Ultra-violet and X-ray Photoelectron Spectroscopy (UPS and XPS) of CO, COz, 0, and H,O on Molybdenum and Gold Films BY S . J. ATKINSON, C. R. BRUNDLE AND M. W. ROBERTS* School of Chemistry, University of Bradford Received 12 June, 1974 Electron spectroscopic (XPS and UPS) studies of the interaction of CO, COz, O2 and H 2 0 with Mo, and H20 and C 0 2 with Au are reported. The object was first to explore how far one could distinguish between physical and chemisorption and between differmt states of chemisorption ; whether there was any relation between bond strength and electron binding energy and whether electron spectroscopy, particularly UPS, could provide unambiguous information on the moleculal- nature of surface species. PhysicaI adsorption data for COz and H20 on ALI at 77 K provide reference binding-energy data.Arguments are given foi believing that, with molybdenum, CO can exist in both molecular and dissociated forms, that C02 dissociates into C and 0 and that water exists in both molecular and dissociated forms. The role of the substrate temperature in determining the relative stability of the diffeient states of the molecules is emphasized. Implicit is the assumption that identical electron- binding data by XPS reflect identical surface-bonding situations. In this context the role of com- plementary UPS data is important. Whze information was available on the relative strengths of surface bonding then for any given adsorbate the electron binding energy was greatest when bonding to the surface was weakest. Electron spectroscopy is an important experimental approach for the study of adsorption at solid surfaces.There is now clear evidence for both good analytical sensitivity and the ability to obtain chemical bonding inf~rmation.”~ The question of the magnitude of electron escape-depths is now largely resolved 8a* although neither the exact dependence on electron energy nor the theoretical foundation of this dependence is completely clear.sc X-ray induced photoelectron spectroscopy (XPS) provides unambiguous informa- tion on the atomic nature of the surface, but less unambiguous information on chemical bonding since we are dependent on a change in the electron configuration of the valence electrons being reflected by a change in the binding energy of the core electron, as has been established by Siegbahn and his co-workers for gas-phase molecules.With vacuum u.-v. radiation (UPS) more direct information is obtained on the valence electrons. By combining both fmms of electron spectroscopy distinct advantages accrue over and above each separately and this is the approach adopted. We are not concerned here with electron-induced Auger spectroscopy although brief mention is made of X-ray induced Auger spectra in one or two specific instances. The systems studied are inherently simple but chosen so that they cover a range of possible surface phenomena (e.g., physical and chemisorption, multilayer formation and oxidation) and for which in many cases there is a reasonably sound background information. We believe this to be essential at this stage of development of electron spectroscopy in surface chemistry.For the carbon monoxide-molybdenum system there is clear evidence for a number 62S. J. ATKINSON, C. R . BRUNDLE A N D M. W. ROBERTS 63 of distinguishable regimes of adsorption.fo* l1 By comparison with the W + CO system,12 these “ regimes of adsorption ” may be analogous to the “ states ” invoked to explain the tungsten + CO data. The term “ states ” originates largely from flash- filament studies and any one state has implied a specific adsorption energy. On the other hand, previous work lo suggested a range of metal-CO bond energies for the Mo+CO system at temperatures below 295 K, the heat of adsorption decreasing linearly with increasing coverage. Similar behaviour has been observed for H2 desorption from dissociatively chemisorbed H2S on Fe These conclusions were based on both isotherm and desorption data for evaporated films.Flash filament studies were carried out with single crystals, ribbons, or polycrystalline wires. We use the terminology generally accepted l 2 to describe the adsorption of CQ on W but do not necessarily imply a direct similarity of surface bonding. For example “ virgin layer ” follows Gomer’s terminology for low temperature adsorption, 77 K or less ; P-state refers to the initial adsorption at room temperature (e.g., at lo-* Torr) ; the a-state is of lower heat of adsorption than the jl and occurs at higher CO pressure (e.g., Torr); y-state refers to the reversible adsorption which occurs lo. l 1 at 77 K subsequent to adsorption at higher temperature (295 K).We have reported briefly electron spectroscopy data for the Mo+CO system 4 9 ’ as also have Madey et aL5 on the W+CO system. In this paper we describe more detailed experimental work and compare the results with the Mo + 02, Mo + CO, and Mo+H20 systems. In the case of XPS we refer to three important parameters : (a) the binding energy of a core electron (b) peak width at half-maximum height (FWHM) and (c) peak intensities. Together with the information obtained from UPS (He I and He 11) studies they form the main basis of the discussion. XPS and UPS studies on the physical and multilayer adsorption (condensation) of CQ2 and H,O at 77 K on Au are also included. They provide “ standard data ” for XPS electron binding energies, and intensities, and for valence levels (UPS) in the undissociated molecules.The role of the substrate in influencing the binding energies of core electrons of the physically adsorbed layer is also explored. and N2 from W. EXPERIMENTAL The electron spectrometer incorporating both XPS and UPS facilities and the experi- mental procedures used for the preparation of Mo and Au substrates in the form of evaporated films have been described.2* l6 The Mo films were never obtained completely free of contamination, the relative peak intensities of O(1s) and C(1s) before and after an adsorption experiment indicating an initial contamination of < 15 % coverage. Specpure gases were admitted directly from bottles through the appropriate temperature traps. Calibration of the spectra was made against the Mo 3d peak of clean M o , ~ which has previously been calibrated against Au 4f,/2, at 83.7k0.1 eV below &.I6 RESULTS ADSORPTION OF co ON MOLYBDENUM ADSORPTION AT 298 K ( b STATE) It is first essential to refer to our previous results for CQ adsorbed at 298 K ; these are given in fig.1. The O( 1 s) peak height as a function of exposure is shown ; the FWHM value was constant at 2.2 eV during the formation of the ad-layer. The O(1s) peak position remained invariant during formation of the ad-layer at the same binding energy, 530eV, as the small oxygen contaminant initially present on the molybdenum film. The Mo(3q) ratio is 1 : 0.9 & 0.06 for CO exposures of cu. 10 L64 ELECTRON SPECTROSCOPIC STUDIES (fig. 1) but is about 10 % greater, 1 : 1.07 k 0.05 at much higher CO exposures ( - 400 L).As mentioned previously,’ monitoring the C( 1s) signal is not particularly satisfactory since it is superimposed on a broad characteristic loss feature (Mo 3d) which appears to be sensitive to adsorption. 1 1 1 1 1 1 1 l l I I 5 3 0 536 524 5 3 0 536 542 eV 2 4 6 8 I0 exposure/L FIG. 1.-(a) O(1s) peak as a function of exposure to CO at 298 K. (6) Mo(3s3) : O(1s) ratio as a function of exposure to CO at 298 K. (c) Formation of the yCO state at 77 K. (1) 1 L exposure ; (2) 3 L exposure ; (3) and (4) 6 and 12 L exposure. After completion of the monolayer, raising the CO pressure from -lo-’ to 1 x Torr resulted in no change in the O(ls) peak either in intensity or width. We estimate that the auerage sticking probability for CO is 0.2 at 298 K since the monolayer is virtually complete after an exposure of 5 L (fig.lb). Table 1 summarises the principal binding-energy, peak-width and peak-height data for this and other systems studied. ADSORPTION AT 77 K (a) SUBSEQUENT TO ADSORPTION AT 298 K.-We are particularly concerned with how do(es) the y state(s) form during exposure to CO and have therefore studied the development of the envelope of the binding energies which make up the y state (fig. lc). The y state is desorbed on warming to 298 K and reformed on cooling in CO to 77 K. We have monitored the O(1s) peak after CO exposures of 1, 3, 6 and 12 L ; above 6 L we observed no further change. The ratio of the areas of the B to the y envelopes was 1.9 :1, 1.8 :1 and 1.76 :1 in three separate experiments.The CO pressure was less than 1 x Torr.TABLE 1.-O(1s) AND C(1s) B.E’s, INTENSITIES AND HALF-WIDTHS FOR ADSORPTION ON Au AND Mo B.E.’s (eV) relative intensities FWHM (eV) adsorption state c (1s) 0 (1s) 0 (ls)/C (1s) 0 (ls)/Mo ( 3 ~ 4 ) c (Is) Mo (298 K)---clean CO/Mo (298 K)-P-state CO/Mo (298 K)+ CO/Mo (77 K) CO/Mo (77 K)-virgin state virgin state warmed to 298 K (P+y states) 02/M0 (298 K) heated to 420 K in Torr O2 heated to 520 K in 10-l Torr O2 Au (77 K)-clean C02/Au (77 K)-physical ads. C02/Au (77 K)-multilayers H20/Au (77 K)-multilayers C02/Mo (298 K) C02/Mo (298 K) cooled to 77 K+ COZ CO;!/Mo (77 K) then warmed to 298 K H20/M0(298 K) then cooled to 77 Kffurther H20 then heated to 523 K in vacuo H20lM0 (298 K)+Oi 284.2( @) 282.7,284.2 282.7, 287.2 f 284.6 282.7, 284.2 284.2( 0 ) 285.1 ( 0 ) 290.9 290.9(s), 292.3 285.1 ( 0 ) 284.0(0), 283.0 284.0(0), 283.0, 291.6 284.5, 291.6 283.5 530.2 530.0 530.0, 534.5 f 53 1.2 530.0 530.0 530.0 530.0 none 534.3 534.3(s), 535.6 532.7 530.0 530, 534.2 53 1 .O, 534.1 530.0 530.0, 532.3(s) 530.0(s), 532.5 530.0, 532.3(s) 530.0 2.0 2.2 ca.2.5 2.2 - large large large - 4.8 4.6 large 2.2 -, 4.4 -, 4.4 3.3 - - - - < 0.2 0.9 0.9 - - 2.5 4.8 6.0 - - - I 0.9 0.9, large - , large 1 1.64, - - , large 1.8 2.06 - 4.5 5e, 5 2.9 g 4.5 e9g - - - 2.1 2.1 2.2 - not taken not taken 3.5 3.0 - - - I 0 (1s) 2.2 2.2 2.2, >5 3.8 3.0 2.4 2.4 2.4 - 2.2 2.2 2.3 2.4 2.4, 2.4 2.4, 2.4 2.7 overlapped overlapped 2.5 overlapped 0 estimated accuracy i 0.3 eV ; b approximate ; C taken at 50 eV resolution unless otherwise stated ; d taken at a few monolayers exposure ; e FWHM of oveLIapped p s k s ;fenties of very broad flat peaks (see spectra) ; g taken at 100 eV analyzing energy ; .contaminant peak ; s shoulder.66 ELECTRON SPECTROSCOPIC STUDIES (b) ADSORPTION AT 77 K (VIRGIN STATE).-Fig.2 (a and b) shows the O(1s) and C( 1s) spectra resulting from adsorption of CO during exposure at a pressure of about 2 x The O(1s) and C(1s) spectra at zero exposure correspond to adsorp- tion on the clean film during and after formation and prior to commencing CO exposure. With increasing CO exposure both the O( 1s) and C( 1s) binding energies Torr. . . . 524 530 536 ' 5 4 0 20L 14L FIG. 2.-(u) O(1s) signal at 77 K as a function of exposure of Mo to CO.(6) C(1s) signal at 77 K as a function of exposure of Mo to CO. (c) O(1s) signal from CO(ads) layer at 77 K and during warming to 295 K. (d) C(1s) signal from CO(ads) layer at 77 K and during warming to 298 K. iiiove gradually to higher values, O( 1s) shifting about 1.2 eV and C( 1s) about 0.4 eV compared to the initial contaminant positions. The O( 1s) peak height increases with exposure until a constant height is reached after about 14 L; the FWHM value at 3.0 L exposure is about 3.0 eV but gradually broadens to a value of about 3.8 eV after 20 L. The C(ls) envelope behaves analogously, the FWHM value increasing from 1.8 to 2.9 eV after 20 L exposure. Adsorption states with much higher O( 1s) binding energies than that correspond- ing to the major adsorption also develop at high exposure (shaded area) ; this behaviour is associated with the formation of the y state(s) which are reversible l 1 and charac- terised by a low heat of adsorption.(c) THERMALLY INDUCED CONVERSION OF THE VIRGIN (77 K) STATE TO THE p (298 K) STATE.-Fig. 2(b and d ) shows the O( 1 s) and C( 1 s) spectra during the gradual warming of the ad-layer formed at 77 to 298 K. The middle O(ls) spectrum corresponds to aS . J . ATKINSON, C . R . BRUNDLE AND M. W. ROBERTS 67 temperature of about 170 K ; at this stage the adsorbed species giving rise to the high binding energy O(1s) states (cp. 7) have desorbed. The total CO coverage at 77 K is some 35 % greater than at 295 K.l0* l 1 The O(1s) peak position after warming has shifted to a lower binding energy, and its FWHM decreased (table 1).The changes observed with the C(1s) spectra are more complex (fig. 2d) in that a gradual broadening of the C(ls) envelope (in contrast to the narrowing of the O(1s) envelope) occurs with emergence of two C(1s) peaks at 282.7 and 284.2 eV. The peak at 284.2 eV is at the same value as the small C( 1s) peak associated with contami- nation present on the “ clean ” film prior to CO adsorption (fig. 2), and the 282 7 V peak is at the same position as for CO in the /I-state. UPS STUDIES OF co INTERACTION WITH M O FILMS We have reported results of UPS studies of the adsorption of CO on molybdenum films at 295 and 77 K and the changes observed on warming the adlayer from 77 to 295 K. The spectra for these states, and also the y-state, are shown in fig.3a. The I I I I I I I I l l 8 E 4 8 1 2 e v E 3 6 9 FIG. 3.-(u) UPS (Hel) of CO adsorbed on Mo. (1) Mo ; (2) CO(ads) 298 K ; (3) CO(ads) 298+ CO (ads) 77 K ; (4) CO(ads) 77 K. (6) XPS of Mo film. (1) Mo film ; (2) Mo+ CO (lo00 L) at 298 K. positions of the observed bands with respect to the Fermi level are summarised in table 2. During the UPS studies, XPS spectra were occasionally taken both to monitor the coverage and to relate unambiguously the two types of spectra. The XPS studies of the valence levels were not particularly informative. Fig. 3b shows the molybdenum valence band before and after adsorption of CO at 295 K ; the CO exposure was about 1000 L. The only notable point was a decrease in the intensity, there being a lack of structure both for the “clean” and CO covered surfaces.The CO features are presumably not observable because of high escape depths and low cross-section for ionization. INTERACTION OF OXYGEN WITH MOLYBDENUM Fig. 4(a) shows the Mo(3s+) and O(ls) peaks for the as-prepared Mo film, after exposure (20 L) at 295 K to oxygen at a pressure of -2 x Torr, at which point the sticking probability had fallen to a low value (< 1/100 of this initial value), and after increasing the oxygen pressure to lo-’ Torr. Both C and 0 were present as surface impurities in concentrations of < 15 % of a monolayer. Both the O( 1s) peak68 ELECTRON SPECTROSCOPIC STUDIES position and FWHM remained constant at 530.0+0.3 and 2.4 eV respectively, during exposure to oxygen.The Mo(3s+) :O(ls) ratio is 1 : 2.5 & 0.05 for saturation exposures at pressures up to about Torr. On increasing the pressure to 10-1 Torr the ratio decreased to 1 : 3.1. Slight broadening of the Mo(3d4,+) peaks (from 1.1 to 1.3 eV) was observed during adsorption. Repitition of the experiment at 77 K produced identical results, and no spectral changes were observed on warming to 295K. We recall that no changes in work function were observed in a similar experiment reported earlier. TABLE 2.-UPS VALENCE LEVELS a FOR C 0 2 AND H20 ADSORBED ON Au (77 K) and CO (p, y, VIRGIN STATES) ADSORBED ON Mo orbital energies/eV b CO~/AU (77 K) H e I { ~ ~ ~ ~ ~ d C02(d He1 orbital assignment 6.9; 9.6; 11.0; 13.1 8.3; 11.6; 12.3; 13.6(?) 6.8; 9.3; 11.0; 12.5 8.2; 11.4; 12.4; 13.7 13.7; 17.6; 18.0; 19.4 ( 7 d 4 ; ; (ad2 ; (d2 H20/Au (77 K) He1 6.3; 10.2; 12.6 H20(8) He1 orbital assignment CO/Mo He1 p Y virgin CO(d He1 orbital assignment 4-7 7.8; 11.2 6.7; 10.2 14.0; 16.8; 19.7 (a,)"; (nu)"; (aJ2 a All the adsorbed levels are broad.The quoted values represent the centre of the bands. b Ad- sorption system energies are ieferenced to EF of the substrate. For position relative to the vacuum level an appropriate work function term (see discussion) should be added. C ref. (31). Although from the known reactivity of molybdenum to oxygen, lattice penetration and oxidation would be unlikely at room temperat~re,~ flash filament-mass spectro- metric data with tungsten have been interpreted l8 in terms of some oxide formation, but the results are open to question.We therefore explored the possibility of " oxidation " of molybdenum at higher temperatures and pressures. First, on heating the chemisorbed layer to 523 K in a background pressure of Torr (" in uacuo ") no change occurred in either the O(1s) or Mo(3d+,+) peaks (fig. 4b). At a higher oxygen pressure (10-l Torr) and after 5 min heating at 420 K, the Mo(3d3) peak was unchanged but the Mo(3d+) peak had broadened from 1.8 to 2.0 eV (FWHM). The MO(%+) : O(1s) ratio decreased to 1 : 4.8t0.2 (cp. 1 : 3.1 at 295 K) but no change occurred in either binding energy or the width (FWHM) of the O(1s) envelope (fig. 4c). On increasing the temperature to 520 K (oxygen pressure 10-1 Torr), a further decrease (fig. 4b) in the intensity of the main Mo peaks occurred with a corresponding increase in the O( 1s) intensity, the Mo(3s3) :0( 1s) ratio approaching 1 :6.0.The Mo(3d3) was broadened even further to 2.55 eV (cp. 1.6 eV, 30 L, O2 at 235 K ; 1.8 eV 0.1 Torr 0, at 295 K) and a distinct broad feature ( N 3 eV wide) appeared at higher binding energy centred around 234 eV (fig. 4b, (iii) shaded area).S . J . ATKINSON, C. R. BRIJNDLE AND M. W. ROBERTS 69 FJG. 4 . 4 3 ) (i) Mo(3q) and O(1s) signals after 20 L exposure to O2 at 298 K. (ii) Signals after further exposure to O2 (0.1 Torr). (b) Mo(3d) peaks after (i) 520 K ‘‘in vacua" ; (ii) 420 K at 0.1 torr O2 : (iii) 520 K at 0.1 torr 02. (c) (i) O(1s) for b(i) above; (ii) O(1s) for b(ii) above ; (iii) high resolution O(ls) for b(iii) above. eV 240 200 160 eV (4 (6) FIG.5.- (a) Mo Auger signals during the interaction of Mo with O2 (1) Mo ; (2) Mo+monolayer of adsorbed oxygen ; (3) Mo+O.l Torr O2 at 573 K. (6) Mo(3d) XPS peaks for (l), (2) and (3) above.70 ELECTRON SPECTROSCOPIC STUDIES The O(ls) binding energy remained at 530 eV throughout these experiments (fig. 4c) though there was a slight broadening to high B.E. at all exposures. No additional oxygen features which might be associated with the broad additional Mo band (fig. 4b(iii)) were observed even when the O(1s) peak was examined under high resolution (fig. 4c(iii)). The X-ray induced Auger signals are shown in fig. 5 with the main Mo XPS peaks for comparative purposes. Clearly, the X-ray induced Mo(3d) peaks are more sensitive to the surface oxygen than are the X-ray induced Auger peaks in that they clearly reveal what we believe are electrons on the high energy side of the 3d peak (shaded in (3), fig.5) associated with oxidation of the metal. There are no shifts in the Auger peaks at this stage, though they become apparent at much later stages of oxidation.8b INTERACTION OF C 0 2 AND H 2 0 WITH Au AT 77 K XPS data for physically adsorbed CO, have been reported briefly.2 If, however, multilayers are formed by exceeding the vapour pressure of COz at 77 K, distinct differences in the O(1s) and C(1s) binding energies are observed (fig. 6). Since the ' ' ' ' ' ' ' 5 5 2 e v 528 5 4 0 a FIG. 6.-XPS for adsorbed COz on Au at 77 K. (a) at equilibrium at 2 x lo-" Torr ; (b) at equilibrium at - lo-' Torr ; (c) 100 s, O(1s) and Au(4s) peaks after- exposure to C 0 2 Torr ; (d) 50 min, 4 x 10-6Torr.change in binding energy is sharp, we conch e that the changes are genuine and do not reflect charging. Charging, resulting in a broadening and shifting of adsorbate levels, does occur at much higher exposures. The binding energies, O(1s) and C(ls), are given in table 1. The UPS data (He I and He 11) for both monolayer and multilayers of C02 are shown in fig. 7a and the " orbital energies " observed given in table 2, together with the free molecule values. The positions of the lowest at 6.9 eV (physical adsorption) and 8.3 eV (multilayer adsorption) are clear but there is less certainty about higher levels. Valence-level data by XPS were not easy to obtain due to the long escape-depthS .J . ATKINSON, C . R . BRUNDLE AND M . W. ROBERTS 71 of the photoelectrons ejected by A1 Ka radiation and also the rather low cross- sections of these levels with 1486 eV energy photons. Data were therefore only obtained from thick multilayers (fig. 7b). We estimate from the shift and broadening , €F 4 0 12 r, 6 12 eV i3 " fF " 2 0 ' 40 ' m e v FIG. 7.-(4 He1 and He11 spectra for C 0 2 on Au (77 K) (1) clean Au ; (2) physisorbed COz (equili- brium at - Torr) ; (3) condensed multilayers COz. (b) XPS for multilayers of C 0 2 on Au (feature at 5 eV represents the X-ray satellite of the 15 eV peak). of the O(1s) peak that there is about a 1.5 eV charging effect present so that the two lower energy peaks observed by XPS (10 and 15 eV) agree fairly well with the lowest level (8.3 ev) recorded by U P S for C 0 2 multilayers and the centre of gravity (ca.13 eV) of the three higher energy orbitals. The orbital at 33.5 eV (XPS) is not observable by UPS. For H20, only one O(1s) peak (fig. 8) and only one set of orbitals are observed in the UPS (see also ref. (19)), the positions of which are given in tables 1 and 2. Fig. 8 also shows an O(1s) spectrum of co-adsorbed C 0 2 and H20 illustrating that a chemical shift exists in the solid-state spectra whereas in the gas phase the O(1s) values reported are virtually identical (540 eV). The valence level and O(2s)XPS data are also reported for H20 multilayers (fig. 8b). A charge-induced shift of about 1 eV is apparently present in this spectrum. ADSORPTION OF CARBON DIOXIDE ON MOLYBDENUM ADSORPTION AT 295 K FOLLOWED BY COOLING TO 77 K The development of the O(1s) and C(1s) signals during the exposure of the M o surface at 295 K to CO,(g), and also the O(ls) signal as a function of exposure, is72 ELECTRON SPECTROSCOPIC STUDIES shown in fig.9. The maximum C 0 2 pressure was about Torr. There is little change in intensity after an exposure of about 2 L which suggests an average sticking probability of 0.5 ; there is no change in the O(ls) binding energy which occurs at the same value as the original oxygen contaminant throughout the adsorption. The O(1s) peak height has been corrected for the O(ls) contamination peak which is I I I I I I zv 20 40 t l l l l l EF 6 12 eV FIG. &-(a) Coadsorbed H20 and C02 on Au at 77 K, O(1s) positions ; (b) XPS band structure of H20 multilayers on Au at 77 K ; (c) UPS He1 of Au and H20 multilayer on Au at 77 K.approximately 10 % of the intensity at CO saturation. At saturation, the Mo(3s+) : O(1s) ratio is 1 : 0.9 and the O(1s)FWHM is 2.4 eV (table 1). The C(ls) peak increased in intensity but remained fixed at an energy of about 283 eV (fig. 9). No change occurred in either the C( 1s) or O(1s) intensities on increasing the C 0 2 pressure to On cooling to 77 K at this pressure, new oxygen and carbon peaks appeared at higher binding energy, O*(ls), (534.2 eV) C*(ls), (291.6 eV) correspond- ing to multilayers of condensed C02 (table 1). Torr. ADSORPTION AT 77 K Adsorbing C 0 2 at 77 K resulted in the appearance (fig. 10) of two O(1s) peaks, one, O(ls), at 531 eV and the other O(ls)* at 534.1 eV.On removing CO,(g) and warming in uacuo to 298 K the 0(1s)* peak disappeared and there was a gradual shiftS . J . ATKINSON, C . R . BRUNDLE A N D M . W. ROBERTS 73 of the O(1s) peak to lower binding energy; at 298 K the value was 530.0 eV, a shift from the initial value at 77 K of about 1 eV. This behaviour is analogous to that observed with CO adsorbed at 77 K and subsequently warmed to 298 K when the I 1 1 I I I I 2 4 6 exposure/L L, 5 i O ' 540 FIG. 9.-(a) Ratio of intensity l o f O(ls) signal to O(1s)signal at monolayer as a function of exposure to C02(g) at 295 K. (b) Development of O(1s) and C(1s) signals as a function of exposure of Mo film to C02 at 295 K. observed O(1s) was virtually identical to the O(ls) value for CO adsorption at 298 K. It is tempting to suggest that the molecular events occurring with CO,(ads) are also analogous, namely, that at 77 K, CO, dissociates into CO(ads) and O(ads) followed by further dissociation of Cotads), which may be considered to be at 77 K in the virgin CO state, to C(aas) and O(ads) at 298 K.INTERACTION OF WATER WITH MOLYBDENUM The O(1s) peak was monitored as a function of exposure to water at 295 K and a nominal pressure of 2 x The height of the O(1s) peak (530.3 eV) became constant at an exposure of 6 L, the Mo(3s3) :O(ls) ratio then being 1 : 1.64 (table 1). This suggests an average sticking probability of 0.16 for the chemisorbed layer. A high-resolution spectrum of the O(1s) peak is shown in fig. 11. With continued exposure (i.e., beyond 6 L) the O(1s) peak broadens and at about 100 L a second signal centred at 532.5 eV has developed (fig. 11).The Mo(3d~) peak behaviour during adsorption was similar to that observed with CO and O2 in that it broadened slightly (1.1 to 1.3 eV) while the intensity of the main Mo peaks decreased. On cooling the surface which had interacted with water at 295K (H,O(g) had been removed by evacuation) to 77 K an incipient second O(1s) peak was more evident. This presumably arises from the enhanced physical adsorption at 77 K Torr (fig. 11).74 ELECTRON SPECTROSCOPIC STUDIES Increasing T I 5 2 7 . 533 ' 539 eV FIG. lO.-(a) Mo(3s) and O(ls+) signals for : (1) Mo film with small oxygen contamination at 77 K. (2) Mo+C02 at 77 K. (b) O(1s) spectra during heating of COz layer at 77 K to 298 K.(compared with 298 K) of the small quantity of water vapour still present (pressure - Torr) ; this contention is supported by the rapid increase in the peak at 532.5 eV on exposure to H20 vapour at 77 K (fig. 11). Continued exposure to water vapour resulted in the 532.5 eV O(1s) peak dominating the O(1s) peak at 530 eV. On warming to 298 K the 532.5 eV peak decreased to become only a shoulder on the O(ls) peak at 530 eV, this being a characteristic of interaction at 298 K (table 1). Heating to 523 K in vacuo removed this shoulder with a consequent sharpening of the O(ls) peak. After further heating in water vapour (523 K, 0.1 Torr, 15 min) and cooling in water vapour, initially at a pressure of lo-' Torr and finally -4 x 10-l' Torr), the Mo(3s3) : O(1s) ratio decreased to 1 : 2.2 and the high binding energy O( 1 s) reappeared.When a surface which had interacted with H,O(g) at 295 K to the stage where the O(ls) peak height was constant (exposure of -6 L) was exposed to O,(g) (1 x Torr for 5 min), the O(1s) signal at 530 eV increased immediately by about 17 % and the Mo(3s3) : O(ls) ratio decreased to 1 : 2.06. There was no apparent change in the O(ls) shoulder. DISCUSSION There are now available experimental data enabling XPS peak heights to be used to provide relative concentrations of surface species present. For example, for COz condensed on Au the C(ls) : O(1s) ratio is 1 : 4.6 which establishes a relativeS . J . ATKINSON, C. R . BRUNDLE A N D M . W . ROBERTS 75 effective ionization cross-section of 1 :2.3 for the C( 1s) and O( 1s) electrons, and which may be used in other systems involving surface carbon and oxygen.I 2 4 6 8 10 L 1 1 1 1 1 1 ' 1 ~ ~ 1 ~ FIG. 1 1 . 4 ~ ) Ratio of the observed intensity Z of O(ls) to the O(1s) intensity at the monolayer during exposure of Mo to H20(g) at 295 K ; (b) O(1s) spectra after 100 L exposure to H20(g) at 295 K ; (c) O(1s) after cooling (b) to 77 K and further exposure to H20(g). 524 530 536 524 530 536 eV CHEMICAL SHIFTS Distinct energies of binding can occur for different state of adsorption. Analogies with gas phase studies, where clear correlations exist between electron density (or charge) and binding energy, suggest that, e.g., for the y-state(Mo + CO system) the charge on the oxygen atom is more positive than the charge on the oxygen atom in the B-state.A complete interpretation of these shifts is not possible and, in particular, the problem of relaxation effects has to be unravelled. Theoretical work on relaxation contributions for gaseous molecules is in progress but the relevance to surface species is uncertain. Demuth and Eastman 2o have assumed that any shift in the energy of non-bonding valence electrons during adsorption is due entirely to relaxation effects ; consequently, if this correction is then applied to the observed shifts in other valence levels the remaining shift will be due to adsorption. In the present work we are less concerned about absolute binding energy values since both XPS and UPS data have mainly been used to distinguish between different states of adsorption.If, however, we compare, e.g., the binding energy of the valence electrons of adsorbed CO with those in CO(g), then since the experimental adsorption values are referred to the Fermi level of the substrate then, in addition to relaxation effects, an appropriate work function correction is necessary. It is not clear what the appropriate correction should be but we have added the clean substrate76 ELECTRON SPECTROSCOPIC STUDIES work function in order to compare our experimental UPS data with gas phase studies (cp. also Yates, Madey and Erickson 9. No attempt is made to account for relaxation effects. PHYSICAL AND MULTILAYER ADSORPTION ON AU There is a shift of - 1.4 eV in the O(1s) binding energy between monolayer and multilayer adsorption of C02 on Au; no similar shift occurred with H20. These facts suggest that for the Au + C02 system the first layer is influenced strongly by the Au substrate but less significantly for multilayer formation. We would expect to observe differences in heat of adsorption during monolayer and multilayer formation and these are probably reflected in the O(1s) binding energy.The apparent absence of a similar effect with H20 is puzzling. The valence levels of physically adsorbed C02 are important for elucidating the adsorbed state in the Mo + C02 system. At present, no UPS data for the Mo + C 0 2 systems are available. The lowest binding-energy feature for the Au + C02 system is clearly the oxygen lone pair orbital (n,), (table 2).The next three orbitals of C02(g), (nu),, (a,)2 and two of which overlap, are probably all present in the region 9-13 eV for C 0 2 (phys. ads.) and 10-14.5 for C02 (multilayer). Both He I and He I1 spectra show structure in these regions and in the He I1 spectra for C02 (phys. ads.), three distinct features are present (fig. 8). The X P S valence and O(2s) levels of “ solid” H20 have been reported by Siegbahn 21 ; with allowance for the 1 eV charging estimated to be present in our spectra, the agreement is reasonable. Presumably all the valence levels (lbl, 2a and 1b2) observed clearly in the UPS spectrum of H20 (ads) are present in the broad band from 6 to 13 eV (fig. 8). Again, the valence level data for H,O/Au (77 K) will prove useful for interpreting UPS data in the H,O/Mo system, when available.CHEMISORPTION ON MOLYBDENUM By observing O( Is), C( 1s) and valence level spectra for the Mo + CO system, three regimes of adsorption have been identified; adsorption at 77 K (the virgin state), adsorption at 295 K (D state) and adsorption at 77 K following adsorption at 295 K (y state). The different “ states ” reflect the difference between the thermal energy available at 295 compared with 77 K and the significance of surface coverage in determining the nature of the adsorption bond. It is therefore the interplay between available sites (O), thermal energy, surface diffusion and the activation energy necessary to acquire a particular bonding configuration that determine the observed spectra. Studies in the temperature range 77-295 K offer a number of possibilities where one or more of the above factors are of more or less significance. Briefly, the 8-state was judged 4 9 ’ to be dissociated in that the O(ls) spectra are identical to O(1s) features during oxygen interaction (table 1 and fig.1) and that UPS did not reveal CO-like orbitals (fig. 3) but features more like atomic C and 0 orbitals were present. In contrast, both the virgin and y states possess CO-like features ; furthermore, when virgin CO was warmed to 295 K these features dis- appeared and were replaced by /3-type spectra. The O(ls) : C(ls) ratios (1 : 2.5, table 1) indicate that in the adsorbed state the C : 0 ratio is 1 : 1 as would be expected. We have no evidence for a-CO state(s) on Mo which contrasts with data reported for the W + CO system.sa Since this work was completed, Viswanath and Schmidt 22 have studied the desorption of adsorbed oxygen and carbon from Mo(100) and W(100) and reported that it is identical to that of adsorbed CO.There is therefore substantial evidenceS. J . ATKINSON, C. R. BRUNDLE AND M. W. ROBERTS 77 that indicates that CO is dissociated in the P-state and this is compatible with the absence of infra-red activity when CO is strongly chemisorbed. 32* 33 However, electron spectroscopy can only show whether there is appreciable bonding between the carbon and oxygen atoms in adsorbed carbon monoxide; this would appear not to be the case. Whether the surface carbon and oxygen exist as independent entities is uncertain. As to the y-state, we have clear evidence for a range of O(1s) states (fig.lc) which correlate with the range of M-CO bond energies corresponding with the known decrease in the heat of chemisorption with increasing coverage at low temperature. Unfortunately, the signal is too weak and overlaps two greatly with the contribution from the O(1s) P-state to be certain that the y states fill up starting with the low binding energy (higher heat of adsorption) surface species, which would be consistent with a heat of adsorption that decreases with increasing coverage.1° The O(1s) features are completely reversible in that they are removed on warming to 295 K and reappear on cooling to 77 K in CO(g) at - Torr pressure. Bonding to the surface is therefore weak. The UPS data for the y-state (fig.3) suggest the presence of molecular CO but in view of the much smaller coverage of y-state than the virgin state the evidence is less convincing. The O(1s) peak associated with the dissociative chemisorption of oxygen on molybdenum remains unchanged in energy and width during the interaction in the temperature range 77-520 K (table 1). This behaviour is similar to that recently observed for the W + 0 2 system and is in keeping with work function l7 and photo- emission studies 23 of the Mo+02 system where no evidence for interaction beyond the chemisorption stage was obtained at 298 K. We conclude from the present work that the electronic configuration of chemisorbed oxygen is essentially invariant with coverage and note that the main O(1s) peak occurs at exactly the same energy as the p-CO state and both the COz chemisorption and the major component of the H20 chemisorption at 298 K.The small tailing (cp. also W+02) of the O(1s) peak to higher binding energy may possibly be attributed to the presence of a small proportion of molecular oxygen of low heat of adsorption. The O(1s) : Mo(3s+) ratio of 2.5 : 1 (table 1) implies that the oxygen adatom coverage is some 1.25 times greater than for CO at the monolayer. This is in general agreement with comparative adsorption studies of O2 and CO on Mo and W surfaces. The emergence of the broad feature centred at 234 eV (fig. 4) in the Mo(3d) spectrum is clearly related to bulk oxidation and compares with the observations of Fraser et ~ 1 . ~ ~ on exposing a clean Mo surface to ambient atmosphere for several hours when a feature also developed at 234 eV.The major O(1s) component arising from the Mo+H,O interaction at 298 K is at 530 eV. The higher binding energy component (532.3 eV) is close to the O(1s) position for H20 adsorption on Au at 77 K (532.7 eV, table 1). We suggest therefore that the minor component (about 25 % of the total signal at 295 K) is due to molecu- larly adsorbed water whereas the major component arises from dissociatively chemi- sorbed water. Unpublished work and also comparison with H2S studies would suggest the presence of OH(ads) (cp. SH with HzS on Ni and Fe13), so that OH species are likely to contribute to the O(1s) spectrum. Kinetic 2 5 and electrical resistance studies 26 suggest " compound formation " at 295 K.At 77 K, Suhrmann et aZ.26 interpreted their electrical resistance data in terms of molecular adsorption, with dissociation occurring in the temperature range 77-298 K. The Mo : main O(1s) ratio of 1 : 1.6 by height indicates about 60 % as many oxygen adatoms present as in the Mo+O, system at 298 K. The minor O(1s) peak accounts for a further 20 "/o. Clearly, the specificity of the Mo surface for dissociative78 ELECTRON SPECTROSCOPIC STUDIES chemisorption of oxygen is appreciably less than for the dissociative chemisorption of water and probably arises from the competition between molecular dissociation and molecular adsorption in the case of H20. Molecular adsorption of the highly polar H 2 0 molecule is much more likely than that of molecular oxygen so that the surface phase after interaction with water vapour probably contains an appreciable fraction (20 %, according to the present data) of molecularly adsorbed water.The 02(g) + H,O(ads) data, although of a preliminary nature, support this viewpoint since they suggest that dissociative chemisorption of 0, occurs (reflected in the increase of the intensity of the 530 eV peak) but without desorption of molecular water (no change in the shoulder at 533.2 eV). Replacement of H,O(ads) by O(ads) with further readsorp- tion of molecular water probably occurs. There is considerable support for the view that CO, is dissociatively chemisorbed on Mo and W at or just above room temperature. It includes isotopic exchange experiments infra-red 28 and field-emission arguments based on thermo- chemical data 30 and the detection of CO(g) arising from dissociative chemisorption of CO,.The present data also suggest dissociative chemisorption but the final chemisorbed state at 298 K is C(ads) and O(ads) and not CO(ads)+O(ads). The basis of this is the identical O(1s) and C(1s) binding energies to flCO(ads). The heavy tailing to higher binding energies is compatible with a small coverage of either CO(ads) or CO,(ads). Some reversible uptake of CO, by Mo, W and Ta has been reported. 30 On cooling the chemisorbed layer formed at 298 K to 77 K, further adsorption occurred but the C(1s) and O(1s) signals are at higher binding energy, comparable with those observed in the Au + CO, system at 77 K where only molecular adsorption occurs (table 1).The area ratios (C(1s) : O(1s)) are consistent with molecular adsorption (table 1). Adsorption of C 0 2 on a " clean " Mo film at 77 K resulted in the immediate growth of the low binding energy O( 1s) and C( 1s) signals at energies observed for the interaction of CO at 77 K (virgin state). On warming to 298 K the peaks moved to energies typical of the p-CO state. We therefore propose a step-wise dissociation model for CO, interaction at 77 K : 71 K C02(g) --+ CO(ads) + O(ads) C(ads) + O(ads). 5- 298K Quantitative deductions based on a comparison of Mo(3.s) : O(1s) ratios for CO and CO, adsorption are not attempted since the ratios depend on the respective surface coverages attained in each case. If we assume, however, that in our experi- ments the CO coverage is 2.3 times the C02 coverage, a ratio based on previous studies,30 then the Mo(3s) :O(ls) is about what is expected.However, at 298 K the C(1s) : O(1s) ratio is only about half that expected (table 1) for a surface containing surface carbon and oxygen in the ratio 1 : 2 .Clearly, our two sets of data (Mo : 0 and C : 0 ratios) are incompatible. We would not expect dissociative chemisorption with concurrent oxygen desorption and favour the Mo(3s) : O(1s) data in view of the rela- tively poor accuracy of measuring C(1s) : O(1.s) ratios. We are grateful to the Science Research Council for support of this work and to Mr. A. Carley for experimental assistance.S . J . ATKINSON, C. R . BRUNDLE AND M. W. ROBERTS 79 W. T. Bordass and J. W. Linnett, Nature, 1969,222,660 ; D.Eastman and J. K. Cashion, Phys. Rev. Letters, 1971, 27, 1520. C. R. Brundle and M. W. Roberts, Proc. Roy. SOC. A, 1972,331, 383. C. R. Brundle arid M. W. Roberts, Chem. Phys. Letters, 1973, 18, 380. S. J. Atkinson, C. R. Brundle and M. W. Roberts, Chem. Phys. Letters, 1974, 24, 175. (a) T. E. Madey, J. T. Yates and N. E. Erickson, Chem. Phys. Letters, 1973, 19,487. (b) J. T. Yates and N. E. Erickson, Surface Sci., to be published M. Barber, E. L. Evans and J. M. Thomas, Chem. Phys. Letters, 1973, 18,423. S . J. Atkirtson, C . R. Brundle and M. W. Roberts, J. Eiectron Spectr., 1973, 2, 105. (a) e.g., C. R. Brundle and M. W. Roberts, Chem. Phys. Letters, 1973, 18, 380; M. P. Seah, Surface Sci., 1972, 32, 703. (b) C. R. Brundle, J. Vuc. Sci. Tech., 1974, 11, 212. (c) C. J. Powell, Surface Sci., to be published. K. Siegbahn et al., ESCA Applied to Free Molecules (North Holland, Amsterdam, 1969). M. W. Roberts, Trans. Faraciuy Soc., 1963, 59,698. lo J. G. Little, C. M. Quinn and M. W. Roberts, J. Catalysis, 1964, 3, 57. l2 R. R. Ford, Adv. Catalysis, 1970, 21, 51. l3 J. R. H. Ross and M. W . Roberts, Trans. Faraday SOC., 1966, 62,2301. l4 D. 0. Hayward, D. A. King and F. C. Tompkins, Proc. Roy. SOC. A, 1967,297, 305. Is R. Gomer and A. A. Bell, J. Chem. Phys., 1966,44,1065 ; D. Menzel and R. Gomer, J. Chem. Phys., 1964,40,1164. l6 C. R. Brundle, M. W. Roberts, K. Yates and D. Latham, J. Electroiz Spectr., 1974, 3, 241 ; G. Johansson, J. Hedman, A. Berndtsson, M. Klasson and R. Nilsson, J. Electron Spectr., 1973,2, 295. D. A. King, T. E. Madey and J. T. Yates, J. Chem. Phys., 1971,55,3236 ; 1971,553247. l7 C. M. Quinn and M. W. Roberts, Trans. Farday SOC., 1964, 60,899. l9 C. R. Brundle and M. W. Roberts, Surface Sci., 1973, 38, 234. 2o J. E. Demuth and D. E. Eastman, Phys. Rev. Letters, 1974, 32, 1123. 21 K. Siegbahn et al, ref. (9) p. 82. zZ Y. Viswanath and L. D. Schmidt, J. Chem. Phys., 1973,59,4184. 23 C. M. Quinn and M. W. Roberts, unpublished data. 24 W. A. Fraser, J. V. Florio, W. N. Delgass and W. D. Robertson, Reu. Sci. Instr., 1973,44,1490. 25 H. Imai and C. Kemball, Proc. Roy. SOC. A, 1968,302,399. 26 R. Suhrmann, J. M. Heras, L. V. De Heras and G. Wedler, 2. Elektrochem., 1964,68,511. 27 D. Brennan, E. Greenhalgh and B. M. W. Trapnell, unpublished data. 28 R. P. Eischens and W. A. Pliskin, Ado. Catalysis, 1957,9,662. 29 D. 0. Hayward and R. Gomer, J. Chem. Phys., 1959,30,1617. 30 D. Brennan and D. 0. Hayward, Phil. Trans., 1965,258, 375. 31 D. W. Turner, A. D. Baker, C. Baker and C. R. Brundle, Molecular Photoelectron Spectroscopy 32 A. M. Bradshaw and J. Pritchard, Proc. Roy. SOC. A, 1970,316,169. 33 D. A. King, C. G. Goymour and J. T. Yates, Proc. Roy. SOC. A, 1972,331, 361. (Academic Press, London, 1970).