The following is an abstract for the selected article. A PDF download of the full text of this article is available here. Members may download full texts at no charge. Non-members may be charged a small fee for certain articles.

X-Ray Fluorescence Calibration with Partial Least-Squares

Volume 44, Number 6 (July 1990) Page 998-1002

Wang, Yongdong; Zhao, Xinna; Kowalski, Bruce R.

X-ray fluorescence (XRF) spectrometry is a popular method for noninvasive elemental analysis of a wide variety of sample types. Unfortunately, applications of XRF are complicated by the occurrence of interelement effects, which make the measured intensities complex functions of multiple-element concentrations. In this work, an implicit modeling method, partial least-squares (PLS), was used to correct for interelement effects in x-ray fluorescence analysis. Instead of fitting XRF peak intensities and concentrations to classical explicit models, the XRF spectra of a set of standard samples were used to build PLS models, implicitly relating the XRF spectra to the concentrations. Methods for the pretreatment of spectra are compared for a set of nickel alloy samples containing Fe, Cr, Ni, Mn, Mo, Ti, and Si. The PLS calibration method using appropriately preprocessed spectra gives the best results, when compared to the results from the literature and from a conventional explicit calibration model (MLR method).