Cu/O Frustrated Lewis Pairs on Cu Doped CeO 2 (111) for Acetylene Hydrogenation: A First-Principles Study

: In this work, the H 2 dissociation and acetylene hydrogenation on Cu doped CeO 2 (111) were studied using density functional theory calculations. The results indicated that Cu doping promotes the formation of oxygen vacancy (O v ) which creates Cu/O and Ce/O frustrated Lewis pairs (FLPs). With the help of Cu/O FLP, H 2 dissociation can firstly proceed via a heterolytic mechanism to produce Cu-H and O-H by overcoming a barrier of 0.40 eV. The H on Cu can facilely migrate to a nearby oxygen to form another O-H species with a barrier of 0.43 eV. The rate-determining barrier is lower than that for homolytic dissociation of H 2 which produces two O-H species. C 2 H 2 hydrogenation can proceed with a rate-determining barrier of 1.00 eV at the presence of Cu-H and O-H species., While C 2 H 2 can be catalyzed by two O-H groups with a rate-determining barrier of 1.06 eV, which is significantly lower than that (2.86 eV) of C 2 H 2 hydrogenated by O-H groups on the bare CeO 2 (111), showing the high activity of Cu doped CeO 2 (111) for acetylene hydrogenation. In addition, the rate-determining barrier of C 2 H 4 further hydrogenated by two O-H groups is 1.53 eV, much higher than its desorption energy (0.72 eV), suggesting the high selectivity of Cu doped CeO 2 (111) for C 2 H 2 partial hydrogenation. This provides new insights to develop effective hydrogenation catalysts based on metal oxide.


Introduction
Partial hydrogenation of acetylene in excess of ethylene is a crucial industry process in the purifying of ethylene produced in steam cracker since polymerization catalysts can be poisoned by acetylene [1]. Pd based catalysts are commonly used in this reaction, which have been extensively studied over the past decades [2][3][4]. However, the lower selectivity caused by over-hydrogenation and oligomerization is a long-standing concern [1,4,5]. The high cost of Pd triggers intense interest in searching for low-cost catalysts with high performance.
The surface oxygens were previously considered as the active sites for acetylene hydrogenation based on the density functional theory (DFT) study on CeO2(111) [26]. It was suggested that acetylene be catalyzed by the homolytic products of H2 dissociation (O-H groups) but with an extremely high barrier of 2.86 eV for the second hydrogenation step. Subsequently, a concerted mechanism was reported by García-Melchor et al., suggesting that propyne is simultaneously hydrogenated by a surface O-H and a H atom in gas H2 with a barrier of about 1.88 eV [27]. Recently, a new mechanism on CeO2(110), CeO2(100), and CeO2(111) was proposed based on the presence of oxygen vacancies (Ovs) [19,28,29]. The Frustrated Lewis acid-base pairs (FLPs), namely, the combination of Lewis acid and Lewis base sterically encumbered, play a vital role in the hydrogenation process due to their high activity in activation of small molecules [30][31][32][33][34][35][36][37]. For example, on CeO2(111), the acetylene is hydrogenated by the Ce-H and O-H groups generated by H2 heterolytic dissociation with a barrier of 0.70 eV, significantly lower than the previous value (2.86 eV) [26]. The FLP formed by the Ce exposed by Ovs and the surface oxygen promotes the H2 heterolytic dissociation with a much lower barrier (0.52 eV) than that (about 1 eV) [38,39] of homolytic dissociation. More importantly, the Ce sites can help stabilizing the Ce-H hydride and avoiding the strong adsorption of C2H3 intermediate, thus leading to a lower hydrogenation barrier. This mechanism concerning the involvement of Ce-H hydride in acetylene hydrogenation was partly confirmed by the experimental observations by different techniques such as neutron scattering spectroscopy [14,40].
Although CeO2 shows high selectivity for alkyne hydrogenation, the reaction requires a high temperature >500 K [20]. Therefore, it is highly desirable to enhance the catalytic activity of CeO2 for alkyne hydrogenation. Riley et al. reported that Ni can be used as a dopant to create Ovs and improve the activity of CeO2 for acetylene hydrogenation [28]. They found that the reaction temperature on Ni doped CeO2 is significantly lower than that on bare CeO2, in which Ni did not take part in the reaction but acted as a promoter by leading to more active sites via increasing the Ovs. Moreover, the catalytic activity for alkyne hydrogenation can also be enhanced by introducing Ga into CeO2 lattice [41]. The Ga/O FLP led by Ga doping plays a vital role in H2 dissociation and reducing the rate-determining barrier of hydrogenation [42]. Cu based catalysts were found to show promising selectivity in hydrogenation reactions [43][44][45][46]. In addition, the previous studies indicated that Cu dopant [47,48] can enhance the formation of oxygen vacancies over CeO2, which might be helpful to improve the catalytic activity of CeO2 for hydrogenation. However, the catalytic role of Cu dopant in the hydrogenation on Cu doped CeO2(111) (denoted as Cu-CeO2(111)) is not clear.
In this work, since CeO2(111) is the most stable facet, the mechanism of H2 dissociation and C2H2 hydrogenation on Cu-CeO2(111) was studied through DFT calculations. The results indicated that Cu doping leads to the formation of Cu/O FLP which promotes the dissociation of H2 and C2H2 hydrogenation can be catalyzed by two surface O-H groups with the assistance of the Cu single atom on Cu-CeO2(111). This is organized as follows: Section II gives the computational details. The results and discussion are shown in Section III. The conclusions are presented in the final section.

Computational Details
All calculations were performed using spin-polarized DFT as implemented in Vienna ab initio simulation package (VASP) [49,50]. The gradient-corrected Perdew-Burke-Ernzerhof (PBE) approximation was used to treat the exchange-correlation potential [51]. The wave functions for the valence electrons were expanded in plane waves with a cutoff energy of 400 eV, while the core electrons were represented by projector augmented-wave (PAW) method [52]. DFT-D3 method of Grimme was employed to describe the van der Waals interaction [53]. The DFT + U method with an effective U = 4.5 eV was used to improve the description of Ce f states in ceria [54][55][56]. A periodic slab with a p(3 × 3) unit cell was selected to simulate the CeO2(111) surface. The CeO2(111) slab consists of nine atomic layers and a vacuum space of 14 Å. For all calculations, the atoms in top six layers were fully relaxed while those in the bottom three layers were fixed at their bulk positions. For Cu-CeO2(111), a surface Ce was replaced by a Cu. A 1 × 1 × 1 Monkhorst−Pack mesh k-point was adopted in Brillouin zone integration, which was tested to converge. Transition states of related elementary reaction steps were determined using the climbing image nudged elastic band (CI-NEB) method [57]. The convergence criteria for forces on each iron and for energy is set to 0.05 eV/Å and 10 −4 eV, respectively. The adsorption energy (Eads) was computed using the equation of Eads = E(adsorbate + surface) − E(free molecule) − E(free surface). The formation energy of an Ov was obtained by the following equation: Derived from homogeneous catalysis, the combination of Lewis acids and bases hindered by steric hindrance can form FLPs [30,31] which are found to be very efficient in activating small molecules such as CO2 and H2 [32][33][34][35][36][37]. For CeO2, a heterogeneous catalyst, the surface Ce and O atom can be considered as Lewis acid and base, respectively. In the presence of surface defects, some of the Lewis acid and base sites can be sterically separated, generating solid FLPs [19,28,29,32,42,58].

Geometry of Cu-CeO2(111) without and with One Ov
The most stable geometry of Cu-CeO2(111) are shown in Figure 1a. The Cu atom interacts with one surface oxygen (O1) and three subsurface oxygens (O2, O3, and O4), forming a square planar with the Cu-O1, Cu-O2, Cu-O3, and Cu-O4 distances of 1.97, 1.92, 1.93, and 1.98 Å, respectively. This is consistent with the result calculated by Guo et al [59]. The lattice distortion leads to two 2-fold Os, which are easy to form Ovs. The calculated formation energy of an Ov by a 2-fold O is −0.15 eV, indicating this process is thermally favorable. This is similar to the situation in Ni doped CeO2(111) [28]. The optimized geometry of Cu-CeO2(111) with one Ov (denoted as Cu-CeO2(111)-Ov) is shown in Figure 1b. For Cu-CeO2(111)-Ov, it was found that Cu and O5 are sterically separated by 3.64 Å, forming a standard FLP. The Cu and nonadjacent O6 with a distance of 3.24 Å also fall in the domain of FLP (denoted as Cu/O6 FLP). In addition, the distance of the nonadjacent Ce1 and O6 is 4.66 Å, forming another FLP candidate (denoted as Ce/O6 FLP).

Reactions
C2H2  [38,39] and is also about 0.1 eV lower than that of heterolytic dissociation on Ni doped CeO2(111) with an Ov [28]. This suggests that Cu/O5 FLP plays an important role in the stabilization of TS. Interestingly, the interaction of H2 with Cu leads to the structural transformation from CuO4 to CuO3, which facilitates the formation of Cu-H and O-H groups. The H2 dissociation via Path I releases an energy of 0.66 eV, showing that it is a thermally favorable process. For Path I, the corresponding geometries are shown in Figure 2 a-c. H2 weakly adsorbs on the surface with the adsorption energy of −0.19 eV with the H-H distance of 0.76 Å. In TS (Figure 2b), the bond distance of H1-O and Cu-H2 decreased to 1.29 and 1.77 Å, respectively, while the bond length of H1-H2 increased to 0.98 Å. Finally, it dissociates into H-O and Cu-H groups (Figure 2c) via a heterolytic path on Cu/O5 FLP by overcoming a barrier of 0.40 eV, which is about 0.6 eV lower than that of homolytic dissociation on CeO2(111) [38,39] and is also about 0.1 eV lower than that of heterolytic dissociation on Ni doped CeO2(111) with an Ov [28]. This suggests that Cu/O5 FLP plays an important role in the stabilization of TS. Interestingly, the interaction of H2 with Cu leads to the structural transformation from CuO4 to CuO3, which facilitates the formation of Cu-H and O-H groups. The H2 dissociation via Path I releases an energy of 0.66 eV, showing that it is a thermally favorable process.
For Path II, the geometries of IS, TS and FS are shown in Figure 2d-f. H2 dissociation proceeds on Ce/O6 sites through a heterolytic path. Before dissociation, H2 weakly adsorbs on the Ov with the adsorption energy of −0.24 eV. In TS (Figure 2e), the distance between H1 and H2 is elongated to 1.02 Å from 0.74 Å in its initial state, while the distances of H1-O2, H1-Ce1, and H2-Ce1 reduced to 1.27, 2.33, and 2.47 Å, respectively. The calculated barrier of H2 dissociation on Ce/O sites is 0.51 eV, which is very close to those on bare CeO2(111)-Ov (0.52 eV) and on Ni-CeO2(111)-Ov (0.50 eV) [28]. However, the barrier is about 0.16 eV higher than that (0.40 eV) on Cu/O5 FLP, indicating that Cu/O5 FLP is more active than Ce/O6 FLP for H2 dissociation. In FS (Figure 2f), H2 breaks into O-H and Ce-H hydride. This process is slightly endothermic (0.08 eV).
Besides the above paths, there is another reaction pathway (Path III) in which H2 was expected to dissociate on Cu/O6 FLP following the heterolytic mechanism. The corresponding geometries are displayed in Figure 2g-i. In this case, H2 is physically adsorbed on the surface with the adsorption energy of −0.19 eV. Observing the geometry of TS (Figure 2h), it is found that the distance of H1-H2 has increased to 0.86 Å and the distances of O1-H1, Cu-H1, and Cu-H2 decreased to 1.59, 1.83, and 1.71 Å, respectively, while the distance of O6-H2 is more than 3 Å. This suggests that the Cu/O6 is not an effective FLP for catalyzing H2 dissociation. Finally, we can see H2 dissociates into two O-H groups (Figure  2i), implying a homolytic mechanism. The calculated barrier is 0.54 eV.
Comparing the barriers of H2 dissociation via the above three discussed reaction paths, it is found that the barrier (0.40 eV) of Path I on Cu/O5 FLP is not only lower than that (0.51 eV) on Ce/O6 FLP in Path II and also lower than that (0.54 eV) in Path III, showing that H2 prefers to dissociate through Path I on Cu/O5 FLP to form Cu-H hydride and O-H groups.
To understand the stability of the Cu-H hydride, the migration of hydride H to a neighbor O is calculated. The geometries of IS, TS and FS are given in Figure 3. The calculated barrier is 0.43 eV, implying that the migration of hydride H is not difficult. This barrier is also lower than those for Path II (0.52 eV) and Path III (0.54 eV), indicating H2 can firstly dissociate via a heterolytic mechanism (Path I) and then the heterolytic products transform to homolytic ones with a barrier of 0.43 eV. This is different from the case of Ni doped CeO2(111) with one Ov, on which the hydride is very stable since the migration of hydride H needs to overcome a relatively large barrier of 0.87 eV [28]. Therefore, for Cu-CeO2(111)-Ov, the homolytic products (O-H groups) from H2 dissociation might be the active species for the subsequent C2H2 hydrogenation.

Acetylene Hydrogenation on Cu-CeO2(111)-Ov
The reaction process of acetylene catalyzed by the Cu-H and O-H groups was firstly studied (denoted as Path I). The corresponding geometries are shown in Figure 4a Based on the above results, it is possible that the migration of hydride H can take place before C2H2 hydrogenation because its barrier (0.43 eV) is very close to that (0.40 eV) of the first hydrogenation step and much lower than that (1.00 eV) of the second hydrogenation step. This is different with the case of Ni doped CeO2(111)-Ov, in which, the hydride is very stable due to the migration barrier of 0.87 eV is much higher than the first and second hydrogenation barriers (0.13 and 0.62 eV) [28].
To further investigate the possibility of acetylene hydrogenation catalyzed by the O-H groups and explore the role Cu dopant in the catalysis, the hydrogenation process was calculated (denoted as Path II). The corresponding geometries are depicted in Figure 4f-j and the reaction and activation energies are shown in Table 1.  (Figure  4i), the distance of C1-Cu and H1-O enlarged to 1.38 and 2.08 Å, respectively, and the distance of C1-H2 decreased to 1.30 Å. The second hydrogenation step is endothermic (0.73 eV). In the FS, C2H4 (Figure 4j) adsorbs on the surface with the C1-Cu and C2-Cu distance of 2.90 and 3.36 Å, respectively. In this configuration, the Cu connects with three subsurface Os, forming a CuO3 configuration. From Figure 5, one can see that CuO3 can facilely transform to a square planar CuO4 configuration via the migration of Cu with a minor barrier of 0.13 eV. In the other words, the Cu-CeO2(111)-Ov catalyst can recover itself, revealing the importance of structural dynamics during the catalysis. In order to investigate the possibility of C2H4 hydrogenated by the O-H groups, the hydrogenation process of C2H4 was also studied. The reaction and activation energies are also shown in Table 1 and the corresponding geometries are given in Figure 6a-e. The C2H4 adsorbs over the surface with an adsorption energy of −0.65 eV. It can react with the H on the O to form C2H5 by surpassing a barrier of 1.53 eV, which is much higher than that the desorption energy of C2H4 (0.72 eV). C2H5 adsorbs on the Cu site with an adsorption energy of −1.27 eV and with the C-Cu distance of 2.00 Å. Then C2H5 can be further hydrogenated to C2H6 by another surface H adsorbed on O with a barrier of 1.33 eV. The results indicate that the barriers for both the first and second steps of C2H4 hydrogenation are much higher than the desorption energy (0.72 eV) of C2H4 implying a good selectivity of Cu-CeO2(111)-Ov for C2H2 partial hydrogenation. Finally, the whole energy pathway is summarized with the energy profiles of C2H2 and C2H4 hydrogenation shown in Figure 7. The results of C2H2 hydrogenation on CeO2(111)-Ov calculated by some of our current authors [28] and on CeO2(111) calculated by Carrasco et al [26]. are also shown in Figure 7. For C2H2 catalyzed by the heterolytic products (Path I), the barrier of the second hydrogenation step is 1.00 eV, which is not only higher than the barrier (0.40 eV) of H2 heterolytic dissociation but also higher than that (0.40 eV) of the first hydrogenation step. Hence, the rate-determining barrier is 1.00 eV, which is much higher than the migration barrier (0.43 eV) of hydride H, suggesting the formation of homolytic products is very possible during the hydrogenation process. For C2H2 catalyzed by the homolytic products (Path II), the barrier of the second hydrogenation step is 1.06 eV. It is higher than that (0.69 eV) of the first hydrogenation step and that (0.43 eV) of H2 homolytic dissociation, showing the second hydrogenation step controls the whole reaction rate of C2H2 partial hydrogenation on Cu-CeO2(111)-Ov. Comparing the results with those on CeO2(111)-Ov and CeO2(111), it is found that the barrier of 1.06 eV is higher than that (0.70 eV) [28] on CeO2(111)-Ov, however, it is 1.80 eV lower than that (2.86 eV) [26] catalyzed by homolytic products on bare CeO2(111), suggesting the activity of CeO2 is effectively enhanced by the Cu dopant. Cu plays a vital role in reducing the barrier of second hydrogenation step, which avoids the strong adsorption of C2H3 intermediate. For C2H4, the barrier of the first hydrogenation step is 1.53 eV, higher than that (0.43 eV) of H2 dissociation and that (1.33 eV) of the second hydrogenation step, suggesting the addition of the first hydrogen is the rate-determining step. This barrier is 0.81 eV higher than the desorption energy (0.72 eV) of C2H4, implying a high selectivity of Cu-CeO2(111)-Ov for C2H2 partial hydrogenation.  Figures 1-6

Conclusions
In this work, the H2 dissociation and C2H2 hydrogenation on Cu-CeO2(111)-Ov were studied by density functional theory calculations. It is found that Cu doping leads to the spontaneous formation of Ov, generating Cu/O FLP which promotes the H2 dissociation and acetylene hydrogenation. On Cu-CeO2(111)-Ov, H2 can firstly dissociate to Cu-H and O-H groups via a heterolytic mechanism by surpassing a barrier of 0.40 eV. The migration of hydride H on Cu leads to the homolytic product of two O-H groups by overcoming a barrier of 0.43 eV, which is much lower than that (about 1 eV) of homolytic dissociation of H2 on bare CeO2(111). In addition, the results indicate that C2H2 can be catalyzed either by the heterolytic products (Cu-H and O-H) or homolytic products (O-Hs). For C2H2 hydrogenated by the heterolytic products, the addition of the second hydrogen controls the whole reaction rate. The rate-determining barrier is 1.00 eV, leading to a higher possibility of hydride H migration to form homolytic products. For C2H2 hydrogenated by the homolytic products, the reaction rate is determined by the second hydrogenation step with a barrier of 1.06 eV, which is about 1.80 eV lower than that (2.86 eV) of C2H2 hydrogenated by O-H groups on CeO2 (111), showing the efficiency of Cu-CeO2(111)-Ov for C2H2 hydrogenation. Moreover, the desorption energy (0.72 eV) of C2H4 is significantly lower than the rate-determining barrier (1.53 eV) for C2H4 hydrogenation, suggesting a high selectivity of Cu-CeO2(111)-Ov for acetylene partial hydrogenation. Cu dopant plays a vital role in promoting the activity of CeO2 for acetylene hydrogenation. Firstly, it creates the Cu/O FLP, the active site for H2 dissociation and C2H2 hydrogenation. Secondly, it provides the adsorption site for C2H3 intermediate, avoiding the strong adsorption on the surface O, thus reducing the energy barrier for the second step of C2H2 hydrogenation. This work provides valuable insights to design effective catalysts based on metal oxide.

Conflicts of Interest:
The authors declare no conflict of interest.