Strong Interaction between Gold and Anatase TiO2(001) Predicted by

Jan 10, 2012 - ARC Centre of Excellence for Functional Nanomaterials, Australian Institute for Bioengineering and Nanotechnology, The University of Qu...
11 downloads 6 Views 7MB Size
Article pubs.acs.org/JPCC

Strong Interaction between Gold and Anatase TiO2(001) Predicted by First Principle Studies Chenghua Sun*,†,‡ and Sean C. Smith*,§ †

Centre for Computational Molecular Science, Australia Institute for Bioengineering and Nanotechnology, The University of Queensland, Qld 4072, Australia ‡ ARC Centre of Excellence for Functional Nanomaterials, Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Qld 4072, Australia § Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Tennessee 37831-6494, United States S Supporting Information *

ABSTRACT: The adsorption of gold clusters (Aun, n = 1−10) on the minority surface, (001), of anatase titanium dioxide (TiO2) has been studied in the framework of density functional theory. Various adsorption geometries of gold (Au) clusters on clean, unreconstructed TiO2(001) have been investigated. It is found the adsorption of gold on TiO2(001) is much stronger than that on the majority surface, (101). Due to the strong interfacial bonding, the valence electrons of gold have been highly delocalized and dominate the highest occupied frontier orbitals of Au/TiO2(001). Consequently, it is predicted that the support of TiO2(001) may offer better catalysis performance than conventionally used TiO2(101).

1. INTRODUCTION Strong metal−support interaction (SMSI), originally introduced to describe the chemisorption properties of group 8 noble metals supported on titanium dioxide (TiO2),1 is an important concept in the design of heterogeneous catalysts.2−15 Generally, it is believed that the SMSI can directly affect the dispersion of metal particles, their adsorption geometries, as well as electronic structures, and as a result change their catalysis performance. A typical example is nanosized gold (Au) particles supported on TiO2, indicated as Au/TiO2, which can oxidize carbon monoxide (CO) under low temperature,3−5 while pure Au clusters are almost inert.16 It is believed that the TiO2 support not only helps to disperse the Au atoms but also helps to activate Au particles due to the SMSI because electrons may be transferred from the cation centers of the support (e.g., Ti3+ for TiO2) to the metal particles.17−19 In the case of CO oxidation, the TiO2 support enhances electron transfer from Au to the antibonding states of O2, and thus, the reaction barriers for O2 dissociation and CO oxidation are reduced dramatically at the Au/TiO2 interface.18,19 This example vividly illustrates the significance of SMSI in the design of advanced catalysts. For specific catalyst support, like TiO2, the metal−support interaction is mainly determined by the surface conditions, including unsaturated atoms, defects, reconstructions, surface orientations, etc. As of now, extensive experimental and theoretical studies of Au/TiO2 have been reported, focusing on the majority surfaces of TiO2, such as rutile TiO2(110) and anatase TiO2(101).20−25 For instance, a combination of STM and DFT modeling indicated that Au adsorbed on the perfect © 2012 American Chemical Society

terrace of anatase TiO2(101) is weak, with an adsorption energy (Eads) of 0.25 eV, while over the oxygen vacancies, much higher values of Eads (2.41−3.16 eV) have been predicted.24 On the basis of the STM images and calculated data, it is further predicted that gold clusters prefer to nucleate over oxygen vacancies. 24 Similarly, the bonding between gold and stoichiometric rutile TiO2(110) is very weak and bridging oxygen vacancies are identified as the active nucleation sites for gold clusters.20 According to those studies, the adherence of gold on TiO2 majority surface interfaces is often weak mainly because those majority surfaces are featured by fully coordinated Ti and O. A typical example is anatase TiO2(101), on which Ti6c and O3c are up to 50% and those atoms hardly form new bonds with gold atoms because the Ti6c−O3c bonding is too strong to break. Consequently, gold atoms tend to form Au−Au bonding, rather than Au−O or Au−Ti bonding, and as a result, strong Au−TiO2 coherency can only be obtained over vacancies or other defects.20,24 To improve the catalysis performance of gold clusters, supports with stronger interfacial interactions (e.g., IrO2 support17) are targeted. In recent years, the high reactivity of minority surfaces of metals and metallic oxides has been paid much attention.26−30 In the case of anatase TiO2, we have good reasons to expect that the adsorption of gold clusters on the minority surface, Received: September 16, 2011 Revised: December 16, 2011 Published: January 10, 2012 3524

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

geometries of gold clusters because, for large clusters, a large substrate surface is requested to reduce the interactions between neighboring images. In addition, large supercells are helpful to fully release the interfacial strain energy and find more stable configurations. To eliminate the interaction between two neighboring images along the vertical direction, a vacuum space of more than 2 nm above the surface has been employed. During the geometric optimization, no atom is fixed, except that lattice parameters are fixed to experimental values. In our extensive tests of Au3 adsorbed on TiO2(001) and TiO2(101) with different starting geometries, a (3 × 3) slab has been employed to speed up the sampling. As illustrated below, all major structural features obtained from (4 × 4) can be obtained well, except the adsorption energy. Using a (3 × 3) slab, the averaged adsorption energies of Au3/TiO2(001) are smaller than those calculated from the (4 × 4) slab because the larger supercells can fully release the strain energy associated with the adsorption of gold clusters. All calculations are spin-polarized and carried out using the DMol3 code.52,53 The generalized gradient approximation with the Perdew−Burke−Ernzerhof functional,54,55 together with effective core potentials with double-numeric quality basis, was utilized for all geometric optimization and single-point energy calculations. During our calculations, the convergence criteria for structure optimizations were set to (1) an energy tolerance of 1.0 × 10−6 Ha per atom, (2) a maximum force tolerance of 1.0 × 10−3 Ha/Å, and (3) a maximum displacement tolerance of 1.0 × 10−3 Å. Due to the large size of the super cells, the kspace is sampled by the gamma point. The Au−TiO 2 interaction is described by the averaged adsorption energy, Eads, which is defined by

(001), may be much stronger than that on (101) because (i) (001) is dominated by unsaturated atoms, which tend to bond strongly with external species; (ii) due to the surface cleavage, Ti−O bonding is relatively weak compared with bulk conditions or (101) surface, and potentially breaks to bond with Au. The challenge is that minority surfaces often diminish rapidly during the crystal growth, and thus, the overall percentage in equilibrium crystals is very small.31,32 Our recent success on the synthesis of a high-ratio (001) surface,33,34 the minority of anatase TiO2, offered an excellent chance to discuss this issue. Using fluorine species as the controlling agent, welldefined anatase TiO2 crystals with 47% or even 90% have been synthesized readily.35−38 Such (001)-dominated crystals are highly reactive and have shown exciting applications, such as lithium barriers,39,40 photocatalytic decomposition,41−46 solarhydrogen production,47,48 etc. Given that all surface atoms on the (001) surfacefive-coordinated titanium (Ti5c) and twocoordinated oxygen (O2c)are unsaturated, it is reasonably expected that a strong Au−TiO2 interface may be obtainable. If it is true, better catalysis performance of gold clusters for CO oxidation can be expected because gold atoms at the Au/TiO2 interface are very effective at reducing the dissociation barrier of oxygen.18 To get insight into the nucleation and growth of gold clusters on anatase TiO2(001), theoretical modeling is performed and reported in this work. Herein, gold clusters, indicated as Aun (n = 1−10), have been introduced on TiO2(001), initially with different configurations. The adsorption energies and the effect of the TiO2 support on the electronic structures of Aun are particularly discussed. As a prediction of Au on the minority surface of anatase TiO2(001), our aim is to provide a primary evaluation on the interaction of Au with TiO2 support. Different from an early report of gold on TiO2(001),49 our focus is not to discuss the diffusion and incorporation of gold into the TiO2 lattice but to explore the possible catalysis application. Hopefully this contribution can attract experimentalists to design novel catalysts based on the Au/TiO2(001) interaction.

Eads = (nE(Au) + E(TiO2 ) − E(Au n − TiO2 ))/n where n is the total number of adsorbed Au atoms and E(Au), E(TiO2), and E(Aun−TiO2) are the total energies of single Au, the clean TiO2 slab, and the interacting Aun/TiO2 system. By this definition, positive Eads indicates that the adsorption is stable. On the basis of the computational setting, the cohesive energy per TiO2 for bulk anatase TiO2 is 21.60 eV, which is consistent with the literature.56 For the clean (001) surface, the calculated surface energy is 0.90 J/m2, in line with refs 33 and 56. As to the Au−TiO2 interaction, the calculated adsorption energy for individual Au on antase TiO2(101) is 0.29 eV (using a 2 × 2 × 1 supercell), being consistent with the result by plane-wave first principle calculations (Eads = 0.25 eV).21

2. COMPUTATIONAL METHODS Calculations have been performed based on a periodic density functional theory (DFT) approach. The (001) surface is modeled by a (4 × 4) slab (dimension size: 1.512 nm × 1.512 nm), with a thickness of four titanium layers based on early publications,30,50,51 as shown in Figure 1. On the basis of our tests, the large (4 × 4) slab is essential for scanning the

3. RESULTS AND DISCUSSION 3.1. Adsorption of Single Gold on TiO2(001). Figure 2 shows the optimized geometries of a single Au on TiO2(001) modeled by a (4 × 4) slab. For the sake of clarity, only atoms around the adsorbed Au in the first layer have been shown. Herein, a gold atom has been introduced to the surfaces, initially with four different geometries, including Au−O2c, Au− Ti5c, Ti5c−Au−Ti5c, and O2c−Au−O2c. For the first two, we got the same result (Ti5c−Au−O2c−Ti5c), and the side view and top view are shown in Figure 2b and c, respectively. The adsorption energy is 5.34 eV. Starting from Ti5c−Au−Ti5c (see Figure 2d), we got the adsorption geometry of Ti5c−O2c−Au−O2c−Ti5c (as shown in Figure 2e and f), with an adsorption energy of 6.14 eV. If we started from O2c−Au−O2c (see Figure 2g), constrained adsorption was obtained, as shown in Figure 2h and i. However, such adsorption is not stable and prefers to be transferred to the Ti5c−O2c−Au−O2c−Ti5c. Therefore, the

Figure 1. Anatase TiO2(001) modeled by a (4 × 4) slab. Titanium and oxygen are indicated as gray and red spheres, respectively. 3525

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

Figure 3. Optimized geometries of Au2 on TiO2(001) with various adsorption energies in units of eV.

more favorable than the other two. (2) Au−Au bonding is not preferred before Au is fully bonded with Ti and O according to Figure 3g. (3) If possible, Au prefers to be stabilized by two Au−O bonds, rather than one Au−O and one Au−Ti, as indicated by Figure 3c. These features confirm our speculation that Au is stabilized by both Au−O and Au−Ti bonds, with the cost of the breakages of Ti5c−O2c bonds at the interface. 3.3. Adsorption of Au3 (n = 3−10) on TiO2(001). The adsorption of Au3 is much more complicated than that of Au2, so extensive tests are carried out to give an extensive sampling of the starting geometries. Twenty tests have been first performed with the (3 × 3) slab and the initial geometries are randomly generated, but we ensured that the typical structures (three separated Au atoms, Au−Au−Au triangle standing and lying down, Au−Au plus one separated) are included in the tests. The tested results are shown in Figure 4, from which we can summarize three major features: (a) Ti−O

Figure 2. Adsorption of a single gold atom on anatase TiO2(001) starting from four geometries, including the initial geometry, side view and top view of the optimized geometries. Titanium, oxygen, and gold are indicated as gray, red, and golden spheres or labeled directly.

geometry of Ti5c−O2c−Au−O2c−Ti5c is the most stable. We also tested the starting geometry with Au being away from the surface with a distance more than 3 Å, and we got the same result with Figure 2e and f. With respect to clean (001), the typical change caused by gold adsorption is the breakage of the Ti5c−O2c bond, leading to the formation of Au−O bonds, which may strongly stabilize Au or Au clusters. The calculated adsorption energy, Eads = 6.14 eV, supports such a speculation too. Except Au−O bonding, Au may also bond with unsaturated titanium. As shown in Figure 2b and c, Au is not exactly at the center of four Ti5c atoms but closer to one with a distance of 2.87 Å, a typical distance for Au−Ti bonding. Therefore, it is reasonably speculated that the single Au is actually stabilized by both Au−O and Au−Ti bonds, which can be confirmed by following studies. 3.2. Adsorption of Au2 on TiO2(001). On the basis of the adsorption of single Au, the Au−Au bonding is further taken into account through the optimization of Au2 on TiO2(001). Figure 3 shows the optimized geometries of Au2 together with the averaged adsorption energies. Herein, eight starting geometries are considered on the basis of the symmetry consideration, representing the major possible combination of adsorption sites for two gold atoms. However, we did not consider the vertical movement of gold because we found that, even though gold is away from the surface for 4 Å, it can get down to be adsorbed, and geometries with separated Au(up)− Au(down) or clustered Au−Au (off the surface) are not energetically preferred. Compared with the case of single Au, the adsorption energies are smaller by around 2−3 eV, due to the coverage effect. On the basis of these geometries and energies, three features can be summarized: (1) The first six of the eight show the breakages of Ti5c−O2c (see green lines in Figure 2), leading to the formations of Au−O and Au−Ti bonding, and obviously are

Figure 4. Adsorption of Au3 on anatase TiO2(001) starting from different geometries. For the above tests, a (3 × 3) slab has been employed. Adsorption energy is inserted in units of eV.

bond breakage has been widely observed; (b) the formation of Ti−Au and O−Au bonds has been observed in all geometries; (c) separated configurations (see Figure 4m, q, and t) are not preferred. In terms of averaged adsorption energy, the clustered (see Figure 4o and s) and dispersed (see Figure 4a, d, i, and r) geometries show similar stabilities. According to our experience 3526

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

Figure 5. Adsorption of Au3 on anatase TiO2(001) starting from different geometries. For the above tests, a (4 × 4) slab has been employed. Adsorption energy is inserted in units of eV.

Figure 6. Optimized geometries of Aun (n = 3−10) on TiO2(001) with Eads in units of eV. For the above tests, s(4 × 4) slab has been employed.

Figure 7. Unfavorable geometries of Au3 and Au4 on TiO2(001) with Eads in units of eV.

another 10 tests are performed on the (4 × 4) slab, and the optimized geometries are shown in Figure 5. The additional tests give the same conclusion that clusters can be strongly

in the adsorption of Au2, unreleased strain energy associated with gold adsorption may dramatically reduce the stabilities, which can be investigated by larger supercells. Therefore, 3527

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

Figure 8. Adsorption of gold clusters on TiO2(001) and TiO2(101) with (a) adsorption energies and geometries of Au3 on (b) TiO2(101) and (c) TiO2(001).

Figure 9. Optimized geometries of Au3 on TiO2(101), starting from different geometries.

adsorbed by TiO2(001), and the Au/TiO2(001) interface is dominated by Ti−Au, O−Au, and Au−Au bonds; moreover, separated adsorption (see Figure 5j) is not favored. Different

from Figure 4, however, clustered adsorption, shown in Figure 5i, is not as stable as those dispersed (but gold atoms are linked together by Au−Au, see Figure 5a−f) adsorption. Therefore, 3528

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

Figure 10. Comparison of Au3 adsorption on TiO2(001) and TiO2(101). The inserted numbers are total adsorption energies in units of eV.

Figure 11. Calculated (a) TDOS and (b) LDOS (Au) for gold clusters.

introduced by gold adsorption has not been released due to the periodic-boundary-condition constraint. Similarly, if several gold atoms are introduced on the (4 × 4) slab and completely separated as shown in Figure 7b and d, strain energies cannot be fully released as well, which is why those separated geometries are not preferable. It is also reasonably consistent with our tests of Au3 on the (3 × 3) slab, in which strain energy is not fully released and thus dispersed adsorption does not show obviously higher stability than clustered adsorption. 3.5. Strong Interaction between Au and TiO2(001). As described above, the highlight of gold clusters on TiO2(001) is the breakage of Ti5c−O2c bonds, while such breakage has never been reported for gold on TiO2(101). Given that such breakage may directly lead to strong Au−TiO2 interaction, it is speculated that TiO2(001) can offer stronger adherence for gold clusters than TiO2(101), which has been supported by the calculated adsorption energies, as shown in Figure 8a. Figure 8b and c show the geometries of three gold atoms on perfect (001) and (101), respectively, in which gold atoms form a triangle partially isolated from the TiO2 surface, rather than closely adhere to the support as that on (001). This is not surprising because all Ti−O bonds on (101) involve fully saturated titanium or oxygen; however, all surface atoms on (001) are undersaturated, Ti5c or O2c, and thus Ti−O bonds are not as strong as those on (101). In Figure 8, we compared the adsorption energies of gold clusters on TiO2(101) and TiO2(001), and the size effect of the slab is considered using coverage density, which may not be accurate for clustered adsorption. To make a better comparison of the adsorption of gold clusters on TiO2(001), 30 tests of Au3 on the (3 × 3) slab of TiO2(101) have been carried out and the optimized geometries are shown in Figure 9, in which some geometries may be similar or even the same, but they are still presented because those geometries are obtained from different

gold atoms prefer to stay together by Au−Au bonding and strongly bond with TiO2(001) via Au−O and Au−Ti bonds. 3.4. Adsorption of Aun (n = 3−10) on TiO2(001). Figure 6 shows the optimized geometries of Aun (n = 3−10) on TiO2(001) with averaged adsorption energies. For Au3 and Au4, the present geometries are selected from several possible configurations (see Figure 5 and Figure S1 in the Supporting Information). For Aun (n = 5−10), additional Au atoms are introduced based on the geometries of Au4 shown in Figure 6a and b. Again, the breakage of the Ti5c−O2c bond has been widely observed in all of these images, which agrees well with our conclusions obtained from Au2 and Au3. Therefore, the Au−TiO2(001) interface is essentially dominated by Au−O, Au−Ti, and Au−Au bonding, with typical bond lengths of 1.98 ± 0.04, 2.73 ± 0.06, and 2.70 ± 0.10 Å, respectively. Given the bond length of Au−Ti is actually close to that of Au−Au, the lattice mismatch at Au−TiO2(001) is not high, which is critically important for the formation of a solid interface. To illustrate the stabilization effect of Au−O and Au−Ti bonding, unfavorable geometries for Au3 and Au4 are shown in Figure 7. Figure 7a and c represent clusters almost isolated from the support, which are less stable than those geometries dominated by strong Au−TiO2 interactions (see Figure 6a and b) by 0.64 and 0.66 eV in terms of averaged adsorption energy. The most unfavorable geometries are shown in Figure 7b and d, in which all gold atoms are completely separated from each other. Although Au−O and Au−Ti bonds are observed, they show much smaller adsorption energies compared with other geometries. Similar results are also observed for Au2, as shown in Figure 3h. To understand such instability, the adsorption energies of single gold on three slabs have been calculated, 1.56 eV for Au-(2 × 2), 4.28 eV for Au-(3 × 3), and 6.15 eV for Au(4 × 4). Particularly, the (2 × 2) slab fails to predict the breakage of the Ti5c−O2c bond and thus much strain energy 3529

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

Figure 12. Calculated (a) LDOS (Au), (b) HOMO, and (c) LUMO for a gold cluster (Au9) on TiO2(001). The blue and green isosurfaces indicate the positive and negative wave functions.

will be dominated by Au5d, as shown in Figure 12b and c. For the oxidation of carbon oxide catalyzed by Au/TiO2, Au5d electrons partially transfer to the antibond π* of oxygen molecules, leading to its activation and thus reducing the dissociation barrier.19 Compared with pure Au9, Au5d electrons dominating the HOMO are more unlocalized when adsorbed on TiO2(001), as revealed by Figure 12a and b, and thus offer better mobility for surface reactions. In addition, oxidized Au, rather than neutral Au particle, has been identified as the active species for CO oxidation,57−61 particularly for CO adsorption;60 as a result, the ultrasmall size of gold clusters is essential for high reactivity.60 As shown above, Au clusters show an excellent distribution on the surface, rather than separately clustering like that on TiO2(101). Therefore, it is predicted that Au/TiO2(001) may show better catalysis performance for the CO oxidation.

initial geometries. According to the results, the following is found: (i) Among the 30 tests, only five geometries do not show the clustered adsorption, although most of the starting geometries are introduced with the breaking of Au−Au bonds and Au atoms are separated on the surface. (ii) For clustered geometries, Au3 has three typical bonding networks, indicated as ①②③ on the top of the figure, and have been assigned in the figure. (iii) The relative stabilities of these geometries is, ③ > ② > ① > unclustered, in terms of adsorption energies. Therefore, it is confirmed that gold clusters prefer to keep the clustered configurations on TiO2(101), in line with ref 24, in which it is found that strong adherence is only available over oxygen vacancies or step edges. While on TiO2(001), as shown in Figures 5 and 6, dispersed adsorption, particularly Ti−O−Au− O−Ti and Ti−O−Au−Ti, is favored. To further compare the adherence strength of gold clusters on TiO2(001) and TiO2(101), calculated averaged adsorption energies of Au3 have been collected using the (3 × 3) slab and shown in Figure 10. In the case of TiO2(001) and TiO2(101), most preferred geometries show adsorption energies of 6.0−6.2 and 5.5−6.0 eV, respectively, suggesting that the adsorption of Au3 on TiO2(001) is stronger, consistent with the conclusion obtained from Figure 8. To understand the above difference between Au-(001) and Au-(101), the total density of states (TDOS) and local DOS (LDOS) for a gold cluster (Au3 as an example) have been calculated and shown in Figure 11a and b, with the Fermi level as the reference of zero energy. In Figure 11a, the TDOS profile for Au-(001) stays at lower energy than that for Au(101) by around 1.0 eV, meaning the system is more stable. And the LDOS of gold on (101) is a sharp peak, suggesting electrons from gold atoms are highly localized and the overlap with TiO2 is weak, consistent with the geometry of isolated triangle shown in Figure 8b. On the contrary, LDOS for gold on (001) is distributed over the whole valence band range, suggesting that electrons from gold atoms extensively overlapped with the support, in line with the Au−O and Au−Ti bonds shown in Figure 8c. 3.6. Effect of Gold Adsorption on Electronic Structures of Gold and TiO2. As illustrated above, electrons from gold atoms can strongly mix with those contributed by titanium and oxygen on TiO2(001), which will undoubtedly affect the electronic structures of both gold and TiO2. Figure 12a shows the LDOS of a gold cluster (Au9 as an example) adsorbed on TiO2(001), and it is clear that the electrons are more unlocalized than those of pure Au9. It is well-known that the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of pure TiO2 are dominated by O2p and Ti3d electrons, respectively. In the case of Au/TiO2(001), the LUMO has no change, but the HOMO

4. CONCLUSION The geometries and energetics of gold clusters adsorbed on anatase TiO2(001) have been studied under the framework of DFT. It is found that (i) gold atoms can be strongly adsorbed through Au−Ti and Au−O bonding, with the breakage of Ti5c− O2c bonds; (ii) with the adsorption of more gold atoms, a solid Au/TiO2(001) interface featured by Au−Au, Au−O, and Au− Ti bonds is predicted, while isolated gold clusters or completed separated adsorption is not favorable; (iii) the Au/TiO2(001) interface is stronger than Au/TiO2(101), essentially due to the highly unsaturated nature of TiO2(001); and (iv) strong Au− TiO2(001) interaction results in the unlocalization of the valence electrons of gold, which may improve the catalysis performance with respect to Au/TiO2(101). The above knowledge may serve as a guideline for experimentalists to design better catalysts on the basis of the high reactivity of minority surfaces of the support.



ASSOCIATED CONTENT

S Supporting Information *

Details of the adsorption geometries of Au4 on TiO2(001) with calculated adsorption energies. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (S.C.S.); [email protected] (C.S.).



ACKNOWLEDGMENTS This work is inspired by collaborations within the Australian Research Council Centre of Excellence for Functional Nanomaterials and has been financially supported by The University 3530

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531

The Journal of Physical Chemistry C

Article

(34) Yang, H. G.; Liu, G.; Qiao, S. Z.; Sun, C. H.; Jin, Y. G.; Smith, S. C.; Zou, J.; Cheng, H. M.; Lu, G. Q. J. Am. Chem. Soc. 2009, 131, 4078. (35) Dai, Y. Q.; Cobley, C. M.; Zeng, J.; Sun, Y. M.; Xia, Y. N. Nano Lett. 2009, 9, 2455. (36) Han, X. G.; Kuang, Q.; Jin, M. S.; Xie, Z. X.; Zheng, L. S. J. Am. Chem. Soc. 2009, 131, 3152. (37) Liu, G.; Yang, H. G.; Sun, C. H.; Cheng, L.; Wang, L. Z.; Lu, G. Q.; Cheng, H. M. CrystEngComm 2009, 11, 2677. (38) Fang, W. Q.; Zou, J. Z.; Liu, J.; Chen, Z. G.; Yang, C.; Sun, C. H.; Qian, G. Q.; Zou, J.; Qiao, S. Z.; Yang, H. G. Chem.Eur. J. 2011, 17, 1423. (39) Sun, C. H.; Yang, X. H.; Chen, J. S.; Li, Z.; Lou, X. W.; Li, C. Z.; Smith, S. C.; Lu, G. Q.; Yang, H. G. Chem. Commun. 2010, 46, 6129. (40) Chen, J. S.; Tan, Y. L.; Li, C. M.; Cheah, Y. L.; Luan, D.; Madhavi, S.; Boey, F. Y. C.; Archer, L. A.; Lou, X. W. J. Am. Chem. Soc. 2010, 132, 6124. (41) Wu, B. H.; Guo, C. Y.; Zheng, N. F.; Xie, Z. X.; Stucky, G. D. J. Am. Chem. Soc. 2008, 130, 17563. (42) Liu, G.; Yang, H. G.; Wang, X. W.; Cheng, L. N.; Pan, J.; Lu, G. Q.; Cheng, H. M. J. Am. Chem. Soc. 2009, 131, 12868. (43) Liu, S. W.; Yu, J. G.; Jaroniec, M. J. Am. Chem. Soc. 2010, 132, 11914. (44) Liu, G.; Sun, C. H.; Smith, S.; Wang, L. Z.; Lu, G. Q.; Cheng, H. M. J. Colloid Interface Sci. 2010, 349, 477. (45) Liu, G.; Sun, C. H.; Yang, H. G.; Smith, S. C.; Lu, G. Q.; Cheng, H. M. Chem. Commun. 2010, 46, 755. (46) Liu, M.; Piao, L. Y.; Zhao, L.; Ju, S. T.; Yan, Z. J.; He, T.; Zhou, C. L.; Wang, W. J. Chem. Commun. 2010, 46, 1664. (47) Yu, J. G.; Qi, L. F.; Jaroniec, M. J. Phys. Chem. C 2010, 114, 13118. (48) Yang, X. H.; Li, Z.; Liu, G.; Yang, C.; Sun, C. H.; Li, C. Z.; Yang, H. G. CrystEngComm 2011, 13, 1378. (49) Mete, E.; Gülseren, O.; Ellialtıoğlu, Ş. Phys. Rev. B 2009, 80, 035422. (50) Bredow, B.; Giordano, L.; Cinquini, F.; Pacchioni, G. Phys. Rev. B 2004, 70, 035419. (51) Kowalski, P. M.; Meyer, B.; Marx, D. Phys. Rev. B 2009, 79, 115410. (52) Delley, B. J. Chem. Phys. 1990, 92, 508. (53) Delley, B. J. Chem. Phys. 2000, 114, 7756. (54) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (55) Kresse, G.; Joubert, J. Phys. Rev. B 1999, 59, 1758. (56) Lazzeri, M.; Vittadini, A.; Selloni, A. Phys. Rev. B 2001, 63, 155409. (57) Bunluesin, T.; Gorte, R. J.; Graham, G. W. Appl. Catal., B 1998, 15, 107. (58) Zalc, J. M.; Sokolovskii, V.; Loffleer, D. G. J. Catal. 2002, 206, 169. (59) Fu, Q.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Science 2003, 301, 935. (60) Liu, Z. P.; Jenkins, S. J.; King, D. A. Phys. Rev. Lett. 2005, 94, 196102. (61) Zhang, C.; Michaelides, A.; King, D. A.; Jenkins, S. J. J. Am. Chem. Soc. 2010, 132, 2175.

of Queensland (Research Excellence Award for C.H.S.), the Australian Research Council, and the Queensland State Government (Smart Future Fellowship for C.S.). We also appreciate the generous grants of CPU time from both the University of Queensland and the Australian National Computational Infrastructure Facility.



REFERENCES

(1) Tauster, S. J.; Fung, S. C.; Garten, R. L. J. Am. Chem. Soc. 1978, 100, 170. (2) Tauster, S. J.; Fung, S. C.; Baker, R. T. K.; Horsley, J. A. Science 1981, 211, 1121. (3) Tauster, S. J. Acc. Chem. Res. 1987, 20, 389. (4) Hong, C. T.; Yeh, C. T.; Yu, F. H. Appl. Catal. 1989, 48, 385. (5) Haller, G. L.; De, R. Adv. Catal. 1989, 36, 173. (6) Vannice, M. A.; Sen, B. J. Catal. 1989, 115, 65. (7) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M. J.; Delmon, B. J. Catal. 1993, 144, 175. (8) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647. (9) Gilb, S.; Weis, P.; Furche, F.; Ahlrichs, R.; Kappes, M. M. J. Chem. Phys. 2002, 116, 4094. (10) Li, J.; Li, X.; Zhai, H. J.; Wang, L. S. Science 2003, 299, 864. (11) Fielicke, A.; Ratsch, C.; von Helden, G.; Meijer, G. J. Chem. Phys. 2005, 122, 091105. (12) Xing, X.; Yoon, B.; Landman, U.; Parks, J. H. Phys. Rev. B 2006, 74, 165423. (13) Bulusu, S.; Li, X.; Wang, L. S.; Zeng, X. C. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 8326. (14) Lechtken, A.; Schooss, D.; Stairs, J.; Blom, M.; Furche, F.; Morgner, N.; Kostko, O.; von Issendorff, B.; Kappes, M. Angew. Chem., Int. Ed. 2007, 46, 2944. (15) Gruene, P.; Rayner, D. M; Redlich, B.; van der Meer, A. F. G.; Lyon, J. T.; Meijer, G.; Fielicke, A. Science 2008, 321, 674. (16) Hammer, B.; Kørskov, J. K. Nature 1995, 376, 238. (17) Liu, Z. P.; Jenkins, S. J.; King, D. A. Phys. Rev. Lett. 2004, 93, 156102. (18) Molina, L. M.; Rasmussen, M. D.; Hammer, B. J. Chem. Phys. 2004, 120, 7673. (19) Liu, Z. P.; Gong, X. Q.; Kohanoff, J.; Sanchez, C.; Hu, P. Phys. Rev. Lett. 2003, 91, 266102. (20) Wahlström, E.; Lopez, N.; Schaub, R.; Thostrup, P.; Rønnau, A.; Africh, C.; Lægsgaard, E.; Nørskov, J. K.; Besenbacher, F. Phys. Rev. Lett. 2003, 90, 026101. (21) Okazaki, K.; Morikawa, Y.; Tanaka, S.; Tanaka, K.; Kohyama, M. Phys. Rev. B 2004, 69, 235404. (22) Liu, L. M.; McAllister, B.; Ye, H. Q.; Hu, P. J. Am. Chem. Soc. 2006, 128, 4017. (23) Matthey, D.; Wang, J. G.; Wendt, S.; Matthiesen, J.; Schaub, R.; Lægsgaard, E.; Hammer, B.; Besenbacher, F. Science 2007, 315, 1692. (24) Gong, X. Q.; Selloni, A.; Dulub, O.; Jacobson, P.; Diebold, U. J. Am. Chem. Soc. 2008, 130, 370. (25) Lee, S.; Molina, L. M.; Lopez, M. J.; Alonso, J. A.; Hammer, B.; Lee, B.; Seifert, S.; Winans, R. W.; Elam, J. W.; Pellin, M. J.; Vajda, S. Angew. Chem., Int. Ed. 2009, 48, 1467. (26) Vittadini, A.; Selloni, A.; Rotzinger, F. P.; Grätzel, M. Phys. Rev. Lett. 1998, 81, 2954. (27) Gong, X. Q.; Selloni, A. J. Phys. Chem. B 2005, 109, 19560. (28) Tian, N.; Zhou, Z. Y.; Sun, S. G.; Ding, Y.; Wang, Z. L. Science 2007, 316, 732. (29) Xie, X. W.; Li, Y.; Liu, Z. Q.; Haruta, M.; Shen, W. J. Nature 2009, 458, 746. (30) Sun, C. H.; Liu, L. M.; Selloni, A.; Lu, G. Q.; Smith, S. C. J. Mater. Chem. 2010, 20, 10319. (31) Berger, H.; Tang, H; Lévy, F. J. Cryst. Growth 1993, 130, 108. (32) Chen, X.; Mao, S. S. Chem. Rev. 2007, 107, 2891. (33) Yang, H. G.; Sun, C. H.; Qiao, S. Z.; Zou, J.; Smith, S. C.; Cheng, H. M.; Lu, G. Q. Nature 2008, 453, 638. 3531

dx.doi.org/10.1021/jp208948x | J. Phys. Chem. C 2012, 116, 3524−3531