Oxygen-modulated engineering of Cu0–Cu+ interfaces for CO2-to-C2H4 photoreduction Authors: Yuqi Ren, Miao Cheng, Hongbin He, Changjun You, Lunbo Duan, Zhenkun Sun, Ruoxuan Peng, Yuan-Hao Zhu, Hongyang Zhang, Jiancheng Zhou, Kai Huang, Jie Guan, Naixu Li Published: 2025 ABSTRACT CO2-to-hydrocarbon fuel conversion via artificial photosynthesis is limited by catalyst poisoning under high O2 partial pressures and sluggish C2-product formation. Herein, Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalysts with adaptive O2 tolerance are prepared by O2-mediated dynamic interfacial reconstruction. Single-atom Ru doping at Ce lattice sites creates a trigonal prismatic coordination configuration, enabling proton-coupled electron transfer and accelerating H2O dissociation. Photothermal effect promotes O2-driven self-assembly of Cu0–Cu1+ charge-gradient interfaces within Ru/Cu alloy clusters, thereby optimizing the adsorption behavior of *CHOCO intermediates and restructuring the C–C coupling pathway. Interfacial charge cascade transfer and geometric site synergy thermodynamically shift the product selectivity from C1 to C2, as determined by operando spectroscopy and electronic structure analysis. Under concentrated solar irradiation, the catalyst produces 549 ± 20 μmol·g−1 C2H4 with 74.3% selectivity and 0.5% solar-to-chemical energy conversion efficiency—25-fold higher than that in non-concentrated systems. Dynamic interfacial regulation facilitates precise carbon chain synthesis in complex reaction networks. Introduction A central constraint impeding the practical implementation of artificial photosynthesis is the efficient conversion of carbon sources under oxygen-rich environments—encountered in industrial flue gas or ambient air1,2. This fundamental challenge arises from cascade attenuation in solar-to-chemical energy conversion. Prevailing strategies for improving energy efficiency employ concentrated light systems to generate transient high-temperature microenvironments and elevated photon-flux densities on catalyst surfaces3,4. However, this approach accelerates water dissociation, triggering explosive proliferation of the O2 byproduct (with molar yields reaching 3–5 times those of the primary CO2 reduction products). Consequently, it induces an exponential rise in system oxygen partial pressure, establishing a self-reinforcing negative feedback loop where intensified light input ↔ microenvironment deterioration mutually amplifies. This loop comprehensively suppresses catalytic activity through dual thermodynamic and kinetic inhibition: (1) Thermally, the oxygen reduction reaction (ORR) competes with the target CO2 reduction reaction (CO2RR) for electron transfer pathways5; (2) Kinetically, persistent accumulation and monolayer coverage of oxygen adsorption states on catalyst surfaces obstruct CO2 chemisorption and activation pathways. Ultimately, conversion efficiency decreases by 2–3 orders of magnitude relative to anaerobic systems6. Current oxygen-tolerance strategies (physical isolation/surface passivation) have remained fundamentally limited by the oxygen-as-poison paradigm and competitive adsorption framework7. These approaches passively suppress ORR but fail to mitigate system-level perturbations induced by drastic oxygen partial pressure fluctuations, creating inherent incompatibility with real-world oxidative environments. Consequently, designing intrinsically oxygen-tolerant gas-solid catalytic interfaces is imperative. Such systems must efficiently mitigate O2 poisoning while productively leveraging oxygen molecules as triggers to induce steady-state self-assembly of active sites. This would enable the directional construction of catalytic microenvironments adaptive to high-oxygen conditions, thereby decoupling efficient C=O bond activation from competitive CO2/O2 adsorption kinetics—a prerequisite for practicaldeployment. During hydrogenation-driven carbon-chain growth, single-atom catalysts (SACs) leverage their unique localized electronic structures and coordinative unsaturated to efficiently dissociate water molecules, continuously supplying active hydrogen species (*H) while optimizing proton-coupled electron transfer (PCET) kinetics8. Ru single atoms stabilized in the 0 < δ+ < 4 oxidation state (Ruδ+) precisely modulate the d-band center via their t2g orbital configuration. Synergistic interaction with the support’s strong electron effects establishes an electron-deficient microenvironment9, enhancing OER efficiency via three mechanisms: (1) Strengthening chemical adsorption of H2O at active sites and lowering the O–H bond dissociation energy barrier; (2) Optimizing adsorption strength of key oxygen intermediates (*OH, *O, and *OOH), circumventing conventional Sabatier principal limitations; (3) Significantly reducing kinetic overpotential in PCET steps. However, spatial confinement at single atomic sites induces competitive adsorption, thermodynamically stabilizing the linear adsorption configuration of *CO intermediates during CO2 activation. This results in symmetry mismatch between the C atom’s px orbital and the metal atom’s dz² orbital, severely inhibiting the sp3 hybridization transition required for C–C coupling. To overcome this limitation, a multi-active-site system integrating single atoms with alloy clusters is essential. The core strategy involves concerted modulation of discrete d-band orbitals and interfacial charge distribution to drive the dynamic transition of *CO from linear to bridging adsorption, thereby facilitating carbon chain growth10. Under aerobic conditions, Cu-based polymetallic interfaces exhibit dual advantages in terms of O2 activation and directional carbon chain construction. The Cu sites dynamically coordinate to form M–O–Cu antioxidation networks, which selectively destruct reactive oxygen species and enable the catalytic microenvironment to self-repair, thereby avoiding catalyst deactivation under oxidative conditions11. Simultaneously, Cu0–Cu1+ charge-gradient interfaces are generated in situ, which induce the reconstruction of *CHOCO adsorption sites through electron redistribution and break the spatial symmetry of linear adsorption configurations. This balances the coverage of intermediates with their reaction kinetics, resulting in a significant enhancement in selectivity for hydrocarbon products12. To address the dual challenges of kinetic hysteresis in C–C coupling and the oxidative deactivation of metal active sites in oxygen-rich systems, we propose an O2-mediated dynamic interfacial reconstruction strategy and successfully construct a Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalytic system with self-adaptive O2 tolerance. Unlike conventional single-site designs, which exhibit inherent limitations owing to their static electronic configurations, this design dynamically self-optimizes the active microdomains during catalytic processes through cross-scale functional decoupling between the Ru single atoms and Ru/Cu sub-nanoscale alloy clusters. Theoretical calculations and in situ characterizations confirmed that the Ru single atoms substitute Ce lattice sites to form a trigonal prismatic coordination configuration, which induces proton-coupled electron transfer to maintain efficient hydrogen spillover. Meanwhile, the photothermal effectinduced O2 accumulation triggers dynamic self-assembly of Cu0–Cu1+ heterointerfaces. These charge-gradient interfaces reduce the formation energy of the critical intermediate *CHOCO. The transition from static sites to dynamic interfaces improves the C2H4 selectivity from 0% to 74.3% under high O2 partial pressures. This study provides new perspectives on dynamic interface regulation for precise carbon chain construction in complex reaction environments. Results Catalyst design for CO2 photoconversion In heterogeneous catalysis, in situ modulation of the electronic structure and coordination environment of metal active sites is essential for improving the catalytic performance13. Single-atom metal species (denoted as M1) have well-defined active sites; however, owing to their isolated active centers, they exhibit inherent limitations in complex redox reactions that require molecular co-adsorption (Fig. 1a). In contrast, metal clusters (denoted as Mn) possess multiple active centers that enable small-molecule activation; however, their performance is constrained by their structural instability and excessive chemical adsorption (Fig. 1b). These limitations can be overcome by strategically positioning Mn sites adjacent to M1 active centers to produce dual-active-site M1 + Mn systems. The synergistic catalytic effects of the M1 and Mn sites enhance the reaction kinetics and significantly improve the overall catalyst performance (Fig. 1c).Fig. 1Mechanistic evolution in multiscale metal-site catalysis.Advantages of M1 (a), Mn (b), and M1 + Mn (c) configurations. d Tackling the catalytic challenges posed by sub-nanoscale metal sites. Stars denote specific advantages, while squares represent existing challenges. e In situ dynamic modulation of sub-nanoscale metal sites in this work. Effective modulation of valence states (i), and electronic structures (ii) of sub-nanoscale metal sites, thereby breaking the activity-selectivity-stability trade-off in C–C coupling (iii). f DFT derived formation energy profiles for 34 distinct Ru-Cu bimetallic configurations, with atomic color-coding reflecting elemental composition (Cu: purple; Ru: silver). The contrast regions correspond to distinct phase microstructures within the Ru3Cu5 intermetallic compound. Mechanistic evolution in multiscale metal-site catalysis. Advantages of M1 (a), Mn (b), and M1 + Mn (c) configurations. d Tackling the catalytic challenges posed by sub-nanoscale metal sites. Stars denote specific advantages, while squares represent existing challenges. e In situ dynamic modulation of sub-nanoscale metal sites in this work. Effective modulation of valence states (i), and electronic structures (ii) of sub-nanoscale metal sites, thereby breaking the activity-selectivity-stability trade-off in C–C coupling (iii). f DFT derived formation energy profiles for 34 distinct Ru-Cu bimetallic configurations, with atomic color-coding reflecting elemental composition (Cu: purple; Ru: silver). The contrast regions correspond to distinct phase microstructures within the Ru3Cu5 intermetallic compound. Despite notable progress in the catalytic studies of metallic species, critical challenges persist (Fig. 1d). First, constructing catalytic sites with spatial/energy multiscale regulation capabilities presents higher-dimensional design challenges compared to traditional M1 systems. According to the Sabatier volcano curve theory, ideal catalysts must precisely modulate the adsorption free energy (ΔG*) of reaction intermediates to overcome the constraints imposed by linear scaling relationships on the catalytic activity maxima14. Second, although M1 + Mn synergistic active sites can integrate multiple reaction pathways, the Mn sites are often prone to excessive chemisorption and exhibit a limited ability to effectively stabilize key intermediates15. Consequently, the design of metallic species near single-atom sites that exhibit high intrinsic activity, selectivity, and stability is critical for enhancing the efficiency of complex catalytic reactions (Fig. 1e(i)). Under high photon-flux density and thermal flow coupled conditions, it may be possible to utilize O2 as a mediator to induce dynamic valence state evolution of the metal species on catalyst surfaces, ultimately enabling the in situ topological restructuring of heterogeneous interfaces (e.g., Cu0–Cuδ+). By regulating the oxidation states of the metal sites, they can more readily adsorb intermediates and catalyze C–C coupling (Fig. 1e(ii)). Consequently, this approach effectively employs byproduct O2 while also enhancing the C2 selectivity and activity (Fig. 1e(iii)). Herein, we fabricated a Ru/Cu composite system on a CeO2 support via wet chemical reduction. Coexisting single atoms and sub-nanoscale alloy clusters were successfully constructed during this dynamic reduction process (Supplementary Fig. 1). Owing to differences in their reduction kinetics, Ru3+ and Cu2+ undergo stepwise reduction to produce metallic atoms, which gradually aggregate through thermodynamically driven diffusion and collision16. The lattice mismatch between Ru and Cu phases, combined with surface energy differences, induces atomic segregation during alloy formation. During the growth of crystal nuclei, migration hysteresis arises due to Ostwald ripening and the higher energy barrier for Ru diffusion17. Some Ru atoms occupy the Cu lattice sites via substitution, whereas others are prevented from entering the Cu lattice owing to localized concentration fluctuations and abrupt Gibbs free energy changes. These atoms, therefore, stabilize in monoatomic states through surface coordination restructuring. Notably, this results in a distinct structure in which isolated Ru atoms coexist with Ru/Cu alloy clusters. We employed density functional theory (DFT) to systematically investigate the formation energies and thermodynamic stabilities of Ru/Cu alloy clusters. A total of 34 supercell models with distinct stoichiometric ratios were constructed (Fig. 1f). Among these, the Ru3Cu5 configuration exhibited the lowest formation energy (0.76 eV). The formation energies of alloy clusters with other atomic proportions gradually increased as the Ru or Cu content increased, confirming the intrinsic stability of Ru3Cu5 as a thermodynamic ground-state structure. Material characterization Transmission electron microscopy (TEM) revealed that the CeO2 support had a highly oriented, regular octahedral morphology, with particle sizes ranging from 20 to 50 nm (Supplementary Figs. 2 and 3). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) revealed lattice parameters corresponding to the characteristic (111) plane of CeO2 (interplanar spacing of 0.31 nm), confirming its single-crystal nature (Fig. 2a). To analyze the distribution patterns of metallic Ru and Cu on the CeO2 support, three-dimensional conformational characterization of Ru0.6Cu1/CeO2 (Co-existence of monatomic Ru and Ru/Cu alloy) was conducted using atomic-resolution HAADF-STEM. The results revealed that Ru atoms were stably anchored on the CeO2 support in a monoatomic dispersion with no detectable agglomeration (Fig. 2b). Direct imaging further visualized Ru atoms occupying Ce lattice sites along the [111] zone axis (Supplementary Fig. 4), with line-scan profiles confirming localized lattice contraction. Additionally, XRD patterns indicated that Ru doping induced a systematic high-angle shift of ~0.15° in the main diffraction peaks (Supplementary Fig. 5). This lattice contraction resulted from substitution of smaller Ru3+ ions for Ce4+ ions, causing anisotropic compression of the fluorite unit cell. Sub-nanoscale alloy clusters (1–3 nm) were also observed on the support surface (Fig. 2c and Supplementary Fig. 6). Their lattice parameters correspond to the (111) plane of RuCu (interplanar spacing of 0.22 nm), which is slightly expanded compared with that of pure Cu (0.21 nm)18. This observation aligns with Vegard’s law and validates the proposed alloying mechanism19, whereby Ru atoms are incorporated into the Cu lattice via atomic substitution. STEM-energy dispersive X-ray spectroscopy (EDX) elemental mapping revealed the spatial co-localization of Ru and Cu within the sub-nanoclusters (Fig. 2d). EDS line scans across representative alloy clusters (Profiles 1 and 2, Fig. 2e) were used to generate superimposed intensity profiles of Ru Lα and Cu Kα signals (Supplementary Fig. 7), confirming atomic-scale electron cloud overlap between the two metallic elements. This is a characteristic feature of intermetallic compounds.Fig. 2Structural and electronic characterization of Ru0.6Cu1/CeO2.a-c, HAADF-STEM image of CeO2 (a) and Ru0.6Cu1/CeO2 (b, c). The illustration presents a locally resolved atomic-scale lattice fringe pattern. d HDAAF-EDX elemental mapping confirms spatial overlap of Cu (red) and Ru (green) species. e Line-scan intensity profiles across positions 1 and 2 in (c). XANES (f) and EXAFS (g) spectra at Ru K-edge of Ru foil, RuO2 and Ru0.6Cu1/CeO2. XANES (h) and EXAFS (i) spectra at Cu K-edge of Cu foil, CuO, Cu2O and Ru0.6Cu1/CeO2. j Wavelet transform EXAFS spectra of Ru foil, RuO2, Ru0.6Cu1/CeO2 at Ru K-edge, and Cu foil, Cu2O and Ru0.6Cu1/CeO2 at Cu K-edge. Ru–O coordination environment in Ru0.6Cu1/CeO2 (k) and Cu–O coordination environment in Ru0.6Cu1/CeO2 (l) catalysts. The illustration presents an enlarged view of the local coordination configuration. Projected density of states analysis of Ru 4d–O 2p (m) and Cu 3d–O 2p (n) hybridization. Crystal orbital Hamiltonian population analysis of Ru–O bonds (o) and Cu–O bonds (p) in Ru0.6Cu1/CeO2. Structural and electronic characterization of Ru0.6Cu1/CeO2. a-c, HAADF-STEM image of CeO2 (a) and Ru0.6Cu1/CeO2 (b, c). The illustration presents a locally resolved atomic-scale lattice fringe pattern. d HDAAF-EDX elemental mapping confirms spatial overlap of Cu (red) and Ru (green) species. e Line-scan intensity profiles across positions 1 and 2 in (c). XANES (f) and EXAFS (g) spectra at Ru K-edge of Ru foil, RuO2 and Ru0.6Cu1/CeO2. XANES (h) and EXAFS (i) spectra at Cu K-edge of Cu foil, CuO, Cu2O and Ru0.6Cu1/CeO2. j Wavelet transform EXAFS spectra of Ru foil, RuO2, Ru0.6Cu1/CeO2 at Ru K-edge, and Cu foil, Cu2O and Ru0.6Cu1/CeO2 at Cu K-edge. Ru–O coordination environment in Ru0.6Cu1/CeO2 (k) and Cu–O coordination environment in Ru0.6Cu1/CeO2 (l) catalysts. The illustration presents an enlarged view of the local coordination configuration. Projected density of states analysis of Ru 4d–O 2p (m) and Cu 3d–O 2p (n) hybridization. Crystal orbital Hamiltonian population analysis of Ru–O bonds (o) and Cu–O bonds (p) in Ru0.6Cu1/CeO2. Synchrotron radiation X-ray absorption spectroscopy (XAS) was employed to further elucidate the structural characteristics of the metallic active centers. Ru K-edge X-ray absorption near edge structure (XANES) analysis demonstrated that the absorption edge at ~22,116 eV was positioned between that of metallic Ru foil (~22,113 eV) and an RuO2 reference material (~22,120 eV), confirming that the average oxidation state of the Ru species lies between the metallic (0) and +4 valence states (Fig. 2f). Using the linear correlation between K-edge shift and oxidation state for transition metals, we determined an average Ru oxidation state of +1.96 (Supplementary Fig. 8). The Ru K-edge Fourier-transform extended X-ray absorption fine structure (FT-EXAFS) spectrum of Ru0.6Cu1/CeO2 exhibited three coordination peaks at bond distances (R) of 1.6–2.3 Å (Fig. 2g). Multi-shell fitting analysis was used to quantitatively resolve the coordination paths (Supplementary Fig. 9). Three coordination structures were identified: short-range Ru–Cu metallic bonding (coordination number [CN] = 0.6 ± 0.2 at R = 2.53 Å), second-nearest-neighbor Ru–Ru coordination (CN = 1.6 ± 0.1 at R = 2.68 Å), and long-range Ru–O interactions (CN = 5.9 ± 0.2 at R = 2.01 Å) (Supplementary Table 1). The Ru–O coordination number and bond length provided strong evidence that a portion of Ru species existed as atomically dispersed single atoms. This CN value closely matched the theoretical coordination number (~6) for Ru in octahedral sites within the fluorite structure of CeO2. Combined with the observed Ru–O bond distance, these results demonstrated that a significant population of Ru atoms occupied CeO2 lattice sites, forming stable, high-coordination bonds with oxygen atoms and thereby anchoring as single-atom species (Ru–O–Ce). This high-coordination environment was distinct from surface-adsorbed oxygen species, which typically exhibited lower coordination numbers (CN < 2). This discovery of a multiscale coordination architecture reveals that Ru atoms primarily exist as isolated single atoms in the Ru0.6Cu1/CeO2 catalyst, with others forming sub-nanoclusters with Cu. In the XANES spectra, the Cu K-edge absorption energy (Fig. 2h) closely aligned with that of metallic Cu foil (8,979 eV), confirming that Cu predominantly remains in the metallic state. Furthermore, EXAFS analysis revealed that the coordination characteristics diverge from those of conventional metallic lattices (Fig. 2i). Compared with metallic Cu with an ideal face-centered cubic structure (theoretical CN = 12 at R = 2.54 Å)20, the sample exhibits a sharp reduction in Cu–Cu coordination number (CN = 1.5 ± 0.1). A Cu–Ru metallic bonding structure was also observed (CN = 1.5 ± 0.3 at R = 2.63 Å). These ultralow coordination numbers (CN < 3) exceed the size-effect limitations typically observed in monometallic Cu nanoparticles, thereby confirming the formation of sub-nanoclusters. The weak Cu–O coordination (CN = 1.1 ± 0.1 at R = 1.88 Å), with bond lengths characteristic of oxygen species adsorbed on metallic Cu surfaces (1.85–1.90 Å), indicates the presence of weak interactions at the support interface rather than oxidized crystalline phases (Supplementary Table 2 and Supplementary Fig. 10). Based on joint time-frequency domain analysis of the wavelet transforms at the bimetallic K-edges, we further provided a definitive criterion for identifying Ru–Cu metallic bonds. The Ru K-edge wavelet transform exhibited a high-intensity signal at coordinate (R, k) = (2.3 Å, 9.3 Å−1) (Fig. 2j). Here, the strong high-k response (>9 Å−1) directly correlates with quantum interference effects in Ru–Cu scattering pathways. Synchronous Cu K-edge wavelet transform analysis captured corresponding Cu–Ru scattering signals at (R, k) = (2.2 Å, 8.0 Å−1). This corroboration of the multiscale structural parameters directly evidences the coexistence of Ru single atoms and RuCu sub-nanoscale clusters in the Ru0.6Cu1/CeO2 catalyst. Considering the selective single-atom dispersion observed during catalyst synthesis, we systematically investigated the doping behavior and stability of Ru and Cu single atoms on CeO2(111) surfaces using DFT calculations21. In the optimized structure, Ru atoms occupying Ce sites via lattice substitution adopt a trigonal prismatic coordination with six neighboring O atoms, exhibiting a formation energy (Eform) of −7.02 eV (Fig. 2k). In contrast, Cu atoms at the same substitution site adopt a square planar coordination configuration, with a significantly higher Eform of −4.12 eV (Fig. 2l). These results confirm that the Ru single-atom/CeO2 system has superior thermodynamic stability. Projected density of states (PDOS) analysis revealed distinct electronic coupling behaviors of the doped Ru and Cu with their seven neighboring O atoms on the CeO2(111) surface. The band center difference between the O 2p and Ru 4d orbitals (Δ = |ε2p(O) − ε4d(Ru)|) is 0.21 eV for spin up and down orbit smaller than that between the O 2p and Cu 3d orbitals (Δ = |ε2p(O) − ε3d(Cu)|, about 0.37 eV), indicating the relatively more substantial rehybridization between the Ru 4d and O 2p orbitals (Fig. 2m–n). Crystal orbital Hamilton population (COHP) analysis further revealed the nature of the Ru–O and Cu–O bonds22. It was found that electrons primarily occupy the bonding states of Ru–O bonds, while the antibonding state was rarely occupied. For Cu–O bonds, however, electrons occupied both bonding and antibonding states (Fig. 2o–p). The much more negative values of the integrated COHP (ICOHP) for Ru–O bonds (−0.63) than that for Cu–O bonds (−0.13) indicated the relatively stronger bonding between Ru and the coordinated O atoms. Overall, integrated DFT modeling and electronic structure analysis consistently demonstrate that doped Ru single atom is a thermodynamically more favorable and stable dopant than Cu atom in the CeO2(111) surfaces. Photothermal CO2 reduction Employing a research strategy that integrates theoretically guided design of catalysts with multiscale experiments, we elucidated the effects of the single-atom/sub-nanoscale alloy cluster systems on the photothermal CO2 reduction performance. First, the catalytic activity of bare CeO2 and single-atom catalysts (Ru0.6/CeO2 and Cu1/CeO2) was evaluated for the co-reduction of CO2 and H2O under non-concentrated solar irradiation (Fig. 3a). The spectral characteristics of the light source and catalyst surface temperature are illustrated in Supplementary Figs. 11 and 12. The Ru0.6/CeO2 catalyst achieved CH4 and CO production rates of 4.7 and 7.7 μmol·g−1·h−1, respectively. This contrasts sharply with the performance of the bare CeO2 and Cu1/CeO2 catalyst, which both exhibited 100% CO selectivity (production rates of 5.2 and 13.6 μmol·g−1·h−1, respectively). This distinct behavior highlights how Ru single-atom sites enhance the H2O dissociation kinetics; single-atom sites facilitate proton-coupled electron transfer, which continuously supplies *H species to the reaction environment, thereby driving the deep hydrogenation of CO2 to CH4. The Gibbs free energy for the O2 evolution reaction is illustrated in Supplementary Fig. 13. The Ru0.6/CeO2 catalyst had a low energy barrier (1.83 eV) for the rate-determining *O → *OOH step, thereby significantly enhancing the O2 evolution kinetics. In contrast, for Cu1/CeO2, the reaction terminated at the CO desorption stage owing to kinetic limitations in *H supply. This corroborates the role of single-atom Ru in optimizing the adsorption configuration of oxygen-containing intermediates. Theoretical screening (Fig. 1f) identifies the Ru3Cu5 cluster as the thermodynamically most stable and kinetically most accessible structure within the RuxCuy/CeO2 system, providing a foundational framework for potential active site configurations in practical catalysts. Guided by this prediction, bimetallic (RuxCuy/CeO2; x = 0.2–1, y = 1–0.2) catalysts with both Ru single-atom sites and sub-nanoscale alloy clusters were prepared (Supplementary Table 3). The Ru0.6Cu1/CeO2 catalyst, developed through precise composition control, demonstrated exceptional catalytic performance, with CO, CH4, and O2 production rates of 40.5, 7.8, and 50.8 μmol·g−1·h−1, respectively (Fig. 3a). This significantly outperformed catalysts with other Ru/Cu ratios. The close stoichiometric match (Ru:Cu ≈ 3:5) to the theoretically predicted optimal Ru3Cu5 structure validates the computational guidance for synthesizing materials with high-activity architectures.Fig. 3O2-mediated photothermal catalytic CO2 reduction performance.a CO2 conversion performance under non-concentrated solar irradiation (1000 mW·cm−2). b CO2 conversion performance under concentrated solar irradiation (5000 mW·cm−2). c Time-dependent evolution rates of CO, CH4, C2H4, and O2 during photothermal catalytic CO2 reduction with H2O over the Ru0.6Cu1/CeO2 catalyst. Error bars represent standard deviations (s.d.) derived from three independent measurements. Data are expressed as mean ± s.d. d CO2 reduction performance under varied conditions. e Exogenous O2 modulates C2H4 formation dynamics. f Quasi-in situ Cu L3VV Auger XPS spectra for Ru0.6Cu1/CeO2 catalyst. g CO2 reduction performance under varying irradiation intensities. h Cycling stability of Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst under concentrated solar irradiation. O2-mediated photothermal catalytic CO2 reduction performance. a CO2 conversion performance under non-concentrated solar irradiation (1000 mW·cm−2). b CO2 conversion performance under concentrated solar irradiation (5000 mW·cm−2). c Time-dependent evolution rates of CO, CH4, C2H4, and O2 during photothermal catalytic CO2 reduction with H2O over the Ru0.6Cu1/CeO2 catalyst. Error bars represent standard deviations (s.d.) derived from three independent measurements. Data are expressed as mean ± s.d. d CO2 reduction performance under varied conditions. e Exogenous O2 modulates C2H4 formation dynamics. f Quasi-in situ Cu L3VV Auger XPS spectra for Ru0.6Cu1/CeO2 catalyst. g CO2 reduction performance under varying irradiation intensities. h Cycling stability of Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst under concentrated solar irradiation. To enhance the direct solar-to-chemical energy conversion efficiency, we performed experiments under concentrated solar irradiation (5000 mW·cm−2) with the aim of increasing high-energy electron transfers and localized thermal effects by harnessing solar-induced photon–phonon coupling (Supplementary Fig. 14). We systematically optimized H2O concentration to maximize performance (Supplementary Fig. 15). The critical design parameters of the CeO2-based and single-metal catalysts were optimized (Supplementary Figs. 16 and 17). Additionally, the catalyst surface temperatures and UV-Vis diffuse reflectance spectra were obtained (Supplementary Figs. 18 and 19). Electronic band structure analyses demonstrated that all catalysts can match the thermodynamic potentials for CO2 reduction and H2O oxidation reactions (Supplementary Figs. 20 and 21). The Ru0.6Cu1/CeO2 catalyst exhibited distinct product selectivity owing to this energy-field coupling, with a C2H4 production rate of 274.5 μmol·g−1·h−1 and concurrent O2 evolution rate of 910.5 μmol·g−1·h−1 (Fig. 3b, Supplementary Fig. 22 and Supplementary Table 4). This sacrificial agent system did not significantly enhance catalytic performance (Supplementary Fig. 23). The fundamental limitation stems from inherent differences between gas–solid systems and conventional liquid–phase reactions. In contrast, a dual single-atom catalyst (Ru SAs + Cu SAs/CeO2) and a catalyst with coexisting Ru single atoms and Cu sub-nanoclusters (Ru SAs + Cu sub-nanocluster/CeO2) yielded only CO and CH4 during photothermal CO2 reduction (Supplementary Figs. 24 and 25). Notably, this corresponds to a significant enhancement in C2H4 selectivity from 0% to 74.3% compared with that under non-concentrated solar irradiation, with electron selectivity reaching 91.4%. Time-resolved kinetic analysis revealed the temporal trend of product evolution (Fig. 3c). During the initial reaction phase (0–30 min), CO accumulated rapidly, reaching a peak concentration of 794 ± 45 μmol·g−1 at 30 min. Some C2H4 evolution (62.6 ± 8 μmol·g−1) and substantial O2 generation (640 ± 25 μmol·g−1) were also observed; therefore, we hypothesized that O2, which is produced as a byproduct of the CO2 reduction reaction, induces selective surface reconstruction of the Ru/Cu alloy clusters through oxidative etching. This process partially converts the Cu0 sites into Cu1+ species, forming dynamic Cu0–Cu1+ charge-gradient interfaces (catalyst designation: Ru0.6Cu1(Cu0–Cu1+)/CeO2). These gradient interfaces enhance the C–C coupling capability. In subsequent reaction stages (30–120 min), C2H4 production exhibited quasi-linear kinetic behavior, with its yield progressively increasing to 549 ± 20 μmol·g−1. Meanwhile, the CO concentration decayed following quasi-first-order kinetics. This C1 → C2 transition aligns with the Langmuir–Hinshelwood surface reaction, indicating that there is dynamic competition between the C2H4 formation pathways and consumption of CO intermediates. To rigorously verify if oxygen is further reduced during CO2 reduction, we performed catalytic experiments under concentrated solar irradiation. Post-extraction analysis detected no quantifiable H2O2 (Supplementary Fig. 26). Mass spectrometry-calibrated and baseline-corrected gas chromatography data validated the concurrent enhancement of C2H4 selectivity with CO conversion (Supplementary Figs. 27 and 28). To establish the importance of photothermal coupling effects in C2H4 generation, we excluded contributions from single-energy modalities (pure photon excitation or thermal activation alone) through multidimensional control experiments (Fig. 3d): (1) Under concentrated UV irradiation or pure thermal activation (462 °C) alone, no C2H4 production was detected, even when the O2 concentration was maintained at 640 ± 25 μmol·g−1 (equivalent to the measured concentration under photothermal conditions); (2) Maintaining temperature at 462 °C while irradiating with lower-intensity UV light (50–250 mW·cm−2) also yielded no C2H4 (Supplementary Fig. 29). Control experiments—including systems without H2O, CO2, or catalytic components—similarly demonstrated no C2H4 signals. Furthermore, the catalytic system can be defined as a thermally assisted photocatalytic system (Supplementary Figs. 30 and 31). The high photon-flux density demonstrated a significant negative correlation with the apparent activation energy (Supplementary Figs. 32 and 33). To elucidate the mechanisms by which O2 dynamically regulates the C–C coupling process, we conducted concentration-gradient experiments coupled with in situ surface analysis. The O2 concentration-response functional relationship (R2 = 0.99) demonstrates high goodness-of-fit of the standard curve (Supplementary Fig. 34). The experimental results revealed that, as the initial O2 concentration increased from 0 to 429 μmol·g−1, the C–C coupling reaction time reduced from 120 to 97 min (Fig. 3e). The valence state of the Cu active sites was monitored using quasi in situ X-ray photoelectron spectroscopy (XPS). During the initial reaction phase under concentrated solar irradiation (0–15 min), the Cu LMM Auger spectra exhibited only the characteristic peaks of metallic Cu (Cu0; 918.6–919.4 eV). At 30 min, Cu1+ species emerged (913.4–913.9 eV), and their relative abundance progressively increased to 58% as the reaction proceeded further (Fig. 3f)23. No Cu2+ species were detected in the higher binding energy range during the quasi in situ XPS analysis (Supplementary Fig. 35). Quasi in situ XANES analysis revealed that the Cu K-edge spectrum matched that of the Cu foil reference at 0 min. The edge shifted to higher energies after 30 min of exposure, indicating oxidation to Cuδ+. This shift intensified by 60 min and stabilized at a position intermediate between the Cu0 and Cu1+ reference spectra by 120 min (Supplementary Fig. 36), indicating a mixed Cu0/Cu1+ oxidation state. This valence evolution correlates well with the results of the O2 concentration gradient experiments; O2 acts as an oxidant, mediating the in situ formation of dynamic Cu0–Cu1+ interfaces, thereby optimizing the adsorption of C–C coupling intermediates. In concentrated solar irradiation regulation experiments (10–60 suns, 1 sun (100 mW·cm−2)), the O2 generation rate was positively correlated with the irradiation intensity (Fig. 3g). However, C2H4 production exhibited light-intensity-dependent nonlinear behavior with a distinct threshold. C2H4 formation commenced at 30 suns (298.6 °C), achieving 7.3% selectivity and O2 yield of 550 μmol·g−1. Under identical thermal conditions with auxiliary illumination (50–300 mW·cm−2), no C2H4 formation was observed (Supplementary Fig. 37). This confirms that O2 concentration is the primary limiting factor. At 50 suns, the C2H4 selectivity had increased 10-fold to 74.3%. This nonlinear amplification effect causes a 25-fold enhancement in the solar-to-chemical energy conversion efficiency (0.5% vs. 0.02% for that under non-concentrated solar irradiation, Supplementary Table 5). The Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst maintains excellent catalytic activity through seven consecutive cycling tests (Fig. 3h). Furthermore, thermogravimetric analysis revealed that minimal mass loss (3.402%) occurred at 500 °C, which was attributed to the desorption of physically adsorbed H2O; no bulk structural decomposition occurred (Supplementary Fig. 38). Differential scanning calorimetry analysis showed no discernible endothermic or exothermic peaks above 300 °C (Supplementary Fig. 39), indicating the absence of significant phase transitions, decomposition, or other thermally induced structural changes. Scanning electron microscopy confirmed that the catalyst morphology was preserved after cycling, indicating that the lattice is structurally stable under high-temperature reaction conditions (Supplementary Fig. 40). In situ XPS analysis revealed that surface Cu species maintain a stable Cu0–Cu1+ configuration upon illumination cessation without reverting to their pre-irradiation state (Supplementary Fig. 41). Critically, this O2induced reconstruction persists post-irradiation, distinguishing it from transient photogenerated metastable states and reversible thermodynamic equilibria. The persistent structural memory—formed non-equilibrium but retained at equilibrium—establishes the Cu0–Cu1+ interface as an intrinsically stable active component. The interfacial charge dynamics of the Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst were elucidated in Supplementary Fig. 42. In situ characterization Product selectivity critically relies on the dynamic adsorption and directional activation of reactants. DFT calculations revealed reactant adsorption dynamics on three structural motifs: the CeO2 substrate, isolated Ru single-atom sites, and RuCu alloy clusters (Supplementary Figs. 43 and 44). RuCu alloy clusters exhibited optimal CO2 adsorption energy (−0.12 eV), significantly higher than values for CeO2 (−0.06 eV) or Ru single-atom sites (−0.08 eV). In contrast, H2O adsorption was strongest on Ru single atoms ( − 1.01 eV), exceeding adsorption on CeO2 (−0.38 eV) or RuCu clusters ( − 0.50 eV), confirming site-specific affinity for polar molecules. Notably, RuCu clusters showed the strongest O2 adsorption energy ( − 2.70 eV), approximately 1.50 eV more negative than those for CeO2 (−1.13 eV) and Ru single atoms ( − 1.81 eV), demonstrating superior oxygen activation capacity. Collectively, these results establish RuCu clusters as synergistic CO2/O2 activation centers. CO2 temperature-programmed desorption (TPD) provided an in-depth analysis of the adsorption behavior of CO2 (Fig. 4a). The Cu1/CeO2 catalyst exhibited a high-intensity desorption peak at 450 °C, demonstrating that the Cu sites have a strong CO2 adsorption capacity. The desorption peak shifted to 250 °C for Ru0.6/CeO2, indicating that the Ru sites exhibit weak CO2 adsorption activity. The Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst had a significantly stronger desorption peak at 515 °C, which corresponds to the formation of monodentate carbonate species (m-CO32−)24. During H2O-TPD (Fig. 4b), the Cu1/CeO2 catalyst displayed only a high-intensity desorption peak at 416 °C, corresponding to the associative dehydration of hydroxyl species under strong metal–support interactions25. Thus, the Cu sites play a limited role in H2O dissociation. In contrast, the Ru0.6/CeO2 catalyst exhibited a characteristic H2O-dissociation-induced desorption peak at 250–350 °C. This phenomenon is attributed to Ru-mediated H2O dissociation on coordinatively unsaturated sites, effectively releasing *H26. Notably, the Ru0.6Cu1/CeO2 and Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalysts also demonstrated distinct desorption peaks at 328 and 340 °C, respectively, which are closely associated with the H2O dissociation activity of Ru single-atom sites. O2-TPD revealed the dynamic oxidation behavior of the Ru/Cu alloy clusters (Fig. 4c). A high-temperature desorption peak was observed at 294 °C for the Cu1/CeO2 catalyst, which reflects the intrinsic self-oxidation characteristics of Cu0 species from Cu0 to Cu1+, driven by strong chemisorption. In contrast, for the bare CeO2 and single-metal Ru0.6/CeO2 catalysts, a desorption peak emerged at a significantly lower temperature of 130 °C (ΔT = − 164 °C vs. that of Cu1/CeO2), which corresponds to the desorption behavior of weakly physisorbed O227. Thus, the Ru sites had limited O2 activation capability. Notably, Ru0.6Cu1/CeO2 exhibited a significantly enhanced desorption signal at 368 °C, suggesting that the bimetallic structure strengthens the adsorption stability of O2 on Ru/Cu alloy, thereby promoting the formation of dynamic Cu0–Cu1+ redox interfaces. Further analysis of the Ru0.6Cu1(Cu0–Cu1+)/CeO2 system revealed a reduced desorption intensity at 377 °C, directly corroborating the notion that the Cu0–Cu1+ heterointerface suppresses excessive O2 adsorption via steric hindrance, thereby maintaining the metastable equilibrium of the Cu species. CO-TPD was performed to elucidate the adsorption behavior of the key reaction intermediate CO (Fig. 4d). The Ru0.6/CeO2 catalyst exhibited a characteristic desorption peak at 278 °C, which corresponds to moderate-strength CO adsorption. Notably, this temperature is near the catalytic reaction temperature (292.3 °C), enabling CO to participate stably in sequential hydrogenation to form CH4. Cu1/CeO2 exhibited a high-temperature desorption peak at 412 °C, indicating significantly stronger CO adsorption. However, this temperature far exceeds the practical reaction conditions (362.5 °C), leading to a sharp decline in CO coverage during the reaction and preventing its deep conversion. Importantly, the Ru0.6Cu1/CeO2 catalyst demonstrated a new desorption signal at 437 °C. The substantial increase in desorption temperature (ΔT = 159 °C vs. that of Ru0.6/CeO2) confirms that Ru/Cu alloy clusters synergistically enhance the CO surface coverage, creating the necessary conditions for C–C coupling. Furthermore, the Ru0.6Cu1(Cu0–Cu1+)/CeO2 system exhibited a high-temperature desorption peak at 449 °C, achieving a marked enhancement in the C2H4 selectivity by strengthening CO adsorption.Fig. 4Active sites verification and operando characterization of C2H4-formation intermediates.a CO2-TPD. b H2O-TPD. c O2-TPD. d CO-TPD. Different background colors represent the adsorption state of reactant molecules in this temperature range. e Operando DRIFTS of CO2/H2O-derived C–C coupling intermediates on Ru0.6Cu1/CeO2. f CO/H2O-derived pathway DRIFTS signatures. g CH4/H2O-derived pathway vibrational modes. Different background colors represent the precipitation process of different active intermediates at this wavelength. Active sites verification and operando characterization of C2H4-formation intermediates. a CO2-TPD. b H2O-TPD. c O2-TPD. d CO-TPD. Different background colors represent the adsorption state of reactant molecules in this temperature range. e Operando DRIFTS of CO2/H2O-derived C–C coupling intermediates on Ru0.6Cu1/CeO2. f CO/H2O-derived pathway DRIFTS signatures. g CH4/H2O-derived pathway vibrational modes. Different background colors represent the precipitation process of different active intermediates at this wavelength. Next, we decoupled the kinetic mechanisms of the C–C coupling pathway by using a reactant substitution strategy combined with diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS). Under constant H2O dosage (1 mL), when the reaction system was converted from CO2/H2O to CO/H2O by substituting CO2 with an equimolar amount of CO, the C2H4 selectivity markedly increased from 74.3% to 89.7%, whereas an equimolar CH4 system completely suppressed C2H4 production (Supplementary Fig. 45). This demonstrates that CO undergoes reversible adsorption-activation on the catalyst surface and serves as the key C1 intermediate for C–C bond formation, whereas CH4 is excluded from the coupling network because of its lack of effective adsorption and dissociation pathways. Using a custom DRIFTS spectrometer (Supplementary Fig. 46), we monitored the real-time evolution of intermediates on Ru0.6Cu1/CeO2 under concentrated solar irradiation (AM 1.5 G (5000 mW·cm−2)) (Fig. 4e). During the initial phase of the CO2/H2O reaction, signals emerged at 1211 and 2017 cm−1, which correspond to activated *COOH species (derived from CO2 activation) and adsorbed *CO species (the key reaction intermediate)28,29, respectively. Concurrently, *CH2OH (1868 cm−1) and *CH3OH (1746 cm−1) signals were observed, which are associated with the CH4 synthesis pathway30. The critical C2H4-forming intermediates, *COCHO (1586 cm−1) and *COCH2O (1314 cm−1), and the final product, *C2H4 (1449 cm−1), emerged at 30 min31. Notably, this time corresponds to the time at which the Cu0–Cu1+ heterointerface formed, as confirmed by in situ XPS (Fig. 3f). This strict temporal coupling between the formation of C–C coupling intermediates and interfacial reconstruction demonstrates that the O2-mediated dynamic interfacial restructuring effectively reduces the energy barrier for C–C coupling, thereby enabling precise regulation of the C2H4 synthesis pathway. Furthermore, the Ru0.6Cu1/CeO2 catalyst surface showed characteristic vibrational modes for surface-bound *OH radicals (3568, 3627, and 3648 cm⁻1) and dissociative H2O adsorption via ʋ(O–H) stretches (3670, 3701, and 3734 cm⁻1; Supplementary Fig. 47). These signatures confirm hole-mediated H2O dissociation pathway that supplies protons for CO2 reduction (Supplementary Figs. 48 and 49). Real-time DRIFTS analyses were also conducted in CO/H2O and CH4/H2O atmospheres instead of CO2/H2O. For the CO/H2O system (Fig. 4f), two adsorbed *CO states (2006 and 2078 cm−1) and a broad-range *COCHO intermediate signature (1586 cm−1) were observed. The splitting of the *CO adsorption states suggests a restructuring of the catalyst surface’s electronic state, substantially enhancing the surface mobility and coupling probability of C1 intermediates32. Sustained peak broadening of the *COCHO intermediate signature was detected as early as 10 min into the reaction, compared with 30 min in the CO2/H2O system. This CO-induced rapid interfacial restructuring further confirms that CO, as a reactant, substantially accelerates C2H4 production. For the CH4/H2O system (Fig. 4g), the main vibrational signature was of CH4 (1306 cm−1). Small peaks of *CH2 (1436 cm−1), *CH3 (1420 cm−1), and C–C coupling (876 cm−1) were also observed, but their intensities did not change over time33,34. This observation reveals the dual kinetic constraints for CH4 molecules on the Ru0.6Cu1/CeO2 surface; it neither achieves C–H bond activation via effective adsorption to generate essential carbon intermediates nor overcomes the energy barrier to surface migration necessary to drive C–C coupling. The C2H4 selectivity enhancement observed in the preliminary experiments and spectroscopic evidence collectively complete the mechanistic validation loop. Mechanistic investigations By comprehensively considering the multimodal electronic structure analysis, thermodynamic evaluations, and kinetic results, we elucidated the mechanisms for regulating the C–C coupling selectivity in the Ru0.6Cu1(Cu0–Cu1+)/CeO2 system. Differential charge density distribution for loading the Ru/Cu cluster on the CeO2 substrate shows an obvious electron transfer from the cluster to CeO2 (Fig. 5a). However, Bader charge analysis reveals that, prior to O2 adsorption, the average charge of the 10 Cu atoms in the loaded Ru/Cu cluster is 11.00 e−. This closely matches the charge of elemental Cu (11.00 e−), thereby confirming that the Cu atoms retain their zero-valent state. Notably, this unique electronic conservation originates from the Ru atoms functioning as an electron buffer at the interface, which donate 1.64 e− to lattice O, while adjacent Ru single-atom sites contribute 1.60 e− to the substrate. This gradient electron donation mechanism effectively shields the Cu active sites from direct electronic interactions with the strongly oxidative substrate, thereby maintaining the metallic character of the Cu clusters.Fig. 5Theoretical calculation.Comparative analysis of side-view differential charge density for adsorption sites on bimetallic clusters under oxygen-deficient (a) and oxygen-rich conditions (b). The yellow and blue regions denote electron accumulation and depletion, respectively, with atomic configurations overlaid for spatial reference. c Localized charge redistribution at ten surface-projected high-symmetry Cu sites in Ru/Cu alloy clusters pre- and post-O2 adsorption. d Adsorption configurations and energies of CO2 at different Cu sites (Cu7, Cu8, Cu10). e Adsorption configurations and energies of CO at different Cu sites (Cu7, Cu8, Cu10). f Gibbs free energy diagram for the conversion of H2O and CO2 into *CH2OCO and *CHOCHO, and side-view optimized structures of key intermediates (*COCO and *CHOCO) along the CO2 pathway between Cu7 and Cu10 sites. Asterisk (*) denotes the active site. Gibbs free energy diagram for the conversion of H2O and CO2 into *CH2CH2 (g) and side-view optimized structures of key intermediates along the CO2 pathway between Cu7 and Cu10 sites (h). Asterisk (*) denotes the active site. The arrow designates the desorption of water molecules. Theoretical calculation. Comparative analysis of side-view differential charge density for adsorption sites on bimetallic clusters under oxygen-deficient (a) and oxygen-rich conditions (b). The yellow and blue regions denote electron accumulation and depletion, respectively, with atomic configurations overlaid for spatial reference. c Localized charge redistribution at ten surface-projected high-symmetry Cu sites in Ru/Cu alloy clusters pre- and post-O2 adsorption. d Adsorption configurations and energies of CO2 at different Cu sites (Cu7, Cu8, Cu10). e Adsorption configurations and energies of CO at different Cu sites (Cu7, Cu8, Cu10). f Gibbs free energy diagram for the conversion of H2O and CO2 into *CH2OCO and *CHOCHO, and side-view optimized structures of key intermediates (*COCO and *CHOCO) along the CO2 pathway between Cu7 and Cu10 sites. Asterisk (*) denotes the active site. Gibbs free energy diagram for the conversion of H2O and CO2 into *CH2CH2 (g) and side-view optimized structures of key intermediates along the CO2 pathway between Cu7 and Cu10 sites (h). Asterisk (*) denotes the active site. The arrow designates the desorption of water molecules. We performed potential energy surface scans of various O2 adsorption configurations on the alloy cluster (Supplementary Fig. 50) to identify the lowest-energy state. It is found that the O2 molecule prefers to bind to the Ru and Cu bimetallic sites with the lowest adsorption energy (ΔEads) of −4.33 eV. Following the adsorption, significant charge redistribution occurs within the cluster (Fig. 5b). The average charge of the 10 Cu atoms decreases from 11.00 to 10.71 e−, corresponding to an overall electron loss of 2.77 e−. The Ru atoms in the alloy and the adsorbed O atoms gain +0.81 and +1.95 e−, respectively. The conservation of charge [ΔQCu (−2.77) ≈ ΔQRu (+0.81) + ΔQO ( + 1.95)] indicates that the adsorption of O2 does not affect the interaction between the cluster and the substrate. Next, we conducted an electronegativity gradient analysis (Cu: 1.9 < Ru: 2.2 < O: 3.44) to elucidate the thermodynamic driving forces for charge transfer. Two pathways were identified: (1) localized electron migration induced by direct polarization of the Cu–O bond, and (2) a Cu → Ru → O (Adsorbed O) cascade transport pathway mediated by interfacial interactions. Dynamic evolution of the Bader electronic structures further corroborated this mechanism; O-coordinated Ru atoms retain their electron relay function while undergoing a 0.39 e− loss35–38. This establishes a coupled mechanism comprising three interfacial interactions: (i) direct charge transfer via Cu–O covalent interactions, (ii) interfacial electron transport through Cu–Ru metallic bonds, and (iii) secondary electron supply governed by Ru–O coordination bonds. These interactions collectively form a hierarchical Cu → Ru → O electron transport pathway. The electron redistribution caused by O2 adsorption leads to the coexistence of 0 and +1 oxidation states for Cu atoms on the surface of the alloy cluster (Fig. 5c), which is crucial for the C–C coupling during the catalytic process. The Cu7, Cu8, and Cu10 sites, which are located at the surface of the O2-mediated alloy cluster, were considered for the adsorption of the intermediates during the reaction. It is found that Cu8, which is in the 0-oxidation state and not directly bonded to O, exhibits a distinct thermodynamic advantage for both the adsorption of CO2 and the formation of the key intermediate *CO (Fig. 5d–e). Afterwards, multiple pathways were considered (Fig. 5f). Both the formation of *COH through O-terminal protonation and the direct desorption of CO encounter high energy barriers (ΔG = 1.55 and 1.02 eV, respectively). In contrast, *COCO can be formed through an exothermic reaction (ΔG = − 0.48 eV), driven by the dual-site combination of Cu with an oxidation state of 0 (Cu10) and +1 (Cu7). However, its geometric configuration is not conducive to C–C bond coupling, leading to a high energy barrier (ΔG = 1.27 eV) for *CHOCO formation. The optimum pathway for C–C bond coupling is *CO → *CHO → *CHOCO, with the sequential ΔG values of 0.61 and 0.18 eV (Supplementary Fig. 51). These results demonstrate the plausibility of the *CHO intermediate-dominated C–C coupling. The hydrogenation of *CHOCO to form *CH2OCO exhibits a lower ΔG of 0.33 eV than *CHOCHO, indicating the favor to *CH2OCO formation. This is strongly corroborated by the stable *CH2OCO species observed via in situ DRIFTS analysis. In *CH2OCO, the 0-oxidation-state Cu10 and +1-oxidation-state Cu7 are bonded to O and C, respectively, due to their different electrophilicity characters. As a result, a bifurcation of the reaction pathway is triggered (Fig. 5g–h). In Route 1, H preferentially attacks the Cu-coordinated O atom in *CH2OCO (ΔG = − 0.38 eV), but the subsequent hydrogenation of *CH2OHCO at the O-terminus encounter an energy barrier (ΔG = 0.61 eV), which is also the rate-limiting step (RDS) of Route 1. In contrast, Route 2 is initiated via an H attack at the distal O atom of the terminal CO group in *CH2OCO. This site exhibits reduced nucleophilicity due to its indirect bonding with the Cu substrate and thus has a positive ΔG of 0.31 eV. Furthermore, the subsequent hydrogenation steps are thermodynamically spontaneous (ΔG = − 0.66 eV), driving the reaction. Notably, due to the significantly low free energy of the *CH2OCH intermediate, the formation of *CH2OHCH has a second-highest ΔG of 0.46 eV, thus serving as the RDS of Route 2. The relatively lower RDS indicates the higher reaction rate of deep hydrogenation through Route 2. Building on an in-depth analysis of the photothermal CO2 reduction dynamic mechanism, we reveal spatiotemporally ordered reconstruction of active interfaces in the Ru0.6Cu1(Cu0–Cu1+)/CeO2 system (Fig. 6). During the initial irradiation phase (t < 30 min), lattice-anchored single-atom Ru sites establish strong electronic metal-support interactions via a trigonal prismatic coordination configuration, accelerating H2O dissociation kinetics. Generated *H migrates to alloy cluster surfaces, coupling with strongly adsorbed *CO intermediates via protonation; this dominates CH4 generation. Concurrently, interfacial oxygen accumulates exponentially with time. Upon exceeding a critical threshold, O2 byproduct adsorb onto Cu0 sites and accept single-electron transfer to form superoxide radicals. Subsequent O–O bond cleavage yields reactive atomic oxygen, which oxidizes Cu0 to Cu1+ via charge transfer. This transformation induces selective oxidative etching and atomic rearrangement on RuCu alloy surfaces, driving self-assembly of a dynamic Cu0–Cu1+ heterointerface with electron gradient distribution. This restructured interface shifts *CO intermediate from linear to bridging adsorption, redirecting products toward C2 products. In the deep reaction phase (30 ≤ t < 120 min), accumulated *CHOCO intermediates stabilize via dual-site chelation: carbonyl carbon coordinates Cu1+, while aldehyde oxygen binds adjacent Cu0 sites. This spatially confined adsorption reduces the *CHOCO → *CH2OCO hydrogenation barrier, dynamically regulates intermediates coverage, and suppresses competing side-reaction pathways. Collectively, this dynamic mechanism shows how the photothermal coupling effect can control the product selectivity by regulating active sites, both spatially and temporally, as well as the distribution of key intermediates.Fig. 6CO2 reduction mechanism.Mechanistic schematic of selective C2H4 production via O2-mediated dynamic Cu0–Cu1+ interface engineering under concentrated solar irradiation. CO2 reduction mechanism. Mechanistic schematic of selective C2H4 production via O2-mediated dynamic Cu0–Cu1+ interface engineering under concentrated solar irradiation. Discussion This study combines multiscale theoretical modeling with experimental analysis to elucidate how photothermal coupling drives dynamic restructuring at metallic active sites. Our findings show that single-atom Ru substitutions at Ce lattice sites create a distorted a trigonal prismatic coordination field, which promotes proton-coupled electron transfer, thereby accelerating H2O dissociation. During catalysis, the generated O2 species trigger the formation of Cu0–Cu1+ charge-gradient heterointerfaces. These dynamically self-assembled sites mediate charge redistribution, which improves the adsorption behavior of *CHOCO intermediates and restructures the C–C coupling pathways. In situ spectroscopy and electronic structure analysis confirm that the combination of charge-gradient interfaces and geometric site synergy helps to steer the product selectivity from C1 to C2. The rationally designed Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst achieves selective C2H4 production at 549 ± 20 μmol·g−1 under concentrated solar irradiation, with 74.3% selectivity and a 25-fold increase in solar-to-chemical energy conversion efficiency (0.5%) compared with that under non-concentrated conditions. This paper presents that photothermal effect-induced O2 accumulation mediates the in situ formation of interfacial charge gradients, overcoming the persistent activity-selectivity-stability trilemma in catalytic carbon chain growth. Methods Materials and reagents All chemical reagents were obtained commercially and used without further purification. Cerium(III) nitrate hexahydrate (Ce(NO3)3·6H2O, 99.95%), sodium hydroxide (NaOH, AR grade), ruthenium(III) chloride hydrate (RuCl3·xH2O, ≥99.95%), copper(II) chloride (CuCl2, ≥99.99%) and ethylene glycol (C2H6O2, ≥99.8%) were purchased from Aladdin Reagent Co., Ltd. Certified standard gas mixtures (CO2/N2 = 1:99 v/v; CH4/N2 = 1:99 v/v; CO/N2 = 1:99 v/v) were supplied by Nanjing Shangyuan Industrial Gas Plant. Ultrapure water (18.2 MΩ·cm resistivity) was used for all aqueous solution preparations. Preparation of catalysts CeO2 The CeO2 catalyst was synthesized through a solvothermal method. In a typical procedure, 1 g of Ce(NO3)3·6H2O was dissolved in 60 mL deionized water under magnetic stirring (600 rpm) for 1 h. Simultaneously, 8 g NaOH was dissolved in 30 mL deionized water with identical stirring conditions. The NaOH solution was then added dropwise (3 mL·min−1) into the Ce(NO3)3·6H2O solution, maintaining continuous stirring (600 rpm) for 2 h to facilitate formation of a Ce(OH)4 precursor suspension. The resulting mixture was transferred to a 100 mL PTFE-lined stainless-steel autoclave and solvothermal treated at 180 °C for 12 h. After cooling to room temperature naturally, the resulting precipitate was collected by vacuum filtration, thoroughly washed with deionized water (3 × 30 mL), then dried under vacuum at 70 °C for 12 h. Finally, the obtained precursor was thermally annealed in a tube furnace under flowing Ar atmosphere (50 mL·min−1) using a temperature-programmed process: The temperature was elevated from ambient to 400 °C at a ramp rate of 5 °C·min−1, followed by isothermal holding at 400 °C for 6 h. Preparation of catalysts Rux/CeO2 The Ru/CeO2 catalyst with 0.6 wt% Ru loading was synthesized through a solvothermal method. In a typical procedure, 800 mg of pre-synthesized CeO2 support was first uniformly dispersed in 30 mL ethylene glycol using magnetic stirring at 600 rpm for 1 h. Subsequently, 13 mg of RuCl3·xH2O was added to achieve the target Ru loading, followed by continuous stirring (600 rpm) for 12 h. The homogeneous suspension was loaded into a 50 mL PTFE-lined stainless-steel autoclave for solvothermal treatment at 160 °C over a period of 28 h. After natural cooling to room temperature, the solid product was separated via vacuum filtration and rinsed exhaustively with deionized water (using two 15 mL aliquots), then dried under vacuum at 70 °C for 12 h. Finally, the obtained precursor was thermally annealed in a tube furnace under flowing Ar atmosphere (50 mL·min−1) using a temperature-programmed process: The temperature was elevated from ambient to 600 °C at a ramp rate of 5 °C·min−1, followed by isothermal holding at 600 °C for 2 h. Catalysts with different Ru loadings (0.1, 0.3, 0.9, 1.2 wt%) were prepared following an identical procedure to that described for the Ru0.6/CeO2 catalyst, with precise adjustment of the RuCl3·xH2O. Preparation of catalysts Cux/CeO2 The Cu/CeO2 catalyst with 1 wt% Cu loading was synthesized through a solvothermal method. In a typical procedure, 800 mg of pre-synthesized CeO2 support was first uniformly dispersed in 30 mL ethylene glycol using magnetic stirring at 600 rpm for 1 h. Subsequently, 18 mg of CuCl2 was precisely added to achieve the target Cu loading, followed by continuous stirring (600 rpm) for 12 h. The homogeneous suspension was loaded into a 50 mL PTFE-lined stainless-steel autoclave for solvothermal treatment at 160 °C over a period of 28 h. After natural cooling to room temperature, the solid product was separated via vacuum filtration and rinsed exhaustively with deionized water (using two 15 mL aliquots), then dried under vacuum at 70 °C for 12 h. Finally, the obtained precursor was thermally annealed in a tube furnace under flowing Ar atmosphere (50 mL·min−1) using a temperature-programmed process: The temperature was elevated from ambient to 600 °C at a ramp rate of 5 °C·min−1, followed by isothermal holding at 600 °C for 2 h. Catalysts with different Cu loadings (0.6, 0.8, 1.2, 1.4 wt%) were prepared following an identical procedure to that described for the Cu1/CeO2 catalyst, with precise adjustment of the CuCl2. Preparation of catalysts RuxCuy/CeO2 The RuCu/CeO2 catalyst with nominal loadings of 0.6 wt% Ru and 1.0 wt% Cu was synthesized through a solvothermal method. In a typical procedure, 800 mg of pre-synthesized CeO2 support was first uniformly dispersed in 30 mL ethylene glycol using magnetic stirring at 600 rpm for 1 h. Subsequently, 13 mg RuCl3·xH2O and 18 mg CuCl2 were co-introduced stoichiometrically to achieve the target metal loading, followed by prolonged stirring (600 rpm, 12 h) to ensure homogeneity. The well-dispersed suspension was hermetically sealed in a 50 mL PTFE-lined stainless-steel autoclave for solvothermal treatment at 160 °C for 28 h. After natural cooling, the precipitate was retrieved by vacuum filtration, washed sequentially with deionized water (2 × 15 mL), and dried under vacuum at 70 °C for 12 h. Finally, the obtained precursor was thermally annealed in a tube furnace under flowing Ar atmosphere (50 mL·min−1) using a temperature-programmed process: The temperature was elevated from ambient to 600 °C at a ramp rate of 5 °C·min−1, followed by isothermal holding at 600 °C for 2 h. Catalysts with varying Ru:Cu mass ratios (1:3, 1:1, 5:3, 3:1) were synthesized using the same impregnation protocol employed for the Ru0.6Cu1/CeO2 catalyst, with controlled variation of RuCl3·xH2O and CuCl2 precursor concentrations. Preparation of catalysts Ru0.6Cu1(Cu0–Cu1+)/CeO2 The Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst was synthesized through in situ activation during photo-thermal CO2/H2O co-reduction under concentrated solar irradiation (5000 mW·cm−2, AM 1.5 G spectrum, 300–2500 nm) for 30 min in a sealed quartz microreactor (100 mL volume). The dynamic evolution of Cu oxidation states was monitored through quasi in situ X-ray photoelectron spectroscopy, revealing coexistence of metallic Cu0 (918.6–919.4 eV) and oxidized Cu1+ (913.4–913.9 eV) as shown in Fig. 3f. This dual-state Cu configuration informed the systematic nomenclature Ru0.6Cu1(Cu0–Cu1+)/CeO2, where subscripts denote nominal metal loadings (wt%) and superscripts indicate quantified oxidation states. Preparation of catalysts CuO/CeO2 The CuO/CeO2 catalysts were synthesized using a glycerol-assisted hydrothermal method followed by calcination. Specifically, 1.5 mmol of Cu(NO3)2·3H2O and 1.0 mmol of Ce(NO3)3·6H2O were dissolved in 40 mL of an aqueous solution containing 2.3 M glycerol with vigorous stirring at room temperature. The resulting solution was loaded into an autoclave constructed of stainless steel and featuring a Teflon liner. This assembly was subjected to a hydrothermal reaction at 180 °C for a duration of 6 h. Upon cooling to room temperature, the solid precipitate was isolated through centrifugation and subsequently rinsed multiple times with anhydrous ethanol ( ≥ 3 times). The precipitate was dried at 80 °C for 12 h under vacuum. The dried precursor was then calcined in static air using a muffle furnace. The temperature was ramped to 350 °C at 5 °C·min−1 and held for 2 h, yielding the final CuO/CeO2 catalyst. Preparation of catalysts Ru SAs + Cu SAs/CeO2 The Ru SAs + Cu SAs/CeO2 catalyst was synthesized via impregnation-calcination. First, 800 mg of pre-synthesized CeO2 support was dispersed in 30 mL of deionized water under magnetic stirring (600 rpm) for 1 h. Subsequently, 13 mg of RuCl3·xH2O and 18 mg of CuCl2 were added to achieve target metal loadings, with stirring (600 rpm) continued for 12 h to ensure thorough precursor adsorption. The precursor solution was vacuum-dried at 70 °C for 12 h. The precursor underwent temperature-programmed reduction in a tube furnace under 50 mL·min−1 of high-purity 5% H2/Ar: heating from 25 °C to 400 °C at 5 °C·min−1, then holding at 400 °C for 2 h to form highly dispersed single-atom sites. After cooling to room temperature under Ar, the sample was collected for use. Preparation of catalysts Ru SAs + Cu sub-nanocluster/CeO2 The Ru SAs + Cu sub-nanocluster/CeO2 catalyst was synthesized via impregnation-calcination. First, 800 mg of pre-synthesized CeO2 support was dispersed in 30 mL of deionized water under magnetic stirring (600 rpm) for 1 h. Subsequently, 13 mg of RuCl3·xH2O and 90 mg of CuCl2 were added to achieve target metal loadings, with stirring (600 rpm) continued for 12 h to ensure thorough precursor adsorption. The precursor solution was vacuum-dried at 70 °C for 12 h. The precursor underwent temperature-programmed reduction in a tube furnace under 50 mL·min−1 of high-purity 5% H2/Ar: heating from 25 °C to 500 °C at 5 °C·min−1, then holding at 500 °C for 2 h to form highly dispersed single-atom sites. After cooling to room temperature under Ar, the sample was collected for use. Characterizations High-resolution TEM (HRTEM) imaging and energy-dispersive spectroscopy (EDS) elemental mapping were performed on a Talos F200X system (Thermo Fisher Scientific). Field-emission SEM (FESEM) characterization was conducted using an Apreo S LoVac (Thermo Fisher Scientific). HAADF-STEM and atomic-resolution imaging were carried out on a Titan Cubed Themis 60–300 microscope (Thermo Fisher Scientific) operated at 300 kV with a probe aberration correction. X-ray absorption spectroscopy measurements including X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) at both Ru K-edge (22117 eV) and Cu K-edge (8979 eV) were acquired at the 14 W beamline of the Shanghai Synchrotron Radiation Facility (SSRF). A Si(311) double-crystal monochromator was employed for energy selection, with reference standards (Ru foil, RuO2, Cu foil, Cu2O) used for energy calibration. All spectra were recorded in transmission mode at ambient temperature and processed using the Athena software package. XPS analysis was performed on an ESCALAB 250Xi spectrometer (Thermo Scientific) with monochromatic Al Kα radiation (1486.6 eV). UV-Vis DRS were obtained using a UV-2450 spectrophotometer (Shimadzu) with an integrating sphere. Thermal analysis included simultaneous TGA-DSC (SDT Q600, TA Instruments) under N2 (10 °C·min−1, 25–520 °C). TPD experiments (CO2/CO/O2/H2O) were conducted on an AutoChem II 2920 system (Micromeritics) with TCD detection. Operando diffuse reflectance infrared Fourier transform spectroscopy measurements were conducted using a Bruker TENSOR II spectrometer coupled with a custom-built solar concentrator reactor system. Temperature monitoring was achieved through combined thermocouple measurements and infrared thermography using an FLIR A65sc camera. Quasi-in situ XPS Ru0.6Cu1/CeO2 samples (50 mg per time point) were irradiated under concentrated solar light (AM 1.5 G spectrum; 5000 mW·cm−2) using a PerfectLight PLS-SXE 300/300UXE solar simulator. Reactions occurred in a 1 vol% CO2/N2 (99.999%) atmosphere with 1 mL H2O for 0, 15, 30, 60, and 120 min. Post-irradiation, samples were transferred to an N2 glovebox, homogenized with 10 mg of spectroscopically pure graphite powder ( <20 μm particle size) in an agate mortar to mitigate charging effects, and densely packed into a sample holder. XPS spectra were acquired at room temperature in a Thermo Scientific ESCALAB 250Xi spectrometer (base pressure ≤ 5 × 10−8 mbar). All binding energies were charge-corrected and referenced to the graphite C 1s peak at 284.8 eV, ensuring cross-comparability across the batch. Operando DRIFTS In an operando DRIFTS system (Bruker TENSOR II spectrometer) equipped with a high-temperature, high-pressure diffuse reflectance cell, 50 mg of Ru0.6Cu1/CeO2 catalyst powder was dispersed on a spectroscopic-grade IR-transparent window. The sample was purged in situ with high-purity (UHP) N2 (≥99.999%, 50 sccm) at 25 °C for 30 min to remove physically adsorbed impurities and moisture; baseline stability was confirmed by real-time IR monitoring. The gas flow was then switched to a reaction mixture of CO2/N2 balanced gas (controlled by calibrated mass flow controllers, MFCs) co-fed with H2O vapor (via a bubbler). All streams were premixed in heated transfer lines before entering the cell. Upon introducing the reaction gas and activating the solar simulator (PerfectLight PLS-SXE 300/300UXE AM 1.5 G spectrum; 5000 mW·cm−2) simultaneously—defining t = 0 min as the reaction start—collimated irradiation was delivered through the cell’s top window onto the catalyst via a Fresnel lens focusing system (focal length: 30 cm). Operando DRIFTS data acquisition commenced immediately using the stabilized N2-purged spectrum as the background reference, collecting single-beam spectra at 4 cm−1 resolution with 256 co-added scans per measurement. Absorbance spectra were recorded at t = 0, 15, 30, 45, 60, 90, and 120 min under continuous reaction/irradiation to track surface intermediate evolution. Details of the photothermal catalytic CO2 reduction Photothermal catalytic CO2 reduction was performed in a custom-designed stainless-steel reactor (100 mL) equipped with a quartz window and Fresnel lens optics (30 cm focal length), integrated with a solar simulator (PLS-SXE 300/300UXE, PerfectLight) providing calibrated AM 1.5 G irradiation (5000 mW·cm−2). Prior to the reaction, the system was purged with N2 (99.999%, 50 mL·min−1, 10 min) followed by charging with CO2/N2 mixture (1 vol%, Praxair). Water vapor was introduced by injecting deionized water quantitatively (1 mL, 18.2 MΩ·cm resistivity) via a gas-tight syringe. During the reaction, the focused light spot generated simultaneous photoexcitation and localized thermal effects on the catalyst surface, enabling photothermal synergistic catalysis. Gaseous products were analyzed online and quantified using gas chromatograph (GC-8860) with thermal conductivity detector (TCD) and flame ionization detector (FID), employing a Porapak Q packed column. All gaseous components were quantified via multi-point calibration using NIST-traceable standards (calibration curves exhibited linear correlation coefficients R2 > 0.999). The catalyst was continuously irradiated for 2 h with periodic sampling. Dark controls and additional control experiments were conducted concurrently to rigorously exclude background interference and verify result reliability. Photocatalytic H2O2 production After photo-thermal catalytic CO2 reduction, 30 mL high-purity deionized water (18.2 MΩ·cm) was injected through the reactor’s injection port. A low-temperature circulator (5.0 ± 0.1 °C) minimized thermal degradation of H2O2. Components adhering to reactor walls were fully dissolved in the aqueous phase by vortex mixing (200 rpm, 3 min), ensuring quantitative product recovery. H2O2 concentration was then quantified via standardized iodometric titration. Measurement of H2O2 concentration H2O2 concentration was determined using a standardized iodometric method39. Briefly, 1 mL 0.1 M potassium hydrogen phthalate (C8H5KO4) aqueous solution was added to the sample, followed by 1 mL of 0.4 M potassium iodide (KI) aqueous solution. After a 30 min reaction period, H2O2 reacted with iodide anions (I−) under acidic conditions according to the reaction: H2O2 + 3I− + 2H+ → I3− + 2H2O, producing triiodide anions (I3−) that exhibit strong absorption at 350 nm. I3− concentration was quantified using UV-Vis spectroscopy, allowing calculation of the H2O2 concentration generated. Details of the thermocatalytic CO2 reduction The performance of the as-prepared catalyst was evaluated in a fixed-bed stainless steel reactor. The temperature error at the catalyst location was controlled within ±5 °C. The catalytic evaluation was performed in the reaction temperature range of 200–600 °C under continuous CO2 and H2O gas flow. Target pressure was achieved through regulation of reaction gas influx. All gas flows were managed by precision-calibrated mass flow controllers. Product analysis was performed via online gas chromatography using a thermal conductivity detector-equipped system with a packed column40. Post steady-state reaction, CO2 conversion and product selectivity were calculated via an area normalization method. Reaction kinetics were investigated under identical conditions with varying temperatures. Solar-to-chemical energy conversion efficiency The solar-to-chemical energy conversion efficiency (STC) could be expressed as the ratio of output to input energy, calculated by ref.41:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\eta }_{{\mbox{STC}}}=\frac{{\mbox{qm}}}{{\mbox{IAt}}}\times 100\%\left(\frac{{\mbox{J}}\cdot {\mbox{mo}}{{\mbox{l}}}^{-1}\cdot {\mbox{mol}}}{{\mbox{W}}\cdot {\mbox{c}}{{\mbox{m}}}^{-2}\cdot {\mbox{c}}{{\mbox{m}}}^{2}\cdot {\mbox{s}}}=\frac{{\mbox{J}}}{{\mbox{W}}\cdot {\mbox{s}}\,}=1\right)$$\end{document}ηSTC=qmIAt×100%J⋅mol−1⋅molW⋅cm−2⋅cm2⋅s=JW⋅s=1Where q is the combustion heat value of CO (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q}_{{CO}}=283\times {10}^{3}{J}\cdot {{mol}}^{-1}$$\end{document}qCO=283×103J⋅mol−1), CH4 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q}_{C{H}_{4}}=890.31\times {10}^{3}{J}\cdot {{mol}}^{-1}$$\end{document}qCH4=890.31×103J⋅mol−1) and C2H4 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${q}_{C_{2}{H}_{4}}=1411\times {10}^{3}{J}\cdot {{mol}}^{-1}$$\end{document}qC2H4=1411×103J⋅mol−1), m denote the mole of produced CO, CH4 and C2H4 (mol), I denote the light intensity (W·cm−2), A is the irradiation area (cm2), and t denote the reaction time (s). Thus, the solar-to-chemical energy conversion efficiency for the Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst was calculated to be 0.50%. The 200 mg Ru0.6Cu1(Cu0–Cu1+)/CeO2 catalyst was tested in a concentrated solar (light) reactor for 2 h. 30 μmol CO, 8 μmol CH4 and 109.8 μmol C2H4 were produced:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\eta }_{{\mbox{STC}}}=\frac{283\,\times {\,10}^{3}\,\times \,30\,\times \,{10}^{-6}+890.31\,\times {\,10}^{3}\,\times \,8\,\times {\,10}^{-6}+1411\times {\,10}^{3}\,\times \,109.8\, \times {\,10}^{-6}}{5000\, \times \,{10}^{-3}\, \times \,1 \, \times \,7200} \\ \times 100{\%}=0.5{\%}$$\end{document}ηSTC=283×103×30×10−6+890.31×103×8×10−6+1411×103×109.8×10−65000×10−3×1×7200×100%=0.5% Product selectivity calculation The product selectivity for CO2 conversion to CO, CH4 and C2H4 were calculated using the following equation42:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Product}}\; {{selectivity}}\; {{of}}\,{C}_{2}{H}_{4}(\%)=[{{n}}({C}_{2}{H}_{4})]/[n(C_{2}{H}_{4})+n({{CH}}_{4})+n({CO})] \times 100\%$$\end{document}ProductselectivityofC2H4(%)=[n(C2H4)]/[n(C2H4)+n(CH4)+n(CO)]×100%3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Product}}\; {{selectivity}}\; {{of}}\,{{CH}}_{4}(\%)=[{{n}}({{CH}}_{4})]/[{{n}}({C}_{2}{H}_{4})+n({CH}_{4})+n({CO})]\times 100\%$$\end{document}ProductselectivityofCH4(%)=[n(CH4)]/[n(C2H4)+n(CH4)+n(CO)]×100%4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Product}}\; {{selectivity}}\; {{of}}{CO}(\%)\,=[{{n}}({CO})]/[{{n}}({C}_{2}{H}_{4})+n({{CH}}_{4})+n({CO})]\times 100\%$$\end{document}ProductselectivityofCO(%)=[n(CO)]/[n(C2H4)+n(CH4)+n(CO)]×100%where n (C2H4), n (CH4) and n (CO) are the amounts of produced C2H4, CH4 and CO. Electron selectivity calculation Electron selectivity was calculated based on the consumption of 2, 2, 8 and 12 electrons per H2, CO, CH4 and C2H4 molecule, respectively, according to the following formulas43:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Sel}}_{{electron}}\left({H}_{2}\right)=\left(\frac{2\times n\left({H}_{2}\right)}{2\times n\left({H}_{2}\right)+2\times n\left({CO}\right)+8\times n\left(C{H}_{4}\right)+\,12\times n\left({C}_{2}{H}_{4}\right)}\right)\times 100\%$$\end{document}SelelectronH2=2×nH22×nH2+2×nCO+8×nCH4+12×nC2H4×100%6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Sel}}_{{electron}}\left({CO}\right)=\left(\frac{2\times n\left({CO}\right)}{2\times n\left({H}_{2}\right)+2\times n\left({CO}\right)+8\times n\left(C{H}_{4}\right)+\,12\times n\left({C}_{2}{H}_{4}\right)}\right)\times 100\%$$\end{document}SelelectronCO=2×nCO2×nH2+2×nCO+8×nCH4+12×nC2H4×100%7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Sel}}_{{electron}}\left(C{H}_{4}\right)=\left(\frac{8\times n\left(C{H}_{4}\right)}{2\times n\left({H}_{2}\right)+2\times n\left({CO}\right)+8\times n\left(C{H}_{4}\right)+\,12\times n\left({C}_{2}{H}_{4}\right)}\right)\times 100\%$$\end{document}SelelectronCH4=8×nCH42×nH2+2×nCO+8×nCH4+12×nC2H4×100%8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{Sel}}_{{electron}}\left({C}_{2}{H}_{4}\right)=\left(\frac{12\times n\left({C}_{2}{H}_{4}\right)}{2\times n\left({H}_{2}\right)+2\times n\left({CO}\right)+8\times n\left(C{H}_{4}\right)+\,12\times n\left({C}_{2}{H}_{4}\right)}\right)\times 100\%$$\end{document}SelelectronC2H4=12×nC2H42×nH2+2×nCO+8×nCH4+12×nC2H4×100%Where n represents the molar yield of each product. DFT-based computational framework for surface adsorption energetics and reaction thermodynamics All density functional theory computations were performed using the Vienna Ab Initio Simulation Package (VASP)44–46. Two-dimensional slab surfaces were modeled with periodic boundary conditions, incorporating a 20 Å vacuum layer to minimize inter-slab interactions. Sampling of the Brillouin zone was conducted using a Monkhorst-Pack k-point mesh of 3 × 3 × 147. The electronic structure calculations employed projector-augmented-wave (PAW) pseudopotentials combined with Perdew-Burke-Ernzerhof (PBE) exchange-correlation functionals48–50. Plane-wave basis sets with a kinetic energy cutoff of 500 eV described the electronic wavefunctions. Self-consistency convergence was achieved when successive iteration energies differed by less than 10−5 eV. Structural relaxations applied the conjugate-gradient technique51, continuing until all Hellmann-Feynman force components were below 10−2 eV/Å. These DFT-derived conclusions require cautious interpretation given inherent challenges in modeling the exact catalyst structure and reactive environment under operando conditions; fundamental limitations arise from approximate exchange-correlation functionals inherent to DFT, which fail to fully capture complex electronic behaviors. Gibbs free energy change (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G$$\end{document}G) of each chemical reaction step was calculated by:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$G={E}_{{DFT}}+{E}_{{ZPE}}-{TS}$$\end{document}G=EDFT+EZPE−TSwhere \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{{DFT}}$$\end{document}EDFT, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{{ZPE}}$$\end{document}EZPE, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$T$$\end{document}T, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$S$$\end{document}S denote total energy obtained from DFT, zero-point energy, temperature, and entropy, respectively. The temperature was set as room temperature, i.e. 298.15 K. The zero-point energy and entropy were calculated in VASPKIT code50. Adsorption energy \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {E}_{{ad}}$$\end{document}ΔEad on the surface of substrates was defined as:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\Delta {E}_{{ad}}={E}_{*{mole}}-({E}_{\!*}+{E}_{{mole}})$$\end{document}ΔEad=E*mole−(E*+Emole)where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{*{mole}}$$\end{document}E*mole, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{*}$$\end{document}E*, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{{mole}}$$\end{document}Emole denote energies of the adsorbed molecule on substrates, the bare substrates, and the molecule. The formation energy (Ef) is defined as11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{f}=E\left({A}_{x}{B}_{y}\right)-x{E}_{A}-y{E}_{B}$$\end{document}Ef=EAxBy−xEA−yEBwhere \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E({A}_{x}{B}_{y})$$\end{document}E(AxBy) refers to the energy of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{x}{B}_{y}$$\end{document}AxBy clusters. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{A}$$\end{document}EA and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${E}_{B}$$\end{document}EB refer the energy per atom of A bulk and B bulk, respectively. The smaller the formation energy, the easier it is to form, and the more stable the structure. Supplementary information Supplementary Infomation Description of Additional Supplementary File Supplementary Data 1 Transparent Peer Review file Supplementary Infomation Description of Additional Supplementary File Supplementary Data 1 Transparent Peer Review file Source data Source data Source data