Synergistic Effect of the Heteronuclear Double Sites in C₉N₄ on the Electrochemical Reduction of CO₂ to CO

Abstract

In response to the detrimental impact of excessive fossil fuel usage on the environment and the looming energy crisis, the electrochemical reduction of carbon dioxide (CO₂RR) has emerged as a promising solution. This study investigates the potential of dual-atom catalysts, specifically boron (B) and transition metal (TM) co-modified C₉N₄, for efficient CO₂RR. The 2 × 2 × 1 supercell of C₉N₄, considering modification with 26 TM and B atoms, demonstrated stability, confirmed by binding and formation energy calculations. Molecular dynamics simulations further supported the thermal stability of the studied catalysts. The modified structures exhibited metallic behavior, suggesting potential facilitation of electron transfer during electroreduction. Furthermore, by conducting Gibbs free energy calculations on CO₂ reduction pathways, seven low overpotential catalysts were screened out. Considering the competitive hydrogen evolution reaction (HER), Sc-B and Hf-B demonstrate excellent selectivity towards CO₂, with Faradaic efficiencies (FE) close to 100%, and possess low limiting potentials of −0.30 and −0.53 eV, showcasing their potential to be excellent catalysts. The introduction of pre-adsorbed hydrogen atoms further optimized the advantage of CO₂RR over HER, with the efficiencies of Ti-B@C₉N₄-H and Hf-B@C₉N₄-H methods increasing from 0% and 28% to over 99%, respectively, providing new insights into overcoming the low selectivity of CO₂ reduction.

1. Introduction

In recent years, with the rapid development of the global economy, the excessive use of fossil fuels has had adverse effects on people’s lives and development. Large-scale emissions of carbon dioxide exacerbate the greenhouse effect. Secondly, an energy crisis is gradually looming [1,2]. To address these challenges, the use of clean energy sources such as solar and wind power to convert into electricity and conduct CO₂ reduction reactions (CO₂RR) under mild conditions has become an increasingly mature method [3,4,5,6,7]. This approach involves the electrochemical reduction of excess CO₂ from the atmosphere to generate industrially valuable C1, C2, or high-carbon products, such as carbon monoxide, formic acid, methane, ethanol, acetic acid, etc. [8,9,10,11,12,13,14]. CO exhibits application value in chemical synthesis (as a reducing agent), materials science (as a refrigerant), and the fuel sector [15,16]. So far, suitable catalysts for electrocatalytic CO₂ reduction to CO are being explored, conducting extensive studies in both experimental and theoretical aspects. Ren et al. proposed that the Ni/Fe-NC dual-atom site catalyst (DAC) exhibited good CO₂RR performance in CO production and that the CO selectivity could remain above 90% over a wide potential range from −0.5 to −0.9 V [17]. Beyond that, Zhang et al. experimentally demonstrated that a supported Pd2 DAC was used for electrochemical CO₂RR. It exhibited superior CO₂RR catalytic performance with 98.2 % CO Faradic efficiency at −0.85 V vs. RHE [18]. As a product, it demonstrates a relatively high Faradaic efficiency.

With the successful isolation of monolayer graphene [19,20], the investigated graphene-based nano-meshes (GNMs) with nanopore arrays show superior electronic properties to graphene. For example, graphitic carbon nitride (g-C₃N₄) has been synthesized experimentally for decades and is still being investigated today. Zhu et al. investigated g-C₃N₄-based single-atom catalysts (SACs), Ti-, Ag-, and Cr-gC₃N₄ with low limiting potentials for producing different C1 products. In addition, Liu et al. reported that the nitrogen-coordinated graphene substrate composed of Cu and different TM double-doped atoms enhances the catalytic performance for producing C1 products [21].

The theoretically predicted synthesis of C₉N₄ has shown a two-dimensional honeycomb-kagome lattice near the Fermi level [22,23]. Its porous environment and nitrogen coordination facilitate the anchoring of TMs and prevent aggregation, making it widely applicable in catalysts and topological materials. SACs based on C₉N₄ have been studied [24]; however, there is still room for improvement in the overpotential and Faradaic efficiency in the electrochemical CO₂RR on SAC systems. Introducing a new type of metal atom into SACs to form bimetallic catalytic sites is an effective method to enhance the catalytic performance of CO₂RR. However, achieving synergistic interactions between different atoms within the same carrier remains a challenge [25,26,27]. In previous reports, B atoms have been proposed as active centers for CO₂ electrocatalytic reduction [28]. The sp³ orbitals of B share similar functionalities with the d orbitals of TMs, and B@g-C₉N₄ materials can effectively suppress the HER reaction [29]. Therefore, the use of B and TM atoms as dual-atom catalysts is a promising strategy. By introducing dual-active sites, better coordination of interactions between atoms can be achieved, regulating the adsorption of intermediates and controlling catalytic activity and selectivity [30,31,32,33]. This strategy holds promise for providing a new pathway for electrochemical CO₂ reduction, laying the foundation for achieving high catalytic performance.

In this study, C₉N₄ was chosen as the substrate, and B, along with 26 TM co-modified catalysts, was investigated. A series of B-TM catalysts were studied for their stability, catalytic activity, and selectivity in CO₂RR. Two catalysts (Sc-B@C₉N₄, Hf-B@C₉N₄) with high reaction activity and selectivity in reducing CO₂ to CO were identified after screening. Subsequently, modifications were made to systems prone to competitive hydrogen evolution reactions, yielding promising results that could provide new insights for improving the low selectivity of CO₂ reduction.

2. Results and Discussion

2.1. The Structure and Stability of TM-B@C₉N₄

The selection of a 2 × 2 × 1 supercell of C₉N₄ was initiated to provide adequate voids for accommodating TM and B atoms while minimizing the impact of other atoms on the calculations (Figure 1a).

The stability of the substrate material is a crucial factor to consider for excellent catalysts, as it directly influences the catalyst’s usage lifespan. To address this concern, the binding energies (E_b) and formation energies (E_f) of 26 TM atoms anchored in B@C₉N₄ pores were calculated using the formula (Figure 1b). Eb < 0 indicates stable binding of TM atoms to the substrate, while E_f < 0 suggests the feasibility of experimental synthesis. Except for Os atoms not meeting the criteria (E_f > 0), highly negative binding energies and relatively negative formation energies confirm the stability of the structure. The optimized substrate structure is shown in Figure S1. Molecular dynamics simulations (Figure S2), using Sc-B@C₉N₄ as an example, were employed to further evaluate the thermal stability of the studied catalysts. At a temperature of 300 K and a time step of 3 fs, no significant deformation or dissociation of the geometric structure was observed within 12 ps. Temperature and energy exhibited minor fluctuations within a small range, indicating the strong stability of this catalyst, warranting further investigation.

Based on the selection of substrate catalytic activity described in the article, TM-B@C₉N₄ (TM = Sc\Ti\V\Mn\Zr\Ag\Hf) was chosen as the object of analysis. We calculated the electron localization function (ELF) to investigate the bonding characteristics of the substrate (Figure 1c) [34,35,36]. The gradient-colored image represents different degrees of electron localization: red (value 1) signifies complete electron localization, blue (value 0) indicates complete electron delocalization, and green (value 0.5) corresponds to a state resembling a free electron gas. Significant electron localization was observed near the N atoms within the cavity, which aids in anchoring metal atoms. The bonding between B atoms and coordinating N atoms exhibited covalent characteristics, highlighting a strong electron-sharing interaction. The ELF of the other six substrates also exhibited similar electronic characteristics and bonding patterns (as shown in Figure S3), demonstrating the rationality and stability of the TM-B@C₉N₄ structure. Additionally, the ELF diagram indicates charge transfer between the TM and B atoms, which is further confirmed by the partial density of states (PDOS) graph in Figure S4. The graph shows coupling between the d orbitals of the TM atoms in the substrate and the p orbitals of the B atoms, suggesting an interaction between TM and B atoms that facilitates charge transfer and redistribution, which is beneficial for subsequent CO₂ adsorption and activation.

Additionally, the band structure of the TM-B@C₉N₄ substrate was studied (Figure 1d) and compared to the original C₉N₄ [23]; the conduction band bottom crosses the Fermi level, exhibiting metallic properties. This electronic characteristic is expected to facilitate electron transfer during the electroreduction process. The band structures of the remaining tested structures also exhibit similar results (Figure S5). In conclusion, C₉N₄ modified with TM-B atoms demonstrates favorable stability and unique electronic properties, holding the potential to become an excellent catalyst.

2.2. Adsorption and Activation of CO₂

The initial step in previous CO₂RR processes involves the adsorption and activation of CO₂, forming the foundation for subsequent CO₂ hydrogenation. Consequently, the structural optimization of CO₂ adsorption on different substrates was conducted, yielding various adsorption configurations (Figure S6). Taking Sc-B@C₉N₄ as an example (Figure 2a), C and O atoms form bonds with TM and B atoms, causing the linear CO₂ molecule to undergo a V-shaped distortion with an ∠OCO angle of 127.70°. The C-O bond lengths are elongated by 0.147 Å and 0.069 Å compared to free CO₂ gas molecules, indicating CO₂ bond activation. Similar V-shaped distortions were observed for CO₂ adsorbed on other substrates (Table S1), confirming CO₂ bond activation.

Furthermore, differential charge density and Bader charge calculations were employed to better analyze the electronic characteristics of CO₂ adsorption and how electrons are transferred from dual-active centers to activate CO₂. As shown in Figure 2a and Table S2, after CO₂ adsorption, Sc and B atoms lose 0.02 and 0.64, respectively. After CO₂ adsorption, Sc loses fewer electrons because, as a transition metal with low electronegativity and vacant d orbitals, it acts as a Lewis acid, attracting the lone pair electrons from oxygen and bonding with the oxygen atom to form a metal–oxygen bond. In contrast, B atoms donate more electrons due to their slightly lower electronegativity compared to C. As an electron acceptor, C interacts with the electrons donated by B, forming a stable B-C covalent bond. Therefore, in the interaction with CO₂, the TM and B atoms function as a Lewis acid-base pair, promoting electron transfer and molecular activation. The remaining C₉N₄ substrate acts as an electron reservoir, transferring electrons from the active centers to the CO₂ molecule, and it loses 0.27. Hence, CO₂ gains a total of 0.94, facilitating subsequent hydrogenation reactions. Similar changes are observed in the adsorption of CO₂ in the other six systems depicted in Figure S7.

We also examined the projected density of states (PDOSs) of the CO₂ adsorption intermediates (Figure 2b). From the PDOSs obtained after CO₂ adsorption on the substrate, it is evident that there exists a strong coupling between the d orbitals of Sc and the p orbitals of B with the p orbitals of CO₂. The simultaneous downward shift of bonding orbitals below the Fermi level indicates the cooperative stabilization of CO₂ by Sc and B atoms. Simultaneously, the partially occupied orbitals of Sc and B provide electrons to the 2p * orbitals of CO₂, causing some partially occupied 2p * orbitals to cross the Fermi level, indicating effective activation of C-O bonds for the subsequent hydrogenation. Similar changes were observed in other systems with CO₂ adsorption (Figure S8).

Finally, the projection of the crystal orbital Hamiltonian population (PCOHP) for the C-O bond in CO₂, both before and after adsorption, was calculated (Figure 2c). Of the two C-O bonds in CO₂, focus was placed on the C-O bond relevant to the subsequent hydrogenation step. The integration of bonding and antibonding states below the Fermi level yields the integral crystal orbital Hamiltonian population (ICOHP) values, indicating the strength of chemical bond formation. More negative ICOHP values indicate stronger bonding. The ICOHP value of the CO bond in the free CO₂ gas molecule is approximately −18.45 eV. After adsorption on the Sc-B@C₉N₄ substrate, the value increases to −15.80 eV, indicating that the CO bond is effectively weakened and activated. In Figures S8 and S9, the adsorption of CO₂ onto other TM-B@C₉N₄ (TM = Sc\Ti\V\Mn\Zr\Ag\Hf) substrates induces positive changes in ICOHP values, specifically, −15.94, −15.46, −15.11, −15.06, −13.4, and −16.27 eV, respectively, further confirming that TM and B-modified C₉N₄ effectively activate C-O bonds, laying the foundation for subsequent CO₂ reduction.

2.3. The Mechanism and Free Energy of CO₂ Reduction to CO

The proposed CO₂RR pathway is illustrated in Figure 3a. It is observed that CO₂ molecules, after adsorption, can undergo attack by a proton/electron pair to form *COOH species. Subsequently, the *COOH species gains a proton/electron pair to form *CO species, releasing a water molecule. Finally, the *CO species desorbs from the substrate. Following this pathway, Gibbs free energy calculations were performed for the selected catalysts, and the maximum free energy step differences during the reduction process were recorded (Figure 3b). A catalyst was considered potentially efficient if the absolute value of the limiting potential was less than 0.61 [37,38]. Accordingly, seven catalysts, Sc\Ti\V\Mn\Zr\Ag\Hf-B@C₉N₄, were screened. Additionally, we observed that systems from different periods exhibit similar trends, possibly related to the outermost electron count of different TM atoms.

To investigate the potential determining steps for CO₂ reduction to CO and differences in activity on various catalysts, Gibbs free energy diagrams were plotted for the seven mentioned catalysts, as shown in Figure 4. Among these, on Ti-B@C₉N₄, the energy required for the first step of CO₂ hydrogenation to form *COOH species is 0 eV, while the uphill energy needed for the second pair (H+/e-) attacking *COOH and dehydrating is 0.26 eV. Finally, the downhill energy for CO desorption is −0.05 eV. Therefore, the potential-determining step is the second hydrogenation step to produce *CO, with a corresponding limiting potential of −0.26 eV. Except for Mn-B@C₉N₄ and V-B@C₉N₄, where the potential-determining step is the first hydrogenation step producing *COOH, the remaining four catalysts involve the second hydrogenation step. The desorption energies of CO for Ti, V, Mn, and Ag-B@C₉N₄ catalysts are all lower, indicating spontaneous desorption. Sc, Zr, and Hf-B@C₉N₄ require only a small increase in energy, suggesting that they have potential as high-efficiency catalysts. Due to the relatively low free energy step of *+CO (0.21 V), the difference between the free energy steps of CO and +CO reflects the ease of CO desorption. For example, Wu et al. employed a graphene-supported nickel hydride cluster (7HNi₁₀-gra) as a catalyst to precisely tune the external potential, thereby regulating the concentrations of formate and methanol intermediates. This approach allowed them to rationally predict the formation of formic acid, methanol, and methane as the three main products. In the CO₂RR process reported in their work, the Gibbs free energy step for *CO is relatively low, which favors further hydrogenation to form other hydrocarbons. In contrast, on TM-B@C₉N₄ (TM = Sc\Ti\V\Mn\Zr\Ag\Hf), the *CO free energy step is significantly higher, promoting CO (g) desorption as the dominant pathway [39].

2.4. Competitive HER and Catalyst Improvements

Generally, catalysts undergo CO₂ reduction in aqueous solutions. To determine the efficiency of obtaining the desired products, it is necessary to consider the competition with hydrogen evolution reaction (HER), as both CO₂RR and HER significantly consume proton/electron pairs and share active sites. The occurrence of HER can lead to a decrease in the Faradaic efficiency of CO₂RR. Therefore, the Gibbs free energy of HER on TM-B@C₉N₄ (TM = Sc, Ti, V, Mn, Zr, Ag, Hf) was initially calculated (Figure 5a), and the corresponding limiting potentials are provided in Table S3. Additionally, the adsorption energies of CO₂ and H were considered, filtering out catalysts with excellent selectivity. As depicted in Figure 5b, this implies that the catalyst predominantly favors the reduction of CO₂ during hydrogenation [40]. The more positive this value, the higher the selectivity of CO₂ reduction over HER. Sc, Ag, and Hf-B@C₉N₄ all have values greater than zero. The competitively unfavorable process during CO₂RR is the hydrogen evolution reaction (HER), which has a significant negative influence on CO₂RR selectivity. As a result, the following assumptions are made to evaluate the selectivity of CO₂RR:

(1)

HER is the only side reaction in the CO₂RR process;

(2)

The rate of HER and CO₂RR do not depend mainly on proton and electron transfer;

(3)

Boltzmann distribution can be used to estimate the selectivity of CO₂RR relative to HER.

The theoretical Faradaic efficiency (FE) of CO₂RR can be calculated according to the Boltzmann distribution:

$$FE={\displaystyle \frac{1}{1+e^{-\frac{\Delta G}{k_{B}T}}}}\times 100\%$$

where ∆G is the Gibbs free energy difference between the HER and CO₂RR potential determination step, k_B represents the Boltzmann constant, and T = 298.15 K [41]. The efficiencies of the three catalysts are all above 99%. While Zr-B@C₉N₄ does not exhibit a clear preference for CO₂RR on CO₂ desorption, it still shows a 28% Faradaic efficiency (Table S3). The more negative CO₂ adsorption energy compared to H also indicates the substrate’s preference for CO₂RR. However, it is not satisfied on the substrate Ag-B@C₉N₄. Therefore, only Sc and Hf-B@C₉N₄ meet the requirements to suppress HER.

Moreover, during practical electrocatalytic processes, significant discrepancies may arise between the local pH near the catalyst surface and the bulk electrolyte environment due to proton consumption and limited diffusion. This local pH gradient can influence proton-coupled electron transfer (PCET) steps, thereby altering the thermodynamic and kinetic competition between the CO₂RR and the HER. Previous studies have demonstrated that under alkaline or high-pH conditions, the formation of *H intermediates is thermodynamically less favorable, which helps suppress HER. Simultaneously, key intermediates along the CO pathway, such as *COOH, become more stabilized, further enhancing the selectivity toward CO production [6]. Although the standard computational hydrogen electrode (CHE) model at pH = 0 was adopted in this study without explicitly considering local pH effects, the thermodynamic data and electronic structure analysis still indicate that the proposed heteronuclear dual sites exhibit favorable CO selectivity. Future investigations could incorporate explicit solvation models or potential-dependent simulations to gain deeper insights into the regulatory role of the local environment on catalytic mechanisms.

Figure 5. (a) Hydrogen evolution reaction (HER) free energy diagrams on these seven catalysts. (b) The difference in limit potential (U_L) between CO₂ reduction to CO and hydrogen evolution reaction to H₂ on TM-B@C₉N₄, as well as the changes in CO₂ adsorption energy and H adsorption energy (* represents substrate material).

Figure 5. (a) Hydrogen evolution reaction (HER) free energy diagrams on these seven catalysts. (b) The difference in limit potential (U_L) between CO₂ reduction to CO and hydrogen evolution reaction to H₂ on TM-B@C₉N₄, as well as the changes in CO₂ adsorption energy and H adsorption energy (* represents substrate material).

There are localized electrons around the N atoms in the substrate cavity, which may adsorb H atoms. Ti, V, and Zr-B@C₉N₄ also have the potential to act as highly selective catalysts. Therefore, the effect on selectivity and activity was explored by pre-adsorbing H atoms onto N atoms not bonded to TM and B in the above materials (Figure 6a). The adsorption energy of CO₂, the limiting potential for CO₂RR, and the Gibbs free energy for HER on Ti, V, and Zr-B@C₉N₄-H were calculated (as shown in Table S4). It was found that, after pre-adsorbing H, the CO₂ adsorption energy remained almost unchanged, while the limiting potential for CO₂RR showed varying degrees of change. Except for V-B@C₉N₄, which exhibited a small change, the absolute values of the limiting potentials for Ti-B@C₉N₄-H and Zr-B@C₉N₄-H decreased to 0.19 eV and 0.41 eV, respectively. The corresponding Gibbs free energy diagram is shown in S10. Furthermore, the pre-adsorption of H on Ti/V-B@C₉N₄ led to a lower limiting potential for HER, hindering its progression. The corresponding H adsorption energy decreased; however, it remained higher than the CO₂ adsorption energy. HER on Zr-B@C₉N₄-H was almost unaffected. It is noteworthy that, during the structural optimization process, there was no displacement observed in the H atom, indicating its stability. Through Figure 5b, it can be visually confirmed that both Ti and Zr-B@C₉N₄-H satisfy and indicate the predominance and priority of CO₂RR.

Bader charge calculations for the substrate with added H to the cavity were performed and compared with the base substrate without added H (

Table 1. The Bader effective charges of TM, B, and H atoms on the surface of TM-B@C₉N₄(-H), where negative numbers represent electron loss and positive signs represent electron acquisition.

System		Ti-B@C9N4(-H)	V-B@C9N4(-H)	Zr-B@C9N4(-H)
TM-B@C9N4	TM	−1.544	−1.329	−1.982
	B	−0.869	−0.938	−0.786
	H	\	\	\
TM-B@C9N4-H	TM	−1.456	−1.226	−1.794
	B	−0.811	−0.776	−0.771
	H	−0.417	−0.431	−0.408

). It was found that in the system with pre-adsorbed H, both TM and B active centers lose fewer electrons when binding to the substrate because H atoms share some electrons with the substrate. This suggests that more electrons may be involved in CO₂ adsorption and activation in subsequent steps, lowering the barrier for CO₂RR. This approach could provide a new perspective for improving the low selectivity of CO₂.

3. Computational Methodology

All computations were carried out utilizing the spin-polarized density functional theory (DFT) methodology implemented in the Vienna ab initio simulation package (VASP 5.4.4) [42]. The electron–ion interactions were modeled using the projector-augmented wave (PAW) approach, while the electronic exchange and correlation effects were described using the Perdew–Burke–Ernzerhof (PBE) functional within the framework of the generalized gradient approximation (GGA) [43,44]. To account for the van der Waals (vdW) interactions between reactants, intermediates, and catalysts, the empirical correction scheme based on the Grimme method (DFT+D3) was employed [45]. A plane wave energy cutoff of 400 eV was employed for structural optimizations, while a value of 500 eV was chosen for electronic structure calculations. The k-point grids were set to 1 × 1 × 1 for geometry optimization and 6 × 6 × 1 for electronic structure calculations. Convergence criteria for electronic energy and forces were established at 10–5 eV and 0.05 eV/Å, respectively. Additionally, an 18 Å vacuum layer was introduced to mitigate interlayer interactions. Catalyst thermal stability was assessed through ab initio molecular dynamics (AIMD) simulations using VASP, with a Nosé–Hoover thermostat employed. A time step of 3 fs and a total simulation duration of 12 ps were selected to verify catalyst thermal stability. To ascertain the Gibbs free energy during proton transfer in CO₂RR, computations were conducted utilizing a computational hydrogen electrode (CHE) model.

Gibbs free energy changes (ΔG) for all intermediate steps were determined using the following equation: $\Delta G=\Delta E+\Delta ZPE-T\Delta S+\Delta G_{pH}$, where ΔE corresponds to the total energy of surface and intermediate species obtained from DFT calculations, ΔZPE represents the variation in zero-point energy, and ΔS accounts for the entropy change of free molecules and adsorbates at 298.15 K. The pH-dependent ΔG values were calculated using the formula: $\Delta G_{pH}=k_{B}T\times ln10\times pH$, where k_B is the Boltzmann constant, and pH was set to zero.

The calculation of U_L is determined by the following formula:$U_L=-\Delta G_{max}/e$, here, ΔG_max is the maximum Gibbs free energy change in the reaction. The negative sign indicates the reduction potential, and a smaller |U_L| signifies higher catalytic activity.

In order to evaluate the stability of the new structure after doping of transition metal (TM), the binding energy (E_b) of the doped metal atoms was calculated using the following equation: $E_b=E_{\mathrm{TM}-B@{\mathrm{C}}_9{N}_4}-E_{{\mathrm{C}}_9{N}_4}-E_{\mathrm{TM}-\mathrm{single}}-E_B$ where $E_{\mathrm{TM}-B@{\mathrm{C}}_9{N}_4}$ represents the total energy associated with the TM and B atom modifications on C₉N₄, denotes the energy of the pristine C₉N₄ substrate structure, and E_TM-single and E_B, respectively, denote the energies associated with the transition metal (TM) atom and the non-metallic boron (B) atom.

Adsorption energy refers to the energy released or consumed when CO₂ adsorbs onto the catalyst surface. It is calculated using the following formula:$E_{ads}=E_{total}-(E_{catalyst}+E_{adsorbate}$), where E_ads is the total energy of the adsorbate on the catalyst surface, where E_catalyst is the total energy of the catalyst itself, and E_adsorbate is the total energy of the adsorbate.

4. Conclusions

This study employed C₉N₄ as the substrate, utilizing 26 transition metal (TM) atoms along with boron (B) atoms as dual active sites for CO₂ reduction reactions (CO₂RR). The calculations of binding energy (E_b), formation energy (E_f), and ab initio molecular dynamics (AIMD) revealed that the majority of TM-B@C₉N₄ systems exhibit excellent thermal stability. The heteronuclear dual active sites synergistically enhance CO₂ adsorption and activation, as elucidated through an in-depth analysis of charge density differences, electron localization function (ELF), local density of states (LDOS), and integrated crystal orbital Hamilton population (ICOHP) for selected systems. Gibbs free energy calculations identified six catalysts, namely, Sc-B@C₉N₄, Hf-B@C₉N₄, Ti-B@C₉N₄, V-B@C₉N₄, Mn-B@C₉N₄, and Zr-B@C₉N₄ Ag-B@C₉N₄, exhibiting low overpotentials for CO generation (U_L < 0.61). Considering the suppression of competitive hydrogen evolution reactions (HERs), only Sc-B@C₉N₄ and Hf-B@C₉N₄ displayed superior selectivity, achieving nearly 100% Faradaic efficiency. Modification of less selective materials resulted in improved selectivity for Ti-B@C₉N₄-H and Zr-B@C₉N₄-H upon hydrogenation, accompanied by minor changes in overpotential. Insight into this phenomenon was gained through Bader charge analysis. This theoretical investigation provides new strategies for the design and synthesis of high-performance catalysts while offering novel insights into the mechanism of CO₂ activation. It is anticipated that this research will serve as a valuable reference for the further development of novel CO₂RR electrocatalysts.