Interaction of Petroleum-Relevant Organosulfur Compounds with TiO2

Jun 13, 2012 - These results highlight the importance of defects in the reactivity of titania, and lay the foundation for the study of larger, refract...
0 downloads 0 Views 1024KB Size
Download PDF
Article pubs.acs.org/Langmuir

Interaction of Petroleum-Relevant Organosulfur Compounds with TiO2(110) Lauren Benz,* Aileen Park, Jenelle R. Corey, Michelle P. Mezher, and Victoria C. Park Department of Chemistry and Biochemistry, University of San Diego, San Diego, California 92110, United States ABSTRACT: The interaction of two sets of structurally related molecules, thiophenol/ thioanisole, and thiophene/tetrahydrothiophene, with vacuum-annealed and ionbombarded TiO2(110) surfaces has been studied using a combination of temperatureprogrammed reaction spectroscopy (TPRS) and X-ray photoelectron spectroscopy (XPS). All thioethers studied were observed to adsorb and desorb from both surfaces without producing reaction products, while thiophenol, the only species studied containing a S−H bond, reacted with both surfaces. Approximately 25% of surface bound thiophenol decomposed over the vacuum-annealed surface. On the bombarded surface, thiophenol both decomposed into surface-bound CxHy/S fragments, and reacted to form benzene, which desorbed from the surface at 400 K. We propose that phenylthiolate formation on the bombarded surface leads to the observed production of benzene. These results highlight the importance of defects in the reactivity of titania, and lay the foundation for the study of larger, refractory sulfur compounds present in fuel.

1. INTRODUCTION The organosulfur compounds in petroleum present a number of challenges. In refinery processes, the presence of sulfur causes deactivation of reforming catalysts1 and corrosion of equipment over time.2 In addition, the combustion of fuels containing such compounds leads to the emission of sulfur dioxide, an environmental pollutant known to cause acid rain.3 As a result, worldwide regulation of sulfur in fuels continues to tighten, and many refineries have been forced to add or modify processing units to decrease sulfur content.4 To compound the problem, as demand for petroleum increases, the availability of cleaner, lighter sources of crude decreases. The removal of sulfur takes place in several steps in current desulfurization processes. Both pretreatment of crude oil before upgrading and post-treatment of individual products such as gasoline and diesel are commonly employed.5,6 In the pretreatment steps, hydrodesulfurization (HDS) occurs prior to cracking and typically employs alumina-supported Ni and/or Co promoted MoS2 catalysts.1 The gaseous H2S produced in HDS is then converted to elemental sulfur using the Claus process, which typically employs alumina or titania as a catalyst, titania being the more active catalyst by a factor of about 4.7 Recent efforts have been made to improve the HDS process as it is ineffective for complex aromatic compounds of sulfur. Typically, these compounds consist of larger, refractory species such as thiophene derivatives and other polyaromatic sulfur heterocycles. A variety of additives including phosphorus and lanthanum have been mixed with alumina to improve HDS activity, and different supports and support mixtures including zeolite, zirconia, and titania have been investigated specifically targeting the removal of sulfur from dimethyldibenzothiophenes (DMDBT).8 The authors of this work speculate that the improvement of HDS activity toward DMDBT is due to an © 2012 American Chemical Society

increase in support acidity upon addition of oxides such as titania. Post-treatment of end products is becoming increasingly relevant as stricter limits are placed on sulfur content in fuels, and a number of recent developments in this area focus on the adsorptive, rather than catalytic, removal of organosulfur species. For example, metal ion exchanged zeolites including Cu+, Ag+, and Ni2+ show selective adsorption of some of these species in diesel and gasoline.9,10 Silica-supported transition metals and alumina-supported silver ions have also shown promise in the adsorptive removal of thiophenes.11,12 Quite recently, Samokhvalov et al. have shown that titania-supported Ag clusters are effective in the adsorptive removal of dibenzothiophene.13 Despite the importance of understanding the interaction of complex organosulfur compounds with various oxidic substrates and oxide-supported clusters, it is surprising that relatively few fundamental studies on well-defined oxide surfaces have been conducted in this area. Many studies, however, have been performed on metal surfaces, providing mechanistic insight into the various desulfurization processes.14,15 Given that these metals are typically supported on oxide substrates and that the choice of oxide is important to overall activity, fundamental studies of oxides and supported clusters with organosulfur compounds are required to form a complete picture. In particular, there is a lack of studies on the larger compounds, likely because of the difficulty of working with these species in vacuum. Early work in the study of small organosulfur compounds includes investigation of the interactions of methanethiol, ethanethiol, dimethyl and diethyl disulfide with titanium dioxide, and zinc oxide single crystalline Received: May 3, 2012 Revised: June 10, 2012 Published: June 13, 2012 10209

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

surfaces.16−21 In addition, thiophene has been studied over titania,22 and reactions of methanethiol with titania-supported Ni clusters have been reported.23 Herein, we report a study of the interaction of a reduced TiO2(110) substrate with thiophenol, thioanisole, tetrahydrothiophene, and thiophene, the latter three of which are known core structures in petroleum.24 Molecular pairs (thiophenol/thioanisole and tetrahydrothiophene/thiophene) with similar structures were selected for comparison purposes. As it is known that the defect chemistry of titanium dioxide is quite rich,25 and that these defects can lead to surprising reactivity,26−28 we examine the reactivity of these compounds with the oxide substrate and its defects as a first step toward further modifying the surface with metallic clusters and other active species for the adsorptive removal of refractory organosulfur compounds.

on two sets of structurally analogous organosulfur compounds: thiophenol/thioanisole and thiophene/tetrahydrothiophene on a reduced TiO2(110) surface. The spectra of thiophenol and thioanisole are very similar as shown in Figures 1 and 2,

2. EXPERIMENTAL SECTION A single crystalline TiO2(110) substrate (Princeton Scientific, 1 cm × 1 cm × 1 mm) was mounted in an ultrahigh vacuum (UHV) chamber with a base pressure of ∼1 × 10−10 Torr. The UHV chamber was equipped with a mass spectrometer (Hiden HAL 301/3F, mass range 1−300 amu), an ion bombardment gun (RBD 04-165), an X-ray photoelectron spectrometer with a cylindrical mirror analyzer (XPS, Phi 04-548), and an Auger electron spectrophotometer (AES, Phi 15255G). The sample was mounted on a tantalum plate with thin tantalum wires, and a k-type thermocouple was glued in a small hole in the side of the crystal with UHV compatible glue (Aremco, Ceramabond 503). The sample was cooled to minimum temperature of 90 K by thermal contact with a liquid nitrogen reservoir and heated to maximium temperature of ∼900 K by passing current through a tungsten filament mounted ∼2 mm directly behind the Ta back plate. After mounting the crystal, the substrate was cleaned and reduced with multiple cycles (∼30) of Ar+ bombarding (2 kV, 15 min, ∼2 μA) and annealing (850−900 K, 10 min) until no impurities were detected with AES. For experiments involving a highly reduced surface, the surface was bombarded (10 min, 1.5 kV, sample current ∼2 μA) before exposure to the organosulfur species. Thiophenol (Alfa Aesar, 98+%), thioanisole (Aldrich, 99%), thiophene (TCI, 98+%), and tetrahydrothiophene (Acros Organics, 98+%) were all prepared for introduction into vacuum with multiple freeze−pump−thaw cycles with liquid nitrogen, and purity was checked using the mass spectrometer. The pure gases were then introduced into the chamber using a directed doser comprised of a leak valve and a 1/4 in. stainless steel tube positioned ∼1 cm in front of the crystal face. All species were dosed at an initial sample temperature of 90 K. Following exposure to a gas of interest, the sample was moved in front of the mass spectrometer and heated linearly at a rate of 2 K/s while monitoring species of various mass as they evolved from the surface. The entrance to the ionizer was covered with a 4 mm aperture plate, in front of which the sample was centered at a distance of ∼1 cm during the temperature-programmed reaction studies. In the resulting spectra, we have set the integrated area of a saturated monolayer to 1.00 ML and scaled the other coverages accordingly, due to the known difficult nature of determining exact coverages for adsorbates on TiO2.29 X-ray photoemission data were collected at 90 K using a Mg Kα X-ray source (hν = 1253.6 eV, maximum resolution of 1.1 eV). The binding energies were calculated using a half-copper (2p3/2, 932.4 eV) half-gold (4f7/2, 83.8 eV) calibration substrate to give a wide range of peak separation and an accurate binding energy scale. This led to a Ti 2p3/2 photoelectron peak at 458.4 eV on the annealed surface, consistent with previous reports for this surface.21 XPS spectra were fit using CasaXPS software following subtraction of a Shirley background and subsequent fitting with combined Gaussian−Lorentzian functions.

Figure 1. Temperature-programmed reaction spectra of thiophenol. The heating rate was 2 K/s.

Figure 2. Temperature-programmed reaction spectra of thioanisole. The heating rate was 2 K/s.

respectively. In both sets of spectra, two main desorption features are apparent: one at higher temperature which saturates with increasing exposure, indicative of a saturating monolayer, and a multilayer desorption feature which is lower in temperature and does not saturate. The multilayer desorption feature for thiophenol was slightly below that of thioanisole, 193 and 204 K, respectively. A higher multilayer desorption for thioanisole is not surprising given the larger overall size of the molecule and thus slightly larger van der Waals forces between molecules. The relative boiling points of thiophenol (442 K) and thioanisole (466 K) are also consistent with the relative temperatures of multilayer desorption observed here.30 The higher temperature peak in the thiophenol spectrum appears at 388 K at a coverage of 0.09 ML and shifts to ∼365 K for a saturated monolayer. Similarly, a maximum desorption temperature of 395 K was observed for 0.10 ML of thioanisole. This feature also shifted to ∼365 K with increasing coverage. Shifting to lower temperature with increasing coverage is indicative of increasing interaction between molecules and decreasing interaction with the substrate. Such shifts are common in many studies of molecular

3. RESULTS 3.1. Temperature-Programmed Reaction Studies on Vacuum-Annealed TiO2(110). TPRS studies were performed 10210

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

adsorption on TiO2(110).28 Activation energies of desorption can be calculated from the maximum desorption temperatures (in the limit of low coverage) assuming first-order desorption kinetics and a prefactor of 1013 s−1.31 These energies are 102 ± 3 and 104 ± 3 kJ/mol for thiophenol and thioanisole, respectively, signifying weak to moderate chemisorption. We attribute the binding of thiophenol and thioanisole to Lewis acid/base type interactions involving the lone pair electrons on sulfur and surface Ti cation sites, similar to that observed for methanethiol on TiO2(110).21 The main difference between the two molecules is in the appearance of a third broad, but clear, feature at ∼220 K at a coverage of 0.68 ML in the thiophenol spectrum. In the desorption spectra of thioanisole, a general broadening to the lower temperature side of the 390 K peak is observed with increasing coverage, rather than a distinct desorption feature. We attribute the appearance of this additional intensity between 200 and 300 K in both molecules to formation of a compression of the first layer, similar to that observed for acetone.32 This compression may be more easily distinguished in the thiophenol spectra due to a more homogeneous molecular packing given the smaller S−H group. Scans over the entire range of the mass spectrometer (all such scans were taken in increments of 10 amu to preserve sensitivity) following saturation exposures revealed no desorption of reaction products on the annealed surface for thiophenol or thioanisole. Furthermore, the surface yielded clean AES spectra (i.e., no carbon or sulfur buildup) following thioanisole desorption experiments; however, some carbon and sulfur remained on the surface following TPRS studies of thiophenol. The latter case was studied further with XPS (to follow). It is important to note that the sample was not bombarded and annealed between TPRS experiments in the traces shown in Figure 1 in order to maintain the reduction state of the sample, despite evidence that some molecular decomposition occurred. A select number of experiments (not shown) with cleaning in between produced data nearly identical to that shown in Figure 1. This approach was also taken in collecting mass scans over the full spectrometer range; that is, a subset of likely reaction masses were scanned following a bombardment/annealing cycle, yet no reactivity was observed. We suspect that this is due to the fact that any residual C and S left on the surface agglomerated upon heating such that most of the surface was still exposed and unchanged. Unlike thiophenol and thioanisole, the spectra of thiophene and tetrahydrothiophene, shown in Figures 3 and 4, are markedly different. Thiophene has been studied previously on a similarly prepared TiO2 (110) surface, 22 and we have reexamined it here for comparison to tetrahydrothiophene. The results obtained for thiophene are similar to the previously reported work with some minor differences, likely based on small differences in sample preparation. At a coverage of 0.02 ML, a desorption feature appears at 266 K which shifts to 240 K with increasing exposure. A second feature at 170 K is apparent at a coverage of 0.41 ML. A third state appears between these features at 200 K by 0.51 ML. Finally, a multilayer desorption peak was visible at 145 K which did not saturate. The main difference between our study of thiophene and that observed previously is the absence of a peak at 450 K. In the previous study by Liu et al.,22 a very weak, broad peak attributed to adsorption at defect sites was observed to saturate at a coverage of 0.04 ML. It is possible that the higher annealing temperature of 950 K employed in these earlier studies lead to a

Figure 3. Temperature-programmed reaction spectra of thiophene. The heating rate was 2 K/s.

Figure 4. Temperature-programmed reaction spectra of tetrahydrothiophene. The heating rate was 2 K/s.

greater number of surface oxygen vacancies and/or subsurface interstitials. The previous work attributes the complex TPRS spectra of thiophene to two competitive binding mechanisms: binding via the aromatic ring, which is predicted to be the more stable geometry at low coverage, and weaker binding via the S atom, which becomes competitive at higher coverage. This agrees with our observation of the binding in comparison to that of tetrahydrothiophene as described below. As the lone pair electrons in S are involved in the aromatic ring, binding via S is substantially weakened. The desorption spectra of tetrahydrothiophene, while clearly different than thiophene, appeared similar in shape to that of thiophenol and thioanisole, with two major desorption features. The multilayer feature at 160 K did not saturate with increasing exposure. The lower multilayer desorption temperature as compared to that of thioanisole and thiophenol is expected given the relatively weaker intermolecular forces of this smaller, nonaromatic molecule in the condensed phase. In addition, the relative boiling temperatures of thiophene (357 K) and tetrahydrothiophene (394 K) are consistent with the relative multilayer desorption temperatures observed here.30 The higher temperature desorption feature shifts from 384 K at 0.04 ML to ∼325 K at saturation coverage. This shift is again indicative of increasing interactions between the surface-bound molecules with increasing coverage. It is interesting to compare the magnitude of the shift of this higher temperature state: ∼60 K for tetrahydrothiophene versus that of ∼30 K for thioanisole 10211

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

and thiophenol. A greater shift indicates greater increasing intermolecular forces between molecules, perhaps due to a more efficient packing of this smaller molecule. Most striking in the comparison of thiophene and tetrahydrothiophene is the difference in the activation energies of desorption. Comparing the maximum desorption temperatures of the lowest exposures studied in each case (266 and 384 K for thiophene and tetrahydrothiophene, respectively), desorption activation energies of 66 ± 3 and 101 ± 3 kJ/mol can be calculated, the latter very close to that of thiophenol and thioanisole. This large difference in activation energies suggests qualitatively different binding mechanisms. As mentioned previously, in the earlier work on thiophene, density functional theory calculations suggest that thiophene binds with its aromatic ring parallel to the surface at low coverages.22 The observed switch from physisorption to chemisorption in going from thiophene to tetrahdrothiophene observed here demonstrates a change in preference from binding via the ring to binding via the lone pair electrons on sulfur. This change is likely due to the enhanced availability and basicity of the electrons on sulfur for tetrahydrothiophene.33 In fact, one can compare the relative desorption temperatures of these molecules to that on Pt(100) and see roughly the same trends, although in the case of adsorption onto this highly active metal surface, significant reactivity was also observed.34 Similar to thiophenol and thioanisole, scans over the entire range of the mass spectrometer following a saturation exposure revealed no desorption of reaction products from the annealed surface for thiophene and tetrahydrothiophene. In addition, AES revealed no accumulation of carbon or sulfur on the surface following the desorption studies of thiophene and tetrahydrothiophene. 3.2. Adsorption and Reaction over Ion-Bombarded TiO2(110): TPRS and XPS. We also investigated the interaction of thiophenol, thioanisole, and tetrahydrothiophene with a heavily reduced TiO2(110) surface prepared using Ar ion bombardment. Thiophene was not investigated due to its previously observed lack of reactivity on a comparably prepared surface.22 Of the species investigated, thiophenol showed appreciable reactivity with the bombarded surface, yielding benzene at 400 K as shown in Figure 5. The desorption of benzene at 400 K is reaction limited, as benzene desorbs molecularly below 300 K, as reported previously35 and subsequently confirmed here (data not shown). The trace shown in Figure 5 has been corrected for the minor 78 fragment of thiophenol, present at 7% of the parent 110 peak. The production of benzene requires C−S bond cleavage and acquisition of H by the remaining C6H5 fragment. Integrated desorption areas, further corrected for typical mass spectrometer transmission, gain, yield, and ionization using a quantification method reported by Ko et al.,36 reveal that ∼45% of the total desorption in Figure 5 is attributable to the formation of benzene. Other species scanned but not detected include H2 (m/z = 2), H2O (m/z = 18), S (m/z = 32), S2/SO2 (m/z = 64), H2S (m/z = 34), diphenyl (m/z = 154), and diphenyl disulfide (m/z = 218). We also considered the possible formation of phenyl radical, as methyl radical was observed in the study of methanethiol decomposition over TiO2. The corresponding m/z = 77 peak, however, was minor, and its relative intensity was consistent with our observed fragmentation pattern of benzene itself; therefore, no phenyl radical was found to desorb from the surface. Subsequent reaction cycles did not produce a significant benzene peak,

Figure 5. Benzene formation following exposure of bombarded TiO2 to thiophenol. The heating rate was 2 K/s.

indicating that the reaction sites are blocked or deactivated after one cycle. It is worthwhile to note that the baselines for benzene and thiophenol (though to a lesser extent) do not fully recover upon heating to 600 K, suggesting that some additional desorption past this temperature is likely, as confirmed in the XPS studies. Furthermore, thiophenol desorption was observed to broaden significantly on the bombarded surface, as expected given the heterogeneous nature of bombarded surfaces. Finally, the integrated area beneath the 110 parent signal was a factor of 2 greater than that on the vacuum-annealed surface, likely due to the increased surface area of the rough, bombarded surface. We also investigated the reaction of thiophenol using X-ray photoelectron spectroscopy by comparing adsorption to the highly defective bombarded surface to that of the annealed surface. It is known that bombarding a TiO2 surface with inert ions leads to the preferential removal of oxygen atoms and the creation of a rough surface containing lower oxidation states of titanium (III, II).28 Titanium 2p spectra of the two types of surfaces investigated here are shown in Figure 6. On the vacuum-annealed surface, the majority of detectable Ti atoms are formally in the +4 oxidation state, with a minority of ∼6% in the +3 state. Lower oxidation states on the vacuum-annealed surface are attributed to surface oxygen vacancies and near subsurface interstitials.28,37 Following argon ion bombardment, the relative concentration of the +3 state increases to 20%, and a +2 state appears at a small concentration of 3%. These surfaces lead to different reactivity as evidenced by the C 1s and S 2p spectra. In the carbon 1s spectrum shown at the top of Figure 7, a peak appears at 284.5 eV on the annealed surface following a multilayer exposure (∼10 ML), consistent with the reported binding energy of multilayer thiophenol on Mo(110).38 The corresponding S peak is given in Figure 8. After correcting for the relative sensitivities of our detector to S and C,39 we obtained a 1:6 area ratio as expected given the stoichiometry of thiophenol. Upon heating to 220 K at a rate of 2 K/s (same rate as TPRS experiments), a decrease in the intensity of this C 1s peak by a factor of ∼2 occurs due to multilayer desorption, with the remaining intensity approximately representative of a thiophenol monolayer. Further heating to 600 K decreases this peak to ∼25% of its monolayer value, indicating that this 10212

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

Figure 8. X-ray photoelectron spectra of the S 2p region following exposure of an annealed TiO2 surface to a multilayer dose of thiophenol at 90 K, followed by subsequent heating to the above temperatures and reacquisition of spectra.

Figure 6. X-ray photoelectron spectra of the Ti 2p region of TiO2(110) on (A) a vacuum-annealed surface and (B) an argon bombarded surface (1.5 kV, 10 min, sample current of 2 μA).

further decreases by a factor of 2 upon heating to 800 K but remains visible on the surface. A set of corresponding sulfur 2p spectra collected on the annealed surface is given in Figure 8. Following a multilayer exposure at 90 K, a binding energy of 163.9 eV was measured for the sulfur peak, also consistent with the previous study on Mo(110).38 This peak decreases significantly in intensity (by a factor of ∼2) after heating the surface to 220 K and is hardly visible upon heating to 600 K. The fact that the sulfur is barely above the background following heating to 600 K, combined with the observation that no S-containing species other than thiophenol in the mass range of 1−300 amu desorbed from the surface in this temperature range suggests two possibilities: (1) the small amount of sulfur remaining on the surface is below the detection limit of our XPS, and/or (2) the sulfur resides on the surface in such a way that its signal is attenuated by other species such as C. Since sulfur was visible in AES after heating to 600 K as mentioned previously, we believe that (1) is the more likely situation. This is also consistent with the fact that the relative sensitivity factor for S is 0.8 for AES42 but only 0.35 for XPS.39 Furthermore, the amount of S expected on the surface given the remaining observed C is small and can be estimated. Given that the size of thiophenol is similar to the previously studied benzaldehdye,41 one can estimate that one molecule binds at approximately every other Ti4+ surface site. The Lewis-acidic Ti4+ sites are present at a concentration of 5.2 × 10 14 sites/cm 2; 28 thus, a monolayer of thiophenol corresponds to roughly 2.6 × 1014 molecules/cm2. Normalizing the area under the carbon signal following heating to 220 K (as this generates a monolayer by removing the multilayer) to this value, and considering that there are 6 C atoms for every S atom deposited for thiophenol, one would expect only ∼2% of all surface Ti sites to be covered with S if it remains on the surface up to 600 K. Thus, we suspect that S is in fact on the

Figure 7. X-ray photoelectron spectra of the C 1s region following exposure of an annealed TiO2(110) surface to a multilayer dose of thiophenol at 90 K, followed by subsequent heating to the above temperatures and reacquisition of spectra.

fraction of the molecules has decomposed into CxHy fragments over the surface21,40 since this temperature is above that of the observed thiophenol desorption in the TPRS spectra (Figure 1). It is likely that reaction with defect Ti3+ sites in the form of surface oxygen vacancies, step edges, and/or near subsurface intersitials21,41 gives rise to this decomposition. The C 1s peak 10213

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

surface and we simply do not clearly resolve it with XPS as it is below our detection limit. XPS spectra of thiophenol on the bombarded surface exhibited some interesting differences and similarities to that of the annealed surface. The multilayer C 1s peak appeared similar to that of the annealed surface, as expected (Figure 9).

Figure 10. X-ray photoelectron spectra of the S 2p region following exposure of a bombarded TiO2(110) surface to a multilayer dose of thiophenol at 90 K, followed by subsequent heating to the above temperatures and reacquisition of spectra.

that is, reduced Ti defects produced upon bombarding the titania surface give rise to phenylthiolate formation and subsequent decomposition into benzene and adsorbed sulfur. The nature of the bombarded surface is such that both surface and subsurface Ti3+ defects are present to a significant degree, as evidenced by previous XPS and ultraviolet photoemission spectroscopy (UPS) studies;44 therefore, both types of defects are candidates for the reactivity observed here. Further studies are required to determine if one or both of these defect types dominate the reaction chemistry of benzenethiol; however, the fact that the vacuum-annealed surface is inactive suggests that surface oxygen vacancies alone (which are openly accessible on such a surface) are not sufficient for removal of S. Recent literature shows an increased awareness of the important role of subsurface interstitial Ti3+ species in reaction chemistry on both bombarded and annealed surfaces.26,27,37,41,45 In considering the potential use of such defects in catalytic processes, it is important to note that some control of the presence of different types of defects is possible by varying, for example, pressure25 and sample preparation.28 It is interesting to consider the subsurface defects in particular since the surface defects, particularly oxygen vacancies, are more vulnerable to poisoning by water, oxygen, and other species. There is precedent in the literature for both enhanced and unique reactivity of other types of molecules on defect-rich titanium dioxide surfaces. Reactivity has been observed which involves surface oxygen vacancies and subsurface interstitials. For example, Farfan-Arribas and Madix et al. found that methanol was either dehydrogenated to form formaldehyde or deoxygenated to form methane over stoichiometric and defective TiO2(110) surfaces, respectively. The fact that both of these pathways are available depending on the surface state highlights the importance of control and understanding of defect chemistry. The defective surface was more reactive

Figure 9. X-ray photoelectron spectra of the C 1s region following exposure of a bombarded TiO2(110) surface to a multilayer dose of thiophenol at 90 K, followed by subsequent heating to the above temperatures and the reacquisition of spectra.

Following desorption of the multilayer upon heating to 220 K, the peak decreased by a factor of 2. The C 1s signal retained ∼70% of its area upon further heating to 600 K, greater than that of the annealed surface, suggesting greater decomposition, consistent with the higher surface area, more defective surface. Heating to 800 K further decreased the C signal. The S 2p region appears similar to that of the annealed surface following a multilayer exposure at 90 K (Figure 10), attenuating in the same manner upon heating to 220 K. Importantly, however, a distinct shift in the S 2p peak of ∼1.6 eV from 163.7 to 162.1 eV occurs upon heating to 600 K, concurrent with desorption of benzene from the surface in this temperature range (Figure 6). Sulfur peaks in this region are attributed to atomic S on Ti rows on the surface.21,43 This atomic sulfur feature persists upon heating to 800 K. We attribute the presence of residual sulfur on the surface at high temperatures and the desorption of benzene to the formation and subsequent decomposition of phenylthiolate on the bombarded surface. Thiolate formation would require the dissociation of the S−H bond. Upon heating, the C−S bond must break, and transfer of H to the phenyl fragment would produce benzene. Such a mechanism is consistent with previous observations of methanethiol, which was shown by Liu et al. to decompose over a similarly defective TiO2(110) surface.21,40 In the methanethiol work, decomposition of thiolates led to the desorption of methane as well as the deposition of surface bound S and CHx. The authors of this work also observed that the amount of methane was proportional to the defect concentration: increasing bombardment time led to increased methane production. We propose a similar mechanism here; 10214

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

(5) Knudsen, K. G.; Cooper, B. H.; Topsoe, H. Catalyst and Process Technologies for Ultra Low Sulfur Diesel. Appl. Catal., A 1999, 189, 205−215. (6) Shorey, S. W.; Lomas, D. A.; Keesom, W. H. Use FCC Feed Pretreating Methods to Remove Sulfur. Hydrocarbon Process. 1999, 78, 43−51. (7) Pieplu, A.; Saur, O.; Lavalley, J. C.; Legendre, O.; Nedez, C. Claus Catalysis and H2S Selective Oxidation. Catal. Rev.Sci. Eng. 1998, 40, 409−450. (8) Bej, S. K.; Maity, S. K.; Turaga, U. T. Search for an Efficient 4,6DMDBT Hydrodesulfurization Catalyst: A Review of Recent Studies. Energy Fuels 2004, 18, 1227−1237. (9) Hernandez-Maldonado, A. J.; Yang, R. T. Desulfurization of Transportation Fuels by Adsorption. Catal. Rev.Sci. Eng. 2004, 46, 111−150. (10) Yang, R. T.; Hernandez-Maldonado, A. J.; Yang, F. H. Desulfurization of Transportation Fuels with Zeolites Under Ambient Conditions. Science 2003, 301, 79−81. (11) Jeevanandam, P.; Klabunde, K. J.; Tetzler, S. H. Adsorption of Thiophenes Out of Hydrocarbons Using Metal Impregnated Nanocrystalline Aluminum Oxide. Microporous Mesoporous Mater. 2005, 79, 101−110. (12) Ma, X.; Lu, S.; Song, C. A New Approach to Deep Desulfurization of Gasoline, Diesel Fuel and Jet Fuel by Selective Adsorption for Ultra-Clean Fuels and for Fuel Cell Applications. Catal. Today 2002, 77, 107−116. (13) Samokhvalov, A.; Duin, E. C.; Nair, S.; Tatarchuk, B. J. Adsorption and Desorption of Dibenzothiophene on Ag-titania Studied by the Complementary Temperature-Programmed XPS and ESR. Appl. Surf. Sci. 2011, 257, 3226−3232. (14) Friend, C. M.; Chen, D. A. Fundamental Studies of Hydrodesulfurization by Metal Surfaces. Polyhedron 1997, 16, 3165− 3175. (15) Rodriguez, J. A.; Dvorak, J.; Jirsak, T. Chemistry of Thiophene on Mo(110), MoCx and MoSx Surfaces: Photoemission Studies. Surf. Sci. 2000, 457, L413−L420. (16) Casarin, M.; Favero, G.; Glisenti, A.; Granozzi, G.; Maccato, C.; Tabacchi, G.; Vittadini, A. An Experimental and Theoretical Study of the Interaction of CH3OH and CH3SH with ZnO. J. Chem. Soc., Faraday Trans. 1996, 92, 3247−3258. (17) Dvorak, J.; Jirsak, T.; Rodriguez, J. A. Fundamental Studies of Desulfurization Processes: Reaction of Methanethiol on ZnO and Cs/ ZnO. Surf. Sci. 2001, 479, 155−168. (18) Halevi, B.; Vohs, J. M. Reactions of CH3SH and (CH3)2S2 on the (0001) and (0001̅) Surfaces of ZnO. J. Phys. Chem. B 2005, 109, 23976−23982. (19) Halevi, B.; Vohs, J. M. TPD Study of the Reaction of CH3CH2SH and (CH3CH2)2S2 on ZnO(0001) and ZnO. Catal. Lett. 2006, 111, 1−4. (20) Halevi, B.; Vohs, J. M. Site Requirements for the Reactions of CH3SH and (CH3)2S2 on ZnO(0101̅). Surf. Sci. 2008, 602, 198−204. (21) Liu, G.; Rodriguez, J. A.; Chang, Z.; Hrbek, J.; Gonzalez, L. Adsorption of Methanethiol on Stoichiometric and Defective TiO2(110) Surfaces: A Combined Experimental and Theoretical Study. J. Phys. Chem. B 2002, 106, 9883−9891. (22) Liu, G.; Rodriguez, J. A.; Hrbek, J.; Long, B. T.; Chen, D. A. Interaction of Thiophene with Stoichiometric and Reduced Rutile TiO2(110) Surfaces: Role of Ti3+ Sites in Desulfurization Activity. J. Mol. Catal. A 2003, 202, 215−227. (23) Ozturk, O.; Park, J. B.; Black, T. J.; Rodriguez, J. A.; Hrbek, J.; Chen, D. A. Methanethiol Chemistry on TiO2 Supported Ni Clusters. Surf. Sci. 2008, 602, 3077−3088. (24) Jaffe, S. B.; Freund, H.; Olmstead, W. N. Extension of StructureOriented Lumping to Vacuum Residua. Ind. Eng. Chem. Res. 2005, 44, 9840−9852. (25) Bak, T.; Nowotny, J.; Rekas, M.; Sorrell, C. C. Defect Chemistry and Semiconducting Properties of Titanium Dioxide: II. Defect Diagrams. J. Phys. Chem. Solids 2003, 64, 1057−1067.

overall, as a greater fraction of adsorbed methanol was converted to reaction products on the defective surface. In these studies, electron bombardment was employed rather than inert ion bombardment as this technique produces mainly surface defects, likely surface oxygen vacancies. An interesting study of formic acid on the (1 × 2) reconstructed rutile TiO2(110) surface demonstrated that subsurface interstitials migrate to the surface upon heating to react with formic acidderived oxygen, leading to (1 × 1) island regrowth within the (1 × 2) terraces.46 Clearly, defects play an important role in the surface chemistry of titanium dioxide. The ability to understand and control defect roles and levels, as also demonstrated here for larger S-containing molecules, is therefore highly desirable. In summary, we have examined the interaction of four organosulfur species with annealed and bombarded TiO2(110) surfaces. Surface binding occurs via the sulfur atoms in most cases, except for thiophene, in which case binding occurs through the aromatic ring. Moderate desorption activation energies (∼100 kJ/mol) result from the interaction of available electrons on S, while a weak desorption activation energy occurs (∼65 kJ/mol) upon binding via the aromatic ring of thiophene. The fact that thiophenol was the only species studied here capable of reacting with the significantly reduced, bombarded TiO2(110) surface points to the stability gained when S−C bonds are present in petroleum-relevant compounds as compared to S−H bonds. Thus, the challenge lies in the activation of these S−C bonds and in the selective binding of such thioethers. The experiments presented here on the reduced TiO2(110) surface lay the groundwork for future studies of supported active clusters in the selective removal, via strong adsorption or reaction, of refractory organosulfur compounds.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Acknowledgement is made to the Donors of the American Chemical Society Petroleum Research Fund for partial support of this research (#50385-UNI5). Additional financial support was provided by the University of San Diego and the Henry Luce Foundation’s Clare Boothe Luce Program. We also thank David Malicky for assistance in the construction of the directed doser used in this work.

■

REFERENCES

(1) Satterfield, C. N. Heterogeneous Catalysis in Industrial Practice; McGraw-Hill: New York, 1991. (2) Speight, J. G. The Chemistry and Technology of Petroleum; Marcel Dekker: New York, 1999. (3) Mason, B. J. Acid Rain: Its Causes and Its Effects on Inland Waters; Clarendron Press: New York, 1992. (4) Haelsig, C.-P.; Turner, J. What are the Experiences Gained from Clean Fuels Projects? Hydrocarbon Process. 2006, 85, 41−48. 10215

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216

Langmuir

Article

(26) Benz, L.; Haubrich, J.; Quiller, R. G.; Friend, C. M. Acrolein Coupling on Reduced TiO2(110): The Effect of Surface Oxidation and the Role of Subsurface Defects. Surf. Sci. 2009, 603, 1010−1017. (27) Benz, L.; Haubrich, J.; Quiller, R. G.; Jensen, S. C.; Friend, C. M. McMurry Chemistry on TiO2(110): Reductive CC Coupling of Benzaldehyde Driven by Titanium Interstitials. J. Am. Chem. Soc. 2009, 131, 15026−15031. (28) Diebold, U. The Surface Science of Titanium Dioxide. Surf. Sci. Rep. 2003, 48, 53−229. (29) Brinkley, D.; Engel, T. Evidence for Structure Sensitivity in the Thermally Activated and Photocatalytic Dehydrogenation of 2Propanol on TiO2. J. Phys. Chem. B 2000, 104, 9836−9841. (30) Weast, R. C. CRC Handbook of Chemistry and Physics, 62nd ed.; CRC Press, Inc.: Boca Raton, FL, 1981−1982. (31) Redhead, P. A. Thermal Desorption of Gases. Vacuum 1962, 12, 202−211. (32) Henderson, M. A. Acetone Chemistry on Oxidized and Reduced TiO2(110). J. Phys. Chem. B 2004, 108, 18932−18941. (33) Laurence, C.; Gal, J. F. Lewis Basicity and Affinity Scales: Data and Measurement; John Wiley & Sons Ltd.: Chinchester, UK, 2010. (34) Lang, J. F.; Masel, R. I. The Adsorption of Thiophene and Tetrahydrothiophene on Several Faces of Platinum. Surf. Sci. 1987, 183, 44−66. (35) Suzuki, S.; Yamaguchi, Y.; Onishi, H.; Sasaki, T.; Fukui, K.; Iwasawa, Y. Study of Pyridine and its Derivatives Adsorbed on a TiO2(110)-(1 × 1) Surface by Means of STM, TDS, XPS and MD Calculation in Relation to Surface Acid-Base Interaction. J. Chem. Soc., Faraday Trans. 1998, 94, 161−166. (36) Ko, E. I.; Benziger, J. B.; Madix, R. J. Reactions of Methanol on W(100) and W(110)-(5 × 1)C Surfaces. J. Catal. 1979, 62, 264−274. (37) Lira, E.; Wendt, S.; Huo, P. P.; Hansen, J. O.; Streber, R.; Porsgaard, S.; Wei, Y. Y.; Bechstein, R.; Laegsgaard, E.; Besenbacher, F. The Importance of Bulk Ti3+ Defects in the Oxygen Chemistry on Titania Surfaces. J. Am. Chem. Soc. 2011, 133, 6529−6532. (38) Roberts, J. T.; Friend, C. M. Spectroscopic Identification of Surface Phenyl Thiolate and Benzyne on Mo(110). J. Chem. Phys. 1988, 88, 7172−7180. (39) Wagner, C. D.; Riggs, W. M.; Davis, L. E.; Moulder, J. F.; Muilenberg, G. E. Handbook of X-ray Photoelectron Spectroscopy; Perkin-Elmer Corporation: Eden Praire, MN, 1979. (40) Rodriguez, J. A.; Dvorak, J.; Jirsak, T. Chemistry of SO2, H2S, and CH3SH on Carbide-Modified Mo(110) and Mo2C Powders: Photoemission and XANES Studies. J. Phys. Chem. B 2000, 104, 11515−11521. (41) Benz, L.; Haubrich, J.; Jensen, S. C.; Friend, C. M. Molecular Imaging of Reductive Coupling Reactions: Interstitial-Mediated Coupling of Benzaldehyde on Reduced TiO2(110). ACS Nano 2011, 5, 834−843. (42) Davis, L. E.; MacDonald, N. C.; Palmberg, P. W.; Riach, G. E.; Weber, R. E. Handbook of Auger Electron Spectroscopy; Physical Electronics Industries, Inc.: Eden Prarie, MN, 1976. (43) Rodriguez, J. A.; Hrbek, J.; Dvorak, J.; Jirsak, T.; Maiti, A. Interaction of Sulfur with TiO2(110): Photoemission and DensityFunctional Studies. Chem. Phys. Lett. 2001, 336, 377−384. (44) Wang, L.; Baer, D. R.; Engelhard, M. H.; Shultz, A. N. The Adsorption of Liquid and Vapor Water on TiO2 Surfaces: The Role of Defects. Surf. Sci. 1995, 344, 237−250. (45) Henderson, M. A. A Surface Perspective on Self-Diffusion in Rutile TiO2. Surf. Sci. 1999, 419, 174−187. (46) Bennett, R. A.; Stone, P.; Smith, R. D.; Bowker, M. Formic Acid Adsorption and Decomposition on Non-stoichiometric TiO2(110). Surf. Sci. 2000, 454, 390−395.

10216

dx.doi.org/10.1021/la301811q | Langmuir 2012, 28, 10209−10216