A Theoretical Study of Water Adsorption and Decomposition on Low

Article pubs.acs.org/Langmuir

A Theoretical Study of Water Adsorption and Decomposition on Low-Index Spinel ZnGa2O4 Surfaces: Correlation between Surface Structure and Photocatalytic Properties Chuanyi Jia,† Weiliu Fan,*,‡ Fei Yang,‡ Xian Zhao,*,†,‡ Honggang Sun,† Pan Li,† and Li Liu† †

State Key Laboratory of Crystal Materials, Shandong University, Jinan 250100, China School of Chemistry and Chemical Engineering, Shandong University, Jinan 250100, China

‡

S Supporting Information *

ABSTRACT: Water adsorption and decomposition on stoichiometrically perfect and oxygen vacancy containing ZnGa2O4 (100), (110), and (111) surfaces were investigated through periodic density functional theory (DFT) calculations. The results demonstrated that water adsorption and decomposition are surface-structure-sensitive processes. On a stoichiometrically perfect surface, the most stable molecular adsorption that could take place involved the generation of hydrogen bonds. For dissociative adsorption, the adsorption energy of the (111) surface was more than 4 times the energies of the other two surfaces, indicating it to be the best surface for water decomposition. A detailed comparison of these three surfaces showed that the primary reason for this observation was the special electronic state of the (111) surface. When water dissociated on the (111) surface, the special Ga3c-4s and 4p hybridization states at the Fermi level had an obvious downshift to the lower energies. This large energy gain greatly promoted the dissociation of water. Because the generation of O3c vacancy defects on the (100) and (110) surfaces could increase the stability of the dissociative adsorption states with few changes to the energy barrier, this type of defect would make the decomposition of water molecules more favorable. However, for the (111) surface, the generation of vacancy defects could decrease the stability of the dissociative adsorption states and significantly increase their energy barriers. Therefore, the decomposition of water molecules on the oxygen vacancy defective (111) surface would be less favorable than the perfect (111) surface. These findings on the decomposition of H2O on the ZnGa2O4 surfaces can be used toward the synthesis of watersplitting catalysts.

1. INTRODUCTION With economic growth, energy crises and environmental pollution are becoming increasingly severe problems.1 As one of the most hopeful technologies for solving these problems, photocatalysis has been investigated both experimentally and theoretically over the past decades.2−6 This technology has been widely used in the field of photoelectrochemical water splitting,7,8 dye-sensitized solar cells,9,10 and decomposition of harmful pollutants.11−13 Because these applications involve a heterogeneous catalytic process, the surface structure of the catalyst particle plays an important role in the reaction performance, which is typically called the structure−reactivity relationship.14 In recent years, chemists have synthesized a large number of catalysts with different morphologies, and the control of catalysts’ surface structures is growing into a new popular field of study.15−21 Although there have been speculations on the mechanism of the heterogeneous catalytic process based on some experimental observations, the underlying relationships between the experimental parameters and the internal microscopic structures remain unclear and © XXXX American Chemical Society

await elucidation. The internal microscopic mechanism that plays an important role in designing efficient catalyst is also rarely clarified. For example, there has been no explanation for different catalytic performances of the same catalyst prepared under different conditions. To help resolve these issues, a theoretical examination of the relationship between the microstructure and reaction activity of the catalyst is necessary. ZnGa2O4 with spinel-type structure has received much attention owing to its potential applications in areas such as water splitting,22,23 degradation of pollutants,24−26 and photoreduction of CO2.27 Because all of these applications involve an aqueous environment, a fundamental understanding of water adsorption behavior on different surfaces is essential for progress in these fields. The interaction of water with the ZnGa2O4 surface is therefore a subject of great interest to scientists in basic research as well as to industrial engineers. Received: December 13, 2012 Revised: May 10, 2013

A

dx.doi.org/10.1021/la4014812 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Figure 1. Optimized structures and the crystallographic planes and directions of the ZnGa2O4 (100), (110), and (111) surfaces. The surface sites and Mulliken charges carried by them are labeled on the top view of each surface. Color coding: red, O atoms; brown, Ga atoms; gray, Zn atoms.

set within a specified energy cutoff, which was chosen to be 300 eV. The Monkhorst−Pack scheme κ-point grid sampling was set as 2 × 2 × 1 for the irreducible Brillouin zone. The convergence criteria for the structural optimization and energy calculation were 2 × 10−6 eV/atom for a self-consistent field (SCF), 2 × 10−5 eV/atom for energy, 0.05 eV/Å for the maximum force, 0.1 GPa for the maximum stress, and 2 × 10−3 Å for the maximum displacement. In order to determine accurate activation barriers of the reaction, we chose the complete LST/QST approach to search for transition states of the reactions. We first calculated the bulk crystal structure of the spinel ZnGa2O4, which yielded the following lattice parameters: a = b = c = 8.460 Å; α = β = γ = 90°. They are in good agreement with the experimental parameters:33 a = b = c = 8.330(5) Å; α = β = γ = 90°. To ensure that our calculation methods were credible, we also compared the surface state of the (111) surface obtained using the PW91 and other GGA methods (rPBE (Revised Perdew−Burke−Ernzerhof) and PBE (Perdew−Burke−Ernzerhof)). As shown in Figure S1 of the Supporting Information, there was nearly no difference in the surface states calculated by these three methods. Because this means that different computational methods have little effect on the electronic structures of ZnGa2O4 surfaces, we chose the PW91 method whose lattice parameters were in better agreement with the experimental parameters. Moreover, because these three surfaces are all polar surfaces, the calculations of their crystal structure and surface states in the slab geometry require particular care, especially regarding the choice of the slab terminations and the convergence with respect to slab thickness. Ga−Zn−O terminated surfaces are often chosen for studies because this type of surface has much lower surface energies than the Ga−O or Zn−O terminated surface (Table S1). Among these three surfaces, the (100) and

However, a clear and general picture is not yet available for the interaction between water molecules and the ZnGa2O4 surface because of the great complexity of the ZnGa2O4 surface structure. Therefore, a theoretical understanding of the influence of the ZnGa2O4 surface structure has become extremely important for the design and synthesis of highefficiency ZnGa2O4 photocatalysts. In the present work, the structural sensitivity of ZnGa2O4 was studied by examining the adsorption of water on three lowindex stoichiometric ZnGa2O4 (100), (110), and (111) surfaces. The results showed that the ZnGa2O4 surfaces were indeed ready for H2O activation and this reaction is structuresensitive. The aim of this paper is to expand our theoretical understanding of the interaction between water molecules and the ZnGa2O4 surface as well as the chemistry of the latter. The rest of the paper is organized as follows: in section 2, the computational methods and the supercell models used are briefly described; in section 3, we present and discuss our results for H2O adsorption on the ZnGa2O4 surfaces. Finally, in section 4, we end our paper with a conclusion of our main results.

2. COMPUTATIONAL DETAILS AND MODELS All of the periodic density functional theory (DFT)28 calculations were carried out using codes from the Cambridge Sequential Total Energy Package (CASTEP).29 An ultrasoft pseudopotential30 was chosen to deal with the interaction between ion cores and valence electrons. For Zn the 3d and 4s states were treated as valence states, whereas for Ga the 3d, 4s, and 4p states were treated as valence states. The exchange correlation energy was described by the Perdew’s and Wang’s 1991 (PW91) functional within the generalized gradient approximation (GGA).31,32 The Kohn−Sham wave functions for the valence electrons were expanded on a plane-wave basis B


Langmuir

Article

(111) surfaces were simulated by the periodic (2 × 2) slab models. After completing the convergence test (Table S2), we chose seven layers and 16 units of ZnGa2O4 (the total number of atoms was 112). Here, one unit is defined as one Zn atom, two Ga atoms, and four O atoms. For simulation of the (110) surface, we used a (2 × 1) slab model. The number of layers and units were 6 and 12 (84 atoms), respectively (the convergence test is shown in Table S2 as well). The slabs were all separated by 20 Å thick vacuums (the convergence tests, shown in Figure S2, also showed that this was the most reasonable thickness). The lengths of the three directions of the models were larger than 10 Å, which was sufficient to screen the self-interaction effects of the periodic boundary conditions. The atomic positions of the top three layers were allowed to relax, whereas the other layers were fixed in space to simulate the bulk effects. For the equilibrium structure, the Mulliken populations were investigated using a projection of the planewave states onto a linear combination of atomic orbital basis sets,34,35 which is a widely used approach to perform charge transfers and population analyses. The binding energy of the adsorbed H2O is defined as

Five models were constructed to determine the adsorption energy of the system (Figure 2). In the first three models, the

Figure 2. All the possible models of adsorbed H2O on the ZnGa2O4 surfaces.

H2O molecule acted as a Lewis base and the Ow−M bond would play an important role in water adsorption. In the last two models, the H2O molecule acted as Lewis acid and the H− O bond would be the main driving force for water adsorption. However, when we use the last two models, the calculation was either not convergent or quite unstable (Figure S3), which means that the H2O molecule mainly acted as Lewis base when it interacted with the ZnGa2O4 surfaces. Thus, we only list the stable adsorption types of the first three models (Figures 2a−c) in the following sections. 3.1.2. Water Adsorption and Decomposition on the Perfect (100) Surface. In this section, we will explore water adsorption on the clean (100) surface. As is shown in Figure 1’s Surf-(100) on the left, among the four types of surface sites, the 2-fold-coordinated Zn (Zn2c), 5-fold-coordinated Ga (Ga5c), and 3-fold-coordinated O (O3c) were from the 4-foldcoordinated Zn (Zn4c), 6-fold-coordinated Ga (Ga6c), and 4fold-coordinated O (O4c) of the bulk ZnGa2O4 molecule, respectively. They were therefore coordinatively unsaturated. Only the 4-fold-coordinated O (O4c), which is from O4c of the bulk ZnGa2O4 molecule, is saturated. The structures of the ZnGa2O4 surface with four types of adsorbed H2O molecules at the Zn2c site are shown in Figure 3; the computed binding energies and the structure parameters for each are listed in Table 1. Four stable structures were obtained, with the binding modes of the H2O molecule perpendicular to the surface (ZN-1), the H2O molecule perpendicular to the surface and the atoms H1 and H2 are rotated around the z-axis from the ZN-1 configuration (ZN-2), one H atom pointing to O3c (ZN-3), and the H2O molecule parallel to the surface (ZN4). From the viewpoint of adsorption energy, ZN-4 adsorption type resulted in the most energetically favorable state (Table 1) because of its special structure. As shown in Figure 3’s ZN-4, its two H−O3c bonds lengths were 1.69 and 2.40 Å, which were close enough to each other to generate hydrogen bonds (about 1.5−2.5 Å) that could significantly enhance the stability of the state. Furthermore, the generation of hydrogen bonds could allow the three atoms of the H2O molecule (H1, Ow, and H2) and the three atoms on the surface (Zn4c and two O3c) to form a six-membered ring, which was also an important reason for the higher stability of the ZN-4 adsorption state.37 After elucidating the molecular adsorption process, we now turn to the dissociation of water (denoted as ZN-dis) on the stoichiometric (100) surface. From the equilibrium configuration in Figure 4, we can see that the adsorbed O atom of the H2O molecule could be attached to the Zn2c atom to form a Zn2c−Ow bond with a bond length of 1.87 Å. Simultaneously, the proton approached an O3c atom near the surface to form a surface hydroxyl at 0.98 Å. Based on the hydroxyl left by the dissociated H2O molecule, this surface is called a bihydroxy-

E b = Esur + E H2O − Eadsorb

where Esur is the total energy of the bare slab of the surface, EH2O is the total energy of free H2O, and Eadsorb is the total energy of the slab with the adsorbed H2O molecule on the surface. Therefore, a positive Eb value corresponds to exothermic adsorption.

3. RESULTS AND DISCUSSION 3.1. H2O Adsorption and Dissociation on the Perfect Surfaces. 3.1.1. Structures of the Perfect ZnGa2O4 Surfaces. The relaxed structure of the ZnGa2O4 (100), (110), and (111) surfaces are shown in Figure 1. On these surfaces, 2-foldcoordinated zinc (Zn2c), 3-fold-coordinated zinc (Zn3c), 3-foldcoordinated gallium (Ga3c), 4-fold-coordinated gallium (Ga4c), 5-fold-coordinated gallium (Ga5c), 4-fold-coordinated O (O4c), and 3-fold-coordinated O (O3c) were exposed, among which Zn2c, Zn3c, Ga3c, Ga4c, Ga5c, and O3c were coordinatively unsaturated. All of the surfaces consisted of metal (Zn and Ga) and oxygen atoms, but they differed in surface structure and composition such as the degree of unsaturation and the surface roughness. This means that they would have similarities and differences for H2O adsorption and activation. To further understand the distribution of active sites on the surfaces, we also calculated the Mulliken charge population for the surface atoms, which is shown in the top view of Figure 1. Although the absolute value of the Mulliken charge has no actual value, its relative value can clearly show us the distribution of charges and the strength of bonds. In this study, our calculations indicated that the low-coordinated metal atoms all carried positive charges, which means these sites would become strong Lewis acids that can adsorb the H2O molecules through their O atoms.6 In contrast, the surface O atoms, which carried negative charges, would act as strong Lewis bases adsorbing the H atoms of the H2O molecules. To correlate the surface atomic and electronic structures with photocatalytic activity, interactions between the H2O molecule and surfaces of the ZnGa2O4 photocatalyst are examined in the following sections. For clarity, the two hydrogen atoms of the adsorbed H2O molecule are labeled as H1 and H2 if they are not in equivalent positions and the oxygen atom is labeled as Ow. C


Langmuir

Article

Figure 4. Top (left) and side (right) views of water dissociative adsorption on ZnGa2O4 (100) surface. See Figures 1 and 3 for color coding.

Figure 5. Local density of states (LDOS) for the molecular and dissociative adsorption of H2O on the ZnGa2O4 (100) surface: (a) the free water molecule, (b) the most stable molecular adsorption (ZN-4), and (c) the dissociative adsorption (ZN-dis). The Fermi level is shown by the vertical dotted line.

on the surface, the H2O molecule’s LDOS had a distinct shift toward lower energies, indicating strong charge transfer between the H2O molecule and the ZnGa2O4 (100) surface. In addition, the hybridization of the H2O molecule with the surface mainly occurred between its lone pairs of electrons (3a1 and 1b1) with the Zn2c atom’s 3d and 4s states. Compared to the ZN-4 state (Figure 5b), the four peaks (1b2, 3a1, 1b1, and 4a1) of the ZN-dis state (Figure 5c) were mixed together, and the base structure of the H2O(dis) LDOS was completely changed. This means that the interactions between the H2O molecule and the surface in the ZN-dis were larger than that in the ZN-4. However, a comparison of the adsorption energy (Table 1) showed that the adsorption energy of the ZN-dis state was lower than that of the ZN-4 state. We will review the reason causing this phenomenon in the Discussion section. In addition, the adsorption of the H2O molecule on the Ga5c site was also calculated. However, no matter how we adjusted the position of the H2O molecule, the calculation was not convergent and the H2O molecule tended to remain far away from the surface, implying that spontaneous H2O adsorption on the Ga5c site was impossible. This phenomenon can be explained by the special structure of the (100) surface. From Figure 1’s Surf-(100), it can be seen that although Ga5c (1.16|e|) on the optimized surface carried more positive charges than Zn2c (0.85|e|), all of them were buried under the surface and surrounded by four oxygen atoms. The H2O molecule must overcome the higher steric effect to reach the Ga5c site. Furthermore, all four surrounding oxygen atoms carried negative charges, which could engender a large repulsive

Figure 3. Top (left) and side (right) views of water molecular adsorption types on ZnGa2O4 (100) surface. ZN-1: the H2O molecule perpendicular to the surface; ZN-2: the H2O molecule perpendicular to the surface and the atoms H1 and H2 are rotated around the z-axis from the ZN-1 configuration; ZN-3: one H atom pointing to O3c; ZN4: the H2O molecule parallel to the surface. Color coding: white, H atoms; others are the same as in Figure 1.

Table 1. Structural Parameters and Adsorption Energies of H2O Adsorbed in Different Configurations on Perfect ZnGa2O4 (100) Surface adsorption type

H1−Ow bond (Å)

H2−Ow bond (Å)

H1−Ow−H2 angle (deg)

ΔEad (eV)

ZN-1 ZN-2 ZN-3 ZN-4 ZN-dis free

0.98 0.98 0.98 1.07 0.98 0.98

0.98 0.98 0.98 0.98 3.10 0.98

105.4 106.1 106.5 106.2 133.0 104.8

0.33 0.29 0.39 1.00 0.55

lated surface compared to the molecular adsorption surface.38 To shed more light on the adsorbate mode, we compared the local density of states (LDOS) between the most stable molecular adsorption type (ZN-4) and dissociative adsorption, which offered qualitative insight into the bonding interaction of water and the surface. As revealed in Figure 5, upon adsorption D


Langmuir

Article

force toward the oxygen atom in the H2O molecule. Inversely, the Zn2c atom was protruding from the surface and surrounded by only two oxygen atoms, so the H2O molecule could be easily adsorbed. The above analysis illustrates that the H2O molecule is more inclined to be adsorbed at a position that can generate hydrogen bonds and further generate a multiple ring at the same adsorption site. For different adsorption sites, the H2O molecule has a higher tendency to be adsorbed on a site that sticks out of the surface and is surrounded by fewer oxygen atoms, so it can avoid the steric effect and repulsive force caused by the surrounding oxygen atoms. Moreover, because dissociative adsorption has lower adsorption energy than the most stable molecular adsorption (ZN-4), molecular adsorption on the (100) surface will be more favorable thermodynamically. 3.1.3. Water Adsorption and Decomposition on the Perfect (110) Surface. For the ZnGa2O4 (110) surface, our calculations showed that the 3-fold-coordinated Zn (Zn3c), 4fold-coordinated Ga (Ga4c), and 3-fold-coordinated O (O3c) surface sites, which come from the 4-fold-coordinated Zn (Zn4c), 6-fold-coordinated Ga (Ga6c), and 4-fold-coordinated O (O4c) of the bulk ZnGa2O4, respectively, were exposed on the surface. As can be seen in Figure 1, these three surface sites are all coordinately unsaturated. After testing all high-symmetry sites offered by molecular adsorption on the (110) surface, we obtained three stable adsorption types on the Ga4c site and two stable adsorption types on the Zn3c site (Figure 6) with the following corresponding configurations: the H2O molecule perpendicular to the surface (GA-1); one H atom pointing to O3c (GA-2); the Ow atom between two Ga4c (GA-3) atoms at the Ga4c site and the H2O molecule perpendicular to the surface (ZN-1); one H atom pointing to O3c (ZN-2). We could clearly see from the structural parameters, as shown in detail in Table 2, that the most stable adsorption type on the Ga4c site was GA-3, and the most stable adsorption type on the Zn3c site was ZN-2. For the Ga4c site, the optimized structure (Figure 6) showed that the GA-3 adsorption type had the highest symmetry. Moreover, because the Ow atom of the GA-3 state was at the bridge site, it could bond with two Ga4c atoms. The high symmetry and the auxiliary adsorption of the other Ga4c atom could both increase the stability of the GA-3 structure. For the ZN-2 adsorption type, because the distance between H1 and O3c was 2.22 Å, the two atoms were close enough to generate a hydrogen bond and further generate a four-membered ring comprising H1, O3c, Zn3c, and Ow. Although this structure was less stable than the six-membered ring, it could enhance the stability of the adsorption system as well, which accounted for the phenomenon in which the Ga4c site had stronger acidity but lower adsorption energy than the Zn3c site. Having discussed the molecular adsorption on the (110) surface, we now turn our efforts to the dissociative adsorption of water. For the Zn3c site, the computed distance between the Zn3c and the O3c atoms was too small for dissociative adsorption (our calculations showed that the OH radical and atomic H recombined to form H2O again after optimization), so dissociative adsorption could only occur on the Ga4c sites. As illustrated in Figure 7, the dissociative adsorption of H2O (denoted as GA-dis) could form a bihydroxylated surface as well, with bond lengths of 0.99 and 1.86 Å for Ob−H2 and Ga4c−Ow, respectively.

Figure 6. Top (left) and side (right) views of water molecular adsorption types on ZnGa2O4 (110) surface. GA-1: the H2O molecule perpendicular to the surface on Ga4c site; GA-2: one H atom pointing to O3c on Ga4c site; GA-3: the Ow atom between two Ga4c atoms at the Ga4c site; ZN-1: the H2O molecule perpendicular to the surface on Zn3c site; ZN-2: one H atom pointing to O3c on Zn3c site. Color coding: white, H atoms; others are the same as in Figure 1. See Figures 1 and 3 for color coding.

We also performed LDOS analyses for the dissociative adsorption and the two most stable molecular adsorption types (GA-3 and ZN-2) to aid in the understanding of the H2O molecule’s adsorption behavior on the (110) surface. As displayed in Figure 8, the LDOS of the H2O molecule had a E


Langmuir

Article

favorable. However, both molecular and dissociative adsorption energies of the (110) surface were less than the energies of the (100) surface because of the different structures of these two clean surfaces. Compared to the (100) surface, the metal adsorption sites of the (110) surface were all buried under the plane of the (110) surface. We have learnt from the earlier analysis that this type of structure could cause stronger steric hindrance that would render water adsorption more difficult. 3.1.4. Water Adsorption and Decomposition on the Perfect (111) Surface. Similar to the (100) surface, our calculations revealed four types of adsorption sites on the (111) surface, as shown in Figure 1’s Surf-(111). Among them, the 3fold-coordinated Zn (Zn3c), 3-fold-coordinated Ga (Ga3c), and 3-fold-coordinated O (O3c), which come from the 4-foldcoordinated Zn (Zn4c), 6-fold-coordinated Ga (Ga6c), and 4fold-coordinated O (O4c) of the bulk ZnGa2O4 molecule, respectively, were coordinatively unsaturated. Thus, they would be good adsorption sites for the O or H atom of the H2O molecule. Only the 4-fold-coordinated O (O4c) was coordinatively saturated in our results. For molecular adsorption, we obtained three stable adsorption types on the Ga3c site and only one stable adsorption type on the Zn3c site (Figure 9) with the following configurations: the H2O molecule perpendicular to the surface (GA-1); the H2O molecule parallel to the surface (GA-2); one H atom pointing to O3c (GA-3) for the Ga4c site and the H2O molecule parallel to the surface (ZN-1) for the Zn3c site. Comparing the adsorption energies (Table 3), we determined that the GA-3 adsorption type was more stable than the other three states owing to the generation of the hydrogen bond. As clearly shown in Figure 9’s (GA-3), the distance of 2.19 Å for H1−O3c was low enough for the formation of a hydrogen bond. Although it could not further generate a multiple ring, this strong interaction between the H1 and O3c atoms could increase the stability of the adsorption system as well. In the case of dissociative adsorption on the Ga3c site, as illustrated in Figure 10, the dissociation of H2O molecule (denoted as GA-dis) could also form a bihydroxylated surface. The bond lengths obtained for Ob−H1 and Ga3c−Ow were 0.98 and 1.77 Å, respectively. However, because of the large steric hindrance and the competition of the neighboring Ga3c atoms, dissociative adsorption on the Zn3c site would be very difficult. Our calculations also indicated that the dissociative adsorption energy for water on the Zn3c site was even lower than molecular adsorption on the Ga3c or Zn3c site (Table 3). This means that dissociative adsorption on the Zn3c site would be less favorable thermodynamically. Figure 11 shows the results of a further LDOS study of the interaction between the H2O molecule and (111) surface for the most stable molecular adsorption state on each site and the most stable dissociative adsorption configurations (GA-3, ZN-2, and GA-dis). For either molecular or dissociative adsorption, the LDOSs of the H2O molecule had a distinct shift toward lower energies when it was adsorbed on the surface. Similar to the (110) surface, hybridization of the Ga3c site mainly occurred in the 4s and 4p states and the hybridization of the Zn3c site mainly occurred in the 3d and 4s states. The most significant changes of the water orbital for the GA-dis state (Figure 11d) indicated that the interaction between Ga3c and Ow for dissociative adsorption was stronger than that for molecular adsorption. Moreover, a comparison of the adsorption energies (Tables 1−3) clearly showed that the dissociative adsorption energies of the (111) surface were more than 4 times the energies of the other two surfaces. Therefore,


H1−Ow bond (Å)

H2−Ow bond (Å)


ΔEad (eV)

GA-1 GA-2 GA-3 ZN-1 ZN-2 GA-dis

0.82 0.98 0.98 0.98 0.98 0.98

0.98 0.98 0.98 0.98 0.98 2.82

107.6 107.7 107.0 104.6 106.9 141.0

0.22 0.20 0.25 0.21 0.28 0.52


Figure 8. Local density of states (LDOS) for the molecular and dissociative adsorption of H2O on the ZnGa2O4 (110) surface: (a) the free water molecule, (b) the most stable molecular adsorption on Ga4c site (GA-3), (c) the most stable molecular adsorption on Zn3c site (ZN-2), (d) the dissociative adsorption (GA-dis). The Fermi level is shown by the vertical dotted line.

distinct shift toward lower energies when it was adsorbed on the surface as well. On the Ga4c site, the H2O molecule mainly interacted with the surface via hybridization between the lone pair state (1b1 and 3a1) and the Ga4c-4s and Ga4c-4p states (Figure 8b). The hybridization of the Zn3c site (Figure 8c) was similar to the Zn2c site on the (100) surface; besides the valence electron in the 4s orbital, the electrons in the 3d orbital also joined in the hybridization. The largest change of the water orbital for the GA-dis state (Figure 8d) signified that the interactions between the H2O molecule and the dissociative adsorption surface were stronger than molecular adsorption. A comparison of the adsorption energies of all the states (Table 2) showed that the dissociative adsorption type was more stable than other types of adsoption. Thus, the dissociative adsorption on the (110) surface would be thermodynamically more F


Langmuir

Article


Figure 11. Local density of states (LDOS) for the molecular and dissociative adsorption of H2O on the ZnGa2O4 (111) surface: (a) the free water molecule, (b) the most stable molecular adsorption on Ga3c site (GA-3), (c) the most stable molecular adsorption on Zn3c site (ZN-1), and (d) the dissociative adsorption (GA-dis). The Fermi level is shown by the vertical dotted line.

Figure 9. Top (left) and side (right) views of water molecular adsorption types on ZnGa2O4 (111) surface. GA-1: the H2O molecule perpendicular to the surface on Ga3c site; GA-2: the H2O molecule parallel to the surface on Ga3c site; GA-3: one H atom pointing to O3c on Ga4c site; ZN-1: parallel to the surface on Zn3c site. See Figures 1 and 3 for color coding.

study of dissociative adsorption on the ZnGa2O4 surfaces becomes especially important. The above discussions have shown us in detail the changes of structure and energy of molecular and dissociative adsorption of water on different surfaces, but the primary source of the large dissociative adsorption energy of the (111) surface has yet to be determined. It is known that the surface states caused by dangling covalent bonds have an obvious effect on the charge density distribution at the solid−vacuum interface and chemical processes at the surface.42,43 Therefore, we need a detailed investigation of the state in which surface atoms participate in dissociative adsorption of water. After comparing the total surface states (Figure 12), we found that there were more states at the Fermi level (0 eV) of the (111) surface. The LDOS of the metal atoms on the (100), (110), and (111) surfaces (Figure 13) clearly showed that there was a special state at the Fermi level of the (111) surface’s Ga3c atom, but there was no such state on the other two surfaces’ atoms. (In order to visualize the scenario, we also calculated the 3D orbital of this


H1−Ow bond (Å)

H2−Ow bond (Å)


ΔEad (eV)

GA-1 GA-2 GA-3 ZN-1 ZN-dis GA-dis

0.98 0.98 0.98 0.98 0.98 0.98

0.98 0.98 1.00 0.99 4.01 3.99

107.8 107.2 107.6 105.8 94.0 83.9

0.40 0.54 0.62 0.60 0.51 2.44

it became clear that an in-depth comparison of the three surfaces was necessary to find the primary cause that favored dissociative adsorption on the (111) surface. 3.1.5. Discussion. In the field of photocatalysis, water not only provides the reaction environment, it is also the key participant as it can decompose to produce hydrogen, oxygen, and hydroxyl radicals for further reactions.39−41 Therefore, the G


Langmuir

Article

the position of its Ga4c-4s and Ga4c-4p states did not show significant changes. Therefore, we call these two surfaces the “single-driving-force” surfaces. We then determined that the special surface state of the (111) surface (Figure 14) at the Fermi level was generated by the hybridization of the Ga3c-4s and Ga3c-4p states. When water dissociated on the (111) surface, besides the energy gain from the downshift of the O2p state, the formation of the Ga3c−Ow bond could greatly reduce the energy of the Ga3c-4s and Ga3c-4p levels. This large energy gain would promote the dissociation of water as well. Our earlier analysis showed that the O3c atom was the primary driving force of the large energy gain on the (100) and (110) surfaces. However, for the (111) surface, besides the O3c atom, the Ga3c atom could also provide a large energy gain. Therefore, we call it a “bi-driving-force” surface, and it would be much more effective for water decomposition than the “singledriving-force” surface. The transition states for dissociative adsorption on these three perfect surfaces are all provided in Table 4. It can be seen that the energy barrier of dissociative adsorption on the (111) surface was the lowest and that of dissociative adsorption on the (100) surface was the highest. This could further provide a kinetics proof for our conclusion that the (100) surface was not a good surface for water decomposition compared to the other two surfaces; the best surface for water decomposition was the (111) surface, which had the highest adsorption energy and the lowest energy barrier. 3.2. H 2 O Adsorption and Dissociation on the Defective Surfaces. 3.2.1. Structures of the Defective ZnGa2O4 Surfaces. In this section, we will explore H2O adsorption on the defective ZnGa2O4 surfaces with an oxygen vacancy. As is shown in Figure S5, our calculations showed five types of oxygen vacancy defects in total on these three surfaces: O3c and O4c vacancies on the (100) surface, O3c vacancy on the (110) surface; O3c and O4c vacancies on the (111) surface. The formation energy of an oxygen vacancy is defined with respect to the energy of an oxygen molecule in the triplet state and calculated according to EVo = −(Eperfect − Edefective − 1/2EO2). Our calculated oxygen vacancy formation energies of these three surfaces are listed in Table S3. The formation energies decreased in the order (111)-Vo3c > (111)-Vo4c > (110)-Vo3c > (100)-Vo4c > (100)-Vo3c, which suggests that it would be easier for the oxygen vacancy defect to emerge on the (111) surface. The Mulliken charges (Figure 15) indicated that the oxygen vacancy defects could cause significant charge redistribution (the number of positive charges carried by the metal atoms beside them would become lower than the number of charges on clean surfaces). This change would decrease the acidity of the metal atoms and make adsorption on these sites less favorable. We will discuss in detail the changes of adsorption behavior on the defective (100), (110), and (111) surfaces in the following sections. 3.2.2. Water Adsorption and Decomposition on the Defective (100) Surface. Variations in the adsorption energies for the configurations in the presence of an oxygen vacancy defect (Vo3c or Vo4c) were examined; the most stable configuration is shown in Figure S6. These configurations are denoted as follows: the H2O molecule parallel to the (100)Vo3c surface (similar to the ZN-4 state in Figure 3) (Vo3c-ZN4); dissociative adsorption on the Zn site (similar to the ZN-dis state in Figure 4) (Vo3c-ZN-dis); dissociative adsorption on the oxygen vacancy site of the (100)-Vo3c surface (Vo3c-dis); the

Figure 12. Total density of states (DOS) for the bulk and the (100), (110), and (111) surfaces. The Fermi level is shown by the vertical dotted line.

Figure 13. Local density of states (LDOS) for the metal atoms which join in the dissociative adsorption of H2O on the (100), (110), and (111) surfaces and the corresponding atoms in the subsurface. The 3D orbital of the special surface state for the Ga3c atom on the (111) surface is shown beside LDOS of the (111) surface. The Fermi level is shown by the vertical dotted line.

special state, as shown in the inset beside the LDOS of the (111) surface in Figure 13.) Furthermore, this special state was located in a projected band gap, and its wave function decayed exponentially to zero when it went from the surface to the subsurface; this phenomenon could be a strong indication that this was the surface state. This type of surface state could easily hybridize with the 1b1 state of the H2O(free) molecule, which stayed at the Fermi level as well, and then promote the dissociation of the H2O molecule. In order to determine the function of this special surface state, we further compared the change of the surface states for all atoms participating in the dissociative adsorption of the H2O molecule on the (100), (110), and (111) surfaces. The active sites on the (100) surface were the Zn2c and O3c atoms (Figure 14). After undergoing dissociative adsorption, only the O3c-2p state had an obvious downshift toward lower energies; the Zn2c3d state even had a shift toward higher energies, which made the energy gain that drove the dissociation of the H2O molecule lower than those of the other two surfaces. The phenomenon of dissociative adsorption on the (100) surface with lower adsorption energies than molecular adsorption was thus readily explained. On the (110) surface, the active sites were the Ga4c and O3c atoms. The main driving force for the dissociation of H2O was provided by the downshift of the O3c-2p states, and H


Langmuir

Article

Figure 14. Changes of the surface state for the metal and O3c atoms on the (100), (110), and (111) surfaces caused by the dissociative adsorption of H2O (using the electrostatic potential in the fixed layer of the pure surface as the standard to make the LDOS of atoms on the clean surfaces aligned with the atoms on the dissociative surfaces, as is depicted in Figure S4). The Fermi level is shown by the vertical dotted line.

and two O3c atoms), which could significantly enhance the stability of the adsorption states. However, the Vo3c-ZN-4 state could no longer generate any multiple ring structure because of the loss of one O3c atom. Generally, oxygen vacancies are also active sites on the surface of metal oxides. Therefore, a further study of the adsorption behavior on the oxygen defect site is necessary. Our calculations indicated that there were only dissociative adsorption states on the O3c and O4c vacancy sites. For molecular adsorption, we discovered that no matter how we adjusted the position of the H2O molecule, the calculation was not convergent and the H2O molecule tended to stay far away from the surface, implying that spontaneous H2O molecular adsorption on these two sites was impossible. The dissociative adsorption energy of the Vo4c-dis state was much lower than the adsorption energy of the Vo3c-dis state because dissociative adsorption on the O4c vacancy site had larger steric hindrance than the O3c vacancy site. The dissociative adsorption state on the 100-Vo3c surface was even more stable than dissociative adsorption on the clean surface. This means that generation of the O3c vacancy defect on the (100) surface made dissociative adsorption more favorable thermodynamically. Moreover, the Vo3c-dis state was also more stable than the most stable molecular adsorption state Vo3c-ZN-4, indicating that unlike the perfect and O4c vacancy defective (100) surfaces, dissociative adsorption on the O3c vacancy defective (100)

Table 4. Energy Barriers for the Dissociative Adsorption States on the Perfect and Oxygen Vacancy Defective (100), (110), and (111) Surfaces (100)

perfect-Zn2c

TS-dis (eV) (110)

1.88

Vo3c-Zn2c

2.63 perfect-Ga4c

TS-dis (eV) 0.70 (111) perfect-Ga3c Vo3c-Ga3c TS-dis (eV)

0.65

3.21

Vo3c

Vo4c-Zn1c

1.92 Vo3c

1.95

Vo4c

2.35 Vo3c-Ga3c

0.86 Vo3c

Vo4c-Ga2c

0.92 Vo4c

1.24

1.62

1.89

H2O molecule parallel to the (100)-Vo4c surface (similar to the ZN-4 state in Figure 3) (Vo4c-ZN-4); dissociative adsorption on the Zn site (similar to the ZN-dis state in Figure 4) (Vo4cZN-dis); dissociative adsorption on the oxygen vacancy site of the (100)-Vo4c surface (Vo4c-dis). We discovered that the presence of such surface vacancy could modify the energy for configurations that adsorb H2O on the perfect surface at sites near the vacancy. The presence of a Vo defect led to a substantial decrease in adsorption energies in our calculations (Figure S6). Moreover, although the Zn1c site on the 100-Vo4c surface carried much fewer positive charges than the Zn2c site on the 100-Vo3c surface (Figure 15), the Vo4c-ZN-4 state was more stable than the Vo3c-ZN-4 state because the Vo4c-ZN-4 states had two hydrogen bonds and could keep the sixmembered ring structure (generated by the H1, Ow, H2, Zn4c, I


Langmuir

Article

Figure 15. Top views of the optimized structures of the defective ZnGa2O4 (100), (110), and (111) surfaces. The surface sites and Mulliken charges carried by them are labeled on each surface as well. See Figure 1 for color coding.

energies for the configurations in the presence of an oxygen vacancy defect (O3c or O4c), we obtained the most stable configuration on each site, as shown in Figure S8. These configurations are denoted as follows: the H2O molecule perpendicular to the (111)-Vo3c surface (similar to the ZN-1 state in Figure 9) (Vo3c-ZN-1); dissociative adsorption on the Ga site (similar to the GA-dis state in Figure 10) (Vo3c-GAdis); molecular adsorption on the oxygen vacancy site of the (111)-Vo3c surface (Vo3c-mol); dissociative adsorption on the oxygen-vacancy site of the (111)-Vo3c surface (Vo3c-dis); the H2O molecule parallel to the (111)-Vo4c surface (similar to the ZN-1 state in Figure 9) (Vo4c-ZN-1); dissociative adsorption on the Ga site (similar to the GA-dis state in Figure 10) (Vo4cGA-dis); molecular adsorption on the oxygen-vacancy site of the (111)-Vo4c surface (Vo4c-mol); dissociative adsorption on the oxygen vacancy site on the (111)-Vo4c surface (Vo4c-dis). Similar to the two surfaces discussed earlier, owing to the decrease in charges on the Zn atoms beside the oxygen vacancy defect (Figure 15), the molecular adsorption states on the (111)-Vo3c and (111)-Vo4c surfaces became less stable than the former state on the perfect surface. For the Ga sites, this significant decrease even prevented molecular adsorption. (Our calculations on the Ga sites were not convergent, and the H2O molecule had a tendency to remain far away from the surface, implying that spontaneous H2O adsorption on the Ga site was impossible.) For dissociative adsorption, besides the above reason, the change of the special surface state at the Fermi level also played an important role in the decrease in adsorption energy. As shown in Figure 16, the special surface state on the defective surfaces had a distinct downshift toward lower energies. This means that when water dissociated on the defective (111) surface, the energy gain from the downshift of the special surface state would be lower than that of the state on the perfect surface, but it could promote the decomposition of water as well. This explains why the dissociative adsorption states on the defective surfaces were both much more favorable than the dissociative adsorption states on the other surfaces.

surface was favored over molecular adsorption. The transition states in Table 4 also show that the energy barrier of the dissociative adsorption on the O3c defective site was similar to that of the perfect surface. Therefore, based on the above thermodynamical and dynamical effects, the generation of the O3c vacancy defect was advantageous for the decomposition of water molecules on the (100) surface. 3.2.3. Water Adsorption and Decomposition on the Defective (110) Surface. After comparing various adsorption states on the O3c vacancy defective (110) surface, we obtained five typical and stable states, as depicted in Figure S7: the Ow atom between the Ga3c and Ga4c sites (similar to the GA-3 state in Figure 6) (Vo3c-GA-3); one H atom pointing to O3c on the Zn2c site (similar to the ZN-2 state in Figure 6) (Vo3c-ZN-2); dissociative adsorption on the Ga3c site (similar to the GA-dis state in Figure 7) (Vo3c-GA-dis); molecular adsorption on the oxygen-vacancy site (Vo3c-mol); dissociative adsorption on the oxygen-vacancy site (Vo3c-dis). As shown in Figure S6, the adsorption energy of the metal sites beside the oxygen vacancy all had a significant decrease compared to the sites on the perfect surface. This can be explained by the decrease in positive charges of these metal sites as well (Figure 15). However, for the molecular adsorption (Vo3c-mol) and dissociative adsorption (Vo3c-dis) states on the O3c vacancy defective site, the adsorption energies were both much higher than those of the other sites on the defective and perfect (110) surfaces. This means that the generation of the O3c vacancy defect on the (110) surface could make dissociative adsorption more favorable thermodynamically. Similar to the (100) surface, the energy barriers of the defective surfaces also had few differences from that of the perfect surface, as shown in Table 4. Therefore, considering the above thermodynamical and dynamical effects, the generation of oxygen vacancy was advantageous for the decomposition of water molecules on the (110) surface as well. 3.2.4. Water Adsorption and Decomposition on the Defective (111) Surface. Comparing all of the adsorption J


Langmuir

Article

surfaces is a structure-sensitive reaction. Both the surface atomic structure and surface electronic structure have distinct effects on the reaction parameters. Our main findings are as follows: (1) After comparing all of the molecular adsorption types on the clean surfaces, we found two main factors that can make molecular adsorption more stable. One is the surface roughness: if the metal sites are higher than the surface plane, the H2O molecule will suffer less steric hindrance and adsorption will become much easier. The other factor is the special structure generated by the H2O molecule and the surface. When the H2O molecule is adsorbed on the surface, it is more inclined to be adsorbed at a position that can generate hydrogen bonds and further form a multiple ring. The formation of this type of structure can significantly increase the stability of the surface state. (2) For dissociative adsorption on the clean surfaces, the driving force for the dissociation of H2O molecules is provided by change of the surface state. On the (100) and (110) surfaces, the main driving force for the dissociation of H2O molecules is provided by the downshift of the O3c-2p states. However, on the (111) surface, besides the energy gain from the downshift of the O2p state, the formation of the Ga3c−Ow bond can greatly reduce the energy of the special Ga3c-4s and Ga3c4p states at the Fermi level. This large energy gain can significantly promote the dissociation of water. Therefore, compared to the (100) and (110) surfaces, we call it a “bi-driving-force” surface, and it will be much more effective for water decomposition than the “singledriving-force” surfaces. (3) For the oxygen vacancy defective surfaces, the order of the surface activity of these low-index surfaces is (111) > (111-Vo3c) > (111-Vo4c) > (110-Vo3c) > (100-Vo3c) > (100) > (110) > (100-Vo4c), as determined by the dissociative adsorption energies. Because the molecule adsorption states on the perfect (100) and (100-Vo4c) surfaces are more favorable than the dissociative adsorption states and have much larger energy barriers, they will not be good surfaces for water decomposition. However, because the generation of O3c vacancy defects on the (100) and (110) surfaces can enhance the stability of the dissociative adsorption states with little change to the energy barriers, they will be good surfaces for the decomposition of water molecules. For the (111) surface, the generation of O3c and O4c vacancy defects will decrease the stability of the dissociative adsorption states as well as significantly increase their energy barriers. The decomposition of water molecules on the oxygen vacancy defective (111) surface will thus be less favorable than the perfect (111) surface. In summary, our results demonstrate that the atomic and electronic surface structures of ZnGa2O4 are very important for H2O adsorption and activation. The large driving force generated by the surface state is the main cause for the decomposition of H2O. Our study has an important implication for understanding the decomposition of H2O on the ZnGa2O4 surfaces and can provide theoretical guidance for chemists to synthesize efficiently sized and shape-controlled ZnGa2O4 catalysts.

Figure 16. Local density of states (LDOS) for the Ga atom which join in the dissociative adsorption of H2O on the (111) surface. The Fermi level is shown by the vertical dotted line.

For dissociative adsorption on the O3c and O4c vacancy sites, their adsorption energies were larger than the energies of the Ga sites, indicating that dissociative adsorption on each defective surface had a tendency to occur on the oxygen vacancy site. Moreover, because the dissociative adsorption states on the vacancy sites were less stable than the state on the perfect surface, generation of oxygen vacancy defects on the (111) surface would make the decomposition of water less favorable thermodynamically. The transition states in Table 4 also show that the energy barriers of the defective surfaces were all much higher than those of the perfect surfaces, which means that the generation of the O3c or O4c vacancy defect would make the decomposition of water less favorable dynamically. Therefore, considering the above thermodynamical and dynamical effects, the generation of an O3c or O4c oxygen vacancy defect would make the decomposition of water molecules on the (111) surface less favorable. Comparing all of the perfect and defective surfaces in the above discussions, we arrived at the following conclusions. If we want to synthesize a ZnGa2O4 catalyst for photocatalytic H2O decomposition, we should expose the (111) surface more than the other surfaces. Moreover, the generation of oxygen vacancy defects on the (111) surface can significantly change the surface state, making the adsorption state less stable. As a result, when we synthesize a ZnGa2O4 catalyst by exposing the (111) surface, we should ensure that fewer oxygen vacancy defects are generated. However, for the (100) and (110) surfaces, the adsorption states on the O3c vacancy defect are more favorable than the other sites (including the sites on the perfect surfaces). This means that if we want to synthesize a ZnGa2O4 catalyst by exposing these two surfaces, they should generate more oxygen vacancy defects. This is therefore an important finding in the decomposition of H2O on ZnGa2O4 surfaces that can guide chemists to synthesize water-splitting catalysts with efficient sizes and controlled shapes.

4. CONCLUSIONS In this article, we present a comprehensive DFT study of the change of atomic structure and adsorption behavior of water molecules on stoichiometrically perfect and oxygen vacancy defect sites of ZnGa2O4 (100), (110), and (111) surfaces. The results show that water adsorption on the low-index ZnGa2O4 K


Langmuir

■

Article

(16) Ertl, G. Reactions at Surfaces: From Atoms to Complexity (Nobel Lecture). Angew. Chem., Int. Ed. 2008, 47, 3524−3535. (17) Semagina, N.; Kiwi-Minsker, L. Recent Advances in the LiquidPhase Synthesis of Metal Nanostructures with Controlled Shape and Size for Catalysis. Catal. Rev. 2009, 51, 147−217. (18) Goesmann, H.; Feldmann, C. Nanoparticulate Functional Materials. Angew. Chem., Int. Ed. 2010, 49, 1362−1395. (19) Burda, C.; Chen, X. B.; Narayanan, R.; El-Sayed, M. A. Chemistry and Properties of Nanocrystals of Different Shapes. Chem. Rev. 2005, 105, 1025−1102. (20) Xiong, Y. J.; Wiley, B. J.; Xia, Y. N. Nanocrystals with Unconventional ShapesA Class of Promising Catalysts. Angew. Chem., Int. Ed. 2007, 46, 7157−7159. (21) Haruta, M. Catalysis Gold Rush. Nature 2005, 437, 1098−1099. (22) Zhong, M.; Li, Y. B.; Takero, T.; Zheng, M. J.; Yamada, I.; Delaunay, J. J. Vertically Aligned ZnO-ZnGa2O4 Core-Shell Nanowires: From Synthesis to Optical Properties. J. Nanopart. Res. 2012, 14, 804−814. (23) Zhong, M.; Li, Y. B.; Yamada, I.; Delaunay, J. J. ZnO-ZnGa2O4 Core-Shell Nanowire Array for Stable Photoelectrochemical Water Splitting. Nanoscale 2012, 4, 1509−1514. (24) Sun, M. S.; Li, D. Z.; Zhang, W. J.; Chen, Z. X.; Huang, H. J.; Li, W. J.; He, Y. H.; Fu, X. Z. Rapid Microwave Hydrothermal Synthesis of ZnGa2O4 with High Photocatalytic Activity toward Aromatic Compounds in Air and Dyes in Liquid Water. J. Solid State Chem. 2012, 190, 135−142. (25) Boppana, V. B. R.; Lobo, R. F. SnOx-ZnGa2O4 Photocatalysts with Enhanced. Visible Light Activity. ACS Catal. 2011, 1, 923−928. (26) Zhang, W. W.; Zhang, J. Y.; Chen, Z. Y.; Wang, T. M. Photocatalytic Degradation of Methylene Blue by ZnGa2O4 Thin Films. Catal. Commun. 2009, 10, 1781−1785. (27) Yan, S. C.; Ouyang, S. X.; Gao, J.; Yang, M.; Feng, J. Y.; Fan, X. M.; Wan, L. J.; Li, Z. S.; Ye, J. H.; Zhou, Y.; Zou, Z, G. A RoomTemperature Reactive-Template Route to Mesoporous ZnGa2O4 with Improved Photocatalytic Activity in Reduction of CO2. Angew. Chem., Int. Ed. 2010, 49, 6400−6404. (28) Payne, M. C.; Teter, M. P.; Allan, D. C.; Arias, T. A.; Joannopoulos, J. D. Iterative Minimization Techniques for ab Initio Total-Energy Calculations: Molecular Dynamics and Conjugate Gradients. Rev. Mod. Phys. 1992, 64, 1045−1097. (29) Segall, M.; Lindan, P.; Probert, M.; Pickard, M.; Pickard, C.; Hasnip, P.; Clark, S.; Payne, M. First-Principles Simulation: Ideas, Illustrations and the CASTEP Code. J. Phys.: Condens. Matter 2002, 14, 2717−2743. (30) Clark, S. J.; Segall, M. D.; Pickard, C. J.; Hasnip, P. J.; Probert, M. J.; Refson, K.; Payne, M. C. First Principles Methods Using CASTEP. Z. Kristallogr. 2005, 220, 567−570. (31) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, A. K.; Pederson, R. M.; Singh, D. J.; Fiolhais, C. Atoms, Molecules, Solids, and Surfaces: Applications of the Generalized Gradient Approximation for Exchange and Correlation. Phys. Rev. B 1992, 46, 6671−6687. (32) Perdew, J. P.; Wang, Y. Accurate and Simple Analytic Representation of the Electron-Gas Correlation Energy. Phys. Rev. B 1992, 45, 13244−13249. (33) Hornstra, J.; Keulen, E. The Oxygen Parameter of the Spinel ZnGa2O4. Philips Res. Rep. 1972, 27, 76−81. (34) Sanchez-Portal, D.; Artacho, E.; Soler, J. M. Projection of PlaneWave Calculations into Atomic Orbitals. Solid State Commun. 1995, 95, 685−690. (35) Gonze, X. First-Principles Responses of Solids to Atomic Displacement and Homogeneous Electric Fields: Implementation of a Conjugate-Gradient Algorithm. Phys. Rev. B 1997, 55, 10337−10354. (36) Wippermann, S.; Schmidt, W. G.; Thissen, P.; Grundmeier, G. Dissociative and Molecular Adsorption of Water on α-Al2O3 (0001). Phys. Status Solidi C 2010, 7, 137−140. (37) Hass, K. C.; Schneider, W. F.; Curioni, A.; Andreoni, W. FirstPrinciples Molecular Dynamics Simulations of H2O on α-Al2O3 (0001). J. Phys. Chem. B 2000, 104, 5527−5540.

ASSOCIATED CONTENT

S Supporting Information *

The test for the exchange correlation function, the theoretical basis and convergence tests for the surface models, the adsorption behaviors of the H2O molecule acting as Lewis acid, the correction method of the LDOSs in Figure 14, the data and figures for the oxygen vacancy defective (100), (110), and (111) surfaces. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel 86-531-88366330, Fax 86-531-88364864, e-mail [email protected] (X.Z.), [email protected] (W.F.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work is supported by the 973 Program of China (Grants 2009CB930103 and 2010CB933504), National Natural Science Foundation of China (Grant 21173131), and Independent Innovation Foundation of Shandong University (Grant 2012TS212).

■

REFERENCES

(1) Heisey, L. V. Energy and Environment: Four Energy Crises. J. Chem. Educ. 1977, 54, A130−A131. (2) Huang, J. H.; Ding, K. N.; Wang, X. C.; Fu, X. Z. Nanostructuring Cadmium Germanate Catalyst for Photocatalytic Oxidation of Benzene at Ambient Conditions. Langmuir 2009, 25, 8313−8319. (3) Yan, S. C.; Li, Z. S.; Zou, Z. G. Photodegradation of Phodamine B and Methyl Orange over Boron-Doped g-C3N4 under Visible Light Irradiation. Langmuir 2010, 26, 3894−3901. (4) Yin, L. F.; Shen, Z. Y.; Niu, J. F.; Duan, Y. P. Degradation of Pentachlorophenol and 2,4-Dichlorophenol by Sequential VisibleLight Driven Photocatalysis and Laccase Catalysis. Environ. Sci. Technol. 2010, 44, 9117−9122. (5) Narasimham, M.; Rossitza, P.; Matthias, S. Coverage-Dependent Adsorption Mode of Water on Fe3O4 (001): Insights from First Principles Calculations. J. Phys. Chem. C 2010, 114, 11148−11156. (6) Zhao, Z. Y.; Li, Z. S.; Zou, Z. G. A Theoretical Study of Water Adsorption and Decomposition on the Low-Index Stoichiometric Anatase TiO2 Surfaces. J. Phys. Chem. C 2012, 116, 7430−7441. (7) Fujishima, A.; Honda, K. Electrochemical Photolysis of Water at a Semiconductor Electrode. Nature 1972, 238, 37−38. (8) Xu, H.; Zhang, R. Q.; Alan, M. C. N.; Aleksandra, B. D.; Chan, H. T.; Chan, W. K.; Tong, S. Y. Splitting Water on Metal Oxide Surfaces. J. Phys. Chem. C 2011, 115, 19710−19715. (9) O’Regan, B.; Grätzel, M. A Low-Cost, High-Efficiency Solar Cell Based on Dye Sensitized Colloidal TiO2 Films. Nature 1991, 353, 737−740. (10) Meyer, G. J. The 2010 Millenium Technology Grand Prize: Dye Sensitized Solar Cells. ACS Nano 2010, 4, 4337−4343. (11) Fujishima, A.; Rao, T. N.; Tryk, D, A. Titanium Dioxide Photocatalysis. J. Photochem. Photobiol., C 2000, 1, 1−21. (12) Diebold, U. The Surface Science of Titanium Dioxide. Surf. Sci. Rep. 2003, 48, 53−229. (13) Zou, C. W.; Rao, Y. F.; Alyamani, A.; Chu, W.; Chen, M. J.; Patterson, D. A.; Emanuelsson, E. A. C.; Gao, W. Heterogeneous Lollipop-like V2O5/ZnO Array: A Promising Composite Nanostructure for Visible Light Photocatalysis. Langmuir 2010, 26, 11615− 11620. (14) Li, Y.; Liu, Q. Y.; Shen, W. J. Morphology-Dependent Nanocatalysis Metal Particles. Dalton Trans. 2011, 40, 5811−5826. (15) Roberts, M. W. Heterogenous Catalysis Since Berezelius: Some Personal Reflections-Foreword. Catal. Lett. 2000, 67, 1−1. L


Langmuir

Article

(38) Yin, S. X.; Ma, X. Y.; Ellis, D. E. Initial stages of H2O Adsorption and Hydroxylation of Fe-Terminated alpha-Fe2O3 (0001) Surface. Surf. Sci. 2007, 601, 2426−2437. (39) Fox, M. A.; Dulay, M. T. Heterogeneous Photocatalysis. Chem. Rev. 1993, 93, 341−357. (40) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Environmental Applications of Semiconductor Photocatalysis. Chem. Rev. 1995, 95, 69−96. (41) Linsebigler, A. L.; Lu, G.; Yates, J. T. Photocatalusis on TiO2 Surfaces: Principles, Mechanisms, and Selected Results. Chem. Rev. 1995, 95, 735−758. (42) Shockley, W. On the Surface States Associated with a Periodic Potential. Phys. Rev. 1939, 56, 317−323. (43) Hadidi, K.; Hancke, R.; Norby, T.; Gunnces, A. E.; Lovvik, O. M. Atomic Study of LaNbO4; Surface Properties and Hydrogen Adsorption. Int. J. Hydrogen Energy 2012, 37, 6674−6685.

M


A Theoretical Study of Water Adsorption and Decomposition on Low

Recommend Documents