Assessment of alternative methods for analyzing X-ray fluorescence spectra

Abstract

When analyzing characteristic peaks in X-ray fluorescence (XRF) spectra, the peak area is the value most often used to quantify peak size. However, some studies have reported the amplitude of the peak instead of the area. When the width of the peak is allowed to vary from trial to trial in order to provide the best possible fit to the data, these two alternative methods can yield slightly different results. In the current study, these two approaches to peak analysis are compared for data obtained from bone reference materials having certified lead concentrations of 1.09 ± 0.03 μg/g, 16.1 ± 0.3 μg/g, 13.2 ± 0.3 μg/g, and 31.5 ± 0.7 μg/g. Measurements were made with an Olympus Innov-X Delta Premium portable XRF system. Using both the area and amplitude methods, lines of best fit were constructed for the lead Lα and lead Lβ signals as a function of lead concentration. Additionally, coefficients of variation were calculated for each reference material and condition of analysis. To assess possible variations over time, the procedure was performed at two points separated by about one year. The amplitude and area methods were found to produce results which were consistent and proportional. Using either method, lead XRF signal plotted as a function of known lead concentration produced adjusted r2 values of ~0.99. The amplitude method provided slightly higher adjusted r2 values overall. Coefficients of variation were generally very similar between the two methods, although more pronounced differences emerged from measurements of the lowest concentration reference material.