J.C.S. Faraday I, 1980,76,126-134Ultraviolet Photoelectron Spectroscopic and ThermalDesorption Studies of the Chemisorption and Decompositionof Benzene, Cylcohexadiene, Cyclohexene andCyclohexane on W( 100)BY ASHOK K. BHATTACHARYADepartment of Physical Chemistry, University of Cambridge,Lensfield Road, Cambridge CB2 1EPReceived 24th August, 1978The chemisorption and decomposition of the cyclic hydrocarbons, viz. benzene, 1,4 (and 1,3)cyclohexadiene, cyclohexene and cyclohexane on a W(100) surface at 300 K have been investigatedusing ultraviolet photoelectron spectroscopy and thermal desorption studies. The hydrocarbonsall decompose on the surface to give an adsorbed layer of hydrogen and the same molecular species.Increase in exposure to the hydrocarbon results in a monotonic increase in the surface concentrationof the molecular fragment, while the surface concentration of the adsorbed hydrogen, in the caseof the unsaturated hydrocarbon, first increases and then decreases so that at saturation coverageonly the molecular species remains on the surface. On heating the adsorbate covered surface themolecular species decomposes and hydrogen is found to be the only observable desorption productwhile the surface is left with an adsorbed layer of carbon.The nature and bonding of the molecularfragment are also discussed.The elucidation of the nature and bonding of the species formed when a gasadsorbs on well characterized surfaces of different metals is of vital importance to abetter understanding of heterogeneous catalysis.With this in mind we have studiedthe chemisorption, decomposition and interaction with hydrogen of benzene, 1 , 4 (and1, 3) cyclohexadiene, cyclohexene, and cyclohexane on a W(100) surface usingultraviolet photoelectron spectroscopy (u.P.s.) and thermal desorption techniques.The chemisorption of benzene has been previously studied on Ni(lll),l Ir(100)2and Pt(100)3 surfaces by U.P.S. and on Ni(l.11) and faces by high-resolutionelectron energy loss spectroscopy (h.r.e.1.s.). These experiments indicated thatbenzene adsorbs on these surfaces with relatively little distortion. The plane of themolecule was thought to lie parallel to the surface and the interaction to take placevia the n electrons of the ring. Similar conclusions were also drawn by Somorjaiand Gland 5 s on the basis of their investigations of the adsorption of benzene onPt(ll1) and (100) surfaces using 1.e.e.d.and work-function measurements. U.P.S.experiments have further indicated that, like benzene, cyclohexane adsorbs onNi(ll1) with relatively little distortion.EXPERIMENTALThe apparatus used in the present investigation, the geometry employed to record thephotoelectron (p.e.) spectra with HeQ) (21.2 eV) radiation and the method of cleaning thecrystal have been described elsewhere?For thermal desorption studies the crystal was heated radiatively by a hlament at aconstant rate m 12 K s-l.12A . K . BHATTACHARYA 127The hydrocarbons (reagent grade) were all dried and purified by multiple freeze-thaw-vaporisation cycles in the gas handling line of the apparatus.The purity of the hydrocarbonvapours was checked mass spectrometrically.RESULTSThe p.e. spectra of a W(100) surface on sequential exposure to benzene at 300 Kare shown in fig. 1. The corresponding adsorbate-induced emissions are shown by thedifference curves in fig. 2. Each of the difference curves has been obtained byplotting the difference between the p.e. spectrum of an adsorbate-covered surfaceand that of the clean surface. The difference curves can be best described by consider-ing two regions: one between 0 and 4eV and the other beyond 4eV below theFermi level (EF). In the first region an exposure of 1 L (= 1.33 x Pa s) producesa peak at - 3 eV with two shoulders at - 1.3 and -2.2 eV (the peak positions inthis paper axe given with respect to EF).Increasing the exposure causes a rapidincrease in the intensity of photoemission in this region and for an exposure ofx 2 L there is a sharp and intense peak at - 1.9 eV and a shoulder at = -3.2 eV.Further exposure causes the intensity of photoemission to decrease. Saturation isobtained at an exposure of = 6 L when two peaks with maxima -2.2 and -3.2 eVare obtained. In the second region, for an exposure of 1 L peaks appear at -5.0,- 7.0, - 9.0, - 10.4 and - 12.4 eV. Increasing the exposure causes the intensity ofphotoemission in this region to increase monotonically until saturation. However,with increasing coverage the position of the second peak shifts towards higherbinding energy, while the positions of the other peaks remain unaltered.At satura-tion coverage the second peak has its maxima at -7.9 eV.hVI I I I I I I I 1 IL18 -16 -14 -12 -1'0 -8 -6 -4 -2 0E- EFleVFIG, 1.-He(1) photoelectron spectra of a W(100) surface on sequential exposure to benzene at300K: (a)(--- ) clean surface ; (b) (-) 1.0 L of benzene ; (c) (- - -) 2.0 L of benzene and(d) (-) 6.0 L of benzene128 ADSORPTION OF CYCLIC HYDROCARBONS ON TUNGSTENrn+d .*3ti;iii!i21 19 17 15 13 11 9i.p./eV Jw I 1 I 1 I I-14 -12 -10 -8. -6 -4 -2 0E- &levFIG. 2 . 4 0 ) Difference curves for varying exposures of benzene at 300 K : (i) (-) 1.0 L ; (ii) (- - -)2.0 L ; (iii) (- - -) 3.0 L and (iv) (-) 6.0 L ; (b) gas-phase photoelectron spectrum of benzene.-0.41 2 3 4 5 6 7 8exposure/LFIG.3.-Change in A 4 as a function of exposure at 300 K: (a) (-) benzene; (b) (. . .) cyclo-hexadiene ; (c) (- - 9 -) cyclohexane and (d) (- - -) cyclohexeneA . K . BHATTACHARYA 129Each of the other hydrocarbons studied in the present investigation upon adsorp-tion on a W(100) surface gives rise to similar p.e. spectra with the same number ofpeaks at the same positions as are obtained on exposing the W(100) surface to benzene.However, two observations can be made: (a) In the case of cyclohexane the p.e.spectrum saturates at an exposure of FZ 2 L when a sharp intense peak is obtained inthe first region. Further exposure to cyclohexane, unlike that of the other unsaturatedhydrocarbons, does not result in a decrease in intensity of the sharp peak.(b) Fora given low exposure (c 2 L) the intensities of photoemission in the first region forthe different adsorbates have the following order of magnitudes :(both 1, 4 and 1, 3 cyclohexadiene behave identically in all the measurements thatwe have carried out).= 4.6 ev)on sequential exposure to each of the hydrocarbons are shown in fig. 3. Theaccuracy of A 4 is k0.05 eV. The values of A+ at the maxima decrease in the order :Hydrogen was the- only observable desorption product when the W(100) surfaceexposed to any of the hydrocarbons was heated. Fig. 4 shows a series of H2 thermaldesorption spectra taken after varying exposure to benzene at 300 K. For comparisonthe H2 thermal desorgtion spectrum obtained from a W(100) surface saturated withpure hydrogen at 300 K is also shown in fig.4. The H2 thermal desorption charac-teristics of a W(100) surface after varying exposures to each of the other hydro-carbons is similar to that of a W(100) surface exposed to benzene. In table 1 we show< C6HS < C6HlO < C6H12The changes in A+ of a W(100) surface at 300 K (A4 = + - &eaa ;C6Hlo > C6Hg > C6H6.A4 00 500 600 700TIKFIG. 4.-Hz flash desorption spectra from a W(100) surface after';varying exposure to benzene at300 K. (a) 0.5, (b) 1.0, (c) 2.0, (d) 6.0 L. The H2 flash desorption_spectrum from a W(100) surfacesaturated with pure Hz is shown by the dashed curve.1-130 ADSORPTION OF CYCLIC HYDROCARBONS O N TUNGSTENthe ratios of the normalised peak(s) areas for different exposures to the unsaturatedhydrocarbons [a normalised peak(s) area is obtained by dividing the total area underthe desorption peak(s) for a given coverage by the total area under the desorptionpeaks obtained when the surface is saturated with pure hydrogen].TABLE RATIOS OF THE NORMALISED PEAK AREAS FOR VARIOUS EXPOSURESratio of the normalisedexposure/L peak areasCsH6 :CGHs :CGH100.5 1 : 1.6 : 1.91 .o 1 : 1.2 : 1.32.0 1 : 0.9 : 16.0 1 : 0.9 : 1(saturated)Fig.5 shows the room temperature pe. spectra obtained after sequentially flashingthe W(100) surface saturated with benzene at 300 K to M 450 K (which is greater thankhe decomposition temperature of the molecular fragment at saturation coverage.See fig.4.) and to = 1100 K. The difference spectrum plotted in fig. 6 shows that-16 -I4 -12 -10 -8 -6 --A -2 0E- EF/evFIG. 5.-He@) photoelectron spectra of a W(100) surface saturated with benzene after flashing todifferent temperatures : (a) (-) clean surface ; (b) (- - -) 450 K and (c) (-) 1100 K.the fragment left after desorption of hydrogen gives rise to emissions at - 1.6, -4.05,- 5 , -6 and -9.7 eV. The peak at -4.05 is very sharp and has a width at halfheight of M 0.55 eV. Heating the crystal from m 450 to = 1100 K results in adecrease in intensity of the p.e. spectrum below -4.5 eV while the peaks at -1.6and -4.05 eV become sharper. The value of A 4 increases from -0.15 to +0.10 eVA . K. BHATTACHARYA 131-!2 -10 -8 -6 -L -2 0E-EF/eVFIG.6.-Difference curve obtained for the carbon overlayer by plotting the difference fig. 5(c)-(a).The p.e. spectra obtained after exposing a W(100) surface preadsorbed withhydrogen to benzene at 300 K axe shown in fig. 7. By cornparing fig. 1 and 7 it canbe concluded that benzene displaces adsorbed hydrogen. Both cyclohexadiene andcyclohexene behave similarly to benzene. Cyclohexane, however, does not displacehydrogen adsorbed on W(100).E-EFleVFIG. 7.--He(I) photoelectron spectra of a W(100) surface at 300 K under different conditions :(a) (. . .) clean surface ; (b) (-) from (a) after saturation with H2 ; (c) (- - -) from (b) after anexposure of 2 L of benzene and (d) (-) from (c) after an exposure of 24 L of benzene4 32 ADSORPTION OF CYCLIC HYDROCARBONS ON TUNGSTENDISCUSSIONThe p.e.spectra shown in fig. 1 can be readily understood in terms of a de-composition reaction giving rise to an adsorbed layer of a molecular fragment whichgives rise to, among others, multiple photoemission peaks in the region below -4 eVand an atomic species which gives rise to intense photoemission in the region between- 1 and -4 eV. Of the two pussible atoms here, C and H, only the latter is known[ref. (8) and (9), as well as from the present investigation] to give rise to intense emissionin the region between - 1 and -4 eV. Therefore, it is concluded that all the cyclichydrocarbons studied in the present investigation decompose upon adsorption on aW(100) surface to give an adsorbed layer of the same molecular fragment and hydro-gen.This continues up to an exposure of = 2 L. On further exposure to the hydro-carbons (excepting cyclohexane) an increase in the surface concentration of the molecu-lar fragment causes a displacement of adsorbed hydrogen. Indeed the results of theU.P.S. experiments (fig. 7) also show that the unsaturated cyclic hydrocarbons alldisplace adsorbed hydrogen while there is simultaneous chemisorption of the molecu-lar fragment, The inability of cyclohexane to displace adsorbed hydrogen is alsoreflected in the failure of the intense photoemission peals at - 1.9 eV to decrease inintensity for exposures of > 2 L of cyclohexane. If we assume that the stickingcoefficients of all the unsaturated hydrocarbons under investigation here are the samethen it is expected that at low coverages the amount of chemisorbed hydrogen, for agiven exposure of the adsorbate, would decrease with increasing unsaturation in thehydrocarbons.The observed intensities of photoemission in the region between- 1 and -4 eV for exposures of < 2 L would seem to support the above-mentionedtrend.Chemisorption of hydrogen is known to increase the work-function of a W(lO0)surface. If we consider the molecular fragment to be an electron donor then thework-function behaviour of a W(100) surface on sequential exposure to each of thehydrocarbons (fig. 3) would also support the above-mentioned mechanism of chemi-sorption of the cyclic hydrocarbons on W(lO0).It is not clear, however, whythe value of A 4 for an exposure of w 1 L of cyclohexane is not more than that for asimilar exposure of cyclohexene, although pee. spectra suggest that there is morechemisorbed hydrogen in the former case than the latter.From the results of the thermal desorption studies we can arrive at the followingconclusions : (a) it is the strongly bound hydrogen that is displaced initially, followedby the displacement of the weakly bound hydrogen, and (b) the chemisorbed hydrogenformed during the initial adsorption of the hydrocarbons and the hydrogen bound tothe molecular fragment both give rise to the desorption peak around z 450 K. Sincehydrogen may be retained in the chemisorbed state on the surface after dissociation,this desorption temperature may, or may not, be the temperature at which the chemi-sorbed molecular fragment dehydrogenates.It does, however, set a maximumlimit to the temperature of the decomposition reaction. In accord with the above-mentioned mechanism of chemisorption it is seen from table 1 that at initial coveragesthe amounts of hydrogen desorbing on heating the crystal exposed to the differentunsaturated hydrocarbons are almost in the ratio of the numbers of hydrogen permolecule of the hydrocarbons. However, as expected, the ratio of the amounts ofdesorbing hydrogen with increasing coverage deviates from that required by themolecular stoichiometry and for exposures of > 2 L the amounts of desorbinghydrogen are the same (within experimental errors) for 411 the unsaturated hydro-carbon adsorbatesA .K . BHATTACHARYA 133From fig. 1 and 2 (and also from the p.e. spectra recorded for other exposuresbut not shown in these figures for clarity), it can be seen that the photoemission fromadsorbed benzene in the region between - 1 and - 4 eV consists of two maxima at-2.2 and -3.2 eV whose positions do not alter with varying coverage. The othermaxima in this region, on the other hand, shift and at first increase and then decreasein intensity. Having regard to the photoemission from hydrogen chemisorbed onW(100) surface it can be concluded that the maxima at -2.2 and -3.2 eV (whichare also the only ones present at saturation coverage in this binding energy region)are due to the molecular fragment.However, from the evidence that is availableit is not possible to say if the molecular fragment is the only species present on thesurface at saturation coverage. A small amount of chemisorbed hydrogen may bepresent which, although not distinctly identifiable by an inspection of the p.e.spectrum, could be responsible for some of the intensity in the photoemission atsaturation coverage. In any case, the amount of such chemisorbed hydrogen canonly be very small. The peak at 12.4 eV (fig. 2) is probably associated with a changein the electronic structure of tungsten following adsorption, as a peak is found toexist in this region in the p.e. spectra of W(100) surfaces having widely differentadsorbates.Unambiguous identification of the molecular fragment is not possible on thebasis of the data available to us.It is clear, however, that the fragment is obtainedby partial dehydrogenation of benzene.A mechanism for the adsorption of benzene on a W(lO0) surface involving asequence of fragmentation of benzene upon adsorption to form acetylene, dissociativeadsorption of the acetylene formed as HC=C--- and H- and finally displacementof H- by HC=C- at higher coverages (> 2 L) would be consistent with some ofthe above-mentioned observations. This mechanism is, however, not consistentwith the mechanism of chemisorption of acetylene. It is believed lo that acetylene,upon adsorption on a W(100) surface, initially decomposes completely to give anadsorbed layer of atomic C and H. Further exposure of the surface to acetyleneHresults in the formation of an ethylenic species, \C_dH, and a simultaneous / \displacement of the adsorbed H.Ethylene and butadiene have also been found loto behave like acetylene on W(100).A possible structure of the molecular fragment in question here could be oneinvolving di-o-bonded benzene shown below : 0 / \Indeed, examples of organo-metallic clusters with benzene bonded at the 1 , 2 positionsto two metal atoms are kn0wn.l The gas phase p.e. spectrum of a benzene moleculea-bonded to two W atoms at the ortho-positions is not accessible. However, on thebasis of the known effects of substitution l2 on the p.e. spectrum of benzene, we expectthat the highest occupied double degenerate n-orbitals of benzene would split by = 1 eV on forming the above-mentioned structure. Moreover, all the three benzenen-orbitals, which extend over a large region, would interact strongly with the metallic&orbitals, which also extend over a large region at the surface.Such n-d interactionswould increase the binding energies of all the n-orbitals, though by differing amounts.The energies of the a-orbitals, on the other hand, are likely to alter only a little134 ADSORPTION OF CYCLIC HYDROCARBONS O N TUNGSTENFig. 2(b) shows the gas phase p.e. spectrum of benzene.12 The vertical ionizationpotentials of the various 71- and a-orbitals are given below :n : 9.4 (doubly degenerate) and 11.5 eVa : 11.8, 12.2, 13.9, 14.2, 14.8, 15.4, 16.9 and 19.2eVOn comparing fig.2(b) and (aiv), it can be seen that the molecular fragment could,indeed, have the di-a-bonded structure. For the peaks at -2.2 and -3.2 eV inthe difference spectra could be due to the doubly degenerate n-orbitals followingsplitting due to ortho-substitution and further stabilization through n--d interaction.The peak at -5 eV could then correspond to the second band in the p.e. spectrumof benzene after a slight shift in the energy of the a-orbitals and a stabilization throughn-dinteraction of the third n-orbital of benzene. The peak at - 7 eV for low coverage[fig. 2(a i)] agrees well with the third band in the p,e. spectrum of benzene. The shiftof this peak to higher binding energy with increasing exposure could be due to achange in the C-W bond energy with increasing coverage.Further, after a slightshift to lower binding energy of the S t h band, the fourth and fifth bands of the p.e.spectrum of benzene could be made to coincide with the peaks at -9 and - 10.4 eVof the difference spectra. It must be emphasized at this point that, in spite of thereasonable agreement between the expected p.e. spectrum for di-a-bonded benzeneand the observed emissions induced by the molecular fragment in question here,one cannot rule out the possibilities of other structures for the fragment. Comple-mentary studies by X.P.S., l.e.e.d., i.r. and h.r.e.1.s. would, therefore, be useful inresolving the problem.The emissions induced by the carbon overlayer (fig. 6) left after complete de-hydrogenation of the molecular fragment formed upon adsorption of the cyclichydrocarbons on a W(100) surface do not agree with the emission of carbided W(100)surface.This is indicative of the overlayer being chemisorbed carbon. Moreover,the sharpness of the emission at -4.05 eV (fig. 6) and the fact that the overlayerremains stable until x 1600 K seems to suggest that the carbon atoms on the surfaceare perhaps strongly bonded to each other and may exist as a graphitic overlayer.L.e.e.d. studies would, of course, be useful in this respect. However, on heating thissurface to a temperature == 1600 K, its p.e. spectrum changes from that shown infig. 6 to the p.e. spectrum of a carbided W(lO0) surface.8The author thanks the Royal Commission for the Exhibition of 1851 for theaward of a Research Fellowship which made this work possible. Useful conversa-tions with Prof. J. Lewis and R. Hoffman and Dr. J. Q. Broughton are gratefullyacknowledged.J. E. Demuth and D. E. Eastman, Phys. Rev. Letters, 1974,32, 1123.G. Broden, T. Rhodin and W. Capehart, Surface Sci., 1976, 61, 143.T. E. Fischer, S. R. Kelemen and H. P. Bonzel, Surface Sci., 1977, 64, 157.J. C. Bertolini, G. Dalmai-Imelik and J. Rosseau, Surface Sci., 1977, 67,478.J. L. Gland and G. A. Somorjai, Surface Sci., 1973, 38, 157.J. L. Gland and G. H. Somorjai, Surface Sci., 1974, 41, 387.W. F. Egelhoff, J. W. Linnett and D. L. Perry, Farday Disc. Chem. SOC., 1974,58, 35.E. W. Plummer, Topics in AppZ. Phys., ed. R. Gomer (Springer-Verlag, Berlin, Heidelberg,New York, 1975), vol. 4, p. 144.' A. J. Bhattacharya, J. Q. Broughton and I). L. Perry, J.C.S. Faraday I, 1979,75, 850.lo A. K. Bhattacharya, J.C.S. Faraday I, 1979,75, 863.l1 J. Lewis and B. F. G. Johnson, Pure AppZ. Chem., 1975,44,43.l2 D. W. Turner, G. Baker, A. D. Baker and C. R. Brundle, Molecular PhotueZectron Spectroscopy(Wiley-Interscience, New York, 1970), p. 264.(PAPER 8/1554