Background subtraction: II. General behaviour of REELS and the Tougaard universal cross section in the removal of backgrounds in AES and XPS

The general properties of reflected electron energy loss spectroscopy (REELS) data from 59 clean metals using a 1 keV electron beam are analysed. The average loss spectrum is close so that for a Tougaard universal cross section with B=681.2 eV2 and C=355.0 eV2. However, individual spectra differ markedly from this but may be approximated by Tougaard backgrounds with a characteristic centroid energy, Ē, in the range 12–50 eV, which depends on atomic number, Z. A defined dependence is found for Ē (Z/NA)1/2 on Z, where NA is the atomic density of atoms of A, with the function rising up to Ba, falling through the lanthanide series and rising again at Hf. An excellent correlation is found between Ē and the intensity of the REELS background for 90 eV loss energy. These dependencies permit the correct Tougaard background to be defined to give the true peak area in AES or XPS for any Z. Measurements for Sn in which the surface plasmon intensity may be clearly measured, show that the surface plasmon intensity increases linearly with the electron path length in the surface region, whereas the intensity for the bulk plasmon correspondingly falls. This anticorrelation leaves the overall spectrum shape for higher energy losses approximately unchanged so that the REELS data are a good approximation to the loss spectra occurring in AES and XPS, despite the double sampling of the surface in REELS. Measurements for the effect of the incident electron beam energy show that the ratio of the total single plasmon intensity to that of the elastic peak is insensitive to the beam energy for energies in the range 120–1900 eV. Furthermore, the relative intensity of the surface and bulk plasmons is only weakly dependent on the beam energy over this range. This supports the view that 1 keV beam energy REELS data may be used by deconvolution, to remove the inelastically scattered electron background over the full energy range of traditional AES and XPS data. These results give a clearer basis for the use of the 1 keV beam REELS data and of the Tougaard method in the background subtraction needed to determine peak areas in AES and XPS.