Development of a Spectral Database for Testing Quantitative Electron Probe Microanalysis of Rough, Bulk Samples

Abstract

The vast majority of applications of electron probe x-ray microanalysis takes place on specimens which deviate significantly from the ideal configuration. Classic quantitative x-ray microanalysis makes the tacit assumption that the only reason the unknown differs in emitted x-ray intensity from standards is that there is a difference in composition between them. While this seems trivial, it forces the analyst to eliminate a major non-compositional source of possible intensity differences, namely the geometric effects associated with surface roughness. An important early paper by Yakowitz demonstrated that deviations from an ideal flat surface due to surface topography could seriously degrade the accuracy of analysis. Yakowitz interrupted the grinding and polishing procedure on pure elements and alloys at various stages and measured the variance in the x-ray intensity as the probe was scanned across the surface as compared to the predicted distribution based on counting statistics.