Cite this Article
Zhang Ji, Zhang De-Ming. First-principles study of oxygen adsorbed on Au-doped RuO2 (110) surface*

Project supported by the Natural Science Foundation of Anhui Province, China (Grant Nos. KJ2018A0588 and KJ2019A0879).

 . Chinese Physics B, 2019, 28(11): 116107  
Permissions
First-principles study of oxygen adsorbed on Au-doped RuO2 (110) surface*

Project supported by the Natural Science Foundation of Anhui Province, China (Grant Nos. KJ2018A0588 and KJ2019A0879).
Zhang Ji1, †, Zhang De-Ming1, 2
An Hui Xin Hua University, Hefei 230088, China
Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei 230031, China

 

† Corresponding author. E-mail: coolfall123@126.com

Project supported by the Natural Science Foundation of Anhui Province, China (Grant Nos. KJ2018A0588 and KJ2019A0879).

Abstract

Density functional theory calculations are carried out to identify various configurations of oxygen molecules adsorbed on the Au-doped RuO2 (110) surface. The binding energy calculations indicate that O2 molecules are chemically adsorbed on the coordinatively unsaturated Ru (Rucus) sites and the bridge oxygen vacancies on the Au sites. Transition state calculations show that O* can exist on the Rucus site by dissociation and diffusion. The calculations of the reaction path of CO indicate that the reaction energy barrier of CO adsorbed on Au with lattice oxygen decreases to 0.28 eV and requires less energy than that on the undoped structure.

PACS: 61.66.Dk;61.72.J-;61.72.Bb;61.72.uj
Keyword:CO;metal oxides;density functional theory;ruthenium oxide;O2 adsoption
1. Introduction

The physical and chemical properties of transition metal oxides are of great interest and importance for efficiently using catalysis, electrochemistry, and gas sensors. Because the surfaces of these oxides perform well compared with the bulk state, the important goal in such studies is to reveal the surface functionality, especially their interaction with a gas phase species on anatomic scale.[1,2]

In recent years, ruthenium dioxide (RuO2) has become a popular and convenient model system for investigating catalytic reactions. And CO oxidation has attracted much attention because it plays an important role in solving the exhaust gas issues generated from auto-mobiles and industries. In CO oxidation, the surface of metallic Ru is inactive, but after being highly exposed to O2, it displays excellent performance due to the formation of a RuO2 (110) film.[3–6] A Mars–van Krevelen mechanism and a Langmuir–Hinshelwood model were successively proposed for the explaining of the CO catalytic oxidation on the chemical surface.[7,8]In the Mars–van Krevelen model, reductants react with metal oxide lattice oxygen, and then the dissociative adsorption of O2 fills the resulting oxygen vacancies. The Langmuir–Hinshelwood model suggests that CO combines with dissociated O* into CO2. Therefore, O2 adsorption and dissociation are important processes for catalytic reactivity.[9,10] Recent studies have experimentally and theoretically reported that oxygen molecules are adsorbed on the stoichiometric RuO2 (110) surface.[11,12] On the other hand, to improve the properties of a catalyst, an effective strategy is to modify the metal oxides with other metal elements. For example, the (111) surface of ceria can be made reactive to CO oxidation by doping the surface with Au theoretically.[13] Because doped gold atoms become gold ions which play an important role in determining the CO catalytic oxidation, gold becomes a significant doping element. Therefore, it is interesting to study the oxygen adsorption and CO oxidation on Au-doped RuO2 (110) surface.

In this study, O2 adsorption, dissociation, and diffusion on the Au-doped Ru (110) surface are investigated by using density functional theory. First, the optimized configurations of molecular O2 adsorbed on the Au-doped RuO2 (110) surface are obtained. Then the adsorption energy and dissociation energy barrier are calculated. The reaction paths of CO adsorbed onto Au with the O* atom are also determined.

2. Computational method

Calculations based on the density functional theory (DFT) were analyzed in the generalized gradient approximation (GGA)[14] by using the Perdew–Burke–Emzerhof function for correlation energy. Plane-wave basis functions were included with a kinetic energy cutoff of 450 eV. The calculations were carried out by using the Brillouim zone sampled with 4 × 4 × 6 and 2 × 2 × 1 Monkhorst–Pack mesh k-points grid for bulk RuO2 and surface calculations, respectively. The structures were relaxed by using the BFGS algorithm until the forces on all atoms were <0.02 eV·Å−1. The first-principles calculations were performed by using the Cambridge Sequential Total Energy Package (CASTEP) code.[15]

Figure 1(a) shows the ball-and-stick model for the stoichiometric RuO2 (110) surface. In the bulk RuO2, there are six-fold coordinated Ru atoms and three-fold coordinated O atoms. However, two types of Ru atoms exist on the RuO2 (110) surface: one is the type of Ru atom, coordinatively unsaturated Rucus, which has a five-fold coordination and is connected by a three-fold coordinated oxygen O3f, and the other is the type of Ru atom, Rubr, which has a six-fold coordination and is linked by a two-fold coordinated bridging oxygen Obr. Because the oxygen-binding energy is insensitive to slab thickness,[16] two (RuO2)4 layer slabs were used for the calculations. The slabs were separated by a 15-Å vacuum in the c-axis direction. In the present study, one of the Rucus atoms was replaced by one Au atom as illustrated in Fig. 1(b).

Fig. 1. (a) Structure of RuO2 (001) surface, showing ball-and-stick representation of stoichiometric RuO2 (001) surface with red ball representing O atom and cyan balls denoting ruthenium. (b) Top-view of Au-doped RuO2 (001) surface.

Oxygen dissociation and diffusion barriers were calculated by using the Climbing Image Nudged Elastic Band (Cl-NEB) method.[17,18] Five intermediate images were generated between the geometry-optimized initial and final state by linear interpolation. Minimum energy reaction paths were obtained by simultaneously relaxing each image. Vibrational frequencies were calculated through the diagonalization of the Hessian matrix by displacing the atoms by 0.01 Å from their equilibrium positions. The adsorption energy of molecular O2, the oxygen vacancy formation energy, and the charge density difference were calculated below.

The adsorption energy was evaluated from

where Emod+surf represents the total energy of the surface combined with the adsorption molecules, Esurf is the total energy of the substrate, and Emol is the total energy of molecules in the gas phase.

The oxygen vacancy formation is described by the reaction .[19] The oxygen vacancy formation energy Ef, was computed from the following equation

where E(Ru1−xAuxO2−δ) indicates that the energy of the bulk or Ru1−xAuxO2 surface, E(O2) is the energy of gas-phase O2, and E(Ru1−xAuxO2) is the energy of the bulk or surface in the presence of one oxygen vacancy.

The charge density difference ρdiff was obtained from the following expression

where ρmol+surf is the charge density of the adsorbed species on the surface, and ρsurf and ρmol are the charge densities of the bare surface and the adsorbate, respectively.

3. Results and discussion
3.1. Au-doped Ru0.875Au0.125O2 and surface

RuO2 crystallizes into a rutile structure with lattice parameters a and c. To ensure the reliability of the computational results, the bulk lattice constants are predicted and compared with the experimental value. The calculated lattice parameters of the pure RuO2 are a = 4.53 Å and c = 3.13 Å, which accord with the results of Sun (a = 4.52 Å and c = 3.13 Å) obtained through a DFT–GGA pseudopotential study.[20] The theoretical values are similar to the experimental values of 4.492 and 3.306.[21,22] The geometries of stoichiometric and reduced-bulk Au-doped RuO2 are optimized until the external and internal degree of freedom can relax to the force and stress vanished in the bulk calculations. The lattice constants change after the Au atoms have been doped into the bulk RuO2, the predicted lattice constants of bulk 12.5% Au-doped RuO2 are a = 4.58 Å and c = 3.20 Å, which are slightly larger than those of pure RuO2 because of larger radius of the Au atom. The optimized Ru0.875Au0.125O2 presents geometric distortion with an average length of the Ru–O bonds, 1.988 Å, while that of Au–O bonds is 2.163 Å. The RuO2 (110) surface that is built below will use these lattice constants.

The relaxed structure of the Au-doped RuO2 (110) surface in which one Ru atom is replaced by one Au atom is shown in Fig. 1(b). The bond length between Ru-cus and the O atom below it decreases from 1.962 Å in the bulk to 1.865 Å, which are similar to the values calculated by Hong et al..[23] On the contrary, the bond-length between Au-cus and the O atom below it increases from 2.049 Å in the bulk to 2.331 Å. Therefore, the Au atom is located at the position above the plane with the Ru-cus and O3f.

3.2. Structures of molecular O2 adsorption

As is well known, O2 can bind to the transition metal center. In the present study, there are four sites for O2 adsorption: Rucus, Au, and O vacancies A and B as shown in Fig. 1(b). The A site is formed by the vacancy of Obr near Rucus, and the B site is related to the vacancy of Obr near the Au atom on the surface. Moreover, the formation energy Ef of the oxygen vacancy is calculated. The Obr vacancy on the A site has much smaller formation energy (EfA = 2.41 eV) than the undoped RuO2 (110) surface (2.89 eV), indicating that the O vacancy formation is very much facilitated by Au doping. However, the formation energy for the B site (EfA = 3.06 eV) is bigger than that in the undoped case. Therefore, the Au dopant may make the oxygen vacancy formed easier at A site than at B site.

Figure 2 illustrates the DFT-optimized bound on the Au-doped RuO2 (110) surface and density for adsorbed at the Rucus and the Au sites. Binding energy values, O–O bond lengths, Bader charges, and O–O stretching frequencies are summarized in Table 1. First, we consider that binds at a single Rucus site in a bent fashion. There are two configurations in this case. The first situation is that is adsorbed vertically on one Rucus across the coordinatively unsaturated row with Ru–O–O at an angle of approximately 122.8°, and the O–O bond length is 1.284 Å, slightly greater than the O–O bond length of gas-phase molecular O2. The O–O stretching vibrational frequency is 1229.4 cm−1 red-shifted by about 352 cm−1 from molecular O2.[24] The Bader charge analysis shows a 0.22-e− charge transfer from the Rucus cation to the . The charge transfer is illustrated in the density plot marked as the blue area between O and Ru atoms. These features are all consistent with those of a superoxo-like which is characterized by vibrational frequencies near 1140 cm−1 and O–O bond lengths of 1.33 Å.[25]

Fig. 2. Representative configurations and charge density differences of molecular oxygen adsorption involving a br vacancy, cus and Au sites.
Table 1.

Values of adsorption energy (ΔEO2), bond length (dO−O), Bader charge of O2 ( ), and O–O stretching vibrational frequency (ν).

.

According to Wang et al.’s study,[16] O2 can be adsorbed molecularly across two adjacent coordinatively unsaturated vacancies. In the present study, O2 is adsorbed molecularly across Rucus and Au as shown in Fig. 2(b). However, the bond length of O–Au is 2.626 Å, larger than that of O–Rucus at 1.927 Å. Additionally, the density plots of this configuration show that more charges transfer from Rucus to O* than those from Au to O*, which indicates that O is chemically adsorbed onto the Rucus stronger than onto the Au atom. The calculated Au–Rucus– binding energy is −1.78 eV, larger than that of the adsorption at a single Rucus site. The features of with a bond length of 1.31 Å and O–O stretching vibrational frequency of 1165.8 cm−1 are consistent with those of a superoxo-like state. Figure 2(c) shows that the DFT-optimized is bound at a single Au site. The O–O distance and stretching vibrational frequency are 1.267 Å and 1317.7 cm−1, respectively. In this case, the binding energy is only −0.44 eV and the density plot in Fig. 2(c) also illustrates that less charges transfer from Au to O2, so this adsorption maybe physical adsorption.

Next, we consider O2 adsorbed across adjacent vacant bridge A or B and Rucus or Au site as shown in Figs. 2(d) and 2(e). The binding energy of adsorbed across O vacancy B and Rucus site is calculated to be −7.45 eV. Bader analysis indicates the bridge- to be reduced by 0.77 e−. Furthermore, the O–O bond is lengthened to 1.413 Å, and O–O stretching vibrational frequency is red-shifted to 740.7 cm−1. These features are consistent with those of peroxide, which typically exhibits vibrational frequency closer to 850 cm−1 and a bond length of 1.49 Å. By contrast, the adsorbed across O vacancy B and Au site is superoxo-like with an O–O bond length of 1.33 Å and a vibrational frequency of 1078.7 cm−1.

3.3. Dissociation of and diffusion of O*

The minimum energy path from molecular to dissociated oxygen O*, calculated by CI-NEB method, is illustrated in Fig. 3. When the adsorbed across vacant bridge B and Au sites dissociates into two O* atoms, the O–O bond is stretched to 2.05 Å at the transition state (TS) as shown in Fig. 3(a). The calculated dissociation barrier for this case is 1.22 eV, much less than the desorption energy. By contrast, the dissociation barrier of across vacant bridge A and Rucus is only 0.13 eV, which implies that dissociation is energetically more facile.

Fig. 3. Minimum energy path for O2 dissociation at br vacancy B–Au and A–Rucus and diffusion of O* between Rucus and Au. Cross-sectional structures are shown for initial state (IS), transition state, and final state (FS), with gray balls representing Ru, golden balls Au, and red balls oxygen.

The dissociation of molecular causes a single oxygen O* to be adsorbed on Au or Rucus site. Furthermore, O* can diffuse from Au to Rucus or in the reverse direction. The minimum energy paths of diffusion are calculated and shown in Figs. 3(c) and 3(d), respectively. The diffusion barrier of O* from Au to Rucus is only 0.19 eV, much less than the diffusion barrier (2.84 eV) of O* from Rucus to Au. The difference in diffusion barrier shows that O* is easier to diffuse from Au to Rucus. Therefore, single oxygen O* is adsorbed on the Rucus with greater probability.

3.5. CO oxidation on Au-doped RuO2 (110) surface

The experimental study indicates that CO oxidizes readily into CO2 on the RuO2 (110) surface,[26–28] which is also corroborated by many theoretical investigations.[29,30] Previous studies show that CO is first adsorbed on the stoichiometric RuO2 (110) surface, then reacts with the O-bridge atom. Finally, the CO2 is desorbed with an O-bridge vacancy remaining. In the present study, two CO oxidation processes happening on the Au-doped RuO2 (110) are considered. One is that the CO adsorbed on Au reacts with O-bridge to form CO2, and the other is that the CO on Au combines with the O* adsorbed on the neighboring Rucus. The calculated relevant energy values and reaction paths are summarized in Fig. 4. There are two stable configurations for CO to be adsorbed on the Au atom. The molecular CO is adsorbed on the Au atom obliquely toward the O-bridge as shown in Fig. 4(a), which is chosen as an initial state to calculate the reaction path. Along the reaction coordinates, the C atoms from CO and the neighboring O-bridge move towards each other, reducing the distance from 3.57 Å to 1.98 Å at the transition state (TS). At the TS, the product-like OCO is formed with the ∠OCO ≈ 132.6°. This conversion of CO to CO2 has an energy barrier of 0.28 V, less than the value for the undoped case (Ea = 0.7 eV).[31] In addition, The reaction of CO adsorbed vertically on Au atom with O* on the Rucus is considered as shown in Fig. 4(b). During the transition of the C atom from CO to O*, the O and Rucus together move close to the C atom. And the distance between C and O* is reduced from 3.38 Å to 1.64 Å at TS. This reaction is much more exothermic than reaction in Fig. 4(a).

Fig. 4. Calculated potential energy diagram for CO oxidation on a gold atom doped in RuO2 (110) surface.
4. Conclusions

The DFT calculation is carried out to describe the molecular oxygen adsorption and reaction chemistry of CO on the Au-doped RuO2 (110) surface. The O2 can be adsorbed on Rucus, Au, and oxygen bridge vacancies with different configurations. In comparison, the O2 binding energy at the oxygen bridge vacancies is the largest, exceeding −6.98 eV, while the binding energy at the Au atom is only −0.44 eV. And it is found that the configurations can all dissociate at available neighboring vacant sites, which leaves a single O* adsorbed at the Rucus and Au site. The calculation results indicate that the dissociated O* is easier to diffuse from Au to Rucus site than from Rucus to Au. Finally, the CO reaction paths are obtained by DFT. The results show that the reaction energy barrier of CO adsorbed at Au with lattice oxygen decreases to 0.28 eV. However, the reaction of CO with O* at Rucus site requires more energy.

Reference
[1] Henrich V E Cox P A 1994 The Surface Science of Metal Oxides Cambridge Cambridge University Press
[2] Noguera C 1994 Physics and Chamistry at Oxide Surfaces Science Cambridge Cambridge University Press
[3] Over H Kim Y C Seisonen A P Wendt S Lundgren E Schmid M Varga P Morgante A Ertl G 2000 Science 287 1474
[4] Fan C Y Wang J Jacobi K Ertl G 2001 J. Chem. Phys. 114 10058
[5] Wang J Fan C Y Jacobi K Ertl G 2001 Surf. Sci. 481 113
[6] Stampfl C Schwegmann S Over H Scheffler M Ertl G 1996 Phys. Rev. Lett. 77 3371
[7] Monkhorst H J Pack J D 1976 Phys. Rev. 13 5188
[8] Assmann J Narkhede V A Breuer N Muhler M Seitsonen A P Knapp M Crihan D Farkas A Mellau G Over H 2008 J. Phys. Condens. Matter. 20 184017
[9] Ertl G Kno H Weitkamp J 1997 Handbook of Hetero-geneous Catalysis New York Wiley
[10] van Santen R Neurock M 2006 Molecular Heterogeneous Catalysis Weinheim Wiley-VCH
[11] Kim Y D Seitsonen A P Wendt S Wang J Fan C Jacobi K Over H Ertl G 2001 J. Phys. Chem. 105 3752
[12] Rössler M Günther S Wintterlin J 2007 J. Phys. Chem. 111 2242
[13] Shapovalov V Metiu H 2007 J. Catal. 245 205
[14] Perdew J P Burke K Ernzerhof M 1996 Phys. Rev. Lett. 77 3865
[15] Segall M D Lindan P J D Probert M J Pickard C J Hasnip P J Clarck S J Payne M C 2002 J. Phys.: Condens. Matter 14 2717
[16] Wang H Schneider W Schnidt D 2009 J. Phys. Chem. 113 15266
[17] Henkelman G Uberuaga B P Jónsson H 2000 J. Chem. Phys. 113 9901
[18] Henkelman G Jónsson H 2000 Chem. Phys. 113 9978
[19] Yang Z Woo T K Hermansson K 2004 Chem. Phys. Lett. 396 384
[20] Sun Q Rarsten K Scheffler M 2004 Phys. Rev. 70 235402
[21] Boman C E 1970 Acta Chem. Scand. 24 116
[22] O’Grady W E Atanasoska L Pollak F H Park H L 1984 J. Electroanal. Chem. 178 61
[23] Hong S Rahman T S 2007 J. Phys. Chem. 111 12361
[24] Huber K P Herzherg G 1079 Molecular Spectra and Molecular Structure IV. Constants of Diatomic: Molicules New York Van Nostrand Reinhold
[25] Kung H 1989 Transitition Metal Oxides: Surface Chemistry and Catalysis New York Elsevier
[26] Reuter K Scheffler M 2003 Phys. Rev. Lett. 90 046103
[27] Wang Y Lafosse A Jacobi K 2002 J. Phys. Chem. 106 5476
[28] Wang J Fan C Y Jacobi K Ertl G 2002 J. Chem. Phys. 106 3422
[29] Reuter K Scheffler M 2003 Phys. Rev. Lett. 90 046103
[30] Reuter K Frenkl D Scheffler M 2004 Phys. Rev. Lett. 93 116105
[31] Wendt S Seitsonen A P Over H 2003 Catal. Today 85 167