*Article* **Hydride Generation on the Cu-Doped CeO2(111) Surface and Its Role in CO<sup>2</sup> Hydrogenation Reactions**

**Zhi-Qiang Wang 1,†, Hui-Hui Liu 1,†, Xin-Ping Wu 1,\* , Peijun Hu 1,2 and Xue-Qing Gong 1,\***


**Abstract:** Ceria-based catalysts exhibit great activity in catalyzing selective hydrogenation of CO<sup>2</sup> to methanol. However, the underlying mechanism of this reaction, especially the generation of active H species, remains unclear. In this work, we performed extensive density functional theory calculations corrected by on-site Coulomb interaction (DFT + U) to investigate the H<sup>2</sup> dissociation and the reaction between the active H species and CO<sup>2</sup> on the pristine and Cu-doped CeO<sup>2</sup> (111) (denoted as Cu/CeO<sup>2</sup> (111)) surfaces. Our calculations evidenced that the heterolytic H<sup>2</sup> dissociation for hydride generation can more readily occur on the Cu/CeO<sup>2</sup> (111) surface than on the pristine CeO<sup>2</sup> (111) surface. We also found that the Cu dopant can facilitate the formation of surface oxygen vacancies, further promoting the generation of hydride species. Moreover, the adsorption of CO<sup>2</sup> and the hydrogenation of CO<sup>2</sup> to HCOO\* can be greatly promoted on the Cu/CeO<sup>2</sup> (111) surface with hydride species, which can lead to the high activity and selectivity toward CO<sup>2</sup> hydrogenation to methanol.

**Keywords:** ceria; hydride; CO<sup>2</sup> hydrogenation; single Cu doping; density functional theory

#### **1. Introduction**

Oxide-based materials are increasingly recognized as highly efficient and selective catalysts for hydrogenation reactions [1–6]. Among numerous metal oxides, ceria (CeO2), which usually contains abundant oxygen vacancies (OV) on the surface and in the bulk, has attracted great attention for its excellent performance in catalyzing reactions such as hydrogenation of CO<sup>2</sup> to methanol [7–12]. The combined experimental and density functional theory (DFT) study of Liu et al. [9] reported that the morphology control of CeO<sup>2</sup> nano-catalysts is important for methanol synthesis. They also proposed that methanol was likely generated via the so-called formate (HCOO) pathway where the adsorbed CO<sup>2</sup> is firstly hydrogenated to the HCOO\* species. Cheng et al. [10] investigated the conversion of CO<sup>2</sup> to methanol on the reduced CeO2(110) surface by performing DFT calculations corrected by on-site Coulomb interaction (DFT + U) and microkinetic analysis. They also found that the HCOO route is the dominant pathway for methanol formation on the reduced CeO2(110). Kumari et al. [11] performed DFT calculations to study the mechanisms of CO<sup>2</sup> reduction to CO and the hydrogenation of CO<sup>2</sup> to methanol on both the stoichiometric and reduced CeO2(110) surfaces. It was found that CO<sup>2</sup> dissociates to CO through the interaction with the oxygen vacancy on the reduced ceria surface, and the produced CO can be further hydrogenated to methanol. Li et al. [12] investigated the mechanism of thermal catalytic hydrogenation of CO<sup>2</sup> to methanol on the reduced CeO2(100) by using DFT calculations, and they found that CO<sup>2</sup> was hydrogenated via

**Citation:** Wang, Z.-Q.; Liu, H.-H.; Wu, X.-P.; Hu, P.; Gong, X.-Q. Hydride Generation on the Cu-Doped CeO2(111) Surface and Its Role in CO<sup>2</sup> Hydrogenation Reactions. *Catalysts* **2022**, *12*, 963. https://doi.org/10.3390/ catal12090963

Academic Editor: Albert Poater

Received: 14 July 2022 Accepted: 26 August 2022 Published: 29 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the HCOO route rather than the COOH route. These results then indicate that oxygen vacancies on the reduced CeO<sup>2</sup> surface are crucial to the conversion of CO<sup>2</sup> to CH3OH.

Cu-based catalysts are particularly active for CO<sup>2</sup> hydrogenation and have been commercially utilized for CO/CO<sup>2</sup> hydrogenation to methanol [13–16]. Xu et al. [17,18] reported the synergetic effect of Cu and ZnO on the catalytic activity and selectivity in the direct hydrogenation of CO<sup>2</sup> to C2+ alcohols. This synergetic effect can regulate the CO<sup>2</sup> and H<sup>2</sup> activation processes to achieve the optimal concentrations of surface CHx\*, CO\* and H\* species for C2+ alcohols synthesis via CHx\*−CO\* coupling and hydrogenation reactions. Jia et al. [19] found that abundant surface oxygen vacancies were in situ generated and consumed during the CO<sup>2</sup> reduction reaction on the Cu/CeO<sup>2</sup> catalyst with a Cu loading of 15 wt.%, and such vacancy sites can be combined with the adjacent copper clusters to promote the activation of CO<sup>2</sup> and increase the catalytic efficiency. Wang et al. [20] proposed that sub-nano Cu cubes supported on the thermally reduced CeO<sup>2</sup> nanorods through strong metal-support interaction (SMSI) can significantly enhance the formation and activity of hydrides species on both Cu and ceria, as compared with the Cu nanoparticles supported at stoichiometric CeO<sup>2</sup> nanorods. Xia et al. [21] performed DFT calculations and illustrated that the formation of oxygen vacancies was suppressed at the interface of the Cu cluster and CeO2(111), while H<sup>2</sup> oxidation can be facilitated and the catalytic activity can be enhanced, in comparison with the pristine CeO2(111). At the same time, single-atom catalysts were often reported to have catalytic properties that surpass those of nanoparticles [22,23]. Huang et al. [24] demonstrated that the C3N4-supported Cu single-atom catalysts with tailored coordination structures can serve as highly active and selective catalysts for CO<sup>2</sup> hydrogenation to methanol at low temperature, though the role of the Cu single atom in such catalytic process is still vague.

The study of the interaction between H<sup>2</sup> and ceria is crucial to understanding the mechanism of ceria catalyzed selective hydrogenation reactions [25–29]. Many studies have found that the homolytic dissociation of H<sup>2</sup> to form two hydroxyl (OH) groups is thermodynamically favored on the stoichiometric surface of ceria [6,30]. On the other hand, a few recent studies suggested that oxygen vacancies can facilitate the formation of active hydrides (H−) species through heterolytic dissociation of H<sup>2</sup> on the reduced surface of ceria [6,31–37]. This agrees with the previous studies that reduced ceria surface is more active than the stochiometric one in CO<sup>2</sup> hydrogenation reactions [10,11]. Moreover, our recent studies proposed that low-coordinated Ce sites can promote the formation of H− species even on the stoichiometric CeO<sup>2</sup> stepped surfaces [38], which is consistent with the finding by García-Melchor et al. [27], and suggests that oxygen vacancy may not be essential for the formation of hydrides on the CeO<sup>2</sup> surfaces. To better drive the hydrogenation reactions such as the CO<sup>2</sup> reduction reaction (CO2-RR) under moderate conditions, the activity and stability of the hydride species would need to be further enhanced on the surfaces of the ceria-based catalysts.

In this work, we investigated the role of the Cu dopant on the CeO2(111) surface in the catalytic dissociation of H<sup>2</sup> to produce hydride species. Specifically, DFT + U calculations have been performed to illustrate the reactivity of H<sup>2</sup> dissociation on the pristine and Cu-doped CeO2(111) (denoted as Cu/CeO2(111)) surfaces. Our results showed that the doping of a single Cu atom is accompanied by the formation of an oxygen vacancy on the CeO2(111) surface, and the formed oxygen vacancy can play a key role in the formation of H<sup>−</sup> species through heterolytic H<sup>2</sup> dissociation. In addition, the single Cu site can also enhance the adsorption of CO2, which is favorable for improving the catalytic activity. Our calculations further indicated that the Cu/CeO2(111) surface with H− species exhibits high activity and selectivity in the hydrogenation of CO<sup>2</sup> to methanol.

#### **2. Calculation Methods**

In this work, all spin-polarized DFT calculations were carried out using the Vienna Ab– initio Simulation Package (VASP) [39]. The projector augmented wave (PAW) method [40] and the Perdew−Burke−Ernzerhof (PBE) [41] functional under the generalized gradient

approximation (GGA) [42] were applied in the calculations. The kinetic energy cut-off was set as 400 eV (Table S1), and the force threshold for structure optimizations was 0.05 eV/Å. A large vacuum gap of 15 Å was used to eliminate the interaction between neighboring slabs. Different *k*-point meshes have been tested during the optimization of the unit cell of bulk ceria. It was shown that the *k*-point mesh of (5 × 5 × 5) can give the converged result (Table S2), and the calculated lattice parameter (a = b = c = 5.456 Å) is in good agreement with the experimental value (5.411 Å), which was then used for the subsequent study [43].

We then built a *p*(3 × 3) surface slab containing three O−Ce−O layers to model the CeO2(111) surface. To model the Cu-doped CeO2(111), one surface Ce atom was replaced by a Cu atom and one surface oxygen nearby was removed accordingly to ensure charge conservation. The top two layers of the slabs were allowed to fully relax, while the bottom layer was kept fixed to mimic the bulk region. Due to the relatively large surface cell (11.56 Å × 11.56 Å), the *k*-point mesh of (2 × 2 × 1) suggested by previous studies was used for Brillouin-zone integrations [44]. Note that the on-site Coulomb interaction correction is necessary for the appropriate description of the localized Ce 4*f* electrons [45–47], and therefore we used an effective U value of 5 eV, as suggested by previous studies [45,46].

The transition states (TSs) of surface reactions were located using a constrained optimization scheme and were verified when (i) all forces on the relaxed atoms vanish and (ii) the total energy is a maximum along the reaction coordination but a minimum with respect to the rest of the degrees of freedom [48–50]. For example, when using this approach to calculate the transition states of H<sup>2</sup> dissociation, we fixed the H-H distance at preselected values and then optimized the whole system with respect to all the remaining degrees of freedom.

The adsorption energy of species X on the surface, Eads(X), was calculated with

$$\mathbf{E\_{ads}(X)} = -(\mathbf{E\_{X/slab}} - \mathbf{E\_{slab}} - \mathbf{E\_{X}}) \tag{1}$$

where EX/slab is the calculated total energy of the adsorption system, while Eslab and E<sup>X</sup> are the calculated energies of the clean surface and the gas phase molecule X, respectively. Accordingly, a positive Eads(X) value indicates an energetically favorable adsorption process, and the more positive the Eads(X) is, the more strongly the adsorbate X binds to the surface, and this definition was also proposed by Somorjai and Li [51].

The neutral oxygen vacancy formation energy (EOV) was calculated according to

$$\rm{E\_{OV}} = \rm{E\_{slab-vac}} + 1/2\rm{E\_{O2}} - \rm{E\_{slab}} \tag{2}$$

where Eslab-vac is the total energy of the surface with a neutral oxygen vacancy and EO2 is the energy of a gas-phase O<sup>2</sup> molecule.

#### **3. Results and Discussion**

#### *3.1. Structural and Electronic Properties*

The CeO2(111) surface is known as the most stable termination of ceria [26,27]. From Figure 1a, one can see that there are two types of lattice oxygen on the CeO2(111) surface, namely, the three-fold coordinated O (denoted as O3c) on the top layer and four-fold coordinated O (denoted as O4c) on the subsurface. The calculated bond lengths of Ce−O3c (2.36 Å) and Ce−O4c (2.37 Å) are nearly identical. Figure 1b shows the optimized structure of the Cu-doped CeO2(111) (Cu/CeO2(111)). As we have explained, the doping of one Cu atom onto the CeO2(111) surface is accompanied by the spontaneous formation of an oxygen vacancy (EOV = −0.94 eV) to ensure charge conservation [52,53]. The calculated bond lengths of Ce−O3c (2.35 Å) and Ce−O4c (2.36 Å) on the doped Cu/CeO2(111) are nearly identical to those of the pristine CeO2(111). Moreover, the calculated bond lengths of Cu−O3c and Cu−O4c on the Cu/CeO2(111) surface are 3.13 and 1.96 Å, respectively, and these results are consistent with previous studies [52–54]. The elongation of the Cu−O3c bond also indicates that the top surface O atoms bonding to Cu are activated.

**Figure 1.** Calculated structures (left: top view; right: side view) of the (**a**) CeO<sup>2</sup> (111) and (**b**) Cu/CeO<sup>2</sup> (111) surfaces, and calculated density of states (DOS) of the (**c**) CeO<sup>2</sup> (111) and (**d**) Cu/CeO<sup>2</sup> (111) surfaces. The blue circles filled with yellow represent the inherent surface oxygen vacancy on the Cu/CeO<sup>2</sup> (111). Red: O atoms; ivory: Ce atoms; blue: Cu atoms. These notations are used throughout the paper. All DOS were aligned with respect to the O 2*s* orbital of a fixed bottom O atom of the surface slabs.

To better understand the doped Cu/CeO2(111), we first calculated the density of states (DOS) and spin density distribution of this surface (Figure 1d). The lowest unoccupied states are mainly composed of Cu 3*d* and O 2*p*. This indicates that the local CuO structure is capable of receiving extra electrons and may act as the catalytically active site. The calculated Bader charge of the Cu dopant is +1.04 |e|, which is close to that of the Cu2+ on the CuO surface (+0.97 |e|, see Table S3), indicating that the oxidation state of the Cu dopant on the CeO2(111) is indeed +2 [52,55]. This is also consistent with the calculated spin density distribution on the Cu site (Figure 1d). In addition, we also found that after Cu doping the energy level of the lowest unoccupied Ce 4*f* orbital decreases by 0.23 eV from 1.56 eV to 1.33 eV (Figure 1c,d), indicating that the electron accepting capacity of Ce 4*f* is also enhanced accordingly.

#### *3.2. H Adsorption*

Previous studies suggested that hydride is the key species for selective hydrogenation of CO<sup>2</sup> over ceria surfaces [10,11]. To verify the nature of various H species including proton, hydride and hydrogen radical (H•), one needs to calculate their Bader charges as well as the spin density distributions. For the hydride species, there is no (net) spin density localized on it, and the calculated Bader charges are usually negative, while for the hydrogen radical, there are spin densities localized on it and the calculated Bader charge is usually close to 0 |e|. Bearing the above in mind, we thus studied the adsorption of one H atom at different sites on the Cu/CeO2(111) surface. The calculated results show that two types of H species can be formed after H adsorption at the Cu site (Figure 2a,b). Electronic structure analyses confirmed that one adsorbed H species is a radical-like species since it has net spin densities (Figure 2a), and the other one has a Bader charge that is close to zero (Figure 2b). For the calculated adsorption energies, they are negative for both cases, though the second type gives higher value (−1.61 vs. −0.45 eV, Figure 2a,b). Moreover, the calculated density of states of the system shown in Figure S1 also indicates that a relatively stronger Cu-H bond is formed for the second adsorption structure (Figure 2b) [53,54]. In addition, we also investigated H adsorption at the surface O site. In this case, a proton (H<sup>+</sup> )

is formed, and the surface is clearly reduced, with either the Cu or Ce taking the electron from the H (Figure 2c,d). However, it needs to be mentioned that for the reduction of Cu2+ to Cu1+ (Bader charge = 0.66 |e|), the corresponding adsorption process is exothermic by 1.99 eV (Figure 2c), and for the reduction of Ce4+ to Ce3+, it is exothermic by 2.36 eV (Figure 2d).

**Figure 2.** Calculated structures (left: side view; right: top view) of H adsorption at (**a**,**b**) Cu and (**c**,**d**) O sites of the Cu/CeO<sup>2</sup> (111) surface. The calculated adsorption energies of H, Bader charges and the magnetic moments of H (left) and Cu (right), and spin density distributions (in yellow) are also shown.

Hermansson and Fernandez-Gacia and their co-workers [52,55] have shown that the four coordinated planar Cu2+ is relatively stable within the Cu/CeO2(111) structure. It is interesting for us to find that with the hydrogen species being adsorbed at the O site on the Cu/CeO2(111) surface, the coordinate number of Cu maintains four when the extra electron is localized at the Ce site. On the other hand, when the extra electron is localized at the Cu, it induces the change of the coordinate number of Cu to three (Figure S2 and Table S4). In fact, we have also estimated and compared the electrostatic interaction energies at such Ce3+ and Cu1+ sites, which are <sup>−</sup>134.05 eV and <sup>−</sup>130.40 eV, respectively (see Figure S2 and Table S4). These results clearly show that the Cu/CeO2(111) surface can be more stable with the extra electron being localized at the Ce site than the Cu due to the more favorable electrostatic interaction. Moreover, the calculated density of states of the corresponding systems further showed that the newly occupied Cu 3*d* orbital is 0.15 eV higher in energy than the occupied Ce 4*f* orbital (Figure S3). These results can indeed support that the excess electron prefers to be localized in the Ce 4*f*, even though the empty Cu(3*d*)-O(2*p*) state is lower (Figure 1d).
