From: Jean-Marc Joubert Subject: Re: How to apply constraint with more than two variables in GSAS? Date: Thu, 14 Jan 1999 07:14:01 -0500 >> Could any of you point me to the good document about the method of applying >> constraints in GSAS? >> I am trying to figure out some complex cationic intermixing in a layered >> cuprate superlatttice compound. >> But I couldn't find a good documentation in the manual. Specially trouble >> with a constraint with >2 variables. >> For example, how do you make a constraint for something like x+2y+2z=1. >> Thanks for your help. >In a few words -- you probably can't do that with GSAS. This is not true >of all Rietveld codes, but in GSAS a variable parameter can only appear >in one constraint equation. One option that might help would be to put a >soft constraint on the total composition. > Yes you can. You have to use a trick which is detailed in (J.-M. Joubert et al., J. Appl. Crystalogr. 31 (1998) 327-332). If you have three atoms on the same site and want to refine occupancy factors (theoretically it is only possible if you have two data sets, for example X-ray and neutrons), you have to introduce a fourth fictive atom : You have, for example : Atom 1, Ni, occupancy x Atom 2, Mn, occupancy y Atom 3, Co, occupancy z and wants to constrain x+2y+2z=1 You introduce : Atom 4, Ni, occupancy u and change occupancy of atom 1 to v in such a way that u+v=x Set the constraints on the parameters shifts : du=-2dy dv=-2dz You will maintain by this way the condition u+v+2y+2z=x+2y+2z=1 Regards. Jean-Marc Joubert --------------------------------------------------------------- Jean-Marc Joubert Laboratoire de Chimie Metallurgique des Terres Rares CNRS - UPR 209 2-8 rue Henri DUNANT, 94320 THIAIS CEDEX - FRANCE Phone: 33 (0)1 49 78 12 11 Fax: 33 (0)1 49 78 12 03 Email: joubert@glvt-cnrs.fr