Determination of numerical data from polychromatic ESCA spectra and their verification

Abstract

The main disadvantages of commonly used pragmatic procedures for the extraction of numerical data from polychromatic ESCA spectra are discussed. It is shown that physically defined models allow a proper description of polychromatic core-level spectra, when new parameters of the excitation function (Al Kα and Mg Kα) and a new background construction are used. The new background construction consists of a modified Tougaard background, with an additional parameter N, which account for no-loss regions, and a linear background. The appropriate line shapes are voigtians (atomic spectra), Doniach—Sunjić shapes convoluted with a gaussian shape (metallic samples) and experimentally determined singlet shapes, or Kuchiev-type shapes for the case of vibrationally non-resolved spectra of free and solid molecules. The data obtained may be verified by comparison with physically well-defined lorentzian and gaussian widths, spin-orbit splitting, etc., determined separately or taken from the literature. A statistical analysis of the residual spectra (here discussed for the first time) allows for the falsification of statistically unrealiable data.