The way of entering the occupancies to Fullprof are described in the manual which you can find in the Docs folder of the program. Briefly, you cannot simply put 1 everywhere, that will work wrongly.

It says "Occupation number i.e. chemical occupancy × site multiplicity", in my case, I have 2 Ba sites with multiplicity 6, I should put 3 and 3 in each of them, following my unit formula?

No, the value should be equal to the multiplicity if the occupancy is equal to 1. So 6 and 6. Or 6 divided by the multiplicity of the general position, this is probably what your cif import procedure made for you.

Thank you very much

Few answers of following question may be useful for your query:

https://www.researchgate.net/post/How_to_define_occupancy_in_FullProf_when_atoms_are_disordered_to_higher_multiplicity_positions

Determining atom site occupancies by Rietveld analysis with XRD data is very challenging due to the influence of parameter correlations. Particularly, correlations between the site occupancies and thermal parameters are substantial due to the angular dependence of the x-ray scattering factors. The problem is exacerbated by sample-related bias due to preferred orientation, microabsorption and extinction. The severity of correlations will be reduced if we Rietveld fit with data measured to the high angle limit of the instrument.

Use of neutron diffraction will generally give far superior site occupancies and thermal parameters as the scattering factors for most atom types do not have angular dependence. And, while synchrotron diffraction data (SRD) suffer from the same bias effects seen with XRD data, the measurement of SRD high angle data of greater quality compared with XRD will help. But neutron diffraction is by far the best way to go.

In summary –

1) Plan data collection carefully to minimise systematic errors in the data. Micronising the specimen and mounting it to minimize preferred orientation effects are very important.

2) Measure XRD data to the high angle limit of the instrument. Parameter correlations usually render site occupancies from low-angle data sets useless.

3) If preferred orientation effects are unavoidable, ensure that these are carefully modeled in the Rietveld fitting.

4) SRD data are preferable to lab XRD data.

5) Neutron diffraction is highly recommended if you have access.

Brian O'Connor

22-April 2018

Thank you very much for you answer and your time.

If the March model does not work well to correct the intensities due to the preferred orientation effect in your Rietveld fit, use the generalized spherical harmonics description.