Sanity check of PDF fits

[Xiaojian Bai]

Hi all,

I’m new to PDF analysis and have a question about a basic sanity check. Suppose I generate simulated diffraction data from a CIF file and then use PDFgetX3 to obtain the corresponding simulated PDF. If I fit this simulated PDF in PDFgui—both over the short-range and the long-range parts of the pattern—should I expect the results to be exactly identical? My understanding is yes, since the average and local structures are the same.

However, in practice I notice that the refined parameters, such as atomic positions, show a slight trend as the fitting range increases (even though the absolute changes are small). Does this behavior make sense, or is it an indication of something I might be missing?

Best,
Xiaojian Bai

[Andrew Yang]

Hi Xiaojian, Welcome to PDF analysis!

When you compute a PDF with a finite Qmax you have termination ripples. This occurs as imposing a finite Qmax is equivalent to multiplying a box function to your F(Q). Your resulting G(r) will thus be convolved with a sinc function.

These termination ripples are quite prominent in the low-r of bulk materials. In the article below, you can also see that termination ripples broaden the low-r peaks more than the higher-r ones.

https://web.pa.msu.edu/cmp/billinge-group/programs/discus/pdffit_jac.pdf

You can check if this is the case by simulating the diffraction patterns with a higher Qmax and seeing if the trend still persists.

-AY