Alan Hewat wrote:
>Actually the ICDD database contains many more patterns, and contrary to
>what was written, does contain calculated patterns from most of the ICSD
>database as well (50479 entries) *including* single crystal entries.
PowBase is playing the same game as PDF-3 (not PDF-2).
As already discussed at the sdpd mailing list :
http://www.egroups.com/group/sdpd/157.html?
About PDF-2 AND PDF-3, John Faber, from ICDD, said in that mail :
>The ICDD maintains the PDF-2, that contains information on approximately
>120,000 powder patterns. In addition, the ICDD is compiling a database of
>full patterns (the database is called PDF-3). At last count there were
>approximately 3,000 full patterns in this forming database. We also have a
>large number of NIST 1976 corundum datasets that were taken as reference
>patterns.
Nobody said (even not me :-) that PDF-2 was containing 3000 patterns.
>BTW Armel, calculated powder patterns for any of the 50479 ICSD structures
>can be generated via WWW (by academics within the subscriber countries),
>and everyone else has free access to the ~2000 entries on our demo server.
Calculated powder patterns are only interesting for search-match
purpose, completing the PDF-2 database. This has been done now.
Interest in experimental raw powder patterns like in PDF-3 or
PowBase is rather different :
- As said by Helio Salim de Amorim, researchers or teachers in
some countries have no access to synchrotron or neutron data which
may serve as material for education. Seing really data with the highest
resolution is better than to only speaking of them.
- A calculated pattern will never show difficulties in fitting. An
experimental pattern does ;-) In PowBase are many real cases of
size and/or microstrain, turbostratic (etc) effects. Trying to fit such
typical data with realistic physical models may interest developers of
Rietveld method software.
- Real examples of published structure determinations by powder
diffractometry may also serve for teaching purposes.
- Even some unsolved structures are in PowBase. People may
want to offer such unknowns to the public (note that PDF-2
gathers a lot of unindexed compounds, this is similar, but having
the complete powder pattern is necessary for trying to do
something).
- If methods evolve, on can do something to improve previous
results if the full pattern is available - which is not possible if
only dobs and Iobs are available.
- etc, you may propose other reasons for helping John Faber
to find money supporting PDF-3 :-).
As you already know, diplomacy is not my best talent.
How to explain with diplomacy that a database (PDF-2 from
ICDD) was (is still) bad, and that the recent improvements
(including calculated patterns from ICSD) could (should)
have been made much before ? Probably by staying silent.
However, telling nothing was clearly not a good option, since
nothing occurred during so many years up to 1998 ! I started
criticism in 1994 only. Many ICDD members started to
criticize much before me, but they were in minority,
unfortunately.
Well, let us continue with another example of bad practice
by ICDD, as promised.
Yesterday, I had to find if crystallographic data
existed about bethanechol chloride. I found quickly the
bethanechol=urecholine=carbamyl-beta-methylcholine
molecule shape on the Web. I did not find any crystal
structure for bethanechol chloride in CSD (If I am
wrong, please tell me). Finally I looked in PDF-2 and found :
43-1748 ICDD Grant-in-aid (1992) - Israel
46-1964 ICDD Grant-in-aid (1995) - Greece
None pattern indexed. Both clearly the same compound.
So, ICDD paid twice ! Is that good practice ? And how
the pattern quality of an unindexed compound can be
checked ? Both patterns have clearly different zeropoints
and discrepancies about more that 50% on some intensities.
Frequently I encountered such double entries (or more) showing
that a new entry was even not checked against the database, in
order to verify if it was not already inside...
Note that single crystals of Bethanechol chloride exist, and that
the compound is sold by many Chemical Companies. But if no
single crystal had been available, I would have been interested
in finding the full raw powder pattern in PDF-3 or PowBase, just
for trying to solve it, you see my point ?;-).
Best,
Armel Le Bail - Universite du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France
http://www.cristal.org/
