Estimating Site Coordinates/Positions when they're not given for QPA

Ramon13

Hi all,

When I'm doing Phase Identification (Using Bruker's DIFFRACEVA software) before QPA on Topas Launch Mode I've noticed that many of the good fitting phases don't have sites displayed in their .str of .cif files. Because of this, there are some phases which I can't use for QPA despite them being great fits otherwise, instead I have to find phases that have sites given as a condition for me being able to do QPA.

I'm just wondering if there's something that can be done that's not too difficult to approximate the sites for a given phase assuming they're not given? Is there some method or estimation for a general phase (e.g. inferring from other phases or doing some other type of refinement to estimate sites (like a LeBail is for unit parameters) etc.)? I presume that if there is such a method it wouldn't be possible for something like a complex solid solution.

Kind regards,
Ramon

Ramon13

I've just realised, mostly from my other thread, that in place of a regular Rietveld refinement with a structure (with the site coordinates and positions), I can just get the phases with no structure and refine them with a LeBail, Pawley or PONKCS method; correct?

johnsoevans

You won't get quant values directly with LeBail or Pawley, but, as you say, there are various tricks like PONKCS you can apply. If there are phases without coordinates in the databases you're looking at, it may be that they're available in other databases like the ICSD. Or it may be that they are related to other phases where the coordinates are known. Using those coordinates might be a more straightforward approach then PONKCS.

Ramon13

johnsoevans In your Rietveld video you use the num_posns function to get the Wyckoff multiplicity/position coefficient of each site and to confirm the composition of the unit cell with the ISODISTORT/SPACEGROUP site.

In the case where a phase has been selected which fits well but does not come with site information/coordinates and cannot be relatively determined via PONKCS, I was wondering if it's reasonable and possible to get the site information for the given phase from ISOSPACEGROUP and refine certain parameters (that would make theoretical and practical sense, I don't just want a good looking fit) within the site information for a good approximation. Any ideas?

rowlesmr

You can do DDM. Use the peak positions and intensities from the card as your constraints. Fit a scale factor to scale everything up and down. Have some sort of peakshape function. Use the pattern intensity as the basis for quant.

https://topas.awh.durham.ac.uk/doku.php?id=direct-derivation_method_qpa