Faraday Discuss. Chem. SOC., 1990, 89, 323-328 Study of Xenon Layers on a Cu(ll1) Surface Jacques Jupille," Jean-Jacques Ehrhardt, Daniele Fargues and Albert Cassuto CNRS Laboratoire Maurice Letort, BP 104, 54600 Villers-Les- Nancy, France The adsorption of xenon has been studied by SEXAFS on a Cu( 11 1) surface in the monolayer and in the multilayer range. At temperatures ranging between 18 and 47 K, the Xe monolayer has been shown to be incommensur- ate because (i) the Xe-Xe spacing differs from the distance between easy three-fold-hollow sites (a,,J3) and (ii) the coefficient of thermal expansion of the Xe adlayer is much higher than that of copper. In addition, we have found an Xe-Cu distance of 0.345 nm, which compares well with the value determined by a hard-sphere model.The behaviour of xenon overlayers adsorbed on metallic surfaces has been shown to depend very much on the substrate concerned. The two systems which have been studied in depth to date, namely Xe/Ag( 11 1)' and Xe/Pt( 11 1),2 show some substantial differen- ces. In both cases, the attractive adatom-adatom interactions induce island formation [however, these interactions are believed to be repulsive on the Pd(100) surface as no ordered structure is observed at a temperature as low as 10 K3]. A key role is played by the lateral modulations of the holding potential. They are quite effective for xenon adsorption on the P t ( l 1 l ) surface, which gives rise to several commensurate and incommensurate solid phases depending on temperature and coverage.- In contrast, these modulations have been assumed to be of marginal importance for the case of an Xe layer on an Ag(ll1) surface, since their only effect is possibly' to align to within a few degrees the commensurate Xe layer with the underlaying hexagonal structure (this has alternatively been suspected to come from extrinsic effects such as surface defects').If one considers the relative strength of the Xe-Xe interactions and the Xe substrate holding potential, the Cu( 11 1) surface offers an interesting opportunity. On this sub- strate, the Xe layer has been shown to form a commensurate ( J 3 x J 3 ) R30" structure at 77 K.4 This is consistent with the very small mismatch existing at this temperature between the Xe-Xe spacing in bulk xenon (0.4388 nm"') and the distance between the three-fold hollow sites separated by 43 times the Cu-Cu spacing (ac-UJ3 = 0.4414 nmi), which are the probable easy sites (the distance between these particular sites will hereinafter simply be referred to as 'the distance between three-fold hollow sites').However, at temperatures lower than 77 K, a greater mismatch is expected because, if the parameter ac.,J3 is virtually unchanged (0.4413 nm at 25 K'), the minimum Xe-Xe distance in bulk xenon decreases to below 0.434 nrn5*' at the 0 K limit. In these conditions, two distinct behaviours can be expected, depending on whether or not the xenon overlayer remains commensurateX: either the Xe-Xe spacing remains equal to the distance between two three-fold hollow sites of the Cu( 11 1 ) surface and the layer then remains commensurate, or it is governed by its own thermal expansion law and the layer becomes incommensurate (this will ensure that neither commensurate nor high- order incommensurate structuresX are formed).The purpose of this paper is to investigate which of these cases is realised. Measurements have been done by surface extended X-ray absorption fine structure (SEXAFS), this technique being used for the first time Distances between three-fold hollow sites on the Cu( 1 1 1 ) surface were determined using ( i ) a lattice parameter of 0.361 SO nm at 298 K'" and ( i i ) the coeficients of linear thermal expansion."' 323324 SEXAFS on Xe/Cu( 1 1 1 ) I ’ I I I I I I I h Y m .I c 3 i” L3 3.1 L2 i I I I I I I I I I 4800 4900 5000 5100 hvleV i Fig. 1.SEXAFS from an Xe multilayer at 20 K. to study rare gas adsorption on a monocrystalline substrate. The Xe-Cu distance is also determined and discussed. Experimental The L,-edge SEXAFS data have been collected at the Laboratoire pour 1’Utilisation du Rayonnement Electromagnktique (LURE) at Orsay, on the DCI storage ring, with a Si(311) or a Ge( 1 1 1) double-crystal monochromator, thus favouring either resolution or sensitivity, respectively. The Cu( 1 1 1 ) crystal was mounted in an ultra-high-vacuum (UHV) chamber.’ It could be cooled down to 18 K by circulating liquid helium in the sample holder. Temperatures were measured with a calibrated platinum resistor. The sample was treated by repeated argon-ion bombardment followed by annealings at 900 K until it showed a clean surface by Auger electron spectroscopy and a ( 1 x 1 ) pattern with sharp spots by low-energy electron diffraction (LEED).SEXAFS measurements were made using the total-electron-yield method. The SEXAFS signal was recorded at the L,-edge of xenon (4782eV) with the vector polarisation of the light parallel to the surface sample. In this way the SEXAFS oscillations are dominated by the in-plane coordination shells. Resu 1 ts To determine experimental phase shifts and amplitudes we have first examined the EXAFS from multilayers (fig. 1 ) which were deposited on the substrate between 20 and 35 K. As judged from the exposure, the deposits were 15-75 layers thick. The height of the Xe L,-edge jump reached a limit which was ca. 10 times that of the monolayer when the deposit was ca.15-20 layers thick, as estimated from the exposure. A 10% contribution of the first layer to the overall signal recorded from the semi-infinite solid corresponds to an escape depth of ca. 3.5-4 nm; in these conditions, 20 layers of xenon (ca. 7 nm) would give rise to a signal ca. eight times bigger than that of the monolayer, which is precisely what we have observed. The behaviour of the edge jump is thus consistent with a layer-by-layer growth up to 15-20 layers, and we have supposed that a thicker deposit would grow in the same way. EXAFS data were also collected above the temperature of deposit, up to 47 K. It was checked, from the height of the L3 edge, that no major desorption occurred during both the annealing and the data acquisitionJ.Jupille et al. 0.06 - 0.00- -3 -0.06 325 - - 4800 4900 E,IeV aJ '0 E on +-a .I E 0.8 0.6 0 . 4 0.2 0 .o 1 2 3 4 5 6 7 R I A I I I 1 1 1 2 3 4 5 6 k / k ' Fig. 2. SEXAFS from an Xe monolayer on a Cu(l11) surface at 30 K using an Si( 111) mono- chromator: ( a ) SEXAFS data; ( b ) Fourier transform of the normalized kx( k ) signal, experimental (-) and calculated ( * * * .); ( c ) inverse Fourier transform of Fourier-transformed data after multiplication by a window function including the Xe-Xe and the Xe-Cu peaks. time. The accessible data range above the L3 edge (4782 eV)? is limited by the occurrence of the L2 threshold (around 5100 eV, see fig. 1 ) . In any case, very little information can be collected beyond 200 eV because of the huge Debye-Waller damping of the SEXAFS oscillations related to the xenon layer.Fortunately, these limitations are somewhat compensated by the large size of the xenon atom, which shortens the period of SEXAFS oscillations, thus making the analysis tractable. Xenon monolayers have been prepared in two different ways: (i) by adsorption of Xe at 60-65 K under a pressure of (2-3) x Pa, i.e. far from the conditions at which the three-dimensional condensation of xenon takes place and within a temperature range where little desorption occurs from the Xe monolayer adsorbed on the Pt( 11 1 ) surface, even in UHV conditions; ( i i ) by adsorption of a multilayer followed by annealing to + 1 eV= 1.602 18x lo-'" J.3 26 SEXAFS on Xe/Cu( 11 1) 0.453 0.42 1 1 I I 10 20 30 40 50 T l K Fig.3. SEXAFS determination of Xe-Xe spacings as a function of temperature in three- dimensional xenon (stars and dotted line); also indicated are the Xe-Xe spacings as determined by X-ray measurement^^.^ (thin line). 60-65 K and cooling to the working temperature. An Xe (d3 x d3) R30" LEED pattern could be seen for more than 2 h when a monolayer was prepared in such conditions and kept in UHV. This corresponds to the longest collection time we needed to study monolayer coverage, i.e. when using an Si(3 11) double-crystal monochromator. The SEXAFS spectrum recorded from an Xe monolayer at 30 K is shown in fig. 2(a). It is worth noting that the data shown herein have been collected through three different beam times, thus increasing the confidence of their reproducibility. Discussion SEXAFS calculations were done using plane waves.For the case of the multilayer we have used theoretical phase shifts and amplitudes." The coordination number was fixed to 12. Values found in this way for the Xe-Xe spacing were very close to those from X-ray mea~urernents.~~~ However, SEXAFS values at different temperatures were correc- ted to bring their centre onto the X-ray curve. This resulted in a decrease of only 0.0007 nm in the calculated values of the Xe-Xe spacing. The variation of this spacing versus temperature is seen in fig. 3. The error bar corresponds to the scattering of the results when varying the parameters involved in the calculation. Our results agree with the X-ray measurements within experimental error (however, one can notice that EXAFS data show a possibly higher thermal expansion than the reference data; we do not explain this trend since the expectation for materials having a large amount of thermal disorder is a shrinkage of the actual thermal expansion"). The purpose of the examination of the multilayer was mainly to determine experimental phase shifts and amplitudes.In fact, that study has proved that the theoretical values of these parameters1' were fairly accurate. This may be connected with the fact that, in the case of a rare gas, one obviously avoids possible variations of phase shifts due to chemical changes (amplitudes are much less sensitive to chemical environment'"). In the forthcoming calculation, both kinds of parameters (theoretical and experimental) were used and the two sets of results did not show any significant difference (the correction of 0.0007 nm was done as in the case of the multilayer).The Fourier transform of the normalized k x ( k ) signal and the inverse Fourier transform of the Xe-Xe first shell are shown in fig. 2, the experimental spectrum being collected by means of an Si( 3 1 I ) double-crystal monochromator to obtain good resol- ution. As the Xe-Cu bond is not exactly normal to surface it contributes to some extent to the SEXAFS spectrum, despite the fact that the vector polarisation of the lightJ. Jupille et al. 2 ... C e, 0.44- * v) .- 'c3 2 0.43- X I 327 0.45 I 0 0.42 ! I I I I I 10 20 30 40 50 Fig. 4. SEXAFS determination of Xe-Xe spacings as a function of temperature and in a xenon monolayer (-0-) deposited on a Cu(ll1) surface; also indicated are the Xe-Xe spacings in bulk xenon as determined by SEXAFS measurements (.- * a , see fig. 1) and distances between three-fold-hollow sites on the Cu( 11 1) surface (a,,d3) (see footnote t ) Table 1. Xe-substrate distances (nm) surface measurement hard-sphere model reference Xe/Ag( 11 1) 0.35 f 0.01 0.361 Cohen et a/.' Xe/Cu( 11 1) 0.345 f 0.01 0.345 this work LEED SEXAFS is parallel to the surface. In these conditions, the Fourier transform of the kx( k ) signal shows two well separated peaks, above 0.4 nm and slightly below 0.3 nm, which are due to the Xe-Xe and the Xe-Cu (first-neighbour distances), respectively. These two peaks have been treated by means of theoretical phase shifts and amplitudes (as seen above this is a fairly good approximation for Xe scatterers), either separately or together, by introducing two types of scatterers (Xe and Cu) in the neighbourhood of the central Xe atom.The result of such calculation including the two peaks is shown by the dotted line in fig. 2(6). For the case of the Xe-Cu distance, the data could be fitted only by means of a large negative Eo value (ca. -12 eV) accounting for the internal potential of the copper substrate. In contrast, this value was only 2 or 3 eV when the Xe-Xe bond was considered. This can be qualitatively understood by the fact that the xenon levels are referenced to the vacuum level. The photoadsorption of L3 edge involves final states of both s and d symmetry," but, as data were collected with in-plane polarisation, the s-d cross term is zero and the coordination number amounts to a 7.5'' for the Xe-Xe interaction and 2.25 for the Xe-Cu interaction.Xe-Xe spacings measured for the case of the monolayer at temperatures ranging between 18 and 47 K are shown in fig. 4. They clearly characterise an incommensurate structure. First, Xe-Xe spacings are shorter than the distances between three-fold hollow sites of the Cu( 11 1) surface, which are also represented in fig. 4. Moreover, the expansion coefficient is much larger than that of the substrate. It means that even the occurrence of a high-order commensur- ate phase' can be discarded. As on the Ag( 11 1) surface, one can suggest that the lateral modulations of the holding potential are negligible with respect to the xenon-xenon interaction since in the temperature range under discussion (18-27 K) a very small mismatch can be preserved between the Xe overlayer and the copper substrate.However, the thermal expansion of the Xe layer varies in such a way that our observations do328 SEXAFS on Xe/Cu( 1 1 1 ) not contradict the finding that this layer could form a (J3 x J3) R30" commensurate structure at higher temperature and especially at 77 K.4 An average value of 0.345*0.01 nm was found for the Xe-Cu distance in the temperature range of interest. This measurement compares nicely with the value deduced from a simple hard-sphere model, the radius of the Xe and Cu atoms being 0.1275 and 0.2175 nm, respectively. A rather similar conclusion has been reached concerning the Xe monolayer on the Ag( 1 1 1 ) surface, as studied by LEED.' The results are summarised in table 1 .Conclusion The first SEXAFS study of a rare-gas monolayer has been performed on the Xe/Cu( 1 1 1 ) system. As evidenced by both the Xe-Xe distance and the thermal expansion of the adsorbate layer with respect to that of the substrate, xenon has been shown to form an incommensurate structure on the Cu( 1 1 1 ) surface. The average Xe-Cu distance was found to amount to 0.345 f 0.01 nm, in good agreement with a hard-sphere model. The authors are very grateful to D. Chandesris and J. Lecante (LURE), C. Brouder and G. Krill (Universite Nancy I) for helpful assistance and discussions about the treatment of the data. References 1 P. I . Cohen, J. Unguris and M. B. Webb, Surj Sci., 1976, 58, 429. 2 K. Kern, R. David, P. Zeppenfeld and G. Comsa, Surf: Sci., 1988, 195, 353. 3 E. R. Moog and M. B. Webb, Surj: Sci., 1984, 148, 338. 4 M. A. Chesters, M. Hussain and J . Pritchard, Surf: Sci., 1973, 35, 161. 5 G. L. Pollack, Rev. Mod. Phys., 1964, 36, 748. 6 M. L. Klein and J . A. Venables, Rare Gas Solids (Academic Press, London, 1977), vol. 11. 7 ( a ) Handbook of Chemistry and Physics (CRC Press, Boca Raton, Florida, 69th edn, 1988-1989); ( b ) 8 K. Kern, P. Zeppenfeld, R. David and G . Comsa, Phys. Rev. Lett., 1987, 59, 79. 9 P. Roubin, D. Chandesris, G. Rossi, J. Lecante, M. C. Desjonqukres and G . TrCglia, Phys. Rev. Lett., American Institute Handbook (McGraw-Hill, New York, 2nd edn, 1963). 1986, 56, 1272. 10 B-K. Teo and P. A. Lee, J. Am. Chem. Soc., 1979, 101, 2815. 1 1 P. A. Lee, P. H. Citrin, P. Eisenberger and B. M. Kincaid, Rev. Mod. Phys., 1981, 53, 769. 12 P. H. Citrin, fhys. Rev. B, 1985, 31, 700. Paper 91055161; Received 22nd December, 1989