A DFT study of water adsorption on rutile TiO 2 (110) surface: the effects of surface steps

Transcription

1 A DFT study of water adsorption on rutile TiO (1) surface: the effects of surface steps Ting Zheng, Chunya Wu, Mingjun Chen*, Yu Zhang, Peter T. Cummings State Key Laboratory of Robotics and System, and School of Mechatronics Engineering, Harbin Institute of Technology, Harbin, 001, P. R. China Department of Chemical and Biomolecular Engineering, and Multiscale Modeling and Simulation Center, Vanderbilt University, Nashville, Tennessee -10, United States * Corresponding author: Mingjun Chen Telephone/Fax number: address: chenmj@hit.edu.cn 1

2 Abstract The associative and dissociative adsorption of water molecules at low-coverage situations on rutile TiO (1) surface with step defects was investigated by the density functional theory (DFT) calculations. Structural optimization of the hydroxylated/hydrated configurations at step edges along the < _ 1> crystal directions, and the dynamic process of water dissociation were discussed to get a better description of the water/tio interface. Our results indicate that steps on TiO (1) surface could be active site for water dissociation. The results of geometry optimization suggest that the stability of hydroxylated configuration is largely dependent on the locations of the H species and the recombination of water molecules from hydroxyls is observed in the fully hydroxylated condition. However, these hydroxyls can be stabilized by the associatively absorbed water nearby by forming competitive intermolecular hydrogen bonds. The dynamics of water dissociation and hydrogen diffusion were studied by the first principle molecular dynamics (FPMD) simulation and our results suggest that the hydrogen released by water dissociation, can be transferred among the adsorbates, such as the unsaturated oxygen atoms H O hydroxyl (TiO H O OH) complex at step edges, or gradually diffuses to the bulk water system in form of hydronium (H O + ) at higher water coverage. 1 1

3 Introduction The surface science of titania (TiO ) has been intensely studied due to its wide applications in fields such as photocatalyst, hydrophilic films and gas sensors 1-. A large numbers of experimental and theoretical studies have concentrated on describing the surface property of TiO, which strongly affects its interactions with other molecules such as short peptides, water, CO, ions, etc -. Meanwhile, water molecules are employed as a favorite probe to understand the metal oxide surface chemistry, since water molecules are always present even in UHV pressures. Consequently, the adsorption of water molecules on rutile TiO surface at low coverage, especially on the (1) surface, has aroused intense attention -. However, as a fundamental interest, water chemistry on TiO surface is not well understood and controversies still remain on whether water molecules on the defect-free (1) surface absorb associatively, or there is dissociation of absorbed water at the water/tio interface that produces hydroxyl sites 1,. The TiO (1) surface is more complex by exposing defects such as oxygen vacancies, steps 1, subsurface oxygen vacancies 1 and Ti interstitials 1,1. Specially, the surface defects and nanostructures have been proven to be active adsorption sites for peptides and polymer monomers 1-1. Regarding the interaction at the water/tio interface, many studies support the idea that water dissociation is a minor pathway on TiO (l) surface. However, structural defects, such as oxygen vacancies, steps and kinks, play a major role in dissociating water,-. It has been well studied that water molecules dissociated at surface oxygen vacancy sites by transferring one proton to the bridging oxygen atom, forming two hydroxyl groups for every vacancy site,. Wang also reported that the two next nearest neighbor bridging-oxygen vacancies (NNN-OVs)

4 are also active sites for water dissociation. In addition to the point defects, extended defects such as steps are also favored for water dissociation. As reported by Hong, the <1 1 _ 0> step, which exposed -fold coordinated Ti atoms along the step edges, is theoretically more active for water dissociation compared with the steps parallel to the <001> direction. However, not many STM experiments observed the steps along the < _ 0> direction on annealed rutile TiO (1) surface, except those under extreme treatment such as ion beam. Similarly to the vacancy defects on flat surface, the oxygen vacancies along the reconstructed < _ 1> steps (< _ 1> R ) are also active sites responsible for water dissociation with lower binding energy,0. The oxide-water interface could be very complicated and the knowledge of water adsorption at low coverage is a basic step to describe the complex oxide-water interface. The surface hydroxylation is such an important factor in describing the water chemistry on TiO surface that it should not be neglected,1. In this work, water adsorption on reduced rutile (1) surface with steps parallel to the <1 1 _ 1> direction, will be investigated using density functional theory calculations. The hydroxylated/hydrated structures and the dynamics of water dissociation at step edges with low water coverage are discussed in detail. Molecular and dissociative adsorption of H O at single unsaturated Ti sites at lower coverage (0. ML) is studied and compared using the geometry minimization, followed by the discussion of water adsorption at hybrid sites with higher coverage (1 ML). The first principles molecular dynamics calculations are also employed to explore adsorption possibilities at different water coverage.

5 1. Simulation Method.1 Model building The rutile TiO (1) surface is the most stable crystal face and its atomic structure has been studied by employing experimental techniques such as scanning tunneling microscope (STM) and atomic force microscopy (AFM), and theoretical methods such as density functional theory (DFT) and classical molecular dynamics simulation (CMD). For the rutile TiO surface, the point of zero charge (pzc) is about.-.. The computed pk a of the bridge hydroxyl (Ti OH + ) and the terminal H O, absorbed on the surface -fold coordinated Ti sites (TiOH ) on rutile (1) surface, are. and., respectively. Many STM studies have shown that steps are clearly extended along the well-defined <1 1 _ 1> and <001> crystallographic directions. The reconstructed <1 1 _ 1> R steps are believed to be theoretically more stable than the < _ 1> steps. Since the exposure of 1 water molecules at the < _ 1> R steps has been investigated,0, the hydroxylated and 1 hydrated conditions of the < _ 1> steps after water adsorption, which aroused little attention, will be studied in this work. The atomic structure of the < _ 1> steps on rutile TiO (1) surface is shown in Figure 1 a) and b). The step structures was created by removing the TiO unit along the < _ 1> direction, exposing the -fold coordinated Ti atoms (Ti, terminating the Ti rows of the upper terrace) and the -fold coordinated Ti atoms (Ti E, terminating the bridging O rows), separated by the - and -fold coordinated oxygen atoms (O f and O f ). Formation of the steps also created -fold coordinated half-

6 bridging oxygen atoms (abbreviated as O b-hf ) at the terminal of Ti rows at upper terrace. The exposure of the < _ 1> steps exhibits the (1) crystal face. The simulation box for DFT calculations (shown in Figure 1 c) has a dimension of. Å and. Å along the [ _ 1] and [ _ ] crystal directions, and is separated by a 0 Å thick vacuum gap in the z direction. The calculated lattice parameters of the bulk unit cell in this work are a=. Å and c=.0 Å, in good agreement with the experimentally measured values a exp =. Å and c exp =. Å, which were selected to build up the slabs. In this work, the numbers of slab layers were defined as the numbers of Ti layers in the simulation, which was Ti layer slabs in Figure 1c) Figure 1 Atomic structures of the steps parallel to < _ 1> directionon rutile (1) surface (a and b) and the periodic TiO (1) slabs with Ti layers c). Bright atoms in b) represented the upper terrace.. Simulation Details In this work, the DFT calculation was carried out using the generalized gradient 1 approximation (GGA) with the Perdew-Burke-Ernzerhof 0 (PBE) exchange and correlation functional, which has been proven to be generally reliable in describing both the rutile and water structures. All calculations were conducted via the SIESTA 1 package (Spanish Initiative for Electronic Simulations with Thousands of Atoms). The

7 double-ζ plus polarization (DZP) localized numerical atomic basis sets were used in all calculations with an energy cutoff of 0 Ry. The Brillouin zone was integrated using the Monkhorst Pack sets of 1 k-points mesh. Electron-core interactions were described by norm-conserving pseudo-potentials. The electrons in the p, s, d shells of Ti atom and the s, p shells of O oxygen atoms in rutile were treated as valence state and calculated explicitly in all calculation. The geometry optimization, also interpreted as static calculations (conjugate gradients, CG), and first principle molecular dynamics (FPMD) calculations were performed to study the hydroxylation/hydration of the step edges. The static calculations were used to determine adsorption energies so as to compare the different configurations with various hydroxylated/hydrated conditions. Geometry structures were optimized until all forces and total energy were converged to 0.0 ev/å and 1 - ev, respectively. All the geometry optimizations were calculated in the system with Ti layers. We also use the FPMD calculations to explore the adsorption possibilities and also to compute the dynamics of dissociative adsorption. At the beginning, water molecules were assumed to be absorbed at step edges with various water coverages. The simulation time of FPMD ranged from 0. ps to.0 ps, during which the hydrogen mass was set to amu to allow for a large time step (0. fs). The temperature was controlled around.1 K by a Nosé thermostat. A slab of Ti layers was chosen for the FPMD simulation. However, in order to extend the FPMD simulation time, an additional system with Ti layer slabs was also conducted.. Hydrated and hydroxylated step edges During the geometry optimization, various hydroxylated/hydrated configurations on Ti

8 and Ti E sites (shown in Figure ) were studied individually (indicated as single sites with lower water coverage 0. ML) or jointly (indicated as hybrid sites with higher water coverage1 ML). For single Ti sites, two water molecules are assumed to be attracted to meet the fully saturated condition (-coordinated) as mentioned in our previous work. Therefore, totally four configurations (named as Ti WW, Ti WH, Ti HW and Ti HH in Figure ) are considered, where the free H species generated after water dissociation are assumed to be attracted by the unsaturated oxygen atoms nearby. Water molecules may be absorbed on Ti sites molecularly (labeled by W, represented water molecule) or dissociatively (labeled by H, represented hydroxyl group). For single Ti E sites, in addition to the hydrated configuration (Ti W), extra three hydroxylated configurations were considered, named as Ti H hf, Ti H f and Ti H b respectively. The free H species are captured by the corresponding unsaturated oxygen atoms, such as O b-hf, O f and O b. For the hybrid configurations, both the unsaturated Ti atoms at step edges were hydrated or (and) hydroxylated by combining the single Ti /Ti E sites jointly. Therefore, the surface water coverage is 1 ML, and totally adsorption conditions (also shown in Figure ) are considered. The step edges varies from fully hydrated (Ti WW-Ti W) to fully hydroxylated (Ti HH-Ti H f ). The hydroxylated or hydrated configurations at both singe and hybrid models were examined by the energy minimization calculation and discussed.

9 Figure Hydrated/hydroxylated conformations on single and hybrid Ti /Ti E sites at step edges. O b-hf, O f and O b are short for the half-bridging oxygen, the -fold coordinated oxygen and the bridging oxygen at step edges. Only the unsaturated Ti/O atoms along the step edges are shown in various conformations.. Results and Discussion.1 Surface Energy Before the water dissociation, surface energies on rutile TiO (1) and (1) crystal face with different slab layers were calculated using Equation 1: E surf E n A slabs unit unit (1) surf E where A surf and E slabs represent the surface area and the total energy of the slabs respectively, and E unit and n unit stand for the energy of the basic TiO unit in bulk and the number of the basic TiO unit in slabs. The calculated surface energy of the (1) and (1) faces, E surf, is shown in Figure. The surface energy for a slab with a thickness of layers on the (1) surface converges to 0. J m -, which is comparable with the value of 0. and 0. J m -, obtained by Kiejna et al. and Lazzeri et al. ( layers), respectively. For the (1) face, the surface energy E surf is generally higher than the (1)

10 surface, expect for the case of layers. However, little work has been reported in literature for the (1) face, and no solid comparison can be made. As described in T. Hardcastle s work, the fluctuation of surface energies on the TiO (1), (1) and (1) surfaces with zig-zag evolutions became smaller as the slab miller index decreased. Therefore, the fluctuation of the surface energy on the TiO (1) surface can be inferred to be much smoother. Our results suggest the surface energy varies slightly from 0. J m - to 0. J m -, with exhibiting a zig-zag pattern. 1 Figure Surface energy on rutile (1) and (1) crystal face. Water dissociation at step edge During the geometry optimization, adsorption energy (E ads ) of water molecules on Ti /Ti E sites were calculated by equation and compared among different configurations. E E E Equation E hydrated dry isolated ads interface TiO HO where dry E TiO represents the total energy of the substrate without water molecules, is the energy in isolated water molecule system and isolated E H O hydrated E is the energy of the substrate interface with water molecules or hydroxyls absorbed on Ti or Ti E sites. Adsorption energies in different configurations at both singe and hybrid models, as shown in Figure, are all listed in Table 1. The adsorption energy is calculated based on the specific configuration with different hydroxylated/hydrated/bared condition on Ti and Ti sites. The stability of molecular and dissociative adsorption of water molecules at single Ti sites and hybrid

11 conditions are discussed below. Table 1 Adsorption energies (ev) of water at single and hybrid Ti sites on rutile TiO (1) surface with steps. Bared-Ti Ti WW Ti WH Ti HW Ti HH Bared-Ti Ti W Ti H hf Ti H f Ti H b indicates such model does not exist; indicates that this model is not necessary to be considered. The bared-ti indicates that no water adsorption on Ti sites. For example, the Ti WW-Ti W model has an adsorption energy of -. ev...1 Adsorption at single Ti sites For Ti sites, the adsorption energy in fully hydrated (Ti WW) and fully hydroxylated configurations (Ti HH) are -. ev and -. ev, respectively. However, after the geometry optimization, hydroxyls absorbed at the lower position of Ti sites (Figure ) in Ti HH, were observed to recombine to water molecules by capturing the H species absorbed at O b-hf sites (shown in Figure S1 of the supplementary material ). Therefore, the Ti HH model transformed to the Ti HW model eventually. The adsorption energy obtained in Ti HW model was -. ev, which is comparable with that in Ti HH. Similarly, there is no need to conduct the optimization in Ti WH configuration since it will finally transform to Ti WW model in the same way. The adsorption energy in stable Ti WW and Ti HW configurations are almost the same, indicating the molecular and dissociative adsorption of H O can both be observed on Ti sites. Diebold also found that both molecular and dissociative H O adsorption can be observed at steps on anatase (1) surface with very similar adsorption energies. For hydroxylation/hydration on Ti E sites, adsorption energies varies from -0. ev to

12 ev as shown in Table 1, depending primarily on the location of the H species. The hydrated Ti W configuration is generally more stable than the hydroxylated configurations, except the hydroxylated Ti H f model with an adsorption energy of -0. ev. As indicated by Dixon and his co-workers, molecular adsorption of H O on Ti was more favored than the dissociative one on the Ti sites with high saturation number in (TiO ) n clusters. We also found that the relocation of the H species, generated after water dissociation, contributes to the stability of the structural configurations. However, in the study of Hammer et al. 0, they found that relocation of the H species from step edges to a distant site had little influence on system energy. The different results here may result from the reconstructed step structures selected in their work. As indicated by Gong, the types of monoatomic steps on anatase (1) surface can influence water adsorption, where dissociative adsorption was more favored at the B(0) steps. In addition, we also performed structural optimizations of water adsorption at the surface -fold coordinated terminal Ti sites (Ti S ) on TiO (1) surface. The adsorption energy is -1. ev, which is much larger than the adsorption energy on Ti E sites, suggesting that the molecularly absorbed water are more favorable on Ti S sites... Adsorption at hybrid Ti sites As shown in Figure, totally hybrid hydroxylated/hydrated Ti and Ti E sites were considered, varying from fully hydrated (Ti WW-Ti W) to fully hydroxylated (Ti HH- Ti H f ). During discussing various hydroxylated/hydrated conditions, all configurations were classified into categories according to the hydroxylation/hydration at Ti sites as discussed in section..1. For the fully hydrated Ti sites (Ti WW, i.e., the hybrid models in rd column in Table 1

13 1), adsorption energies varies from -. ev to -. ev in different configurations, also depending on the location of the H species. The Ti WW-Ti H b configuration (shown in Figure a) is more stable, compared with the other two hydroxylated Ti sites such as Ti WW-Ti H bf and Ti WW-Ti H f. We can also infer that the fully hydrated site, i.e., TiWW-TiW, was stable after the geometry optimization Figure Optimized conformations of a) Ti WW-Ti H b, b) TiHH-Ti W c) Ti WH-Ti W and d) Ti HW-Ti H hf model. As mentioned in section..1, the Ti WH and Ti HH models transformed to Ti WW and Ti HW models eventually. The fully hydroxylated configuration, Ti HH-Ti H f, also seemed to be unstable with the low adsorption energy of -.0 ev and water combination from hydroxyl group at Ti sites was observed. As indicated by the green arrow in Figure S j) in the supplementary material, the fully hydroxylated model, i.e. Ti HH-Ti H f, transformed to the Ti HW-Ti H f configuration (-.0 ev) eventually. However, the Ti HH-Ti W model (-.0 ev) was stable after optimization without water recombination and the structural conformation is shown in Figure b). The hydroxyl group at Ti sites, highlighted by the green circle, forms H-bonds with both of the protonated O b-hf sites and the associative water molecule on Ti E sites (labeled by the blue arrow). The length of the H-bonds converged to 1. Å and 1. Å, respectively. Different from the fully hydroxylated configuration of Ti HH-Ti H f, it seems that the associatively absorbed

14 water molecules on Ti E sites in Ti HH-Ti W, may help to prevent the hydroxyls from recombining to water molecules. Similarly, the structure in Ti WH-Ti W configuration (-. ev, as shown in Figure c), was stable after optimization. Therefore, we may infer that the participation of the extra H-bonds could help to stabilize the interaction by restricting water recombination. However, in Ti HH-Ti H f model, lacking of this stabilization from extra H-bonds may result in the recombination of water molecules. Therefore, it is unnecessary to conduct structural optimizations in Ti WH-Ti H f and Ti WH-Ti H b, since they may probably transform to Ti WW-Ti H f and Ti WW-Ti H b, respectively. For another hydroxylated Ti (Ti HW) sites, adsorption energies also varies in different configurations. The Ti HW-Ti W model is the most stable configuration compared with other models. The Ti HW-Ti H hf model is less stable and the structural conformation is shown in Figure d). The H species at the O b-hf site is still attracted by the water molecule absorbed on Ti site via H-Bonds. The adsorption energy in Ti HW- Ti H f is largely decreased by 0. ev compared with the fully hydrated configuration (Ti WW-Ti W), thus interprets as an unstable configuration. A summary of the averaged bond length of the hydroxylated groups at step edges after static optimization is listed in Table S1 of the supplementary material. However, the Ti HW-Ti H f and the fully hydroxylated models (Ti HH-Ti H f ), which were assumed to be less stable, were not included. Generally, the length of Ti-O bond extends after the Ti atom sites are hydroxylated.. First Principle MD of dissociation procedure In this section, the dynamic process of water dissociation was simulated using the first 1

15 principle molecular dynamics simulation (FPMD) method. Here we studied the adsorption process under different water coverage, Θ=%, 1% and 00%. The water coverage here is defined as the ratio of water numbers over the total instauration degree of surface Ti atoms. In this work, three unsaturated Ti atoms (Ti, Ti E and Ti S ) are exposed on the (1) surface in the simulation box and the total unsaturation degree is. At Θ=%, initially water molecules were associatively absorbed along step edges and the FPMD simulation lasted for 0. ps. Water dissociation was not observed during the entire FPMD process, which may be due to the short simulation time and the different initial conformation. After increasing the coverage to 1%, water dissociation occurred spontaneously on Ti site at t=0. ps and the dissociation procedure is shown in Figure. Initially the associatively absorbed H O (labeled as W ) formed H-bond with the nearby O b-hf atom, as highlighted by the blue arrow in the 1 st detailed figure. Gradually, water W became dissociated and the free H species was captured by O b-hf atoms, resulting in two hydroxyls formed at step edges as indicated in the rd detailed figure. Meanwhile, the hydroxyl from W was stabilized by the H-bond from another water molecule (W S ) absorbed at surface Ti S site, which was highlighted by the green arrow. As the dissociative adsorption of H O on Ti sites is observed in a very short time, we may infer that the -fold coordinated Ti atoms could show high reactivity for water dissociation compared with Ti E sites. The result here suggested that dissociative adsorption of H O on Ti sites with high unsaturation is much more favored as also indicated by Dixon and his co-workers. In this part, since the dissociation was observed very soon during the FPMD process, the energy barrier during water dissociation should be relatively small. 1

16 Figure Procedures of water dissociation on Ti sites at Θ=1%. Detailed structural conformations (1, and ) at t=0.0 ps, 0. ps and 0. ps were also exhibited. The blue curves represented the distance between the proton in water W (H W ) and the half-bridging oxygen (O b-hf ) nearby, as labeled by the blue arrow. At Θ=1%, water dissociation on rutile TiO (1) slabs with layers was also studied. The FPMD simulations lasted for about.0 ps. A very similar process of water dissociation (W ) on Ti sites, as shown in Figure S (1 st detailed figure), was observed as in the case with -layer slabs. The hydroxyl (labeled as OH ) was stabilized by the hydrogen bond interaction with the nearby water molecule (W ) adsorbed on Ti E sites (shown in the nd detailed figure). However, this interaction seemed unstable and the water W began to share its proton (labeled as H in the rd figure) with the OH. The proton H was observed to be transferred among the H O OH complex in the following 0. ps, and this interaction may be interrupted by extra water molecules above the 1 st water layer. The evidence of proton transfer among the adsorbates was also observed by Lindan 0 and co-workers, where protons were transferred between the terminal hydroxyl group (TOH) and the pre-adsorbed H O on perfect rutile TiO (1) surface. However, compared with the dissociation occurred on slabs with layers, the hydroxyl OH in 1

17 figure was stabilized by the associative water on the surface Ti S sites. As discussed in section..1, molecular adsorption of water on Ti S sites is much more stable than the adsorption on Ti E sites. Therefore, water dissociation on Ti S sites may not be favored in Figure, and the different results in these two situations may cause by different positions of water molecules at the initial stage. As the water coverage increased to 00%, we performed a FPMD run lasted for about 0. ps and spontaneous water dissociation was observed during the simulations. As indicated by Aschauer 1, a higher surface concentration of water thus facilitates dissociation on the reduced anatase (1) surface. Here, water dissociation was occurred by releasing their protons into the water system and the dynamic procedure is shown in Figure. Initially, free water molecules in bulk (labeled as and ) formed tight hydrogen bonds with the associative water molecules on Ti E sites (labeled as 1). Gradually, water 1 began to share its proton (labeled as H 1 ) with water in nd water layer at t=0.1 ps. The evolution of distances between the proton H 1 and water oxygens in water 1 and is shown in Figure d). Under the strong attraction from water, the proton H 1 began to shake off the restriction from the water 1 at t=0.1 ps (conformations shown in Figure b). The dissociation of water 1 eventually resulted in the formation of the hydroxyl-hydronium (Ti-OH H O + ) complex, at step edges. However, hydrogen transfer among water molecules seems to be continuous based on the Grotthuss mechanism. At t=0. ps, the extra proton (labeled as H in Figure c) in the newly formed hydronium ion was attracted by water. The distance between the proton H and water oxygens in and was illustrated in Figure e). Unlike the pathway here, 1

18 the diffusion of protons at lower water coverage was observed to be transferred among the unsaturated oxygen atoms near step edges. Therefore, at higher water coverage, hydrogen diffusion may be facilitated by water molecules above the 1 st water layer and may transfer to the bulk system eventually. As reported by Lindan and his co-workers 0, the third water molecule in the second water layer on perfect rutile TiO (1) surface facilitated proton transfer among the terminal hydroxyl (TOH) rd H O pre-adsorbed H O complex. The DFT results of Kowalski and co-workers also indicated that the hydrogen prefers diffusing into the bulk of rutile over desorbing from the surface into the gas phase. Moreover, the diffusion of hydrogen can be quite different on reduced rutile (1) surface. As reported by Matthiesen et al, protons may be transported by the water dimers on rutile (1) surface. Li and his co-workers found that the hydrogen in the geminate hydroxyl (OH-OH) pairs formed at the vacancy site could hop directly along the bridging oxygen row. Wendt et al., meanwhile, reported a different diffusion pathway assisted by the nearby water molecules the geminate hydroxyl (OH-OH) pairs could split into single hydroxyl group by transferring proton to adjacent oxygen rows in the [ _ 0] direction when interacting with absorbed water molecules. However, the dissociation of water molecules at hydroxylated sites was not taken in to consideration in this work. As described in Walle s work,, the vacancies at step edges are saturated by hydroxyls and were not likely to be active toward additional dissociation. They also found that the capping H atoms in the hydroxyls at step sites, compared with the hydroxyl pairs formed at terrace vacancies, was more difficult to remove by O. Therefore, we may infer that the hydroxyls formed at the step edges could 1

19 be stable at the TiO /water interface Figure Procedure of water dissociation on Ti E sites at Θ=00%. Adsorption conformations at t=0 ps, 0.1 ps and 0. ps were exhibited in detailed figures 1 to, respectively. The curves in d) represented the distance between the shared proton (H 1 ) in water 1 and the water oxygen atoms (O 1 and O ) in water 1 and, while curves in e) represented the distance of the shared proton (H in figure c) with respect to the two oxygen atoms (O and O ) in water and.. Conclusions In this work, we presented a DFT study of water adsorption at step edges on rutile (1) surface. The step edge was designed to parallel to the < _ 1> direction. The structures of the hydrated/hydroxylated step edges and the dynamics of water dissociation at step edges were studied. The results of the geometry optimization suggested that stable hydroxyls at step edges were highly dependent on the location of the H species generated after water dissociation. Hydroxyls at step edges, especially absorbed at the lower part of Ti sites, may sometimes be unstable and recombine to water molecules. For instance, the fully hydroxylated model was regarded as unstable and water recombination was observed at Ti sites. However, the hydroxyls can be stabilized by extra hydrogen bonds 1

20 from the interaction with the nearby associatively absorbed water molecules. The first principle molecular dynamics method was employed to investigate water dissociation at step edges and the diffusion of protons. The FPMD results indicate that the free H species can be captured by the unsaturated oxygen atoms nearby or transferred to the bulk system in the form of hydronium at higher water coverage. These results elucidated the role of step edges on water adsorption and dissociation on TiO (1) surface, and the importance of surface steps on the surface chemistry of metal oxides should be stressed. Notes The authors declare no competing financial interest. Acknowledgement This work was supported by the State Key Laboratory of Robotics and System, Harbin Institute of Technology (grant number SKLRS-0-ZD-0) and the Fundamental Research Funds for the Central Universities (No. AUGA01). This research used resources of the National Energy Research Scientific Computing Center, which is supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC0-0CH1. Peter T. Cummings was supported by the U.S. Department of Energy (DOE), Office of Basic Energy Sciences, Geoscience Research Program, through Grant ERKCC to Oak Ridge National Laboratory, which is managed for DOE by UT Battelle, LLC under Contract No. DE-AC0-00OR. Reference 1. U. Diebold, Surf. Sci. Rep., (00).. M. Grätzel, Nature 1, (001).. R. Wang, K. Hashimoto, A. Fujishima, M. Chikuni, E. Kojima, A. Kitamura, M. Shimohigoshi and T. Watanabe, Nature, 1 (1). 0

21 . G. Eranna, B. C. Joshi, D. P. Runthala and R. P. Gupta, Crit. Rev. Solid State Mater. Sci., 1 1 (00).. C. Wu, M. Chen, A. A. Skelton, P. T. Cummings and T. Zheng, ACS Appl. Mater. Interfaces, (0).. R. Schaub, P. Thostrup, N. Lopez, E. Laegsgaard, I. Stensgaard, J. K. Nørskov and F. Besenbacher, Phys. Rev. Lett., (001).. D. C. Sorescu, W. A. Al-Saidi and K. D. Jordan, J. Chem. Phys.,, 1 (0).. M. Předota, M. L. Machesky, D. J. Wesolowski and P. T. Cummings, J. Phys. Chem. C, (0).. Z. Zhang, O. Bondarchuk, B. D. Kay, J. M. White and Z. Dohnálek, J. Phys. Chem. B 1, 1 (00).. P. J. D. Lindan, N. M. Harrison and M. J. Gillan, Phys. Rev. Lett. 0, (1).. S. Wendt, R. Schaub, J. Matthiesen, E. K. Vestergaard, E. Wahlström, M. D. Rasmussen, P. Thostrup, L. M. Molina, E. Lægsgaard, I. Stensgaard, B. Hammer and F. Besenbacher, Surf. Sci., (00). 1. L. M. Liu, C. Zhang, G. Thornton and A. Michaelides, Phys. Rev. B, 1 (0).. D. Wesolowski, J. Sofo, A. Bandura, Z. Zhang, E. Mamontov, M. Předota, N. Kumar, J. Kubicki, P. R. Kent, L. Vlcek, M. Machesky, P. Fenter, P. Cummings, L. Anovitz, A. Skelton and J. Rosenqvist, Phys. Rev. B, 101 (01). 1. H. Cheng and A. Selloni, Phys. Rev. B, 01 (00). 1. S. Wendt, P. T. Sprunger, E. Lira, G. K. H. Madsen, Z. Li, J. Ø. Hansen, J. Matthiesen, A. Blekinge-Rasmussen, E. Laegsgaard, B. Hammer and F. Besenbacher, Science 0, 1 1 (00). 1. E. Lira, S. Wendt, P. Huo, J. Hansen, R. Streber, S. Porsgaard, Y. Wei, R. Bechstein, E. Lægsgaard and F. Besenbacher, J. Am. Chem. Soc., (0). 1. C. Wu, M. Chen and C. Xing, Langmuir, 1 (0). 1. D. P. Song, M. J. Chen, Y. C. Liang, Q. S. Bai, J. X. Chen and X. F. Zheng, Acta Biomater., (0). 1. H. P. Zhang, X. Lu, Y. Leng, F. Watari, J. Weng, B. Feng and S. Qu, J. Biomed. Mater. Res. - Part A A, (0). 0. M. Chen, T. Zheng, C. Wu and C. Xing, Colloids Surfaces B Biointerfaces, 1 (01). 1. C. Melis, A. Mattoni and L. Colombo, J. Phys. Chem. C,, 01 0 (0).. M. A. Henderson, Surf. Sci., 1 (1). 1

22 . M. A. Henderson, Langmuir 1, 0 0 (1).. O. Bikondoa, C. L. Pang, R. Ithnin, C. A. Muryn, H. Onishi and G. Thornton, Nat Mater, 1 1 (00).. N. G. Petrik and G. A. Kimmel, J. Phys. Chem. C, 0 0 (01).. H. Shi, Y. C. Liu, Z. J. Zhao, M. Miao, T. Wu and Q. Wang, J. Phys. Chem. C, 0 0 (01).. F. Hong, Y. H. Ni, W. J. Xu and Y. Yan, J. Chem. Phys., 1 (01).. T. Luttrell, W. K. Li, X. Q. Gong and M. Batzill, Phys. Rev. Lett., (00).. U. Martinez, L. B. Vilhelmsen, H. H. Kristoffersen, J. Stausholm-møller and B. Hammer, Physics Rev. B, 0 (0). 0. H. H. Kristoffersen, J. Hansen, U. Martinez, Y. Y. Wei, J. Matthiesen, R. Streber, R. Bechstein, E. Lægsgaard, F. Besenbacher, B. Hammer and S. Wendt, Phys. Rev. Lett. 1, (0). 1. J. G. Chang, H. T. Chen, S. P. Ju, C. S. Chang and M. H. Weng, J. Comput. Chem., 1 (0).. H. Onishi, K. Fukui and Y. Iwasawa, Bull. Chem. Soc. Jpn., (1).. R. B. and C. G. and J. S. and P. J. and R. P. and A. Kühnle, Nanotechnology 0, 00 (00).. S. P. Bates, G. Kresse and M. J. Gillan, Surf. Sci., (1).. M. Předota, A. V. Bandura, P. T. Cummings, J. D. Kubicki, D. J. Wesolowski, A. A. Chialvo and M. L. Machesky, J. Phys. Chem. B 0 (00).. J. W. Bullard and M. J. Cima, Langmuir, 1 (00).. J. Cheng, X. Liu, J. VandeVondele and M. Sprik, Electrochim. Acta 1, (01).. T. P. Hardcastle, C. R. Seabourne, R. M. D. Brydson, K. J. T. Livi and A. J. Scott, J. Phys. Chem. C, 0 (0).. C. J.Howard, T. M. Sabine, and F. Dickson, Acta Cryst. B, - (). 0. J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett., (1). 1. J. M. Soler, E. Artacho, J. D. Gale, A. Garcia, J. Junquera, P. Ordejon and D. Sanchez-Portal, J. Phys: Condens. Matter. 1, - (001).. J. D. Pack and H. J. Monkhorst, Phys. Rev. B 1, 1 1 (1).. T. Zheng, C. Wu and M. Chen, Surf. Sci. 1, 1 (0).. A. Kiejna, T. Pabisiak and S. W. Gao, J. Phys. Condens. Matter 1, 0 1 (00).. M. Lazzeri, A. Vittadini and A. Selloni, Phys. Rev. B, (00).

23 . See supplementary material at [URL] for the atomic conformations after energy minimization Figure S1 and S. It also contains the averaged bond length for hydroxyls Table S1 and the dynamics of water adsorption on substrates with layer slabs at Θ=1% Figure S.. X. Q. Gong, A. Selloni, M. Batzill, and U. Diebold, Nat. Mater., 0 (00).. M. Chen, T.P. Straatsma, and D. a. Dixon, J. Phys. Chem. A, 1 1 (01).. X. Gong and A. Selloni, J. Catal., (00). 0. C. Zhang and P. J. D. Lindan, J. Chem. Phys., 0 0 (00). 1. U. Aschauer, Y. He, H. Cheng, S.C. Li, U. Diebold, and A. Selloni, J. Phys. Chem. C, 1 1 (0).. P. M. Kowalski, B. Meyer and D. Marx, Phys. Rev. B, (00).. J. Matthiesen, J. Hansen, S. Wendt, E. Lira, R. Schaub, E. Lægsgaard, F. Besenbacher and B. Hammer, Phys. Rev. Lett., 1 (00).. S. C. Li, Z. Zhang, D. Sheppard, B. D. Kay, J. M. White, Y. Du, I. Lyubinetsky, G. Henkelman and Z. Dohnálek, J. Am. Chem. Soc. 0, 00 0 (00).. S. Wendt, J. Matthiesen, R. Schaub, E. K. Vestergaard, E. Lægsgaard, F. Besenbacher and B. Hammer, Phys. Rev. Lett., 0 (00).. L. E. Walle, A. Borg, P. Uvdal and A. Sandell, Phys. Rev. B 0, (00).

24

25

26

27

28

29