Article pubs.acs.org/JPCC
Water Adsorption on TiOx Thin Films Grown on Au(111) M. H. Farstad,† D. Ragazzon,‡ L. E. Walle,† A. Schaefer,§ A. Sandell,‡ and A. Borg*,† †
Department of Physics, Norwegian University of Science and Technology, NO-7491 Trondheim, Norway Department of Physics and Astronomy, Uppsala University, P.O. Box 530, SE-75121 Uppsala, Sweden § Institute for Applied and Physical Chemistry and Center for Environmental Research and Sustainable Technology, University of Bremen, Leobener Str. UFT, D-28359 Bremen, Germany ‡
ABSTRACT: High resolution photoelectron spectroscopy has been used to investigate water adsorption on four different TiOx ultrathin film structures, grown on Au(111) by chemical vapor deposition. Two of the structures are reduced TiOx single layer phases, forming a honeycomb (HC) and a pinwheel (PW) structure, respectively. The other two phases have TiO2 stoichiometry, one in the form of islands and one in the form of a TiO2(B)(001) extended layer. Partial water dissociation is observed for all phases but the HC phase, and the dissociation propensity and adsorbate thermal stability structure result from interplay between the atomic structure of the particular TiOx phase and defects formed in the preparation. The dissociation on the TiO2(B) film is mainly related to different types of defect sites. The TiO2 islands, interpreted as surface reconstructed rutile TiO2(100), generate the highest amount of hydroxyls with a behavior consistent with reconstruction into a mixed (100) and (110) termination. Water dissociation on the PW layer can be assigned to particular sites of the structure and it stands out by leading to oxidation of Ti species.
■
INTRODUCTION TiO2 is a material with a wide range of applications, most of which rely on the special properties of the titania surfaces.1 Examples are wastewater remediation,2−4 catalytic CO oxidation,5 H2 production,6 and as disinfectant agent.7−9 For this reason, substantial efforts have been directed to the development of TiO2 materials that expose as much surface area as possible. This has led to the synthesis of various nanodimensional forms of TiO2 and investigations of the novel properties that appear when reducing the size. Furthermore, a very large number of industrial catalysts comprise several components and a typical configuration is metal particles dispersed over an oxide surface. A system that has attracted considerable interest from both an applied and fundamental point of view is Au/TiO2. Au nanoclusters depostited on TiO2 has been found to be catalytically active for CO oxidation5,10 and for H2 production from ethanol− water mixtures.6 The inverse system, that is, TiOx nanostructures on a Au support, is active toward the water−gas shift (WGS) reaction (CO + H2O ↔ CO2 + H2). The synergetic effects displayed by mixed systems can be ascribed to special sites formed at the boundary between the materials and to bifunctionality of the materials.11 In the case of WGS, titania is responsible for dissociation of water and gold for the subsequent oxidation of CO. Thus, comparing the surface chemical properties of Au/TiO2 with TiO2/Au is expected to provide complementary understanding of the reaction mechanisms involved12 and enable further tuning of the catalyst. In the present work, we address the ultrahigh vacuum (UHV) water adsorption chemistry of various TiOx nanostructures deposited on the Au(111) surface. A critical issue is the © 2015 American Chemical Society
capability of the TiOx structures to dissociate water since, as noted above; this is the key to the activity toward WGS. However, even when water dissociation is not a vital part in the reaction, H+ and OH− species may serve as charge carriers or oxidation agents improving the reaction rate.2,10,13,14 The dissociation of water on a nondefective titania surface entails adsorption of a water molecule on an acid site (surface Ti ions) and subsequent dissociation by proton donation to an adjacent base site (surface O ions). Previous studies have pointed out the correlation between the particular Ti and O arrangement at the surface and activity toward water dissociation.15 Another important aspect is the occurrence of structural imperfections in the form of oxygen vacancies, Ti interstitials and steps. All these can contribute to water dissociation. To exemplify, on the benchmark surface, rutile TiO2(110), it is by now well established that water dissociates at oxygen vacancies formed at the surface16 and dissociation has also been proposed to occur at steps.17 The most recent experimental and theoretical efforts also show that partial water dissociation occurs on the nondefective TiO2(110) surface.18−23 In the process of characterizing water adsorption on new titania nanostructures it is therefore convenient to build on the wealth of results available for water on TiO2 single crystal surfaces. We have investigated the formation of TiOx nanostructures on Au(111) upon Chemical Vapor Deposition (CVD). CVD is a very efficient and cost-effective thin film deposition method Received: January 19, 2015 Revised: March 5, 2015 Published: March 5, 2015 6660
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C
showed the herringbone reconstruction, indicating a clean and well ordered surface. CVD was used for TiOx film growth with TTIP (SigmaAldrich, purity 99.999%) as precursor. The TTIP was kept in a glass tube connected to the vacuum chamber by a leak valve, and was dosed onto the sample through a stainless-steel tube which opening was positioned a few centimeters from the Au(111) surface. Dosing pressures ranged from 1 × 10−9 mbar to 5 × 10−8 mbar. Prior to deposition, the TTIP was thoroughly pumped and care was taken to avoid contamination by water. The HC and Star TiOx thin films were prepared by depositing TTIP at 300 and 500 °C, respectively. For depositions at 500 °C the growth of the Star phase is predominant, but still a small initial growth of HC is observed. Similarly, HC is the predominant phase upon growth at 300 °C, but disordered TiO2 clusters are also formed. A detailed description of the growth as well as structural and chemical properties of these phases is reported by Ragazoon et al.25 The PW wetting layer phase was formed by annealing the HC phase to 500 °C.27 The TiO2(B) phase was prepared in two consecutive steps. First a thick layer of TiOx was grown at 290 °C, displaying only a faint 1 × 1 LEED pattern, then the disordered film was annealed to 500 °C resulting in a film consisting of TiO2(B)(001) domains with different orientations.26 The water experiments were performed by dosing 5 L D2O through a leak valve, keeping the sample at about 90 K, followed by heating at the selected annealing temperature for 1 min, then cooling down again before measurements. This annealing procedure was performed in steps of increasing annealing temperature until no traces of water related species were detected on the surface. This sequence is referred to as a water annealing series. D2O has been used in these experiments in order to minimize the possibility of any beam induced water dissociation on the thin films studied. However, detailed studies of water on rutile TiO2(110) have shown that beam induced dissociation is not an issue at the conditions offered by the beamline used (moderate flux and large spot size) and we refer to this work for details.29 Neither do we observe beam induced reduction of the oxide films, which would otherwise be clearly visible in the Ti 2p spectra. In addition, D2O due to the different zero point energy compared to H2O, may display an isotope effect. Such an effect has not been seen comparing D2O and H2O adsorption on rutile TiO2(110) at 210 K.29 To avoid possible isotope effects D2O adsorption only has been present work only D2O desorption is considered. Data Analysis. In the following, the procedure for analyzing the HRPES data is briefly described. The binding energy (BE) calibration of the HRPES spectra was performed by recording either the Fermi edge or the Au 4f7/2 peak when the Fermi edge was indiscernible, immediately after the core level region. A linear sum of Voigt or Doniach-Sunjic profiles was used for fitting the spectra. A Lorentzian fwhm of 0.15 eV was applied to all peaks. All spectra were normalized at the low binding side of the respective core level regions, with exception of the valence band region, which was normalized to the ring current and number of sweeps. Background subtraction for the O 1s spectrum was done by subtracting a modeled background obtained from the clean preparations. Above BE 533 eV a spectrum of the clean preparation was used to model the region below the water peak, below 533 eV the Shirley background of the O 1s peaks of the
often used in industry. In particular, CVD is a suitable method for decoration of substrates with high surface-area-to-volume ratio. We have for instance demonstrated the feasibility of CVD to modify nanoporous gold by titania, thereby increasing its activity toward CO oxidation.24 We have also reported on CVD using titanium(IV) isopropoxide (TTIP) as precursor for growing TiOx thin film structures on Au(111).25,26 By means of low energy electron diffraction (LEED), scanning tunneling microscopy (STM), and high resolution photoelectron spectroscopy (HRPES), the structure of four different TiOx thin film phases on Au(111) were identified: Two phases consist of reduced wetting layers, which on the basis of STM images, form a honeycomb (HC) and a pinwheel (PW) structure, respectively.25,27 The two other phases are fully oxidized, that is, with TiO2 stoichiometry. One of these, labeled Star phase, based on its LEED pattern, consists of islands which possibly exposes a reconstructed rutile TiO2(100) surface. This phase is thermodynamically favored.25 The other TiO2 phase is metastable and it has been shown to consist of a TiO2(B)(001) structure.26 Here, we first present the most important characteristics of the four different TiOx thin film phases on Au(111) grown by TTIP-mediated CVD. Next, the adsorption and subsequent heating of water layers on the four phases are discussed on the basis of HRPES results. HRPES is a powerful tool to explore water chemistry at oxide surfaces as the O 1s spectrum presents resolvable components from intact water, hydroxyls and the oxide substrate, respectively. A particular advantage when using synchrotron light (as in the present work) is that the photon energy can be chosen as to maximize the surface sensitivity. On the basis of O 1s spectra we can therefore provide an overview of the interaction of water with various TiOx/Au(111) situations. We discuss the inherent properties of the four different phases versus defect related properties. To aid the interpretation direct comparisons are made to results for water adsorbed on rutile TiO2(110).
■
EXPERIMENTAL SECTION HRPES and STM experiments were conducted in separate chambers under ultrahigh vacuum (UHV) conditions, at pressures in the low 10−10 mbar range. All chambers were equipped with LEED, ion gun, and sample heating capabilities. The HRPES measurements were performed at the MAX IV Laboratory in Sweden, at beamline D1011 of the MAX II ring. This beamline is equipped with a modified Zeiss SX-700 monochromator and a Scienta SES200 electron analyzer.28 Ti 2p spectra were measured using a photon energy of 590 eV, O 1s spectra were recorded at 650 eV, these regions have an energy resolution of about 400−500 meV. Au 4f spectra were measured at 220 eV with an energy resolution of 100 meV. STM experiments were performed at room temperature in home laboratories at Uppsala University and the Norwegian University of Science and Technology (NTNU). In both cases an Omicron Variable Temperature STM system was used and STM images recorded using tungsten tips. In the NTNU system, PES data were recorded using nonmonochromatized Al Kα radiation and a SPECS electron energy analyzer. The PES spectra were used to relate the situations recorded by STM to those attained at the synchrotron. As substrate for the TiOx film growth an Au(111) crystal (Surface Preparation Laboratory) was used. The Au(111) surface was prepared by cycles of Ar-sputtering and annealing until no impurities were detected by HRPES and LEED 6661
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C preparation was used. A Voigt line shape was used for all O 1s peaks. The TiOx film contributions have an asymmetry, which has been modeled by an extra peak at 1 eV higher BE than the main film contribution. The width of this peak was kept constant and the amplitude kept fixed at the same fraction of the main peak, as found from the clean preparation, throughout the water annealing series. Figure 1 shows an example of how
Figure 1. Fitted O 1s spectrum for the Star preparation using a modeled background, and four Voigt line shape peaks, two peaks from the as prepared phase, Star and asymmetry, and two water related peaks, OD and D2O. This spectrum was recorded after the 200 K annealing step.
Figure 2. STM images of the four different TiOx phases grown on Au(111). (a) Domains of the TiO2(B) phase, separated by disordered domain boundaries and holes. The domains themselves have very few defects. (b) Star islands distributed on the Au(111) surface. (c) The honeycomb pattern of the HC phase and (d) the structure of the PW phase.
the O 1s spectra recorded for the Star phase were fitted. The spectrum was recorded after a dose of 5 L D2O at 90 K followed by heating to 200 K. The peaks around 531 eV originates from the TiO2 phase and the peaks at 532 and 534 eV correspond to hydroxyls and molecular water, respectively. For the Ti 2p3/2 spectra a normal Shirley background subtraction has been applied. Voigt lineshapes were used for all peaks except for the peak at 455.5 eV originating from the PW phase where a convolution of an asymmetric Doniach-Sunjic and Gaussian line shape was used. The damping of the Au 4f surface component has been used to determine the total coverage of the preparations, and Ti 2p3/2 peaks have been used to obtain the partial coverages of the mixed phases. The total coverages have an uncertainty of ±0.5% and the partial coverages have an uncertaincy of ±1%. In order to determine the monolayer (ML) coverages of water and hydroxyls both O 1s and valence band spectra have been used. It should be noted that the background at the low binding energy side of the O 1s spectra is rather insensitive to the water coverage except for the larger coverages at temperatures of 160 K and below.
Ti ions in the topmost layer are bonded to oxygen atoms, forming titanyl groups, Vittadini et al. have labeled these terminations, type I and II, respectively.30,31 The STM images suggest that the termination is of type I, in accordance with theoretical predictions that this is the favored termination in vacuum.30,31 Moreover, on the basis of the attenuation of the Au 4f photoemission signal, we estimate that the film needs to be at least three layers thick for the TiO2(B) phase to form.26 Figure 2b shows a STM image recorded of the Star phase similar to the preparation used for the water annealing series. The Star phase consists of TiO2 islands evenly distributed on the Au(111) surface. Images with higher resolution25 are in close resemblance to images obtained for a phase grown on Pt(111) identified as a quasi (2 × 1) rutile TiO2(100) surface featuring an additional (7 × 1) reconstruction.32 However, the STM and LEED results for the Star phase on Au(111) indicate an (8 × 1) reconstruction.25 A possible reconstruction of the rutile TiO2(100) surface has been found to consist of microfacets of the rutile (110) surface.33 Even if the exact atomic structure of the Star phase is not known it appears clear that the surface has a certain corrugation with ridges separated by approximately 25 Å.25 The HC phase is a flat, homogeneous single layer of Ti2O3 with the honeycomb pattern clearly observed by STM, as displayed in Figure 2c. This phase is always accompanied by clusters of oxidized TiO2 islands. The HC islands are large, up to several tens of nanometers wide, and practically no defects are observed within the HC structure. The other wetting layer, the PW phase, shown in Figure 2d, has a more compact structure, which in the STM images indeed resembles a pinwheel pattern with six spokes. Along the spokes and in the hub of the pinwheels there are dislocation sites resulting in
■
RESULTS Clean TiOx Preparations. Representative STM images of the four different thin film TiOx phases grown on Au(111) are displayed in Figure 2. The TiO2(B) phase, seen in Figure 2a, consists of domains which have merged to form a sheet covering the Au(111) surface almost completely. The domains are separated by clear domain boundaries and occasional holes or openings. The interior of the domains exhibit only a few defects. STM and LEED results reveal an atomic structure of the domains consistent with TiO2(B)(001).26 There are two possible terminations for this surface, one that exposes two types of 5-fold coordinated Ti ions and one in which half of the 6662
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C larger variation in the electronic structure.27 Also, this phase forms islands that are up to several tens of nanometers wide. The predominant TiOx phase can be determined with LEED by virtue of having a distinctive diffraction pattern, but to get a more complete picture of each preparation HRPES spectra are imperative. Also, STM data (not shown) reveals coexistence of more than one phase dependent on the preparation parameters. The Ti 2p spectra for the four different preparations used in the water annealing series of the present work are displayed in Figure 3.
regime of Ti3+, which is in accordance with the HC phase having Ti2O3 stoichiometry.25,34 Also, the binding energy of the O 1s peak for this phase is different from the other phases, 529.4 eV compared to 530.5 eV. In addition, there are contributions from peaks A and B in the Ti 2p3/2 spectrum. These are assigned to TiO2 clusters with associated defects. Overall, this preparation covers 37% of the Au(111) surface, of which 79% is HC phase and 21% TiO2 clusters. At the bottom in Figure 3 the Ti 2p spectrum for the PW preparation is shown. This phase is characterized by a peak at BE 455.5 eV, labeled D, in good agreement with the existence of Ti2+ species.26,34 As for the HC preparation, TiO2 clusters are also present, giving rise to contributions from peaks A and B. There is no evidence for the HC phase coexisting with the PW phase, which is expected given that the HC phase transforms to the PW phase upon annealing. The PW preparation covers 25% of the Au(111) surface, of which 92% is the PW phase itself and the remaining 8% is TiO2 clusters. Water Adsorption and Desorption. When 5 L of D2O is deposited at 90 K on the TiOx/Au(111) surfaces, a thick multilayer of water (i.e., ice) is formed, covering the entire surface, including the clean Au(111) areas. By heating the sample above 160 K, the water on the clean Au(111) surface areas desorbs, leaving water only on the TiOx phases. Thus, it is possible to selectively study the water interaction with the TiOx phases at 170 K and above. Water annealing series were carried out for the four different TiOx preparations, and for each temperature step in the series O 1s spectra were recorded. In Figure 4 the results of the annealing series are illustrated by
Figure 3. Ti 2p spectra for the four TiOx thin films, as prepared. The spectra for the HC and PW phases are multiplied by a factor 5 and 10 respectively. Only the Ti 2p3/2 components have been fitted, they are labeled A, B, C, and D.
The two topmost spectra show the Ti 2p spectra of the TiO2(B) and Star preparations, respectively. In both cases, a peak at BE 459 eV dominates. This state, labeled A, corresponds to Ti4+ species in accordance with these phases consisting of TiO2. In addition to peak A, there is also a small contribution at 457.5 eV, labeled B. This peak lies within the BE region, which is reported for Ti3+, 455.6−458.1 eV,34 and has been found to be connected to defects in fully oxidized phases.25 The TiO2(B) preparation covers 94% of the Au(111) surface and the Star preparation covers 67%. The two TiO2 phases are pure in the sense that no trace of HC or PW phases can be found in the Ti 2p spectra. The third spectrum in Figure 3 is the one recorded for the HC preparation. The peak at BE 456.3 eV, labeled C, is characteristic for the HC phase.25 The BE falls within the
Figure 4. O 1s spectra from the water annealing series for all four preparations. The spectra were recorded using grazing emission (GE) for the TiO2(B) phase, while normal emission (NE) was used for the other phases. The background has been subtracted and the spectra have been normalized to the thin film peak height of the 500 K (400 K) annealing step spectra for each preparation.
fitted spectra recorded after annealing to 180, 220, 300, and 500 or 400 K. The modeled background was subtracted before the spectra were normalized in peak height to aid visual comparison. For TiO2(B) preparations, generally consisting of several layers, the spectra were recorded using grazing emission (GE) in order to enhance surface sensitivity and 6663
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C
TiO2(B)(001) surface of type I termination was used. In the case of the Star phase, where the exact atomic structure is unknown, two estimates have been made: one that relates the adsorbate density to the unreconstructed rutile TiO2(100) surface, Star(100), and one related to the rutile TiO2(110) surface, Star(110). The reason for this choice is that the Star phase is assumed to be surface reconstructed rutile TiO2(100), and this reconstruction can occur by formation of rutile TiO2(110) facets.33 The results of the coverage estimates are presented in Figure 5, subdivided into D2O and OD coverages on fully oxidized TiO2 phases and reduced phases, TiOx, x < 2. The D2O and OD coverages on rutile TiO2(110) are taken from Walle et al.36 Summarizing the results for the reference surface, rutile TiO2(110), the first layer comprises about 0.8 ML D2O and 0.4 ML OD. This corresponds to adsorption of 1 ML water as one water dissociation event yields two hydroxyls.37 The molecular water within the first layer desorbs between 210 and 300 K. Desorption of hydroxyls (by recombination) overlaps to a large extent with desorption of molecular species and the last remainder of hydroxyls desorbs between 300 and 400 K. Fully Oxidized Phases. Water interacts weakly with the TiO2(B)(001) surface. At 170 K, the lowest temperature at which we selectively probe water on the TiO2(B) phase, the D2O coverage amounts to 0.5 ML, Figure 5a, filled circles. Assuming that the interaction is strongest with Ti, implies that 50% of the Ti ions are coordinated to D2O molecules at this point. The coverage drops to 0.2 ML at 180 K after which further desorption is slow. The last D2O has desorbed after annealing to 300 K. Thus, there are D2O molecules with similar
hence the water induced contributions. For the other preparations, normal emission (NE) was chosen to maximize the count rate. Molecular water is associated with an O 1s peak at a BE of 534 eV, while the O 1s peak due to hydroxyl contributions appears at a BE of 532 eV. As seen in Figure 4, both water and hydroxyls are present on all four preparations and both species gradually disappear with increasing temperature. In order to make a quantitative comparison between the different phases, the coverages of water and hydroxyls have been determined. The results are compared to a rutile TiO2(110) surface, prepared to obtain a ideal, defect free stoichiometric surface.19,35 On TiO2 surfaces, it is common to define an adsorbate coverage relative to the number of Ti atoms in the surface layer. For example, one ML on rutile TiO2(110) corresponds to the density of 5-fold-coordinated Ti-sites on the surface, which amounts to 5.2 × 1014 cm−2. The Ti densities used for the coverage estimates in the present work are summarized in Table 1. For the TiO 2 (B) phase the value for the Table 1. Density of Surface Ti Sites for Selected TiOx Structures structure a
TiO2(B)(001), type I unreconstructed TiO2(100), Star(100) rutile TiO2 (110), Star(110) HC PW a
Ti density (1014 cm−2) 8.70 7.34 5.20 7.10 10.80
Terminated by two types of 5-coordinated Ti atoms.
Figure 5. (a) Molecular water and (b) hydroxyl coverage as a function of annealing temperature for the fully oxidized phases along with the corresponding results for stoichiometric rutile TiO2(110).36 (c) Molecular water and (d) hydroxyls as a function of annealing temperature for the reduced HC and PW phases. 6664
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C
The origin of the dip in the OD coverage found below 200 K for both the reduced phases and the TiO2(B) phase is uncertain. One possibility is attenuation of the signal from OD species due to formation of 3D D2O−OD complexes; another possibility is a heating-induced dissociation between 180 and 200 K. Water-Induced Effects on the TiOx Films. The effect of a water adsorption/desorption cycle on the four different TiOx preparations has also been investigated. No changes were observed in any of the LEED patterns after the water annealing series, indicating that the overall structures remain unaltered. However, monitoring the oxidation state of Ti provides evidence for changes occurring during water exposure. This is shown for the HC preparation in Figure 6. Before the water
bonding strength as on rutile (110) but these represent minority species. A small degree of water dissociation is observed for the TiO2(B) phase. The maximum OD coverage is 0.1 ML implying that 0.05 ML D2O dissociates on the surface of the TiO2(B) phase, Figure 5b. This in turn suggests that only about 5% of the Ti sites are active toward dissociation. The hydroxyls that form on the TiO2(B) preparation are however comparatively stable; 40% of the maximum amount remains at 400 K and an annealing to 500 K is required to completely remove the hydroxyls. On the Star surface, the water coverage at 170 K is estimated to 0.75 ± 0.15 ML, with the uncertainties given by the different Ti densities considered (see Figure 5a filled and open diamonds). The coverage decreases continuously with increasing temperature and the last D2O species desorb at 280−300 K. At every temperature, the D2O coverage is significantly lower than the coverage measured on the rutile (110) surface. The OD coverage for the Star phase is 0.47 ± 0.08 ML at 170 K, Figure 5b. Up to 200 K only a small decrease in the OD coverage is observed, after which it decreases progressively with increasing annealing temperature. The last hydroxyls desorb between 350 and 400 K. The average OD coverage on the Star phase is observed to be comparable to the amount of hydroxyls present on the rutile (110) surface up to 250 K. However, above 250 K the OD coverage on the Star islands is higher. Reduced Phases. The HC preparation covers 37% of the Au(111) surface and consists of a combination of areas with pure HC phase (79% relative amount) and TiO2 clusters (21% relative amount). The O 1s spectra shown in Figure 4 include the contributions from both phases. Since the Star preparation represents a pure phase with TiO2 islands only, those results can be used to estimate the contribution from clusters for the HC preparation. Consequently, the D2O and OD coverages shown in Figure 5c,d are those obtained after subtraction of the estimated contribution from TiO2 clusters covering 8% of the Au(111) surface (21% of 37% coverage) and should therefore represent an estimate of the coverages solely associated with the HC phase. At 170 K, it is clear that a water multilayer is still stable on the HC structure. Increasing the temperature to 180 K leaves about 0.9 ML D2O. A continuous decrease in coverage is noted for higher temperatures, and all D2O is desorbed by 250 K. Only a small amount of hydroxyls are observed on the HC phase, a maximum coverage of about 0.15 ML is observed at 200 K. Half of the OD species has desorbed at 250 K and the last OD desorbs between 300 and 400 K. Also, the investigated PW thin film represents a mixed situation but to a lesser degree; the relative amount of pure PW phase is 92% at 25% surface coverage. Similar to the HC case the amount of water and hydroxyls on the PW phase has therefore been obtained after first subtracting the contribution associated with TiO2 clusters covering 2% of the Au(111) surface. The fractions of water and hydroxyls associated with the PW structure are displayed in Figure 5c,d. The behavior of D2O is very similar to that found for the HC phase; multilayer formation occurs ≤170 K, and about 0.8 ML D2O is observed upon heating to 180 K. There are a limited number of data points in this series, but the last D2O is likely to desorb around 300 K, as there is only a very small amount left at this point. The maximum amount of hydroxyls is just below 0.4 ML, observed at 180 K. This is considerably higher coverage than observed for the HC phase. About 0.1 ML OD remains at 300 K, and all hydroxyl groups are gone after heating to 500 K.
Figure 6. Ti 2p3/2 spectra of the HC phase as prepared, after water adsorption and subsequent annealing to 160 K and after water desorption at 400 K.
treatment, the peak due to the HC phase, peak C, is clearly dominating the Ti 2p3/2 spectrum (shown in green). After water adsorption and annealing to 160 K, this contribution is reduced and has a spectral weight, which is comparable to peak A, which in turn is only slightly damped. After annealing to 400 K, all D2O and OD species have desorbed but the spectrum differs from that of the pristine film in that intensity has been redistributed toward the high BE side. That is, a certain degree of oxidation has taken place. However, curve fitting indicates that the intensity of peak C is essentially unchanged compared to that of the pristine film and that the effect is rather described by a shift of intensity from peak B to peak A. This means that the HC phase itself is not affected by water, while a change in the properties of the coexisting clusters appear as a decrease in the contribution from defects. The intensity changes of the Ti 2p 3/2 peak components for all four preparations are summarized in Table 2. The results for the two fully oxidized phases, TiO2(B) and Star, show a similar tendency for loss in the intensity from defect states but the changes are less accurately determined due to the lower relative intensities. For the PW phase, however, oxidation upon water interaction is evident through pronounced reduction in the intensity of peak D. It is finally important to note that no reduction of the oxide films were observed, neither prior to water adsorption nor during the water adsorption−desorption sequence. This demonstrates that the synchrotron light in this case does not lead to the formation of oxygen vacancies, something that has 6665
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C
500 K. One possible explanation for this difference lies in the morphology of the preparations. On Pt(111) TiO2(B) grows on top of a reduced TiOx wetting layer and exhibits a tendency to form 3D islands. Due to the underlying TiOx wetting layer, the Ti3+ contribution seen in the Ti 2p PES spectrum is much larger (20−25%) than for the TiO2(B) phase on Au(111). Judging from the STM images,39 about 50% of the surface is covered by TiO2(B)(001) islands after deposition of 4 ML (equiv). On Au(111) about 90% of the surface is covered already after 3 ML (equiv). Consequently, TiO2(B) on Au(111) wets the surface more efficiently, leading to merging of islands and formation of a predominantly continuous film with domain boundaries. Still, the domain sizes are comparable in the two cases. It is likely that the behavior of water at domain boundaries is different from the behavior at the edges of islands, providing an explanation for the observed difference in desorption temperature. The TiO2 structure associated with the Star preparation has previously been proposed to be similar to the rutile (100) surface, albeit with a surface reconstruction, (7 × 1) when formed on Pt(111)32 and (8 × 1) when formed on Au(111).25 On the Star preparation on Au(111), Figure 5a,b shows that all D2O has desorbed at 300 K and that only water due to recombination of OD desorb between 300 and 400 K. No desorption is found above 400 K. These results are in agreement with TPD and PES studies for TiO2(100)15,40 where it was argued that only molecularly adsorbed water desorbs up to 290 K, while desorption above 290 K is associated with recombinative desorption.15 However, for the Star phase, we also observe a significant loss of OD already between 200 and 300 K, where the OD coverage decreases from about 0.44 to 0.19 ML, Figure 5b. Thus, there is a temperature range with an overlap between recombinative desorption of hydroxyls and desorption of molecularly adsorbed D2O. This behavior is similar to that found for H2O deposited on rutile TiO2(110).19 From Figure 5a,b, the D2O−OD speciation for the Star preparation can be derived. If the Ti surface density is assumed to be close to that of the rutile TiO2(110) surface, labeled Star(110) in Table 1, and we use that one water molecule creates two hydroxyls upon dissociation, it follows that the surface coverage is 1.125 ML at 170 K with 24% of the water molecules being dissociated. At 180 K, the coverage is 0.88 ML, of which 31% is dissociated. From this follows that the ML point is located between 170 and 180 K at which 27% is dissociated. Choosing the surface density of the unreconstructed TiO2(100) surface, Star(100), gives a ML point below 170 K and a slightly lower amount of hydroxyls formed. In comparison with the results reported for the rutile TiO2(100)(1 × 1) surface, the temperature of the ML point is similar, while the degree of dissociation is lower. In the TPD study by Henderson et al.,15 uptake is limited to 1 ML at 170 K, and it was estimated that about 50% of the water in the first layer had dissociated. On rutile TiO2(110), the ML point is located at 210 K at which 20% dissociation is found.19 This degree of dissociation is only slightly lower than that found for the Star phase. From this it can be concluded that the behavior of water on the Star phase can be well described in terms of a superposition of the behaviors found on the stoichiometric rutile TiO2(100) and (110) surfaces. Therefore, the results are consistent with the previously proposed concept of an (nx1) surface reconstruction of rutile (100) into (110) facets,33 in this case a (8 × 1) structure.25 Similarly, TiO2(100) structures on
Table 2. Changes in Peak Intensities of the Ti 2p3/2 Contributions after Water Adsorption and Subsequent Heating to 500 K (400 K) Relative to the Total Intensity of the As Prepared Film Phasesa
a
preparation
peak A (%)
peak B (%)
peak C/D (%)
HC PW Star TiO2(B)
12 12 5 3
−9 4 −4 −3
−3 −16 −1 −0
The HC and Star phases are annealed to 400 K.
been observed to happen upon much more extensive photon exposures.38
■
DISCUSSION Fully Oxidized Phases. On the TiO2(B) phase a low amount of hydroxyls (200 K can be ascribed to defects it follows that the interaction of water with the TiO2(B)(001) surface itself is very weak, featuring only molecular adsorption. This conclusion is in good agreement with theoretical predictions, where the water interaction with the type I termination of TiO2(B)(001) was found to be weak, with no tendency for dissociation.30 Water adsorption on TiO2(B)(001) grown on Pt(111) has recently been found to yield a small amount of hydroxyls.39 No surface Ti3+ species were assigned to the TiO2(B) structures, which would exclude Ti3+ sites as active sites for the dissociation.15 Instead, edge sites are suggested to be active in dissociating water.39 On TiO2(B)/Pt(111) the hydroxyls were found to be present on the surface up to 400 K, while complete desorption on TiO2(B)/Au(111) requires heating to 6666
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C Pt(111)39 have been suggested to undergo (7 × 1) surface reconstruction, and a dissociation degree of 30% was reported. The finding of a water dissociation degree below 50% would then be a signature of a reconstructed rutile (100) surface. The effect of defects should also be considered. The Ti3+ contribution seen in Figure 3 for this Star preparation has a relative intensity of about 15%. The location of the defects is unknown, but the many small Star phase islands provide a large amount of rim between Au(111) and TiOx, which represents possible dissociation sites. From Table 2 it is clear that some of the Ti3+ defects are oxidized upon water exposure, but this is a very small effect compared to the observed amount of hydroxyls for the Star preparation. However, it stands clear that further studies are required to clarify the correlation between the structure and defectivity of the Star phase islands. Reduced Phases. For the HC preparation a maximum OD coverage of 0.15 ML is found. Assuming adsorption on Ti ions only, the OD coverage indicates that 8% of the Ti sites are active for dissociation. The HC phase forms a (2 × 2) superstructure on the Au(111) surface. If the dissociation was an inherent property of the HC phase itself the observed dissociation degree corresponds to one dissociation event per six (2 × 2) unit cells. Although such a OD−D2O structure cannot be completely ruled out, it appears more likely that this small amount of hydroxyls is rather due to defects or uncertainty in the subtraction procedure undertaken to eliminate the contribution from TiO2 clusters. Peak B is much larger compared to peak A for the HC preparation than for the Star preparation, see Figure 3, indicating that there are more defects associated with the clusters in the HC preparation. This may be caused by the lower deposition temperature of 300 °C used for the HC preparation compared to the Star preparation, formed at 500 °C. These extra defects may account for the remaining hydroxyls. Other possible sites for hydroxyl formation on this structure are sites along the boundary between the HC phase and the TiO2 clusters. The changes in the Ti 2p spectra, listed in Table 2, indicate that it is the defects represented by peak B that interacts with the water species and are therefore likely to cause the observed hydroxyls. This supports the notion of the dissociation being related to TiO2 clusters rather than the HC phase. Further support for this conjecture is the finding that no hydroxyls are left on the surface after annealing to 400 K, which makes the thermal stability of the hydroxyls observed for the HC preparation comparable to that of the Star phase, but different from both the TiO2(B) phase and the bridging oxygen vacancies on rutile (110). This leads to the conclusion that the HC phase itself does not dissociate water, and the observed hydroxyls are attributed to the high defect contribution in the coexisting Star clusters of this preparation. The water ML point for the PW phase is observed at about 180 K, with the composition 0.8 ML D2O and 0.4 ML OD. This amount of hydroxyls corresponds to 20% of the Ti sites being able to dissociate water. Since the PW preparation is almost pure (92%, with the small amount of clusters corrected for) and very few defects observed within the PW domains it is feasible that the partial dissociation is an inherent property of the PW structure. The STM images reveal a clear contrast between the Ti ions within the structure. Judging from the STM image shown in Figure 2d the spokes and hubs of the PW phase make up 20−30% of the phase, in reasonable correspondence with the amount of hydroxyls observed. This finding indicates that these bright features may serve as active
sites for water dissociation on the PW phase. For this phase, some oxidation occurs upon water adsorption and desorption. The decrease in the Ti 2p peak intensity is about 16%. This number compares well to the number of active Ti sites for dissociation, which is 20% based on the formation of 0.4 ML OD. It is therefore very likely that the dissociation process is related to the oxidation process. However, the particular nature of the oxidation process is not fully understood, one possibility may be abstraction of H2.
■
CONCLUSIONS We have presented the first comprehensive study of water interaction with TiOx nanostructures on Au(111). Four different TiOx structures on Au(111) have been addressed: TiO2(B)(001), TiO2 islands (presumably surface reconstructed rutile TiO2(100)), a honeycomb (HC) Ti2O3 layer and a TiOx (x < 1.5) pinwheel (PW) structure. The presence of adsorbed water and hydroxyls formed by water dissociation has been established by the use of HRPES of the O 1s level. The amount of water and hydroxyls has been related to densities of Ti sites obtained from STM and LEED results. Coverages expressed in this way facilitate the comparison to results obtained for single crystal surfaces, providing a measure of the activity toward water adsorption and dissociation. The aim has been to disentangle the inherent properties of the ordered phases from effects due to defects. The TiO2(B)(001) surface is found to be inactive toward water dissociation; the small amounts of hydroxyls formed is consistent with dissociation occurring at defect sites at TiO2(B)(001) domain boundaries. The TiO2 islands, or Star phase, display a considerable ability to dissociate water. The thermal stability of the hydroxyls can be described in terms of a superposition of TiO2(110) and TiO2(100) surfaces, supporting the notion of having rutile TiO2(100) islands reconstructing into TiO2(110) facets. However, the TiO2 islands are also likely to feature defect induced effects on the water adsorption. No dissociation could be observed for the honeycomb layer while the pinwheel structure is proposed to feature dissociation inherent to its atomic structure. The different behavior of the four different TiOx thin film phases on Au(111) toward water dissociation and thermal stability of D2O and OD species makes these systems highly interesting for model reactions involving water. For example, in the water−gas shift reaction water dissociation is often considered to be the rate-determining step.41 It is interesting in this respect to note that the TiO2 islands, which is the thermodynamically favored phase, displays the highest tendency for dissociation. This is consistent with the high water−gas shift activity found for TiO2 particles on Au, a bifunctional catalyst where TiO2 dissociate water, while Au oxidizes CO. The results here also demonstrate the possibility to experiment with different dissociation mechanisms, oxidation states, and perimeters in order to further the understanding of this reaction.
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +47 41662984. Fax: +47 73591410. E-mail: anne.
[email protected]. Notes
The authors declare no competing financial interest. 6667
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C
■
rutile TiO2(110) surface without oxygen vacancies. Phys. Rev. B 2009, 80, 235436. (20) Sebbari, K.; Domain, C.; Roques, J.; Perron, H.; Simoni, E.; Catalette, H. Investigation of hydrogen bonds and temperature effects on the water monolayer adsorption on rutile TiO2(110) by firstprinciples molecular dynamics simulations. Surf. Sci. 2011, 605, 1275− 1280. (21) Duncan, D. A.; Allegretti, F.; Woodruff, D. P. Water does partially dissociate on the perfect TiO2(110) surface: a quantitative structure determination. Phys. Rev. B 2012, 86, 045411. (22) Patel, M.; Mallia, G.; Liborio, L.; Harrison, N. M. Water adsorption on rutile TiO2(110) for applications in solar hydrogen production: a systematic hybrid-exchange density functional study. Phys. Rev. B 2012, 86, 045302. (23) Amft, M.; Walle, L. E.; Ragazzon, D.; Borg, A.; Uvdal, P.; Skorodumova, N. V.; Sandell, A. A molecular mechanism for the water-hydroxyl balance during wetting of TiO2. J. Phys. Chem. C 2013, 117, 17078−17083. (24) Schaefer, A.; Ragazzon, D.; Walle, L. E.; Farstad, M. H.; Wichmann, A.; Bäumer, M.; Borg, A.; Sandell, A. Controlled modification of nanoporous gold: chemical vapor deposition of TiO2 in ultrahigh vacuum. Appl. Surf. Sci. 2013, 282, 439−443. (25) Ragazzon, D.; Schaefer, A.; Farstad, M.; Walle, L.; Palmgren, P.; Borg, A.; Uvdal, P.; Sandell, A. Chemical vapor deposition of ordered TiOx nanostructures on Au(111). Surf. Sci. 2013, 617, 211−217. (26) Ragazzon, D.; Farstad, M. H.; Schaefer, A.; Walle, L. E.; Uvdal, P.; Borg, A.; Sandell, A. Growth of TiO2(B)(001) on Au(111) by chemical vapor deposition. Surf. Sci. 2015, 633, 102−108. (27) Ragazzon, D.; Farstad, M. H.; Schaefer, A.; Walle, L. E.; Uvdal, P.; Palmgren, P.; Borg, A.; Sandell, A. Structure of a pinwheel-like TiOx single layer phase on Au(111). Manuscript in preparation, 2015. (28) Nyholm, R.; Svensson, S.; Nordgren, J.; Flodström, A. A soft Xray monochromator for the MAX synchrotron radiation facility. Nucl. Instrum. Methods A 1986, 246, 267−271. (29) Walle, L. E.; Ragazzon, D.; Borg, A.; Uvdal, P.; Sandell, A. Photoemission studies of water dissociation on rutile TiO2(110): aspects on experimental procedures and the influence of steps. Appl. Surf. Sci. 2014, 303, 245−249. (30) Vittadini, A.; Casarin, M.; Selloni, A. Structure and stability of TiO2-B surfaces: a density functional study. J. Phys. Chem. C 2009, 113, 18973−18977. (31) Vittadini, A.; Sedona, F.; Agnoli, S.; Artiglia, L.; Casarin, M.; Rizzi, G. A.; Sambi, M.; Granozzi, G. Stability of TiO2 polymorphs: exploring the extreme frontier of the nanoscale. ChemPhysChem 2010, 11, 1550−7. (32) Cavaliere, E.; Artiglia, L.; Rizzi, G. A.; Gavioli, L.; Granozzi, G. Structure and thermal stability of fully oxidized TiO2/Pt(111) polymorphs. Surf. Sci. 2013, 608, 173−179. (33) Zschack, P.; Cohen, J. B.; Chung, Y. W. Structure of the TiO2(100) 1 × 3 surface determined by glancing angle X-ray diffraction and low energy electron diffraction. Surf. Sci. 1992, 262, 395−408. (34) Sedona, F.; Rizzi, G. A.; Agnoli, S.; Xamena, F. X. L. i.; Papageorgiou, A.; Osterman, D.; Sambi, M.; Finetti, P.; Schierbaum, K.; Granozzi, G. Ultrathin TiOx films on Pt(111): a LEED, XPS, and STM investigation. J. Phys. Chem. B 2005, 109, 24411−24426. (35) Wendt, S.; Sprunger, P. T.; Lira, E.; Madsen, G. K. H.; Li, Z.; Hansen, J. O.; Matthiesen, J.; Blekinge-Rasmussen, A.; Laegsgaard, E.; Hammer, B.; et al. The role of interstitial sites in the Ti3d defect state in the band gap of titania. Science 2008, 320, 1755−1759. (36) Walle, L. E.; Ragazzon, D.; Borg, A.; Uvdal, P.; Sandell, A. Competing water dissociation channels on rutile TiO2(110). Surf. Sci. 2014, 621, 77−81. (37) Lindan, P. J. D.; Harrison, N. M. Mixed dissociative and molecular adsorption of water on the rutile (110) surface. Phys. Rev. Lett. 1998, 80, 762−765. (38) Andersson, K.; Nikitin, A.; Pettersson, L. G. M.; Nilsson, A.; Ogasawara, H. Water dissociation on Ru(001): an activated process. Phys. Rev. Lett. 2004, 93, 196101.
ACKNOWLEDGMENTS Financial support was received from the Swedish Energy Agency (STEM), the Swedish Science Council (VR), the Trygger foundation, the Crafoord foundation, NordForsk, and the Göran Gustafsson foundation. D.R. has been supported by the Anna Maria Lundins scholarship from Smålands Nation in Uppsala and L.E.W. has been supported through the Strategic Area Materials at NTNU. We thank the staff at MAX-lab for their support.
■
REFERENCES
(1) Diebold, U. Structure and properties of TiO2 surfaces: a brief review. Appl. Phys. A: Mater. Sci. Process. 2003, 76, 681−687. (2) Konstantyinou, I. K.; Albanis, T. A. TiO2-assisted photocatalytic degradation of azo dyes in aqueous solution: kinetic and mechanistic investigations. Appl. Catal., B 2004, 49, 1−14. (3) Habibi, M. H.; Vosooghian, H. Photocatalytic degradation of some organic sulfides as environmental pollutants using titanium dioxide suspension. J. Photochem. Photobiol., A 2005, 174, 45−52. (4) Maldonado, A. Titanium dioxide thin films deposited by the solgel technique starting from titanium oxy-acetylacetonate: gas sensing and photocatalyst applications. Phys. Status Solidi C 2010, 7, 2316− 2320. (5) Valden, M.; Pak, S.; Lai, X.; Goodman, D. W. Structure sensitivity of CO oxidation over model Au/TiO2 catalysts. Catal. Lett. 1998, 56, 7−10. (6) Jovic, V.; Chen, W. T.; Sun-Waterhouse, D. Effect of gold loading and TiO 2 support composition on the activity of Au/TiO 2 photocatalysts for H2 production from ethanol-water mixtures. J. Catal. 2013, 305, 307−317. (7) Maness, P. C.; Smolinski, S. Bactericidal activity of photocatalytic TiO2 reaction: toward an understanding of its killing mechanism. Appl. Environ. Microbiol. 1999, 65, 4094−4098. (8) Blake, D. M.; Maness, P. C. Application of the photocatalytic chemistry of titanium dioxide to disinfection and the killing of cancer cells. Sep. Purif. Rev. 1999, 28, 1−50. (9) McCullagh, C.; Robertson, J. M. C. The application of TiO2 photocatalysis for disinfection of water contaminated with pathogenic micro-organisms: a review. Res. Chem. Intermed. 2007, 33, 359−375. (10) Gao, F.; Wood, T. E.; Goodman, D. W. The effects of water on CO oxidation over TiO2 supported Au catalysts. Catal. Lett. 2010, 134, 9−12. (11) Rodríguez, J. A.; Ma, S.; Liu, P.; Evans, J.; Perez, M.; Hrbek, J. Activity of CeOx and TiOx nanoparticles grown on Au(111) in the water-gas shift reaction. Science 2007, 318, 1757−1760. (12) Rodríguez, J. A.; Hrbek, J. Inverse oxide/metal catalysts: a versatile approach for activity tests and mechanistic studies. Surf. Sci. 2010, 604, 241−244. (13) Henderson, M. A. A surface science perspective on TiO2 photocatalysis. Surf. Sci. Rep. 2011, 66, 185−297. (14) Farnesi Camellone, M.; Marx, D. On the impact of solvation on a Au/TiO2 nanocatalyst in contact with water. J. Phys. Chem. Lett. 2013, 4, 514−518. (15) Henderson, M. Structural sensitivity in the dissociation of water on TiO2 single-crystal surfaces. Langmuir 1996, 12, 5093−5098. (16) Pang, C. L.; Lindsay, R.; Thornton, G. Chemical reactions on rutile TiO2(110). Chem. Soc. Rev. 2008, 37, 2328−2353. (17) Kristoffersen, H. H.; Hansen, J. Ø.; Martinez, U.; Wei, Y. Y.; Matthiesen, J.; Streber, R.; Bechstein, R.; Laegsgaard, E.; Besenbacher, F.; Hammer, B.; et al. Role of steps in the dissociative adsorption of water on rutile TiO2(110). Phys. Rev. Lett. 2013, 110, 146101. (18) Oviedo, J.; Sánchez-de Armas, R.; San Miguel, M. A.; Sanz, J. F. Methanol and water dissociation on TiO2(110): the role of surface oxygen. J. Phys. Chem. C 2008, 112, 17737−17740. (19) Walle, L. E.; Borg, A.; Uvdal, P.; Sandell, A. Experimental evidence for mixed dissociative and molecular adsorption of water on a 6668
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
Article
The Journal of Physical Chemistry C (39) Artiglia, L.; Zana, A.; Rizzi, G. A.; Agnoli, S.; Bondino, F.; Magnano, E.; Cavaliere, E.; Gavioli, L.; Granozzi, G. Water adsorption on different TiO2 polymorphs grown as ultrathin films on Pt(111). J. Phys. Chem. C 2012, 116, 12532−12540. (40) Muryn, C. A.; Hardman, P. J.; Crouch, J. J. Step and point defect effects on TiO2(100) reactivity. Surf. Sci. 1991, 251−252, 747−752. (41) Liu, P.; Rodríguez, J. A. Water-gas-shift reaction on metal nanoparticles and surfaces. J. Chem. Phys. 2007, 126, 164705.
6669
DOI: 10.1021/acs.jpcc.5b00561 J. Phys. Chem. C 2015, 119, 6660−6669
