Abstract: The methodology of direct hair analysis by energy-dispersive x-ray fluorescence was studied. The effect on the XRF result of having a non-homogeneous radial distribution of the analyte in a single hair strand and the macroscopic effects in a bundle of hairs were calculated to evaluate possible systematic errors. The detection limits were mapped as a function of the target thickness and surface fraction. It appeared that a 10 mg cm−2 sample thickness, i. e. a target with about four layers of hair strands, is recommended. The standard deviation of this simple direct analytical method is 619% for some important elements. Discrepancies with neutron activation analysis had a mean value of around 15%. About twelve elements can be determined simultaneously on a routine basis.

Abstract: Expressions are presented for calculating the matrix effect and the pure particle size effect in the XRF analysis of particulate samples with a discrete particle size. The equations are based on the absorption-weighted radiometric diameter concept. Two excitationdetection geometries are considered, with the angles between the sample plane and both the incident and emerging radiation being either 90° (π geometry) or 45° (π/2 geometry). Calculations were made for different sample loadings and exciting radiation energies. The influence of these parameters on the matrix and pure particle size effects is shown. From the results, it is possible to predict the performances of alternative experimental correction procedures for the particle size effect, involving dual measurements at different excitation energies or in different excitationdetection geometries.

Abstract: Elemental composition of aerosols is important to source apportionment studies and to understand atmospheric processes that influence aerosol composition. Energy dispersive X-ray fluorescence spectroscopy was applied for measuring the elemental composition of Amazonian atmospheric aerosols. The instrument used was a spectrometer Epsilon 5, PANalytical B. V., with tridimensional geometry that reduces the background signal with a polarized X-ray detection. The measurement conditions were optimized for low-Z elements, e. g. Mg, Al, Si, that are present at very low concentrations in the Amazon. From Na to K, our detection limits are about 50% to 75% lower than previously published results for similar instrument. Calibration was performed using Micromatter standards, except for P whose standard was produced by nebulization of an aqueous solution of KH2PO4 at our laboratory. The multi-element reference material National Institute of Standards and Technology-2783 (air particulate filter) was used for evaluating the accuracy of the calibration procedure of the 22 elements in our standard analysis routine, and the uncertainty associated with calibration procedures was evaluated. The overall performance of the instrument and validation of our measurements were assessed by comparison with results obtained from parallel analysis using particle-induced X-ray emission and another Epsilon 5 spectrometer. The elemental composition in 660 samples collected at a pristine site in the Amazon Basin and of 1416 samples collected at a site perturbed by land use change was determined. Our measurements show trace elements associated with biogenic aerosols, soil dust, biomass burning, and sea-salt, even for the very low concentrations as observed in Amazonia. Copyright (C) 2014 John Wiley & Sons, Ltd.

Abstract: An evaluation of the accuracy and suitability of the LachanceTraill algorithm using effective α-coefficients, and the De Jongh α approach for x-ray fluorescence analysis of borax glass beads prepared from geological material was performed. By using theoretical α-coefficients obtained from the same standard and calculated elemental intensities, the two algorithms were compared under identical conditions. When only the monitor standard is used, the LachanceTrail1 approach is found to be inferior to the De Jongh basic α method. If multi-standard calibration is performed for both approaches, the accuracies are almost identical and no significant difference is found in the results. The LachanceTraill algorithm has the advantage of being easier to implement on a minicomputer and, when used without standards, it is found to give results within a few percent of the true values, which is a useful indicator in the choice of appropriate standards for more accurate analytical work.