Tuning binding strength between single metal atoms and supports enhances electrochemical CO2 methanation

Li, Linbo; Lei, Xin; Zheng, Zhilong; Dong, Yingjun; Chen, Haohui; Chen, Jun; Zhong, Yi; Zheng, Yongping; Tang, Yongbing; Zhang, Xiaolong; Cheng, Hui-Ming

doi:10.1038/s41467-025-63781-9

Download PDF

Article
Open access
Published: 30 September 2025

Tuning binding strength between single metal atoms and supports enhances electrochemical CO₂ methanation

Nature Communications volume 16, Article number: 8641 (2025) Cite this article

329 Accesses
Metrics details

Subjects

Abstract

Single-atom catalysts (SACs) with tunable site density and activity are promising for catalytic processes. However, the relationship between interacting sites and the catalytic mechanism, as well as the effect of the support on this relationship, remains incompletely understood. Here we report a support geometry engineering strategy to control the inter-site distance (d_site) of Cu–N–C (CuNC) SACs via strong interactions between CuNC and a secondary support (ss). This process allows tuning of the binding strength (that is Cu–N bond length) between individual Cu atoms and the N-doped primary supports, concomitantly suppressing defect formation and Cu atom detachment in the CuNC framework. The continuous optimization of the electronic and coordination structure of individual active Cu sites, achieved by reducing the d_site to approximately 0.7 nm, enhances their inherent CO₂-to-methane selectivity and activity. As a result, the ss-engineered CuNC with a moderate d_site of 0.68 nm exhibits enhanced methane selectivity of 70% and a partial current density of 303.9 mA cm⁻², over 1.5 times higher than that of unmodified CuNC.

Substrate strain tunes operando geometric distortion and oxygen reduction activity of CuN₂C₂ single-atom sites

Article Open access 03 November 2021

Insights on forming N,O-coordinated Cu single-atom catalysts for electrochemical reduction CO₂ to methane

Article Open access 26 January 2021

Manipulating d-orbital of Cu single atom site by coordination engineering for selective oxidation of benzene

Article Open access 01 July 2025

Introduction

The electrochemical CO₂ reduction reaction (CO₂RR) powered by renewable energy is a promising strategy to achieve efficient and sustainable production of green fuels and chemical feedstocks^1,2,3,4. Among CO₂RR electrocatalysts, copper (Cu)-based catalysts can efficiently catalyse CO₂RR for the production of hydrocarbons and oxygenates^5,6,7,8,9. Of these products (such as methanol, methane, ethylene, ethanol, etc.), methane (CH₄) stands out as a fuel with high combustion efficiency and the highest energy density (~56 kJ/g), and is widely used in the energy and chemical industries¹⁰. Green energy-powered electrochemical CO₂ methanation (ECM) eliminates the high emissions associated with fossil CH₄ extraction and offers a promising opportunity to close the anthropogenic carbon cycle.

However, bulk Cu exhibits limited CO₂RR selectivity toward CH₄ under ambient conditions^11,12. As the dimensions of Cu catalysts gradually decrease to the atomic scale, their selectivity for CH₄ increases correspondingly, due to the high energy barrier for C–C dimerization and favourable protonation of the key *CHO intermediate at Cu-single-atom (Cu SA) sites^{13,14,15,16,17}. Results from previous theoretical and experimental studies show that the modification of the coordination and electronic structures of Cu SA sites is critical for highly efficient and selective CH₄ production over Cu SA catalysts (Cu SACs, Supplementary Table 1)^{15,16,17,18,19,20,21}. Inspired by the role of Cu–C bonds in altering organic reaction pathways in organometallic coordination chemistry, Wang et al. constructed Cu–C bonds to stabilize Cu SA sites on graphdiyne via a unique electronic regulation effect between graphdiyne and Cu. This strategy accelerates charge transfer, changes the electronic structure of Cu atoms, and promotes the formation of *OCHO intermediates and the reaction pathway to CH₄¹⁸. The introduction of a second non-metal element (O/B/N) into the first coordination shell of Cu SA changes the oxidation state and optimizes the adsorption of *CO and *CHO intermediates for highly selective CH₄ conversion^19,20. Despite significant advances in ECM, the correlation of the density and proximity effect of active Cu sites in supported Cu SACs with their activity and selectivity toward CH₄ has rarely been investigated^22,23,24,25. Furthermore, a deep atomic-level understanding of the support effect on this correlation is required.

Recently, a proof-of-concept study of Fe–N₄ SACs for oxygen reduction reaction (ORR) unveiled the connection between the origin of enhanced ORR activity and the kinetic property of isolated active sites as well as the proximity effect of neighbouring metal atoms using a descriptor of inter-site distance (d_site)²⁶. However, the support effect on the d_site effect and the resulting reactivity should also be taken into account. Here, we report a support geometry engineering strategy to control the d_site effect, as well as the electronic and coordination structure of CuN₄ SAs anchored on a N-doped carbons (CuNC), where N-doped carbons (NC) serves as the primary support. This is achieved via conformal coating of polydopamine on the secondary support (graphite flakes, also denoted as ss), which plays a crucial role in shaping the geometry and curvature of CuNC. This process facilitates the continuous and uniform anchoring of homogeneous CuN₄ SAs over a broad range of site density. Strong interactions between CuNC and ss efficiently steer the binding strength (that is Cu–N bond length) between individual Cu atoms and their primary supports, thereby mitigating sintering and further tuning d_site in ss-engineered CuNC (ssCuNC). Such interactions also enable charge transfer from the CuNC to the ss, leading to an elevated oxidation state of the CuN₄ SAs in the ssCuNC. Theoretical modelling suggests that reducing the d_site of CuN₄ SAs to ~0.7 nm enhances their structural stress without changing the ECM reaction pathway. This enhancement shortens the Cu–N bond length and induces a positive shift in the 3d_x²−y² orbitals of Cu, thus improving the adsorption capacity and catalytic activity of individual active Cu sites for CH₄ generation. Computational results also indicate that a low d_site value not only stabilizes *CO adsorption due to the elevated d_x²−y² orbital level but also favours protonation of *OCH₃ to form CH₄. This is consistent with in situ spectroscopic results showing that selective electrosynthesis of CH₄ occurs at the expense of *CO, which shows a strong adsorption signal on ssCuNC. As a result of this strategy, ssCuNC with a moderate d_site of 0.68 nm and enhanced intrinsic site reactivity exhibits increased CO₂-to-CH₄ selectivity of 70% and a partial current density of 303.9 mA cm⁻², which is at least 1.5 times higher than that of all other CuNC materials.

Results

Identification of the role of support geometry engineering in the synthesis of Cu SACs with different d_site values

The synthesis principle related with support geometry engineering was first studied. Polydopamine (PDA), derived from the self-polymerization and oxidation of dopamine, was used as a metal carrier to chelate Cu²⁺ ions via its rich oxygen-containing functional groups, producing uniformly dispersed discrete Cu-chelated PDA nanoparticles (PNP–Cu) under alkaline conditions (Fig. 1a, b and Supplementary Fig. S1)^27,28. A series of PNP–Cu precursors were synthesized with designated CuSO₄ concentrations, such as 0.29 mg ml⁻¹, 0.57 mg ml⁻¹, 0.71 mg ml⁻¹, and 1.14 mg ml⁻¹ (Supplementary Fig. S1 and Methods). To verify the role of support geometry in fabricating Cu SACs with different d_site values, chemically inert planar graphite flakes were introduced as a secondary support (ss) to engineer the geometry of PDA (Supplementary Fig. S2). Strong covalent and noncovalent interactions between PDA and the substrate enable the formation of a PDA film conformally coated on ss²⁹. Oxidation of dopamine by Cu²⁺ produces integrated, continuous Cu-chelated PDA nanoparticles and films conformally coated on ss (ss–PNF–Cu; Supplementary Fig. S3). For these amorphous PNP–Cu and ss–PNF–Cu precursors, no metallic Cu or Cu-related carbides/nitrides were identified in powder X-ray diffraction (XRD) pattens, suggesting that PDA was complexed with Cu ions via its catecholate groups rather than reducing Cu²⁺ to metallic Cu (Supplementary Fig. S4).

**Fig. 1: Support geometry engineering strategy and characterisation of the morphology, chemical, and electronic structures of ssCuNC and CuNC.**

As the CuSO₄ concentration increased from 0.29 to 1.14 mg ml⁻¹, the obtained PNP–Cu precursors exhibited comparable PNP size (0.12–0.15 μm), while the weight of PNP–Cu nanoparticles increased significantly from 0.042 to 0.28 g (Supplementary Figs. S5a–d, Fig. S6a, c and Supplementary Table 2). This finding implies that increasing the Cu²⁺ concentration promotes the generation of more PNP–Cu nanoparticles with similar sizes, rather than substantially increasing the size of individual PNP–Cu nanoparticles. In contrast, the average PNP size of as-prepared ss–PNF–Cu precursors increased linearly from 0.07 μm to a peak of 0.15 μm, and the weight of ss–PNF–Cu precursors increased significantly from 0.2 to 0.45 g (Supplementary Figs. S5e–h and S6b, d). The difference in average PNP size between PNP–Cu and ss–PNF–Cu precursors is ascribed to the competitive growth of PNF in ss–PNF–Cu precursors. This is supported by the observation that the coverage area of PNF–Cu coated on ss increased with increasing CuSO₄ concentration (Supplementary Fig. S3). These geometric features of PNP–Cu and ss–PNF–Cu are depicted in Fig. 1a, b. These results also indicate that redox-active Cu²⁺ contributes to the self-polymerization and oxidation of dopamine but has only a slight effect on the kinetics of dopamine polymerization under basic conditions^28,30,31. In contrast to the similar geometry of spherical PNP–Cu precursors, the varied geometry combinations of spherical PNP–Cu and planar Cu-chelated PDA films in ss–PNF–Cu precursors provide a platform for modulating site density diversity, which is enabled by the support engineering.

The as-prepared PNP–Cu precursors were further uniformly ground with dicyandiamide powder (DCD, N source). The ground mixture was then pyrolyzed to generate atomic Cu sites anchored to N-doped carbon (CuNC; Fig. 1a, c; see Methods). Using ss–PNF–Cu precursors and the same synthesis procedure as for CuNC preparation, the ss-engineered CuNC (ssCuNC) catalysts were generated (Fig. 1b, d). According to the CuSO₄ concentration used, the resulting CuNC catalysts were designated as CuNC40, CuNC80, CuNC100, and CuNC160 (Supplementary Fig. S7). Correspondingly, the resulting ssCuNC catalysts were designated as ssCuNC40, ssCuNC80, ssCuNC100, and ssCuNC160 (Supplementary Fig. S8). No metallic Cu species were detected in any of the CuNC and ssCuNC catalysts (Supplementary Fig. S9).

Elemental analysis of the precursors was carried out using inductively coupled plasma mass spectrometry (ICP-MS) and X-ray photoelectron spectroscopy (XPS). Supplementary Fig. S10 shows that the Cu content in PNP–Cu and ss–PNF–Cu precursors increases almost linearly with increasing Cu²⁺ concentration, demonstrating the robust Cu-complexing capability of PDA over the entire range of Cu²⁺ concentration range used (Supplementary Table 2). This ensures that Cu²⁺ ions are homogeneously distributed throughout the PDA-based precursors within the specified concentration range. The correlation between Cu content in PDA-based precursors and in the corresponding Cu SACs was also studied. When the applied Cu²⁺ concentration is less than 0.71 mg ml^-1, a positive linear correlation exists between Cu loading in both CuNC and ssCuNC catalysts and that in PNP–Cu and ss–PNF–Cu precursors. However, further increasing Cu loading in PDA-based precursors reduces Cu content in CuNC, due to the formation of significant amounts of Cu nanoparticles in raw CuNC160 samples (Supplementary Figs. S11, S12 and Supplementary Table 3). These results demonstrate the crucial role of Cu²⁺ loading in the metal carrier in steering Cu content in Cu SACs. Compared with the CuNC160 catalyst, the reduced formation of Cu nanoparticles and increased Cu loading in ssCuNC160 suggest that support engineering for CuNC using flat ss enhances Cu single atomisation in CuNC.

The Cu loadings of ssCuNC catalysts are found to be lower than those of CuNC catalysts because the weight of pure ss is included in the sample weight during ICP-MS quantification. In fact, XPS analysis reveals that the Cu and N contents (atomic percentage, at.%) on the surfaces of ssCuNC100 and ssCuNC160 are higher than those on CuNC100 and CuNC160 (Fig. 1e, f, Supplementary Figs. S13, S14 and Supplementary Table 4). In contrast to the relatively constant Cu and N contents in CuNC catalysts, the significant variation in Cu and N contents in ssCuNC further confirms the enhanced tunability of Cu content in ssCuNC catalysts (Supplementary Fig. S14). This finding is partly because planar PNF, with lower curvature than spherical PNP, promotes uniform site distribution and close contact between PNF and DCD, resulting in a gradient increase in Cu site density with Cu ion concentration. These observations are consistent with the concept that support geometry engineering applied to the Cu carrier results in Cu SACs with enhanced tunability in site density, as depicted in Fig. 1a, b.

The single-atom nature of Cu sites in the as-prepared Cu SACs was investigated. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) coupled with energy-dispersive X-ray spectroscopy (EDS) mapping displayed uniform dispersion of C, O, N, and Cu across all CuNC and ssCuNC SACs (Fig. 1c, d, Supplementary Figs. S15–S18). The coordination environment and chemical states of Cu sites were confirmed by electron energy loss spectroscopy (EELS), X-ray spectroscopy, and fitting analysis. EELS data imply that the Cu–O moiety contributes negligibly to the local coordination environment in the representative CuNC100 and ssCuNC100 catalysts (Supplementary Fig. S19). Extended X-ray absorption fine structure (EXAFS) spectroscopy plots in R space for all ssCuNC and CuNC SACs display a main peak at ~1.45 Å, attributed to the Cu–N shell (Fig. 1g, h and Supplementary Table 5)^32,33. EXAFS results also confirmed that no Cu–Cu bonds were detected in any Cu SACs, consistent with results from patterns, TEM images, selected area electron diffraction (SAED) patterns, and scanning electron microscopy (SEM) images (Supplementary Note 1).

To reveal the local structure of Cu SACs, comprehensive fitting analyses were carried out on Fourier-transformed EXAFS spectra in both radial (R) and wave vector (k) spaces. A single Cu–N scattering path and mixed scattering paths (including Cu–N/O, Cu–N/Cu, and Cu–N/O/Cu combinations) were used (Fig. 1i, Supplementary Figs. S20–S23). Supplementary Table 5 summarizes fitting parameters for the Cu SACs of interest, where the coordination number (CN) of N ranges from 3.8 to 4.3, with Cu–N bond length between 1.91 Å and 1.95 Å. The shortened or elongated Cu–N bond lengths suggest geometric distortion in Cu SACs, further supported by the geometric curvature observed in TEM images (see the following section). The fitting data also suggest that the Cu sites adopt a planar CuN₄ structure. Note that the average Cu–N bond lengths in ssCuNC are shorter than those in CuNC counterparts, indicating improved binding strength between single Cu atoms and primary supports, which stabilizes the isolated Cu sites in ssCuNC catalysts against demetallization and sintering (Fig. 1j). These results suggest that Cu sites feature single atomic distribution, and the isolated Cu site (CuN₄) are coordinated with an average of four N atoms in both ssCuNC and CuNC SACs.

The chemical states of Cu in the Cu SACs were also studied using XPS and X-ray absorption near edge structure (XANES) spectroscopy. Compared with CuNC catalysts, ssCuNC counterparts exhibit higher N 1s and Cu 2p binding energies, along with the lower C 1s binding energies. This finding demonstrates strong interactions between CuNC and ss in ssCuNC catalysts, leading to charge transfer from CuNC to ss and a higher oxidation state of Cu in ssCuNC compared to CuNC (Fig. 1e, f, Supplementary Fig. S13)^34,35. The larger blue shift of the Cu 2p peak compared to the N 1s peak is attributed to the greater electronegativity difference between Cu and N. The average oxidation state of Cu in the Cu SACs was also quantified by XANES analyses via linear combination fitting (Supplementary Fig. S24)^36,37,38. Results show that the Cu valence in both CuNC and ssCuNC catalysts decreases with increasing Cu loading, due to electronic interactions between neighbouring sites induced by the d_site effect. Thus, engineering the geometry of PDA-based supports using ss not only governs Cu contents in Cu SACs but also shortens the Cu–N bond length and improves the oxidation state of Cu in Cu SACs.

Quantitative comparison of the d_site effect in ssCuNC and CuNC SACs

The effect of support geometry engineering on the d_site effect in Cu SAC was investigated. The average d_site value was calculated via statistical analysis of aberration-corrected (AC)-HAADF-STEM images for all catalysts. Each measured d_site value, as a statistical sample, refers to the shortest distance between a selected atom and its neighbour atom. AC-HAADF-STEM images of CuNC and ssCuNC catalysts show Cu atoms as bright spots with an average size of ~0.2 nm on the carbon substrate (Fig. 2a–h). The statistical average d_site values for the Cu catalysts were 2.2 nm (ssCuNC40), 0.93 nm (ssCuNC80), 0.68 nm (ssCuNC100), 0.36 nm (ssCuNC160), 0.89 nm (CuNC40), 0.85 nm (CuNC80), 0.81 nm (CuNC100), and 0.79 nm (CuNC160) (Fig. 2, Supplementary Figs. S25, S26 and Supplementary Table 6). Examination of STEM images showed a similar density of active Cu atoms in each CuNC catalyst. However, site densities in ssCuNC catalysts exhibited a progressive increase with Cu loading.

**Fig. 2: Correlation of the d_site effect with Cu loading in ssCuNC and CuNC SACs.**

Structural distortions of Cu SACs were also studied via TEM characterization. CuNC SACs with similar Cu loadings exhibit comparable high substrate curvatures (Fig. 2I–IV and Supplementary Fig. S27a–d). In contrast, the substrate curvature of ssCuNC SACs increases progressively with increasing Cu loading (Fig. 2V–VIII and Fig. S27e–h). This finding indicates that Cu SACs with high site density or low d_site are more prone to inducing structural distortions. Note that the substrate of CuNC exhibits a distinctive discrete island-shaped distribution, whereas that of ssCuNC exhibits a continuous ribbon-shaped pattern (Supplementary Fig. S27). This difference in substrate geometry between CuNC and ssCuNC is closely related to the geometric differences in Cu-chelated PDA precursors used to synthesise CuNC and ssCuNC. Specifically, the former uses discrete Cu-chelated PDA nanoparticles (PNP–Cu), while the latter uses integrated Cu-chelated PDA nanoparticles and films conformally coated on ss (ss–PNF–Cu) (Fig. 1a, b, Supplementary Figs. S1, S3). During mixing of Cu-chelated PDA precursors with DCD to prepare Cu SACs, the larger-scale, continuous, planar ss–PNF–Cu with significantly lower curvature than spherical PNF–Cu, facilitates uniform Cu atoms distribution and provides substantial Cu atoms for direct interaction with DCD. Due to the high curvature of spherical PNF–Cu nanoparticles, a significant portion of Cu atoms within PNF–Cu fail to interact with DCD. Furthermore, compared to the uniform geometry of spherical PNP–Cu precursors, the varied geometry combinations of spherical PNP–Cu and planar Cu-chelated PDA films in ss–PNF–Cu precursors promote diverse interactions between Cu atoms and DCD. This phenomenon not only explains the heterogeneous site density in ssCuNC SACs and relatively consistent Cu site density in CuNC SACs with increasing Cu²⁺ concentration but also rationalizes the Cu site-density/d_site-dependent ECM performance shown in Fig. 3 (Fig. 2i and Supplementary Table 6). These findings are again consistent with the concept of an optimized d_site effect induced by support geometry engineering (Fig. 1a, b).

**Fig. 3: Comparison of ECM performances between ssCuNC and CuNC SACs.**

To correlate the d_site value with Cu site density on the catalyst surface, Cu site densities (n_s, Cu atoms per 100 nm²) were calculated from STEM image statistics. The variation trend of n_s values is analogous to that of d_site values, with all CuNC catalysts exhibiting minimal differences (55–79), whereas ssCuNC catalysts show a broader n_s distribution (15–339, Supplementary Table 6). For a d_site value of ~0.7, the corresponding n_s of ssCuNC100 (112.5) is over 1.4 times larger than that of all CuNC catalysts, indicating that the activity of ssCuNC100 is supposed to be superior to that of all CuNC catalysts (Fig. 2j). The higher n_s values in ssCuNC compared to CuNC are partly attributed to enhanced anti-sintering ability, which is triggered by shorter Cu–N bond lengths (Supplementary Table 5). Figure 2k shows that d_site values and Cu loading in ssCuNC catalysts can be controlled over a significantly wider range than in CuNC catalysts.

The effect of the support on the charge conductivity of Cu SACs was also examined via electrochemical impedance spectroscopy (EIS) to compare the charge transfer resistances (R_CT) of CuNC and ssCuNC SACs (Supplementary Fig. S28). CuNC SACs exhibit a volcano-shaped trend in R_CT, with CuNC100 showing the lowest R_CT, while R_CT values for ssCuNC SACs decrease with increasing Cu loading (Supplementary Fig. S28c, d and Supplementary Table 3). R_CT values of both CuNC and ssCuNC SACs show a consistent negative correlation with Cu loading. This suggests a significant effect of Cu loading/site density of Cu SACs on charge transfer. Compared with other Cu SACs, the significantly higher Rct value of ssCuNC40 is mainly attributed to reduced CuNC coverage on graphite flakes and lower Cu loading (Supplementary Figs. S3, S8 and Supplementary Table 3). Excluding ssCuNC40, the comparable R_CT values between CuNC and ssCuNC SACs rule out a dominant effect of ss on the charge transfer capability of Cu SACs. These findings evidence the crucial role of support geometry engineering in effectively controlling d_site, n_s, Cu loading, local structural distortion and charge conductivity in the as-prepared Cu-based catalysts.

ECM performance

ECM performance evaluation on Cu SACs with different d_site effects was conducted in a flow cell with a gas diffusion electrode (GDE) using 1 M KOH as the supporting electrolyte (used throughout this study)⁴. Catalysts were drop-cast on a GDE with a loading of ~0.2 mg cm^-2. ECM activities of ss, pure NC, ss-supported NC (ssNC), and ssNC modified with Cu nanoparticles (CuNPs/ssNC) were also investigated as control experiments (Supplementary Figs. S29 and S30). The sole product H₂ generated on ss, NC, and ssNC under high reductive current densities rules out ECM activity contributions from both ss and NC substrates (Supplementary Fig. S29 and Methods). CuNPs/ssNC exhibits an optimal Faradaic efficiency (FE) towards CH₄ (FE_CH4) of 36% at −1.4 V (Supplementary Fig. S30). This finding implies that dense Cu sites with the lowest d_site or highest n_s have limited selectivity toward CH₄, due to the competing carbon-carbon coupling reaction pathway on metallic Cu catalyst^11,12. We then evaluated the ECM performance of CuNC catalysts. The total current density (j), CH₄ selectivity, and partial current density of CH₄ (j_CH4) for each CuNC catalyst exhibited minor changes as the applied potentials (all referenced to the reversible hydrogen electrode (RHE)) became increasingly negative (Fig. 3a–c). Specifically, the optimal FE_CH4 values for CuNC40, CuNC80, CuNC100 and CuNC160 were 57% at −0.81 V, 58% at −0.75 V, 58% at −0.83 V and 56% at −0.76 V, respectively (Fig. 3b and Supplementary Fig. S31a–c, g). The peak j_CH4 values for CuNC40, CuNC80, CuNC100 and CuNC160 was 196.8 mA cm⁻² at −1.1 V, 208.0 mA cm⁻² at −1.0 V, 214.4 mA cm⁻² at −0.97 V, and 204.8 mA cm⁻² at −0.92 V, respectively (Fig. 3c).

In contrast, ssCuNC catalysts showed significant variations in ECM kinetics, with ssCuNC100 exhibiting the highest j and j_CH4 across all applied potentials (Fig. 3a, c, d, f). A similar trend is observed in the variation of FE_CH4 for ssCuNC catalysts (Fig. 3e and Supplementary Fig. S31d–f, h). The optimal FE_CH4 of ssCuNC100 (70% at −0.96 V) is at least 10% higher than that of all CuNC and other ssCuNC catalysts (30% for ssCuNC40 at −0.97 V, 59% for ssCuNC80 at −0.93 V, and 51% for ssCuNC160 at −0.98 V). At ~−1.0 V, j_CH4 of ssCuNC100 (303.9 mA cm⁻²) is 7.0-, 2.0-, 1.6-, and ~1.5-fold higher than that of ssCuNC40 (43.3 mA cm⁻²), ssCuNC80 (152.9 mA cm⁻²), ssCuNC160 (186.7 mA cm⁻²), and CuNC catalysts, respectively (Supplementary Fig. S32). The effect of electrochemical surface area (ECSA) on ECM performance was also evaluated via cyclic voltammetry measurements for Cu SACs in an Ar-saturated 1 M KOH electrolyte using a flow cell setup (Supplementary Fig. S35 and Supplementary Table 7). The optimal j_CH4 after ECSA normalization (j_{CH4_ECSA}) was further examined. The optimal j_{CH4_ECSA} values for CuNC40, CuNC80, CuNC100, CuNC160, ssCuNC40, ssCuNC80, ssCuNC100, and ssCuNC160 were 0.67, 0.65, 0.72, 0.67, 0.44, 0.71, 1.3, and 0.57 mA cm⁻², respectively. These j_{CH4_ECSA} values show a similar trend to those determined by geometric area, as shown in Fig. 3. This finding rules out a predominant effect of ECSA on ECM performance. Thus, these differences in ECM performance between CuNC and ssCuNC indicate that support geometry engineering effectively enhances the selectivity and activity of CuNC over a wider range.

To reveal the origin of performance differences between CuNC and ssCuNC, the correlation between ECM selectivity/activity and the d_site effect of Cu catalysts was investigated. The correlation of FE_CH4 with Cu loading shows that optimal FE_CH4 depends on Cu loading across all Cu specimens, particularly for ssCuNC (Fig. 3g). However, all CuNC catalysts with higher Cu loadings exhibit lower CH₄ selectivity than ssCuNC80 and ssCuNC100, suggesting that Cu loading is not crucial for determining CH₄ selectivity in Cu SACs. Figure 3h and Fig. 3i, respectively, display a volcano-shaped trend relating optimal FE_CH4 to d_site and optimal FE_CH4 to n_s, with the highest CH₄ selectivity observed at a d_site of 0.68 nm and an n_s of 112.5 Cu atoms per 100 nm⁻². Note that CuNC catalysts with similar optimal FE_CH4 values have a similar d_site of 0.8–0.9 nm or an n_s of 60 ~ 80 Cu atoms per 100 nm². Furthermore, the optimal FE_CH4 of ssCuNC100 is much higher than that of CuNC160 despite their similar d_site, due to the higher n_s of ssCuNC100 relative to CuNC160 (Fig. 3i and Supplementary Table 6). A correlation analysis of j_CH4 (at ~−0.1 V) with Cu loading, d_site and n_s yielded similar conclusions to those discussed above (Supplementary Fig. S32). These results, coupled with the linear relationship between FE_CH4/j_CH4 and n_s, suggest that Cu site density is a crucial descriptor for ECM selectivity and activity on Cu SCAs. The relationship between the d_site effect and ECM activity described above was also verified using ssCuNC100 (denoted as ssCuNC100-scale-up) prepared via a scale-up synthesis procedure (see Methods). The synthesized ssCuNC100-scale-up SAC shows a similar Cu-site coordination environment and d_site effect, achieving an optimal FE_CH4 of 68% at −0.96 V (Supplementary Figs. S33, S34 and Supplementary Table 5). The comparable structure-activity relationship observed for both ssCuNC100 and ssCuNC100-scale-up further confirms the scalability of our synthesis strategy.

The effect of electrolyte pH on ECM for CuNC100 and ssCuNC100 SACs was investigated. CO₂RR electrolysis using CuNC100 and ssCuNC100 SACs was conducted in a liquid-flow electrolyser with 1 M KHCO₃ electrolyte. Results showed that the optimal FE for CH₄ production was 49.8% for CuNC100 and 59.2% for ssCuNC100 at 300 mA cm⁻², respectively (Supplementary Fig. S31i, j). These values are ~10% lower than those obtained in 1 M KOH, which is attributed to suppressed hydrogen evolution reaction (HER) in alkaline environments, as is evidenced by the lower FE of H₂ formation detected in 1 M KOH electrolyte (Supplementary Fig. S31g, h). ECM stability measurements using CuNC100 and ssCuNC100 were conducted at a current density of 300 mA cm⁻² in a flow cell with 1 M KHCO₃ as the supporting electrolyte. ssCuNC100 displays stable CH₄ production, with FE remaining above 50% over 20 h. In contrast, the FE for CH₄ formation on CuNC100 rapidly decreased from ~42% at 6 h to 25% at 18 h (Fig. 3j, k). The stability difference between CuNC100 and ssCuNC100 was further studied via spectroscopy analysis.

Spectroscopic investigations

To investigate the relationship between ECM performance and the d_site effect for ssCuNC and CuNC, we examined the selectivity of combined C₂ products (ethylene, acetate and ethanol) and carbon monoxide (CO) as a function of applied potential (Fig. 4a, b). CO is a crucial intermediate for hydrocarbon production. Figure 4a shows that the FE_CO of ssCuNC100 is lower than that of CuNC100 across all applied potentials. This FE_CO trend was also observed for other CuNC and ssCuNC catalysts (Supplementary Fig. S36a–c). In contrast, FE_C2 is higher for CuNC100 than for ssCuNC100 under the same conditions. Note that FE_C2 (<15%) on ssCuNC160 at high j exceeds that on CuNC160, which is ascribed to improved carbon-carbon coupling activity on highly dense Cu sites (Supplementary Fig. S36f). Product distribution analysis suggests that ssCuNC enables a higher degree of CO₂ hydrogenation to CH₄ at the expense of more CO than CuNC.

**Fig. 4: Relationship between ECM selectivity and adsorbed intermediates on ssCuNC100 and CuNC100 SACs.**

We further investigated the origin of improved CO₂-to-CH₄ selectivity on ssCuNC at the molecular level using in situ attenuated total reflection Fourier-transform infrared spectroscopy (ATR-FTIR), with ssCuNC100 and CuNC100 as model catalysts. Four regions of interest, identified at 2120–2070 cm⁻¹, 1750–1640 cm⁻¹, 1640–1475 cm⁻¹ and 1470–1398 cm⁻¹, are associated with linearly adsorbed *CO (ν(*CO_L))_, stretching vibrations of H₂O (δ(H₂O)), strongly adsorbed *CO₂ (ν(*CO₂)), and *OCH₃ adsorbed at the ssCuNC100 and CuNC100 interfaces, respectively (Fig. 4c, d)^39,40,41,42. The band associated with *OCH₃ adsorbed at atomic CuN₄ site is also confirmed via computational simulation (Supplementary Fig. S37). The δ(H₂O) and ν(*CO₂) bands show potential-dependent positive and negative peaks. Both peak intensities increase with applied potential, with those in ssCuNC100 being significantly stronger than those in CuNC100. These results suggest rapid consumption of adsorbed *CO₂ and a higher mass transfer rate on ssCuNC100 than on CuNC100, accounting for the enhanced ECM activity of ssCuNC100 relative to CuNC100.

The peak positions of ν(*CO_L) in both ssCuNC100 and CuNC100 shifted to lower wavenumbers as the applied potential became more negative, showing a Stark effect (Fig. 4e)⁴³. Over the entire potential range of −0.3 to −1.5 V, ssCuNC100 displayed a lower wavenumber than CuNC100 and a higher Stark tuning rate (23.3 cm^-1 V^-1) in comparison to CuNC100 (22.3 cm⁻¹ V⁻¹), indicating a stronger interaction between *CO_L and ssCuNC100⁴¹. This interprets the suppressed CO liberation on ssCuNC100. Under identical CO₂ electrolysis conditions, the intensity of *CO_L on CuNC100 exceeds that on ssCuNC100, suggesting higher *CO coverage on CuNC100 and thus accounting for the increased C₂ production observed on CuNC100 (Fig. 4b, f). The reduced *CO coverage and improved peak intensity of the *OCH₃ band on ssCuNC100 also indicate favourable hydrogenation of *CO to form *OCH₃ at the expense of *CO on ssCuNC100 (Fig. 4c, d, f). This hydrogenation process is identified as the dominant selectivity-determining step in CH₄ production^3,40.

To elucidate the effect of structural alterations of active sites on ECM performance, we carried out in situ and ex situ XAS measurements and analyses for ssCuNC100 and CuNC100 SACs under different current densities (Supplementary Figs. S38–S43, Supplementary Table 8; See Methods). Both in situ and ex situ R-space Cu K-edge EXAFS spectra show a positive shift of the main peak, which—based on fitting results—is attributed to an increase in Cu–N bond lengths for both CuNC100 and ssCuNC100 during CO₂RR (Fig. 4g, h). Specifically, during 5 h of CO₂RR, the Cu–N bond length of CuNC100 increased significantly from 1.93 Å to 1.96 Å, 0.03 Å longer than the Cu–N bond length of ssCuNC100 (1.93 Å). This finding implies more pronounced geometric structure distortion of CuN₄ sites in CuNC100 compared to those in ssCuNC100 during CO₂RR. After 20 h of CO₂RR, the Cu–N bond length of CuNC100 recovered to 1.94 Å, still 0.01 Å longer than that of ssCuNC100 (1.93 Å). Note that a new peak assigned to Cu–Cu bond emerged in CuNC100. These results indicate that active sites in ssCuNC100 remained as single-site CuN₄ structures, whereas those in CuNC100 transformed from single-site CuN₄ structures to a combination of CuN₄ and metallic Cu after 20 h of CO₂RR. This phenomenon rationalizes the observed differences in ECM performance between CuNC100 and ssCuNC100, as depicted in Fig. 3. Thus, in situ and ex situ XAS results offer direct experimental evidence that CuN₄ is the active site in Cu SACs for ECM and that the Cu–N bond in ssCuNC100 is more stable against structure distortion than that in CuNC100 during ECM. This also accounts for the lower defect density in the ssCuNC100 framework relative to that in CuNC100 (Supplementary Fig. S44).

Theoretical study using density functional theory calculations

We further performed density functional theory (DFT) calculations to understand the origin of catalytic CO₂-to-CH₄ activity on Cu SACs with different d_site values. Decreasing d_site from 1.68 nm (1Cu) to 0.66 nm (8Cu) shortens the average Cu−N coordination bond length from 1.94 Å to 1.91 Å, consistent with XAS fitting results for Cu SACs (Fig. 5a, b, Supplementary Table 5 and Supplementary Data 1). Concurrently, the interlayer height difference (Δh) increased from 0.72 to 1.48 Å (Fig. 5c), suggesting potential graphene twisting under dense CuN₄ motifs, which aligns with TEM observations (Fig. 2I-VIII, Fig. 5a and Supplementary Fig. S27). Electronic restructuring was evidenced by the splitting and upward shift of Cu 3d_x²−y² orbitals toward the Fermi level, elevating the d-band centre (Fig. 5d). This, in turn, changes their electron distribution, thereby influencing catalytic activity⁴⁴.

**Fig. 5: Identification of d_site effect-induced structure and coordination environment in Cu SACs.**

Supplementary Fig. S45 shows that the highest occupied orbital of planar four-coordinated Cu ions is 3 d_x²−y², and the 2p orbitals of adsorbates hybridize to form bonding and antibonding orbitals. When the energy level of Cu 3d_x²−y² orbitals shifts to the Fermi level, the antibonding orbitals also increase in energy^45,46. Accordingly, the number of electrons in the antibonding orbitals decreases, strengthening the bonding and enhancing the adsorption capacity of adsorbents at Cu sites⁴⁵. This enhancement agrees with experimental results, where promoted CH₄ production is achieved at the expense of CO, accompanied by a stronger *CO adsorption signal on ss-engineered Cu SACs (Fig. 4c, d). Moreover, the increase in density of states (DOS) values at the Fermi level with increasing Cu loading indicates enhanced electron transfer kinetics between catalytic intermediates and high-density catalytic centres (Supplementary Fig. S46)⁴⁷. This d_site effect establishes a distinct electron distribution at the Fermi energy level by modulating Cu 3d orbital splitting, thereby enhancing intrinsic ECM activity⁴⁸.

The DFT-based climbing-image nudged elastic band (CI-NEB) method was used to calculate kinetic barriers for Cu atom detachment from a CuN₄ SA site, subsequent defect formation, and cluster aggregation (Supplementary Fig. S47 and Supplementary Data 1). Calculations reveal that removing a Cu atom from ss-free CuNC requires an energy barrier of 3.68 eV, whereas ss-supported CuNC necessitates a higher energy of 4.02 eV. Furthermore, the defect formation energy of CuNC (2.47 eV) is lower than that of ssCuNC (2.87 eV). These results demonstrate enhanced structural stability of CuN₄ single-atom sites with ss incorporation. The suppression of Cu atom detachment is likely attributed to the stronger Cu–N bond (with shorter bond lengths in ssCuNC100, as confirmed by XAS) and optimized support geometry induced by ss. These findings align with experimental observations of reduced defect density, increased active site density, and enhanced electrochemical stability for the ssCuNC100 SAC.

To further clarify the effect of high-density CuN₄ sites on boosting CH₄ selectivity, we performed DFT calculations coupled with a computational hydrogen electrode (CHE) model. We elucidated the free energy diagram of the lowest-energy pathway for CO₂-to-CH₄ conversion at the active CuN₄ site, along with atomic configurations of each elementary step (Fig. 5e, Supplementary Fig. S48 and Supplementary Data 1). The potential-determining step (PDS) for CO₂-to-CH₄ conversion at the CuN₄ site involves protonation of *OCH₃ to *O + CH₄, requiring cleavage of the C–O bond in *OCH₃. The associated free energy change (∆E) is 1.03 eV for the 1Cu catalyst, decreasing to 0.99, 0.94 and 0.88 eV for the 2Cu, 4Cu and 8Cu catalysts, respectively (Fig. 5e). In the explicit solvent model, where solvent molecules are included in calculations, the change in the energy barrier is more significant. Furthermore, the trend of the energy barrier change is consistent with results from the CHE model (Supplementary Figs. S49, S50 and Supplementary Data 1). These results indicate that the thermodynamic favourability of CH₄ production at the CuN₄ site enhances with increasing site density. Adsorption of the *CO intermediate promotes the formation of deep reduction products, including CH₄. Interestingly, high-density CuN₄ sites with a lower d_site value can stabilize *CO adsorption due to their relatively elevated d_x2-y2 orbital level (Supplementary Fig. S45 and Fig. 5e, inset). For the competitive HER, the calculated free energies for HER on 1Cu, 2Cu, 4Cu, and 8Cu are 1.16, 1.40, 1.45, and 1.61 eV. In contrast, the Cu(100) counterpart (d_site of 0.26 nm) exhibits a significantly lower free energy of only 0.24 eV (Supplementary Fig. S51a). We further found that Cu(100) enables CO₂ adsorption and exhibits a comparably low free energy barrier (0.4 eV) for the C–C coupling PDS, indicating that Cu(100) offers extra active sites for the CO₂-to-C₂H₄ process (Supplementary Fig. S51b). These results suggest that protonation/deprotonation on different Cu SACs is considerable challenging, while agglomerated Cu sites exhibit superior HER and competitive C–C coupling activities, in line with CO₂RR performance observed for CuNPs/ssNC, CuNC and ssCuNC (such as ssCuNC160) SACs (Supplementary Figs. S30, S31, S36).

We also examined the charge density difference and adsorption free energies of reaction intermediates for ssCuNC SACs to explore the effect of ss on CuNC SACs. Examination of the charge density difference in ssCuNC reveals significant charge transfer from CuNC to ss (Fig. 5f, Supplementary Fig. S52 and Supplementary Data 1). This accounts for the positively charged Cu atoms found in both XPS and XAS results for ssCuNC SACs (Fig. S13 and Fig. S24). Compared to CuNC, ssCuNC exhibits a lower free energy barrier for the PDS of CO₂-to-CH₄ conversion, confirming that ss also helps improve CH₄ selectivity over CuN₄ sites (Fig. 5g). These findings show the versatile role of support engineering in tuning the d_site effect, coordination environment, chemical structure, and intrinsic reactivity of interacting sites in SACs.

Discussion

In summary, we report a support geometry engineering strategy to optimize the d_site effect and intrinsic electrocatalytic performance of active sites in SACs by tuning the binding strength between single metal atoms and a catalytically inert secondary support. Flat graphite was used as a secondary support and model to verify our method. Engineering PDA with ss leads to the formation of differentiated geometric combinations of spherical (particulate) and flat (film) PDA-based metal precursors, which are used to synthesise ss-engineered CuNC SACs. Strong interactions between CuNC and ss enable regulation of the Cu−N bond length. This strategy enables effective control of Cu loading, d_site, and n_s in Cu SACs over a wide range. Reducing d_site to ~0.7 nm optimizes both the coordination environment and electronic structure of Cu sites, significantly enhancing their intrinsic ECM reactivity in Cu SACs.

We found that the d_site effect, particularly n_s, of Cu SACs plays a critical role in determining ECM selectivity and activity. Specifically, at ~−1.0 V, the ssCuNC SAC with a moderate d_site of 0.68 nm and an n_s of 112.5 Cu atoms 100 per nm² shows increased CH₄ selectivity of 70% and a j_CH4 of 303.9 mA cm⁻², at least 1.5 times higher than that of all CuNC catalysts without ss. In-situ spectroscopic and computational studies confirmed that ssCuNC SACs with modified electron distribution and Cu−N bond length, show optimized d_site and n_s values. These modifications improve adsorption capacity, stabilize *CO adsorption, and favours cleavage of the C–O bond in *OCH₃, thereby promoting ECM selectivity and activity. This approach sheds light on the relationship between the support effect, the proximity effect of interacting metal sites, and the CO₂RR catalytic mechanism. It also provides guidance for developing highly efficient materials with tunable site densities and reactivities.

Methods

Chemicals and materials

Sigracet 28BC gas diffusion layer (28BC GDL, SGL Carbon, Germany); anion-exchange membranes (Fumasep FAA-3-PK-130 (thickness: 130 μm, FuMA-Tech BWT GmbH) was activated in a 0.5 M aqueous NaCl solution for 24 h, washed with water, and used in liquid-electrolyte flow cells. Sustainion X37-50 grade 60 (thickness: ~50 μm, Dioxide Materials) was activated in a 1 M aqueous KOH solution for 48 h, washed with water, and used in membrane electrode assembly electrolysers); dimethyl sulfoxide (DMSO, ≥99.9%, Thermo Scientific, USA); deuteroxide (D₂O, 99.9%, Sigma-Aldrich, USA); aqueous ammonia solution (30 wt%, Macklin, China); dopamine hydrochloride (Sigma-Aldrich, Germany); anhydrous cupric sulfate (CuSO₄, Sigma-Aldrich, USA); Nafion solution (5 wt%, Sigma-Aldrich, Germany); 2-propanol (≥99.5%); absolute ethanol (≥99.0%). High purity (99.99%) CO₂, H₂, He, and Ar (Air Liquide, China); nickel plate (0.1 mm thick, Dongguan Shengze Metal Material Co., LTD., China). All chemicals were used as received. All aqueous solutions were prepared fresh as needed using deionized (DI, Milli-Q) water with a resistivity of 18 MΩ cm^-1, and all prepared solutions were stored in a refrigerator at ~5 °C.

Synthesis of ss–PNF–Cu precursors (Cu-chelated polydopamine films and nanoparticles coated on ss): 0.2 g of graphite flake powder was added to a conical flask containing 90 mL of ultrapure water and 5 mL of ethanol (99.7%, Shanghai Lingfeng Chemical Reagent Co., Ltd., China) as solvents, followed by ultrasonication for 30 min. Next, 2 mL of ammonia solution (30 wt%) and 35 mL of ethanol (99.9%) were injected into the flask with stirring at 500 rpm. Subsequently, 10 mL of an aqueous solution containing dopamine hydrochloride (DAH) and CuSO₄ was added dropwise. The flask was sealed, and the mixture was stirred at 1500 rpm under ambient conditions for 24 h. The resulting black precipitate was collected by centrifugation at 12,000 rpm for 30 min, then dried overnight in a vacuum oven at 60 °C. The weight ratio of DAH to CuSO₄ is described in the “Synthesis of ssCuNCX (X = 40, 80, 100, 160) and CuNCX”.

Scale-up synthesis of ss–PNF–Cu precursor (Supplementary Fig. S33a): 8.0 g of graphite flake powder was added to a reagent bottle containing 1.7 L of ultrapure water and 0.70 L of ethanol (99.7%) as solvents, followed by ultrasonication for 30 min. Next, 40 mL of ammonia solution (30 wt%) and 0.5 L of ethanol (99.9%) were injected into the bottle with stirring at 700 rpm. Then, 0.25 L of an aqueous solution containing DAH (10.0 g) and CuSO₄ (4.0 g) was added dropwise. The bottle was sealed, and the mixture was stirred at 700 rpm under ambient conditions for 24 h. The resulting black precipitate was collected by vacuum filtration and dried overnight in a vacuum oven at 60 °C. The remaining synthesis steps follow those described in “Synthesis of ssCuNCX (X = 40, 80, 100, 160) and CuNCX”.

Synthesis of PNP–Cu precursors (Cu-chelated polydopamine nanoparticles): Precursors synthesis followed the same procedure as “Synthesis of ss–PNF–Cu precursors”, except that no ss powder was added.

Synthesis of ssCuNCX (X = 40, 80, 100, 160) and CuNCX: ssCuNCX was prepared by uniformly mixing and grinding dry ss–PNF–Cu powder with dicyandiamide (DCD) at a weight ratio of 1:11. The ground powder was placed in a ceramic boat and transferred to a temperature-programmed furnace for pyrolysis under an inert gas stream. The pyrolysis program was as follows: (1) purge the furnace with high purity Ar gas at 50 SCCM for 40 min; (2) heat from room temperature to 800 °C at a ramp rate of 5 °C min⁻¹; (3) hold at 800 °C for 90 min; (4) cool naturally to room temperature. The resulting catalyst was ground, collected, and stored in an Ar-purged glove box at room temperature. ssCuNC160, ssCuNC100, ssCuNC80, and ssCuNC40 were prepared with DAH-to-CuSO₄ weight ratios of 0.25 g:0.16 g, 0.25 g:0.10 g, 0.25 g:0.08 g, and 0.25 g:0.04 g, respectively. The same procedure was used to synthesise the ssCuNC100-scale-up SAC, except that the controlled atmosphere box furnace (LC-SA3-20-12TP, LICHEN, China) was purged with high-purity Ar gas at 300 SCCM.

CuNCX catalysts (X = 40, 80, 100, 160) were synthesized using the same procedure, except that dry G–PNF–Cu powder was used instead of dry PNP–Cu powder.

Synthesis of ssNC (ss-supported N-doped carbons): ssNC was synthesised using the same procedures as “Synthesis of ss–PNF–Cu precursors” and “Synthesis of ssCuNCX”, except that no CuSO₄ was added.

Synthesis of NC: NC was synthesised using the same procedures as “Synthesis of ssNC”, except that no ss powder was added.

Synthesis of CuNPs/ssNC (ssNC modified with Cu nanoparticles): CuNPs/ssNC were synthesised using the same procedures as “Synthesis of ss–PNF–Cu precursors” and “Synthesis of ssCuNCX”, except that no DCD was added.

Characterisation

Scanning electron microscopy (Thermofisher Apero 2 S HiVac), transmission electron microscopy (FEI-Talos F200S), and probe-corrected scanning transmission electron microscopy (ThermoFisher Themis Z, operating voltage: 300 kV) were used to characterize the morphology of the samples. Prior to collecting SEM images of PNP–Cu and G–PNF–Cu precursors, the samples were coated with a thin layer of platinum particles using a Quorum EMS150TES Plus precision coating system. X-ray diffraction (XRD) patterns were recorded using a Bruker D8 ADVANCE X-ray powder diffractometer (Cu Kα radiation, λ = 0.15406 nm). Surface analysis of catalysts was performed using X-ray photoelectron spectroscopy (XPS) with a Thermo Scientific Escalab xi⁺ system equipped with a focused monochromatic Al Kα X-ray source (hν = 1486.6 eV, 650 μm spot size). Survey scans (wide) and high-resolution scans (narrow) were recorded at analyser pass energies of 150 and 20 eV, with step sizes of 1.0 eV and 0.1 eV, respectively. All XPS data were calibrated against the C 1s peak at 284.8 eV unless otherwise specified. All X-ray absorption spectroscopy (XAS) measurements were conducted at the XAS Beamline of the Australian Synchrotron, ANSTO. A Si(111) double-crystal monochromator was used to monochromatize the white beam, and the energy resolution of the setup is approximately 0.2 eV at the Cu K-edge. A 36-element Canberra silicon drift detector (SDD) was used to collect the fluorescence signal. The beam was collimated to a vertical size of ~0.3 mm at the sample position. XAS spectra were collected at room temperature. For the pre-edge region, a 2 eV energy step with 1 s integration time per point was used. The X-ray absorption near-edge structure (XANES) region was scanned using a 0.2 eV step with the same dwell time. Extended X-ray absorption fine structure (EXAFS) data were collected up to k = 13 Å⁻¹ with a variable energy step corresponding to Δk ≈ 0.05 Å⁻¹. The integration time per point increased from 2 to 5 seconds as a function of k to improve data quality at high k values. For in-situ measurements, XAS spectra were recorded in slow-scan (continuous energy) mode, with a typical acquisition time of 5 min per scan. The in-situ electrochemical cell was fabricated from Polyetherketoneketone (PEKK). It employed nickel foam as the anode, a Hg/HgO reference electrode, and carbon paper as the cathode, with ~5.6 mg cm⁻² of catalyst loaded onto the carbon paper (active area: 1 cm²). The X-ray window was sealed with Kapton tape, and the total thickness of the cathode layer was 370-450 μm. EXAFS data processing was performed using the ATHENA module from the IFEFFIT software suite. The k³-weighted χ(k) spectra were transformed from k-space (3–10.5 Å⁻¹) to R-space using a Hanning window with dk = 1.0 Å⁻¹. To extract quantitative structural information around Cu atoms, curve fitting of the R-space data was performed using the ARTEMIS module of IFEFFIT. Fitting parameters and additional details are provided in Supplementary Tables 5 and 8. For in-situ measurements using a flow cell, the setup (Supplementary Fig. S38c) was mounted on the beamline X-Y stage at a 45° angle to the incoming beam to maximize the fluorescence signal. All in-situ measurements were collected in fluorescence mode. Prior to measurement, the gas compartment (2-mm thick, 0.2 cm³, cathodic chamber) was filled with He/N₂. Both the anolyte and catholyte were 1 M KOH. During CO₂RR, a 20 SCCM CO₂ gas flow was passed through the gas compartment. References data for oxide powders or metal foil were collected in transmission mode using ionization chambers, with the sampling chamber filled with N₂/He. For ex-situ samples, N₂/He was used to fill the cathodic chamber during XAS signal collection. ¹H NMR experiments were performed with a Bruker 400 MHz NMR spectrometer using a 5 mm broadband auto-tuneable probe with Z–gradients. Quantitative analysis of Cu loading in relevant samples was conducted via inductively coupled plasma mass spectrometry (ICP-MS, PerkinElmer, OPTIMA 7000DV). X-band EPR measurements (microwave frequency: 9.45597 GHz; power: 10 mW) were performed using an EPR MS-5000 spectrometer (Bruker, Germany). Samples were loosely packed into low-background 4 mm quartz paramagnetic tubes. Spectra were recorded with a modulation amplitude of 0.2 Gauss and a modulation frequency of 100 kHz. Raman measurements were conducted using a LABRAM HR EVO Raman microscope (equipped with a 50× objective, Leica, Germany) and a 532 nm laser. Acquisition parameters included an integration time of 6 s, 2 accumulations, a 600 nm grating, 10% laser power, and averaging over 3 scans per region.

CO₂ electrolysis measurement and product analysis

Electrochemical CO₂ reduction reaction (CO₂RR) experiments were performed in a thin liquid-electrolyte flow cell with a three-electrode configuration, controlled by a potentiostat (CHI1140C, data were exported as.txt files). The electrocatalyst (~4.0 mg) was mixed with 1 mL of 2-propanol and 10 μL of 5 wt% Nafion solution, followed by ultrasonication for 10 min to form an ink. The resulting ink (100 μL) was drop-cast onto a 28BC GDL (effective geometric area: 1 cm²) and dried under an infrared lamp. The catalyst-coated 28BC GDL was used as the cathode, contacted with copper foil tape for conductivity. A Hg/HgO (1 M KOH) and a nickel plate (0.1-mm thick) served as the reference and counter electrodes, respectively. 1 M KOH electrolyte was pumped into the catholyte (~0.2 cm³) and anolyte (~1 cm³) compartments, separated by an anion exchange membrane (Fumasep FAB-PK-130). The electrolyte flow rates for the cathodic and anodic sides were 5–10 and 35–70 mL min⁻¹, respectively. The cathodic gas chamber (~0.15 cm³) was continuously fed with 15–20 SCCM of CO₂ gas, controlled by a mass flow controller (CS200, Sevenstar). All applied potentials (unless stated otherwise) were calibrated to the reversible hydrogen electrode (RHE) with iR_u drop compensation using the equation: E_RHE = E_Hg/HgO + 0.14 + 0.0591 pH + η_IRu. R_u was measured via electrochemical impedance spectroscopy (EIS) at open circuit potential (OCP). EIS parameters were set with an AC potential of 10 mV and frequency sweeps from 10⁵ to 0.1 Hz. EIS data for Cu SACs were recorded in the same flow cell with a three-electrode configuration, using 1 M KOH as the supporting electrolyte. Impedance spectra were simulated using Zview.

For the calibration of the Hg/HgO electrode (1 M KOH), all electrochemical measurements were performed in a Teflon jar containing a redox solution to avoid any change in electrochemical activity caused by trace silicate contamination under ambient conditions. The redox solution was prepared by adding equimolar amounts of K₄[Fe(CN)₆] and K₃[Fe(CN)₆] to 1 M KOH (with a final concentration of ~1 mM for each). The Hg/HgO electrode and a Pt wire (serving as a counter electrode, though no current is required) were immersed in the redox solution. The OCP of the Pt wire versus the Hg/HgO electrode was measured using a potentiostat (CHI1140C). This OCP equals the formal potential of the redox couple (0.262 V) versus Hg/HgO. The measured OCP of the Hg/HgO electrode was verified to match this value (within ±5 mV) prior to use.

Gas product quantification was performed using an online gas chromatograph (Shimadzu, GC-2014C) equipped with a thermal conductivity detector (TCD) and a flame ionization detector (FID). Carrier gases included argon, nitrogen and hydrogen. Gas products were detected every 13 minutes using an automatic sampler to determine catalyst selectivity and corresponding partial current densities. The faradic efficiency (FE) of gas product was calculated using the following equation:

$${{{\rm{FE}}}}\left(\%\right)=\frac{{nc}\vartheta {{{\rm{F}}}}}{{IV}}*100\%$$

(1)

where n is the number of electrons transferred, c is the product concentration (ppm) detected by gas chromatography, F is Faraday’s constant, ϑ is the gas flow rate (SCCM) at the GC outlet, V is the molar volume (24.5 L mol⁻¹ at 298.15 K), and I (mA) is the total current at the start of sampling.

Liquid products were quantified using an NMR spectrometer with water suppression and pre-saturation methods. NMR samples were prepared by mixing 0.5 mL of catholyte solution, 0.10 mL of D₂O, and 100 ppm (v/v) DMSO (added as an internal standard).

The electrochemical active surface area (ECSA) of the Cu SACs was calculated using the following equation: ECSA = R_fS, where S is the ideal areal capacitance (10 μF cm⁻²) of a smooth carbon surface (in this work, S = 1 cm²). The roughness factor (R_f) was estimated relative to the ideal areal capacitance of a smooth carbon surface using the relation: Rf = C_dl/(10 μF cm⁻²). The double-layer capacitance (C_dl) was determined by measuring the capacitive current associated with double-layer charging from the scan rate dependence of cyclic voltammograms (CVs) (Supplementary Fig. S35). Specifically, C_dl was estimated by plotting j_capacity = (j_a − j_c)/2 at 1.058 V versus RHE against the scan rate, where the slope corresponds to C_dl. Here, j_a and j_c represent anodic and cathodic current densities, respectively.

In situ ATR-SEIRAS measurement

In situ ATR-SEIRAS tests were performed using a Fourier transform infrared (FTIR) spectrometer (Thermo Fisher Nicolet iS50) equipped with a Pike Technologies VeeMAX III ATR accessory and an MCT-A detector. A gold film was deposited on the Si prism via magnetron sputtering. The electrocatalyst (~3.2 mg) was mixed with 1 mL of 2-propanol and 20 μL of 5 wt% Nafion solution, followed by ultrasonication for 15 min to form an ink. The resulting ink was drop-cast onto the Au surface and dried under an infrared lamp. The Si prism was assembled into a home-made cell, with a graphite rod and an Ag/AgCl (saturated KCl) electrode serving as the counter and reference electrodes, respectively. During measurements, 0.5 M KHCO₃ electrolyte was pre-saturated and continuously bubbled with CO₂ gas. All spectra were recorded at designated potentials from 0 to −1.3 V vs. RHE, with a resolution of 4 cm⁻¹ and 64 scans.

Calculations

All DFT calculations were performed using the projector-augmented wave (PAW) method as implemented in the Vienna ab initio simulation package (VASP 5.4.4)^49,50,51,52. The revised Perdew–Burke–Ernzerh (PBE) functional (generalized gradient approximation) was used as the exchange-correlation functional, with long-range dispersion correction from Grimme’s DFT-D3 method, and a kinetic energy cutoff of 450 eV was employed⁵³. To describe localized Cu 3d electrons, the Hubbard model (DFT + U) was used with a U–J value of 3.0 eV⁵⁴. The CuN₄ SAC was modelled in a supercell with a vacuum thickness >20 Å along the z-direction. The Brillouin zone was sampled using gamma-cantered 1 × 1 × 1 k-points generated via the Monkhorst-Pack scheme. Structural optimization was considered converged when the residual force on each ion was <0.01 eV Å⁻¹ and the energy difference between consecutive iterations was <10⁻⁵ eV per atom.

On this basis, we simulated the adsorption behaviour of the *COOH, *CO, *CHO, *OCH₂, *OCH₃, *O and *OH intermediates for each catalyst, with each model optimized to convergence. The free energy change (ΔG) for each ECM step was calculated using the computational hydrogen electrode (CHE) model via the equation:

$$\Delta G=\Delta E+\Delta {{{\rm{ZPE}}}}-T\Delta S+\Delta {G}_{U}+\Delta {{{\rm{GpH}}}}$$

(2)

where ΔE is the total energy change from DFT calculation, ΔZPE is the zero-point energy change, T is temperature (298.15 K), and ΔS is the entropy change. Entropy was calculated using a harmonic oscillator approximation. The effect of applied bias on the free energy of electron transfer steps was assessed using ∆G_U = −neU (where U is the electrode potential relative to RHE, e is the elementary charge, and n is the number of electrons transferred). ∆GpH = 0.0591*pH. For the CHE model, acidic (pH = 0) and alkaline (pH = 14) conditions yielded identical results. Explicit solvent model calculations were performed to evaluate solvent effects in 1 M KOH (containing 55 H₂O molecules and 1 KOH molecule). All DFT energies, zero-point energies, and temperature-entropy corrections (T*S, T = 298.15 K) for relevant species are provided in Supplementary Tables 9 and 10.

The DFT-based climbing-image nudged elastic band (CI-NEB) method was employed⁵⁵. The initial structure was a Cu SA model with a 7-Cu-atom cluster, and the final structure was a defect model supporting an 8-Cu-atom cluster. After optimizing the initial and final structures, six transition state points were inserted using the VTST tool, and CI-NEB calculations were performed with a force convergence criterion of 0.05 eV Å⁻¹.

Data availability

All data generated in this study are provided in the Supplementary Information/Source data file. Source data are provided with this paper.

References

Artz, J. et al. Sustainable conversion of carbon dioxide: an integrated review of catalysis and life cycle assessment. Chem. Rev. 118, 434–504 (2018).
Article PubMed CAS Google Scholar
Li, F. et al. Molecular tuning of CO₂-to-ethylene conversion. Nature 577, 509–513 (2020).
Article PubMed CAS Google Scholar
Fan, M. et al. Single-site decorated copper enables energy- and carbon-efficient CO₂ methanation in acidic conditions. Nat. Commun. 14, 3314 (2023).
Article ADS PubMed PubMed Central CAS Google Scholar
Li, L. et al. Hydrophobicity graded gas diffusion layer for stable electrochemical reduction of CO₂. Angew. Chem. Int. Ed. 61, e202208534 (2022).
Article ADS CAS Google Scholar
Qiao, J., Liu, Y., Hong, F. & Zhang, J. A review of catalysts for the electroreduction of carbon dioxide to produce low-carbon fuels. Chem. Soc. Rev. 43, 631–675 (2014).
Article PubMed CAS Google Scholar
Nitopi, S. et al. Progress and perspectives of electrochemical CO₂ reduction on copper in aqueous electrolyte. Chem. Rev. 119, 7610–7672 (2019).
Article PubMed CAS Google Scholar
Wu, H. et al. Selective and energy-efficient electrosynthesis of ethylene from CO₂ by tuning the valence of Cu catalysts through aryl diazonium functionalization. Nat. Energy 9, 422–433 (2024).
Article ADS CAS Google Scholar
Tan, X. et al. Stabilizing copper by a reconstruction-resistant atomic Cu–O–Si interface for electrochemical CO₂ reduction. J. Am. Chem. Soc. 145, 8656–8664 (2023).
Article ADS CAS Google Scholar
Li, L. et al. Hydrophobicity graded gas diffusion electrode with enhanced CO intermediate coverage for high-performance electroreduction of CO₂ to ethylene. Appl. Catal. B Environ. 331, 122597 (2023).
Article CAS Google Scholar
Sun, L., Wang, Y., Guan, N. & Li, L. Methane activation and utilization: current status and future challenges. Energy Technol. 8, 1900826 (2020).
Article CAS Google Scholar
Hori, Y., Wakebe, H., Tsukamoto, T. & Koga, O. Electrocatalytic process of CO selectivity in electrochemical reduction of CO₂ at metal electrodes in aqueous media. Electrochim. Acta 39, 1833–1839 (1994).
Article CAS Google Scholar
Hori, Y., Kikuchi, K., Murata, A. & Suzuki, S. Production of methane and ethylene in electrochemical reduction of carbon dioxide at copper electrode in aqueous hydrogencarbonate solution. Chem. Lett. 15, 897–898 (1986).
Article Google Scholar
Ju, W. et al. Understanding activity and selectivity of metal-nitrogen-doped carbon catalysts for electrochemical reduction of CO₂. Nat. Commun. 8, 944 (2017).
Article ADS PubMed PubMed Central Google Scholar
Rong, W. et al. Size-dependent activity and selectivity of atomic-level copper nanoclusters during CO/CO₂ Electroreduction. Angew. Chem. Int. Ed. 60, 466–472 (2021).
Article CAS Google Scholar
Zhang, Y. et al. Coordination environment dependent selectivity of single-site-Cu enriched crystalline porous catalysts in CO₂ reduction to CH₄. Nat. Commun. 12, 6390 (2021).
Article ADS PubMed PubMed Central CAS Google Scholar
Zou, H. et al. Electronic perturbation of copper single-atom CO₂ reduction catalysts in a molecular way. Angew. Chem. Int. Ed. 62, e202217220 (2023).
Article CAS Google Scholar
Osella, S. & Goddard, W. A. CO₂ Reduction to methane and ethylene on a single-atom catalyst: a grand canonical quantum mechanics study. J. Am. Chem. Soc. 145, 21319–21329 (2023).
Article ADS PubMed PubMed Central CAS Google Scholar
Shi, G. et al. Constructing Cu−C bonds in a graphdiyne-regulated Cu single-atom electrocatalyst for CO₂ reduction to CH₄. Angew. Chem. Int. Ed. 61, e202203569 (2022).
Article ADS CAS Google Scholar
Cai, Y. et al. Insights on forming N,O-coordinated Cu single-atom catalysts for electrochemical reduction CO₂ to methane. Nat. Commun. 12, 586 (2021).
Article ADS PubMed PubMed Central CAS Google Scholar
Dai, Y. et al. Manipulating local coordination of copper single atom catalyst enables efficient CO₂-to-CH₄ conversion. Nat. Commun. 14, 3382 (2023).
Article ADS PubMed PubMed Central CAS Google Scholar
Zhao, P. et al. Construction of low-coordination Cu−C2 single-atoms electrocatalyst facilitating the efficient electrochemical CO₂ reduction to methane. Angew. Chem. Int. Ed. 62, e202314121 (2023).
Article CAS Google Scholar
Zhang, Z. et al. Distance effect of single atoms on stability of cobalt oxide catalysts for acidic oxygen evolution. Nat. Commun. 15, 1767 (2024).
Article ADS PubMed PubMed Central CAS Google Scholar
Jin, H. et al. Regulating the electronic structure through charge redistribution in dense single-atom catalysts for enhanced alkene epoxidation. Nat. Commun. 14, 2494 (2023).
Article ADS PubMed PubMed Central CAS Google Scholar
Wang, B. et al. A site distance effect induced by reactant molecule matchup in single-atom catalysts for fenton-like reactions. Angew. Chem. Int. Ed. 61, e202207268 (2022).
Article ADS CAS Google Scholar
Shan, J. et al. Metal-metal interactions in correlated single-atom catalysts. Sci. Adv. 8, eabo0762 (2022).
Article PubMed PubMed Central CAS Google Scholar
Jin, Z. et al. Understanding the inter-site distance effect in single-atom catalysts for oxygen electroreduction. Nat. Catal. 4, 615–622 (2021).
Article CAS Google Scholar
Lee, H. A., Park, E. & Lee, H. Polydopamine and its derivative surface chemistry in material science: a focused review for studies at KAIST. Adv. Mater. 32, 1907505 (2020).
Article CAS Google Scholar
Salomäki, M. et al. Polydopamine nanoparticles prepared using redox-active transition metals. J. Phys. Chem. B 123, 2513–2524 (2019).
Article PubMed PubMed Central Google Scholar
Lee, H., Dellatore, S. M., Miller, W. M. & Messersmith, P. B. Mussel-inspired surface chemistry for multifunctional coatings. Science 318, 426–430 (2007).
Article ADS PubMed PubMed Central CAS Google Scholar
Ball, V. et al. Comparison of synthetic dopamine-eumelanin formed in the presence of oxygen and Cu²⁺ cations as oxidants. Langmuir 29, 12754–12761 (2013).
Article PubMed CAS Google Scholar
Bernsmann, F. et al. Dopamine-melanin film deposition depends on the used oxidant and buffer solution. Langmuir 27, 2819–2825 (2011).
Article PubMed CAS Google Scholar
Han, L. et al. A single-atom library for guided monometallic and concentration-complex multimetallic designs. Nat. Mater. 21, 681–688 (2022).
Article ADS PubMed CAS Google Scholar
Qu, Y. et al. Direct transformation of bulk copper into copper single sites via emitting and trapping of atoms. Nat. Catal. 1, 781–786 (2018).
Article CAS Google Scholar
Sui, X. et al. Advanced support materials and interactions for atomically dispersed noble-metal catalysts: from support effects to design strategies. Adv. Energy Mater. 12, 2102556 (2022).
Article CAS Google Scholar
Li, J. et al. Highly active and stable metal single-atom catalysts achieved by strong electronic metal-support interactions. J. Am. Chem. Soc. 141, 14515–14519 (2019).
Article ADS PubMed CAS Google Scholar
Arčon, I., Mirtič, B. & Kodre, A. Determination of valence states of chromium in calcium chromates by using X-ray Absorption Near-Edge Structure (XANES) spectroscopy. J. Am. Ceram. Soc. 81, 222–224 (1998).
Article Google Scholar
Li, Q. et al. Fe isolated single atoms on S, N codoped carbon by copolymer pyrolysis strategy for highly efficient oxygen reduction reaction. Adv. Mater. 30, 1800588 (2018).
Article ADS Google Scholar
Kumar, A. et al. Moving beyond bimetallic-alloy to single-atom dimer atomic-interface for all-pH hydrogen evolution. Nat. Commun. 12, 6766 (2021).
Article ADS PubMed PubMed Central CAS Google Scholar
Yin, J. et al. Customizable CO₂ electroreduction to C1 or C2+ products through Cuy/CeO₂ interface engineering. ACS Catal. 12, 1004–1011 (2022).
Article CAS Google Scholar
Zhao, J. et al. Modulation of *CH_xO adsorption to facilitate electrocatalytic reduction of CO₂ to CH₄ over Cu-based catalysts. J. Am. Chem. Soc. 145, 6622–6627 (2023).
Article ADS PubMed CAS Google Scholar
Li, Z. et al. Interface-enhanced catalytic selectivity on the C2 products of CO₂ electroreduction. ACS Catal. 11, 2473–2482 (2021).
Article Google Scholar
Chen, X. et al. Highly stable layered coordination polymer electrocatalyst toward efficient CO₂-to-CH₄ conversion. Adv. Mater. 36, 2310273 (2024).
Article CAS Google Scholar
Handoko, A. D., Wei, F., Jenndy, Yeo, B. S. & Seh, Z. W. Understanding heterogeneous electrocatalytic carbon dioxide reduction through operando techniques. Nat. Catal. 1, 922–934 (2018).
Article CAS Google Scholar
Su, J. et al. Strain enhances the activity of molecular electrocatalysts via carbon nanotube supports. Nat. Catal. 6, 818–828 (2023).
Article CAS Google Scholar
Hammer, B. & Norskov, J. K. Why gold is the noblest of all the metals. Nature 376, 238–240 (1995).
Article ADS CAS Google Scholar
Nørskov, J. K., Abild-Pedersen, F., Studt, F. & Bligaard, T. Density functional theory in surface chemistry and catalysis. PNAS 108, 937–943 (2011).
Article ADS PubMed PubMed Central Google Scholar
Wang, T. et al. Single-atom anchored curved carbon surface for efficient CO₂ electro-reduction with nearly 100% CO selectivity and industrially-relevant current density. Adv. Mater. 35, e2205553 (2023).
Article PubMed Google Scholar
Han, G. et al. Substrate strain tunes operando geometric distortion and oxygen reduction activity of CuN₂C₂ single-atom sites. Nat. Commun. 12, 6335 (2021).
Article ADS PubMed PubMed Central CAS Google Scholar
Kresse, G. & Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 47, 558–561 (1993).
Article ADS CAS Google Scholar
Kresse, G. & Hafner, J. Ab initio molecular-dynamics simulation of the liquid-metalamorphous- semiconductor transition in germanium. Phys. Rev. B 49, 14251–14269 (1994).
Article ADS CAS Google Scholar
Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953–17979 (1994).
Article ADS Google Scholar
Joubert, D. From ultrasoft pseudopotentials to the projector augmented-wave method. Phys. Rev. B Condens. Matter Mater. Phys. 59, 1758–1775 (1999).
Article ADS Google Scholar
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).
Article ADS PubMed CAS Google Scholar
Dudarev, S. L., Botton, G. A., Savrasov, S. Y., Humphreys, C. J. & Sutton, A. P. Electron-energy-loss spectra and the structural stability of nickel oxide: An LSDA + U study. Phys. Rev. B 57, 1505–1509 (1998).
Article ADS CAS Google Scholar
Sheppard, D., Terrell, R. & Henkelman, G. Optimization methods for finding minimum energy paths. J. Chem. Phys. 128, 134106 (2008).
Article ADS PubMed Google Scholar

Download references

Acknowledgements

The authors gratefully acknowledge financial supports from the National Natural Science Foundation of China (52201237, 52125105, 22502229), Shenzhen Science and Technology Innovation Bureau (KQTD2022110109364705, ZDSYS20210706144000003, KCXFZ20240903093801003), the Talent Introduction Project of Chinese Academy of Sciences (E344011), the Joint Research Project of China Merchants Group and SIAT (E2Z1521), the Cross Institute Joint Research Youth Team Project of SIAT (E25427), and the China Postdoctoral Science Foundation (2024M753384). This research was conducted at the XAS beamline at the Australian Synchrotron, part of ANSTO.

Author information

These authors contributed equally: Linbo Li, Xin Lei.

Authors and Affiliations

Shenzhen Key Laboratory of Energy Materials for Carbon Neutrality, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
Linbo Li, Xin Lei, Zhilong Zheng, Yingjun Dong, Haohui Chen, Yi Zhong, Yongping Zheng, Yongbing Tang, Xiaolong Zhang & Hui-Ming Cheng
Faculty of Materials Science and Energy Engineering, Shenzhen University of Advanced Technology, Shenzhen, China
Linbo Li, Yongbing Tang, Xiaolong Zhang & Hui-Ming Cheng
Shenzhen College of Advanced Technology, University of Chinese Academy of Sciences, Shenzhen, China
Xin Lei
School of Physics, Huazhong University of Science and Technology, Wuhan, China
Zhilong Zheng
Intelligent Polymer Research Institute, Innovation Campus, University of Wollongong, Squires Way, North Wollongong, NSW, Australia
Jun Chen
Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, Shenyang, China
Hui-Ming Cheng

Authors

Linbo Li
View author publications
Search author on:PubMed Google Scholar
Xin Lei
View author publications
Search author on:PubMed Google Scholar
Zhilong Zheng
View author publications
Search author on:PubMed Google Scholar
Yingjun Dong
View author publications
Search author on:PubMed Google Scholar
Haohui Chen
View author publications
Search author on:PubMed Google Scholar
Jun Chen
View author publications
Search author on:PubMed Google Scholar
Yi Zhong
View author publications
Search author on:PubMed Google Scholar
Yongping Zheng
View author publications
Search author on:PubMed Google Scholar
Yongbing Tang
View author publications
Search author on:PubMed Google Scholar
Xiaolong Zhang
View author publications
Search author on:PubMed Google Scholar
Hui-Ming Cheng
View author publications
Search author on:PubMed Google Scholar

Contributions

X.Z., Y.T. and H.C. supervised this project. L.L. carried out material synthesis, characterization, electrochemical experiments, and data analyses. X.L. implemented DFT calculations. Z.Z., H.C. (Haohui Chen), Y.D., and Y.Z. performed electrochemical measurements. J.C. performed characterization. L.L. drafted the manuscript with input from all other authors. All authors discussed the results and contributed to the final manuscript.

Corresponding authors

Correspondence to Yongbing Tang, Xiaolong Zhang or Hui-Ming Cheng.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary Data 1

Transparent Peer Review file

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Li, L., Lei, X., Zheng, Z. et al. Tuning binding strength between single metal atoms and supports enhances electrochemical CO₂ methanation. Nat Commun 16, 8641 (2025). https://doi.org/10.1038/s41467-025-63781-9

Download citation

Received: 08 December 2024
Accepted: 28 August 2025
Published: 30 September 2025
DOI: https://doi.org/10.1038/s41467-025-63781-9