24
Density Functional Theory Study of the Adsorption and Reaction of H 2 S on TiO 2 Rutile (110) and Anatase (101) Surfaces Wen-Fei Huang, Hsin-Tsung Chen, and M. C. Lin J. Phys. Chem. C 2009, 113, 20411–20420 Rewritten by Yao Cui The interaction of H 2 S with TiO 2 rutile (110) and anatase (101) surfaces has been studied by using period density functional theory (DFT) method. It was found that H 2 S, HS, S, and H preferentially adsorb at the Ti 5c , O 2c , (Ti 5c ) 2 , and O 2c sites, respectively, on the rutile (110) surface; but at the Ti 5c , (Ti 5c ) 2 , (-O 2c )(-Ti 5c ), and O 2c sites, respectively, on the anatase (101) surface. The mechanisms for these reactions were analyzed. Based on the nudged elastic band (NEB) method, two possible pathways are constructed: one is H 2 S(g) + TiO 2 (s) H 2 O(g) + S(v-O 2c ) ; the other one is H 2 S(g) + TiO 2 (s) H 2 (g) + S(-Ti 5c ) 2 (a). S(v-O 2c ), which is the product of O<->S exchange, is harder to form at TiO 2 rutile (110) surface due to the higher barrier of 35.5 kcal/mol and the endothermicity of 15.4 kcal/mol. However, at TiO 2 anatase (101) surface, it is easier to form due to the lower

Density Functional Theory Study of the Adsorption and …python.rice.edu/~brooks/Chem515/Lectures/ClassReports/Yao-report.pdf · Density Functional Theory Study of the Adsorption

Embed Size (px)

Citation preview

Density Functional Theory Study of the Adsorption and Reaction of H2S

on TiO2 Rutile (110) and Anatase (101) Surfaces

Wen-Fei Huang, Hsin-Tsung Chen, and M. C. Lin

J. Phys. Chem. C 2009, 113, 20411–20420

Rewritten by Yao Cui

The interaction of H2S with TiO2 rutile (110) and anatase (101) surfaces has been studied

by using period density functional theory (DFT) method. It was found that H2S, HS, S,

and H preferentially adsorb at the Ti5c, O2c, (Ti5c)2, and O2c sites, respectively, on the rutile

(110) surface; but at the Ti5c, (Ti5c)2, (-O2c)(-Ti5c), and O2c sites, respectively, on the

anatase (101) surface. The mechanisms for these reactions were analyzed. Based on the

nudged elastic band (NEB) method, two possible pathways are constructed: one is H2S(g)

+ TiO2(s) à H2O(g) + S(v-O2c) ; the other one is H2S(g) + TiO2(s) à H2(g) + S(-Ti5c)2(a).

S(v-O2c), which is the product of O<->S exchange, is harder to form at TiO2 rutile (110)

surface due to the higher barrier of 35.5 kcal/mol and the endothermicity of 15.4

kcal/mol. However, at TiO2 anatase (101) surface, it is easier to form due to the lower

barrier of 12.4 kcal/mol and the lesser endothermicity of 5.0 kcal/mol. Also the rate

constants in the temperature range 300-2000 K were predicted.

1. Introduction

Titanium dioxide (TiO2) has become attractive due to its widely technical applications

in photoelectrochemistry, catalysis,1,2polymers3,4 and semiconductors.5-7Since the study of

photoelectrochemical splitting of H2O on the TiO2 rutile surface in the early 1970s,8 small

molecule-TiO2 surface interactions have arisen people’s interest. 9-15

To enhance its photocatalytic activity, people have found that doping with nonmetal

elements (S, C, N, P, B, F)16-24 is a good way. For instance, the S-doped TiO2 has reduced

the band gap successfully. 25 Also, the interaction of H2S and its fragments on TiO2

surfaces has been studied. 26-27 Experimentally, Smith et al. annealed TiO2 with TiS2 to

obtain the S-doped TiO2,28 and Chen et al. used H2S as a precursor to produce a S-doped

TiO2 surface.29 However, as we know, the mechanisms for these reactions are not clear.

According to the research by Chen et al., the possible pathway for this reaction is due to

the dissociation of H2S. As H2Sà HS+H and then HSà S+H, H2O and O vacancies are

produced, which let S atom move into them to form the S-doped TiO2 surface. The TiO2

nanoparticle film is a polycrystalline material with different TiO2 surface structures.

Among them, rutile (110) and anatase (101) surfaces are the most stable ones with the

lowest surface energies. 30,31

In our research, we try to illustrate the possible mechanism for the reaction of H2S on

the TiO2 rutile (110) and anatase (101) surfaces by periodic density functional (DFT)

theory. The geometries and adsorption energies H2S and its fragments (HS, S, and H) on

the TiO2 surfaces have been studied. Also, the rate constants for these reactions are

calculated.

2. Computational Methods and Models

Density functional theory (DFT) computations were performed by using the plane-

wave technique implemented in the Vienna ab initio simulation package (VASP)32,33 in

calculating reactants, transition states, intermediates, and products during the reactions of

H2S and its fragments on both surfaces. The ion-electron interaction is described with the

projector augmented wave (PAW) method. 34 The exchange-correlation potential is

approximated by the generalized gradient approximation (GGA) using the PW91

functional35 since it goes well with surfaces.36,37 The Brillouin zone was sampled with

(2×4×2) and (4×4×1) Monkhorst-Pack k-points with a 500 eV cutoff energy for the

plane-wave basis set was used for the rutile bulk and (110) surfaces, respectively; and the

(4 × 4 × 4) and (6 × 4 × 1) k-points with the same cutoff energy were chosen The

convergence threshold was set as 10-4 eV in energy and 10-3 eV/Å in force. In these

calculations spin polarization was considered for the radicals.

As shown in Figure 1a and b, periodically repeated slabs consisting of 16 and 12

[TiO2] unit cells were used to model the rutile and anatase surfaces, respectively. The

adsorption energies were calculated by the following equation:

where Etotal, Esurf, and Egas are the calculated electronic energies of the adsorbed species on

the surface, a clean surface, and a gas-phase molecule, respectively. To try to elucidate

the mechanism of the reaction, the nudged elastic band (NEB) method was used. 38,39

Also, frequency calculations were applied to verify all transition states. To calculate the

rate constants, the ChemRate program40 based on microcanonical Rice-Ramsperger-

Kassel-Marcus (RRKM) theory using was carried out. Bader charge analysis41 was

applied to explain the hydrogen effect.

3. Results and Discussion

To make sure our calculations corresponding to the experiments and able to explain

the experimental results, we carried out the calculation of bulk lattice constants for these

structures. For rutile TiO2, our calculations show that a=4.593 Å and c=2.958 Å, which

are in reasonable agreement with the experimental results: a=4.587-4.593 Å and

c=2.954-2.959 Å; for anatase TiO2, our calculations show that a= 3.824 Å and c=9.678

Å which agree well with the experimental values of a=3.782-3.785 Å and c= 9.502-

9.514 Å. 4243 Also as shown in Table 1, the geometrical parameters of H2S and HS

calculated in our research agree well with the previous theoretical4445 and experimental42,43

values.

As shown in Figure 1, there are four different adsorption sites on TiO2 rutile (110)

and TiO2 anatase (101) surfaces. They are 5-fold coordinated titanium (Ti5c), 6-fold

coordinated titanium (Ti6c), 2-fold bridging oxygen (O2c), and 3-fold coordinated oxygen

(O3c).

3.1 Adsorption of H2S, HS, S, and H on the Rutile (110) Surface.

On the Rutile (110) surface, our calculations show that the Ti5c and O2c sites are more

active than the Ti6c and O3c sites, which are in good agreement with previous theoretical

studies.14 As shown in Figure 2, the most stable adsorption structures are illustrated. After

calculations, we found that for H2S adsorption, H2S-Ti5c(a), H2S-O2c(a), H2S-(Ti5c)2(a),

HSH-Ti5c(a), HSH-O2c(a), and O2c-HSH-O2c(a) are the most stable structures among lots

of adsorption structures. The adsorption energies for these structures are 12.9, 2.3, 3.5,

1.3, 3.0, and 4.2 kcal/mol, respectively, as shown in Table 2. For H2S adsorption, H2S-

Ti5c(a) has the largest adsorption energy, which means it is the most stable structure. The

S-Ti5c bond length is 2.757 Å, which is the shortest length for H2S adsorption between the

adsorbate molecule and the surface. There is a H···O2c bond which decreases the energy

of this structure.

Also we calculated the adsorption sites and energies for HS, S, and H on the surface

since they are important to illustrate the reaction mechanism. For HS adsorption, we

found HS-O2c(a) is the most stable one, with the adsorption energy of 18.7 kcal/mol. The

reason that this particular structure is so stable is the S-O covalent bond, which results

from the strong overlap between p orbitals of the surface O anion and the S atom. For S

adsorption, S-(Ti5c)2(a) has the largest adsorption energy (38.6 kcal/mol). It means that S

prefers to form two bonds with the neighboring Ti5c atoms at the same time. In addition,

the situation that one of the bridging site O2c atoms is removed was considered. This O

vacancy with an adsorption of S atom has large adsorption energy of 90.0 kcal/mol.

For adsorption of H, our calculations show that H-O2c(a) is the most stable one, which

agrees well with previous theoretical work 11. The absorption energy is 63.5 kcal/ mol.

3.2 Adsorption of H2S, HS, S, and H on the Anatase (101) Surface.

There are four energetically more favorable structures: H2S-Ti5c(a), H2S-O2c(a), HSH-

O2c(a), and O2c-HSH-O2c(a) with adsorption energies of 11.4, 1.1, 3.2, and 2.5 kcal/mol,

(as shown in Table 2) respectively. H2S-Ti5c(a) is the most stable configuration among

them. Compared to the rutile (110) surface, the bond lengths of H···O2c (2.322 Å) and S-

Ti5c (2.828 Å) (as shown in Figure 3) are longer on the anatase (101) surface.

Similar to rutile (101) surface, there are three configurations for HS, S, and H

adsorption considered. Their adsorption energies are illustrated in Table 2. Unlike the

rutile surface, HS(-Ti5c)(-O3c)(a) owns the adsorption energy of 11.5 kcal/mol, which

predicts that this structure is preferential to be formed. The bond lengths of S-Ti5c, S-O3c,

and H···O2c are 2.616, 2.581, and 2.233 Å, respectively. The adsorption energies of HS-

Ti5c(a), HS(-Ti5c)(-O3c)(a), and HS-O2c(a) are 11.4, 11.5, and 9.8 kcal/mol, respectively.

When it goes to S atom adsorption, we found that S(-O2c)(-Ti5c)(a) is most favorable

(as shown in Figure 3) with an adsorption energy of 46.5 kcal/mol. In addition, the

adsorption of S on a reduced TiO2 surface was also considered. We found that with

absorption energy of 115.9 kcal/mol, the S atom prefers to adsorb on the O2c vacant site:

S(v-O2c).

Like the rutile surface, H-O2c(a) is favorable with an absorption energy and O-H

distance of 55.4 kcal/mol and 0.975 Å, respectively.

3.3 HS, S, and H Co-Adsorption.

From the discussion above, we know that HS-Ti5c, S-Ti5c,H-O2c, S-Ti5c, HS-(Ti5c)2,

S-(Ti5c)2,H-O2c, and S-(Ti5c)2,H2-O2c(a) are among the most stable structures for

adsorption. However, what happens when there is Hydrogen atom adsorbed on the TiO2

suface at the same time? As illustrated in Table 3, HS-Ti5c,H-O2c(a), S-Ti5c,H-O2c,H-

O2c(a), S-Ti5c,H2-O2c(a), HS-(Ti5c)2,H-O2c(a), S-(Ti5c)2,H-O2c,H-O2c(a), and S-(Ti5c)2,H2-

O2c(a) are considered as possible structures for adsorption. The adsorption energies for

these structures are calculated, as shown in Table 3. For the rutile surface, the adsorption

energies for HS-Ti5c,H-O2c(a), S-Ti5c,H-O2c,H-O2c(a), S-Ti5c,H2-O2c(a) are 106.2, 183.9,

and 177.7 kcal/mol, respectively; however, the adsorption energies for HS-(Ti5c)2,H-

O2c(a), S-(Ti5c)2,H-O2c,H-O2c(a), and S-(Ti5c)2,H2-O2c(a) are 96.2, 167.3, and 165.8

kcal/mol, respectively. For the anatase surface, the adsorption energies for HS-Ti5c,H-

O2c(a), S-Ti5c,H-O2c,H-O2c(a), S-Ti5c,H2-O2c(a) are 101.6, 188.4, and 164.6 kcal/mol,

respectively; however, the adsorption energies for HS-(Ti5c)2,H-O2c(a), S-(Ti5c)2,H-

O2c,H-O2c(a), and S-(Ti5c)2,H2-O2c(a) are 83.8, 172.3, and 159.5 kcal/mol, respectively.

After comparison, we found that for both rutile and anatase surfaces, the adsorption

energies for HS-Ti5c,H-O2c(a), S-Ti5c,H-O2c,H-O2c(a), S-Ti5c,H2-O2c(a) are much larger

than those for HS-(Ti5c)2,H-O2c(a), S-(Ti5c)2,H-O2c,H-O2c(a), and S-(Ti5c)2,H2-O2c(a).

Thus, when there is hydrogen atom co-adsorption, the sulfide atom prefers to form one

bond with on Ti5c atom instead of two bonds with the neighboring two Ti5c atoms.

Also, compared to these fragments adsorbed on the surface separately, the hydrogen

co-adsorption makes the structures more stable. The adsorption energies for these co-

adsorption structures are larger than the sum of adsorption energies of the individual

fragments. The possible reason is the hydrogen effect. For example, for HS-Ti5c,H-O2c(a)

structure, there forms hydrogen bond between the first H atom and O2c atom; also,

between S atom and the second H atom, there is another hydrogen bond. Thus the whole

energy decreases and structure is more stable. To make sure about this effect and try to

explain it, we carried out the Bader Charge analysis for these structures.

3.4 Reaction Mechanism of H2S-TiO2 Rutile (110) Interactions.

To predict the mechanism of the H2S adsorption/dissociation processes on the TiO2

rutile (110) surface, the most stable configurations obtained from the H2S, HS, and S

adsorptions were used in the potential energy profiles (as shown in Figure 5). The nudged

elastic band (NEB) method was applied by connecting the local minima. In Figure 6, the

optimized structures of transition states, intermediates, and products were shown. Two

possible pathways are illustrated as follows:

where (g), (s), (a), and (v) denote gas, surface, adsorption, and vacancy, respectively.

Path 1. As shown in Figure 5, H2S first adsorbs on the rutile (110) surface to form

H2S-Ti5c(a) with an exothermicity of 12.9 kcal/mol. Then the H2S dissociation process

happens: one H atom tends to attach to an adjacent bridging O2c anion to form an O-H

bond. During this process, TS1 is necessary to overcome a little reaction barrier of 3.1

kcal/mol. The second dissociation process follows with an overcome of 18.2 kcal/mol

reaction barrier at TS2. S-Ti5c,H2-O2c(a) configuration is formed with an endothermicity

of 17.6 kcal/mol compared to HS-Ti5c,H-O2c(a). However, this is not the only way to

form S-Ti5c,H2-O2c(a) from HS-Ti5c,H-O2c(a). In Figure 5, we can see another stepwise

pathway. TS5 is necessary for forming S-Ti5c,H-O2c,H-O2c(a) intermediate by

overcoming a 30.6 kcal/mol reaction barrier. However this intermediate tends to undergo

a H-migration via TS6 with a reaction barrier of 26.1 kcal/mol, also forming the H2O-

containing S-Ti5c,H2-O2c(a) configuration with an endothermicity of 6.1 kcal/mol. Due to

the higher reaction barriers (at TS5 and TS6), this stepwise pathway is less favorable.

Once the H2O-containing S-Ti5c,H2-O2c(a) configuration is formed, the H2O tends to

go away from the original structure. As a result, the O vacancy is generated and the S

atom goes into this vacant site. These processes accompany with an overcome of reaction

barriers of 29.6 and 39.7 kcal/mol, followed by H2O desorption from LM1 and LM2.

After these processes the first product H2O(g)+S(v+O2c) is formed.

Figure6. Optimized geometries of intermediates, transition states, and products for the H2S + TiO2 rutile

(110) reaction.

Path 2. As presented in Figure 5, the two H atoms in S-Ti5c,H-O2c,H-O2c(a)

intermediate can bound together and leave the original structure by overcoming the

highest reaction barrier of 55.2 kcal/mol to form TS7. This dehydrogenation process is

less favorable because of the highest barrier.

Kinetics. Rice-Ramsperger-Kassel- Marcus (RRKM) theory calculations using the

ChemRate program40 were used to predict the rate constants based on the PES shown in

Figure 5. Two possible pathways are illustrated:

where the rate constants, k1 and k2, are controlled by TS3 and TS7, respectively. The

predicted rate constants (in molecular units, cm3/s) in the temperature range 300-2000 K

can be represented by

The value of k1 is much larger than k2, which means that the first pathway is more

favorable.

3.5 Reaction Mechanism of H2S-TiO2 Anatase (110) Interactions.

Similar to the rutile surface, two possible pathways for the anatase (101) surface are

also considered: (1) formation of H2O(g) and an oxygen vacancy; (2) formation of H2(g).

Figure 7 shows the potential energy profiles for this reaction and Figure 8 shows the

optimized structures of transition states, intermediates, and products.

Path 1. Similar to the rutile surface, H2S-Ti5c(a) is formed at first when there is the

adsorption of H2S on the anatase (101) surface with an exothermicity of 11.4 kcal/mol.

Then also the dissociation of H2S happens. The dissociating H atom tends to form bond

with an adjacent bridging O2c anion via TS8 with a reaction barrier of 7.1 kcal/mol. Also

like the rutile surface, to form S-Ti5c,H2-O2c(a) from HS-Ti5c,H-O2c(a) there are two

possible pathways: (1) to form TS11 by overcoming the barrier of 27.2 kcal/mol; (2) a

stepwise pathway to form S-Ti5c,H-O2c,H-O2c(a) by overcoming a 10.4 kcal/mol reaction

barrier at TS9 and then follows the H-atom migration process to form HS-Ti5c,H-O2c(a)

by overcoming a 32.5 kcal/mol energy barrier. Comparing these two mechanisms, we

found that the first one is favorable due to the lower energy barrier.

Once HS-Ti5c,H-O2c(a) configuration is formed, the H2O in this structure tends to go

away from the original structure to form H2O molecule. By overcoming a high-energy

barrier of 33.4 kcal/mol at TS12, the weakly physisorbed H2O is removed from LM4 and

the S atom goes into the O vacancy. Different from the rutile surface, this is not the only

way to form H2O(g)+S(v-O2c) product. Another way is that: in the intermediate HS-

Ti5c,H-O2c(a), the OH radical can easily rotate to another direction to form LM6, by

overcoming energy barrier of 3.1 kcal/mol at TS15. Then the H atom on the rotating OH

radical can form bond with the neighboring OH radical—forming a H2O species in LM7.

By overcoming energy barrier of 19.4 kcal/mol at TS16, the H2O species is removed to

form a H2O molecule and the S atom goes into the O vacant site. Due to the formation of

an O-Ti5c bond (2.233 Å) and hydrogen bond (2.187 Å), LM7 is more stable than the

other H2O containing intermediates, LM1, LM2, and LM4.

Path 2. As shown in Figure 7, in the OH-containing S-Ti5c,H-O2c,H-O2c(a)

intermediate the two H atoms can bound together and try to go away just like on the rutile

surface. LM5 via TS13 with a reaction barrier of 46.6 kcal/mol is formed followed by the

remove of H2 molecule. Compared with the H2O(g) formation process, this H2-

elimination pathway is less favorable due to the higher reaction barrier.

Figure 8. Optimized geometries of intermediates, transition states, and products for the H2S + TiO2 anatase

(101) reaction.

Kinetics. Similar to the rutile surface, RRKM theory using the ChemRate program40

was carried out based on the PES for the reaction of H2S on TiO2 anatase (101) surface.

Also two possible pathways are illustrated:

where the rate constants, k1 and k2, are controlled by TS6 and TS13, respectively. The

predicted rate constants (in molecular units, cm3/s) in the temperature range 300-2000 K

can be represented by

The value of k1 is much larger than k2, which means that the first pathway is more

favorable.

4. Conclusion

The adsorption and reaction of H2S and TiO2 surface have been studied by using

period DFT calculations. The adsorptions of H2S, HS, S, and H on TiO2 surface were

separately considered. We found that the H2S, HS, S, and H preferentially adsorb at the

Ti5c, O2c, (Ti5c)2, and O2c sites, respectively on the rutile (110) surface and at the Ti5c,

(Ti5c)2, (-O2c)(-Ti5c), and O2c sites, respectively, on the anatase (101) surface. According to

the potential energy profiles for the reactions of H2S on the rutile (101) and anatase (101)

surface respectively, two possible products were illustrated: one is H2O(g) + S(v-O2c),

and the other one is H2(g) + S-(Ti5c)2(a). The rate constants for these two pathways in the

temperature range 300-2000 K were predicted.

                                                                                                               1 Hadjiivanov, K. I.; Klissurski, D. G. Chem. Soc. ReV. 1996, 25, 61.

2 Linsebigler, A. L.; Lu, G.; Yates, J. T. Chem. ReV. 1995, 95, 735.

3 Huisman, C. L.; Goossens, A.; Schoonman, J. Chem Mater 2003, ���15, 4617.

4 Li, D.; Gu, C.; Guo, C.; Hu, C. Chem Phys Lett 2004, 385, 55.

5 Yu, P.; Zhu, K.; Norman, A. G.; Ferrere, S.; Frank, A. J.; Nozik, A. ���J. J phys Chem B 2006, 110,

25451.

6 Blackburn, J. L.; Celmarten, D. C.; Nozik, A. J. J Phys Chem B ���2003, 107, 14154.

7 Vogel, R.; Hoyer, P.; Weller, H. J Phys Chem 1994, 98, 3183.

8 Fujishima, A.; Honda, K. Nature 1972, 138, 37.

9 Gratzel, M. Nature 2001, 414, 338.

10 Asahi, R.; Morikawa, T.; Ohwaki, T.; Aoki, K.; Taga, Y. Science 2001, 293, 269.

11 Wang, J.-H.; Lin, M. C.; Sun, Y.-C. J. Phys. Chem. B 2005, 109, 5133.

12 Rego, L. G. C.; Batista, V. S. J. Am. Chem. Soc. 2003, 125, 7989.

                                                                                                                                                                                                                                                                                                                                         13 Duncan, W. R.; Stier, W. M.; Prezhdo, O. V. J. Am. Chem. Soc. 2005, 127, 7941.

14 Harris, L. A.; Quong, A. A. Phys. ReV. Lett. 2004, 93, 086105.

15 Fahmi, A.; Ahdjoudj, J.; Minot, C. Catal. Today 1996, 352, 529.

16 Li, D.; Haneda, H.; Hishita, S.; Ohashi, N. Chem. Mater. 2005, 17, 2596.

17 Zhao, W.; Ma, W.; Chen, C.; Zhao, J.; Shuai, Z. J. Am. Chem. Soc. 2004, 126, 4782.

18 Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B., Jr. Science 2002, 297, 2243.

19 Lin, Z.; Orlov, A.; Lambert, R. M.; Payne, M. C. J. Phys. Chem. B 2005, 109, 20948.

20 Valentin, C. D.; Pacchioni, G.; Selloni, A.; Livraghi, S.; Giamello, E. J. J. Phys. Chem. B 2005,

109, 11414.

21 Sathish, M.; Viswanathan, B.; Viswanath, R. P.; Gopinath, C. S. Chem. Mater. 2005, 17, 6349.

22 Diwald, O.; Thompson, T. L.; Goralski, E. G.; Walck, S. D.; John T. Yates, J. J. Phys. Chem. B

2004, 108, 52.

23 Sakthivel, S.; Kisch, H. ChemPhysChem 2003, 4, 487.

24 Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B., Jr. Science 2002, 297, 2243.

                                                                                                                                                                                                                                                                                                                                         25 Tian, F.; Liu, C. J. Phys. Chem. B 2006, 110, 17866.

26 Fahmi, A.; Ahdjoudj, J.; Minot, C. Surf. Sci. 1996, 352-354, 529.

27 Rodriguez, J. A.; Hrbek, J.; Chang, Z.; Dvorak, J.; Jirsak, T. Phys. ReV. B 2002, 65, 235414.

28 Smith, M. F.; Setwong, K.; Tongpool, R.; Onkaw, D.; Na- phattalung, S.; Limpijumnong, S.;

Rujirawat, S. Appl. Phys. Lett. 2007, 91, 142107.

29 Chen, Y.; Jiang, Y.; Li, W.; Jin, R.; Tang, S.; Hu, W. Catal. Today 1999, 50, 39.

30 Burnside, S. D.; Shklover, V.; Barbe, C.; Comte, P.; Arendse, F.; Brooks, K.; Gratzel, M.

Chem. Mater. 1998, 10, 2419.

31 Labat, F.; Baranek, P.; Adamo, C. J. Chem. Theory Comput. 2008, 4, 341.

32 Kresse, G.; Furthmuller, J. J. Comput. Mater. Sci. 1996, 6, 15.

33 Kresse, G.; Furthmuller, J. Phys. ReV. B 1996, 54, 11169.

34 Blochl, P. Phys. ReV. B 1994, 17, 953.

35 Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244.

36 Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.; Penderson, M. R.; Singh, D. J.;

Fiolhais, C. Phys. ReV. B 1992, 46, 6671.

                                                                                                                                                                                                                                                                                                                                         37 Herman, G. S.; Dohnalek, Z.; Ruzycki, N.; Diebold, U. J. Phys. Chem. B 2003, 107, 2788.

38 Henkelman, G.; Uberuaga, B. P.; Jo ̈nsson, H. J. Chem. Phys. 2000, 113, 9901.

39 Mills, G.; Jonsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305.

40 Mokrishin, W. B., V.; Tsang, W.; Zachariah, M.; Knyazev, V. ChemRate, version 1.5.2;

National Institute of Standards and Technology: Gaithersburg, MD, 2006.

41 Henkelman, G.; Arnaldsson, A.; Jonsson, H. Comput. Mater. 2006, 36, 354.

42 Burdett, J. K.; Hughbanks, T.; Miller, G. J.; Richardson, J. W.; Smith, J. V. J. Am. Chem. Soc.

1987, 109, 3639.

43 Howard, C. J.; Sabine, T. M.; Dickson, F. Acta Crystallogr. 1991, 47, 462.

44 Jiang, D. E.; Carter, E. A. J. Phys. Chem. B 2004, 108, 19140.

45 Choi, Y. M.; Compson, C.; Lin, M. C.; Liu, M. Chem. Phys. Lett. 2006, 421, 179.