Theoretical investigations of transition metal atom-doped MoSi₂N₄ monolayers as catalysts for electrochemical CO₂ reduction reactions

Abstract

Following the principle of single-atom catalysts (SACs), the fourth-period transition metals (TM) were designed as active sites on a MoSi₂N₄ monolayer surface with N vacancy, and the catalytic mechanisms of these single-atom active sites for the conversion of CO₂ to CO were investigated by first-principles calculations. Our results showed that the doped TM atoms on the MoSi₂N₄ surface significantly enhanced the CO₂ reduction reaction (CO₂RR) activity compared with the pristine MoSi₂N₄ monolayer. Our findings after analyzing all the doped structures in our work were as follows: (1) the Sc-, Ti-, and Mn-doped structures exhibited very low limiting potentials; (2) out of Sc-, Ti- and Mn-doped structures, the Mn@MoSi₂N₄-N_v structure showed the best catalytic performance with a limiting potential of only –0.16 V, exhibiting an advantage over the hydrogen evolution reaction, which is a competitive reaction of CO₂RR. However, the positive binding free energy at 298.15 K of the intermediate reactant *COOH on the Mn@MoSi₂N₄-N_v surface indicated its unstable state, which hinders the CO generation process. This is contrary to the results derived from the adsorption/binding energy at 0 K, which indicated that the effect of temperature cannot be ignored when considering adsorption/binding energy. Our work provides insights into the effects of temperature on the catalytic mechanisms for CO₂RR through TM-doped MoSi₂N₄ monolayers.

---

1. Introduction

The increasing carbon dioxide (CO₂) emissions have led to various environmental problems. Thus, there is an urgent need to decrease CO₂ emissions and develop techniques for CO₂ capture and utilization. The electrochemical carbon dioxide reduction reaction (CO₂RR) is one of the promising solutions that can convert CO₂ into valuable chemical fuels through multiple protonation steps.^1–3 However, there are several challenges to overcome as follows: first, the large C=O bond energy in CO₂ molecules indicates that breaking this bond requires a significant amount of energy, which results in a slow reaction rate and Faraday inefficiency of CO₂RR;^4–6 second, CO₂RR can result in multiple reaction products, which makes the reaction path very complex and yields various final products; third, the presence of side reactions, such as hydrogen evolution reaction (HER), further results in low product selectivity.^7–10 So, developing high-performance catalysts for electrochemical CO₂RR is crucial, which explains the volume of work on this topic. For example, noble metals, such as Au and Ag, have been designed as effective catalysts for selectively reducing CO₂ to CO, but their high cost limits their wide applications.^11,12 Therefore, developing noble metal-free (low-cost), efficient, and stable electrocatalysts for CO₂RR is one of the current research focus. New strategies, such as single-atom catalysts (SACs), have also been adopted to design new catalysts. The SACs exhibit excellent properties, including high atomic utilization efficiency, product selectivity, and uniform distribution of catalytic active sites.^13–17 The activity of SACs is closely related to the coordination environment of the central atom.^18–20 In 2011, Zhang et al.²¹ demonstrated the excellent catalytic performance of Pt₁/FeO_x single-atom catalysts for CO production. Two-dimensional (2D) materials have been considered promising SAC substrates, which can load highly active transition metal atoms (TMs).²² The unique physical, chemical, and electronic properties make them suitable for the development of novel CO₂RR catalysts that can replace noble metal catalysts.^23–26 The recently synthesized MSi₂N₄ (M = Mo, W),²⁷ which have been confirmed as novel 2D semiconductor materials without knowing their 3D parents, provide us new candidates for catalyst designs. Different modifications have been adopted to tune the active sites on the surface of 2D MSi₂N₄ for high electrochemical catalytic performance.^28–35 In the line of intrinsic defect engineering, Luo et al.³⁰ found that the pre-formed N vacancies on the MSi₂N₄ surface can expose the active Si atoms on the surface, thus enhancing the N₂ adsorption capacity, which can effectively promote the nitrogen reduction reaction (NRR). TiSi₂N₄ and TaSi₂N₄ were identified as high-potential and promising NRR catalysts, with calculated limiting potentials of only 0.41 and 0.46 V. Extraneous TM atom doping is another important method to design catalysts. For example, the TM atoms doped on the surface of MoSi₂N₄ show excellent activity in CO₂RR.³¹ In the case of the Co atom-doped MoSi₂N₄, HCOOH is the preferred product for CO₂ reduction with a rate-determining step (RDS) height of 0.89 eV; on the other hand, the surface modification using Sc, Ti, Fe, and Ni atoms inclined the catalysts towards reducing CO₂ to CH₄, with an RDS barrier of 0.81–1.24 eV. The Cu–MoSi₂N₄ catalyst, where a Cu atom substitutes one surface Si atom, shows excellent reaction performance for C₂ products.³² The products of the Cu–MoSi₂N₄ catalytic reaction are valuable C₂₊ chemicals, which were further identified as CH₄, C₂H₄, C₂H₆, C₂H₅OH, and C₃H₇OH. Similar TM-doped catalysts were screened theoretically to realize efficient nitric oxide electroreduction reaction (NOER), and Zr- and Pt-doped MSi₂N₄ monolayers exhibited the best catalytic activities and selectivity for generating NH₃ at limiting potentials of 0 V and 0.10 V.³³ Special TM–N₄ structures were constructed by embedding a series of TM atoms into defective MoSi₂N₄ monolayers, which demonstrates the oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) activities.³⁴ Through high-throughput computational screening of 3d, 4d, and 5d TMs doped into the M-, Si-, and N-sites of MSi₂N₄ monolayers, Zang et al.³⁵ discovered that the doping of V, Fe, Nb, Tc, and Ta atoms to MSi₂N₄ significantly enhances its HER activities. Finally, 10 out of 124 candidates have been successfully identified as excellent HER catalysts. The outstanding performance of defective or doped MoSi₂N₄ as electrocatalysts in various important reactions raises another interesting question: are there design strategies based on doped MSi₂N₄ monolayers to explore good catalysts for the conversion of CO₂ into CO with high efficiency and high selectivity? There have been only very few studies about this so far. In previous work, single TM-anchored MoSi₂N₄ monolayers with no defects on the surface were explored as CO₂RR catalysts.³¹ Then, in the Cu–MoSi₂N₄ catalyst, Si vacancy was introduced, so the TM atoms were coordinated with N atoms on the surface.³² In the present work, we will explore the synergistic effect of N vacancy and TM atoms, where the TM atoms are coordinated with Si atoms on the surface. We take SACs as a novel design strategy and theoretically screen 10 TM (Sc–Zn)-doped MoSi₂N₄ monolayers as electrocatalysts for CO₂RR. The TM atoms work as active sites and break the C=O bond. Based on the complexity of the product selectivity and reaction paths, CO is selected as the target product because CO has a very high industrial value as a feedstock for various chemical products and industrial reactions.^36–38 Catalyzing CO₂ to CO is considered a more economically viable pathway due to its high selectivity and relatively low energy consumption among multiple product conversions.^16,18,19 Currently, the available CO electrocatalysts suffer from issues of high overpotentials, low current efficiency, low product selectivity, slow reaction rates (compared to HER), poor stability, and high application costs.^39,40 This puts forward higher requirements of selectivity for the designed catalysts. Another issue related to the catalytic reaction rate is the presence of side reactions, such as hydrogen evolution reaction (HER), so the competition between CO₂RR and HER also will be discussed. In all, we conduct a theoretical investigation into the CO₂RR performance of 10 TM (Sc–Zn)-doped MoSi₂N₄ monolayers (denoted by TM@MoSi₂N₄-N_v) during CO generation using the density functional theory (DFT) method. We hope our work will be helpful in finding CO₂RR electrocatalysts of high performance based on our obtained insights into the catalytic process of TM@MoSi₂N₄-N_v catalysts and the related catalytic mechanism.

2. Computational methods

All spin polarization calculations were performed using the DS-PAW software.⁴¹ The Perdew–Burke–Ernzerhof (PBE) functional based on the generalized gradient approximation (GGA) and PAW pseudopotential were used to deal with the exchange–correlation interaction for the ion-electron interaction.^42,43 The initial spin configurations were taken as the ferromagnetic state. The Grimme's semiempirical scheme (DFT-D3) was incorporated to address the van der Waals (vdW) interactions.⁴⁴ The k-mesh was set to be 3 × 3 × 1 for structural optimization and 5 × 5 × 1 for accurate electronic density of state (DOS) calculations. In all the calculations, the thickness of the vacuum layer was set to be 30 Å along the z direction to eliminate the interaction between adjacent periodic supercells. The energy convergence criteria were set to 10⁻⁵ eV for geometry optimization and 10⁻⁶ eV for electronic DOS calculations and the criteria of force convergence was set to 0.02 eV Å⁻¹. The cut-off energy was set to 500 eV. The local three-dimensional electron correlation of the fourth-period TMs can be described by considering the on-site Coulomb repulsion (U) and exchange interactions (J) in the DFT+U method.⁴⁵ The optimized U–J values for different transition atoms are listed in Table S1 in the ESI.†

On the surface of a 3 × 3 × 1 supercell of a 2D MoSi₂N₄ monolayer, one N atom is substituted by a TM atom (TM@MoSi₂N₄-N_v). The binding energy (E_bind) of the TM atom on the MoSi₂N₄-N_v monolayer indicates the thermal stability of TM@MoSi₂N₄-N_v monolayers and is calculated by the following formula:

E_bind = E(TM@MoSi₂N₄-N_v) − E(MoSi₂N₄-N_v) − E(TM) (1)

E(TM@MoSi₂N₄-N_v), E(MoSi₂N₄-N_v), and E(TM) represent the DFT total energies of TM@MoSi₂N₄-N_v, MoSi₂N₄ monolayer with N-vacancy, and an isolated TM atom, respectively. The E_f was calculated to evaluate the difficulty of doping TM in the MoSi₂N₄ monolayer and is defined as:

E_f = E(TM@MoSi₂N₄-N_v) − E(MoSi₂N₄-N_v) − μ(TM) (2)

μ(TM) is calculated as E_tot/N, where E_tot and N are the DFT total energies of the TM atom in the bulk phase and the number of atoms in it, respectively. Similar to E_bind, the CO₂ adsorption energy (E_ads) was calculated by:

E_ads = E(CO₂_TM@MoSi₂N₄-N_v) − E(TM@MoSi₂N₄-N_v) − E(CO₂) (3)

E(CO₂_TM@MoSi₂N₄-N_v), E(TM@MoSi₂N₄-N_v), and E(CO₂) represent the DFT total energies of CO₂ adsorbed by the TM@MoSi₂N₄-N_v monolayer, TM@MoSi₂N₄-N_v substrate, and CO₂ molecule, respectively. The binding energy (E_b) for all intermediates on the TM@MoSi₂N₄-N_v surface was calculated by:³⁶E(*COOH) and E(*CO) are the DFT total energies of TM@MoSi₂N₄-N_v adsorbed with the *COOH intermediate and *CO intermediate, and E(*) is the energy of the clean TM@MoSi₂N₄-N_v surface. E(CO₂), E(H₂O) and E(H₂) are the DFT total energies of the gas-phase CO₂, H₂O, and H₂, respectively.

The computational hydrogen electrode (CHE) model was employed to determine the reaction energy as a function of the potential with a proton–electron pair in the reactant.^46–48 Thus, the chemical potential of the proton-electron pair (H⁺ + e⁻) in an aqueous solution is equal to that of half of an isolated H₂ molecule. According to the CHE model, the Gibbs free energy change (ΔG) for all the electrochemical steps is calculated by:

ΔG = ΔE = ΔZPE − TΔS + ΔG_U + ΔG_pH (6)

ΔE refers to the reaction energy change of each hydrogenation step calculated by DFT. ΔZPE and ΔS are the changes in zero-point energy and entropy, respectively. ΔZPE can be calculated by summing the vibrational frequencies (ΔZPE = 1/2Σhv), the details of which are given in Section 1 in the ESI.† In the phonon frequency calculation, the finite displacement method is used with a set of supercells containing small displacements. The first principle code, DS-PAW, is used to calculate the total energies of these supercells. Then, the force constant and dynamical matrix can be derived from the displacement–total energy results, and the phonon modes and frequencies will be the eigenvectors and eigenvalues of the dynamical matrix in the harmonic approximation. The theory and calculation details and examples of input files for DS-PAW in phonon frequency calculations are given in Section 2 in the ESI.† The results of ZPE and TS corrections for *COOH and *CO are listed in Tables S2 and S3 (ESI†), respectively. When performing vibrational frequency calculations for adsorption intermediates, only the adsorbate in the catalyst was allowed to move while the other atoms were fixed. For free molecules, the vibrational frequencies were calculated directly without fixing the atoms. The temperature T is set to 298.15 K.

ΔG_U = −neU (7)

n is the number of transferred electrons, and U is the electrode potential relative to the reversible hydrogen electrode (RHE). ΔG_pH is the free energy correction of pH, which can be expressed as:

ΔG_pH = k_BT × ln 10 × pH (8)

Here, k_B is the Boltzmann constant, and the pH is set to zero for simplicity in this work. If the electrolyte pH is known, the RHE can be related to the standard hydrogen electrode (SHE) scale by the Nernst equation: U_SHE = U_RHE + 0.059·pH. The best absolute value for the standard hydrogen electrode in water is 4.28 V. If the absolute value of SHE (4.28 V) is included explicitly, all calculated Gibbs free energies will have a constant shift. The catalytic activity for the CO₂RR can be described by the limiting potential (U_L), which is defined as:

U_L = −ΔG_max/e (9)

ΔG_max is the ΔG of the elementary step with the maximum ΔG.

The above computational methods are widely adopted to explore the underlying catalytic mechanisms, which have been successfully applied in different kinds of catalytic reactions. It provides us insights into the possible reaction paths at the microscopic level, where the Gibbs free energy change for all electrochemical steps are compared. However, there are still some drawbacks in binding energy calculations of molecular adsorbates. The binding energy calculated by eqn (4) and (5) and the CO₂ adsorption energy calculated by eqn (3) are derived from the DFT total energies, and all of these energies are the values corresponding to the ground states at 0 K. These physical quantities cannot describe the chemical reactions taking place at room temperature (298.15 K) because the effects of room temperature were excluded in DFT calculations. To take into account the temperature effects, the DFT total energy E should be replaced by the Gibbs free energy G, where the zero-point energy and temperature-related entropy corrections are included. Now, we can get different insights from the binding energy at 0 K and binding free energy at 298.15 K. For example, the CO₂ adsorption free energy (G_ads) is calculated by:

G_ads = G(CO₂_TM@MoSi₂N₄-N_v) − G(TM@MoSi₂N₄-N_v) − G(CO₂) (10)

G(CO₂_TM@MoSi₂N₄-N_v), G(TM@MoSi₂N₄-N_v), and G(CO₂) represent the Gibbs free energies for CO₂ absorption by the TM@MoSi₂N₄-N_v monolayer, TM@MoSi₂N₄-N_v substrate, and CO₂ molecule, respectively. The Gibbs free energies were calculated using eqn (6).

3. Results and discussions

3.1 Structure and stability of TM@MoSi₂N₄-N_v monolayers

As shown in Fig. 1(a) and (b), 10 TM atoms (Sc–Zn) are considered to substitute the N atoms on the surface of the pristine MoSi₂N₄ monolayer. The percentage of TM doping can be calculated by N_TM/N_tot × 100%, where N_TM and N_tot are the number of doped atoms and the total number of atoms in the doped substrate, respectively. The 3 × 3 × 1 MoSi₂N₄ supercell comprises a total of 63 atoms. As a TM atom replaces an N atom, the total number of atoms remains unchanged. Therefore, the percentage of TM doping should be 1/63, approximately 1.6%. Due to the larger radius of the TM atoms compared to that of the N atoms, the anchoring space generated by the N vacancy is insufficient to accommodate the TM atoms. Consequently, the TM atoms are extruded from the pristine N layer, forming a cone-shaped structure. As a result, the bond lengths of all three TM–Si bonds in the TM@MoSi₂N₄-N_v (TM = Sc–Zn) monolayers exceed those of the N–Si bonds in the pristine MoSi₂N₄ monolayer (1.74 Å). In comparison to other monolayers, the Zn atoms in the Zn@MoSi₂N₄-N_v monolayer are far away from the surface of the MoSi₂N₄-N_v monolayer. Out of the ten TM atoms, only the bond lengths of the Zn–Si bonds are significantly larger than those of the TM–Si bonds in the other monolayers. As shown in Table S4 in ESI,† the Zn–Si bond length is about 3.2 Å, which is longer than that of the Zn–Si bond in silicene/(0001) ZnS interfaces. This is also similar to the result obtained when Zn atoms are directly adsorbed onto the pristine MoSi₂N₄ monolayer.⁴⁹ The optimized stable configurations and structural parameters of all TM@MoSi₂N₄-N_v monolayers are shown in Fig. S1 and Table S4 in the ESI,† respectively.

Fig. 1 (a) Top and (b) side views of a 2D TM@MoSi₂N₄-N_v monolayer. The selected location of the doped TM atoms and the selected TM atoms are labeled. Calculated (c) binding energy (E_b) and (d) formation energy (E_f) of the TM@MoSi₂N₄-N_v monolayers. (e) d-band centers of TM atoms (TM = Sc–Zn).

The thermal stability of TM@MoSi₂N₄-N_v (TM = Sc–Zn) monolayers is a crucial parameter for evaluating the catalyst performance. As shown in Fig. 1(c), the binding energies range from −4.42 eV to −0.44 eV. The Ni@MoSi₂N₄-N_v monolayer has the highest binding energy of −4.42 eV, while the Zn@MoSi₂N₄-N_v monolayer has the lowest binding energy of −0.44 eV. In general, catalysts with larger negative binding energies exhibit better thermal stability. Comparing the binding energy of the optimized structures, the Zn@MoSi₂N₄-N_v monolayer is unstable. As a result, it will not be taken into account in the subsequent CO₂RR performance analysis. Fig. 1(d) shows that only the Mn@MoSi₂N₄-N_v monolayer has a negative formation energy of −0.38 eV. However, a positive formation energy does not necessarily rule out the possibility of its experimental synthesis. For example, Co–MoS₂ has a calculated formation energy of 1.80 eV, but it can be synthesized under mild hydrothermal conditions.⁵⁰ The formation energies of the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers are all lower than this value, indicating that they can be synthesized without harsh experimental conditions. Secondly, the orbital overlaps in the partial density of states (PDOS) can further confirm the difference in binding energy for these 10 TM atoms on the surface of MoSi₂N₄-N_v monolayers. The 3d orbital of TM atoms and the 2p orbital of Si atoms overlap near the Fermi level in all monolayers except for the Zn@MoSi₂N₄-N_v monolayer. The transfer of part 3d orbital electrons from the TM atoms to the pristine MoSi₂N₄ surface during doping results in the formation of stable TM–Si bonds. At the same time, the resulting TM-3d empty orbitals provide channels for electron acceptors and donors in the catalytic process. The more the atomic orbitals overlap, the stronger the orbital hybridization. The stronger orbital hybridization will induce more negative binding energy and more stable structures. The Ti, V, Cr, Mn, Fe, Ni and Cu-doped systems possess magnetic ground states, indicated by the asymmetric spin-up and spin-down states in PDOS calculations. Also, the magnetic moments for single TM atoms and total magnetic moments for TM @ MoSi₂N₄-N_v are given in Table S9 in the ESI.† The unbalanced magnetic moment induced by the N vacancy in MoSi₂N₄-N_v will interact with the magnetic moment caused by unpaired electrons in TM atoms. The PDOS of TM-3d and Si-2p for TM@MoSi₂N₄-N_v (TM = Sc–Zn) monolayers are given in Fig. S2 in the ESI.† Thirdly, the chemical bond length of the Si–TM bond will be another criterion. As given in Table S4 in the ESI,† the TM–Si bond lengths in their compounds are bigger or close to the values calculated in our present work. This means that the chemical bonds have been created between TM atoms and Si atoms nearby, which is helpful for stability.

3.2 Initial adsorption of CO₂ molecule

The adsorption capacity of TM atoms can be described by the d-band center. Based on the classic d-band center theory,⁵¹ a d-band center value closer to the Fermi level corresponds to a stronger adsorption capacity. Fig. 1(e) shows the d-band center values of TM atoms in the TM@MoSi₂N₄-N_v (TM = Sc–Zu) monolayers. The d-band center values of the Sc, Ti, Mn, Fe@MoSi₂N₄-N_v monolayers are closer to the Fermi energy level (blue dashed box). Therefore, we predict that their adsorption capacity will be strong. The first and most critical step in CO₂RR is the initial adsorption of CO₂ molecules, and the adsorption configuration of CO₂ can significantly influence the subsequent reduction processes. The present study considers two adsorption configurations: vertical adsorption of CO₂ (denoted by *CO₂_1) and horizontal adsorption of CO₂ (denoted by *CO₂_2). The optimized configurations for *CO₂_1 and *CO₂_2 are shown in Fig. S3 and S4 in the ESI,† respectively. The results indicate that, except for the Mn@MoSi₂N₄-N_v monolayer, CO₂ maintained a linear shape in the optimized *CO₂_1 adsorption configuration for all monolayers. However, in the *CO₂_2 adsorption configurations of TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers, the configurations of CO₂ changed significantly, and the originally linear CO₂ became V-shaped (∠O–C–O < 180°). This suggests that the CO₂ molecules were activated. Table S5 in the ESI† presents the structural parameters of the *CO₂_2 configuration of the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers.

In the case of the pristine MoSi₂N₄ monolayer, the CO₂ molecules in two optimized CO₂ adsorption configurations always exhibit a linear structure. This suggests that activating CO₂ on its surface is a challenging process. The *CO₂_1 and *CO₂_2 optimized configurations of TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers and pristine MoSi₂N₄ monolayer are given as Fig. S3 and S4 in ESI,† respectively. Firstly, the surface substitution defect, TM on N vacancy, enhances the adsorption of CO₂ on the surface. Table 1 compares the CO₂ adsorption energies/adsorption free energies of the *CO₂_1 and *CO₂_2 configurations, further illustrating the enhanced CO₂ adsorption capability achieved through surface doping with TM atoms. The CO₂ adsorption energies of the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers (−1.32 to −0.16 eV) are all lower than those of the pristine MoSi₂N₄ monolayer (0.13 to 0.20 eV). In view of the adsorption free energy, the same trend is observed, which indicates that TM doping enhances the adsorption of molecular CO₂. Secondly, the two configurations, *CO₂_1 and *CO₂_2, show different adsorption abilities for CO₂.

Table 1 Adsorption energies (E_ads/eV) at 0 K and adsorption free energies (G_ads/eV) at 298.15 K of *CO₂_1 and *CO₂_2 on TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers and pristine MoSi₂N₄ monolayer

Transition metals	*CO2_1		*CO2_2
	E ads (0 K)	G ads (298.15 K)	E ads (0 K)	G ads (298.15 K)
MoSi2N4	0.13	0.55	0.20	0.64
Sc	−0.18	0.27	−1.32	−0.81
Ti	−0.21	0.12	−0.78	−0.32
V	−0.47	−0.02	−0.76	−0.25
Cr	−0.22	0.32	−0.28	0.16
Mn	−0.47	−0.05	−0.46	0.02
Fe	−0.27	0.14	−0.71	−0.20
Co	−0.23	0.21	−0.31	0.15
Ni	−0.18	0.02	−0.45	0.00
Cu	−0.16	0.37	−0.45	0.08

---

In the *CO₂_1 optimized configuration, when the CO₂ molecule is adsorbed on the surface, one terminal O atom of CO₂ binds with the TM atom on the catalytic surface. In the *CO₂_2 optimized configurations, both the O and C atoms can bind with the TM atom on the catalytic surface. The *CO₂_2 optimized configurations should be more stable than *CO₂_1 optimized configurations because there are two possible chemical bonds (O–TM and C–TM) between CO₂ and TM atom in the *CO₂_2 optimized configurations. Our calculations demonstrated in Table 1 confirm this surmise. Thirdly, the configuration of the molecular CO₂ depends on the TM atoms. Except for the Mn atom, the *CO₂_2 configuration has lower adsorption energy (−1.32 to −0.31 eV) than those of the *CO₂_1 configurations (−0.47 to −0.16 eV) for TM doping. The adsorption energy in the case of the Mn atom is almost the same for the *CO₂_1 configuration and the *CO₂_2 configurations. This is the first reason why the *CO₂_2 configurations will be used in the subsequent CO₂RR process.

Considering the temperature effects into account, values of the adsorption free energies at 298.15 K for the TM doping are positive. Usually, it is assumed that a positive adsorption free energy indicates unstable adsorption; only V and Mn atoms have negative adsorption free energies for the *CO₂_1 configuration, and only Sc, Ti, V and Fe atoms have negative adsorption free energies for the *CO₂_2 configurations. These results are very different from those inferred from the adsorption energy at 0 K. The *CO₂_2 configuration has a lower adsorption energy than that of the *CO₂_1 configuration, except for Mn doping. Finally, based on the adsorption energies at 0 K and adsorption free energies at 298.15 K, only Sc, Ti, V, Mn and Fe-doped MoSi₂N₄-N_v are suitable as catalysts for CO₂RR. From the relation between free energy and the equilibrium constant, ΔG₀ = −RT ln(K_p), the equilibrium constant K_p is calculated from the Gibbs free energies at 298.15 K listed in Table 1. As shown, only the equilibrium constants of Sc, Ti, V and Fe in the *CO₂_2 configurations are larger than 10³, and the values are 4.9 × 10¹³, 2.6 × 10⁵, 1.7 × 10⁴, and 2.4 × 10³ for Sc, Ti, V and Fe, respectively. The equilibrium constants of V and Mn in the *CO₂_1 configurations are 2.2 and 7.0, respectively. Other equilibrium constant values are not bigger than 1, so they can be omitted. The calculations of equilibrium constants further confirm the conclusion inferred from the adsorption free energy.

Besides the adsorption free energies at room temperature, the charge transfer between CO₂ and the catalytic surface can supply another insight. As shown in Fig. 2, the charge transfer occurs between the CO₂ molecules and the TM atoms in the *CO₂_2 configurations, donating electrons to the CO₂via the C atom and significantly activating the C=O bond and the CO₂ molecules. The Bader charge (Table S5 in the ESI†) reveals that the charges obtained by CO₂ decrease with increasing atomic number, consistent with the change in adsorption energy. Furthermore, the adsorption capacities of the Sc, Ti, Mn, Fe@MoSi₂N₄-N_v monolayers align with the predicted results of the d-band center, where the bond angles of CO₂ after adsorption change from 180° to 132.1°–143.5°, adsorption energies of CO₂ range from −1.32 to −0.46 eV, and CO₂ molecules gain charges of 0.39–0.88 e. The *CO₂_2 configurations will be used for further studies in the subsequent CO₂RR process.

Fig. 2 Differential charge density diagrams of CO₂ adsorbed on (a) Sc@MoSi₂N₄-N_v, (b) Ti@MoSi₂N₄-N_v, (c) V@MoSi₂N₄-N_v, (d) Cr@MoSi₂N₄-N_v, (e) Mn@MoSi₂N₄-N_v, (f) Fe@MoSi₂N₄-N_v, (g) Co@MoSi₂N₄-N_v, (h) Ni@MoSi₂N₄-N_v, (i) Cu@MoSi₂N₄-N_v monolayers, respectively. The isosurfaces are 0.005 e Å⁻³. The charge accumulation and depletion are marked by the yellow and cyan regions, respectively.

3.3 Reaction pathway analysis of CO₂RR

Previous studies have demonstrated that CO₂RR can produce a range of products when CO₂ molecules are adsorbed on the catalyst surface. These products include CO and HCOOH (two-electron products), CH₃OH (six-electron product), CH₄ (eight-electron product), and C₂₊ (generated by C–C coupling).^52,53 The present work excludes considering the C₂₊ products due to the difficulty of achieving C–C coupling with only one active site in SAC.^17,54Fig. 3(a) shows that CO and HCOOH are the two-electron products generated during CO₂RR, via the CO₂ → *COOH → *CO → CO and CO₂ → *OCHO → *HCOOH → HCOOH pathways, respectively. The *CO and *HCOOH can produce the *CHO intermediate, which is crucial for the production of multi-electron products like CH₄, in addition to desorption.

Fig. 3 (a) Schematic of the analyzed reaction pathways of CO₂RR and HER in this work. (b) Gibbs free energy diagrams of CO₂ to CO on the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers (U^RHE = 0 V). (c) U_L of CO₂RR to CO on the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers. (d) ΔG of *COOH and *OCHO formations. (e) ΔG of *OCHO + H⁺ + e⁻ → *HCOOH on TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers.

The proton-coupled electron transfer (PCET) is a crucial step in the electrocatalytic process. At a basic level, PCET is defined as any process that entails the combined movement of at least one electron and one proton.⁵⁵ The electron and proton could move in the same direction or in opposite or orthogonal directions. In heterogeneous electrochemical PCET, The electrode participates in chemical bond breaking and forming, serving as a reservoir for electrons. Simultaneously, protons are transferred to the electrode surface by H₃O⁺ and facilitate the formation of reaction intermediates or products.⁵⁶ Many electrocatalytic reaction mechanisms require multiple electron and proton transfers, and these transfers could be coupled to each other or other chemical steps. In the CO₂RR process, the next challenge following CO₂ adsorption on catalyst surfaces is the first hydrogenation step: * + CO₂ + H⁺ + e⁻ → *COOH. This step represents the first PCET step of CO₂RR and is a significant obstacle due to the substantial energy required. Effective CO catalysts can reduce the free energy barrier in this step, thereby increasing the reaction rate of the entire CO₂RR process. Furthermore, HER can deplete the active sites on the catalyst surface, thereby reducing the reaction rate. It is the main competing reaction for CO₂RR.^25,57 Therefore, an ideal CO catalyst should not only have high CO₂RR activity but also inhibit other side reactions, such as the production of HCOOH, further hydrogenation of CO, and HER. In this study, we evaluate the activity of CO production on monolayers of TM@MoSi₂N₄-N_v (TM = Sc–Cu). Based on the above analysis, we consider the CO₂RR with CO generation in the following steps:

* + CO₂ + H⁺ + e⁻ → *COOH (11)

*COOH + H⁺ + e⁻ → *CO + H₂O (12)

*CO → * + CO (13)

The pristine MoSi₂N₄ monolayer has a low capacity for activating CO₂ molecules for adsorption, so forming the *COOH intermediate requires overcoming a high ΔG of 4.07 eV, which directly impedes subsequent reactions. The surface structure of the pristine MoSi₂N₄ monolayer needs to be properly modified, i.e., doped with TM atoms, due to its poor CO₂RR performance. According to the Brønsted–Evans–Polanyi relationship, reactions with lower ΔG have smaller thermodynamic barriers and are therefore kinetically favorable.^58,59 As shown in Fig. 3(b), after comparing the ΔG_max for CO production on TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers, the results indicate that the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers have lower ΔG_max values of 0.28, 0.34, and 0.16 eV, respectively, suggesting excellent CO catalytic activity. In fact, the Gibbs free energy diagrams of CO production from the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers (Fig. S5 in the ESI†) indicate that generating *COOH becomes easier in all the systems when compared to the pristine MoSi₂N₄ monolayers. This suggests that doping TM atoms in the pristine MoSi₂N₄ monolayers reduces the ΔG of generating *COOH, which significantly enhances the CO₂RR performance and increases the catalytic rate. However, the ΔG for CO production and the potential determining steps (PDS) vary with different TM@MoSi₂N₄-N_v monolayers. The PDS for the Cr, Mn, Co@MoSi₂N₄-N_v monolayers is the activation process of CO₂: * + CO₂ + H⁺ + e⁻ → *COOH. The PDS for the Sc, Ti@MoSi₂N₄-N_v monolayers is the dehydration and the CO product generation process: *COOH + H⁺ + e⁻ → *CO + H₂O. In contrast, the RDS for the V, Fe, Ni, Cu@MoSi₂N₄-N_v monolayers is the CO desorption process: *CO →* + CO. On the other hand, for the Fe, Ni, Cu@MoSi₂N₄-N_v monolayers, the formation of *COOH is the PDS. The details on the intermediate configurations are provided in Fig. S6 (ESI†). Detailed information on the distances between substrates and intermediates, as well as the binding energy and binding free energy of the intermediates, are given in Tables S6 and S7 in the ESI,† respectively. From Table S6 in the ESI,† the *COOH binding energies of nine TM, @MoSi₂N₄-N_v monolayers are all negative, indicating that the COOH product can adsorb on the monolayer surface after being generated. When the temperature is 298.15 K, the COOH binding free energies of Sc, Ti, V and Ni@MoSi₂N₄-N_v monolayers are −0.11 eV, −0.25 eV, 0.21 eV and −0.01 eV, respectively. The other five TM@MoSi₂N₄-N_v monolayers have positive binding free energies, which indicates the *COOH on the TM@MoSi₂N₄-N_v monolayers tends to desorb. Because the COOH is a reactant for the CO product generation process, *COOH + H⁺ + e⁻ → *CO + H₂O, the desorption of *COOH will hinder the CO generation. According to this, only Sc, Ti, V and Ni@MoSi₂N₄-N_z monolayers are suitable as catalysts for CO generation.

Theoretical U_L has been shown as an effective metric for evaluating catalyst activity.⁴⁶Fig. 3(c) shows that the U_L values for CO production on the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers are more positive, which are −0.28, −0.34, and −0.16 V, respectively, than those of the other monolayers. To further evaluate the CO₂RR performance of these catalysts, we have summarized the CO-generating performance of other SACs reported in previous studies for comparison (Table S8 in the ESI†). The comparison shows that the U_L of the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers is still larger than that of most catalysts, demonstrating their excellent catalytic performance. The U_L of the Mn@MoSi₂N₄-N_v monolayer (−0.16 V) is larger than that of the Mn–N₃ SAC embedded in graphitic carbon nitride (−0.24 V).⁶⁰ As shown in Fig. 3(c), the maximum value of the U_L (−0.20 V) for the SACs summarized in Table S8 in the ESI† is marked with a red dashed line. Only the U_L of the Mn@MoSi₂N₄-N_v monolayer is still larger than this value, indicating its higher catalytic activity.

It is important to consider the generation of HCOOH as a side reaction. Upon adsorption of CO₂ molecules on the catalyst surface, the proton–electron pair can attack the C atom to form the *OCHO intermediate and then attack the O atom to form the *HCOOH intermediate, which ultimately yields the HCOOH product.^61,62 Thus, *COOH and *OCHO are both intermediates in the first protonation step of CO₂RR and correspond to CO and HCOOH, respectively. To assess the selectivity of the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers between CO and HCOOH, we first examine the propensity of CO₂ to produce *COOH and *OCHO on these monolayers. As shown in Fig. 4(d), all monolayers prefer to produce *OCHO than *COOH, i.e., ΔG(*OCHO) < ΔG(*COOH). However, Fig. 4(e) shows that the production of *HCOOH requires overcoming very large ΔG for almost all the monolayers. Although it is easier to produce the *OCHO compared to the *COOH, the CO will still be the preferred product for these monolayers according to the Brønsted–Evans–Polanyi relationship.^58,59 However, for the Co@MoSi₂N₄-N_v monolayer, the ΔG for *OCHO → *HCOOH is only 0.34 eV, which is much lower than that of the other monolayers and lower than the ΔG for forming *COOH (0.71 eV). After generating *HCOOH, the HCOOH product can be produced by overcoming the ΔG of 0.49 eV. Thus, the Co@MoSi₂N₄-N_v monolayer will preferentially generate the HCOOH product over the CO product. Additionally, the PDS height is only 0.49 eV, which is lower than that of the HCOOH product generated by directly modifying the Co atoms on the MoSi₂N₄ surface (0.89 eV).³¹

Fig. 4 (a) ΔG for the first protonation step in the CO₂RR (ΔG(*COOH)) and HER (ΔG(*H)). (b) U_L difference between CO₂RR and HER (U_L(CO₂RR) − U_L(HER)).

The analysis above indicates that the ΔG_max for CO production in the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers is lower than that for HCOOH production, suggesting that *CO is easily generated from their surface. At this stage, either *CO desorbs to produce CO or continued hydrogenation of *CO generates C₁ products (Fig. 3(a)). The *CHO is crucial in the production of C₁ products such as CH₄. Therefore, we focus on the selectivity of the reaction between *CO desorption and *CO hydrogenation to produce *CHO. The free energy diagrams for CH₄ generation from the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers, where the reaction pathway is *CO → *CHO → *OCH₂ → *OCH₃ → *O + CH₄ → *OH → * + H₂O, are given as Fig. S7–S9 in the ESI.† For the Sc, Ti@MoSi₂N₄-N_v monolayers, despite the exothermic nature of the *CO → *CHO step with respect to the *CO desorption step, to complete the reaction process to produce CH₄, the subsequent hydrogenation needs to overcome the ΔG of 1.52 eV and 1.64 eV, respectively. The Mn@MoSi₂N₄-N_v monolayer only requires 0.07 eV for the protonation of *CO to form *CHO. However, this is still larger than the thermodynamic barrier for CO desorption (0.04 eV). The ΔG for the formation of CH₄ is 1.12 eV, which is unfavorable for the production of CH₄. This is consistent with the Sc, Ti@MoSi₂N₄-N_v monolayers. Furthermore, according to the Sabatier principle,⁶³ the adsorption of *CO is less stable, which favors the desorption of *CO over further reduction. In Table S7 in the ESI,† the *CO binding energies of Sc, Ti, Mn@MoSi₂N₄-N_v monolayers are −0.21, −0.24, and −0.19 eV, respectively. These values are higher than the CO binding energy of −0.75 eV calculated on edge-hosted Fe–N₄ SAC.⁶⁴ The Faraday efficiency of CO on edge-hosted Fe–N₄ SAC is up to 94%, indicating that the CO product can be smoothly desorbed after being generated on its surface. Therefore, *CO on the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers tend to desorb, producing CO products and inhibiting the generation of other products. When the temperature is 298.15 K, the *CO binding free energies G_b of Sc, Ti, Mn@MoSi₂N₄-N_v monolayers are 0.27, 0.19, and 0.26 eV, respectively. These positive values further indicate that *CO on the Sc, Ti, Mn@MoSi₂N₄-N_v monolayers is unstable and tends to desorb from the surface, which is helpful for producing CO at room temperature.

3.4 Selectivity for the CO₂RR versus the HER

It is widely acknowledged that the most probable competitive reaction in the CO₂RR process is the HER, which is * + H⁺ + e⁻ → *H,^65,66 because the H atom absorption is inevitable for CO₂RR applications. The initial hydrogenation step should be more favourable to the formation of *COOH or *OCHO than *H.^57,67 The calculated U_L of HER on the TM@MoSi₂N₄-N_v surface ranges from −0.6 to −0.03 V (Fig. S10, ESI†). In order to gain insights into the competition between HER and CO₂RR at the first protonation step, two cases with or without applied potential should be considered. In the case without the applied potential, if the ΔG of HER and that of CO₂RR are equal, the two reactions occur simultaneously. This corresponds to the 45-degree line in the coordinate system where horizontal and vertical coordinates are selected as ΔG of HER and that of CO₂RR, respectively. As shown in Fig. 4(a), the majority of data are above the line, meaning that these catalysts prefer the first protonation step of HER at an applied potential of zero. In the presence of an applied potential, the U_L is taken as an indicator of the preference for HER and CO₂RR. If the difference between U_L(CO₂RR) and U_L(HER) is a positive value, it indicates the catalyst easily generates a CO₂RR product. A more positive result indicates a higher CO₂RR selectivity of the catalyst.^14,20 This result is more reflective of the selectivity of the actual reactions because actual reactions involve the applied potential. As shown in Fig. 4(b), the result of U_L(CO₂RR) − U_L(HER) is 0.16 V on the Mn@MoSi₂N₄-N_v monolayer, which indicates that CO can be preferentially produced on its surface over H₂ in the presence of an applied potential. Therefore, considering all the competing reactions, CO will be the primary product on the Mn@MoSi₂N₄-N_v monolayer.

3.5 Origin of the catalytic activity of Mn@MoSi₂N₄-N_v

Fig. 5(a) shows that the Mn@MoSi₂N₄-N_v monolayer adsorbs CO₂, requiring a ΔG of 0.16 eV to produce *COOH. Subsequently, *CO is produced with a ΔG of 0.01 eV, which then desorbs to produce the CO product with a low ΔG of 0.04 eV. In contrast, the Cr and Co@MoSi₂N₄-N_v monolayers, which have the same PDS as the Mn@MoSi₂N₄-N_v monolayer, produce *COOH with much larger ΔG of 0.64 eV and 0.71 eV, respectively. In order to investigate the reasons behind the exceptional CO₂RR performance of the Mn@MoSi₂N₄-N_v monolayer, it is essential to analyze its origins. The Mn@MoSi₂N₄-N_v monolayer exhibits a higher CO₂ adsorption energy and transfers more charges than the other two monolayers, as shown in Fig. 5(b). This indicates that it can adsorb CO₂ molecules more stably. Additionally, Fig. 5(d)–(f) displays the PDOS of the Cr, Mn, Co@MoSi₂N₄-N_v monolayers after CO₂ adsorption, with the PDOS of CO₂ molecules included for comparison in Fig. 5(c). Upon the adsorption of CO₂, the bonding orbital of CO₂ splits and overlaps with the 3d orbital of the TM atoms. This results in the transfer of electrons from the π_g orbital of CO₂ to the empty d orbital of the TM atoms. The 3d orbital of Mn@MoSi₂N₄-N_v fully overlaps with the 2p orbital of CO₂, and the bonding orbitals are hybridized compared to those of the Cr, Co@MoSi₂N₄-N_v monolayers. This synergistic effect of electron acceptance, feedback and d-orbital occupation ensures the stable adsorption and effective activation of CO₂ and contributes to the difference in catalytic activity.^14,16,31 Moreover, the presence of an empty antibonding orbital near the Fermi energy level makes it easier for the Mn@MoSi₂N₄-N_v monolayer to accept electrons and continue the hydrogenation, which reduces the difficulty of generating *COOH.

Fig. 5 (a) Gibbs free energy diagrams of CO₂ to CO on the Cr, Mn, Co@MoSi₂N₄-N_v monolayer (U^RHE = 0 V). (b) Comparison of CO₂ adsorption energy and CO₂ transferred charges on the Cr, Mn, Co@MoSi₂N₄-N_v monolayers. (c) Molecular orbitals (MO) of the free CO₂ molecule. The PDOS of CO₂ adsorbed on the (d) Cr@MoSi₂N₄-N_v, (e) Mn@MoSi₂N₄-N_v and (f) Co@MoSi₂N₄-N_v monolayer.

In order to confirm the relationship between adsorption energy and catalytic activity, we illustrate the limiting potential U_L of the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers for CO production with the binding free energy G_b at 298.15 K of the *COOH by plotting it as a volcano plot.^2,18,66 As shown in Fig. 6, the Mn@MoSi₂N₄-N_v monolayer is closest to the summit position of the activity volcano, while the Sc, Ti@MoSi₂N₄-N_v monolayers are closer to the summit of the activity volcano, and the Cr, Co@MoSi₂N₄-N_v monolayers are far away from the summit of the volcano. This is in accordance with the Sabatier principle that catalysts with appropriate intermediate adsorption strengths have superior catalytic performance.⁶³ Also, the volcano plot using the limiting potential U_L of the TM@MoSi₂N₄-N_v (TM = Sc–Cu) monolayers for CO production and the binding energy (0 K) of the *COOH is given as Fig. S11 (ESI†), where a similar trend can be found.

Fig. 6 Activity volcano plot for two-electron reduction of CO₂ to CO, and the relationship between the binding free energy G_b (*COOH) and U_L(CO).

4. Conclusions

In this study, following the principle of single atomic catalysts, we investigate the potential of 10 TM atoms (Sc–Zn) doped on the MoSi₂N₄ monolayer as CO₂RR catalysts by first-principles calculations. Based on our screening results, we found: (1) the Sc, Ti, and Mn-doped structures exhibit very low limiting potentials; (2) out of Sc, Ti and Mn doping, the Mn@MoSi₂N₄-N_v structure shows the best catalytic performance with a limiting potential of only −0.16 V and also has an advantage over the HER. However, when Mn@MoSi₂N₄-N_v monolayers act as catalysts, the positive binding free energy of the intermediate reactant *COOH indicates the unstable adsorption of COOH on the catalytic surface, which will hinder the CO product generation process: *COOH + H⁺ + e⁻ → *CO + H₂O. So, a low temperature will be favorable for CO generation using Mn@MoSi₂N₄-N_v as catalysts. The effect of temperature will play an important role in the electrochemical catalytic reactions and our work provides some examples. Our work provides potential CO catalysts based on TM-doped MoSi₂N₄, and it is more important that the temperature effects be taken into account to get realistic insights into the electrochemical reaction process.

Data availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (NSFC, no. U20A20145 and 11974004). We gratefully acknowledge HZWTECH for providing computation facilities. This work was also supported by the High-performance Computing Platform of UESTC.

References

1. R. Wu, D. Liu, J. Geng, H. Bai, F. Li, P. Zhou and H. Pan, Appl. Surf. Sci., 2022, 602, 154239.
2. Y. Zhang, R. Yang, H. Li and Z. Zeng, Small, 2022, 18, 2203759.
3. H. Chen, A. D. Handoko, T. Wang, J. Qu, J. Xiao, X. Liu, D. Legut, Z. Wei Seh and Q. Zhang, ChemSusChem, 2020, 13, 5690–5698.
4. Y. Y. Birdja, E. Pérez-Gallent, M. C. Figueiredo, A. J. Göttle, F. Calle-Vallejo and M. T. M. Koper, Nat. Energy, 2019, 4, 732–745.
5. D. Gao, T. Liu, G. Wang and X. Bao, ACS Energy Lett., 2021, 6, 713–727.
6. Y. Chen, S. Ji, C. Chen, Q. Peng, D. Wang and Y. Li, Joule, 2018, 2, 1242–1264.
7. N. Li, X. Wang, X. Lu, P. Zhang and W.-J. Ong, Chem. – Eur. J., 2021, 27, 17900–17909.
8. T. Liu, G. Wang and X. Bao, J. Phys. Chem. C, 2021, 125, 26013–26020.
9. S. Cao, Y. Hu, C. Yang, J. Li, H. Chen, S. Wei, S. Liu, Z. Wang, D. Sun and X. Lu, J. CO2 Util., 2022, 62, 102074.
10. B. Xiong, Y. Yang, J. Liu, Z. Hua and Y. Yang, Fuel Process. Technol., 2022, 233, 107315.
11. M. Liu, Y. Pang, B. Zhang, P. De Luna, O. Voznyy, J. Xu, X. Zheng, C. T. Dinh, F. Fan, C. Cao, F. P. G. de Arquer, T. S. Safaei, A. Mepham, A. Klinkova, E. Kumacheva, T. Filleter, D. Sinton, S. O. Kelley and E. H. Sargent, Nature, 2016, 537, 382–386.
12. Q. Lu, J. Rosen, Y. Zhou, G. S. Hutchings, Y. C. Kimmel, J. G. Chen and F. Jiao, Nat. Commun., 2014, 5, 3242.
13. H. Xu, D. Cheng, D. Cao and X. C. Zeng, Nat. Catal., 2018, 1, 339–348.
14. P. Hou, Y. Huang, F. Ma, X. Wei, R. Du, G. Zhu, J. Zhang and M. Wang, Appl. Surf. Sci., 2023, 619, 156747.
15. C. Zhao, X. Su, S. Wang, Y. Tian, L. Yan and Z. Su, J. Mater. Chem. A, 2022, 10, 6178–6186.
16. Z. Lou, W. Li, H. Yuan, Y. Hou, H. Yang and H. Wang, J. Mater. Chem. A, 2022, 10, 3585–3594.
17. X.-F. Yang, A. Wang, B. Qiao, J. U. N. Li, J. Liu and T. Zhang, Single-atom catalysts: a new frontier in heterogeneous catalysis, 2013.
18. X. Cui, W. An, X. Liu, H. Wang, Y. Men and J. Wang, Nanoscale, 2018, 10, 15262–15272.
19. Y. W. Sun and J. Y. Liu, Phys. Chem. Chem. Phys., 2023, 25(6), 4773–4779.
20. F. Xu, X. Wang, X. Liu, C. Li, G. Fan and H. Xu, Mol. Catal., 2022, 530, 112571.
21. B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li and T. Zhang, Nat. Chem., 2011, 3, 634–641.
22. K. S. Novoselov, A. K. Geim, S. V. Morozov, D.-E. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666–669.
23. W. Fan, C. Liu, C. Hu, X. Liu, X. Wang, J. Wu, Z. Yu, P. Cheng, T. Yang, Q. Liu and Y. Qi, Appl. Surf. Sci., 2023, 635, 157611.
24. Z. Lu, Y. Cheng, S. Li, Z. Yang and R. Wu, Appl. Surf. Sci., 2020, 528, 147047.
25. Y. Pan, R. Lin, Y. Chen, S. Liu, W. Zhu, X. Cao, W. Chen, K. Wu, W.-C. Cheong, Y. Wang, L. Zheng, J. Luo, Y. Lin, Y. Liu, C. Liu, J. Li, Q. Lu, X. Chen, D. Wang, Q. Peng, C. Chen and Y. Li, J. Am. Chem. Soc., 2018, 140, 4218–4221.
26. Z. Yu, H. Xu and D. Cheng, Design of Single Atom Catalysts, Report null, Taylor & Francis, 2021.
27. W. Ding, Z. Xue, J. Li, M. Li, L. Bai, Q. Zhou, X. Zhou, Y. Peng and L. Miao, ECS J. Solid State Sci. Technol., 2022, 11, 016001.
28. W. Shi, G. Yin, S. Yu, T. Hu, X. Wang and Z. Wang, J. Mater. Sci., 2022, 57, 18535–18548.
29. C. Xiao, R. Sa, Z. Cui, S. Gao, W. Du, X. Sun, X. Zhang, Q. Li and Z. Ma, Appl. Surf. Sci., 2021, 563, 150388.
30. Y. Luo, M. Li, Y. Dai, X. Zhang, R. Zhao, F. Jiang, C. Ling and Y. Huang, J. Mater. Chem. A, 2021, 9, 15217–15225.
31. W. Xun, X. Yang, Q.-S. Jiang, M.-J. Wang, Y.-Z. Wu and P. Li, ACS Appl. Energy Mater., 2023, 6, 3236–3243.
32. Y. Linghu, T. Tong and C. Wu, Appl. Surf. Sci., 2023, 609, 155332.
33. T. Tong, Y. Linghu, G. Wu, C. Wang and C. Wu, Phys. Chem. Chem. Phys., 2022, 24, 18943–18951.
34. S. Lu, Y. Zhang, F. Lou, K. Guo and Z. Yu, Appl. Surf. Sci., 2022, 579, 152234.
35. Y. Zang, Q. Wu, W. Du, Y. Dai, B. Huang and Y. Ma, Phys. Rev. Mater., 2021, 5, 045801.
36. S. Cao, H. Chen, Y. Hu, J. Li, C. Yang, Z. Chen, S. Wei, S. Liu, Z. Wang, D. Sun and X. Lu, Chem. Eng. J., 2023, 461, 141936.
37. J. M. Spurgeon and B. Kumar, Energy Environ. Sci., 2018, 11, 1536–1551.
38. S. Jin, Z. Hao, K. Zhang, Z. Yan and J. Chen, Advances and Challenges for the Electrochemical Reduction of CO₂ to CO: From Fundamentals to Industrialization, Report, 2021, 1433–7851.
39. Z. Gao, Y. Meng, A. Koso, J. Mishima, B. Xie, Z. Ni and S. Xia, Colloids Surf., A, 2022, 648, 129365.
40. X. Guan, C. Zhao, X. Liu, S. Liu, W. Gao and Q. Jiang, J. Phys. Chem. C, 2020, 124, 25898–25906.
41. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953.
42. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758.
43. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
44. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J. Chem. Phys., 2010, 132, 154104.
45. S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys and A. P. Sutton, Phys. Rev. B: Condens. Matter Mater. Phys., 1998, 57, 1505–1509.
46. A. A. Peterson, F. Abild-Pedersen, F. Studt, J. Rossmeisl and J. K. Nørskov, Energy Environ. Sci., 2010, 3, 1311–1315.
47. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. R. K. J. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jonsson, J. Phys. Chem. B, 2004, 108, 17886–17892.
48. J. Rossmeisl, A. Logadottir and J. K. Nørskov, Chem. Phys., 2005, 319, 178–184.
49. Z. Cui, K. Yang, K. Ren, S. Zhang and L. Wang, Mater. Sci. Semicond. Process., 2022, 152, 107072.
50. G. Liu, A. W. Robertson, M. M. J. Li, W. C. H. Kuo, M. T. Darby, M. H. Muhieddine, Y. C. Lin, K. Suenaga, M. Stamatakis, J. H. Warner and S. C. E. Tsang, Nat. Chem., 2017, 9, 810–816.
51. B. Hammer and J. K. Norskov, Nature, 1995, 376, 238–240.
52. T. K. Todorova, M. W. Schreiber and M. Fontecave, ACS Catal., 2020, 10, 1754–1768.
53. S. Xu and E. A. Carter, Theoretical Insights into Heterogeneous (Photo)electrochemical CO₂ Reduction, Report 0009-2665, American Chemical Society, 2019.
54. H. Li, T. Liu, P. Wei, L. Lin, D. Gao, G. Wang and X. Bao, Angew. Chem., Int. Ed., 2021, 60, 14329–14333.
55. S. Hammes-Schiffer, Energy Environ. Sci., 2012, 5, 7696–7703.
56. R. E. Warburton, A. V. Soudackov and S. Hammes-Schiffer, Theoretical Modeling of Electrochemical Proton-Coupled Electron Transfer, Report 0009-2665, American Chemical Society, 2022.
57. Y. Tian, T. Zhao, C. Zhao and Y. Likai, Appl. Surf. Sci., 2022, 597, 153724.
58. J. N. Brønsted, Chem. Rev., 1928, 5, 231–338.
59. M. G. Evans and M. Polanyi, Trans. Faraday Soc., 1938, 34, 11–24.
60. J. Feng, H. Gao, L. Zheng, Z. Chen, S. Zeng, C. Jiang, H. Dong, L. Liu, S. Zhang and X. Zhang, Nat. Commun., 2020, 11, 4341.
61. Y. Liu, J. Zhao and Q. Cai, Phys. Chem. Chem. Phys., 2016, 18, 5491–5498.
62. Z. Jiang, T. Wang, J. Pei, H. Shang, D. Zhou, H. Li, J. Dong, Y. Wang, R. Cao, Z. Zhuang, W. Chen, D. Wang, J. Zhang and Y. Li, Energy Environ. Sci., 2020, 13, 2856–2863.
63. M. Che, Catal. Today, 2013, 218–219, 162–171.
64. F. Pan, B. Li, E. Sarnello, Y. Fei, X. Feng, Y. Gang, X. Xiang, L. Fang, T. Li, Y. H. Hu, G. Wang and Y. Li, ACS Catal., 2020, 10, 10803–10811.
65. N. Li, X. Chen, W.-J. Ong, D. R. MacFarlane, X. Zhao, A. K. Cheetham and C. Sun, ACS Nano, 2017, 11, 10825–10833.
66. Y. Xiao and W. Zhang, Nanoscale, 2020, 12, 7660–7673.
67. X. Wang, H. Niu, Y. Liu, C. Shao, J. Robertson, Z. Zhang and Y. Guo, Catal. Sci. Technol., 2020, 10, 8465–8472.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cp03493g

‡ These authors share first authorship.

---