Reaction and Diffusion Paths of Water and Hydrogen on Rh Covered Black Titania
- 101 Downloads
The reactions of H2O, H2 D2 and CO with clean and rhodium covered black titania have been investigated by TDS, AES and sensitive temperature programmed work function (TP-WF) measurements to elucidate the complex interactions with this narrow bandgap material promising for visible light energy harvesting. Water formed molecular and dissociative adsorption states with positive outward dipole moments on the reduced, r–TiO2 (110). Surface hydroxyl groups decomposed to H2 and recombined to H2O in a broad temperature range, characterized by TDS peaks at 300, 355–377 and 470 K, which has been associated with surface inhomogeneity. On a strongly reduced, sr–TiO2 (110), a part of H atoms arising from OHa species dissolved in the titania at 200–500 K, and desorbed as H2O with Tp = 570, 670 and 750 K. Sub-monolayer TiOx films produced by stepwise heating on r–TiO2 (110) supported Rh particles suppressed the adsorption of hydrogen, but allowed its spillover to the support. Co-adsorption experiments with D2, H2, H2O and CO on the Rh covered r–TiO2 (110) were also performed. Saturating the Rh by CO at 330 K blocked the uptake of hydrogen on the metal, eliminating its spillover to the support. At 270 K saturation CO exposure removed the pre-adsorbed hydrogen, while at 200 K replaced a part of it decreasing the adsorption bond energy of the rest remained adsorbed. Co-adsorption data proved that the hydrogen desorption states with Tp = 470 and 570 K belong to the decomposition of hydroxyl groups on the r–TiO2 (110) support and at the Rh–TiOx interface, respectively. It is remarkable that the latter state can evolve in the presence of adsorbed CO, which exhibits reactivity towards the same interface.
KeywordsBlack titania Hydrogen dissolution Hydrogen spillover Rh–TiOx interface OH decomposition CO coadsorption
Titania has outmost practical importance in solar energy harvesting, catalysis, gas sensing, self-cleaning surfaces, air/water purification devices, corrosion protection, etc. which promoted detailed single crystal studies of the associated elementary processes . In its photochemical applications it is to be considered that pure titania has a bandgap of over 3 eV, which means that photo-excitation needs UV photons. To exploit the photon energies of visible light, tremendous efforts have been made to narrow the bandgap or produce sub-bandgap states in titania. The bandgap tailoring can be accomplished by means of metallic and non-metallic modifiers, through structural modification, and so on. It was discovered recently, that the highly reduced, disordered, so-called black titania nanomaterials show enhanced reactivity in photocatalytic reactions due to their narrow bandgap allowing enhanced solar light absorption [2, 3, 4]. The reduction can be performed for example with reactive metals and hydrogen . Considering that many catalytic reactions on titania proceeds in the presence of water and hydrogen, we address the interaction of these compounds with the pure and rhodium covered oxide.
Although the spillover of different species from the active part of a catalyst to the support and the inverse process have been studying for long, there still exist some controversies regarding the spillover of hydrogen. The hydrogen spillover is of great practical importance, studied widely experimentally and theoretically. Beside a catalytic relevance, this phenomenon is also of interest in making gas-sensors and hydrogen storage materials. It has been established that this process can take place on reducible oxide supports, but is hindered on non-reducible ones . Hydrogen spilt-over to the support can take part in a broad range of reactions, for example it can remarkably improve the photocatalytic activity of titania for water-splitting . The extremely high surface/volume ratio of titania nanotubes and nanorods makes them more efficient sensor material as compared to macroscopic titania. Titania nanotubes in contact with Pd particles sense selectively H2 at 300 K . The storage of hydrogen as a fuel on a solid adsorbent is a promising method for environmentally friendly transportation. One way to accomplish the onboard hydrogen storage is to make molecular hydrogen dissociate on finely dispersed metal particles, from where H atoms can spill over the high area support, which is preferably carbon. Application of high area titania can also be advantageous in hydrogen storage. The impregnation of carbon nanotubes (CNTs) with titania nanotubes was found to enhance the hydrogen uptake of CNT-s considerably , enhancing their hydrogen storage capacity by a factor of five, reaching 0.4 wt% at 298 K and 18 atm.
Beside the reactions of hydrogen atoms with titania, that of water are also of considerable interest associated with water–gas shift reaction (WGS) promoted by metals over titania  and with hydrogen production through photocatalytic water splitting . The reactions of TiO2 (110) surface with water are well-studied by many experimental and theoretical facilities. However, controversial issues still remained, for example regarding the formation of OH species on a structurally well-defined, stoichiometric TiO2 (110) surface , and detecting the dissolution of hydrogen atoms into the bulk on strongly hydroxylated, stoichiometric TiO2 (110) surfaces [10, 11].
It is also of practical importance in catalysis that not only the species adsorbed on the metal particles can migrate to the titania, but titania can also diffuse onto the active metal. This phenomenon has different terms in the literature, like encapsulation, decoration and in heterogeneous catalysis, strong metal support interaction (SMSI). The catalytic activity of metal oxide nanoparticles on metal surfaces attracted longstanding interest, addressed also by model studies involving metal single crystals [9, 12]. Importantly, TiOx overlayers on metals can exert a substantial catalytic promotional effect, representing a system termed as inverse catalyst. Rh particles covered partially by atomically thin TiOx [13, 14], and MoOx  layers have been investigated as active model catalysts in our laboratory under UHV conditions, which studies are extended here to the reactions of hydrogen, water and CO on clean titania, on titania supported rhodium clusters and on TiOx layer covered Rh particles. Taking into account the potentially high photocatalytic activity of hydrogenated titanium suboxides [2, 3, 4], we focus on the properties of strongly reduced, dark gray/black colored TiO2 (110) samples.
The experiments were performed in an ultrahigh vacuum (UHV) chamber (base pressure < 5 × 10−8 Pa) equipped with facilities for Auger electron spectroscopy (AES), thermal desorption spectroscopy (TDS) and temperature programmed work function (TP-WF) measurements. AES data were collected with a Physical Electronics coaxial-gun single pass cylindrical mirror analyzer (CMA), while mass spectrometric and TDS measurements were performed by a Balzers QMS 200 quadrupole mass spectrometer. Smooth polynomials were subtracted from TDS spectra as baseline correction, and care was taken to avoid the generation of any spurious TDS feature by this procedure. AES spectra were taken in the differential mode with 3 keV primary electron energy, 3 V modulation, and 1–2 µA beam current. AES data were evaluated either plotting the absolute peak-to-peak heights of main peaks (Mo: 186, Rh: 302, Ti: 387, O: 503, C: 272 eV) or Auger ratios calculated from these peaks. WF data were recorded as a function of temperature  by means of a Besocke type Kelvin-probe with a sensitivity of 1 meV, applying the same heating rates as for TDS measurements.
The TiO2 (110) single-crystal was the product of PI-KEM. The sample was attached to a Ta plate with an oxide glue (Aremco, Ceramabond 571), and could be heated with a W filament placed behind the Ta plate. The sample temperature was measured by a chromel–alumel thermocouple, attached to the side of the sample with the same adhesive material. To avoid the fracture of titania single crystal, the heating and cooling rates were always less than 2 K/s, regulated by a computer-controlled circuit. The duration of annealing in stepwise heating experiments was equal to or less than 1 min.
The typical cleaning procedure, applied several hundred times for the same titania sample, consisted of Ar+-ion bombardment (1.5 keV, 1 × 10−5 A cm−2, 300 K, 30 min) and annealing at 1000 K for 30 min. The absence of oxygen-treatments resulted in a dark gray/black, defective crystal. It was characterized by an O/Ti AES ratio of 0.9 and in a later stage by 0.8, being much lower than the 1.5 ratio for a nearly stoichiometric TiO2 (110) plane , mentioned here as s–TiO2 (110). A comparison with the O/Ti ratio of 0.75, characteristic of TiOx (x ~ 1.2) encapsulation layer on a 20 monolayer (ML) thick Rh film  verifies the strongly reduced state of the black titania surface. Note that the extended ion sputtering/annealing cycles resulted samples reduced not only on their surfaces, but in their bulk as well. The samples mentioned throughout this article as reduced titania (0.90 O/Ti AES ratio), and strongly reduced titania (0.80 O/Ti AES ratio), are designated as r–TiO2 (110) and sr–TiO2 (110), respectively, while atomically thin titania overlayers  as TiOx. The dark gray/black color of r–TiO2 (110) and sr–TiO2 (110) crystals is indicative of strong absorption in the visible spectral region. The preparation of samples was highly reproducible as verified by the good reproducibility of TDS and TP-WF data taken in adsorption–desorption experiments.
The above treatment ensured appropriate electrical conductivity for electron spectroscopy and high enough defect-density to observe 2D-like growth of Rh-particles in extended coverage range, which allowed the estimation of surface coverage through AES measurements . 1 ML Rh coverage corresponds to a Rh surface concentration, where, supposing a 2D film growth on a TiO2 (110) single crystal, the Rh Low Energy Ion Scattering (LEIS) signal would disappear . Since on an Ar+-ion sputtered TiO2 (110) surface 2D Rh particles form at low and 3D clusters at 1 ML Rh coverage , on the less reduced r–TiO2 (110) and sr–TiO2 (110) surfaces 3D particles are present at 1 ML Rh coverage. An EGN4 e-beam evaporator of Oxford Applied Research was used for the deposition of Rh (Goodfellow, purity 99.9%) by physical vapor deposition (PVD) at a sample temperature of 180–200 and 300–330 K. Distilled water was purified by freezing and pumping cycles. Hydrogen, deuterium and CO gases were the products of Linde, with purities of 5.0, 3.0 and 3.7, respectively.
3 Results and Discussion
3.1 Reaction of Water with Reduced TiO2 (110) Surfaces
The reaction of water with reduced titania surfaces was investigated by TDS and WF measurements. In Fig. 1a TDS traces at m/z = 18 taken after H2O exposures to the r–TiO2 (110) at 200 K are depicted. After the smallest H2O uptake, originating from the background gas, a broad, asymmetric H2O desorption state appears at Tp = 377 K with a shoulder at 470 K. Enhancing the water exposure, new wide features appear at ~ 300 and 243–227 K. The broadening of desorption states at high H2O exposures is due to water desorption from the sample holder, the effect of which is eliminated in TP-WF measurements. According to previous TDS data , on an ideal, defect-free TiO2 (110) surface the water adsorbs molecularly, while the presence of structural defects greatly enhances its dissociation. It has been verified with atomic scale resolution that the oxygen vacancies reacted with H2O give surface hydroxyl groups , which recombine to water on annealing to 500 K, reproducing the oxygen-deficient surface . The desorption states distinguishable by means of Gaussian fitting with Tp = 227–243, ~ 300, 377 and 470 K in Fig. 1a resemble of former findings for a slightly reduced TiO2 (110) plane , where the states below 300 K were attributed to molecular, and those above 300 K to recombinative H2O desorption. Accordingly, the peaks at 377 and 470 K can be associated with the recombination of OH groups, while those at 227–243 K with molecular desorption. A water desorption state on a stoichiometric TiO2 (110) with Tp = 265 K was attributed to the bonding of intact H2O molecules to fivefold coordinated Ti ions . Similarly to the appearance of TDS states with Tp = 243 and 377 K in Fig. 1a, on a reduced TiO2 (110) surface produced by Ar+-ion sputtering and annealing to 700 K H2O desorption states attributed to molecular and recombinative desorption were detected , although at somewhat lower temperatures, with Tp = 200 and 330 K, respectively. According to XPS data , a part of 300 K water desorption state may belong to intact H2O molecules attached preferentially to the OH moieties. The TDS curves of Fig. 1a are similar to those presented for a strongly reduced titania film , on which isotope scrambling experiment revealed that the water desorption state at Tp ~ 300 K stemmed from the recombination of surface hydroxyl groups. Corresponding to the above findings, the water desorption state at 300 K in Fig. 1a probably belongs partly to molecular and partly to recombinative desorption.
The above statements are in harmony with the observations for r–TiO2 (110) that at low water coverages, where the sample is only slightly oxidized on annealing, a H2/H2O ratio of ~ 3 was determined, while for the water saturated surface which is more oxidized during TDS run a value of ~ 1.5 was found (for recombinative water desorptions). It is to be considered that on an ion-sputtered TiO2 (110) migrations of Ti3+ and O2− ions between the reduced surface and the stoichiometric bulk region started at ~ 400 K . Accordingly, we suppose that at higher temperatures, possibly above 400 K, the Ti3+ and Ti4+ ions formed in reaction (1) and O2− ions produced in reaction (2) can diffuse to the more reduced surface and bulk regions, altering the local surface structure, composition and reactivity, which complicates the overall reaction scheme. In addition, although the formation of Ti–H species was excluded by HREELS on a stoichiometric TiO2 (110) − (1 × 1) surface even at 0.48 ML hydrogen atom coverage , reaction routes for the production of hydrogen and water involving the formation of Ti–H cannot be ruled out on the reduced r–TiO2 (110) and sr–TiO2 (110) surfaces.
The changes in the time derivative of WF as a function of temperature for an r–TiO2 (110) surface exposed to different amounts of water at 200 K are displayed in Fig. 2a. At 5 min of H2O exposure, two asymmetric, broad state appeared with Tp = 210 and 452 K, with shoulders at 300 and 535 K, respectively. At higher H2O coverages these states intensified, accompanied by the shift of 452 K peak to 410 K. The asymmetric, broad peak at 210 K may be associated with the removal of molecularly adsorbed water (i.e. the state with Tp = 227–243 K, Fig. 1a), while the features above 300 K with the reactions of OH groups. At saturation water exposure, the WF change determined by integration of the dWF/dt curve over time up to 300 K is 350 meV, which can be attributed partly to molecular desorption. The integration corresponding to the 300–600 K region gives 1300 meV WF enhancement, which relates mostly to the reactions of surface hydroxyl groups. It is worth comparing this 1.30 eV WF change to the 1.15 eV decrease in WF due to ~ 4 L of water exposure to a reduced TiO2 (110) crystal at 300 K . Although the values agree reasonably well, it must be taken into account that the WF decrease of H2O exposed TiO2 (110) at 300 K was found to depend on the extent of sample reduction, being around 0.5 eV for a nearly stoichiometric and 1.2 eV for a reduced surface . Moreover, a WF decrease of 1.2 eV was reported for a nearly perfect TiO2 (110) surface at 1.4 ML of H2O coverage , attributed to molecular water adsorption. Considering that saturating the titania surface either with water molecules or covering it by OH groups stemming from H2O decomposition can decrease the WF by more than 1 eV, the WF change experienced during adsorption and heat treatments depend on the relative amounts of the two adsorbates determined by the adsorption temperature and the oxidation state of sample. These factors may contribute to the difference between the observed 1.30 eV (derived from Fig. 2a) and 1.15 eV  WF changes, which varies if the integration limit is altered. Making comparison with an adsorption experiment performed at 300 K, some enhancement of the integration limit would be reasonable, since a part of the states populating above, but near to 300 K desorbs already at 300 K.
The TDS traces for H2 and H2O (Fig. 1a, b) and − dWF/dt vs. T curves (Fig. 2a) deserve more detailed comparison. For this purpose, they are shown in Fig. 2b where the sum of H2O and H2 desorption signals (corrected by ionization probabilities in the mass spectrometer, positive region) are plotted in the company of − dWF/dt vs. T curve (negative values). The moderate increase in WF corresponding to the change in the latter plot up to 250 K can be ascribed to molecular water desorption, which, according to previous study , has less pronounced effect on the WF. At 280–360 K the water and hydrogen desorption have comparable rates, accompanied by slight variation in − dWF/dt values. Note, however, that integrating the dWF/dt signal over time in the 280–360 K temperature range, the resulting WF enhancement, associated with the hydrogen and water desorption, is 276 meV. Above 360 K, an intense feature can be seen at 410 K, which is not resolved on the TDS spectra. At 410 K the desorption rates of both H2 and H2O are lower than at around 300 K, hence the surface species removed from the surface at 410 K have higher dipole moments than those at 300 K. This can be associated with the desorption of intact water molecules near to 300 K, as it was discussed interpreting the TDS curves of Fig. 1a, which molecules have lower dipole moment than the surface OH groups .
The diffusion of H atoms from highly hydroxylated stoichiometric rutile TiO2 (110) surface into the bulk was suggested to occur on the base of TDS data, showing negligible water and hydrogen desorption , but this process was ruled out by others, detecting H2O desorption with ~ 100% yield . To study the dissolution of H atoms into the bulk of titania we performed TDS experiments on a strongly reduced, sr–TiO2 (110) surface (Fig. 3a). In this case, the source of water exposure was the background gas, resulting in narrower TDS features due to less desorption from the sample holder. The smallest water exposure (1000 s) to the sr–TiO2 (110) at 200 K gave water desorption peak with Tp = 355 K (β1 state), corresponding to recombination of surface OH groups. Enhancing the exposure time to 5000 s, this peak increased and shifted to 330 K, while a new feature appeared at 575 K (β2 state). After 10000 s of H2O exposure, the β2 Η2Ο peak intensified, broadened significantly and shifted to 565 K. Interestingly, the β1 state appearing with Tp = 340 K, decreased at the same time. On the effect of further 10000 s water exposure it was even more suppressed and approximated its initial intensity and peak temperature recorded at 1000 s exposure time, while the β2 state peaked at Tp = 670 K. The corresponding, broad H2 TDS traces (not shown) could be decomposed to peaks at 330, 390 and 470 K. Noticeably, the amounts of H2 and β1 H2O states were maximized at around 5000 s H2O exposure, where the β2 H2O starts to intensify, as the integrated peak areas plotted in Fig. 3b also illustrate. These observations suggest that the H atoms of surface hydroxyl groups formed after water exposure at 200 K followed different reaction routes: (i) a part of them reacted with surface OH groups and desorbed in the β1 H2O state with Tp = 330–355 K (ii) another part diffused into the subsurface region of titania. (iii) The dissolved hydrogen reentered the surface above 500 K to produce gaseous water. Considering that the total desorbed amount of H-containing molecules, H2O and H2 represents a saturation-type curve (Fig. 3b), it can be concluded that the formation of these products is limited by the concentration of surface and/or near-surface oxygen defect sites being reactive towards water. The dissolution of hydrogen into the bulk of TiO2 (011) − (2 × 1) plane was also experienced , although at higher temperatures, between 300 and 500 K. The deviation in the temperature ranges can be due to the differences in the structure of TiO2 (011) and sr–TiO2 (110) surfaces. Noticeably, on the TiO2(011) − 2 × 1 plane the formation of surface OH groups was found to be reversible process. The surface hydroxylated by atomic hydrogen at 300 K lost its surface H content an annealing to 500 K due to diffusion of hydrogen into the bulk, but on cooling again to 300 K the dissolved hydrogen reappeared on the surface. According to calculation , the activation energy for the diffusion of H atoms in rutile along the  direction, in channels consisting of Ti4O4 square units, has as low activation energy as 0.29–0.50 eV. From the TiO2 (011) − 2 × 1 surface, being rectangular to these channels, H atoms can diffuse into the bulk by annealing to 500 K. On the stoichiometric TiO2 (110) plane, being parallel to  channels, H atoms cannot diffuse from OHbridge moieties to the subsurface region , they rather form gaseous water. On our strongly reduced, sr–TiO2 (110) sample the O vacancies represent openings of these channels, making possible for the H atoms to enter them below 340 K, as the suppressed H2O desorption state with Tp = 340 K (Fig. 3a) suggests. Dissolved H atoms remain in the bulk up to 500 K and may migrate along the  channels, while above 500 K they react with the surface O atoms and desorb as water.
The question arises, why the amounts of H2O and H2 desorbed below 500 K (Fig. 3) maximized after 5000 s of H2O exposure. A reasonable explanation is that at lower H2O exposures associated with lower surface concentrations of OH groups, H atoms prefer surface reactions to diffusion into the bulk, while above 5000 s of water exposure resulting in higher hydroxyl coverages, the driving force for the dissolution of H atoms into bulk enhanced. This explanation seems to be supported by the observation that exposing the sr–TiO2 (110) to water at 330 K, water desorption was not detected above 500 K. The reason is that the majority of OH groups decompose to gaseous H2 at the adsorption temperature, 330 K, in consistency with the hydrogen desorption peaked at 330 K shown in Fig. 1b.
The interpretation of water desorption states observed above 500 K (Fig. 3a) requires comparison with literature data. The maximal desorption peak temperature of water from defective TiO2 (110) surface is ~ 500 K , while those for hydrogen from nearly perfect TiO2 (110) samples are 510–580 K , 750 K  and 626 K . Noticeably, the OH coverage was found to be constant between 275 and 800 K on a TiO2 (110) sample kept in 0.4 mbar H2O vapor , indicating the high thermal stability of OH groups. Hydrogen can bind to titania in the form of subsurface OH moiety  and Ti–H species . On the TiO2(110) − (1 × 1) surface Ti–H species was tentatively suggested to be present up to 900 K , but in a later work its formation failed to be detected by vibration spectroscopy , rather the presence of subsurface OH was suggested. Considering these findings, the water desorption states above 500 K on our hydroxylated titania sample (Fig. 3a) probably arise from subsurface OH groups. It is worth mentioning that the reduction of rutile and anatase titania  performed in hydrogen atmosphere was characterized by TPR peaks at 640 and 780 K, respectively, suggesting that the rate determining step of these processes was the rupture of surface Ti–O bonds. Accordingly, the H2O desorption peaked at 565 and 670 K after 10,000 and 20,000 s H2O exposure, presented in Fig. 3a, indicates the surface reaction between dissolved hydrogen and titania, resulting in the reduction of the oxide. The reaction of titania with C atoms was also proved in our previous study through the observation of a high temperature CO desorption state with Tp ~ 780 .
In summary, our strongly reduced, sr–TiO2 (110) sample showed considerably altered reactivity towards water as compared to less reduced  and stoichiometric TiO2 (110) surfaces , for which the release of hydrogen from surface OH groups were not detected at all. sr–TiO2 (110) decomposed the water to hydrogen, which is characterized by a higher H2/H2O ratio at the low coverage limit, ~ 5 (Fig. 3b), than that found for the less reduced r–TiO2 (110) sample, ~ 3. Noticeably, the sr–TiO2 (110) displayed a high hydrogen dissolution capability, reaching a maximum amount of 67% for the H-content of surface OH groups (Fig. 3b, high coverage limit). The low temperature, 340 K, where the dissolution of H atoms in the sr–TiO2 (110) sample occurred, is in harmony with a previous finding  concerning the diffusion of hydrogen atoms from the surface of titania powder towards bulk with as low activation energy as 0.09 ± 0.01 eV.
3.2 The Interaction of H2O, H2, D2 and CO with Rh Covered r–TiO2 (110)
Considering that the Rh/s–TiO2 (110) interface has a profound role in the dissociation process of CO , we address the effect of Rh/r–TiO2 (110) system on the decomposition of water as well. Expecting a similar behaviour of CO and H2O, the coverage of Rh was set to 1 ML where maximal reactivity towards adsorbed CO was found . In the following, the r–TiO2 (110) covered by 1 ML of Rh is pre-annealed to different temperatures to adjust the encapsulation of metal particles and control the length of Rh–TiOx interface. The release of water could not be detected from these surfaces after water exposures at 300 K. This can be rationalized taking into account that Rh prefers to bind to oxygen vacancies , thereby decreases the number of active sites on the r–TiO2 (110) surface and suppresses strongly the adsorption of water in a competitive manner. The desorption of H2 is displayed in Fig. 4a. From the Rh layer formed at 300 K not subjected to further annealing, 10 min of water exposure gave a H2 TDS feature centered at around 615 K. After pre-annealing to 500 K a state with Tp = 570 K, while to 700 and 750 K broad TDS spectra containing components at around Tp = 390, 470 and 570 K were detected. Further annealing to 1000 K caused a strong suppression in H2 desorption, and the appearance of features at ~ 390 and ~ 570 K. A comparison with Fig. 1b reveals that the 570 K state, attributable to the decomposition of OH groups, can only form in the presence of rhodium. In Fig. 4b the desorbed amount of H2 is plotted as a function of pre-annealing temperature, showing maximum at ~ 700 K. The state with Tp ~ 570 K in Fig. 4 displays maximum also at this temperature. Qualitatively similar trends were observed for s–TiO2 (110)  and molybdena  supported Rh particles (1 ML) after saturation with CO; the amount of recombinatively desorbed CO was maximized approximately at 700 K pre-annealing temperature. The present and former findings support that the Rh–TiOx interface plays an active role in the dissociation of H2O and CO molecules. Considering that the broad H2 desorption features in Fig. 4a may stem from the applied H2O exposures and from the H2 content of the residual gas, further investigations were carried out to reveal the details of surface processes.
In the following, hydrogen was exposed to Rh covered r–TiO2 (110) at 200 K. Rh prefers to bind to oxygen vacancies , thereby decreases the number of active sites on the r–TiO2 (110) surface. Hydrogen dissociates in negligible amounts on a reduced titania film , but it adsorbs and dissociates on rhodium with low activation energy . In Fig. 5a it can be seen that H2 exposure to the r–TiO2 (110) covered by 1ML Rh resulted in much more intense hydrogen desorption (curve 1) than that found for the Rh-free surface (curve 5). Three H2 desorption states can be suggested for the H2 exposed surfaces; with Tp = 390, 470 and 590 K. Exposing 20 L of H2 to the r–TiO2 (110) sample, there was detected some H2 desorption (curve 5). Since molecular hydrogen has rather low reactivity towards a reduced titania film , this state can be associated with water uptake from the residual gas, as a comparison with Fig. 1b suggests. The appearance of desorption state with Tp = 390 K is in close resemblance with the thermal desorption state at Tp = 340 K for H2 on Rh (111) single crystal . Accordingly, the H2 desorption feature at ~ 390 K on curve 1 can be attributed to hydrogen adsorbed on Rh. To identify the origin of other states, it is to be considered that hydrogen ions exposed to the clean s–TiO2 (110) surface gave recombinative H2 desorption peaked at 380 and 510–580 K . Hence, the features at 470 and 590 K on curve 1 of Fig. 5a possibly belong to the recombinative desorption of hydrogen spilt-over to r–TiO2 (110). A hydrogen state appearing with Tp = 460 K on a TPR spectrum, taken for hydrogen exposed titania powder covered with Pd particles , was attributed to spilt-over hydrogen. To check the assignment of the above-mentioned desorption states, CO was exposed to the hydrogen covered surfaces. TDS trace 2 in Fig. 5a refers to 10 min, trace 3 to 30 min of CO exposure, exhibiting substantial downward shift in Tp-s to 360 and 310 K, respectively, accompanied by a decrease in the amount of desorbed H2 of about 15% in both cases. To interpret these changes, it should be considered that the bonding of CO is weak to reduced titania , characterized by a TDS peak at ~ 200 K. On the Rh (100) single crystal CO post-adsorption shifted down the Tp of hydrogen desorption by about 100, from 330 to 230 K . The significant downward shift of hydrogen desorption peak temperature due to CO exposure, as reflected by TDS traces 1 and 3, suggests that the underlying process is the same, namely the co-adsorption of CO and hydrogen on rhodium which results in the decrease of hydrogen adsorption bond energy due to strong lateral repulsive interaction operating between Ha and COa . However, the shoulders in curve 3 at ~ 470 and ~ 590 K cannot be attributed to desorption from the Rh surface , they rather indicate hydrogen desorption from the titania support . A straightforward way to eliminate the adsorption of hydrogen on Rh is its poisoning with pre-adsorbed carbon monoxide. After saturating the Rh nanoparticles by 20 L CO at 330 K and exposing them to 20 L H2 at 200 K, curve 4 was recorded, which coincided well up to 500 K with curve 5, belonging to H2 desorption from the Rh-free titania exposed to H2 (and water from the residual gas). This indicates that the source of desorption up to 500 K in both cases is the titania support. Remarkably, above 500 K a desorption state could be resolved with Tp = 590 K, which can be recognized even better in curve 6 difference spectrum, provided by the subtraction of curve 5 from curve 4. The 590 K feature can be associated with hydrogen spilt-over at 330 K  during CO adsorption and/or the reaction of background water with the Rh–TiOx interface (Fig. 4a). To decide about these possibilities further investigations were performed, including the co-adsorption of deuterium, hydrogen, water and carbon monoxide as presented below.
To determine the components of broad hydrogen TDS feature (curve 1) in Fig. 5a, Gaussian fits were calculated which resulted in variable results. For the sake of an unambiguous decomposition, further co-adsorption experiments were carried out at Ta = 270 K, see Fig. 5b. The Rh covered (1 ML) titania was exposed to 20 L H2 (and water from the background gas), then post-exposed to 10, 30 and 60 min of CO at 270 K. In contrast to the case of CO adsorption at 200 K (Fig. 5a), the amount of desorbed hydrogen decreased considerably due to CO uptake and after the highest CO exposure (curve 4) the H2 TDS traces displayed three features at around 310, 470 and 570 K. The TDS spectrum in dotted line (curve 5) represents H2 desorption from the Rh-free titania surface exposed to H2 (and water from the residual gas). On the difference spectrum (curve 6) produced by subtracting curve 5 from curve 4 two hydrogen desorption states can be distinguished, with Tp = 470 and 570 K, the amounts of which are comparable. Taking into account that hydrogen desorbs from the titania supported Rh nanoparticles with Tp = 390 K (Fig. 5a), and co-adsorbed CO decreases the desorption peak temperature of hydrogen on Rh , it is clear that the desorption features with Tp = 470 and 570 K belong to the decomposition of OH groups formed on the titania.
Since during H2 exposures water from the residual gas has also been exposed to the r–TiO2 (110), which is very reactive towards the latter compound, the question arises about the concerted and separate role of adsorbed hydrogen and water in the hydroxylation process of titania. To distinguish the impact of hydrogen and water on the development of hydrogen desorption features, adsorption experiments with deuterium were carried out. 1 ML Rh deposited on r–TiO2 (110) was pre-annealed to different temperatures, saturated with 20 L D2 at 200 K (accompanied by inevitable H2 and H2O exposure from the residual gas), then characterized by TDS spectra plotted in Fig. 6a. Desorption spectra were recorded at m/z = 2, 3 and 4, corresponding to the release of H2, HD and D2. Red curves correspond to Rh covered titania exposed to 20 L of D2 at 200 K without pre-annealing. The broad desorption spectra of H2, HD and D2 could be decomposed to peaks at 310, 370, and 450–460 K. Enhancing the pre-annealing temperature to 700 K, the state with Tp = 370 K attributable to D2 desorption from rhodium is suppressed. This can be associated with the spillover of TiOx species to the Rh nanoparticles, hindering the uptake of both D2 and H2, alike that was found for carbon monoxide . TDS peak positions remained constant up to 800 K pre-annealing. Importantly, for HD and D2 the highest peak temperature resolvable in Fig. 6a is at around 450 K, that is the spillover of D2 and HD does not result a state with Tp = 570 or 590 K (Fig. 5), supporting that the latter features originate from the surface reaction of water. However, on the H2 TDS spectrum a shoulder appeared at 590 K after pre-annealing to 800 K. This state may originate from water being present in trace amounts in the vacuum chamber. Incomplete encapsulation of 1 ML thick Rh layer was observed after pre-annealing to 800 K, which treatment still preserved the catalytic activity of metal particles, manifested in the dissociation of adsorbed CO . After heating to 1000 K (Fig. 6a), H2 desorption could hardly be detected and the TDS traces for HD and D2 peaked at around 310 and 370 K can be associated with the OH and OD groups formed on titania, as the data of Fig. 1b and the similar desorption states for atomic deuterium exposed TiO2 (110) suggest .
The total desorbed amounts of deuterium and hydrogen calculated from the spectra of Fig. 6a, as a function of pre-annealing temperature are presented in Fig. 6b. The uptake of H2 is about 60% that of D2, indicating a substantial H2 adsorption from the background gas. After pre-annealing to 550 K, the coverage of deuterium somewhat increased due to removing the hydrogen adsorbed from the residual gas during the heat treatment, while the total uptake of D2 + H2 decreased by about 15% (uppermost curve), corresponding to the onset of encapsulation of Rh particles by TiOx . After pre-annealing to 550, 700, 800 and 1000 K, a steeper decrease was found for the desorbed amounts of H2, HD and D2, related to the higher extent of encapsulation. However, the desorbed amounts did not drop to zero, that is the Rh particles remained reactive due to incomplete encapsulation even after annealing to 1000 K. This is also supported by a previous observation of CO dissociation on titania supported Rh particles pre-annealed to 1000 K . Accordingly, the desorption states for H2, HD and D2 with Tp = 310, 370 and 450 K (Fig. 6a) relate to the presence Rh particles, i.e. to the adsorption and dissociation of D2 and H2 on Rh particles, followed by spillover to the support. The H2 desorption state with Tp = 590 K could only be detected after pre-annealing to 800 K. A comparison with Fig. 4a, b showing that the maximization of 570 K H2 desorption state needed pre-annealing to 700–750 K before H2O exposure to the r–TiO2 (110) + Rh surface, suggests that the 590 K state originates from water exposure. This assumption is justified in the following section.
In Fig. 7 the upper TDS spectra (thin curves) for D2, HD and H2 were taken for r–TiO2 (110) supported Rh (1 ML) particles heated to 700 K (where the length of Rh–TiOx borderline is expected to be maximized), then saturated with 20 L D2 during cooling in the 500–270 K temperature range. These curves are similar to those presented in Fig. 6a with the same pre-annealing temperature, although the lowest temperatures of gas exposures were 270 and 200 K, respectively. The lower (thick) TDS traces were recorded after the following treatments. First, the sample pre-annealed at 700 K was saturated with CO at 400 K, to suppress the hydrogen uptake of Rh particles. This was followed by cooling to 270 K in D2 stream and by 20 L of D2 exposure at 270 K. At last, 10 min of H2O exposure was applied at 270 K. As a result, a hydrogen (m/z = 2) TDS peak was detected at ~ 540 K, which feature was missing on the D2 and HD TDS spectra. The H2 desorption trace for Rh-free r–TiO2 (110) exposed to 10 min of H2O is also presented for comparison (dotted curve in violet), on which the 540 K feature is also absent. These observations prove unambiguously that the formation of this feature requires the presence of rhodium, titania and water. Accordingly, the TDS features at 570 K in Figs. 4a and 5b and at 590 K in Fig. 5a are associated with H2O adsorption and not with hydrogen spillover. The maximization of 570 K H2 desorption state at around 700 K pre-annealing temperature (Fig. 4a), where the length of Rh-TiOx borderline was found to be maximized , indicates that the reaction centers for the decomposition of water are at the Rh–TiOx interface. The reactivity of TiOx film supported by Au (111) substrate towards H2O was related to the number of defect sites [9, 38], which can be present in large amounts in the incomplete TiOx overlayer formed on rhodium  as well.
It is important to emphasize, that saturating the Rh (1 ML) with CO at 400 K (Fig. 7), which leads to the dissociation of molecule , does not cease the reaction of water with the Rh/r–TiO2 (110) interface. This observation can be understood considering the following factors. It is reasonable to suppose that the decomposition of water and CO needs different type of active centers at the Rh/r–TiO2 (110) interface. This seems to be supported by the observation that the reactivity of H2O is higher towards the oxygen defect sites of r–TiO2 (110) than that of CO. Our strongly reduced (heavily ion-sputtered) TiO2 (110) surface could be oxidized by saturation H2O exposure followed by a heat treatment, while CO failed to dissociate at high CO exposures on this surface and on a strongly reduced titania film . The reversibility of water–TiOx reaction in the presence of CO also deserves attention. The reversible reaction of water with the oxygen defect sites of a nearly stoichiometric, s–TiO2 (110) surface has been established by STM . Accordingly, water desorption at around 500 K leads to the regeneration of defect sites. Annealing the r–TiO2 (110) supported Rh particles repeatedly to 700 K to form the TiOx overlayer, the intensity of 570–590 K H2 peak (Fig. 5) did not change after water exposures at 300 K (not shown). This supports that the adsorption-annealing cycles recovered the reactivity of the interface which may be interpreted supposing that the desorption of hydrogen is followed by the dissolution of oxygen into the r–TiO2 (110) sample through the TiOx overlayer. Since the diffusion of oxygen in highly reduced, non-stoichiometric rutile TiO2−x is facilitated as compared to stoichiometric one , the repeated annealing of the r–TiO2 (110) + 1 ML Rh layer to the same, high enough temperature can restore the reduced state of surface and hence the activity of TiOx overlayer. For CO molecules dissociated on TiO2 (110) − (1 × 2) supported rhodium particles, the diffusion of carbon atoms was observed from the Rh–TiOx interface towards the titania support by STM even at 500 K . These processes free up interfacial sites for the decomposition of H2O well below the recombination temperature of C and O atoms, 780 K , making possible to regenerate the CO pre-covered surface on annealing to 700 K. It must be noted that the production of O vacancies, responsible for H2O dissociation on TiOx particles supported by Au (111) surface was performed by reduction in CO at 575 K . However, the deliberation of CO2 from the CO-exposed TiO2 (110) + Rh surfaces was not observed by TDS neither in the present, nor in our former  study, indicating the absence of CO oxidation under our experimental conditions.
It is of importance that the composition of TiOx overlayer with x ~ 1.2 on the Rh particles  corresponds to an even more reduced state of titania than that established for an overlayer on Ag nanoparticles  with a stoichiometry of Ti4O7 and Ti3O5 (x = 1.75 and 1.67, respectively). Accordingly, a narrow bandgap may belong to our encapsulation layer, allowing IR photon driven photochemical reactions on it.
Two black TiO2 (110) samples have been investigated, characterized by O/Ti AES ratios of 0.9 and 0.8, designated as reduced, r–TiO2 (110) and strongly reduced, sr–TiO2 (110) surfaces, respectively. r–TiO2 (110) exposed to water at 200 K produced OHa species, which recombined to H2O with Tp = 300, 377 and 470 K and decomposed releasing H2 with similar peak temperatures, attributable to surface inhomogeneity. Temperature-programmed work function measurements revealed the presence of adsorption states with positive outward dipole moments. sr–TiO2 (110) could hold the H atoms, stemming from OHa species, dissolved up to 500 K, which reentered the surface producing β2 H2O states with Tp = 570, 670 and 750 K. On the Rh covered r–TiO2 (110) surface the reaction of water at the Rh–TiOx interface was characterized by a H2 TDS sate with Tp = 570–590 K. Remarkably, this state could be formed in the presence of pre-adsorbed CO which is also reactive towards the interface. 1 ML Rh on r–TiO2 (110) saturated with deuterium at 270 K gave D2 TDS features with Tp = 370 and 450 K, corresponding to desorption from the metal and from the spilt-over state, respectively. On the r–TiO2 (110) + 1 ML Rh particles the adsorbed CO exerted variable influence on the surface reactions of hydrogen, depending on the adsorption temperature. (i) Saturating the Rh by CO at 330 K blocked the uptake of hydrogen on the metal, consequently eliminated the spillover of hydrogen. (ii) At 270 K, CO replaced the pre-adsorbed hydrogen on the metal, while the spilt-over hydrogen could be detected. (iii) At 200 K, CO removed a part of pre-adsorbed hydrogen from the Rh particles; a decrease in the adsorption bond energy of the part remained adsorbed and the presence of spilt-over hydrogen were observed. TiOx overlayers produced by stepwise heating of the r–TiO2 (110) supported Rh particles suppressed the D2 and H2 uptake of the metal at 200 K, but TiOx sub-monolayers allowed the spillover of hydrogen towards the support.
Supports through grants of the Hungarian Scientific Research Fund (OTKA) K120115, GINOP-2.3.2-15-2016-00013 and COST Action CM1104 are gratefully acknowledged.
- 3.Liu X, Zhu G, Wang X, Yuan X, Lin T, Huang F (2016) Adv Energy Mater 1600452Google Scholar
- 4.Fujiwara K, Deligiannakis Y, Skoutelis CG, Pratsinis SE (2014) Appl Catal B - Environ 154–155 9–15Google Scholar