The B factors of each element in my refinement are around 26.5. But the overall B factor is -22.8. Is it possible that the overall B factor is negative? But my refinement looks very good and Rwp is only 8.5. Could anybody give me some advice? Thanks sincerely.
As far as I understand mathematics of FULLPROF (in fact, I do not have much experience with it, I usually use GSAS), overall B is used as an average value when one does not expect that it will be possible to refine B values for every atom separately. So when you refine B's for each atom, overall B should be set to 0 and left unchanged. Otherwise equal decrease of the overall B and increase of each atom's B will give the same intensities. In your case B for each element will be around 3.7. In fact, even this value seems for me not very satisfactory, but much better than 26.5.
Thank Alexander for your directions! Could you please further clarify why B for each element should be around 3.7? How did you get this value?
Please read a Georg Will book:
Powder Diffraction: The Rietveld Method and the Two-Stage Method.
Springer, 2006.
Thank Alexander and Amitabh for your answers. I checked more literatures and found I cannot refine the B factors for each atom if I don't have the data at high Q. Just fix B factors for each atom to be 0.3 and overall B factor to be 0. But I cannot get the good refinement. Once I refine the overall B factor while fixing B factors to be 0.3 for each atom, the refinement will be better. Does this refinement make sense?
Thank all of you for your attention to my further question and I appreciate your further help very much!
Is your data HQ?
What values did you get from Rwp and GOF?
Which step/size are you using?
Thank Lucas for your response! The 2theta of my data is only up to 100 degree, so it is not high Q. If I fix each B to be 0.3 and didn't refine B factors, RWP is 25.0 and GoF-index is 3.7.
Once I fix B of each atom to be 0.3 and only refine overall B factor, I get RWP to be RWP and GoF-index is 3.7 and GoF-index: 2.3. In this case, scale is also changed, which will influence the magnetic moments after I add the magnetic phase. Could you advise me if my refinement on overall B factor correct or not?
Alternatively, should I fix overall B factor to 0 and refine individual B factors for each atom? Or, just don't refine any B factor?
I would start refinement with fixed B=2 for atoms from 2nd period and B=1 for all other atoms, overall B=0. Then allow refinement of B for atoms one by one starting from those with highest contribution (as far as I understand, you have neutron diffraction data, so refer to http://www.ncnr.nist.gov/resources/n-lengths/list.html) and check each time whether resulting values are reasonable and R's changed significantly. Depending on the quality of data and composition of your compound, 1-3 atoms will give the most important contribution, while for all others B values can be left unchanged.
I usually do not cross refine the overall B and isotropic B.
If the data quality is not of sufficient, I will fix isotropic B to be a small value.
The Debye-Waller factor B in crystal structure refinement is considered to be a dust bin meaning that all measurement and modeling errors enter in this value. So if you get negative values of B then you make assume that there is somewhere some mistake either in the intensity measurements or in modeling. The absorption correction is something that you should take care.
As practice shows, in ca, 95 % of cases the negative isotropic displacement parameters are originated from the not properly accounted absorption problem. This especially holds for Bragg-Brentano geometry. Another possible reasons could include mixed occupations, disorder, strain and size and so on. Of course it is only true when there is a fair agreement between the model and experimental diffraction pattern achieved. Concerning the dust bin: the displacement parameters (when properly accounted) deliver unique information about distribution of electron and nuclear densities and to my best knowledge there is no real alternative.
In general is not a good strategy vary both the isotropic B and the overall due to the strong correlation between them. I think that in the negative overall value are collected all the intensities problems of your refinement. Also the isotropic B are very high. Maybe you can fix the isotropic B at very low values and use the only the overall factor. Otherwise, as I do, you can use only the isotropic B, I do not like the overall B parameter.
This B is a square of some quantity. So it can not be negative.
Please can you give me the pdf file of
The Rietveld Method Edited by R. A. Young because I am not able to access.
- Badges
- Science method
- Similar topics
- Methodology
- Crystallography
- X-ray Diffraction