At 05:56 PM 10/28/98 -0800, you wrote:
>Dear All,
>
>When using GSAS to refine and analyze multiphase (2 phases in my case)
>powder data, is that "crystal phase fraction" the relative molar fraction
>of one phase? If so, do I need to put constraints to make sure the sum is
>one? If not so, how do I calculate the phase molar ratio from the
>parameters used by GSAS (I have a rough idea that some scale factors have
>to be weighted by mass absorption and other things to get phase molar
>ratio) ?
>
>I am relatively familiar with GSAS refinement of single phase data, but
>have never done multiple phase analysis and would like to listen to some
>one with experience before rushing to "expedt". Any general comments,
>advice and hints would also be greatly appreciated. Thank you very much
>and in advance.
Dear Tao (and others),
The phase fractions are proportional to "mole fraction unit cells" of each
phase in the sample. They need not sum to unity but can be rescaled by hand
if desired. The .lst file output from GENLES converts the phase fractions
to wt. fractions based upon the actual atom types, positions and atom
fractions given for each phase. The "phase composition" given elsewhere in
the .exp file is not used for this purpose. If one has just two phases the
best refinement strategy is to refine the histogram scale factor(s) and the
two phase fractions with constraints that make them the same over all
histograms and phases so that the sum of the phase fractions remains fixed
at unity. If one has more than two phases and one histogram (usual case for
x-ray diffraction) then fix the histogram scale to some convenient value
(1.0 maybe) and vary all the phase fractions with no constraints. They will
sum to some nonunitary value but the wt. fractions will be properly
normalized by GENLES. For multiple histograms this same strategy still
works. Fix the histogram scale and refine all the phase fractions
independently.
Bob Von Dreele
