Mechanism and structure–activity relationship of H₂ and CO₂ activation at the ZnO/Cu catalyst interface

Abstract

Cu/ZnO/Al₂O₃ catalysts are the most well-known heterogeneous catalysts for the hydrogenation of CO and CO₂ into methanol. Herein, density functional theory calculations were performed to investigate the mechanism of H₂ activation and the effects of hydrogen spillover on CO₂ adsorption and activation at the interfacial site of the ZnO/Cu model catalyst, which was simulated by loading ZnO ribbons of different sizes on the Cu(111), Cu(100), and Cu(211) surfaces. The ZnO/Cu interface is found to facilitate the formation of H adsorbates from the dissociation of H₂ molecules, which promotes the facile formation of oxygen vacancy (V_O) sites in the ZnO component due to its reducibility and the hydrogen spillover effect. The resulting interfacial structure of the ZnO/Cu model catalyst can contain perfect, hydroxylated, and oxygen-vacancy-present ZnO sites, which may act as the adsorption and activation sites for CO₂. Further calculations show that molecular CO₂ adsorbed at the V_O site can be efficiently activated by direct dissociation or hydrogenation to the HCOO* species. In addition, the smaller ZnO structure and less exposure of the Cu(211) facet facilitate hydrogen spillover and the formation of the interfacial V_O site. This study provides important insights into the structure–activity relationship for the active sites of the ZnO/Cu model catalyst and the mechanisms of CO₂ activation and hydrogenation.

---

1 Introduction

The use of fossil energy led to excessive emission of greenhouse gases especially carbon dioxide (CO₂), and the resulting global warming effect is threatening the sustainable development of mankind.^1–4 In recent years, CO₂ capture, utilization, and storage (CCUS) technologies have received increasing attention because of their potential for mitigating the severe consequences of excessive CO₂ emission and at the same time producing necessary fuels and valuable chemicals.^5–7 CO₂ hydrogenation is an effective approach for the synthesis of methanol,^8–11 which can be directly used as a fuel or fuel additive, or indirectly used as a chemical intermediate to further produce more advanced fuels and chemicals. Cu/ZnO/Al₂O₃ is the most widely studied catalyst for CO₂ hydrogenation to methanol due to its low cost and high activity, and the industrial Cu/ZnO/Al₂O₃ catalyst has been employed for over 50 years.^12–15 However, CO₂ conversion and methanol selectivity at low temperature and pressure still have room for improvement,^16–18 and the tendency for the conventional Cu/ZnO/Al₂O₃ catalyst to deactivate due to sintering also makes it necessary to further improve its stability, by searching for better support materials,^19–21 adding suitable promoters,^22–24 and changing the synthesis method,^25,26 all of which can lead to enhanced catalyst performance. In addition, the unique role of and interaction between the Cu and ZnO components^27–29 were also examined to improve our understanding of the CO₂ hydrogenation mechanism for the rather complex Cu/ZnO/Al₂O₃ catalyst system, but our understanding on its active site and reaction mechanism remains very limited.

There remain debates on the nature of the active site of the Cu/ZnO/Al₂O₃ catalyst, although the Cu/ZnO interface or the CuZn alloy has been generally recognized as the active site for methanol synthesis.^12,30–35 Despite intensive efforts, it remains difficult for experimental methods alone to provide a thorough understanding on the active site and reaction mechanism of the Cu/ZnO/Al₂O₃ catalyst for methanol synthesis. For instance, X-ray photoelectron spectroscopic (XPS) experiments have difficulty in clarifying the valence states of the Zn species due to its limited ability to distinguish the 2p peaks of the different Zn species.^30,36,37 In addition, introducing a reactive atmosphere can lead to significant changes in the already complex structure of the Cu/ZnO/Al₂O₃ catalyst, which makes it greatly challenging to elucidate the structure, properties, and reaction mechanism of the catalyst active site.^38–41

The metal-oxide interface has unique electronic properties due to its strong interaction,^42–45 which has been proposed by experimental studies to promote the CO₂ hydrogenation reaction. Kattel et al. compared the activities of the oxide-on-metal ZnO/Cu and bimetallic ZnCu catalysts for CO₂ hydrogenation built from a single crystal model system, which illustrated the pivotal role of ZnO on the Cu surface and suggested the metal-oxide interface as the active site for methanol synthesis.³¹ Wu et al. used the higher valent metal oxide ZrO₂ to construct the ZrO₂/Cu model catalyst, which greatly enhanced the reactivity of the Cu catalyst and showed excellent catalytic performance in methanol synthesis.⁴⁶ Recently, Liu et al. further proposed the interface of the oxygen-deficient ZnO and Cu to favour methanol synthesis from CO₂ hydrogenation through the HCOO pathway,³⁴ and ZnO_1−x/Cu catalysts with abundant oxygen vacancies prepared by Zhang et al. showed excellent methanol selectivity of above 90%.⁴⁷ These studies demonstrate the rational design of reverse oxide/metal configurations as efficient CO₂ hydrogenation catalysts and the importance of the oxygen vacancy in the CO₂ hydrogenation reaction.

Density functional theory (DFT) calculations can now provide important physical insights, and thus can potentially reduce the number of experimental trials. Previous computational studies suggest that methanol can be synthesized from CO₂ hydrogenation through the HCOO pathway, or via the reverse water gas shift (RWGS) reaction followed by CO hydrogenation, known as the RWGS + CO-hydro pathway.^31,34,42,48 Previous theoretical studies on the Cu/ZnO/Al₂O₃ catalyst for CO₂ hydrogenation to methanol mainly focused on the nature of the active site, as well as the morphology and metal coordination of the catalyst.^{12,34,48–53} Catalyst models, including Cu-supported ZnO and ZnO-supported Cu, have been employed to simulate the ZnO/Cu interface, and the size and shape of the supported Cu cluster and ZnO cluster or nanoribbons were examined by several studies.^{31,34,45,47,49,51–55} Some recent theoretical investigations examined the mechanism of CO₂ hydrogenation at the ZnO/Cu interface, which is generally modelled by placing ZnO clusters or nanoribbons on the Cu surface. Kattel et al.³¹ and Liu et al.³⁴ constructed such ZnO/Cu catalyst models using different ZnO clusters or nanoribbons, and calculated the reaction pathway of CO₂ hydrogenation to methanol through the above two reaction pathways at the perfect or oxygen-deficient ZnO/Cu interface, but only focused on the interfacial site between the ZnO and the Cu(111) surface. It is worth noting that in these reverse oxide-on-metal catalyst models, the metal is modified by the reducible oxide, and there are metal, redox, and Brønsted and Lewis acid sites,⁵⁶ so the reaction mechanism can be more complex than that on the pure metal or pure oxide surface. To this end, DFT calculations can greatly improve our understanding on the properties of the active site of the Cu/ZnO/Al₂O₃ catalyst as well as the reaction mechanism for CO₂ hydrogenation especially when combined with results from various experimental characterization techniques.

In this work, extensive DFT calculations were performed to reveal the mechanisms of H₂ and CO₂ activation at the interface of the ZnO/Cu model catalyst during CO₂ hydrogenation. Our results show that the ZnO/Cu interface plays a vital role in H₂ dissociation, which provides H adsorbates for CO₂ hydrogenation. Hydrogen spillover from the Cu surface to the terminal O of the ZnO ribbons results in Brønsted acid and redox sites at the ZnO/Cu interface. The interfacial V_O site formed upon hydrogen spillover from Cu to ZnO facilitates both the direct dissociation of CO₂ and the formation of the HCOO intermediate by CO₂ hydrogenation. The size of the ZnO ribbon and the exposed Cu crystal plane have important effects on H₂ activation, which is relevant to the formation of interfacial V_O sites. Our work thus provides significant physical insights on the mechanisms of H₂ and CO₂ activation at the ZnO/Cu interface, which are crucial towards the rational design of more active Cu-based catalysts for CO₂ hydrogenation to methanol.

2 Computational methods

All periodic DFT calculations were carried out with the Vienna ab initio simulation package (VASP)^57,58 using the Perdew–Burke–Ernzerhof (PBE)⁵⁹ exchange–correlation functional and the projector-augmented wave (PAW) potentials.^60,61 An energy cutoff of 400 eV and a Gaussian smearing width of 0.05 eV were used. The electronic energy in the self-consistent field (SCF) iteration was set to 10⁻⁴ eV, and the forces on all unconstrained atoms were converged to 0.03 eV Å⁻¹. A vacuum space of 15 Å was used to separate adjacent slabs along the Z direction.

Recently, inverse ZnO/Cu catalyst models with ZnO clusters or nanoribbons supported on the Cu slab surface were constructed to simulate the ZnO/Cu surface.^31,34,49,55 Considering the ZnO overlayer observed in experiments of Cu/ZnO/Al₂O₃ catalysts,^34,62–64 catalyst models with ZnO ribbons are used in this work. The interface between Cu and ZnO was modelled by placing a layer of (3 × 3) ZnO(0001) ribbon on top of the three-layer (4 × 8) Cu(111) slab, the three-layer (3 × 5) Cu(100) slab, or the five-layer (3 × 4) Cu(211) slab. For comparison, interface models were also built by loading the smaller one-layer (2 × 3) ZnO(0001) ribbon on the above Cu surfaces. We use W^x_y(z) to represent the six models, where x, y, and z refer to the number of ZnO layers, the number of columns of ZnO ribbons, and the Miller index of the Cu surface, respectively. For instance, Fig. 1(a) shows the catalyst model denoted by W¹₃(111), where one layer of the three columns of ZnO ribbons is supported by the Cu(111) surface. Our calculations show that catalyst models with three Cu layers can give results as accurate as with five Cu layers including adsorption energies, reaction energies, and energy barriers as shown in Fig. 4, 5 and S1.† For the ZnO/Cu catalyst models containing ZnO ribbons employed in this work, the mismatch along the extended direction of the ZnO ribbons may result in some strain between these two components, and the misfit can be calculated using the following equation:

where a_{Cu slab} represents the lattice parameters of (4 × 8) Cu(111), (3 × 5) Cu(100), (3 × 4) Cu(211) slabs along the direction of the extended ZnO ribbons, and a_{ZnO ribbon} represents the lattice parameter of ZnO(0001) along the interface region. We find that the misfits between the ZnO ribbon and Cu slab in the W¹₃(111), W¹₂(111), W¹₃(100), W¹₂(100), W¹₃(211), and W¹₂(211) catalyst models are 5.25%, 5.25%, 10.71%, 10.71%, 5.25%, and 5.25%, respectively, so only the misfits of the ZnO ribbon and Cu(100) slab are relatively large, and the resulting catalyst models may be less stable compared to the others. Thus, we focus our discussion on the computational results for the reasonably stable W¹₃(111), W¹₂(111), W¹₃(211), and W¹₂(211) catalyst models.

Fig. 1 Top and side views of (a) perfect and (b) defective W¹₃(111) surfaces; comparison of (c) the calculated energy barriers (E_a/eV); and (d) BEP relationship for H₂ dissociation at the ZnO/Cu interface and the Cu surface.

For W¹₃(111) and W¹₂(111), the Brillouin zone was sampled with a Monkhorst–Pack k-point grid mesh of (3 × 1 × 1), whereas for W¹₃(100) and W¹₂(100), it was sampled with that of (2 × 1 × 1); in these calculations, the bottom two atomic layers were fixed, whereas the top atomic layer and the adsorbed species were relaxed. For W¹₃(211) and W¹₂(211), the Brillouin zone was sampled with a Monkhorst–Pack k-point grid mesh of (1 × 2 × 1), and the bottom three atomic layers were fixed, whereas the top two atomic layers and the adsorbed species were relaxed. The k-points for the different ZnO/Cu catalyst models were generated using the VASPKIT program with the recommended KPT-resolved value of 0.04 2 × π/Å.⁶⁵

The Hubbard U correction for the Zn atoms was included in our DFT calculations to treat the on-site Coulomb repulsion of their 3d electrons using an effective U value of 4.7 eV (U_eff = U − J) based on the previous work of Liu et al.³⁴ The climbing-image nudged elastic band (CI-NEB) method^66,67 and the improved dimer method (IDM)⁶⁸ were employed to locate the transition states, which were further verified by vibrational analysis showing one and only one imaginary mode. All structures were built and visualized using Materials Visualizer from Materials Studio.⁶⁹

The formation energy of a V_O site (ΔE_f,VO,O2) with respect to gas phase O₂ is defined as the reaction energy of the thermal desorption of molecular O₂:

In addition, the formation energy of a V_O site (ΔE_f,VO,H2/H2O) with respect to gas phase H₂/H₂O is defined as:
where E_VO-surface, E_perfect, E_O2, E_H2 and E_H2O denote the total energies of the defective surface, perfect surface, and gas phase O₂, H₂ and H₂O. The adsorption energy of an adsorbate A on the ZnO/Cu slab surface S is defined as:

E_ad,A = E_total − (E_slab + E_A)

where E_total, E_slab, and E_A are the total energies of the slab with the adsorbate, the clean slab, and the adsorbate as a free molecule, respectively.

3 Results and discussion

3.1 H₂ dissociation at the ZnO/Cu interface and the Cu surface

We first compare the dissociative adsorption of H₂ at the ZnO/Cu interface and the traditional Cu surface sites.⁷⁰ As shown in Fig. 1 and 2, unlike the homolytic dissociation of H₂ on the Cu surface (TS-H_Cu), the heterolytic dissociation of H₂ at the perfect ZnO/Cu(*_P) interface (TS-H_int) leads to two H adsorbates at the O and Cu sites, where one H atom (H__O*) is adsorbed at the terminal O site of ZnO resulting in a Brønsted acid site, and the other H atom (H__Cu*) is adsorbed on the Cu surface. For the perfect W¹₃(111) surface, the O–H_A bond formed after H₂ dissociation has a length of 0.98 Å, while the other H atom is adsorbed at the three-fold hollow site surrounded by three Cu atoms. H₂ dissociation at this interfacial site has a lower energy barrier (E_a) of 0.26 eV than that on the Cu surface of 0.47 eV. In its transition state, the distance between H_A and H_B is 0.91 Å, similar to that on the Cu surface of 0.92 Å. H₂ dissociation on the Cu(100) and Cu(110) surfaces was also investigated by Higham et al.,⁷¹ whose energy barriers were calculated to be 0.54 and 0.42 eV; these values are in good agreement with our results for the homolytic dissociation of H₂ on the Cu surfaces. Besides, our calculated energy barrier for the heterolytic dissociation of H₂ at the ZnO/Cu interface is comparable to 0.17 eV for the homolytic dissociation of H₂ at the Zn sites over the perfect Zn-terminated ZnO(0001) surface.⁷² The energy barriers of H₂ dissociation on all studied ZnO/Cu catalyst models are compared in Fig. 1(c), and those at the interface are generally much lower than those on the Cu surfaces. In addition, H₂ dissociation at the interface is strongly exothermic, indicating that it is favoured both thermodynamically and kinetically. Further details in the calculated energy barriers and reaction energies (E_r) are given in Table S1.†

Fig. 2 Potential energy profiles for H₂ dissociation at the interface and the subsequent hydrogen spillover process with their energy barriers (E_a/eV) shown. The images display structures of the intermediates and transition states on the perfect W¹₃(111) surface. Brown: Cu, grayish blue: Zn, red: O, white: H.

Fig. 1(d) further shows the Bell–Evans–Polanyi (BEP)^73–75 relationship between the energy barrier and the reaction energy for H₂ dissociation at the perfect ZnO/Cu interface and on the Cu surface. The BEP relationship for H₂ dissociation on the Cu surface shows a reasonably good linear fit with a coefficient of determination (COD) of R² = 0.89, which makes it possible to estimate the energy barrier of H₂ dissociation from its reaction energy on the Cu surface. However, the BEP relationship for H₂ dissociation at the perfect ZnO/Cu interface is more complex than linear, as indicated by the low COD of R² = 0.46 when a linear fit is attempted even by including only four catalyst models and excluding the perfect W¹₂(211) interface, which is marked by the blue dashed circle and is clearly an outlier. Thus, although a roughly linear trend appears between the energy barrier and the reaction energy of H₂ dissociation at the interfacial sites of ZnO and different Cu crystal planes, the linearity is not nearly as good as that on the Cu surface.

3.2 Hydrogen spillover mechanism

To study the mechanism and effect of hydrogen spillover on oxygen vacancy formation, we then calculated the formation energy of oxygen vacancies (ΔE_f,VO,O2) with respect to gas phase O₂ at the interfaces of the different ZnO/Cu catalyst models. For the perfect W¹₃(111) interface model shown in Fig. 1(a), we find that the O sites from the rightmost ZnO ribbon are more likely to be hydrogenated to form oxygen vacancies due to their lower oxygen vacancy formation energies (with an average oxygen vacancy formation energy of 2.94 eV as listed in Table S2†) than those from the other two ZnO ribbons. For the other five catalyst models, the oxygen vacancy formation energies of the O sites from the rightmost ZnO ribbon are generally also lower, indicating the more likely location of oxygen vacancy formation at the ZnO/Cu interface. Thus, in our subsequent calculations, the defect site at the ZnO/Cu interface is chosen to be located at the O site from the rightmost ZnO ribbon.

Fig. 2 and Tables S1 and S3† show H₂ dissociation at the ZnO/Cu interface and on the Cu surface, leading to two hydrogen spillover mechanisms and oxygen vacancy formation at the interface. After H₂ dissociation at the interface, the H atom adsorbed on the Cu surface is further transferred to the Brønsted acid site (OH site) (TS-H_sp2), which is itself formed from H₂ dissociation. The energy barrier of this step is between 0.79 and 0.94 eV, and an adsorbed H₂O molecule is formed after the hydrogen spillover. Upon H₂O desorption, an oxygen vacancy site is formed leading to the defective ZnO/Cu surfaces (*_D).

In contrast, after H₂ dissociation on the Cu surface, an additional hydrogen spillover step is required to first form the aforementioned Brønsted acid site followed by the further formation of the oxygen vacancy site by another hydrogen spillover step. Over the perfect ZnO/Cu surfaces, after H₂ dissociation on the Cu surface, both H atoms are adsorbed on the hollow Cu sites. For the perfect W¹₃(111) surface, the first H adsorbate diffuses from the Cu surface to the terminal O of ZnO to form a Brønsted acid site (TS-H_sp1) with an energy barrier of 0.74 eV and an exothermicity of −0.54 eV. Then, the second H atom also spills over from the Cu surface to the above Brønsted acid site to form an adsorbed H₂O molecule (TS-H_sp2), desorption of which results in an oxygen vacancy and the defective surface as shown in Fig. 1(b).

As discussed in the previous section, H₂ tends to dissociate at the ZnO/Cu interface. Nevertheless, the above-mentioned two hydrogen spillover processes are both likely to occur, so the ZnO/Cu interface can promote H₂ dissociation to form a Brønsted acid site, as well as the formation of a Brønsted acid site and an oxygen vacancy site after hydrogen spillover. As shown in Table S2,† the formation energies of an oxygen vacancy (ΔE_f,VO,H2/H2O) with respect to gas phase H₂/H₂O for the interfacial O atoms in ZnO are mostly negative, ranging from −0.28 to −0.04 eV, except for that in W¹₃(211) of 0.18 eV, indicating that the formation of the interfacial V_O sites in these catalyst models are usually thermodynamically favorable. Under the actual reaction conditions, H₂O molecules, which are also products of CO₂ hydrogenation, are also present in the reaction atmosphere, so we also consider the possible dissociation of H₂O at the ZnO component to form Brønsted acid sites. As shown in Fig. S2 and Table S3,† H₂O dissociation at the ZnO component involves a rather low energy barrier of <0.30 eV, leading to the formation of two hydroxyl groups (Zn–OH). This indicates that H₂O produced during CO₂ hydrogenation may also promote the formation of Brønsted acid sites, and as the reaction proceeds, the ZnO component will likely undergo continuous hydrogenation and redox reactions, leading to dynamic structural changes.

3.3 CO₂ adsorption at the ZnO/Cu interface

CO₂ can potentially adsorb at the metal site or the oxygen vacancy site on the ZnO/Cu model catalyst. For the perfect and defective W¹₃(111) surfaces, CO₂ can adsorb at the interfacial O site of ZnO on the perfect ZnO/Cu surface to form the carbonate configuration as shown in Fig. 3(a), or physisorb in a linear configuration as shown in Fig. 3(b), or chemisorb at the interfacial oxygen vacancy site on the defective W¹₃(111) surface to form the bent configuration as shown in Fig. 3(c). The different CO₂ adsorption structures on the different ZnO/Cu catalyst models are shown in Fig. S3 and S4.†

Fig. 3 Adsorption structures and adsorption energies (E_ads/eV) of (a) , (b) on the perfect W¹₃(111) surface, (c) on the defective W¹₃(111) surface, and (d) in the presence of the Na promoter. Brown: Cu, greyish blue: Zn, red: O, dark grey: C, green: O in CO₂ molecules, purple: Na.

For the perfect W¹₃(111) surface without an oxygen vacancy site at the interface, CO₂ can combine with a terminal O atom of ZnO to form the , where the C–O bond lengths in the adsorbed CO₂ are elongated to 1.27 and 1.29 Å with the formation of a new C–O bond of 1.37 Å. During the adsorption, the original Zn–O bond is broken, and this Zn atom binds one O atom in the adsorbed CO₂, while the other O atom in CO₂ binds the surface Cu atom. CO₂ can also physisorb in the at the perfect ZnO/Cu interface, which undergoes little changes from the free CO₂ molecule. For the defective W¹₃(111) surface with an oxygen vacancy site at the interface, CO₂ can chemisorb in the , which undergoes significant deformation compared to the free CO₂ molecule, resulting in two C–O bonds of 1.26 and 1.33 Å and an O–C–O bond angle of 120.5°. In the , the oxygen vacancy is essentially occupied by one O atom in CO₂, while the other O and the C atoms bind different surface Cu atoms.

As shown in Tables 1 and S4,† after CO₂ adsorption in the on the perfect W¹₃(111) surface, the Bader charges of the Cu and Zn atoms bound to the adsorbate increase by 0.17 and 0.11 |e|, respectively, while the Bader charges of the O atoms in the adsorbate decrease by 0.09 |e|. Furthermore, after CO₂ adsorption in the on the defective W¹₃(111) surface, the Bader charges of the two Cu atoms around the adsorbate increase by 0.22 and 0.13 |e|, whereas the Bader charges of the two Zn atoms around the adsorbate also increase by 0.29 and 0.23 |e|.

Table 1 Calculated Bader charges and corresponding changes for the Cu, Zn and O atoms around the adsorbate before and after CO₂ adsorption on the perfect surface (*_P) and defective surface (*_D). Q2 and Q4 represent the positive charges of Cu, Zn atoms and the negative charges of O atoms around the adsorbed CO₂ molecule, Q1 and Q3 represent the corresponding charges of these atoms before CO₂ adsorption

W^13(111) surface	Q/|e|	Cu	Zn	O
*_P	Q1 (*_P)	0.00	1.08	−1.08
	Q2	0.17	1.19	−1.17
	ΔQ1 (= Q2 − Q1)	0.17	0.11	−0.09
*_D	Q3 (*_D)	−0.03/−0.05	0.83/0.80	—
	Q4	0.19/0.08	1.12/1.03	—
	ΔQ2 (= Q4 − Q3)	0.22/0.13	0.29/0.23	—

---

In addition, as shown in Table S5,† the negative charge on ZnO over the defective surface is significantly lower than that over the perfect surface, so the presence of the oxygen vacancy sites greatly reduces the number of electrons transferred from Cu to ZnO. We note that an opposite direction of charge transfer was predicted by Heenemann et al. for the ZnO-supported Cu nanoparticle models.⁷⁶ When CO₂ is adsorbed on both the perfect and defective surfaces, CO₂ acquires some electrons, and that in the obtains about one electron, which is significantly more than that in the of 0.17 electrons on average. For example, after CO₂ adsorption on the defective W¹₃(111) surface, the CO₂ adsorbate obtains 0.99 |e|, and the Bader charges of the other adsorbed CO₂ are shown in Table S5.† This suggests that the presence of the oxygen vacancy site can significantly promote electron transfer from the defective surface to the CO₂ adsorbate. Thus, the may be more likely hydrogenated to the HCOO intermediate, because it carries more negative charge and is partially reduced upon the formation of the oxygen vacancy.^77,78

It is worth noting that although the obtains more electrons than the , the adsorption energy of the is significantly less negative than that of the . As shown in Table S6,† the adsorption energy of the on the perfect W¹₃(111) surface is −0.29 eV, which is much more negative than that of on the defective W¹₃(111) surface of 0.31 eV. Reichenbach et al. also reported a similar observation that adsorption of the activated CO₂ molecule on the ZnO cluster models with different oxidation states for Zn is relatively unstable with CO₂ adsorption energies of 0.28–0.66 eV.⁴⁹ We conjecture that introducing the Na₂O promoter may further enhance the adsorption of the , and as shown in Fig. 3(d), our calculations show that adding Na₂O can indeed change its adsorption energy from 0.31 to −0.04 eV, thus leading to stronger adsorption. Besides, although the carbonate adsorption structures are much stable at the ZnO/Cu interface compared to the , recent computational works suggest that the carbonate species, which is easy to form and stable, will block the active sites of the ZnO surface;^72,79 the role of the carbonate will be further discussed in the next section.

In summary, the CO₂ adsorbates in both the and the appear to be activated considering the decrease in the C–O–C angle and the increase of the C–O bond length, compared with that in the . However, the absence or presence of the oxygen vacancy at the ZnO/Cu interface shows a distinct effect on the stability of the and the , in which the activated is more stable than the activated at the ZnO/Cu interface. Although the is thermodynamically more stable and the has an endothermic adsorption energy, the oxygen vacancy site promotes the charge transfer in the latter leading to the accumulation of more electrons on the CO₂ adsorbate, which may facilitate its further conversion.

3.4 HCOO formation at the ZnO/Cu interface

It is generally believed that CO₂ can be hydrogenated to methanol via the formate (HCOO) pathway,^77,80–83 where H can either come from direct H₂ dissociation or from the Brønsted acid site resulting from hydrogen spillover as discussed in the earlier section. We thus examined CO₂ adsorption at the ZnO/Cu interface, and its further hydrogenation to the HCOO intermediate.

As shown in Table S7,† in the absence of an oxygen vacancy site at the ZnO/Cu interface, hydrogenation of the to the HCOO intermediate needs to overcome a very high energy barrier of >2.00 eV, and the energy barrier for the hydrogenation of the by H* from the Brønsted acid site to the HCOO intermediate remains substantial at 1.42–1.62 eV, whereas the energy barrier of the hydrogenation of the by H* adsorbed on the Zn site is the lowest at only 0.24–0.34 eV. At the perfect W¹₃(111) interface, the detailed reaction pathway for the hydrogenation of the and adsorbates to form the HCOO* is shown in Fig. 4(a). The can react with the H* adsorbed at the hollow Cu site with a very high energy barrier of 2.02 eV and an exothermicity of −0.53 eV, where the dissociated O atom binds the Zn site, confirming the inactivity of the carbonate at the ZnO/Cu interface for methanol synthesis, similar to that on the ZnO surface.^72,79 The can react with the H* adsorbed at the O site with a substantial energy barrier of 1.62 eV and an endothermicity of 0.26 eV. The same reaction step of CO₂ hydrogenation to HCOO* through H atoms in the Brønsted acid site was calculated by Liu et al. to have a lower energy barrier of 1.15 eV,³⁴ which is likely due to the adsorption of the reactant CO₂ molecule at the Cu sites near the interface, different from the physisorbed CO₂ in our work. Furthermore, the can also react with the H* adsorbed at the Zn site to form the HCOO* with a very low energy barrier of 0.26 eV and an exothermicity of −1.05 eV. The significant difference in the reaction barrier and reaction energy for the hydrogenation of the by the H__O* and H__Zn* can be attributed to the different affinity of the H* adsorbate to the O and Zn sites. This can be inferred from the rather close effective energy barriers (E_a,eff) for the hydrogenation of the to the HCOO* in these two pathways, which are 0.94 and 0.88 eV involving the H__O* and H__Zn*, respectively.

Fig. 4 Potential energy profiles for CO₂ hydrogenation to the HCOO* intermediate at the interface of the (a) perfect and (b) defective W¹₃(111) surfaces. Structures of the transition states on the perfect and defective W¹₃(111) surfaces are shown.

In the presence of an oxygen vacancy site at the interface, hydrogenation of the to form the HCOO* only needs to overcome a very low energy barrier of 0.18–0.30 eV, indicating that CO₂ can be easily converted into the HCOO* intermediate at the oxygen vacancy site. The detailed energy barriers and reaction energies are listed in Table S7,† and for the defective W¹₃(111) surface, the detailed reaction pathway for the hydrogenation of the adsorbed at the interface to form the HCOO* is shown in Fig. 4(b). The can react with the H* adsorbed at the hollow Cu site to form the HCOO* with a very low energy barrier of only 0.18 eV and an exothermicity of −0.84 eV, where the C–H distance shrinks from 2.54 Å in the initial state to 1.66 Å in the transition state, and the resulting C–O bond lengths in the HCOO* are 1.25 and 1.29 Å. Over other catalyst models such as Cu supported ZnO clusters, similar energy barriers of no more than 0.30 eV were predicted for hydrogenation to HCOO* of the bent CO₂ adsorbates with one O atom adsorbed at the Zn site,^31,51 comparable with that on the ZnO surface of 0.11 eV,^72,79 while hydrogenation to HCOO* of the linear CO₂ adsorbates with one O atom adsorbed at the Zn sites needs to overcome a slightly higher energy barrier⁴⁹ of about 0.60 eV. Thus, these catalytically active sites all exhibit higher activity for HCOO* formation than the Cu surface.^71,81

3.5 CO₂ dissociation at the ZnO/Cu interface

Methanol may also form from hydrogenation of the CO intermediate, which is produced from the reverse water gas shift (RWGS) reaction. CO can form either by the direct dissociation of the adsorbed to CO* and O*, or by the indirect dissociation pathway via the carboxylic acid (COOH*) intermediate, which can further dissociate into CO* and OH*. We examined the direct dissociation and the COOH* pathways from the different CO₂ adsorption structures, and the detailed energy barriers and reaction energies are listed in Tables S8 and S9.† In the absence of an oxygen vacancy on the surface, the can be directly hydrogenated to the COOH*, while the must be first hydrogenated into the bicarbonate (CO₃H*) intermediate prior to its further conversion to the COOH*. On the other hand, in the presence of an oxygen vacancy on the surface, the can be converted into CO by both the direct and indirect pathways.

For the perfect W¹₃(111) surface, the detailed reaction pathways for the hydrogenation of the and to form the COOH* are shown in Fig. 5(a). The O in the can be first hydrogenated with the H* adsorbed on the Cu site to form the CO₃H* species with a relatively low energy barrier of 0.54 eV and an exothermicity of −0.27 eV. However, breaking the C–O bond in the CO₃H* to form the COOH* must overcome a very high energy barrier of 1.82 eV, indicating that converting the to CO* by hydrogenation is difficult. Similarly, our calculations show that it also seems difficult for hydrogenation of the by an H atom adsorbed at the Zn site to form the COOH* to occur due to the rather high energy barrier of 1.42 eV. Therefore, in the absence of an oxygen vacancy on the surface, the adsorbed CO₂ molecule can only be converted into CO by the indirect pathway, which is expected to be slow due to the high energy barrier, and similar conclusions can be drawn for the indirect dissociation of CO₂ at the ZnO/Cu interface for the other studied catalyst models. This suggests that the perfect ZnO/Cu interface is rather inactive for the conversion of CO₂ into either the HCOO or COOH intermediate, so the absence of oxygen vacancy sites may lead to low CO₂ activity.

Fig. 5 Potential energy profiles for CO₂ direct and indirect dissociations at the interface of the (a) perfect and (b) defective W¹₃(111) surfaces. Structures of the transition states on the perfect and defective W¹₃(111) surfaces are shown.

In the presence of an oxygen vacancy at the interface, our calculations using the different catalyst models show that CO₂ dissociation to form CO easily occurs. For the defective W¹₃(111) surface, Fig. 5(b) shows the detailed potential energy diagrams for the direct dissociation of the adsorbed at the interface and for its hydrogenation to form the COOH* intermediate. As indicated by the red line in this figure, the can directly dissociate at the oxygen vacancy site to yield CO* and O* occupying the oxygen vacancy site with a very low energy barrier of 0.25 eV and an exothermicity of −0.23 eV. In this reaction, one O atom in the binds two Zn atoms in the ZnO ribbon, and the C–O distance in the transition state is 1.74 Å. The resulting CO molecule is adsorbed at the Cu site with C–Cu and C–O bond lengths of 1.85 and 1.16 Å, respectively. There is a lack of theoretical investigation on direct dissociation of CO₂, especially on the inverse ZnO/Cu catalyst models. Our results show that the V_O sites at the interface favor this process with an energy barrier of 0.25 eV, which is much lower than those of about 0.60–0.79 eV on the CuZn alloy.^31,48 In addition, the can also be hydrogenated by the H* adsorbed at the Zn site to form the COOH* with an exothermicity of −0.39 eV, and its energy barrier of 0.63 eV is significantly higher than that to the HCOO* of 0.18 eV likely due to the oxophilicity of the Zn site. Thus, based on the above analysis, oxygen vacancy formation is not only conducive to CO₂ hydrogenation to form the HCOO* intermediate, but can also promote the direct dissociation of CO₂ to form CO, which is crucial for the subsequent CO hydrogenation reaction pathway. Furthermore, our calculations show that the presence of oxygen vacancies in ZnO is mostly to increase CO₂ reactivity, whereas its effect on the product selectivity may be rather limited.

3.6 Effects of the ZnO size and Cu crystal plane on H₂ activation

In our simulations, we considered catalyst models with different sizes of ZnO ribbons loaded on the Cu surface, including ZnO ribbons of three and two columns with all Zn atoms exposed. For ZnO ribbons of different sizes loaded on the Cu(111) surface, namely, the W¹₃(111) and W¹₂(111) surfaces, we further compared the energy barriers of H₂ dissociation, hydrogen spillover, CO₂ hydrogenation, and dissociation. As shown in Fig. 6(a), ZnO size is shown to mainly influence H₂ dissociation and hydrogen spillover during CO₂ hydrogenation. When decreasing the ZnO size, the energy barrier of H₂ dissociation at the ZnO/Cu interface slightly decreases from 0.26 to 0.18 eV, but that on the Cu surface slightly increases, indicating that a better exposure of the Zn atoms may benefit H₂ dissociation at the ZnO/Cu interface. In addition, for oxygen vacancy formation, the energy barrier of the first hydrogen spillover step decreases from 0.74 to 0.50 eV, while that of the second hydrogen spillover step to form an H₂O molecule also slightly decreases. Thus, reducing the ZnO size is also conducive to the occurrence of hydrogen spillover.

Fig. 6 (a) Comparison of the energy barriers of H₂ dissociation at the interface, Cu surface, and two hydrogen spillover processes on the W¹₃(111) and W¹₂(111) surfaces. (b) Comparison of energy barriers of the two hydrogen spillover processes at the interface of ZnO and different Cu crystal planes.

Using model catalysts with well-defined structures, such as specific metal surfaces,^12,51,84,85 may simplify the mechanistic investigation of complex catalysts. We thus also inspected the influence of the Cu crystal plane at the ZnO/Cu interface on the CO₂ hydrogenation reaction. As shown in Fig. 6(b), we find that at the interface between the Cu(211) facet and ZnO, the energy barrier of the first hydrogen spillover step is significantly higher, while that of the second hydrogen spillover step is slightly lower than those for the other studied Cu crystal facets. This indicates that to some extent, interfaces formed on the Cu(211) facet is less favourable for spillover of the H adsorbates formed by H₂ dissociation on the Cu surface, leading to less efficient formation of the Brønsted acid site (OH site).

3.7 Discussion

At present, our theoretical understanding on the ZnO/Cu interface for CO₂ hydrogenation to methanol remains lacking, and based on previous experimental and theoretical studies, we constructed a series of catalyst models with ZnO ribbons of different sizes loaded on different Cu crystal planes and examined the influence of the ZnO/Cu interfacial structure on the CO₂ hydrogenation reaction. Previously, Kattel et al. compared the activities of the ZnCu and ZnO/Cu model catalysts for methanol synthesis from CO₂ hydrogenation, and their DFT calculations on the ZnCu(211) and ZnO/Cu(111) surfaces identified the formate pathway and the RWGS + CO-hydro pathway for methanol synthesis from CO₂.³¹ In this work, besides the Cu(111) facet, we also examined interfaces formed between ZnO ribbons and the slightly less stable Cu(211) and Cu(100) facets.⁸⁶ In addition, their studies suggest that O* formed by dissociation accumulates at the Zn sites on the ZnCu(211) surface, and hydrogenation of the resulting ZnO is hindered due to the relatively high energy barrier of 1.21 eV. Our calculations show that hydrogenation of the terminal O at the ZnO/Cu(211) interface to form the Zn-OH species also incurs a relatively high energy barrier of 1.08 eV, which is significantly higher than our other studied catalyst models. This suggests that the Cu(211) facet is less favourable for spillover of the H adsorbate at the ZnO/Cu interface. Furthermore, our results show more favourable CO₂ hydrogenation to the HCOO* than that to the COOH*, consistent with the theoretical studies of Kattel et al.³¹ and Xiong et al.⁵¹ using ZnO cluster models of the ZnO/Cu(111), ZnO/Cu(110), and ZnO/Cu(100) surfaces.

Recent experimental studies showed that the reverse ZnO/Cu model catalyst prepared by atomic layer deposition exhibits excellent performance for CO₂ hydrogenation to methanol, which is much higher than that using the Cu catalyst itself.³⁴In situ X-ray absorption spectroscopy (XAS) further showed that the Zn species in the highly active catalyst are present as oxygen-deficient ZnO aggregates. To elucidate the promotional effect of the Zn species on the Cu-based catalysts, theoretical calculations were performed to show that the oxygen-deficient ZnO_1−x/Cu(111) interface has high activity for methanol formation from CO₂ hydrogenation through the formate pathway, while indirect CO₂ dissociation incurs a relatively high energy barrier. In addition to the above-mentioned indirect CO₂ dissociation pathway, we also considered direct CO₂ dissociation at the defective ZnO/Cu interface. Our calculations show that the direct dissociation of the has the same low energy barrier as its hydrogenation to form the HCOO* intermediate. Thus, the defective ZnO/Cu interface can be expected to have rather similar reactivity to the ZnCu(211) surface for the competition between the direct dissociation and hydrogenation of CO₂.^31,48 HCOO*, the key reaction intermediate for methanol synthesis from CO₂ hydrogenation, can also be easily formed at the ZnO/Cu interface. With the accumulation of CO in the gas phase, it may undergo further hydrogenation to give methanol. However, previous isotope-labelling studies show that methanol is formed from the hydrogenation of CO₂ rather than CO,^82,87–89 whereas CO is regarded as a mere reducing agent. As Shi et al. reported,⁹⁰ under different H₂, CO, CO₂ pressure conditions, structures of the ZnO/Cu catalyst undergo dynamic changes, and CO can even reduce the Zn species. Furthermore, the HCO* species, which is involved in the CO hydrogenation route, are demonstrated as an unstable adsorbate, as the hydrogenation of HCO* usually incurs a higher energy barrier than the reverse reaction of CO* hydrogenation to HCO*. Many of the aforementioned studies reached the conclusion that the HCO* species should quickly dissociate to CO* and H* using quite different ZnO/Cu catalyst models, including the ZnO/Cu interface, ZnCu alloy, and pure Cu surfaces.^{12,48,49,80,81,90–92} Besides, Liu et al. reported that CO₂ hydrogenation to methanol at the stoichiometric ZnO/Cu(111) interface through the H adatom strongly adsorbed at the edge oxygen site encounters rather high energy barriers. Although we find hydrogenation of the to the HCOO* by the weakly adsorbed H__Zn* to incur a very low energy barrier, the effective energy barrier for CO₂ hydrogenation by the H__Zn* is still quite high. Furthermore, our study shows that CO₂ can also adsorb on the perfect ZnO/Cu surface to form a carbonate species, although hydrogenation of this carbonate configuration is rather difficult, suggesting that the carbonate species is likely just a spectator.

Conclusions

In this study, we performed a systematic study on the effect of the interfacial structure of the ZnO/Cu catalysts for methanol synthesis from CO₂ hydrogenation using the reverse oxide-metal catalyst models with different sizes of ZnO ribbons loaded on different Cu crystal planes. The interfacial oxygen vacancy (V_O) sites on the ZnO/Cu model catalysts, formed by H₂ dissociation and subsequent hydrogen spillover at the ZnO/Cu interface, were found to be the effective active site for CO₂ activation and hydrogenation. Our main observations are the following:

(1) The formation of interfacial V_O sites from H₂ dissociation and hydrogen spillover is investigated. Our results show that H₂ molecules tend to dissociate at the ZnO/Cu interface with lower energy barriers instead of the Cu surface, and hydrogen spillover from Cu surfaces to the terminal O atoms of the ZnO occurs with moderate energy barriers to form V_O sites. These processes are kinetically favorable on most of the catalyst models with different sizes of ZnO ribbons and Cu crystal planes, consistent with the experimental observation of the promotion of methanol production by oxygen-deficient ZnO at the interface.

(2) Studies on the further conversion of the CO₂ adsorbates at the ZnO/Cu interface, including the hydrogenation, indirect, and direct dissociation, indicate that the interfacial V_O site accelerates the hydrogenation of the through the HCOO* pathway as well as the direct dissociation of CO₂, whereas the carbonate adsorption structure may act as a spectator. Thus, the ZnO/Cu interface containing V_O sites promotes the formation of the HCOO* intermediates and CO, similar to the CuZn alloy in previous studies.

(3) Catalyst models with smaller ZnO ribbons are more conducive to the formation of the interfacial V_O active site due to the lower reaction barrier of interfacial H₂ dissociation and hydrogen spillover from the Cu surface to the ZnO ribbon. However, the presence of the Cu(211) facet disfavours the first hydrogen spillover step during the formation of the Brønsted acid site, which may result in a lower coverage of the interfacial V_O site.

Our work reveals the mechanism of oxygen vacancy formation at the ZnO/Cu interface, and demonstrates the promotion of CO₂ activation and reaction at the ZnO/Cu interface, thus providing significant insights into the crucial role of ZnO/Cu interface sites in the CO₂ hydrogenation reaction.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (22293023, 22172188, 22172189, 22293025), CAS Youth Interdisciplinary Team, Program of Shanghai Academic Research Leader (22XD1424100), and Science and Technology Commission of Shanghai Municipality (23YF1453400, 23ZR1481700).

References

1. C. F. Shih, T. Zhang, J. Li and C. Bai, Joule, 2018, 2, 1925–1949.
2. D. A. Lashof and D. R. Ahuja, Nature, 1990, 344, 529–531.
3. P. Friedlingstein, R. M. Andrew, J. Rogelj, G. P. Peters, J. G. Canadell, R. Knutti, G. Luderer, M. R. Raupach, M. Schaeffer, D. P. van Vuuren and C. Le Quere, Nat. Geosci., 2014, 7, 709–715.
4. A. G. Olabi and M. A. Abdelkareem, Renewable Sustainable Energy Rev., 2022, 158, 112111.
5. J. F. D. Tapia, J. Y. Lee, R. E. H. Ooi, D. C. Y. Foo and R. R. Tan, Sustain. Prod. Consum., 2018, 13, 1–15.
6. K. Jiang, P. Ashworth, S. Y. Zhang, X. Liang, Y. Sun and D. Angus, Renewable Sustainable Energy Rev., 2020, 119, 109601.
7. S. Y. Chen, J. F. Liu, Q. Zhang, F. Teng and B. C. McLellan, Renewable Sustainable Energy Rev., 2022, 167, 112537.
8. R. P. Ye, J. Ding, W. Gong, M. D. Argyle, Q. Zhong, Y. Wang, C. K. Russell, Z. Xu, A. G. Russell, Q. Li, M. Fan and Y. G. Yao, Nat. Commun., 2019, 10, 5698.
9. S. Kar, A. Goeppert and G. K. S. Prakash, Acc. Chem. Res., 2019, 52, 2892–2903.
10. J. Zhong, X. Yang, Z. Wu, B. Liang, Y. Huang and T. Zhang, Chem. Soc. Rev., 2020, 49, 1385–1413.
11. S.-T. Bai, G. De Smet, Y. Liao, R. Sun, C. Zhou, M. Beller, B. U. W. Maes and B. F. Sels, Chem. Soc. Rev., 2021, 50, 4259–4298.
12. M. Behrens, F. Studt, I. Kasatkin, S. Kuhl, M. Havecker, F. Abild-Pedersen, S. Zander, F. Girgsdies, P. Kurr, B. L. Kniep, M. Tovar, R. W. Fischer, J. K. Norskov and R. Schlogl, Science, 2012, 336, 893–897.
13. A. Alvarez, A. Bansode, A. Urakawa, A. V. Bavykina, T. A. Wezendonk, M. Makkee, J. Gascon and F. Kapteijn, Chem. Rev., 2017, 117, 9804–9838.
14. X. Jiang, X. Nie, X. Guo, C. Song and J. G. Chen, Chem. Rev., 2020, 120, 7984–8034.
15. A. Beck, M. A. Newton, L. G. A. van de Water and J. A. van Bokhoven, Chem. Rev., 2024, 124, 4543–4678.
16. Z. Chen, J. Wen, Y. Zeng, M. Li, Y. Tian, F. Yang, M. M.-J. Li, P. Chen, H. Huang, D. Ye and L. Chen, Appl. Catal., B, 2024, 340, 123192.
17. J. Wang, Y. Song, J. Li, F. Liu, J. Wang, J. Lv, S. Wang, M. Li, X. Bao and X. Ma, Appl. Catal., A, 2024, 674, 119618.
18. Y. Xu, Z. Gao, Y. Xu, X. Qin, X. Tang, Z. Xie, J. Zhang, C. Song, S. Yao, W. Zhou, D. Ma and L. Lin, Appl. Catal., B, 2024, 344, 123656.
19. J. Zhu, D. Ciolca, L. Liu, A. Parastaev, N. Kosinov and E. J. M. Hensen, ACS Catal., 2021, 11, 4880–4892.
20. Y. Ou, T. Zhang, L. Lv, W. Tang and S. Tang, Colloids Surf., A, 2023, 676, 132167.
21. C. Han, L. Qin, P. Wang, H. Zhang, Y. Gao, M. Zhu, S. Wang and J. Li, Fuel, 2024, 357, 129945.
22. T. Qi, W. Li, H. Li, K. Ji, S. Chen and Y. Zhang, Mol. Catal., 2021, 509, 111641.
23. R. Dalebout, L. Barberis, N. L. Visser, J. E. S. van der Hoeven, A. M. J. van der Eerden, J. A. Stewart, F. Meirer, K. P. de Jong and P. E. de Jongh, ChemCatChem, 2022, 14, e202200451.
24. F. Chen, W. Gao, K. Wang, C. Wang, X. Wu, N. Liu, X. Guo, Y. He, P. Zhang, G. Yang and N. Tsubaki, Fuel, 2022, 315, 123272.
25. B. Hu, Y. Yin, Z. Zhong, D. Wu, G. Liu and X. Hong, Catal. Sci. Technol., 2019, 9, 2673–2681.
26. Y. Xu, Z. Gao, L. Peng, K. Liu, Y. Yang, R. Qiu, S. Yang, C. Wu, J. Jiang, Y. Wang, W. Tan, H. Wang and J. Li, J. Catal., 2022, 414, 236–244.
27. R. M. Palomino, P. J. Ramírez, Z. Liu, R. Hamlyn, I. Waluyo, M. Mahapatra, I. Orozco, A. Hunt, J. P. Simonovis, S. D. Senanayake and J. A. Rodriguez, J. Phys. Chem. B, 2017, 122, 794–800.
28. A. Arandia, J. Yim, H. Warraich, E. Leppäkangas, R. Bes, A. Lempelto, L. Gell, H. Jiang, K. Meinander, T. Viinikainen, S. Huotari, K. Honkala and R. L. Puurunen, Appl. Catal., B, 2023, 321, 122046.
29. C. Li, K. Chen, X. Wang, N. Xue and H. Yang, Acta Phys.-Chim. Sin., 2020, 37, 2009101.
30. S. Kuld, M. Thorhauge, H. Falsig, C. F. Elkjaer, S. Helveg, I. Chorkendorff and J. Sehested, Science, 2016, 352, 969–974.
31. S. Kattel, P. J. Ramirez, J. G. Chen, J. A. Rodriguez and P. Liu, Science, 2017, 355, 1296–1299.
32. J. Nakamura, T. Fujitani, S. Kuld, S. Helveg, I. Chorkendorff and J. Sehested, Science, 2017, 357, eaan8074.
33. R. Dalebout, L. Barberis, G. Totarella, S. J. Turner, C. La Fontaine, F. M. F. de Groot, X. Carrier, A. M. J. van der Eerden, F. Meirer and P. E. de Jongh, ACS Catal., 2022, 12, 6628–6639.
34. X. Liu, J. Luo, H. Wang, L. Huang, S. Wang, S. Li, Z. Sun, F. Sun, Z. Jiang, S. Wei, W. X. Li and J. Lu, Angew. Chem., Int. Ed., 2022, 61, e202202330.
35. X. Liu, H. Wang and J. Lu, J. Catal., 2024, 436, 115561.
36. P. Amann, B. Klotzer, D. Degerman, N. Kopfle, T. Gotsch, P. Lomker, C. Rameshan, K. Ploner, D. Bikaljevic, H. Y. Wang, M. Soldemo, M. Shipilin, C. M. Goodwin, J. Gladh, J. H. Stenlid, M. Borner, C. Schlueter and A. Nilsson, Science, 2022, 376, 603–608.
37. D. Laudenschleger, H. Ruland and M. Muhler, Nat. Commun., 2020, 11, 3898.
38. W. Tu, P. Ren, Y. Li, Y. Yang, Y. Tian, Z. Zhang, M. Zhu, Y. C. Chin, J. Gong and Y. F. Han, J. Am. Chem. Soc., 2023, 145, 8751–8756.
39. L. Xu, K. G. Papanikolaou, B. A. J. Lechner, L. Je, G. A. Somorjai, M. Salmeron and M. Mavrikakis, Science, 2023, 380, 70–76.
40. R. Wang, H. Wang, X. Weng, J. Dai, Z. Gong, C. Zhao, J. Lu, Y. Cui and X. Bao, J. Energy Chem., 2021, 60, 150–155.
41. S. Jensen, M. H. R. Mammen, M. Hedevang, Z. Li, L. Lammich and J. V. Lauritsen, Nat. Commun., 2024, 15, 3865.
42. S. Kattel, P. Liu and J. G. Chen, J. Am. Chem. Soc., 2017, 139, 9739–9754.
43. J. A. Rodriguez, P. Liu, D. J. Stacchiola, S. D. Senanayake, M. G. White and J. G. Chen, ACS Catal., 2015, 5, 6696–6706.
44. J. Y. Han, J. Y. Yang, Z. X. Zhang, X. Z. Jiang, W. Liu, B. T. Qiao, J. J. Mu and F. Wang, J. Am. Chem. Soc., 2023, 145, 22671–22684.
45. C. R. A. Catlow, S. A. French, A. A. Sokol, M. Alfredsson and S. T. Bromley, Faraday Discuss., 2003, 124, 185–203.
46. C. Wu, L. Lin, J. Liu, J. Zhang, F. Zhang, T. Zhou, N. Rui, S. Yao, Y. Deng, F. Yang, W. Xu, J. Luo, Y. Zhao, B. Yan, X.-D. Wen, J. A. Rodriguez and D. Ma, Nat. Commun., 2020, 11, 5767.
47. F. Zhang, B. Li, X. Quan, K. Wang, J. Xu, T. Wu, Z. Li, M. Yan, S. Liu, Y. He, Y. Shi, Y. Su and P. Xie, ACS Catal., 2024, 14, 7136–7148.
48. X.-K. Wu, G.-J. Xia, Z. Huang, D. K. Rai, H. Zhao, J. Zhang, J. Yun and Y.-G. Wang, Appl. Surf. Sci., 2020, 525, 146481.
49. T. Reichenbach, K. Mondal, M. Jäger, T. Vent-Schmidt, D. Himmel, V. Dybbert, A. Bruix, I. Krossing, M. Walter and M. Moseler, J. Catal., 2018, 360, 168–174.
50. K. Mondal, Megha, A. Banerjee, A. Fortunelli, M. Walter and M. Moseler, J. Phys. Chem. C, 2021, 126, 764–771.
51. W. Xiong, Z. F. Wu, X. Y. Chen, J. Q. Ding, A. A. I. Ye, W. H. Zhang and W. X. Huang, Sci. China: Chem., 2023, 67, 715–723.
52. X. Z. Wang, H. Zhang, H. Qin, K. M. Wu, K. Wang, J. F. Ma and W. D. Fan, Fuel, 2023, 346, 128381.
53. Z.-J. Zuo, P.-D. Han, Z. Li, J.-S. Hu and W. Huang, Appl. Surf. Sci., 2012, 261, 640–646.
54. S. T. Bromley, S. A. French, A. A. Sokol, C. R. A. Catlow and P. Sherwood, J. Phys. Chem. B, 2003, 107, 7045–7057.
55. K. Yao, S.-S. Wang, X.-K. Gu, H.-Y. Su and W.-X. Li, Chin. J. Catal., 2013, 34, 1705–1711.
56. J. Fu, S. Liu, W. Zheng, R. Huang, C. Wang, A. Lawal, K. Alexopoulos, S. Liu, Y. Wang, K. Yu, J. A. Boscoboinik, Y. Liu, X. Liu, A. I. Frenkel, O. A. Abdelrahman, R. J. Gorte, S. Caratzoulas and D. G. Vlachos, Nat. Catal., 2022, 5, 144–153.
57. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
58. G. Kresse and J. Furthmuller, Comput. Mater. Sci., 1996, 6, 15–50.
59. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
60. P. E. Blochl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
61. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
62. T. Lunkenbein, J. Schumann, M. Behrens, R. Schlögl and M. G. Willinger, Angew. Chem., Int. Ed., 2015, 54, 4544–4548.
63. T. Song, R. Li, J. Wang, C. Dong, X. Feng, Y. Ning, R. Mu and Q. Fu, Angew. Chem., Int. Ed., 2023, 63, e202316888.
64. M. Mahapatra, J. Kang, P. J. Ramírez, R. Hamlyn, N. Rui, Z. Liu, I. Orozco, S. D. Senanayake and J. A. Rodriguez, J. Phys. Chem. C, 2018, 122, 26554–26562.
65. V. Wang, N. Xu, J.-C. Liu, G. Tang and W.-T. Geng, Comput. Phys. Commun., 2021, 267, 108033.
66. G. Henkelman, B. P. Uberuaga and H. Jonsson, J. Chem. Phys., 2000, 113, 9901–9904.
67. G. Henkelman and H. Jonsson, J. Chem. Phys., 2000, 113, 9978–9985.
68. A. Heyden, A. T. Bell and F. J. Keil, J. Chem. Phys., 2005, 123, 224101.
69. Materials Studio, Biovia Software Inc., 5005 Wateridge Vista Drive, San Diego, CA 92121 USA, 2018.
70. Q.-L. Tang, W.-T. Zou, R.-K. Huang, Q. Wang and X.-X. Duan, Phys. Chem. Chem. Phys., 2015, 17, 7317–7333.
71. M. D. Higham, M. G. Quesne and C. R. A. Catlow, Dalton Trans., 2020, 49, 8478–8497.
72. Y.-F. Zhao, R. Rousseau, J. Li and D. Mei, J. Phys. Chem. C, 2012, 116, 15952–15961.
73. V. Pallassana and M. Neurock, J. Catal., 2000, 191, 301–317.
74. T. Bligaard, J. K. Nørskov, S. Dahl, J. Matthiesen, C. H. Christensen and J. Sehested, J. Catal., 2004, 224, 206–217.
75. J. K. Nørskov, T. Bligaard, J. Rossmeisl and C. H. Christensen, Nat. Chem., 2009, 1, 37–46.
76. M. Heenemann, M.-M. Millet, F. Girgsdies, M. Eichelbaum, T. Risse, R. Schlögl, T. Jones and E. Frei, ACS Catal., 2020, 10, 5672–5680.
77. K. Larmier, W. C. Liao, S. Tada, E. Lam, R. Verel, A. Bansode, A. Urakawa, A. Comas-Vives and C. Coperet, Angew. Chem., Int. Ed., 2017, 56, 2318–2323.
78. Y. H. Wang, S. Kattel, W. G. Gao, K. Z. Li, P. Liu, J. G. G. Chen and H. Y. Wang, Nat. Commun., 2019, 10, 1166.
79. A. J. Medford, J. Sehested, J. Rossmeisl, I. Chorkendorff, F. Studt, J. K. Nørskov and P. G. Moses, J. Catal., 2014, 309, 397–407.
80. Y. Yang, J. Evans, J. A. Rodriguez, M. G. White and P. Liu, Phys. Chem. Chem. Phys., 2010, 12, 9909–9917.
81. L. C. Grabow and M. Mavrikakis, ACS Catal., 2011, 1, 365–384.
82. Y. Yang, C. A. Mims, D. H. Mei, C. H. F. Peden and C. T. Campbell, J. Catal., 2013, 298, 10–17.
83. S. Kattel, B. Yan, Y. Yang, J. G. Chen and P. Liu, J. Am. Chem. Soc., 2016, 138, 12440–12450.
84. L. Jiang, K. Liu, S.-F. Hung, L. Zhou, R. Qin, Q. Zhang, P. Liu, L. Gu, H. M. Chen, G. Fu and N. Zheng, Nat. Nanotechnol., 2020, 15, 848–853.
85. T. Yang, T. Gu, Y. Han, W. Wang, Y. Yu, Y. Zang, H. Zhang, B. Mao, Y. Li, B. Yang and Z. Liu, J. Phys. Chem. C, 2020, 124, 27511–27518.
86. L. Song, X. Tian, Y. Yang, J. Qin, H. Li and X. Lin, Front. Chem., 2020, 8, 607.
87. K. C. Waugh, Catal. Lett., 2012, 142, 1153–1166.
88. G. C. Chinchen, P. J. Denny, D. G. Parkera, M. S. Spencer and D. A. Nhan, Appl. Catal., 1987, 30, 333–338.
89. G. C. Chinchen, P. J. Denny, J. R. Jennings, M. S. Spencer and K. C. Waugh, Appl. Catal., 1988, 36, 1–65.
90. Y.-F. Shi, P.-L. Kang, C. Shang and Z.-P. Liu, J. Am. Chem. Soc., 2022, 144, 13401–13414.
91. F. Studt, M. Behrens, E. L. Kunkes, N. Thomas, S. Zander, A. Tarasov, J. Schumann, E. Frei, J. B. Varley, F. Abild-Pedersen, J. K. Nørskov and R. Schlögl, ChemCatChem, 2015, 7, 1105–1111.
92. Y.-F. Zhao, Y. Yang, C. Mims, C. H. F. Peden, J. Li and D. Mei, J. Catal., 2011, 281, 199–211.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cy00604f

---