Hi everyone,
I'm relatively new to working with EDX spectra but currently we're using a SEM setup with an EDX detector to analyze both bulk materials and thin metallic films. Our usual workflow involves calculating K-Ratios and Atomic % from standardless EDX spectra. For the thin films, we estimate the layer thickness using Stratagem, that works with K-Ratios measured at different acceleration voltages.
Now we're thinking about switching to a different software — mainly because we need scripting support to process many spectra at once, which our current setup doesn't allow. Before making that move, we want to be sure the K-Ratios and Atomic % values we get from the new software match what we're used to.
What I'm trying to figure out is: how do I actually extract K-Ratios from standardless spectra in the new program? So far, I've only been able to plug in the K-Ratios we already calculated in our old software. I know that I can calculate Atomic % via Tools > Quantification Alien > Determine the composition from K-Ratios, but that assumes I already have the K-Ratios—which I'd like to generate directly from the spectrum.
Another thing I'm unclear about is how the background is handled. I've managed to fit a background, but when I try to quantify, it seems like the program still uses the one from the "strip background" option. That gives noticeably different results compared to when I manually subtract the fitted background. So I'm wondering: how can I make sure the fitted background is actually used during quantification?
Any tips or experiences would be really helpful. Thanks!
Leon
Sorry, didn't see this for a while... Hope the answer is still relevant.
When you fit an unknown using the "Fit unknown with standards using MLLSQ..." the resulting spectra contain the k-ratios as a property. You can tabulate them using the "tabulate(...)" function (use "help(tabulate)" to get details.) If you are careful and use carefully collected standards, your k-ratios will be much better than any standardless method.
The background is handled using a filter fit. This is by far the best way. It doesn't involve modeling the background which is a major source of error in some vendor's software.