Two-step detection of analytical signals in X-ray analysis at an unknown background intensity

The growing interest in digital methods for treating analytical signals was prompted by the need to automate observations and by the growth in the number of the latter. The possibilities of detecting traces of chemical elements are usually restricted by the loss of visual signs of the analytical signals and the subjective capabilities of the observer. The detection and measurement of analytical signals are usually improved by means of linear and nonlinear filtration, least-squares smoothing, differentiation, Fourier transformation, the method of moments, clustering, and determination of the principal components of the measurement matrix. Such preliminary treatment not only does not solve the problems of detecting traces of chemical elements in the samples investigated (qualitative analysis) and of detecting differences in the chemical composition of those samples (signature analysis), but can significantly alter the original analytical information. In particular, in the X-ray spectrometry of multicomponent samples with complicated structures in their characteristic X-ray spectra due to matrix effects, where the possibilities of identifying weak signals, deconvoluting them, and compensating for the background are restricted, direct statistical interpretation of the spectrometric data obtained in the information system consisting of the sample being investigated and the analytical instrument becomes necessary