Specific construction of asymmetric carbon-nickel-chlorine single-atom sites via carbon vacancy engineering for efficient CO₂ electroreduction

Abstract

Breaking the structural symmetry of active sites in single-atom catalysts (SACs) allows efficient regulation of the electron distribution around the metal centers, holding great promise for promoting their performance in electrocatalytic carbon dioxide reduction reaction (ECO₂RR). Herein, we propose a vacancy-engineering strategy for constructing asymmetric carbon-nickel-chlorine (C–Ni–Cl) sites in Ni SAC (Ni₁-C/Cl). In strongly acidic media (pH=1), Ni₁-C/Cl achieves Faradaic efficiency over 98% for carbon monoxide (CO) product at the operated current density of 500 mA cm⁻². In situ X-ray absorption spectra reveal that during electrocatalysis, the C₃–Ni–Cl sites exhibit potential-dependent structure evolutions, which can optimize their adsorption configurations for the reactive species. Theoretical calculations demonstrate that the Ni–C/Ni–Cl co-coordination induces the asymmetric electron distribution in C₃–Ni–Cl sites, resulting in the regulation of the electronic properties of the Ni centers, thereby optimizing the reaction pathway of CO₂-to-CO on these single-atom sites. This work extends the synthesis of SACs containing asymmetric single-atom sites, provides insights into designing industrial-oriented electrocatalysts toward other important electrocatalytic reactions.

Introduction

Electrocatalytic carbon dioxide reduction reaction (ECO₂RR) offers a promising way for the sustainable conversion of CO₂ into valuable chemicals and fuels, facilitating the closing of carbon cycle and accelerating the carbon neutrality process^1,2,3. Single-atom catalysts (SACs), as new-generation electrocatalysts, have attracted great attention for electrocatalytic CO₂ conversion, owing to their maximized atom utilization and fully exposed active sites^4,5,6. Among various SACs, carbon-supported nickel (Ni) SACs, especially those containing single-atom Ni sites with Ni–N₄ coordination structure, have exhibited favorable selectivity in ECO₂RR to carbon monoxide (CO)^7,8,9,10,11. However, the symmetric electron distribution in the Ni–N₄ moieties leads to the unfavorable adsorption and activation of *COOH intermediate on central Ni atoms, resulting in unsatisfactory activities for ECO₂RR on these Ni SACs¹².

In order to break this bottleneck, great efforts have been devoted in engineering the local coordination environments of atomic Ni sites^13,14. Recently, the construction of coordination-asymmetric Ni single-atom sites provides an effective way to regulate the electronic properties of Ni centers and enhance their adsorption/activation capacity for *COOH intermediate^14,15,16. Moreover, during electrocatalysis, the geometric distortion in these asymmetric sites can induce potential-dependent evolutions of their coordination structures, resulting in the rearrangements of electron distribution at these sites and thus promoting the adsorption/activation/desorption of reactive species¹⁷. Heteroatom doping, is widely applied in breaking the structural symmetry of the Ni–N₄ hosts through forming Ni–N_x–M_y (M = O, S, Cl,…) asymmetric coordination structures^18,19,20. The electronegativity differences between the dopant atoms and the host N atoms regulate the electron arrangement around the Ni centers, thus influencing their adsorption and activation capacities for the reactive species. However, in these asymmetric single-atom sites, the strong electron transfer between the Ni centers and the adjacent Lewis acidic N ligands induces the electron localization at the N ligands, which limits the flexible modulation of the electronic structures of the Ni centers from the dopant atoms. In our opinion, due to the weaker electronegativity of C element than N element, the Ni–C₄ configuration can be an alternative of Ni–N₄ as the host for constructing the asymmetric single-atom sites. In this host, the moderate electron transfer in the Ni–C bonding reduces the electron localization, which allows the dopant atoms to more effectively modulate the asymmetry of the electron distribution in these single-atom sites, to achieve more efficient electrochemical CO₂ conversion.

As a proof of concept, in this work, we selected Cl element as a coordination symmetry regulator for constructing asymmetric C–Ni–Cl single-atom sites in Ni SAC (Ni₁-C/Cl). The significant electronegativity difference between C and Cl element (0.61, Pauling electronegativity) can strengthen the asymmetry of the electron distribution in the C–Ni–Cl single-atom sites. The synthesis of Ni₁-C/Cl was based on a carbon vacancy engineering strategy. In this strategy, abundant Cl-doped carbon vacancies served as trapping and anchoring sites for binding the migrated Ni species and forming asymmetric C–Ni–Cl moieties. X-ray absorption fine structure (XAFS) spectroscopy and aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (AC HAADF-STEM) revealed that the Ni species in Ni₁-C/Cl existed as isolated single atoms, with coordination structure of C₃–Ni–Cl. In strongly acidic media (pH = 1), Ni₁-C/Cl achieved Faradaic efficiency (FE) over 98% for CO product (FE_CO) at the operated current density of 500 mA cm⁻². Even in an acidic medium containing 0.02 M K⁺, Ni₁-C/Cl still achieved FE_COs greater than 95% in the current density range of 150 to 500 mA cm⁻². In situ XAFS spectra revealed that during electrocatalysis, the C₃–Ni–Cl single-atom sites exhibited potential-dependent structure evolutions, which can self-optimize their adsorption configurations for the reactive species, thus accelerating the generation and desorption of CO product¹⁷. Density functional theory (DFT) calculations and in situ attenuated total reflectance-surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) measurements indicated that the asymmetric electron distribution in C₃–Ni–Cl sites optimized the reaction pathway of CO₂-to-CO and promoted CO desorption on the active Ni centers. We also extended the application of Ni₁-C/Cl in a Ni₁-C/Cl@cuprous oxide tandem system for ethylene (C₂H₄) electrosynthesis. This system achieved C₂H₄ selectivity of ~33%, with a corresponding C₂H₄ partial current density above 0.37 A cm⁻². Our work demonstrated a vacancy-engineering strategy for the specific construction of asymmetric Ni single-atom sites with C₃–Ni–Cl coordination structure, and revealed their catalytic mechanisms in promoting CO₂ electroreduction. These findings provided insights into the development of industrial-oriented electrocatalysts for other important electrocatalytic reactions.

Results

Synthesis and characterization

The theoretical criterion and schematic diagram of this vacancy-engineering strategy for specific constructing C–Ni–Cl asymmetric single-atom sites were shown in Fig. 1a. In this strategy, the construction of abundant Cl-doping carbon vacancies facilitated the trapping and anchoring of isolated Ni atoms. In order to verify the feasibility of this strategy, we systematically synthesized Cl-doped vacancy-rich/vacancy-deficient carbon supported Ni catalysts, named as Ni₁-C/Cl and Ni_n-C/Cl, respectively. The synthesis of Ni₁-C/Cl involved the encapsulation of Ni salt in agarose aerogel, and a dual molten salt-assisted (ZnCl₂/KCl) pyrolysis of the Ni/agarose complex. Unlike Ni₁-C/Cl, in the synthesis of Ni_n-C/Cl, only KCl was selected as the molten salt template. We also prepared Ni-free catalyst through similar synthetic procedure with Ni₁-C/Cl, denoted as Zn₁-C/Cl, for comparison.

Fig. 1: Synthesis strategy and structural characterizations.

a Pauling electronegativity scales of different elements and schematic diagram of the vacancy-engineering strategy for constructing asymmetric single-atom sites. b XRD patterns of Ni₁-C/Cl, Zn₁-C/Cl, and Ni_n-C/Cl. c EPR spectra of Ni₁-C/Cl, Zn₁-C/Cl, and Ni_n-C/Cl. d Raman spectra of Ni₁-C/Cl, Zn₁-C/Cl, and Ni_n-C/Cl. In order to obtain an accurate I_D/I_G, the two broad peaks in the Raman spectra were deconvoluted into five Lorentzian peaks⁴³. e HAADF-STEM image of Zn₁-C/Cl. f HAADF-STEM image of Ni₁-C/Cl. g HAADF-STEM image of Ni_n-C/Cl. Source data are provided as a Source data file.

The X-ray diffraction (XRD) patterns in Fig. 1b only exhibited two broad graphite peaks, indicating the amorphous state of the metal species in these three samples²¹. In the electron paramagnetic resonance (EPR) spectra (Fig. 1c), the single Lorentzian lines with a g value of 2.002 were derived from the unpaired electrons in the carbon vacancies²². The stronger peak intensities in the Lorentzian lines of Zn₁-C/Cl and Ni₁-C/Cl indicated their higher vacancy concentrations than that of Ni_n-C/Cl²³. Moreover, the higher vacancy concentrations in Zn₁-C/Cl and Ni₁-C/Cl were also confirmed by the Raman spectra (Fig. 1d), in which Zn₁-C/Cl and Ni₁-C/Cl demonstrated higher ratios of disorder band to graphite band (I_D/I_G) compared to Ni_n-C/Cl²³. The increased vacancy concentrations in Zn₁-C/Cl and Ni₁-C/Cl stemmed from the sufficient activation and etching of the carbon precursors by ZnCl₂ with a low melting point during pyrolysis²⁴. Then we conducted TEM measurements to characterize the microscopic morphologies of these samples. As can be seen from Supplementary Fig. 1, the carbon carrier in Ni_n-C/Cl exhibited a three-dimensional porous structure. Different from Ni_n-C/Cl, the carbon carriers in Zn₁-C/Cl and Ni₁-C/Cl were assembled from stacked carbon fragments (Supplementary Figs. 2, 3). The corresponding high-resolution TEM images of these samples demonstrated the absence of Ni nanoparticles on their carbon carriers (Supplementary Figs. 1d, 2d, and 3d). The AC HAADF-STEM technique enabled identifying the dispersion of metal species at atomic-level resolution^25,26. In the HAADF-STEM images of Zn₁-C/Cl (Fig. 1e and Supplementary Fig. 4), isolated bright dots assigned to Zn single atoms were observed on the carbon carriers. The energy-dispersive X-ray spectroscopy (EDS) mapping of Zn₁-C/Cl (Supplementary Fig. 5) showed the uniform distribution of C, Cl, and Zn elements on the carbon carriers. These observations suggested that during pyrolysis, the molten or evaporated Zn species were captured by the vacancy-rich carbon carriers, and consequently formed single-atom Zn sites. In Ni₁-C/Cl, Ni/Zn single atoms were homogeneously dispersed on the carbon carriers (Fig. 1f and Supplementary Figs. 6, 7), which further demonstrated the strong capture effect of the vacancy-rich carbon substrates to these metal species. Unlike Zn₁-C/Cl and Ni₁-C/Cl, the vacancy-deficient carbon carriers in Ni_n-C/Cl induced the formation of atomically dispersed Ni clusters and trace amounts of isolated Ni single atoms. These Ni clusters were assembled from dozens of Ni atoms, and exhibited an amorphous state (Fig. 1g, Supplementary Figs. 8, 9). An inductively coupled plasma optical emission spectrometer (ICP-OES) was used to quantify the metal loadings in these samples (Supplementary Table 1). The Zn loadings in Zn₁-C/Cl and Ni₁-C/Cl were 0.32 and 0.25 wt%, respectively. Ni₁-C/Cl and Ni_n-C/Cl exhibited similar Ni loadings of 1.88 and 1.71 wt%, respectively.

The aforementioned characterizations validated the feasibility of our vacancy-engineering strategy for preparing carbon-supported SACs, as well as confirmed the importance of vacancy sites in carbon carriers for anchoring the migrated metal species and constructing single-atom sites.

X-ray photoelectron spectroscopy (XPS) was applied to investigate the chemical states of different elements in these three samples. The high-resolution C 1s XPS spectra in Supplementary Fig. 10 were deconvoluted into four peaks at 284.4, 286.1, 288.6 and 290.8 eV, assigning to C = C/C–C, C–O/C–Cl, C = O moieties and C 1s satellite peaks, respectively²⁷. The C KLL first-derivative Augar spectra of these three samples were shown in Supplementary Fig. 11. The energy difference between the intensity maximum and minimum in the C KLL first-derivative Augar spectrum was defined as D-parameter for evaluating the degree of C sp²/sp³ hybridizations in the carbon materials²⁸. The vary sp³-like materials possessed D-parameter values from 11 to 13 eV, the D-parameter values of vary sp²-like materials located in the range of 21 to 23 eV²⁹. The calculated D-parameter values of these three samples were demonstrated in Supplementary Table 2. Apparently, both Zn₁-C/Cl and Ni₁-C/Cl had lower D-parameter values than Ni_n-C/Cl, indicating the higher proportion of sp³-hybridized carbon atoms in their substrates, which represented the higher degree of intrinsic defects in their carbon carriers³⁰. In the high-resolution Cl 2p XPS spectra (Supplementary Fig. 12), the binding energies of Ni/Zn-Cl moieties in Ni₁-C/Cl and Zn₁-C/Cl were much lower than that in Ni_n-C/Cl, indicating the existence of strong electron transfer between the doped Cl species and the Ni/Zn single atoms in Ni₁-C/Cl and Zn₁-C/Cl³¹. The high-resolution O 1 s XPS spectra were deconvoluted into two peaks, representing C = O and C–O moieties, respectively (Supplementary Fig. 13), which excluded the presence of Ni/Zn-O bonding in these samples. As can been seen from the high-resolution Zn 2p XPS spectra in Supplementary Fig. 14, the Zn species in Ni₁-C/Cl and Zn₁-C/Cl possessed close chemical states. Due to the different existence form of Ni species in Ni₁-C/Cl and Ni_n-C/Cl, in the high-resolution Ni 2p XPS spectra, Ni₁-C/Cl showed a higher binding energy of Ni 2p_3/2 compared to that of Ni_n-C/Cl. (Supplementary Fig. 15).

We then applied XAFS spectroscopy to acquire more information about the electronic properties and the local coordination environments of the metal species in these three samples. The Zn K-edge X-ray absorption near-edge structure (XANES) spectra were shown in Fig. 2a. The absorption edges of Zn₁-C/Cl and Ni₁-C/Cl located between those of Zn foil and ZnO, which indicated that the valence states of Zn species in these two samples were between 0 to +2. In the Zn K-edge Fourier-transformed extend X-ray absorption fine structure (FT-EXAFS) spectra (Fig. 2b), the broad peaks of Zn₁-C/Cl and Ni₁-C/Cl at ~1.75 Å represented the co-coordination of C/Cl atoms with the central Zn atoms in the first shell³². No peaks attributed to Zn-Zn bonds were observed in the FT-EXAFS spectra of both Zn₁-C/Cl and Ni₁-C/Cl, suggesting that the Zn species in Zn₁-C/Cl and Ni₁-C/Cl existed in the form of single atoms. The fitted EXAFS results at Zn K-edge indicated that in Zn₁-C/Cl and Ni₁-C/Cl, most single-atom Zn sites were existed with asymmetric coordination structure of C₃-Zn-Cl (Fig. 2c, Supplementary Fig. 16, and Table 3). As can been seen from the Ni K-edge XANES spectra (Fig. 2d), in Ni_n-C/Cl, the valence state of Ni species was slightly higher than zero valence. In Ni₁-C/Cl, The Ni species showed a higher valence state than those in Ni_n-C/Cl, stemming from the strong electron transfer between the Ni single atoms and the surrounding C/Cl atoms. The corresponding Ni K-edge FT-EXAFS spectra were shown in Fig. 2e. For Ni foil, the main peak at ~2.19 Å was attributed to Ni-Ni bond in the first shell. Different from Ni foil, the average Ni-Ni bond length in Ni_n-C/Cl was shorter, with the location of primary peak at ~2.09 Å in the FT-EXAFS spectra. For Ni₁-C/Cl, two peaks located at ~1.25 Å and ~1.91 Å were assigned to Ni-C and Ni–Cl bonds, respectively, demonstrating the co-coordination of C/Cl atoms with the central Ni atoms in single-atom Ni sites²⁷. The fitted EXAFS results of Ni₁-C/Cl and Ni_n-C/Cl at Ni K-edge verified that the single-atom Ni sites in Ni₁-C/Cl existed with C₃–Ni–Cl coordination structure (Fig. 2f, Supplementary Fig. 17, and Table 4). The Ni–Ni coordination number for Ni_n-C/Cl was much lower than 12 (≈7.4, Supplementary Table 4), due to the structural disorders in Ni clusters. This distinction suggested that the Ni clusters in Ni_n-C/Cl had a different structure from typical crystalline Ni metal, in agreement with the HAADF-STEM results. The wavelet transform (WT) spectra of Zn₁-C/Cl and Ni₁-C/Cl at Zn K-edge exhibited similar intensity maximums at ~5.46 Å⁻¹, shifted toward a negative direction compared to that of Zn foil (~6.53 Å⁻¹), indicating the absence of aggregated Zn species in Zn₁-C/Cl and Ni₁-C/Cl (Fig. 2g–i). In the Ni K-edge WT spectra of Ni foil and Ni_n-C/Cl, their intensity maximums located at a similar position (~7.30 Å⁻¹), showing the existence of Ni-Ni bonds in Ni clusters (Fig. 2j, k). In the WT spectrum of Ni₁-C/Cl at Ni K-edge, only one intensity maximum at ~3.85 Å⁻¹ was observed, further confirming that in Ni₁-C/Cl, the Ni species existed as isolated single atoms (Fig. 2l). Taking C₄ and C₃Cl vacancy sites as examples, we investigated the contribution of Cl-doping to the anchoring of Ni single atoms on the carbon vacancies. The binding energies between Ni atoms and the C₄/C₃Cl vacancy sites were 0.538 and −0.451 eV, respectively (Supplementary Fig. 18). Obviously, the Cl-doping enhanced the interaction between the Ni atom and the carbon vacancies, resulting in a transition from a non-spontaneous to a spontaneous process for the anchoring of Ni single atoms to vacancy sites. Therefore, in the synthesis of Ni₁-C/Cl, the migrating Ni species preferentially bonded to Cl-containing carbon vacancies to form single-atom Ni sites. The lack of vacancy sites in the carbon carriers of Ni_n-C/Cl led to the formation of Ni clusters during pyrolysis. The above detailed observations demonstrated the feasibility of our vacancy-engineering strategy for the specific construction of asymmetric single-atom sites.

Fig. 2: Electronic property and local coordination environment investigations.

a XANES spectra of Zn foil, ZnO, zinc phthalocyanine (ZnPc), Zn₁-C/Cl, and Ni₁-C/Cl at Zn K-edge. b FT-EXAFS spectra of Zn foil, ZnO, ZnPc, Zn₁-C/Cl, and Ni₁-C/Cl at Zn K-edge. c Experimental and fitted FT-EXAFS spectra of Zn₁-C/Cl and Ni₁-C/Cl at Zn K-edge. d XANES spectra of Ni foil, NiO, NiPc, Ni₁-C/Cl, and Ni_n-C/Cl at Ni K-edge. e FT-EXAFS spectra of Ni foil, NiO, NiPc, Ni₁-C/Cl, and Ni_n-C/Cl at Ni K-edge. f Experimental and fitted FT-EXAFS spectra of Ni₁-C/Cl and Ni_n-C/Cl at Ni K-edge. g–i WT spectra of Zn foil (g), Zn₁-C/Cl (h), and Ni₁-C/Cl (i) at Zn K-edge. j–l WT spectra of Ni foil (j), Ni_n-C/Cl (k), and Ni₁-C/Cl (l) at Ni K-edge. Source data are provided as a Source data file.

The evaluation of ECO₂RR performances

The ECO₂RR performance evaluations of these three catalysts were conducted in a typical H-type cell. As can be seen from the linear sweep voltammetry (LSV) curves in Fig. 3a, Ni₁-C/Cl achieved higher polarization current densities in the potential rage of −0.54 to −0.90 V versus the reversible hydrogen electrode (vs. RHE) compared to Zn₁-C/Cl and Ni_n-C/Cl, suggesting that the introduced Ni single atoms in Ni₁-C/Cl contributed to the higher electrocatalytic activity. Moreover, Ni₁-C/Cl and Zn₁-C/Cl exhibited different onset potentials for electrochemical reactions (CO₂ reduction reaction or hydrogen evolution reaction (HER)), which represented a switch in the active sites from Zn single atoms to Ni single atoms or possibly existed Ni-Zn dual atoms in Ni₁-C/Cl. Due to the significant difference in Ni and Zn loadings, we believed that the possibly existed Ni–Zn interactions had a weak influence on the catalytic activity of Ni₁-C/Cl. Chronoamperometry was applied for evaluating their catalytic selectivity and activity for CO₂ electroreduction, the gas-phase products were detected and analyzed by an online gas chromatography (GC), the liquid-phase products were detected by an offline nuclear magnetic resonance (NMR) spectrometer. Among these three catalysts, Ni₁-C/Cl showed the optimal selectivity for ECO₂RR-to-CO, with FE_COs higher than 90% from −0.5 to −0.9 V vs. RHE (Fig. 3b). Only gaseous products of CO and H₂ were detected during the whole electrocatalysis process (Supplementary Figs. 19, 20). The maximum CO partial current density (j_CO) of Ni₁-C/Cl reached ~51.7 mA cm⁻² at −0.9 V vs. RHE, much higher than that of Zn₁-C/Cl (~18.12 mA cm⁻²) and Ni_n-C/Cl (~0.3 mA cm⁻²) (Fig. 3c). These results verified that the single-atom Ni sites served as the active sites for ECO₂RR. We then conducted the electrochemical capacitance measurements to quantify the electrochemically active surface area (ECSA) of these catalysts (Supplementary Fig. 21a–c). Ni₁-C/Cl showed a higher double layer capacitance (C_dl) of 261.9 mF cm⁻² compared to that of Ni_n-C/Cl (123.3 mF cm⁻²) and Zn₁-C/Cl (217.8 mF cm⁻²), which suggested that Ni₁-C/Cl had the maximum ECSA, favoring for the exposure of more active sites for ECO₂RR (Fig. 3d)³³. The ECSA normalized j_COs of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl in Fig. 3e indicated that the introduced single-atom Ni sites in Ni₁-C/Cl contributed to the intrinsic activity for electrocatalysis of CO₂ to CO. The comparisons of the ECO₂RR performances between Ni₁-C/Cl and the recently reported Ni catalysts were shown in Fig. 3f and Supplementary Fig. 21d. Ni₁-C/Cl demonstrated advanced catalytic activity for CO production over other single-atom Ni catalysts in ECO₂RR. In a 190-h continuous electrolysis at −0.6 V vs. RHE, Ni₁-C/Cl maintained FE_COs higher than 95%, indicating the structural stability of the asymmetric C₃–Ni–Cl active sites in long-term electrolysis (Fig. 3g).

Fig. 3: ECO₂RR performance evaluation in H-cell.

a LSV curves of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl under CO₂ atmosphere. b FE_COs of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl under different potentials. c j_COs of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl under different potentials. d C_dls of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl. e ECSA normalized j_COs of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl. f ECO₂RR performance comparisons of Ni₁-C/Cl with other recently reported electrocatalysts in H-cell (0.5 M KHCO₃ as the electrolyte)^{33,35,44,45,46,47,48,49}. g Long-term stability test of Ni₁-C/Cl. The flow rate of CO₂ inlet was 30 mL min⁻¹ (sccm). The measured resistance of the H-cell containing 0.5 M KHCO₃ electrolyte was 5.89 ± 0.04 Ω. The potentials were presented without iR compensation. The error bars correspond to the standard deviations were acquired based on three independent experiments under the same testing conditions. Source data are provided as a Source data file.

The clarification of the catalytic mechanisms

In situ XAFS spectroscopy was applied to monitor the structure evolution of the single-atom sites in Ni₁-C/Cl during the ECO₂RR process. The schematic diagram of the in situ electrochemical cell was shown in Fig. 4a. Figure 4b demonstrated the Ni K-edge XANES spectra of Ni₁-C/Cl recorded at different applied potentials and after electrochemical reaction. When cathodic potentials were applied from open-circuit voltage (OCV) to −0.4, −0.6, and −0.8 V vs. RHE, significant shifts of the absorption edge toward lower energies were observed, representing a distinct decrease in the valence state of the central Ni atoms in single-atom sites. This phenomenon indicated that when cathodic potentials were applied for driving ECO₂RR, the transported electrons were enriched at the Ni centers, thereby forming electron-rich catalytic sites to enhance the adsorption and activation of reactive species. After the electrochemical reaction, the absorption edge shifted back to a higher energy, reflecting a rebound in the valence state of these active Ni centers. The local coordination structure evolutions of the single-atom Ni sites during ECO₂RR were revealed by the EXAFS spectroscopy (Fig. 4c and Supplementary Fig. 22). Figure 4c showed the experimental and fitted Ni K-edge FT-EXAFS spectra of Ni₁-C/Cl at different potentials and after electrochemical reaction. In the FT-EXAFS spectra, peaks located at ~1.30 Å and ~1.94 Å were assigned to Ni–C and Ni–Cl bonds, respectively. The local coordination structure of the active Ni sites showed potential-dependent evolutions during ECO₂RR, with a ~0.02 Å and a ~0.015 Å variations in Ni-C and Ni–Cl bond lengths, respectively (Supplementary Table 5). The structure evolutions in these asymmetric sites at various potentials can optimize their absorption configurations for reactants and intermediates, thus accelerating the ECO₂RR process¹⁷. In particular, after electrocatalysis, the C₃–Ni–Cl sites exhibited similar coordination structure with that at OCV, confirming their structural stability during ECO₂RR.

Fig. 4: The clarification of the catalytic mechanisms.

a The schematic diagram of the electrochemical cell for in situ XAFS measurements. b In situ Ni K-edge XANES spectra of Ni₁-C/Cl recorded at different potentials. c Experimental and fitted FT-EXAFS spectra of Ni₁-C/Cl recorded at different potentials. d Charge density map of NiC₄ site. e Charge density map of NiC₃Cl site. f Calculated free energies of CO₂-to-CO and HER-Volmer step on NiC₄ and NiC₃Cl sites. g The total DOS of the NiC₄ site, as well as the corresponding projected DOS of Ni-3d and C-2p. h The total DOS of the NiC₃Cl site, as well as the corresponding projected DOS of Ni-3d, Cl-2p, and C-2p. (i) IR spectra of Ni₁-C/Cl recorded at different potentials. The red and green color in Fig. 4d and Fig. 4e represent charge accumulation and charge loss, respectively. Source data are provided as a Source data file.

DFT calculations were performed to theoretically understand the CO₂-to-CO conversion on C₃–Ni–Cl sites. The C₃–Ni–Cl structural model (denoted as NiC₃Cl) embedded in a graphene substrate was built based on the XAFS results, a symmetric NiC₄ structural model was also established for comparison (Supplementary Figs. 23–25, Supplementary Data 1). The charge density maps of NiC₄ and NiC₃Cl sites are shown in Fig. 4d and Fig. 4e, respectively. Apparently, a symmetric electron distribution can be observed in the NiC₄ site. In the NiC₃Cl site, the asymmetric local coordination environment induced a symmetry-broken electron distribution. The calculated free energies of CO₂-to-CO and HER-Volmer step on NiC₄ and NiC₃Cl sites indicated that both NiC₄ and NiC₃Cl sites had a preference for ECO₂RR over HER (Fig. 4f). However, the desorption of CO product on the NiC₄ site was unfavorable, with a free energy for the rate-determining step (RDS) of 0.540 eV. In contrast, an optimized reaction pathway of CO₂-to-CO conversion was achieved on the asymmetric NiC₃Cl site, with a RDS free energy of 0.386 eV. The electronic density of states (DOS) of NiC₄ and NiC₃Cl sites were shown in Fig. 4g and Fig. 4h. The d-band center of the NiC₃Cl site showed a lower shift toward Fermi energy than that of the NiC₄ site, suggesting a weaker adsorption capacity of the NiC₃Cl site to the reactant/intermediates for CO₂-to-CO conversion³⁴. These DFT calculations theoretically indicated that the Cl doping induced the formation of an asymmetric electron distribution in the single-atom Ni site, which optimized the adsorption strength of the reactants/intermediates on the Ni center in ECO₂RR, resulting in a moderate reaction pathway for CO₂-to-CO conversion.

The in situ ATR-SEIRAS measurements were conducted to identify the reaction intermediates on Ni₁-C/Cl during the ECO₂RR process (Supplementary Fig. 26). Figure 4i showed the IR spectra recorded at different potentials. The bands located at 1364-1391 cm⁻¹ were assigned to the C-O stretching in *COOH intermediate³⁵. The bands located at ~1650 cm⁻¹ were attributed to the H–O–H bending of water¹². The increased bands of *COOH intermediates and decreased bands of water from −0.4 to −0.9 V vs. RHE represented the efficient proton supply from water dissociation for the formation of *COOH intermediates during ECO₂RR. No distinct bands referring to *CO intermediate were observed, indicating the rapid desorption of CO products from the active sites. These investigations in ATR-SEIRAS measurements experimentally confirmed the fast reaction kinetics of CO production and desorption on the asymmetric C₃–Ni–Cl single-atom sites during ECO₂RR.

ECO₂RR in strongly acidic media

We first conducted ECO₂RR in an acid-fed flow cell electrolyzer, in which 0.05 M K₂SO₄ solution (pH=1) was selected as the cathodic and anodic electrolytes. Figure 5a and Supplementary Fig. 27 showed the schematic diagrams of the flow cell. Chronopotentiometry was applied for evaluating the ECO₂RR performances of Ni₁-C/Cl, Zn₁-C/Cl, and Ni_n-C/Cl in the acidic flow cell. Ni₁-C/Cl exhibited lower full-cell voltages (E_cells) in the current density range of 50 to 500 mA cm⁻² compared to Zn₁-C/Cl and Ni_n-C/Cl, indicating the higher electrochemical activity of Ni₁-C/Cl for ECO₂RR (Supplementary Fig. 28). Moreover, Ni₁-C/Cl also realized FE_COs greater than 90 % within the current density range of 50 to 500 mA cm⁻². (Fig. 5b). In contrast, Ni_n-C/Cl was barely selective for electrocatalytic CO₂-to-CO conversion in the acidic flow cell, with a maximum FE_CO of ~15.7% at the current density of 50 mA cm⁻². The limited selectivity of Zn₁-C/Cl in acidic ECO₂RR was identified from the dramatic decrease in the FE_COs on Zn₁-C/Cl at current densities higher than 150 mA cm⁻². These investigations further revealed the irreplaceable contribution of the asymmetric C₃–Ni–Cl active sites in Ni₁-C/Cl for highly selective electrocatalytic CO₂ conversion, endowing Ni₁-C/Cl with j_CO of ~491.4 mA cm⁻² at the applied current density of 500 mA cm⁻² (Fig. 5c). In a 40-h continuous electrolysis at the current density of 200 mA cm⁻², Ni₁-C/Cl consistently maintained FE_COs above 97 %, which represented the stability of Ni₁-C/Cl for industrial-level ECO₂RR (Fig. 5d). We then investigated the effect of different K⁺ concentrations on the activity and selectivity of ECO₂RR on Ni₁-C/Cl. The corresponding E_cells recorded at different current densities in Supplementary Fig. 29 showed that increasing the K⁺ concentrations promoted the kinetics of the cathodic CO₂ reduction and the anodic oxygen evolution reactions, thus reducing the E_cells for CO electrosynthesis. Notably, even in a strongly acidic medium (pH = 1) containing 0.01 M K₂SO₄, Ni₁-C/Cl still achieved FE_COs greater than 95% in the current density range of 150 to 500 mA cm⁻² (Supplementary Fig. 30).

Fig. 5: ECO₂RR in strongly acidic media.

a Schematic diagram of the flow cell electrolyzer. b FE_COs of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl at different applied current densities. c j_COs of Ni_n-C/Cl, Zn₁-C/Cl, and Ni₁-C/Cl at different applied current densities. d Long-term stability test of Ni₁-C/Cl in flow cell. e E_cells in flow cell and MEA at different applied current densities using Ni₁-C/Cl as the cathodic electrocatalyst. f FE_COs of Ni₁-C/Cl at different applied current densities in MEA. g Long-term stability test of Ni₁-C/Cl in MEA. The flow rate of CO₂ inlet was 20 sccm. The measured resistances of flow cell and MEA electrolyzers containing 0.05 M K₂SO₄ electrolyte were 4.37 ± 0.31 Ω and 0.60 ± 0.01 Ω, respectively. The E_cells were recorded without iR compensation. The error bars correspond to the standard deviations were acquired based on three independent experiments under the same testing conditions. Source data are provided as a Source data file.

The ECO₂RR performance evaluations of Ni₁-C/Cl were also carried out in an acid-fed membrane electrode assembly (MEA) electrolyzer (Supplementary Fig. 31), using 0.05 M K₂SO₄ solution (pH=1) as the anodic electrolyte, humidified CO₂ as the feedstock for CO production. In the acid-fed MEA, Ni₁-C/Cl achieved much lower E_cells for ECO₂RR than those in the flow cell, due to the lower resistance and faster mass transport in MEA (Fig. 5e and Supplementary Fig. 32). In the current density range of 50 to 350 mA cm⁻², the FE_COs were consistently above 90% (Fig. 5f). The single-pass carbon efficiency in MEA for CO₂-to-CO conversion (SPCE_CO) reached about 70% at the industrial current density of 200 mA cm⁻², under the flow rate of CO₂ inlet of 2 sccm. (Supplementary Fig. 33). The SPCE_CO comparisons of Ni₁-C/Cl with other reported electrocatalysts in MEA demonstrated the higher carbon utilization on Ni₁-C/Cl during CO₂-to-CO conversion (Supplementary Fig. 34). Moreover, the CO selectivity on Ni₁-C/Cl remained above 90% throughout a 10-h continuous electrolysis at the current density of 200 mA cm⁻² (Fig. 5g).

In order to minimize the formation of (bi)carbonates during CO₂ electrolysis, we further used 0.01 M K₂SO₄ solution (pH = 1) as the anodic electrolyte to evaluate the long-term stability of Ni₁-C/Cl for ECO₂RR in MEA. As can be seen from Supplementary Fig. 35a, in a ~80 h continuous electrolysis at 200 mA cm⁻², Ni₁-C/Cl consistently maintained the FE_CO over 93% and the SPCE_CO over 28%. The long-term stability comparisons between Ni₁-C/Cl and other reported electrocatalysts proved the advances of Ni₁-C/Cl toward industrial CO₂ electrolysis (Supplementary Fig. 35b). The ex situ XRD, XPS, and AC HAADF-STEM measurements disclosed that the Ni species in Ni₁-C/Cl still existed in the form of single atoms after the stability test, verifying the structural stability of the single-atom sites in Ni₁-C/Cl (Supplementary Figs. 36, 37). The in situ ATR-SEIRAS measurements were conducted to monitor the ECO₂RR process on Ni₁-C/Cl after long-term stability test in MEA. Obvious bands assigned to the *COOH intermediates were observed in the recorded IR spectra, which indicated that the C₃-Ni–Cl single-atom sites still retained a good activation capacity for CO₂ reactants during ECO₂RR, even after a long period of continuous electrolysis (Supplementary Fig. 38).

To compare the catalytic activity and selectivity differences between C–Ni–Cl and N–Ni–Cl single-atom sites, we introduced nitrogen source in the synthesis of Ni₁-C/Cl and changed the coordination structure of the single-atom Ni sites from C–Ni–Cl to N–Ni–Cl, the prepared electrocatalyst was named as Ni₁-N/Cl. The XRD pattern and TEM images of Ni₁-N/Cl in Supplementary Figs. 39, 40 verified the absence of nanoparticle species in Ni₁-N/Cl. The HAADF-STEM images and the corresponding EDS mapping in Supplementary Fig. 41 confirmed that the Ni species in Ni₁-N/Cl were anchored on the carbon carriers as single atoms. The XPS spectra of Ni₁-N/Cl indicated the existence of Ni–N and Ni–Cl moieties in Ni₁-N/Cl (Supplementary Fig. 42). The XAFS measurements further revealed the N₃–Ni–Cl coordination structure of most single-atom Ni sites in Ni₁-N/Cl (Supplementary Fig. 43 and Supplementary Table 6). The ECO₂RR performance evaluation of Ni₁-N/Cl in the H-cell confirmed the higher activity of Ni₁-C/Cl than Ni₁-N/Cl (Supplementary Fig. 44). In the flow cell, Ni₁-C/Cl also achieved lower E_cells for ECO₂RR to CO than Ni₁-N/Cl at industrial current densities, further validating the advanced performance of C₃–Ni–Cl single-atom sites in CO₂ electrolysis (Supplementary Fig. 45). The normalizations of partial current for CO production by the concentration of electrochemical active Ni sites were shown in Supplementary Figs. 46, 47. Obviously, the active Ni sites in Ni₁-C/Cl exhibited higher electrocatalytic activity than those in Ni₁-N/Cl and Ni_n-C/Cl. The recorded Nyquist plots during ECO₂RR under different potentials in Supplementary Fig. 48 demonstrated that the constructed Ni single-atom sites facilitated the electron transfer from the active centers to the reactants/intermediates during ECO₂RR, endowing Ni₁-C/Cl and Ni₁-N/Cl with higher activity for CO₂ conversion compared with Ni_n-C/Cl and Zn₁-C/Cl.

Extended application of Ni₁-C/Cl in tandem electrocatalysis

The combination of CO₂-to-CO and CO-to-C₂₊ in tandem CO₂ electrolysis provided a transit route for improving the selectivity of C₂₊ products in acidic ECO₂RR^36,37. Herein, we extended the application of Ni₁-C/Cl in a tandem electrode to investigate the promotion of Ni₁-C/Cl for C₂₊ production on the cuprous oxide (Cu₂O) electrocatalyst in a flow cell. 0.05 M K₂SO₄ solution (pH = 1) was used as the cathodic/anodic electrolyte. The tandem electrode contained two distinct catalyst layers, Ni₁-C/Cl layer and Cu₂O nanocubes (Cu₂O NCs) layer, respectively (Fig. 6a). The structural characterization of Cu₂O NCs was shown in Supplementary Figs. 49, 50. ECO₂RR in this tandem system exhibited lower E_cells compared to those in the system containing sole Cu₂O catalyst, representing the faster reaction kinetics on the Ni₁-C/Cl@Cu₂O tandem catalysts (Supplementary Fig. 51). Notably, the Ni₁-C/Cl@Cu₂O tandem catalysts exhibited effective promotion for selective C₂H₄ generation at high current densities (Fig. 6b, c and Supplementary Figs. 52–54). These advances may stem from the regulation of the local pH over the Cu₂O catalyst layer by the generated OH^- during CO₂–CO conversion on Ni₁-C/Cl³⁶. Moreover, the optimized coverage of *CO and *H species also facilitated the improvement of C₂H₄ selectivity. This tandem system realized maximum C₂H₄ Faradaic efficiency (FE_C2H4) of ~33%, with a considerable C₂H₄ partial current density (j_C2H4) over 0.37 A cm⁻². The XRD patterns in Supplementary Fig. 55 revealed that in the Cu₂O NCs layer, the Cu₂O-derived Cu (111) planes contributed to the catalytic activity and selectivity for ECO₂RR to C₂H₄. DFT calculations were then conducted to reveal the intrinsic synergy in the coupled Cu (111)@C₃–Ni–Cl sites (Supplementary Figs. 56–59). The calculated free energies for C₂H₄ formation on Cu (111) plane and coupled Cu (111)@NiC₃Cl site in Fig. 6d and Fig. 6e indicated that the coupled Cu (111)@NiC₃Cl site reduced the energy barrier for C-C coupling compared with bare Cu (111) plane, with an optimized free energy of 1.679 eV for the formation of *OCCO intermediate on Cu (111)@NiC₃Cl site. Moreover, the spontaneous formation of *OCCOH intermediate from *OCCO intermediate on coupled Cu (111)@NiC₃Cl site facilitated the formation and hydrogenation of *CCO intermediate, to favor the C₂H₄ production. This application of Ni₁-C/Cl@Cu₂O tandem electrocatalysis revealed the neighboring synergy between asymmetric C₃–Ni–Cl single-atom sites and active Cu species for selective electrosynthesis of C₂H₄ products.

Fig. 6: Extended application of Ni₁-C/Cl in tandem electrocatalysis.

a Schematic diagram of Ni₁-C/Cl@Cu₂O NCs tandem electrocatalysis for C₂H₄ production. b FE_C2H4s of Ni₁-C/Cl@Cu₂O tandem electrocatalysts and sole Cu₂O electrocatalysts at different applied current densities. c j_C2H4s of Ni₁-C/Cl@Cu₂O tandem electrocatalysts and sole Cu₂O electrocatalysts at different applied current densities. d Calculated free energies for C₂H₄ formation on Cu (111) plane. e Calculated free energies for C₂H₄ formation on Cu (111)@NiC₃Cl coupled site. The flow rate of CO₂ inlet was 10 sccm. The measured resistance of the flow cell electrolyzer containing 0.05 M K₂SO₄ electrolyte was 4.37 ± 0.31 Ω. The error bars correspond to the standard deviations were acquired based on three independent experiments under the same testing conditions. Source data are provided as a Source data file.

Discussion

In summary, we developed a vacancy-engineering strategy for constructing asymmetric C₃–Ni–Cl single-atom sites on carbon supports. Detailed characterizations revealed that the abundant Cl-doping vacancy sites facilitated the trapping and anchoring of Ni atoms, to form asymmetric C₃–Ni–Cl single-atom sites. The symmetry-broken electron distribution in C₃–Ni–Cl coordination structure regulated the reaction pathway of CO₂-to-CO on the Ni center, promoting the selective CO electrosynthesis on Ni₁-C/Cl. In situ experiments clarified the potential-dependent structure evolutions of the C₃–Ni–Cl active sites during ECO₂RR, the C₃–Ni–Cl sites with self-optimized coordination structure at various potentials can modulate their adsorption configurations for reactant and intermediates to promote CO production. Benefitting from the highly selective ECO₂RR on Ni₁-C/Cl, even in the strongly acidic media (pH = 1), Ni₁-C/Cl still exhibited near 100% selectivity for CO products under industrial-level current densities. The extended application of Ni₁-C/Cl in a Ni₁-C/Cl@Cu₂O tandem electrocatalysis system revealed the neighboring synergy between single-atom Ni sites and active Cu species for facilitating the selective production of C₂H₄ from CO₂ electrolysis. This work highlighted the significance of coordination engineering in the Ni SACs for promoting CO₂ electroreduction, offering new insights for designing industrial-oriented electrocatalysts toward other important electrocatalytic reactions.

Methods

Chemicals

ZnCl₂ (MACKLIN, Z885069, 99%), KCl (Aladdin, P112134, AR, 99.5%), agarose (Aladdin, A104062, for biochemistry), nickel(II) chloride (NiCl₂, N433839, 99.99% trace metals basis), sodium hydroxide (NaOH, Aladdin, S580606, ≥ 98%), copper(II) chloride (CuCl₂, Aladdin, C106774, ≥98%), ascorbic acid (Aladdin, A103533, AR, ≥99%), potassium bicarbonate (KHCO₃, Aladdin, P110485, 99.5%), sulfuric acid (H₂SO₄, Sinopharm Chemical Reagent Co., Ltd, GR), potassium sulfate (K₂SO₄, Aladdin, P112580, AR, 99%), cyanuric acid (Aladdin, C106082, 98%), Ar and CO₂ (Hangzhou BESTGAS Co., Ltd, 99.999%). The Nafion D520 dispersion (alcohol based 1000 EW at 5 wt%), proton exchange membranes (212 µm-thickness Nafion 212 membrane and 115 µm-thickness Nafion 115 membrane), and gas diffusion electrode (GDE, Sigracet, 39BB) were purchased from SCI Materials Hub (www.scimaterials.cn). The IrO_x/Ti felt with IrO_x loading of 2 mg cm⁻² was purchased from Suzhou Sinero Technology Co., Ltd. The H-cell (CH2010-S), Pt mesh electrode (2 × 2 cm², Pt purity of 99.99%), working electrode clip (J110), as well as Ag/AgCl electrode (R0303-Φ6 mm) were purchased from Ada Hengsheng Technology Co., Ltd.). The flow cell and MEA electrolyzers were designed and fabricated by Suzhou Sinero Technology Co., Ltd. Ultrapure water (Millipore, 18.2 MΩ·cm) was used throughout the experimental process.

Preparation of Ni₁-C/Cl

First, 800 mg of KCl was dissolved in 7 mL of water and heated at 120 °C with continuous stirring. Then 200 mg of agarose powder was added to the stirred solution. After the agarose powder was dissolved, 1 mL of aqueous solution containing 12.5 mg NiCl₂ was added dropwise with continued stirring for 10 min. At the end of the stirring period, the above solution was poured onto a Petri dish and left at room temperature to transform into a hydrogel. The obtained hydrogel was frozen at −100 °C for 30 min and freeze-dried at −50 °C for 24 h to remove the water content. The dried sample was grounded into powder before mixed with 2 g of ZnCl₂. Subsequently, the above mixture was treated in a gradient pyrolysis at 300 °C for 3 h and 1000 °C for 1 h in a tube furnace under Ar atmosphere (Ar flow rate, 80 sccm). The harvested product was washed with water several times to remove the salt templates, and vacuum dried at 60 °C for 24 h to obtain Ni₁-C/Cl. The Ar flow rate was controlled by an electronic flowmeter (HORIBA, SEC-Z714AGX).

Preparation of Ni_n-C/Cl

The preparation of Ni_n-C/Cl was similar to that of Ni₁-C/Cl, with the absence of ZnCl₂ salt template.

Preparation of Zn₁-C/Cl

The preparation of Zi₁-C/Cl was similar to that of Ni₁-C/Cl, with the absence of NiCl₂ aqueous solution.

Preparation of Ni₁-N/Cl

The preparation of Ni₁-N/Cl was similar to that of Ni₁-C/Cl, accompanied by the addition of 2 g of cyanuric acid as the nitrogen source.

Preparation of Cu₂O NCs

First, 30 mL of 2 M NaOH aqueous solution (denoted as Solution A), 300 mL of 10 mM CuCl₂ aqueous solution (denoted as Solution B), and 30 mL of 0.6 M ascorbic acid solution (denoted as Solution C) were prepared, respectively. Then Solution A was added dropwise into Solution B, with continuous stirring at 55 °C for 30 min. Subsequently, Solution C was added dropwise into the above solution and kept stirring at 55 °C for 5 h. The harvested product was washed with water and ethanol several times, and vacuum dried at 60 °C for 24 h to obtain Cu₂O NCs.

Characterizations

The XRD patterns were collected from a micro-X-ray diffractometer (Rigaku) and a powder X-ray diffractometer (Bruker D8 Advance). The TEM measurements were performed on a transmission electron microscope (Thermo Fisher, Talos F200X G2) at an accelerating voltage of 200 kV. The AC-HAADF-STEM measurements were conducted on a spherical aberration corrected transmission electron microscope (Thermo Fisher, Themis Z) at an accelerating voltage of 300 kV. The EPR spectra were collected from a pulsed electron paramagnetic resonance spectrometer (Bruker, A300). The Raman spectra were collected from a VIS-NIR confocal Raman microscope system (WITec, Alpha300R). The XPS spectra were collected from an XPS spectrometer (Thermo Fisher, ESCALAB Xi + ). The XAFS spectra at Zn K-edge and Ni K-edge of Ni₁-C/Cl, Ni_n-C/Cl and Zn₁-C/Cl were obtained at the 1W1B beamline of Beijing Synchrotron Radiation Facility. The XAFS spectra at Ni K-edge of Ni₁-N/Cl were collected at the BL14W1 beamline of Shanghai Synchrotron Radiation Facility. The XAFS data were recorded under fluorescence mode, processed and analyzed with Athena and Artemis software codes.

Electrode preparation for evaluating the ECO₂RR performances of Ni₁-C/Cl, Zn₁-C/Cl, and Ni_n-C/Cl

First, the catalyst (5 mg) was homogeneously dispersed in a mixed solution containing isopropanol (300 μL), water (100 μL), and Nafion D520 solution (50 μL) through 30-min sonication. Then the catalyst dispersion was drop-casted onto the GDE (1 × 2 cm² for electrochemical tests in H-cell, 1.5 × 1.5 cm² for electrochemical tests in flow cell and MEA electrolyzers) and dried at room temperature, with the catalyst loading of 0.5 and 1 mg cm⁻² for electrochemical tests in H-cell and flow cell/MEA electrolyzers, respectively.

Electrode preparation for evaluating the ECO₂RR performances of Ni₁-C/Cl@Cu₂O tandem electrocatalysts

First, 20 mg of Cu₂O catalyst was homogeneously dispersed in a mixed solution containing isopropanol (300 μL), water (100 μL), and Nafion solution (100 μL) through 30-min sonication, and 5 mg of Ni₁-C/Cl catalyst was homogeneously dispersed in a mixed solution containing isopropanol (300 μL), water (100 μL), and Nafion solution (50 μL) through 30-min sonication. Then the Cu₂O catalyst dispersion was drop-casted onto the GDE (1.5 × 1.5 cm²) and dried at room temperature, with the Cu₂O catalyst loading of 5 mg cm⁻². The Ni₁-C/Cl catalyst dispersion was drop-casted onto the Cu₂O catalyst layer and dried at room temperature, with the Ni₁-C/Cl catalyst loading of 1 mg cm⁻².

Electrochemical measurements

The electrochemical measurements for evaluating the ECO₂RR performances were performed on a Biologic VMP3 electrochemical workstation under ambient conditions (room temperature, 22 °C). In the typical H-cell system, 0.5 M KHCO₃ solution was used as the electrolyte. Specifically, 50.3 g of KHCO₃ were transferred into a volumetric flask (1 L), then adding 800 mL of ultrapure water to the volumetric flask and stirred until KHCO₃ was completly dissolved. Finally, added ultrapure water to adjust the volume of the solution to 1 L to obtain 0.5 M KHCO₃ solution. Before ECO₂RR tests, the KHCO₃ electrolyte was stored at room temperature. The volume of the single cell in the H-cell was 100mL, with 50mL of electrolyte in each cell. The cathode and anode chambers were separated by a 212 µm-thickness Nafion 212 membrane. Before using, the Nafion 212 membrane was sequentially treated with 5% H₂O₂ solution and 5% H₂SO₄ solution at 80 °C for 1 h. Pt mesh electrode and Ag/AgCl electrode (saturated KCl) were used as counter electrode and reference electrode, respectively. Before ECO₂RR performance evaluation H-cell, continuous CO₂ stream was fed into the cathodic KHCO₃ electrolyte for 1 h to obtain CO₂-saturated KHCO₃ electrolyte. The pH value of the CO₂-saturated KHCO₃ electrolyte was measured using a pH meter (METTLER TOLEDO, SevenExcellence), with pH value of 7.5 ± 0.02 in three independent measurements. The polarization curves were acquired from the linear sweep voltammetry, with a scan rate of 10 mV s⁻¹. The cyclic voltammetry was used in the electrochemical capacitance measurements. The potentio electrochemical impedance spectroscopy with a frequency range from 200 kHz to 10 mHz was carried out to obtain the Nyquist plots of the catalysts during ECO₂RR under various potentials.

The potential conversion from Ag/AgCl reference electrode scale to the RHE reference scale was performed in a three-electrode system, with hydrogen-saturated 0.5 M H₂SO₄ solution as the electrolyte. The preparation of 0.5 M H₂SO₄ solution was in a 1 L volumetric flask. First, added 500 mL of ultrapure water into the volumetric flask, then slowly added 27.17 mL of concentrated H₂SO₄ into the volumetric flask. After the solution temperature dropped to room temperature, added ultrapure water to adjust the volume of the solution to 1 L to obtained 0.5 M H₂SO₄ solution. In the three-electrode system, two Pt wire electrodes were used as the working and counter electrodes, respectively. An Ag/AgCl electrode (saturated KCl) was used as the reference electrode. The calibration was performed using cyclic voltammetry, with a scan rate of 1 mV s⁻¹. The average of two potential values at zero current represented the thermodynamic potential of the hydrogen electrode reaction (0.197 V). Therefore, the potential conversion from Ag/AgCl reference electrode scale to the RHE reference scale was acquired based on the following equation¹²:

$$E({{\rm{V}}}\,{{\mathrm{vs}}}.\,{{\mathrm{RHE}}})=E({{\rm{V}}}\,{\mathrm{vs}}.\,{\mathrm{Ag}}/{\mathrm{AgCl}})+0.05916\times pH+0.197$$

(1)

In the flow cell system, K₂SO₄ solutions with various concentrations were used as the cathodic/anodic electrolyte (0 M, 0.01 M, 0.05 M, 0.1 M and 0.25 M, pH = 1, adjusted by concentrated H₂SO₄). The preparation process of K₂SO₄ electrolyte was the same as that of KHCO₃ electrolyte. The pH value of the electrolytes was measured using a pH meter to ensure the pH value of the electrolyte was within 1 ± 0.05. Before using, the electrolytes were stored at ambient conditions. The schematic diagram of the flow cell electrolyzer was shown in Supplementary Fig. 27. GDE loaded with target catalyst was used as the working electrode (1.5 × 1.5 cm²), commercial IrO_x/Ti felt catalyst (1.5 × 1.5 cm²) was used as the counter electrode. A piece of 115 µm-thickness Nafion 115 membrane (2 × 2 cm²) was selected as the proton exchange membrane. Before using, the Nafion 115 membrane was sequentially treated with 5% H₂O₂ solution and 5% H₂SO₄ solution at 80 °C for 1 h. The flow rate of CO₂ inlet was controlled by an electronic flowmeter (HORIBA, SEC-Z714AGX). The flow rate of the anodic and cathodic electrolytes was controlled by a peristaltic pump (Gaoss Union, EC200-01), with a specific flow rate of 8 mL min⁻¹. Chronopotentiometry was applied in the evaluation of the acidic ECO₂RR performances, the E_cells were recorded without iR compensation. The area contacting electrolyte in the flow cell was 1 × 1 cm².

In MEA system, 0.05 M K₂SO₄ solution (pH=1, adjusted by concentrated H₂SO₄) was used as the anodic electrolyte. The schematic diagram of the MEA electrolyzer was shown in Supplementary Fig. 31. GDE loaded with target catalyst was used as the working electrode (1.5 × 1.5 cm²), commercial IrO_x/Ti felt catalyst (1.5 × 1.5 cm²) was used as the counter electrode. A piece of 115 µm-thickness Nafion 115 membrane (2 × 2 cm²) was selected as the proton exchange membrane. Before using, the Nafion 115 membrane was sequentially treated with 5% H₂O₂ solution and 5% H₂SO₄ solution at 80 °C for 1 h. Humidified CO₂ was used as the cathodic feedstock, the humidifier was purchased from SCI Materials Hub. The flow rate of CO₂ inlet was controlled by an electronic flowmeter. The flow rate of the anodic electrolyte was controlled by a peristaltic pump, with a specific flow rate of 8 mL min⁻¹. Chronopotentiometry was applied in the evaluation of the acidic ECO₂RR performances, the E_cells were recorded without iR compensation. The active area in MEA was 1 × 1 cm².

Before ECO₂RR performance evaluations, the resistance value of the electrochemical cell was measured by ZIR tool in EC-Lab software, with a specific frequency of 100 KHz at open-circuit voltage.

The gaseous products were detected and analyzed by a chromatography equipped with a methanizer, a thermal conductivity detector, and a flame ionization detector (GC2014, Shimadzu). The liquid-phase products were detected and analyzed by a 500 MHz solution NMR spectrometer (Cryo Probe, AVANCE NEO, Bruker BioSpin).

The FEs of CO, H₂, and C₂H₄ products were calculated based on the following equation²¹:

$${{\mathrm{FE}}(\%)=\frac{Q}{{Q}_{total}}}=\frac{z\times n\times F}{{\rm{Q}}_{total}}\times 100\%$$

(2)

where z is the number of electrons needed for H₂, CO and C₂H₄ formation, respectively (z = 2 for CO and H₂, z = 12 for C₂H₄), n is the molar amount of the corresponding products, F is Faraday’s constant (96485 C mol⁻¹), and Q_total is the recorded charge flowing through the working electrode during ECO₂RR process.

The partial current density of a specific product (j_product) was calculated as follows:

$${j}_{{\mathrm{product}}}({{\mathrm{mA}}}\, {{\mathrm{cm}}}^{-2})={{\mathrm{FE}}}_{{\mathrm{product}}}\times j({\mathrm{mA}}\, {{\mathrm{cm}}}^{-2})\div100\%$$

(3)

where j is the total current density.

In situ ATR-SEIRAS measurements

The in situ ATR-SEIRAS measurements were conducted on an FT-IR spectrometer (Thermo Fisher, Nicolet iS50) equipped with an MCT detector and an in situ IR optical accessory (SPEC-I, Shanghai Yuanfang Tech.). A single-cell containing 20 mL of CO₂-saturated 0.5 M KHCO₃ electrolyte was used as the reactor. The schematic diagram of the electrochemical cell is shown in Supplementary Fig. 26. During the ECO₂RR tests, CO₂ gas was continuously fed to the reactor. A Pt wire was selected as the counter electrode, an Ag/AgCl reference electrode was used as the reference electrode. A monocrystalline silicon coated with a gold film was applied as the working electrode. The Ni₁-C/Cl catalyst was sprayed on the gold film for ECO₂RR tests. A CHI 660 electrochemical workstation was applied for potential control. An 8 cm⁻¹ resolution and 16 scans were set for collecting the IR spectra during ECO₂RR.

In situ XAFS measurements

The in situ XAFS spectra at Ni K-edge were recorded at the 1W1B beamline of Beijing Synchrotron Radiation Facility. The schematic diagram of the electrochemical cell was shown in Fig. 4a. The GDE loaded with Ni₁-C/Cl catalyst was sealed in a cell by Kapton film for electrochemical tests. CO₂-saturated 0.5 M KHCO₃ solution was used as electrolyte. During the ECO₂RR tests, CO₂ gas was continuously fed to the reactor. A CHI 660 electrochemical workstation was applied for potential control. The in situ XAFS spectra were collected during the ECO₂RR process, under fluorescence mode.

Computational Details

The first principle DFT calculations were performed by Vienna Ab initio Simulation Package with the projector augmented wave method^38,39. The exchange functional was processed using the generalized gradient approximation of Perdew–Burke–Ernzerhof functional⁴⁰. The spin polarizations were employed for all calculations. The energy cutoff for the plane wave basis expansion was set to 450 eV. The convergence criterion for geometric relaxation was a force less than 0.03 eV/Å per atom. The k-points in the Brillouin zone were sampled in a 3× 3 × 1 grid. The convergence energy threshold for self-consistent calculations was 10^–5 eV. The DFT-D3 method was employed to consider the van der Waals interaction⁴¹. In order to avoid interactions between the periodic structures, a 15 Å vacuum was added along the z-axis.

The free energies were calculated as follows⁴²:

$$\Delta {{\rm{G}}}=\Delta {{\rm{E}}}_{{\mathrm{DFT}}}+\Delta {{\rm{E}}}_{{\mathrm{ZPE}}}-{{\rm{T}}}\Delta {{\rm{S}}}$$

(4)

where ΔE_DFT is the DFT electronic energy difference of each step, ΔE_ZPE is the correction of zero-point energy, and ΔS is the variation of entropy, obtained by vibration analysis. T is the temperature (T = 300 K).