I have tried to refines to structure but not matter what strategy (parameter secuense to refinement), or program to refines (GSAS or FULL Prof Suite) i never obtain to chi2 value lower 9,0. By the way the appereance of the refinement look that the secuence is appropiated, beacause the profiles show to good adjust. The process do not allow refines occupancies, and only isotropic factor in several atoms of the structure.
Thaks for your help and suggestions.
The problem could be virtually everywhere. In example, it might be that everything is ok, but your data quality is so good that even minor deviations in the model result in a high chi2. On the other hand, the data might be full of strong, random intensity deviations from too large grain size in your sample. The way to tackle this is to try one refinable parameter, one correction after the other, until you find the one that significantly increases the fit quality. As a first step, I propose a LeBail fit of the data. As integrated peak intensities are allowed to refine freely in this, the only remaining source of problems are from the profile - peak position, FWHM, shape and background. These can be corrected without having to bother about structure and intensity errors. Also, the final LeBail fit marks the best possible agreement for any structure model, allowing for a better assessment of the progress.
I do not know what parameter you refine. The first you must seach-match what your phase, and you can try to refine step by step, for example background until you get significantly fit quality and with scale factor, width and shape, atom position, zero point and unit cell,
In general, already at the first step, with the only structural model with all fixed parameters you can see a good result, if the model is correct. If you have problems, first try to check the profile parameters, background, lattice parameters and zero or displacement errors. The occupancies or thermal factor could help only to adjust the intensities of peaks, but these parameters could be varied at the end of the refinement, in the last step. You can post the graph with comparison experimental-calculated pattern, so we can suggest more precise possible solutions
Hola Vladimir, que tipo de material estas trabajando?, cuantas fases tiene tu XRD? ... Enviame tus datos, hare un ajuste rapido usando el FullProf y te aviso cuales podrian ser los problemas.
Nos vemos.
Only because the opportunity of a refinement of the occupancy or anisotropic temperature factors are part of a software does not automatically mean that it will work for all phases. I would even claim that at least the anisotropic temperature factors are highly questionable for a powder diffraction pattern refinement. Please follow D. Toebbens recommendation, and if then everything looks fine start to refine isotropic temperature factors and occupations as proposed by M. Bini. Good luck!
As others have noted, do pay attention to Daniel M. Többens. I also emphasize the common problem of large grain size and add the problem of nonrandom orientations. Counting longer to get good counting statistics will not help in when you have either of those two issues - in fact chi^2 will get worse - sorry. Remember that ch^2 indicates how well the deviations observed are explainable by chance (statistical noise), so very good counting statistics with poor particle statistics will may give a poor chi^2.
If you rerun you sample, going all the way back to reloading the sample in its holder, then compare the two scans, you will get an idea if particle statistics are a problem (i.e., large grains sizes).
