Dear Rietveld-mailing list users,
As I just subscribed this e-list, let me present myself:
My name is Edouard RENY, and I'm a fresh doctor is material Science
(I obtained my Doctor degree in France this february). Now I'm a post-doc
in Japan working on silicon clathrate structures.
the Rietveld refinement program I use is Fullprof (with a DBWD viewer),
And I have a problem in refining the X-ray peak profiles of one of my samples:
Whatever I do, the refinement leads always to Gaussian type profiles
which we can see on the .prf file, at high 2 teta, don't match at all with
the experimental profile. This sample contains some silicon, and there, no
problem to refine the
peak profiles (even if intensities are not perfect). However at low two
tetas (for 2tetas < 50deg) ,
calculated peak shapes match quite well with experimental data.
(see fig1.gif. At higher 2tetas, it's even worse)
I use a PseudoVoigt, it leads usually to eta=0.35 (mixing parameter)
approximatively.
If I fix it to more lorentzienne values (0.6 or 0.8 for instance),
residuals become too high.
U,V, W combination refining converges but if we free eta,
I always finish by having Gaussians peaks mixing the Ka1 and Ka2
contributions as one peak.
What is strange is that there is no problem for the impurity silicon phase
(peak shapes are
well defined). The profile parameters of the two phases are not commonly
refined.
Has anyone been through this problem? has anyone have an idea of refinement
procedures which would lead to a better peak profile refinement and avoid
this false minimum
where the program always seem to converge?
Thanks in advance for your help
Sincerely
Edouard RENY
