[IUCr Home Page]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

AW: further proposal



Dear David,

	thank you for introducing me into this group. We (Crystal Impact) are
currently developing a new database management system called 'Voyager',
which serves as universal interface between existing crystallographic
databases (ICSD, PDF, Pauling File, etc.). One of the most challenging tasks
is the development of algorithms for the automatic assignment of database
entries (describing one structure/compound from one bibliographic reference)
to distinct phases. We found that if the structure is fully determined, it
is always possible to assign entries to unique phases. But the resulting
phase identifier has - at least in some cases - an 'external' component,
which cannot be derived from properties 'internal' to the material.

	Our approach is only applicable on fully determined crystal structures
which are of course only a part of the target. It starts with the formula as
described as item 7 in your proposal. As it is completely redundant and more
informative than the chemical system, I think we don't need the chemical
system.

	The second component is the Wyckoff sequence, consisting of the space group
number and a sorted sequence of the numbers and Wyckoff letters of all
occupied atom sites, e.g. '174, k3 j3 e d b'. This descriptor is a very
convenient parameter for the classification of structure types and permits a
much finer subdivision of structure types than the Pearson symbol. A more or
less complex standardization procedure is necessary to get a unique Wyckoff
sequence, but this can be done by computer programs.

	By the way, it is true that wrong assigments of space groups appears
frequently, but we should not suppose that in general. There already exists
software which helps to find those errors and it should be applied
systematically.

	In rare cases we find compounds with the same chemical composition and the
same Wyckoff sequence but nevertheless describing different modifications
(Example: ICSD entries 24674 and 27116; both H7 N O6; Wyckoff sequence: 19,
a7). In these cases we need a third, empirical component for the phase
identifier, e.g. 'H7 N O6, 19, a7, I' and H7 N O6, 19, a7, II'. The
resulting identifier allows the selection of entries belonging to the same
phase and this is the key to use databases like the ICSD as 'phase-oriented
databases'.

	So as a conclusion I propose to consider the Wyckoff sequence as additional
parameter for the phase identifier - maybe as a replacement for the Pearson
symbol, since the informations are redundant.

	We are currently at the implementation and testing stage for the automatic
phase assignment. It will be interesting to see, how much it can reproduce
e.g. the assignments in Pauling File which are made by editors. I will
report on the results when they are available...

	Best regards

	Michael

------------------------------------------
Dr. Michael BERNDT
Crystal Impact GbR
Postfach 1251
D-53002 Bonn
Germany
Phone: +49 - 228 - 9813643
Fax: +49 - 228 - 9813644
Email: berndt@crystalimpact.com
http://www.crystalimpact.com
------------------------------------------


Reply to: [list | sender only]


Copyright © International Union of Crystallography

IUCr Webmaster