Hi Tony,
Sorry, I know the documentation on this is not very good.
In that "Results" page for the XRF fits, the reported values for "Amplitude" will have an arbitrary scale. For example, we don't try to accurately account for the solid angle of the detector (sort of challenging for multi-element detectors) or bother trying to get a value for the incident flux (which, I do believe we do pretty well with now).
But: there should be (maybe "ideally") a single scaling factor from those reported Amplitudes to elemental concentration (in the thin-film limit) for the various elements. That is, we try hard to account for all the efficiencies, cross-sections, absorption, and attenuation factors (well, again, in the thin-film limit, so not accounting for self-absorption).
So: if you know the concentration of one of the elements in ppm or the areal density in ng/mm^2, you can enter it there: "I know that Fe is 10,000 ppm", for example. That should then give you decent values (say, factor of 2) for the concentrations of the other elements (well, I would not trust "Ar", as that's from the air).
With a fit to an XRF spectrum extracted from an XRF Map (your screenshot looks like it is), you can apply that scaling (and use the eigen-vector energy responses for each element) to decompose the XRF map into maps of abundances for the elements you have fit (in your scaled units). This can be a great help for strongly overlapping lines, and gives something like reasonable concentrations. The decomposition is way faster than trying to fit each pixel (which will be pretty noisy anyway). The main caveat is that it tries to linearize pileup, which is, of course, not linear. So, if the pileup changes dramatically in a map, the elements with peaks near the pileup will still have artifacts. I think that might be unavoidable.
We use this often on samples where we have decent abundances, and it sort of works reasonably well in the thin film limit and for the thin-film standards we have. I have some code to account for self-absorption, but it needs an overhaul. Doing this correctly would be great, but my To Do list is long ;).
Let me know if you have any more questions,
--Matt
________________________________
From: Tian, Tony via Ifeffit
