On the Mechanism of Low-Temperature CO Oxidation on Ni(111) and

Nov 22, 2010 - that CO oxidation at the perimeter of O islands on Ni(111) cannot occur at low ..... adsorb above the center of three Ni atoms, to form...
0 downloads 0 Views 3MB Size
J. Phys. Chem. C 2010, 114, 21579–21584

21579

On the Mechanism of Low-Temperature CO Oxidation on Ni(111) and NiO(111) Surfaces Guowen Peng,† Lindsay R. Merte,‡ Jan Knudsen,‡ Ronnie T. Vang,‡ Erik Lægsgaard,‡ Flemming Besenbacher,‡ and Manos Mavrikakis*,† Department of Chemical and Biological Engineering, UniVersity of WisconsinsMadison, Madison, Wisconsin 53706, United States, and Interdisciplinary Nanoscience Center (iNANO) and Department of Physics and Astronomy, Aarhus UniVersity, DK-8000 Aarhus C, Denmark ReceiVed: September 6, 2010; ReVised Manuscript ReceiVed: October 30, 2010

Through an interplay between density functional theory and scanning tunneling microscopy, we investigated the mechanism of low-temperature CO oxidation on Ni(111) and NiO(111) surfaces. We systematically examined CO oxidation on different possible active sites. We find that sub- and full monolayers of O chemisorbed on Ni(111) surfaces play no significant role in low-temperature CO oxidation. We further show that CO oxidation at the perimeter of O islands on Ni(111) cannot occur at low temperatures. In contrast, we suggest that oxidized Ni(111) surfaces, i.e., NiO(111), can catalyze low-temperature CO oxidation when NiO(111) is saturated by O2. Our findings can rationalize low-temperature CO oxidation on Ni(111) surfaces that have been predosed with large amounts of oxygen, as observed in recent experiments. 1. Introduction CO oxidation on transition-metal surfaces is one of the most extensively studied reactions because of its technological importance (e.g., in automotive and industrial exhaust cleaning) and its use as a model system for addressing general scientific questions in heterogeneous catalysis.1-13 Under ultrahigh vacuum (UHV) and low gas pressure conditions, it is believed that CO oxidation on transition metals proceeds via the traditional Langmuir-Hinshelwood mechanism, where coadsorbed CO and O react to form CO2. At high pressures and in the presence of excess O2, it was found that Pt group metals, such as Ru,5,6,14 Pt,9,15 Pd,15,16 and Rh,15,17,18 are more catalytically active for CO oxidation. To understand the underlying mechanism under these conditions, much effort has been put and different active phases possibly responsible for the enhanced reactivity for CO oxidation were proposed.5-7,9,15-18 More specifically, Goodman and coworkers suggested that the active phase at atmospheric pressure and in excess O2 is a surface covered with a full monolayer (ML) of atomic O.5,14,15 The oxidation of gas-phase CO with one monolayer chemisorbed O on a Ru(0001) surface via the Eley-Rideal mechanism was studied by Stampfl and Scheffler using density functional theory (DFT).7 Over and co-workers, on the other hand, proposed that the O-rich Ru phase is ruthenium oxide, RuO2(110),6 and that CO oxidation on ruthenium oxide follows the Mars-van Krevelen mechanism,19 i.e., CO chemisorbs on RuO2(110) and reacts with neighboring lattice O of ruthenium oxide to produce CO2; the consumed lattice O is subsequently restored by oxygen from the gas phase. Similarly, surface oxides were proposed as the active phases for the enhanced CO oxidation reactivity on Pt,9,10 Pd,16 and Rh18 surfaces under O2-rich conditions. Recently, using a combination of scanning tunneling microscopy (STM), temperature-programmed desorption (TPD), and X-ray photoelectron spectroscopy (XPS), we studied the lowtemperature CO oxidation on Ni(111) and Au/Ni(111).20 We * To whom correspondence should be addressed. E-mail: manos@ engr.wisc.edu. † University of WisconsinsMadison. ‡ Aarhus University.

showed that an oxide is formed on both Ni(111) and Au/Ni (111) surfaces when the surfaces were predosed with large amounts of O2. CO can be oxidized at ∼120 K on both of these surfaces under these conditions. In this article, we present a detailed discussion of our density functional theory calculations on low-temperature CO oxidation mechanisms on different oxygen-covered Ni(111) and on NiO(111) surfaces. The CO oxidation activity on structures with a chemisorbed oxygen coverage between 1/3 and 1 ML, at the perimeter of local oxygen islands on Ni(111), and on oxidized NiO(111) surfaces are compared. We find that sub- and full monolayers of O chemisorbed Ni(111) surfaces play no significant role in lowtemperature CO oxidation. We further show that CO oxidation at the perimeter of O islands cannot occur at low temperatures. By using DFT+U calculations, we show that though the clean NiO(111) surface alone cannot catalyze low-temperature CO oxidation, an oxygen-saturated NiO(111) surface can play a role in catalyzing CO oxidation at low temperatures. 2. Methods Calculations were performed using the VASP code21,22 based on spin-polarized density functional theory. For CO oxidation on Ni(111) surfaces, the standard DFT calculations using ultrasoft pseudopotentials23 for electron-ion interactions and the generalized gradient approximation (GGA-PW91) for the exchange-correlation functional24 were performed. For CO oxidation on strongly correlated NiO surfaces, DFT+U calculations25-27 using the projector augmented wave (PAW) potentials28,29 and the generalized gradient approximation (GGA-PW91) were performed. The parameters describing the on-site Coulomb repulsion between the Ni 3d orbitals were chosen as U ) 6.3 eV and J ) 1 eV, which give the optimized properties for bulk NiO.30-32 The electron wave function was expanded using plane waves with an energy cutoff of 400 eV. A ferromagnetic (FM) and an antiferromagnetic (AFM) phase was adopted for Ni(111) and NiO(111), respectively. The climbing image nudged elastic band (CI-NEB) method33 was used to calculate the reaction barriers unless stated otherwise. Transition states were verified

10.1021/jp108475e  2010 American Chemical Society Published on Web 11/22/2010

21580

J. Phys. Chem. C, Vol. 114, No. 49, 2010

by calculating the Hessian matrix and identifying a single imaginary frequency. The Ni(111) surface was modeled by a four-layer slab with a (3 × 3)R30° surface unit cell separated from the next slab in the z direction by a vacuum equivalent to six atomic layers. A Ni(111)-(6 × 6) surface unit cell with four O atoms adsorbed on it was used to investigate CO oxidation at the perimeter of local oxygen atom islands on Ni(111). The NiO(111) surface was modeled by a slab with six NiO double layers and a vacuum equivalent to six double layers between two successive slabs. The p(2 × 2) O-terminated octopolar NiO(111) reconstruction with 3/4 ML O missing at the top layer and 1/4 ML Ni missing at the second layer34,35 was utilized in our study. The Brillouin zones of Ni(111)-(3 × 3)R30° and NiO(111)-(2 × 2) were sampled using (6 × 6 × 1) and (3 × 3 × 1) k-point meshes based on the Monkhorst-Pack scheme,36 respectively. The large Ni(111)-(6 × 6) surface unit cell was sampled with the Γ point only. Molecules were adsorbed on only one of the two exposed surfaces of the slabs. The electrostatic potential was adjusted accordingly.37 The two bottom-most Ni layers or NiO double layers were fixed during the relaxation. All structures were relaxed until the HellmannFeynman forces acting on the atoms were smaller than 0.05 eV/Å. The adsorption energy is defined as BE ) Eads Eclean - Egas, where Eads, Eclean, and Egas are the calculated total energy of the slab with adsorbate, the clean slab, and the adsorbate species in the gas phase, respectively. STM images were acquired at 100-120 K in constant-current mode using a home-built Aarhus scanning tunneling microscope38 mounted in an ultrahigh vacuum chamber with basepressure ∼1 × 10-10 Torr. The Ni(111) crystal was prepared by Ar+ sputtering and annealing at 1000 K. Oxygen exposure was carried out by backfilling the UHV chamber with the sample in the scanning tunneling microscopy (STM) block. 3. Results and Discussion 3.1. CO Oxidation on Ni(111) Predosed with 1/3 ML O. We studied CO oxidation on a Ni(111)-(3 × 3)R30°-O surface, which is an O-chemisorbed structure with the highest O coverage (1/3 ML) observed in experiments and reported in the literature.39,40 For the Ni(111)-(3 × 3)R30°-O surface, our calculations showed that O prefers to adsorb on a facecentered cubic (fcc) site, in agreement with experiments.41 The calculated O adsorption energy on that site is -2.19 eV; herein, O adsorption energy is referenced to the gas-phase energy of O2. O on a hexagonal close packed (hcp) site is slightly less stable, with an adsorption energy of -2.06 eV. We considered the adsorption of one CO molecule on the (3 × 3)R30°-O surface, which corresponds to a CO coverage of 1/3 ML. On the clean Ni(111) surface, CO prefers to adsorb on hollow sites (fcc and hcp) with an adsorption energy of ca. -1.92 eV, in excellent agreement with the result of a previous study.42 On the Ni(111)-(3 × 3)R30°-O surface, we find that CO prefers to adsorb on an unoccupied fcc site. The calculated CO adsorption energy on the 1/3 ML O predosed Ni(111) surface, -0.88 eV, however, is much smaller in magnitude than the adsorption energy (-1.92 eV) of CO on the clean Ni(111) surface. This indicates that the preadsorbed O destabilizes CO adsorption on Ni(111). Figure 1 shows the calculated potential energy surface for CO oxidation on the 1/3 ML O-predosed Ni(111) surface. Selected geometries along the reaction pathway are also illustrated. As shown in Figure 1, CO oxidation consists of three elementary processes, with two intermediate states involved.

Peng et al.

Figure 1. Potential energy surface for CO oxidation on a Ni(111) (3 × 3)R30° surface predosed with 1/3 ML atomic oxygen (O). Blue, red, and gray spheres represent Ni, O, and C atoms, respectively. The zero energy refers to the gas phase energy of CO and 1/2 O2. Top and side views of initial, final, intermediate, and transition states are shown as insets.

In the first intermediate state [inset (iii)], the longer C-O bond (1.32 Å) is above a top-fcc-bridge site and is nearly parallel to the surface. The O-C-O bond angle in that state is 132°. This intermediate structure is 0.19 eV less stable than the initial state (i). In the second intermediate structure [inset (iv)] the longer C-O bond parallel to the surface is slightly shortened to 1.26 Å, and the O-C-O bond angle is enlarged to 140°. The longer C-O bond in structure (iv) is nearly parallel to one Ni-Ni bond. Structure (iv) is 0.11 eV less stable than the initial structure (i). This structure is characteristic of a bent CO2 and is similar to the CO2 chemisorbed structure on Pt(111).8 As clearly shown in the potential energy surface, the most likely rate-determining step for CO oxidation is the elementary step from structure (i) to structure (iii), with an energy barrier of 0.83 eV. In the transition state [inset (ii)] of this step, CO is nearly at an atop site and O moves close to a bridge site. The activation of O from a hollow site to a bridge site was previously proposed as a decisive step for CO oxidation on O-chemisorbed Pt group surfaces.8,43 The elementary step transforming structure (iii) to (iv) involves the rotation of one C-O bond and is practically spontaneous. The desorption of chemisorbed CO2 from structure (iv) in the third step [i.e., from structure (iv) to (vi)] is also quite facile, with an energy barrier of 0.09 eV. Overall, the oxidation of CO on the 1/3 ML O-predosed Ni(111) is exothermic, with an energy gain of 0.33 eV. The reaction barrier of CO oxidation on Ni(111) (3 × 3)R30° predosed with 1/3 ML O is 0.83 eV, which suggests that the 1/3 ML O-predosed Ni(111) (3 × 3)R30° surface cannot be responsible for the low-temperature (∼120 K) CO oxidation observed experimentally.20 For a reaction to take place at a reasonable rate at ∼120 K the barrier would have to be ∼0.2 eV or less. 3.2. CO Oxidation on Ni(111) Predosed with 2/3 ML O. Considering the fact that a high O coverage, larger than 1/3 ML, was observed at the boundary regions between small domains in the STM experiments,20 we investigated CO oxidation on Ni(111) (3 × 3)R30° predosed with 2/3 ML O. For O adsorption on the Ni(111) surface, we find that both O atoms prefer to adsorb on fcc sites, with an adsorption energy of -1.53 eV per O atom. On the 2/3 ML O-predosed Ni(111) surface, CO prefers to adsorb on the free fcc site, with an adsorption energy of -0.32 eV, which is 0.56 eV smaller in

Low-Temperature CO Oxidation on Ni(111) and NiO(111)

J. Phys. Chem. C, Vol. 114, No. 49, 2010 21581

Figure 2. Potential energy surface for CO oxidation on a Ni(111) (3 × 3)R30° surface predosed with 2/3 ML atomic oxygen (O). The zero energy refers to the gas-phase energy of CO and O2. Top and side views of initial, transition, and final states are shown as insets.

magnitude than on the 1/3 ML O-predosed Ni(111). This suggests the higher the coverage of O preadsorbed on Ni(111) surfaces, the weaker the CO adsorption. Figure 2 shows the potential energy surface for CO oxidation on the 2/3 ML O-predosed Ni(111) surface. Interestingly, no intermediate states similar to structures (iii) and (iv) shown in Figure 1, pertaining to the 1/3 ML O-predosed surface, were found. The reaction is a one step process: CO moves from the initial fcc site to a bridge site to react with one O (see the transition state in the inset of Figure 2). The energy barrier for CO oxidation on Ni(111) surface predosed with 2/3 ML O is 0.79 eV, which is slightly smaller than on Ni(111) surface predosed with 1/3 ML O. This is expected because CO adsorption is destabilized by the second adsorbed O. Though CO oxidation on the 2/3 ML O-predosed Ni(111) is exothermic with an energy gain of 2.2 eV, the high energy barrier suggests that the Ni(111) surface predosed with 2/3 ML O cannot be responsible for the observed low-temperature CO oxidation.20 3.3. CO Oxidation on a Full ML O-Predosed Ni(111). At full monolayer O coverage, all fcc sites of the Ni(111) surface are occupied by O, with an adsorption energy of -0.81 eV per O atom. The full monolayer O is expected to completely block CO adsorption on the surface, and it is likely that the gas phase CO directly reacts with the predosed O to produce CO2 via an Eley-Rideal mechanism. To address this possibility, we investigated CO oxidation on a full monolayer O-covered Ni(111) surface with CO starting directly above an fcc site. We followed the procedure described in ref 7. We constructed a high-dimensional potential energy surface using two variables ZC and ZOsthe vertical position of C (of the gas-phase CO molecule) and that of the surface O (below the C atom) relative to the remaining oxygen atoms, as schematically illustrated in the inset of Figure 3a. To construct the high-dimensional potential energy surface, the CO was put directly above one O adatom, and the coordinates of the C atom and the adsorbed O underneath it were fixed. The remaining atoms, except Ni atoms in the two bottom-most layers of the slab, were allowed to relax. Figure 3a is the resulting contour plot of the high-dimensional potential energy surface for CO oxidation on 1 ML O-covered Ni(111) via the Eley-Rideal mechanism with CO approaching one surface O above an fcc site. It is interesting to notice that as CO approaches the surface the O below the incoming CO molecule moves inward toward the surface due to repulsion between CO and the surface. As CO continues to move further

Figure 3. Contour plot of high-dimensional potential energy surface for CO oxidation on 1 ML O-predosed Ni(111) via the Eley-Rideal mechanism with CO approaching O (a) above an fcc site and (b) above a neighboring bridge site. The potential energy surface is constructed using two variables ZC and ZO, as shown in the insets (a and b). A cross-section view of the transition state is also shown. The estimated energy barriers for CO oxidation are 1.2 and 0.5 eV, in (a) and (b), respectively. (c) Two-dimensional potential energy surface for CO oxidation along the channels shown in (a) and (b).

toward the surface, the repulsion becomes attraction and the O below CO moves outward. As shown in Figure 3a, the energy barrier for CO approaching the surface above an fcc site (i.e., directly above an adsorbed O atom) via the Eley-Rideal mechanism is ca. 1.2 eV. In the transition state (see Figure 3a),

21582

J. Phys. Chem. C, Vol. 114, No. 49, 2010

Figure 4. Potential energy surface for CO oxidation at the perimeter of a local (2 × 2) O island in a large Ni(111)-(6 × 6) surface unit cell. The elementary step with the largest barrier is the reaction of CO with one adsorbed O, resulting in CO2 adsorbed on the surface with a structure similar to that shown in the inset (iv) of Figure 1.

the reactive surface O moves outward from the surface 0.3 Å and the distance between C and the surface O is 1.8 Å. We further examined CO oxidation via an Eley-Rideal mechanism with CO starting above a bridge site. The calculated contour plot of the potential energy surface is shown in Figure 3b. As in the previous case, we found that when CO comes close to the surface (ca. 3 Å), the reactive surface O moves toward the surface due to the repulsion between CO and the surface. As CO moves closer to the surface the repulsive interaction becomes attractive and the surface O moves outward from the surface. Figure 3b shows that the energy barrier for CO oxidation via the Eley-Rideal mechanism when CO approaches the surface above a bridge site is about 0.5 eV, which is 0.7 eV smaller than when CO reacts with the fcc O directly below. The side view of the transition-state geometry is shown in Figure 3b. In the transition state, the reactive surface O moves ca. 0.2 Å out from the surface; the distance between C and the active O is about 1.76 Å, and the O-C-O bond angle is 121°. Summarized in Figure 3c is the two-dimensional potential energy surface of CO oxidation on a full monolayer O-covered Ni(111) surface through these two Eley-Rideal channels. The overall reaction of CO oxidation on a full monolayer O covered surface is exothermic with an energy gain of 3.76 eV which is larger than on 1/3 and 2/3 ML O-predosed surface. The energetically favorable route for CO oxidation consists of CO approaching the surface above a bridge site, with an energy barrier of 0.5 eV. This barrier is smaller than the barrier calculated for lower-coverage O-predosed Ni(111) surfaces but is still too large to account for CO oxidation at ∼120 K, as observed experimentally.20 3.4. CO Oxidation on Ni(111) with Local O Islands. Previous studies4,11 showed that, on Pt(111), CO oxidation occurs at the boundary between CO-c(4 × 2) and O(2 × 2) domains. CO is more mobile than O and moves to the perimeter of O islands where the reaction takes place. To examine if local O islands might facilitate low temeprature CO oxidation on Ni(111), we considered a local O island formed by four O atoms in a large Ni(111)-(6 × 6) surface unit cell. We compared the relative stability of different O patches on the surface and found that the most stable patch is a local (2 × 2) O island (geometry shown in the inset of Figure 4). This structure is 0.17 eV more stable than the structure with four O uniformly distributed on the Ni(111)-(6 × 6) surface. The adsorption energy of O at the

Peng et al.

Figure 5. (a-d) STM images of the Ni(111) surface following exposure to the indicated quantities of O2 at 100 K. The inset in (c) shows a NiO(111) island formed upon oxidation at room temperature. (a) 200 × 200 Å2, (b) 200 × 200 Å2, (c) 1000 × 1000 Å2, inset 25 × 25 Å2, (d) 1000 × 1000 Å2.

fcc sites forming the local (2 × 2) O island is -1.31 eV per O atom. For CO adsorption on the Ni(111) surface with a local (2 × 2) O island, we found the most favorable site is a next nearest-neighbor hcp site (denoted as h in the (2 × 2) island structure shown in Figure 4), with a CO adsorption energy of -1.88 eV. This value is very close to the CO adsorption energy on clean Ni(111), which is expected because CO and the nearest adsorbed O are next nearest-neighbors and the repulsion between them is almost negligible. When CO adsorbs on a nearestneighbor fcc site (f), the magnitude of the adsorption energy decreases by 0.19 eV due to the repulsion between CO and O. On Ni(111)-(6 × 6) with a local (2 × 2) O island, we found that CO2 can adsorb on the surface, with a geometry similar to that shown in the inset (iv) of Figure 1. The calculated potential energy surface for CO oxidation on the Ni(111) surface with a local (2 × 2) O island is shown in Figure 4. The reaction begins with the diffusion of CO toward the O island. The CO diffusion barrier is 0.20 eV, indicating that CO is rather mobile. The most likely rate-determining step is the reaction of CO with O leading to adsorbed CO2. The energy barrier of that step is 1.43 eV, which is much larger than in the respective elementary step on the 1/3 ML O predosed Ni(111) surface. Desorption of CO2 is an easy step, with an energy barrier of 0.15 eV. The reaction of CO with O on this model surface is endothermic when an energy penalty of 0.82 eV. The high energy barrier together with the endothermic nature of the reaction suggests that the local (2 × 2) O island on Ni(111) surfaces plays no role in the low-temperature CO oxidation observed experimentally.20 3.5. CO Oxidation on NiO(111). STM images of the Ni(111) surface following exposure to different quantities of O2 at ∼100 K are displayed in Figure 5. Initially, as depicted in Figure 5a, following a dose of 10 L O2 (1 L ) 1 × 10-6 Torr · s), the oxygen forms a dissociatively chemisorbed O overlayer, characterized by patches of the (3 × 3)R30° phase separated by less-well-ordered patches of higher O density. After exposure to 20 L O2, we observe that the O-adatom phase is still dominant on most of the surface, but islands begin to form predominantly at step edges and also on the Ni(111) terraces. Further exposure to O2 leads to continued growth of this phase away from the step edges and across the surface, as shown in

Low-Temperature CO Oxidation on Ni(111) and NiO(111) Figure 5d. Although atomic resolution could not be obtained on the islands formed upon oxidation of Ni(111) at low temperatures, the characteristic 3 Å hexagonal structure of NiO(111) could be observed on similar islands formed by exposure to O2 at room temperature, as depicted in the inset of Figure 5c. It is thus clear that even at temperatures as low as 100 K, a poorly ordered film of NiO can be grown on the Ni(111) surface (for further details, see ref 20). We have, therefore, examined in our calculations whether NiO may be responsible for the observed low-temperature CO oxidation. The rock-salt NiO bulk has alternating cationic and anionic layers along the (111) direction. From the electrostatics perspective, the simply cleaved NiO(111) surface, with an ideal stoichiometry, is polar and has a diverging electrostatic surface energy and is thus unstable.44 A p(2 × 2) octopolar reconstruction, however, can cancel the divergence of the electric field in the crystal and stabilize the NiO(111) surface.34,35 In this study, we used a p(2 × 2) O-terminated octopolar NiO(111) surface with 3/4 ML O missing in the top layer and 1/4 ML Ni missing in the second layer (Figure 6). To examine if a clean NiO(111) surface can catalyze CO oxidation by itself, we first studied CO adsorption on a clean NiO(111)-p(2 × 2) surface. Our calculations showed that CO prefers to adsorb on a Ni atop site, as shown in Figure 6a, in agreement with experiments.45 The calculated adsorption energy is -0.46 eV. To oxidize the adsorbed CO, the lattice apex O atom of the NiO(111)-p(2 × 2) octopolar surface might be used. However, the sum of the total energy of a fully relaxed NiO(111)-p(2 × 2) surface without the lattice apex O and that of a gas phase CO2 is 0.9 eV higher than that of NiO(111)-p(2 × 2) with CO adsorbed at an atop Ni site. This indicates that the reaction is endothermic and the energy barrier of CO oxidation on the clean NiO(111)p(2 × 2) is at least 0.9 eV, if such a reaction channel exists. Therefore, we suggest that the clean NiO(111)-p(2 × 2) surface cannot catalyze low-temperature CO oxidation. At the high O2 exposures (600 L) used in the low-temperature CO oxidation experiments,20 O2 molecules adsorb on NiO(111)p(2 × 2) surfaces prior to CO adsorption. We calculated O2 adsorption on NiO(111)-p(2 × 2) and found that O2 prefers to adsorb above the center of three Ni atoms, to form three Ni-O

J. Phys. Chem. C, Vol. 114, No. 49, 2010 21583

Figure 6. Top and side views for (a) CO and (b) O2 adsorbed on a NiO(111)-p(2 × 2) octopolar reconstruction. The parallelogram drawn with dashed lines indicates the (2 × 2) supercell. Blue, gray, and red (pink) spheres represent Ni, C, O atoms, respectively.

bonds (Figure 6b). The calculated O2 adsorption energy is -0.74 eV. We began examining CO oxidation on the O2 preadsorbed NiO(111)-p(2 × 2) surface by studying CO adsorption on the surface. Interestingly, our calculations showed that the incoming CO spontaneously reacts with the lattice apex O, with a large energy gain of -2.5 eV, because all Ni atoms are blocked by the preadsorbed O2. The relaxed structure is characteristic of CO2 weakly adsorbed on the O2 preadsorbed NiO(111) surface without the lattice apex O [referred to hereafter as the defective NiO(111)]. By use of a fully relaxed defective NiO(111)-p(2 × 2) and gas-phase CO2 as reference systems, the calculated adsorption energy of CO2 on the defective NiO(111) is -0.48 eV. Without any assistance, desorption of the CO2 from the defective NiO(111) surface is moderately activated, with a desorption barrier of 0.47 eV according to our CI-NEB calculations. Free O2 molecules can heal the lattice O vacancy created upon CO2 desorption. The calculated adsorption energy of

Figure 7. (a-e) Schematic of the CO oxidation catalytic cycle O2 + 2CO f 2CO2 on O2 preadsorbed NiO(111) surface. Blue and gray spheres represent Ni and C atoms, respectively. Purple and pink spheres are used to signify gas-phase and adsorbed O species, to distinguish those from lattice O (red) of the NiO substrate. For details, see text.

21584

J. Phys. Chem. C, Vol. 114, No. 49, 2010

O2 healing a lattice O vacancy is -0.94 eV. Interestingly, if O2 adsorption occurs at the same time as CO2 desorption, namely, under an adsorption-assisted-desorption mechanism,46-49 desorption of CO2 will be very facile. The energy gain of 0.46 eV obtained by switching the positions of O2 and CO2, i.e., from O2(g) + CO2(a) to O2(a) + CO2(g), is the driving force of the adsorptionassisted-desorption mechanism. With the assistance of O2 adsorption, desorption of CO2 is very facile and can occur at low temperature, as observed in our experiments.20 Following a healing of the O-apex vacancy by CO2 desorption and O2 adsorption, the NiO(111) surface has one extra O atop the apex O. Our calculations suggest that this extra O can catalyze a second CO oxidation easily. An incoming second CO spontaneously reacts with the extra O, forming a gas phase CO2. The reaction is exothermic with an energy gain of 3.63 eV. After the oxidation of the second CO, one catalytic cycle is complete, and the active site returns to its original phase with O2 preadsorbed on NiO(111). To illustrate the catalytic CO oxidation on O2 predosed NiO(111) surfaces, we schematically show the whole catalytic cycle (O2 + 2CO f 2CO2) in Figure 7. 4. Conclusion In summary, we performed a systematic first-principles study of the CO oxidation mechanism on Ni(111) and NiO(111) surfaces predosed with oxygen. Different possible active phasess1/3 ML O-covered Ni(111), 2/3 ML O-covered Ni(111), 1 ML O-covered Ni(111) surface, Ni(111) surface with a local O (2 × 2) island, and clean and O2-predosed NiO(111)-p(2 × 2) surfacesswere all examined for their ability to catalyze the low-temperature CO oxidation. Our results suggest that low coverage (1/3, 2/3 ML) O-chemisorbed Ni(111) surfaces and Ni(111) with local O islands cannot catalyze the low-temperature reaction. We showed that CO oxidation on a full monolayer O-covered Ni(111) surface can occur through the Eley-Rideal mechanism by CO approaching the adsorbed surface O, with a minimum activation energy barrier of 0.5 eV. Furthermore, clean NiO(111) surfaces cannot catalyze the low-temperature CO oxidation. O2-predosed NiO(111) surfaces, however, can facilitate the low-temperature CO oxidation with the assistance of nearby O2 through an adsorption-assisted-desorption mechanism. The experimentally observed low-temperature (∼120 K) CO oxidation is, therefore, rationalized by CO oxidation on O2predosed NiO(111) surfaces. Acknowledgment. We thank Prof. Mark A. Barteau (University of Delaware) for helpful discussions and Carrie Farberow for comments on the manuscript. Work at University of WisconsinsMadison was supported by the Department of Energy (DOE) Basic Energy Sciences, Division of Chemical Sciences. CPU time was utilized at facilities located at National Energy Research Scientific Computing Center, Oak Ridge National Laboratory, Pacific Northwest National Laboratory, and Argonne National Laboratory, all supported by the DOE. Work at iNANO was supported by the Danish Research Agency, the Strategic Research Council and the Danish Council for Independent, the Carlsberg and Villum Kahn Rasmussen Foundations, and the European Research Council. References and Notes (1) Engel, T.; Ertl, G. In AdVances in Catalysis; Eley, D. D., Pines, H., Weisz, P. B., Eds.; Academic Press: 1979; Vol. 28, p 1-78. (2) Engel, T.; Ertl, G. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis; King, D. A., Woodruff, D. P., Eds.; Elsevier, Amsterdam: 1982; Vol. 4, pp 73-79.

Peng et al. (3) Peden, C. H. F. In Surface Science of Catalysis: In Situ Probes and Reaction Kinetics; Dwyer, D. J., Hoffman, F. M., Eds.; American Chemical Society, Washington, DC, 1992, p 143. (4) Wintterlin, J.; Volkening, S.; Janssens, T. V. W.; Zambelli, T.; Ertl, G. Science 1997, 278, 1931–1934. (5) Peden, C. H. F.; Goodman, D. W. J. Phys. Chem. 1986, 90, 1360– 1365. (6) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Lundgren, E.; Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science 2000, 287, 1474– 1476. (7) Stampfl, C.; Scheffler, M. Phys. ReV. Lett. 1997, 78, 1500–1503. (8) Alavi, A.; Hu, P. J.; Deutsch, T.; Silvestrelli, P. L.; Hutter, J. Phys. ReV. Lett. 1998, 80, 3650–3653. (9) Hendriksen, B. L. M.; Frenken, J. W. M. Phys. ReV. Lett. 2002, 89, 046101. (10) Ackermann, M. D.; Pedersen, T. M.; Hendriksen, B. L. M.; Robach, O.; Bobaru, S. C.; Popa, I.; Quiros, C.; Kim, H.; Hammer, B.; Ferrer, S.; Frenken, J. W. M. Phys. ReV. Lett. 2005, 95, 255505. (11) Gland, J. L.; Kollin, E. B. J. Chem. Phys. 1983, 78, 963–974. (12) Lahr, D. L.; Ceyer, S. T. J. Am. Chem. Soc. 2006, 128, 1800– 1801. (13) Falsig, H.; Hvolbaek, B.; Kristensen, I. S.; Jiang, T.; Bligaard, T.; Christensen, C. H.; Nørskov, J. K. Angew. Chem., Int. Ed. 2008, 47, 4835– 4839. (14) Goodman, D. W.; Peden, C. H. F.; Chen, M. S. Surf. Sci. 2007, 601, L124–L126. (15) Chen, M. S.; Cal, Y.; Yan, Z.; Gath, K. K.; Axnanda, S.; Goodman, D. W. Surf. Sci. 2007, 601, 5326–5331. (16) Hendriksen, B. L. M.; Bobaru, S. C.; Frenken, J. W. M. Surf. Sci. 2004, 552, 229–242. (17) Westerstrom, R.; Wang, J. G.; Ackermann, M. D.; Gustafson, J.; Resta, A.; Mikkelsen, A.; Andersen, J. N.; Lundgren, E.; Balmes, O.; Torrelles, X.; Frenken, J. W. M.; Hammer, B. J. Phys.: Condens. Matter 2008, 20, 184018. (18) Gustafson, J.; Westerstroem, R.; Mikkelsen, A.; Torrelles, X.; Balmes, O.; Bovet, N.; Andersen, J. N.; Baddeley, C. J.; Lundgren, E. Phys. ReV. B 2008, 78, 045423. (19) Mars, P.; van Krevelen, D. W. Chem. Eng. Sci. 1954, 3, 41. (20) Knudsen, J.; Merte, L. R.; Peng, G.; Vang, R. T.; Rest, A.; Lægsgarrd, E.; Anderson, J. N.; Mavrikakis, M.; Besenbacher, F. ACS Nano 2010, 4, 4380. (21) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15–50. (22) Kresse, G.; Furthmüller, J. Phys. ReV. B 1996, 54, 11169–11186. (23) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892–7895. (24) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244–13249. (25) Anisimov, V. I.; Zaanen, J.; Andersen, O. K. Phys. ReV. B 1991, 44, 943–954. (26) Liechtenstein, A. I.; Anisimov, V. I.; Zaanen, J. Phys. ReV. B 1995, 52, R5467–R5470. (27) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Phys. ReV. B 1998, 57, 1505–1509. (28) Blöchl, P. E. Phys. ReV. B 1994, 50, 17953–17979. (29) Kresse, G.; Joubert, D. Phys. ReV. B 1999, 59, 1758–1775. (30) Rohrbach, A.; Hafner, J.; Kresse, G. Phys. ReV. B 2004, 69, 075413. (31) Rohrbach, A.; Hafner, J. Phys. ReV. B 2005, 71, 045405. (32) Cinquini, F.; Giordano, L.; Pacchioni, G.; Ferrari, A. M.; Pisani, C.; Roetti, C. Phys. ReV. B 2006, 74, 165403. (33) Henkelman, G.; Uberuaga, B. P.; Jonsson, H. J. Chem. Phys. 2000, 113, 9901–9904. (34) Wolf, D. Phys. ReV. Lett. 1992, 68, 3315–3318. (35) Barbier, A.; Mocuta, C.; Kuhlenbeck, H.; Peters, K. F.; Richter, B.; Renaud, G. Phys. ReV. Lett. 2000, 84, 2897–2900. (36) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188–5192. (37) Neugebauer, J.; Scheffler, M. Phys. ReV. B 1992, 46, 16067–16080. (38) Lægsgaard, E.; Besenbacher, F.; Mortensen, K.; Stensgaard, I. J. Microsc. Oxf. 1988, 152, 663–669. (39) Holloway, P. H.; Hudson, J. B. Surf. Sci. 1974, 43, 141–149. (40) Tyuliev, G. T.; Kostov, K. L. Phys. ReV. B 1999, 60, 2900–2907. (41) Mendez, M. A.; Oed, W.; Fricke, A.; Hammer, L.; Heinz, K.; Mu¨ller, K. Surf. Sci. 1991, 253, 99–106. (42) Eichler, A. Surf. Sci. 2003, 526, 332–340. (43) Zhang, C. J.; Hu, P. J. Am. Chem. Soc. 2000, 122, 2134–2135. (44) Tasker, P. W. J. Phys. C 1979, 12, 4977–4984. (45) Bandara, A.; Dobashi, S.; Kubota, J.; Onda, K.; Wada, A.; Domen, K.; Hirose, C.; Kano, S. S. Surf. Sci. 1997, 387, 312–319. (46) Tamaru, K. Bull. Chem. Soc. Jpn. 1996, 69, 961–962. (47) Tamaru, K. Appl. Catal., A 1997, 151, 167–177. (48) Takagi, N.; Yoshinobu, J.; Kawai, M. Phys. ReV. Lett. 1994, 73, 292–295. (49) Boudart, M. J. Mol. Catal A: Chem. 1999, 141, 1–7.

JP108475E