Pamela, At 10:24 07.06.2004 -0400, you wrote: >In a perfect world this would be the case, but given that very few people >run systems that conform to the requirements for true fundamental parameters >(Bragg-Brentano with NO monochromator, mirrors, etc), then one is not really >using fundamental parameters per se, which is why using the term >convolution-based profile fitting is a better terminology.
sounds like a reasonable definition. >Some of the >effects of things like monochromators are amenable to modelling - they act >in the same way as a Soller slit, as well as having a slight effect on the >radiation characteristics reaching the detector. At least for standard diffractometers equipped with secondary beam pyrolitic graphite monochromators we did never observe such problems. You are right for single crystal monochromators (they cut the wavelength distribution profile in the feets), and of course for mirrors having unknown divergence and sometimes a dubious lateral intensity distribution. But what is a parameter to be refined in such a case? Any virtual "divergence"? This sounds a little bit "empiric". >In the case of the conventional >Bragg-Brentano, these functions are known, and their dependence on the >geometry known. For other systems is it a good start, but other, user >defined functions (or modifications of the existing ones) may be necessary >to correctly model the instrument function, which, at the end of the day is >what you need. O.K., this is really "learning" the instrumental function by fitting of empirical parameters. >I don't know if you've any experience with them, but if you can model the >effects of two mirrors on the profiles of a system (including minor errors >in alignment in both) then you are a better person than I. We never tried to invent any "divergencies" of these optics. But we tried to get the geometric profile by "learning" from a standard for a D5000 with 2 Goebel mirrors (http://www.bgmn.de/learnt.html) and also for another device using a position sensitive detector. No problem to get the instrumental profile, but a high quality standard measurement is necessary. And this must done again after realignement ... :-) >This is true - I don't know about BGMN, but Topas has a very stable algorithm. However, Topas will also use more conventional profile characteristics, and can suffer from some similar and some different problems from conventional programs. I have been a long time user of GSAS >and appreciate the stability of Topas, but one must be careful to make sure >the parameters make sense at the end of the day. Only having one or two >variables in the profile shape is a constraint in itself, additionally >contributing to a stable refinement in complex systems. > Absolutely, this is the same deep impression I got when I was coming from DBWS to BGMN. The program ALWAYS converges. But the choice of a reasonable model stays the task of the user. Reinhard Dr. R. Kleeberg TU Bergakademie Freiberg Institut für Mineralogie Brennhausgasse 14 D-09596 Freiberg Germany Tel. +49 (0) 3731-39-3244 Fax. +49 (0) 3731-39-3129
