Quantitative electron microprobe analysis of flat polished samples under conditions of normal electron beam incidence is a relatively mature field. When crystal diffraction spectrometers are used serious peak overlaps do not occur, reined versions of the ZAF and φ(ρz) methods now make it possible to do analyses with better than 2% relative accuracy for most major components in a specimen. Unfortunately, the validity of these methods is more questionable when applied to situations of non-normal electron beam incidence or in the examination of rough surfaces, thin films or porous materials, In these cases Monte Carlo calculations offer a lot of promise because geometrical effects can be more easily and completely included. The accuracy of Monte Carlo calculations has not, however, been studied extensively. It was felt that it might be useful to test the method against high quality data sets collected with the conventional normal incidence geometry described above. This type of data has been used extensively to refine ZAF and φ(ρz) methods, and a good collection can be found in the work of Pouchou and Pichoir, and also in the extremely valuable summary of work on gold-copper and gold-silver published by NBS (now NIST). Myklebust et. al., in fact, later did some comparisons of Monte Carlo calculations with the NBS data. The current study provides additional comparisons with a new program developed by Gauvin and Lifshin.
展开▼
