Identité Message:<[EMAIL PROTECTED]>
X-Sender: [EMAIL PROTECTED]
X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0.1
Date: Tue, 04 Jan 2000 15:37:04 +0100
To: [EMAIL PROTECTED]
From: Armel Le Bail <[EMAIL PROTECTED]>
Subject: Re: ICDD discussion
References: <[EMAIL PROTECTED]>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Long reply - delete if you have no time !-).
Dear Mr. Lengauer,
You are a representative ICDD member which
was happy with that PDF-2 database containing 59800 "active
patterns" in the 1994-95 release, up to set 44. You apparently
do not care about the notion of database completeness that is
a constant dream of analysts who observed >200000 crystal
structures determined at the same time. Thus, ICDD
represented less than 1/3 of what it should have been if a
correct job had been made.
I am not a member and was and continue to be unhappy with
PDF-2 (though, ICDD proposed to me several times to become
so, and the first time they propose was the first time I critisize
PDF-2, in a paper submitted to Powder Diffraction -1994 - which
had difficulties to be publish, hence my permanent angry - I may
resort the referee comments, just for kidding more ;-). I have
some doubts in the efficiency of an organization obtaining high
percentage agreement from its members by fast hands up voted
projects, as I could see at the Denver X-ray Conference, 1998.
I guess too that I will be unhappy with PDF-3 and try to
produce an alternative for free :
http://sdpd.univ-lemans.fr/powbase/
I do not understand how industrials can be satisfied if academic
researchers are not. But I can see how the "nonprofit" ICDD
could consider not necessary to give satisfaction to academy if
industrials are the main clients (how can you expect that an
analyst working for an industrial could complain ? He would
be fired...).
I would be curious to read the details of the secret arrangement
between ICDD and ICSD or CSD. What exactly are doing ICSD
and CSD when they add one entry in their database. Do they
have to pay taxes to Elsevier or other publishers ? Or do they
just copy for free the atomic coordinates and crystallographic
characteristics ? You tell me now that ICDD has signed a
promise to not add atomic coordinates. Well, this is not
necessarily the interest of the user which may have to buy
both databases. I continue to believe that it would have been
of great interest to users to have a unified database, to be paid
once. DVD technology allows the storage of the whole stuff
on one disc.
>Concerning the aggreement with the FIZ (ICSD) one should note,
>that the ICDD is not allowed to add the single crystal data to their entries.
Well, this is surely the interest of FIZ. But if the copy can be made
for free, then ICSD may (should) have made the choice to do the
same. In fact, if ICDD had made that choice when it was founded
in 1941, as the "Joint Commitee on Chemical Analysis by X-ray
Diffraction Methods", then none of the ICDD and CSD databases
would have existed, very probably. We would have in hands ONE
crystallography database, complete... Was there some war between
powder and single crystal specialists that precluded this to occur ?
So, please don't ask me to admire the choices that conducted us
where we are, and to be happy with the current situation, which
I recall, is presented by ICDD as a great advance !!! Sure, it is,
but so late, and in what conditions...
>Besides that, it is useless to compare the production rates of single crystal and
>powder diffractionists:
I really disagree when the question is a database for search-match
recognition.
>- In many cases it is easy to seperate a suitable single crystal out of a
>synthesis product, however, it is a different story to establish a reproduceable
>synthesis route to obtain several grams of a specified powder material.
I do not see your point. Why this could justify to do not have a
complete database ? That is simply unjustifiable, don't try to do it.
>- It is known, that many entries in the ICSD database are single crystal data
>from non-ambient conditions.
They are interesting data too because powder diffraction may also
be sometimes under non-ambient conditions.
>- For many organic and organometallic materials, e.g. large cells, the simulation
>of powder patterns is possible, but I do not think that in these problems a
>routine powder diffraction is the right choice.
I think the contrary. Search-match process remains very efficient
even for quite complex compounds.
>People always forget, and I have the impression
>You are one of them, that the submission of a PDF entry is a part of a quality
>control process, i.e. each step of the process should be standardized and as
>reproducable as possible.
Clearly, more than 70% of the 59800 patterns proposed in
1994-95 were not submitted to that quality control process.
I can show you an example taken in PDF-2 each day up to
my retirement. Let us just start with the files concerning
[Pd(NH3)4]Cr2O7. Lot of info about it :
39-1422 triclinic cell, huge preferred orientation
40-1486 was exactly the same as 39-1422 up to the
structure publication from powder data. Then,
that file was modified in a curious way.
The correct monoclinic cell was added, removing
the false P1. Accordingly, the new indices hkl
are given. But the old Iobs and dobs are kept,
and used to recalculate an F30 = 6 which is
quite bad if compared to the F value given in the
publication. But ICDD did not wanted of the
dcalc and "Iobs" produced by the Rietveld method.
They absolutely preferred those badly estimated dobs
and Iobs produced by peak fitting or second
derivative method. The paper authors (including
me of course) said that they would not deliver such
values because of the compound complexity and
high overlapping degree which make them inaccurate...
85-1942 : fortunately, that card from ICSD characterizes now
well that compound.
So, is that good ICDD practice ? Not understandable to me.
>5. You wrote, You collected about 500 patterns in the last two decades, why didnt
>You submit them to the ICDD. It is known to the scientific community, at least
>for 10 years, that anyone is cordially invited to submit GIA-proposals of his/her
>substances, provided that they are new.
Well, I am in the current PDF-2 (including ICSD ;-) for near of 70 entries.
My point has always been that calculated data had to be included in
PDF-2. Each time I submitted something to ICDD, it was
calculated (Rietveld) data, despite I possessed a raw powder pattern
(which ICDD don't wanted) and I had dobs and Iobs estimated by
classical derivative or individual peak fitting methods : such
Iobs and dobs are simply less reliable than "Iobs" and dcalc obtained
from the whole pattern fitting with structure by the Rietveld
method. You would be convinced if you tried to determine a
structure from powder d
