Water Dissociates at the Aqueous Interface with Reduced Anatase TiO2 (101).

Elucidating the structure of the interface between natural (reduced) anatase TiO2 (101) and water is an essential step toward understanding the associated photoassisted water splitting mechanism. Here we present surface X-ray diffraction results for the room temperature interface with ultrathin and bulk water, which we explain by reference to density functional theory calculations. We find that both interfaces contain a 25:75 mixture of molecular H2O and terminal OH bound to titanium atoms along with bridging OH species in the contact layer. This is in complete contrast to the inert character of room temperature anatase TiO2 (101) in ultrahigh vacuum. A key difference between the ultrathin and bulk water interfaces is that in the latter water in the second layer is also ordered. These molecules are hydrogen bonded to the contact layer, modifying the bond angles.

Ever since Honda and Fujishima[1] demonstrated photoassisted water splitting on titanium dioxide (TiO2), it has been widely investigated for hydrogen fuel production.[2] Determining the interface structures of well-defined TiO2 surfaces and water is a crucial step toward understanding this process on an atomic scale. Rutile TiO2 (110) (R110) and anatase TiO2 (101) (A101) have been the focus of numerous surface science studies. While the structure of the R110/H2O interface has been studied in a number of environments,[3,4] studies of A101 have so far been largely restricted to ultrahigh vacuum (UHV).[5−13] A surface science perspective of water on brookite[14] is limited to simulations.

The A101 surface consists of 5-fold (Ti5c) and 6-fold (Ti6c) coordinated Ti atoms and 2-fold (O2c) and 3-fold (O3c) coordinated O atoms in a sawtooth geometry (see Figure ).[15,16] Water does not adsorb on A101 in UHV conditions at room temperature, although it adsorbs molecularly on Ti5c at low temperature.[5] Dissociative adsorption to form terminal OH (OHt)[6,7] (i.e., OH adsorbed to Ti5c) and/or bridging OH (OHbr)[12,13] has been reported following electron[12,13] and photon excitation[9] as well as coadsorption with O2[6,7] at low temperature. There is evidence from photoemission spectroscopy of mixed molecular-dissociative adsorption at room temperature at a higher pressure of water (0.6–6.0 mbar).[17]

Figure 1

Ball and stick model of A101 (1 × 1). The numerical labeling of the atoms serves as identification for the atomic displacements shown in Table . The indicated azimuth defines the x, y, and z directions along which the atomic coordinates are defined as positive.

Table 1

Experimental (SXRD) and Theoretical (DFT) Surface Atomic Displacements Away from the Bulk Terminated Structure of A101a

	displacements (Å)
	A101/UHV (as-prepared)		A101/ultrathin water film		A101/bulk water
atom label	Δ [101̅] [[16]SXRDb:SXRDc:DFT]	Δ [101] [[16]SXRDb:SXRDc:DFT]	Δ [101̅] [SXRD:DFT]	Δ [101] [SXRD:DFT]	Δ [101̅] [SXRD:DFT]	Δ [101] [SXRD:DFT]
O-1	0.11:0.14:0.23	0.07:0.10:0.02	0.05:–0.01	–0.01:0.08	–0.03:–0.05	0.03:0.05
Ti-1	0.03:0.02:–0.01	0.01: −0.01:–0.12	0.02:0.00	0.07:0.15	–0.02:–0.06	0.11:0.13
O-2	0.11:0.13:0.14	0.15:0.14:0.25	–0.08:–0.02	0.12:0.16	–0.03:–0.04	0.09:0.17
O-3	0.18:0.16:0.11	0.08:0.05:0.06	–0.08:–0.04	0.10:0.00	–0.04:–0.09	0.04:0.01
Ti-2	0.12:0.11:0.12	0.15:0.16:0.21	0.08:–0.03	0.09:0.01	0.02:–0.07	0.06:0.03
O-4	–0.01:0.01:0.13	0.01:0.01:–0.02	0.15:–0.04	0.00:0.07	0.04:–0.05	0.01:0.08
O-5	–0.07:–0.04:0.05	0.06:0.06:0.06	–0.03:–0.03	–0.03:0.04	0.05:–0.04	0.04:0.04
Ti-3	0.01:–0.01:–0.05	0.04:0.03:–0.05	0.06:–0.02	0.06:0.07	0.05:–0.03	0.05:0.07
O-6	–0.06:–0.05:–0.01	0.05:0.07:0.02	–0.13:–0.03	0.07:0.06	–0.04:–0.03	0.05:0.07
O-7	0.13:0.14:0.01	0.08:0.05:0.04	0.01:–0.03	0.04:0.05	0.05:–0.04	0.02:0.05
Ti-4	0.06:0.08:0.01	0.08:0.09:0.10	0.07:–0.03	0.07:0.03	0.03:–0.04	0.03:0.04
O-8	–0.05:–0.04:–0.01	0.00:0.04:0.02	0.03:–0.03	0.00:0.04	0.01:–0.03	0.01:0.05

Positive or negative displacements indicate those parallel or anti-parallel to the directions of the coordinate axis defined in Figure . Experimental errors correspond to ±0.01 Å as obtained from the fitting procedure.

Represents as-prepared surface before formation of the ultrathin water film interface (10 ± 2 layers).

Represents as-prepared surface before formation of the bulk water interface.

The reduced room temperature reactivity of A101 to water in UHV compared with R110 is thought to be due to the lack of surface oxygen vacancies.[18] These vacancies, which promote dissociation on R110,[19] are absent on A101 because they are more stable in subsurface sites.[18] However, the unreactive character of reduced A101 in UHV is predicted to be modified when a liquid interface is formed by trapping excess electrons at bound hydroxyl complexes.[10] Here we test this idea through a quantitative structure determination of the A101 surface covered by an ultrathin water film or bulk water, complemented with density functional theory (DFT) calculations. We find that a mixture of molecular and dissociated water is present in the contact layer, pointing to a significantly enhanced reactivity of the substrate compared with that observed in UHV.

The interface structures for A101 with an ultrathin film and bulk water were obtained from surface X-ray diffraction (SXRD) data in comparison with DFT calculations. SXRD data recorded from the clean surface in UHV prior to the interface measurements are essentially identical to those published previously.[16] Labeling of titanium and oxygen atoms used here is identical to that used in our previous work[16] (see Figure ). The atomic displacements on the as-prepared surface, given in Table , indicate a relaxation of atoms away from the bulk, a phenomenon previously observed on R110 in UHV.[20] As discussed in our previous work,[16] surface roughness has been modeled with a “terraced roughness”[21] approach, which allows better simulation of the step-related surface roughness. Modeling is performed with two surface domains with identical terminations that differ in the relative height from the bulk at which the termination occurs. Occupancy of the two domains is in a 1:3 ratio as in our previous work.[16]

Three experimental crystal truncations rods (CTRs) and the best fits for A101 covered with a 10 ± 2 monolayer water film (see SI for details) and bulk water are shown in Figure , with the complete data sets in Figures S1 and S2, respectively. The A101 surface atomic displacements before and after formation of the water interface are shown in Table . These optimized atomic displacements indicate mixed associative and dissociative water adsorption with a normalized χ2 (χ2n) of 1.12 and 1.05 for the ultrathin water film and bulk water, respectively. The nonuniform agreement between certain experimental and DFT displacements is largely attributed to calculation limitations. Displacements were determined by considering the difference of the optimized atomic positions without sampling different atomic configurations, as would be more appropriate at finite temperature, especially at the aqueous interface.

Figure 2

Comparison of experimental CTRs for as-prepared A101 in UHV[16] (red), and for the A101 interface with an ultrathin water film (blue) and bulk water (gray). CTRs are offset for clarity. A full set of CTRs and their respective best fit is given in the SI (see Figures S1 and S2). FExperiment: experimental structure factor. FCalculated: calculated structure factor.

The best-fit SXRD model for the A101 interface with the ultrathin water film suggests ordering in the contact layer only, with a complete coverage of adsorbed H2O/OH species on Ti5c (Ti-1). There are two distinct Ti5c–OH2/OH species with 25% and 75% coverage and bond lengths of 2.21 ± 0.04 Å and 1.94 ± 0.01 Å, respectively (see Figure and see Figure S3a for a graph of χ2n against a change in surface adsorbate coverage). The best-fit SXRD model for the interface with bulk water has an additional ordered layer above the contact layer. Similar to the ultrathin case, the contact layer contains two distinct Ti5c–OH2/OH species with 25% and 75% surface coverage and bond lengths of 2.18 ± 0.03 Å and 1.95 ± 0.01 Å, respectively (see Figure and see Figure S3a for a graph of χ2n against a change in surface adsorbate coverage). The second layer appears to consist of H2O molecules that are hydrogen bonded to molecules in the contact layer based on the bond distances (see Figure ). Interestingly, as is shown in Figure , the bond angle of the H2O and OHt species on Ti5c (Ti-1) varies depending on whether the surface is contacted with the ultrathin film or bulk water. This can be attributed to the presence of the ordered second layer in the case of the bulk water interface (see Figure S3b for a graph of χ2n against a change in second monolayer coverage).

Figure 3

Ball and stick model of the proposed A101 interface with (a) an ultrathin water film and (b) bulk water. Experimental (SXRD) bond lengths and angles are presented in black, with DFT calculations denoted in red. The black arrows represent the relative magnitude and direction of atom displacements with respect to bulk lattice positions. Hydrogen atoms were not included in the experimental fitting procedure due to their small X-ray scattering strength, and so are only displayed for illustrative purposes. A complete coverage of adsorbed H2O/OH on Ti5c is proposed. However, for presentation purposes, this figure shows only one adsorbed H2O and OHt.

Previous DFT and molecular dynamics (MD) simulations[22−31] predict a Ti5c–OH bond length in the range 2.15–2.30 Å, while the Ti5c–OOH bond length is predicted to be 1.80–1.90 Å. Our current DFT calculations predict the Ti5c–OH2 bond length to be 2.27–2.28 Å and the Ti5c–OHt bond length to be 1.85–1.89 Å (see Figure ). Experimental measurements of the R110/H2O(l) interface show a Ti5c–OHt bond length at 1.95 ± 0.03 Å.[4] On this basis, the Ti5c-O bond lengths of 2.21 ± 0.05 Å (ultrathin film) and 2.18 ± 0.03 Å (bulk water) can be attributed to associative H2O surface adsorption on Ti5c, while the bond lengths of 1.94 ± 0.01 Å (ultrathin film) and 1.95 ± 0.01 Å (bulk water) correspond to dissociative adsorption to form Ti5c–OHt. A 25% occupation of Ti5c sites by molecular water was also found in UHV scanning tunneling microscopy (STM) images following exposure of A101 to water vapor at 6 K.[5] At low temperature this forms a locally ordered 2 × 2 overlayer, which could in principle be present at the ultrathin water film and bulk water interface. The small domain size would prevent fractional order rods (FORs) from being observed.

The surface atomic displacements after formation of the aqueous interfaces are in general close to zero. In other words, the expansion of the surface observed in UHV is reversed with the formation of the interface. This behavior has been previously observed at the R110/H2O interface[4] and is well reproduced by our DFT calculations. Interestingly, experiment and theory suggest an expansion away from the bulk for the Ti5c (Ti-1) atom. This movement is attributed to the formation of Ti5c (Ti-1) bonds to OH2/OH in the contact layer. The experimental bond angles associated with molecules in the contact layer (see Figure ) are reproduced reasonably well by our theoretical calculations. Any discrepancies can be attributed to limitations associated with the optb88-vdw DFT functional. This functional has been shown to simulate the aqueous environment better than the Perdew, Burke, and Ernzerhof (PBE) functional[32,33] and has been used to describe several semiconductor/water interfaces accurately, although its suitability to reproduce bond angles is as yet unclear.[32−35]

Previous calculations of the A101/water interface predict that the formation of OHt species from water dissociation is coupled with the formation of OHbr species that trap excess electrons from the selvedge.[10] In principle, this can be probed experimentally by the position of the O2c to which a H atom is bound to form OHbr.[27] Our SXRD results indicate that, after H2O exposure, the Ti5c–O2c bond length increases from 1.90 ± 0.02 Å to 1.95 ± 0.01 Å and 1.88 ± 0.01 Å to 1.95 ± 0.01 Å, respectively for the ultrathin water film and bulk water interfaces. Earlier calculations[27] predict that the Ti5c–O2c bond length is 1.86 Å for the clean surface, which can increase up to 1.88 Å in the presence of OHt and H2O species at the Ti5c (Ti-1) site. However, in the presence of both OHbr and OHt species, the Ti5c–O2c bond length can increase up to 2.01 Å. This is supported by our current DFT calculations, which show that the presence of OHbr species can result in a Ti5c-O2c bond length of ∼2 Å, whereas in the absence of OHbr and with only H2O or OHt adsorption at the Ti5c site, the Ti5c-O2c bond length is ∼1.85 Å. Given that our experimental findings indicate an expansion of the Ti5c-O2c bond length after aqueous interface formation, it can be inferred that the interface consists of OHbr species formed via H2O dissociation to form OHt and OHbr species.

The influence of the water layer thickness on the contact layer structure has been discussed in the literature, although there has been a lack of experimental evidence.[36] In our work, we observe that our ultrathin water film and bulk water on A101 induces a similar contact layer with differences arising from an ordered second monolayer at the A101/bulk water interface. This indicates that the ultrathin water film thickness of 10 ± 2 monolayers[37] is below that required for it to behave as bulk water. The activation of A101 to induce water dissociation at the aqueous interface while being inert in UHV can be explained in terms of the interplay between excess electrons and adsorbed water. Although little electron trapping is observed at the surface of as-prepared A101 in UHV, an excess electron at the aqueous interface can trigger water dissociation to form surface OH species.[10] The catalytic activation of A101 under aqueous conditions can be explained by the interaction of excess electrons with multiple water layers and the subsequent electron trapping at the resultant OH species.

In conclusion, this study has shown that room temperature aqueous interfaces with reduced A101 have a mixture of molecular H2O (25%) and OHt (75%) bound to Ti5c in the contact layer. OHt formation from water dissociation is accompanied by the formation of OHbr. On the basis of previous calculations,[10] the reduced state of the anatase will play a crucial role in the formation of this contact layer since it provides the excess electrons needed for dissociation. Upon water exposure to the as-prepared surface, the surface atoms contract toward the bulk and adopt a relatively more bulk-like appearance when compared to the as-prepared surface. This study highlights the importance of the substrate environment in determining its reactivity. For A101, the aqueous interface is reactive, whereas the UHV substrate is inert at room temperature. Since the aqueous interface is relevant in photocatalysis, it also highlights the importance of studies in realistic environments. This behavior of A101 is likely to be observed on other reducible metal oxides; however, this will depend on its surface electronic structure. For instance, in contrast to the (101) termination, the (001) termination of anatase does not trap electrons.[10]