I would like to know if there is any tutorial available to perform multiphase refinement in GSAS. I did not find it in GSAS manual. I need to know the important strategies while refining two phases (just a hint).
In my mind you could start to refine the first phase separate from the other. This makes sense if one phase is clearly the major phase fraction. Lets say you have a phase with around 80-90% and the other forms the rest. Then you could start with refining only the first phase. In this case use the overall scaling factor first.
As the first phase is about fitted, you can load the second phase. Under "Scaling" uncheck the overall scale factor and check the factors for both phases. They start both with one. If you already have a rough estimate about the phase ratio, lets say 90:10, then change the scale factor values to 0.9 and 0.1 accordingly. Sometime it can be problematic for the refinement, if you start with the same value for both phases. Especially if one phase is clearly more abundant in your sample. In this case it can help to start with a good initial guess. (I am sure you might know but: do not attempt to refine the overall scale factor and scale factors for the different phases at once. Use one or the other or the refinement will not work.)
In the final refinement both phases have to be flaged and you have to reach convergence. If you do not reach convergence, it can sometimes happen that there is a correlation with the phase fractions and another parameter or set of parameters (i.e. background). Then I would rather uncheck the other parameter instead of a phase flag.
Just search online!
Some good material here:
http://www.aps.anl.gov/Xray_Science_Division/Powder_Diffraction_Crystallography/
and here:
http://www.fhi-berlin.mpg.de/acnew/department/pages/teaching/pages/teaching__wintersemester__2013_2014/frank_girgsdies__phase_analysis_and_structure_refinement__131129.pdf
Thanks Edward
Let me be more specific in my question. What I wanted was, I am carrying out two phase refinement. I have a confusion in phase fraction refinement. Whether flag should be on for both phases if we want to refine phase fractions or should be refined always separately. And they are strongly correlated while carrying out other refinements say profile. I have not found such information online !!!. I wanted hints on such minor things. Anyway, I will find it out soon since I recently started working on multiphase refinement.
I appreciate your reply and for the links.
Good questions!
In GSAS you are basically scaling a model to best fit the experimental data. You can refine the overall scale, and the scale for each phase. You really cannot refine each phase separately, in that they are highly correlated, but if you have an overall scale and one of the two phases scale turned on, then the refinement should work. Similarly if you have the overall scale set at any arbitrary value (say 1) with the refinement flag off and both phases scale factors turned on, that should also work. You should get the same phase fraction (in wt%) from either refinement.
Thank you Edward.
This is what I was not knowing properly.I follow your suggestions. I think it will greatly help in my refinement.
In my mind you could start to refine the first phase separate from the other. This makes sense if one phase is clearly the major phase fraction. Lets say you have a phase with around 80-90% and the other forms the rest. Then you could start with refining only the first phase. In this case use the overall scaling factor first.
As the first phase is about fitted, you can load the second phase. Under "Scaling" uncheck the overall scale factor and check the factors for both phases. They start both with one. If you already have a rough estimate about the phase ratio, lets say 90:10, then change the scale factor values to 0.9 and 0.1 accordingly. Sometime it can be problematic for the refinement, if you start with the same value for both phases. Especially if one phase is clearly more abundant in your sample. In this case it can help to start with a good initial guess. (I am sure you might know but: do not attempt to refine the overall scale factor and scale factors for the different phases at once. Use one or the other or the refinement will not work.)
In the final refinement both phases have to be flaged and you have to reach convergence. If you do not reach convergence, it can sometimes happen that there is a correlation with the phase fractions and another parameter or set of parameters (i.e. background). Then I would rather uncheck the other parameter instead of a phase flag.
Great...!!!! I am trying it out.....I will get back to you soon if I find something not clear.
Thanks Dirk Friedrich. It is a very good suggestion.
In principle the scale factor for all the phases should be varied in multiphase refinement. However the reliable phase fractions are obtained only when there is less overlap of the Bragg peaks for various phases. When the diffraction profiles are overlapping the profile shape parameters can significantly affect the phase fractions if their values are not same for both the phases.
The Rietveld refinement basically quantifies the phases in terms of area under diffraction peaks for various phases. If diffraction peaks for a particular phase are very broad it may be overestimated than its actual phase fraction.
Try to use the MAUD program to perform multiphase Rietveld Refinement.
http://www.ing.unitn.it/~maud/
- Badges
- Science method
- Similar topics
- Methodology
- Crystallography
- X-ray Diffraction