A First-Principles Screening for Axial Ligand Regulation of Electrocatalytic Carbon Dioxide Reduction on Dual-Metal Atomic Catalysts (M₁M₂N₆‑R)

Abstract

A primary challenge in the carbon dioxide reduction reaction (CO₂RR) is the rational design and engineering of high-efficiency electrocatalysts. A series of M₁M₂N₆ catalysts (M₁M₂ = NiNi, CoNi, CoFe, CoCo) with precisely tailored axial ligands (R = –OH, –COH, –CN) have been high-throughput screened out to exhibit optimal electrocatalytic activity, which is extended to further estimate their CO₂RR performance in this work. The adsorption energies of three distinct ligands at the M₁–M₂ bridge site are evaluated to quantitatively assess the ligand stabilization. On pristine and ligand-engineered M₁M₂N₆ catalysts, the free energy variation along CO₂RR pathways leading to C1 products reveals that the initial proton-coupled electron transfer to form the *HCOO/*COOH intermediate is the main potential-limiting step of yielding the key intermediate CO*. The formation barrier energy difference of <0.06 eV between *HCOO and *COOH intermediates on pristine CoCo/CoFe/CoNi and CN-functionalized CoFe/CoCo catalysts facilitates *CO intermediate generation and enables the subsequent *CO–*CO coupling to C2 products for formation of C₂H₅OH and C₂H₆. However, –COH and –OH modification excludes *CO-intermediate formation and directs the reaction toward CH₄ and CH₃OH production due to the large kinetic energy difference of 0.96–1.11 eV between *HCOO and *COOH. Our results provide a possible axial ligand engineering strategy of regulating C1/C2 product selectivity on different dual-atom catalysts.

Introduction

The electrochemical CO₂ reduction reaction (CO₂RR) has become a promising strategy to convert CO₂ into valuable chemicals and fuels by coupling the CO₂RR with renewable electricity to mitigate carbon emissions and produce a sustainable carbon cycle. − One of the challenges of the CO₂RR at room temperature is the low selectivity and efficiency due to the produced multicarbon products. It is highly demanded to develop highly efficient CO₂RR electrocatalysts for improving the high-value-added hydrocarbon selectivity. Although commonly used metal-based catalysts like Cu have exhibited unique catalytic activity of CO₂ conversion, Cu has been particularly notable for its ability to facilitate multicarbon products, while Al and Au have demonstrated high selectivity toward CO or formate. −

Through tailoring metals into atomic-level single-atom catalysts (SACs), the conversion efficiency and product selectivity have been obviously improved by maximum atomic utilization and specific active atom regulation. Single-atom nitrogen-doped carbon catalysts (M–N–C) are a typical SAC and receive considerable attention due to their controllable synthesis and tunable electronic structures. − For example, CO₂RR catalysts of Ni-SAC with unsaturated Ni–N coordination have been experimentally synthesized, achieving a 96% CO Faradaic efficiency at 577 mA·cm^–2 and an 85% CO selectivity at 200 mA·cm^–2 under neutral pH due to weak *CO binding on Ni–N₂ motifs. , It is well known that the CO–CO coupling ability determines the selectivity toward C2 products. Dual-metal atomic catalysts (DACs) have attracted growing interest due to their enhanced tunable dual-metal synergistic effects. A DAC of Ni–Ag on nitrogen-rich porous carbon was reported to achieve nearly 100% CO selectivity at −0.8 V vs RHE. Fe–Ni single-atom pairs on MOF-derived N-doped carbon were also demonstrated to enhance CO₂RR performance through nonbonding metal interactions.

Axial ligand engineering has been utilized to regulate the activity of SAC catalysts along CO₂RR processes. − The introduction of axial ligands into Fe–N–C catalysts with –NH₂, –COH, and –F was demonstrated to influence O₂ adsorption and activation ability via altering Fe-3d electron distribution. , OH-ligand coordination on FeNiN₆ provided an electron-rich catalytic environment around FeNi dual-metal sites to synergistically enhance CO₂ activation of boost the CO₂RR while suppressing the HER. The modification of axial ligands including –O, –N, –F, and –Cl into M–N₄ catalysts has been demonstrated to enhance their CO₂RR activity, outcompeting the original M–N₄ structures. − Axial O-coordinated FeN₄ has also shown better activity and selectivity for CO production along CO₂RR than pristine FeN₄ via altering Fe-3d electron distribution. This axial ligand engineering serves as an effective strategy for enhancing CO₂RR performance on electrocatalysts.

Through high-throughput screening of 13 distinct ligands on 55 randomly combined M₁M₂N₆ (M₁M₂ = 3d TMs) DAC catalysts, we have previously screened out four typical M₁M₂N₆ dual-metal catalysts and three optimal ligands of –OH, –COH, and –F with excellent ORR performance when compared to Pt/C. Herein, these four M₁M₂N₆ catalysts (M₁M₂ = NiNi, CoNi, CoFe, and CoCo) modified with three axial ligands are extended to evaluate the axial ligand effect on CO₂RR conversion efficiency and product selectivity. The adsorption energies of three ligands on the M₁–M₂ bridge site are calculated to evaluate the ligand-binding stability. The adsorbed CO*-intermediate is an essential precursor related to the selectivity of the C2 product, which has been compared on pristine and ligand-modified M₁M₂N₆ catalysts. The subsequent CO–CO coupling kinetic process was probed to evaluate the ligand effect on C2 products through dual-metal synergies. Along with the calculated free energy variation of the CO₂RR to main C1 products on pristine M₁M₂N₆ and ligand-modified catalysts, the first hydrogenation on C or O (CO₂) to form the *HCOO or *COOH intermediate is a key step of producing the CO* intermediate and the following CO*–CO* coupling. The electron-donation ability of M₁M₂ is found to be highly related to the binding strength with the axial ligands. We expect to provide an efficient ligand engineering strategy for DAC to affect the key CO* intermediate formation and hence regulate C1 and C2 product selectivity.

Computational Details

Computational Parameter Setting

All DFT calculations were performed by using the Vienna ab initio Simulation Package (VASP5.4) , based on the plane wave functional. The core electrons and electron interactions were evaluated by the projected augmented wave (PAW) and Perdew–Burke–Ernzerhof (PBE) generalized gradient approximation (GGA), respectively. The BEEF-vdW functional was used to describe the long-range dispersion interactions. The plane-wave truncation energy was set to 500 eV, and the sampling of the Brillouin zone was based on a Gamma scheme with a 2 × 2 × 1 K-point grid centered on the gamma-point for structural optimization. The electron energy and force converged to 1.0 × 10^–4 eV and −0.03 eV/Å, respectively. The search for the transition states (TS) of CO*–CO* coupling was conducted with the climbing-image nudged elastic band (CI-NEB) method.

Model Construction

Based on our previously screened results, axial ligand (R)-modified M₁M₂N₆(-R) models (M₁M₂ = CoCo, CoNi, CoFe, NiNi, R = –OH, –COH, and –CN) are constructed from 6 × 6 pristine graphene containing 2 metal atoms, 6 nitrogen atoms, and 62 carbon atoms, as presented in Figure S1. The adsorption structures of all reaction intermediates on M₁M₂N₆ are shown in Figure S2. The M₁–M₂ bridge is identified as the favorable site of binding with the axial groups (−OH, –COH, and –CN) by evaluating their adsorption energy (E _ads). The free energy variations (ΔG) associated with each elementary step of proton and electron (H⁺ + e^–) transfer along the electrochemical CO₂RR are determined by employing the computational hydrogen electrode (CHE) model. The free energy variations are obtained using ΔG = ΔE + ΔE _ZPE – TΔS, where ΔE is the calculated electronic total energy difference and ΔE _ZPE and ΔS are zero-point energy corrections obtained by vibrational analysis and entropy changes between the reactants and products. T is the temperature (T = 298 K).

Mechanism of the CO₂RR for C1 and C2 Products

Scheme presents the possible mechanistic pathways governing the CO₂RR on M₁M₂N₆-R, elucidating the selectivity toward C1 and C2 products. Along the C1 formation pathway, two hydrogenation sites on the adsorbed CO₂ are involved: one is the C-site, yielding the *HCOO intermediate, followed by either HCOOH desorption or dehydration to *HCO, and the other is the O-site, which produces the *COOH intermediate and converts to *CO after removing H₂O. Due to the general high desorption energy barrier of the produced gas CO, *CO is more favorable to couple with another *CO to initiate the C2 pathway via C–C coupling. The alternative path will undergo sequential hydrogenation to form the CH₃OH product via the steps of *HCO → *H₂CO → *CH₃OH. The final C1 product (CH₄) is obtained after one hydroxyl group is lost, accompanied by hydrogenation on *CH₃OH. Along the C2 pathway, the *CO–CO dimer hydrogenates to *COHCH and then branches into O-retained intermediates (*CHCHOH → C₂H₅OH) or deoxygenated *CCH to yield C₂H₆.

1. Possible Reaction Pathways of the CO₂RR on M₁M₂N₆-R Catalysts.

Results and Discussion

The Stability of M₁M₂N₆-R

The adsorption energy (E _ads) is a crucial parameter for understanding the stability of M₁M₂N₆-R. To evaluate the stability of M₁M₂N₆-R, E _ads is calculated by using $E_{\mathrm{a}\mathrm{d}\mathrm{s}}=E_{{\mathrm{M}}_1{\mathrm{M}}_2{\mathrm{N}}_6‐\mathrm{R}}-E_{{\mathrm{M}}_1{\mathrm{M}}_2{\mathrm{N}}_6}-E_{\mathrm{R}}$ , where $E_{{\mathrm{M}}_1{\mathrm{M}}_2{\mathrm{N}}_6‐\mathrm{R}}$ and $E_{{\mathrm{M}}_1{\mathrm{M}}_2{\mathrm{N}}_6}$ are the total energies of M₁M₂N₆-R and pristine M₁M₂N₆, respectively, and E _R denotes the energy of the axial ligands. As illustrated in Figure , the adsorption energies of –OH, –COH, and –CN ligands on dual-metal catalysts reveal distinct ligand–catalyst interaction trends. Their structural stabilities are highly related to the electron-donation abilities of M₁ or M₂ to the surrounding nitrogen atoms. For the OH-ligand, the calculated negative E _ads of −0.46, −0.36, and −0.22 eV corresponding to CoCo, CoFe, and CoNi imply the favorable ligand coordination, while the calculated positive E _ads = 0.36 eV indicates unfavorable OH coordination on NiNi. Similarly, COH-ligand coordination also appears to stabilize CoCo, CoFe, and CoNi with negative E _ads of −0.58, −0.12, and −0.26 eV, respectively, except for NiNi with positive E _ads = 0.90 eV. The CN ligand universally enhances the binding stabilities for the four M₁M₂N₆, with all negative E _ads values ranging from −1.18 (CoFe) to −0.53 eV (NiNi) in Figure b. Although the CN-ligand could stably coordinate on NiNiN₆, the calculated total charge distribution (ρ = ρ_M1 + ρ_M2 ) in Figure c indicates that Ni–Ni possesses the smallest total positive charge (1.20 e). The relatively poor electron-donation ability of NiNiN₆ is ascribed to the weak adsorption strength toward the axial ligands and hence is not discussed in the following CO₂RR catalytic performance. Only three DAC catalysts, CoCoN₆, CoFeN₆, and CoNiN₆, with three ligands of –OH, –COH, and –CN, are further studied to evaluate the ligand effect on CO₂RR selectivity toward different C1 and C2 products.

1.

(a) The optimized structural models of M₁M₂N₆-R. (b) Calculated formation energies of M₁M₂N₆-R structures. (c) The calculated total charge distribution of pristine M₁M₂N₆.

CO₂RR to C1 Products on Pristine M₁M₂N₆/C

As discussed in the evolution of the C1 pathway in Scheme , two important intermediates of *HCOO and *COOH are possibly formed via protonation on the C or O site in the initial *CO₂ adsorption state. It is notable that both intermediates undergo nearly barrierless hydrogenation (<0.04 eV) on pristine CoCoN₆, CoFeN₆, and CoNiN₆ without axial R-groups, as plotted in Figure a–c. Proceeding from the *COOH intermediate, the subsequent dehydration yields the adsorbed *CO state barrierlessly. In contrast, hydrogenation of the *HCOO intermediate to form *HCOOH will require significant Gibbs free energy (ΔG) ranging from 0.66 to 0.88 eV. Specifically, CoNi achieves the lowest ΔG of 0.66 eV, while CoFe yields the highest activation barrier of 0.88 eV in forming *HCOOH. The subsequent reaction will adopt two pathways: one is direct desorption to form the C1 product of HCOOH with ΔG = 0.07–0.42 eV for CoCoN₆ to CoNiN₆ catalysts, the other one is dehydration to form *HCO intermediates with no barrier for all three catalysts.

2.

Reaction free energy variations (ΔG) along the process of CO₂RR to C1 products on the pristine three M₁M₂N₆/C catalysts of (a) CoCoN₆, (b) CoFeN₆, and (c) CoNiN₆. The background in orange emphasizes the formation of CO and HCOOH, while the blue color emphasizes the further formation of CH₃OH and CH₄. (d) Summarizes the maximum ΔG (ΔG _max) for yielding the possible C1 product.

Starting from the key intermediate *HCO, further hydrogenation will generate *H₂CO through the activation barriers ranging from 0.09 to 0.79 eV. The *H₂CO intermediate is selectively driven into *H₃CO or *CH₂OH depending on the protonation on C or O sites. Notably, CoCo and CoNi enable barrierless *H₃CO formation, whereas CoFe requires a moderate barrier of 0.12 eV. In comparison, only CoNi achieves barrierless *CH₂OH formation, with CoCo and CoFe necessitating activation energies of 0.25 and 0.35 eV, respectively. Both *H₃CO and *CH₂OH intermediates converge to *CH₃OH through barrierless hydrogenation. Two distinct barrierless pathways following *CH₃OH are divided: one is direct CH₃OH desorption to form the CH₃OH product, and the other is dehydration to *CH₃ and then hydrogenation to form the CH₄ product. The thermodynamic stability of the adsorbed *CO intermediate determines that the rate-determination step (RDS) for CH₃OH/CH₄ occurs at either the *CO → *HCO transition state or the subsequent *HCO–*H₂CO conversion with a ΔG _max of 0.79 to 1.10 eV. Figure d summarizes the maximum ΔG (denoted as ΔG _max) of four possible C1 products; it is observed that the free gas CO molecule is difficult to form due to the highest activation barriers of 1.87–2.42 eV, which allows the following hydrogenation of *CO to *HCO intermediates or C–C coupling to initiate the C2 pathway. HCOOH is a favorable C1 product with a relatively smaller ΔG _max of 0.66–0.88 eV, comparable to further hydrogenation of *HCO with 0.55–1.10 eV, and then forms CH₃OH and CH₄.

CO₂RR to C1 Products on R-Ligand-Modified M₁M₂N₆/C

Ligand modification of –OH, –COH, and –CN is generally accompanied by experimental solvents, which have been screened out to improve ORR performance after modification of these four M₁M₂N₆ catalysts. Since –COH is a moderate electron-withdrawing group due to its smallest charge density of M₁M₂ among the three ligands (Figure d), it serves as a representative model to elucidate how the ligand effect influences selectivity toward distinct C1 products, as plotted in Figure a–c. The corresponding reaction pathways on –OH- and –CN-modified catalysts are comparatively presented in Figures S3–S8. COH-functionalized DAC catalysts (CoCo/CoFe/CoNi) enable negligible barrierless *HCOOH formation (≤0.03 eV) while imposing substantial activation barriers for *COOH generation in the order of 0.69 (CoFe) < 0.83 (CoCo) < 1.14 eV (CoNi). It is obviously noticed that selective hydrogenation at the CO₂ carbon site preferentially stabilizes the *HCOO intermediate, effectively suppressing *COOH formation and subsequent pathways toward *CO derivatives or C2 products via CO coupling. Since COH ligand modification typically influences the formation energy difference of *HCOO and *COOH, we compare d-band center (ε_d) variation between CoCoN₆, CoFeN₆, and CoNiN₆ with and without COH modification, as presented in Figure S9. For example, the downshift ε_d = −1.58 eV on metal Co after –COH modification (pristine Co: −1.18 eV) weakens the hybridization between the metal Co-3d orbital and C-2p antibonding orbital of CO₂ and thereby enlarges the formation energy barrier for *COOH to 0.96–1.10 eV. A similar trend could be observed for CoFeN₆ and CoNiN₆ before and after COH modification, which aligns with the recent reports on d-band-mediated selectivity control. ,

3.

Reaction free energy diagrams of the CO₂RR to C1 product comparison on R-modified M₁M₂N₆/C. (a) CoCoN₆, (b) CoFeN₆, and (c) CoNiN₆; (d) the calculated total charge distribution of pristine M₁M₂N₆-R. (e) The *COOH and *HCOO formation barriers are summarized to illustrate the ligand effect.

Although *HCOO formation exhibits negligible energy barriers, subsequent hydrogenation requires substantial activation energies (0.68–0.84 eV), proceeding through two competing pathways: direct HCOOH desorption with low barriers (0.06–0.20 eV) or exothermic *HCO formation via H₂O removal. Energetic analysis reveals barrierless *HCO formation on all catalysts except CoNi (ΔG = 0.28 eV). The subsequent multistep hydrogenation on *HCO shows enhanced thermodynamic accessibility with activation barriers ≤0.28 eV. COH-modified CoCo, CoFe, and CoNi catalysts demonstrate dominant C1 product selectivity (CH₃OH/CH₄). The absence of *CO intermediate formation and accompanied *CO-*CO coupling in COH-modified catalysts effectively suppresses C2 product generation, thereby enhancing C1 product selectivity compared to pristine counterparts. A similar trend could be found for the three OH-modified catalysts, as plotted in Figures S3–S5, where a larger activation barrier of *COOH (0.17–0.60 eV) than no barriers of *HCOO obviously excludes the *CO formation and the following *CO–*CO coupling. CN modification exhibits a similar effect on CoCo and CoNi catalysts with significantly larger barriers of *COOH than no barriers of *HCOO, while CN-ligand modification on CoFe exhibits nearly negligible effect on formation of *COOH and *HCOO due to only a 0.06 eV barrier of *HCOOH, which leads to a possible *CO intermediate and the following C–C coupling, as shown in Figures S6–S8.

Figure e quantifies the ligand-modulated pathway selectivity through comparative analysis of *COOH vs *HCOO formation barriers. Hydrogenation at the C-site exhibits nearly barrierless *HCOO generation (ΔG ≤ 0.03 eV) for all systems. Except for three pristine catalysts and CN-ligand CoCo/CoFe systems, *COOH exhibits a competitive pathway with barriers <0.06 eV, while other ligand-modified catalysts show significant *COOH formation activation energies. Specifically, the barriers of *COOH are about 0.96–1.11 eV higher than *HCOOH on COH-modified CoCo/CoFe/CoNi catalysts. The ligand coordination of COH and OH significantly affects the formation kinetic barrier of *COOH and *HCOO intermediates and hence produces different pathways to C1 or C2 products.

Ligand Effect on the CO₂RR to C2 Products via *CO*CO Coupling

C–C coupling along the CO₂RR process serves as an essential precursor for multicarbon product formation, which is critically dependent upon the availability of *CO intermediates. As discussed above, the barrier-free formation of *COOH intermediates enables subsequent *CO generation, thereby facilitating *CO dimerization pathways on pristine CoCo, CoFe, and CoNi catalysts. Ligand-modulated pathways reveal that only CoCo–CN (barrierless) and CoFe–CN (ΔG = 0.06 eV) enable feasible *COOH → *CO conversion among COH/OH/CN modifications, as evidenced by plotted energy profiles in Figures S6–S7. The following possible CO*–CO* dimerization from coadsorbed configurations on pristine dual-metal sites (M₁–M₂) and CoCo–CN/CoFe–CN is hence carried out by quantifying C–C coupling energetics via CI-NEB calculations in Figures a and S10–S13. It is noticeable that CoCo–CN is identified as the optimal C–C coupling catalyst with the lowest energy barrier of 0.70 eV, contrasting sharply with the prohibitive 1.38 eV barrier for CoFe–CN, accompanied by moderate activation energies of 1.00 (CoCo), 0.97 (CoFe), and 1.02 eV (CoNi).

4.

(a) The C–C coupling reaction process on CoCo–CN; (b) the summarized C–C coupling energy barriers (E _b) related to d _C–C (Å) of the TS corresponding to pristine CoCo/CoFe/CoNi, CoCo–CN, and CoFe–CN.

Along the CO–CO coupling process, the related kinetic barriers and C–C bond lengths (d _C–C) of TS are correlated in Figure b, where larger bond lengths of C–C correspond to higher coupling barriers. For example, CoCo–CN shows a C–C bond length of 1.570 Å at the TS, resulting in the lowest coupling barrier of 0.70 eV, while a C–C bond length of 1.872 Å corresponds to the highest barrier of 1.38 eV. Based on Hammond’s postulate, − a TS energy similar to that of the reactant/product in a chemical reaction leads to a similar structure to the reactant/product. CO–CO coupling is an exothermic reaction for these DAC catalysts; all TS emerge near the product with a C–C bond length of ∼1.54 Å in the order of 1.570 Å (CoCo–CN) < 1.617 Å (CoFe) < 1.775 Å (CoCo) < 1.852 Å (CoNi) < 1.872 Å (CoFe–CN). Therefore, the CoCo–CN catalyst produces the most similar TS to the CO–CO product, which leads to the lowest energy barrier.

Figure presents the step-by-step hydrogenation of discrete carbon centers versus conjugated C–C bonds to probe formation pathways of C2 products of ethanol (C₂H₅OH) and ethane (C₂H₆), with selectivity governed by intermediate stabilization strength and bond activation site accessibility. As illustrated in Figure , the distinct selectivity observed across the five *CO-generating catalysts including pristine CoCo/CoFe/CoNi and CN-modified CoCo/CoFe is caused by the differences in their rate-determining step (RDS) and unique electronic interactions. Specifically, the *COCHOH → *COHCHOH step serves as the RDS for pristine CoCo and CoFe–CN catalysts, with activation barriers of 0.35 and 0.55 eV, respectively. In contrast, pristine CoNi, CoFe, and CoCo–CN catalysts have the COCHO → COCHOH step as the RDS, corresponding to activation energies of 0.80, 0.92, and 0.59 eV.

5.

Reaction free energy diagrams of the CO₂RR to C₂H₅OH and C₂H₆ on (a) CoCoN₆, (b) CoCoN₆–CN, (c) CoFeN₆, (d) CoFeN₆–CN, and (e) CoNiN₆; (f) the free energy of the CO₂RR to C2 for the catalyst.

For the final C2 product formation, C₂H₅OH is generated via oxygen-retaining hydrogenation pathways, exhibiting lower activation barriers (≤0.24 eV) on pristine CoCo, CoFe, and CoFe–CN catalysts. In contrast, C₂H₆ formation requires slightly higher energy barriers of E _b = 0.60, 0.35, and 0.38 eV, respectively, on these same catalysts. Notably, C₂H₆ formation on CoCo–CN and pristine CoNi displays significantly reduced energy demands, with no barrier and an E _b of 0.21 eV, respectively, compared to the higher barriers of 0.18 eV (Figure b) and 0.35 eV (Figure e) observed for competing pathways. However, the comparatively higher energy barriers (E _b) in the first two reaction steps promote the simultaneous formation of both C2 products (C₂H₅OH and C₂H₆) across all catalysts except CoCo.

Conclusion

This study demonstrates that axial ligand engineering enables precise control over CO₂ reduction pathways on dual-metal nitrogen-coordinated catalysts (M₁M₂N₆-R) by modulating intermediate stabilization and kinetic bottlenecks. All three ligands enhance the binding stabilities on CoCo, CoFe, and CoNiN₆ catalysts, with negative adsorption energies ranging from −1.18 (CoFe–CN) to −0.12 eV (CoFe–COH). Although the CN-ligand could stably coordinate to NiNiN₆, the smallest positive charge indicates its relatively poor electron-donation ability with ligands. Two important intermediates of *HCOO and *COOH are possibly formed via protonation on the C or O site of the initial *CO₂ adsorption state. Their kinetic energy barrier difference determines *CO formation and hence influences the following *CO–*CO coupling to form multicarbon products. Hydrogenation at the C-site exhibits nearly barrierless *HCOO generation (ΔG ≤ 0.03 eV) for all systems. Except for the three pristine catalysts and CN-ligand CoCo/CoFe systems, *COOH exhibits a competitive pathway without barriers of <0.06 eV; other ligand-modified catalysts show significant *COOH formation activation energies; especially, the barriers of *COOH are about 0.96–1.11 eV higher than *HCOOH on COH-modified CoCo/CoFe/CoNi catalysts. For the final C2 product formation, C₂H₅OH is generated via oxygen-retaining hydrogenation pathways, exhibiting lower activation barriers (≤0.24 eV) on pristine CoCo, CoFe, and CoFe–CN catalysts. In contrast, C₂H₆ formation requires slightly higher energy barriers (E _b = 0.60, 0.35, and 0.38 eV, respectively) on these same catalysts. Our study is expected to provide a rational axial ligand engineering regulation strategy to improve C1/C2 product selectivity on different dual-atom catalysts (DACs).

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c05014.

• Thirteen additional figures, including the structure of M₁M₂N₆-R catalysts, optimized adsorption structures of all reaction intermediates, reaction free energy diagrams, and calculated band center of pristine M₁M₂N₆-R and C–C coupling barriers (PDF)

The authors declare no competing financial interest.