vlad
Dear All,
What type of error (random, systematic or absolute) is the error that the Topas calculates? If this is an absolute error (the sum of a random and systematic error), can I estimate a random error knowing the systematic error of my diffractometer? For example, after Rietveld refinement of LaB6, I obtain that the value of unit cell parameter is 4.156333A. The certified value is 4.156468A. The difference is 0.000135A. The TOPAS gives an error 0.000046A. How to estimate the obtained results?
I have an additional question. It is known that at Rietveld refinement it is necessary to include either a "zero shift" or "sample displacement". I found that the results of refinement depend on what is included and differ by about 0.0002-0.0008 A. How can this be explained?
rowlesmr
The errors from the Rietveld fit are errors in the model, and don't necessarily represent anything in real life.
Re the lattice prm difference: What temperature was the data taken at? Could it just be thermal expansion?
Re zero point/sample displacement. It isn't necessary to include either, and definitely don't refine both at once. If your instrument is aligned well, you either know the zero error, or it is =0,, and you don't need it. If your gonio is aligned well, and your sample height is correct, then you don't need specimen displacement.
By including these prms in your refinement, you are giving an extra degree of freedom to your peak positions. If you aren't correctly modelling absorption, or other things that affect peak position, then those errors can be masked by a specimen displacement.
Matthew
vlad
Thanks for your reply. But I can not fully agree that "the errors from the Rietveld fit are errors in the model". Perform, please, a simple experiment. You have a model (you have a CIF). Create a XRD pattern based on this CIF (model!) and process it using Rietveld software. The errors ,which the your software will calculate, will not be zero...Concerning the alignment of the diffractometer.... I think that every measurement is a "new" measurement. Therefore, we cannot assume that nothing (alignment of the diffractometer, sample displacement, etc.) does not change.In other words, I do not think that we can completely eliminate a systematic error.
Vlad