Theoretical studies on the monomeric vanadium oxides supported by ceria: the atomic structures and oxidative dehydrogenation activities

Abstract

The relative stabilities of different CeO₂(111)-supported VO_x are compared by calculating the phase diagrams, and a thermodynamically more stable VO₂ type monomeric species is located. Studies based on H adsorption and O vacancy formation suggest high activities of the determined VO_x (x = 2–4) species in oxidative dehydrogenation reactions.

---

Catalysts based on ceria-supported vanadium oxides are extremely active in selective oxidative dehydrogenation (ODH) reactions.^1–4 Great efforts have been devoted to resolving the structure–activity correlation of the vanadia/ceria (VO_x/CeO₂) catalysts. High-resolution scanning tunneling microscopy (STM) studies found that the vanadium oxide species are highly dispersed as monomers on ceria support at very low vanadia loadings, indicating that strong interaction exists between vanadia and ceria.⁵ It has also been found that increasing the loadings gives rise to a higher density of monomers firstly, and then the monomers will aggregate to larger oligomers at further increased loadings. Subsequent studies further pointed out that the ODH activity of VO_x/CeO₂ is affected by the loadings of vanadia, with VO_x monomers giving much better performance than other bigger oligomers.^2–4 However, even the VO_x monomers are rather complicated with different experimental conditions leading to much different structures and the associated activities.^6,7

Metiu and co-workers first systematically studied the atomic structures of CeO₂(111) supported monomeric VO_x.⁶ The charges of the different atoms in the vanadia clusters and the vanadium 2p core-level energies were calculated. However, the pristine density functional theory (DFT) calculations used in the above study often fails in accurately describing Ce 4f electrons.⁸ Sauer and co-workers then calculated the structures of similar systems and provided the phase diagram of the relative stabilities of different VO_x species by using DFT with a Hubbard U correction (DFT + U).^7,9 In their work,⁷ sub-surface O-vacancies of the ceria substrate were also considered in studying the possible compositions of the supported system, while the cases related to top surface O-vacancies were not systematically studied. It should be noted that occurrence of top surface O-vacancies beside vanadia clusters would induce strong relaxations to the clusters. Such relaxations can be probably beneficial to the stabilities of some configurations of supported vanadia. Therefore, in this paper, we intend to further study the structures of CeO₂(111) supported monomeric VO_x and refine the phase diagram by using DFT + U calculations. To avoid the interactions between periodic surface vanadia clusters, we used a 4 × 4 surface supercell for CeO₂(111) support, which ensures large distances (>10 Å) between neighboring clusters. van der Waals (vdW) interaction is also important to accurately describe the interaction between vanadia clusters and the CeO₂(111) support. So we employed the DFT-D2 scheme¹⁰ which can introduce good London dispersion corrections in the calculation (see computational details in ESI†). Finally, properties of such supported catalysts related to the ODH activity were also studied to illustrate the structure–activity correlation at atomic level.

A systematic search for stable structures of CeO₂(111) supported monomeric VO_x (hereafter called VO_x) needs to consider the synergetic change of the configuration and the oxygen content of the surface vanadia cluster. By taking into account these issues, we constructed different vanadia clusters sitting at stoichiometric CeO₂(111). In the up panels of Fig. 1a, we illustrate the most stable structures with given compositions of VO_x (x = 1–4) directly located in this work (VO_x-A structures). These structures are quite similar to those previously reported^6,7 (several second most stable structures are also reported in Fig. S2 and Table S1†). As one can see from the calculated structure, VO-A occurs in a way that a VO group sits on-top of one sub-surface O, and as the result, three top surface O atoms binding to the VO group are slightly lifted up. Our calculations also showed that the V atom loses all its valence electrons to become a V⁵⁺ and three Ce³⁺ cations are generated nearby (see Fig. S1†). For the bent VO₂ group sitting on the ceria surface (see VO₂-A of Fig. 1a), its formation actually significantly raises the two surface O atoms it binds with, generating two the so-called pseudovacancies.¹¹ The V atom of such VO₂-A maintains its oxidation state of +5, and accordingly, only one Ce³⁺ is generated on the surface (see Fig. S1†). Previously reported stability analyses proposed that this structure is the most stable one for CeO₂(111) supported monomeric VO_x within the temperature range of 190–425 K at low oxygen pressure of 10⁻⁹ atm,⁹ in which a high methanol ODH activity was also observed.² VO₂-A was then taken as a key species of active VO_x, and many studies^2,7,11–13 employed this structure to explain the observed high ODH activity of vanadia/ceria. Larger VO_x clusters, i.e. VO₃-A and VO₄-A (see Fig. 1a), were determined to bind to the ceria surface through formation of V–O and O–Ce bonds, and their interface structures are quite similar. The only difference is that VO₃-A is terminated by a V=O group, while VO₄-A has an O₂-like species on the top. No Ce³⁺ is generated in these two structures.

Fig. 1 (a) Calculated structures of CeO₂(111) supported monomeric VO_x (x = 1–4). Ce³⁺ distributions of VO-A, VO-B, VO₂-A and VO₂-B are shown in Fig. S1.† Some selected bonds are labelled with their length in pm. V atoms are in grey, Ce in ivory and O above the CeO₂ surface in blue. O atoms of top- and sub-surface are in red and cyan, respectively. Some selected O atoms are numbered for further studies. O-vacancy is marked with “O_v”. These notations are used throughout the paper. (b) Oxygen pressure-temperature phase diagram of CeO₂(111) supported monomeric VO_x. (c) LDOS of the vanadia clusters (V and its neighboring O atoms) of VO₂-A (black) and VO₂-B (red).

The ceria support can affect the composition of the system via generating O-vacancies. In this work, only top surface O-vacancies were considered since sub-surface O-vacancies have been shown to have almost no effect on the stabilities of supported VO_x.⁷ Specifically, we removed the top surface O beside the vanadia clusters of VO_x-A to generate the corresponding new structures, i.e. VO_x−1-B (see Fig. 1a). Compared to the VO-A and VO₃-A structures, VO-B and VO₃-B obtained in this way are 1.16 and 1.05 eV less stable, respectively. By contrast, the newly obtained VO₂-B is 0.34 eV more stable in total energy than the well-known VO₂-A structure. As one can see from its structure (Fig. 1a), the formation of VO₂-B can be also taken as the result of the strong relaxation of one surface O to bind with the VO₂, leaving a pseudovacancy nearby. In the atmosphere with O₂, such O-vacancy of VO₂-B can adsorb the O₂ molecule with calculated adsorption energy of 0.68 eV, giving rise to the VO₄-C structure (see Fig. 1a). DFT calculations showed that VO₄-C is 0.52 eV less stable in energy than VO₄-A.

Then, for the most stable VO_x species with given compositions, i.e. VO-A, VO₂-B, VO₃-A and VO₄-A, their relative stabilities are illustrated through the phase diagram with variable temperatures (T) and oxygen pressures (p). The original VO₂-A structure was also taken into account when the diagram was plotted. From the calculated phase diagram (Fig. 1b), one may notice two obvious trends: (i) at one specific oxygen pressure, the most stable VO_x species shifts from VO₄-A to VO₂-B, and then to VO-A with increasing temperatures; and (ii) at one specific temperature, the most stable VO_x species shifts from VO-A to VO₂-B, and then to VO₄-A with increasing oxygen pressure. It is interesting to note that the structural evolution of supported VO_x skips over VO₂-A and VO₃-A according to this phase diagram, indicating that they are not the most favorable species at any conditions. It needs to be mentioned that some discrepancies exist for the relative stabilities of different VO_x species obtained in this work and those reported in previous thermodynamic study.⁹ Compared to the results of our calculations, stabilities of VO₃-A and VO₄-A (especially VO₄-A) were underestimated in previous study,⁹ which can be seen from the calculated reaction energies of VO₂-A + 0.5O₂ → VO₃-A and VO₃-A + 0.5O₂ → VO₄-A listed in Table 1. Accordingly, intersections among VO₂-A, VO₃-A and VO₄-A in the temperature phase diagram (p = 10⁻⁹ atm) of CeO₂(111) supported monomeric VO_x are all shifted to higher temperatures in this work compared to those in ref. 9 (see Fig. S3†). Repeated calculations of ref. 9 and calculations based on Heyd–Scuseria–Ernzerhof (HSE) hybrid functional¹⁴ gave similar reaction energies for the oxidative processes of different VO_x species (VO_x−1 + 0.5O₂ → VO_x) (see Table 1), which then further confirms the reliability of our new diagram. It also needs to be mentioned that the energy of VO₂-A is higher by ∼0.4 eV than that of VO₂-B in all these calculations. Moreover, the calculated electronic structures (local density of states, LDOS) in Fig. 1c also shows that the electronic states of the supported VO₂-B distribute at lower energies compared to those of VO₂-A (LDOS of surface O and Ce atoms are shown in Fig. S4†). In fact, in the current work, we also calculated the structure transformation between VO₂-A and VO₂-B. The calculated profile in Fig. 2 illustrates that the transformation from VO₂-B to VO₂-A through rotation of the vanadia cluster (see the six images in Fig. 2) has an energy barrier of 0.39 eV, while the inverse process from VO₂-A to VO₂-B is much easier with a barrier of only 0.05 eV, which further supports the significant stability of VO₂-B.

Table 1 Calculated reaction energies (ΔE) with different methods for the oxidative processes between VO_x species (VO_x−1 + 0.5O₂ → VO_x) and the transformation between two VO₂-type species

		This work	Ref. 9	This work	This work
Method		PBE + U (5.0 eV) + D	PBE + U (4.5 eV)	PBE + U (4.5 eV)	HSE
a0 (CeO2) (Å)		5.440	5.490	5.490	5.398
Surface cell		4 × 4	2 × 2	2 × 2	2 × 2
Cut-off energy (eV)		400	400	400	400
K-points		1 × 1 × 1	3 × 3 × 1	3 × 3 × 1	2 × 2 × 1
Number of atomic layers		9	9	9	9
ΔE (eV)	VO-A + 0.5O2 → VO2-A	−0.79	−0.79	−0.70	−0.88
	VO2-A → VO2-B	−0.34	—	−0.40	−0.44
	VO2-A + 0.5O2 → VO3-A	−0.86	−0.25	−0.78	−0.87
	VO3-A + 0.5O2 → VO4-A	−0.67	−0.40	−0.76	−0.73
	VO2-A + O2 → VO4-A	−1.53	−0.65	−1.54	−1.60

---

Fig. 2 Calculated structure transformation from VO₂-B to VO₂-A. TS represents the transition state. All six CI-NEB optimized images are given.

It should be noted that in surface reactions, it is possible that molecule adsorption can cause the structural transformation from VO₂-B to the less stable VO₂-A.¹⁵ Therefore we also systematically calculated the adsorptions at these two VO₂-type structures by using methanol as a probe molecule (see details in Fig. S5†) and located the optimal molecular and dissociative adsorption configurations (see Fig. 3). According to our calculations, methanol prefers to adsorb at the O-vacancy site of VO₂-B with its O atom partially filling the vacancy and H of OH group pointing toward the interface O atom. This configuration has the calculated adsorption energy of 1.46 eV, and the dissociation of methanol can readily occur with the activation barrier of 0.04 eV and the adsorption energy of dissociatively adsorbed methanol is 1.47 eV, nearly the same as that of molecular one. By contrast, methanol has rather low molecular adsorption energy of 1.00 eV at VO₂-A and it has to overcome a barrier of 0.47 eV to evolve to an endothermic dissociation state (dissociative adsorption energy: 0.58 eV). Accordingly, it can be suggested that molecule adsorption may not promote the structural transformation from VO₂-B to VO₂-A either.

Fig. 3 Calculated structures (top view) of molecular and dissociative adsorbed methanol at VO₂-A (left) and VO₂-B (right). C atoms are in brown and H in pink.

Atomic scale insights into the catalytic activities are crucial to better understanding these VO_x species. Recent theoretical studies on ODH reactions at vanadia/ceria suggested that adsorption energy of H may determine the reaction energy in the process of dehydrogenation of adsorbates.^2,4 Therefore, it can be expected to influence the reaction barrier according to the Bronsted–Evans–Polanyi (BEP) principle.¹⁶ Accordingly, adsorption energy of H can be used as one descriptor for ODH reactivities. Another descriptor for quick screening of highly active ODH catalysts is O-vacancy formation energy as it is crucial in removal of adsorbed H through H₂O formation and desorption. In addition, H diffusions also occur during the processes of H abstraction and H₂O formation. Therefore, in the current work, we systematically calculated the H adsorption energies, diffusion barriers as well as the O-vacancy formation energies at the most stable VO_x species with given compositions.

As one can see from the results listed in Table 2, the calculated adsorption energies of H at two different surface O of VO-A are 0.92 and 0.45 eV, respectively, which are both lower than that at clean CeO₂(111) (1.21 eV), while VO₂-B has higher H adsorption capacities than clean CeO₂(111) at several surface O. These results show that dehydrogenation may perhaps occur more easily at VO₂-B than at clean CeO₂(111). Nevertheless, the O-2 site (labeled in Fig. 1a) of VO₂-B exhibits rather weak H adsorption capacity. This is due to the combined effect of missing hydrogen bond with neighboring top-surface O and appearance of electrostatic repulsion between H⁺ and the bare Ce ions after H adsorption. Interestingly, VO₃-A gives much higher H adsorption energies than other VO_x species. It could be simply due to the fact that this surface is “oxidized” by the surface vanadia cluster as one V atom can only provide five electrons in maximum to the excess three O atoms, which may not be able to form the formal O²⁻, and the surface then has a strong tendency to capture electrons through H adsorption. For VO₄-A, the O₂ species on top of V was determined to be an O₂⁻ with a calculated Bader charge of −0.57e relative to a neutral O₂. VO₄-A then also has an excellent H adsorption capacity due to its strong electron affinity.

Table 2 Calculated H adsorption energies (E_ad[H]) at different O sites labelled in Fig. 1a, H diffusion barriers (E_b), and O-vacancy formation energies (E_ov) of the most stable VO_x species with given compositions (in eV). Note that the most favorable O-vacancy for VO species occurs as a sub-surface O-vacancy below the VO group. Results of VO₂-A are listed in Table S2

		VO-A	VO2-B	VO3-A	VO4-A
Ead[H]	O-1	0.92	1.35	2.52	0.99
	O-2	0.45	0.88	2.46	1.44
	O-3	—	1.30	2.46	1.44
	O-4	—	1.41	2.57	1.77
Eb	O-1 → O-2	1.71	1.69	1.66	0.54
	O-2 → O-1	1.24	1.22	1.60	0.99
	O-2 → O-3	—	1.36	1.79	1.88
	O-3 → O-2	—	1.78	1.79	1.88
	O-3 → O-4	—	0.02	0.06	0.01
	O-4 → O-3	—	0.12	0.17	0.34
Eov		2.58	1.13	0.52	0.67

---

For H diffusion, we found it is difficult to occur among different O within the vanadia clusters. By contrast, at the interfaces between vanadia clusters and the CeO₂ support, fast H diffusions (E_b < 0.40 eV) can occur, largely due to the fact that short O–O distances at these sites can avoid the complete breaking of O–H bonds.¹⁷ Since the reactants, such as methanol, prefer to adsorb at the interface sites as well (see Fig. 3 and S5†), such fast H diffusions may therefore contribute to the overall ODH activity through fast H diffusion after hydrogen abstraction to promote H₂O formation.

From the calculated O-vacancy formation energies listed in Table 2, we can conclude that O-vacancies at VO₂-B, VO₃-A and VO₄-A are more favorable to occur compared to that at top-surface of CeO₂(111) (E_ov = 2.58 eV). So, it can contribute to the elimination of adsorbed H from H abstraction through H₂O formation. Temperature-programmed desorption (TPD) measurements found a high activity for ODH of methanol to formaldehyde at low vanadia loadings where surface vanadia species are primarily monomers, indicated by a low temperature peak (around 370 K) for formaldehyde production under oxygen pressure of 10⁻⁹ atm.^2,3 From the thermodynamic analyses illustrated in Fig. S3†, we can find that VO₂-B is the most stable species at the corresponding condition of the aforementioned desorption peak. Nevertheless, the species which have excellent H adsorption and O-vacancy formation capacities, e.g. VO₃-A species, are only slightly less favorable than VO₂-B at such condition. So the contributions of them to the overall activity observed by the TPD experiment^2,3 should not be simply neglected. In addition, higher oxygen pressures in practical applications may further stabilize the active species with high oxygen content, and they could also be important for ODH activity.

In summary, we have systematically calculated the stable monomeric VO_x species with different O content, namely VO, VO₂, VO₃ and VO₄, supported at CeO₂(111). A new thermodynamically more stable VO₂-type species (VO₂-B) is located, and the relative stabilities of different supported VO_x are compared by plotting the phase diagrams. Detailed studies on two VO₂-type species indicate that the previously reported VO₂-A can easily convert to the more stable VO₂-B through rotation of its vanadia cluster, while VO₂-B does not prefer to transform to VO₂-A even when methanol adsorbs on the surfaces. Therefore, we suggest that VO₂-B is crucial for studying the VO_x species instead of VO₂-A. The calculated energetics of H adsorption and O-vacancy formation indicate higher ODH activities of the CeO₂(111) supported VO_x (x = 2–4) species compared to clean CeO₂(111). Specifically, VO₃-A has a rather high H adsorption energy, indicating its unique role in dehydrogenation processes. For H diffusion, it occurs easily at the interface sites, while it is inhibited among O of the vanadia clusters. The fast H diffusion at the interface sites, together with the low formation energies of O vacancies, also contribute to the H₂O formation and desorption in ODH. These findings further reveal the structure–activity relationship of the vanadia/ceria system, and can provide assistance to the model construction in illuminating the unusual catalytic activities of VO_x species.

Acknowledgements

This work was supported by the National Basic Research Program (2011CB808505), National Natural Science Foundation of China (21421004, 21322307), “Shu Guang” project of Shanghai Municipal Education Commission and Shanghai Education Development Foundation (13SG30) and the Fundamental Research Funds for the Central Universities (WD1313009). The authors also thank the computing time in the National Super Computing Center in Jinan.

References

1. I. E. Wachs, Catal. Today, 2005, 100, 79; W. Daniell, A. Ponchel, S. Kuba, F. Anderle, T. Weingand, D. H. Gregory and H. Knözinger, Top. Catal., 2002, 20(1–4), 65; M. N. Taylor, A. F. Carley, T. E. Davies and S. H. Taylor, Top. Catal., 2009, 52, 1660; T. Feng and J. M. Vohs, J. Catal., 2004, 221, 619.
2. M. V. Ganduglia-Pirovano, C. Popa, J. Sauer, H. Abbott, A. Uhl, M. Baron, D. Stacchiola, O. Bondarchuk, S. Shaikhutdinov and H.-J. Freund, J. Am. Chem. Soc., 2010, 132, 2345.
3. H. L. Abbott, A. Uhl, M. Baron, Y. Lei, R. J. Meyer, D. J. Stacchiola, O. Bondarchuk, S. Shaikhutdinov and H.-J. Freund, J. Catal., 2010, 272, 82.
4. J. Liu, X.-P. Wu, S. Zou, Y. Dai, L. Xiao, X.-Q. Gong and J. Fan, J. Phys. Chem. C, 2014, 118, 24950.
5. M. Baron, H. Abbott, O. Bondarchuk, D. Stacchiola, A. Uhl, S. Shaikhutdinov, H.-J. Freund, C. Popa, M. V. Ganduglia-Pirovano and J. Sauer, Angew. Chem., Int. Ed., 2009, 48, 8006.
6. V. Shapovalov and H. Metiu, J. Phys. Chem. C, 2007, 111, 14179.
7. C. Popa, M. V. Ganduglia-Pirovano and J. Sauer, J. Phys. Chem. C, 2011, 115, 7399.
8. S. Fabris, G. Vicario, G. Balducci, S. D. De Gironcoli and S. Baroni, J. Phys. Chem. B, 2005, 109, 22860; M. Nolan, S. C. Parker and G. W. Watson, Surf. Sci., 2005, 595, 223; M. Nolan, S. Grigoleit, D. C. Sayle, S. C. Parker and G. W. Watson, Surf. Sci., 2005, 576, 217.
9. C. Popa, M. V. Ganduglia-Pirovano and J. Sauer, J. Phys. Chem. C, 2012, 116, 18572.
10. S. Grimme, J. Comput. Chem., 2006, 27, 1787.
11. T. Kropp, J. Paier and J. Sauer, J. Am. Chem. Soc., 2014, 136, 14616; J. Paier, T. Kropp, C. Penschke and J. Sauer, Faraday Discuss., 2013, 162, 233.
12. B. Beck, M. Harth, N. G. Hamilton, C. Carrero, J. J. Uhlrich, A. Trunschke, S. Shaikhutdinov, H. Schubert, H.-J. Freund, R. Schlögl, J. Sauer and R. Schomäcker, J. Catal., 2012, 296, 120.
13. C. Penschke, J. Paier and J. Sauer, J. Phys. Chem. C, 2013, 117, 5274.
14. J. Heyd, G. E. Scuseria and M. Ernzerhof, J. Chem. Phys., 2003, 118(18), 8207.
15. Q. Cuan, J. Tao, X.-Q. Gong and M. Batzill, Phys. Rev. Lett., 2012, 108, 106105.
16. T. Bligaard, J. K. Nørskov, S. Dahl, J. Matthiesen, C. H. Christensen and J. Sehested, J. Catal., 2004, 224, 206; J. K. Nørskov, T. Bligaard, A. Logadottir, S. Bahn, L. B. Hansen, M. Bollinger, H. Bengaard, B. Hammer, Z. Sljivancanin, M. Mavrikakis, Y. Xu, S. Dahl and C. J. H. Jacobsen, J. Catal., 2002, 209, 275; A. Logadottir, T. H. Rod, J. K. Nørskov, B. Hammer, S. Dahl and C. J. H. Jacobsen, J. Catal., 2001, 197, 229.
17. X.-P. Wu, X.-Q. Gong and G. Lu, Phys. Chem. Chem. Phys., 2015, 17, 3544.

---

Footnote

† Electronic supplementary information (ESI) available: Computational details, structures and energies for the second most stable structures, oxidative dehydrogenation activities of VO₂-A, Ce³⁺ distributions around VO- and VO₂-type species, temperature phase diagram, LDOS of two VO₂ species, methanol adsorption at two VO₂ species. See DOI: 10.1039/c5ra08962j

---