Supported Bimetallic Trimers Fe2M@NG: Triple-Atom Catalysts for CO2 Electroreduction.

Excessive accumulation of carbon dioxide in the atmosphere has become a serious environmental problem due to the increasing consumption of fossil fuels in modern society. Reasonably reducing CO2 in the atmosphere has become a new research hotspot. Electrocatalytic CO2 reduction reaction (CO2RR) offers an appealing strategy to reduce the atmospheric CO2 concentration and to produce value-added chemicals simultaneously. In this paper, two-dimensional (2D) N-decorated graphene (NG)-supported bimetallic trimers (Fe2M@NG) were designed as triple-atom catalysts (TACs). Theoretical calculations showed that Fe2M@NG can effectively activate CO2, and among the 23 TACs examined, Fe2Ir@NG not only has a good catalytic activity for CO2RR (limiting potential is 0.49 V for CH4 formation) but also limits the competing side reaction of the hydrogen evolution reaction (HER). Our theoretical study not only further extends the triple-atom catalysts, but also opens a new door to boost the sustainable CO2 conversion.

Introduction

For centuries, the utilization of fossil fuels, coal, oil, and natural gas has led to the rapid development of human society.[1−3] However, the ever-increasing energy demands and serious environmental problems require the “optimization” of these traditional energy sources.[4] The massive consumption of these carbonaceous substances results in the continuous increase and accumulation of carbon dioxide (CO2) in the atmosphere.[5] There are mainly three strategies to prevent increasing atmospheric CO2 concentration: prohibition of CO2 release, CO2 storage, and CO2 conversion.[6] Among them, the CO2 conversion technology has attracted the attention as the most promising approach to slow down or even reverse the rising trend of atmospheric carbon dioxide concentration.[7,8] The electrochemical CO2 reduction reaction (CO2RR) can directly convert CO2[9] into high value-added chemicals and fuels,[10,11] and renewable electricity can be used for this electrocatalytic process, resulting in a “net-zero emission” sustainable development.[12] As early as the 1990s, Hori et al. demonstrated the ability of different pure metal catalysts for the electrochemical reduction of CO2, providing a solid foundation for subsequent CO2RR.[13] However, CO2 molecules are chemically inert and the hydrogen evolution reaction (HER) competes fiercely with CO2RR, and it is challenging to design suitable CO2RR electrocatalysts with low limiting potential, high current density, high selectivity, and low cost.[14]

Recent studies showed that nitrogen-doped carbon (N-C)-supported transition-metal atom catalysts can effectively reduce CO2 to single-carbon and multicarbon species.[15−18] N-C-based materials have attracted attention because of their high-temperature resistance, acid and alkali resistance, poison resistance, and environmental protection.[19] Compared to the low activity of pure carbon materials,[20] on the one hand, the N-doped carbon materials as a support can provide stable anchoring sites for the metal atoms and thus endow the system with the characteristics of high stability, selectivity, and low coordination state;[21] on the other hand, its strong interactions with the supported metal atoms can regulate the electronic structure of the catalyst[22] and thus enhance the adsorption of CO2 and facilitate CO2RR.[23] The single-atom catalysts (SACs) were considered as the minimum use of metal,[24−27] and expansively, double-atom catalysts (DACs)[27−29] and triple-atom catalysts (TACs)[17,30−33] have gained enormous attention from both theoretical and experimental aspects owing to their synergistic effect and tunable composites in the metal dimers or trimers. For example, the experimentally obtained uniform Ru3 clusters stabilized by nitrogen species (Ru3/CN) were found to exhibit excellent catalytic activity for the oxidation of alcohols.[32] Ma and collaborators fabricated Pt3 clusters on a core–shell nanodiamond@graphene (ND@G) hybrid support, which inhibits the side reactions and enhances catalytic performance in the direct dehydrogenation of n-butane at a low temperature (450 °C) toward olefin products with a selectivity >98%.[33]

Inspired by the theoretical finding that the stable Fe3@NG (iron trimer-embedded N-decorated graphene) exhibits high catalytic activity to convert CO2 through the C2 and C3 pathways[17] and considering the large space of mediating the combination and electronic coupling of the bimetallic trimers, we explored the stability and catalytic performance of 23 Fe-based bimetallic TACs (Fe2M@NG) for CO2 reduction to C1 products by means of first-principles calculations and elucidated that the Fe2Ir@NG has excellent catalytic activity for the electroreduction of CO2 to CH4 with a small limiting potential of 0.49 V and can also inhibit the competing hydrogen evolution reaction (HER). Our theoretical explorations provide an effective approach for designing bimetallic trimers of high-performance for electrocatalyzing CO2 reduction.

Computational Methods

All of the spin-polarized density functional theory (DFT) calculations were carried out using the Vienna Ab initio Simulation Package (VASP).[34] The exchange–correlation functional was described by the Perdew–Burke–Ernzerhof (PBE) parameterization of the generalized gradient approximation (GGA).[35] The cutoff energy of 500 eV was adopted. The van der Waals interactions were described using the empirical correction in the Grimme scheme (DFT + D3).[36] The convergence parameters of geometric optimization of the maximum force and energy were designated as 0.01 eV/Å and 10–4 eV, respectively. The Brillouin k-point grid was sampled using a 2 × 2 × 1 γ centered Monkhorst–Pack scheme.[37] To weaken the interaction between the layers, a 20 Å vacuum layer was applied. The binding energy (Eb) of Fe2M@NG based on Fe2@NG was determined by the following equationwhere EM represents the energy of a single metal atom, EFe2M@NG and EFe2@NG are the energies of Fe2M@NG and Fe2@NG, respectively. The free energy G was calculated based on the hydrogen electrode (CHE) model proposed by Nørskov and colleagues[38]where E is the reaction energy, which can be directly obtained from the DFT calculation. G(T) = EZPE – TS, where EZPE and S are the zero-point energy and entropy, respectively, which can be obtained by calculating the vibration frequencies.[39]T is 298.15 K.[40] The calculated values of gas molecules and intermediates are presented in Tables S1 and S2, respectively. Free energy change (ΔG) of the elementary reaction represents the free energy difference between the product and the reactant (ΔG = Gproduct – Greactant).[41] The limiting potential (η) of the entire reduction process is determined by the potential limiting step, which has the most positive (ΔGMax), as computed by η = ΔGMax/e.[42]

Results and Discussion

Structure and Stability of Fe2M@NG

A 7 × 7 graphene supercell doped with six N atoms, forming a hole structure similar to that of a pyridine hole, as shown in Figure a, was taken as the support for bimetallic trimers.[17] The optimized configurations of the 23 TACs examined are displayed in Figure S1, and the geometric parameters are given in Table S3. The binding energies were in the range of 1.25–7.70 eV (Figure b). With similar treatment in ref (17), we performed a first-principles molecular dynamics (FPMD) simulation for Fe2Zn@NG, whose binding energy is the smallest among the 23 Fe-based TACS, in the NVT ensemble with a temperature of 500 K, a time step of 0.5 fs, and the total time scale of 10 ps. The Fe2Zn@NG can maintain its original structure through the 10 ps’s FPMD simulation at 500 K (Figure S2), indicating the high thermal stability in Fe2Zn@NG and other bimetallic TACs.

Figure 1

(a) Top view of Fe2M@NG and the considered transition metal M. (b) Formation energy of Fe2M@NG.

CO2 Adsorption

The capture of CO2 molecules on the catalyst is the first step in the CO2RR process.[43] Due to the strong interaction between CO2 and the catalyst, the C–O–C angle of the adsorbed CO2 in the most stable configuration on these TACs is reduced to 121.36–139.78° compared to 180° of the isolated CO2 molecule. We denoted the O of CO2 closer to the TAC as O2, and the other as O1. The lengths of the C–O1 and C–O2 bonds were elongated to 1.20–1.26 and 1.29–1.45 Å, respectively, compared with 1.18 Å of free CO2. For comparison, C–O1 and C–O2 bonds are 1.21 and 1.33 Å for the adsorbed CO2 on Fe3@NG. The adsorption energy (Ead) of CO2 was estimated by the following formulawhere E, E*, and ECO represent the energies of CO2-adsorbed Fe2M@NG, clean Fe2M@NG, and a free CO2 molecule, respectively. The Ead values on these stable TACs ranged from −3.21 to −0.86 eV, and the adsorption energy of CO2 on Fe3@NG was calculated to be −1.57 eV, the same as the value reported in ref (17). The charge density difference of the CO2-adsorbed Fe2M@NG (Figure ) showed that the electron depletion (cyan) is around bimetallic trimers, suggesting that the metal trimers donate electrons to the CO2, and Bader charge analysis confirmed that the electrons are transferred from Fe2M@NG to CO2 by 0.57–0.98 e–. The DOS (density of states, Figure S3) clearly shows the couplings between the d orbitals of the transition metal and the molecular orbitals of CO2 near the Fermi level. The above results indicate that the TACs well activate the adsorbed CO2, which is beneficial to facilitate CO2 activation and reduction.

Figure 2

Differential charge diagram of CO2-adsorbed Fe2M@NG. The isosurface value was set as 0.002 e/Å3. Yellow and cyan regions indicate electron accumulation and depletion, respectively.

CO2 Reduction Pathways

To find the optimal CO2RR pathway, the free energies of the reaction over these TACs were calculated.[44] The reaction pathways of all catalysts are given in Table S3, and the ΔG of each elementary step is illustrated in free energy diagrams (Figures and S4–S8). The CO2 electroreduction reaction involves multiple electron transfer paths, and the general form is CO2 + nH+ + ne– → products + yH2O.[45] In the formula, n is usually equal to 2, 4, and 8, and the corresponding products are CO/HCOOH, methanol, and methane.[46] The C1 products and pathways are discussed in the following subsections, and the potential limiting steps (PLSs) and limiting potentials (η) of our examined TACs are illustrated in Table S4.

Figure 3

(a) Possible C1 pathways of CO2 reduction on the Fe2Ir@NG, and (b) the corresponding free energy diagrams. Data denote the ΔG of each elementary step.

First Proton-Coupled Electron Transfer (PCET) Process

For n = 1, the first proton-coupled electron transfer (PCET) process involves the hydrogenation of the C or O atom.[47] We calculated the free energies of the first hydrogenation of *CO2 in Figure S4. Therefore, the reaction pathway in this study was along * → *OCO → *H + *(O1/C/O2), i.e., hydrogenation occurs at different positions (O1, C, O2 atoms). After structural optimization, we noticed that it is easier to form *OCOH on the five Fe2M@NG (M = Co, Ni, Rh, Pd, Re), whose free energy changes are 0.16, 0.16, 0.44, −0.16, and 0.24 eV, respectively. *OCHO is favored to be generated over two Fe2M@NG (M = Cr, Mn) with ΔG values of −0.44 and −0.34 eV, respectively. In addition, on the other 16 Fe2M@NG (M = Sc, Ti, V, Cu, Zn, Y, Zr, Nb, Mo, Ru, Hf, Ta, W, Os, Ir, and Pt), protons preferentially react with O2 atom, thereby promoting breaking of the O1C–O2 bond. Therefore, compared with other products, the free energy for forming *CO + *OH on these TACs exhibits a downhill trend (−0.42, −1.15, −0.45, −1.12, −0.67, −0.55, −0.64, −0.85, −1.42, −0.90, −0.69, −0.81, −0.71, −0.72, −0.74, and −0.61 eV, respectively).

CO2 Reduction to CO

After the first hydrogenation of *CO2, the intermediates (*CO + *OH, *OCHO, and *OCOH) became the targets of the next proton–electron pair attack. Based on our computations, the following intermediates may be formed in the subsequent hydrogenation process: *COH + *OH (M = Sc), *CHO + *OH (M = Cr, Cu, W), *CO (after releasing a water molecule, M = Ti, V, Co, Ni, Zn, Y, Zr, Nb, Mo, Ru, Pd, Hf, Ta, Os, Ir, Pt), and *OCHOH (M = Mn, Rh, Re). The free energies of releasing the adsorbed CO were in the range of 0.92–2.62 eV; thus, the strong interaction between CO(g) and the TACs (Ti, V, Co, Ni, Zn, Y, Zr, Nb, Mo, Ru, Pd, Hf, Ta, Os, Ir, Pt) makes these TACs suffer CO poisoning; however, strong adsorption is beneficial to further hydrogenation of *CO on the other hand.

CO2 Reduction to CH3OH(g)

Regarding the reduction of CO2 to CH3OH(g), we will continue the discussion based on the above results. For the third hydrogenation, the *COH + *OH intermediate on the Fe2Sc@NG released a water molecule to form the *COH intermediate. The *CO species on the TACs may further form *COH (M = Ti, V, Zn, Y, Nb, Mo, Ru, Pd, Hf, Ta) or *CHO (M = Zr, Os, Ir, Pt). The *OCHOH on Fe2M@NG (M = Mn, Rh, Re) may also be converted into *COH (M = Re) or *CHO species (M = Mn, Rh) by releasing an H2O molecule. The *CHO + *OH species on Fe2M@NG (M = Cr, Cu, W) were further reduced to *CHOH + *OH (M = Cu, W) and *CH2O + *OH (M = Cr). From the point of energy calculations, these hydroxyl intermediates will remain until the final hydrogenation into *H2O. For the fourth hydrogenation, *CHOH (M = Sc, Ti, V, Zn, Y, Nb, Mo, Ru, Pd, Hf, Ta, Os, Ir, Pt), *CH3O + *OH (M = Cr), *CH2OH + *OH (M = Cu, W), and *CH2O (M = Mn, Zr, Rh, Re) will be formed on the Fe2M@NG. Subsequently, these intermediates will be further reduced to *CH2OH (M = Sc, Ti, V, Zn, Y, Zr, Nb, Mo, Ru, Rh, Pd, Hf, Ta, Re, Os, Ir, Pt) and *OH (M = Cu, W, Cr) on Fe2M@NG. For the final hydrogenation, *CH3OH(g) (M = Sc, Ti, V, Zn, Y, Zr, Nb, Mo, Ru, Rh, Pd, Hf, Ta, Re, Os, Ir, Pt) and *H2O (M = Cu, W, Cr) will be formed. Moreover, the Fe2M@NG (M = Cr, Cu, Rh, Pd, Ir) have small limiting potentials (0.76, 0.75, 0.55, 0.76, 0.60 V, respectively) in the pathway for CH3OH formation. Their PLSs in the generation of CH3OH(g) are *OH + (H+ + e–) → *H2O, (*CO + *OH) + (H+ + e–) → *CHO + *OH, *CHOH + (H+ + e–) → CH2OH, *COH + (H+ + e–) → *CHOH, and *CHOH + (H+ + e–) → *CH2OH, respectively. In comparison, the η to form CH3OH(g) over the Fe3@NG is 1.09 eV (PLS: *COH→*CHOH, see Table S4 and Figure S8), largely because *COH is adsorbed on the hollow position of the Fe3 cluster, making it unsuitable to attach H to C of *COH.

CO2 Reduction to CH4(g)

The first few elementary steps are the same for the formation of CH4(g) and CH3OH(g); we thus mainly discuss the variant hydrogenating steps. For these Fe2M@NG (M = Sc, Ti, V, Co, Ni, Zn, Y, Nb, Mo, Ru, Pd, Hf, Ta), the hydrogenation of *COH follows the pathway of *C → *CH → *CH2 → *CH3 → CH4(g) to generate CH4(g); the *CHO intermediates on the three Fe2M@NG (M = Os, Ir, Pt) will be converted into CH4(g) as follows, which is further hydrogenated to form *CH, and then follows the pathway: *CHO → *CHOH → *CH → *CH2 → *CH3 → CH4(g). On the Fe2Cr@NG and Fe2W@NG, the pathways are *CH3O + *OH → *O + *OH+CH4(g) → *OH+OH → *OH → *H2O and *CHOH + *OH → *CH + *OH → *CH2+OH → *CH3 + *OH → *OH → *H2O, respectively. The three Fe2M@NG (M = Sc, Ni, Ir) have small limiting potentials (0.71, 0.77, 0.49 V) in producing CH4(g), and the PLSs are *CH3 + (H+ + e–) → CH4(g), *CO + (H+ + e–) → *COH, and *CHO + (H+ + e–) → *CHOH, respectively. Overall, for CO2RR on the 23 bimetallic Fe2M@NG catalysts, the Fe2Ir@NG has the smallest limiting potential (η = 0.49 V) on the path toward CH4 generation (* → *CO2 → *CO + *OH → *CO → *CHO → *CHOH → *CH → *CH2 → *CH3 → CH4(g)) among the three C1 conversion routes (Figure ), lower than the η value of the Fe3@NG at the same level of theory (Figure S8, η = 0.79 V).

Essential Analysis of Catalytic Performance

Previous studies reveal that an ideal catalyst should provide moderate adsorption strength for all reactants, intermediates, and products.[48] Therefore, we evaluated the relationship between the limiting potential of the C1 pathways over Fe2M@NG and the CO2 adsorption energy (Figure a). The results revealed that there is a volcanic relationship between the limiting potentials of the C1 pathways and the CO2 adsorption energies. By comparison, we found that six Fe2M@NG have moderate Ead for CO2 (ranging from −1.40 to −1.13 eV) and have smaller η (0.77, 0.75, 0.55, 0.76, 0.70, and 0.49 V for Ni, Cu, Rh, Pd, Re, and Ir, respectively) compared with the Fe3@NG (η = 0.79 V, Ead = −1.57 eV). As the main competitive reaction of CO2RR, the hydrogen evolution reaction (HER) also consumes proton–electron pairs (H+ + e–),[49] we considered the HER activity of the Fe2M@NG. We first calculated the Ead of *H on Fe2M@NG and compared with the Ead of *CO2 (Figure b) and found that on the Fe2M@NG (M = Rh, Ni, Pd, Mn, Pt), whose η values are lower than 0.79 V of Fe3@NG, the Ead of *H is more negative than that of CO2, indicating that it is not conducive for the progress of CO2RR on these TACs.[17] However, the Fe2Ir@NG has a stronger interaction with *CO2 than *H, and the ΔG (0.94 eV) for the HER (Figure S9) is much greater than the maximum ΔG for CO2RR toward CH4(g) (0.49 eV). Thus, the Fe2Ir@NG has high selectivity for CO2RR.

Figure 4

(a) Variation of limiting potentials of Fe2M@NG catalysts for CO2 reduction to C1 products with respect to the CO2 adsorption strength, and (b) comparison of the adsorption energies of *CO2 and *H on Fe2M@NG.

According to the density of states (Figure S3), compared to the case of Fe3@NG, in bimetallic TACs, the coupling between M and Fe2 weakens or strengthens the hybridization with CO2 near the Fermi level and thereby modulates the catalytic capability of the TACs. Among the 23 bimetallic TACs, the Fe2Ir@NG has a medium adsorption strength for CO2 and a minimum limiting potential and can effectively suppress the HER. All in all, the Fe2Ir@NG is a very promising catalyst for CO2 reduction to CH4.

Conclusions

In summary, by performing comprehensive first-principles calculations, we explored the stability and catalytic behavior of 23 bimetallic triple-atom catalysts (Fe2M@NG) for CO2 reduction to C1 products (including CO, CH3OH, and CH4). Our results showed that the supported bimetallic trimers provide sufficient adsorption sites to adsorb CO2, thereby effectively activating the CO2 and promoting the breaking of the C–O bond. The synergy between Ir and Fe atoms in the outstanding TAC of Fe2Ir@NG improves the catalytic activity (low limiting potential 0.49 V for CH4 formation) and selectivity (reluctant for the competing reaction HER) for CO2RR. Our theoretical study offers guidance to design bimetallic TACs with high catalytic performance beyond CO2RR.