Dear Santu,
GSAS being the most advanced and the most updated Rietveld refinement program (from my "narrow" point of view) is not quite user friendly -- it may not always converge. If too many parameters that correlate to so degree are refined it may diverge. Therefore parameters should be refined/added in some sequences which cannot be strictly defined, and therefore some experience is needed. Roughly its something like this:
1) before refinement make sure that position of calculated reflections is as much as possible close to the observed ones (or at least in the vicinity);
2) calc. peak width is more or less close to observed.
3) refinement order (very approximate, when model is almost complete):
background & scale;
then + cell parms & sample shift;
+ peak shape parameters (1 or 2 usually LX or LY and then LU),
then care should be taken about pref. orient., absorption, etc.
+ xyz for heavy atoms
+ Uiso common for all atoms
+ xyz of lighter atoms,
+ more profile parameters, anisotropy of PSF, etc.
+ pref. orientation (perhaps 2nd axis or spherical harmonics if complex)
etc.
There are some tutorials in GSAS & EXPGUI (graphic user interface for GSAS by Toby Brian) and on-line at http://www.iucr.org/iucr-top/comm/cpd/rietveld.html.
Also the recently published (by myself and V. Pecharsky) book on powder diffraction (see http://www.wkap.nl/prod/b/1-4020-7365-8) contains numerous examples on GSAS Rietveld refinement emphasizing sequence or order in refinement/adding parameters.
Of course there are possible other causes of refinement instability, e.g. non-centrosymmetric group usually requires fixed origin, accidental pseudo-relation between various parameters (e.g. cell dimensions), etc. All that should be considered specifically for each case...
Peter Zavalij
University Crystallographer
Institute for Materials
Research
and Chemistry Department
Binghamton University, SUNY, Vestal
Pkwy, East
Binghamton, NY 13902-6000, USA
Tel:
(607)777-4298 Fax:
(607)777-4623
E-mail:[EMAIL PROTECTED]
http://materials.binghamton.edu/zavalij
From: santu chakraborty [mailto:[EMAIL PROTECTED]
Sent: Friday, July 09, 2004 1:25 AM
To: [EMAIL PROTECTED]
Subject: Problem: GSAS
Dear all,We are trying to solve a powder crystal structure containing Praseodymium ( At. No. 59) as heavy atom. We have index the powder pattern by using the programme TREOR and DICVOL . Both gives the same results with a very figure of merits ( 23, 45). We have got the Pr co-ordinate by using the programme TOPAS and it supports the profile fitting in TOPAS. But the problem is when we are trying to refine this structure using this Pr coordinate with the help of programme GSAS, it's unable to refine and the refinement diverges.Can any one tell me how can we proceed from here with GSAS or what is wrong with us ?Thanking you in advance.Santu ChakrabortyDept.of PhysicsJadavpur UniversityKolkata-700032IndiaYahoo! India Careers: Over 50,000 jobs online.
