SURFACE AND INTERFACE ANALYSIS Surf. Interface Anal. 2002; 33: 410–413 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/sia.1224 Experimental estimation of surface excitation parameter for surface analysis G. Gergely, 1* M. Menyhard, 1 S. Gurban, 1 A. Sulyok, 1 J. Toth, 2 D. Varga 2 and S. Tougaard 3 1 Research Institute for Technical Physics and Materials Science H-1525 Budapest, PO Box 49 Hungary 2 Research Institute for Nuclear Physics, Hungarian Academy of Sciences, ATOMKI, PO Box 51, H-4001 Debrecen, Hungary 3 Physics Department, University of Southern Denmark, DK-5230 Odense, Denmark Received 7 September 2001; Revised 14 December 2001; Accepted 8 January 2002 The surface excitation produced by impinging or escaping electrons is a competitive process to elastic backscattering. It affects the intensity of Auger and XPS peaks and is characterized by the surface excitation parameter P se . This appears in the surface loss peak I.E pls /, and possibly a surface plasmon. In our work P se is defined as the ratio of the probability to create a surface plasmon/elastic scattering and is deduced from the integrated surface loss peak and elastic peak, respectively. Our procedure is based on reflection electron energy loss spectroscopy and elastic peak electron spectroscopy spectra and Kl i spectra (K is the inelastic scattering cross-section and l i is the inelastic mean free path), determined using Tougaard’s method. The normalized Kl i curves are fitted to the spectrum at E = 5 keV (primary energy) and E L = E pl1 (volume plasmon), approximated with the Lorentzian type three-parameter formula of Tougaard. For E > E pl1 , the normalized curves are overlapping and P se is deduced from the integrated difference spectra fitted with the Tougaard cross-sections. The procedure was applied on materials exhibiting surface and volume plasmons: III–V semiconductors (GaAs, InSb) and In and Sb metals. The REELS experiments were carried out with an ESA 31 HSA spectrometer covering the E =0.2 – 5 keV energy range. Copyright  2002 John Wiley & Sons, Ltd. KEYWORDS: surface excitation; experiments; EPES; REELS INTRODUCTION Electrons impinging and penetrating into the solid and escap- ing, respectively, lose a part of their energy in the surface layer to produce surface excitation. This appears in several parameters: the elastic reflection coefficient r e calculated by Monte Carlo simulation using the material parameters of the NIST database and the elastic peak intensity studied by elastic peak electron spectroscopy (EPES) are larger than the experimental data, various surface losses can be observed by reflection electron energy-loss spectroscopy (REELS); and the Auger (AES), XPS, low-energy cathodoluminescence and EBIC (electron bombardment induced conductivity) effi- ciency values are reduced with respect to their intrinsic data. The surface excitation parameter P se is defined as the number of surface plasmons excited per electron moving across a solid surface. 1 In AES and XPS the spectrum adjacent to the Auger or photoemission (XPS, UPS) peaks exhibits losses, as well as showing surface plasmons in some cases. The surface excitation is strongly affected by the input and escape angle of electrons and the roughness of the surface. L Correspondence to: G. Gergely, Research Institute for Technical Physics and Materials Science, H-1525 Budapest, PO Box 49, Hungary. E-mail: gergely@mfa.kfki.hu Contract/grant sponsor: EU; Contract/grant number: COPERNICUS ERBIC 15CT96800. Contract/grant sponsor: OTKA (Hungarian Natural Research Fund); Contract/grant numbers: T026524; T030430; T030433. Abundant literature has been published on the surface excitation processes, including the surface dielectric function. In latter years some fundamental papers contain compre- hensive reviews of the literature. 1–5 Much less has been published on experiments 6–13 using EPES 7,8 , REELS 6,9,10,12,13 and angular REELS. 11 Very few P se experimental data have been published. Apart from the surface monolayer (ML), the inelastic mean free path (IMFP) is nearly constant in the bulk of the solid. 1,2 Calculating the energy loss in the first 2 ML, it is 40% for E D 0.2 keV and 9% for 5 keV on Ag. In this paper a new approach is described for experimental estimation of P se PROCEDURE FOR EXPERIMENTAL ESTIMATION OF P se Reflection electron energy-loss spectra are produced by backscattered electrons, subjected to losses either before or after elastic backscattering. Our experimental estimation uses REELS and the calculation of P se is based on some assumptions and conditions: the REELS spectrum contains some peaks associated with surface losses and our procedure is confined to materials exhibiting dominating plasmon loss peaks, the surface losses increase with the input or escape angle; 11 – 13 and the contribution of surface losses decreases with increase of primary energy E. 9,12,13 Our procedure is based on comparison of the experimental K⊲E⊳ i inelastic Copyright  2002 John Wiley & Sons, Ltd.