Dear colleagues, I am helping some people from the University of Frankfurt with their Rietveld refinements, and usually their powder patterns are measured on their own instrument and the refinements are excellent. This time, however, they have to work with a powder pattern that was sent to them. It is laboratory X-ray data, copper radiation, visible peaks out to 60 degrees 2theta, reasonable counting statistics, almost certainly flat plate (reflection). The crystal structure is known. For the refinement, they have TOPAS available.
The Rietveld refinement progresses very smoothly, but when the global Biso is refined, it becomes very large and negative (-4). If we allow the Biso to become negative, the difference curve is a flat line, in other words, the fit is perfect. If we fix the global Biso at, say, 3.0, the fit is still very good, but not as good as we are used to and the difference curve starts to show some small residuals. We have no experience with this, but some Google searches suggest "absorption" as the problem. We have already modelled anisotropic peak broadening and in our experience the background is modelled very well in TOPAS and does not correlate with other parameters. So our main two questions are: 1. Is absorption indeed the most likely cause of this problem or are there other possibilities? 2. How do we model absorption in TOPAS? Which keyword do we use and what are reasonable values for the parameters in the expression? Best wishes, -- Dr Jacco van de Streek Senior Scientist Avant-garde Materials Simulation Freiburg im Breisgau, Germany
