\rm{vib}}}$$

(2)

$$\Delta {\rm{S}}=\Delta {{\rm{S}}}_{{\rm{t}}{\rm{ran}}}+\Delta {{\rm{S}}}_{{\rm{rot}}}+\Delta {{\rm{S}}}_{{\rm{vib}}}$$

(3)

Where ΔE is the reaction energy acquired from DFT calculations of the optimized structure of reactants and products at each intermediate step, calculated as ΔE = E(products) − E(reactants). ΔZPE is the difference in zero-point energy due to the reaction. ΔU and ΔS are the change in internal energy and entropy, contributed by the translation, rotation, and vibration, respectively, which could be given by:

$${{\rm{U}}}_{{\rm{tran}}}={{\rm{U}}}_{{\rm{rot}}}=\frac{3}{2}{k}_{B}{\rm{T}}$$

(4)

$${{\rm{U}}}_{{\rm{vib}}}=\sum _{{\rm{i}}}\frac{h{v}_{i}}{\left({{\rm{e}}}^{h{v}_{i}/{k}_{B}{\rm{T}}}-1\right)}$$

(5)

$${{\rm{S}}}_{{\rm{tran}}}={k}_{B}\left\{{\mathrm{ln}}\left[{\left(\frac{2{\rm{\pi }}{\rm{m}}{k}_{B}{\rm{T}}}{{h}^{2}}\right)}^{3/2}\frac{{k}_{B}{\rm{T}}}{{\rm{P}}}\right]+\frac{5}{2}\right\}$$

(6)

$${{\rm{S}}}_{{\rm{rot}}}={k}_{B}\left\{{\mathrm{ln}}\left[\frac{\sqrt{{\rm{\pi }}{{\rm{I}}}_{{\rm{A}}}{{\rm{I}}}_{{\rm{B}}}{{\rm{I}}}_{{\rm{C}}}}}{{\rm{\sigma }}}{\left(\frac{8{{\rm{\pi }}}^{2}{k}_{B}{\rm{T}}}{{h}^{2}}\right)}^{3/2}\right]+\frac{3}{2}\right\}$$

(7)

$${{\rm{S}}}_{{\rm{vib}}}={k}_{B}\sum _{{\rm{i}}}\left(\frac{h{v}_{i}}{{k}_{B}{\rm{T}}\left({{\rm{e}}}^{h{v}_{i}/{k}_{B}{\rm{T}}}-1\right)}-{\mathrm{ln}}\left(1-{{\rm{e}}}^{-h{v}_{i}/{k}_{B}{\rm{T}}}\right)\right)$$

(8)

Where T is temperature, v_i are vibrational frequencies that could be calculated by DFT calculations. Then ΔZPE, ΔU, and ΔS are determined from v_i associated with the normal modes of the adsorbed species calculated in the harmonic oscillator approximation, which can be calculated using VASPKIT⁷⁴ after vibration frequency calculation. The temperature (T = 298.15, 373.15, 423.15, 473.15, 523.15, 573.15, 623.15, 673.15 K) dependence of entropic and vibrational contributions to the Gibbs free energy has been explicitly taken into account. It is worth mentioning that this computational strategy for analyzing the influence of temperature effects on surface reaction mechanisms from the thermodynamics perspective has been applied to some reported research in recent years^62,75,76.

Computational models

The modified metal oxide semiconductor systems were constructed based on our previous theoretical and experimental studies^56,77,78. Typically, TiO₂-based catalysts can maintain a stable crystal form of anatase at temperatures below approximately 450 °C. Furthermore, it has been observed that a temperature increase within this range results in a slight decrease in the band gap of TiO₂ due to the expansion of atomic spacing, though this influence is almost negligible²³. In the meantime, taking the temperature range employed in our photothermal catalysis experiments into account, the temperature range of this study is set as 298.15~673.15 K. Consequently, the supporting catalyst, that is, the pure anatase TiO₂, whose most stable surface - (101) surface was directly introduced into the calculation. The aTiO₂(101) surface slab was optimized with three layers in a 4 × 4 supercell (144 atoms), keeping fixed the bottom layer to the bulk atomic positions and the top two layers allowed to relax. A vacuum space of 15 Å along the surface normal direction is applied to minimize unphysical interaction between periodic images and simulate the exposed surface. One bridge O atom was removed to represent the defective surface with oxygen vacancy (V_O). Furthermore, according to the reported literature, for the metal oxides (M_xO_y) decorating surface, the model of M_xO_y doped TiO₂ catalysts was built by incorporating one metal atom to replace one five-coordination Ti atom^79,80,81. And as the previously extensively accepted 4-atom cluster model catalyst^73,82,83,84, the M_T4 cluster consists of four M_T atoms that were put onto the supporting catalyst surface, to signify the transition metal (M_T) loaded surface. The most stable adsorption configurations of all reaction intermediates at the surface reaction sites, were identified by the lowest calculated energy values among all possible configurations. The catalysts’ models and the optimized reaction intermediate structures on corresponding surface-active sites can be seen in Supplementary Figs. 1 and 2, respectively.