I am trying to refine High pressure(HP) xrd data of my sample using GSAS+ EXPGUI package. I have observed a structural phase transition,speculating that there are two phases originating at the transition pressure.the problem is the peaks of both phases are overlapping partially or completely. Refinement with single highpressure phase doesnt give a good fit but with two phases showed a good agreement values but the equation of state fit doesnt fits well due to the irregular trend in pressure-volume data, could be due to the wrong refined lattice parameters. So what can be the strategy one can follow to solve this issue? Do any constraints/restraints refinement will help? If so How can i proceed with that?
I would try two phase refinement at the same lattice constants with fixed profile functions but you should be more specific about the nature of the phase change. do you know both structures? Is it a completely new symmetry, lowered symmetry...
Thank you for the suggestions and comments.
From the literature, I got the two possible HP phases. One is the low symmetry monoclinic phase and the other is the hexagonal phase, both are low symmetry structures compared to the parent phase. can you be more specific about what it means the two-phase refinement at the same lattice constant? and what criteria i have to adopt to fix the profile function of individual phases?
Right, now I get a better picture. First of all you should be aware that if you have a dominant phase and some other phase that overlaps with the first one you can see it very easily in a good quality data. Even if overlapped, one phase refinement will not match intensities of all reflections. Just do the refinement with one, primary phase and check intensities and peak profiles. Regions where the second phase emerges will be marked by the increased mismatch on the residual line (intensity, peak broadening or shoulders...). I think you know this already so just make another step:
1) If you are not sure what kind of new structure you have you could try to index peaks and find the unit cell. Unfortunately GSAS + EXPGUI cannot help here. I would use GSAS II here but be aware that low symmetry structures will require a broad 2-Theta range to get enough reflections.
2) If you trust literature then just add one of the possible phases (starting from hexagonal one). Both phase flags should be selected. See if you can match the pattern through the refinement. Remember to uncheck the global refinement scale factor and select refinement of phase fractions (both are in the tab "Scaling"). Fix the lattice constant of the original phase (assuming stable pT conditions) and profile functions (these should be established with some standard and if really, really necessary, additionality refined). Try first refining the unit cell of the new phase using profile functions from the parent phase (it is a first guess that assumes the same crystallinity and may need some later change to correct for .e.g. strain, small crystal size). If the fraction of the new phase is too small or your initial guess is incorrect try to adjust the lattice constants manually. If peak positions agree with observations and you account for all unknown reflections then you are safe.
If you have some unfitted peaks you may want to check for a third phase (if possible) or another new phase with different symmetry. If some of the calculated peaks are not matched by observed reflections you may consider a preferred orientation (if possible) or try another model.
First, please post an image of the diffractogram so that we have an imagination about the degree of asymmetry in your reflections. To me it sounds like e.g. perovskite phase transformations, i.e., it will be more or less impossible without any "speculative" constraints. But this becomes easier if we see the pattern. The problem will be that your monoclinic phase only slightly differ from a hexagonal, e.g. by a tiny shift of only one atomic position. This forces crystallography to switch into a lower (or higher) symmetry which often also needs to describe the lattice parameters entirely different. In very bad cases you even do not recognize the cell anymore, and also the atomic arrangement within the structure looks totally different. However, in nature this usually does not happens and your diffraction pattern also demonstrates this: it does not change remarkable, i.e. the atomic positions (crystal structure) is still very similar, anyway how you have to describe this by crystallographic rules (and these are only RULES and can be changed). Therefore, send us a XRD pattern that we can give you a useful advise.
I think you can refine the parameters of the pattern step by step by using MAUD Rietveld Program. to do this you must know your phases well and have an initial guess for your phases, and respectively refine the lattice parameters, bkg, texture,scale factor. to refine the line broadening of microstructure and size- strain model, use respectively Popa LB and Popa rules.
@ H. Naji: you can refine a lot describing the observed asymmetry, broadening by different profile functions, texture, anisotropic broadening, more instead of 3 phases etc. The problem is, you have to make some assumptions which are probably wrong, and then you don't know more than in the beginning. Therefore, before starting to vary different parameters, think critically what do you really know about a sample and exclude everything you can. A better R-value does not tell you that your model is better. it only tells you that mathematically and without any meaning the deviation between theoretically curve and experimental signal is smaller. It is better to use a result of a worse R-value where you still have some imagination (belief) what's going on.
@ Andrzej Falenty,
Thanks for your suggestions Sir, i am following your comments and trying to refine it. If any problems faced i will post it here.
@Gert Nolze,
Sir, i have attached a HPXRD pattern. Phase 1(parent face) Phase 2 (Pressure calibrant) Phase 3&4 are the confusing one.