Direct electrosynthesis of acetamide from CO2 and nitrate via an atomically engineered dual-site catalyst Authors: Shuai Xia, Hao Tan, Jianfang Zhang, Miao Han, Chang Xu, Cuiping Yu, Kui Chen, Yong Zhang, Yucheng Wu, Yan Wang Published: 2026 ABSTRACT AbstractThe electrocatalytic synthesis of amides from abundant small molecules offers a sustainable route for green chemical production, yet faces fundamental challenges due to kinetic competition between C–C and C–N bond formation. Here we show an atomically engineered dual-site catalyst featuring nickel single atoms adjacent to copper nanoclusters (Ni-SA/Cu-NCs) on a nitrogen-doped carbon matrix for efficient CO2 and NO3− co-reduction to acetamide. This architecture enables complementary functions, with Ni sites selectively converting CO2 to CO and neighboring Cu nanoclusters promoting C–C coupling to form the *CCO intermediate while concurrently reducing NO3− to form the *NH2 intermediate. The resulting synergy facilitates rapid intermediate transfer and C–N coupling, delivering an acetamide yield rate of 257.3 mmol h−1 gcat.−1 at an industrial current density of 215.7 mA cm−2, with stable operation over 160 h. In situ spectroscopic studies and theoretical calculations suggest that strong Ni–Cu electronic coupling promotes reactant adsorption and reduces the activation barriers for critical steps, including *CO dimerization and *CCO–*NH2 coupling. This work provides an atomic-level design strategy for multi-site catalysts to steer complex electrocatalytic reactions toward value-added products.Subject terms: Electrocatalysis, Electrocatalysis, Electrocatalysis, Heterogeneous catalysis Introduction Amides, as fundamental structural motifs in polymers and fine chemicals, represent a vital class of organic nitrogen compounds. Their significance is particularly evident in the pharmaceutical industry, where over 25% of marketed drugs contain amide bonds1,2. Consequently, the development of sustainable synthetic routes for amides represents a key objective in advancing green chemical manufacturing. Conventional thermochemical methods, however, often require harsh reaction conditions. For example, the industrial synthesis of acetamide via condensation of acetic acid and ammonia typically operates at 150–180 °C (Fig. 1a), involving energy-intensive processes accompanied by substantial carbon emissions3–5. Therefore, establishing clean and energy-efficient synthetic pathways under ambient conditions remains a highly desirable and pressing goal. Routes for acetamide synthesis from conventional thermochemistry to direct CO2 and nitrate electrosynthesis. a Traditional thermochemical synthesis of acetamide from acetic acid and ammonia. b Electrocatalytic synthesis of acetamide using carbonaceous and nitrogenous feedstocks. b1 Previously reported electrocatalytic CO + NH3 pathway. b2 Direct electrocatalytic CO2 + NO3− pathway investigated in this work. Source data for Fig. 1 are provided as a Source Data file. Electrocatalytic synthesis, powered by renewable electricity, has emerged as a sustainable strategy for chemical production. A promising route involves the co-reduction of carbonaceous (e.g., CO2, CO) and nitrogenous (e.g., NO3−, NO2−, NH3) feedstocks (Fig. 1b), which enables the direct conversion of abundant small molecules into value-added nitrogen-containing organics, thereby offering a potential pathway to close both the carbon and nitrogen cycles6–8. While significant advances have been achieved in electrocatalytic C–N coupling, current research has predominantly centered on the co-reduction of CO2 with nitrogenous species such as NO3− for urea synthesis9–11. Extending this methodology to synthesize more structurally complex amides, such as acetamide, remains substantially challenging. A primary obstacle lies in the difficulty of simultaneously governing C–C and C–N bond formation at the atomic level. In electrocatalytic systems, these coupling pathways often compete kinetically and spatially, which hinders the selective coupling of key intermediates like *CO and *NH212,13. Furthermore, competing side reactions, including the hydrogen evolution reaction (HER) and over-reduction processes, severely impair selectivity. This underscores the critical need for advanced catalysts capable of concurrently activating both carbon and nitrogen species while steering their selective coupling toward the desired amide products. Previous studies have shown that acetamide formation is critically dependent on both reactant composition and catalyst structure, and a feasible pathway involving CO2 reduction to CO followed by nucleophilic attack of NH3 on the carbonyl carbon has been proposed (Fig. 1b1). For instance, it has been demonstrated that during the co-electrolysis of CO2 and NH3 on copper-based catalysts, acetamide can form via CO dimerization into a ketene (CCO) intermediate, followed by coupling with NH37,14. However, such systems often suffer from sluggish CO2-to-CO conversion and inefficient C–C coupling, resulting in limited yield and selectivity. Moreover, the reliance on NH3 derived from the energy-intensive Haber-Bosch process diminishes the overall sustainability of this route. In contrast, electrocatalytic reduction of nitrogen oxides offers a more practical and environmentally benign alternative, as it enables nitrogen recycling while mitigating nitrogen pollution. Building on this concept, our group previously developed a catalytic system capable of efficiently co-reducing CO and NO3− to acetamide, achieving industrially relevant current density and production rates15. However, that system relied on CO as the carbon feedstock and therefore bypassed the kinetically demanding CO2-to-CO conversion step. By contrast, direct acetamide synthesis from CO2 and NO3− is not only more sustainable from a feedstock perspective, but also substantially more challenging. The fundamental bottleneck lies in the intrinsic kinetic mismatch between intermediate generation and downstream coupling. Specifically, CO2 reduction to CO is kinetically demanding, whereas NO3− reduction to NH2 requires substantial electron input and appropriately tuned adsorption strength. Efficient acetamide formation therefore requires effective kinetic matching between intermediate generation and the subsequent C–C and C–N coupling steps. These parallel pathways compete for active sites and charge carriers, often resulting in an insufficient supply of either *CO or *NH2, and consequently, limited efficiency in both C–C and C–N coupling16–18. To overcome the intrinsic limitations arising from active-site competition, the construction of synergistic dual-site catalysts represents a promising strategy. Such architectures incorporate spatially distinct yet functionally integrated active centers, enabling capabilities such as the selective conversion of CO2 to CO at low overpotentials. By combining different metal sites at a shared interface, these catalysts facilitate the cooperative adsorption and coupling of carbon and nitrogen intermediates19–21. Among candidate metals, copper (Cu) is particularly advantageous owing to its distinctive d-band structure, which promotes both C–C and C–N bond formation22,23. Unlike single-atom catalysts, dual-site configurations leverage electronic coupling and charge transfer between adjacent metal atoms to precisely modulate the local electronic environment. This interaction induces asymmetric charge redistribution and optimizes the d-band configuration, thereby circumventing the linear scaling relationships that typically constrain adsorption energies on uniform surfaces24–27. The resulting electronic reconstruction establishes a favorable electronic microenvironment conducive to the cooperative adsorption and conversion of key intermediates, ultimately facilitating C–N coupling. Therefore, the efficacy of such synergy is critically governed by the specific atomic configuration of the dual-site catalyst, as different structural arrangements dictate charge interaction patterns and, consequently, catalytic performance in reactant activation and C–N coupling28–31. Herein, we report an alternative pathway for direct electrosynthesis of acetamide from CO2 and NO3− using an atomically engineered Ni–Cu catalyst (Fig. 1b2), which features nickel single atoms adjacent to copper nanoclusters (Ni-SA/Cu-NCs) on a nitrogen-doped carbon matrix. This specific architecture was designed to spatially and electronically match the distinct mechanistic requirements of direct acetamide electrosynthesis. In this configuration, Ni sites selectively reduce CO2 to CO, providing a localized carbon feedstock, while neighboring Cu nanoclusters promote *CO dimerization to electrophilic CCO species and concurrently reduce NO3− to generate nucleophilic NH2. The close spatial proximity between these sites facilitates rapid intermediate transfer and localized C–N coupling, thereby effectively suppressing competing side reactions. Leveraging this atomic-level synergy, the Ni-SA/Cu-NCs catalyst delivers a high acetamide yield rate of 257.3 mmol h−1 gcat.−1 at a current density of 215.7 mA cm−2, with stable operation exceeding 160 h. In situ spectroscopic studies corroborate the dynamic formation and cooperative interaction of key *CCO and *NH2 intermediates on the catalyst surface. Density functional theory (DFT) and transition-state calculations further suggest that strong Ni–Cu electronic coupling enhances CO2/NO3− adsorption, lowers the activation barrier for *CO dimerization, and thermodynamically favors *CCO–*NH2 coupling, collectively improving acetamide formation. This work exemplifies a rational design strategy for multi-site catalysts with atomic precision, providing a sustainable and versatile route for synthesizing valuable amides from abundant small molecules. Results Catalyst synthesis and characterization The Ni-SA/Cu-NCs catalyst was fabricated via a stepwise impregnation-pyrolysis approach (Fig. 2a). Initially, the Ni single-atom (Ni-SA) precursor was prepared through a two-step pyrolysis process using L-cysteine, melamine, and nickel nitrate. Copper nanoclusters (Cu-NCs) were subsequently anchored onto the Ni-SA substrate to yield the final catalyst. For comparative evaluation, a series of control catalysts, including Ni-SA/Cu-SA, Cu-NCs, Fe-SA/Cu-NCs, and Co-SA/Cu-NCs, were synthesized under analogous conditions (see methods section). Transmission electron microscopy (TEM) images of Ni-SA/Cu-NCs revealed a porous architecture of thin carbon nanosheets, with no observable metallic nanoparticles (Supplementary Fig. 1), suggesting the high dispersion of both Ni and Cu species. This morphology remained consistent across all control samples (Supplementary Figs. 2, 3). Brunauer-Emmett-Teller (BET) measurements further showed that Ni-SA/Cu-NCs exhibited the largest specific surface area (372.47 m2 g−1) and mesopore volume among the catalysts studied (Supplementary Fig. 4). Such a porous framework is conducive to enhanced mass transport and increased accessibility of active sites. Structural and compositional characterization of Ni–Cu dual-site catalysts. a Schematic of the stepwise impregnation–pyrolysis synthesis. b–d Aberration-corrected HAADF-STEM images of Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs; orange, yellow and red circles denote isolated Ni single atoms, isolated Cu single atoms and Cu nanoclusters, respectively. e EDX elemental mapping of Ni-SA/Cu-NCs, showing the spatial distribution of Ni, Cu, C and N. Scale bars, 5 nm. f XRD patterns and g Raman spectra of the three catalysts. h Cu 2p and i Ni 2p XPS spectra of Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs. Intensities in (f–i) are given in arbitrary units (a.u.). Source data for Fig. 2 are provided as a Source Data file. To resolve the atomic-scale structure of the catalysts, aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) was employed. As shown in Fig. 2b–d, both Ni and Cu species were discernible as bright dots against the lighter carbon-nitrogen support. In the control samples Ni-SA and Ni-SA/Cu-SA, these bright dots appeared uniformly isolated, consistent with the presence of atomically dispersed Ni (orange circles) and Cu (yellow circles) single atoms, respectively32,33. The homogeneous elemental distribution in these samples was further verified by energy-dispersive X-ray (EDX) mapping (Supplementary Figs. 2c and 3d). Notably, in the Ni-SA/Cu-NCs catalyst, HAADF-STEM images not only revealed the preservation of isolated Ni single atoms but also showed uniformly distributed high-density bright regions with an average diameter of ~0.83 nm, corresponding to ultrasmall Cu nanoclusters (red circles) composed of several atoms34–36. Elemental mapping further confirmed the homogeneous co-distribution of Ni, Cu, C, and N throughout the Ni-SA/Cu-NCs catalyst (Fig. 2e). Additionally, inductively coupled plasma-mass spectrometry (ICP-MS) measurements quantified the metal loadings: the Ni contents in Ni-SA, Ni-SA/Cu-SA, and Ni-SA/Cu-NCs were 0.49 wt%, 0.47 wt%, and 0.5 wt%, respectively, while the corresponding Cu contents were undetectable, 0.51 wt%, and 0.83 wt% (Supplementary Table 1). The crystallinity and carbon structure of the catalysts were further examined by X-ray diffraction (XRD) and Raman spectroscopy. As shown in Fig. 2f, all samples exhibited two broad diffraction peaks at 26.2° and 43.6°, corresponding to the (002) and (101) planes of graphitic carbon, respectively37,38. No discernible diffraction peaks from crystalline Ni, Cu, or related metal phases were observed, consistent with the high dispersion of metal species indicated by STEM analysis. The Raman spectrum of Ni-SA/Cu-NCs (Fig. 2g) showed a slightly higher defect-to-graphitic carbon ratio (ID/IG = 1.070) than those of Ni-SA (1.063) and Ni-SA/Cu-SA (1.064). Given the small differences among these values, this result does not indicate a fundamentally different carbon framework, but rather suggests a subtle increase in local defect-rich carbon environments in Ni-SA/Cu-NCs. Such local defect environments may provide anchoring sites for ultrasmall Cu nanoclusters and atomically dispersed Ni species within the N-doped carbon matrix, helping to stabilize the highly dispersed metal sites39,40. X-ray photoelectron spectroscopy (XPS) was employed to probe the electronic states of surface species. In the Cu 2p spectrum of Ni-SA/Cu-NCs (Fig. 2h), the two main peaks at 933.1 eV and 952.6 eV were assigned to Cu 2p3/2 and Cu 2p1/2 of Cu+/Cu0 species, respectively. Accompanying satellite peaks at 934.5 eV and 955.3 eV, characteristic of Cu2+, further indicated the mixed valence state of Cu within the nanoclusters. By contrast, only Cu2+ features were observed in Ni-SA/Cu-SA, reflecting the dominance of positively charged, isolated Cu single atoms, likely a result of electronic modification induced by atomic dispersion. No Cu signal was detected in the Ni-SA sample. The high-resolution Ni 2p spectra (Fig. 2i) consistently exhibited Ni2+ signatures across all three catalysts, with peaks at 854.9 eV (Ni 2p3/2) and 872.2 eV (Ni 2p1/2), supporting the stabilization of Ni as atomically dispersed sites. Soft X-ray absorption near-edge structure (XANES) spectroscopy offered deeper insights into the electronic configuration of the catalysts. In the C K-edge absorption spectra (Fig. 3a), the nitrogen-doped carbon (NC) support exhibited three characteristic peaks at 285.83, 293.3, and ~298.5 eV, corresponding to C 1 s core-level electronic transitions to π*(C = C), σ*(C–C), and σ*(C = O) orbitals, respectively41. A similar spectral profile was observed for Ni-SA, Ni-SA/Cu-SA, and Ni-SA/Cu-NCs. Notably, the σ*(C–C) peak progressively decreased in intensity and broadened with increasing Cu content, implying enhanced electronic interaction between Cu nanoclusters and the carbon matrix. In the N K-edge spectra (Fig. 3b), all catalysts displayed characteristic absorption features related to different nitrogen species. Slight shifts and intensity variations among the samples indicated that metal incorporation modulates the electronic environment of the N-doped carbon framework42. Consistently, the N 1 s XPS spectra (Supplementary Fig. 5) showed a distinct peak at 397.2 eV in all metal-containing samples, attributable to M–N (M=Ni, Cu) coordination, supporting the formation of metal-nitrogen bonding environments. The Cu K-edge XANES spectrum of Ni-SA/Cu-NCs (Fig. 3c) exhibited an absorption edge energy situated between those of Cu foil (Cu0) and CuO (Cu2+), supporting the mixed valence state of Cu identified by XPS. In contrast, the edge position of Ni-SA/Cu-SA closely aligned with that of CuO, affirming the dominant +2 oxidation state of Cu in the single-atom form. Extended X-ray absorption fine structure (EXAFS) analysis further elucidated the local coordination environment (Supplementary Fig. 6a). The Fourier-transformed Cu K-edge EXAFS spectrum of Ni-SA/Cu-NCs (Fig. 3d) showed two prominent peaks at ~1.48 Å and ~2.50 Å, attributable to Cu–N and Cu–Cu scattering paths, respectively43,44. This result supports the coexistence of Cu atoms coordinated to N in the carbon matrix and adjacent Cu atoms within nanoclusters. By comparison, only the Cu–N coordination peak was observed in Ni-SA/Cu-SA, consistent with its single-atom structure. Quantitative Cu K-edge EXAFS fitting further revealed weak Cu–Cu scattering in Ni-SA/Cu-NCs, with a low Cu–Cu coordination number of ~2.0 and a Cu–Cu distance of ~2.51 Å (Supplementary Fig. 7 and Supplementary Table 2), supporting the presence of ultrasmall Cu clusters comprising only a few atoms. Cu K-edge wavelet transform (WT) analysis further distinguished the Cu–N and Cu–Cu contributions in k space (Fig. 3g). Ni-SA/Cu-SA showed only a Cu–N-related low-k feature at k ≈ 4.6 Å−1, whereas Ni-SA/Cu-NCs exhibited an additional weaker high-k feature at k ≈ 6.3 Å−1, assignable to Cu–Cu scattering. The attenuated Cu–Cu-related WT intensity is consistent with the low Cu–Cu coordination and sub-nanometer cluster size of Ni-SA/Cu-NCs. X-ray absorption spectroscopy identifies the local coordination and electronic structures of Ni and Cu sites. a C K-edge and b N K-edge XANES spectra of NC, Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs. c Cu K-edge XANES and d Cu K-edge FT-EXAFS spectra of Cu foil, Ni-SA/Cu-SA, Ni-SA/Cu-NCs, CuPc, Cu2O and CuO. e Ni K-edge XANES and f Ni K-edge FT-EXAFS spectra of Ni foil, Ni-SA, Ni-SA/Cu-SA, Ni-SA/Cu-NCs, NiPc and NiO. g Cu and Ni K-edge wavelet-transform (WT) spectra of the corresponding samples. R denotes the radial distance from the absorbing atom and k the photoelectron wavevector. Absorption and FT-EXAFS intensities are given in arbitrary units (a.u.). Source data for Fig. 3 are provided as a Source Data file. The Ni K-edge XANES spectra (Fig. 3e) indicated an oxidation-state trend in the order of Ni-SA > Ni-SA/Cu-NCs > Ni-SA/Cu-SA, suggesting that the electronic structure of Ni single atoms is modulated by adjacent Cu species, likely through intermetallic electronic interactions. The FT-EXAFS spectra at the Ni K-edge (Fig. 3f) exhibited only one dominant peak at 1.59 Å in all three catalysts, corresponding to the Ni–N coordination shell, with no detectable Ni–Ni contribution45. Quantitative Ni K-edge EXAFS fitting further confirmed the preserved Ni–N4-like coordination environment across Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs, with Ni-N coordination numbers of ~4.2–4.6 (Supplementary Fig. 8 and Supplementary Table 3). Consistently, the Ni K-edge WT plots showed only a Ni–N-related maximum at k ≈ 4.1 Å−1 for all samples (Fig. 3g), supporting the preservation of atomically dispersed Ni sites46. Together with the Cu K-edge analysis, these Ni K-edge results support the successful construction of the intended Ni single-atom/Cu nanocluster dual-site architecture in Ni-SA/Cu-NCs. Electrochemical performance of acetamide synthesis The electrocatalytic performance for acetamide synthesis through CO2 and NO3− co-reduction was systematically investigated in a flow cell. Preliminary pH-dependence evaluations revealed a volcano-shaped activity trend, indicating that a near-neutral environment optimally balances the reaction by suppressing excessive hydrogen evolution in acidic media and mitigating competitive C2+ formation as well as acetamide hydrolysis in alkaline conditions (Supplementary Figs. 9, 10). Accordingly, 1.0 M KCl containing 0.02 M KNO3 was used as the catholyte for subsequent CO2 + NO3− co-reduction measurements. Gaseous and liquid products were quantified by gas chromatography and 1H nuclear magnetic resonance (NMR), respectively, with additional validation provided by colorimetric assays (Supplementary Figs. 11–13). Linear sweep voltammetry (LSV) measurements showed that Ni-SA/Cu-NCs delivered higher current densities than Ni-SA and Ni-SA/Cu-SA at comparable potentials under co-reduction conditions (Fig. 4a), indicating enhanced co-reduction activity. Cyclic voltammetry (CV) in the non-Faradaic region further showed that Ni-SA/Cu-NCs exhibited the largest Cdl-derived electrochemical active surface area (ECSA) among the catalysts (Supplementary Fig. 14), suggesting more electrochemically accessible active sites. Electrochemical performance for acetamide synthesis via CO2 and nitrate co-reduction. a LSV curves of Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs under CO2 + NO3⁻ co-reduction. b, c Total Faradaic efficiency and current density of Ni-SA/Cu-NCs under CO2 + NO3⁻ co-reduction and CO2 electroreduction, respectively. d Acetamide yield rate and Faradaic efficiency over different catalysts. e, f Intermediate-equivalent partial current densities for NH2 (jNH2) and CO (jCO) generation and the corresponding normalized NH2 and CO utilization, respectively. g ¹H NMR spectra under different reaction conditions. h Long-term stability of Ni-SA/Cu-NCs at −1.2 V vs. RHE, with the corresponding acetamide yield rate and Faradaic efficiency. Data are presented as mean ± s.d. from three independent measurements (n = 3). Individual data points are overlaid in (d–f). Electrochemical measurements were performed in a flow cell using 1.0 M KCl + 0.02 M KNO3 catholyte (1.0 M KCl for CO2 reduction), with a 1 cm2 GDE, a catalyst loading of 0.25 mg cm⁻2, a CO2 flow rate of 30 sccm and a catholyte flow rate of 0.8 mL min⁻1 at 25 °C. For Ni-SA/Cu-NCs under standard CO2 + NO3⁻ co-reduction conditions, the uncompensated resistance was 1.413 ± 0.003 Ω (n = 3), and potentials are reported without iR correction. Source data for Fig. 4 are provided as a Source Data file. A comparison with nitrate-free CO2 reduction further showed that Ni-SA/Cu-NCs exhibited an approximately 100 mV lower overpotential under CO2 + NO3− co-reduction conditions (Supplementary Fig. 15), indicating more favorable electrochemical kinetics in the coupled system. As shown in Fig. 4b and Supplementary Figs. 16, 17, Ni-SA/Cu-NCs achieved the highest acetamide yield rate and FE among the tested catalysts, with a volcano-shaped dependence on applied potential. The optimal performance was obtained at −1.2 V vs. RHE, reaching an acetamide yield rate of 257.3 mmol h−1 gcat.−1 and an FE of 12.8%, whereas Ni-SA/Cu-SA delivered only 3.7 mmol h−1 gcat.−1 with an FE of 0.4%, and Ni-SA showed negligible acetamide formation. At the same potential, the acetamide partial current density increased from 0.4 mA cm−2 over Ni-SA/Cu-SA to 27.6 mA cm−2 over Ni-SA/Cu-NCs, with the total current density reaching 215.7 mA cm−2. This high-current-density performance was reproducible across three independently synthesized Ni-SA/Cu-NCs batches, each fabricated into an independent gas-diffusion electrode, giving an acetamide FE of 12.8 ± 0.1%, a yield rate of 257.3 ± 3.3 mmol h−1 gcat.−1 and a total current density of 215.7 ± 1.9 mA cm−2 at −1.2 V vs. RHE (Supplementary Table 4). Together, these results highlight the ability of Ni-SA/Cu-NCs to sustain high acetamide productivity under high-current co-reduction conditions, which is competitive with previously reported electrocatalytic acetamide syntheses (Supplementary Table 5). To clarify the role of CO-derived intermediates in the enhanced acetamide productivity of Ni-SA/Cu-NCs, individual CO2RR was compared with CO2 + NO3− co-reduction. In CO2 reduction using 1.0 M KCl electrolyte, Ni-SA and Ni-SA/Cu-SA mainly generated CO, with Ni-SA showing near-unity CO selectivity (Fig. 4c and Supplementary Fig. 18). By contrast, Ni-SA/Cu-NCs exhibited a potential-dependent shift from CO to C2+ products, including ethylene, ethanol, acetate, and propanol, at more negative potentials. This behavior is consistent with the dual-site architecture of Ni-SA/Cu-NCs, in which CO generated on Ni single atoms can be locally supplied to adjacent Cu nanoclusters for subsequent CO–CO coupling. Upon switching to CO2 + NO3− co-reduction, C2+ products were still detected on Ni-SA/Cu-NCs but with lower efficiencies than in individual CO2RR, suggesting that some CO-derived C2 intermediates are redirected from C2+ product formation toward acetamide synthesis. In contrast, neither Ni-SA nor Ni-SA/Cu-SA showed detectable C2+ products under individual CO2RR or CO2 + NO3− co-reduction conditions, indicating limited *CCO availability for subsequent C–N coupling. The role of nitrogen-derived intermediates was further examined by individual NO3RR tests in 1.0 M KCl + 0.02 M KNO3 electrolyte (Supplementary Figs. 19–21). Ni-SA converted NO3− to NH3 with a yield rate of 175.4 mmol h−1 gcat.−1 and an FE of 17.3%, indicating that Ni single atoms can participate in NO3− activation. Incorporation of Cu single atoms further increased the NH3 yield and FE, while Ni-SA/Cu-NCs showed the highest NO3RR performance, suggesting that the adjacent Cu nanoclusters facilitate nitrogen-intermediate generation. Under CO2 + NO3− co-reduction conditions, however, both NH3 selectivity and the corresponding partial current density over Ni-SA/Cu-NCs decreased substantially (Supplementary Figs. 22 and 23), consistent with partial consumption of NH2-related intermediates through C–N coupling with carbon-derived species to form acetamide. Together with the CO2RR results, these observations support the requirement for coordinated generation of both *CCO- and NH2-related intermediates during acetamide electrosynthesis. To further assess the structural factors governing intermediate coupling and acetamide formation, control catalysts including Fe-SA/Cu-NCs, Co-SA/Cu-NCs, and Cu-NCs were synthesized and evaluated. As shown in Fig. 4d and Supplementary Fig. 24, the M-SA/Cu-NCs catalysts (M = Ni, Fe, and Co) showed substantially higher acetamide yield and FE than Cu-NCs alone or catalysts without Cu nanoclusters, such as Ni-SA and Ni-SA/Cu-SA, at −1.2 V vs. RHE. This comparison indicates that coupling isolated metal sites with contiguous Cu nanoclusters is beneficial for driving the tandem generation and coupling of carbon- and nitrogen-derived intermediates. Among these M-SA/Cu-NCs catalysts, Ni-SA/Cu-NCs delivered the highest performance, suggesting that the Ni–Cu interfacial configuration provides a more favorable balance among CO2-to-CO conversion, CO–CO coupling, and nitrogen-intermediate formation. The effect of Cu loading was further examined at a fixed Ni content of ~0.5 wt%. Acetamide performance showed a volcano-type dependence on Cu content, with negligible formation on Ni-SA, only trace activity on Ni-SA/Cu-SA, and optimal performance at ~0.83 wt% Cu in Ni-SA/Cu-NCs. Further increasing the Cu loading to ~1.36 wt% in Ni-SA/Cu-NCs-H led to slightly larger Cu clusters and reduced acetamide performance (Supplementary Figs. 25–28), suggesting that excessive Cu loading may partially disrupt the optimized Ni–Cu interfacial configuration and increase competition from parallel reduction pathways. The influence of these structural factors on reactive-intermediate generation and utilization was further evaluated using intermediate-equivalent partial current densities for NH2 and CO generation (jNH2 and jCO) and their normalized utilization efficiencies in C–N coupling (Fig. 4e, f). Ni-SA/Cu-NCs showed the highest utilization of both NH2- and CO-related intermediates among the tested catalysts, consistent with more efficient CO dimerization to CCO and improved steering of carbon- and nitrogen-derived intermediates toward acetamide formation. Product identification by ¹H NMR further showed a characteristic signal at ~2.0 ppm, matching the acetamide standard, only under CO2 + NO3− co-electrolysis conditions (Fig. 4g), supporting its formation through the coupled C–N pathway. Ni-SA/Cu-NCs also exhibited stable operation over 160 h at −1.2 V vs. RHE, with only minor decreases in acetamide yield rate and FE of 4.4% and 1.8%, respectively (Fig. 4h). Post-reaction HAADF-STEM, XPS and XRD analyses supported the preservation of the catalyst morphology, chemical states and phase structure after prolonged electrolysis. Time-resolved in situ XRD at −1.6 V vs. RHE for 300 min and HAADF-STEM of the recovered catalyst further showed no crystalline Cu-derived phases or detectable Cu-cluster aggregation, consistent with the structural robustness of the highly dispersed Ni-SA/Cu-NCs architecture under strongly cathodic co-reduction conditions (Supplementary Figs. 29–31). In an enlarged-area flow cell with an active area of 25 cm2, Ni-SA/Cu-NCs maintained an average acetamide FE of 12% and a yield rate of 5.89 mol h−1 gcat.−1 over 15 h at a cell voltage of 2.5 V and a total current of 5 A (Supplementary Fig. 32), providing an initial demonstration of enlarged-cell operation. Reaction mechanism investigation Control experiments using different carbon and nitrogen sources were performed over Ni-SA, Cu-NCs and Ni-SA/Cu-NCs to clarify the key intermediates and the functional roles of the dual-site architecture (Fig. 5a and Supplementary Fig. 33). The catalytic performance was strongly dependent on both feedstock composition and catalyst structure. Ni-SA exhibited negligible acetamide yield under all tested feed conditions, indicating that isolated Ni sites alone are insufficient to drive the downstream C–C and C–N bond-forming steps required for acetamide synthesis. When CO2 was used as the carbon source, Ni-SA/Cu-NCs delivered substantially higher acetamide yield and FE than Cu-NCs. By contrast, when CO was used as the carbon source, both catalysts showed similarly high performance, with yields exceeding those obtained under CO2-fed conditions. This contrast supports CO as a key carbon intermediate and is consistent with a functional division within the dual-site architecture, in which Ni single atoms mainly serve as local CO-generation sites that supply adjacent Cu nanoclusters for subsequent C-C and C-N bond-forming steps. Consistently, a physical mixture of Ni-SA and Cu-NCs at the same metal loadings gave lower acetamide yield and FE than the atomically integrated catalyst (Supplementary Fig. 34), indicating that this functional division requires atomic-scale Ni–Cu proximity. Replacing NO3− with NH3 further increased acetamide formation, supporting the participation of NH2-related species in the coupling pathway. The effect of CO availability on *CCO formation and subsequent acetamide synthesis was further examined over Cu-NCs using mixed CO/CO2 feeds with varying molar ratios (Fig. 5b). Both acetamide yield and FE increased progressively with increasing CO content and reached a plateau at a CO/CO2 ratio of 10:20, suggesting that local *CO coverage on Cu nanoclusters approaches saturation and that the reaction then becomes more limited by downstream coupling kinetics than by *CO supply. Although pure CO feeding affords high performance by bypassing the initial CO2 activation step (Fig. 5a), the key advance of Ni-SA/Cu-NCs lies in enabling the more challenging CO2 + NO3− pathway through local CO generation on Ni sites and subsequent *CO utilization on adjacent Cu nanoclusters. This interpretation is further supported by acetanilide electrosynthesis experiments (Supplementary Fig. 35), where the activity trends of Cu-NCs and Ni-SA/Cu-NCs closely mirrored those observed for acetamide synthesis, with Ni-SA/Cu-NCs showing higher performance than Cu-NCs when CO2 was used as the carbon source. This generality extends beyond acetanilide. Using CO2 as the carbon source, Ni-SA/Cu-NCs further converted methylamine, dimethylamine, and ethylamine into the corresponding N-substituted acetamides, and sodium 4-nitrobenzoate into 4-acetamidobenzoic acid (Supplementary Fig. 36), indicating that the *CCO-mediated C-N coupling accommodates diverse aliphatic and aromatic nitrogen sources. To further probe the CO relay, additional CO-feeding experiments were conducted to assess whether CO also affects the nitrogen-reduction branch beyond serving as a carbon precursor. Compared with pure CO2 feeding, adding a controlled amount of CO to the CO2 stream (CO/CO2 = 5:25) enhanced both acetamide yield and selectivity over Ni-SA/Cu-NCs (Fig. 5c). This enhancement suggests that increased local CO availability not only promotes *CCO formation but also facilitates the NO3− reduction cascade toward NH2-related intermediates, thereby identifying local CO availability as a key factor governing the acetamide pathway. To confirm that this behavior reflects the intrinsic reaction mechanism rather than local microenvironmental effects, the influence of local pH and mass transport was further examined. The in situ CO32−/HCO3− Raman signals were converted into an apparent local pH using calibrated K2CO3/KHCO3 standards and the Henderson-Hasselbalch relationship (Supplementary Figs. 37–39)47,48. The apparent local pH was nearly identical under individual CO2 reduction and CO2 + NO3− co-reduction, remaining within 11.00–11.03 across the examined potential range. Mass transport was further assessed by varying the CO2 flow rate (20–40 sccm) and the catholyte flow rate (0.6–1.5 mL min−1). The acetamide yield rate and FE remained essentially unchanged (Supplementary Fig. 40), indicating that external transport is not rate-controlling. The surface-intermediate evolution observed during co-reduction is therefore unlikely to originate mainly from local pH or transport effects, but is more consistent with the feedstock-dependent mechanism discussed above. Identification of key intermediates in acetamide synthesis. a Acetamide yield rate and Faradaic efficiency on Ni-SA, Cu-NCs and Ni-SA/Cu-NCs with different carbon and nitrogen sources. b, c Acetamide yield rate and Faradaic efficiency of Cu-NCs and Ni-SA/Cu-NCs at varying CO/CO2 molar ratios, respectively. d–f Potential-dependent changes in the integrated in situ Raman peak areas of COatop, δNO2⁻ and δsNH2, respectively, over the various catalysts under different reaction conditions. g–i In situ Raman peak areas of CO–CO, δsNH2 and νs(C–N), respectively, over the various catalysts at applied potentials during CO2 + NO3⁻ co-reduction. Data are presented as mean ± s.d. from three independent measurements (n = 3). Individual data points are overlaid in (a–c). Raman peak intensities are given in arbitrary units (a.u.). In situ Raman measurements were performed in a custom flow-type electrochemical cell coupled to a 532 nm laser; the electrolyte and feed conditions were as in Fig. 4. Source data for Fig. 5 are provided as a Source Data file. This feedstock-dependent intermediate evolution was further examined by potential-dependent in situ Raman spectroscopy over Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs under individual and co-reduction conditions (Figs. 5d–i, 6a and Supplementary Figs. 41–45). Relative to individual CO2RR, CO2 + NO3− co-reduction attenuated the adsorbed atop CO (COatop) signal across the catalysts (Fig. 5d and Supplementary Figs. 43–45), consistent with competitive adsorption and further consumption of *CO-derived intermediates. In parallel, the nitrogen-related bands evolved differently under co-reduction compared with individual NO3RR (Fig. 5e, f). For Ni-SA/Cu-NCs, the NO2− bending vibration (δNO2−) decreased most prominently, accompanied by an increase in the symmetric NH2 bending vibration (δsNH2). These changes suggest that adsorbed *CO is not only involved in C–C coupling but may also facilitate the NO3− reduction cascade toward NH2-related intermediates49. This interpretation is consistent with the enhanced intermediate-equivalent partial current densities under co-reduction conditions (Supplementary Fig. 46). Spectroscopic and theoretical investigation of the reaction mechanism. a In situ Raman and b in situ ATR-FTIR spectra of CO2 and NO3⁻ co-reduction on Ni-SA/Cu-NCs at different applied potentials. c Three-dimensional differential charge-density isosurfaces of Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs; yellow and cyan isosurfaces denote electron accumulation and electron depletion, respectively. Atoms are colored as follows: C, brown; N, light blue; Ni, grey; Cu, blue. d Surface electron localization function (ELF) and e CM5 charge distribution of Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs; in d, the color scale ranges from 0 to 1, with higher values indicating greater electron localization, and in e, the color scale represents the CM5 atomic charge in units of electron charge, as indicated by the color bars. f Calculated adsorption free energies of *CO2 and *NO3 on Ni-SA, Ni-SA/Cu-SA and Ni-SA/Cu-NCs. g–i Calculated free-energy diagrams for acetamide formation over Ni-SA/Cu-SA and Ni-SA/Cu-NCs. The computational models are representative of the active sites rather than definitive atomic configurations under operating conditions. Spectral intensities in a, b are given in arbitrary units (a.u.). In situ measurement conditions were as in Fig. 4. Source data for Fig. 6 are provided as a Source Data file. The correlated evolution of carbon- and nitrogen-derived intermediates was further evaluated by comparing the integrated peak areas of the CO–CO stretching, δsNH2 bending and C–N stretching vibrations (Fig. 5g-i). The CO–CO-related signal, obtained by deconvoluting the C ≡ O atop region at 2000–2100 cm−1 (Supplementary Fig. 47)50,51, and the δsNH2 signal were both more pronounced on Ni-SA/Cu-SA and Ni-SA/Cu-NCs than on Ni-SA. This trend suggests that Ni–Cu dual-site configurations promote the accumulation of key carbon- and nitrogen-containing intermediates under co-reduction conditions. Notably, the evolution of the ν(C–N) band followed the changes in the CO–CO- and NH2-related signals, supporting the involvement of CO-derived C2 intermediates and NH2-related species as precursors for C–N bond formation7,8. The stronger intermediate signals observed on Ni-SA/Cu-NCs are therefore consistent with its higher acetamide formation activity. To assess whether these Raman features originate from interfacial species rather than dissolved bulk species, potential-dependent Raman peak shifts were analyzed using the vibrational Stark effect (Supplementary Figs. 48–51). The linearly adsorbed *CO band showed a Stark tuning rate of 25.2 cm−1 V−1, while the NO2−-, NH2- and C–N-related bands exhibited tuning rates of 5.7, 9.5 and 3.9 cm−1 V−1, respectively. These potential-dependent shifts support the assignment of these features to interfacial species under electrochemical control, rather than to homogeneous bulk-solution contributions52,53. Complementary in situ ATR-FTIR measurements provided additional spectroscopic evidence for the evolution of surface intermediates during individual and co-reduction of CO2 and NO3− (Fig. 6b and Supplementary Figs. 52–56). During CO2 electrolysis on Ni-SA/Cu-NCs, several carbon-related bands appeared in the 1000–4000 cm−1 region, including features assignable to *CO (2080 cm−1), *COOH (1367 cm−1), *COH (1320 cm−1), *OCCOH (1252 cm−1), *HOCCOH (1164 cm−1) and *C–O (1062 cm−1). Their potential-dependent growth is consistent with the formation of CO-derived C–C-coupled intermediates along the CO2-to-C2 pathway (CO2 → *COOH → *CO → *COCO → *OCCOH → *HOCCOH). In parallel, NO3− reduction produced characteristic nitrogen-related bands, including features associated with *NO (1050 cm−1), *NH2 (1162 cm−1), symmetric * NO2− stretching (1254 cm−1), *NH2 bending (1634 cm−1), and symmetric/asymmetric *NH2 stretching modes (3224/3384 cm−1), consistent with stepwise NO3− reduction toward NH2-related intermediates. Because several CO2- and NO3−-derived bands overlap under co-reduction conditions, these assignments were based on potential-dependent spectral evolution, catalyst- and feed-dependent controls, catalytic trends, literature-reported vibrational assignments, and complementary Raman results, rather than on peak positions alone. To further evaluate the NHx-related assignments, in situ ATR-FTIR measurements were conducted in D2O -based electrolytes together with a reactant-free D2O control under otherwise identical conditions (Supplementary Fig. 57). In D2O, the overlapped band near 1641 cm−1 was redistributed into an ND2-related feature at ~1146 cm−1 and a D2O-derived band at ~1191 cm−1, while the broad features at 3224/3384 cm−1 shifted to the overlapping N-D/O-D stretching region at ~2320–2500 cm−1. By contrast, representative hydrogen-free bands, including the symmetric *NO2− stretching mode at ~1251 cm−1 and OCCO-type carbonaceous features at ~1545 cm−1, remained largely unchanged54,55. Compared with the reactant-free control, the D2O + NO3− spectra showed additional growth in the corresponding deuterated regions, supporting contributions from NDₓ surface intermediates formed during nitrate reduction. This assignment is also consistent with the lower-frequency NHₓ-related band at ~1156 cm−1, where water interference is less pronounced, and with the evolution of the δsNH2 signal in the in situ Raman spectra. Under co-reduction conditions, the attenuation of certain intermediate-related bands is consistent with competitive adsorption and the further consumption of surface-bound intermediates. Meanwhile, additional features at ~1057 and ~1410 cm−1 emerged and can be assigned to C–N-related vibrational modes, supporting the formation of C–N-coupled surface intermediates along the acetamide pathway. Their potential-dependent evolution, together with the feed-control experiments, further supports the involvement of C–N-coupled intermediates in acetamide formation. Notably, the C–N-related bands evolved in parallel with the NH2- and OCCO-related features across the catalyst series, consistent with the participation of NH2-related species and CO-derived C2 intermediates as precursors for C–N coupling. Similar intermediate features were also observed over Ni-SA and Ni-SA/Cu-SA, but their weaker intensities indicate a shared yet less efficient reaction pathway compared with Ni-SA/Cu-NCs. Although *CO-related features were clearly resolved by in situ Raman spectroscopy across the catalyst series, the corresponding CO bands in the ATR-FTIR spectra were weak or unresolved, likely owing to different Raman/infrared selection rules, adsorption-geometry effects and the relatively low CO sensitivity of the present Au-free ATR-FTIR configuration. Comparative in situ ATR-FTIR measurements of Cu-NCs and Ni-SA/Cu-NCs further showed that NH2-related features emerge at less negative potentials on Ni-SA/Cu-NCs (Supplementary Fig. 58), suggesting that locally generated *CO facilitates NO3− deoxygenation and hydrogenation. These spectroscopic observations support the cooperative generation of carbon- and nitrogen-containing intermediates on Ni-SA/Cu-NCs, which is a prerequisite for subsequent C–N coupling. This kinetic cooperation was further reflected in product-specific Tafel analysis performed in the low-overpotential region from −0.60 to −0.80 V vs. RHE (Supplementary Fig. 59). The NH2-generation slope decreased from 237 mV dec−1 for individual NO3−RR to 171 mV dec−1 for CO2 + NO3− co-reduction, while the CO-generation slope decreased from 253 mV dec−1 for individual CO2RR to 231 mV dec−1 under co-reduction conditions. These lower apparent slopes suggest accelerated formation of NH2- and CO-related intermediates in the coupled system. Together with the in situ spectroscopic results, this behavior is consistent with locally generated *CO facilitating the NO3− reduction cascade toward NH2-related species, while the subsequent consumption of *CO-derived intermediates through *CCO formation and C–N coupling may help maintain favorable CO-generation kinetics. Thus, CO2 + NO3− co-reduction proceeds through a kinetically coupled surface reaction network rather than a simple superposition of independent CO2RR and NO3RR pathways. The atomic-level origin of this spectroscopic and kinetic synergy was further examined by DFT calculations. Three representative active-site models were constructed: Ni–N4, Ni–N4/Cu–N4, and Ni–N4/Cu5 cluster (Supplementary Fig. 60). The thermodynamically optimized Cu5 cluster was selected as a representative model for the contiguous active sites and structural complexity of sub-nanometer Cu ensembles anchored on the defective N-doped carbon matrix, rather than as a definitive atomic configuration under operating conditions. All simulations were performed with the CP2K package, and data were processed and visualized using Multiwfn program and VMD software56,57. Differential charge density (DCD), electron localization function (ELF) and CM5 charge analyses were used to compare the local electronic structures of these models. The DCD isosurfaces and corresponding planar maps showed interfacial charge redistribution, with electron depletion mainly localized at the metal centers and electron accumulation distributed around the coordinating N atoms and interfacial region (Fig. 6c and Supplementary Fig. 61). Compared with Ni–N4 and Ni–N4/Cu–N4, the Ni–N4/Cu5 cluster exhibited more spatially extended charge redistribution, suggesting enhanced interfacial electronic coupling. The ELF results showed directional electron localization along the N–M (M = Ni, Cu) bonds (Fig. 6d), consistent with pronounced M–N interactions that stabilize the isolated metal sites. CM5 analysis further indicated electron transfer from the metal centers to the coordinating N atoms, resulting in partially oxidized metal sites (Fig. 6e). These electronic-structure characteristics modulate the local adsorption environment and are consistent with the distinct adsorption behavior of key CO2- and NO3−-derived intermediates. The calculated Gibbs free energies (ΔG) showed thermodynamically favorable CO2 adsorption on the examined Ni and Cu sites (Fig. 6f). By contrast, Ni–N4 and Ni–N4/Cu–N4 showed relatively weak *NO3 binding, whereas Cu sites in the Ni–N4/Cu5 cluster exhibited stronger *NO3 affinity. This trend suggests that Ni–Cu electronic coupling in Ni-SA/Cu-NCs tunes the local charge distribution and promotes NO3− adsorption/activation. Consistent with these adsorption trends, the calculated free-energy profiles for acetamide formation (Fig. 6g–i) and the corresponding optimized intermediate configurations (Supplementary Figs. 62–64), together with comparative DFT analysis of competing C2 pathways (Supplementary Figs. 65 and 66), provide energetic insight into the origin of acetamide selectivity. On Ni-SA/Cu-NCs, nucleophilic attack of *NH2 on *CCO to form *CCONH2 (ΔG ≈ − 1.38 eV) is thermodynamically more favorable than the competing protonation of *CCO to *CHCO (ΔG ≈ − 0.99 eV). Beyond these thermodynamic descriptors, transition-state calculations were performed to evaluate the kinetics of the key coupling steps on both catalysts (Supplementary Fig. 67). The activation barrier for *CO–CO dimerization decreased from 1.588 eV on Ni-SA/Cu-SA to 1.159 eV on Ni-SA/Cu-NCs, and that for *CCO–*NH2 coupling decreased from 1.851 eV to 0.524 eV. These lower barriers provide kinetic evidence that the Ni–Cu synergy accelerates both the C–C and C–N bond-forming steps. The reduction is consistent with the extended interfacial charge redistribution revealed by the differential charge-density and ELF analyses (Fig. 6c, d). The contiguous multi-atom Cu sites within the nanoclusters, together with the strong Ni–Cu electronic coupling, stabilize the coupled transition-state configurations more effectively than the isolated single-atom sites in Ni-SA/Cu-SA, thereby helping to rationalize the lower barriers and the higher acetamide activity on Ni-SA/Cu-NCs. This thermodynamic preference makes C–N coupling competitive with further hydrogenation of *CCO toward oxygenated C2 products, consistent with the suppressed C2+ selectivity observed under co-reduction conditions. Targeted low-potential and pulsed electrolysis experiments (Supplementary Fig. 68) further suggest a potential-dependent pathway shift mediated by local *CO coverage. At low *CO coverage (e.g., −0.2 V vs. RHE), direct C–N coupling involving isolated *CO-derived species remains accessible, whereas at −1.2 V vs. RHE, enhanced CO2-to-*CO conversion on Ni single atoms increases local *CO availability on adjacent Cu nanoclusters, favoring *CO dimerization to *CCO before C–N coupling with NH2-related species. Product-specific reaction orders extracted by varying the CO partial pressure at a fixed CO2 supply confirm this shift. Using the CO-feed-dependent partial current densities (Δj, referenced to the CO-free baseline), urea formation is first-order in the CO feed at −0.2 V (slope 0.96 ± 0.03, R2 = 0.995), whereas acetamide formation is second-order at −1.2 V (slope 1.94 ± 0.08, R2 = 0.993) (Supplementary Fig. 69), providing direct evidence that bimolecular *CO dimerization precedes C–N coupling along the acetamide pathway. This pathway shift is consistent with the feed-control experiments (Fig. 5a–c) and the correlated evolution of CO–CO, NH2, and C-N-related intermediates in the in situ spectra (Fig. 5g–i). Together with the in situ spectroscopic and feed-control results, these DFT calculations further support a tandem pathway involving *CO dimerization followed by *CCO–*NH2 coupling toward acetamide formation. This integrated experimental-theoretical picture indicates that the synergistic Ni–Cu dual sites optimize adsorption energetics, lower the activation barriers for *CO dimerization and *CCO–*NH2 coupling, and thereby help to rationalize the enhanced acetamide synthesis performance of Ni-SA/Cu-NCs. Discussion In summary, we have developed a synergistic dual-site Ni-SA/Cu-NCs catalyst that enables highly efficient electrocatalytic synthesis of acetamide from CO2 and NO3−. Structural characterization confirms the well-defined architecture comprising atomically dispersed Ni–N4 sites alongside Cu nanoclusters. Through comprehensive electrochemical evaluation and in situ spectroscopic analysis, we demonstrate that Ni single atoms selectively catalyze CO2-to-CO conversion, while adjacent Cu nanoclusters concurrently drive C–C coupling (from *CO to *CCO) and NO3− reduction to *NH2. The close spatial proximity of these functionally distinct sites facilitates efficient intermediate transfer and coupling, effectively mitigating active-site competition and suppressing side reactions. DFT calculations further suggest that the Ni–Cu electronic synergy enhances reactant adsorption and lowers the activation barriers for both *CO dimerization and the subsequent C–N bond-forming step. This work provides a practical green route for amide synthesis and a rational design strategy for multi-site catalysts that steer complex tandem electrosynthesis toward value-added organonitrogen products, while also suggesting that further gains in FE may be achieved by integrating this atomic-level dual-site design with microenvironment engineering or dynamic electrolysis regulation. Methods Chemicals Melamine (C3H6N6, 99%), L-alanine (C3H7NO2, 99%), nickel(II) nitrate hexahydrate (Ni(NO3)2·6H2O, 98%), copper(II) nitrate trihydrate (Cu(NO3)2·3H2O, 99%), ethanol (C2H6O, analytical reagent) and Nafion solution (5 wt%) were purchased from Sigma-Aldrich. Nitric acid (HNO3, 68%), hydrochloric acid (HCl, 37%), potassium chloride (KCl, analytical reagent, ≥99.5%), potassium nitrate (KNO3, ≥99%), potassium bicarbonate (KHCO3, ACS reagent, ≥99.7%), potassium hydroxide (KOH, analytical reagent, 85%), isopropanol (C3H8O, 99.8%), deuterium oxide (D2O, ≥99 atom% D), sodium 3-(trimethylsilyl)propionate-2,2,3,3-d4 (TSP, ≥98 atom% D), acetamide (CH3CONH2, analytical reagent, ≥99%), sodium nitrite (NaNO2, analytical reagent, 99%), ammonium chloride (NH4Cl, ≥99%), methylamine (CH3NH2, 40 wt% in H2O), dimethylamine ((CH3)2NH, 40 wt% in H2O), ethylamine (C2H5NH2, 68.0–72.0 wt% in H2O) and sodium 4-nitrobenzoate (C7H4NNaO4, ≥95%) were purchased from Aladdin. The gas diffusion layer (Sigracet 39BC) was obtained from Aida Hengsheng. Nickel foam was purchased from Suzhou Sinero Technology Co., Ltd. The Ag/AgCl reference electrode and platinum wire were obtained from Wuhan Gaoss Union Technology Co., Ltd. The Fumasep FAA-3-PK-75 anion-exchange membrane was purchased from Fuel Cell Store. CO2, CO, H2 and Ar gases (99.999% purity) were supplied by Shanghai Material Special Gas Co., Ltd. Deionized water (18.2 MΩ cm) was used in all experiments. All chemicals were used as received without further purification unless otherwise specified. Synthesis of catalysts The Ni single-atom catalyst (Ni-SA) was prepared via a two-step pyrolysis process. Briefly, a homogeneous precursor was obtained by ball-milling a mixture of melamine (4 g), L-alanine (2 g), and Ni(NO3)2·6H2O (5 mg) for 2 h, followed by the addition of an ethanol/HCl solution (10 mL, 5:1 v/v) and further milling for 1 h. After drying at 80 °C, the resulting powder was subjected to pyrolysis under Ar atmosphere, first at 550 °C for 2 h, then at 900 °C for another 2 h. The product was subsequently treated with 1 M HNO3 for 24 h to remove metallic nanoparticles, then washed and dried to yield the final Ni-SA material. The Ni-SA/Cu-NCs catalyst was synthesized through an impregnation-annealing method. In a typical procedure, 30 mg of as-prepared Ni-SA support was dispersed in an aqueous solution containing 10 mg of Cu(NO3)2·3H2O, stirred for 12 h, dried, and then annealed at 550 °C under 5% H2/Ar for 1 h. The final product was collected after washing and drying. For comparison, the Ni-SA/Cu-SA catalyst was prepared following a similar impregnation-annealing route but with a lower Cu precursor amount (5 mg). To ensure atomic dispersion of Cu, the pyrolyzed material was further treated with 1 M HNO3 to remove any Cu nanoparticles or clusters. For comparative evaluation, Fe-SA and Co-SA were synthesized following the same protocol as Ni-SA, except that the nickel precursor was replaced with 5 mg of Fe(NO3)3·9H2O or 5 mg of Co(NO3)2·6H2O, respectively. After two-step pyrolysis and acid washing, Fe-SA and Co-SA were obtained. Fe-SA/Cu-NCs and Co-SA/Cu-NCs were subsequently prepared using the same impregnation-annealing procedure described above, with 30 mg of the corresponding single-atom support and 10 mg of Cu(NO3)2·3H2O, followed by annealing at 550 °C under 5% H2/Ar. Pure Cu-NCs were synthesized using a metal-free nitrogen-doped carbon (NC) support in place of the single-atom support. The NC support was prepared by ball-milling melamine (4 g) and L-alanine (2 g), followed by pyrolysis without any transition-metal precursor. Subsequently, 30 mg of NC was subjected to the same Cu impregnation-annealing procedure using 10 mg of Cu(NO3)2·3H2O, followed by annealing at 550 °C under 5% H2/Ar. Materials characterization The morphology of the catalysts was examined using transmission electron microscopy (TEM, JEM-2100F) and aberration-corrected TEM (AC-TEM, JEM-ARM200F). X-ray diffraction (XRD) patterns were collected on a PANalytical X-Pert PRO MPD diffractometer with Cu Kα radiation (λ = 1.5406 Å) to identify crystalline phases. X-ray photoelectron spectroscopy (XPS) measurements were performed on an ESCALAB 250Xi system with Al Kα radiation, and all binding energies were referenced to the C 1 s peak at 284.8 eV. Specific surface area, pore-size distribution, and pore volume were determined from N2 adsorption-desorption isotherms measured on a QUADRASORB EVO analyzer, with the Barrett-Joyner-Halenda (BJH) model applied to the desorption branch. X-ray absorption spectroscopy (XAS) was carried out at the BL13SSW beamline of the Shanghai Synchrotron Radiation Facility (SSRF). XANES and EXAFS data were processed using Athena, and quantitative EXAFS fitting was performed using Artemis. The EXAFS spectra were Fourier transformed and fitted in R space using Cu–N/Cu–Cu or Ni–N/Ni–Ni scattering paths to extract the local coordination parameters of the metal sites. The resulting fitting parameters are summarized in Supplementary Tables 2 and 3. In situ Raman spectroscopy In situ Raman measurements were performed on a LabRAM HR Evolution confocal Raman spectrometer equipped with a 532 nm laser and coupled to a custom flow-type electrochemical cell. The laser was focused through a 50× long-working-distance objective, and spectra were collected using a 600 gr mm−1 grating. The laser output was set to 10% of the nominal instrument power. This low-power setting was maintained throughout the measurements to reduce local heating and beam-induced damage. Spectra were acquired from −0.8 to −1.6 V versus RHE in 0.2 V increments after 30 s of potential stabilization, using 10 accumulations with an exposure time of 5 s per spectrum over the 200–4000 cm−1 range. Owing to the approximate 1/λ4 dependence of Raman scattering intensity, 532 nm excitation affords higher scattering efficiency than longer wavelengths such as 633 nm, thereby enabling shorter acquisition times and improved temporal resolution for tracking surface-intermediate evolution during CO2/NO3− co-reduction (Supplementary Fig. 70). No external Raman-enhancing substrate was introduced. The investigated catalysts feature atomically dispersed sites and ultrasmall nanoclusters, which are not expected to exhibit classical localized surface plasmon resonance. Nevertheless, the observed signals of adsorbed intermediates can be reasonably attributed to an intrinsic electrochemical surface-enhanced Raman spectroscopy (EC-SERS) effect58–61, arising primarily from potential-dependent chemical enhancement associated with charge-transfer interactions between adsorbates and active sites under electrochemical reduction conditions, together with the relatively high surface coverage of key intermediates under reaction conditions. Quantitative analysis of in situ Raman spectra To quantitatively evaluate the potential-dependent evolution of reaction intermediates (Fig. 5d–i), the raw in situ Raman spectra were processed using LabSpec 6 software (HORIBA Scientific). Baseline correction and smoothing were first applied to all spectra to remove background fluorescence and solvent interference. For well-resolved peaks, peak areas were obtained by direct integration within defined wavenumber windows. For overlapping spectral regions, particularly the C ≡ O atop region (2000–2100 cm−1), multi-peak fitting was performed using Gaussian-Lorentzian mixed functions. The peak positions, widths (FWHM), and intensities were iteratively optimized until the fitted curve closely matched the experimental spectrum (R2 > 0.99). The integrated areas of the fitted components were then extracted and used for the semi-quantitative comparison of key intermediates across different catalysts. For apparent local-pH analysis, K2CO3/KHCO3 standard solutions with known CO32−/ HCO3− concentration ratios were measured under the same Raman configuration. The integrated areas of the CO32− band at 1072 cm−1 and the HCO3− band at 1356 cm−1 were obtained after baseline correction. The Raman sensitivity factor S was determined from the slope of the origin-constrained linear fit between CCO32−/CHCO3− and ACO32−,1072/AHCO3−,1356. The apparent local pH was calculated according to the Henderson-Hasselbalch relationship: where ACO32−,1072 and AHCO3−,1356 are the integrated Raman peak areas of carbonate and bicarbonate, respectively, and pKa,2 is the second dissociation constant of carbonic acid. In situ FTIR measurements In situ FTIR measurements were carried out on a Nicolet iS50 spectrometer equipped with an H-type electrochemical cell (Gaossunion Technology Co., Ltd.) fitted with a ZnSe window to track surface intermediates during electrocatalysis. To avoid background catalytic interference and spectral overlap, the working electrode for the in situ FTIR measurements was fabricated without an Au enhancement layer or polymer binders. The catalyst powder was dispersed in pure isopropanol, directly airbrushed onto the reflective surface of the ZnSe crystal with a catalyst loading of 0.1 mg cm−2, and dried naturally in air. Conductive copper tape was attached to the outer edge of the crystal to connect the catalyst layer to the external circuit without interfering with the optical path, thereby forming the working electrode. Spectra were collected over the range of 650–4000 cm−1 with a resolution of 8 cm−1, each representing an average of 64 scans. A background spectrum was recorded under open-circuit conditions prior to each measurement. Potential-dependent spectra were acquired from −0.8 to −1.6 V vs. RHE in 0.2 V increments. Electrochemical measurements All electrochemical experiments were conducted using an Autolab PGSTAT302N workstation with NOVA software, in a custom flow cell with a 1 cm2 geometric electrode area. Unless otherwise stated, all electrochemical and in situ spectroscopic measurements were performed at room temperature (25 °C). The working electrode was prepared by airbrushing catalyst ink onto a gas diffusion layer (Sigracet 39BC). Typically, 5 mg of catalyst was dispersed in 0.9 mL of isopropanol and 0.1 mL of 5 wt% Nafion solution, followed by ultrasonication to form a homogeneous ink. After coating and drying, the catalyst loading was determined from the mass difference of the gas diffusion electrode before and after catalyst deposition. The catalyst loading was 0.25 mg cm⁻2 unless otherwise specified. For the enlarged-area flow-cell test, a 25 cm2 gas diffusion electrode was prepared using the same procedure, with a catalyst loading of 0.5 mg cm⁻2; CO2 was supplied at 50 sccm while maintaining the same electrolyte flow rate as used in the standard flow-cell measurements. Nickel foam and Ag/AgCl (saturated KCl) were used as the counter and reference electrodes, respectively. The cathodic and anodic compartments (with internal volumes of ~0.5 cm3 and 2 cm3, respectively) were separated by an anion-exchange membrane (Fumasep FAA-3-PK-75, thickness ~75 μm). The membrane size slightly exceeded the electrode geometric area to ensure adequate sealing, and was soaked in the standard catholyte for 24 h before use. For standard CO2 + NO3− co-reduction measurements, the catholyte was 1.0 M KCl containing 0.02 M KNO3, and 1.0 M KHCO3 was used as the anolyte. Catholytes and anolytes were freshly prepared on the day of each experiment by dissolving the corresponding salts, acids or bases in deionized water under magnetic stirring at room temperature until fully dissolved. The KHCO3 anolyte was saturated with CO2 before use to stabilize its pH. All electrolytes were used within the same day of preparation and were not stored for extended periods. For pH-dependence investigations, acidic (0.1 M HCl + 1.0 M KCl + 0.02 M KNO3) and alkaline (0.1 M KOH + 0.9 M KCl + 0.02 M KNO3) catholytes were prepared with comparable supporting-electrolyte backgrounds, using the neutral 1.0 M KCl + 0.02 M KNO3 catholyte as the reference. Both electrolytes were circulated at 0.8 mL min−1 using peristaltic pumps, with CO2 supplied to the cathode chamber at 30 sccm. The gas flow rate was controlled and monitored using a calibrated mass-flow controller, and the set flow rate was used for gaseous-product quantification. Control experiments for individual CO2 reduction used nitrate-free 1.0 M KCl catholyte, while NO3− reduction tests employed 0.02 M KNO3 + 1.0 M KCl without CO2 feeding. Electrocatalytic performance was evaluated by chronoamperometry at potentials ranging from −0.8 to −1.6 V vs. RHE in 0.2 V increments, with each electrolysis conducted for 10 min before product collection. All potentials were converted to the RHE scale using the equation: ERHE = EAg/AgCl + 0.197 V + 0.0591×pH, where 0.197 V is the standard potential of the saturated-KCl Ag/AgCl reference electrode versus the standard hydrogen electrode at 25 °C. The freshly prepared catholyte and 1.0 M KHCO3 anolyte had bulk pH values of 6.2 ± 0.1 and 8.3 ± 0.1, respectively, measured with a calibrated pH meter; the catholyte pH was used for RHE conversion. CO2 was supplied through the gas diffusion electrode rather than by purging the catholyte. The uncompensated resistance was determined from the high-frequency intercept of Nyquist plots recorded before electrolysis. For Ni-SA/Cu-NCs under standard CO2 + NO3− co-reduction conditions, the resistance was 1.413 ± 0.003 Ω (n = 3). All potentials are reported without iR correction. Mass-transport control experiments were performed at −1.2 V vs. RHE by varying either the CO2 gas flow rate from 20 to 40 sccm or the catholyte flow rate from 0.6 to 1.5 mL min−1, while keeping the other electrolysis conditions unchanged. Acetamide FE and yield rate were quantified using the same procedures described above. LSV was performed from −0.2 to −1.8 V vs. RHE at 10 mV s−1. The relative electrochemical active surface area was evaluated from the double-layer capacitance (Cdl), obtained from cyclic voltammetry in the non-Faradaic region (0.2 to 0.3 V vs. RHE) at scan rates of 10–200 mV s⁻¹. Cdl was determined as the slope of the charging current density against scan rate, and was used as a relative measure of the electrochemical active surface area to compare the catalysts. Substrate-scope experiments The substrate scope was evaluated in the same flow-cell configuration under CO2 co-reduction conditions. Unless otherwise specified, the nitrogen source was replaced by methylamine, dimethylamine, ethylamine or sodium 4-nitrobenzoate (0.08 M), while the other electrolysis parameters were kept consistent with the standard measurements. For sodium 4-nitrobenzoate, a mildly alkaline catholyte was used to ensure adequate substrate solubility. The amide products were identified and quantified by 1H NMR using TSP as the internal standard, following the same procedure as for acetamide, and the corresponding Faradaic efficiencies were calculated accordingly. Product quantification and analysis Liquid products were analyzed immediately after electrolysis. NH3 was quantified by the indophenol blue method (UV-vis, 655 nm), and NO2− was determined via the Griess reaction (UV-vis, 540 nm). Acetamide, formate, acetate, ethanol and propanol were characterized by 1H NMR (600 MHz, Bruker). For NMR, 500 μL of electrolyte was mixed with 50 μL of D2O containing 5 mM TSP as an internal standard. Spectra were collected using presaturation, with concentrations determined by integrating peaks relative to TSP. Gaseous products (CO, H2, CH4, C2H4) were monitored in real time using an online gas chromatograph (Agilent 7890B) equipped with TCD (for H2) and FID with a methanizer (for hydrocarbons, CO, and CO2). The effluent gas was automatically sampled every 10 min with argon as the carrier gas, and the chromatogram corresponding to each electrolysis was used for gaseous-product quantification. All measurements were performed in triplicate and quantified using calibration curves from standard solutions or reference gases. The yield and Faradaic efficiency for liquid products were calculated using the following equations: Where n is the number of electrons transferred to form one molecule of product. F is the Faraday constant (96485 C mol−1). C is the measured product concentration (mol L−1). V is the total catholyte volume (L). Q is the total charge passed during electrolysis (C). t is the total electrolysis time (h), and m is the catalyst loading (g). To evaluate how efficiently the generated carbon- and nitrogen-containing intermediates were directed toward acetamide formation, we further defined the partial current densities for NH2/CO generation and consumption and calculated their normalized utilization values. For the nitrogen pathway, the partial current density corresponding to NH2-equivalent generation is defined as: where jNH3 is the partial current density for NH3 production and jNH2,consumption is the partial current density of NH2-equivalent consumed in acetamide formation. The normalized NH2 utilization is then calculated as: For the carbon pathway, the partial current density corresponding to CO-equivalent generation is defined as: where jCO is the measured partial current density for gaseous CO production, and jCO,consumption is the total partial current density of the downstream CO-derived products. The jCO,consumption is calculated according to the following equation: Thus, jCO,generation represents the total partial current density of CO-equivalent intermediates generated from CO2 reduction, including both released gaseous CO and the downstream CO-derived products. The normalized CO utilization is calculated as: Reaction-order analysis Apparent reaction orders with respect to CO were determined by varying the CO flow rate (0.5–4.0 sccm, below the CO saturation regime) while keeping the CO2 supply (25 sccm) and total flow rate (30 sccm) constant, with Ar as the balance gas. Measurements were performed at −0.2 V vs. RHE for urea and −1.2 V vs. RHE for acetamide. For each condition, the product-specific partial current density was obtained from the quantified product amount and the electron-transfer number, and the baseline-corrected value (Δjᵢ) was calculated by subtracting that measured under CO-free conditions (CO2-only, with Ar replacing CO at the same total flow rate). The apparent reaction order was extracted from the slope of the linear fit of log(Δjᵢ) versus log(pCO), where pCO is the CO partial pressure in the feed. The Tafel slopes, vibrational Stark-effect tuning rates and apparent local pH values were each obtained from a single measurement; the reported uncertainties represent the standard error of the linear regression fit, not the standard deviation across independent repeats. Computational details DFT calculations were carried out using the CP2K package with the QUICKSTEP module. The Perdew-Burke-Ernzerhof (PBE) functional and DZVP-MOLOPT-SR-GTH basis set were employed to describe electron interactions56,62,63. The plane-wave energy cutoff was set to 800 Ry, and geometry optimizations were considered converged when the maximum force fell below 4.5 × 10−4 Hartree/Bohr and the total energy variation was within 1 × 10−6 Ry. Van der Waals interactions were incorporated using the DFT-D3(BJ) method with Becke-Johnson damping64. For post-processing analysis, Multiwfn software was utilized to generate local integral curves and charge displacement profiles from the computed electronic structures57. Transition-state calculations for *CO–CO to *COCO and *CCO–*NH2 to *CCONH2 were performed using the climbing-image nudged elastic band (CI-NEB) method with the same DFT settings as the geometry optimizations. The optimized initial and final states were used to construct the reaction pathways. The activation barrier was defined as the energy difference between the highest-energy image along the optimized pathway and the corresponding initial state. The atomic coordinates of the optimized structures for all computational models are provided in Supplementary Data 1. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Source Data