Please be more specific...because this is a scientific forum. Which XRF format ED or WD? what is your sample? what do you want to know? relative or absolute quantity of elements? sample is alloy/soft matter/or what?....Do a little bit of web search before asking.

The Software is usually available from the instrument you got the data. If done separately, first the peaks need to be deconvoluted, There has to be a calibration file from standards/certified ref. std. similar to sample. Using empirical co-eff. method the sample data is processed to give you quantification....relative or absolute depending on what you want. There are free softwares for EDXRF (QXAS: available from IAEA for non-commercial use and another one from Denmark...sorry I don't remember the name)...but for WDXRF, I did not come across any free software.

If you're having trouble finding software for easy and accurate XRF analysis, you could estimate the fractional composition of the heavier elements of a material as described on pages 141-142 of my dissertation, linked here (bedzyk.mccormick ...). Instead of using a separate calibrated reference sample like I did, you could just get the relative fraction of different elements in one sample.

If you assume that the detection efficiency (DE) of the XRF photons is nearly 100% for those elements (which should be valid for fluorescent X-rays up to 9 keV with a 350 um thick Si drift detector, like the Vortex) and that the detector is not near saturation (i.e. LTF = 1), then all you need are the total XRF intensities (found by fitting the XRF peaks as Gaussians and taking the total area) and XRF cross-sections for each element at the incident X-ray energy. These can be found for K and L XRF emission at select incident X-ray energies in the attached publication by Puri et al., and other tables and XRF cross-section estimation methods may be found elsewhere. You may also have to consider the attenuation of the fluorescent X-rays through the sample, air, or the beryllium window of the detector, which can be calculated at the link below (henke.lbl.gov, "X-ray transmission of a solid / gas"). You can also use that link to estimate the detection efficiency, which is just the fraction of X-rays absorbed through the thickness of the detection element.

Just be sure you consider everything the outgoing fluorescent X-ray "sees" between the excited atom and the detector, and how X-rays of different energies would be attenuated differently along that path.

http://henke.lbl.gov/optical_constants/

http://bedzyk.mccormick.northwestern.edu/files/thesis/McBriarty_dissertation_final.pdf

Article K and L Shell X-Ray Fluorescence Cross Sections

Thank you Very much Dear Dhrubajyoti Gupta and Martin E. McBriarty for your valuable information.

Basically you have two free choices: QXAS (MSDOS) available from the IAEA and PyMca (Open Source, Windows, Linux, MaxOS). I am the developer of the later and its web site is pymca.sourceforge.net

Yes, Armando,

Sorry I forgot to mention PyMCA is a free software too.

I thought its used for imaging/X-ray mapping in the synchrotron -u-XRF beamlines.

Is it also applicable for quantitative analysis or just spectrum deconvolution in commercial XRF systems? 

@ Dhrubajyoti: PyMca can do both, imaging and quantitative analysis. It supports many different data formats. Despite what the 2007 article says, its use is not any longer restricted to non-commercial applications.

@Armando

Thank you for the update. I have been away from XRF for 3 years now.  

So, we can use empirical coeff. methods for automated matrix corrections in PyMCA  too or only the FP method with matrix corrections using scatter peak ratios?

@ Dhrubajyoti:

For matrix effect corrections PyMca can use two approaches:

- The Monte Carlo code XMI-MSIM

- The formulas of de Boer (X-Ray Spectrometry 19, (1990) 145-154.

Since version 5.X.X, the program can refine the user supplied matrix automatically.

The current limitation is that the program does not variate the sample density/thickness (yet!). I guess that is the use you had foreseen when mentioning the use of scatter peak ratios.

@Armando

So you mean that samples are considered as infinitely thick? 

What I also meant was the "dark matrix correction" through scatter peak ratios (effective atomic number or closely representing dark matrix) in case of soft materials.

@ Dhrubajyoti:

In PyMca the samples are not considered infinitely thick. They are considered to have the thickness and density as provided by the user. That is what I meant when I said the program does not variate the sample density and/or matrix. I have plans to take into account the sample transmission if measured, but it is not done yet.

When refining the matrix you can provide the completing material, but the program does not make use of the scatter peak ratios. I know some users are working on that direction, but I am not directly involved. Perhaps I am mistaken, but I think in order to exploit the scatter peak ratios I should have a better description of their peak shapes.

"Do a little bit of web search before asking". Did I really read that? Ridiculous.

Anyway, great answer from Dr. McBriarty.

DTSA 2 it's free , just google it, and you can use it to analyze your XRF peaks

Peakaboo is also a convenient XRF analysis tool， it's free，visit this website for details .

https://peakaboo.org/