Optimization of the identification of chemical compounds by energy-dispersive x-ray fluorescence spectrometry and subsequent multivariate analysis

Chemical compounds were measured by EDXRF (with three different types of instrumentation). The spectra were used to evaluate discriminant analysis classification. The chemical compounds were primarily classified by their fluorescence signals. Valuable additional information was obtained from the spectral range of the Compton and the Rayleigh scattering regions by the use of multivariate methods (e.g. PCA and RDA). An improvement over earlier work was achieved considering different parameters such as the position of the sample, the grain size of a chemical compound and the thickness of the package wall. Copyright (C) 2002 John Wiley Sons, Ltd.