Structural and electronic investigations of Cu islands on graphite

Surface

352

Science

156 (1985) 352-35X

North-Holland.

STRUCTURAL AND ELECTRONIC ON GRAPHITE M. DE CRESCENZI. Dipartimento

P. PICOZZI

INVESTIGATIONS

Amsterdam

OF Cu ISLANDS

and S. SANTUCCI

di Fisrca, Unioersita’ de Lilquila.

I - 67100 L’Aquila, Ita!v

and C. BATTISTONI

and G. MATTOGNO

Istituto di Teoria e Struttura Elettronlca Monterotondo Starione, I-00016 Rome, Italy Received

8 August

del

CNR,

Area

della

Rlcerca

dr

Romu.

1984

Thin copper films with thicknesses 5, 10. 20, 50 and 300 A prepared by evaporation in UHV conditions onto graphite substrates have been studied by different surface analysis spectroscopies (Auger, XPS and ELS). Particular attention has been devoted to the investigations of fine structures observed beyond the M,,, copper edge in the energy loss spectra in the reflection mode. Their Fourier analysis, in the form of the radial distribution function F(R), has shown a great sensitivity to (i) the lattice parameter contractions occurring in clusters. and (ii) the effects of the surface thermal disorder enhanced by the reduced cluster dimension. Moreover the F( R ) allowed us to check that the local crystalline structure in the cluster does not change from the fee bulk metal structure.

1. Introduction During the last years a great deal of experimental and theoretical work has been done on small metal clusters in order to obtain a full understanding of their physical properties [l]. The use of surface spectroscopy methods [2-41 has greatly contributed to the knowledge of the electronic structure of clusters and the relaxation process occurring when the “d” bands of metal clusters go towards those of the isolated atoms [3]. Moreover, the surface effects due to the high surface/volume ratio play a fundamental role in the cluster properties. For instance, they are responsible for (i) the high chemical reactivity in the oxidation process and in heterogeneous catalysis [5] and (ii) the lowering of the melting temperature [6] due to the high mobility of the surface atoms. In the past, structural studies of metal clusters have been done by means diffraction methods but broadening and diffusion effects limit these techniques to cluster size greater than 10 A [7]. Recently, these obstacles seem to be 0039-6028/85/$03.30 0 Elsevier Science Publishers (North-Holland Physics Publishing Division)

B.V.

M. De Crescenzi et al. / Cu islands on graphite

353

overcome by the EXAFS (extended X-ray absorption fine structure) investigations [8] due to the three-dimensional local character of this technique which does not require a long-range order. From an analysis of the Au EXAFS spectra it results that a nearest neighbour distance contraction of about 2.5% occurs for cluster diameter of about 11 A [9]. In this paper we report an investigation on the energy loss spectrum in reflection mode of Cu clusters evaporated on graphite, combining electronic with surface structural techniques. The electronic information on the interband transitions is limited to the energy region O-30 eV of the elastic peak. Beyond Cu absorption edge some different structures [lO,ll] have been the I%,, observed. The Fourier transform of these structures furnishes the radial distribution function around each copper atom of the cluster, similar to that obtained from the EXAFS spectra. The most important results of the present study are (a) a sizeable narrowing of the lattice parameter (3 2% for film with mass thickness of 5 A), (b) a greater vibration of the atoms of the cluster (about three times with respect to the bulk) which reduces the multiple scattering effects occurring in the bulk, (c) the local structure of the cluster is still fee like the bulk one, and (d) the interband transitions are shifted towards higher energies. The energy loss studies have been compared with XPS and Auger studies [12]. In this way it is possible to carry out, on the same system prepared in situ, electronic and structural investigations without using different excitation sources.

2. Experimental results and discussion Pure polycrystalline graphite was cleaned and checked in a VG ESCA-3 chamber equipped with an Al K’a source for the XPS measurements and an electron gun which allows energy loss (ELS) measurements. The copper evaporations were monitored by a quartz microbalance. Fig. 1 shows the Auger signals for different film thicknesses. The formation of islands is ensured by the presence of the graphite peak also for a 50 A deposit. A careful analysis of the Cu peak around 920 eV (LW transitions) shows for the 5 A deposit a marked broadening of the structure and a shift towards low kinetic energies of about 1 eV. The X-ray photoemission (XPS) valence band spectra show a continuous shift towards higher binding (1 eV) energies of the d-band centroid. This valence shift, together with the Auger shift, is comparable with those observed in other metallic clusters. The ELS spectra were taken using a primary electron beam of 200 eV impinging on the surface at an angle of 50” in order to have a better surface (Cu clusters)to volume (graphite) ratio. Fig. 2 shows the second derivative in the range O-30 eV of the electron yield N(E), for our set of copper deposits. Fig. 3 shows the EELFS (extended energy loss

M. De Crescenrr ef al. / Cu mlands on graphire

354

graphlte

0..

0..

t

“‘“‘Vh

300

A

‘i”

1000

KINETIC Fig. 1. Auger spectra

0

500

ENERGY

of different

(eV)

Cu thicknesses

deposited

t::;;::;:::..:;;:;::.:

0

10 20 ENERGY

30 40 LOSS (eV)

on clean graphite.

Fig. 2. Electron energy loss (ELS) spectra of different Cu thicknesses on clean graphite. The primary beam energy was 200 eV. The ELS spectrum of clean Cu bulk is reported for comparison.

fine structure) spectrum, above the M core edge, for 5 .& of copper deposited sample. The primary beam energy was 2000 eV. The modulation (6 V) and the experimental settings were identical for both spectra. Modulations up to 10 VP,

IU. De Crescenri et al. / Cu islands on graphite

355

Ep=2000

eV

iii -

To w “w

NS u I II

cu bulk

0

W

‘;;

0 Graphite K edge

I..“...‘..‘.“.‘,.‘.‘.“.

50

100

150 ENERGY

200 LOSS

250

300

350

(eV)

Fig. 3. Extended energy loss fine structure (EELFS) spectra of 5 A Cu on graphite (lower curve) and clean Cu bulk (medium curve) above the M2.3 Cu core level. The upper curve is the Cu bulk EELFS spectrum obtained with a modulation of 10 VPP. The K edge of graphite substrate (285 eV) for Cu cluster is also shown.

(upper curve of fig. 3) do not show a sizeable broadening effect on the period of the EELFS features. The ELS results give information about single-particle and collective transitions which involve band levels over a wide energy range above E,. Through the loss function [lo]; N(E)cx

-Im(l/C)=~,/(~~+~:)

it is possible to relate the observed features in the transitions described by the dielectric function 2(w) = C, + ie,. The 5 and 20 A samples show the presence of all the structures and the lineshape which characterize the thicker clusters and the bulk. We note, however, a shift of some higher energy loss features (D, E and F). This observation excludes the change of the local structure of the particles and we interpret the observed peak shift as due mainly to a reduction of the lattice parameter of copper in the cluster without a substantial distortion of the fee

A4. De Crescenri et al. / Cu rsiands on graphite

/ I._

I..

,

‘-/

I,

0 Fig. 4. Fourier

transforms, F(R), of the EELFS features of fig. 3. The integral was limited to due to the presence of the K edge of graphite. The inset shows the envelope functions as obtained by back-Fourier inversion of the first peak (a) and (b). Note the shift towards low k values of A(k) for the cluster.

k max = 7.5 10-l

cage. The influence of the lattice parameter change is more sizeable for the higher energy loss transitions involving the successive Brillouin zones. The ELS results agree with the EELFS analysis reported in fig. 4. Direct structural information can be derived from EELFS spectroscopy [lo]. EELFS features above a core edge in an energy loss spectrum come from an interference phenomenon on the wavefunction of the ejected core electron which is backscattered from the atoms in the neighbourh~d of the excited atom. The theoretical formalism underlying the EELFS technique follows the same scheme as the EXAFS theory [S]. The data manipulation by means of the Fourier transform is also very similar. Fig. 4 shows the radial distribution functions, F(R), of (a) Cu bulk and (b) a 5 A Cu layer. For Cu bulk, the main peak is located at 2.20 f 0.02 A which corresponds, after the proper EXAFS phase-shift correction, to the first nearest neighbours and a broadened structure is seen at 4.32 A which corresponds to the third and

M. De Crescenri et al. / Cu islands on graphite

0

1

2

3

4

357

5

Fig. 5. (a) Theoretical F(R) obtained for 4 shells of Cu bulk using the single-scattering EXAFS formula (solid line) [8]. Experimental phase shifts are used. The dotted line shows the enhancement of the fourth Cu shell due to the focusing effect generated by the first shell [12,13]. (b) Experimental F(R) for Cu bulk.

fourth shells. Due to the reduced k range we cannot resolve the second neighbours in the F(R). For the 5 A Cu layer the first peak is at 2.16 k 0.02 A while the outer shells are located at 4.05 A. The comparison between the two F(R) gives directly a measurement of the contraction of the lattice parameter in the cluster. We evaluated a lattice parameter contraction of about 2% for clusters with mass thickness of 5 A in good agreement with EXAFS measurements [9] on similar systems. The inverse Fourier transform of the first F(R) peak furnishes the backscattering amplitude of the copper atom and information about the Debye-Waller factor. We note a shift towards low k values (3 0.7 A-‘) of the backscattering amplitude for the 5 A Cu layer compared to the bulk (inset of fig. 4). We ascribe this observation to the higher mean displacement of the Cu atoms on the cluster surface. Furthermore, we compute a theoretical EXAFS model in order to understand the role of the vibrations on the outer shells. Fig. 5 shows the theoretical F(R) obtained taking into account the first four copper shells. The calculation (dotted line) has been done considering for the fourth shell the raising of the structure due to the multiple scattering for the focusing effect of the first shell [12]. A fair overall agreement between measured and computed data has been obtained for the bulk taking into account also this effect [13].

M. De Crescenri t-1al. / Cu rslands on graphrte

358

Due to the high mobility of nearest neighbours around the absorbing atom in the cluster, the multiple scattering effect is strongly reduced because their forward scattering amplitude decreases dramatically for angles close to zero. A similar effect has been observed in the EXAFS spectra of bulk copper at different temperatures [14]. In conclusion, the combination of electronic and structural techniques furnishes a powerful tool for a complete analysis of the Cu clusters deposited on graphite. The ELS spectra show for all deposits the characteristic fingerprint and lineshape of Cu bulk. We have interpreted the shifts of some higher energy loss structures as mainly due to a narrowing of the cluster lattice parameter without structural change. The above conclusion are supported by the EELFS structural analysis up to the fourth coordination shell.

Acknowledgements We are particularly assistance.

indebted

to G. Cossu and G. Righini

for their technical

References [l] [2] [3] [4] [5] [6] [7] [8] [9]

[IO] [ll] [12] (131 1141

M.G. Mason, Phys. Rev. B27 (1983) 748. and references therein. W. Egelhoff and G.G. Tibbetts, Phys. Rev. B19 (1979) 5028. S.-T. Lee, G. Apai, M.G. Mason, R. Benbow and Z. Hurych, Phys. Rev. B23 (1981) 505. L. Oberli, R. Monot, H.J. Mathieu, D. Landolt and J. Buttet, Surface Sci. 106 (1980) 301. J. Sinfelt, Rev. Mod. Phys. 51 (1979) 569. J.-P. Borel, Surface sci. 106 (1981) 1. L. Paoletti, P.A. Rosa, M. Diociaiuti, P. Picozzi and S. Santucci, Vuoto, in press. P.A. Lee, P.H. Citrin, P. Eisenberger and B.M. Kincaid, Rev. Mod. Phys. 53 (1981) 671. G. Apai, J.F. Hamilton, J. Stohr and A. Thompson, Phys. Rev. Letters 43 (1979) 165; A. Balerna, P. Picozzi, S. Santucci, A. Reale, E. Bernieri, E. Burattini and S. Mobilio, Phys. Rev. B31 (1985). G. Chiarello, E. Colavita, M. De Crescenzi and S. Nannarone, Phys. Rev. B29 (1984) 4878. and references therein. M. De Crescenzi,F. Antonangeli, C. Bellini and R. Rosei, Phys. Rev. Letters 50 (1983) 1949. B.K. Teo, J. Am. Chem. Sot. 103 (1981) 3990. N. Motta. M. De Crescenzi and A. Balzarotti, Phys. Rev. B27 (1983) 4712. R.B. Greegor and F.W. Lytle, Phys. Rev. B20 (1979) 4902.