Hello Hamid, it seems like a bad idea to refine Beq right at the start. I would leave these to the very end, and only if your data are sufficient to warrant refining these parameters. I don't know what your sample is, what your data are like or what software you are using, so it's all rather guesswork. Unit cell first, if it is stable there should be no need to fix. Then look at any issues such as preferred orientation. Atomic positions, keeping special positions fixed, if you can, do monitor the structure with a viewer at this stage, as it can have unexpected results. I would work with heavy atoms first and fix then move to lighter and fix. BUT working with very light atoms is difficult and may not give realistic results. Leave Beq to the very end of the process. If you have data up to 90 2-theta or above -AND- you have used variable counting times to enhance the high angle intensities then you can consider refining the Beq values. If not, then don't. Look for realistic values, there should be no negative B values, values should be lower for heavier elements. 

Dear Prof. Ian J Slipper

Thanks for your message. My data is up to 90 Theta degree.

What is the reason that Beq values refinement needs high angle data ? and also by "variable counting times" do you suggest longer counting time at higher angles? 

Thanks,

Hamid

Read up on the Debye-Waller factor. You will see that it accounts for reduction in measured intensity due to atomic vibrations; because of the form of the Debye-Waller equation at the higher angles the reduction is greater. The issue is that at higher angles the signal also drops off due to reduced scattering, so in order to obtain data on the more important part of the diffractogram with regard to information on atomic displacement parameters it is necessary to increase the counting time at higher angles. This is what is called variable counting time.

 Ian J Slipper has provided excellent advice which I endorse entirely.

I also suggest that you become familiar with the underlying mathematics, especially with the approximations made and also with parameter uncertainties and parameter correlations. This will enable you to understand the underlying principles of structural refinement.

Kind regards.

Brian O'Connor