I am trying to refine my XRD data with Fullprof software. Although I got low values for R Bragg and Chi 2 factors, when I insert new cell parameters as phase, they don't fit all peaks. In fact, the literature data that I use for creating cif file is more reliable than my results. It seems to me I refined parameters with the wrong sequence. But why I got low values ??? How can I know that the overall refining process is good?
Dear Elnur N. Orujlu ,
If the cell parameters did not fit well, check how they were obtained, extract them again to confirm. It seems that your model fits better with the experimental data. Start to fit using a less symmetry space group description and as the fit improves, check for systematic peak absences.
There is no simple way to distinguish a reasonably good fit from one that is just a set of wrong based on R factors or other discrepancies. A large number of Rietveld indices (R-factors) have been proposed; there is no way, to pick up one that can be used as an absolute measure of refinement quality.
A good fit not always means a low Bragg-R factor. A good fit means a physically meaningful and acceptable description of the material. There are several ways to get relatively low Bragg-R factors completely meaningless.
Diffraction data are a set of intensities measured at a set of specific momentum transfer (Q) values, usually expressed as 2θ scanning. In general, the wrong space group symmetry description is responsible for such an effect. You may be used to fit your data a high symmetric set of crystallographic operators instead of a general space group symmetry.
Please read the reference below.
It is an excellent discussion on how good a fit is good enough.
https://www.cambridge.org/core/journals/powder-diffraction/article/r-factors-in-rietveld-analysis-how-good-is-good-enough/17439A1F889B689C495549A234D53682/core-reader
Best regards,
WNM
Dear Elnur N. Orujlu ,
If the cell parameters did not fit well, check how they were obtained, extract them again to confirm. It seems that your model fits better with the experimental data. Start to fit using a less symmetry space group description and as the fit improves, check for systematic peak absences.
There is no simple way to distinguish a reasonably good fit from one that is just a set of wrong based on R factors or other discrepancies. A large number of Rietveld indices (R-factors) have been proposed; there is no way, to pick up one that can be used as an absolute measure of refinement quality.
A good fit not always means a low Bragg-R factor. A good fit means a physically meaningful and acceptable description of the material. There are several ways to get relatively low Bragg-R factors completely meaningless.
Diffraction data are a set of intensities measured at a set of specific momentum transfer (Q) values, usually expressed as 2θ scanning. In general, the wrong space group symmetry description is responsible for such an effect. You may be used to fit your data a high symmetric set of crystallographic operators instead of a general space group symmetry.
Please read the reference below.
It is an excellent discussion on how good a fit is good enough.
https://www.cambridge.org/core/journals/powder-diffraction/article/r-factors-in-rietveld-analysis-how-good-is-good-enough/17439A1F889B689C495549A234D53682/core-reader
Best regards,
WNM
Dear Wagner Da Nova Mussel , thanks for your detailed answer. I am trying to learn this analysis method and it seems it will take some time.
Elnur N. Orujlu You last sentence reads 'How can I know that the overall refining process is good?' No, no, no... You're putting the cart before the horse...
You need to confirm that your experimental data are excellent/perfect long before you start playing with models (and adjusting the acquired experimental data). The idea is to accurately describe your system. Good fits do not necessarily mean that this is a faithful or accurate description of your material.
How do you decide that your data are worthy of investigation and that your apparatus is performing correctly? How can you trust the data you're processing?
Dear Alan F Rawle, thanks for your answer. The only problem relating to XRD data is its high step size. Otherwise, it is a phase pure sample. I already worked with this form of isostructural compounds and this one has quite identical reflection lines similar to other compounds with the same structure. My main problem is related to modeling. I am trying to learn Fullprof and I am totally confused while comparing calculated and measured datas.
Dear Elnur N. Orujlu ,
A large step size is a step down for the analysis you intend to do.
An empirical rule for structure analysis is to have a small step size, let say 0,01 degree 2theta at maximum, and for noise to peak good relationship, a minimum counting per step of about 10.000.
Those conditions aligned with a well-filling sample holder and a right collimated beam provide an excellent experimental condition for structure analysis.
Generally, use spinning of the sample, let say 40-60 rmp is desirable since it is hard to control preferred orientation over sample preparation, even in small particle size materials. At this particular, the best sample preparation you may get is by a spray dryer; the resulting sample will be composed of small spherical particles, ideal for X-ray analysis.
Those experimental conditions will provide the best reliable data set for structural analysis, considering you are using lab X-ray equipment.
It is not worth spending time fitting high step experimental data; in the beginning, you already lost several details in the structural description; this data set may compromise even the suitable space group symmetry correct determination and use.
I hope it helps somehow.
Best regards,
WNM
Dear Elnur N. Orujlu ,
just one correction, the 10.000 counts at minimum is at the 100% peak, not at each step.
Best regards
WNM
- Badges
- Science method
- Similar topics
- Methodology
- Crystallography
- X-ray Diffraction