English

• With atmospheric CO₂ concentrations in the atmosphere reaching unacceptably high levels and continuing to rise, capturing atmospheric CO₂ and converting it into value-added fuels has become both essential and sustainable [1–3]. The electrochemical CO₂ reduction reaction (CO₂RR) stands out as an effective and promising strategy for transforming CO₂ into high-value chemicals and liquid fuels [4]. Among the various products of electrochemical CO₂RR, CO serves as a fundamental feedstock for methanol synthesis, Fischer-Tropsch synthetic oils, and diverse carbonylation reactions [5]. However, this conversion process involves two proton-coupled electron transfer steps, posing challenges such as identifying active sites and competing with the hydrogen evolution reaction (HER) [6–8]. In recent years, single-atom catalysts (SACs) have attracted significant attention due to their well-defined active sites and excellent selectivity for C₁ products [6,9–11].

  Carbon-based SACs play a pivotal role in electrocatalysis due to their versatile structures, exceptional stability, and high conductivity [12–14]. Their high activity in CO₂RR systems is typically associated with symmetric M-N₄ sites, such as Cu-N₄ [15], Ni-N₄ [16], or Co-N₄ [17]; yet, challenges regarding activity, stability, and product selectivity persist. For instance, excessively strong ^*CO adsorption energy or a high ^*COOH energy barrier can result in unsatisfactory selectivity and stability [18,19]. To address these issues, dual-atom catalysts (DACs) with diverse metal combinations have been developed, leveraging interactions between the two metals to effectively lower reaction barriers, accelerate reaction rates, and thus achieve superior catalytic performance [20–22]. However, to the best of our knowledge, recent research has predominantly focused on comparing DACs with SACs, with limited studies investigating how intermetallic interactions influence CO₂RR performance in DACs systems [23,24].

  This study explores the impact of intermetallic interactions on the CO₂RR performance of Cu/Ni-NC DACs. Theoretical simulations were conducted on Cu/Ni-NC DACs models to assess how intermetallic interactions affect electronic structure, CO₂RR activity, and selectivity, ultimately identifying the reason for the selectivity shift in Cu-N-C catalysts after Ni doping. These findings highlight that intermetallic interactions are a critical and reliable parameter influencing catalytic activity in DACs systems. Aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC—HAADF-STEM) and X-ray absorption fine structure (XAFS) analyses further confirmed that the synthesized Cu/Ni-N-C catalyst features an atomic configuration with neighboring Cu/Ni sites coordinated with four nitrogen atoms, respectively. The resulting catalyst efficiently facilitates the selective conversion of CO₂ to CO, achieving nearly 100% CO Faradaic efficiency (FE_CO) at −1.3 V vs. RHE in an H-type cell. Density functional theory (DFT) calculations reveal that the Cu/Ni dual-atom catalyst synergistically electroreduces CO₂ to CO: The Cu atom, with low charge density, adsorbs the oxygen atom of the CO₂ molecule; while the Ni atom, with high charge density, adsorbs the carbon atom of the CO₂ molecule, collectively enhancing overall CO₂RR activity. This study underscores the importance of tuning intermetallic interactions in DACs to maximize synergistic effects and optimize CO₂RR performance.

  The fabrication of Cu-Ni dual-atom catalysts is schematically illustrated in Fig. 1. 1,3,5-Benzenetricarboxylic acid, copper(Ⅱ) nitrate trihydrate, and PVP were used as starting materials to synthesize HKUST-1 metal-organic framework (MOF) via a hydrothermal reaction (see Experimental Section in Supporting information). When copper(Ⅱ) nitrate trihydrate was partially substituted with nickel nitrate hexahydrate, with all other experimental procedures kept unchanged, (CuNi)HKUST-1 MOFs with different Cu/Ni ratios were successfully obtained. The crystal phase and morphology of the as-synthesized samples were characterized by X-ray diffraction (XRD, Fig. S1 in Supporting information) and scanning electron microscopy (SEM, Figs. 2a-c and Fig. S2 in Supporting information). Results confirmed the successful synthesis of HKUST-1 MOF and showed no changes in the structure or morphology of HKUST-1 after partial substitution of Cu atoms with Ni atoms. Finally, HKUST-1 and (Cu_xNi_y)HKUST-1 MOFs were mixed with dicyandiamide and then calcined; the resulting products were designated as Cu_xNi_y-N-C-BL (where x:y = 1:0, 0.8:0.2, 0.5:0.5 and 0:1; BL means before acid leaching). After acid leaching under an oxygen atmosphere, metal particles in the Cu_xNi_y-N-C-BL samples were removed, yielding target products of Cu_xNi_y-N-C. For comparison, nitrogen-doped carbon (N—C) was also synthesized without adding any metal components.

  Figure 1

  

  Figure 1.  Schematic of the procedure to synthesize Cu-N-C and Cu_0.5Ni_0.5—N-C catalysts. Orange sphere: Cu atom, purple sphere: Ni atom, blue sphere: N atom, black sphere: C atom.The XRD patterns of all Cu_xNi_y-N-C-BL samples exhibit sharp diffraction peaks at 2θ = 43.3°, 50.4°, and 74.1° (Fig. 2d, Figs. S3 and S4 in Supporting information), corresponding to the (111), (200), and (220) crystal planes of the cubic Cu structure (JCPDS No. 04–0836). After acid leaching, the resulting Cu_xNi_y-N-C samples show only a broad peak centered at 26.4°, which is attributed to the (002) plane of the hexagonal graphitic structure (Fig. 2d and Figs. S3-S5 in Supporting information) and confirms the complete removal of metallic Cu via acid leaching. This is because metal particles had weak interactions with the carbon matrix and were thus dissolved in the oxygen-saturated acid solution. For comparison, the XRD pattern of the nitrogen-doped carbon (N—C) sample was also examined (Fig. S6 in Supporting information), where no Cu diffraction peaks are observed and only the carbon-related peak is identifiable.

  DownLoad: Full-Size Img PowerPoint

  Figure 2

  

  Figure 2.  (a) High-magnification SEM image of HKUST-1. (b) Low-magnification SEM image of (Cu_0.5Ni_0.5) HKUST-1 MOF. (c) High-magnification SEM image of (Cu_0.5Ni_0.5) HKUST-1 MOF. (d) XRD patterns of Cu_0.5Ni_0.5—N-C-BL and Cu_0.5Ni_0.5—N-C samples. (e) SEM image of Cu-N-C. (f) Low-magnification TEM image of Cu-N-C. (g) High-magnification TEM image of Cu-N-C, with the inset showing the corresponding selected area electron diffraction pattern. (h) EDS elemental distribution map of Cu-N-C. (i) SEM image of Cu_0.5Ni_0.5—N-C. (j) Low-magnification TEM image of Cu_0.5Ni_0.5—N-C. (k) High-magnification TEM image of Cu_0.5Ni_0.5—N-C, with the inset showing the corresponding selected area electron diffraction pattern. (l) EDS analysis of Cu_0.5Ni_0.5—N-C. (m) EDS elemental distribution map of Cu_0.5Ni_0.5—N-C. (n) AC—HAADF-STEM image of Cu_0.5Ni_0.5—N-C (dual-atom sites (circled in red) and single-atom sites (circled in yellow)). (o) 3D surface plot from region 1 of (n). (p) Statistical distribution of single sites and dual sites in Cu_0.5Ni_0.5—N-C. (q) Intensity profiles obtained from areas 2–3 in (n).

  DownLoad: Full-Size Img PowerPoint

  Scanning electron microscopy (SEM) images show that all Cu_xNi_y-N-C-BL samples consist of interconnected foam structures with wrinkled, entangled flakes (Fig. S7 in Supporting information), which is similar to the Cu-N-C sample (Fig. 2e). The low-resolution transmission electron microscopy (TEM) image of the Cu-N-C catalyst reveals cross-linked, ultrathin folded-sheet features (Fig. 2f) and the corresponding high-resolution TEM (HRTEM) image confirms no observable copper particles in the Cu-N-C catalyst (Fig. 2g). The selected-area electron diffraction (SAED) pattern (inset of Fig. 2g) further confirms the sample’s non-crystalline nature. Energy-dispersive spectroscopy (EDS) elemental mapping shows the existence of Cu, N, and C elements and the uniform distribution of N and Cu across the entire carbon matrix (Fig. 2h). As presented in Figs. 2i-k and Figs. S8-S11 (Supporting information), the Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C and Ni-N-C samples display a morphology analogous to that of the Cu-N-C sample, with no distinct metal particles observed on their surfaces. EDS elemental mapping further verifies the homogeneous distribution of N, Cu, and Ni across the carbon support (Figs. 2l-m and Figs. S9-S11). As illustrated in the AC—HAADF-STEM image (Fig. 2n), numerous bright spot pairs are distinctly observed, which are indicative of potential atomic-scale active sites. Furthermore, the 3D surface plot derived from region 1 of Fig. 2n offers compelling evidence to corroborate the existence and structural features of these atomic pairs (Fig. 2o). In addition, statistical analysis presented in Fig. 2p clearly demonstrates that over 88.73% of the detected sites correspond to isolated dual-atom configurations (circled with red dashed lines) with a distance around 2.81–2.92 Å (Fig. 2q). The electron energy-loss spectroscopy (EELS) diagram in Fig. S12 (Supporting information) further confirmed the co-existence of Ni and Cu atomic pairs. The small interatomic distance between Cu and Ni induces intense through-space and through-bond electronic coupling interactions, effectively expediting the migration of electrons between Cu and Ni. Collectively, these characterizations confirm that Cu/Ni dual-atom sites are not only the dominant active species but also uniformly distributed throughout the Cu_0.5Ni_0.5—N-C samples, laying a solid foundation for their enhanced catalytic performance. The Raman spectra of Cu-N-C, Cu_0.8Ni_0.2—N-C, and Cu_0.5Ni_0.5—N-C samples exhibit two prominent characteristic peaks at approximately 1355 and 1570 cm^-1 (Fig. S13 in Supporting information), which are ascribed to the d-band and G-band, respectively [25,26]. Defect concentrations were quantified by normalizing the intensity of the d-band relative to that of the G-band (i.e., the I_D/I_G ratio, Table S1 in Supporting information) [25]. Notably, the I_D/I_G values were calculated as 1.09, 1.08 and 1.07 for Cu-N-C, Cu_0.5Ni_0.5—N-C and Cu_0.8Ni_0.2—N-C samples, respectively, which are comparable and indicate that the defect content in these Cu_xNi_y-N-C samples is not the critical factor governing their catalytic performance. N₂ adsorption-desorption isotherms and corresponding pore size distribution analyses (Fig. S14 in Supporting information) confirm that all samples share comparable Brunauer-Emmett-Teller (BET) specific surface areas and total pore volumes. For example, the Cu_0.5Ni_0.5—N-C sample achieves a BET specific surface area of 137.71 m²/g and a total pore volume of 0.62 cm³/g, which is similar to that of the Cu-N-C and Ni-N-C sample. Such favorable textural characteristics are capable of optimizing the mass transfer behavior of CO₂ and electrolyte at the triple-phase interface, thereby creating favorable conditions for the smooth progression of the CO₂ electroreduction reaction.

  X-ray photoelectron spectroscopy (XPS) measurements were conducted to systematically investigate the chemical composition and elemental valence states of the Cu_xNi_y-N-C and Cu-N-C samples. As revealed by the XPS survey spectra (Fig. S15 in Supporting information), Cu_xNi_y-N-C contained four main elements, i.e., C, N, Cu and Ni, confirming the successful incorporation of N, Cu and Ni into the carbon sheet matrix. Notably, the N dopant is inherently derived from the dicyandiamide precursor. As shown in Fig. S16 (Supporting information), the high-resolution C 1s spectrum of the Cu_0.5Ni_0.5—N-C sample deconvolutes into four major peaks centered at 284.6, 286, 287.3 and 289.1 eV, which correspond to sp²-bonded C=C, sp³-bonded C—N, C—O, and C=O functional groups, respectively [27]. Fig. 3a presents the high-resolution N 1s XPS spectrum of Cu_0.5Ni_0.5—N-C, revealing five distinct sub-peaks at 398.7, 399.2, 400.1, 401.1 and 403.1 eV. These peaks are assigned to pyridinic N, Cu/Ni-N_x, pyrrolic N, graphitic N, and oxidized N, respectively. The presence of the Cu/Ni-N_x peak in the deconvoluted N 1s spectrum provides direct evidence for the formation of Cu/Ni-N_x coordination groups in the sample [27–29]. For the Cu 2p region (Fig. 3b), the peaks at 932.50 and 934.8 eV are attributed to the 2p_3/2 orbitals of Cu(Ⅰ) and Cu(Ⅱ) species, while those at 952.2 and 954.7 eV correspond to the 2p_1/2 orbitals of Cu(Ⅰ) and Cu(Ⅱ), respectively. Additional satellite peaks at 942.8 and 944.9 eV further confirm the presence of Cu oxide phases in the carbon matrix, which is further confirmed by the Cu Auger spectra in Fig. S17 (Supporting information) [27,30]. In the high-resolution Ni 2p spectrum of Cu_0.5Ni_0.5—N-C (Fig. 3c), the peaks at 855.29 and 872.78 eV are assigned to Ni 2p_3/2 and Ni 2p_1/2, respectively. The absence of the metallic Ni 2p_3/2 peak (at 852.6 eV) indicates that the dispersed Ni species observed in HAADF-STEM images exist as Ni⁺ or Ni²⁺ cations [31]. For comparison, XPS measurements were conducted on Cu-N-C, Ni-N-C and Cu_0.8Ni_0.2—N-C under the same conditions (Figs. S18-S20 in Supporting information). The results indicated that the chemical states of the major elements are analogous to those of Cu_0.5Ni_0.5—N-C. Notably, the divergent orbital hybridization behaviors and electron-donating properties of Cu and Ni in their N-coordination interactions trigger an asymmetric charge redistribution in the Cu-Ni diatomic pair. As observed, the binding energy of Cu 2p_3/2 in Cu-N-C is 932.35 eV, and that of Ni 2p_3/2 in Ni-N-C is 855.47 eV. While the Cu 2p_3/2 binding energy in Cu_0.5Ni_0.5—N-C shifts to 932.50 eV (an increase of 0.15 eV), and the Ni 2p_3/2 binding energy shifts to 855.29 eV (a decrease of 0.18 eV) (Fig. S21 in Supporting information). These shifts directly confirm the proposed electron transfer from Cu to Ni in the dual-atom catalyst.

  Figure 3

  

  Figure 3.  High-resolution (a) N 1s, (b) Cu 2p, and (c) Ni 2p XPS spectra of Cu_0.5Ni_0.5—N-C. (d) Cu K-edge and (e) Ni K-edge XANES spectra of Cu_0.5Ni_0.5—N-C. FT-EXAFS spectra of (f) Cu K-edge and (g) Ni K-edge of Cu_0.5Ni_0.5—N-C. (h) EXAFS fitting of Cu_0.5Ni_0.5—N-C in the R space at the Cu K-edge and (i) Ni K-edge of Cu_0.5Ni_0.5—N-C. (j) Wavelet transform images of EXAFS data at Cu K-edge with the optimized Morlet parameter for CuPc (left) and Cu_0.5Ni_0.5—N-C (right). (k) Wavelet transform images of EXAFS data at Ni K-edge with the optimized Morlet parameter for NiPc (left) and Cu_0.5Ni_0.5—N-C (right).

  DownLoad: Full-Size Img PowerPoint

  X-ray absorption near-edge spectroscopy (XANES) and Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) were employed to characterize the structural and electronic properties of Cu and Ni sites in the Cu_0.5Ni_0.5—N-C sample. As shown in Fig. 3d, the Cu K-edge XANES spectrum of the Cu_0.5Ni_0.5—N-C catalyst places the Cu absorption peak between those of CuPc and Cu foil, confirming the oxidized state of Cu. Similarly, Fig. 3e displays the Ni K-edge XANES spectrum of the Cu_0.5Ni_0.5—N-C sample, where the Ni absorption peak is positioned between those of NiO/NiPc and a Ni foil, indicating that Ni is present in an oxidized state. For the FT-EXAFS analysis, Fig. 3f displays the Cu K-edge FT-EXAFS spectrum of the Cu_0.5Ni_0.5—N-C sample, which is characterized by a dominant peak at 1.54 Å and can be assigned to Cu-N coordination environments. Notably, a small peak positioned at 2.12 Å was observed, which is due to the weak interaction between Cu and adjacent Ni atoms. Correspondingly, the Ni K-edge FT-EXAFS spectrum (Fig. 3g) shows a major peak at 1.56 Å for the sample, associated with Ni-N bonding. To determine the detailed structural configuration of Cu and Ni in Cu_0.5Ni_0.5—N-C sample, the FT-EXAFS spectra were further fitted (Figs. 3h and i, Figs. S22 and S23 in Supporting information). The fitting results (Tables S2 and S3 in Supporting information) reveal that the coordination numbers (CN) of Cu-N and Ni-N in the catalyst are 4.03 and 4, respectively. Wavelet transforms (WT) were used to validate the above conclusions, as they provide resolution in both R and k space. The WT contour plots of the Cu_0.5Ni_0.5—N-C catalyst exhibit a strong signal at 2.68 Å^-1, which corresponds to Cu-N contributions and is consistent with the signal of CuPc (Fig. 3j and Fig. S24 in Supporting information). A comparable finding was obtained from the WT-EXAFS analysis of the Ni K-edge (Fig. 3k and Fig. S25 in Supporting information).

  To evaluate the performance of Cu_xNi_y-N-C catalysts, each catalyst powder was drop-coated onto a carbon paper with a Nafion binder to fabricate the working electrode. CO₂ reduction reaction tests were conducted in a three-electrode H-cell configuration using 0.1 mol/L KHCO₃ aqueous electrolyte saturated with CO₂. Linear sweep voltammetry (LSV) curves were recorded for all samples in both Ar- and CO₂-saturated 0.1 mol/L KHCO₃ solutions. As shown in Fig. 4a, the current densities were notably higher under CO₂ atmosphere compared to Ar, indicating their strong catalytic activities toward CO₂RR. The potential-dependent gas product distribution of all samples was measured between −1 V and −1.5 V vs. RHE, with in-line gas chromatography (GC) used for gas product quantification. For the Cu-N-C sample, CO₂RR yielded main products including CO and CH₄ with maximum Faradaic efficiency (FE) of CO 38.35% at −1 V vs. RHE and maximum FE_CH4 21.89% at −1.4 V vs. RHE (Fig. 4b). In contrast, Cu_0.8Ni_0.2—N-C and Cu_0.5Ni_0.5—N-C exhibited distinct catalytic selectivity. The Cu_0.8Ni_0.2—N-C catalyst achieved a maximum FE_CO of 66.69% at −1.3 V vs. RHE, corresponding to a partial current density of 10.06 mA/cm² (Fig. S26 in Supporting information).

  Figure 4

  

  Figure 4.  (a) LSV curves of Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C in 0.1 mol/L KHCO₃ saturated with Ar or CO₂. Faradaic efficiency (FE) of all CO₂RR products at different potentials of (b) Cu-N-C, (c) Cu_0.5Ni_0.5—N-C, (d) N—C. (e) Tafel plots for CO₂RR over Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C. (f) Nyquist curves of Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C. (g) CV fitting curves of Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C. (h) Stability assessment of Cu_0.5Ni_0.5—N-C at −1.3 V vs. RHE. High-resolution (i) N 1s, (j) Cu 2p, and (k) Ni 2p XPS of Cu_0.5Ni_0.5—N-C catalyst after CO₂RR.

  DownLoad: Full-Size Img PowerPoint

  Interestingly, increasing the Ni content enhanced CO generation, making it the dominant product. For Cu_0.5Ni_0.5—N-C, the maximum FE_CO reached 96.78% with a partial current density of 24.80 mA/cm² at −1.3 V vs. RHE and negligible CH₄ and liquid products were detected (Fig. 4c, Figs. S27 and S28 in Supporting information), exhibiting a different product selectivity from that of Cu-N-C catalyst. Notably, the optimal Cu_0.5Ni_0.5—N-C dual-atom catalyst displays a superior ability for CO formation relative to the Ni-N-C single-atom catalyst, where the latter yields a maximum FE_CO of 86% at −1.3 V vs. RHE with a partial current density of 20.59 mA/cm² (Fig. S29 in Supporting information). The distinct shift in selectivity arises from the formation of Cu-Ni atomic pairs after Ni doping into the Cu-N-C catalyst. As Ni has a higher electronegativity than copper, the electron cloud density of Cu atoms is reduced, which weakens the adsorption of intermediate species during the CO₂ reduction reaction. This effect hinders the subsequent protonation process of the CO₂RR intermediates (e.g., ^*CO), thereby suppressing CH₄ formation and elevating CO selectivity. The reduced electron cloud density around Cu atoms in the Cu/Ni-N-C catalyst, in comparison with the Cu-N-C catalyst, is further verified by the calculations of charge density difference and electron localization function (Figs. 5a-c, Figs. S36 and S37 in Supporting information). For comparison, the N—C catalyst was tested and showed negligible CO₂RR activity. It exhibited nearly full Faradaic efficiency toward HER across various applied potentials (Fig. 4d). These results confirm that Ni and Cu species serve as the main active sites for CO₂RR in Cu_xNi_y-N-C catalysts. To gain deeper insight into reaction kinetics, Tafel analysis was performed as presented in Fig. 4e and the potential window adopted for Tafel fitting ranges from −0.75 V to −1.3 V vs. RHE. This range was chosen because FE_CO stays above 90% to guarantee CO₂RR dominance, current density increases linearly with overpotential (a signature of kinetic control free of mass transfer constraints), and negligible side reactions like hydrogen evolution take place. The Cu_0.5Ni_0.5—N-C catalyst had a Tafel slope of 67.43 mV/dec, lower than those of Ni-N-C (77.57 mV/dec), Cu-N-C (122.41 mV/dec), and Cu_0.8Ni_0.2—N-C (97.60 mV/dec). This indicates that Cu_0.5Ni_0.5—N-C effectively promotes the first electron transfer to form the adsorbed ^*COOH intermediate, accelerating overall CO₂RR dynamics. Electrochemical impedance spectroscopy (EIS) measurements further validated the reaction kinetics. As shown in Fig. 4f, the R_ct values of N—C, Cu-N-C, Ni-N-C, Cu_0.8Ni_0.2—N-C, and Cu_0.5Ni_0.5—N-C are approximately 3055, 368.5, 265.7, 279.5 and 167.7 Ω, respectively, and the Cu_0.5Ni_0.5—N-C displays the smallest semicircle, which means faster electron transfer kinetics compared to the other catalysts, demonstrating efficient interfacial charge transfer.

  Figure 5

  

  Figure 5.  (a, b) The charge density difference of Cu/Ni-N-C and Cu-N-C. The yellow and cyan region indicates electron accumulation and depletion, respectively. (c) ELF analysis for Cu/Ni-N-C with color indicating where red and blue indicate electron localization and delocalization regions, respectively. (d) Proposed CO₂ configuration adsorbed on the Cu/Ni-N-C due to the electron distribution differences between the localized positive charge on the Cu sites and negative Ni sites. (e) Projected density of states demonstrating the Cu d-states of Cu/Ni-N-C and Cu-N-C. The Fermi energy is illustrated by a black dashed line, while d-band centers (ε_d) for Cu-N-C and Cu/Ni-N-C are marked by orange and purple dashed lines, respectively. Adsorption energy diagram of (f) H₂O dissociation and (g) hydrogen ad-desorption for Cu-N-C, Ni-N-C and Cu/Ni-N-C. (h) Gibbs free energy diagrams for the formation of CO on Cu-N-C, Ni-N-C and Cu/Ni-N-C. (i) Structural comparison of COOH^* adsorbed on Cu-N-C and Cu/Ni-N-C, showing atomic configurations and bond lengths. Color codes: Cu (orange), Ni (purple), N (blue), C (gray), O (red), H (white).

  DownLoad: Full-Size Img PowerPoint

  The electrochemically active surface area (ECSA), a key factor influencing intrinsic catalytic activity, is proportional to the double-layer capacitance (C_dl). Cyclic voltammetry (CV) measurements were used to determine C_dl values for N—C, Cu-N-C, Ni-N-C, Cu_0.8Ni_0.2—N-C, and Cu_0.5Ni_0.5—N-C (Fig. S30). The calculated C_dl values were 2.11 mF/cm² (N—C), 6.57 mF/cm² (Cu-N-C), 12.07 mF/cm² (Ni-N-C), 9.94 mF/cm² (Cu_0.8Ni_0.2—N-C), and 13.63 mF/cm² (Cu_0.5Ni_0.5—N-C) (Fig. 4g). These results further support the significantly enhanced CO₂RR activity of Cu_0.5Ni_0.5—N-C. Furthermore, the superior intrinsic activity of Cu_0.5Ni_0.5—N-C is further highlighted by calculating the turnover frequency (TOF). The developed Cu_0.5Ni_0.5—N-C exhibits an exceptionally high TOF of 15,372 h^-1 at −1.3 V vs. RHE, outperforming most of the reported state-of-the-art single atom catalysts evaluated under similar conditions (Table S4 in Supporting information). Cu_0.5Ni_0.5—N-C catalyst shows excellent durability with negligible decrease in current density, and FE_CO approximately remains about 90% after continuous electrocatalysis for 10 h (Fig. 4h). Post-stability characterizations via XRD, SEM, AC—HAADF-STEM image and XPS (Figs. 4i-k and Figs. S31-S35 in Supporting information) revealed no significant changes in the catalyst’s phase structure, morphology, or elemental composition, confirming its good stability. This outstanding durability is due to the strong coupling interaction of N-coordinated dual atomic sites in the carbon support. This interaction reduces ^*CO poisoning and accelerates CO desorption from active sites, thereby enhancing stability during CO₂RR.

  To understand the origin of the enhancement in the catalytic activity of Cu/Ni-N-C catalyst, DFT calculations were conducted to analyze its electronic properties. Charge density difference (CDD) analysis reveals pronounced electron accumulation (yellow region) on Ni atoms and electron depletion (cyan region) on Cu atoms. This directly visualizes the electron density bias toward Ni and thus indicates a stronger interaction between Ni atoms and the intermediates of the electrochemical CO₂ reduction reaction (Fig. 5a). More electron accumulation around Ni active centers than around Cu atoms is also observed in both Cu-N-C and Ni-N-C configurations (Fig. 5b and Fig. S36 in Supporting information), further verifying that the catalytic activity of Ni-N sites is higher than that of Cu-N sites. This observation is further corroborated by electron localization function (ELF) analysis, which reveals localized electron density between the Cu-N and Ni-N moieties and demonstrates that Ni-N sites exhibit higher electron density than Cu-N sites (Fig. 5c and Fig. S37 in Supporting information). This electron distribution is likely to have a substantial influence on the adsorption behavior of the intermediates and the reaction selectivity. When CO₂ is polarized, its lowest unoccupied molecular orbital is concentrated mainly on the carbon atom, making it vulnerable to nucleophilic attacks. Nevertheless, the presence of regions with localized positive charge on the catalyst surface can impede nucleophilic attacks on the carbon atom. This, in turn, raises the thermodynamic barrier for ^*COOH adsorption. As a result, for the Cu/Ni-N-C catalyst, the carbon atom of polarized CO₂ molecules preferentially adsorbs onto the electron-enriched Ni active sites, while the oxygen atoms of CO₂ tend to bind more strongly to the electron-deficient Cu active sites (Fig. 5d). This unique adsorption mode is a direct consequence of the electron density bias toward Ni and serves as theoretical evidence for the charge transfer from Cu to Ni, which is consistent with the XPS results.

  The shift from monometallic Ni-N-C and Cu-N-C to bimetallic Cu/Ni-N-C brings about remarkable alterations to the electronic structure of the catalysts. To establish how such electronic structural changes are associated with CO₂RR catalytic activity, orbital analysis was further performed in this work. As Cu/Ni-N-C delivered boosted catalytic performance with metal sites acting as the active sites, this specific configuration was prioritized for systematic investigation and compared against Cu-N-C, aiming to unravel the underlying reasons for the disparities in their catalytic behaviors. Nickel possesses a valence electron configuration of 3d⁸4s², whereas copper has that of 3d¹⁰4s¹. Within the Cu-Ni dual-atom architecture, the unoccupied 3d orbitals of Ni (3d⁸) are able to accept electron density from the fully filled 3d orbitals of Cu (3d¹⁰), thereby enabling orbital hybridization-driven interatomic electron transfer. PDOS calculations demonstrate that the HOMO-LUMO band gap is narrowed in Cu/Ni-N-C, a feature that points to improved electronic conductivity and consequently higher catalytic activity potential, which stems from more efficient electron transfer throughout the reaction process (Fig. 5e and Fig. S38 in Supporting information). In addition, the introduction of Ni into the Cu-N-C matrix causes the d-band center of Cu 3d to positively shift from −3.396 eV to −2.844 eV (Fig. 5e), which infers stronger binding interactions with reaction intermediates. An analogous positive shift is also detected in the PDOS spectrum for the Ni 3d d-band center (Fig. S38).

  The surface coverage of ^*H species over catalysts with distinct hybrid heterostructures was quantified by evaluating the free energy variations related to water dissociation (ΔG_H2O) and hydrogen adsorption-desorption (ΔG_H) on Cu-N-C, Ni-N-C and Cu/Ni-N-C catalysts (Figs. 5f and g). Specifically, Cu/Ni-N-C exhibits a ΔG_H2O value of 0.61 eV, which is considerably lower than those of Cu-N-C (1.65 eV) and Ni-N-C (1.20 eV) (Fig. 5f). This result manifests that the incorporation of Ni endows Cu/Ni-N-C with more thermodynamically favorable water dissociation kinetics, thus yielding an elevated ^*H concentration that is available for the subsequent electrocatalytic reactions. On the contrary, the absolute ΔG_H value of Cu/Ni-N-C is 1.08 eV, which is markedly lower than the corresponding values of Cu-N-C (1.96 eV) and Ni-N-C (1.86 eV) (Fig. 5g). To further clarify the CO formation mechanism over Cu/Ni-N-C, DFT calculations were utilized to simulate the Gibbs free energy profiles of the complete reaction sequence. As illustrated in Fig. 5h, the ^*COOH formation on Cu/Ni-N-C is energetically more favorable (ΔG = 1.78 eV) than on Cu-N-C (ΔG = 2.52 eV) and Ni-N-C (ΔG = 2.24 eV). This suggests that Cu/Ni-N-C was more conducive to CO formation than the Cu-N-C and Ni-N-C catalysts. Furthermore, the thermodynamic barrier for ^*COOH formation is found to be higher on Cu-N-C (ΔG = 2.52 eV) than on Ni-N-C (ΔG = 2.24 eV), a result that verifies Ni (rather than Cu) as the preferred metal site for anchoring the C atom of ^*COOH moieties in Cu/Ni-N-C, which aligns with the findings in Fig. 5d.

  To clarify the fundamental reason for this catalytic selectivity, we analyzed the structural configurations of the key ^*COOH intermediate on the surfaces of Cu-N-C, Ni-N-C and Cu/Ni-N-C catalysts (Fig. 5i). On Cu-N-C, the Cu-C bond length in the ^*COOH intermediate is 2.027 Å, and its C—OH bond length is 1.376 Å. In the same manner, the Ni-N-C surface exhibits a Ni-C bond length of 2.020 Å and a C—OH bond length of 1.360 Å for the adsorbed ^*COOH. In sharp contrast, the Cu/Ni-N-C surface shows a shortened Ni-C bond length of 1.921 Å for ^*COOH, as well as a corresponding C—OH bond length of 1.388 Å. Such structural changes are of great significance to the underlying reaction mechanism of CO₂RR. A longer C—OH bond on Cu/Ni-N-C points to a weaker interaction between C and O atoms, thus making the C—OH bond easier to break. On the other hand, the shorter Ni-C bond on Cu/Ni-N-C means a stronger binding between the metal site and carbon atom, which is more favorable for the formation of the ^*CO intermediate. This analysis of intermediate configurations offers a molecular-level interpretation for the improved ^*CO formation efficiency observed on the Cu/Ni-N-C catalyst, and these structural understandings also confirm the validity of our earlier energy calculations.

  This study systematically investigates the structure and CO₂ electroreduction performance of Cu/Ni dual-atom catalysts (Cu/Ni-N-C) through a combination of experiments and theoretical calculations, clarifying the regulatory mechanism of intermetallic interactions on catalytic selectivity. Experimental results confirm that Cu and Ni atoms in Cu/Ni-N-C are each coordinated with four nitrogen atoms, forming uniformly distributed dual-atom active sites. The obtained Cu-N-C catalyst exhibits maximum FE_CO of 38.35% at −1 V vs. RHE and maximum FE_CH4 of 21.89% at −1.4 V vs. RHE, while the Cu_0.5Ni_0.5—N-C catalyst exhibits exceptional CO₂RR performance: At −1.3 V vs. RHE, the CO Faradaic efficiency (FE_CO) reaches 96.78% with a partial current density of 24.80 mA/cm². The observed selectivity variation is caused by the electron-withdrawing effect of Ni atoms on Cu atoms in the Cu/Ni-N-C catalyst, which lowers the electron cloud density around Cu atoms and thus weakens the adsorption of CO₂RR intermediates on Cu sites. DFT calculations reveal the core mechanism underlying the enhanced catalytic performance for CO formation: Charge regulation between Cu and Ni forms an asymmetric adsorption mode. Electron-deficient Cu atoms act as O-terminal adsorption sites to bind the oxygen atoms of CO₂; while electron-enriched Ni atoms serve as C-terminal adsorption sites to anchor the carbon atoms. This synergistic adsorption promotes CO₂ polarization and activation, narrows the HOMO-LUMO band gap for efficient electron transfer, and reduces the thermodynamic barrier for ^*COOH intermediate formation. The Cu/Ni-N-C not only strengthens the binding between the metal site and carbon atom of CO₂ but also weakens the C—OH interaction, facilitating CO generation. This work provides new insights into the rational design of dual-atom catalysts and highlights the potential of regulating intermetallic interactions to optimize CO₂RR performance, advancing their industrial applications.

  CRediT authorship contribution statement

  Jielian Yang: Writing – original draft, Software, Formal analysis, Data curation. Yue Shen: Methodology, Investigation. Bing Chen: Methodology, Investigation. Qi Wu: Methodology, Investigation. Yuemei Liao: Methodology, Investigation. Liya Zhou: Writing – review & editing, Project administration, Funding acquisition. Jing Luo: Methodology, Investigation. Xue Mao: Methodology, Investigation. Jin Guo: Methodology, Investigation. Binbin Luo: Methodology, Investigation. Qi Pang: Methodology, Investigation. Chunyan Zhou: Methodology, Investigation. Peican Chen: Writing – review & editing, Project administration, Funding acquisition. Anxiang Guan: Writing – review & editing, Resources, Project administration, Funding acquisition.

  Declaration of competing interest

  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

  Acknowledgments

  We thank the following funding agencies for supporting this work: the Natural Science Foundation of Guangxi Province (No. 2025GXNSFBA069409), 2025 Annual Project for Improving the Basic Scientific Research Capabilities of Young and Middle-aged Teachers in Colleges and Universities of Guangxi (No. 2025KY0035), National Natural Science Foundation of China (No. 22465006); the Guangxi Science and Technology Major Program (No. Guike AA24263003); Guangxi Youth Talent Inclusive Support Policy Research Launch Fund (No. ZX02080030425009); the launch funding for high-level talent researchers of Guangxi University (No. ZX01080020124003).

  Supplementary materials

  Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2026.112638.

  
  
• 1. [1]

     I. Song, Y. Eom, M. P, et al., Small 19 (2023) 2300049. doi: 10.1002/smll.202300049

  2. [2]

     F.M. Mota, D.H. Kim, Chem. Soc. Rev. 48 (2019) 205–259. doi: 10.1039/C8CS00527C

  3. [3]

     Y. Birdja, E. Pérez-Gallent, M. Figueiredo, et al., Nat. Energy 4 (2019) 732–745. doi: 10.1038/s41560-019-0450-y

  4. [4]

     Y. Chen, X. Pan, L. Li, et al., Small 21 (2025) 2411249. doi: 10.1002/smll.202411249

  5. [5]

     G. Jeong, Y. Tan, J. Song, et al., Chem. Eng. J. 426 (2021) 131063. doi: 10.1016/j.cej.2021.131063

  6. [6]

     B. Chen, D. Shi, R. Deng, et al., ACS Catal. 14 (2024) 16224–16233. doi: 10.1021/acscatal.4c05169

  7. [7]

     X. Zhang, X. Zhu, S. Bo, et al., Nat. Commun. 13 (2022) 5337. doi: 10.1038/s41467-022-33066-6

  8. [8]

     M. Hulsey, V. Fung, X. Hou, et al., Angew. Chem. Int. Ed. 61 (2022) e202208237. doi: 10.1002/anie.202208237

  9. [9]

     H. Fei, J. Dong, Y. Feng, et al., Nat. Catal. 1 (2018) 63–72. doi: 10.1038/s41929-017-0008-y

  10. [10]

      C. Xia, Y. Qiu, Y. Xia, et al., Nat. Chem. 13 (2021) 887–894. doi: 10.1038/s41557-021-00734-x

  11. [11]

      X. Liu, F. He, Y. Lu, et al., Chem. Eng. J. 453 (2023) 139833. doi: 10.1016/j.cej.2022.139833

  12. [12]

      J. Pei, H. Shang, J. Mao, et al., Nat. Commun. 15 (2024) 416.

  13. [13]

      X. Niu, J. Wei, D. Xu, et al., Nano Res. 17 (2024) 4702–4710. doi: 10.1007/s12274-023-6317-3

  14. [14]

      J. Leverett, T. Tran-Phu, J. Yuwono, et al., Adv. Energy Mater. 12 (2022) 2201500. doi: 10.1002/aenm.202201500

  15. [15]

      A. Guan, Y. Fan, S. Xi, et al., ACS Mater. Lett. 5 (2023) 19–26. doi: 10.1021/acsmaterialslett.2c01023

  16. [16]

      D. Koshy, D. Chen, D. Lee, et al., Angew. Chem. Int. Ed. 59 (2020) 4043–4050. doi: 10.1002/anie.201912857

  17. [17]

      M. Fang, X. Wang, X. Li, et al., ChemCatChem 13 (2021) 603–609. doi: 10.1002/cctc.202001667

  18. [18]

      S. Liang, J. Xiao, T. Zhang, et al., Angew. Chem. Int. Ed. 62 (2023) 202310740. doi: 10.1002/anie.202310740

  19. [19]

      X. Li, M. Qin, X. Wu, et al., Small 19 (2023) 2302530. doi: 10.1002/smll.202302530

  20. [20]

      S. Lu, M. Mazur, K. Guo, et al., Small 20 (2024) 2309251. doi: 10.1002/smll.202309251

  21. [21]

      X. Wang, N. Zhang, S. Guo, et al., J. Am. Chem. Soc. 146 (2024) 21357–21366. doi: 10.1021/jacs.4c03576

  22. [22]

      L. Zhang, N. Zhang, H. Shang, et al., Nat. Commun. 15 (2024) 9440. doi: 10.1038/s41467-024-53871-5

  23. [23]

      D. Yao, C. Tang, X. Zhi, et al., Adv. Mater. 35 (2023) 2209386. doi: 10.1002/adma.202209386

  24. [24]

      Y. Chen, J. Zhao, X. Pan, et al., Angew. Chem. Int. Ed. 63 (2024) e202411543. doi: 10.1002/anie.202411543

  25. [25]

      B. Hu, Y. Cao, Front. Energy. 19 (2025) 757–766. doi: 10.1007/s11708-025-1021-5

  26. [26]

      H. Xu, S. Fan, H. Liu, et al., Energy Storage Mater. 82 (2025) 104604. doi: 10.1016/j.ensm.2025.104604

  27. [27]

      A. Guan, Z. Chen, Y. Quan, et al., ACS Energy Lett. 5 (2020) 1044–1053. doi: 10.1021/acsenergylett.0c00018

  28. [28]

      L. Zhou, C. Yang, J. Wen, et al., J. Mater. Chem. A 5 (2017) 19343–19350. doi: 10.1039/C7TA05522F

  29. [29]

      Y. Qian, P. Du, P. Wu, et al., J. Phys. Chem. C 120 (2016) 9884–9896. doi: 10.1021/acs.jpcc.6b02670

  30. [30]

      Z. Gu, N. Yang, P. Han, et al., Small Methods. 3 (2019) 1800449. doi: 10.1002/smtd.201800449

  31. [31]

      H. Wang, H. Zou, Y. Liu, et al., Sci. Rep. 11 (2021) 21414. doi: 10.1038/s41598-021-00776-8

• 

  Figure 1  Schematic of the procedure to synthesize Cu-N-C and Cu_0.5Ni_0.5—N-C catalysts. Orange sphere: Cu atom, purple sphere: Ni atom, blue sphere: N atom, black sphere: C atom.The XRD patterns of all Cu_xNi_y-N-C-BL samples exhibit sharp diffraction peaks at 2θ = 43.3°, 50.4°, and 74.1° (Fig. 2d, Figs. S3 and S4 in Supporting information), corresponding to the (111), (200), and (220) crystal planes of the cubic Cu structure (JCPDS No. 04–0836). After acid leaching, the resulting Cu_xNi_y-N-C samples show only a broad peak centered at 26.4°, which is attributed to the (002) plane of the hexagonal graphitic structure (Fig. 2d and Figs. S3-S5 in Supporting information) and confirms the complete removal of metallic Cu via acid leaching. This is because metal particles had weak interactions with the carbon matrix and were thus dissolved in the oxygen-saturated acid solution. For comparison, the XRD pattern of the nitrogen-doped carbon (N—C) sample was also examined (Fig. S6 in Supporting information), where no Cu diffraction peaks are observed and only the carbon-related peak is identifiable.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) High-magnification SEM image of HKUST-1. (b) Low-magnification SEM image of (Cu_0.5Ni_0.5) HKUST-1 MOF. (c) High-magnification SEM image of (Cu_0.5Ni_0.5) HKUST-1 MOF. (d) XRD patterns of Cu_0.5Ni_0.5—N-C-BL and Cu_0.5Ni_0.5—N-C samples. (e) SEM image of Cu-N-C. (f) Low-magnification TEM image of Cu-N-C. (g) High-magnification TEM image of Cu-N-C, with the inset showing the corresponding selected area electron diffraction pattern. (h) EDS elemental distribution map of Cu-N-C. (i) SEM image of Cu_0.5Ni_0.5—N-C. (j) Low-magnification TEM image of Cu_0.5Ni_0.5—N-C. (k) High-magnification TEM image of Cu_0.5Ni_0.5—N-C, with the inset showing the corresponding selected area electron diffraction pattern. (l) EDS analysis of Cu_0.5Ni_0.5—N-C. (m) EDS elemental distribution map of Cu_0.5Ni_0.5—N-C. (n) AC—HAADF-STEM image of Cu_0.5Ni_0.5—N-C (dual-atom sites (circled in red) and single-atom sites (circled in yellow)). (o) 3D surface plot from region 1 of (n). (p) Statistical distribution of single sites and dual sites in Cu_0.5Ni_0.5—N-C. (q) Intensity profiles obtained from areas 2–3 in (n).

  下载: 全尺寸图片 幻灯片
  

  Figure 3  High-resolution (a) N 1s, (b) Cu 2p, and (c) Ni 2p XPS spectra of Cu_0.5Ni_0.5—N-C. (d) Cu K-edge and (e) Ni K-edge XANES spectra of Cu_0.5Ni_0.5—N-C. FT-EXAFS spectra of (f) Cu K-edge and (g) Ni K-edge of Cu_0.5Ni_0.5—N-C. (h) EXAFS fitting of Cu_0.5Ni_0.5—N-C in the R space at the Cu K-edge and (i) Ni K-edge of Cu_0.5Ni_0.5—N-C. (j) Wavelet transform images of EXAFS data at Cu K-edge with the optimized Morlet parameter for CuPc (left) and Cu_0.5Ni_0.5—N-C (right). (k) Wavelet transform images of EXAFS data at Ni K-edge with the optimized Morlet parameter for NiPc (left) and Cu_0.5Ni_0.5—N-C (right).

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) LSV curves of Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C in 0.1 mol/L KHCO₃ saturated with Ar or CO₂. Faradaic efficiency (FE) of all CO₂RR products at different potentials of (b) Cu-N-C, (c) Cu_0.5Ni_0.5—N-C, (d) N—C. (e) Tafel plots for CO₂RR over Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C. (f) Nyquist curves of Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C. (g) CV fitting curves of Cu_0.5Ni_0.5—N-C, Cu_0.8Ni_0.2—N-C, Cu-N-C, Ni-N-C and N—C. (h) Stability assessment of Cu_0.5Ni_0.5—N-C at −1.3 V vs. RHE. High-resolution (i) N 1s, (j) Cu 2p, and (k) Ni 2p XPS of Cu_0.5Ni_0.5—N-C catalyst after CO₂RR.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  (a, b) The charge density difference of Cu/Ni-N-C and Cu-N-C. The yellow and cyan region indicates electron accumulation and depletion, respectively. (c) ELF analysis for Cu/Ni-N-C with color indicating where red and blue indicate electron localization and delocalization regions, respectively. (d) Proposed CO₂ configuration adsorbed on the Cu/Ni-N-C due to the electron distribution differences between the localized positive charge on the Cu sites and negative Ni sites. (e) Projected density of states demonstrating the Cu d-states of Cu/Ni-N-C and Cu-N-C. The Fermi energy is illustrated by a black dashed line, while d-band centers (ε_d) for Cu-N-C and Cu/Ni-N-C are marked by orange and purple dashed lines, respectively. Adsorption energy diagram of (f) H₂O dissociation and (g) hydrogen ad-desorption for Cu-N-C, Ni-N-C and Cu/Ni-N-C. (h) Gibbs free energy diagrams for the formation of CO on Cu-N-C, Ni-N-C and Cu/Ni-N-C. (i) Structural comparison of COOH^* adsorbed on Cu-N-C and Cu/Ni-N-C, showing atomic configurations and bond lengths. Color codes: Cu (orange), Ni (purple), N (blue), C (gray), O (red), H (white).

  下载: 全尺寸图片 幻灯片
•