This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes. Article Cite This: ACS Omega 2019, 4, 2989−2999 http://pubs.acs.org/journal/acsodf Water Adsorption on MO2 (M = Ti, Ru, and Ir) Surfaces. Importance of Octahedral Distortion and Cooperative Effects Danilo Gonzaĺ ez,† Javier Heras-Domingo,† Stefano Pantaleone,† Albert Rimola, Luis Rodríguez-Santiago, Xavier Solans-Monfort,* and Mariona Sodupe* Departament de Química, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain * S Supporting Information ABSTRACT: Understanding metal oxide MO2 (M = Ti, Ru, and Ir)−water interfaces is essential to assess the catalytic behavior of these materials. The present study analyzes the H2O−MO2 interactions at the most abundant (110) and (011) surfaces, at two different water coverages: isolated water molecules and full monolayer, by means of Perdew−Burke− Ernzerhof-D2 static calculations and ab initio molecular dynamics (AIMD) simulations. Results indicate that adsorp- tion preferably occurs in its molecular form on (110)-TiO2 and in its dissociative form on (110)-RuO2 and (110)-IrO2. The opposite trend is observed at the (011) facet. This different behavior is related to the kind of octahedral distortion observed in the bulk of these materials (tetragonal elongation for TiO2 and tetragonal compression for RuO2 and IrO2) and to the different nature of the vacant sites created, axial on (110) and equatorial on (011). For the monolayer, additional effects such as cooperative H-bond interactions and cooperative adsorption come into play in determining the degree of deprotonation. For TiO2, AIMD indicates that the water monolayer is fully undissociated at both (110) and (011) surfaces, whereas for RuO2, water monolayer exhibits a 50% dissociation, the formation of H3O2− motifs being essential. Finally, on (110)-IrO2, the main monolayer configuration is the fully dissociated one, whereas on (011)-IrO2, it exhibits a degree of dissociation that ranges between 50 and 75%. Overall, the present study shows that the degree of water dissociation results from a delicate balance between the H2O−MO2 intrinsic interaction and cooperative hydrogen bonding and adsorption effects. ■ INTRODUCTION Rutile-like TiO2, RuO2, and IrO2 are isostructural metal oxides dissociation, and thus, surface preparation appears to be extremely delicate to determine the intrinsic water−surface with a large number of applications. As an example, they have interface.14−16,26 Moreover, results are also sensitive to the all been studied for energy conversion and storage from water experimental conditions and techniques used for the analysis.9 oxidation either photo- or electrocatalytically.1−6 Many of In this context, several contributions based on high-resolution these applications imply the existence of a material−water electron energy loss spectroscopy, temperature-programmed interface, whose understanding is crucial for the rationalization desorption, infrared reflection absorption spectroscopy, and of the global catalytic process. Therefore, the knowledge of the scanning tunneling microscopy (STM) data are in agreement intrinsic water−MO2 interaction as well as how this interaction with water adsorbed molecularly on (110),27−31 particularly at varies when increasing the water coverage is of high low water coverages. Other experiments performed at higher importance. In this context, several works dealing with coverages with X-ray photoelectron spectroscopy (XPS) and H2O−TiO2,7−17 H2O−RuO2,12,13,18−23 and H2O−IrO213,24,25 photoelectron diffraction techniques suggest that water interfaces have been reported in the literature. The obtained partially dissociates.32,33 The reader is referred to the excellent results allowed determining the different behaviors as a review of U. Diebold for a detailed survey of water adsorption function of the material and the exposed surface. However, on TiO2 surfaces, the performance of different techniques, and some controversy still exists on the amount of water that surface preparation.34 On the other side, density functional dissociates after adsorption on the different surfaces and theory (DFT) calculations show that the energy difference materials.7 between the molecular and dissociative adsorption forms is Among the three materials, titanium oxide is the most very small and sensitive to the surface model and level of intensively explored one. The most stable (110) surface has theory used, particularly, the thickness of the slab model, which centered most of the available investigations, which mainly focused on understanding whether water adsorption on this Received: November 30, 2018 surface dissociates generating H+/OH− species.7,9 The O- Accepted: January 25, 2019 defective surfaces have been observed to favor water Published: February 11, 2019 © 2019 American Chemical Society 2989 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article exhibits an even−odd behavior, the DFT functional, or the (i) the H2O−MO2 intrinsic interaction, (ii) the different inclusion of the U Hubbard correction.9,10,13,28,35−40 This turns octahedral distortion of TiO2 with respect to RuO2 and IrO2, several contributions in favor of one or the other situation. The and (iii) the presence of hydrogen bonding and adsorption influence of water coverage on the degree of water dissociation cooperative effects. on the (110)-TiO2 surface has been less addressed and most of the calculations including a water monolayer or more suggest that water adsorbs mainly in its molecular form.41,42 ■ RESULTS AND DISCUSSION As mentioned above, the main goal of this paper is to compare Although the interaction between water and the (011) the properties of MO2 (M = Ti, Ru, and Ir) upon interacting surface of titania has been much less studied, water adsorption with water. For that, we first addressed the structural and at this surface is also relevant because of its large contribution electronic properties of the bulk and main crystallographic to the Wulff construction. STM, X-ray diffraction measure- surfaces. Second, we considered the adsorption of one single ments, and DFT calculations have shown that the (011) water molecule at each of the two selected surfaces, with the surface suffers an important reconstruction when prepared in ultimate goal of comparing the intrinsic water interaction ultrahigh vacuum conditions.11,43−45 This reconstruction is, among the three metal oxides. Third, we studied the water however, reversed when exposing the distorted (011) surface monolayer adsorption and evaluated the degree of deprotona- to liquid water and the original facet is recovered.11,45 From a tion in each case. For that, we carried out, in addition to static computational point of view, calculations suggest that water calculations, AIMD simulations to address the influence of adsorption at low coverages occurs through the dissociative thermal effects. mode, although increasing the water coverage to the full MO2 Bulk and Surfaces. All three considered metal oxides monolayer seems to decrease the preference for the MO2 (M = Ti, Ru, and Ir) crystallize in a rutile structure, dissociative form.11,45−47 tetragonal with space group P42/mnm (see Figure 1a).49 Ruthenium oxide has also been largely studied and again most studies focus on the (110) surface.12,13,18−23,48 For this (110) surface, it is accepted that at very low coverages, single water adsorbs on unsaturated Ru centers, establishing an equilibrium between molecular and dissociated water mole- cules.18−20 Moreover, STM and DFT calculations also indicate that increasing the water coverage allows the formation of water dimers that are adsorbed in contiguous unsaturated Ru centers.18,19,22 These dimers enclose one molecular and one dissociative water, leading to the formation of H3O2− motifs or hydrogen-bonded (H3O2−)n chains at higher water coverages. The other (011), (100), and (001) surfaces have only been studied in detail very recently.21,22 XPS, in situ surface diffraction, and DFT calculations suggest a mixed molecular/ dissociative arrangement at high coverages, the ratio between Figure 1. (A) Bulk structure and (B) main surfaces considered in the the two forms varying depending on the surface. The two most present work. stable surfaces [(110) and (011)] are more prone to dissociate the adsorbed water, whereas the (001) and (100) surfaces Titanium dioxide exhibits two other (thermodynamically mostly present the molecular form. In our previous paper,22 metastable) crystalline phases in nature: anatase (tetragonal, observed trends were rationalized by the combination of three I41/amd) and brookite (rhombohedral, Pbca). However, for factors: (i) the intrinsic acid−base properties of each surface; comparison, the present work will only consider the water (ii) the presence of strong cooperative effects; and (iii) an adsorption on the rutile polymorph of TiO2. In the bulk increase of the surface oxygen bridge (Obr) basicity by the structure, metal cations, M4+, show a distorted octahedral adsorption of water. coordination and O2− atoms display a trigonal planar Finally, H 2O−IrO 2 interaction has been much less environment. Main distances, cell parameters, and computed studied.13,24,25 To our knowledge, only the adsorption of net charges of the metal and oxygen atoms for the three metal water on the most stable (110) facet has been addressed by oxides are given in Table 1. means of DFT calculations. Results suggest that the interaction Concerning the bulk, it can be observed that both M−O energy between H2O and IrO2 is significantly stronger than in distances and cell parameters are in very good agreement with the other two materials and that the adsorbed molecules tend the experimentally determined values,49 deviations being less to dissociate. than 1.5%. The M−O distances range between 1.95 and 2.01 With the aim of analyzing how the nature of the metal oxide Å, and the largest ones correspond to IrO2, as expected. As influences the water adsorption and, in particular, the degree of found experimentally, Ti4+ exhibits a distorted octahedral dissociation, we studied the rutile-like H2O−MO2 interaction environment with four shorter Ti−O distances in the (M = Ti, Ru, or Ir) at the (110) and (011) surfaces, the ones equatorial plane and two longer Ti−O axial distances that contribute the most in the Wulff construction of the three (tetragonal elongation), whereas the opposite situation with materials, considering two different water coverages: isolated four longer equatorial M−O distances and two shorter axial water molecules and full monolayer. Moreover, thermal effects ones (tetragonal compression) is observed for RuO2 and IrO2. and proton mobility on the surface are analyzed by performing These differences may be related to the electronic config- ab initio molecular dynamics (AIMD) of the full monolayer uration of the metal in each metal oxide. That is, the electronic coverage. Results show that there are three key factors for configuration of Ti is 4s23d2, that of Ru is 5s24d6, and that of Ir determining the degree of dissociative water over the surface: 6s15d8. Although the metal ion can be formally considered as 2990 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article Table 1. MO2 (M = Ti, Ru, and Ir) Bulk Computed Distances (in Å), Cell Parameters, and Net Atomic Charges (a.u)a d(M−O)eq d(M−O)ax d(M−M) a=b c q(M) q(O) TiO2 1.945 1.990 2.977, 3.579 4.603 2.977 2.240 −1.118 1.945 1.986 4.594 2.959 RuO2 1.997 1.945 3.149, 3.542 4.487 3.149 1.600 −0.750 1.984 1.942 4.492 3.107 IrO2 2.013 1.980 3.193, 3.584 4.497 3.193 1.659 −0.829 1.998 1.958 4.505 3.159 a Experimental values49 in italics. Table 2. Surface Energies (γ in eV Å−2), % of the Surface’s Contribution to the Wulff Shape, Main Surface Distances (in Å), and Net Atomic Charges (in a.u) (h k l) γ % Wulff d(M−Obr) d(M5/4c−O)eq d(M5/4c−O)ax d(M−M) q(M5/4c) q(Obr) TiO2 (1 1 0) 0.042 70.4 1.833 1.965 1.792 2.977 2.16 −0.99 (0 1 1) 0.074 29.3 1.810, 1.888 1.810, 2.164 1.886, 2.036 3.293 2.19 −1.04 (1 0 0) 0.060 0.0 1.854 1.854, 2.073 1.833 2.977 2.17 −1.02 (0 0 1) 0.085 0.3 1.845, 1.901 1.901 1.787 4.603 2.13 −1.08 RuO2 (1 1 0) 0.098 40.0 1.918 1.982 1.881 3.131 +1.6 −0.75 (0 1 1) 0.105 51.8 1.873 1.873, 2.086 1.955 3.670 +1.64 −0.78 (1 0 0) 0.119 8.2 1.925 1.925, 2.028 1.900 3.130 +1.65 −0.75 (0 0 1) 0.134 0.0 1.842, 1.932 1.918 1.842 4.480 +1.54 −0.81 IrO2 (1 1 0) 0.150 40.3 1.982 2.012 1.943 3.187 1.47 −0.70 (0 1 1) 0.158 53.1 1.897, 1.926 2.000, 2.077 2.000 3.708 1.54 −0.74 (1 0 0) 0.185 6.6 1.960 1.960, 2.034 1.985 3.187 1.50 −0.71 (0 0 1) 0.206 0.0 1.905, 1.992 1.955 1.905 4.541 1.45 −0.77 M4+ and thus there are no d electrons in Ti4+, Ru4+ has four d Computed values for the surface energies are similar to those electrons, and Ir4+ has five d electrons, there is a certain previously reported in the literature and follow the same covalent character, and in an octahedral ligand field, this can trend.53 That is, the smallest surface energy corresponds to the lead to different geometrical distortions for early and late (110) facet, whereas the largest one corresponds to the (001) transition metals. Furthermore, for both RuO2 and IrO2, spin- one with M4c unsaturated metal centers. The remaining two polarized Perdew−Burke−Ernzerhof (PBE)-D calculations surfaces show intermediate values, and their relative order indicate that magnetization of each metal ion is equal to depends on the material. For TiO2, (100) is more stable than zero, as found previously at this level of theory.50−52 Finally, the (011) one, whereas for RuO2 and IrO2, the relative stability is reversed. Despite that, the contribution of the (100) surface and as expected, computed charges indicate that TiO2 is to the TiO2 Wulff shape is zero, whereas (011) accounts for significantly more ionic than RuO2 and IrO2. 29.3% because of symmetry equivalences. Overall, it can be Slab models for different crystallographic orientations observed in Figure 2 and Table 2 that the (110) and (011) [(110), (011), (100), and (001)] were built cutting out the slab from the optimized bulk structure. Surface energies, main M−O distances of the outermost layer, and net atomic charges are given in Table 2. Values corresponding to the internal layers are very similar to the bulk values and thus have not been included and will not be discussed further. In all cases, bulk cutting to generate slab models leads to two-coordinated bridging oxygen Obr at the surface, as well as five-coordinated M5c sites for (110), (011), and (100) and four-coordinated M4c for (001). The vacant site of M5c in the (110) and (100) surfaces is axial, whereas the vacant site of Figure 2. Wulff constructed equilibrium shapes for rutile-like MO2 M5c in the (011) surface is equatorial. For the M4c sites in the (M = Ti, Ru, and Ir). (001) facet, both vacant sites are equatorial. Surface M−O distances corresponding to unsaturated M sites vary depending facets contribute by more than an 80% of the total surface, and on the coordination environment, the shortest values being thus, these two surfaces are the ones considered to analyze the found for the axial M4c−O bonds of the (001) surface. M−M water adsorption. distances largely vary depending on the crystal facet and range Adsorption of Isolated Water Molecules. As mentioned between 3 and 4.5 Å. For all materials, the shorter values are above, the aim of the present work is to get insights into the found in the (110) and (100) surfaces, whereas the larger ones different behavior of MO2 (M = Ti, Ru, and Ir) upon are found for (001). interacting with water. For that, we first studied the adsorption 2991 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article of a single water molecule onto the two surfaces that interacting with the undercoordinated M5c sites of the surface contribute the most to the Wulff shape (Figure 2): the exhibits a larger increase when it is adsorbed on IrO2 than on (110) and (011) surfaces (Figure 1B). This corresponds to a RuO2 or on TiO2. This is in agreement with the computed water coverage of 1/4 for (110) and of 1/8 for (011). The charge of the water molecule in H2O−MO2 (M = Ti, Ru, and preferred adsorption configuration at the (110) and (011) Ir), which increases from M = Ti to Ir. Thus, despite the higher surfaces of MO2 is that in which the water molecule binds basicity of Obr in TiO2, as indicated by the net atomic charge of through its O atom with the undercoordinated M5c sites.10 Obr (Table 3) and the density of states (DOS), which shows This interaction increases the water acidity, leading to the that p bands of Obr in TiO2 lie at higher energy (see Figure 4), formation of a hydrogen bond between one H atom of the the increase of water acidity upon adsorption is not sufficient water molecule and the nearest undercoordinated Obr. This to favor the dissociative form. interaction can also lead to a dissociative OH−/H+ adsorbed form, with water deprotonated and the Obr protonated. Thus, both adsorbed forms, molecular (mol) and dissociative (diss), have been considered in each of the two surfaces (see Figure 3). Adsorption energies, relative stabilities between the two forms, and main structural parameters are given in Table 3. Figure 3. Molecular and dissociative adsorbed forms in the (110)- Figure 4. Projected DOS (pDOS) associated with the Obr 2p bands MO2 and (011)-MO2 surfaces. and metal d bands. Regarding the (110) surface, both the molecular and The computed preference of a single water to be molecularly dissociative minima were localized for TiO2 and RuO2, adsorbed form on (110)-TiO2 is in agreement with the recent whereas only the adsorbed dissociative form was located in carefully conducted molecular beam/STM experiments by the case of IrO2 in agreement with previous calculations.9,13,18 Wang et al.,9 which determine that molecular adsorption is Indeed, all attempts to optimize the molecular adsorbed preferred by 0.035 eV. It should be noted that whether or not minimum collapsed to the dissociative adsorbed species. water adsorbs in a molecular or a dissociative form on a defect- Furthermore, results show that while the molecular form is free TiO2 (110) surface has been a subject of intense debate.7 more stable for TiO2, the dissociative form is more stable for From a computational point of view, different works agree with RuO2 and the only minimum for IrO2. This is related to the the fact that the energy difference between mol and diss is small strength of the M−H2O interaction, which follows the trend: and sensitive to the computational approximation.9,10,36,37 H2O−TiO 2 < H2 O−RuO2 < H2O−IrO2; that is, the A significantly different behavior is observed for the (011) adsorption energy increases (in absolute value) from TiO2 to surface. At this surface, the diss form is the most stable IrO2. As a consequence, the acidity of the water molecule upon adsorbed species on TiO2, whereas the mol form is the Table 3. Adsorption Energies (ΔEads in kJ mol−1), Relative Stabilities between the Molecular and Dissociated Forms (ΔErel in kJ mol−1), Main Distances (in Å), and Net Atomic Charges ΔEads ΔErel Ow−Hw Obr−Hw M−Ow q(Hw) q(Ow) q(Obr) q(M) (110) TiO2 mol −86.9 0.0 1.001 1.794 2.251 0.643 −1.22 −1.06 2.23 diss −52.2 34.7 1.954 0.994 1.876 0.642 −1.12 −1.22 2.24 RuO2 mol −131.3 0.0 1.038 1.588 2.172 0.634 −1.17 −0.87 1.67 diss −139.6 −8.3 1.754 1.011 1.993 0.660 −1.06 −1.08 1.68 IrO2 diss −211.5 1.614 1.014 2.018 0.725 −1.13 −1.09 1.68 (011) TiO2 mol −136.2 0.0 1.044 1.510 2.103 0.654 −1.23 −1.13 2.22 diss −140.0 −3.7 1.555 1.028 1.911 0.665 −1.18 −1.24 2.22 RuO2 mol −115.4 0.0 1.021 1.615 2.177 0.624 −1.17 −0.89 1.73 diss −102.8 12.6 1.754 1.011 1.993 0.640 −1.08 −1.04 1.79 IrO2 mol −152.9 0.0 1.043 1.549 2.108 0.703 −1.24 −0.86 1.65 diss −145.9 7.0 1.372 1.110 2.025 0.690 −1.12 −1.01 1.69 2992 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article Table 4. Adsorption Energy (ΔEads in kJ mol−1) per Water Molecule and Main Distances (in Å)a config ΔEads Ow1−Hw1 Ow1−Hw2 Obr1−Hw1 Obr2−Hw2 M1−Ow1 M1−Ow2 (110) TiO2 4mol −99.9 1.005 2.132 1.737 1.738 2.309 2.309 2mol/2diss −85.8 1.574 1.733 1.042 1.880 2.045 2.232 RuO2 4mol −138.0 0.981 2.281 1.728 1.889 2.231 2.253 2mol/2diss −147.7 1.524 1.689 0.995 2.117 2.131 2.215 IrO2 4diss −210.5 1.759 2.350 1.019 1.019 2.002 2.000 (011) TiO2 8mol −126.7 1.025 1.898 1.561 1.561 2.242 2.231 4mol/4diss −123.7 1.557 1.736 1.026 1.664 1.970 2.257 RuO2 6mol/2diss −130.4 1.510 1.514 1.035 2.705 2.067 2.173 4mol/4diss −130.8 1.591 1.531 1.024 2.049 2.099 2.175 IrO2 4mol/4diss −156.7 1.598 1.503 1.021 2.410 2.030 2.141 3mol/5diss −160.7 1.512 1.499 1.042 2.505 2.057 2.135 2mol/6diss −172.8 1.458 1.679 1.069 1.534 2.060 2.102 a See Figure 5 for water labels. preferred one on RuO2 and IrO2. This can be related to the ultimately related to (i) the kind of octahedral distortion different behavior observed regarding the water adsorption observed in the bulk of these materials (tetragonal elongation energies on (110) and (011). The interaction energy of water for TiO2 and tetragonal compression for RuO2 and IrO2) and on TiO2 (110) is smaller than on TiO2 (011), whereas for (ii) the different nature of the vacant sites created (axial or RuO2 and IrO2 the reverse trend is observed. Such differences equatorial) on these surfaces. arise from the nature of the vacant site (axial or equatorial) on Adsorption of Water Monolayer. Previous section has each surface and on the octahedral distortion observed in the shown that the relative stability between the mol and diss bulk of each material: tetragonal elongation for TiO2 (axial adsorbed forms of water on the (110) and (011) surfaces of bonds are larger than equatorial ones) and tetragonal MO2 (M = Ti, Ru, and Ir) depends on the increase of water compression for RuO2 and IrO2 (axial bonds are shorter acidity upon adsorption and on the basicity of Obr. These than equatorial ones). As mentioned above, the vacant site of factors depend on the metal oxide and on the different nature M5c in the (110) surface is axial, whereas the vacant site of M5c of the M5c vacant at each surface. At higher water coverages, in the (011) surface is equatorial. Thus, for TiO2, interaction however, cooperative H-bond interactions can come into play of water with the axial vacant site of M5c in (110)-TiO2 is in determining whether deprotonation occurs or not. Thus, we smaller than the interaction with the equatorial vacant site of have analyzed the structure of a water monolayer on each (011)-TiO2, whereas the opposite is observed for RuO2 and surface and metal oxide. For that, we have added one water IrO2. The larger the interaction is, the larger the increase of molecule at each of the unsaturated metal centers; that is, four acidity of the water molecule, which would explain that the molecules per unit cell at the (110) surface and eight water proton transfer to Obr occurs more easily on (011) than on molecules at the (011) one. We considered all possible (110) for TiO2 and on (110) than on (011) for RuO2 and combinations of mol and diss water molecules as initial IrO2. Among these latter materials, relative energies indicate structures. For instance, for the (110) surface, we considered that deprotonation on (011), although unfavorable in both seven possible structures: (i) 4 undissociated (molecular) cases, is less difficult on IrO2 (7.0 kJ mol−1) than on RuO2 water molecules (4mol), (ii) 3 molecular and 1 dissociated (12.6 kJ mol−1) as found for (110). (3mol/1diss), (iii) 2mol/2diss, (iv) 1mol/3diss, and (v) 4diss. Differences in structural parameters and net atomic charges Note that for the 2mol/2diss configuration three different are consistent with this explanation. For TiO2, M−Ow and starting situations are possible, two in which the two equal Obr−Hw distances are larger on the (110) surface (2.251 and molecules (mol or diss) are neighbors and another one in 1.794 Å) than on the (011) one (2.103 and 1.510 Å). In line which they are not. All possible combinations were also with these findings, the net charge of the water molecule when considered for the (011) surface. Adsorption energies per adsorbed in (110)-TiO2 is 0.069 a.u and 0.075 in (011)-TiO2. water molecule and structural parameters of main config- In contrast, for RuO2, the values of these distances follow the urations are given in Table 4. Figure 5 shows the optimized opposite trend; that is, M−Ow and Obr−Hw distances on the structures of the most stable configuration of each material and (110)-RuO2 surface (2.172 and 1.588 Å) are shorter than on surface. the (011)-RuO2 one (2.177 and 1.615 Å) and the net charge of First of all, it can be observed that the adsorption energy per water is 0.104 when adsorbed on (110)-RuO2 and 0.076 on water molecule in the monolayer is in all cases except (110)- (110)-RuO2. For IrO2, we could not locate the mol form on IrO2 and (011)-TiO2 larger (in absolute value) than that of a the (110) surface. The trends observed on the electron transfer single water molecule, which indicates the presence of from water to the metal oxides are in agreement with the fact cooperative effects as a result of the formation of H-bond that metal d-bands above the Fermi level are higher for TiO2 chains of moderate strength. Indeed, Figure 5 shows that two than for RuO2 and IrO2 (see Figure 4), which would explain parallel H-bond chains separated by Obr are formed for both the smaller electron donation from water to the metal and the (110) and (011) surfaces. Increases on the adsorption weaker bond in the former case. energy per water molecule range from 10 to 30 kJ mol−1 and Overall, the ability of (110) and (011) surfaces to induce result from a subtle balance between the changes induced on dissociation of an interacting water molecule seems to be the water−surface interaction, which decreases as indicated by 2993 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article per water molecule is essentially the same as that of an isolated molecule. For the (011) surface, we observe similar trends; that is, the most stable arrangement (011)-TiO2 is that in which no water molecules are dissociated (8mol), whereas for (011)-RuO2 and (011)-IrO2, the preferred configurations exhibit a 50% (4mol/ 4diss) and 75% (2mol/6diss) degree of deprotonation, respectively. In this case, the observed trends are not in agreement with that found for the adsorption of a single water molecule, which shows that the dissociated form is the preferred situation for TiO2 and the molecular form the preferred one for RuO2 and IrO2. This is due to the fact that the presence of the monolayer modifies the water−surface interaction as compared to that with a single molecule. For TiO2, the adsorption energy per water molecule in the monolayer is smaller than in the single water adsorption. Indeed, the Ti−O distance is significantly larger in the former (2.24 vs 2.10 Å). Such an increase in the M−O distance is produced to establish an efficient H-bond network but weakens the water−surface interaction, leading to a smaller Figure 5. Most stable configuration at the PBE-D2 level of theory for increase of water acidity that hinders deprotonation. Still, the the water monolayer interacting with MO2 (M = Ti, Ru, and Ir). The configuration with 50% dissociation, that is, with two H3O2− numbers 1 and 2 refer to interacting water molecules and define the and two protonated Obr, is only 3 kJ mol−1 less stable than the main structural parameters included in Table 4. fully undissociated one (Table 4), and thus, the present results are not conclusive about whether water monolayer at the (110)-TiO2 surface is dissociated or not. Furthermore, thermal the increase of the M−O distance, and the stabilizing H-bond effects need to be taken into account. interactions between the water molecules in the monolayer. At For (011)-RuO2, the most stable configuration encloses four the (110)-IrO2 surface, the adsorption energy per water molecular and four dissociated waters; that is, it shows a 50% molecule is essentially the same than that obtained for the degree of deprotonation, as in the (110) surface. Note that we isolated water molecule because the H-bonding at this surface have not been able to localize a minimum corresponding to the is the weakest one; that is, the Ow1−Hw2 distance (2.350 Å) is fully undissociated monolayer. Attempts to optimize such a the largest one. At the (011)-TiO2 surface, the adsorption structure collapsed to (6mol/2diss), the second most energy is smaller in the monolayer because there is a significant representative structure. This behavior is in contrast with the increase of the M−O distance (see Tables 3 and 4). fact that isolated water prefers a molecular adsorbed form. Most stable monolayer configuration depends on the However, formation of H3O2− species, with a strong H-bond material and on the surface. For (110)-TiO2, the most stable between the OH− and the undissociated water molecule, is arrangement is that in which no water molecules are particularly favorable. A 50% deprotonation is the preferred dissociated (4mol), whereas for (110)-RuO2, the preferred situation as it maximizes the number of H3O2− units and hence configuration has a 50% degree of deprotonation (2mol/2diss) H-bond cooperative effects. and for (110)-IrO2, the only configuration located exhibits a For (011)-IrO2, the preferred configuration shows a 75% 100% degree of deprotonation (4diss). This trend is in deprotonation (2mol/6diss), despite the molecular form being agreement with what was found for the adsorption of a single the most stable for the isolated water molecule, although with a water molecule, which showed that the preference for lower relative energy as compared to RuO2. This 75% dissociation increases from TiO2 to RuO2 and to IrO2. Indeed, deprotonation does not maximize the number of H3O2−, as for the latter material only the dissociated form was localized one would expect, because water surface interaction dominates (see Table 3). For RuO2, the 2mol/2diss situation is more over H-bonding. Furthermore, as already seen for RuO2,21 stable than the 4diss, despite the dissociation of a single water adsorption cooperative effects may induce deprotonation. That molecule being the preferred situation (Table 3), because it is, deprotonation of one water molecule favors deprotonation allows forming very stable H3O2− species. That is, this 2mol/ of a neighbor adsorbed water because of the increase of the 2diss configuration encloses two H3O2− species resulting from metal Lewis acidity as a result of the protonation of Obr. deprotonation of two water molecules to two Obr. This Overall, the degree of deprotonation results from a subtle deprotonation leads to OH− species that, due to their higher balance between H-bond cooperativity and adsorption basicity, establish strong hydrogen bonds (∼1.7 Å) with the cooperativity. In the case of IrO2, the latter effect is larger undissociated water molecules. Furthermore, as already seen because of the larger interaction with the metal sites. previously for H2O−RuO2,21 the M−OH interaction involves Present results show that relative energies corresponding to a significant electron donation to the surface that accumulates lower mol/diss configuration arrangements per water molecule on the Obr so that the charge of the H3O2− species is smaller are small (3−16 kJ mol−1) and thus may contribute to the than 1. For IrO2, the water−surface interaction dominates, in behavior of the water−metal oxide interface. On the other agreement with its much larger interaction energy (see Table hand, thermal effects may modify the relative stability of these 3) and higher surface energy (Table 2) and all water molecules configurations. Because of that, we have run AIMDs up to 8 ps dissociate. In this situation, H-bonding between metal- (1 ps equilibration) for all metal oxides and the two surfaces coordinated OH− is very weak, and thus, adsorption energy starting from the most stable monolayer obtained with static 2994 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article Figure 6. H-bond distances of two neighbor water molecules (in Å) along the dynamics. See Figure 5 for labeling. calculations. Figure 6 shows the H-bond distances correspond- of the 1mol/3diss (11.4%) and of the 2mol/2diss (4.8%). This ing to two interacting water molecules and those between indicates that thermal effects disfavor deprotonation, because these water molecules and Obr along the simulation. M−O of the increase of the M−Ow distances, which reduces the distances are reported in Figure S1 of the Supporting water−surface interaction and thus the water acidity. Information. Table 5 shows the frequency of each possible The (011) surface shows a larger proton mobility compared configuration, considering that proton transfer to an Obr occurs to the (110) one. For TiO2 and RuO2, most stable if the H-bond distance is smaller than 1.2 Å. configurations (8mol and 4mol/4diss, respectively) obtained from static calculations remain the most frequent arrangement Table 5. Percentage of Configurations with Different (94 and 72%). However, for IrO 2 , the most stable Degree of Water Deprotonation on the (110) and (011) configuration (2mol/6diss) is no longer the main one when Surfaces of MO2 (M = Ti, Ru, and Ir) including thermal effects. Indeed, three configurations account for a frequency of 95%: 4mol/4diss with 33%, 3mol/5diss with TiO2 RuO2 IrO2 43%, and 2mol/6diss with 19%. Note that the most frequent (110) configuration (3mol/5diss) does not correspond to the most 4mol 100.0 0.0 0.0 stable one obtained from static calculations (2mol/6diss). 3mol/1diss 0.0 0.0 0.0 These results again show that thermal effects tend to decrease 2mol/2diss 0.0 99.9 4.8 the M−OH2 interactions, thereby increasing the percentage of 1mol/3diss 0.0 0.1 11.4 molecular water. 4diss 8mol 0.0 94.3 (011) 0.00 0.0 83.7 0.0 ■ CONCLUSIONS The present study analyzes the H2O−MO2 (M = Ti, Ru, and 7mol/1diss 5.7 0.0 0.0 Ir) interactions by means of periodic DFT (PBE-D2) 6mol/2diss 0.0 1.11 0.0 calculations. Adsorption of both an isolated water molecule 5mol/3diss 0.0 22.7 2.6 and a full monolayer on the two surfaces that mostly 4mol/4diss 0.0 71.8 33.5 contribute to the Wulff shape, the (110) and (011) surfaces, 3mol/5diss 0.0 4.4 42.7 has been addressed. Results indicate that the adsorption of a 2mol/6diss 0.0 0.0 19.5 single molecule preferably occurs in its molecular form on the 1mol/7diss 0.0 0.0 1.4 (110)-TiO2 surface and in its dissociative form on (110)-RuO2 8diss 0.0 0.0 0.3 and (110)-IrO2. However, the opposite trend is observed on the (011) surface; that is, water prefers to adsorb in its For (110)-TiO 2 and (110)-RuO 2 , the most stable dissociative form on (011)-TiO2 and in its molecular form on configurations (4mol and 2mol/2diss, respectively) remain (011)-RuO2 and (011)-IrO2. This is related to the kind of along the 7−8 ps simulation. For (110)-TiO2, the Obr−Hw H- octahedral distortion observed in the bulk of these materials bond distance oscillates around 1.7 Å, whereas the H-bond (tetragonal elongation for TiO2 and tetragonal compression for distance between the two water molecules oscillates around 2.1 RuO2 and IrO2) and to the different nature of the vacant sites Å. In the latter case, oscillations are larger because of the created on these surfaces, axial on (110) and equatorial on weaker H-bond. For (110)-RuO2, H-bond distances are (011). Thus, water adsorption on TiO2 leads to longer M−O consistent with the presence of H3O2− species and a distances on (110) than on (011), and consequently, the protonated Obr almost all along the simulation. Note that the increase of water acidity (and possible dissociation) is larger on frequency of the 2mol/2diss is 99.9% and only the 1mol/3diss (011). The opposite is observed for RuO2 and IrO2 with arrangement appears in 0.1%. For IrO2, we observed a larger longer M−O distances on (011). Furthermore, adsorption proton mobility. As predicted by static calculations, the main energies (in absolute value) increase from TiO2 to RuO2 and configuration is a fully dissociated monolayer (4diss) with an IrO2, along with the electron transfer from the water molecule 84% frequency. Noticeably, there is a non-negligible frequency to MO2. 2995 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article For the monolayer, in addition to the intrinsic water process. Surface energies of the (110) and (011) facets were adsorption, other effects such as cooperative H-bond computed through the following equation: interactions, particularly the formation of H3O2− species, and cooperative adsorption come into play in determining whether Eslab − (N ·E bulk ) γ(hkl) = deprotonation occurs or not. Furthermore, thermal effects 2·A (1) seem to favor configurations with a smaller degree of where Eslab is the energy corresponding to the relaxed surface dissociation because of an enlargement of M−O distances, without optimizing the bulk cell parameters; Ebulk is the fully which leads to a smaller increase of water acidity. For TiO2, relaxed bulk energy; N is the number of formula units in the water monolayer is fully undissociated on both (110) and slab per units in the bulk unit cell; and 2A is the corresponding (011) surfaces, whereas for RuO2, water monolayer exhibits a two cross-sectional area of the slab. 50% dissociation, the formation of H3O2− motifs being Water adsorption on the two most stable facets of the rutile essential. Finally, on (110)-IrO2, the main monolayer polymorph, that is, (110) and (011), was simulated with a (2 × configuration is the fully dissociated one, whereas on (011)- 2) supercell in both the low and high coverage regimes. The IrO2, it exhibits a degree of dissociation that ranges from 50 to low coverage regime corresponds to the adsorption of only one 75%. Overall, the present study shows that several effects, in water molecule per unit cell, whereas in the high coverage addition to the intrinsic water adsorption, are responsible for regime, all the outermost (undercoordinated) metal atoms the degree of dissociation of adsorbed water on MO2 (M = Ti, were saturated with water molecules, leading to a water Ru, and Ir), IrO2 being the one more prone to induce monolayer. Reported adsorption energies are normalized per dissociation. ■ water molecule, according to the following equation: COMPUTATIONAL DETAILS E(hkl) + H2O − E(hkl) − (n H2O·E H2O) Periodic boundary DFT calculations were carried out with the ΔEads = n H 2O (2) Vienna ab initio simulation package (VASP) code.54,55 All calculations were performed with the GGA PBE functional56 where E(hkl)+H2O is the total energy of the slab with the plus D2 Grimme’s correction57 for dispersion and using the adsorbed water, E(hkl) is the total energy of the slab model, projector augmented wave pseudopotentials58,59 to describe EH2O is the total energy of an isolated water inside a 15 × 15 × ionic cores and valence electrons through a plane wave basis with a kinetic energy cutoff equal to 500 eV. The above 15 Å3 cubic box, and n(H2O) is the number of adsorbed waters computational parameters ensure a good agreement with onto the surfaces. The properties of these materials and their experimental cell parameters of the bulk structures for all the propensity to induce water deprotonation are discussed in studied materials (TiO2, RuO2, IrO2). Moreover, the inclusion terms of pDOS and Bader charge analysis.66−68 of dispersion corrections is essential to describe properly At this point, it is worth mentioning that TiO2 is particularly adsorption processes60 and bulk water.61,62 Bulk calculations sensitive to the computational model used.7 Concerning the were performed considering a K-point mesh for the Brillouin slab thickness of the (110) surface, our calculations show, as zone of (8,8,8), (15,15,15), and (9,9,9) for TiO2, RuO2, and found previously,36 an even odd oscillation on the water IrO2, respectively, employing the Monkhorst−Pack (MP) adsorption energy with the number of TiO2 layers (see Table grid,63 whereas slab calculations were performed considering a S2 of the Supporting Information). Most accurate results, MP K-point mesh of (3,3,1), (6,6,1), and (4,4,1) for TiO2, given by a six- or seven-layer TiO2 slab, provide the molecular RuO2, and IrO2, respectively. The cutoff and K-point mesh form as the more stable one, as found experimentally.9 Because were chosen according to the best cost/accuracy strategy of of computational reasons, and for consistency with RuO2 and both cell parameters and surface energies. The energy IrO2, present calculations correspond to a four-layer TiO2 slab. convergence criteria for electronic and geometry relaxations This model provides the correct relative stability between the were fixed to 10−5 and 10−4 eV, respectively. molecular and dissociative forms, although with a relative Because the water adsorption mode on TiO2 has been energy (34.7 kJ mol−1) that is significantly higher than that controversial and sensitive to the computational approach,7,10 recently determined by combining supersonic molecular beam, we have performed additional calculations for this system with STM, and AIMD (3.5 kJ mol−1).9 Regarding the (011) surface, a hybrid functional PBE064 and considering the D3 correction it is worth mentioning that depending on the number of layers, for dispersion65 (see Table S1 of the Supporting Information). the surface may present a significant reconstruction, as Results obtained at the PBE0-D2 and PBE-D3 levels of theory observed experimentally.11 This involves the cleavage of two show that although adsorption energies can vary up to 13 kJ internal Ti−O bonds to strengthen the Ti−O bonds with the mol−1 with respect to PBE-D2, relative energies between the surface undercoordinated Ti5c sites (see Figure S3 of the molecular and dissociative forms follow the same trend and Supporting Information). The energy difference between the vary less than 3.5 kJ mol−1, which shows the robustness of the non-reconstructed and the reconstructed one is small (Figure chosen approximation. S4), and the former surface has a higher surface energy. Surface models of the main crystallographic orientations AIMDs were carried out on the most stable water monolayer were built by cutting out the slab from the optimized bulk structures (i.e., the one with the most stable deprotonation structure. Slabs were constructed considering a (2 × 1) degree) for both (110) and (011) surfaces for all the materials. supercell and a four-layer thickness, the minimum one for a The energy convergence criteria were fixed to 10−4 eV. AIMDs reasonable converged surface energy (see Figure S2 of the were carried out considering an equilibration period of 1 ps Supporting Information). The c value was set to 35 Å ensuring (1000 steps of 1 ft) and a production period of 7 ps (7000 an interlayer distance of at least 21 Å to minimize the steps of 1 ft) in the NVT ensemble. During both the interaction between replicas in the (h k l) perpendicular equilibration and the production periods, only the water direction. Atom positions were fully relaxed in the optimization monolayer and the first layer of the surface were allowed to 2996 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article move according to the motion’s equations, while atoms of the (7) Diebold, U. Perspective: A Controversial Benchmark System for remaining surface layers were maintained at fixed positions. Water-Oxide Interfaces: H2O/TiO2(110). J. Chem. Phys. 2017, 147, This option was chosen in order to avoid unrealistic 040901. deformation of the structure of the slabs and to simulate the (8) Serrano, G.; Bonanni, B.; Di Giovannantonio, M.; Kosmala, T.; actual rigidity of the material. Schmid, M.; Diebold, U.; Di Carlo, A.; Cheng, J.; VandeVondele, J.; ■ Wandelt, K.; Goletti, C. Molecular Ordering at the Interface Between ASSOCIATED CONTENT Liquid Water and Rutile TiO2(110). Adv. Mater. Interfaces 2015, 2, 1500246. * S Supporting Information (9) Wang, Z.-T.; Wang, Y.-G.; Mu, R.; Yoon, Y.; Dahal, A.; Schenter, The Supporting Information is available free of charge on the G. K.; Glezakou, V.-A.; Rousseau, R.; Lyubinetsky, I.; Dohnálek, Z. ACS Publications website at DOI: 10.1021/acsome- Probing equilibrium of molecular and deprotonated water on ga.8b03350. TiO2(110). Proc. Natl. Acad. Sci. U.S.A. 2017, 114, 1801−1805. PBE-D2 surface energies with different slab thicknesses; (10) Sun, C.; Liu, L.-M.; Selloni, A.; Lu, G. Q.; Smith, S. C.; Smith, optimized structures for the TiO2 (011) rutile surface; S. C. Titania-Water Interactions: A Review of Theoretical Studies. J. Mater. Chem. 2010, 20, 10319. M−O distances along the AIMD; and relative energies (11) Balajka, J.; Aschauer, U.; Mertens, S. F. L.; Selloni, A.; Schmid, between molecular and dissociative adsorbed forms on M.; Diebold, U. Surface Structure of TiO2 Rutile (011) Exposed to the TiO2 (110) rutile surface, as a function of the Liquid Water. J. Phys. Chem. C 2017, 121, 26424−26431. number of layers (PDF) (12) Mu, R.; Zhao, Z.-j.; Dohnálek, Z.; Gong, J. Structural Motifs of ■ AUTHOR INFORMATION Corresponding Authors Water on Metal Oxide Surfaces. Chem. Soc. Rev. 2017, 46, 1785− 1806. (13) Siahrostami, S.; Vojvodic, A. Influence of Adsorbed Water on the Oxygen Evolution Reaction on Oxides. J. Phys. Chem. C 2014, *E-mail: xavier.solans@uab.cat (X.S.-M.). 119, 1032−1037. *E-mail: mariona.sodupe@uab.cat (M.S.). (14) Zhang, Z.; Bondarchuk, O.; Kay, B. D.; White, J. M.; Dohnálek, ORCID Z. Imaging Water Dissociation on TiO2(110): Evidence for Javier Heras-Domingo: 0000-0002-4322-3146 Inequivalent Geminate OH Groups. J. Phys. Chem. B 2006, 110, Stefano Pantaleone: 0000-0002-2457-1065 21840−21845. Albert Rimola: 0000-0002-9637-4554 (15) Bikondoa, O.; Pang, C. L.; Ithnin, R.; Muryn, C. A.; Onishi, H.; Thornton, G. Direct visualization of defect-mediated dissociation of Luis Rodríguez-Santiago: 0000-0003-4983-4228 water on TiO2(110). Nat. Mater. 2006, 5, 189−192. Mariona Sodupe: 0000-0003-0276-0524 (16) Ketteler, G.; Yamamoto, S.; Bluhm, H.; Andersson, K.; Starr, D. Author Contributions E.; Ogletree, D. F.; Ogasawara, H.; Nilsson, A.; Salmeron, M. The † Nature of Water Nucleation Sites on TiO2(110) Surfaces Revealed by D.G., J.H.-D., and S.P. have equally contributed. Notes Ambient Pressure X-Ray Photoelectron Spectroscopy. J. Phys. Chem. C 2007, 111, 8278−8282. The authors declare no competing financial interest. ■ (17) Björneholm, O.; Hansen, M. H.; Hodgson, A.; Liu, L.-M.; Limmer, D. T.; Michaelides, A.; Pedevilla, P.; Rossmeisl, J.; Shen, H.; ACKNOWLEDGMENTS Tocci, G.; Tyrode, E.; Walz, M.-M.; Werner, J.; Bluhm, H. Water at The authors gratefully acknowledge financial support from Interfaces. Chem. Rev. 2016, 116, 7698−7726. MINECO (CTQ2017-89132-P) and the Generalitat de (18) Mu, R.; Cantu, D. C.; Glezakou, V.-A.; Lyubinetsky, I.; Catalunya (2017SGR-1323). X.S.-M. is grateful for the Rousseau, R.; Dohnálek, Z. Deprotonated Water Dimers: The Professor Agregat Serra Hunter position. A.R. is indebted to Building Blocks of Segmented Water Chains on Rutile RuO2(110). “Ramon y Cajal” program. J. Phys. Chem. C 2015, 119, 23552−23558. ■ (19) Mu, R.; Cantu, D. C.; Lin, X.; Glezakou, V.-A.; Wang, Z.; REFERENCES Lyubinetsky, I.; Rousseau, R.; Dohnálek, Z. Dimerization Induced Deprotonation of Water on RuO2(110). J. Phys. Chem. Lett. 2014, 5, (1) Suen, N.-T.; Hung, S.-F.; Quan, Q.; Zhang, N.; Xu, Y.-J.; Chen, 3445−3450. H. M. Electrocatalysis for the Oxygen Evolution Reaction: Recent (20) Nguyen, M.-T.; Mu, R.; Cantu, D. C.; Lyubinetsky, I.; Development and Future Perspectives. Chem. Soc. Rev. 2017, 46, 337−365. Glezakou, V.-A.; Dohnálek, Z.; Rousseau, R. Dynamics, Stability, and (2) Paoli, E. A.; Masini, F.; Frydendal, R.; Deiana, D.; Schlaup, C.; Adsorption States of Water on Oxidized RuO2(110). J. Phys. Chem. C Malizia, M.; Hansen, T. W.; Horch, S.; Stephens, I. E. L.; 2017, 121, 18505−18515. Chorkendorff, I. Oxygen evolution on well-characterized mass- (21) Rao, R. R.; Kolb, M. J.; Hwang, J.; Pedersen, A. F.; Mehta, A.; selected Ru and RuO2 nanoparticles. Chem. Sci. 2015, 6, 190−196. You, H.; Stoerzinger, K. A.; Feng, Z.; Zhou, H.; Bluhm, H.; Giordano, (3) Reier, T.; Oezaslan, M.; Strasser, P. Electrocatalytic Oxygen L.; Stephens, I. E. L.; Shao-Horn, Y. Surface Orientation Dependent Evolution Reaction (OER) on Ru, Ir, and Pt Catalysts: A Water Dissociation on Rutile Ruthenium Dioxide. J. Phys. Chem. C Comparative Study of Nanoparticles and Bulk Materials. ACS Catal. 2018, 122, 17802−17811. 2012, 2, 1765−1772. (22) Heras-Domingo, J.; Sodupe, M.; Solans-Monfort, X. Interaction (4) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; between Ruthenium Oxide Surfaces and Water Molecules. Effect of Mi, Q.; Santori, E. A.; Lewis, N. S. Solar Water Splitting Cells. Chem. Surface Morphology and Water Coverage. J. Phys. Chem. C 2018, Rev. 2010, 110, 6446−6473. DOI: 10.1021/acs.jpcc.8b06438. (5) Song, F.; Bai, L.; Moysiadou, A.; Lee, S.; Hu, C.; Liardet, L.; Hu, (23) Lobo, A.; Conrad, H. Interaction of H2O with the RuO2(110) X. Transition Metal Oxides as Electrocatalysts for the Oxygen Surface Studied by HREELS and TDS. Surf. Sci. 2003, 523, 279−286. Evolution Reaction in Alkaline Solutions: An Application-Inspired (24) Jaimes, R.; Vazquez-Arenas, J.; González, I.; Galván, M. Renaissance. J. Am. Chem. Soc. 2018, 140, 7748−7759. Theoretical evidence of the relationship established between the HO (6) Yin, W.-J.; Wen, B.; Zhou, C.; Selloni, A.; Liu, L.-M. Excess radicals and H2O adsorptions and the electroactivity of typical electrons in reduced rutile and anatase TiO2. Surf. Sci. Rep. 2018, 73, catalysts used to oxidize organic compounds. Electrochim. Acta 2017, 58−82. 229, 345−351. 2997 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article (25) Kuo, D.-Y.; Kawasaki, J. K.; Nelson, J. N.; Kloppenburg, J.; density functional theory, X-ray diffraction, and scanning tunneling Hautier, G.; Shen, K. M.; Schlom, D. G.; Suntivich, J. Influence of microscopy study. Surf. Sci. 2009, 603, 138−144. Surface Adsorption on the Oxygen Evolution Reaction on IrO2(110). (45) Aschauer, U.; Selloni, A. Structure of the Rutile TiO2(011) J. Am. Chem. Soc. 2017, 139, 3473−3479. Surface in an Aqueous Environment. Phys. Rev. Lett. 2011, 106, (26) Sasahara, A.; Tomitori, M. An Atomic-Scale Study of 166102. TiO2(110) Surfaces Exposed to Humid Environments. J. Phys. (46) Barnard, A. S.; Zapol, P.; Curtiss, L. A. Modeling the Chem. C 2016, 120, 21427−21435. Morphology and Phase Stability of TiO2 Nanocrystals in Water. J. (27) Brookes, I. M.; Muryn, C. A.; Thornton, G. Imaging Water Chem. Theory Comput. 2005, 1, 107−116. Dissociation on TiO2(110). Phys. Rev. Lett. 2001, 87, 266103. (47) Di Valentin, C.; Tilocca, A.; Selloni, A.; Beck, T. J.; Klust, A.; (28) Matthiesen, J.; Hansen, J. Ø.; Wendt, S.; Lira, E.; Schaub, R.; Batzill, M.; Losovyj, Y.; Diebold, U. Adsorption of Water on Lægsgaard, E.; Besenbacher, F.; Hammer, B. Formation and Diffusion Reconstructed Rutile TiO2(011)-(2×1): TiO Double Bonds and of Water Dimers on Rutile TiO2(110). Phys. Rev. Lett. 2009, 102, Surface Reactivity. J. Am. Chem. Soc. 2005, 127, 9895−9903. 226101. (48) Rao, R. R.; Kolb, M. J.; Halck, N. B.; Pedersen, A. F.; Mehta, (29) Lee, J.; Sorescu, D. C.; Deng, X.; Jordan, K. D. Water Chain A.; You, H.; Stoerzinger, K. A.; Feng, Z.; Hansen, H. A.; Zhou, H.; Formation on TiO2(110). J. Phys. Chem. Lett. 2012, 4, 53−57. Giordano, L.; Rossmeisl, J.; Vegge, T.; Chorkendorff, I.; Stephens, I. (30) Kimmel, G. A.; Baer, M.; Petrik, N. G.; VandeVondele, J.; E. L.; Shao-Horn, Y. Towards Identifying the Active Sites on Rousseau, R.; Mundy, C. J. Polarization- and Azimuth-Resolved RuO2(110) in Catalyzing Oxygen Evolution. Energy Environ. Sci. Infrared Spectroscopy of Water on TiO2(110): Anisotropy and the 2017, 10, 2626−2637. Hydrogen-Bonding Network. J. Phys. Chem. Lett. 2012, 3, 778−784. (49) Bolzan, A. A.; Fong, C.; Kennedy, B. J.; Howard, C. J. Structural (31) Henderson, M. A. An HREELS and TPD Study of Water on Studies of Rutile-Type Metal Dioxides. Acta Crystallogr., Sect. B: TiO2(110): The Extent of Molecular versus Dissociative Adsorption. Struct. Sci. 1997, 53, 373−380. Surf. Sci. 1996, 355, 151−166. (50) Ping, Y.; Galli, G.; Goddard, W. A. Electronic Structure of IrO2: (32) Walle, L. E.; Borg, A.; Uvdal, P.; Sandell, A. Experimental The Role of the Metal d Orbitals. J. Phys. Chem. C 2015, 119, 11570− Evidence for Mixed Dissociative and Molecular Adsorption of Water 11577. on a Rutile TiO2(110). Phys. Rev. B: Condens. Matter Mater. Phys. (51) Berlijn, T.; Snijders, P. C.; Delaire, O.; Zhou, H.-D.; Maier, T. 2009, 80, 235436. A.; Cao, H.-B.; Chi, S.-X.; Matsuda, M.; Wang, Y.; Koehler, M. R.; (33) Duncan, D. A.; Allegretti, F.; Woodruff, D. P. Water Does Kent, P. R. C.; Weitering, H. H. Itinerant Antiferromagnetism in Partially Dissociate on the Perfect TiO2 (110) Surface: A Quantitative RuO2. Phys. Rev. Lett. 2017, 118, 077201. Structure Determination. Phys. Rev. B: Condens. Matter Mater. Phys. (52) Torun, E.; Fang, C. M.; de Wijs, G. A.; de Groot, R. A. Role of 2012, 86, 045411. Magnetism in Catalysis: RuO2 (110) Surface. J. Phys. Chem. C 2013, (34) Diebold, U. The Surface Science of Titanium Dioxide. Surf. Sci. 117, 6353−6357. Rep. 2003, 48, 53−229. (53) Novell-Leruth, G.; Carchini, G.; López, N. On the Properties of (35) Gutiérrez Moreno, J. J.; Fronzi, M.; Lovera, P.; O’Riordan, A.; Binary Rutile MO2 Compounds, M = Ir, Ru, Sn, and Ti: A DFT Nolan, M. Stability of Adsorbed Water on TiO2-TiN Interfaces. A Study. J. Chem. Phys. 2013, 138, 194706. First-Principles and Ab Initio Thermodynamics Investigation. J. Phys. (54) Kresse, G.; Furthmüller, J. Efficient iterative schemes forab Chem. C 2018, 122, 15395−15408. initiototal-energy calculations using a plane-wave basis set. Phys. Rev. (36) Kowalski, P. M.; Meyer, B.; Marx, D. Composition, Structure, B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. and Stability of the Rutile TiO2. Phys. Rev. B: Condens. Matter Mater. (55) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid Phys. 2009, 79, 115410. metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561. (37) Kumar, N.; Kent, P. R. C.; Wesolowski, D. J.; Kubicki, J. D. (56) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Modeling Water Adsorption on Rutile (110) Using van Der Waals Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. Density Functional and DFT+U Methods. J. Phys. Chem. C 2013, 117, (57) Grimme, S. Accurate Description of van Der Waals Complexes 23638−23644. by Density Functional Theory Including Empirical Corrections. J. (38) Nilsson, J. O.; Leetmaa, M.; Wang, B.; Ž guns, P. A.; Pašti, I.; Comput. Chem. 2004, 25, 1463−1473. Sandell, A.; Skorodumova, N. V. Modeling Kinetics of Water (58) Blöchl, P. E. Projector Augmented-Wave Method. Phys. Rev. B: Adsorption on the Rutile TiO2 (110) Surface: Influence of Condens. Matter Mater. Phys. 1994, 50, 17953−17979. Exchange-Correlation Functional. Phys. Status Solidi B 2017, 255, (59) Kresse, G.; Joubert, D. From Ultrasoft Pseudopotentials to the 1700344. Projector Augmented-Wave Method. Phys. Rev. B: Condens. Matter (39) Hosseinpour, S.; Tang, F.; Wang, F.; Livingstone, R. A.; Mater. Phys. 1999, 59, 1758−1775. Schlegel, S. J.; Ohto, T.; Bonn, M.; Nagata, Y.; Backus, E. H. G. (60) Rimola, A.; Costa, D.; Sodupe, M.; Lambert, J.-F.; Ugliengo, P. Chemisorbed and Physisorbed Water at the TiO2/Water Interface. J. Silica Surface Features and Their Role in the Adsorption of Phys. Chem. Lett. 2017, 8, 2195−2199. Biomolecules: Computational Modeling and Experiments. Chem. (40) Calegari Andrade, M. F.; Ko, H.-Y.; Car, R.; Selloni, A. Rev. 2013, 113, 4216−4313. Structure, Polarization, and Sum Frequency Generation Spectrum of (61) Ohto, T.; Dodia, M.; Imoto, S.; Nagata, Y. Structure and Interfacial Water on Anatase TiO2. J. Phys. Chem. Lett. 2018, 9, 6716− Dynamics of Water at the Water-Air Interface Using First-Principles 6721. Molecular Dynamics Simulations within Generalized Gradient (41) Liu, L.-M.; Zhang, C.; Thornton, G.; Michaelides, A. Structure Approximation. J. Chem. Theory Comput. 2018, 15, 595−602. and Dynamics of Liquid Water on Rutile TiO2(110). Phys. Rev. B: (62) Gillan, M. J.; Alfè, D.; Michaelides, A. Perspective: How Good Condens. Matter Mater. Phys. 2010, 82, 161415. Is DFT for Water? J. Chem. Phys. 2016, 144, 130901. (42) Agosta, L.; Brandt, E. G.; Lyubartsev, A. P. Diffusion and (63) Monkhorst, H. J.; Pack, J. D. Special Points for Brillouin-Zone Reaction Pathways of Water near Fully Hydrated TiO2 Surfaces from Integrations. Phys. Rev. B: Solid State 1976, 13, 5188−5192. Ab Initio Molecular Dynamics. J. Chem. Phys. 2017, 147, 024704. (64) Adamo, C.; Barone, V. Toward Reliable Density Functional (43) Torrelles, X.; Cabailh, G.; Lindsay, R.; Bikondoa, O.; Roy, J.; Methods without Adjustable Parameters: The PBE0 Model. J. Chem. Zegenhagen, J.; Teobaldi, G.; Hofer, W. A.; Thornton, G. Geometric Phys. 1999, 110, 6158−6170. Structure of TiO2(011)(2×1). Phys. Rev. Lett. 2008, 101, 185501. (65) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and (44) Gong, X.-Q.; Khorshidi, N.; Stierle, A.; Vonk, V.; Ellinger, C.; accurate ab initio parametrization of density functional dispersion Dosch, H.; Cheng, H.; Selloni, A.; He, Y.; Dulub, O.; Diebold, U. The correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010, 2×1 reconstruction of the rutile TiO2(011) surface: A combined 132, 154104. 2998 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999 ACS Omega Article (66) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Oxford University Press: New York, 1994. (67) Yu, M.; Trinkle, D. R. Accurate and Efficient Algorithm for Bader Charge Integration. J. Chem. Phys. 2011, 134, 064111. (68) Tang, W.; Sanville, E.; Henkelman, G. A Grid-Based Bader Analysis Algorithm without Lattice Bias. J. Phys.: Condens. Matter 2009, 21, 084204. 2999 DOI: 10.1021/acsomega.8b03350 ACS Omega 2019, 4, 2989−2999