> -----Original Message-----
> From: Armel Le Bail [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, April 17, 2001 8:19 AM
> To: [EMAIL PROTECTED]
> Subject: RE: Size/Strain Round Robin - 1st Report
> If the Round Robin stops on that first sample, it will be a Size Round
> Robin, not a Size/Strain Round Robin. I think that, even if some Fourier
> methods needing at least 2 well separated harmonics would disqualify,
> a more "difficult" sample should be introduced in the Round.
Well, although the specimen was nearly strain-free, strictly speaking this
was still a size-strain round robin; all the participants extracted (or
refined) strain parameters, didn't they :-)
>
> I have one with a small orthorhombic cell, volume ~300 A**3... It is up
> to you to receive 300g of that sample, self evolution of a
> commercial product
> (no heating, no human intervention, just aging). "Only" 350 reflections
> in the 15-150 3-theta degrees range, wavelenght 1.54056 Angstroms.
The plan is to continue with other samples, but I think it is a good idea to
take some time, finish up the first phase and hopefully learn something from
the results, referees, etc. to make the second phase better?
>
> But the problem will certainly be : if the reference methods cannot
> be applied, then which result will be recognized as the good answer...
>
> In other words, even if Rietveld-derived methods for size-strain
> analysis have shown some efficiency in a basic and simple case,
> can we consider that they will systematically work with the same
> efficiency in a complex case (i.e. a normal case ;-). Remember
> that most Size/Strain experts disqualified these Rietveld-derived
> methods as giving dubious results, in their opinion.
My feeling is that a comparison of different methods would not make much
sense any more. Why? Because majority of methods that can be applied in that
case (Rietveld, pattern decomposition followed by integral-breadth methods,
etc.) are based on similar premises (Scherrer for the size and Stokes and
Wilson relation for the strain). Therefore, they might lead not to very
different but similar (!) results, but we would never know how serious are
systematic errors. Several notable exceptions to this rule that are based on
a different approach (ARIT, for instance, has a potential to give different
strain values, thank you Armel:-). However, who is to say which one gives a
correct answer, because the bottom line for all is that some sort of
analytical model is presumed for size and strain-broadened profiles. Of
course, there is also problem of a true background, which would tend to be
overestimated in a cluttered pattern and consequently systematically
affecting both size and strain values.
Davor Balzar
