Inter-laboratory comparison of a WDS-EDS quantitative X-ray microanalysis of a metallic glass

TL;DR

This study compares WDS and EDS microanalysis techniques across multiple laboratories for a nickel-based metallic glass, highlighting the influence of experimental parameters and the inherent advantages of WDS for accurate quantification.

Contribution

It provides a comprehensive inter-laboratory comparison of WDS and EDS methods for metallic glass analysis, identifying factors affecting result dispersion and suggesting WDS's potential superiority.

Findings

Results from WDS and EDS are close to reference values.

Dispersion is larger at 5 kV due to L line use and lower intensities.

WDS generally shows less dispersion than EDS.

Abstract

We conducted an inter-laboratory study of a metallic glass whose main component is nickel. Two determinations of the mass fractions of the different elements present within the sample were asked to the participants: one at an acceleration voltage of 15 or 20 kV and another one at 5 kV. We compare the mass fractions obtained from wavelength dispersive (WDS) and energy dispersive spectrometries (EDS) and also try to find an influence of the kind of EDS detector and its entrance window, the background subtraction method, the use or not of standards as well as the quantification method. Both means of WDS and EDS mass fractions are close to the reference values. The dispersion of the results was larger at 5 kV than at 15-20 kV owing to the use of the L lines rather than K lines and to the lowest collected intensities. There is an exception with the case of boron because at the lowest voltage, the excitation condition is more favourable for the production of the K line. It appears that the dispersion of the results is larger with EDS than with WDS but it was not possible to find a correlation between the large dispersion and one of the considered experimental parameters and quantification factors. Thus, one can think that electron microprobes are inherently better for the determination of mass fractions or that the implementation of quantitative analysis must be optimized for some cases, especially in SEMs.