Faraday Discuss. Chem. SOC., 1990, 89, 77-89 Sulphur-induced Structural Chemistry of Oxide Surfaces Christopher A. Muryn, Douglas Purdie, Peter Hardman, Allen L.. Johnson,? N. S. Prakash,$ Ganesh N. Raiker and Geoffrey Thornton Interdisciplinary Research Centre in Surface Science and Chemistry Department, Manchester University, Manchester M13 9PL Daniel S-L. Law SERC Daresbury Laboratory, Warrington WA4 4AD Surface EXAFS, NEXAFS and photoemission have been used to study the structure of surface phases resulting from reaction of H,S with NiO( 100) and SO2 with TiO,( 110). S K-edge SEXAFS data, in conjunction with the LEED pattern symmetry, suggest that the NiO( 100)/H2S reaction at 570 K results in reduction of the substrate to form an azimuthally aligned Ni( 100)c(2 x 2 ) s raft, with S occupying the fourfold Ni( 100) hollow site.S K-edge NEXAFS results for SO, adsorption on low-temperature TiO,( 110) identify two surface SO, species: chemisorbed SO, and SO:-. The polarisa- tion dependence of the NEXAFS suggests that the SO2 molecular plane lies close to parallel t o the surface. Photoemission data also evidence chemisorp- tion at 110 K and show that this precursor phase reacts further in a thermally activated process to form a stable surface sulphate-like species. A mechanism is suggested which involves SO, bonding to a terrace Ti site, with an activated hop t o an adjacent site in which sulphur bonds to two oxygen atoms on the raised row. 1. Introduction Although the reactivity of transition-metal oxide surfaces has received increased attention in recent years, the underlying physics and chemistry remain poorly undzrstood.’ In particular, the lack of structural information about clean substrates and adsorbate complexes provides a major hinderence to theoretical studies aimed at understanding factors which influence the reactivity.With the limited information available, it has been thought that surface defects in the form of steps and oxygen vacancies play a major role. More recent work on well defined single-crystal surfaces has borne out this idea in some cases; for instance we found H,O dissociation on SrTi0,(001) to be catalysed by step sites,’ and steps on a TiO,( 110) surface have been found to activate SO, dissociation.3 There are, however, examples of reactions in which defects do not play a significant part, such as in the reaction of HzO with TiOJ In this work our interest lies in the structural chemistry associated with the reaction of S-containing molecules with transition-metal oxides.Sulphur poisoning of oxide and oxidised metal catalysts and catalyst supports, and the potential use of oxide gas sensors provide a motivation for these studies over and above their intrinsic interest. Previous work in this area includes X-ray photoelectron spectroscopy (XPS), ultraviolet photoelectron spectroscopy (UPS) and low-energy electron diffraction (LEED) studies of SO, reaction with TiO,( 1 10),3*5 a-Fe_703(0001),h Ti,03( lOl2),’ and an LEED study ? Present address: Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 :I: Present address: Department Chemie Appliquee et Genie Chimique, CNRS UA417, 43 Bd du 1 1 1 EW.Novembre 191 8, F-69622 Villeurbanne Cedex, France. 7778 Sulphur-induced Structural Chemistry of Oxide Surfaces of the reaction of H,S with NiO( loo).' The chemistry observed is, as might be expected, strongly substrate dependent. On Ti203( 10i2), SO2 oxidises the substrate to a mixture of TiS2 and TiO,;' a-Fe203(0001) forms a sulphate-like adsorbate complex;' and on room-temperature Ti02( 110) there have been reports of both minimal reaction' and reaction to form a surface sulphite species.' Here we focus on the reaction of NiO( 100) with H2S, and the reaction of SO2 with Ti0,(110). In the case of the TiO1(110)-SO2 reaction we explore to a limited extent the influence of 0 vacancies on the reactivity.We employ three synchrotron radiation techniques: surface extended X-ray absorption fine structure ( SEXAFS);9 near edge X-ray absorption fine structure (NEXAFS);"' and photoemission. The first is used to determine the surface structure resulting from the H2S surface reduction of NiO( 100); NEXAFS is used to fingerprint the adsorbate complexes formed on reaction of SO2 with Ti02( 110); and valence-region photoemission is used to follow the temperature dependence of this reaction. 2. Experimental X-ray absorption measurements employed a Ge( 11 1) crystal pair in the monochromator of station 6.3 at the SRS, Daresbury Laboratory, giving an energy range 2000 d hv/eVJ- d 8400.'' During this work the base pressure of the instrument was ca.4 x loplo mbar. At the S K-edge (2472eV) the monochromator resolution is ca. 3 eV full width at half maximum (FWHM) and the degree of polarisation is close to 100%. A double-pass cylindrical mirror analyser (CMA) (Physical Electronics Inc.) was used to record the yield of S KLL Auger electrons (kinetic energy = 2100 eV) as a monitor of the surface absorption coefficient above the S K-edge. Using this Auger yield method necessarily limits the data range to ca. 150eV above the S K-edge, because at this photon energy the substrate 0 1s photoelectron peak enters the analyser energy-window. Normalisation to the incident X-ray flux was accomplished by measuring the drain current from a thin Al foil placed between the monochromator and the sample.The photoemission measurements were performed using the grazing incidence mono- chromator ( 3 0 s hv/eVc 200) and CMA on station 6.1 at the SRS," the instrument base pressure during this work being ca. 1 x 10p'O mbar. The valence-band photoemission experiments were carried out at 110 eV, the combined resolution (monochromator+ analyser) being ca. 0.7 eV (FWHM). To prevent sample charging effects, the rutile TiQ2 sample had been pre-reduced so that it was blue-grey in colour, indicating an n-type carrier concentration of about 10" cmp3. Ti0,(1 10) surfaces were prepared in si?u by cycles of Ar' bombardment (800 eV, 17 PA, 5 min) and annealing at 925 K for 20 min. The surface stoichiometry could be restored by additional annealing in 10p6mbar O2 (BOC, research grade) at 925 K for 20 min.The success of this procedure was determined by noting the lack of band-gap photoemission," although this method could not be used in the NEXAFS experiments. NEXAFS experiments investigated only an 0, annealed surface, photo- emission measurements being carried out on both a stoichiometric surface and a reduced surface which contains 0 point defects." These cleaning procedures resulted in a surface contamination level below the detection level of Auger spectroscopy, and a 1 x 1 LEED pattern. SO, (K&K, 99.98%) exposure was carried out at 100 K (NEXAFS) and 110 K (photoemission). After subsequent annealing, NEXAFS spectra were recorded at 100 K and photoemission spectra were recorded at the anneal temperature. The LEED pattern was monitored during exposure in the case of the photoemission measurements, becom- ing more diffuse on adsorption.The pattern gradually returned towards that of the clean surface on annealing. .; 1 eV- 1.602 x 10 "'J.C. A. Muryn et al. 79 0 p I [oil] Fig. 1. Schematic LEED pattern resulting from reaction of HIS with NiO(100) at 570 K. The interpretation of the pattern is that given in ref. (8). (a) NiO( 100); (0) Ni( 100)c(2 x 2)s. A clean NiO( 100) 1 x 1 surface was prepared by cleaving in situ. Exposure to 2 x lo4 L of H2S (B.D.H. 99.6%) was carried out with the sample at 570 K,* the measurements being performed at 155 K. Model compound NEXAFS data shown in fig. 6 (later) were obtained using a Ge(ll1) crystal pair in the monochromator of SRS station 3.4 (SOXAFS)14 and a fluorescence detector.Except for NiS the compounds were obtained commercially and were of at least 96% purity. a-NiS was kindly supplied by Dr R. A. D. Pattrick of the Geology Department, Manchester University. Spectra were calibrated in energy by aligning the SO:- white line to 2482.7 eV." 3. Results and Discussion Ni0(100)+ H2S A number of surface phases can be obtained by reacting NiO( 100) with H2S at elevated temperatures, ultimately resulting in the formation of a surface sulphide.' Reaction at 570 K results in the LEED pattern shown schematically in fig. 1. If the pattern is taken to represent a c(2 x 2) overlayer, the overlayer principal-order beams have position vectors which are 1.19 times larger than those of the NiO(100) substrate. Since this corresponds to the ratio of NiO and Ni metal cell constants, the LEED pattern can be interpreted in terms of an azimuthally aligned Ni( 100)c(2 x 2)s overlayer.x The Auger spectrum is consistent with this interpretation.Fig. 2 shows the corresponding S K-edge SEXAFS spectrum recorded in a normal incidence geometry after background subtraction and normalisation of the data to the edge jump. The experimental EXAFS oscillations x ( k ) with a k' weighting and their Fourier transform are shown as solid lines in fig. 2. EXAFS spectra were analysed using a curve-fitting routine based on the rapid curved-wave computational scheme.16 The phase shifts used in the data analysis were obtained by adjusting theoretical S-Ni phase shifts to fit a S K-edge EXAFS spectrum of bulk NiS, for which the bond distances are known accurately." This model compound data set was obtained by Brennan et a/." and used in their SEXAFS study of Ni( 100)c(2 x 2)S, and subsequently by us in a SEXAFS study of Ni( 110)c(2 x 2)s." The best theoretical fit to the SEXAFS data is shown in fig.2 as dashed lines. This fit was achieved with ,a single-shell nearest-neighbour S-"i bond distance of 2.26* 0.04A (using a 0.06 A correction to the theoretical phase shiftix) and an effective80 Sulphur-induced Structural Chemistry of Oxide Surfaces 1.0 - I 0.0 2 Y, * -1.0 0.50- - 5 4 - 0.2 5 - I I I 4 5 6 k l A ' Fig. 2. S KLL Auger yield SEXAFS spectra of NiO( 100) following H2S adsorption at 570 K and cooling t o 155 K, recorded at normal incidence. The EXAFS function x ( k ) weighted by k 2 (solid line) and the best theoretical fit (dashed line) are compared in the upper part of the figure.The lower section contains the modulus of the Fourier transform. coordination number of 3.8 f 1.5. It is of particular note that the analysis routine rejected models which included an oxygen backscattering atom, the presence of which would strongly influence our short k-range data. The bond distance and the coordination number results are consistent with those found in a SEXAFS study of Ni( 100)c(2 x 2)S, where S was found to occupy the fourfold hollow site: S- Ni bond distance 2.23 f 0.02 A; effective coordination number 3.94 f 0.75." This suggests the model shown in fig. 3 for the surface structure resulting from the NiO( 100)/H2S reaction at 570 K, where S lies in the fourfcdd hollow of an Ni(!OO) plane in an Ni(lOO)c(2x2)S raft.As for the thickness of the raft, this cannot be obtained from our limited SEXAFS data. However, preliminary Stacding X-ray Wavefield (SXW)'" data point to a single Ni layer thickness of the raft. An interesting comparison can be made between this result and a surface structure obtained on reacting H,S with Ni( 111). In the latter case, reaction at 440 K produces a (5J3 x 2 ) s overlayer, which SEXAFS data" show to have the pseudo-Ni( 100)c(2 x 2 ) s structure shown in fig. 4. In this case the S-Ni bond distance is 2.27i0.02 A.'' It seems that the Ni( 100)c(2 x 2)s structure offers particular stability, providing the thermodynamic driving force for the reaction.Hence, it is not surprising that the H2S reduction of NiO( 100) would limit to one monolayer given the relatively mild reaction conditions employed. Since this overlayer is incommensurate with the substrate,C. A. Muryn et al. 81 Fig. 4. The pseudo-Ni( 100)c(2 x 2 ) s overlayer model of Ni( 11 1 ) ( 5 J 3 x 2).*’ Ni( 11 1) substrate atoms are shown as open circles, ‘surface’ Ni atoms as grey and S atoms as black. azimuthal alignment could arise from pinning of the structure at steps, as suggested for dioxygen physisorbed on the basal plane of graphite.22 This behaviour differs from the (5J3 x 2)s overlayer on Ni( 11 I), where a relatively small super-cell can be formed by a distortion of the overlayer from fourfold symmetry. Ti02( 110) + SO2 The results of previous work on this system are to some extent contradictory.In a room-temperature adsorption study using UPS, XPS and LEED, Smith et ~ 1 . ~ found essentially no reaction with the nearly perfect surface. This is in contrast with the conclusions reached in a similar study, where surface sulphite (SO:-) was evidenced following room-temperature adsorption.’ Formation of a surface sulphide phase was detected when SO2 was reacted with a defected surface created by Ar+ b~mbardrnent.~ In contrast to the earlier work, we have concentrated on lower temperature adsorption and the effect of thermal activation. S K-edge NEXAFS spectra recorded at a 20” (grazing) angle of incidence are shown in fig. 5 . Condensing SO2 at 100 K gives rise to NEXAFS identical to that obtained in the gas-phase spectrum, from which the two major features observed can be assigned to an S 1s -+ TT* (2473 eV) bound-state resonance and a (T* (2478 eV) shape resonance.23 On annealing the sample incrementally, the SO2 features decrease in intensity and appear to be replaced by similar peaks shifted to ca.1-2 eV lower photon energy (the (T* component is observed at normal incidence) and a new resonance at 2482 eV. At higher anneal temperatures the ‘shifted SO, features’ decrease in intensity leaving the 2482 eV resonance. Peaks due to condensed SOz appear Fig. 3. The structural model for NiO( 100) + HIS at 570 K derived from SEXAFS data and the LEED pattern. In this model substrate Ni atoms are represented by the small shaded circles, 0 atoms by the lightly shaded large circles.Ni atoms in the Ni( 100) plane are darkly shaded and S atoms are black.82 Sulphur-induced Structural Chemistry of Oxide Surfaces 1 7-- -i T-1 -~ --i -__I- -2 247 0 2490 2510 photon energy/eV Fig. 5. S KLL Auger yield NEXAFS spectra of TiO,( 110) following SO2 adsorption at 100 K and annealing to the indicated temperatures. All measurements were recorded with the sample at 100 K and at 20" (grazing) angle of incidence. in all spectra because of re-adsorption from the residual vacuum at the measurement temperature (100 K). A significant contribution to the shifted SO, features will also arise from re-adsorption. We can identify the surface species responsible for the 2482 eV resonance by com- parison with S K-edge NEXAFS spectra of model compounds shown in fig.6. The Na,S,O, spectra are consistent with those previously reported by Sekiyama et ~ l . , ~ ~ differing only in the relative intensities and a 2 eV shift of the energy scale. The latter arises from the use in the earlier work of NiS to calibrate the energy scale. Sekiyama et al.'4 interpret the rich structure of these spectra in terms of bound-state transitions to antibonding orbitals derived from S 3p, and d-type shape resonances, through which a chemical-shift trend can easily be discerned. When compared with our 500 K data in fig. 5, the model compound spectra clearly fingerprint sulphate (SO: ) species as the origin of the 2482 eV resonance. Although the two features between 2490 and 2500 eV in the Na2S04 spectrum are not reproduced in the surface spectrum, this region of the spectrum matches that of single-crystal CuSO, ." The SO, features which are shifted to lower photon energy of the condensed SO, peaks indicate the presence of a chemisorbed SO2 species with a longer S-0 bond distance than that in the gas phase.25 These peaks are obscured by the condensed SO, features in the spectra recorded after lower-temperature anneals.However, in the 300 K anneal spectrum the u* peak is sufficiently clear to allow its polarisation dependence to be investigated. Spectra recorded at X-ray incidence angles of 20" and 90" with the photon E-vector in the [OOl] azimuth are shown in fig. 7. In order to clarify the polarisation dependence, we have attempted to subtract the background due to the SO: 2482 eV resonance.Although this procedure is not likely to be particularly accurate,C. A. Muryn et al. 83 / // I I 2480 2 500 2520 photon energy/eV Fig. 6. S K-edge NEXAFS spectra of ( a ) Na2S0,, ( b ) Na,SO,, ( c ) Na2S20, and ( d ) a-NiS recorded at ca. 120 K. the data clearly indicate that the higher-energy resonance is considerably reduced at grazing incidence. We can first interpret this data on the basis of the commonly employed building-block model of molecular resonances.Io This would consider the low-( high)- energy near edge features to be due to scattering into S-0 T* (S-0 a*) final-state resonances. In this model a transition into T* (a*) from S 1s is allowed if a component of the E-vector lies perpendicular (parallel) to the molecular bond axes. On this basis the a* polarisation dependence indicates that the SO2 molecular plane is close to parallel to the surface.The building-block model can give rise to errors;'6 the data are better analysed in terms of excitation into SO, molecular orbitals and by employing standard group-theory techniques. The gas-phase SO, NEXAFS spectrum has been assigned in these terms by comparison with electronic structure c a l c ~ l a t i o n s . ~ ~ This identifies the 2473.8 eV peak with excitation into the 3b, virtual orbital, with the 'a*' feature comprising two peaks due to excitation into 9a, and 6b, lying at 2478.4 and 2478.9 eV, respectively. The appropriate C2\. molecular coordinate system has the x axis perpendicular to the SO2 plane, with the z axis parallel to the C', rotation axis.In C2\, which we will assume to be the effective symmetry of the surface species, x, y and z transform as B , , B2 and A , , respectively. From group theory the small 9a, and 6b, intensity at grazing incidence (20") (fig. 7 ) indicates that in this geometry the E-vector must lie close to x, i.e. perpendicular to the molecular plane. At normal incidence we expect to observe the 9a, and/or the 6b, transition, depending on whether the molecule is azimuthally ordered on the surface. The width of the 2475eV peak in fig. 7 compared with the 9a,/6b2 doublet of condensed SO, indicates that only one component is present. However, this84 Sulphur- induced Structural Chemistry of Oxide Surfaces Fig. 7. Polarisation dependence of the S K-edge NEXAFS spectra recorded from Ti02( 1 10)-S02 at 100 K after annealing to 300 and 500 K.Data are shown for angles of X-ray incidence of 20" and 90" with the E-vector in the [OOl] azimuth, the data being normalised to the edge step. The lower part of the figure shows the 300 K annealed data after removing a fourth-order polynomial background fitted to the region 2461-2481 eV. Fig. 8. Model of Ti02( 110) with the suggested site of SO2 chemisorption shown on the left and surface sulphate on the right. The black circles represent Ti atoms, large circles represent 0 atoms with the open large circles being the bridging oxygen atoms of the raised row. The hatched circles represent SO2. component cannot be identified in the absence of normal incidence data recorded with the E-vector in the [ l i O ] azimuth.If the molecule was parallel to the surface then no 3b, contribution is expected at normal incidence. That there is indicates that the molecule is tilted upwards from the surface in a geometry such as that shown in fig. 8. The anomalously large 3b, contribution in this geometry (it should have a c o d 8 dependence, where 8 is the angle between the E-vector and x ? ~ ) may well arise from a condensed SO? peak contribution. Having gleaned an idea of the orientation of SO2 on Ti02( 1 lo), albeit in the presence of a co-adsorbed SO:- phase, we can consider possible adsorption sites. In principle, SO2 could bond either to the fivefold coordinate Ti site on the terrace or to the terraceC. A. Muryn et al. 85 or bridging oxygen atoms (see fig.8). However, the minor perturbation apparent in the NEXAFS suggests the Ti site since SO, bonding to 0 atoms would be expected to result in the formation of a complex anion, which as we have seen above have very different NEXAFS signatures. Although there is no direct analogy in the molecular coordination chemistry of SOz,28 the surface geometry proposed probably corresponds to a type of '77'-pyramidal' coordination, as shown in fig. 8. A similar SOz coordination has been proposed for Ag(l10).29 In this bonding configuration SO, would bond to a Ti atom through the S atom, the M-S bond lying along the surface normal. In transition-metal square-planar complexes which incorporate SO, in this fashion, the bond angle between M-S and the sum of the S - 0 vectors is typically 120"." With this type of SO, coordination on TiO,( 1 lo), the molecular plane would be tilted at 30" up from the surface.On the basis of ionic bonding of SO:- to the surface, the 2482 eV resonance of the surface sulphate, which would correspond to an S 1s- t? transition," is not expected to show a polarisation dependence even with the 'adsorbate' oriented on the surface. In a tetrahedral geometry all three Cartesian axes are identical, transforming as t2, which gives rise to the polarisation independence. Covalent bonding will lower the symmetry, for instance bidentate bonding of SO:- to Ti will reduce the symmetry at least to CZb, which will split the t: resonance and introduce a polarisation dependence. Assuming the presence of an ordered SO:- species on the surface, a close to Td geometry is suggested by the polarisation independence of the 500 K annealed TiO,( 1 lO)-SO, spectra shown in fig.7. Although the NEXAFS cannot help us in identifying the SO:- adsorption site, it seems likely that it involves incorporation of SO, into a bridge site on the raised row of 0 atoms, as shoown in fig. 8. Assuming bulk termination, the 0-0 distanceoon the raised row is 2.96 A, somewhat larger than that in uncoordinated SO:-, 2.43 A.30 This difference could be accommodated by moving together the two raised 0 atoms bound to SOz, putting an upper limit of the SO%- occupation on the raised row of one half, which corresponds to a coverage of 1/6 of the surface. The edge jump observed in the '500 K annealed' NEXAFS spectrum is consistent with this figure.Valence region photoemission data for low-temperature SO, adsorption on both a reduced and stoichiometric TiOJ 1 10) surface are consistent with molecular adsorption on a Ti site, with SO, bonding through S. Fig. 9 shows photoemission spectra of TiO,( 110) before and after adsorption of 100 L of SO2 at 110 K on the stoichiometric surface, along with the difference spectrum and the gas-phase He" ( h v = 40.8 eV) spectrum.3'.32 A band-bending correction was not needed to align the surface spectra. Essentially the same result is achieved on the reduced surface (see fig. 10) except that the additional Ti 3d contribution to the valence-band and band-gap regions are quenched, giving rise to band bending and indicating preferential bonding of SO, to the 0 vacancy sites.Aligning the difference and gas-phase spectra at the 0 lone pair la2/5b, peak indicates a relaxation/polarisation shift of 2.1 eV to lower ionisation energy, and a bonding shift of 0.4eV for the 2bI/7a,/4b, features with the same shift for 6al. These shifts are consistent with the orbital the 2b,/7al/4b2 set containing a significant contribution from the S 3p orbitals, which would form CT and T bonds with substrate Ti atoms. The 6aI is the S 3s-0 2s antibonding orbital, the S 3s component of vrhich could provide additional CT bonding to the surface. This feature is larger relative to the other components in the 110 eV difference spectrum, as expected from the variation in the S 3s and 0 2p photoionisation cross-section.'3 The 0 2p,-containing 8a, orbital is probably represented by the shoulder at ca.6 eV in the difference spectrum, although its position is difficult to determine accurately in the presence of the 5bz/la, features. The feature in the difference spectrum at 4.1 eV binding energy has no counterpart in the gas-phase spectrum and probably evidences formation of surface sulphate species. Annealing both SO,-adsorbed surfaces results in a clear change to the difference spectra. Fig. 10 shows the results of annealing the initial SO, phase on the reduced86 Sulphur-induced Structural Chemistry of Oxide Surfaces '1 a2:5b, I I r~~ -~1 .----.T--d 30 20 10 EF binding energy/eV Fig. 9. Photoelectron spectra ( h v = 110 eV) of ( b ) clean stoichiometric Ti02( 110) at 110 K, ( a ) the clean surface dosed with 100 L SO, at 110 K, (c) the difference of the dosed and clean spectra, and ( d ) an He" ( h v = 40.8 eV) spectrum of gas-phase SOz aligned at the la2/5b7 peak position.surface to 150 K. The data are presented in the form of difference spectra, where we have simplified the spectra by subtracting a spectrum of the clean stoichiometric surface rather than that of the reduced surface. This only has the effect of elivinating the substrate-related changes described above and removing the need for a band-bending correction t o the alignment of data. The difference spectra in fig. 10 are compared with XPS ( h v = 1486.6 eV) spectra of Li,S04 and Li2S03 34 and an He" spectrum of Li2S04." Being guided by the NEXAFS data, we will interpret the 150 K difference spectrum in terms of a surface SO:--like species. However, we cannot rule out the possibility that the different reaction conditions employed in the NEXAFS and photoemission experi- ments give rise to different products.This is a point to s t r w s i n c e the difference between the valence region spectra of SO: and SO: is sufficiently small to prevent a positive assignment on the basis of these spectra alone. Indeed, the match between the surface spectrum and that of 'ionic' SOf- is better than that for sulphate. Aligning the difference spectra at the 2t2 S - 0 bonding-orbital peak of SO: ' 5 ~ 2 h is the only arrangement which does not give a negative bonding shift for this orbital. This corresponds to a negligible polarisation/relaxation shift when compared with the He" spectrum of Li2S04," and is the alignment employed for sulphate on ~ - F e ~ 0 , ( 0 0 0 1 ) .~ On this basis the I t , mainly 0 2s orbital experiences a bonding shift of ca. 0.9 eV, in line with our expectation that the negative charge on the 0 atoms bound to Ti will be less than in an ionic sulphate. Similar arguments can be used to explain the 1.1 eV bonding shift of the 0 2s S 3s, 3p 2a, orbital. Unfortunately, there does not appear to be a published theoretical analysis of the effect on the electronic structure of the covalent coordination of SO: , such as that carried out for the NO, ion." It is therefore difficult to make a detailed assignment ofC. A. Muryn et al. 87 30 20 10 E F binding energy/eV Fig. 10. The photoemission ( h v = 110 eV) difference spectrum (S02-clean) for a reduced Ti02( 110) surface ( a ) after exposure to 200 L at 100 K ( b ) after warming to 150 K (see text for details).Surface spectra are compared with suitably aligned XPS ( h v = 1486.6 eV) spectra of ( d ) Li2S04 and ( e ) Li2S03 from ref. (34) and ( c ) an He" ( h v = 40.8 eV) spectrum of a Li2S0, from ref. (35). The orbital assignments are from ref. (36). discrepancies between the difference spectra and the spectrum of 'isolated' SO:-. We would, for instance, expect 3d-orbital density to be introduced by adsorption since formally two electrons are transferred to the surface as SO2 is oxidised to SO:-. This charge can be nominally assigned to create one Ti3+ species and two Ti3.'+ species per SO:- moiety created. With saturation coverage of SO:-, all Ti atoms under the raised rows would be in oxidation state 1 1 1 .At least one of the low binding-energy peaks should carry this 3d density since it does not go into band-gap states. Otherwise, one is forced to consider a model in which raised row 0 atoms carry a formal charge of -1. Annealing the surface to higher temperatures simply causes the adsorbate-induced features to decrease in intensity; no further change to the shape of the spectrum is observed above 150 K. The intensities of adsorbate features decreased uniformly up to room temperature, falling to ca. 50% of the 150 K values. All adsorbate features could be removed by heating to 800 K. Taken together, the NEXAFS and photoemission data indicate chemisorption at 110 K with a further, activated reaction to form a stable surface sulphate-like species.The mechanism suggested would involve SO2 bonding to a terrace Ti site, with an activated hop to an adjacent site in which sulphur bonds to two oxygen atoms on the raised row. In other words, the reaction proceeds from the left to the right of fig. 8.88 Sulphur-induced Structural Chemistry of Oxide Surfaces In the low-temperature adsorption regime examined here, the effect of oxygen vacancies on the reaction mechanism appears to be small. While this in line with results for the Ti02( and TiOJ 1 10)"-H20 reaction, there is an essential difference. Whereas SO2 quenches the defects on Ti02( 1 lo), oxygen vacancies are relatively unaffected by H 2 0 adsorption on the (100) and (1 10) surfaces.Some knowledge of the initial species formed on reacting SO2 with Ti02( 110) oxygen vacancies would provide valuable insight into this difference in behaviour. The He' difference spectrum reported for SO2 adsorption on room temperature Ti02( 110)' is not dissimilar to that of the annealed surface in fig. 10, although the He' spectrum was interpreted as indicating surface SO:- formation. Given the discussion above, it is possible that the species formed in the earlier work' was SO:-; this could be tested by NEXAFS measurements. Observation of room-temperature reaction is restricted to the work of Onishi et al.' In our work, reaction of a stoichiometric surface was found to be minimal at this temperature, consistent with the report of Smith et aL5 A possible explanation for this discrepancy lies in the method of sample preparation; Onishi et al.' annealed only in vacuum which leaves the surface with a small concentration of 0 vacancies.Such defects are known to activate H,O dissociation at room temperature, in contrast to their irlert character at 160 K.13 4. Summary This paper has described the application of synchrotron radiation techniques in the study of two oxide surface reduction reactions. A NEXAFS and synchrotron radiation photoemission study of the low-temperature Ti02( 1 lo)/ SO2 reaction indicates molecular adsorption at ca. 110 K, being a precursor to incorporation of SO2 into the bridging oxygen atoms of the raised row to form a sulphate-like species. A SEXAFS study of the Ni0(100)/H2S reaction at 570K suggests the reduction of the substrate to form an azimuthally aligned Ni( 100)c(2 x 2)s raft, with S in the fourfold hollow site of Ni( 100).In general terms, this work illustrates the potential of synchrotron radiation tech- niques in studies of the quite complicated structural chemistry asociated with oxide surface reactions. We plan future NEXAFS and SEXAFS work which will focus on a structure determination of the separated precursor SO2 and product SO:- phases on Ti02( 1 lo), and the adsorbate structure associated with 0 vacancies. We are grateful to P. A. Cox for useful discussions, R. G. Egdell for the loan of the Ti02(110) sample and to J. R. Drabble for the NiO crystal. This work was funded by the S.E.R.C. (U.K.), including the provision of studentships to C.A.M., D.P. and P.H.Additional support was provided by Johnson Matthey PIC. References 1 V. E. Henrich, in Surface and Near-Surface Chemistry of Oxide Materials, ed. J. Nowotny and L-C. Dufour (Elsevier, Amsterdam, 1988) pp. 23-60; Rep. Prog. fhys., 1985, 48, 1481. 2 N. B. Brookes, F. M. Quinn and G. Thornton, Solid State Commun., 1987, 64, 383. 3 H. Onishi, T. Aruga, C. Egawa and Y. Iwasawa, S u r - Sci., 1988, 193, 33. 4 C. A. Muryn, G. Tirvengadum, J. Crouch, D. R. Warburton, G. N. Raiker, G. Thornton and D. S-L. Law, J. Phys.: Condensed Matter, 1989, I , SB127. 5 K. E. Smith, J. L. Mackay and V. E. Henrich, fhys. Rev. B, 1987, 35, 5822. 6 R. L. Kurtz and V. E. Henrich, Phys. Rev. B, 1987, 36, 3413. 7 K. E. Smith and V. E. Henrich, fhys.Reu. B, 1985, 32, 5384. 8 A. Steinbrunn, P. Dumas and J. C . Colson, Surf: Sci., 1978, 74, 201. 9 P. H. Citrin, J. Physique, 1986, 47, C8-437. 10 J. Stohr and D. A. Outka, Phj1.Y. Rev. B, 1987, 36, 7891. 1 1 A. A. MacDowell, D. Norman and J. B. West, Ret.. Sci. Instr., 1986, 57. 2667. 12 M. R. Howells, D. Norman, G. P. Williams and J. B. West, J . Phjx E, 1978, I I , 199.C. A. Muryn et al. 89 13 R. L. Kurtz, R. Stockbauer, T. E. Madey, E. Roman and J. L. De Segovia, Surf: Sci., 1989, 218, 178. 14 A. A. MacDowell, J. B. West, G . N. Greaves and G . van der Laan, Rev. Sci. Instr., 1988, 59, 843. 15 T. A. Tyson, L. Roe, P. Frank, K. 0. Hodgson and B. Hedman, Phys. Rev. B, 1989. 39, 6305. 16 S. J. Gurman, N. Binsted and 1. Ross, J. Phys. C: Solid State 1984, 17, 143. 17 J. Trahan, R. G. Goodrich and S. F. Watkins, Phjx Rev. B, 1970, 2, 2859. 18 S. Brennan, J. Stohr and R. Jaeger, Phys. Rev. B, 1981, 24, 4871. 19 D. R. Warburton, G . Thornton, D. Norman, C. H. Richardson, R. McGrath and F. Sette, Surf Sci., 1987, 189/190, 495. 20 D. P. Woodruff, D. L. Seymour, C. F. McConville, C. E. Riley, M. D. Crapper, N. P. Prince and R. G . Jones, Phys. Rev. Lett., 1987, 58, 1460. 21 D. R. Warburton, P. L. Wincott, G. Thornton, F. M. Quinn and D. Norman, Surf Sci., 1989, 2111212, 71. 22 M. F. Toney and S. C . Fain Jr, Phys. Rev. B, 1984, 30, 1115. 23 A. P. Hitchcock, S. Bodeur and M. Tronc, Chem. Phys., 1987, 115, 93. 24 H. Sekiyama, N. Kosugi, H. Kuroda and T. Ohta, Bull. Chem. Soc. Jpn, 1986, 59, 575. 25 J. Stohr and R. Jaeger, PhJx Rep. B, 1982, 26, 4111. 26 J . Somers, A. W. Robinson, Th. Lindner and A. M. Bradshaw, Phys. Rev. B, 1989, 40, 2053. 27 S. Bodeur and J. M. Esteva, Chem. P h p . , 1985, 100, 415. 28 R. R. Ryan, G . J. Kubas, D. C. Moody and P. G. Eller, Strucr. Bond., 1981, 46, 47. 29 D. A. Outka and R. J. Madix, SurJ Sci., 1983, 137, 242. 30 A. F. Wells, Structural Inorganic Chemistry (Oxford University Press, Oxford, 5th edn, 1986). 31 I . H . Hillier and V. R. Saunders, Mol. Phys., 1971 22, 193. 32 I>. R. Lloyd and P. J. Roberts, Mol. Phys., 1973, 26, 225. 33 S. M . Goldberg, C. S. Fadley and S. Kono, J. Electron Spectrosc. Relat. Phenom., 1981, 21, 285. 34 R. Prins, J. Chem. Phys., 1974, 61, 2580. 35 J . A. Connor, I . H. Hillier, M. H. Wood and M. Barber, J. Chem. Soc., Faraday Trans. 2, 1974,6, 1040. 36 N. Kosuch, G. Wiech and A. Faessler, J. Electron Spectrow. Relat. Phenom., 1989, 20, 11. 37 A. A. MacDowell, C. D. Garner, I . H. Hillier, C. Demain, J. C. Green, E. A. Seddon and M. F. Guest, J. Chem. Soc., Chem. Cornmun., 1979, 427. Paper 0/0050l K; Received 2nd Fehruary, 1990