Analysis of cement using low-resolution energy-dispersive x-ray fluorescence and partial least-squares regression

Abstract

A bench-top energy-dispersive x-ray fluorescence instrument equipped with a low-power x-ray tube and a gas-filled proportional counter was used to determine CaO, SiO2, SO3, Al2O3 and Fe2O3 in cement. Spectrum evaluation and quantitative analysis were performed using partial least-squares (PLS) regression. A mean relative error of 5% or better was achieved for all constituents determined. It is demonstrated how the PLS method uses both explicit (characteristic peaks of the analyte of interest) and non-explicit information (correlation between concentrations of different species) to build the regression model. It is also shown how PLS is able to combine data originating from spectra that are recorded under different measurement conditions. Copyright © 2000 John Wiley & Sons, Ltd.