Prof. Ray Young wrote :
>-We leave that question open. Some people do still feel that way.
>Consequently, a really good pattern of a really well characterized material
>would be considered very seriously even if a pattern calculated from
>single crystal data were already in the file. Further, the fact that a
>structure has been reported as determined from single crystal data does
>not guarantee that the material is correctly or sufficiently characterized.
> Further, there are all sorts of departures from ideal stoichiometry that
>might be of great interest and would result in a distinguishably different
>powder pattern.
Seems to me that powder diffractionists in the 1950-70 years, had too
much confidence in powder diffraction : much more than in single
crystal diffraction results.
This is the only reason I see for having waited so long for the
systematic inclusion of calculated powder patterns in PDF-2,
thus delivering a poor uncomplete database to real users all those
past long years. Those ideas still continued to dominate up
to the 80-95 years !
That some people still feel that way is not really a surprise.
However, we are scientists, in principle. So, we should try
to demonstrate the powder diffraction superiority against
single crystal results, if really this is what we believe :
1-Better estimated intensities ?
2-Better estimated cell parameters ?
3-Diffusion and line broadening that cannot be seen from
single crystal data ?
That may be true for old single crystal data and/or poor quality
single crystal data jobs if compared to the recent highest quality
powder diffraction jobs. But nowadays, plate detectors produce
a global image of the single crystal allowing to see diffusion
as well as line broadening.
Or is it antagonism between two groups of crystallographers
that produced that long belief ? I am pretty sure that some
powder diffractionists are still angry about those ICSD and
CSD data in PDF-2. But you know what ? Almost each time
I perform a search-match process, the ICSD data come first,
and the identification is confirmed, from those single crystal
data, not from powder data (I would say that this is observed
in a large majority of cases).
I would be interested in hearing about different feeling when
phase X exists both as pure old fashion ICDD data and pattern
calculated from ICSD data in the new PDF-2 file : failure
of matching with ICSD data and success with old pure powder
data.
Was a systematic comparison produced between data coming
from ICSD and old PDF-2 that lead to the conclusion of powder
data superiority ? Where is it ?
I think that cases where powder diffraction quality outperform
a single crystal work are anecdotal (<5%). But the contrary
certainly occurs very frequently because, for instance, of
systematic errors on intensities due to preferred orientation
which remain undetected if the structure is not refined by the
Rietveld method from those powder data (justifying that
discussion in the Rietveld mailing list ;-).
My estimation is that more than 70% of the olld pure powder
data are distorted data. I would not say thay of single crystal
data.
Last provocative opinion of the 1999 year.
Best,
Armel Le Bail
http://www.cristal.org/
