1 Introduction
In recent years, renewable energy driven chemical processes, such as CO
2 reduction reaction [
1,
2], nitrogen reduction reaction [
3], overall water-splitting reaction [
4,
5], hydrogen evolution reaction [
6,
7], and oxygen evolution reactions [
8] are now viewed as environmentally friendly and sustainable pathways for producing valuable chemical products. The viability of the electrocatalytic synthesis hinges largely on the performance of the electrocatalysts involved. Hence, the development of high-performance and economically viable electrocatalysts is crucial for enabling their widespread practical implementation [
9]. Single-atom catalysts (SACs) exhibit significant advantages in the field of electrocatalytic CO
x reduction due to their unique metal-ligand atomic electronic structure [
10,
11]. However, the spatial isolation properties between SAC active sites result in a product selectivity that is usually limited to the C
1 product, such as CO [
12,
13] and CH
4 [
14]. The C
2 products, including C
2H
4 [
15], C
2H
5OH [
16,
17], and C
2H
6 [
18], are characterized by their higher energy density and industrial application value. However, the synthesis of C
2 products involves complex processes with multi-electron transfer and C−C coupling, and its reaction energy barrier is significantly higher than that of the C
1 product generation pathway [
19]. In recent years, it has been shown that the issue can be effectively solved by introducing synergistic catalytic sites through the construction of dual-atom catalysts (DACs) [
20−
23]. Unlike SACs where isolated sites typically limit products to C1 species, the paired active sites in DACs can accommodate more complex reaction pathways, providing the spatial flexibility and multi-site synergy crucial for C−C coupling. This unique advantage is evidenced by recent experimental advances. For instance, dual-copper-site catalysts have been shown to promote ethylene formation with a Faradaic efficiency of 52% during CO
2 electroreduction [
24], while asymmetric In−Cu and Ni−Co dual-atom pairs have achieved high selectivity (up to 92%) for C
2 products in photocatalytic CO
2 reduction [
25,
26]. Since synergistic effect of metal atoms with different valence states can promote C−C coupling [
27−
29], it is believed that dimers composed of different metal atoms in bimetallic DACs can provide functional active sites for C
2 production. By selecting different metals, the net charge difference between metal atoms in the dimer can be regulated to facilitate C−C coupling and potentially enhance selectivity toward C
2 products. Currently, the dominant bimetallic DACs are two-dimensional nitrogen-doped carbon materials such as Fe−Co bisites [
30], Ni−Fe bisites [
31], and Co−Pt bisites [
32]. Their catalytic performances are highly dependent on the microstructure and coordination of the DACs, and can be improved by external fields or doping, which is however a high energy costing and challenging process. Recent theoretical and experimental works have demonstrated that, the adjustable polarization direction in ferroelectric materials is an alternative and visible strategy to modulate the catalytic activity of their surfaces and the final reaction products [
4,
33−
35]. Different from single atom anchored ferroelectric substrate, in DACs the polarization is expected to not only modulate the charge transfer from substrate to the metal, but also regulate the asymmetric charge distributions among two metal sites to facilitate C−C coupling and tune C
2 product selectivity.
In this study, we anchor dual transition metal atoms on the surface of ferroelectric α-In2Se3 monolayer to form ferroelectric DAC. The theoretical calculations demonstrate that the positive and negative charge center pair (PNCCP) is formed between the two transition metal (TM) atoms, which can effectively lower the energy barrier of C−C coupling for C2 productions. Besides, the switched polarization can not only tune the reaction barrier and path of CO reduction reaction (CORR), but also lead to different final products due to the modulated charge states among the metal sites. These findings unveil the potential of ferroelectric DACs as tunable and selective catalysts for CORR, offering a feasible strategy for rational electrocatalysts design to produce high-value fuels.
2 Methods
The density functional theory (DFT) calculations are performed with the Vienna ab initio simulation package (VASP) code [
36,
37]. The spin-polarized generalized gradient approximation (GGA) in the form of Perdew−Burke−Ernzerhof (PBE) treats the exchange−correlation interactions, while the frozen-core projector augmented wave (PAW) approximation describes the interaction between the ion and electron [
38−
40]. The van der Waals (vdW) interactions are described with the DFT-D3 method in Grimme’s scheme [
41]. Two TM atoms are distributed on two adjacent hexagonal centers of 4 × 4 In
2Se
3 supercells, which act as catalytic active sites. The dipole correction is taken into account for all the asymmetric structures [
42]. The 2D Brillouin sampling is presented with a 1 × 1 × 1 gamma-centered Monkhorst-pack k-mesh. To ensure that the periodic images in the non-periodic direction do not interact with each other, a vacuum region of more than 20 Å is reserved. The cutoff energy for plane-wave basis sets is 500 eV. The convergence thresholds for force and total energy are 10
−2 eV/Å and 10
−5 eV, respectively. The site-specific charge differences are obtained using Bader analysis. Besides, when calculating the Gibbs free energy of CORR, we consider the solvent effect by using the implicit water solvent model implemented in VASPsol [
34,
43], with a relative dielectric constant of 80 to represent the aqueous environment. More details of the simulations on Gibbs free energy can be found in the Supporting Information. In particular, for all calculations in this work, the thermodynamic state of the product H
2O is defined as the liquid phase at 298.15 K.
3 Results and discussion
3.1 Configurations of ferroelectric DACs
The van der Waals layered α-In
2Se
3, a two-dimensional room-temperature ferroelectric semiconductor, is selected as the substrate. This material exhibits robust structural stability under ambient conditions and has been successfully synthesized experimentally [
44]. Notably, its switchable spontaneous polarization with reversible behavior at 300 K has been unequivocally validated through piezoresponse force microscopy [
44]. As shown in Fig. S1, the polarization orientation, locked by the inherent asymmetry in the quintuple-layer stacking configuration, can be inverted via displacement of the central selenium atoms within the lattice [
35,
45]. The heteronuclear and homonuclear ferroelectric DACs (TM
1−TM
2@In
2Se
3 and TM
1−TM
1@In
2Se
3) are constructed with the ferroelectric In
2Se
3 substrate anchored with TM atoms. According to our previous work [
35], only five types of single TM atom anchored In
2Se
3 monolayer (TM@In
2Se
3, TM = Ni, Pd, Rh, Nb, and Re) are stable [
35]. Therefore, in this work we select these five TM atoms to construct ten heteronuclear TM
1−TM
2@In
2Se
3 and five homonuclear TM
1−TM
1@In
2Se
3 models. Under the configuration of downward polarization (P↓), we investigate the stability of these ferroelectric DACs initially based on the formation energy (
Efor), which is obtained with Eq. (ES1) in the Supporting Information. Based on the definition, the configurations with a negative formation energy are synthesizable. Therefore, three TM
1−TM
2@P↓-In
2Se
3 hetero-configurations (Ni−Nb, Pd−Nb, and Rh−Nb) and Nb−Nb@P↓-In
2Se
3 homo-configuration are preliminarily screened out as candidate catalysts [Fig. 1(a)]. As listed in Table S1 for the P↑ situation, although the TM
1−TM
2@P↑-In
2Se
3 hetero-configurations (Ni−Nb, Pd−Nb, Rh−Nb) and the Nb−Nb@P↑-In
2Se
3 homo-configuration exhibit positive formation energies, their desorption energies are very high (>10 eV). Consequently, once synthesized in the P↓ state, polarization reversal cannot dislodge the single TM atoms from the In
2Se
3 surface in these heteronuclear and homonuclear ferroelectric DACs. To further demonstrate the structural stability of these candidates, we employ molecular dynamics simulation (AIMD) calculations on Ni−Nb@P↓-In
2Se
3 model, which has the highest formation energy among these four ferroelectric DACs in the P↓ state. AIMD simulations are conducted at temperature of 500 K, with a time step of 1 fs, and a total time length of 10 ps. During the AIMD simulations, the geometrical structure remains intact, indicating that Ni−Nb@P↓-In
2Se
3 has good thermal stability [see Figs. 1(b) and S2]. Overall, both the formation energy and molecular dynamics simulations promise the structural stability of these four ferroelectric DAC configurations. Regarding experimental realization, the synthesized DACs (e.g., Pd−Nb@In
2Se
3 and Rh−Nb@In
2Se
3) could potentially be achieved via advanced techniques such as atomic layer deposition [
46,
47] or wet chemical synthesis [
48], which have proven effective in creating tailored dual-metal sites on various substrates.
3.2 Positive and negative charge center pair in the ferroelectric DACs
Previous reports have indicated that the occurrence of C−C coupling reactions can be effectively promoted if the bimetallic sites in DAC can form PNCCP on the substrate. Therefore, we next screen the samples that can form the PNCCP, from the three TM1−TM2@In2Se3 configurations (Ni−Nb, Pd−Nb, and Rh−Nb). As listed in Table S2, in the Pd−Nb@In2Se3 and Rh−Nb@In2Se3 configurations, the adsorbed TM1 and TM2 single atoms gain and lose electrons respectively, to form PNCCP. Taking Pd−Nb@P↓-In2Se3 as an example, the adsorbed Nb single atom donates 2.01 electrons to the substrate, forming a positive charge center, while the Pd single atom accepts 0.16 electrons from the substrate, forming a negative charge center (as shown in Fig. 2). Consequently, the Nb−Pd dual-atom pair constitutes an opposite charge center pair. Similarly, in Pd−Nb@P↑-In2Se3, the adsorbed Nb atom donates 1.87 electrons to form a positive charge center, whereas the Pd atom accepts 0.15 electrons to form a negative charge center. Therefore, while polarization reversal modulates the intensity of the opposite charge center pair via reversal of the internal electric field, it does not alter the charge sign of the pair. This is clearly demonstrated in the Rh−Nb system, where the electron gain at the Rh site decreases from 0.24 in the P↓ state to 0.11 in the P↑ state, while the electron donation from its Nb partner changes from −1.96 to −2.07. Subsequently, we employ the Pd−Nb@In2Se3 and Rh−Nb@In2Se3 systems to investigate whether PNCCP could effectively promote C−C coupling, along with examining the effect of polarization switching on their CORR catalytic performances.
3.3 Tunable CORR catalytic performance by polarization switching
The switchable polarization within ferroelectric materials significantly affects the electronic structure of the catalyst by modulating the
d-band center position [
35]. When the ferroelectric substrate undergoes a polarization flip (see Fig. S3), its surface stoichiometry ratio and the electronic environment of the adsorption site are reconfigured, resulting in the significantly changed adsorption energy of CO molecules (Δ
G*CO) and energy barrier of the C−C coupling. Therefore, it is expected that the reaction barriers, reaction paths, and even the intermediate and final products of CO reduction could be controlled by ferroelectric switching. As illustrated in Fig. S4, we systematically explore the lowest-energy pathways for CORR across Pd−Nb and Rh−Nb DACs under different polarization states. Table 1 summarizes the effect of polarization switching on CORR in terms of C−C coupling type, potential-determined hydrogenation steps, limiting potential and final products.
The electrochemical CORR over the Pd−Nb@P↓-In2Se3 catalyst follows the reaction pathway illustrated in Fig. 3(a). The process begins with the adsorption of CO molecules onto the Pd site of catalyst surface (denoted as CO*), where the Pd−Nb bimetallic sites play a critical role in stabilizing the adsorbed species. Subsequently another CO adsorption at the Nb site (CO*+CO → CO*CO*) is facilitated by synergistic electronic interactions between Pd and Nb, but the two C atoms from the two adsorbed CO molecules are not bonded. First hydrogenation of the CO group is preferred at the Pd site to form a CO*CHO* intermediate (CO*CO* + H+ + e− → CO*CHO*). It is interesting to notice that the hydrogenation at Pd site will enable C−C coupling reaction (CO*CHO* → COCHO*), which is thermodynamically favorable (exothermic). After that, the second hydrogenation of the intermediate species COCHO* leads to the formation of glyoxal (CHOCHO*) (COCHO* + H+ + e− → CHOCHO*). Then one carbonyl group (C=O) in CHOCHO* undergoes selective hydrogenation. A hydrogen atom binds to the oxygen atom, forming a hydroxyl group (−OH), which generates the intermediate of CHOHCHO* (CHOCHO* + H+ + e− → CHOHCHO*). Subsequently, the remaining carbonyl group is fully hydrogenated, leading to the formation of the diol intermediate CHOHCHOH* (CHOHCHO* + H+ + e− → CHOHCHOH*), and CHOHCH2OH* (CHOHCHOH* + H+ + e− → CHOHCH2OH* + H2O). This CHOHCH2OH* intermediate then progresses through further deoxygenation and condensation steps: CHOHCH2OH* loses an oxygen atom to form CHCH2OH* (CHOHCH2OH* + H+ + e− → CHCH2OH* + H2O). H2O will be produced and released after hydroxyl group combining with H+. CHOHCH2OH* generate CHCH2OH*(CHCH2OH* + H+ + e− → CHCH2* + H2O). Finally, structural stabilization occurs via C-H bond reorganization and hydrogenation, producing the ethylene precursor CH2CH2* (CHCH2* + H+ + e− → CH2CH2*). Subsequently, CH2CH2* continues the hydrogenation reaction to produce CH3CH2* (CH2CH2* + H+ + e− → CH3CH2*). A second H atom then binds to the adjacent carbon, fully saturating the C−C bond to yield ethane (CH3CH3*) (CH3CH2* + H+ + e− → CH3CH3*). Throughout this pathway, the synergistic interaction between Pd−Nb bimetallic sites optimizes the electronic environment, accelerating hydrogen migration and bond cleavage. These combined effects drive the highly selective formation of ethane (C2H6). The PDS is identified as the transformation of CH2CH2* to CH3CH2* (Ul = −1.06 V).
When the polarization direction is switched, the reaction of CORR on Pd−Nb@P↑-In
2Se
3 catalyst [see Fig. 3(b)] is significantly different, with a energetically minimum reaction path of CO* → CO*CO* → COCO* → COCOH* → CCO* → CHCO* → CHCHO* → CHCHOH* → CH
2CHOH* → CH
2CH
2OH* → CH
2CH
2* → CH
3CH
2* → CH
3CH
3*. Although the final product is also ethane, the PDS is CH
2CHOH* → CH
2CH
2OH* with the
Ul of −1.47 V. The significantly higher overpotential indicates that the reaction efficiency of CORR is polarization tunable via ferroelectric switching. Moreover, the C−C coupling type could also be adjusted by ferroelectric switching. On Pd−Nb@In
2Se
3 catalyst, the C−C coupling type is CO*CHO*→COCHO* for the P↓ state, it becomes CO*CO*→ COCO* for the P↑ state. Notably, the C−C coupling reaction on both P↓ and P↑ states of Pd−Nb@In
2Se
3 catalyst is exothermic reaction. Therefore, in terms of the ability to promote C−C coupling, Pd−Nb@In
2Se
3 catalyst is superior to many catalysts for C
2 products with higher C−C coupling energy barriers, such as
θ-Fe
3C(031) (0.99 eV for CH
2+CH
2) [
49], Co
2C(001) (0.52 eV for CH
2+CH
2) [
50], and
χ-Fe
5C
2 (510) (0.96 eV for CH+CH) [
51].
To gain deeper insight into the electronic origin of the polarization-dependent C−C coupling behavior, we analyzed the projected density of states and the d band centers for the Pd−Nb@In2Se3 under both P↓ and P↑ states (see Fig. S5). The results reveal that switching the polarization from P↓ to P↑ induces the d band centers of both Pd and Nb shift closer to the Fermi level. This shift signifies an enhanced catalytic activity at both metal sites, which directly rationalizes the facilitated C−C coupling. This point could be verified by the phenomenon that, the released energy of C−C coupling reaction in P↑ state is 0.27 eV more than the one in P↓ state. The d band centers shift caused by the polarization switching also change the C−C coupling style, such as CO*CHO*→COCHO* in P↓ state, and CO*CO*→ COCO* in P↑ state.
The CORR on Rh−Nb@P↓-In2Se3 catalyst follow a minimum reaction path of CO* → CO*CO* → CO*CHO* → COCHO* → CHOCHO* → CHOHCHO* → CHCHO* → CH2CHO* → CH2CHOH* → CH2CH2OH* → CH3CH2OH* [see Fig. 4(a)]. The final product is ethanol,and PDS is CH2CHO* → CH2CHOH* with the Ul of −0.98 V. In sharp contrast, CORR over Rh−Nb@P↑-In2Se3 catalysts follows a minimum reaction path [see Fig. 4(b)] of CO* → CHO* → HCOH* → CH* → CH2* → CH3* → CH4*. The final product is methane, and the PDS is CHO* → HCOH* with the Ul of −0.35 V. The different limiting potential and final products under different polarization states reveal that, the reaction kinetics and the product selectivity of the CORR can be effectively regulated by adjusting the polarization state of the substrate in ferroelectric DACs.
3.4 Opposite charge center pair promotes C−C coupling
As mentioned before, in Pd−Nb@In2Se3 and Rh−Nb@In2Se3, both Pd and Rh single atoms gain electrons from the three Se atoms connected to them, while Nb single atoms lose electrons. In contrast, in the Nb−Nb@In2Se3, both Nb atoms lose electrons to surrounding three Se atoms. Therefore, PNCCP can only be formed between the two heteronuclear dual atoms (Pd−Nb, Rh−Nb), but not homonuclear dual atoms (Nb−Nb). In order to investigate whether the PNCCP can lower the C−C coupling energy barrier, CORR on Nb−Nb@P↓-In2Se3 is studied for comparison.
As shown in Figs. 3 and 4, the C−C coupling reactions on Pd−Nb@P↓-In2Se3, Rh−Nb@P↓-In2Se3, and Pd−Nb@P↑-In2Se3 are thermodynamically favorable, with the released energies of 0.16, 0.23, and 0.43 eV, respectively. Since the final reaction product on Rh−Nb@P↑-In2Se3 tends to be C1 without C−C coupling process, it is not considered here. As displayed in Fig. S6, the C−C coupling step in Nb−Nb@P↓-In2Se3 is CO*CHO*→COCHO* with a energy barrier of 1.28 eV, which is significantly higher than that of Pd−Nb@In2Se3 and Rh−Nb@In2Se3 with opposite charge center pairs, indicating that the PNCCP favors the occurrence of C−C coupling.
As illustrated in Fig. S7, for the intermediates before C−C coupling reaction, we separately extract the single TM atoms and their attached CxHyOz groups, naming as moiety 1 and moiety 2, respectively. When two moieties possess opposite charges, the Coulomb force between them is attractive with a negative vector value. This attractive force accelerates C−C coupling. A more negative Coulomb force correlates with a lower energy barrier for C−C coupling, thereby facilitating the reaction. Thus, we employ the Coulomb force as a descriptor to quantify the promoting effect of PNCCP on C−C coupling. The Coulomb force () could be obtained through the following equation:
where k is electrostatic constant (9 × 109 N∙m2·C−2),r is distance between the gravity centers of moiety 1 and moiety 2, while q1 and q2 are the total charge of moiety 1 and moiety 2, respectively. The greater the positive and negative charges on the dual TM single atoms, the more negative the value of the Coulomb force will be. The relationship between Coulomb force and the C−C coupling energy barrier are displayed in Fig. 5. The fitted straight line states that the more negative the value of the Coulomb force is, the smaller the energy barrier for C−C coupling will be, indicating that the C−C coupling reaction will be easier to occur. It is perfectly in line with our expectations, and further substantiates that PNCCP can facilitate the C−C coupling reaction. We note that the linear correlation (R2 = 0.82) has some deviation, primarily arising from the point-charge approximation used in our Coulomb force model, which simplifies the actual diffuse charge distribution.
3.5 Selectivity for CORR vs. HER
We also explore the competing relationship between CORR and hydrogen evolution reaction (HER), which as an important side reaction can significantly inhibit the Faraday efficiency of CORR by depleting proton-electron pairs in the electrolyte solution. To check whether CORR is more favorable, the Gibbs free energy changes of the formation of CO* (ΔGCO*) and H* (ΔGH*) are firstly calculated. As shown in Fig. 6, the values of ΔGCO* on both Rh−Nb@In2Se3 and Pd−Nb@In2Se3 catalysts are smaller than the ones of ΔGH*. This indicates that the active sites of both Rh−Nb@In2Se3 and Pd−Nb@In2Se3 catalysts exhibit a higher propensity for adsorbing CO molecules over H atoms, which results in superior CORR selectivity relative to the HER.
4 Conclusion
Our DFT study demonstrates that charge-asymmetric Pd−Nb and Rh−Nb dual-atom catalysts on ferroelectric In2Se3 monolayer provide an efficient and dynamically tunable platform for electrocatalytic CO reduction toward C2 products. The key to this performance lies in the formation of the PNCCP between the heteronuclear metal sites, which generates a powerful Coulombic attraction, that significantly enables thermodynamically favorable C−C bond formation. The Coulomb force serves as a quantitative descriptor, exhibiting a clear linear relationship, where more negative forces directly correlate with lower C−C coupling energy barriers. Crucially, ferroelectric polarization switching acts as a powerful external knob to reconfigure the electronic structure. By shifting the d band centers of the metal sites, polarization reversal not only modulates the reaction energy barriers but also decisively switches the C−C coupling mechanism and the final product selectivity. Furthermore, both Pd−Nb and Rh−Nb dual-atom catalysts exhibit superior affinities for CO adsorption over H adsorption, ensuring high selectivity for CORR against the competing HER. This work predicts ferroelectric DACs as a promising class of electrocatalysts, where the synergistic combination of intrinsic charge asymmetry and external polarization control paves the way for the rational design of highly efficient and electrically tunable systems for on-demand C2 synthesis.